Vol.3 N.2 - Journal of Aerospace Technology and Management by Journal of Aerospace Technology and Management

Vol. 3 N. 2 May/Aug. 2011

ISSN 1984-9648 ISSN 2175-9146 (online) www.jatm.com.br

Journal of Aerospace Technology and Management V.3, n. 2, May/Aug., 2011

General Information

Correspondence

Journal of Aerospace Technology and Management (JATM) is a techno-scientific publication serialized, edited, and published by the Institute of Aeronautics and Space (Instituto de Aeronáutica e Espaço-IAE). It contains articles that have been selected by an Editorial Committee composed of researchers and technologists from the scientific community. The magazine is published every four months, and its main objective is to show the scientific and technological research results related to the aerospace field, as well as promote an additional source of diffusion and interaction, providing public access to all of its contents, following the principle of making free access to research and generate a greater global exchange of knowledge. JATM is added/ indexed in the following databases; SCOPUS - Elsevier; CAS - Chemical Abstracts Service; DOAJ - Directory of Open Access Journals; J-GATE - The e-journal gateway from global literature; LIVRE Portal to Free Access Journals; GOOGLE SCHOLAR; SUMÁRIOS.ORG - Summaries of Brazilian Journals; EZB- Electronic Journals Library; ULRICHSWEB - Ulrich´s Periodicals Directory; SOCOL@AR - China

All correspondence should be sent to: Journal of Aerospace Technology and Management Instituto de Aeronáutica e Espaço (IAE) Praça Mal. Eduardo Gomes, 50- Vila das Acácias CEP 12228-901 São José dos Campos/ São Paulo/Brazil Contact Phone: (55)12-3947-5115/6444 E-mail: editor@jatm.com.br Order your copy (for free): secretary@jatm.com.br Web: http://www.jatm.com.br

Educational Publications; LATINDEX - Regional Cooperative Online Information System for Scholarly Journals, PERIÓDICOS-CAPES and is under analysis in other major indexing databases. JATM is affiliated to ABEC - Brazilian Association of Scientific Editors and all published articles contain DOI numbers attributed by Crossref.

Journal of Aerospace Technology and Management

Published and distributed by: Institute of Aeronautics and Space

Vol. 3, n.2 (may/aug. 2011) – São José dos Campos: Zeppelini Editorial, 2011

Desktop publishing and printing: Zeppelini Editorial

Four monthly issue

Edition: 750 São José dos Campos, SP, Brazil

1. Aerospace sciences

ISSN 1984-9648

2. Technologies

Note: This publication is sponsored by the Institute of Aeronautics and Space (IAE) to whom the copyright on

3. Aerospace engineering CDU:629.73

all published material belongs. Permission must be requested prior to use

Journal of Aerospace Technology and Management J. Aerosp. Technol. Manag. Vol. 3, Nº. 2, May - Aug., 2011

Editor in Chief

Executive Editor

Francisco Cristovão Lourenço de Melo Institute of Aeronautics and Space São José dos Campos - Brazil editor@jatm.com.br

Ana Marlene Freitas de Morais Institute of Aeronautics and Space São José dos Campos- Brazil secretary@jatm.com.br

ASSOCIATE EDITORS Adriana Medeiros Gama - Institute of Aeronautics and Space - São José dos Campos - Brazil Ana Cristina Avelar - Institute of Aeronautics and Space - São José dos Campos - Brazil André Fenili - Universidade Federal do ABC- São Paulo - Brazil Angelo Pássaro - Institute for Advanced Studies - São José dos Campos - Brazil Antonio Fernando Bertachini - National Institute for Space Research - São José dos Campos - Brazil Antonio Pascoal Del’Arco Jr.- Institute of Aeronautics and Space - São José dos Campos - Brazil Carlos de Moura Neto - Technological Institute of Aeronautics - São José dos Campos - Brazil Cynthia C. Martins Junqueira - Institute of Aeronautics and Space - São José dos Campos - Brazil Eduardo Morgado Belo - University of São Paulo - São Carlos - Brazil Elizabeth da Costa Mattos - Institute of Aeronautics and Space - São José dos Campos - Brazil Flaminio Levy Neto - Federal University of Brasília - Brasília - Brazil Gilberto Fisch - Institute of Aeronautics and Space - São José dos Campos - Brazil João Luiz F. Azevedo - Institute of Aeronautics and Space - São José dos Campos - Brazil José Márcio Machado- Univ. Estadual Paulista - São José do Rio Preto - Brazil José Roberto de França Arruda - State Universiy of Campinas- Campinas - Brazil Marco Antonio Sala Minucci - Institute for Advanced Studies - São José dos Campos - Brazil Marcos Pinotti Barbosa - Federal University of Minas Gerais- Belo Horizonte - Brazil Mischel Carmen N. Belderrain- Technological Institute of Aeronautics - São José dos Campos - Brazil Paulo Tadeu de Melo Lourenção - Embraer - São José dos Campos - Brazil Valder Steffen Junior - Federal University of Uberlândia - Uberlândia - Brazil Waldemar de Castro Leite - Institute of Aeronautics and Space - São José dos Campos - Brazil

Editorial Production Ana Cristina C. Sant’Anna Glauco da Silva Helena Prado A. Silva Janaina Pardi Moreira Mônica Elizabeth Rocha de Oliveira

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 105-108, May-Aug., 2011

105

Editorial Board

Editorial Board Acir Mércio Loredo Souza - Federal University of Rio Grande do Sul - Porto Alegre - Brazil Adam S. Cumming - Defence Science and Technology Laborator - Fort Halstead - UK Adrian R. Wittwer - National University of the Northeast - Resistencia - Argentine Alain Azoulay - Superior School of Eletricity - Paris - France Alexandre Queiroz Bracarense - Federal University of Minas Gerais- Belo Horizonte - Brazil Antonio Henriques de Araujo Jr - State University of Rio de Janeiro - Rio de Janeiro - Brazil Antonio Sérgio Bezerra Sombra - Federal University of Ceará - Fortaleza - Brazil Bert Pluymers - Catolic University of Leuven - Leuven - Belgium Carlos Eduardo S. Cesnik - University of Michigan - Ann Arbor - USA Carlos Henrique Marchi - Federal University of Paraná - Curitiba - Brazil Charles Casemiro Cavalcante - Federal University of Ceará - Fortaleza - Brazil Cosme Roberto Moreira da Silva - University of Brasília - Brasília - Brazil Edson Aparecida de A. Querido Oliveira - University of Taubaté - Taubaté - Brazil Edson Cocchieri Botelho - Univ. Estadual Paulista - Guaratinguetá - Brazil Fabrice Burel - National Institute of Applied Sciences - Lion - France Fernando Luiz Bastian - Federal University of Rio de Janeiro - Rio de Janeiro - Brazil Francisco Souza - Federal University of Uberlândia - Uberlândia - Brazil Frederic Plourde - Superior National School of Mechanics and Aerotechnics - Poitiers - France Gerson Marinucci - Institute for Nuclear and Energy Research São Paulo - Brazil Gilson da Silva - National Industrial Property Institute - Rio de Janeiro - Brazil Hazin Ali Al Quresh - Federal University of Santa Catarina - Florianópolis - Brazil Hugo P. Simão - Princeton University - Princeton - USA João Amato Neto - University of São Paulo - São Paulo - Brazil Joern Sesterhenn - University of Munich - Munich - Germany Johannes Quaas - Max Planck Institute for Meteorology - Hamburg - Germany John Cater - The University of Auckland - Auckland - New Zealand Jorge Carlos Narciso Dutra Institute of Aeronautics and Space - São José dos Campos - Brazil José Alberto Cuminato - São Carlos School of Engineering - São Carlos - Brazil José Ângelo Gregolin - Federal University of São Carlos - São Carlos - Brazil José Atílio Fritz Rocco - Technological Institute of Aeronautics - São José dos Campos - Brazil José Carlos Góis - University of Coimbra - Coimbra - Portugal José Leandro Andrade Campos - University of Coimbra - Coimbra - Portugal José Maria Fonte Ferreira - University of Aveiro - Aveiro - Portugal José Rubens G. Carneiro - Pontifícia Univers. Católica de Minas Gerais- Belo Horizonte- Brazil Juno Gallego - Univ. Estadual Paulista - Ilha Solteira - Brazil Ligia M. Souto Vieira - Technological Institute of Aeronautics - São José dos Campos - Brazil Luis Fernando Figueira da Silva - Pontifical Catholic University - Rio de Janeiro - Brazil Luiz Antonio Pessan - Federal University of São Carlos - São Carlos - Brazil Márcia Barbosa Henriques Mantelli - University of Santa Catarina - Florianópolis - Brazil Maurizio Ferrante - Federal University of São Carlos - São Carlos - Brazil Michael Gaster - University of London - London - UK Mirabel Cerqueira Resende - Institute of Aeronautics and Space - São José dos Campos - Brazil Nicolau A.S. Rodrigues - Institute for Advanced Studies - São José dos Campos - Brazil Paulo Celso Greco - São Carlos School of Engineering - São Carlos - Brazil Paulo Varoto - São Carlos School of Engineering - São Carlos - Brazil Rita de Cássia L. Dutra - Institute of Aeronautics and Space - São José dos Campos - Brazil Roberto Costa Lima - Naval Research Institute - Rio de Janeiro - Brazil Roberto Roma Vasconcelos - Institute of Aeronautics and Space - São José dos Campos - Brazil Samuel Machado Leal da Silva - Army Technological Center - Rio de Janeiro - Brazil Selma Shin Shimizu Melnikoff - University of São Paulo - São Paulo- Brazil Tessaleno Devezas - University of Beira Interior - Covilha - Portugal Ulrich Teipel - University of Nuremberg - Nuremberg - Germany Vassilis Theofilis - Polytechnic University of Madrid - Madrid - Spain Vinicius André R.Henriques -Institute of Aeronautics and Space - São José dos Campos - Brazil Wim P. C. de Klerk - TNO Defence - Rijswijk - The Netherlands 106

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 105-108, May-Aug., 2011

ISSN 1984-9648 ISSN 2175-9146 (online)

Journal of Aerospace Technology and Management Vol. 03, N. 02, May - Aug., 2011

CONTENTS Editorial 109 UFABC: an Interdisciplinary University Helio Waldman Technical Papers 111

Computation of air chemical equilibrium composition until 30000k - Part I Carlos Alberto Rocha Pimentel, Annibal Hetem Jr.

127

Experimental results of a mach 10 conical-flow derived waverider to 14-X hypersonic airspace vehicle Tiago Cavalcanti Rolim, Paulo Gilberto de Paula Toro, Marco Antonio Sala Minucci, Antônio de Carlos de Oliveira, Roberto da Cunha Follador

137

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability Leonardo Ostan Bitencourt, Gregori Pogorzelski, Ramon Morais de Freitas, João Luiz F. Azevedo

147

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers Isabel Lima Hidalgo, Airton Nabarrete, Marcelo Santos

159 Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine Emerson Andrade Santos, Wilton Fernandes Alves , André Neves Almeida Prado, Cristiane Aparecida Martins 171

Investigação da distribuição do filme de resfriamento em um motor foguete à propulsão líquida Investigation of the cooling film distribution in liquid rocket engine Luís Antonio Silva

179

Methodology for DSC calibration in high heating rates Carlos Isidoro Braga, Mirabel Cerqueira Rezende, Michelle Leali Costa

193

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara Statistical analysis of wind profile in the surface layer at the Alcântara launching center Carlos Alberto Ferreira Gisler, Gilberto Fisch, Cleber de Sousa Correa

203

Identification and analysis of explanatory variables for amulti-factor productivity modelof passenger airlines Antonio Henriques deAraújo Jr, Flávio Hegenberg, Isabel Cristina dos Santos, José Glênio Medeiros de Barros

215

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil Cognitive mapping applied to Brazilian aerospace decision problem Paloma Ribeiro dos Santos, Rocio Soledad Gutierrez Curo, Mischel Carmen Neyra Belderrain

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 105-108, May-Aug., 2011

107

Thesis abstract 227

Environmental conditioning effects on the shear and damping properties of fiber metal hybrid composites César Augusto Damato

227

Study of curing and water sorption on glass-epoxy composites by luminescence and infrared spectroscopy Rita de Cássia Mendonça Sales

228 Hybrid solutions for heat transfer on ablative thermal protection system Daniel Fraga Sias 228

Surface hardening of a VC131 tool steel using a fiber laser Flávia Aline Goia

229

Manufacture and characterization of carbon fiber/PEKK composites with aeronautical applications Rogério Lago Mazur

229

Law airborne enforcement helicopters technological updating Márcio Luiz Ramos Pereira

230 Instructions to the Authors

108

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 105-108, May-Aug., 2011

Helio Waldman*

Rector of Universidade Federal do ABC Santo André/SP – Brazil reitoria@ufabc.edu.br

Editorial UFABC: An Interdisciplinary University

UFABC stands for “Universidade Federal do ABC”, a young Brazilian Federal University in the ABC region, which is a 2.5-million people district outlying the Great São Paulo City. During the 20th century, it was in the ABC industrial belt that Brazil leveraged its thriving automobile industry. Legally created in 2005, UFABC started its activities in 2006 and is now celebrating five years of teaching. During this period, its name has been increasingly associated with innovation in higher learning, commitment to academic excellence with an interdisciplinary outlook and, last but not least, a firm belief in affirmative action as a means to play an effective role in society. The interdisciplinary approach emerged in the 1970s and 1980s as the Cold War neared its end and the world enhanced its awareness of pressing social and environmental issues. The new issues were (are) much more complex than even the most formidable disciplinary challenges. However, they were neglected during most of the 20th century, when the disciplinary approach brought about new developments in science and technology at an unprecedented rate. The work effort aimed at such developments was much accelerated during the two World Wars and kept up by the risk of global conflict, ever since. However, after the collapse of the Soviet Union, the new threats that challenge the prevailing powers have changed their nature; thus, requiring a new approach to current conflicts and deadlocks. During the 1950s and 1960s, the two competing powers committed themselves with some remarkable technological challenges, with the purpose of inspiring awe on all peoples of the world and gaining their “hearts and minds”. During the 1950s, the Soviets launched the Sputnik satellite, which was the first manmade object ever to orbit our planet. The first manned orbital flight happened in the early 1960s. It was then that Yuri Gagarin, the first astronaut, uttered his famous sentence “The Earth is blue”, a thoroughly new, but rather detached way of looking at the world. The Americans responded in the late 1960s by having an astronaut walk on the moon and bringing him back alive. Officially, all these events were motivated by scientific interest. However, they were also heavily covered by the media and followed by all peoples in a mediatic fashion. The underlying message was that if the big powers were able to achieve such amazing feats, they would also be able to solve the problems that afflict the lives of most men and women in the world, such as poverty, violence, environmental degradation, malnutrition, and so on. Unfortunately, it is now clear that such expectations were mistaken, if not misgiven. Most social and environmental issues cannot be successfully addressed from the perspective of only one discipline or disciplinary field, such as science and technology. Instead, the scientific perspective must join into the wider social, economic and environmental debate in the search for feasible and sustainable solutions in a democratic, educated decision-making framework. In order to do this, we need scientists with a new frame of mind, interacting with society in social organizations, in government, in academia, in the productive sector, in regulating agencies, and so on. These people should be ready to approach complex issues incorporating social, political, and technical requirements that may have to be negotiated to produce a feasible solution. UFABC is aware of this emerging need and is eager to address it. At UFABC, undergraduate education starts necessarily with a three-year Interdisciplinary Bachelorship (IB). Two IBs are currently offered: Science and Technology, and Science and Humanities. Upon completion of the IB on Science and *Helio Waldman was born in 1944 in São Paulo, Brazil. In 1966, he graduated as Electrical Engineer at Instituto Tecnológico da Aeronáutica (ITA), Brazil. He received the M.Sc. and Ph.D. degrees from Stanford University, USA, in 1968 and 1972 respectively. His Ph.D. work used satellite signals and numerical simulation to address some issues in ionospheric physics. He came back to Brazil in 1972 as an Associate Professor at COPPE/UFRJ, Rio de Janeiro. From 1974 through 2005, he was with Unicamp, in Campinas, where he became a Full Professor in 1980, headed the Campinas School of Engineering from 1982 to 1986, was the first Research Provost from 1986 to 1990, and is now retired. In 2006, he joined the precursor team that organized the Federal University of ABC (UFABC), a pioneering Brazilian University for the 21st. century, where he is now the Rector for the 2010-2014 term. He has authored or coauthored three books, 33 papers in international j ournals, two book chapters, and 101 papers published in proceedings of scientific events. He has also supervised 34 M.Sc. and 11 doctoral Theses, is an IEEE Senior Life Member, and bears the insignia of the Brazilian National Order of Scientific Merit as a commendation holder (“comendador”). J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 109-110, May-Aug., 2011

109

Helio Waldman

Technology, the student is entitled to a degree and may register in 1 of 17 courses leading to professional standing: eight two-year Engineering courses leading to Engineering degrees, or one-year B.S in Computer Science, or a B.S. and/or a license degree for high school teaching in Mathematics, Physics, Chemistry, or Biology. The profile of “classical”, 20th century engineering education favored the definition of disciplinary specializations, such as Electrical, Mechanical, and Chemical Engineering. Given the interdisciplinary approach to general education through the sophomore level at UFABC, it was only natural to adopt an innovative engineering education profile, in which specializations are defined according to application domains, instead of scientific disciplines. Energy Engineering deals with a critical, but environmentally sensitive variable of the infrastructure underlying modern economies. Information Engineering focuses on how information is processed in the emerging Knowledge Society, an issue that is central to the productivity of modern organizations and to the new ways of organizing labor and cultural activities. Materials Engineering is crucial to produce new materials needed to support emerging functionalities in new devices, starting from the molecular level. Biomedical Engineering supplies more and more technology to the medical profession, thus helping it in improving the treatment of patients. Instrumentation, Automation and Robotics add more intelligence to the operation of systems, thus raising the efficiency of labor and machines. Aerospace Engineering supports the exploration of space for scientific purposes, as well as for monitoring the planet itself and providing communication capabilities. Management Engineering looks for effective ways for organizations to achieve their strategic objectives, in highly competitive environments. Finally, Urban and Environmental Engineering deals with the crucial problem of mediating the needs of urban development and environmental protection. In order to encourage the shaping of interdisciplinary ventures and to avoid any kind of disciplinary entrenchment, UFABC is not organized in departments. Instead, the teaching staff is assigned to three large centers: the Center for Engineering, Modeling and Applied Social Sciences (CECS); the Center for Natural and Human Sciences (CCNH); and the Center for Mathematics, Computation and Cognition (CMCC). Notice that these Centers are not defined along disciplinary lines. Rather, they draw their identity from professional cultures and teaching responsibilities. This arrangement has already resulted in some very exciting interdisciplinary ventures. A remarkable one is the creation of new programs in the emerging area of human cognition, leading to B.S., M.Sc., and Ph.D. degrees. These programs will provide the first regular source of neuroscientists to the thriving São Paulo labor market. Also, in the pipeline, there are graduate and undergraduate programs in Planning and Management of Territories, which address the complex problem of managing the resources and spatial occupation of a territory impacted, for example, by the presence of a large metropolitan area. Interdisciplinarity is not a new idea in Brazilian academic scene. Since the 1980s, leading Brazilian Universities have been establishing interdisciplinary special units to address timely issues. However, these initiatives were and still are peripheral to the hard core of the academic structure: the undergraduate courses, which remain essentially untouched for at least one half century. UFABC is the first Brazilian University to place interdisciplinarity at the center of its vision for the 21st century and to build a new institutional architecture to accommodate this vision. Of course, UFABC should be seen as an experiment, albeit an important one. In the coming years, it is vital that this experiment receives continued support while being followed up and critically evaluated. As of today, about 15 other Federal Universities are following suit and starting similar experiments. On the other hand, disciplinary teaching and research must still go on, both for their own sake and to strengthen the disciplinary foundation of interdisciplinary learning. It is important that this be done in close correlation with industry, so that Brazil may overcome the current gap between scientific and industrial endeavors. In the ABC Region, and especially in its largest city (São Bernardo), the municipal authorities are keenly interested in the attraction of R&D initiatives by leading corporations in the area of air defense. UFABC may be a key component of this effort, not only by using its expertise in the search for new solutions for innovative industrial products, but also by using its interdisciplinary approach to address the defense issue as a whole, in its strategic, geopolitical, and technological dimensions.

110

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 109-110, May-Aug., 2011

doi: 10.5028/jatm.2011.03021011

Carlos Alberto Rocha Pimentel*

Federal University of ABC Santo André/SP – Brazil carlos.pimentel@ufabc.edu.br

Annibal Hetem Jr.

Federal University of ABC Santo André/SP – Brazil annibal.hetem@ufabc.edu.br *author for correspondence

Computation of air chemical equilibrium composition until 30000K - Part I Abstract: An algorithm was developed to obtain the air chemical equilibrium composition. The air was considered to be composed of 79% N2 and 21% O2, and the models are 5 chemical species,!! ! !! ! !"! !! !!, and 7 chemical species, !! ! !! ! !"! !! !! !"! ! !! !, respectively. The air chemical equilibrium composition is obtained through the equilibrium constants method and it was used the Absolute Newton method for convergence. The algorithm can be coupled as a subroutine into a Computational Fluid Dynamics code, given the ﬂow ﬁeld over an atmosphere reentry vehicle where, due to high velocities, dissociative chemical reactions and air ionization can occur. This work presents results of air chemical equilibrium composition for pressures of 1, 5, 10, 50 and 100 atm in a temperature range from 300 to 30000K. Keywords: Air composition, Nitrogen chemistry, High temperature, Chemical equilibrium problem, Absolute Newton method.

LIST OF SYMBOLS

INTRODUCTION

A k: f: G i0: H i0: Jf : k: Mi : m: nir or p: N: p 0: p: R u: r: S i0: T: Xi : Yii: vi : ρ:

The utilization of Computational Fluid Dynamics (CFD) techniques to compute the flow field around atmospheric reentry vehicles needs a robust and able algorithm to obtain the air chemical composition in a wide range of pressure and temperature. There are two strategies to deal with the air chemical composition in this kind of flow. The first one considers the air not in chemical equilibrium, and the other considers chemical equilibrium. Each approach has its particularities and difficulties on how to obtain the air chemical composition. In this first part of the work, only the equilibrium method is used, and the non-equilibrium computations shall be presented in the second part.

coefficients of thermodynamic data fits () function free energy of species i [J/kg mole] specific enthalpy of species i [J/kg mole K] Jacobian matrix of function f equilibrium constant molecular weight [kg/mole] number of ionized species mole number for species i number of equations standard state pressure, products universal gas constant, reactants entropy of species i [J/kg mole K] temperature [K] mole fraction of species i mass fraction of species i stoichiometric coefficient of species i [kg moles] mass density of mixture [kg/m3]

Received: 15/03/11 Accepted: 01/06/11

Due to its iterative codification nature, a CFD code needs to perform many times these chemical computations. This code could be inserted into the main program as a subroutine or library and be called many times during the convergence process. So, this implies the code should be very efficient in processing time terms. This efficiency is reflected in the CFD computational code, and its ability to obtain more quickly the flow field solution in study. There are many techniques to compute the chemical equilibrium composition. The mostly know and used is the Gibbs free energy minimization (Gordon and McBride, 1994), but due to its inherent generality, it is used to solve problems that involve a high number

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

111

! !! !!

Pimentel C.A.R., Hetem Jr. A.

of chemical species as, for example, hydrocarbon combustion. A more simple method, and also widely used, is the equilibrium constants method (Prabhu and Erickson 1988; Sabetta, Favini and Onofri, 1993). This method combines an algebraic combination of the equilibrium constants, mass balance and the electron charge conservation. Besides its simplicity, this method is robust and efficient enough to compute the flow field around reentry vehicles, whose medium is considered to be in chemical equilibrium. The method presented herein does not divide the study dominium in four regions to obtain the air chemical equilibrium composition in high temperatures, where each region is limited by a series of predominant species as in Prabhu and Erickson (1988) or Smith, Erickson and Eastwood (1967). What is wanted is to obtain the equilibrium chemical composition prefixing the existent chemical species, independently of region. This work presents results of an algorithm that applies the equilibrium constants method to obtain the chemical equilibrium composition for the air, considered to be composed of 79% N2 and 21% O2, and the models are five chemical species, , and seven chemical species, , and it is assumed to be a perfect gas.

Considering the concentration of the chemical species i, one has (Eq. 4 and 5):

!!!

!!!! !! !!" !! ! ! !! ! ! ! ! !" ! (4) !!! !!! !!" !! !!" !

!!!! !! !!" !! ! ! !!! !! ! ! ! ! !" ! (5) !!! !!! !!" !! !!" ! Where the fraction of initial mass of the reagent species, N2 and O2, are defined as (Eq. 6 and 7):

!!!! !

!!! !!! ! !!! !!! ! !!! !!!

(6)

!!!! !

!!! !!! ! !!! !!! ! !!! !!!

(7)

Where: XN2 = 0,79 and XO2= 0,21

THE GOVERNING EQUATIONS The describing expressions of the chemical equilibrium composition for a given pressure and temperature condition are built in a set of N equations, where N depends on the number of chemical elements present in the considered model. For the seven chemical species model, considering, the equations are (Eq. 1 to 3):

so,

XN2 + XO2= 1 For an ionized gas, the computation of the equilibrium requires a complementary equation (Eq. 8) on the charge conservation of electrons (William, 2000):

chemical reaction equation !

!!! ! !! ! !!! ! !! !

!!! ! !!

!!!! !!

! !!" !"

! ! ! ! ! ! !!" !" ! !!! ! ! !!! ! ! !!" ! !" ! !! ! ! !

! (1)

! ! ! ! !!" ! !! ! ! ! !! ! ! ! ! !! ! !" ! !! ! ! ! !" !! ! !

!!!

(8)

Where is the number of ionized species.

atom conservation equations

! ! ! ! !!!! ! ! !!!! ! ! !!" ! !!! ! !!" !

(2) !

! ! ! ! !!!! ! ! !!!!! ! !!" ! !!! ! !!" !!

(3)

112

!! ! ! !!! !! ! !!!!! !

For the completeness of the equation set solution, complementary equations are necessary. These equations are defined from independent chemical reactions that involve the considered chemical species. The chemical expressions are (Eq. 9 to 12):

!! ! !!!

(9)

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

!! ! !!!

(10)

!! ! !! ! !!"!

(11)

!"! ! ! ! ! !"!

(12)

!! ! ! !

!!" !! !!!!

!! ! ! ! !!! Now, it is necessary to express the equilibrium constants for each chemical reaction equation. These expressions are (Eq. 13 to 16): !! ! !

!! ! !!! !!! !!

!!" ! !

!!! !

!!" !! !!!!

!! ! !!! !!! !! !!! !!! !! ! !!!!

!!! ! !!

! !! (13)

! !!

(14)

!!! ! !!

! !!

!!" ! ! ! ! !!" !!! !!! !! !!

!! ! ! ! !!" !

(21)

! !! (22)

!!" ! ! !! ! ! !!" ! !! (23) !!" ! !! ! !"

Where the equilibrium constants, !! !, the considered chemical species molecular weights, !! !, and the mixture density, !!, are known. To obtain the equations for a five chemical species model

!!! ! !! ! !"! !! !!, one just eliminates the chemical species !"! !and ! ! !from the model described above. THERMODYNAMIC DATA

!!!

!!! !!! !!" ! !!" ! ! ! ! ! ! !!" ! ! ! !!" !!" ! !!! !!! !!! !!! !! !!

! !! (15)

! ! ! !! ! !!" !!" !!" ! ! !" ! ! ! ! !!" ! ! !!" !!" ! !! ! !!" ! !! ! !!" !!" (16) ! ! ! !!" ! !!" ! !! ! ! !!" ! !! !!" ! !! ! !!" ! !! ! The non-linear equation system is given by the equation set (Eq. 17 to 23):

!! ! ! !

!!! !!!

!!! !!" !! !!" ! ! ! ! ! ! ! !!(17) !!" !! !!" ! !!!

!! ! ! !

!!! !!! !!" !! ! ! ! ! ! !" ! ! ! ! !! (18) !!! !!" !! !!" ! !!!

!! ! !

!!" ! !! ! ! ! !! !!" ! !! !

!! !! !! ! ! !!!!

entropy

!!! !! !! !! ! !! !" ! ! !! ! ! ! ! ! ! ! ! ! ! ! !! !(24) !! ! ! !

specific enthalpy

!!! !! !! !! !! !! ! !! ! ! ! ! ! ! ! ! ! ! ! ! !(25) !! ! ! ! ! ! !

(19) Gibbs free energy

!!! ! !!

To obtain the equilibrium composition, it is necessary to compute the thermodynamic properties enthalpy ( !! !), entropy ( !! !) and the Gibbs free energy (!!! )! for each chemical species included in the model. The polynomial !!" ! !! ! !that !!"are ! used !! for these computations are expressions based on the works from Gupta et al. (1990) and Gordon and McBride (1994):

! !!

(20)

!!! !!! !!! ! ! ! !! ! !! ! !!

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

(26)

113

Pimentel C.A.R., Hetem Jr. A.

The polynomial expressions coefficients, to , are those given by Gupta et al. (1990). They were computed for 5 temperature ranges, from 300 to 30000 K (300 K≤T≤1000 K, 1000 K≤T≤6000 K, 6000 K≤T≤15000 K, 15000 K≤T≤25000 K, 25000 K≤T≤30000 K). As these curves do not present continuous behavior in their frontiers, the coefficients A1 to A7 were linearly smoothed in the borders to avoid an abrupt variation of thermodynamic properties. EQUILIBRIUM CONSTANTS According to Pinto and Lage (2001) and Carvalho and McQuay (2007), the equilibrium constant for a chemical reaction at constant pressure and temperature is a function of the Gibbs free energy:

!!!!

!!!! ! !"# ! ! !"

(27)

Reaction 11: Δn=-1 SOLUTION METHOD A much known method for solving a non-linear equation set is the Newton method (Pinto and Lage, 2001), whose convergence process is described by the iteration, as follows. One starts with:

!! ! !! ! !! !

(30)

with

! ! !!!! ! ! !

(31)

Where J is the Jacobian for the partial derivatives of the function f(Y). In this work f(Y) set of Eq. 17 to 23.

Where

!!! !!! !

!!!! ! !

!!! !!! !

(28)

and vi is the stoichiometric coefficient of chemical species i. To compute the equilibrium constants for the 4 independent chemical reactions in the equation set, it is necessary to express Eq. 27 as function of molar concentration. So,

!! ! !!!!

!! !"

!! !

As it is widely known, a non-linear equation system can present a number of solutions, both negative and positive, or a solution that involves negative and positive values. Due to the complexity in obtaining the equilibrium chemical composition for a gas mixture reacting chemically, the Newton method must be modified in such a way that values which do not correspond to physical meaning are avoided. For the model described above, only positive values shall be allowed as results from the Newton method. We adopted the modified method proposed by Meintjes and Morgan (1989), and the iteration process suffered a simple modification:

(29)

!! ! !! ! !! ! !!! !

Reactions 9 and 10: Δn=1

(32)

!!! !

Where for each independent ! ! chemical reaction. Then, for the reactions in the model, one has:

114

Reaction 11: Δn=0

That means, through the iteration process, if one of the possible solutions is negative, it is replaced by its absolute value. The applied method, with this modification, was called the Absolute Newton Method.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

RESULTS AND CONCLUSIONS The developed numerical algorithm was able to obtain the air equilibrium chemical composition for pressures of 1, 5, 10, 50 and 100 atm, for a temperature range varying from 300 to 30000 K. The convergence process needs an initial condition for the mass fractions Y of each chemical species that compose the model. So, the iterative process is described by the following steps: 1. a constant temperature is set. 2. initial values for the chemical species mass fractions are chosen. 3. the iterative process is started with loops until the desired precision is reached. To obtain the results presented herein the adopted precision was. The implemented method computes the model mass balance of the chemical species for a given temperature. To be strictly correct, the temperature itself needs to be re-evaluated through the energy equation, which should be coupled with non-linear equations set. The results are presented as plots of the logarithm of mol fraction and mass fraction as function of the temperature for the chemical composition of the models. Under 1 atm, oxygen (02) starts dissociate around 2000 K, being fully dissociated just above 4000 K. On the other side, nitrogen (N2) starts its dissociation process near 4000 K and finishes it around 9000 K, while the nitrogen monoxide (N0) starts forming on 1800 K and is completely dissociated near 6000 K, having its maximum on 3300 K. This behavior is in agreement with results from Anderson (1989) and William (2000). One observes that, in the studied temperature range, from its beginning to the end, the dissociation processes for both five and seven chemical species models are very likely. This occurs due to the ionized species present in the seven species model show significative mass only above 10000 K. It is important to observe the pressure effect over the dissociation mechanism. If the pressure increases the

dissociation processes became slower even under high temperatures. This is motivated by the fact that at high pressures the molecules that compose the mixture are closer one to the others, making the mechanisms that pull out the outer layer electrons less effective. The result is that the plotted curves move to the high temperature as the pressure increases, and the opposite occurs when the pressure drops, because the molecules are sparser and the dissociation process is facilitated. This kind of phenomena is observed in all presented plots, both for the five chemical species as for the seven chemical species model. When comparing our results to those of William (2000) for the 7 species, one can notice a slight deviation in the 5, 10, 50 and 100 atm models. In the worst case, the deviation is about 6% in temperature (~ 615 degrees) in the curves of atomic nitrogen (N) formation at 100 atm. These small deviations can be due to many factors, being the most relevant the method used to obtain the polynomial coefficients to that have a very important role in the thermodynamical properties of chemical species present in the models, and these influence the equilibrium constants evaluation. Figures 1 to 10 present the evolution of each model for 100, 50, 10, 5 and 1 atm. The 4 first models for 7 chemical species are also compared to the same calculations of William (2000). These values were computed in a 3.33 GHz Intel Core i5 computer, and the maximum time needed to compute 1 convergence was 0.04 seconds, for a case where 16 iterations were used. The presented results can diverge slightly from those found on literature, but these small differences can be due to many factors. One cause is related to the method used to obtain the polynomial coefficients, A1 to A7, that influences the computation of chemical species thermodynamic properties in the model, and these the equilibrium constants. The developed algorithm is the first step in the direction of a free library for the simulation of the field flow on reentry vehicles studies.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

115

Pimentel C.A.R., Hetem Jr. A.

0 -1

-3

-4

NO +

-5

-3 -4 -5 -6

-7

-6

-8

-2 Log (Xi)

Log (Xi)

-2

NO + e-

-8

5000 10000 15000 20000 25000 30000 Temperature (K)

0.8

0.7

0.6

0.5

0.4

0.3 0.2 0.1 0

-0.1 0

5000 10000 15000 20000 25000 30000 Temperature (K)

4000 6000 8000 Temperature (K)

0.3

N O

0.1

0 -0.1

10000

0.2 NO +

2000

2500 5000 7500 Temperature (K)

NO + e-

10000

Figure 1. Logarithm of mole fraction and mass fraction to the pressure of the model of seven chemical species. The panels on the right present the details for the range from 0 to 10000 K. The dots on last panel represent the results form William (2000) for the same parameters.

116

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

-1

-2

-3

-4

NO + e-

-5

-7

0.8

-8

0.6

0.5

0.4

0.7

0.6

0.2 0.1 0 -0.1

NO + e-

NO +

5000 10000 15000 20000 25000 30000 Temperature (K)

2000

10000

0.3

0.1

0 -0.1

4000 6000 8000 Temperature (K)

0.2

0.8

0.7

0.3

-5 -6

5000 10000 15000 20000 25000 30000 Temperature (K)

-4

-6

-1

-2

-8

Log (Xi)

Computation of air chemical equilibrium composition until 30000K - Part I

NO + e-

2500 5000 7500 Temperature (K)

10000

Figure 2. Logarithm of mole fraction and mass fraction to the pressure of the model of seven chemical species. The panels on the right present the details for the range from 0 to 10000 K. The dots on last panel represent the results form William (2000) for the same parameters.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

117

Pimentel C.A.R., Hetem Jr. A.

0 -1 -2

-3

-4 -5

Log (Xi)

-2

NO + e-

-7

0.8

-8

5000 10000 15000 20000 25000 30000 Temperature (K)

0.7

0.6

0.5

0.4

0.2 0.1

0 -0.1 0

5000 10000 15000 20000 25000 30000 Temperature (K)

0.3 0.2

NO +

NO + e-

0.7

0.3

0.8

-5 -6

-4

-6

-8

Log (Xi)

-1

0.1

2000

4000 6000 8000 Temperature (K)

N2 N

NO + e-

0 -0.1 0

1000

2500

5000 7500 Temperature (K)

10000

Figure 3. Logarithm of mole fraction and mass fraction to the pressure of the model of seven chemical species. The panels on the right present the details for the range from 0 to 10000 K. The dots on last panel represent the results form William (2000) for the same parameters.

118

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

0 -1

-3

-4

NO + e-

-5

Log (Xi)

-2

-3 -4 -5

-6

-7

-8

5000 10000 15000 20000 25000 30000

NO NO + e-

2000

0.8

0.7

0.6

0.5

0.4

0.3

0.2 0.1

0 -0.1

8000

0.3

1000

NO +

5000 10000 15000 20000 25000 30000 Temperature (K)

0.1

0 -0.1

0.2

6000

Temperature (K)

0.8

4000

NO + e-

2500

5000

7500

1000

Temperature (K)

Figure 4. Logarithm of mole fraction and mass fraction to the pressure of the model of seven chemical species. The panels on the right present the details for the range from 0 to 10000 K. The dots on last panel represent the results form William (2000) for the same parameters.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

119

Pimentel C.A.R., Hetem Jr. A.

0 -1

-2

-3

-4

NO + e-

-5

Log (Xi)

-1

-5 -6

-7

-7 -8

5000 10000 15000 20000 25000 30000

-4

-6

-8 0

NO NO + e-

4000

2000

Temperature (K)

0.8

0.7

0.6

0.5

0.4

0.3 0.2

0.1 0 -0.1 0

5000 10000 15000 20000 25000 30000 Temperature (K)

0.3

10000

0.1

0 -0.1

0.2 NO +

8000

Temperature (K)

0.8

6000

NO + e-

2500

5000

7500

1000

Temperature (K)

Figure 5. Logarithm of mole fraction and mass fraction to the pressure of the model of seven chemical species. The panels on the right present the details for the range from 0 to 10000 K.

120

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

-2

-3

-4 NO

-5

-7

0.8

-5 -6

-8

5000 10000 15000 20000 25000 30000 Temperature (K)

0.8

0.7

0.6

0.5

0.4

0.7

0.3

0.2 0.1

2000

4000 6000 8000 Temperature (K)

10000

0.3

0.2

0.1

0 -0.1

-4

-6

-1

-2

-8

Log (Xi)

-1

0 0

5000 10000 15000 20000 25000 30000 Temperature (K)

-0.1

2500

5000 7500 Temperature (K)

10000

Figure 6. Logarithm of mole fraction and mass fraction to the pressure of the model of five chemical species. The panels on the right present the details for the range from 0 to 10000 K.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

121

Pimentel C.A.R., Hetem Jr. A.

-1

-2

-3

-4 NO

-5

-7

0.8

-8

5000 10000 15000 20000 25000 30000 Temperature (K)

0.6

0.5

0.4

0.7

0.2 0.1

2000

4000 6000 8000 Temperature (K)

10000

0.3

0.2 0.1

O NO

0 -0.1

0.8

0.7

0.3

-5 -6

O N

-4

-6

-8

Log (Xi)

-1

5000 10000 15000 20000 25000 30000 Temperature (K)

-0.1

2500

5000 7500 Temperature (K)

10000

Figure 7. Logarithm of mole fraction and mass fraction to the pressure of the model of five chemical species. The panels on the right present the details for the range from 0 to 10000 K.

122

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

-1

-2

-3

-4 NO

-5

-7

0.8

-8

5000 10000 15000 20000 25000 30000 Temperature (K)

0.7

0.6

0.5

0.4

0.7

0.3

0.2

0.1 0 -0.1

4000 6000 8000 Temperature (K)

10000

0.3

0.2

-0.1

2000

0 5000 10000 15000 20000 25000 30000 Temperature (K)

0.1

0.8

-5 -6

-4

-6

-8

Log (Xi)

-1

2500

5000 7500 Temperature (K)

10000

Figure 8. Logarithm of mole fraction and mass fraction to the pressure of the model of five chemical species. The panels on the right present the details for the range from 0 to 10000 K.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

123

Pimentel C.A.R., Hetem Jr. A.

0 -1

-2

-3

Log (Xi)

-1

-4 -5 NO

-6

-4

-5

-8 0

5000 10000 15000 20000 25000 30000 Temperature (K)

0.8

0.7

0.6

0.5

0.4

0.4 Yi

-7

0.8

0.3

0.2

0.3 0.2 0.1

0.1 NO

0 -0.1

-6

-7 -8

-0.1

4000 6000 8000 Temperature (K)

10000

0 5000 10000 15000 20000 25000 30000 Temperature (K)

2000

2500 5000 7500 Temperature (K)

1000

Figure 9. Logarithm of mole fraction and mass fraction to the pressure of the model of five chemical species. The panels on the right present the details for the range from 0 to 10000 K.

124

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

Computation of air chemical equilibrium composition until 30000K - Part I

0 -1

-2

-3

-4 -5

-7

0.8 0.6

0.6

0.5

0.4

0.4 Yi

0.7

0.2 0.1

0.3 0.2 0.1

0 -0.1

-0.1

2000

4000 6000 8000 Temperature (K)

10000

0 5000 10000 15000 20000 25000 30000 Temperature (K)

0.8

0.7

0.3

-8 0

5000 10000 15000 20000 25000 30000 Temperature (K)

-5 -6

-4

-6

-8

Log (Xi)

-1

2500 5000 7500 Temperature (K)

1000

Figure 10. Logarithm of mole fraction and mass fraction to the pressure of the model of five chemical species. The panels on the right present the details for the range from 0 to 10000 K.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

125

Pimentel C.A.R., Hetem Jr. A.

REFERENCES Anderson Jr., J.D., 1989, ‘‘Hypersonic and high temperature gas dynamics’’, Series in Aeronautical and Aerospace Engineering, McGraw-Hill, Singapore, 690p.

Pinto, J.C., Lage, P.L.C., 2001, ‘‘Método numéricos em problemas de Engenharia Química’’, Série Escola Piloto em Engenharia Química – COOPE/Federal University of Rio de Janeiro, Ed. E-papers, Rio de Janeiro, Brazil, 316p.

Carvalho Jr., J.A.C., McQuay, M.Q., 2007, ‘‘Princípios de combustão aplicada’’, Ed. Federal University of Santa Catarina, Florianópolis, Brazil, 175p.

Prabhu, R., Erickson, W.D., 1988, ‘‘A Rapid Method for the Computation of Equilibrium Chemical Composition of Air to 15000 K’’, NASA Technical Paper 2792.

Gordon, S., McBride, B.J., 1994, ‘‘Computer Program for Calculation of Complex Chemical Equilibrium Compositions and Applications. I. Analysis’’, NASA Reference Publication 1311.

Sabetta, F., Favini, B., and Onofri, M., 1993, ‘‘Equilibrium and nonequilibrium modeling of hypersonic inviscid flows’’, Computers & Fluids, Vol. 22, No. 2-3, pp. 369-380. doi:10.1016/00457930(93)90066-I

Gupta, R.N., Yos, J.M., Thompson, R.A., and Lee, K.P., 1990, ‘‘A review of reaction rates and thermodynamic and transport proprieties for an 11-species air model for chemical and thermal nonequilibrium calculations to 30000 K’’, NASA Reference Publication 1232. Meintjes, K., Morgan, A.P., 1989, ‘‘Elements Variables and the Solution of Complex Chemical Equilibrium Problems’’, Combustion Science and Technology, Vol. 68, pp. 35-48.

126

Smith, G.L., Erickson, W.D., and Eastwood, M.R., 1967, ‘‘Equations for the rapid machine computation of equilibrium composition of air and derivatives for flowfield calculations’’, NASA Technical Note D-4103. William, J., 2000, ‘‘Étude des Processus PhysicoChimiques dans les Écoulements Détendus à Haute Enthalpie: Application à la Soufflerie à Arc F4’’, ONERA NT 2000-5, Vol. I and II.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 111-126, May-Aug., 2011

doi: 10.5028/jatm.2011.03027510

Tiago Cavalcanti Rolim*

Instituto de Estudos Avançados São José dos Campos /SP – Brazil tiagorolim@ieav.cta.br

Paulo Gilberto de Paula Toro

Instituto de Estudos Avançados São José dos Campos/SP – Brazil toro@ieav.cta.br

Marco Antonio Sala Minucci

Instituto de Estudos Avançados São José dos Campos/SP – Brazil sala@ieav.cta.br

Antônio de Carlos de Oliveira

Instituto de Estudos Avançados São José dos Campos/SP – Brazil acoc@ieav.cta.br

Roberto da Cunha Follador

Instituto de Estudos Avançados São José dos Campos/SP – Brazil follador@ieav.cta.br

Experimental results of a Mach 10 conical-flow derived waverider to 14-X hypersonic aerospace vehicle Abstract: This paper presents a research in the development of the 14-X hypersonic airspace vehicle at Institute for Advanced Studies (IEAv) from Department of Science and Aerospace Technology (DCTA) of the Brazilian Air Force (FAB). The 14-X project objective is to develop a higher efficient satellite launch alternative, using a Supersonic Combustion Ramjet (SCRAMJET) engine and waverider aerodynamics. For this development, the waverider technology is under investigation in Prof. Henry T. Nagamatsu Aerothermodynamics and Hypersonics Laboratory (LHTN), in IEAv/DCTA. The investigation has been conducted through ground test campaigns in Hypersonic Shock Tunnel T3. The 14-X Waverider Vehicle characteristic was verified in shock tunnel T3 where surface static pressures and pitot pressure for Mach number 10 were measured and, using Schlieren photographs Diagnostic Method, it was possible to identify a leading-edge attached shock wave in 14-X lower surface. Keywords: Hypersonics, Hypersonic systems, Shock tunnel, Schlieren, Waverider.

*author for correspondence

A waverider vehicle could be defined as a supersonic or a hypersonic vehicle which uses a leading-edge attached shock wave to form a high pressure zone on its lower surface to generate lift. The interest in hypersonic waverider vehicles lies on the promise of a high lift-todrag ratio vehicle able to deploy a payload into earth orbit. Until now, research has shown that a waverider vehicle has superior aerodynamic performance compared with other hypersonic aerodynamic concepts as accelerators and as aerogravity-assisted maneuvering vehicles (Rault, 1994). They are also being considered for high-speed long-range cruise vehicles, since their high lift-to-drag ratio becomes important in achieving global range. Furthermore, with horizontal takeoff and landing capability, they could reduce the turn around time of the current space missions. A waverider which uses air breathing propulsion would not need to carry the oxidizer, which results in weight saving, reduced complexity, and less ground support (Javaid, et al., 2005). A strong candidate for hypersonic propulsion is the Hydrogen fueled scramjet engine. That is because this air breathing engine cycle is capable to provide the thrust Received: 14/06/10 Accepted: 11/07/11

required for a hypersonic vehicle more efficiently than conventional rocket propulsion. As a matter of fact, at hypersonic speeds, a typical value for the specific impulse of a H2-O2 rocket engine is about 400 s while for a H2 fueled scramjet is between 2,000 s and 3,000 s (Heiser, et al., 1994). In fact, the use of atmospheric air as oxidizer permits air breathing vehicles to substantially increase payload weight (Figs. 1 and 2).

Fuel Mass fraction

INTRODUCTION

Range = 10,000km CL/CD = 4.0 Cruise Speed = 3km/s

0.8 0.6 0.4 0.2 0

2000 1000 Specific Impulse [s]

3000

Figure 1. Fuel mass fraction variation with specific impulse (Isp) using Breguet equations for a hypersonic cruise mission.

The waverider’s concept was introduced by Terence Nonweiler (Nonweiler, 1963), as a delta shaped reentry vehicle. This concept was named caret shaped waverider

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

127

Rolim, T.C. et al

Aircraft

Rocket Empty 7% Fuel 30%

Empty 55%

Payload 4%

Fuel 24%

Oxidizer 0% Payload 15%

Oxidizer 65% Figure 2. Typical takeoff mass fractions for current aircrafts and rockets, based on data given by Heiser (Heiser, et al., 1994).

due to its resemblance with the typographic symbol (^). While studying the flow over the vehicle, Nonweiler realized that the high pressure on the under surface due to a shock wave could be used for generating lift. Also, the attached shock wave in a sharp leading edge isolates the high pressure zone from the low pressure zone, which inhibits the flow spillage. Regarding the aerodynamic design of a waverider, its surface is constructed from a base body. The streamlines of the flow over a body are traced to generate the lower surface. The upper surface is generally aligned with the free stream flow. In general, the goal of each method of construction is to attain high values of the lift-to-drag ratio as well as a high package’s capacity. Several works regarding the waverider construction, engine integration and geometric optimization have been carried on during recent years.

to better assess the engine integration, to the pure waverider surface, a compression ramp, a flat and an expansion surface were added, in order to simulate a scramjet. To improve the efficiency of these engines, the flow must be two-dimensional and uniform, and must have adequate pressure and temperature for the supersonic combustion. Furthermore, the inlet must be large enough so as to generate thrust even in high altitudes, where there’s a low air density. To simulate a free expansion nozzle, a 15o-ramp was integrated on the rear of the model. Thirteen shock tunnel tests were conducted to acquire experimental data for the hypersonic flow over a waverider vehicle’s compression surface and the scramjet combustor inlet at high Mach numbers and high total enthalpies. The main goals of the research were: i) to design and build a waverider model with instrumentation; ii) to measure surface static pressures and Pitot pressure for Mach number 10 with high reservoir enthalpies; iii) to take schlieren photographs to support data analysis.

An important work on waverider design area was performed by Rasmussen (Rasmussen, et al., 1990), which presented an aerodynamic surface that used the shock wave to generate lift. It was derived from a supersonic flow past a cone. The Rasmussen’s surface obtained a superior overall performance to the classical Nonweiler’s waverider. Since then, various families of cone derived waveriders as well as their hybrid variations like conewedge and multiple cone derived waveriders have been studied. The objectives of these studies have been mainly to increase the low lift-to-drag ratio due to viscous effects and to improve the package capability of such vehicles (Wang, et al., 2007; Kim, et al., 1983).

For the present investigation, the waverider model was built according to the Rasmussen method. In that work, the hypersonic small disturbance theory was applied to analyze waveriders derived from axisymmetric flows past circular cones.

The focus of the analysis described in this work was to investigate the flow field over a waverider derived from a Mach 10 conical flowfield, as described by Rasmussen (Rasmussen, et al., 1990). Also, in order

As stated above, the design was based on a known flow field over a conical body. Figure 3 shows many parameters of the applied method. Given a conical body (the basic body) and a parabolic upper surface trailing

128

MODEL DESIGN a) Pure waverider surface

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

Experimental results of a Mach 10 conical-flow derived waverider to 14-X hypersonic aerospace vehicle

edge, the supersonic flow streamlines are traced back until they reach the shock wave formed by the base body. The generated curve is the vehicle’s leading edge. The localization of the leading edge then permits the flow stream lines to be traced downward defining the lower surface trailing edge. Finally, the lower and upper trailing edges along with the leading edge were used to calculate the entire lower surface; the upper surface was aligned to the free stream flow.

base cone z

க

O x

waverider I

where rs (�) defines the conical shock (θ=β), the intersection of the upper and lower surface, which is also the leading edge. Similarly, the compression surface is defined by:

! ! !! ! !!

! !

! !! ! ! ! ! ! ! !

! !

Also, since we assume that the upper surface is aligned to the free stream, we just need the upper trailing edge to define the entire upper surface. In this work, the upper trailing edge was assumed as a parabola, described in Cartesian coordinates in the base plane by:

! ! !! ! !! !

ࢥ1

(3)

(4)

where

freestream surface

Α and R0 are constants; shock

compression surface

Figure 3: Construction of a general cone-derived waverider (Rasmussen, et al., 1990).

X = x/ιδ; Y = y/ ιδ.

Since, at the shock we have: X=Xs and Y=Ys . Thus, Let us assume a slender cone subjected to a steady inviscid supersonic flow. For this analysis, a spherical coordinate system (r,θ,�) was applied, with origin in the cone vertex (Fig. 3). The free stream velocity V∞ points in the positive z direction. In the scheme presented in Fig. 3, the necessary parameters to describe a waverider are: the free stream Mach number M∞, the ratio of specific heats γ, the semi-vertex cone angle δ, the semi-vertex shock angle β, the dihedral angle �ι and the base body length ι. For slender cones, the shock wave angle can be calculated with the relation:

! !!! ! ! ! ! ! ! ! !! !

! !

(1)

It is also important to define the three surfaces that are used to describe the waverider: the free stream surface (upper surface), the compression surface (lower surface) and the base plane (z=l). For small angles, the free stream surface is defined by:

! ! ! ! !! ! ! !

(2)

!! ! !! !!!

(5)

where

!! ! !!!"# !!

(6)

!! ! !!!"# !!

(7)

And σ = β/δ. From the coordinate system analysis, the radial distance projected in the base plane is given by Rb = r sin θb. For small angles, Rb ≈ r θb and z = r cos θ ≈ r. In the base plane, z=l. Thus, one can define the dimensionless distance of a point in the free stream trailing edge from the axis center projected in the base plane as R∞b = θ∞b (�)/δ. In a similar fashion, for the compression surface, Rcb(�) = θcb (�)/δ . Using Eqs. 2 and 3 we can relate the compression surface trailing edge with the free stream trailing edge as following:

!!" !

! ! ! ! !!! !!! !! !

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

(8) 129

Rolim, T.C. et al

(Rasmussen, et al., 1990).

The Hypersonic Small Disturbances Theory permit us to write the lift coefficient as:

!! ! ! ! ! !! ! !! ! ! ! !

!! !

!!" !!! !! !"#$!!" !

(9)

Similarly, the drag coefficient can be written as the sum of the pressure drag and the friction drag:

!! !

!! ! !! !! ! !!! ! !! !! !! !!

(10)

where Sp and Sw are the waverider plan form and wetted ! !! ! !! !!! areas, respectively. With ! and,

!!!

!! ! ! ! ! ! !! ! ! ! !

!! !

! ! !!" !!" ! ! ! ! !" ! !!

!!"

!!!!"!! !! ! !!!!

(12)

!! !! !! !!

(13)

where

!!!! ! and

!! ! ! !! !! ! !!! !

! !

!! !

!!! !! !

! !

! !!!

!!!! ! !"

! ! !

7.8 7.7

0.16 0.15

7.6 7.5 7.4

0.14 0.13

7.3 7.2

(11)

The friction drag coefficient was evaluated based on a laminar flow, thus:

!! !

As a general result, for a fixed δ, the lift-to-drag ratio is maximum near 30o this behavior is depicted in Fig. 4, for δ = 5.5o. On the other hand, the volumetric efficiency was minimum at 37o but from �ι = 20o to 50o its variation was irrelevant, less than 5%.

0.12

KI° Figure 4. Effect of the dihedral angle on the waverider characteristics, with δ = 5.5o.

Furthermore, the influence of the base cone angle δ was investigated (Fig. 5). One can conclude that the lift-todrag ratio is maximum at 5.5o, and it varies slightly over the considered range of δ. Also, the volumetric efficiency is significantly improved as δ increases. Although a large volumetric efficiency is desired, the drag increases with δ2, as shown in Fig. 6. Thus, to avoid the drag penalty to a large volumetric efficiency, and maximize the liftto-drag ratio, we chose δ = 5.5o. and �ι = 30o. The final configuration is shown in Fig. 7.

(14)

(15

The design involved a tradeoff analysis in order to find the dihedral angle �ι and the base cone angle δ that would !! and the volumetric maximize the lift-to-drag ratio !! ! efficiency ! ! !!!. This analysis is summarized in Figs. 4 and 5. It must be pointed out that our design option ! was !! ! !" . Moreover, since !!! ! ! ! ! we chose !! 130

Lift -to-Drag ratio

! ! ! ! !! !!! ! ! ! !!! !" !! ! !!

Volumetric Efficiency

!! !!! ! !!!" . In fact, examining Eqs. 4 and 5 one can see that low values of !! !!! could result in final geometries with excessive bluntness and, ultimately, large drag. On the other hand, values of !! !!! close to 1 produce very slender geometries.

8.0 7.8 7.6 7.4 7.2 7.0 6.8

0.20 0.18 0.16 0.14 0.12 4

6 I°

0.10

Volumetric Efficiency

!!! !

Lift -to-Drag ratio

Moreover, it was shown that θ∞b is single valued for

Figure 5. Effect of the base cone angle on the waverider characteristics, with ϕι = 30o.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

Drag coefficient

Experimental results of a Mach 10 conical-flow derived waverider to 14-X hypersonic aerospace vehicle

of investigation. Figure 8 shows the flow schematics, the free stream flow is compressed by the forebody which produces a conical shock wave, and then the flow undertakes turn in the scramjet inlet The flow behind the conical shock was calculated using the oblique shock relationship, which is a fairly good approximation for the conical shock at stations far from the base body surface.

0.006 0.004 0.002 0

6 I°

free stream inlet shock wave

Figure 6. Drag coefficient for several base cone angles, with �ι = 30o .

body z

conical shock wave

cowl

Figure 8. Scheme for the flow on the compression surface.

y x

z x

Figure 7. Generated waverider, for � = 30°, M = 10, δ = 5.5° and Ro/Xs = 0.75.

b) Scramjet inlet The design goal of the compression system is to provide the desired pressure and temperature for the supersonic combustion over the entire flight range with minimum losses. Bearing this fact in mind, one must consider that after the conical shock wave produced by the leading edge and after the compression ramp, the flow must reach the adequate conditions, or close to them, for autoignition of the Hydrogen-Air mixture, pressures between 25-100kPa and temperatures between 1000-2000K (O’neill, et al., 1992). Furthermore, due to the fact that a hypersonic waverider in this study would be used for transatmospheric missions, the change of the air properties with flight altitude must be accounted, this was used to define the limits for the scramjet operation in the present analysis. The inlet ramp adopted was a single turn ramp. Also, twodimensional, calorically perfect and inviscid flow, were assumed in that region. These assumptions substantially simplify calculations for the ramp geometry to be matter

60 Altitude (km)

The results are shown in Fig. 9. As one can see, as the ramp angle increases, the maximum allowed altitude increases while the minimum increases also, the operating corridor (upper minus lower limits) remains quite constant of the order of 10 km. As a design option, we chose an operating altitude range of 40 to 50 km. Consequently, the ramp shock wave angle was found to be 25°, and the ramp angle was calculated giving 20° with respect to the z-axis.

upper limit

operating corridor

40 30 20

lower limit

10 20 30 40 Ramp shock wave angle (degrees)

Figure 9. Relationship between the secondary-shock wave angle and the required altitude limits. According to the International Standard Atmosphere model.

c) Combustor and nozzle Although the combustor and expansion systems were not the aim of the work presented here, a 108 mm x 270 mm long flat surface and a 115 mm long constant slope ramp of 15° were integrated to the compression surface. d) Final conﬁguration The model tested in the T3 tunnel is shown in Fig. 10. In this figure one can see the location of a set of 7 piezoelectric pressure transducers used to assess the

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

131

Rolim, T.C. et al

pressure field on the compression surface. The 781 mm x 327.5 mm model was machined with a CNC milling machine and made by stain less steel, Fig. 11 shows photographs of the actual model.

111.1 64.2

33.3

724.8 675.7

405.7 455.7 619.3 706.3

conditions for the supersonic nozzle located at the end of the driven section. The hypersonic shock tunnel T3 was used in the present work. This shock tunnel is located at the Laboratory Prof. Henry. T. Nagamatsu of Institute for Advanced Studies (IEAv) from Department of Science and Aerospace Technology (DCTA). That hypersonic facility comprises of two other shock tunnels T2 and T1, 11.50 m and 7.80 m long, respectively. The T3 is 17.50 m long with a test section diameter of 610 mm. The T3 has a moveable sting aligned with the nozzle axis that permits an easy longitudinal alignment of the models; a commercial masonry level is generally used for lateral alignment.

dimensions in mm

b) Test campaign

Figure 10: Location of pressure transducers.

The test conditions were controlled by the driver todriven pressure ratio along with the driver gas choice (Helium or dry air). That is because Helium produces higher stagnation enthalpies than air. The Mach number of 10 was fixed by the nozzle geometry. The test matrix is shown in Table 1. Piezoelectric pressure transducers were used for measuring the reservoir pressure and the incident shock wave speed in driven section. These data were used to estimate the total temperature and free stream conditions. The last four runs, with lower Mach numbers, were made with the model inside the nozzle, in order to better visualize the flow at the inlet.

c) Airflow visualization Figure 11: Actual machined model.

EXPERIMENTAL WORK a) T3 shock tunnel facility The large stagnation enthalpy and pressure required for simulation of hypersonic flow make the shock tunnel facilities the only applicable tools for this purpose. The shock tunnel consists of a high pressure section (driver section) separated by means of a diaphragm section from a low pressure section (driven section). When the diaphragm bursts a shock wave propagates toward the driven, while an expansion wave propagates into the driver. The shock wave interacts with the cold air in the driven section, increasing pressure and temperature. Once the shock wave reaches the nozzle diaphragm, it is reflected and interacts with the contact surface – the interface between the gases. After few interactions and reflections, the resulting high temperature and high pressure are used as stagnation

132

As shown in Table.1, the free stream static temperature varied from 38 to 122K and static pressure from 0.13 to 0.89kPa. The Reynolds number ranged between 2.25 x 106 to 8.76 x 106 (m-1). These Reynolds numbers indicates that the flow is laminar over a considerable distance from the leading edge of the model. Furthermore, the low Knudsen numbers suggest continuum flow conditions to be expected throughout the investigated model, except inside the slip region. Although the forebody compression surface presents some longitudinal angle variation, detailed observation of the model sketches can show that air flow turning is so slight – in fact the turning angle reaches 3° as a maximum before the inlet – that the flow in this region can be compared with the flow over a flat plate under the same conditions. Several works (Nagamatsu, et al., 1961; Nagamatsu, et al., 1960; Hayes, et al., 1959) have shown the existence of the three different flow field regions downstream the leading edge of a

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

Experimental results of a Mach 10 conical-flow derived waverider to 14-X hypersonic aerospace vehicle

Table 1. Test matrix.

Stagnation conditions Freestream conditions Pressure Temperature Static Static Mach Reynolds Number Knundsen (MPa) (K) pressure (kPa) temperature (K) Number (M) (Re)(m-1) Number (km) 15.0 2150 0.30 121.0 10.0 2.25 x 106 0.19 18.2 1624 0.35 87.9 10.0 4.36 x 106 0.11 6 16.0 1535 0.31 82.5 10.0 4.20 x 10 0.12 18.2 1820 0.48 109.2 9.4 3.96 x 106 0.11 20.3 1558 0.48 87.9 9.6 5.73 x 106 0.08 5.3 795 0.17 41.6 9.7 7.03 x 106 0.11 5.4 715 0.19 38.0 9.6 8.76 x 106 0.09 5.0 834 0.13 41.3 10 5.58 x 106 0.15 20.3 1706 0.40 92.8 10.0 4.48 x 106 0.11 6 20.0 1840 0.74 122 8.9 4.91 x 10 0.08 19.1 1730 0.12 114.3 8.9 5.35 x 106 0.07 20.3 1757 0.11 114.8 9.0 5.41 x 106 0.07 19.6 1702 0.13 116.9 8.7 6.17 x 106 0.06 flat plate. The three distinct flow regions are the slip region, the strong and weak interaction regions (Toro, et al., 1998). Even under continuum flow conditions (kn<<1) the existence of slip near leading edge for very large Mach numbers was reported elsewhere (Nagamatsu, et al., 1961). According to the theory developed by those authors, the length of the slip region can be assumed as proportional to the free stream Mach number. Indeed, following the same procedure as taken by Minucci (Minucci, 1991) it was possible to estimate the slip region length as between 0.03 and 0.1 mm. In this region, which can be considered a rarefied flow, the Boltzmann equations are more suitable. As it can be noted in the schlieren photograph in Fig. 12, the flow pattern around the leading edge no longer presents a bow shock as in the continuum case, instead, near the leading edge the shock wave starting point was almost imperceptible, a main characteristic of a slip condition.

Furthermore, the region that follows the slip region presents shock-wave/boundary-layer interactions, the extent of the influence of this region downstream depends on the size of the subsonic portion of the boundary layer and on the strength of the shock wave (Bertin, 1994). When the rate of growth of the boundary layer is large, the boundary layer and the shock wave are merged within a limited region. In this situation, the outer inviscid flow is strongly affected by the displacement thickness, which in turn substantially affects the boundary layer. This mutual process is called strong viscous interaction (Anderson, 1989). As stated before, due to the low Reynolds number, one can assume that the boundary layer is laminar. The similarity parameter that governs laminar viscous interactions is given by:

! !!

!!!

!! !! !! !!

(16)

Rez is based on the free stream properties, calculated at a distance z from the leading edge. While M∞ is the free stream Mach number, μ is the dynamic viscosity ρ is the density and the subscript w relates to the wall and e to the edge of the boundary layer, respectively. The effects of the hypersonic viscous interactions on the pressure distribution over a flat plate as function of the parameter χ were presented in several works (Nagamatsu, et al., 1961; Hayes, et al., 1959; Minucci, 1991; Anderson, 1989). A common result is that the induced pressure change varies linearly with χ .Thus, one can write: Figure 12: Mach 10 flow past the model leading edge. Reservoir conditions: 20.3 MPa and 1706 K.

!! ! ! ! ! ! !!

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

(17)

133

Rolim, T.C. et al

ρ is the static pressure; ∆ρ is the static pressure perturbation; F is a function of the ratio of the specific heats and the wall temperature condition. Another important result relates the leading-edge shockwave angle β with the local Reynolds number inside the strong interaction region as below: !!

!! ! !! !! !!! ! !!!! !" !! !!

Shock Wave Angle (degrees)

where: 25 20 15 10

inviscid, L = 1.4 value

5 0

0.7

1.2

2.2

1.7

2.7

!!!

(18)

Figure 13: Primary shock wave angle variation with the parameter !! !! . Same flow conditions indicated in Fig. 12. ! ! !! !! !!

where δ* is the displacement thickness. Following, in the weak interaction region, the displacement effects are small enough so the inviscid flow does not interact with the boundary layer. Also, small pressure gradients inside the boundary layer permit the use of the Blausius solution for a viscous flow over a flat plate, adapted for compressible flows. Thus, the displacement thickness variation with the local Reynolds number within the weak interaction region is so as (Anderson, 1991):

!! ! !!!" !! !" !!!

(19)

where G accounts for the compressibility effects, but also it is a function of M∞, the wall temperature, the Prandtl number and the boundary-layer-edge temperature. After careful data analysis of a high definition version of Fig.12, the shock wave angle was measured using a computational grid on the image, thus its dependence on the viscous interaction parameter was investigated. The results are shown in Fig.13. It is evident that the curve follows the same pattern as seen in a strong interaction region at foremost stations. In fact, as previously stated, the small shock-wave angle and the low Reynolds number imply a merged shock-wave/boundary-layer. Moreover, the measured angles were quite larger than the predicted by the inviscid, calorically perfect theory (with ratio of specific heats, γ = 1.4). This result is reasonable since the displacement thickness δ* modifies the effective body; the effect is exacerbated by the adverse pressure gradient due to the slight turning of the flow along the forebody region. Besides, the author believes that the nozzle’s nonequilibrium effects certainly impart some deviations from the calculated inviscid value at stations far from the strong interaction zone, since γ must be considered a function of the local pressure and temperature. 134

Another aspect of the hypersonic flow which was observed concerns the flow profile over the inlet ramp. Like the flow over a flat plate, the shock-wave/boundarylayer interaction depends largely on the length of the subsonic portion of the boundary layer and on the shock wave strength. The adverse pressure gradient causes the boundary layer to thicken as it approaches the ramp deflection. If the conditions are such that the boundary-layer separates, as depicted in Fig. 14, a series of compression waves is formed in the separation point and they coalesce into a single curved shock-wave. Downstream that point, the region of separated flow features unsteady nature and large gradients. In the reattachment point, the inviscid flow along the effective ramp encounters the actual ramp, a phenomenon which causes an incremental compressive turning. The compression waves formed in the reattachment point coalesce to another shock wave. Finally, the separation and the recompression shock waves interact downstream. In this model, the inviscid flow impinges two successive compression ramps, the

Intersection separation shock wave

reattachment shock wave reattachment point

separation point

recirculation zone

Figure 14: Main characteristics of a separated flow over a compression ramp. Adapted (Bertin, 1994).

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

Experimental results of a Mach 10 conical-flow derived waverider to 14-X hypersonic aerospace vehicle

first one represents the separated region and the second one the region of reattached flow. Occasionally, the shock waves intersect at some point.

be fully expanded, such as at station z/Lw = 0.58, the pressure is slightly lower than at the initial part of the expansion fan, z/Lw = 0.52.

Figure 15 shows a schlieren photograph of the flow over the rear region of the compressive ramp. With the fins mounted, the complete flow development over the inlet ramp could not be observed. However, the coalescent compression waves, that seem to intersect downstream the frame region, can be seen. Their extensions indicate a large separated region compared with the ramp length. In addition to that, it was possible to infer the angle of its linear portion as being 27°, only 8% larger than the inviscid, calorically perfect value, 25°.

Furthermore, in Fig. 16, from stations z/Lw = 0.58 to 0.86, there is a pressure increase despite the fact that station z/Lw = 0.86 was situated over the expansion ramp. Although a more complete pressure survey is needed, it is believed that flow separation is also possible to take place on the flat portions between stations z/Lw = 0.58 and z/Lw = 0.86, since the boundary layer thickness is increasing along the model length. Thus, it is entirely possible that a “bubble” shaped shock forms between these stations (Bertin, 1994). Also, from stations z/ Lw = 0.86 to 0.93, located over the expansion ramp, a pressure increase can be noted, probably indicating that at the nozzle the flow separation and reattachment occur before point z/Lw = 0.93 . CONCLUSIONS

Figure 15. Mach 8.7 flow past the inlet ramp. Reservoir conditions: P0 = 19.6 MPa , T0 = 1702 K.

d) Static pressure measurements One can see in Fig. 16 the static pressure variation with the distance from the nose leading edge along the centerline of the model during the first three ! tests. The correspondent parameter !! ! !!! was indicated in that figure. The inviscid solution for the pressure distribution over the pure waverider surface was also inserted. Regarding that figure it was found that the pressure decreases from z/Lw = 0.52 to 0.58, just after the compression ramp. This result is believed to be consequence of the existence of a non-centered expansion fan formed at the end of the ramp, caused by a separated region. Thus, where the flow is believed to 3.0 2.5 2.0

Moo3/Reoo1/2 0.6726 0.4849 0.4813 Inviscid

p/p

1.5 1.0 0.5 0.0 0.5

0.6

0.7 0.8 z/Lw

0.9

1.0

Figure 16. Pressure distribution along the model center line. Lw is the model length.`

An experimental investigation of a 781mm waverider model was performed at the Laboratory Prof. Henry. T. Nagamatsu/IEAv/DCTA. Schlieren photographs and static pressure measurements were made for further flow analysis. The waverider surface was constructed according to the Hypersonic Small Disturbances Theory (Rasmussen, et al., 1990), scramjet compression and expansions ramps were integrated to the pure waverider surface. The stagnation conditions as well as the free stream properties were estimated using the numerical codes. The tunnel operated at Mach number ranges of 8.9 to 10, Re = 2.25 x 106 to 8.76 x 106 (m-1) and Kn = 0.06 to 0.19. As results, the schlieren photographs demonstrated the attached shock wave, a key fact of the waverider concept, and the formation of the compression waves in the inlet region. Furthermore, the pressure distribution over the compression surface was measured. The pressure rise found downstream the centerline is believed to have been caused by the boundary layer thickness increase and consequently flow separation. However, it’s believed that the entire flow complexity was not properly depicted by this investigation and shall be subject of future works. ACKNOWLEDGEMENTS The authors hereby express their gratitude to: Agência Espacial Brasileira (AEB), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq),

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

135

Rolim, T.C. et al

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (Capes), Fundação de Amparo à Pesquisa do Estado de São Paulo (Fapesp), and Financiadora de Estudos e Projetos (Finep). REFERENCES Anderson, J. D., 1991, “Fundamentals of Aerodynamics”, New York , MacGraw-Hill, 2nd Edition. Anderson, J. D., 1989, “Hypersonic and High Temperature Gas Dynamics”, New York , McGraw-Hill. Bertin, J. J., 1994, “Hypersonic Aerothermodynamics”, New York, AIAA Education Series. Hayes, W. D. and Probstein, R. F., 1959, “Hypersonic Flow Theory”, New York, Academic Press. Heiser, W. H., et al., 1994, “Hypersonic Airbreathing Propulsion”, AIAA Education Series. Javaid, K. H., Serghides, V. C., 2005, “Airframe Propulsion Integration Methodology for Waverider Derived Hypersonic Cruise Aircraft Design Concepts”, Journal of Spacecraft and Rockets, AIAA, No. 5 , Vol. 42. , pp. 663-671. Kim, B. S., Rasmussen, M. L., Jischke, M. S., 1983, “Optimization of Waverider Configurations Generated from Axisymmetric Conical Flows” Journal of Spacecraft, AIAA, No. 5 , Vol. 20. , pp. 461-469. Minucci, M. A. S., 1991, “Experimental Investigation of 2-D Scramjet Inlet at Flow Mach Numbers 8 to 25 and Stagnation Temperature of 800K to 4100K”,

136

Ph.D. dissertation, Aeronautical Engineering Dept. Ressenlaer Polytechnique Institute, Troy, NY. Nagamatsu, H. T., Sheer, R. E. Jr, 1961, “Hypersonic Shock Wave-Boundary layer Interaction and Leading Edge Slip”, American Rocket Society Journal, No. 5, Vol. 30, pp. 454-462. Nagamatsu, H.T., Li., T. Y., 1960, “Hypersonic Viscous Flow Near the Leading Edge of a Flat Plate” Physics of Fluid., No. 1, Vol. 3, pp. 140-141. Nonweiler, T. R. F., 1963, “Delta wings of Shape Amenable to Exact Shock Wave Theory”, Journal of the Royal Aeronautical Society, Vol. 67, p. 39. O’neill, M. K., Lewis, M. D., 1992, “Optimized Scramjet Integration on a Waverider”, Journal of Aircraft, AIAA, No. 6 , Vol. 29, pp. 1114-1121. Rasmussen, M. L., He, X., 1990, “Analysis of ConeDerived Waveriders by Hypersonic Small-Disturbance Theory”, Proceedings of the First International Hypersonic Waverider Symposium. Rault, D. F. G., 1994, “Aerodynamic Characteristics of Hypersonic Viscous Optimized Waverider at High Altitudes”, Journal of Spacecraft and Rockets, No. 5, Vol. 31, pp. 719-727. Toro, P. G. P., et al, 1998, “Hypersonic Flow Over a Flat Plate”, 36th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV. Wang, Y., Zhang, D., Deng, X., 2007, “Design of Waverider Configuration with High Lift-Drag Ratio” Journal of Aircraft, No. 1, Vol. 44, pp. 144-148.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 127-136, May-Aug., 2011

doi: 10.5028/jatm.2011.03021711

Leonardo Ostan Bitencourt

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability

Gregori Pogorzelski

Abstract: The work reported in the present paper was performed to honor the centennial of the flight by Alberto Santos Dumont with his 14-Bis aircraft. The paper describes results for a computational fluid dynamics (CFD) analysis of the 14-Bis aircraft aerodynamics and flight stability. The 14-Bis aircraft geometry was generated from historical sources and observations. CFD computations were performed using well-established commercial codes for calculation of the historical flight conditions. Simulations considered a Reynolds-averaged Navier-Stokes formulation, in which turbulence closure was achieved by using Menter’s model. The flight conditions investigated were primarily concerned with historical observations regarding flight speeds and the need for a more powerful engine, as well as flight stability characteristics of the 14-Bis airplane, which are unknown up to the present day. The results led to qualitative agreement with historical reports, although quite interesting conclusions could be drawn with regard to the actual aerodynamic flight speeds and the aircraft stability parameters. Keywords: Aerodynamics, CFD, Centennial of flight, Santos Dumont, 14Bis aircraft.

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil lostan@gmail.com

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil gpogor@gmail.com

Ramon Morais de Freitas

Instituto Nacional de Pesquisas Espaciais São José dos Campos/SP – Brazil aeroramon@gmail.com

João Luiz F. Azevedo*

Instituto de Aeronáutica e Espaço São Jose dos Campos/SP – Brazil joaoluiz.azevedo@gmail.com *author for correspondence.

INTRODUCTION The year 2006 marked as the centennial of the historical, heavier-than-air flight by Alberto Santos Dumont, with his 14-Bis aircraft. The present work, which was actually performed around 2005–2006 (Bitencourt et al., 2005; 2006; and Freitas et al., 2006), but was never published in an archival journal, came about as an attempt to honor Santos Dumont’s flight centennial. Since the authors work with computational fluid dynamics (CFD), it seemed appropriate that a way to celebrate Santos Dumont’s accomplishments was to study some of his historical flights using CFD technology. Hence, in this context, the present paper describes the results of a CFD-based analysis of the 14-Bis aircraft aerodynamics and flight stability. The flight conditions investigated were primarily concerned with historical observations regarding flight speeds and the need for a more powerful engine, as well as flight stability characteristics of the 14-Bis airplane. It must be emphasized that such stability characteristics are unknown up to the present day. Therefore, there is no way of actually validating the present simulations, but it became clear that results of the effort here undertaken led to qualitative agreement with historical reports about the 14-Bis aircraft flights. Moreover, some quite interesting conclusions could be drawn with regard Received: 10/05/11 Accepted: 06/07/11

to actual aerodynamic flight speeds and aircraft stability parameters, as this paper will attempt to convey. On October 1906, in the Bagatelle Field, Paris, France, Santos Dumont flew the 14-Bis aircraft and won the Deutsch-Archdeacon Prize for the first officially observed heavier-than-air powered flight. The 14-Bis was constructed from pine wood, bamboo poles, and covered with Japanese silk. The aircraft had a complex canardbiplane configuration, which was a construction based on Hargrave’s box kites. The Hargrave cell in the nose pivoted up and down to act as an elevator and from side to side in the role of a rudder. The wings were rigged with 10 deg. of dihedral, and the first flights were made without ailerons. The preliminary flight tests happened with the 14-Bis aircraft held by Santos Dumont’s No. 14 dirigible. The 14-Bis flew without the dirigible on September 13, 1906, making a hop between 6 and 13 m. The original power plant was a 24 hp Antoinette engine, but this was later upgraded to the 50 hp Antoinette engine on the October 23 flight, when Santos Dumont managed to fly for 60 m. Such flight is indicated in Fig. 1. Then, on November 12, flying 220 m in 21 1/2 seconds, with members of the Aero-Club de France in attendance, he won a prize of 1,500 Francs for performing the first powered flight of over 100 m in Europe. Since he was observed by officials from what would become the Federation Aeronautique Internationale, Santos Dumont was credited with making the first heavier-than-air powered flight (Vilares, 1956).

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

137

Bitencourt L.O. et al

where u, v and w are the velocity vector Cartesian components and ei is the internal energy.

Source: Museu Aeroespacial, Rio de Janeiro, Brazil.

Figure 1. 14-Bis in flight on October 23, 1906.

CFD techniques have emerged as a serious alternative for aerodynamic analysis in the last 30 to 40 years. Such techniques are able to reduce project costs, since time and money spent with wind tunnel testing are substantially reduced. In addition to this, CFD has the advantage of numerically solving the fluid equations in the entire flow field, thus allowing for local analysis of the flow properties in a much more detailed way than any wind tunnel visualization techniques could show. The main objective of the present paper is to apply CFD techniques for aerodynamic analysis of the 14-Bis aircraft. The approach, here adopted, involves the computation of the aerodynamic characteristics for the aircraft, at the presumed flight conditions, in order to assess and clarify some controversial points regarding stability, flight speed, ground effect, and power plant performance. The study also explores angle of attack and velocity variations around the historical data. THEORETICAL FORMULATION The Navier-Stokes equations constitute the most general flow formulation for which the fluid continuum hypothesis can be assumed. The Navier-Stokes equations, for a perfect gas, without the generation of heat and with negligible field forces can be written as yW y( Wu j ) " 0, yx j yt y( Wui ) yt

y( Wui u j ) yx j

(1)

yp yY ij "0, yxi yx j

ye y[( e p )u j Y ij ui q j ] "0, yx j yt

(2) (3)

where ρ, p and ui are the fluid density, pressure and velocity, respectively, τij represents the viscous stress tensor components, qj is the heat flux vector and t is the time. The e term is the total energy per unit of volume, given by ¼ ¬ 1 e " W ei (u 2 v 2 w2 ) ½ 2 ® ¾ 138

(4)

In the formulation actually solved in the present work, two additional assumptions are adopted: the absence of heat transfer, i.e., the heat flux vector terms are equal to zero, and the flow is treated as incompressible, due to the low Mach number values here considered. Velocities achieved by the 14-Bis aircraft during flight correspond to, at most, a Mach number of 0.05. Furthermore, since turbulent effects can be important in the present case, flow analysis is performed using the Reynoldsaveraged Navier-Stokes equations. These equations contain the mean variables and a certain number of terms representing the turbulence effects, which must be modeled. Turbulence closure is achieved using Menter’s shear-stress transport (SST) turbulence model (Menter, 1994). NUMERICAL APPROACH Flow solver The present computations considered unstructured grids, and they have been constructed using the CFX code (CFX, 2005), which is a well-known commercial code currently available. The solutions of the turbulent flows of interest are based on the Reynolds-averaged Navier Stokes (RANS) equations, supported by Menter’s SST turbulence model (Menter, 1994). In the present case, the CFX solver simulated steady, viscous and incompressible flows around the 14-Bis model. This code uses a cellvertex, finite element-based control volume method. An iterative, second order, time marching scheme is used to numerically solve the RANS equations. To decrease the computational time, some convergence acceleration techniques, such as the algebraic multigrid (MG) procedure, and parallel computations are used in the simulations. Grid generation The 14-Bis computer-aided design (CAD) geometry is generated from planform and historical source observations (Greco and Ribeiro, 2003). The authors have tried to express the real forms of the airplane as much as possible. As discussed, the aircraft had a complex canard-biplane configuration, which was a construction based on Hargrave’s box kites. Sketches of the airplane geometry and configuration can be seen in Fig. 2. The main 14-Bis geometric characteristics are presented in Table 1. The flow domain about the geometry is discretized using unstructured grids.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability

Since memory and processing capabilities are limited, the geometry is simplified keeping only the main components, i.e., wings, canard and fuselage. Figures 3, 4 and 5 show a parallel between the original geometry and the simulated one. The grid generator software used, ICEM-CFD (2005), allows the automatic creation of the hexahedral grid. Initially, it was thought that such grids would be preferable because viscous solutions are being sought. However, the resulting surface

Figure 4. Mesh view of idealized configuration.

Table 1. 14-Bis geometric characteristics.

Total Canard Area Canard Chord Canard Span Length Wing Chord Wing Span Historical Flight Speed Wing Chord Re Wing Dihedral Canard Chord Re Total Wing Area Canard-Wing Distance Weight with Pilot Engine Power

8 m2 2m 2m 10 m 2.5 m 11.50 m 9 to 12 m/s 107 10 deg. 107 50 m2 5m ~ 300 kg 24 hp (initially)-50 hp (afterwards) Figure 5. 14-Bis CFD model with streamlines.

2.5

1.5

mesh over the airplane has a poor arrangement, when such mesh generation methodology is applied to the present configuration. Hence, the strategy adopted was to first create a structured 2-D grid over the geometric surface.

11.55

9.56

Figure 2. Sketches of airplane geometry and configuration.

Source: EESC-USP.

Figure 3. Original CAD model.

Afterwards, the Delauney method (Field, 1987) is applied, generating the desired unstructured volumetric grid. To assure a faster convergence and a good solution, the mesh quality must be taken into account. Therefore, the element size transitions are gradually performed. Furthermore, regions of leading edges, trailing edges and the ones containing wakes received appropriate grid refinement to avoid spurious solutions. The guidelines used to define and construct such regions of additional refinement, as well as the overall volume mesh distribution, followed the best practices that had been developed during an innovation project, which involved several research institutions in the country and Embraer. The innovation project was conducted in the 2002â&#x20AC;&#x201C;2006 period with FAPESP sponsorship (Azevedo, 2006). It must be understood that there has been no attempt, in the course of the present work, to perform systematic grid refinement studies for the solutions reported herein.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

139

Bitencourt L.O. et al

Furthermore, there are no claims in the paper that the presented solutions are grid independent. Clearly, the authors acknowledge that both concepts are very important issues in CFD studies. However, as anyone, who has had the opportunity to use such CFD techniques in an actual industrial environment, i.e., for simulations over real life configurations, will certainly report, both conditions are truly difficult to be fully satisfied. As a matter of fact, the concept of what exactly constitutes a systematic grid refinement for a fully unstructured grid over a complex configuration is simply not yet defined by the CFD community. Moreover, the concept of a grid independent solution must be approached with care, in particular in view of recently reported calculations that use over three billion grid points (Pulliam, 2011), and it is not completely clear that the solutions could be called grid independent. Boundary conditions The correct application of boundary conditions is vital to properly close the numerical problem, assuring correct modeling. The INLET condition is applied along the computational domain entrance surface. In this boundary, the freestream speed and its direction are specified. The NO-SLIP WALL condition is used on the aircraft surface, as usual for a viscous simulation, and it assures that neither tangential nor normal velocity components are present along the airplane surfaces. The OUTLET condition is used to model the fluid flow at the domain exit. In the simulations here performed, the atmospheric pressure was specified as an exit pressure at this particular domain boundary. The SLIP WALL condition is used on the surface just below the airplane in order to model the ground effect. It should be observed that the normal velocity component is kept zero and the surface moves with the freestream speed, i.e., the tangential velocity component of the surface is equal to the freestream velocity, under this condition. Finally, the OPENING condition models a boundary condition which allows entrance and exit of fluid freely. This boundary condition is used for all other external boundary surfaces of the computational domain. It should also be pointed out that the atmospheric pressure is also specified for such boundaries. The nomenclature used here is the one adopted by the CFX solver (CFX, 2005). A general overview of the computational domain can be seen in Fig.6. Post-processor for aerodynamic forces The post-processor, by means of simple and useful tools, allows the evaluation of aerodynamics forces and the observation of the flow field variables as, for example, 140

Aircraft

Exit

Entrace

Lateral Sides

Figure 6. General overview of the computational domain.

pressure contours, streamlines, as indicated for instance in Fig. 5, or boundary layer velocity profiles. The resultant force in the airplane, when projected into the wind axis, results in drag, lift and yaw force components. The evaluation of these aerodynamic forces is performed by integrating the surface pressure distributions and shear stresses, as shown in Eq. (5). Such methodology for the calculation of aerodynamic forces and moments is called the near-field approach. A more detailed description of these force integration methods can be found in van der Vooren and Slooff (1990). I Fnear " µ S

near

I ¬( p " p ) I Y ¼ ndS h ® ¾½

(5)

The aerodynamic drag is a force exerted, by the flow field, on the body surface in a direction contrary to its movement. The drag is the summation of the tangential or skin friction forces, and surface pressure or normal forces, projected into the freestream direction. The drag breakdown with the near-field drag computation approach, as described in van der Vooren and Slooff (1990), comprises the pressure and the friction drag components. From the evaluation of forces and moments over the airplane for several flight conditions, i.e., varying the angle-of-attack or the canard angle, the authors are able to extract the relevant aerodynamic coefficients. With such data, one can analyze details of the 14-Bis flight conditions and possible stability range. The aerodynamic coefficients evaluated in the present work are only valid for small angles-of-attack because, since steady flow conditions are assumed, the calculations beyond stall would be incorrect. TEST CASES The chosen test cases explore the main aerodynamic characteristics of the 14-Bis airplane. This parametric study includes 46 simulations, involving five major objectives, namely:

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability

•

to verify the speed influence over the aerodynamic coefficients;

•

to verify the overall aerodynamic behavior at different angles-of-attack and to determine the drag polar;

•

to analyze the canard deflection influence over the aircraft;

•

to study the aircraft aerodynamics when submitted to sideslip angles;

•

to verify the extent of the ground effect.

Velocity variation studies allow the verification that aerodynamic coefficients do not change with the flow speed. It must be clear that there has been no attempt, in the present investigation, to conduct a study of Reynolds number influence in the aerodynamic results. The variations in flight speed are merely attempting to ascertain that, in the range of flight speeds here considered, the aerodynamic coefficients are essentially insensitive to such variations. Moreover, such studies also allow finding the most probable flight speed, which is not exactly known because historical sources are not in agreement. Through the angle of attack variation studies, it is possible to estimate lift, drag, and moment derivatives. The canard incidence angle variation allows the estimation of some stability derivatives. Moreover, ground effect influence is verified through variation of the distance from the airplane to the ground. All the historical registries only take into account the airplane velocity relative to the ground, but it would be more interesting, in an aerodynamic point of view, to obtain the wind relative velocity. Hence, a range of velocities was tested. The interference between the main airplane parts is also addressed. A summary of the test cases analyzed is presented in Table 2.

coefficients. As it is well-known, for low speed subsonic flight, the general aerodynamic characteristics of an aircraft must have a weak dependence on the flight velocity. Table 3 shows the results concerning the longitudinal aerodynamic coefficients to four different simulated speeds: 7.5, 9.5, 11.5, and 14.0 m/s. The other important flight parameters, namely angle of attack, sideslip angle and elevator deflection, are set to 0 deg., as indicated in Table 2. The results in Table 3 show maximum relative differences of 0.29, 2.48 and 6.35% for lift (CL), drag (CD) and pitching moment (CM) coefficients, respectively, in the speed range analyzed. The relatively small differences encountered indicate that the aerodynamic coefficients can be treated as independent of the flight velocity. This is further supported by the fact that the range of aerodynamic coefficient variations are probably inside the uncertainty range induced by the model geometrical simplifications adopted as, for example, the ignored aircraft elements, such as the wheels, which certainly would increase the drag coefficient. Therefore, in the following analyses and discussions, the consideration of aerodynamic coefficient independence with respect the flight speed is adopted. This is especially important when a linear aerodynamic model of the aircraft is developed and applied with constant control and stability derivatives over different flight speed values. The next set of simulations is concerned with the aerodynamic characteristics of the aircraft under an angle of attack variation with no canard deflection. The flight speed of 11.5 m/s is adopted as the default value for such simulations, because this is the value of flight speed closest to the reported historical one. A range of angles of attack, varying from 5.0 to 6.5 deg., is considered. The presence of nonlinear effects, probably related to the growth of the separated regions, together with time and computational resource constraints did not allow the exploration of the flow under higher angles of attack. Figure 7 presents the lift coefficient behavior as a Table 3. Longitudinal aerodynamic coefficients at different flight speeds.

RESULTS AND DISCUSSION

Speed (m/s)

General aerodynamic results The first set of simulations performed had the objective of verifying the influence of flight speed over the aerodynamic

7.5 9.5 11.5 14.0

0.8501 0.8511 0.8516 0.8526

0.1002 0.0988 0.0979 0.0977

-0.0606 -0.0623 -0.0632 -0.0645

Table 2. Simulated test cases for the parametric study of the main aerodynamic characteristics of the 14-Bis airplane.

Parameter V∞ α δp β ∆

Variation 7.5 to 14 m/s -5 to +6.5 deg 0 to 7.5 deg. 1 to 7.0 deg. 0 to 6 m

Fixed Conditions Variation of V∞ for α = 0 deg. Variation of angle of attack with V∞ = 11.5 m/s. Variation of canard incidence angle, α = 0 deg. Variation of the slide slip angle. Variation of the airplane distance from the ground.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

141

Bitencourt L.O. et al

The aircraft drag polar is shown in Fig. 8. To analytically represent the data, a polynomial curve fit of second degree is fitted to the data points. As indicated in the figure, the polynomial is a good approximation and it is adopted as the drag polar model. The fitted polynomial is given by

The aerodynamic efficiency behavior, in terms of the lift-to-drag ratio, at different angles of attack can be seen in Fig. 9. A considerable loss of efficiency can be observed as the angle of attack increases. For instance, a variation of 61% in the L/D values is found between the two extreme points represented. The explanation for such behavior can be found in Figs. 7 and 8. In other words, whereas CL grows linearly with the angle of attack, the drag coefficient increases significantly after CL values of approximately 0.6, which correspond to an angle of attack of 3 deg., as one can see in Fig. 7. In fact, in Fig. 9, one can verify that, for angles of attack higher than -3 deg.,

-4.00

-2.00

0.00

2.00

Alfa (deg.) Reference area: Sref = 28.75 m2.

Figure 7. Aircraft CL x α curve. 142

0.2 0.15 0.1 0.05

(6)

An important observation, with respect to the drag polar and the general drag results obtained, is concerned with the geometric simplifications assumed in the simulated model. One can expect that the missing components, such as struts, landing gear and even the pilot, should increase the drag when compared to the current calculations. Nevertheless, it is hoped that the computational drag polar still gives enough information to a first analysis of the airplane, allowing reasonable drag predictions when simulating the historical flight conditions.

-6.00

0.25

4.00

6.00

8.00

0.2

0.4

0.6

0.8

1.2

1.4

1.6

Figure 8. Aircraft drag polar.

L/D

CD = 0.089 – 0.206 CL + 0.252 C2L

0.3

function of angle of attack. As can be noted, a general linear pattern is observed, except maybe for the last two points, where some nonlinear effects might be beginning to appear. From the results in Fig. 7, the value of the CL∞ derivative can be extracted as CL∞ = 4.85 rad -1.

-5.00

-3.00

-1.00

1.00

3.00

5.00

7.00

Alpha (deg)

Figure 9. Aircraft lift-to-drag ratio (L/D) curve.

L/D values are quite reduced due to, most probably, the fairly large induced drag produced by the aircraft. It is estimated in Vilares (1956) that the 14-Bis aircraft first flight speed was about 11.5 m/s. It must be pointed out that this is a mean speed value using the ground as reference. The wind influence over the airplane speed is not considered in such estimate. Therefore, the aerodynamic speed could be different from the historical measured value of 11.5 m/s. In addition, there is also the speed variation during the acceleration procedure. From this information, it is possible to conclude that the true air speed could actually have been higher than the estimated mean value. With the objective of having a better estimation of the most probable speed value, a parametric analysis of this variable influence over the aircraft lift and drag was performed. According to Greco and Ribeiro (2003), the aircraft mass was about 300 kg. Therefore, a lift force larger than 3,000 N must have been generated to allow the flight. The process of obtaining, therefore, the relationship between the necessary flight speed and corresponding angle of attack for sustained flight is illustrated in Fig. 10. In this figure, the lift curves, as a function of the flight speed, are shown for some different angles of attack. The

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability

figure also indicates the minimum lift for sustained flight. A summary of all the results from such an analysis can be compiled as in Fig. 11, which presents the angle of attack necessary to allow sustained flight for each flight speed. From Fig. 11, it is possible to verify that the minimum lift for sustainable flight is reached with speeds of 14.5 and 11.5 m/s, for 0 and 5 deg. of angle of attack, respectively.

L(N)

However, another important parameter that should be analyzed in order to define the flight envelope is the thrust availability. In other words, as the drag force varies as a function of aerodynamic speed, the required force to balance drag must be available from the aircraft engine, using the propeller capability to transform the shaft power into traction. Historical sources, mentioned by Vilares (1956), indicate that Santos Dumont initially used a 24 hp nominal power engine. The power deficiency of this engine became evident on September 1906 during a flying

6000 5500 5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0 -500

F = 0s F = 3s F = 5s

Enough Lift

u (m/s)

Figure 10. Lift values as a function of flight speed for some flight attack angles.

attempt, when the aircraft, in spite of some jumps, was unable to take off. During the following experiments, a new and more powerful engine was selected. Its nominal power was 50 hp at 1,500 rpm. The 14-Bis aircraft performance in terms of propulsive efficiency (ηp) is unknown. However, it is important to note that the thrust produced by the engine varies with flight speed, decreasing with the speed increment. As the propeller blades do not completely convert the given engine shaft power into thrust, three isolines of different propulsive efficiencies, namely ηp = 20, 30 and 40%, are considered in the present paper. Figures 12 and 13 indicate the results of such analysis, respectively, for the 24 and 50 hp engines. In other words, the figures show the drag dependence with speed and, hence, the required thrust dependence with speed, and the three available thrust curves considering the different assumed propulsive efficiencies. The propulsive analysis for the 24 hp engine, shown in Fig. 12, indicates that flight may be viable with this engine, but only under very restrictive conditions. For instance, according to these curves, the maximum possible flight speed would be just a little over 12 m/s, if the 20% efficiency curve were used. However, as already pointed out, the drag results here obtained are probably lower than the actual drag in flight, due to the geometric simplifications adopted. Therefore, the drag curves in Figs. 12 and 13 should actually be shifted upwards, further restricting the admissible flight speed range. As also already discussed in the paper, the power deficiency of the 24 hp engine became evident in Santos Dumont’s flight attempts during September 1906, when the aircraft, in spite of some jumps, was actually unable to take off. Hence, the current calculations are completely supported by the historical accounts. Furthermore, the current analysis clearly indicates

Thrust, Drag (N)

A0A (deg.)

10 5 0 -5 8

Thrust 24hp vs. Drag

950 900 850 800 750 700 650 600 550 500 450 400 350 300 250 200

Drag Thrust (Mef=20%) Thrust (Mef=30%) Thrust (Mef=40%)

V (m/s)

u (m/s)

Figure 11. Attack angle of and flight speed necessary to allow sustained flight.

Figure 12. Drag dependence with velocity and 24 hp engine available thrust curves.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

143

Bitencourt L.O. et al 1.33 Drag Thrust (Mef=20%)

1800 1600

Thrust (Mef=30%)

0.22

1.3 CL

Thrust, Drag (N)

1200

1.29

1000

1.28

800

0.21

1.27

600

1.26

400

1.25

200 8

0.5 1 Distance of Ground (m)

0.2

Figure 14. Ground effect influence.

V (m/s)

Figure 13. Drag dependence with velocity and 50 hp engine available thrust curves.

that, with the 50 hp engine, the propulsive restrictions are overcome, as the historical accounts again report. Another aspect that should be mentioned is the fact that, during take-off, the ground causes additional drag forces. On the other hand, there are also lift increments due to ground effect. An initial analysis of ground effect is presented in Fig. 14, in which the influence of the distance to the ground in both airplane lift and drag coefficients is indicated. A 5deg. angle of attack was considered in the simulations that led to the results shown in Fig. 14, and a 11.5 m/s flight speed was assumed. One can see in this figure that, as the aircraft approaches the ground, lift increases faster than drag. This behavior can again justify the hopping-type flight observed on the September 1906 flight attempts. It is clear, however, that a more detailed analysis of all effects is necessary in order to better quantify the influence of all parameters involved. Such more detailed analysis, however, is beyond the intended scope of the present work Static stability analysis Stability is possibly the most critical part of the 14-Bis aircraft flight due to the complex canard-biplane configuration. The canard function is to generate enough lift to compensate the nose-down moment caused by the distance between the wing neutral point and the aircraft center of gravity (CG). The canard surface is placed well ahead of the center of gravity, creating an extensive destabilizing influence. Consequently, it was vital that, despite the forward motion of the aircraft neutral point due to the canard lift contribution, the aircraft CG position is still situated ahead of aircraft neutral point for longitudinal static stability. The exact CG position of the 14-Bis aircraft is unknown (Greco and Ribeiro, 2003). Therefore, conclusions 144

0.23

1.31

Thrust (Mef=40%)

1400

Cl CD

1.32

Thrust 50 hp vs. Drag

2000

concerning the aircraft stability are based on estimates of such CG position obtained from the observation of photos from tests Santos Dumont performed, in which he hanged the aircraft presumably by the CG. The stability criterion states that an airplane is stable if, when perturbed from its equilibrium condition, restorative moments bring the airplane back to the equilibrium condition. Therefore, based on the cited historical photos and according to an estimation of the mass of each airplane component, it is possible to find a range for the CG position. Such estimates indicate that it must be situated between 7.0 and 7.5 m from the aircraft nose. The test cases here considered explore the flight conditions in which the airplane has a linear aerodynamic behavior, i.e., the aerodynamic coefficients change linearly with the angle of attack and canard deflections. It must be emphasized that, for higher or lower angles of attack, unsteady CFD solutions were found in the present investigation. Another aspect that should be pointed out is the wing incidence angle, with regard to the fuselage, of approximately 5 deg. used in the 14-Bis aircraft. The authors further note that all moment coefficients are calculated using the CG as the reference point, and the CG was assumed to be at 7.25 m from the aircraft nose, which is precisely the half point in the previously identified range. Moreover, all aerodynamic derivatives with respect to the aircraft angle of attack were calculated assuming that the canard is kept with zero deflection. On the other hand, all aerodynamic derivatives with respect to the canard incidence, p, were calculated assuming a zero angle of attack of the aircraft. A summary of the aircraft most relevant aerodynamic coefficients and aerodynamic derivatives is presented in Table 4. The numerical results also indicate that the 14-Bis aircraft would have an unstable condition in pitch for CGs situated farther than 7.05 m from the aircraft nose. In other words, the neutral point is located at 7.05 m from the aircraft nose. Therefore, the results show that

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

Table 4. Aerodynamic coefficients and derivatives of the airplane and control surfaces.

Canard

0.85 4.85 rad-1 0.85 rad-1 -0.86 rad-1 -1.12 rad-1

CLδp CMδp

0.45 rad-1 1.31 rad-1

the 14-Bis airplane would most probably be a statically unstable airplane, if the current estimated range for CG positions is correct. Nevertheless, the authors emphasize that unstable airplanes can fly, despite the more difficult controllability. Moreover, it is also possible that Santos Dumont could have changed the CG position by adding weights in the frontal part of the aircraft. In any event, the current results indicate that pitch static stability and, hence, controllability of the 14-Bis aircraft was certainly an issue. Furthermore, even if the plane were stable, small variations on the CG position could make it dangerously approach an unstable flight condition. The results are also indicating that the relative values of CMρp and C Mα, and of CLρp and CLα indicate that the canard seems to be effective to perform its main function, which is the aircraft pitch control. However, since the aircraft resultant moment increases with the angle of attack, the airplane is unstable and the pilot would have to do more work to keep the airplane trimmed. The canard downwash effect over the wing was also verified and, as expected, negligible effects were detected. Hence, it seems that it is fairly safe to discard the effect of the canard over the wing for all practical purposes. The canard lift coefficient curve is shown in Fig. 15. It is important to observe that only the canard lift is plotted in this figure. It is clear from the figure that the canard contribution is not relevant in terms of the total lift, but a significant pitch moment is added, due to the canard position well ahead of the aircraft CG. Moreover, even at zero canard incidence, one can see that 0.07 0.06 0.05 CL

0.04 0.03 0.02 0.01 0

Canard deflection (deg.) Figure 15. Canard lift coefficient curve in the linear range.

7 CI Cn

Aircraft

CL0 CLα C Mα Cnβ Clβ

0 -0.01 -0.02 -0.03 -0.04 -0.05 -0.06 -0.07 -0.08 -0.09 -0.1 -0.11 -0.12 -0.13 -0.14 -0.15

Beta (deg.)

0 -0.01 -0.02 -0.03 -0.04 -0.05 -0.06 -0.07 -0.08 -0.09 -0.1 -0.11 -0.12 -0.13 -0.14 -0.15

A CFD-based analysis of the 14-Bis aircraft aerodynamics and stability

Figure 16. Cl x β and Cn X β curves for the aircraft.

some residual lift is being generated by the canard, which amounts to approximately 9.7 N of lift force. The influence of lateral flow on the airplane was also studied in the present work by varying the sideslip angles. Figure 16 shows the airplane rolling moment, Cl, and yawing moment, Cn, as a function of the sideslip angle, β. The results in Fig. 16 indicate that the linear approximation for the lateral stability derivatives seems perfectly reasonable in the range of sideslip angles considered. As can be observed in the figure, the sideslip angle induces significant and equally important roll and yaw moments, since both coefficients have the same order of magnitude. Such behavior points out to a coupling between roll and yaw motion, which seems to be an underlying characteristic of the airplane. The numerical results have also shown that the lateral flow has negligible influence on the longitudinal coefficients, for instance, CL and CD, causing a maximum relative variation of 3% in these coefficients within the tested sideslip angle range. CONCLUDING REMARKS The present work has used CFD techniques to perform an aerodynamic evaluation of the 14-Bis aircraft configuration. The historical flight conditions were simulated using a finite volume method and solving the RANS equations with the Menter SST turbulence model. A geometrically simplified model of the aircraft is used, and the results obtained seem to corroborate many of the historical reports. For instance, the results have confirmed that the 24 hp Antoinette engine would probably yield an underpowered aircraft, thus making the 14-Bis airplane unable to take off during the first flight attempt on September 1906. Therefore, the engine change made by Santos Dumont for the succeeding flights, selecting a more powerful 50 hp engine, is clearly justified. Furthermore, based on the present calculations, it is difficult to believe that 11 m/s was the true airspeed of the aircraft in the historical flights of October 23 and/or

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

145

Bitencourt L.O. et al

November 12, 1906. The present simulations have shown that the lift versus speed curve indicates very restrictive conditions for flight at such flight airspeed. An acceptable speed, assuming a 5 deg. angle of attack for the aircraft, seems to be between 12 and 14 m/s. Such speeds could be reached more easily when flying against the wind direction. In any event, it can be stated, based on the present numerical results, that the actual flight airspeed must have been higher than 12 m/s. The present calculations were also able to obtain a welldefined range of flight conditions, namely angles of attack between 5 and 10 deg., canard deflections between -5 and 5 deg. and flight speeds between 11 and 14 m/s. The results seem to indicate that the viable flight conditions were, in fact, wider than the historical values usually cited. Moreover, other important aircraft characteristics were identified, as the roll and yaw coupling when subjected to lateral flow. As in all aircraft, stability was certainly an important concern for the 14-Bis airplane. The analysis of longitudinal static stability considered the linear regime and it has shown that the present estimate for the position of the neutral point is coherent with the reality of historical reports. However, the parametric tests demonstrated that the aircraft was either aerodynamically unstable or had a very small positive static margin. Hence, even small center of gravity position variations, around the historical point, could have important impacts on the ability to fly the 14-Bis aircraft. Finally, the authors are aware that there are quite a few additional studies that could have been performed in order to better understand the aerodynamics of the 14-Bis airplane. The analyses presented here are the studies that were possible at the time. In any event, the authors feel that the major thrust intended with the work, which was to honor the centennial of Santos Dumont historical 14-Bis flight, was fully accomplished. It is hoped that the calculations and the information here reported could be useful in the future if others decide to revisit the aerodynamics of this peculiarly interesting aircraft. If nothing else is deemed useful, the authors would hope that the present effort could serve as a reminder of the unmistakably important contributions of Alberto Santos Dumont to heavier-than-air flight. ACKNOWLEDGMENTS The authors are indebted to Professor Paulo Greco, from Escola de Engenharia de São Carlos, Universidade de São Paulo, who provided the geometrical CAD model, and to Mr. Marcus Reis, from Engineering Simulation and Scientific Software, ESSS, who provided support and licenses for all used software. The authors also gratefully acknowledge the partial support provided by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), under the Integrated Project Research Grant No. 312064/2006-3. 146

REFERENCES Azevedo, J.L.F., 2006, “Aplicações Avançadas de Mecânica dos Fluidos Computacional para Aeronaves de Alto Desempenho,” Final Scientific Report, Process FAPESP No. 2000/13768-4, Instituto de Aeronáutica e Espaço, São José dos Campos, SP, Brazil. Bitencourt, L.O., Freitas, R.M., Pogorzelski, G. and Azevedo, J.L.F., 2005, “Advanced CFD Analysis of 14Bis Aircraft,” Proceedings of the 18th ABCM International Congress of Mechanical Engineering, COBEM 2005, ABCM, Ouro Preto, MG. Bitencourt, L.O., Freitas, R.M., Pogorzelski, G., and Azevedo, J.L.F., 2005, “CFD-Based Analysis of the 14Bis Aircraft Aerodynamics and Stability,” Proceedings of the 11th Brazilian Congress of Thermal Sciences and Engineering, ENCIT 2006, Paper CIT06-0249, ABCM, Curitiba, PR, Dec. 2006. CFX, 2005, www.waterloo.ansys.com/cfx/. Field, D.A., 1987, “Laplacian Smoothing and Delaunay Triangulations,» Communications of Applied Numerical Methods, Vol. 4, pp. 709-712. Freitas, R.M., Bitencourt, L.O., Pogorzelski, G., and Azevedo, J.L.F., 2006, “A CFD Analysis of the 14Bis Aircraft Aerodynamics,» Proceedings of the 25th Congress of the International Council of the Aeronautical Sciences, ICAS 2006, DGLR, Hamburg, Germany. Greco, P.C., Ribeiro, M.L., 2005, “Estudo das Características Aerodinâmicas, de Estabilidade e de Controle do 14-Bis”, Technical Report FAPESP No. 01/11158-7, Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos, SP, Brazil, 2003. ICEM-CFD, 2005, http://www.icemcfd.com/icemcfd.html. Menter, F.R., 1994, “Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications,” AIAA Journal, Vol. 32, No. 8, pp. 1598-1605. Pulliam, T.H., 2011, “High Order Accurate Finite Difference Methods: as Seen in OVERFLOW,” AIAA Paper No. 2011-3851, Proceedings of the 20th AIAA Computational Fluid Dynamics Conference, Honolulu, Hawaii. Van der Vooren, J., and Slooff, J.W., 1990, “CFD Based Drag Prediction: State of the Art Theory and Prospects,” Report TP 90247L, National Aerospace Laboratory, The Netherlands. Vilares, H. D., 1956, “Quem Deu Asas ao Homem”, Instituto Nacional do Livro, Rio de Janeiro.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 137-146, May-Aug., 2011

doi: 10.5028/jatm.2011.03021611

Isabel Lima Hidalgo

EMBRAER São José dos Campos/SP – Brazil isabel.lima@embraer.com.br

Airton Nabarrete*

Technological Institute of Aeronautics São José dos Campos/SP – Brazil nabarret@ita.br

Marcelo Santos,

EMBRAER São José dos Campos/SP – Brazil marcesantos@embraer.com.br, *author for correspondence

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers Abstract: In the aircraft industry a great practical relevance is given to the extensive use of vibration dampers between fuselage and interior panels. The proper representation of these isolators in computer models is of vital importance for the accurate evaluation of the vibration transmission paths for interior noise prediction. In general, simplified models are not able to predict the component performance at mid and high frequencies, since they do not take into account the natural frequencies of the damper. Experimental tests are carried out to evaluate the dynamic stiffness and the identification of the material properties for a damper available in the market. Different approaches for its modeling are analyzed via FEA, resulting in distinct dynamic responses as function of frequency. The dynamic behavior, when the damper natural modes are considered jointly with the high modal density of the plate that represents the fuselage, required the averaging of results in the high frequency range. At this aim, the statistical energy analysis is then used to turn the comparison between models easier by considering the averaged energy parameters. From simulations, it is possible to conclude how the damper natural modes influence the dynamic response of aircraft interior panels for high frequencies. Keywords: Vibration damper, Fuselage structures, Vibroacoustic, Dynamic stiffness, SEA.

INTRODUCTION In the aircraft manufacturing, interior panels are fastened to the fuselage structure by means of mountings designed to permit the easy disassembly in the case of maintenance. These mountings should also minimize the vibration transmission between the internal panels and fuselage and they are frequently called vibration dampers. Conceptually, these isolators are resilient elements that have been applied to structures aiming at minimizing the vibration transmission (Beranek and Vér, 1992). In aircraft design, isolators or dampers are of great practical relevance by their extensive use between fuselage and interior panels for minimizing the structure-borne vibration. Their proper representation in computer models helps the engineer on obtaining the accurate evaluation for vibration transmission paths between fuselage and interior panel, and consequently, the interior noise prediction. Commonly, vibration dampers are based on rubber materials, due to some mechanical properties, such as high damping, low stiffness, and high bearing capacity (Downey et al., 2001). As the dynamic behavior of the rubber changes with load and environment conditions, the characterization of employed materials plays an important Received: 03/05/11 Accepted: 14/06/11

role in the performance prediction of dampers. The temperature is an important parameter to be considered during the damper characterization, since the rubber material at low temperatures tends to be stiffer and to have higher damping. In opposite sense, when tested at high temperatures, it tends to have low stiffness and damping. In addition, the rubber properties are also dependent to frequency and strain (Jones, 2001). Some instances of rubber materials applied for vibration control are the natural rubber, neoprene, and silicone. Different calculations are used to evaluate the damper behavior. A classical method represents the set damperstructure as a mass-spring-damper system and it is useful at low frequency predictions (Nashif et al., 1985). Nevertheless, this representation does not include some real system features, such as material nonlinearity and the component dynamic behavior. These features, associated with the fact that an isolated structure does not behave as a rigid body, influence the damper attenuation at mid and high frequencies (Snowdon, 1979; Beranek and Vér, 1992; Weisbeck, 2006). The dynamic behavior of dampers is considered in some published calculation methods. One of them, the fourpole method, described by Molloy (1957), incorporates dynamic characteristics of isolators and structure as frequency dependent through transfer matrices. As well

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

147

Hidalgo, I.L., Nabarrete, A., Santos, M.

reported by Snowdon (1979), Weisbeck (2006) and Weisbeck et al. (2009) the parameters used in this method are usually obtained experimentally. An alternative prediction method can assess the damper attenuation considering its dynamic behavior through a detailed finite element model (Jones, 2001). This is a robust and wellestablished method to solve engineering problems related to solid structures (Bathe, 1996). In this research, solid finite elements are used to model one damper which is an assembly of metallic and rubber parts. Parameters regarding to geometry and material properties applied to the damper are detailed and investigated. A very refined mesh is employed to model one damper in order to achieve accurate modal frequencies and modes. The results from finite element analysis (FEA) are compared to the ones obtained from an experimental modal test and the finite element model is updated by finding the material properties that adjust it in accordance with experimental results. The dynamic analysis of the set fuselage-isolator-panel is the motivation for this work. The fuselage-isolator-panel structural system is modelled by the component mode synthesis (CMS), in order to test different configurations for the vibration dampers. For damped structures the analytical approach to CMS (Craig Jr., 1987) is used to solve for the component undamped modes and frequencies, to assemble those modes into the coupled model, and to supplement the resulting system with assumed modal damping. CMS is usually considered for the analysis of structures that are built-up of several components (Ewins, 2000). The structure-borne transmissibility to interior panels and the consequent noise radiation are considered as significant until 10 kHz. However, when dealing with the high-frequency range, FEA brings concerns such as the large model size and dynamic properties with some uncertainty. As alternative to these issues, the traditional method of statistical energy analysis (SEA) (Lyon and DeJong, 1995), is considered. SEA represents a field of study in which statistical descriptions of a system are a)

employed in order to simplify the analysis of complicated structural-acoustic problems, especially in the highfrequency range. Searching for a solution to the entire frequency range, some authors have also considered the possibility of enriching a SEA model with information from FEA or measurements. One has proposed a coupled solution scheme considering the “stiff” components modelled with finite elements and the “soft or flexible” ones modelled with SEA (Shorter and Langley, 2005). Using engineering experience or by performing a large number of subcomponent calculations one can investigate whether a specific subsystem should be considered as ‘‘stiff’’ or ‘‘soft’’. In the present work, the finite element model of the damper is considered as a substructure in the FEA approach and also as a subcomponent model to the SEA. A comparison of results for both analyses is commented for the entire frequency range. FINITE ELEMENT MODELING The finite element model used in this research aims at discussing the influence of the high frequency vibration modes from damper and panels in the structure-borne transmissibility. A finite element mesh very refined with solid elements is proposed for the isolator model, in order to achieve accurate modal frequencies and modes. Other models considering the vibration damper as a rigid connector or a spring is considered for comparison by the advantage of the simplified modeling. Figure 1(a) shows a rendered image of a damper available in the market with a refined solid mesh that contains at total 9,626 finite elements (HEXA and PENTA) and 11,115 nodes. In Fig. 1(b), the same mesh appears with different colors for the material properties. The colors red and green represent the external structure made in aluminum; yellow and blue represent the component parts made in rubber and steel, respectively. The vibration damper selected to this work has the material properties described in Table 1. Also, as per the supplier catalogue (Lord Corporation, 2010), the modelled isolator has equivalent static stiffness equal to 88,000 N/m. b)

Y Y

X Z

Finite element mesh

Damper materials represented: aluminum in red and green, rubber in yellow and steel in blue

Figure 1. Isolator finite element model 148

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers

Table 1. Damper, fuselage and internal panel material properties

Material

Density (kg/m³)

Poisson coefficient

Loss factor (%)

Young modulus [Pa]

Aluminum

2,700

0.33

7.1 x 1010

Steel 304 SS

8,300*

0.28

2.0 x 1011

Silicone

1,200

0.40

2.5 x 106

Aluminum

2,700

0.33

7.1 x 1010

Part Damper external structure Damper central fixture Damper rubber Fuselage and internal plates *

density to updated isolator mass.

There are several possibilities to model the aircraft fuselage, but for this work just a simple uniform panel, or a rectangular aluminum plate, is used to represent the equivalent stiffness and mass distribution of a unit cell of fuselage. The same configuration is used to the internal panel model. Fuselage and internal panels represented by uniform panels are modelled with plate finite elements. The detail for the connection between meshes from the damper and panel is depicted in Fig. 2. The nodes of panel considered for the connection with the solid finite element model of the damper are the same considered for other simulations, connecting the panels with the rigid or spring elements.

Figure 2. FEM model detail: isolator-plate interface

The panels are modelled, each of them as a substructure, with QUAD4 elements. The modal analysis of each one is performed by Nastran solver (MSC Software, 2008). The isolator mesh with solid finite elements is defined in order to guarantee the minimum number to correctly represent the vibration modes. As a minimum amount recommended by Fahy and Gardonio (2007), six elements per wavelength are applied to the isolator mesh. The calculation of the wavelength λ is based on the propagation speed cl of the quasi-longitudinal wave on solid, and evaluated as presented in Eq. 1. 6d = h =

cf = flim

E/l flim

(1)

In this expression, d is the finite element dimension, E is the Young modulus, ρ is the density, and flim is the mesh frequency limit. The lower wave speed leads to the lower mesh frequency limit, and it occurs for the rubber material. Hence, taking into account the rubber properties (Table 1) and an element dimension of 0.7 mm, the frequency limit for the current mesh is approximately 10 kHz. Component mode synthesis In CMS, the substructure or component models are transformed from physical to modal coordinates, using a set of normal modes after solving the component eigenvalue problem. The component models are assembled together to form the global dynamic problem for the entire structure. The equation of motion of a component r, neglecting damping, is described by Eq. 2, where the motion is represented by a vector of physical displacements. In this expression, written in a partitioned form, the indices j and i relate the vectors and matrices to boundary and internal nodes, respectively. M and K are the partitioned mass and stiffness matrices, while f represents the partitioned force vectors. M ³ ii ³M ji 

M ij ¨« u i ¬« K ii K ij µ© +³ M jj µ « u j « ³ K ji K jj ® ª

µ µ

¨ u ¬ ¨ f ¬ « i « « i « =© © «ª u j «® «ª f j «® (2)

In CMS with fixed interface, the response of a system is represented in terms of a set of ‘component’ modes and ‘constraint’ modes. The component modes are taken as a subset of the local modes when the boundary degrees of freedom are clamped. The constraint modes are given by the static response of the substructure when a unit displacement or rotation is applied to a given boundary degree of freedom while all other boundary degrees of freedom remain fixed (Craig Jr., 1981). The normal modes with fixed-interface represent the component modes in this work. The size of this eigenvalue problem equals the number of internal degrees of freedom. Each component model is transformed from physical to modal coordinates, using a set of normal modes.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

149

Hidalgo, I.L., Nabarrete, A., Santos, M.

The reduction in the model size is achieved by truncating the number of component modes included in the analysis (Craig and Bampton, 1968). In CMS with fixed-interface, the modal matrix is formed by the combination of the kept number of normal modes with fixed-interface and the static constraint modes for the component. The constraint modes assure the compatibility of the component displacements at the interfaces, improve convergence and also yield the exact static solution. Equation 3 shows the relation between the physical coordinates and modal coordinates for the component model. In this equation, Φik and qk represent, respectively, the kept normal modes and the modal coordinates with fixedinterface. The constraint coordinates are represented by qc and Ijj is an identity matrix. r

¨ u ¬ \ « i « ³ ik = © «ª u j «® ³³ 0

-K -1ii K ij ¨« q k µ © µ q I jj µ «ª c

¬ « «®

(3)

The equation of motion based in modal coordinates of the component r is then presented in Eq. 4. The matrices mcc and kcc are the constraint modal mass and stiffness respectively, mkc is the coupling matrix and Λkk is a diagonal matrix of kept modal eigenvalues. I ii

³ mT kc

 r

0 ¨« q k ¬« k cc

µ © =³ µ « qc « ³ 0 ª ® ³

m kc ¨« q k © m cc µ « q c ª

-K ii K ij -1

I jj

¬ R « ³ kk + «® ³ 0

¨ f µ «© i µ f µ «ª j

¬ «

«®

0 ¨« q k µ © k cc µ « q c ª

¬ F « ik =³ ³ 0 «® ³

2 £ F¥ k= ² m - ´ 2/ f x¦ ¤ 

(

)

(5)

The dynamic stiffness k from the tested vibration damper is calculated by the Eq. 5. In that, f is the frequency and m is the mass from the moving parts of damper and the attached set, such as bolt and other parts. By the dimensions of the damper, the rubber material is considered as massless. The experimental result from this test is presented on Fig. 3. The first peak is associated to test set. The test data is considered valid where the dynamic stiffness is constant. Above this value the response becomes very large since it is controlled by inertia, explaining the decrease in stiffness. From this test, the real part value of dynamic stiffness is 1.50x105 N/m, and the imaginary part value is 2.89x104 N/m. Calculated as the quotient of the imaginary part by the real part, the loss factor is approximately 0.2. Afterwards, frequency response analyses are employed as the identification process for the isolator FE model, in order to fit the experimental data.

Frequency-constant properties are used for the damper rubber -1 -K ii K ij ¨« f i ¬« material. µ © Although the rubber properties are dependent on µ I jj frequency, « f j «® for typical applications of the damper studied in µ ª r

the current work, it is possible to consider a constant Young Modulus value since the damper elastomeric material is designed to work in the viscoelastic rubbery region.

vibration damper. It is axially excited by the shaker at one end while the other end is completely fixed, approximating to the one-degree-of-freedom experiment. The ratio between acceleration and applied force is taken as test result.

(4)

Considering the synthesis of two or more components and the continuity of the modal displacements at their common interface, a transformation matrix is written to impose the coupling conditions among them (Craig Jr., 1987). Due to the simplicity of the transformation matrix, the component synthesis is straightforward and the system matrices have the same structure as the component matrices.

In addition, dynamic analyses consider the unitary axial concentrated force applied on the damper central fixture. Concentrated mass is not considered, which differs from the experiment, and generates the later decreasing in real dynamic stiffness. Also, no perturbation at low frequency is observed in the numerical analysis. Finally, the updated Young Modulus of 2.5 MPa and loss factor of 20%, resulted in the best dynamic stiffness curve adjustment, as can be verified in Fig. 3. DAMPER DYNAMIC ANALYSIS

Rubber material properties identification The rubber material is of great importance for the analysis of the isolator model, and in the present research, the rubber property identification is based on dynamic stiffness measurements and later in the numerical adjustment. The test procedure developed by Clark and Hain (1996) for measuring the axial dynamic stiffness is applied to the selected 150

With the rubber properties updated in the FE model, the modal analysis of the damper in a free-free condition is performed, in order to assess its dynamic behavior. The first six vibration modes are omitted from the results discussion as they represent the rigid body modes. With the following six modes is noted the spring-mass behavior as depicted in Fig. 4. The central fixture pin works as a concentrated mass and the rubber material as a spring element. The behavior of a six degree-of-freedom spring

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers

4.0E + 05

Test assembly resonances

Real_Experimental Imaginary_Experimental Real_Numerical

Valid region for test data

{

Dynamic Stiffness [N/m]

3.0E + 05 2.0E + 05

Imaginary_Numerical

1.0E + 05 0.0E + 05 -1.0E + 05 -2.0E + 05

100

10000

1000 Frequency [Hz]

Figure 3. Comparison between numerical (FEA) and experimental dynamic stiffness

Mode 1. Radial mode 512.8 Hz

Mode 2. Radial mode 514.7 Hz

Y Z

Mode 3. Axial mode 738.0 Hz

Y X

Mode 4. Torsion mode 755.3 Hz

Mode 5. Rocking mode 1138.1 Hz

Mode 6. Rocking mode 1148.9 Hz

Figure 4. Spring-Mass Modes J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

151

Hidalgo, I.L., Nabarrete, A., Santos, M.

is seen in Fig. 4 and, how told by Gardner et al. (2005), these modes are described as spring modes. Also, it is possible to observe the symmetrical radial modes as a consequence of the geometry in this specific damper. The modes that follow the spring modes are called rubber modes, in which this part of the damper responds for higher frequencies. For these vibration modes, the external metallic structure and the central fixture do not influence the natural damper motion. In the current case, these modes occur at very high frequency, from 4000 Hz and on, where each one is very close to the other in frequency. Table 2 describes the rubber modes until 6500 Hz. Figure 5 shows some of these rubber mode shapes.

FUSELAGE-ISOLATOR-PANEL ANALYSES The complexity of modeling the internal panels connected to the fuselage of an aircraft is evaluated by considering, as first model, two rectangular plates centrally connected by a vibration damper. This model recalls the concept of fuselage-isolator-panel, where one damper supports the internal panel attached to the fuselage. Then, two identical aluminum plates with thickness equal to 2 mm and dimensions equal to 0.5 m and 0.7 m are modelled with QUAD4 finite elements. The material properties selected to this model are described on Table 1. The plates are separated by the damper height or approximately 8.25 mm. The rotation degree of freedom around each plate edge is restricted aiming at simulating the fuselage continuity.

Table 2. Rubber material modes of the damper

Mode 7 8 9 10 11 12 13 14 15 16 17 18

Nat Freq (Hz) 4,086.4 4,492.5 4,505.5 4,550.0 4,586.8 4,587.6 4,662.2 4,662.4 4,717.3 4,717.3 4,746.0 4,796.0

Mode 19 20 21 22 23 24 25 26 27 28 29 30

Nat Freq (Hz) 4,804.9 4,804.9 4,813.8 4,814.8 4,931.5 4,931.5 5,053.3 5,053.5 5,060.7 5,094.5 5,094.5 5,288.1

Mode 31 32 33 34 35 36 37 38 39 40 41 42

Mode 43 44 45 46 47 48 49 50 51 52 53 54

Nat Freq (Hz) 6,091.9 6,096.2 6,104.9 6,104.9 6,240.9 6,240.9 6,275.2 6,275.4 6,403.9 6,403.9 6,496.7 6,496.7

Y X

Nat Freq (Hz) 5,288.1 5,396.5 5,396.6 5,505.7 5,505.7 5,740.8 5,740.8 5,761.2 5,761.2 5,987.6 5,987.6 6,028.2

Mode 8

Mode 12

Y X

Mode 19

Mode 52

Figure 5. Instances of rubber material mode shapes 152

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers

Modal frequency responses using the substructure modal solutions are obtained for the entire structure considering different approaches for the isolators. Justified by the necessary simplification, distinct connection elements can be modelled between both plates, when dealing with the complete aircraft structure. Aiming at the response comparison, in this research, five approaches for isolators are modelled to connect both plates at the center, as follows: 1.

rigid connection: The damper infinite stiffness is used as reference to this approach;

linear axial spring connection: The damper static stiffness value of 88,000 N/m (Lord Coorporation, 2010) is considered as the constant dynamic stiffness;

linear axial spring connection: The dynamic stiffness value considered here is equal to 148,000 N/m, which is obtained from the test described in section 2.2, for the axial direction of the damper;

six degrees-of-freedom spring connection: The dynamic stiffness value per degree of freedom is considered as presented on Table 3. These values are obtained from other experiments performed in similar manner as described in section 2.2;

Table 3. Dynamic stiffness and loss factor

Dynamic stiffness (N/m)

Loss factor Îˇ

1.0E+05

0.18

1.0E+05

0.18

1.5E+05

0.20

2.8E+01

0.20

2.8E+01

0.20

2.7E+01

0.14

Direction

updated FE solid model for the damper, as described in the previous sections. The damper connection to the upper and bottom plates is performed through rigid elements. Figure 2 shows the rendered image for this connection.

In these analyses, the excitation force ranges in frequency from 1 to 8,000 Hz. Initially, only one concentrated harmonic force is applied perpendicular to the bottom panel. Subsequently it is proposed to create an average spectrum for each response. This is performed by applying six different excitations arbitrarily distributed and applied not simultaneously, as sketched in Fig. 6. Hence, the dynamic response of the upper panel is obtained by averaging velocity in space considering chosen nodes, and afterward by averaging these results for the six different excitations. This data is calculated for each isolator approach. The velocity magnitude is considered as the response parameter, since it characterizes the vibration level. FUSELAGE-ISOLATOR-PANEL SIMULATION RESULTS After identifying the damper properties through experimental tests, the update is performed to the FE solid model. Then, the simulation result with this model is considered as the most accurate to represent the real damper behavior. Figure 7 brings the low frequency results (1 to 100 Hz) in narrow band when only one excitation is applied. In this range, the behavior of the six-dof spring approach is the closer to the updated FE solid model. Significant differences among models can be noted in the antiresonance behavior, when analyzing both axial spring models around 60 Hz, while other models show a resonance at 60 Hz. This fact indicates that this mode is related to the coupling with other directions, different from the axial one.

Response nodes on the upper plate Excitation Force on the bottom plate Concentrate force at coordenates (0.126, 0.124, 0.0) y

Connection between plates

z x

x F

Figure 6. FEM model sketch: response nodes positions and excitation points J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

153

Hidalgo, I.L., Nabarrete, A., Santos, M.

Rigid Axial spring k = 88e3N/m Axial spring k = 148e3N/m 6 DOF spring Updated FE solid model

1.0E + 00

|Velocity| (m/s)

1.0E - 01 1.0E - 02 1.0E - 03 1.0E - 04 1.0E - 05 1.0E - 06 10

Frequency [Hz]

100

Figure 7.Upper plate velocity average in low frequency range (10 to 100 Hz)

Due to the high modal density of the plate at high frequency, the influence of rubber modes of the damper on the plate response may not be properly assessed. This fact justifies the need of an average spectrum for each response. The average spectrum is calculated for each isolator approach and presented in Fig. 8, considering frequency band of one-third octave from 100 to 6,300. Based on Fig. 8, it is possible to notice a similar behavior for all models until 800 Hz, except for a decrease for the updated FE isolator model in 500 Hz. The rigid connection and the six-dof spring model follow similar trend along the entire frequency range, showing high vibration levels at high frequency when compared to the other models. Conversely the updated FE model shows that vibration level decreases with frequency from 1,000 Hz and on, such as both axial spring models. A difference however appears in the 4,000 Hz frequency band, where a vibration level increase is verified. Facing the different approaches for isolator model, it is visible they result in distinct responses depending on the frequency of concern. Therefore, the simplicity of modeling the damper as axial springs when compared to the updated FE model is a worthy discussion. This is true since they show responses comparable to the 154

FE model for a proper frequency range. Although, in the high frequency range, the axial spring with the measured dynamic stiffness shows the same trend and values comparable to the updated FE model response, the difference related to the damper rubber modes can overestimate the damper attenuation. Furthermore, an additional drawback to evaluate these high frequency results is related to the feature of FEA. The responses are deterministic and taken at a specific point, which insert influence of plate local modes. In addition, depending on the force position, it may not be enough to excite the mode along the frequency range of interest. Although the results are averaged and more than one excitation is applied, in order to minimize this effect, this influence can affect the results interpretation. An alternative to overcome this issue and make easier the comparison between models considers the energy parameters, such as spectral density applied in the random analysis, or the averaged energy parameters in the statistical energy analysis (SEA). Moreover, the current work is interested in vibration transmission from fuselage to interior panels and the consequent noise radiation, which can be significant until 10 kHz. Thus, the high frequency range is of great relevance and concern.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers

Rigid

1.0E + 00

Axial spring k = 88e3N/m

|Velocity| (m/s)

Axial spring k = 148e3N/m 6 DOF spring

1.0E - 01

Updated FE solid model

1.0E - 02

1.0E - 03

1.0E - 04 100

1000 Frequency [Hz]

Figure 8. Upper plate velocity magnitude (1/3 octave frequency band)

COMPARISON BETWEEN FEA AND SEA As presented in the previous section, simplified models for dampers are not feasible for predicting the high frequency vibration transmission. In this frequency range, although FEA can be employed, its results bring a complex task in interpreting the influence of isolator modes, due to the high modal density of the plate, requiring the need for averaging the results. At the same time, the effort required to define an FE model increases with size and with the geometric and material complexity of the model. Thus in order to obtain the response of complex structures, coupled through dampers, the use of FEA can be cost prohibited due to the high level of discretization required for the high frequency range. This issue for large models can be lessened by applying techniques such as component modal synthesis as described previously. However, in addition, there is uncertainty related to the dynamic properties of complex structures, regarding, for instance, the damping distribution, joints and connections between components, or the material properties. Consequently, it is granted that natural modes at high-frequencies based on a deterministic model with nominal material properties and dimensions may vary with real values and, as result, some statistical evaluation is required. FEA using Monte Carlo numerical simulation performs such analysis by repeating calculations for randomly generated sets of system

properties, which is, for high frequencies, a simulation with much time consuming. Also, excitation forces are not usually precisely known (Langley and Fahy, 2004). As an alternative for vibroacoustic problems in the high frequency range, SEA is largely used, including in the aerospace industry (Lyon and DeJong, 1995). SEA allows the calculation of the flow and storage of dynamic energy in a system. In SEA, a system is divided into subsystems, which are groups of similar energy storage modes. Each mode type in a SEA subsystem acts as a separate store of vibroacoustic energy and is therefore represented by a separate degree of freedom in the SEA equations. For each subsystem, as showed by Eq. 6, the energy dissipated internally (Pi,diss) by damping is proportional to the subsystem vibrational energy Ei. The proportionality rate is given by the internal loss factor ηii, which can result from structural damping (material property), acoustic radiation, subsystem interface, or friction mechanisms. Pi,diss= Mii \ E i

(6)

The power flow between two different subsystems is supposed to be proportional to the difference in modal energy, as demonstrated in Eq. 7. In this equation, Pij is the power flow, E and n are respectively, the total energy and the modal density of each subsystem, and ηij is the

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

155

Hidalgo, I.L., Nabarrete, A., Santos, M.

coupling loss factor (CLF) between system i and j. CLF depends on the characteristics of the junction between the subsystems and damping (Lyon and DeJong, 1995). Pij = t (dij E i - d ji E j ), n idij = n jd ji

(7)

Equation 8 considers Pi as the power injected into the system, by the conservation of energy for n subsystems. There, the power losses, resulted from the dissipation (Pi,diss ) and the coupling (Pi→j ) of the subsystem, is diminished from the power gains (Pj→i ), coming from the coupled subsystems. n

j i

Pi = Pi,diss + - PiA j - u - PjA j

(8)

Equations 7 and 8, in a matrix form, is written in Eq. 9. In this equation, [η0] represents the total loss factor matrix of the system defined by the internal and coupling loss factors, which are represented individually in Eq. 10. Finally, in Eq. 11, the power input Pi , due to a point structural source, considers force and velocity at the excitation point (Lyon and DeJong, 1995). t u[d 0 ]{E}={P}

(9)

dij0 = -d ji , dii0 = - dim m =1

1 Pi = Re F u v * 2

(10)

(11)

In SEA, the local modes of a subsystem are described statistically and the average response of the subsystem is predicted with respect to frequency and space (Lyon and DeJong, 1995). Since average parameters are considered, it is not necessary to have a detailed model. On the contrary, it is only required the overall length, width or volume of a subsystem along with approximate estimate of properties that govern the wave propagation within the subsystem. The response energy, from the model solution, is usually related to a particular quantity of interest such as acceleration, velocity, or sound pressure level. The number of modes within a subsystem represents the capacity of the storage energy, hence the modal density characterizes the energy storage, and then, it is a restrictive parameter for SEA. The modal density of the isolator is low along the frequency range of interest, when compared to a typical subsystem in SEA, such as the rectangular plate. This fact restrains the capability of equivalent energy storage for proper calculation of the subsystem 156

parameters. The connection between subsystems is considered where impedance discontinuities exist. The measure of the energy rate flowing out of a subsystem through the coupling to another subsystem defines the CLF (Lyon and DeJong, 1995). Usually, within SEA models, the damper is represented by a CLF, since so far, there is not a specific formulation for an isolator subsystem in SEA. The comparison between FEA and SEA is achieved by modeling in SEA the same two simple panels connected by an isolator at the center point, as depicted in Fig. 6. SEA is performed using the software VaOne (ESI Group, 2009). In this software, two distinct connections are possible to choose for this comparison, the first one with a rigid link, and the second one with a six-dof spring. For the last, it is considered the same dynamic stiffness values per direction given by Table 3. The power input in SEA model is calculated by Eq. 11, with the same driving point velocity adopted by FEA. In SEA, the power input is based on the average response resulting from six concentrated and harmonic forces, randomly distributed and applied not simultaneously. As required for comparison with SEA, the finite element response is obtained from the spatial average of velocity (using nine nodes), which is subsequently averaged for six different excitations. The comparison between FEA and SEA is depicted in Fig. 9. It shows similar behavior for all data up to 630 Hz. In the high frequency range, updated FE solid model has the same trend of the SEA model with the spring connection. However, the updated FE model presents that vibration level increase in the 4000 Hz frequency band, what coincides with the band where the rubber modes of the damper are concentrated. Hence, in order to not overestimate the damper performance in the high frequency, it is recommended to include the damper dynamic behavior into the SEA model. An alternative to accomplish the inclusion of the damper dynamic behavior is got by employing a hybrid FEA-SEA (Gardner et al., 2005). This method, which encompasses the advantages of FEA and SEA, is described by Langley and Bremner (1999) and allows including details of a system or coupling through a finite element component into the SEA model. For the present case, the damper is modelled by FEA, allowing the prediction of its modes, and joined to the calculation of the panel response by SEA, resulting in a hybrid coupling loss factor. Based on the fact that the structure borne path, and consequently the vibration damper, is of great importance at high frequency, it is essential to model properly the damper at this frequency range. This includes accounting for the damper internal resonances, which tend to reduce the vibration attenuation.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

Structure-borne transmissibility evaluation through modeling and analysis of aircraft vibration dampers

1.0E + 00

Updated FE solid model SEA_rigid link SEA_6 DOF spring

|Velocity| (m/s)

1.0E - 01

1.0E - 02

1.0E - 03

1.0E - 04 100

1000

6300

Frequency [Hz] Figure 9. Upper plate velocity magnitude from SEA model and update solid FE model

CONCLUSIONS Within this paper the FE model of a typical vibration damper applied to the aircraft fuselage is developed. Based on this model, it is possible to identify the damper dynamic behavior: spring modes, and the rubber internal modes. Different modeling forms of damper are applied between two plates. The result comparison demonstrates that simplified models, such as axial spring, can be easily and well employed depending on the frequency range of interest. Although, in low frequency, some agreement is verified, in high frequency, the simplified model can not reach a satisfactory result when compared to the updated FE solid model. On the other hand, in order to model the damper through finite elements with fidelity, the rubber material properties must be reliable and be available, what is frequently a challenger. The FE model development is time consuming, mainly due to the refined mesh, as required for the high frequency limit. In order to highlight the differences between models, mainly in high frequency, where the plate modal density is high, FE results are averaged. They show a vibration increase due to the damper rubber modes. In addition, applying simplified models, SEA is performed for comparison.

The vibration transmission through the dampers is of great importance in aircraft applications such as noise radiation from interior panels. For high frequency, it requires a proper modeling of the component, taking into account its internal vibration modes. REFERENCES Bathe, K. J., 1996, Finite element procedures, PrenticeHall, New Jersey, USA. Beranek, L. L., Vér, I. L., 1992, Noise and vibration control engineering - principles and application, Jonh Wiley & Sons Inc., New Jersey, USA. Clark, M. D., Hain, H. L., 1996, A systematic approach used to design floor panel isolation for commercial aircraft, Proceedings of Noise-Con 96, Seattle, USA, pp. 455-460. Craig Jr., R. R., 1981, Structural dynamics, an introduction to computer methods, Wiley, New York. Craig, R. R., Bampton Jr., M., 1968, Coupling of substructures for dynamic analysis, AIAA Journal 6, pp. 1313-1319. Craig Jr., R.R., 1987, A review of time-domain and frequency-domain component-mode synthesis methods,

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

157

Hidalgo, I.L., Nabarrete, A., Santos, M.

Journal of Analytical and Experimental Modal Analysis, Vol. 2, pp. 59-72. Downey, P., Bachman, G., Diffendall, C., 2001, Microcellular resilience for optimised insertion loss using an rubber insulating material, Proceedings of Noise-Con 2001, Portland, USA. ESI Group, 2009, VaOne 2009 Guide. Ewins, D. J., 2000, Modal testing: theory, practice and application, 2nd edition, John Wiley & Sons Inc., New York.

Lyon, R. H., DeJong, R. G., 1995, Theory and application of statistical energy analysis, Butterworth-Heinemann, Boston, EUA. Lord Corporation, <http://www.lord.com>, in June 6, 2010. Molloy, C. T., 1957, Use of four-pole parameters in vibration calculations, Journal of the Acoustical Society of America, Vol. 29, pp. 842-853. MSC Software, 2008, MSC Nastran 2008 Quick Reference Guide.

Fahy, F., Gardonio, P., 2007, Sound and structural vibration, 2nd edition, Elsevier, Oxford, UK.

Nashif, A. D., Jones, D. I., Henderson, J. P., 1985, Vibration damping, John Wiley & Sons Inc., New York, USA.

Gardner, B. K., Shorter, P. J., Cotoni, V., 2005, Modeling vibration isolators at mid and high frequency using Hybrid FE-SEA Analysis, Proceedings of InterNoise 2005, Rio de Janeiro, Brazil.

Shorter, P. J., Langley, R. S., 2005, Vibro-acoustic analysis of complex systems, Journal of Sound and Vibration, Vol. 288, pp. 669-699.

Langley, R. S., Bremner, P. G, 1999, A hybrid method for the vibration analysis of complex structural-acoustic systems, Journal of the Acoustical Society of America, 105, pp.1657-1671. Langley, R. S., Fahy F. J., 2004, Noise and Vibration, 1st edition, Chapter 11 - Advanced Applications in Acoustics, Spon Press, London UK. Jones, D. I. G., 2001, Handbook of viscoelastic vibration damping, Jonh Wiley & Sons Inc., West Sussex, England.

158

Snowdon, J. C., 1979, Vibration isolation: use and characterization, Journal of the Acoustical Society of America, Vol. 66, pp. 1245-1274. Weisbeck, J. N., 2006, Effect of stiffness, damping, and design on side panel isolator noise attenuation characteristics, Proceedings of InterNoise 2006, Honolulu, USA. Weisbeck, J. N., Sanetick, R. M., Zmijevski, T. R. L., 2009, Structure borne noise control of oscillating pumps, Proceedings of InterNoise 2009, Ottawa, Canada.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 147-158, May-Aug., 2011

doi: 10.5028/jatm.2011.03021111

Emerson Andrade Santos

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil andradesos@yahoo.com.br

Wilton Fernandes Alves

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil wfernandes@iae.cta.br

André Neves Almeida Prado

Instituto de Aeronáutica e Espaço São José dos Campos/SP – Brazil andrenap@iae.cta.br

Cristiane Aparecida Martins*

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil cmartins@ita.br *author for correspondence

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine Abstract: The main objective of this work was to present the specification of an experimental firing test stand for liquid rocket engines (LRE) and develop a program for control and acquisition of data. It provides conditions to test rocket engines with thrust from 50 to 100 kgf. A methodology for laboratory work implementation using information technology, which will allow the automatic and remote functioning of the test stand, permits users to input the necessary data to conduct tests safely, achieve accurate measurements and obtain reliable results. The control of propellant mass flow rates by pressure regulators and other system valves, as well as the test stand data acquisition, are carried out automatically through LabVIEW commercial software. The test stand program is a readable, scalable and maintainable code. The test stand design and its development represent the state of art of experimental apparatus in LRE testing. Keywords: Experimental firing test, Liquid rocket engine, Data acquisition.\

LIST OF SYMBOLS

INTRODUCTION

Aa: outlet area of nozzle, m

Acr: critical section area, m

C*: characteristic velocity, m/s F: thrust, kgf FH: thrust in atmospheric condition, kgf

GN2: gaseous nitrogen GOX: gaseous oxygen Isp: specific impulse, s

k: adiabatic exponent km: mixture ratio

: mass flow rate, kg/s m m  F : fuel mass flow rate, kg/s m  O : oxidizer mass flow rate, kg/s Pa: outlet static pressure, bar

Pch: pressure in combustion chamber, bar

PH: ambient pressure, bar

R: gas constant, J/kg.K

Tch: temperature in combustion chamber, K

W0: velocity in the combustion chamber, m/s Wa: outlet velocity of nozzle, m/s

Received: 25/03/11 Accepted: 20/05/11

The Brazilian Space Agency (AEB) through National Plan of Space Activities (PNAE) (2005) has invested in the formation of specialists in technology of calculation, design and construction of liquid rocket engines. In the next scheduled version of the Brazilian Vehicle Launcher of Satellite (VLS), the liquid rocket engine (LRE) will only be utilized in the upper stages, but in future versions, it will be employed in other stages of the vehicle. The advantages of LRE in relation to the solid propellant rocket motor justify the investment in this area. Among the advantages of LRE are its long operating time, thrust control and high achievable specific impulse. LREs are subjected to tests in installations known as test stands before they are put into operation. Test stands are oriented in determining the specific operating parameters and achievable performances of LREs. On September 17th, 2005, the Institute of Aeronautics and Space (IAE) performed the first firing test of a liquid rocket engine of 5 kN of thrust (named L5 engine). The tests were carried out in a test stand located at the Liquid Propulsion Laboratory (LPL) in São José dos Campos, which has capacity to test engines up to 20 kN. The L5 engine was designed to operate with liquid oxygen (LOX) and kerosene but in the preliminary phase of the tests it used ethyl alcohol as fuel. The injector head

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

159

Santos, E.A. et al

consists of bipropellant centrifugal liquid injectors of holes in the periphery of the fire bottom responsible for the film cooling formation. Currently, the IAE has been working in two new projects in the liquid propulsion field. The L15 engine is a bipropellant LRE of 15 kN of thrust that operates with liquid oxygen and ethyl alcohol, which will be used in the VS-15 sounding rocket. The other project under development is the L75, a bipropellant LRE of 75 kN of thrust, which will operate with a turbopump feed system using the propellants: liquid oxygen and kerosene. This work has resulted in two master’s dissertations in the Aerospace Engineering course of liquid propulsion area of Aeronautics Institute of Technology (ITA). The first one developed test stand specifications, design of a LRE, and a test methodology (Alves, 2008). The second one developed a program for a test stand data acquisition and control system (Andrade, 2008). The IAE decided to build this test stand, which is in the final phase of assembly. Once completed, it will be used as an educational tool in the formation of new groups of the master’s degree course of ITA, in order to train IAE technical personnel, to evolve LRE research and to acquire liquid propulsion knowledge for application in satellite launch vehicles. The apparatus under development will make available for laboratories to verify the influence of mass flow rate (object of this work) and nozzle expansion ratio in the thrust force of the LRE. With small changes in the thrust chamber, it will be possible to carry out additional tests to study the heat flow through the engine. Using an automated data acquisition and control system, the user will be able to remotely operate the engine and do measurements of several parameters, like pressure, temperature, mass flow rates and thrust. THEORETICAL BASIS OF LIQUID ROCKET ENGINES The LRE consists basically of one thrust chamber, tanks to store the propellants, a feed system to force the propellants into the thrust chamber, a power source to supply the energy for the feed system, piping to transfer the liquids, a structure to transmit the thrust, and control devices to initiate and regulate the propellant flow and control thrust (Sutton, 2001). The thrust chamber is the main part of a rocket engine (Sutton, 2001). It is usually formed by an injector head, a combustion chamber, a nozzle, a cooling jacket, and an ignition system according to Fig. 1. 160

Injector Head Ignition System

Nozzle

Cooling Jacket

Combustion Chamber

Figure 1. Thrust chamber.

The injector head consists of injectors distributed along the surface of a plate placed at the inlet of the combustion chamber and a set of ducts that guarantee a uniform distribution of propellants. The injectors are the key elements of the thrust chamber because they determine the behavior of propellants in the combustion chamber. It injects and atomizes the propellants into the combustion chamber, mixing them homogeneously and in well defined ratios of fuel and oxidizer, before it is vaporized and quickly ignited. The injectors are classified as centrifugal or jet. They can be monopropellant or bipropellant (in case of bipropellants, for a better homogenization of the mixture) and can be used as a coaxial centrifugal injector. The combustion chamber is the part of thrust chamber in which the combustion of propellants occurs at high pressure and temperature. The process of combustion can be characterized in three zones (Kessaev, 2006): 1) gasification zone (warming-up, evaporation), 2) burning zone, and 3) combustion product mixing zone. Internal cooling of the combustion chamber is made to protect the inner shell from contact with the high temperature gases. They can be of two types: by wall layer, which is constituted by a flow of combustion products with lower burning temperature along the wall; and a wall screen formed by the liquid fuel film. The hot gases of combustion are accelerated from the stagnation to transonic velocity in the throat, reaching supersonic velocity in the exit of the nozzle (Barrère, 1960). There are different types of nozzles, including conical, contoured, and ring nozzle. Conical nozzles are simple and relatively easy to manufacture but they are not the most efficient ones in terms of thrust for a given length. The type of nozzle chosen in this work was the conical one, due to simplicity and easiness to manufacture it. The basic objective of the cooling jacket in a thrust chamber is to prevent its walls from becoming too hot, and enable them to withstand the imposed thermal loads and stresses. Most materials lose strength and become

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine

weaker when temperature is increased. Cooling, thus, reduces the wall temperatures to an acceptable value (Sutton, 2001). The regenerative cooling is carried out by a cooling jacket around the thrust chamber by circulating one of the liquid propellants through it before it is fed to the injector head. The propellants are combined inside the combustion chamber where they chemically react to form hot gases which are then accelerated and ejected at high velocity through the nozzle (Huzel and Huang, 1992; Kessaev, 2005). The equation of thrust in atmospheric conditions (Eq. 1), as shown in Fig. 2, is given by: FH = m.Wa + A a (Pa - PH )

(1)

Where Aa is the outlet area of the nozzle, Pa is the outlet static pressure of the nozzle and PH is the ambient pressure. This equation shows a balance between the created forces due to the mass ejected by the rocket and the force due to the effects of pressure at the exit plane of the nozzle. In space applications, the ambient pressure is considered zero (vacuum).

k-1 2kRTch ³ £ Pa ¥ k µ I sp = 1(k-1) ³ ²¤ Pch ´¦ µ ³ µ

Equation 2 also shows that the Isp decreases with the increase of atmospheric pressure and the same is directly proportional to the thrust. Equation 3 shows that when the LRE is operating in vacuum, with the value of Pa tending to zero, the value of Isp is independent on the combustion chamber pressure. The characteristic velocity, C*, is a figure of thermochemical merit for a particular propellant and may be considered as an indicative of the combustion efficiency. It can be calculated in two ways: experimentally through the chamber pressure measurement (Pch), critical area section (Acr) and mass flow rate (m ) according to Eq. 4 and from the thermodynamics properties of propellants used according to Eq. 5: C* = 

Pch A cr m

F H

Wa Wo = 0

F H

£ 2 ¥ k-1 2k ²¤ k+1´¦ . k+1

(5)

R: gas constant in (J/Kg.K); Tch: temperature of combustion gases in chamber in (K); Pa

k: adiabatic exponent. The value of C* empirically calculated considers a loss due to the friction and the movement of the combustion gases in the throat of the nozzle. This loss depends directly on the profile of the nozzle.

PH Figure 2. Thrust in atmospheric conditions.

The specific impulse Isp is the main performance measurement of an LRE. It is used as a base for comparison among propellants, combinations of propellants, and overall performance of LREs. The Isp can be calculated in two ways: experimentally, measuring the thrust (F) and the of the propellant according to Eq. 2 and mass flow rate (m) from the thermodynamics properties of the propellants for a given expansion ratio Pa/Pch, according to Eq. 3: (P-P -PH )) a (P a -P) F AaA A (P A Isp = = a (P a aa -P a HH )H F F F Isp Isp = = = = Isp = m m  = m m m m m m  

P H

Where: PH

F = mWa P H

(4)

RTch

C* = F

(3)

(2)

METHODOLOGY OF LABORATORY WORK The experimental test stand was initially designed to carry out firing and cold flow tests of liquid propellant rocket engines using a pressurized propellant feed system. The LRE will use ethyl alcohol (C2H5OH) as fuel and gaseous oxygen as oxidizer. The choice of the propellants (ethyl alcohol and gaseous oxygen) was made based on requirements for low cost, ease of acquisition, and non toxic

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

161

combustion product. Gaseous oxygen can be readily and inexpensively obtained in pressurized cylinder in almost all communities because it is used in oxyacetylene welding. With reasonable precaution, it is safe to handle and for rocket test stand use. The alcohol is readily available in some communities. Safety precautions are already known by the most responsible personnel due to wide use of the fuel in internal combustion engines for automobiles and other power machines. The combustion chamber will be water cooled. The pressurization of the fuel tank and the pilot lines that feed the pressure regulators, as well as the purging of the fuel lines, will be made with gaseous nitrogen (GN2) from pressurized cylinders at 200 bar. The test stand allows tests of rocket engines in different regimes of operation by enabling the variation of parameters such as chamber pressure, mass flow rate, and firing time. It provides measurements of several physical variables associated with these regimes. The test stand will be controlled automatically by a computer (PC) that will enable real-time measurements during the rocket engine firing, allowing remote control and more safety to the operator. The tests will be carried out in four different regimes of operation with constant mixture ratio (km) â&#x20AC;&#x201C; ratio between the mass flow rate of oxidizer and mass flow rate of fuel. The change of regimes will be obtained by the variation of propellant feed line pressures, mass flow rates, and chamber pressures. The firing time will also be varied, limited only by the size of the fuel and oxidizer tanks. The number of operational regimes can be extended calibrating the system with additional values of feed line and chamber pressures. The tests will be conducted sequentially according to a pre-defined operational regime. The system will also allow the user to choose one unique regime of operation among four available options. Figure 3 shows a test with four regimes of operation carried out sequentially, and Fig. 4 illustrates an example of choice for an only regime that can be sometimes replied to survey statistics.

Trust (Kg)

100

9 10

Time (s) Figure 4. Representation of one unique regime.

The control system will basically work in three different conditions: set up, operation, and shut down; operation condition is subdivided into starting, test, wait and finishing. The test is initiated with actions of opening and closing synchronized feed line valves, firstly to ignite the gasdynamic igniter and then to provide the conditions to start the engine firing. During engine firing, the mixture ratio is kept constant by controlling the propellant line pressures with automatic pressure regulators (PID system). Initially, the pressures related to mass flow rates (fuel and oxidizer) are obtained by calculations but verified experimentally. DEVELOPMENT OF TEST STAND INSTALLATIONS The main requirements of the test stand and engine developed are: a) propellants are ethyl alcohol and oxygen gas; b) capacity of generating a 50 to 100 kgf range of thrust;

d) four regimes of operation with chamber pressure of 8, 10, 12 and 15 bar;

80 60

e) maximum admissible pressure of combustion chamber: 20 bar;

40 20 0

10 15 20

25 30 35

Time (s) Figure 3. Sequential regimes. 162

45 40 35 30 25 20 15 10 5 0

c) constant mixture ratio (km) for maximum specific impulse;

120

Trust (Kg)

Santos, E.A. et al

40 45

f) duration of each regime of operation equal to 10 seconds; g) constant mass flow rate using pressure regulators in the propellant feed lines;

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine

h) gas dynamic igniter for multiple ignitions; i) water-cooled jacket; j) automated data acquisition and control system; k) acquisition of temperature values, pressure, mass flow rate and thrust; l) system safety is automatic, safety routine program is to ensure that safety, consistent with mission requirements, is designed into systems, subsystems, equipment, facilities, and their interfaces. Considering the mixture ratio km equal 1.6 to maximize specific impulse, the values for mass flow rates of propellants of the combustion chamber and their respective thrusts for four operational regimes were calculated and the results are shown in Table 1.

Table 1. Pressures in the combustion chamber (Pch), mass flow O ,m f ) and thrust (FH) values. rates (m



Pch (bar)

m  O (kg/s)

m  f (kg/s)

FH (kgf)

1º

0.200

0.125

56.05

2º

0.250

0.157

70.22

3º

0.301

0.188

84.43

4º

0.376

0.235

105.78

Regimes

The thrust measurement system is part of the test stand frame and it is constituted of an interface support that fixes the engine in the frame, a compression load cell (which measures the thrust force) and a calibration mechanism, as the system shown schematically in Fig. 6. Strain gages, load cell, were placed on the beam to convert strain to a voltage proportional to the thrust force. The calibration is easily effected by adding weights. In order to obtain a proper zero reading for all thrust measurements, it is necessary that the whole measurement system be pre-stressed with a pre-load of 5-10% of the expected thrust. The frame of the test bench is constituted of the parts shown in Fig 6.

Beggining of ribs 5 10 11 9

Figure 5 provides an AutoCAD drawing in cross section of a thrust chamber, detailing the ignition system (detail 1) and the mechanical interface of the engine with the frame of the test stand (detail 2). The injector head will be made of stainless steel except the base plate (detail 12) and igniter ducts (detail 7), which will be made of copper. The combustion chamber will be made of stainless steel. On the inner shell (detail 6), there will be milling ribs that begin in the cylindrical part and finish in the throat; the cooling jacket is constituted of a double wall shell, without ribs. The inlet and outlet of the cooling system can also be seen in details 15 and 4. The nozzle is segmented in parts that can be removed during the laboratory work as configuration makes it possible to verify the influence of expansion ratio on thrust. Consequently, three conditions of expansion in the nozzle can be tested: under expanded, over expanded, and adapted. In the planned work, the tests to determine LRE thrust characteristics will be carried out only in three segments (details 16, 17 and 18 of Fig. 5) corresponding to an adapted nozzle at sea level.

End of ribs

14 15

16 17

Figure 5. General drawing of longitudinal cross section thrust chamber. J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

163

Santos, E.A. et al

b) 990

Thrust

Calibration mechanism

Frame

Load cell Air spring Interface

3 4

Engine

1836

Weights

300

Figure 6. Thrust measurement system: (a) lateral view of scheme test stand and (b) design in CAD. 1: load cell (thrust force measurement); 2: air spring (pneumatic spring); 3: support to fix the engine in the frame; 4: rocket engine; 5: damper; 6: balance beam; 7: body; 8 and 9: counterweights (5 and 10 kg).

A structural analysis of the test stand frame, using the finite element method, was carried out to evaluate the linear static behavior and modal to steel ASTM-A36. The main results of the analysis are listed below and Fig. 7 shows the mesh and Von Mises stress. The values of stress are below the admissible stress in the analyzed models and no problem was found in the screws of the frame and in the bearings of the load cell. Table 2 shows the natural frequency of the frame test stand. Table 2. Natural frequency.

Frequency number 1 2 3 4 5

Frequency (seconds) 0.7136914E-01 0.4997758E-01 0.3089837E-01 0.2763159E-01 0.1987037E-01

According to Fig. 7, the maximum stress was obtained where the engine was fixed. The results to stress, displacement, and values to steel ASTM-A36 are: •

admissible stress=150 MPa;

•

maximum vertical displacement=0.45 mm;

•

maximum stress (Von Mises)=83.9 MPa;

•

safety margin=(admissible stress – maximum stress)/ maximum stress=79%.

164

The feed system is constituted of an hydropneumatic installation illustrated in Fig. 8, composed of piping lines, a series of valves, provisions for filling and removing (draining and flushing) the liquid propellants, and control devices to initiate, stop, and regulate their flow and operation. GOX passes from a stored cylinder through a pressure regulator, where its pressure is reduced and feeds the engine when electric valves are opened. Part of the GOX is also used in the ignition unit. GN2 from a stored cylinder is reduced by the pressure regulator and pressurizes the fuel tank to feed the engine. The scheme also shows the cooling water inlet in the throat region of the engine and the exit from the mixing head. Figure 8 also shows temperature, pressure, and thrust measurement cell load sensors. The layout of the test complex is shown in Fig. 9, which gives an overview of the arrangement, equipment of control and data acquisition system. The interface between the test stand and the control room will be carried out by signal transmission cables represented by the black dotted line in Fig. 9. The propellant and pressurization gas will be located in three separated bays, and the engine rocket will be installed in the frame of test stand. Figure 9 also shows the control room which not only serves to protect test personnel from a possible combustion chamber or tank explosion, but also contains the equipment controlling for test operation and data acquisition equipment.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine

b) Von Mises 8.3925E + 007 7.3434E + 007 6.2944E + 007 5.2453E + 007 4.1963E + 007 3.1472E + 007 2.0901E + 007 1.0491E + 007 0.000000000

Figure 7. Results – (a) Mesh and (b) Stress (Von Mises).

A VM2

B C

VS1

RP3

PG2

VA2

RA2 VM14

RA1

VM5 F1

VS5

VS9 P10

VS10

VA5

PG6 P8

MV4

MV2 VR2 P2

VR1

P11 P12 P13 P14

LC1

VM10 T3

Water Inlet GOX VM7

VM9

RP4

VS8

igniter

VS7

PG5

PG7

MV1

RP5 VM11 VA6

MN1 Fuel tank

VS4

VM12

VM4

GN2

VM3

P1 VM15

VM6

7 VM13

VS2 VS3

PG3

Ignition Unit

RP1

RP2 VA3

VM1

VA1

PG1

PG4

VA4

Combustion Chamber

VM8

VS6 MV3

P6 P15 P16 P17

Figure 8. Hydropneumatic scheme. RP(n): pressure regulator (n=1,2,3,4,5), T(n): thermocouple (n=1,2,3,4,5), VA(n): relief valve (n=1,2,3,4,5,6), VM(n): manual valve (n=1,2,3,...,14), VS(n): solenoid valve (n = 1,2,3,...,10), MV(n): flowmeter (n=1,2,3,4). J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

165

Santos, E.A. et al

BAY 1 (FUEL AND WATER)

BAY 2 (PRESSURIZATION GAS) BAY 3 (GAS OXYGEN) INSTRUMENTATION ROOM TEST STAND

CONTROL AND DAQ LINES

IGNITION SYSTEM

FUEL (ALCOHOL) PRESSURIZATION GAS (GN2) GAS OXYGEN (GOX)

50 m

WATER CONTROL ROOM

Figure 9. Layout of test complex.

The control and acquisition system in use in the test stand is composed by the elements presented in the system architecture (Fig. 10). The pressure transducers (PT), mass flow rate (FT), strain or thrust (ST) and temperature (TT), shown in Fig. 10, are sensitive elements that convert the physical phenomena of the experiment in electrical signal that will be conditioned and later analyzed. The signals will be conditioned and processed in the PXI platforms of instrumentation for measurement and automation of National Instruments.

EXPERIMENT INSTR. ROOM PT FT ST TT

BLI SENSORS

SWITCH

Ethemet

RACK PXI 1 BLI ACTUAT.

REGULATORS

Setpoint HMI

Regulator pressure inlet

Feedback transmitter

Pressure to be controlled

HMI: human-machine interface.

PT: pressure transducers; FT: mass flow rate; ST: strain or thrust; TT: temperature; HMI: human-machine interface. 1: controller system only for temperature data; 2: temperature peripheral module; 3: controller for other data acquisition and control; 4 and 5: pressure regulator. Figure 10. Data acquisition system architecture.

In Fig. 10, item (1) is a controller system only for temperature data, (2) is the temperature peripheral module, (3) is a controller for other data acquisition and 166

Electropneumatic pressure controller

HMI

TB ACTUATORS

It will be possible through the automatic system to control the pressure in the lines of fuel and oxidizer to keep the mass flow rate constant. It is done by a microcontroller-based device that implements a digital control algorithm (PID controller) to regulate pressure. PID controller involves three parameters of control: proportional, integral and derivative values that are obtained by test. The weighted sum of these three actions is outputted to a control element such as the position of the control valve. Figure 11 shows a scheme of control in pressure regulators using computer: the user is responsible for fixing a setpoint value to control the pressure.

CONTROL ROOM 3 45 RACK PXI 2

1 2

control and (4) is the pressure regulator. The interface with user or human-machine interface (HMI) is carried out by means of a personal computer connected by ethernet and a software that, in this case, will be the LabVIEW program. The control will be carried out through the opening and closing of the solenoid valves, also commanded by PXI racks.

Figure 11. Control in regulators by computer to keep the pressure constant.

METHODOLY and Development of program The program developed for test stand control and data acquisition for LRE test uses the software LabVIEW because of its user interface and the readable, scalable and maintainable algorithm. The development of this program was possible due to a well defined development methodology. It consists of the phases shown in Fig 12.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine

Design 9Analysys

In this phase, the interface between the user and the available infrastructure was described to meet the functioning of the LRE and installation of the hydropneumatic scheme, as shown in Fig. 8.

9Design

Methodology for the development of program

9Building 9Verification 9Implantation

Building

Figure 12. Phases of the methodology for the program development.

Analysis In this phase, the investigation of the necessary data for the understanding of the design requirements was done. Hydropneumatic scheme was planned and developed with the schedule of all its activities from activation through shutdown of the test stand. Based on the hydropneumatic scheme shown in Fig. 8, the state machine was prepared, with the purpose of supplying and focusing on the development of the program (Shaw, 2003). Figure 13 shows the state machine of the hydropneumatic scheme of test stand.

Start

Firing test working Set up Stand-by Wait a new test

Choose a test Operation

Meet requirements

Finish test

In this phase, there was the interpretation of performed requirements resulting in the development of the LabVIEW codification. All codification sequence was made in a modular and scalable way, with comments inserted in the program. Figures 14, 15 and 16 show some LabVIEW screens developed to control the test stand. Figure 14 shows the first screen of the program with its guides. This screen, called “Gravação”, allows the user: •

to choose the name and the folder where the acquired test data will be saved;

•

to monitor and save the generated sensor data;

•

to finish the check of sensors when they are being monitored;

•

to determine the time of countdown of the test.

The guide screen “Canais” allows the user to choose the type of configuration and adjust the data channels for the valves or keep the default. Figure 15 shows the guide screen “Seleção das Pressões”, displaying chamber pressures available to the test stand.

Check Safety Shut Down End

Figure 13. Test stand state diagram.

According to Fig. 13, during the set up phase, the test stand is checked, i.e, valves, pressure regulators, gas lines and so on, until the stand-by state, when it will be ready to choose a test. During the test stand work, safety routines are used to monitor some system parameters like pressure and temperature, in order to avoid incidents due to super heating or pressure higher than the one specified in the design. The safety routines are programmed to perform determined actions of system protection by executing routines like closing of valves, depressurization of lines, banks etc. The test is finished in the shut down state.

Finally, Fig. 16 shows the guide screen “Ensaio”, where the user can perform and observe other options such as an elapsed time clock, a clock with countdown to the test, and the pressures of the combustion chamber carried out at the moment of the test. There is one visual alert when the internal pressure during tests exceeds the pre-established values. Verification This phase enables the user to identify and correct errors in the program. The following verification tests were performed: a) software function test, in which one can verify if all requirements were contemplated in the codification;

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

167

Santos, E.A. et al

Figure 14. Initial screen of interface with user.

Figure 15. Guide screen “Seleção das Pressões”. 168

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

Development of test stand for experimental investigation of chemical and physical phenomena in Liquid Rocket Engine

Figure 16. Guide screen “Ensaio”.

Implantation

black box testing of the functionality of software and its interaction with the user;

white box testing, in which the source code was tested in detail and all transitions of the program demands were verified;

In this phase, it was defined the position of the control hardware on the test stand inside the control room in the LPL (Fig. 9). The responsibility level of each operator during the test is in accordance with the standards of IAE.

d) integration testing, in which the architecture of the program and its communication with external interfaces were tested.

This standard requires an IAE qualified professional to be the test supervisor and manage the step operations of the test stand.

Description of test

CONCLUSION

The objective of test was to simulate the parameters through the computer in ambient Windows using the National Instruments Data Acquisition System. In order to simulate the opening and closing of the solenoid valves, a schematic panel of the hydropneumatic system was used with leds representing the state of the valves. These leds turn on/turn off at a determined time, representing the open/close state of the solenoid valves during the tests. To verify the data acquisition system, a device called DAQ-Acessory was used to supply a square wave as analogical input and was displayed and stored by program.

This work presented the specification of an experimental firing test stand for a LRE and the apparatus to its development in order to carry out the laboratory work. The design of a hydropneumatic scheme and a LRE led to the development of a program to control and acquire data, providing conditions to the engine work in agreement with parameters of design. All the phases of this project were done with two major aims: the excellence required in the field of Space Engineering and the state of the art of several Engineering fields like electricity, mechanics, thermodynamic, heat transfer, propulsion, combustion. These aims result in

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

169

Santos, E.A. et al

higher knowledge and aggregate value to the development and investigation of LREs.

Barrère, M., et al., 1960, “Rocket propulsion”, Elsevier Publishing Company, London.

ACKNOWLEDGEMENTS

Brazilian Space Agency, 2005, “National Plan of Space Activities. PNAE 2005-2014”. Brasília, DF, 44p.

This work was supported in part by the Institute of Aeronautics and Space (IAE) and Aeronautics Institute of Technology (ITA).

Huzel, D.K., Huang, D.H., 1992, “Modern engineering for design of Liquid Propellant Rocket Engines”. AIAA, Washington.

REFERENCES

Kessaev, K.V., 2006, “Theory and Calculation of Liquid Rocket Engine”, In: “Fundamental Course in Engine Course Design”, ITA/MAI, São José dos Campos.

Alves, W.A.F., 2008, “Development of experimental firing test stand to study the rocket engine thrust characteristics”, Masters Thesis, Aeronautics Institute of Technology, São José dos Campos, 198f. Andrade, E., 2008, “Graphical Programming applied to the control firing test stand of liquid rocket engine”. Masters Thesis, Aeronautics Institute of Technology, São José dos Campos, 133f.

170

Kessaev, K.V., 2005, “Introduction to Liquid Rocket Engine Design”, In: “Fundamental Course in LRE Introduction”, CTA/ITA, São José dos Campos. Shaw, A.C., 2003, “Systems and software in real time”, Porto Alegre: Bookman. Sutton, G.P., Biblarz, O., 2001, “Rocket propulsion elements”, John Wiley & Sons, New York.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 159-170, May-Aug., 2011

doi: 10.5028/jatm.2011.03022711

Luís Antonio Silva*

Instituto de Aeronáutica e Espaço São José dos Campos/SP – Brasil silvalas@iae.cta.br *autor para correspondência

Investigação da distribuição do filme de resfriamento em um motor-foguete à propulsão líquida Resumo:O presente estudo apresenta os resultados da investigação de um método de resfriamento amplamente utilizado em câmaras de combustão, denominado filme de resfriamento, que é aplicado a um motor-foguete à propulsão líquida. Esta utiliza como propelentes oxigênio líquido e querosene. Partindo de um motor cujo filme de resfriamento é formado por meio da introdução de combustível pelos injetores posicionados na periferia do sistema de injeção, analisou-se experimentalmente o filme formado pelo líquido que escoa pela parede interna da câmara de combustão. O parâmetro utilizado para validação e refinamento dos dados teóricos foi o comprimento do filme de resfriamento, pois esse parâmetro é de suma importância para que se obtenha uma proteção térmica eficiente internamente à câmara de combustão. Os ensaios a frio confirmaram um comprimento suficiente do filme de resfriamento para a câmara de combustão do motor estudado. Palavras-chave: Filme de resfriamento, Motor-foguete, Propulsão líquida.

Investigation of the cooling film distribution in liquid rocket engine Abstract: This study presents the results of the investigation of a cooling method widely used in the combustion chambers, which is called cooling film, and it is applied to a liquid rocket engine that uses as propellants liquid oxygen and kerosene. Starting from an engine cooling, whose film is formed through the fuel spray guns positioned on the periphery of the injection system, the film was experimentally examined, it is formed by liquid that seeped through the inner wall of the combustion chamber. The parameter used for validation and refinement of the theoretical penetration of the film was cooling, as this parameter is of paramount importance to obtain an efficient thermal protection inside the combustion chamber. Cold tests confirmed a penetrating cold enough cooling of the film for the length of the combustion chamber of the studied engine. Keywords: Cooling film, Rocket engine, Liquid engine.

INTRODUÇÃO A atividade espacial contribui de maneira significativa para o desenvolvimento do Brasil, seja pelas informações que disponibiliza, sob a forma de imagens e dados coletados sobre o território nacional, ou pelo efeito indutor de inovação que decorre dos esforços na aquisição, desenvolvimento de tecnologias e de conhecimentos críticos para atender às necessidades do Programa Nacional de Atividades Espaciais (PNAE) de 2005 a 2014. Tais esforços resultam em um benefício para a indústria e para a sociedade (Agência Espacial Brasileira, 2005). A construção de veículos lançadores não apenas garante e preserva a necessária autonomia para o acesso ao espaço, Received: 09/06/11 Accepted: 04/07/11

como possibilita também a exploração comercial dos serviços de lançamento. Para alcançar seu objetivo, o PNAE propõe, entre outras diretrizes, a capacitação do país na área de propulsão líquida, inicialmente para equipar os estágios destinados a garantir precisão de inserção em órbita dos satélites e, subsequentemente, para integrar os grandes lançadores, objetivando o crescimento da capacidade dos veículos lançadores em disputar o mercado internacional de transporte espacial (Agência Espacial Brasileira, 2005). O atual veículo lançador de satélites (VSL) brasileiro, denominado VLS-1, possui quatro estágios com a missão de lançar satélites de massa entre 100 a 350 kg, em altitudes de 200 a 1.000 km. Com o domínio da tecnologia de propelente líquido, será possível o desenvolvimento

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

171

Silva, L.A.

dos lançadores de médio e grande porte, visando à inserção de satélites em órbita geoestacionária (Agência Espacial Brasileira, 2005). O veículo com a tecnologia de propulsão líquida é mostrado na Fig. 1.

MFPL

3 1 – Gerador de gás; 2: unidade de turbobomba; 3: agregados; 4: câmara de empuxo.

Figura 2. Componentes de um MFPL (Silva, 2009).

Existem diversos métodos de resfriamento que permitem ao material das paredes da câmara de combustão manter sua integridade estrutural. Os mais comuns são: o resfriamento regenerativo (regenerative cooling), resfriamento por transpiração (transpiration cooling), filme de resfriamento (ﬁlm cooling) e o revestimento (coating). Tais métodos são brevemente apresentados a seguir (NASA, 1972). Resfriamento regenerativo: Um ou mais propelentes circulam como fluido refrigerante em torno da superfície externa da parede a ser resfriada. Este método também é conhecido por jaqueta de refrigeração.

Fonte: IAE, modificado pelo autor. Figura 1. VLS.

Assim, para a obtenção do motor-foguete à propulsão líquida (MFPL), que equipará o VLS, é indispensável um método eficaz de resfriamento das paredes que compõem o corpo da câmara de empuxo, devido às elevadas taxas de transferência de calor envolvidas. Com este intuito, este artigo apresenta uma investigação do método de resfriamento denominado filme de resfriamento (ﬁlm cooling), por meio da análise do comportamento do filme ao longo da câmara de combustão. MFPL Em termos gerais, pode-se admitir que os MFPL pressurizados por turbobomba, como o que se pretende estudar, são dispostos conforme apresentado na Fig. 2. A câmara de empuxo é um dos subsistemas principais de um motor-foguete, pois nela os propelentes são dosados, injetados, atomizados, vaporizados, misturados e queimados, gerando gases a altas temperaturas. Na sequência, tais gases são acelerados e expelidos a velocidades supersônicas (Sutton, 1986; Huzel, 1992). 172

Resfriamento por transpiração: Uma parede porosa é resfriada internamente pelo escoamento forçado do fluido refrigerante por meio dos poros do material. Filme de resfriamento: Uma fina camada de fluido refrigerante é mantida sobre a superfície interna da parede da câmara de combustão. Este método é o objetivo deste estudo investigatório e está detalhado no decorrer deste artigo. Revestimento: Uma camada de material de baixa condutividade térmica é depositada como barreira térmica no lado interno da parede. É importante ressaltar que a utilização de apenas um método de resfriamento não é suficiente, o que implica no uso combinado de duas ou mais técnicas de resfriamento. Este trabalho concentra-se no método de resfriamento, por meio do filme de resfriamento, porque este representa uma técnica muito utilizada em câmaras de empuxo como a que se pretende estudar. O método estudado apresenta como vantagens: o resfriamento da parede interna da câmara de empuxo, utilizando o próprio

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

Investigação da distribuição do filme de resfriamento em um motor-foguete à propulsão líquida

propelente fornecido para o funcionamento do motor, pois não necessita de adição de novos componentes que acarretariam em acréscimo de massa ao estágio; o fornecimento de líquido para a formação do filme de resfriamento é realizado pelos injetores posicionados na periferia do cabeçote, os quais são componentes do sistema de injeção da câmara de empuxo e possibilitam a utilização dos mesmos conceitos já aplicados na injeção dos propelentes na câmara de combustão. Neste método, o líquido que resfria a câmara participa da queima dos propelentes. Como desvantagens da utilização do método de resfriamento por filme de resfriamento podem ser citadas: a adição de propelente na superfície da parede da câmara de combustão modifica a razão de mistura do par propelente, reduzindo a eficiência do motor; em alguns casos, este método não é suficiente para ser utilizado como único recurso de resfriamento; devido ao filme de resfriamento ser consumido ao longo da câmara de empuxo, não é possível, em alguns casos, o resfriamento de toda a câmara de empuxo, principalmente a garganta, em que ocorre a maior taxa de transferência de calor.

Filme de resfriamento Conforme foi apresentado, há alguns métodos para resfriar a câmara em projetos de motores foguete. Assim, o filme de resfriamento pode ser produzido pelo sistema de injeção por meio da aspersão de um dos propelentes pelos injetores periféricos ou pelos canais, dispostos no fechamento inferior do cabeçote de injeção, direcionados à parede da câmara de combustão. Para que o material da câmara de empuxo possa resistir às altas temperaturas dos produtos de combustão, organiza-se a mistura dentro da câmara de empuxo de modo a formar regiões de gases com temperaturas decrescentes, na direção do núcleo para a periferia. Para tanto, usualmente, os injetores periféricos possuem razão de mistura mais rica em combustível do que os injetores centrais, proporcionando temperatura de estagnação mais baixa junto à parede. Este comportamento gera regiões que podem ser observadas na Fig. 3 (Kessaev, 2006). m

Para a investigação de quaisquer métodos de resfriamento, é primordial o conhecimento do motor a que se pretende estudar, por meio de seus requisitos e parâmetros técnicos de funcionamento. O motor em desenvolvimento para aplicação no VLS-1 denomina-se L75 e seus parâmetros técnicos de funcionamento principais estão apresentados na Tabela 1.

1 2 3 4 T Núcleo

T Periféria T Parede T Líquido

Tabela 1. Parâmetros técnicos de funcionamento do L75.

Parâmetro Descrição P  global m  ko m k m

km global km k Is global Fonte: IAE.

Unidade dimensional 75 kN 23,341 kg/s Valor

Empuxo Vazão mássica global Vazão mássica de 15,806 oxidante na câmara de combustão Vazão mássica 7,201 de combustível na câmara de combustão Relação de mistura 2,17 global Relação de mistura 2,19 da câmara de combustão Impulso específico 327,7 do motor

kg/s

1: Núcleo da câmara de combustão, em que as temperaturas variam tipicamente entre 3.500 e 4.000 Kelvin aproximadamente (TNúcleo); 2. Região de mistura entre o fluxo pertencente ao núcleo e a região 3; 3: Wall layer é a região próxima à parede da câmara de combustão com temperaturas que chegam a atingir de 2.300 a 2.800 Kelvin aproximadamente (TPeriferia); 4: Região denominada Camada Limite, na qual geralmente atingem-se temperaturas entre 1.200 e 1.300 Kelvin (TParede).

Figura 3. Regiões de mistura (SILVA, 2009).

kg/s

Nas principais câmaras de empuxo existentes que utilizam o método de resfriamento regenerativo, observa-se a aplicação simultânea do ﬁlm cooling, o qual é normalmente aplicado por meio dos injetores periféricos (NASA, 1972). A Tabela 2 mostra a aplicação do filme de resfriamento em alguns motores regenerativos conhecidos. O sistema de resfriamento deve ser projetado visando o preenchimento homogêneo de líquido no perímetro da câmara de combustão. Para tanto, há necessidade de se investigar a distribuição do líquido aspergido pelos injetores

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

173

Silva, L.A.

Tabela 2. Film cooling em motores regenerativos (NASA, 1972).

Propelentes Líquido refrigerante Motores empregados para film cooling F-1 LOx/RP-1 RP-1 J-2 LOx/LH2 GH2 RL 10 LOx/LH2 GH2 Agena IRFNA/UDMH UDMH Titan I family LOx/RP-1 RP-1 Titan II family N2O4/A-50 A-50 Titan III family N2O4/A-50 A-50 X-15 LOx/NH3 NH3 periféricos, com vistas à isenção das áreas de deficiência na proteção térmica que permitem pontos quentes. Ocasionalmente, o filme pode ser utilizado para solucionar superaquecimentos, localizados pelo fornecimento do filme de líquido diretamente na área afetada. Os principais problemas, que envolvem o projeto do sistema de resfriamento por ﬁlm cooling, resultam da dificuldade em (NASA, 1972): •

fornecer a quantidade exata de líquido de resfriamento (vazão mássica), pois o fornecimento de líquido deve ser suficiente para o resfriamento, mas não deve degradar acentuadamente o impulso específico;

•

determinar o comprimento e o arranjo do fluxo de líquido na quantidade requerida de líquido refrigerante para que este seja usado de forma eficiente, proporcionando um recobrimento homogêneo da parede da câmara de combustão.

Com relação ao fornecimento de líquido refrigerante à parede da câmara de combustão, há um entendimento – um tanto qualitativo – de que este fornecimento deve (NASA, 1972): minimizar a mistura entre os reagentes buscando tornar o filme de resfriamento com um único componente; e maximizar a cobertura das superfícies a serem protegidas. A formação do filme de resfriamento, no caso do motor L75, ocorre pela inserção de líquido rico em combustível por meio dos injetores periféricos. Esse líquido colide com a parede da câmara de combustão, fazendo com que uma parcela se desprenda e componha a parcela de líquido definida como atomização, enquanto que outra porção do líquido fornecido adere à parede formando uma película denominada ﬁlm cooling. Este mecanismo de formação é mostrado na Fig. 4. O líquido aspergido pelo injetor periférico forma um cone, com ângulo de abertura diferente do cone formado pelos 174

filme

belt screen

garganta

qmax x é a distância entre a face do injetor periférico em que há a formação do cone de combustível e o ponto onde ocorre a colisão do líquido (ponto vermelho na Figura 4) e a parede da câmara de combustão; e é a espessura do filme de resfriamento; y é a distância entre a face do injetor periférico em que há a formação do cone de combustível e a parede da câmara de combustão; 2α é o ângulo formado pelo cone de combustível do injetor periférico; β é o ângulo formado pela trajetória do líquido que, após colidir com a parede da câmara de combustão, segue para a região do núcleo e a parede da câmara de combustão; lfilme é o comprimento ou penetração do filme de resfriamento ao longo da câmara de combustão; L é o comprimento da parte cilíndrica da câmara de combustão. Esse comprimento foi determinado teoricamente e, para o motor L75, é de 190 mm.

Figura 4. Formação do filme de resfriamento (SILVA, 2009).

injetores centrais. Esse procedimento visa proporcionar ao líquido injetado pelos injetores periféricos um maior alcance, pois na região próxima à face, na qual estão posicionados os injetores, não há necessidade de resfriamento, visto que as temperaturas geradas nessa região são facilmente suportadas pelo material da câmara de combustão. Entretanto, nas proximidades da garganta, onde acontece a maior transferência de calor, há a necessidade de uma combinação dos sistemas de resfriamento. A distância percorrida pelo líquido aspergido dos injetores periféricos, após a sua colisão com a parede da câmara de combustão, recebe o nome de penetração e pode ser determinada teoricamente. A validação dos cálculos que envolvem a penetração do filme de resfriamento ocorre por meio de ensaios a quente. Portanto, não fazem parte do presente estudo, visto que este limita-se à investigação por meio do ensaio a frio do sistema de injeção. Outro parâmetro de suma importância na investigação do filme de resfriamento é a razão de mistura do filme de resfriamento, pois ocorre a interferência do líquido aspergido pelos injetores centrais na composição do filme de resfriamento. Para a análise desse parâmetro, é possível comparar os dados teóricos e os reais, obtidos por meio de ensaios a frio, com o auxílio de um espectrofotômetro. Este procedimento é proposto como trabalho futuro para a continuação do estudo apresentado.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

InvestigaĂ§ĂŁo da distribuiĂ§ĂŁo do filme de resfriamento em um motor-foguete Ă propulsĂŁo lĂquida

PenetraĂ§ĂŁo do filme de resfriamento

onde:

Ă&#x20AC; medida que o filme de resfriamento percorre a parede da cĂ˘mara de combustĂŁo, ele ĂŠ submetido a elevadas temperaturas provenientes da combustĂŁo, sendo consumido e, portanto, havendo uma diminuiĂ§ĂŁo da eficiĂŞncia da proteĂ§ĂŁo tĂŠrmica na parede da cĂ˘mara de combustĂŁo. Para que o filme de resfriamento percorra uma distĂ˘ncia suficiente, ĂŠ necessĂĄria a determinaĂ§ĂŁo da penetraĂ§ĂŁo, denominada Liquid ďŹ lm length (lfilme). Essa penetraĂ§ĂŁo ĂŠ determinada por meio do balanĂ§o de energia. A expressĂŁo (Eq. 1) que representa a penetraĂ§ĂŁo do filme de resfriamento na cĂ˘mara de combustĂŁo ĂŠ (Kessaev, 2006):

Th ĂŠ a temperatura inicial do lĂquido e corresponde a 298 K;

l filme "

m filme resfr Cp Ts Th Qs x D q aquec q evap

(1)

onde, Îˇ ĂŠ o coeficiente que contabiliza a porcentagem de lĂquido do filme consumido pela combustĂŁo. Esse coeficiente depende do nĂşmero de Reynolds, calculado para o filme de resfriamento, ReL. Esta relaĂ§ĂŁo ĂŠ mostrada no grĂĄfico da Fig. 5.

0.8 M(ReL)

0.6 0.4

1000

2000

3000

4000

5000

ReL

Figura 5. Consumo do filme de resfriamento (KESSAEV, 2006).

O nĂşmero de Reynolds deve ser determinado levando-se em consideraĂ§ĂŁo a vazĂŁo mĂĄssica no filme de resfriamento (m ď&#x20AC;Ś filme resfr ) e a viscosidade do lĂquido empregado nesse filme. Pela utilizaĂ§ĂŁo do grĂĄfico da Fig. 5, o valor do coeficiente Îˇ ĂŠ de 0,98 para o nĂşmero de Reynolds correspondente a 168,2. D ĂŠ diĂ˘metro da cĂ˘mara de combustĂŁo. O motor L75 apresenta esta dimensĂŁo igual a 0,18 m; Cp ĂŠ a capacidade tĂŠrmica do lĂquido Ă temperatura Tcp. Para a temperatura considerada, este valor ĂŠ de 2.805 kJ/kg K (Raznjevic, 1976); Tcp ĂŠ a temperatura mĂŠdia do lĂquido, determinada pela EquaĂ§ĂŁo 2:

Tcp "

Th TS 2

(2)

TS ĂŠ a temperatura de ebuliĂ§ĂŁo do lĂquido Ă pressĂŁo de cĂ˘mara, pk. Para o lĂquido empregado, esta temperatura ĂŠ de 351,45 K (Raznjevic, 1976); qaquec ĂŠ o fluxo total de calor para o aquecimento do filme lĂquido, determinado por meio da anĂĄlise tĂŠrmica; qevap ĂŠ o fluxo total de calor para a evaporaĂ§ĂŁo do filme lĂquido, determinado por meio da anĂĄlise tĂŠrmica; Qs ĂŠ o calor de evaporaĂ§ĂŁo do lĂquido aspergido para a formaĂ§ĂŁo do filme. Para o presente estudo, considera-se o combustĂvel;

ď&#x20AC;Ś filme resfr m

ĂŠ a vazĂŁo mĂĄssica de combustĂvel destinada Ă formaĂ§ĂŁo do filme de resfriamento. Na expressĂŁo utilizada para a determinaĂ§ĂŁo do filme de resfriamento, a vazĂŁo mĂĄssica no filme ĂŠ a variĂĄvel mais importante. Tal fato ocorre, pois, na investigaĂ§ĂŁo do ďŹ lm cooling proposta neste trabalho, a mediĂ§ĂŁo dessa vazĂŁo tornarĂĄ possĂvel verificar se os valores determinados teoricamente para a formaĂ§ĂŁo do ďŹ lm cooling sĂŁo suficientes a uma penetraĂ§ĂŁo que preencha o comprimento da parte cilĂndrica, L, da cĂ˘mara de combustĂŁo atĂŠ o recebimento de uma nova porĂ§ĂŁo de lĂquido de resfriamento, por intermĂŠdio do belt screen, conforme mostra a Fig. 4. Para o motor L75, a distĂ˘ncia y, apresentada na Fig. 4, ĂŠ de 8,65 mm e o Ă˘ngulo de cone, 2Îą, formado pelos injetores perifĂŠricos, de 94Ë&#x161;. A partir desses valores, ĂŠ possĂvel determinar a distĂ˘ncia x como sendo 8,1 mm. Por meio da expressĂŁo que representa o comprimento do filme e considerando a distĂ˘ncia x que o lĂquido aspergido alcanĂ§a antes de tocar na parede da cĂ˘mara de combustĂŁo, hĂĄ penetraĂ§ĂŁo do filme de resfriamento de 124 mm. O valor da penetraĂ§ĂŁo do filme de resfriamento estĂĄ relacionado Ă quantidade de lĂquido destinado a sua formaĂ§ĂŁo. O valor de 124 mm, determinado para a penetraĂ§ĂŁo, estĂĄ considerando que 50% da vazĂŁo mĂĄssica que escoa pelos injetores perifĂŠricos fazem parte da formaĂ§ĂŁo do lĂquido. Esta porcentagem ĂŠ uma estimativa e pode ser investigada pelo ensaio a frio, utilizando um dispositivo adequado. A correĂ§ĂŁo do valor da vazĂŁo mĂĄssica do filme de resfriamento possibilita uma avaliaĂ§ĂŁo mais precisa do

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

175

Silva, L.A.

comportamento do filme de resfriamento, principalmente no que diz respeito à penetração do filme.

dimensionado para a geometria deste motor, implicando em prejuízo para o tempo de operação, já que a vazão mássica do motor L75 é superior à mássica do L15.

INVESTIGAÇÃO DO FILME DE RESFRIAMENTO

Para melhor identificação do posicionamento dos injetores, usa-se um esquema que divide-os em camadas, como mostra a Fig. 7.

As simplificações consideradas no desenvolvimento desta investigação são: •

utilização de água como propelentes – oxidante e combustível;

•

ausência de contrapressão internamente ao dispositivo de ensaio;

•

tempo máximo de operação do dispositivo de ensaio proposto de quatro segundos.

Com a finalidade de garantir maior fidelidade dos valores colhidos, o ensaio é realizado no banco de testes utilizado para ensaios a quente do Laboratório de Propulsão Líquida (LPL), da Divisão de Propulsão Espacial – APE, do Instituto de Aeronáutica e Espaço (IAE). Este banco de ensaio é utilizado para o ensaio a quente dos motores L5 e L15. O banco de testes é capaz de fornecer vazões mássicas e pressões nos intervalos de tempo exigidos pelo experimento. O sistema de injeção utilizado no ensaio opera com as vazões mássicas e pressões especificadas para o motor L15, sendo que, para a operação com o motor L75, não há banco de testes com a capacidade requerida disponível até a data de término deste estudo. A montagem do dispositivo de ensaio no banco de 20 kN está apresentada na Fig. 6.

Injetores periféricos

Injetores centrais

Figura 7. Identificação dos injetores em camadas (SILVA, 2009).

Os injetores são ensaiados individualmente para a determinação de parâmetros, tais como a vazão mássica e o ângulo de cone. Cada injetor é identificado e sua posição no cabeçote é mapeada. A Fig. 8 apresenta o cabeçote de injeção utilizado para o ensaio.

Figura 8. Cabeçote de injeção (SILVA, 2009).

Figura 6. Dispositivo de ensaio no banco de 20 kN (SILVA, op. cit.).

O dispositivo de ensaios poderá ser reaproveitado para o mesmo teste hidráulico com o cabeçote de injeção do MFPL L75 quando houver disponível um banco de testes com maior capacidade, visto que tal dispositivo foi 176

O líquido aspergido pelos injetores periféricos escoa pela parede do dispositivo de ensaio e é coletado em sua parte inferior, na qual são posicionadas válvulas que conduzem o líquido para um recipiente calibrado (proveta), onde serão realizadas as medições. A medição também pode ser realizada com o auxílio de uma balança (massa). A Fig. 9 ilustra o esquema de funcionamento do dispositivo para tal ensaio.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

Investigação da distribuição do filme de resfriamento em um motor-foguete à propulsão líquida

Assim, para a análise dos resultados obtidos no presente estudo, são considerados os seguintes casos: Tanques

Sistema de controle Estrutura do banco de ensaios Estrutura de fixação do motor ao banco

Dispositivo de ensaio e sistema de alimentação

Experimento 1: Uma primeira configuração do dispositivo de ensaio é montada com um cilindro posicionado na camada 5, possibilitando a separação do líquido aspergido pelos injetores periféricos em 50%. Experimento 2: Para o segundo experimento, o cilindro empregado na condição anterior é substituído por uma aba cônica que permite a coleta do líquido, o qual escoa pela parede interna do dispositivo. Os valores obtidos para o Experimento 1, plotados em gráfico que representa o escoamento na parede da câmara de combustão, apresentam o comportamento mostrado na Fig. 11. 0

Figura 9. Esquema do ensaio (Silva, 2009).

São executados dois tipos de ensaios: um com a utilização de um cilindro que separa as camadas 4 e 5 para verificar a porção de líquido fornecida pelos injetores periféricos, sem a influência dos injetores centrais da camada 4; e outro com uma aba cônica que permite a captação exclusiva do líquido que escoa pela parede interna do dispositivo de ensaio. Em ambos os casos ocorre a separação do líquido em cavidades localizadas no perímetro da camada 5, conforme mostra a Fig. 10.

Penetração [mm]

Proveta 100

150

200

250 300

F30

Figura 11. Penetração do filme de resfriamento - Experimento 1 (Silva, 2009).

Para os ensaios que utilizam a aba cônica para captação exclusiva do líquido que escoa pela parede (Experimento 2), os resultados obtidos também estão apresentados em forma de vazões mássicas nas posições para coleta de líquido, apresentando o comportamento mostrado na Fig. 12.

Penetração [mm]

100 150 200 250 300

Figura 10. Cavidades da coleta de líquido (SILVA, 2009).

F30

Figura 12. Penetração do filme de resfriamento – 2 (SILVA, 2009).

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

177

Silva, L.A.

A Tabela 3 reúne os resultados do percentual de vazão mássica média no filme de resfriamento com relação à vazão mássica total da câmara de combustão, as vazões mássicas médias no filme de resfriamento e a penetração média do filme de resfriamento para os dois casos analisados neste estudo. Tabela 3. Resumo dos resultados.

Vazão Penetração Percentual da mássica filme do filme Casos vazão mássica resfriamento resfriamento [%] [g/s] [mm] 1 11,3 711 210 2 9,00 564 183 Fonte: Elaborado pelo autor.

O desenvolvimento do dispositivo de ensaio, para um motor que não requer desempenho elevado e de baixo empuxo, permitiu acumular conhecimento de projeto que será útil na concepção de dispositivos mais complexos, de maior desempenho e empuxo. Para motores de maior porte, a vazão mássica certamente excederá a capacidade do dispositivo, sendo necessária a construção de um dispositivo dedicado para cada motor. REFERÊNCIAS

CONCLUSÕES

A presente investigação apresenta uma análise preliminar do funcionamento de um motor em fase de concepção, com ênfase nos parâmetros que mais afetam a temperatura e a geração de calor. Inicialmente, tinha-se como objetivo que o motor em questão fosse o L75, porém devido a limitações das instalações de teste do IAE, optou-se por considerar uma versão limitada do L75. Esta versão estudada mantém as mesmas dimensões da câmara de combustão (que são idênticas às do já existente L15), porém, com o sistema de injeção diferente, o que implica numa degradação de seu desempenho, no que diz respeito ao empuxo gerado. A realização dos ensaios para a investigação do filme de resfriamento confirma que os valores limites recomendados pelos especialistas russos, com relação à vazão mássica na formação do filme de resfriamento, fornecem um valor de penetração suficiente para o comprimento da câmara de combustão do motor L75. Entretanto, os valores adquiridos nos ensaios demonstram que esses podem ser refinados para as aplicações específicas a que se destinam. O procedimento de ensaio se mostrou viável para aplicação no desenvolvimento de outros motores que possam surgir no âmbito do IAE. Porém, outros aspectos se mostraram

178

relevantes na qualidade dos dados analisados, como, por exemplo, a qualidade da manufatura dos componentes do sistema de mistura, principalmente dos injetores.

Agência Espacial Brasileira, 2005, “Programa Nacional de Atividades Espaciais: PNAE/Agência Espacial Brasileira”, Brasília, AEB. Sutton, G.P.A., 1986, “Rocket propulsion elements”, New York, John Wiley & Sons. Huzel, D.K.;Huang, D.H., 1992, “Modern Engineering for Design of Liquid-Propellant Rocket Engines”, Washington, AIAA, Process in Aeronautics and Astronautics, v. 147. National Aeronautics and Space Administration, 1972, “Liquid rocket engine fluid-cooled combustion chambers”, Ohio, NASA. Kessaev, K., 2006, “Theory and calculation of liquid propellant rocket engines in: Fundamental course in engine design”, São José dos Campos, CTA/IAE. Raznjevic, K., 1976, “Handbook of thermodynamic tables and charts”, Washington, Hemisphere Pub. Silva, L.A., 2009, “Investigação da Distribuição do Film Cooling em um Motor Foguete a Propelente Líquido de 75 kN de Empuxo”, São José dos Campos, ITA, dissertação de mestrado.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 171-178, May-Aug., 2011

doi: 10.5028/jatm.2011.03021911

Carlos Isidoro Braga

Instituto Tecnológico de Aeronáutica São José dos Campos/SP – Brazil braga@srgrupo.com.br

Mirabel Cerqueira Rezende

Instituto de Aeronáutica e Espaço São José dos Campos/SP – Brazil mirabelmcr@iae.cta.br

Michelle Leali Costa*

São Paulo State University Guaratinguetá/SP – Brazil michelle@fastline.com.br

*author for correspondence

Methodology for DSC calibration in high heating rates Abstract: Despite the large use of differential scanning calorimetry (DSC) technique in advanced polymer materials characterization, the new methodology called DSC in high heating rates was developed. The heating rate during conventional DSC experiments varying from 10 to 20ºC.min-1, sample mass from 10 to 15mg and standard aluminum sample pan weighting, approximately, 27mg. In order to contribute to a better comprehension of DSC behavior in different heating rates, this work correlates as high heating rate inﬂuences to the thermal events in DSC experiments. Samples of metallic standard (In, Pb, Sn and Zn) with masses varying from 0.570mg to 20.9mg were analyzed in multiples sample heating rate from 4 to 324°C. min-1. In order to make properly all those experiments, a precise and careful temperature and enthalpy calibrations were performed and deeply discussed. Thus, this work shows a DSC methodology able to generate good and reliable results on experiments under any researcher choice heating rates to characterize the advanced materials used, for example, for aerospace industry. Also it helps the DSC users to ﬁnd in their available instruments, already installed, a better and more accurate DSC test results, improving in just one shot the analysis sensitivity and resolution. Polypropylene melting and enthalpy thermal events are also studied using both the conventional DSC method and high heating rate method. Keywords: DSC, High heating rate, Calibration, Thermal analysis, Polymers.

LIST OF SYMBOLS

PRTs:

DSC:

Differential Scanning Calorimeter

ASTM:

Association Standards Testing Materials

In:

Indium Metal

Pb:

Lead Metal

Sn:

Tin Metal

Zn:

Zinc Metal

Characteristic Glass Transition DSC Curve Shape

Platinum Resistances Temperature, temperature direct proportional to the resistance

NATAS: North American Thermal Analysis Society x:

axis “x” in a particular graphic

axis “y” in a particular graphic

Tmax:

maximum peak temperature

T0:

ideal first order temperature transition

∆T:

difference between Tmax – T0

∆ Hm:

melting enthalpy

DSC system total thermal resistance

β:

heating rate applied to the sample

R :

DSC system total thermal conductivity

τlag:

time constant due to the thermal delay effect

Rsample:

The sample thermal resistance or the thermal resistance related to the sample

Tpeak:

peak temperature

Tonset:

onset temperature

sample mass

dQ/dt:

derivative of the heat as function of the time or heat flow

cp:

specific heat

factor to correct the sample mass as function of the heating rate

Dp:

standard deviation of a specific parameter

PP:

polypropylene sample

-1

Rinstrument: The thermal instrument

resistance

the

Rsample pan: The thermal resistance related to the sample pan tan α:

Alpha angle is proportional to the total system thermal resistance

Received: 16/05/11 Accepted: 04/07/11

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

179

Braga, C.I., Rezende M.C., Costa M.L.

INTRODUCTION

DSC is a thermal analysis technique that measures the difference in energy provided to a sample and a reference material in function of a controlled temperature programming. This technique keeps constant the heat supplied to the sample and reference. A control system (servo system) immediately increases the energy supplied to the sample or the reference, depending on if the process involved is endothermic or exothermic. Therefore, the equipment keeps the sample and the reference at the same temperature. The record of the DSC curve is expressed in terms of heat flow versus temperature or time (Vasconcelos, 2010). For the proposal methodology presented in this paper does not matter the DSC principle of operation, heat flow or power compensation. The thermal events which generate the DSC curves are mainly first and second order transitions (Canevarolo, 2004). Figure 1 shows the typical DSC curve and represents a standard metal melting, the indium. The melting point maximum peak split the curve in left side of the peak, called here as low temperature side and, right side of the peak, called here as high temperature side. The angle Îą related to the low temperature side is proportional to R-1, where R is the total thermal resistance of the DSC thermal system, and it is the addition of the sample resistance (Eq. 1), instrument resistance and sample pan resistance (Mathot, 1994; Poel and Mathot, 2006). Or, in another way, R-1 is the system total thermoconductivity. Rtotal = Rsample+ Rinstrument + Rsample pan

(1)

Usually, the worldwide suppliers for DSC purity data analysis software use the curve slope, where tan Îą â&#x2030;&#x2026; R-1 (vanâ&#x20AC;&#x2122;t Hoff equation) and it is obtained experimentally during the instrument calibration procedure. The â&#x20AC;&#x153;Râ&#x20AC;? value maybe modified during the furnace ageing, reaching unacceptable values, and this may affect the curve area which is the enthalpy related to the thermal event.

180

(mW)

heat flow endo

In recent years, the fast development in science and technology of materials has improved the production of new products for the aerospace industry. Among them the advance in polymeric composites are an example of recognized success. So, it is also necessary there are techniques to be able to characterize these new polymeric materials in the adequate way. The most commonly technique used in characterization of polymeric matrices is the thermal analyses, specifically, the differential scanning calorimetry (DSC). However, nowadays new generation of methodology in DSC technique is being used in the study of polymers, the DSC in high heating rates (Poe and Mathot, 2006).

low temperature side high temperature side

tan ÄŽ 5

> 7SHDN 7RQVHW 5@.e W 5&VDPSOH

ÄŽ start up hook Csample.heating rate

Temperature (ÂşC)

Figure 1. Theoretic considerations about a typical DSC curve (Braga, 2009).

The portion Rsample represents the sample thermal resistance and is heat capacity, sample mass and diffusivity, dependent. While, Rinstrument is the thermal resistance related to the instrument itself and is furnace geometry, furnace mass, furnace material, purge gas type and humidity, dependent. And, finally, Rsample-pan representes the thermal resistances sum between: (a) the sample pan and the equipment, which is sample pan contact area dependent, (b) the sample and sample pan, which is sample contact area dependent, and (c) the thermal resistance related to the sample pan itself, which is mass, material type and purity, dependent (Poe and Mathot, 2006). Figure 1 shows also, from the left to the right side, the DSC curve beginning, which can appears as descendent or ascendant, indicating how the thermal equilibrium process for both systems takes place: the reference system sample pan and the sample system sample pan. This thermal equilibrium process usually appears during the initial portion of the DSC curve and the reason is a non-appropriate mass balance between both systems, the sample and the reference. The thermal equilibrium above mentioned can also be denominated instrument thermal lag or start up hook. If the DSC equipment into the laboratory is a heat flow principle of operation, the instrument thermal inertia is caused by factors as: mass difference between sample and reference thermocouples, a non homogeneity of the heat distribution into the furnace, and a non homogeneity of the alloy utilized in the detector system. In a power compensation DSC, those factors are related to difference in mass between the platinum resistances

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

thermometer (PRTs), difference in mass between the sample and reference furnace, furnace electronic control response, difference in purity between the furnaces and related parts. This thermal equilibrium lag interferes in the initial temperature utilized in the experiment and will force changes in the sample analysis methodology, mainly in experiments utilizing high heat transfer rates. Figure 1 shows the DSC indium melting point, the right side of the curve (high temperature side) is system total thermal resistance “R” dependent and, also, sample specific heat capacity (C) dependent. For an uniform and perfect crystalline molecular structure, the specific heat is theoretically infinite, during the whole phase transformation process. Although, that is not applicable to semicrystalline polymers, which show the materials melting point, in the DSC curve, as a peak with certain width (related to the sample: purity, mass, heat capacity, diffusivity and heating rate). As narrower the DSC peak more uniform is the sample crystalline morphology. Above the melting point, the heat capacity increase or decrease smoothly as function of the temperature and those changes are no greater than 5% (Mathot, 1994). If took into consideration semicrystalline polymers, its thermal conductivity is crystallinity dependent and, obviously, the material thermal history is very important. This property is also affected by the sample anisotropy and, when anisotropy is present, it is recommended appropriate and additional measurements (Santos, 2005).

In the year 2000, in the 28° North American Thermal Analysis Society Congress (NATAS) (Pijpers et al., 2000), a work utilizing DSC under high heating rates was presented for the first time. This study showed how easy is carry out weak transitions experiments, increasing the instrument sensitivity, utilizing high heating rates. To carry out this kind of experiments, Pijpers et al. (Pijpers et al., 2000; Pijpers et al., 2004) suggested to use low mass and small dimensions furnaces. In 2002, the first publication in periodic about DSC happens, running into high heat transfer, heating or cooling, employed to polymers (Pijpers et al., 2004). In 2004, the academy showed the polymorphs in the pharmaceutical industry being solved by DSC under high heating rates (Gramaglia et al., 2005; Hurtta and Pitkänen, 2004; McGregor et al., 2004; Saunders et al., 2004). Thus, a conventional DSC analysis, employed to polymers, utilizing 10 and 20°C.min-1 as heating rates, nitrogen or air as purging gas, sample mass around 10mg, aluminum sample pan (~27mg) and temperature range from room temperature to 300°C, the experiment total time is around 28 minutes. However, this experiment can be carried out into high heating rate and take only 1.4min, increasing the number of DSC analysis by day (Poel and Mathot, 2006; Gill and Sauerbrunn, 1993; Pijpers et al., 2000; Pijpers et al., 2004).

The polymers thermal conductivity is very low when compared with metallic materials, or some ceramic materials. Taking into consideration the material processing, low thermal conductivity create some real problems: the polymer can be heated and processed in a lower speed and this reduce the production (Santos, 2005). During the cooling, the low conductivity can result in final products not uniform and shrinkable. It can result cooling stress, extruded deformation, delaminating, molded void etc. (Santos, 2005).

Considering to the use of DSC technique in high heat rates, as showed previously, the literature presents some little works in the pharmaceutical area (Gramaglia et al., 2005; Hurtta and Pitkänen, 2004; McGregor et al., 2004; Saunders et al., 2004) and in the characterization of polymer processing in real time (Poel and Mathot, 2006; Gill and Sauerbrunn, 1993; Pijpers et al., 2000; Pijpers et al., 2004; Pijpers et al., 2002). These studies, despite of being rare, show the good potential of this technique in the study of material thermal behavior. Thus, the present work aims to contribute to nationalization and enlargement of the DSC technique in high heating rates use. For this, the present study shows in details the basic principle and the development of this methodology, useable in any DSC equipment and any material, utilizing as proof of concept, the polypropylene polymer characterization.

According to Illers (1974) the heating rate is considered as conventional up to 36°C.min-1, and heating rates higher than 36°C.min-1 will be considered high heating rates for DSC experiments or hyper-DSC. Higher heating rates do not mean new DSC equipment, it is a new operation mode for DSC utilizing a proper methodology capable to make possible high heat transfer, cooling or heating the sample utilizing the current equipments already used into the laboratories.

For a better understanding of the instrument limitations, metallic standards, as indium, tin, lead and zinc, were used for calibration purpose. The DSC furnace, linearity and symmetry are also studied. Besides the point, indium masses from 0.570mg to 20.9mg were submitted to different heating rates from 4°C.min-1 to 324°C.min-1. These experiments permit comparing the melting point temperatures and enthalpies values along with those available in the literature.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

181

Braga, C.I., Rezende M.C., Costa M.L.

EXPERIMENTAL

the onset melting temperature was obtained, and the peak temperature and the enthalpy associated to each fusion were determined.

Materials The materials used, in the first step of this work, are metallic traceable Standards, indium 99.99% pure, tin 99.96% pure, lead 99.98% pure and zinc 99.99% pure. These standards were utilized to perform and study the DSC furnace linearity and symmetry, as well a better DSC understanding, running into high heat transfer. Also, a polypropylene sample was experimented, kindly offered by Polibrasil Resina S/A company. Equipment The instrument utilized in this study is a Perkin Elmer model Pyris 1 connected to a cooling system model Intracooler 2P. It is a power compensation DSC along with low mass furnace, around 1g, which minimize the thermal lag effect.

Trying to evaluate the thermal resistance decrease, between the DSC furnace and the sample, experiments were carried out utilizing aluminum foil sample pan with 15µm in thickness. The indium metal was used to calibrate the instrument and also used as sample. The calibrations were made in two different heating rates (10 and 100°C.min-1). Multiple heat transfer was tested utilizing the same 12 heating rates performed before. Both, calibration and experiments were carried out with 10.2mg sample mass. After the instrument calibration being performed and the furnace linearity being verified, polypropylene experiments were carried out along with 1.00mg sample mass and a set of heating rates of 10, 50, 100, 200, 300, 400, 500 and 600°C.min-1. RESULTS AND DISCUSSION

Calibration

Experiments utilizing metallic standards

Initially an usual calibration was carried out, utilizing indium standard, following the instrument supplier instructions, ASTM 967 (2008) and ASTM 968 (2008), suggesting to calibrate the temperature and the enthalpy utilizing metallic standards.

A proper temperature calibration for high heating rates experiments consists of performing a conventional DSC calibration in an extended temperature range, utilizing primary metallic standards, which have a precise and clear thermal transition in the temperature range of interest. After this first step calibration, a matrix of calibration has to be filled, experimenting the previously defined standard into different masses and different heating rates, simulating the standard to be a sample, as suggest the literature (Poe and Mathot, 2006).

Experiments After perform the calibration utilizing indium standard, carried out under the ASTM-967 (2008) and ASTM968 (2008) conditions, different metallic standards (Zn, In, Pb and Sn) were weighted, approximately, with the same 10mg mass, and tested as they were sample. Those samples were placed, separately, into conventional aluminum sample pan, 27mg mass, and the purpose was to verify the instrument calibration made earlier, also, study the DSC furnace linearity and symmetry. These analyses are performed in the same experimental conditions, sample mass 10mg, heating rate 10°C.min-1, identical test methodology and the same nitrogen purge gas flowing. In order to study the influence of mass in the DSC curves obtained in high heating rates, nine different masses of the same sample (indium) were selected, ((20.9; 15.8; 12.7; 9.10; 6.40; 4.03; 1.70; 1.00 and 0.570) mg). These samples were submitted to twelve different heating rates ((4; 9; 16; 25; 36; 64; 100; 144; 196; 225; 256 e 324)°C·min-1). From each DSC curve 182

Once carried out the calibration in the choose heating rates, the DSC furnace linearity and symmetry need to be evaluated. When the DSC furnace presents good symmetry and linearity, the same temperature calibration may be used either for the heating rate or for the cooling rate. If the DSC furnace response is not symmetric in terms of temperature another calibration in the cooling mode is unavoidable (Poe and Mathot, 2006). These long terms calibration procedure seems to be very slow if the user want to make conventional DSC experiments, but it is mandatory if carring out high heat transfer analysis is the user’s choice. Figure 2 exhibits the onset melting temperature, obtained experimentally in this work, for the metallic standards (In, Sn, Pb and Zn measured as sample) as function of the onset melting temperature observed in the literature (Canevarolo, 2004). Even so, the DSC instrument was calibrated with just one metallic

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

Experimental onset temperature (ºC)

For DSC equipment used in present work, Fig. 2 also exhibits a good Pearson correlation coefficient (0.99), indicating that the carried out calibration with just one point of temperature standard is enough to guarantee a proper instrument operation, in a large temperature range. In the case the experimental points do not match with the media straight line, but if the Pearson linear correlation coefficient is good, between 0.98 and 1.00, it means the furnace has a linear behavior.

450 400 350

Y = a + bx a = 0.814 b = 0.99 r = 1.00

Zinc Lead

300 250

Tin

200 Indium

150 150

200 250 300 350 400 Literature onset temperature (ºC)

450

Figure 2. Experimental melting temperature as function of literature melting temperature using the standards In, Sn, Pb and Zn.

In spite of these good results, it is recommended, according to the literature suggestion (Poe and Mathot, 2006), to utilize at least three temperature standard calibrations, in the range of interest, with the purpose to get reliable experimental results. Those extended calibration procedures were carried out only in the beginning of this work, for a better knowing of the instrument response in an extensive temperature range. A fast way to verify and validate or not the carried out calibration is to make a DSC run with indium metal considering it as a sample. The onset melting temperature indicates if the instrument, already calibrated, is proper to initialize the experiments. Figure 3 exhibits the enthalpy of fusion calibration, in which the Pearson linear correlation coefficient is 0.98, the

enthalpy values for the DSC analysis is linear, suggesting only one standard material for calibration is enough in a extended temperature range (in this work). In the same way, utilization of at least three temperature standard calibrations, in the range of interest, is recommended, according to the literature suggestion (Poe and Mathot, 2006), with the purpose to get a reliable experimental result. So, after the experimental procedures, the equipment is considered calibrated, in temperature (“x” axis) and in enthalpy (“y” axis).

values of experimental melting enthalpy (J.g-1)

standard, the indium. The experimental values obtained for the onset melting point temperatures, considering all metallic standard studied, are very similar for those respective literature values (Canevarolo, 2004). These results indicate the DSC furnace utilized in this study and presents linearity and symmetry, in the studied temperature range.

120 100

y = a + bx a = 0.22 b = 0.94 r = 0.99 sd = 0.002

60 40

In 20

100

120

values of literature melting enthalpy (J.g-1)

Figure 3. Values of experimental melting enthalpy as function of the values of literature melting enthalpy for the standards In, Sn, Pb and Zn

Tables 1 and 2 exhibit the onset melting temperature and enthalpy of fusion, respectively, obtained from several analyses, several indium masses experimented in different heating rates and the equipment was already calibrated, previously, utilizing 9.1mg of indium mass at 9°C.min-1. All the thermal analyses tests were conducted according to ASTM 3418 (2008), which mention to start the experiment 50°C below the thermal transition studied and the final temperature 30°C over the studied thermal transition. In the tables, the bold values indicate which the instrument was calibrated. It is very easy to verify in Table 1 the theoretical and classic thermal analysis behavior (Canevarolo, 2004; Poe and Mathot, 2006), when the same amount of sample (specific mass) is tested in higher heating rate the transition temperature shifts to higher temperatures when compared with literature values. It happens due to the thermal lag or thermal inertial effect, in which as the higher the heating rate the slower the instrument and sample response (Canevarolo, 2004; Poe and Mathot,

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

183

Braga, C.I., Rezende M.C., Costa M.L.

Table 1 - Experimental values for onset melting temperature using indium sample into the conventional aluminum sample pan, different masses and different heating rates. The calibration is shown in the table as bold letters

β(°C.min-1) 20.90mg 15.88mg 12.73mg 9.10 mg 6.40 mg 4.03 mg 1.70 mg 1.00 mg 0.57 mg

4.00 156.24 155.99 155.80 156.87 156.18 155.79 155.86 155.94 155.68

9.00 157.03 156.27 156.02 156.54 156.40 155.96 155.98 156.06 155.75

16.0 157.03 156.73 156.36 156.93 156.71 156.15 156.10 156.21 155.83

25.0 157.70 157.29 156.79 157.48 157.07 156.37 156.21 156.35 155.90

36.0 158.39 157.83 157.28 158.13 157.46 156.59 156.33 156.49 155.99

64.0 159.55 158.72 158.16 159.71 158.28 157.04 156.56 156.84 156.15

100 160.65 159.44 158.88 161.45 159.19 157.49 156.80 157.26 156.38

144 161.68 160.12 159.49 163.35 160.18 157.95 157.04 157.57 156.60

196 162.78 160.73 160.00 165.55 161.18 158.43 157.31 158.01 156.76

225 163.37 161.06 160.23 166.67 161.84 158.69 157.60 158.23 156.91

256 163.90 161.35 160.42 167.66 162.50 158.93 157.70 158.34 157.17

324 164.67 161.83 160.79 169.42 163.78 159.50 158.00 158.62 157.42

Table 2 - Experimental values for enthalpy of fusion using indium sample into the conventional aluminum sample pan, different masses and different heating rates. The calibration is shown in the table as bold letters

β(°C.min-1) 20.90mg 15.88mg 12.73mg 9.10 mg 6.40 mg 4.03 mg 1.70 mg 1.00 mg 0.57 mg

4.00 29.51 29.66 29.84 28.68 28.56 29.06 25.19 27.54 27.72

9.00 29.65 29.74 29.96 28.71 29.12 29.09 25.10 27.41 27.40

16.0 29.65 29.89 30.09 28.80 29.22 29.24 25.21 27.57 27.33

25.0 29.78 30.10 30.27 28.89 29.30 29.29 25.19 27.62 26.88

36.0 30.09 30.31 30.46 29.04 29.44 29.42 25.24 27.73 27.05

2006). Also an increase in the enthalpy value is verified when the same sample is submitted to higher heating rates (keeping the same material and same mass) (Table 2). Being both thermal events, onset melting temperature and enthalpy of fusion, essentially thermodynamic events, they should not change, neither with mass changes nor with heating rates. And so we can conclude that the variations presented in Tables 1 and 2 suggest a possible problem with the instrument or with the sample. In respect to the first suspect, the instrument is operating under the supplier specification and its response is linear, according to the exhausted instrument calibration carried out. Supported by Illers (1974), Poel and Mathot (2006) and Neuenfeld and Schick (2006) suggestions in the literature, the shift effect in the material melting point temperature, to higher temperature values, is caused by the time the heat takes to diffuse into the sample homogeneously (thermal diffusivity and conductivity). Even so, the sample is not an ideal material, and most of the time it is not 100% pure, and also, the possible presence of polymorphism in the sample can contribute to this behavior. In spite of it all, the DSC

184

64.0 30.59 30.71 30.86 29.33 29.86 29.56 25.37 27.93 27.11

100 31.05 31.34 31.49 29.48 29.57 30.32 26.52 27.46 26.44

144 31.71 31.75 31.96 30.08 30.63 30.01 26.08 27.99 28.19

196 32.13 32.25 32.59 30.77 30.46 31.00 26.68 28.30 28.86

225 32.22 32.55 32.85 30.78 31.05 31.06 26.19 28.65 30.96

256 32.74 32.69 33.25 31.22 31.12 31.47 25.30 29.88 27.51

324 33.51 33.30 33.91 32.13 31.73 33.19 25.38 32.51 28.33

temperature sensor which is located in direct contact with the sample pan and can have its reading affected by the DSC sample pan, which has three times more mass than the DSC sensor. In that situation, the sample pan acts like a thermal resistance between the furnace sensor and the sample, and, an expected delay in the heat transfer takes place. Among the above-mentioned situations, the sample mass increase helps the thermal inertial factor, which shift the sample onset melting temperature to values greater than those exhibited in the literature. According to Illers (1974), Poel and Mathot (2006) and Neuenfeld and Schick (2006), the maximum peak temperature concerning the first order transition, like the melting point temperature, is shifted to higher temperatures with the heating rate increase, due to the sample latent heat be added to the process of fusion, which happen thermodynamically at a constant temperature. Illers (1974) suggests the Eq. 2 to explain these effects. ¨T = Tm ax - T0 = 2.¨H m .R.ȕ + Ylag ȕ

(2)

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

It is very important to mention the thermal inertial effect observed in the thermodynamic experimental measurements, is huge in polymer materials, in which the thermal conductivity and diffusivity are very low, when compared along with metallic materials (Illers, 1974). According to the ASTM D3418 (2008), the onset melting temperature values for reproducibility should be within ±4.2°C and ±7.3% for the enthalpy. The data obtained during the experiments, and exhibited in Tables 1 and 2, have different experimental values (changing the mass and the heating rate). Considering the differences cited in the ASTM D3418, these both onset melting point temperature and enthalpy of fusion are partially reproducible and may be accepted. Keeping the same idea, based on Table 1, the onset melting point temperature values are satisfactory for the entire list of studied masses up to 100°C.min-1 heating rate. Evaluating the enthalpy values, they also attend the permitted error cited in ASTM up to 64°C.min-1 heating rate. Thus, considering a conservative criterion for choice, the studies by DSC may be carried out with the conventional DSC calibration (usually performed at 10°C.min-1 heating rate and 10mg sample mass) up to 64°C.min-1 and the sample mass must to be between 0.50 and 20mg. Two sample masses were chosen as representing the studies in this work, 9.10mg sample mass (Fig. 4) and 1.0mg sample mass (Fig. 5), both are indium samples. Figures 4 and 5 exhibit the curves for the indium sample (heat flow as function of temperature) in different heating rates using conventional aluminum sample pan. It can be verified that the heating rate increment shifts the thermal event peak temperature to higher values and the peak becomes higher and wider. These effects are caused by sample mass, sample diffusivity and sample purity. Figure 4 shows the left side of the indium melting point peaks (low temperature side) in which the left sides are parallel between each other, in other words, the α angle does not change, it is kept basically equal for different heating rates studied. In a similar way, in the literature is found α angle not modified up to 36°C. min-1 (Mathot, 1994; Poe and Mathot, 2006). In this

case, the sample thermal resistance presents a greater contribution to the system total thermal resistance (R), while the instrument thermal resistance and the sample pan thermal resistance has an insignificant contribution to the system thermal resistance, and this information matchs with the literature (Mathot, 1994).

120 9.1 mg 324ºC/min 110 256ºC/min 100 225ºC/min 90 196ºC/min 144ºC/min 80 100ºC/min 70 64ºC/min 60 36ºC/min 25ºC/min 50 16ºC/min 40 9ºC/min 4ºC/min 30 20 10 140 150 160 170 180 190 200 210 220 230 240 Temperature (ºC)

heat flow undo (mW)

where: Tmax is the maximum peak temperature; T0 is the ideal first order transition, real melting point temperature; ∆Hm is the transformation enthalpy; R is the system thermal resistance; τlag is the time constant due to the thermal delay effect, which depends on the heat capacity, the thermal resistance and the thermal system; where β is the heating rate.

Figure 4. DSC curves showing 9.1mg of indium sample experimented at different heating rates and basically the (α) alpha angle is the same

Confronting Figs. 4 and 5, the peaks in the left sides (low temperature sides) show different behaviors, in both figures. The peak slope, to the low temperature side (α angle), related to the lower sample mass in the test (1.00mg), presents greater heating rate dependence than the larger sample mass in the test (9.10mg). The purity data analysis (purity software for DSC experiments) utilizes the van’t Hoff equation and considers in its algorithm the α angle (Mathot, 1994). So, the purity studies carried out by DSC equipments have to take a special attention to chose each test sample mass (1 to 3mg) as recommended by ASTM E928 (2008). This procedure is very important to obtain reliable and repetitive results. Smaller sample mass increases the probability to get purity results non-reproducible and different from the initial expectations (Mathot, 1994). Thus, if the sample mass is smaller than a specific value, the system total thermal resistance will be lightly sample mass dependent (Rinstrument + Rsample-pan), in other words, the sample purity value is going to be heating rate dependent, and, this is something undesirable. As the thermal resistance is an instrument intrinsic value and, this value can change to each instrument, so, all the procedures

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

185

Braga, C.I., Rezende M.C., Costa M.L.

utilized in this work shall be verified each time a new equipment, model or brand. Figures 4 and 5 show, clearly, increasing the heating rate in the DSC experiments, when the sample mass is around 10mg, the onset melting temperature shifts to higher values. However, decreasing the sample mass for values around 1mg, very low shift in the onset melting temperature is promoted (as shown in Table 1). Lower sample mass smaller onset melting temperature shift is observed. The diffusivity and thermal conductivity are the main responsible for the thermal delay, and, into this low mass experimental conditions, the sample response is very close to the thermodynamic theory (Neuenfeld and Schick, 2006; Turi, 1981).

25 24 16ºC/min 9ºC/min 23 4ºC/min 22 21 20 19

1.00 mg 324ºC/min 256ºC/min 225ºC/min 196ºC/min 144ºC/min 100ºC/min 64ºC/min 25ºC/min

18 17 140 150 160 170 180 190 200 210 220 Temperature (ºC) Figure 5. DSC curves showing 1.00mg of indium sample experimented at different heating rates and a great difference between the (α) alpha angles

In an experiment, in which the same sample type and quantity is submitted to different heating rates, and the DSC curve shift only in the “x” axis (this means, time or temperature), this behavior means the sample thermal resistance has greather impact in the thermal system response, or, the thermal resistance external to the sample has a minimum contribution to the thermal system response. If the opposite case happens, it means, the DSC curve shifts only in the “y” axis, heat flow, so, the thermal resistance external to the sample has the major impact in the thermal system response and the sample thermal resistance is meaningless. The relationship (Tpeak - Tonset) shows how DSC curve depends on the heating rate and the sample mass. Considering that the DSC curve shape affects the total area of the thermal transition, so, the shape of the

186

Figure 6 shows a linear relationship between the sample mass and the heating rate up 36 °C min-1 does exist. And, after this value, a more complicate relationship between these results takes place. So, as cited in the literature (Illers, 1974; Neuenfeld and Schick, 2006; Poe and Mathot, 2006), the heating rate is considered conventional up to 36°C.min-1. Above this value this parameter is assumed to be high, and the curve shows to be strongly sample mass dependent. For small samples (<1mg) the ∆T increase is directly proportional to the heating rate square mean root (Poe and Mathot, 2006) and the heat flow amplitude is higher compared with the linear response. 11 10 9 8 7 6 5 4 3 2 1 0 -1

20.90 mg 15.88 mg 12.73 mg 9.10 mg 6.40 mg 4.03 mg 1.70 mg 0.57 mg 1.00 mg

ǻ7 7peak 70 7 &

heat flow endo up (mW)

27 26

indium melting curve in the high temperature side is important to calculate the DSC curve area, and the DSC curve area is proportional to the enthalpy of fusion in the DSC experiments.

8 10 12 14 16 18 20 ȕ 1/2 & PLQ 1/2

Figure 6. Indium peak temperature minus onset melting temperature as function of heating rate (β) square root

Figure 7 shows the DSC curve for a 9.10mg indium sample, studied in three different heating rates (4, 36 and 196) °C.min-1 as function of time. For the same type of sample, three distinct duration times of the tests, for different heating rates, are observed. The indium melting temperature occurs in less than 1min when the heating rate is set to 196°C.min-1. Also, the melting point peak becomes narrower, heighter and proportional to the heating rate increase. However, the peak area, related to the energy necessary to melt the indium metal, is the same for any heating rate used (Canevarolo, 2004; Ionashiro and Giolito, 1980; Wendlandt, 1986; Wunderlich, 1990). Equation 3, a simple derivative heat flow equation, exhibits the dependence of heat flow on the mass, the sample heat capacity and the heating rate. This equation shows also the

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

instrument sensitivity is increased proportionally to the heating rate increase. This affirmative is confirmed, experimentally, in Figure 7. On the other hand, the heating rate decreasing improves the instrument resolution and decreases the instrument sensitivity (Kasap, 1997; Chagas, 1999). dQ = m.c p .` dt

(3)

where: m is the sample mass, cp is the sample specific heat, β is the heating rate and dQ/dt é the derivative of heat as function of time which is the heat flow.

100

heat flow end up (mW)

9.10 mg

196 ºC/min

dQ/dt = m x Cp x heating rate

70 60

36 ºC/min

50 4°C/min

40 30 20 0

6 8 time(min)

Figure 7:Heating rate affecting the instrument resolution and sensitivity, from the equation dQ/dt = m.cp.β

According to Pijpers et al. (2002), in order to minimize the thermal gradient inside the sample, when the sample heating rate be increased of a “z” factor, the sample mass should be reduced by the same factor “z” (Pijpers, 2002; Pijpers, 2004).

Even so, the sample pan mass, used in the experiment is fundamental to be considered. So, the aluminum foil sample pan has been considered in this experiment (~5mg) instead of conventional aluminum sample pan (~27mg). Thus, the aluminum foil sample pan will promote a much better heat transfer in the system: furnace, sample pan and sample. Table 3 exhibits the experimental results using 10.2mg indium sample mass in an aluminum foil sample pan which was submitted to different heating rates. The DSC instrument was calibrated using 10mg of indium at 10°C.min-1 and, both, calibration and experiments were executed using aluminum foil sample pan. The experiments were all done under ASTM E 3418 (2008). Although the conventional aluminum sample pans were replaced by aluminum foil sample pan, the obtained results kept the same tendency, shifting the DSC thermal events to higher temperature values when compared to the available literature. The same tendency happens to the melting enthalpy values. Table 4 exhibits the experimental values for enthalpy of melting, peak temperature and onset temperature for melting, for a 10.2mg-indium sample, heated at different heating rates, with the instrument calibrated at 100°C.min-1 and either the calibration and the experiments were performed in aluminum foil sample pans. The results obtained, either the measured onset melting temperature or the measured enthalpy of melting, are inside the error permitted by the ASTM 3418 (2008). This behavior suggests that experiments performed in high heating rates and using aluminum sample pan improve the results. Another approach is related to sample pans, which can be made of different material types (copper, aluminum, stainless steel, platinum, gold, glass and others) and can change the sample pan thermal conductivity, the heat capacity or the thermal diffusivity during the experiments, which modify the heat transmission in the sample direction and, consequently, the final DSC curve shape.

Table 3 - Experimental values for 10.2mg indium sample and the calibration was carried out at 10°C.min-1, both, calibration and experiments carried out into aluminum foil sample pan

9.00 16.0 25.0 36.0 64.0 100 144 196 225 256 324 β(°C.min-1) 4.00 Tonset (°C) 156.31 156.55 156.88 157.25 157.83 158.44 159.33 160.25 160.88 161.28 161.60 162.19 Tpeak (°C) 157.99 159.19 158.95 159.79 160.73 163.19 164.77 166.44 168.37 168.85 169.92 171.53 -1 ∆H (J.g ) 28.19 28.43 28.55 28.70 28.73 28.96 29.29 29.48 29.96 30.00 29.99 30.71 Table 4 - Experimental values for 10.2mg indium sample and the calibration was carried out at 100°C.min-1, both, calibration and experiments carried out into aluminum foil sample pan

9.00 16.0 25.0 36.0 64.0 100 144 196 225 256 324 β(°C.min-1) 4.00 Tonset (°C) 153.51 153.71 153.31 153.72 154.20 155.15 156.71 157.59 158.41 158.55 158.79 159.42 Tpeak (°C) 154.86 155.52 155.85 156.64 157.54 159.08 161.92 163.92 165.93 164.51 165.37 166.52 ∆H (J.g-1) 27.47 27.52 27.70 27.76 27.87 28.24 28.39 28.53 28.88 28.41 28.89 29.20 J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

187

Braga, C.I., Rezende M.C., Costa M.L.

Besides this, using the aluminum foil sample pan in the experiments, the temperature thermal lag has decrease and, as consequence, the onset melting temperature standard deviation (Dp). Comparing the onset melting temperature presented in Table 2 in the line of 9.10mg with the values presented in Table 3, the standard deviation, Dp, changed from 4.51 to 2.04 and to 2.27 for the values presented in Table 4. Analogous to the studies already mentioned and carried out for the onset melting temperature, the standard deviation of 1.13 calculated for the melting enthalpy, presented in Table 2, line of 9.10mg, goes down to 0.75 (Table 3) and to 0.55 according to Table 4. These decreases observed in the standard deviation values show experimental results nearer to the literature results, either for onset melting temperature or for enthalpy of melting, as both are thermodynamic values and should not be heating rate dependent. According to Table 4, the onset melting temperature experimental values are inside the permitted error, in accordance with ASTM D3418 (2008), up to 144°C. min-1 heating rate. On the other hand, the melting peak temperature values are inside the ASTM D3418 (2008) tolerance limit up to 36°C.min-1. And, finally, the values of enthalpies of melting are inside the permitted error for the heating rates studied in the present work. The heat generated by the DSC furnace is transmitted to the sample environment, reachs the sample pan, the interface area between the sample and the sample pan, and finally, propagates across the sample. This propagation of energy does not occur instantaneously, it takes a time to reach the entire sample and, consequently, the sample thermal equilibrium. In the DSC instruments, usually, the heat transfer occurs mainly by thermal conduction, approximately, from -150 to 600oC, and, in higher temperatures, the thermal radiation process takes place and becomes the main source of heat in the DSC thermal system. From above this temperature in which the type of heat transfer changes, it is also important to select a proper purge gas before start the experiment, taking into consideration the gas thermal conductivity. As the smaller the sample, easier to reach the thermal equilibrium across the sample in a shorter time. This thermal equilibrium is dependent of sample characteristics and this phenomenon is known as thermal lag or thermal inertia, caused by the sample thermal diffusivity process. As bigger the ratio between the sample pan mass and the sample mass utilized in a specific experiment,

188

larger is the shifting effect in the thermal events (onset melting temperature and peak temperature) due to the sample pan thermal diffusivity effect. These statements drive to three different situations which can happen during the DSC experiment: sample mass loss during the heating process, sample mass gain in oxidative atmosphere or the sample mass stays constant during the experiment. In each case the ratio between the sample pan mass and sample mass has different values (Braga, 2009). Also this ratio affects directly the instrument resolution. The shift in both onset melting temperature and peak temperature during the DSC experiments is caused by the sum of diffusivities: sample diffusivity, furnace diffusivity and sample pan diffusivity. The DSC temperature calibration purpose is to minimize these thermal diffusivity effects. For a better understanding, an analogy is propose: considering the heat as a fluid flowing to the sample direction, the DSC sample pan acts like a “screen” avoiding the heat to reach the sample; the sample pan absorbs heat before sample. As the higher the sample pan heat capacity (copper, aluminum, stainless steel, platinum, gold, silver etc.), smaller is “screen”, becoming more difficult the heat to reach the sample. The opposite is true (Braga, 2009). Thus, if the “screen” size is dependent sample pan temperature, so, the quantity of heat which reaches the sample varies, which suggests it can lightly change the heating rate established, by the DSC operator, in his working plan. Thus, variations in the DSC curve shape can be observed consequently in the curve’s area, which is numerically equal to the enthalpy of melting. This condition explains why the melting enthalpies values vary during experiments with high heating rates or high heat transfer (Table 2 to 4) (Braga, 2009). Polypropylene experiments After the detailed study utilizing several metallic standards for a better DSC instrument understanding, its limitations and responses when running into different heating rates, experiments in a polymeric sample, the polypropylene (PP), which was submitted to high heat transfer trying to simulate industrial processing, as extrusion or injection, were carried out. This polymer has been widely studied in the literature and there are many information about its molecular structure, processing, crystallinity, fusion and morphology (Braga, 2009; Canevarolo, 2004).

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

Figure 8 exhibits the DSC curve which is related the second PP melting using 1mg sample mass and, both, heating and cooling rate at 10°C.min-1 and aluminum foil sample pan. Table 5 exhibits the melting and crystallization temperatures, as well the respective enthalpies of melting are in Figure 8. Thus, the melting temperature for this PP sample is 142°C and its enthalpy is 45J.g-1, suggesting low crystallinity for this polymer, when compared with the literature values (Canevarolo, 2002). The supercooling degree, which is expressed as being the difference between the onset melting temperature and crystallization temperature, is 29°C. The small difference between the melting and crystallization peak areas (15%) is following the literature, as good as 20% for polymeric material (Canevarolo, 2002) Figure 9 exhibits the DSC crystalline melting curves, at the second PP heating, obtained at different heating rates, and 1mg sample mass inside of an aluminum foil sample pan. The instrument calibration was performed for each heating rate studied. All the samples were, at the begging, heated and cooled at 10°C.min-1. Table 6 presents the melting temperature and enthalpy for each heating rate tested. It is observed, as expected, the onset melting temperature varies between 121 to 140°C. The melting peak temperature shifts very lightly between 145 to 148°C and the enthalpies of fusion obtained are inside of an acceptable range of 33%. The error founded in the enthalpy values, may be explained based in the low thermal conductivity values imputed for polymers materials 0.03 to 0.61 (W.m1 .°C-1) (Chagas, 1999; Halliday and Resnick, 1992). This

1.2 heat flow (mW) endo up

ASTM D3418 (2008), ASTM E793 (2006) and ASTM E794 (2006) describe the procedures to determine the melting temperature, melting enthalpy and enthalpy of crystallization for polymers. As suggest the ASTM D3418 (2008), the polymer should be submitted to a first heating at a higher heating rate, to erase the sample thermal history, followed by a lower cooling rate to allow the polymer crystal to be organized, following by another heat in which the polymer melting temperature is registered.

attribute impute a limit in the speed which the polymer may be heated or cooled. The results presented suggest the 1mg PP mass follows the heating rate up to 100°C. min-1, in a satisfactory way, since the heating rate affects the enthalpies of fusion values very lightly. Thus, the results obtained show the 1mg sample mass utilized

1.0 0,8 0,6 0,4 0,2 0,0 80

100

120

140

160

180

Temperature (ºC) Figure 8. Polypropylene DSC curve in the second heating, experimenting 1mg sample mass and 10 °C.min-1 heating rate Table 5 - Values of polypropylene thermal behavior, studied at 10°C.min‑1

Event First heating Second heating Cooling

heat flow endo up (mW)

The polypropylene melting temperature which is associate with the material crystalline portion is in the range of 112 to 208°C (the more frequent value is 160°C) and the melting enthalpy in the range of 40 and 209 J.g-1 depends on the crystallinity, that can change from 40 to 70% (Canevarolo, 2004; Wellen and Rabello, 2005; Mothé and Azevedo, 2002; ASTM E 793, 2006; ASTM E 794, 2006).

Tonset (°C)

Tpeak (°C)

∆H (J.g-1)

140 133 104

146 142 101

42 45 -53

80 60

600 ºC.min-1 500 ºC.min-1 400 ºC.min-1 300 ºC.min-1 200 ºC.min-1 100 ºC.min-1 50 ºC.min-1 10 ºC.min-1

40 20 0 80

100 120 140 160 180 Temperature (ºC)

200

220

Figure 9. Polypropylene DSC melting curves under different heating rates

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

189

Braga, C.I., Rezende M.C., Costa M.L.

during the experiments for different heating rates are not appropriate for the tests performed in heating rates greater than 100°C.min-1, as the enthalpies of fusion values do not match each other. Table 6 - Onset melting temperatures, peak temperatures, and melting enthalpies, obtained in 1mg of polypropylene sample into an aluminum foil sample pan. Samples were submitted to different heating rates

Onset β(C.min‑1) Temperature (°C) 10 140 50 139 100 137 200 136 300 121 400 115 500 108 600 92

Peak Temperature (°C) 146 145 145 146 148 147 148 149

ΔH (J.g‑1) 42 41 40 30 31 72 93 115

Supported in the literature results, it is known that the sample mass reducing is inversely proportional to the heating rate increase (Poe and Mathot, 2006; Poe and Mathot, 2007) which can minimize the thermal inertial effect inside the sample, helping to obtain good enthalpy results. This study was not carried out in this work, due to the laboratory where the experiments were executed do not have an analytical balance limitation to measure masses lower than 1mg. However, it can be confirmed that DSC in high heating rates mode guarantees a better accuracy to characterize polymers, specially to measure the peak temperatures of the crystalline fusion as this value is lightly affected by the heating rate imposed to the thermal system, applying 1mg sample mass. It is mandatory to be mentioned that the DSC instruments in high heating rate mode do not replace the conventional DSC mode, but the high heating mode appears as an additional tool to amplify the researchers capability to investigate a sample, in low or high heating rates, utilizing the same instrument already installed in the laboratory. CONCLUSIONS This work presents successfully a DSC methodology in high heating rates (up to 100oC.min-1) to reproduce temperature of fusion and enthalpy of fusion, with better sensitivity, better resolution and precision in a short period of time, following the ASTM specifications. This condition amplifies the DSC investigation capability using the

190

same instrument and it permits that the better and deep characterization of advanced material used in aerospace field be done. As expected by the conventional thermal analysis theory, for the same sample mass if the heating rate increase occurs the shift of the transition temperature for higher values, caused by the thermal inertial effect. Similar behavior is observed for the enthalpy of fusion values. The indium sample study, utilizing different masses and different heating rates, shows that conventional analyses are considered up to 36 °C.min-1. Above this heating rate is observed the mass dependence. A greater dependence with the heating rate, proportional to the total system thermal resistance, is also observed for lower mass values. Same sample mass and different heating rates promote curve shifts only in the “x” axis (time or temperature), which means the sample thermal resistance affect predominantly the thermal system response. To minimize the thermal gradient inside the sample, if the heating rate is increased by a “z” factor, the sample mass should be reduced by the same factor “z”. The use of aluminum foil sample pan (~5mg) is recommended instead of conventional aluminum sample pan (~27mg), to promote a better heat transfer among furnace, sample pan and sample. The high heating rate mode in the DSC may be utilized to characterize advanced polymer materials, preferentially to find the melting peak temperature and crystallization temperature, but these measurements depend greatly on the polymers thermal properties. acknowledgment The authors express gratitude to the National Council for Scientific and Technological Development (CNPq) by the financial support processes 305478/2009-5 and 152384/2007-3. References ASTM D3418-08, Association Standards Testing Materials, 2008, ASTM D3418-08: “Test method for transition temperatures of polymers by differential scanning calorimetry”, Philadelphia, USA. ASTM E793, Association Standards Testing Materials, 2006, ASTM E793: “Test method for enthalpies of fusion and crystallization by differential scanning calorimetry”, Philadelphia, USA. ASTM E794, Association Standards Testing Materials, 2006, ASTM E794: “Test method for melting and crystallization temperatures by thermal analysis”, Philadelphia, USA. ASTM E928, Association Standards Testing Materials, 2008, ASTM E928: “Test method for purity by differential scanning calorimetry”, Philadelphia, USA.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

Methodology for DSC calibration in high heating rates

ASTM E967, Association Standards Testing Materials, 2008, ASTM E967: “Standard practice for temperature calibration of differential scanning calorimetry and differential thermal analyzers” ,Philadelphia, USA. ASTM E968, Association Standards Testing Materials, 2008, ASTM E968: “Standard practice for heat flow calibration of differential scanning calorimeters” Philadelphia, USA. Braga, C. I., 2009, “Desenvolvimento de metodologia para análises de DSC em altas taxas de transferência de calor” (in Portuguese), Master Thesis, Instituto Tecnológico de Aeronáutica, S.J.Campos, S.Paulo, Brazil. Canevarolo Jr, S. V., 2002, “Ciência dos Polímeros”, Artliber Editora, São Paulo, Brazil. Canevarolo Jr, S. V. et al., 2004, “Técnicas de Caracterização de polímeros”, Artliber Editora, São Paulo, Brazil, pp.209285. Chagas, A. P., 1999, “Termodinâmica Química”, Editora da Unicamp, Campinas, São Paulo, Brazil. Gill, P. S.; Sauerbrunn, S. R. & Reading, M., 1993, Journal of Thermal Analysis and Calorimetry, Vol. 40, pp. 931939. Gramaglia, D., Conway, B.R., Kett, V.L., Malcolm, R. K., Batchelor, H. K. 2005, “International Journal of Pharmaceutics”, Vol. 301, pp. 1-5 . Halliday, D., Resnick, R., Walker, J., 1992, “Fundamentos de Física Gravitação, Ondas e Termodinâmica” (in Portuguese), Vol. 2, São Paulo, Brazil, LCT. Hurtta, M. Pitkänen, I., 2004, “Thermochimica Acta”, Vol. 419, pp. 19-29. Illers, K. H., 1974, “DSC calibration during cooling. A survey of possible compounds”, European Polymer Journal, Vol. 10, pp. 911-916. Ionashiro, M.; Giolito, E. I.,1980, “Nomenclatura, Padrões e Apresentação dos Resultados em Análise Térmica”, Parte II. Cerâmica, Vol. 34, pp. 225-230. Kasap, S.O.,1997, “Principles of Electrical Engineering Materials and Devices”, New York, USA, Irwin Professional Publishing.

Mathot, V. B. F., 1994, “Calorimetry and Thermal Analysis of Polymers”, Hanser Publishers, New York. Nascimento, M. L. F., 2000, “Condutividade elétrica de vidros de boratos, silicatos e sílico-sulfatos de íons alcalinos” (in Portuguese), Master Thesis, Universidade de São Paulo, Physics Institute, São Paulo, Brazil. Neuenfeld, S., Schick, C., 2006, “Verifying the symmetry of differential scanning calorimeters concerning heating and cooling using liquid crystal secondary temperature standards”. Thermochimica Acta, Vol. 446, pp. 55-65. Pijpers, F.J. et al., 2000, “Metastability in polymer systems studied under extreme conditions: high pressure, scan-iso T-t ramps and high scanning rates”, Proceedings NATAS (North American Thermal Analysis Society) 28th Conference, October 4-6,Orlando, USA. Pijpers, F.J. et al., 2002, “High-Speed calorimetry for the study of kinetics of (de)vitrification; crystallization and melting of macromolecules”, Macromolecules, Vol. 35, pp. 3601-3613. Pijpers, F.J. et al., 2004, “Metastability In Polymer Systems Studied At Extreme Conditions,Including Low to Very High Scanning Rates”, Proceedings NATAS (North American Thermal Analysis Society) 30th Conference. Poel, G. V. & Mathot, V. B. F., 2006, Thermochimica Acta, Vol. 446, pp. 41-54. Poel, G. V., Mathot, V. B. F., 2007, Thermochimica Acta, Vol. 461, pp. 107-21. Santos, W. N., 2005, “Polímeros: Ciência e Tecnologia”,Vol. 15, No. 4, pp. 289-295. Saunders, M., Podluii, K., Shergill, S., Buckton, G., Royall, P., 2004, “International Journal of Pharmaceutics”, Vol. 274, pp. 35-40. Turi, E., 1981, “Thermal Characterization of Polymer Materials”, Editor Academic Press Inc, Boston, USA. Vasconcelos, G. C., Mazur, R.L; Botelho, E.C, Rezende, M.C, Costa, M.L. 2010, “J. Aerosp.Technol. Manag.”, Vol. 2, No. 2, pp. 155-162. Wellen, R., Rabello, M., 2005, “Journal of Materials Science”, Vol. 40, No. 23, pp. 6099-6102.

McGregor, C., Saunders, M. H., Buckton, G., Saklatvala, R. D., 2004, “Thermochimica Acta”, Vol. 417, pp. 231-237.

Wendlandt, W. W., 1986, “Thermal Analysis”, Wiley, edition3, New York, USA.

Mothé, C. G., Azevedo, A. D., 2002, “Análise térmica de materiais”, (in Portuguese), iEditora, Rio de Janeiro, Brazil.

Wunderlich, B., 1990, “Thermal Analysis”, Academic Press Inc., Boston, USA.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 179-192, May-Aug., 2011

191

doi: 10.5028/jatm.2011.03022411

Carlos Alberto Ferreira Gisler

Instituto do Controle do Espaço Aéreo São José dos Campos/SP – Brasil gisler@icea.gov.br

Gilberto Fisch*

Instituto de Aeronáutica e Espaço São José dos Campos/SP – Brasil gilbertoﬁschgff@iae.cta.br

Cleber de Sousa Correa

Instituto do Controle do Espaço Aéreo São José dos Campos/SP – Brasil cleber@icea.gov.br *autor para correspondência.

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara Resumo: A análise estatística do perfil de vento obtido junto à torre anemométrica do Centro de Lançamento de Alcântara foi realizada com base nos dados de vento (direção e velocidade), coletados no período de 1995 a 1999, obtidos em seis níveis diferentes: 6,0; 10,0; 16,3; 28,5; 43,0 e 70,0 m. Nesta análise, foram considerados os meses característicos das estações chuvosa (março) e seca (setembro), na região do Centro de Lançamento de Alcântara. O total de dados analisados para o período chuvoso (março) foi de 76.882 perfis do vento (intervalo de tempo de dez minutos) e para o período seco (setembro), foi de 109.809. Foram obtidos os valores médios da velocidade (ou intensidade) do vento, desvio padrão, mediana, moda e predominância da direção do vento. Obteve-se a como direção predominante NE com 33 e 40% de frequência de ocorrência para os períodos chuvoso e seco, respectivamente. Os valores médios da velocidade do vento apresentam dependência com a altura, sendo observado que os níveis mais elevados da torre anemométrica apresentam as maiores intensidades no período seco (8,2 ms-1). Os valores da velocidade média dos ventos observada foram de 6,4 ms-1 para o período seco e de 4,1 ms-1 para o chuvoso. Na análise das distribuições de dados, foram realizados ajustes para as distribuições estatísticas normal e Weibull. Os resultados mostram que a velocidade do vento está ajustada para um nível de significância de 95% (α=0,05) à distribuição estatística normal e Weibull. A distribuição de Weibull, para o período completo, apresentou ajuste para valores entre 3,0 e 9,0 ms-1 e a distribuição normal apresentou um bom ajuste para valores entre 4,0 e 9,0 ms-1. Palavras-chave: Torre Anemométrica, Distribuição estatística de Weibull, Distribuição Normal, Anemômetros.

Statistical analysis of wind proﬁle in the surface layer at the Alcântara launching center

Received: 03/06/11 Accepted: 21/07/11

Abstract: Statistical analysis of the wind profile made at an anemometric tower installed at the Alcântara Launching Center was based on wind data (direction and wind speed) collected between 1995 up to 1999, which was carried out at six different levels: 6.0, 10.0, 16.3, 28.5, 43.0, and 70.0 m. This analysis was made for typical rainy months (March) and dry (September) season, in the Alcântara Launching Center area. The analyzed data total during the wet season (March) was 76,882 wind profiles (time interval of ten minutes) and during the dry season (September) was of 109,809 profiles. It was computed the mean wind speed (or intensity) of the wind, standard deviation, median, mode, and the prevailing wind direction. The predominant direction was from NE with 33 and 40% for wet and dry seasons, respectively. The average values of wind speed showed a dependency with height and it was observed that the highest levels of the anemometric tower have the strongest wind speed in the dry period (8.2 ms-1). The values of average wind speeds observed were 6.4 ms-1 for the dry season and 4.1 ms-1 during the wet one. The normal and Weibull statistical distributions were adjusted to the observed data set. Results show that the wind speed is adjusted to a 95% level (α=0.05) for the normal and Weibull statistical distributions. The Weibull distribution for the entire period presented and adjust to values between 3.0 and 9.0 ms-1, and the normal one showed a good fit for values between 4.0 and 9.0 ms-1. Keywords: Anemometric tower, The Weibull statistical distribution, Normal distribution, Anemometers.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

193

Gislei, C.A.F., Fisch, G., Correa, C.S.

INTRODUÇÃO O Centro de Lançamento de Alcântara (CLA) tem, por missão, o lançamento de cargas úteis (satélites e/ ou experimentos científicos) brasileiros, por meio do veículo lançador de satélites (VLS) e/ou foguetes de sondagem, tal como o VSB-30. O lançamento de um satélite brasileiro, por um veículo nacional, a partir de um centro espacial no Brasil, compõe a denominada Missão Espacial Completa Brasileira (MECB). Para o sucesso de qualquer lançamento ou operação aeroespacial, é de vital importância o conhecimento das condições meteorológicas da região. Dentre tais variáveis, podem ser ressaltados os ventos (de superfície e altitude) como possíveis agentes de influência e/ou modificação na trajetória dos veículos. A Fig. 1 mostra a área do Setor de Preparação e Lançamento (SPL) do CLA, notandose a torre móvel de integração, a casamata e a torre anemométrica (TA), bem como a proximidade da costa litorânea. O conhecimento da variação temporal da velocidade e direção do vento é de grande interesse para as atividades operacionais nos lançamentos de foguetes, que são realizados nos Centros de Lançamento (como, por exemplo, no Kennedy Space Center – EUA e/ou no Centre Spacial Guianense, na Guiana Francesa), devido às flutuações, vórtices e turbilhões que compõem a turbulência atmosférica do local. O conhecimento do perfil vertical do vento possibilita identificar os processos da troca de momentum e as características do vento em níveis diferentes. Recentemente, Johnson (2008) publicou um relatório, que mostrava os critérios do vento a serem considerados para estudos de pesquisa e desenvolvimento (P&D) na área de veículos espaciais, bem como as informações do vento (tanto de superfície como de altitude) a serem consideradas para o lançamento de foguetes. No caso específico do CLA, os estudos desenvolvidos pelo

Fonte: Gisler (2009). Figura 1. Vista parcial da área do SPL, mostrando a disposição da Casamata, a Torre Móvel de Integração (TMI) e Torre Anemométrica (TA).

194

projeto de meteorologia aeroespacial (Pires et al., 2009; Fisch et al., 2010) mostram a complexidade do escoamento atmosférico gerado pela presença da falésia no local. O clima de Alcântara apresenta grande complexidade do ponto de vista meteorológico, com a influência de vários sistemas atmosféricos comuns à região tropical, tais como: a brisa marítima (BM); a zona de convergência intertropical (ZCIT); o distúrbio ondulatório de Leste (DOL); o vórtice ciclônico em altos níveis (VCAN); as linhas de instabilidade (LI) e os sistemas frontais (SF). Apesar disto, as características climáticas evidenciam dois macrorregimes pluviométricos: uma época seca, que compreende o período de agosto a dezembro, e outra chuvosa de janeiro a julho (Fisch, 1999; Guedes e Oyama, 2004). Os ventos apresentam-se com características de forte intensidade na época seca, em parte devido ao aumento do contraste térmico entre continente e oceano, intensificandose as brisas marítimas e terrestres e atingindo a velocidade de 7,0 ms-1 (Coutinho e Fisch, 2007). Na época chuvosa, o contraste térmico entre o continente e o oceano é menor; em consequência, os ventos possuem menor intensidade, com velocidades de aproximadamente 3,0 ms-1 (Guedes e Oyama, 2004). Considerando-se o vetor médio do vento para as épocas chuvosa e seca (Fig. 2), observa-se que a direção é de NE com velocidade entre 4,0 e 5,0 ms-1 na época chuvosa, sendo que esta velocidade aumenta para 5,0 e 6,5 ms-1 na época seca. Ressalta-se que esta configuração é oriunda da saída dos produtos de previsão numérica de tempo (reanálises). O regime dos ventos no CLA tem predominância de NE, sendo que sua velocidade apresenta uma variação com a altura do perfil, ou seja, segue a lei logarítmica de aumento da velocidade com a altura (Fisch, 1999; Roballo e Fisch, 2008). O ajuste do campo de vento por distribuições estatísticas tem sido estudado por vários autores. Dentre eles, Souza e Granja (1997) fizeram o ajuste do vento para a região do Mato Grosso do Sul, encontrando bons resultados para o ajuste com a distribuição de Weibull. Também salientam que a distribuição de Weibull tem dificuldades em ajustar os dados do vento para valores muito baixos (inferior a 2,0 m/s). Por outro lado, Yim et al. (1999) ajustaram os valores de vento em intervalos de dez minutos para as distribuições estatísticas de Gumbel e Weibull e também concluíram sobre a adequabilidade do uso da distribuição de Weibull. Resultados similares também foram encontrados por Silva et al. (2002) para uma série climatológica (15 anos) dos dados de vento no NE do Brasil.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara

NOAA/ESRL Physical Sciences Division

10N EQ 10S Jan a Jun: 1995 a 1999 a 2

Jul a Dez: 1995 a 1999 b 4

Figura 2. Reanálises do vetor médio da época chuvosa (janeiro a junho) e seca (julho a novembro), em ms-1.

O objetivo deste trabalho consistiu na análise dos perfis verticais do vento (PVV) dos dados coletados na TA, do CLA. As informações possibilitam a caracterização estatística detalhada do vento na região do CLA, nos períodos chuvosos e seco, e visam aumentar o conhecimento da climatologia da região, aumentando a segurança dos lançamentos de foguetes. DADOS E METODOLOGIA A área de estudo é o CLA, localizada no município de Alcântara (2°24’S; 44°25’W; 40 m), como pode ser visto na Fig. 3, situada a aproximadamente 20 km ao Sudeste de São Luis, no Maranhão, na região Nordeste do Brasil – NEB (CLA, 2008). Em relação à climatologia do local, a nebulosidade apresenta uma variação sazonal muito acentuada, com média superior a cinco oitavos na época chuvosa e sendo reduzida para um valor inferior a três oitavos na época seca. Tal comportamento está associado com o regime pluviométrico, que apresenta chuvas intensas (período chuvoso) entre janeiro e junho, com picos em março e abril, e um período de forte estiagem (seca) de julho a dezembro. A temperatura média mensal apresenta uma pequena amplitude anual de 1,4º C; sendo que o mês mais quente é novembro com média de 28,1º C e o mais frio, março com média de 26,7º C. (Pereira, 2002).

Figura 3. Localização do município de Alcântara – MA (esquerda) e vista parcial do CLA, com destaque para o Setor de Meteorologia (direita).

Figura 4. Vista da TA e dos sensores anemométricos.

As séries de dados estudadas foram obtidas pelos sensores instalados na TA do CLA, em uma estrutura metálica de 70 m. A Fig. 4a mostra uma vista panorâmica da TA e dos sensores de direção e velocidade do vento, sendo que a Fig. 4b mostra um aerovane em detalhes.

– MI, USA), que mede a velocidade horizontal (hélices) e a direção (aerovane) do vento (Fig. 4b). Estes anemômetros foram instalados em seis níveis, nas alturas de 6,0 (nível 1), 10,0 (nível 2), 16,3 (nível 3), 28,5 (nível 4), 43,0 (nível 5) e 70,0 m (nível 6), e são rotineiramente calibrados no túnel de vento do Instituto de Aeronáutica e Espaço/Divisão de Aerodinâmica (IAE/ALA), seguindo procedimentos usuais de calibração dos sensores de vento.

O equipamento utilizado foi um anemômetro (aerovane) da R.M Young (anemômetro de Modelo 05106, Traverse City

A coleta dos dados foi realizada automaticamente por meio de um sistema de aquisição de dados CR-7, da

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

195

Gislei, C.A.F., Fisch, G., Correa, C.S.

Campbell Scientific Instrument (Logan – UT, USA), transmitindo os dados por meio da fibra ótica para o Setor de Meteorologia do CLA. A taxa de amostragem dos sensores é de 0,5 Hz, sendo que os valores médios e estatísticos são armazenados em intervalos de dez minutos para cada nível da TA. A constante de distância do aerovane é de 3,3 m (para uma recuperação de 63%), o que produz uma constante de tempo de 0,5 segundos para um vento típico de 6 m/s (Fisch, 1999). A partir dos dados coletados na TA do CLA, foram gerados arquivos mensais com as alturas, direção e intensidade do vento a cada dez minutos, as quais são denominadas PVV. O período considerado para a análise foi de 1995 a 1999, com um total de 859.468 perfis. Para analisar as características estatísticas das variáveis de velocidade e direção do vento, os dados foram agrupados em diferentes escalas de tempo (meses e anos), sendo determinados os valores médios, o desvio padrão (DP) e a variância da amostra, os valores de mediana, moda e extremos (máximos e mínimos), bem como a distribuição de freqüência, na qual os dados se ajustaram melhor. Para a determinação dos parâmetros das distribuições normal e Weibull, utilizou-se o software MATLAB© 6.5. Para avaliar o ajuste das distribuições ao conjunto de dados, foram usados três métodos: o teste do qui-quadrado, o de Kolmogorov-Smirnov e gráficos de probabilidades empíricas acumuladas versus teóricas acumuladas. Na análise das distribuições estatísticas, os valores com velocidade igual a zero e os superiores a 18,0 ms-1 foram desconsiderados, por apresentarem pouca representatividade na série (menos de 0,01% dos dados) e por não serem representativos dos ventos que ocorrem na região do CLA. A equação da distribuição normal é especificada usando dois parâmetros, a saber: a média populacional, μ, e o DP populacional, σ, ou equivalentemente a variância populacional σ2. Denota-se N (μ, σ2) à curva normal com média μ e variância σ2. A distribuição normal é simétrica em torno da média, o que implica que os valores de tendência central (média, mediana e moda) são todos coincidentes. A função de densidade de probabilidade (f(x)) é dada por (Eq. 1): (2)

! As funções de densidade de probabilidade f(x) e de probabilidade acumulada F(x) para a distribuição de Weibull podem ser representadas como (Eq. 2 e 3):

196

– ax f(x) = a b x b-1 exp � b

– ax F(x) = 1- exp �

(2) (3)

Nessas, os coeficientes a e b são, respectivamente, os parâmetros de escala e forma. Os seus momentos de ordem 1 em relação à origem (µ1´) e centrado de ordem 2 (µ2) são dados por (Eq. 4 e 5): _

µ1´= E(x) = x = a-1/b Γ(1+b-1)

(4)

e µ2 = S2 = a–2/b [Γ(1+2b-1) - Γ2(1+b-1)]

(5)

Os parâmetros de ajuste para as distribuições foram computados pelo método da máxima verossimilhança, o qual, na maioria dos casos, leva às melhores estimativas (menores valores do erro médio quadrático), usando as ferramentas estatísticas do pacote MATLAB. RESULTADOS Predominância do vento A Fig. 5 mostra a frequência da direção do vento, observado em março (característico da época chuvosa) e setembro (característico da época seca). Observa-se, na Fig. 5a, que a rosa dos ventos para março apresenta maior predominância de NE (com 33%), seguido de NNE (18%) e ENE (13%). Observase, também, ocorrência de direções do vento de E e W com percentual inferior a 5%. Para setembro (Fig. 5b), a predominância da direção do vento é dividida entre NE (50%) e ENE (40%), com apenas 6% para as direções de NNE e E. Observa-se que, para estas condições (estação seca), não há registro de ocorrência do vento nas demais direções. A Tabela 1 apresenta os valores estatísticos (média, DP, moda, e mediana) do vento para março e setembro. No caso do período chuvoso (março), observa-se que o valor médio da velocidade varia verticalmente entre 3,3 a 5,4 ms-1 e o DP entre 1,4 e 2,2 ms-1. Não obstante a variação logarítmica do vento com a altura, é possível calcular um valor médio da camada. Nestas condições, o valor médio da velocidade do vento no período total é de 4,3 ms-1, DP de 1,8 ms-1, sendo a moda igual a 4,0 ms-1 e a mediana, a 3,9 ms-1. A moda da direção do vento é de 38º (NE). Para o período seco (setembro), os valores médios da velocidade do vento variam verticalmente entre 5,2 e 8,4 ms-1 e o DP

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara

entre 1,1 e 1,5 ms-1. O valor médio da velocidade do vento para setembro é de 6,8 ms-1, DP de 1,4 ms-1, sendo a moda igual a 6,0 ms-1 e a mediana igual a 6,2 ms-1. A moda da direção do vento permanece NE. Análise das distribuições estatísticas As Fig. 6 e 7 apresentam distribuição estatística normal e de Weibull para o conjunto completo dos dados de velocidade do vento. Com base nos testes de aderência do qui-quadrado a um nível de significância de 5%, podese constatar que a distribuição dos dados de velocidade do vento se ajusta bem às distribuições estudadas. Para a distribuição normal, os valores se ajustam à distribuição estatística até um determinado valor entre 8,0 e 8,5 ms-1, ocorrendo um desvio para valores maiores (em outras palavras, as velocidades máximas observadas são inferiores àquelas calculadas pela distribuição normal). Esse desvio está associado ao fato dos valores extremos não seguirem fielmente a distribuição normal, bem como um número menor de perfis também (representado pelo tamanho da amostra). Para o valor limite da velocidade do vento de 8,0 ms-1, existe probabilidade de 95% de o vento ser inferior a este valor crítico. No caso da distribuição de Weibull (Fig. 7), o ajuste é bom para velocidades no

intervalo entre 3,0 e 9,0 ms-1, sendo que 99% dos valores situam-se abaixo de 9,0 ms-1. Ressalta-se que, para os veículos de sondagem (foguetes do tipo VSB30, por exemplo), o limite operacional para o lançamento é de 10,0 ms-1, o que garante uma boa janela de lançamento do veículo, pelo menos para os ventos na camada limite superficial. As distribuições estatísticas (normal e de Weibull) para março (época chuvosa) e setembro (época seca) são apresentadas nas Fig. 8 a 11. Para março (Fig. 8), observa-se que os dados apresentam-se bem ajustados a uma distribuição normal para os valores de velocidade entre 1,0 e 5,0 ms-1, sendo que este último valor possui uma probabilidade cumulativa de 98% dos casos serem inferiores a ele. No caso da distribuição de Weibull (Fig. 9), valores entre 1,0 e 9,0 ms-1 ajustam-se bem à distribuição, apontando uma probabilidade cumulativa de 99%. Nas Fig. 10 e 11, mostram-se os ajustes a distribuições estatísticas normal e de Weibull para setembro. No caso da distribuição normal (Fig. 10), o ajuste é bom entre os valores de 4,5 a 9,0 ms-1, indicando que, para uma velocidade do vento de 9,0 m/s, há probabilidade de 90% dos valores serem inferiores a este. A Fig. 11 apresenta

Figura 5. Frequência da direção do vento para os meses de março (a) e setembro (b).

Tabela 1. Valores estatísticos: direção (moda) e velocidade (média, DP, moda e mediana), por níveis da TA para março (chuvoso) a setembro (seco). MARÇO SETEMBRO MODA MEDIA (DP) MODA MED MODA MEDIA (DP) MODA MED (dir) (vel) (dir) (vel) Nível 2 40 3,3 (1,4) 3,7 3,5 67 5,2 (1,4) 5,3 5,3 Nível 3 38 3,7 (1,6) 3,3 3 ,9 69 6,1 (1,1) 5,6 6,0 Nível 4 36 4,3 (1,8) 4,6 4,5 46 6,7 (1,3) 6,2 6,6 Nível 5 39 4,7 (1,9) 5,1 5,0 39 7,6 (1,5) 7,1 7,4 Nível 6 50 5,4 (2,2) 6,0 6,0 54 8,4 (1,5) 7,9 8,3

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

197

Gislei, C.A.F., Fisch, G., Correa, C.S. Distribuição Normal Janeiro 1996 - Dezembro 1999

Probabilidade

0,999 0,997 0,99 0,98 0,95 0,90 0,75 0,50 0,25 0,10 0,05 0,02 0,01

0,5

Velocidade (m/s)

Figura 6. Distribuição normal: a linha tracejada representa a extrapolação da curva ajustada e a linha azul, os pontos em distribuição cumulativa.

as informações para a distribuição de Weibull, sendo que há um bom ajuste entre 4,5 e 9,0 m/s. As probabilidades de ocorrência para valores inferiores a 9,0 m/s são equivalentes para as duas distribuições (90%). COMENTÁRIOS FINAIS E CONCLUSÕES Na análise da frequência do vento, considerando os meses característicos como chuvoso (março) e seco (setembro), pode-se concluir que a predominância dos ventos é de NE. A predominância do vento no período seco apresenta maior persistência com aproximadamente 90% e, para o período chuvoso, a persistência do vento de NE é de aproximadamente 65%. Estes resultados evidenciam que pode ocorrer uma maior variabilidade da direção do vento associada à presença de nuvens e chuvas oriundas de sistemas meteorológicos na região, característicos da época chuvosa. As velocidades do vento apresentam-se diferentes em relação ao período considerado: os valores de velocidade variam entre 3,3-3,7 a 5,4-6,0 ms-1, para o período chuvoso, e de 5,2-5,3 a 7,9-8,4 ms-1, para o seco, com aumento de quase 2,5 ms-1 na época seca do que na chuvosa, considerando-se o nível de 70 m. Em todas as situações, existe uma dependência do vento com a altura. Evidenciase que o contraste térmico entre oceano e continente no

198

período seco é mais acentuado e pode estar colaborando na determinação da intensidade maior do vento. Os dados observacionais do período estudado ajustam-se bem às distribuições estatísticas normal e de Weibull para os valores de velocidade do vento entre 3,0 e 8,0 ms-1. Na estação chuvosa (março), o ajuste às distribuições foi observado entre 3,0 e 8,0 ms-1 e, para a época seca (setembro), o ajuste está entre 3,0 e 9,0 ms-1. Além disso, constatou-se que ambas as distribuições falham em representar a velocidade do vento para as velocidades superiores a 10,0 ms-1. A distribuição de Weibull mostra um ajuste melhor do que a distribuição normal para velocidades acima de 8,0 ms-1. Para a estação chuvosa, que apresenta ventos mais fracos, o ajuste de ventos acima de 5,0 m/s ocorreu somente para a distribuição de Weibull. Entretanto, são poucos os casos, pois o aumento da velocidade do vento entre 5,0 e 9,0 m/s representou somente 1% na probabilidade cumulativa. Na época seca, as duas distribuições se mostraram equivalentes, com probabilidades cumulativas iguais. Este estudo contribui para um melhor entendimento das estruturas do PVV no CLA, visando aumentar a segurança de voo e diminuindo os riscos no lançamento de foguetes e veículos lançadores.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara Distribuição de Weibull Janeiro 1996 - Dezembro 1999

0,999 0,99 0,96 0,90 0,75

Probabilidade

0,50 0,25 0,10 0,05 0,02 0,01

0,5

4 Velocidade (m/s)

Figura 7. Distribuição de Weibull, período de 1996 a 1999. Distribuição Normal Março 1996 - 1999

Probabilidade

0,999 0,997 0,99 0,98 0,95 0,90 0,75 0,50 0,25 0,10 0,05 0,02 0,01

0,5

Velocidade (m/s)

Figura 8. Distribuição normal para março. J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

199

Gislei, C.A.F., Fisch, G., Correa, C.S.

Distribuição de Weibull Março 1996 - 1999

0,999 0,99 0,96 0,90

Probabilidade

0,75 0,50 0,25 0,10 0,05 0,02

Velocidade (m/s)

Figura 9. Distribuição Weibull para março.

Distribuição Normal - Setembro 1996 - 1999

Probabilidade

0,999 0,997 0,99 0,98 0,95 0,90 0,75 0,50 0,25 0,10 0,05 0,02 0,01 0,003 0,001

Velocidade do vento (m/s)

Figura 10. Distribuição normal para setembro. 200

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

Análise estatística do perfil de vento na camada limite superficial no centro de lançamento de Alcântara Distribuição Weibull Setembro 1996 - 1999 0,999 0,99 0,96 0,90 0,75

Probabilidade

0,50 0,25

0,10 0,05

0,02 0,01 1,0

2,0

3,0

6,0

8,0

10,0

Velocidade do vento (m/s)

Figura 11. Distribuição Weibull para setembro.

REFERÊNCIAS Centro de Lançamento de Alcântara (CLA), 2008, “Catálogo CLA: história”, Retrieved in July, 2008, from http://www.cla.aer.mil.br/. Coutinho, E. C., Fisch, G., 2007, “Distúrbios Ondulatório de Leste (DOLs) na região do Centro de Lançamento de Alcântara – MA”, Revista Brasileira de Meteorologia, Vol. 22, No. 2, pp. 193-203. Fisch, G., 1999, “Características do Perfil Vertical do Vento no Centro de Lançamento de Foguetes de Alcântara (CLA)”, Revista Brasileira de Meteorologia, Vol. 14, No.1, pp. 11-21. Fisch, G., Avelar, A.C., Pires, L.B.M, Gielow, R., Girardi, R.M. and Souza, L.F., 2010, “The internal boundary layer at the Alcântara Space Center: winds measurements, wind tunnel experiments and numeric simulations”, Proceedings 5th International Symposium on Computational Wind Engineering (CWE2010), Chapel Hill, NC/USA, 6 p., CD-ROM. Guedes, R. L. and Oyama, M. D., 2004, “Aspectos Observacionais das Oscilações Intra-sazonais de Intensidade do Vento em Alcântara usando ondeletas: análise preliminar”, Anais do XIII Congresso Brasileiro de Meteorologia, Fortaleza-CE, Brasil, 8 p., CD-ROM. Gisler, C.A.F., 2009, “Análise do Perfil de Vento na Camada Limite Superficial e Sistemas Meteorológicos atuantes no Centro de Lançamento de Alcântara”, Dissertação de Mestrado - INPE, from http://urlib.net/sid. inpe.br/mtc-m18@80/2009/04.24.12.33.

Johnson, D.L., 2008, “Terrestrial Environmental (Climatic) Criteria Guidelines for Use in Aerospace Vehicle Development”, NASA/TM-2008-215633, CD-ROM. Pereira, E. I. Atlas climatológica do Centro de Lançamento de Alcântara – MA. São José dos Campos: CTA/IAE – Divisão de Ciências Atmosféricas, 2002. Pires, L.B.M., Avelar, A.C., Fisch,G., Roballo, S.T., Souza, L.F., Gielow, R. and Girardi, R.M., 2009, “Studies using Wind Tunnel to Simulate the Atmospheric Boundary layer at the Alcantara Space Center”, Journal of the Aerospace and Technology Management, Vol. 1, No. 1, pp. 91- 98. Roballo, S. T. and Fisch, G., 2008, “Escoamento Atmosférico no Centro de Lançamento de Alcântara (CLA): Parte I – Aspectos observacionais”, Revista Brasileira de Meteorologia, Vol. 23, No. 4, pp. 510-518. Silva, B.B., Alves, J.J.A., Azevedo, F.G.B., Cavalcanti, E.P., Dantas, R.T., 2002, “Potencial eólico na direção predominante do vento no Nordeste brasileiro”, Revista Brasileira de Engenharia Agrícola e Ambiente, Vol. 6, No. 3, pp. 431-439. Souza, A., Granja, S.C, 1997, “Estimativa dos parâmetros “C” e “K” do modelo de Weibull e da direção dos ventos para Campo Grande e Dourados, MS, Brasil”, Revista Brasileira de Agrometeorologia, Santa Maria, Vol. 5, No.1, pp. 109-114. Yim, J.Z., Lin, J.G., Hwang, C.H., 1999, “Statistical properties of the wind field at Taichung harbour, Taiwan”, Journal of Wind Engineering and Industrial Aerodynamics, Vol. 83, pp. 49-60.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 193-202, May-Aug., 2011

201

doi: 10.5028/jatm.2011.03020111

Antonio Henriques de Araújo Jr*

State University of Rio de Janeiro Rio de Janeiro – Brazil anthenriques2001@yahoo.com.br

Flávio Hegenberg

University Center Oswaldo Aranha Volta Redonda – Brazil ﬂavio.leeds@gmail.com

Isabel Cristina dos Santos

University of Taubaté Taubaté – Brazil isa.santos.sjc@gmail.com

José Glênio Medeiros de Barros

State University of Rio de Janeiro Rio de Janeiro – Brazil glenio.barros@gmail.com *author for correspondence

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines ABSTRACT: The paper aimed to identify and analyze the explanatory variables for airlines productivity during 2000 2005, by testing the Pearson correlation between the single factor productivity capital, energy and labor of a sample of 45 selected international airlines (4 Brazilian carriers among them) and their productivity explanatory variables like medium stage length, aircraft load factor, hours flown and cruise speed for selected routes besides aircraft seat configuration and airlines number of employees. The research demonstrated, that a set of variables can explain differences in productivity for passenger airlines, such as: investment in personnel training processes, automation, airplane seat density, occupation of aircraft, average flight stage length, density and extension of routes, among others. Keywords: Multifactor productivity, Multifactor productivity model, Airline productivity, Passenger airline productivity.

INTRODUCTION In the age of deregulation, great disparities exist between airlines in their ability to reduce unit costs by improving productivity and also to generate adequate revenues despite increasing price competition. Substantial differences exist, for example, between The United States of America’s (US) airlines and non-US airlines in terms of cost efficiency, revenue generation and, in turn, profitability. Usually, measures of airline productivity to the extent they are used in the industry are limited to relatively simple ratios – such as passenger enplanements per employee and Available Seat Miles (ASM) produced per labor dollar spent – which does not allow reliable conclusions and comparisons among the productivity of airlines. RESEARCH OBJECTIVES This paper focuses on the productivity analysis of the main production factors for airlines: (a) labor, (b) capital and (c) energy and the identification and analysis of variables that can statistically explain these single productivities factors labor and, consequently, their Total Factor Productivity. Little or no research has been done to identify variables that explain productivity of scheduled passenger airlines in order to develop a model of multiple variables, more

Received: 09/02/11 Accepted: 10/05/11

complex to measure productivity and compare productivity between airlines. The paper aimed to identify and discuss the explanatory variables for the productivity of scheduled international airlines by testing the Pearson correlation between the productivity changes of airlines and their explanatory variables with the objective of proposing a productivity model. The research demonstrated that an extensive set of variables can explain differences in productivity of airlines. These variables include: investment in personnel training, process automation, airplane seat configuration, occupation of the aircraft (load factor), flight stage length, density and extension of routes, among others. The aim of this paper was not to formulate the model itself, but to allow, from the identification of these variables, the creation of conditions to formulate such a model. THEORETICAL FRAMEWORK The demonstrable effects of successful US deregulation and ongoing inefficiency in the industry may have influenced the European Commission to introduce certain reforms to promote competition and thus increase the efficiency and productivity of European airlines. Much of the literature has concentrated on productivity in the United States compared to that in Europe according to the McKinsey Global Institute (1992) and Good et al. (1993), whereas only a small proportion of papers

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

203

Araújo Jr, A.H.A. et al.

present productivity estimates for European countries individually, as reported by Encaoua (1991). Moreover, many authors prefer to concentrate on Total Factor Productivity (TFP), as stated by Windle (1991), in favor of labor productivity measurements. The productivity of air transport has been extensively studied over the last two decades, using different methods. Bailey, Graham and Kaplan (1985) proposed, when considering the deregulation of air transportation in the United States, a method to measure the productivity of US airlines, based on the relationship between the average costs per ton x km and two time periods, and an index of input prices for airlines, according to the following formula (Eq. 1): PR t =

CR t / CR t-1 Pt / Pt-1

(1)

Where CRt is the average cost per ton transported in period t and Pt is a price index of inputs in period t. These authors estimated, in 1985, the total productivity due to changes in the occupation of 18 airline fleets in the domestic US market. Windle (1991) compared the TFP and costs of 41 companies (14 American, 27 European and Asian), between 1970 and 1983, using the translog multilateral output index, as proposed by Caves, Christensen and Diewert (1982). In this study, five categories of inputs were utilized: (a) labor, (b) fuel, (c) flight equipment, (d) ground equipment and (e) materials. The author pointed to the evidence of a relationship between TFP and Multifactor Productivity (hereafter, MFP) input categories, such as Revenue Ton Miles (RTM) per employee. Distexhe and Perelman (1994) aimed, in their study, to evaluate the consequences of the deregulation in the US market. This was done by measuring efficiency and productivity of airlines (during the period between 1977 and 1988). The sample consisted of 33 companies operating in 3 groups of markets: (a) Asia and Oceania, (b) Europe and (c) North America. The authors used the Data Envelopment Analysis (DEA) method to construct efficient frontiers for these companies, using Färe’s approach to estimate the Malmquist Productivity Index (MPI), by breaking it down into technical progress and efficiency gains, and using labor and capital as inputs. In the above mentioned research, Distexhe and Perelman (1994) showed that European airlines were less efficient than the surveyed US carriers. Among the European airlines, Lufthansa, KLM and Air France had the highest efficiency 204

score, while British Airways, Alitalia and Swissair failed to reach more than 80% of the efficient frontier. Sickles, Good and Getachew (2002) examined the productive performance of a group of 3 East European carriers and compared them to 13 of their West European competitors during the period 1977-1990. The authors first modeled the multiple output/multiple input technology with a stochastic distance frontier using semi-parametric efficient methods. The endogenous character of multiple outputs is addressed, in part, by introducing multivariate Kernel estimators for the joint distribution of the multiple outputs and potentially correlated firm random effects. They augmented estimates from semi-parametric stochastic distance function with nonparametric distance function methods, using linear programming techniques, as well as with extended decomposition methods, based on the Malmquist index number. Both semi- and nonparametric methods indicated significant slack in resource utilization in the East European carriers studied relative to their Western counterparts, and limited convergence in efficiency or technical change between them. Kune, Mulder and Poudevigne (2000) evaluated air transport productivity in France, Germany, United Kingdom and the United States for the period 1970-1998, using the TFP method. The objective of this study was to evaluate and compare the productivity of labor, capital and the TFP of air transport in these countries. The MFP was estimated by means of production functions and with the utilization of variables such as Value-added, Labor, and Capital. The authors suggested in their study that, if the costs of production factors are equal to their marginal productivities, according to the neo-classical assumptions for competitive markets, the increase of TFP can be estimated with the Divisia-Tornquist index. The above mentioned authors concluded that capital is a key production factor in the airline industry, and a large part of the improvement of this economic sector depends on investments in infrastructure and equipment. The differences of the capital stock per worker are also important variables for explaining performance differences between economic sectors and countries. The labor and capital productivity between France, Britain, Germany and the United States was compared in this study. Färe, Grosskopf and Sickles (2001) examined a sample of 13 US companies between 1979 and 1994 based on the

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines

generalization of Shephard directional distance functions, by using the TFP of US airlines, whereby this author employed the Malmquist-Luenberger Productivity Index, constructed from directional distance functions. Oum and Yu (2001) produced interesting research in empirical and conceptual terms, evaluating the performance and productivity of the largest Canadian airlines for the period 1995-2000, in comparison to the eight largest American companies, using Kendrick’s arithmetic index and performance metrics such as average load factor and medium stage length, evaluating also the economic and financial performance of these companies. In Brazil, Araújo Junior (2004) studied the productivity of Brazilian airlines, during 1996-2002, evaluating the performance of the five largest Brazilian airlines, also using Kendrick’s arithmetic index and concluded that the TFP of these carriers, surpassed the average productivity of the Brazilian industry sector. METHODOLOGICAL PROPOSITION A Multi-factor Productivity (hereafter, MFP) index, which includes the main production factors (i.e. labor, capital and energy), was used to measure the productivity of companies surveyed during the 2000-2005 time period.

MFP =

AVt / AV0 x100 a 0 (L t / L0 ) + b0 (K t / K 0 ) + c0 (E t / E0 ) (3)

Equation 3 is derived from Eq. 1, which makes possible to calculate productivity growth in physical terms in a time period (0, t), where AVt is the number of passengers transported or the Revenue Seat-km (RSK); Lt represents the number of employees at the end of period t (31st December); Kt is the number of aircraft operating at the end of the same period and Et is the amount of fuel spent also at end of period t. Different labor and capital productivity weights, taken from Economic Report (IATA, 2001), take into account the share of input in the operational costs of carriers, according to the airline of origin, as shown in Table 1. Equation 3 gives the productivity change from a reference period 0 to a future time t. Some authors, among them Moreira (1994), propose that weights a0 and b0 should be substituted, periodically, in order to reflect alterations in the production structure and changes in relative prices of capital and labor. Some organizations, such as the National Bureau of Economic Research (NBER), recommend changes every five years.

Multi-factor productivity TFP or MFP is defined as the ratio in the quantities/ volumes produced and a weighted combination of quantities and volumes of the different inputs used in the production process. Kendrick’s productivity measurement method was used with changing-weight indices of outputs and inputs according to Kendrick (1996). The MFP index is represented as the ratio between the output and input, where inputs are weighted by their share in production costs (Eq. 2). MFP =

AVt x100 a0 (Lt ) + b0 (K t )

(2)

In Eq. 2, MFP indicates the MFP index measured in monetary terms, according to Kendrick’s method, which, in this case, is calculated from the ratio between the added value of the airlines in year t and the weighted relationship of labor, e.g. salaries (Lt) and capital, e.g. capital assets (Kt) in the same year, where a0 and b0 represent labor and capital weights, respectively.

Pearson product-moment correlation coefficient The Pearson product-moment correlation coefficient (denoted by r) is a measure of the correlation (linear dependence) between two variables X and Y, taking values from -1 through 0 to +1. It has been used in the sciences as a measure of the strength of linear dependence between two variables. The correlation coefficient is sometimes called “Pearson’s r.” Pearson correlation coefficient between two variables is defined as the covariance of the two variables (X and Y) divided by the product of their standard deviations (Eq. 4): l X ,Y =

cov( X ,Y ) E[( X < + X )(Y < +Y )] = m X mY m X mY

(4)

Equation 4 defines the population correlation coefficient, commonly represented by the Greek letter ρ (rho). If we substitute estimates of the covariances and variances based on a sample, we obtain the sample correlation coefficient r (Eq. 5):

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

205

AraĂşjo Jr, A.H.A. et al.

Data collection

n - i 1 ( X i < X )(Yi < Y ) n n - i 1 ( X i < X )2 - i 1 (Yi < Y )2

(5)

An equivalent expression gives the correlation coefficient as the mean of the products of the standard scores. Based on a sample of paired data (Xi, Yi), the sample Pearson correlation coefficient is (Eq. 6): r"

1 n Xi X 8 n 1 i"1 sX

Yi Y sY

(6)

Where X i < X , X and s X are the standard score, sample sX mean, and sample standard deviation. Several authors have offered guidelines for the interpretation of a correlation coefficient. Cohen (1988) has observed, however, that all such criteria are, in some ways, arbitrary and should not be observed too strictly. The interpretation of a correlation coefficient depends on the context and purposes. A correlation of 0.9 may be very low if one is verifying a physical law using high-quality instruments, but may be regarded as very high in the social sciences where there may be a greater contribution from complicating factors. Pearsonâ&#x20AC;&#x2122;s correlation intervals are disclosed in Table 2. Statistical inference based on Pearsonâ&#x20AC;&#x2122;s correlation coefficient often focuses on one of the following two aims. One aim is to test the null hypothesis that the true correlation coefficient is Ď , based on the value of the sample correlation coefficient r. The other aim is to construct a confidence interval around r that has a given probability of containing Ď .

Table 1. Adopted weights for labor and capital productivity.

Airlines (a0) North American 0.66 European 0.72 Asian 0.57 South American 0.61 a0: labor weight; b0: capital weight. Source: IATA (2001).

(b0) 0.34 0.28 0.43 0.39

Table 2. Pearson correlation intervals.

Correlation None Small Medium Large

206

Negative -0.09 to 0.0 -0.3 to -0.1 -0.5 to -0.3 -1.0 to -0.5

Positive 0.0 to 0.09 0.1 to 0.3 0.3 to 0.5 0.5 to 1.0

The information and data like medium stage length, load factor, hours flown, airplane model configuration, number of employees, for the period of 2000-2005 were collected from international and Brazilian publications: World Air Transport Statistics (IATA), the Digest of Statistics (ICAO); Fleet and Personnel Series (ICAO), the Financial Data Series (ICAO) and the Brazilian National Civil Aviation Agency (ANAC) commercial aviation yearbook. Three categories of inputs were used: (a) labor, (b) capital, represented by flight equipment and (c) energy. Labor The labor productivity index is calculated as a multilateral index of 5 categories: pilots, co-pilots, other cockpit personnel, cabin attendants and other personnel. Output is composed of two separate components: scheduled revenue (passenger/km), and passengers transported. Flight equipment It is represented by the number of aircraft used to transport passengers and cargo. In the index of aggregate capital, the percentage change in the number of aircraft was considered, adjusting it by the number of seats offered, so as to take into consideration the size of aircraft. Energy The aggregate index of energy was constructed considering the percentage change in consumption of fuel (jet fuel, since only the fleet of jets was considered). Sampling criteria Forty-five carriers were selected and grouped as follows: â&#x20AC;˘

26 full service;

â&#x20AC;˘

7 low-cost/low fare; and

â&#x20AC;˘

12 regional airlines.

The airlines were sampled according to the following criteria: (i) the presence and importance of the airlines in their markets (North and South American, European and Asian airlines); (ii) carriers whose data availability and previous studies indicated good operational performance and productivity were chosen.

J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines

The airlines included in the sample are detailed below:

Production output

•

full service: Aeroflot, Aerolineas, Aeromexico, Air Canada, Air France, Alitalia, Austrian American Airlines, British Airways, China Southern Continental, Delta, Iberia, JAL, Korean, Lan, Lufthansa, Malev, SAS Singapore, Airlines, Swiss, TAP, Thai, Turkish Airlines, TAM and VARIG;

•

low-cost/low-fare: Air Berlin, Air Europa, America West, GOL, Jet Airways, Ryanair, and Virgin Express;

Output in the airline industry is comprised of passenger services, as measured by Revenue Passenger Miles (RPMs), and cargo services, as measured by ton x miles. Passenger miles are by far the largest component, making up more than 90% of total revenue, with the remainder attributable to ton x miles. Although the output measure does not account for changes in service quality, such as flight delays, some recent studies seem to indicate that these changes did not significantly affect output and productivity.

•

regional: Alaska, Nordeste, Oceanair, Pantanal, Passaredo, Penta, Portugalia Airlines, Rico, Riosul, TAF, Total, US Airways.

In the case of the Brazilian airlines, a survey was undertaken through field research to collect the necessary information and data via a questionnaire specially designed to include the main outputs and production inputs. This was sent by e-mail to: •

•

the four largest Brazilian airlines: TAM, GOL, VARIG, and WebJet; three of them operate in domestic and international markets, and one in the regional market; the Brazilian Regulatory Agency.

The single factor productivity of each of the researched companies was calculated: capital, energy and labor. These airlines single factor productivities were then compared with the explanatory variables like medium stage length, aircraft load factor, hours flown, aircraft size, aircraft seat configuration, cruising speed, and aircraft engine performance for selected routes. AIRLINE PRODUCTIVITY AND EXPLANATORY VARIABLES The purpose of this research was to understand the main variables which explain the air transport productivity, namely: labor, capital and energy productivity. These variables influence and are influenced by others, such as investment in personnel training, processes automation, aircraft load factor, flight stage length, fleet mix, among others. Some variables impact more than a single productivity factor. Investment in training of pilots, for instance, affects both labor and energy productivity. The flight stage length might influence both the capital and energy productivity. Airlines, however, have only limited control over some of these explanatory variables, as explained below.

An airline may increase or decrease its output level through management actions, but it is usually more influenced by economic conditions, such as the demand for passengers (over which they have no control). Average stage length This variable depends on the route, the market structure and the air network operated by the company which, in turn, depends on the country or territory extension served, the extent of regulatory control and the attitude of government towards bilateral agreements. Output composition or “output mix” This variable is strongly influenced by the geographic location, the regulatory control and the different demands placed on commercial airlines. In the case of Brazilian airlines, there are small variations in the output mix. Most of them transport passengers, with a smaller share of cargo and mail. Aircraft load factor Some researchers, among them those of the International Labor Organization (2001), argue that the load factor is largely determined by the market demand and the extent of control the airline has over the choice of the aircraft type and the flight frequency. These researchers argue further that an airline can only manage the load factor of its fleet by adjusting the flight frequency and the aircraft size, with permission of the regulatory authorities. Determinants of labor productivity The labor productivity is influenced, for instance, by the amount of investment in the training of crew members (pilots and co-pilots) and maintenance teams, the outsourcing of some functions and activities, and also

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

207

Araújo Jr, A.H.A. et al.

by the automation of some processes, such as computer ticketing. Duke and Torres (2005) reported that, although flight crew members (which include pilots and flight attendants) are highly visible employees in the airline industry (comprising about 30% of total employment in the industry), the majority of employees work in “ground occupations”. In addition to travel reservation agents and transportation ticket and customer service representatives, their occupations include aircraft mechanics, service technicians, and baggage handlers, among others. Yet according to Duke and Torres (2005), during the decade of 1990, employment growth in the air transportation industry slowed markedly to an average 1.8% per year. Employment declined by a slight 0.2% in 2001, which then dropped to a substantial 11.6 % in 2002. Part of the slowdown in the 1990’s was spurred by increased customer use of Internet web sites for air travel planning. These web pages became increasingly more sophisticated, allowing travelers to do almost everything related to their travel, from checking the status of their frequent-flyer accounts, to booking flights and selecting their own seats. With increased Internet use by customers, airlines have been able to reduce the number of customer service agents required to handle bookings and flight information questions. In addition to being able to book their own flights, once travelers arrive at the airports across the country, they can take advantage of the self-service kiosks provided by the airlines, which have grown in popularity since their introduction in 1995. These kiosks allow the passengers, for example, to get boarding passes, select seats, check baggage, and change flights. The increased use of self-service kiosks has given airline carriers the flexibility to lower their costs by using fewer employees at the airports. The outsourcing of certain functions and activities, particularly those that are not concerned with the core competence of airlines, have contributed to improving labor productivity (especially so in the airline industry), as they transfer to specialized firms the rationalization of activities and processes in pursuit of a reduction in operation and service costs, such as aircraft maintenance services, ground operation support and catering. One of the first outsourced activities in the airline industry was food preparation. While 10 years ago most airlines produced and distributed their own food on board, according to the study of the International Labor

208

Organization (2001), currently only 2 companies control around 60% of the catering market (with annual revenues of US$ 11 billion). Aircraft maintenance is currently undergoing a restructuring process. A growing number of carriers are hiring service and selling out maintenance workshops and equipment. The new technologies required for the maintenance of modern aircrafts make this activity extremely costly and a highly specialized business. Sophisticated aircraft models that require less frequent maintenance, make it increasingly difficult for an airline, individually, to justify high investments in workshops and equipment. Currently, 75% of aircraft maintenance, according to the International Labor Organization (2001), is undertaken by airlines, while the rest is performed by specialized firms or by aircraft manufacturers. Maintenance of engines (a more specialized service), is performed, in most cases, by the manufacturers. The IATA, in its annual report (2001), forecasts for the coming years increased outsourcing for ground handling services. Currently, 75% of these services are performed by airports or airlines. In 2010 (it is estimated that) 50% of this US$ 27 billion business will be in the hands of specialists. A global company was created with the sale of GlobeGround (a Lufthansa subsidiary), to the French Penauille Polyservices, which operates in 199 airports and 39 countries, employing over 30,000 employees. The automation of some processes (such as office activities and ticketing) is another important factor influencing labor productivity. Reservation systems and computerized ticketing were shared between different companies by cost and emission time reductions. According to the International Labor Organization (2001), “the Internet and aviation were made for each other. Flights are expensive highly perishable products and the information via Internet, can be quickly available to customers”. Airlines have another important reason for adopting the Internet: to generate savings in marketing and distribution costs, that are currently responsible for 25% of the operating expenses. The Internet has enabled, in 2001, according to IATA (2001), to generate savings of up to 5% on tickets sales, eliminating the printing and distribution costs of tickets and also computer reservation fees (approximately US$ 11 per ticket), thus reducing labor. The IATA, in its 2001 annual report, estimates that electronic ticketing (“e-ticketing”) is already generating savings to airlines, every year, of about US$ 1 billion

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines

in distribution costs. The e-ticketing of airlines has now the largest sales volume on the Internet. Although the electronic sales represent a share of 5% to the conventional airlines in the United States, to some US low cost carriers, they already account for 90% of total sales. Determinants of capital productivity The productivity of capital is strongly influenced by the way the airlines operate their flight equipment, which, in general, represent their most important asset. The capital productivity is affected by variables such as aircraft seat density, fleet composition or mix, aircraft load factor, use of aircraft, and flight stage length. It can be recognized in a simplified form that the main cost factors of air transport are represented by labor, depreciation and leasing of aircraft and fuel consumption, which had, in 2002, considering the case of Brazilian airlines, an average share of 72% of the direct costs. This proportion has had practically no change within the period between 2000-2005. The main fixed asset item of carriers is represented by flight equipment. The fixed assets of Brazilian airlines represented, on average, around 40% of the total assets in 2004. The main variables that impact the capital productivity of airlines are: - average seating configuration of aircrafts, an important measure implemented by the airlines to improve productivity has been the increase of seats per aircraft. American companies (since the beginning of the 1990’s) have increased the seating configuration in trunk lines. Average seat numbers per aircraft increased by 15% in the United States, according to the Civil Aeronautics Board (between 1989 and 1994). Contrary to the trend observed in the American carriers, the exact opposite occurred in the case of Brazilian aviation between 1995 and 2002: there was a decrease in the aircraft seat density by 16%;

Large load factors indicate an efficient use of the aircraft and crew, leading to favorable economic results. Douglas and Miller (1974) examined the relationship between aircraft load factor, route length and route density, using data from the US market in 1969, and concluded that the aircraft load factor was negatively correlated with the medium stage length, i.e. it increased with the decrease of flight distances. This is exactly contrary to what would be consistent with the economic theory of welfare maximization, but consistent with the theory that airline deregulation forced the carriers to offer a capacity excess. Bailey, Graham and Kaplan (1985) analyzed the change of this relationship after the US market deregulation. When comparing the statistics of this study, it became evident that this relationship had changed over time, as predicted by theoretical studies. The aircraft load factor grew with the increase in the flight stage length, according to Bailey, Graham and Kaplan (1985) based on US market data for the period between1976-1981 period, exactly the opposite of what occurred during the regulation period. A large portion of the costs of the airlines is fixed costs, such as crew wages and aircraft depreciation aircraft leasing. The better occupation of the aircraft reduces unit costs (unit costs per passenger). In a regulated market, a load factor increase is very difficult to implement, as an airline depends on authorization from the regulators to eliminate flights. The American experience has shown that the fleet load factor has grown considerably since deregulation occurred in the late 1970’s, due to greater pricing flexibility permitted by the regulator, and the freedom of airlines to match supply and demand. Also in Brazil, the load factor of airlines increased for the period of 2000-2005, whereby the average load factor of Brazilian airlines reached 62.1% within this same period;

- fleet composition or fleet mix: Brazilian carriers, similarly to the American ones, have substantially reduced the use of aircrafts that are less efficient in fuel consumption. The increase of aviation input prices (especially fuel) have forced this procedure. The fleet adequacy in terms of aircraft size, efficiency and engine output has contributed to the increase in the capital and energy productivity, and, consequently, in the multi-factor productivity;

- use of aircraft: the operating objective of airlines is to use aircrafts more intensively by increasing the number of flight hours/day. The American literature based on reports of The Civil Aeronautics Board states that the equipment utilization rate increased by an average of half an hour per day in the post-deregulation period. Within Brazilian aviation (between 1995 and 2002), there was an absolute increase in the number of hours flown, despite the reduction in the number of hours flown per aircraft;

- fleet/aircraft operation: one of the variables with the strongest impact on the productivity of air transport is the aircraft load factor, which represents the relationship between demand and supply of passenger or freight.

- average stage length: the flight stage is one of the operating parameters that most influence the unit cost and productivity of an airline. Airlines with flight stage beyond average have lower operating costs per unit of production.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

209

Araújo Jr, A.H.A. et al.

A rapid decline in unit costs, with the increase in the average flight, is a characteristic of air transport. This is due to the fact that airport charges and other associated costs such as fees for landing and takeoff are fixed, regardless of the flight distance. Therefore, a larger flight stage length has as result a better use of aircraft and crew. In the case of US firms, the experience has shown that those in pursuit of operational efficiency in both trunk routes (trunk carriers) and “feeder” routes (local service carriers) have increased the proportion of long distance flights in order to increase their operational efficiency. Larger stage length means, in practice, more efficient use of flight equipment by reducing proportionally to the distance traveled the fuel consumption, since the largest specific consumption occurs during takeoff and landing. Doganis (1985) states that an aircraft burns a significant amount of fuel during the aircraft maneuver on the ground, the landing and takeoff (on average, 20 to 30 minutes). During takeoff and on a smaller scale on landing, the fuel consumption is high (relative to the distance traveled horizontally). Ground maneuvers, takeoffs and landings become proportionately smaller when the medium flight stage length increases. The Canadian experience, as reported by Oum and Yu (2001), showed that an airline with a 10% longer flight stage length had its multi-factor productivity increased in the order of 1.7%. Determinants of energy productivity Doganis (1985) affirms, by examining the determinants of air transport costs, that the main variables that influence fuel consumption and, consequently, the energy productivity of an airline are: - cruise speed: the cruising speed of an aircraft affects its operating cost, regardless of its size. This effect can be expressed in terms of its hourly productivity. Since the hourly productivity of an aircraft is the product of its payload in ton times its speed, the higher the cruising speed, the greater the production and the productivity per hour. As, in practice, faster aircraft are also larger, the advantages of speed and size reinforce each other; - aircraft size: some technological aspects have a direct effect on productivity and operating costs of each type of aircraft. Most importantly, the economic point of view is probably the size of aircraft, its cruising speed and flight range with full payload. The significance of size, speed and range of an aircraft is reinforced by the 210

fact that these variables affect its hourly productivity, which in turn affects its operational costs. As a general rule, the larger the aircraft the lower the operating costs per unit of production, i.e. per ton x mile or passenger x mile. The operating cost per hour flown of a larger aircraft will be higher than that of a smaller aircraft, but this cost will be even lower when converted to cost per seat-km or tonne-kilometers. Doganis (1985) states that the size of an aircraft affects cost and productivity in two ways: the larger aircraft has a proportionally lesser aerodynamic drag, allowing it to carry more pay-load per unit of weight. At the same time, larger aircrafts use larger and more efficient engines; - engine performance: the basic characteristic of an aircraft is its engine. The same type of engine may have different performances on different aircrafts and routes. The performance of an engine also depends on variables beyond the operating control of the airline: altitude and temperature of airports served, flight stage length, aircraft aerodynamics, cruising altitude etc. The type of aircraft operated has a significant effect on the operating costs. Taking into account this premise, the key question is the extent to which an airline is free to select the type of aircraft it wants to operate, or to what extent the choice is conditioned by the extent and density of traffic in its routes. Since the company made the choice of aircraft and its engine for the different segments of its transport network, and due to high investment in maintenance, facilities, training of pilots, engineers and mechanics, it is unlikely to replace it in short-term period. The correlation analysis between the single productivity factors and the explanatory variables of an airline is detailed in the following section. CORRELATION BETWEEN SINGLE-FACTOR PRODUCTIVITY AND THEIR EXPLANATORY VARIABLES It is intended to identify and understand in the context of this research the main variables that explain the multifactor productivity and the single-factor productivity of airlines (e.g. labor, capital and energy). We calculated the correlation matrices for each company separately and a joint matrix for the complete sample of airlines studied. Table 3 shows the correlation between the single-factor productivity of airlines and the different explanatory variables studied.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines

These single-factor productivities (labor, capital and energy) are dependent on other variables such as investment and training of crew members and maintenance teams, outsourcing of activities and processes, automation of administrative and operational processes, average seating configuration of operating fleet of an airline, load factor, cruise speed, stage length, among others (Fig. 1).

both labor and energy productivity. The flight stage length influences both capital and energy productivity.

Some variables affect more than a productivity factor. The investment in the training of pilots, for instance, influences

The study showed that the correlation capital productivity versus cruise speed is larger among carriers with larger

As it can be seen in Table 3, there is, in descending order, a large positive correlation, according to the intervals defined in Table 1, between capital productivity and cruise speed (r2=0.9405), capital productivity growth and seat configuration/density (r2 = 0.9062).

Table 3. Correlation matrix between single-productivity factor of airlines and its main explanatory variables.

(km)

(%)

(hours flown)

Seat configuration (seats/aircraft)

Capital productivity

0.3039

0.6575

0.6320

0.9062

0.9405

Energy productivity

0.4774

0.2623

0.4033

-0.8229

-0.8764

Labor productivity

Very low correlation

0.2397

Very low correlation

Stage length Load factor

Single-productivity variables

Aircraft use

Cruise speed

Employees

(km/h)

(unit) Very low correlation Very low correlation -0.6078

Explanatory variables Training

Labor

Employees (-0.6068)

(Exogeneous variables) Marketing influence

Process automation Seat config. (+0.9062) Output composit. Multifactor productivity

Capital

Competitiveness

Load factor (+0.6575)

Route length

Aircraft use (+0.6320)

Route density

Stage length (+0.3039)

Cruise speed (-0.8764) Energy

Seat config. (-0.8229) Engine performance

Figure 1. Productivity model for scheduled airlines, deduced from the collation and analysis of labor, capital and energy productivity and their explanatory variables. Config.: configuration; composit.: composition. J. Aerosp.Technol. Manag., SĂŁo JosĂŠ dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

211

Araújo Jr, A.H.A. et al.

average stage length and, in the case of the correlation capital productivity versus aircraft size (seats/aircraft), the larger correlation occurs among airlines with larger aircraft size. There is also a large positive correlation between capital productivity and load factor (+0.6775), and capital productivity versus aircraft utilization in hours flown (+0.6320). In the case of the productivity of energy, a large but negative correlation between energy productivity and cruise speed (-0.8764) and energy productivity and seat density (-0.8229), and between labor productivity and number of employees (-0.6078) was verified. The study showed also in the case of the correlation energy productivity versus cruise speed that this correlation is larger among airlines operating with lower cruise speed that can be explained by the aircraft engines consumption, which increases proportionally higher with the increasing speed of aircraft. A medium positive correlation, according to the criteria defined in Table 2, can be inferred from energy productivity growth and stage length (+0.4774) and energy productivity growth and aircraft utilization (+0.4033). And, finally, a small positive correlation is verified between energy productivity growth and load factor (+0.623) and labor productivity and load factor (+0.2397). The correlation labor productivity versus number of employees of airlines was negative (-0.6078), which was expected. Regarding the correlation of productivity labor and other variables like stage length, load factor, aircraft utilization, aircraft size and cruising speed, it can be considered low (correlation with load factor) and very low. The analysis of the determinants of the single productivity factors labor, capital and energy and their explanatory variables led us to the conceptual model as shown in Fig. 1, which reproduces the inter-relationship between the main productivity elements labor, capital and energy and their explanatory variables. The numbers in brackets, in Fig. 1, represent the Pearson correlation between explanatory variables and the respective single-factor productivity, as shown in Table 3. CONCLUSION The survey was conducted with 41 international airlines within the categories Full Service Companies (FSC), Low Cost/Low Fare (LCC) and Regional Companies (RC) 212

between 2000 and 2005 (and as part of this sampling, the four major Brazilian airlines were included). Kendrick’s productivity method was used to measure the multiple-factor productivity growth of linear dependence between the single-factor productivity of labor, capital and energy and the productivity explanatory variables of the airlines analysed. The results of the research confirmed conclusions from analysis carried out by other researchers such as Bailey, Graham and Kaplan (1985), Douglas and Miller (1974) and Oum and Yu (2001). The largest positive correlation was verified between capital productivity and cruise speed (correlation of 94%), and capital productivity growth and seat configuration (correlation of 90%), which has also been confirmed by Kune, Mulder and Poudevigne (2000) in the conclusion: “that capital is a key production factor in the airline industry and a large part of the improvement of this economic sector depends on investments in infrastructure and equipment”. Kune, Mulder and Poudevigne (2000), and Windle (1991) also identified among the most important explanatory variables of airlines productivity flight and ground equipment and materials. The largest negative correlation was found between energy productivity and cruise speed (correlation of 88%), and between labor productivity and number of employees (correlation of 60%), confirming Oum and Yu’s study (2001), that identified, among the important productivity explanatory variables, the average stage length and the average load factor. Acknowledgements The authors would like to thank the National (Brazilian) Research and Development Council (CNPq, Conselho Nacional de Pesquisa e Desenvolvimento) for Project Research support (Grant no. 154 203/2006-8), which made this research possible. REFERENCES Bailey, E.E., Graham, R.D. and Kaplan, D.P., 1985, “Deregulating the Airlines”, MIT Press, Cambridge, Massaschussets. Caves, D.W., Christensen, L.R. and Diewert, 1982, “Multilateral comparisons of output, input, and productivity using superlative index numbers”, The Economic Journal, Vol. 92, p.73- 86.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

Identification and analysis of explanatory variables for a multi-factor productivity model of passenger airlines

Cohen, J., 1988, “Statistical power analysis for the behavioral sciences”, 2nd ed., Routledge Academic, London. Distexhe, V., Perelman, S., 1994, “Technical efficiency and productivity growth in an era of deregulation: the case of airlines”, Swiss Journal of Economics and Statistics, Vol. 130, No. 4, p. 668-669. Doganis, R., 1985, “Flying off course: the economics of international airlines”, New York Press, New York. Douglas, G.W., Miller C.J., 1974, “Economic Regulation of Domestic Air transport: Theory and Policy”, The Brookings Institution, Washington D.C. Duke, J, Torres, V., 2005, “Multifactor Productivity Change in the Air Transportation Industry”, Monthly Labor Review, p. 32-45. Encaoua, D., 1991, “Deregulating European Airlines”, International Journal of Industrial Organisation, 1991, Vol. 9, p. 61-81. Färe, R., Grosskopf, S. and Sickles, R.C., 2001, “Productivity of U.S. airlines after deregulation”, Department of Agricultural and Resource Economics, Oregon State University. Good, D.H., Nadiri, M.I., Röller, L.H. and Sickles, R.C., 1993, “Efficiency and Productivity Growth Comparisons of European and US air carriers: a first look at the data”, The Journal of Productivity Analysis, Vol. 4, p. 115-125. International Air Transport Association (IATA), 2001, “Annual Report (various years)”, Geneve, Switzerland.

International Labor Organization (ILO), 2001, “Restructuring of Civil Aviation: Consequences for Management and Personnel”, Reference Document, Geneve, Switzerland. Kendrick, D.A., 1996, “Handbooks in Economics, Sectoral Economics, Chapter 6”, Vol. 1, 13, Elsevier, Amsterdam, Netherlands. Kune, B.C., Mulder, N., and Poudevigne, P., 2000, “La mesure du capital et de la productivité dans les Transports: le cas du transport aérien”, Centre d’Etudes Prospectives et d’ Informations Internationales, Notes de Syntheses du Sen, Paris, France. McKinsey Global Institute, 1992, “Service Sector Productivity”, McKinsey and Company Inc., Washington, D.C. Moreira, D.A., 1994, “Os benefícios da Produtividade Industrial”, Pioneira, São Paulo, Brazil. Oum, T.H., Yu, C., 2001, “Assessment of Recent Performance of Canadian Carriers: Focus on Quantitative Evidence for evaluating”, Report to the Canada Transportation Act. Sickles, R.C., Good, D.H., Getachew, L., 2002, “Specification of Distance Functions Using Semi- and Nonparametric Methods with an Application to the Dynamic Performance of Eastern and Western European Air Carriers”, Journal of Productivity Analysis, Vol. 17, p. 133-155. Windle R. J., 1991, “The World’s Airlines: Cost and Productivity Comparison”, Journal of Transport Economics and Policy, Vol. 25, No. 1, p. 31-49.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 203-214, May-Aug., 2011

213

doi: 10.5028/jatm.2011.03021211 Paloma Ribeiro dos Santos*

Instituto Tecnológico de Aeronáutica São José dos Campos/SP - Brazil paloma@ita.br

Rocio Soledad Gutierrez Curo

Instituto Tecnológico de Aeronáutica São José dos Campos/SP - Brazil rocio@ita.br

Mischel Carmen Neyra Belderrain

Instituto Tecnológico de Aeronáutica São José dos Campos/SP - Brazil carmen@ita.br *autor para correspondência

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil Resumo: Atualmente, as empresas têm interesse por adotar novas estratégias metodológicas, colocando em prática a cognição administrativa como ferramenta de crescente interesse para captar a natureza dos seus problemas. O presente trabalho teve como objetivo aplicar o mapa cognitivo como apoio à decisão a partir dos depoimentos extraídos do trabalho abordado por Silva (2008), aplicado a um problema de decisão em grupo da Força Aérea Brasileira (FAB) sobre quais ações adotar como prioritárias, inseridas no contexto entre comprar ou desenvolver tecnologias para o Setor Aeroespacial de Defesa do Brasil. O artigo tem como principal contribuição, a estruturação do problema do setor de Defesa Brasileiro, indicando os critérios mais importantes que devem ser considerados para avaliação das possíveis ações. Palavras-chave: Mapa cognitivo, Tomada de decisão, Setor aeroespacial de defesa.

Cognitive mapping applied to Brazilian aerospace decision problem Abstract: Currently, companies are interested in adopting new methodological strategies, using the administrative cognition as a tool of growing interest for grasping the nature of their problems. The present work aims to apply the cognitive map as decision support based on the statements taken from the work discussed by Silva (2008), applying to the Brazilian Air Force (FAB) group decision problem, about what actions to take as a priority, set in the context of either buying or developing technologies for the aerospace defense of Brazil. The contribution of this paper is a Brazilian defense industry problem structuring deﬁ ning the relevant criteria that should be considered for the evaluation of possible actions to be taken by the Defense Sector in Brazil. Keywords: Cognitive mapping, Decision making, Aerospace sector of defense.

INTRODUÇÃO A tomada de decisões por indivíduos ou grupos, considerando aspectos de desenvolvimento científico ou tecnológico, mostra-se difícil de desenvolver-se por possuir objetivos que comumente são conflitantes. Portanto, o que faz uma situação problemática para os tomadores de decisão é o sentimento de pressão para escolher uma decisão em que o problema ainda não está claro para eles. (Keeny e Raiffa, 1999 apud Gomes, Rangel e Jerônimo, 2010). O mapa cognitivo é uma ferramenta para estruturar problemas de tomada de decisões (Gomes, Rangel e Jerônimo, 2010), e para organizar e representar o Received: 28/03/11 Accepted: 21/06/11

conhecimento (Novak e Cañas, 2008). Este mapa proporciona assistência na explicação dos objetivos estratégicos e na identificação dos valores fundamentais dos tomadores de decisão (Montibeller Neto, 1996). Esta ferramenta tem sido aplicada amplamente em vários tipos de problemas: pesquisa em administração de projetos (Edkins et al., 2007), acadêmicos (Rieg e Araujo Filho, 2003), meio ambiente (Jardim e Silva, 2009), entre outros. O problema abordado neste trabalho está inserido na decisão de comprar ou desenvolver tecnologias para o setor Aeroespacial de Defesa do Brasil. Através da estruturação do problema, o presente trabalho objetivou levantar critérios, chamados de pontos de vista fundamentais (PVF) que possam servir de base para adotar medidas que contribuam com o progresso do país neste setor.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

215

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

O artigo apresenta a identificação do contexto decisório e o uso do mapa cognitivo no foco do conceito, na construção de mapas cognitivos individuais, mapa agregado e congregado, e por último, a análise do mapa congregado. O resultado desta aplicação é a definição dos critérios a serem considerados no processo de tomada de decisão. A metodologia considerada para a elaboração dos mapas cognitivos foi a proposta por Ensslin, Montibeller Neto e Noronha (2001), com uma modificação na geração dos conceitos, pois estes foram extraídos de textos e não de entrevistas pessoais. FUNDAMENTAÇÃO TEÓRICA

Mapas Cognitivos

Identificação do contexto decisório Para a identificação do contexto decisório, é necessário conhecer os atores envolvidos no processo de decisão, a escolha dos decisores, a definição das ações disponíveis e da problemática de referência (Ensslin, Montibeller Neto e Noronha, 2001). Os autores Roy (1981; 1996) e Bana e Costa (1995) explicam os tipos de problemáticas de referência, as quais podem ser: •

Problemática da escolha (Pα): esta abordagem visa selecionar uma alternativa ou um conjunto de alternativas dentro de várias propostas.

•

Problemática da alocação em categorias (Pβ): se classificam as alternativas em categorias pré-definidas disponíveis, de acordo com elementos semelhantes de classificação delas.

•

Problemática da ordenação (Pγ): é utilizada quando há o propósito de estabelecer a prioridade das alternativas existentes.

•

Problemática da descrição (Pδ): descreve formalmente as alternativas e suas características.

•

Problemática da rejeição absoluta (Pβº): reduz o número de ações a serem avaliadas e permite considerar aspectos que não são compensatórios.

A definição de qual problemática de referência usar depende do tipo de problema que os decisores têm, influenciando na definição de quais critérios serão utilizados. Depois da exploração do contexto decisório, pode-se começar o desenvolvimento dos mapas cognitivos. Estruturação do problema A técnica dos mapas cognitivos é muito usada para estruturação de problemas complexos em grupo. O 216

objetivo desta estruturação do problema não é alcançar um consenso entre os decisores sobre a base de conhecimento e critérios normativos, mas que os participantes alcancem um acordo sobre a formulação do problema, suas soluções e resultados significantes. Neste processo de estruturação, o facilitador tem tarefa importante em ajudar aos tomadores de decisão a explorar seus pensamentos sobre o problema e expressar seus pontos de vista, mesmo divergentes, proporcionando um ambiente criativo onde os objetivos de cada decisor são considerados e usados na avaliação das opções de decisão (Franco e Montibeller, 2010b).

Os mapas cognitivos são representações gráficas, resultados da interpretação mental sobre um problema, baseados na teoria da construção da personalidade e que compreende como os seres humanos pensam e raciocinam a respeito de sua experiência (Kelly, 1955). Um mapa cognitivo é definido como uma hierarquia de conceitos, relacionados por ligações de influência entre conceitos meios e fins (Montibeller Neto, 1996), e é usado para estruturar, analisar e dar sentido aos problemas (Ackermann, Éden e Cropper, 2004). Hart (1977) explica que os mapas cognitivos podem ser realizados por meio de três formas: sistematicamente da codificação de documentos que representam os escritos ou declarações do indivíduo; da codificação das transcrições integrais das reuniões privadas em que o indivíduo seja um participante; e ao extrair das crenças causais através de questionários ou entrevistas. De forma similar, Ackermann, Éden e Cropper (2004) disseram que o mapa cognitivo pode ser construído através de transcrições de entrevistas ou outros documentos que permitam questionar, analisar e entender as informações; Iederan et al. (2011) coincidem em considerar transcrições das entrevistas analisando-as de acordo a cinco dimensões: organização, processos, causas, obstáculos e consequências. Para construir um mapa cognitivo formalmente, optouse por seguir a sequência de etapas descrita por Ensslin, Montibeller Neto e Noronha (2001): •

Definição de um rótulo para o problema: encontrar o nome que descreva o problema em questão, de modo que os decisores o considerem adequado, delimitando assim o contexto decisório de acordo com os aspectos mais relevantes envolvidos com a resolução do problema do decisor.

•

Definição dos elementos primários de avaliação (EPA): servem como início para o desenvolvimento do mapa e estão constituídos por objetivos, valores

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil

dos decisores, metas, ações, opções e alternativas. Quanto mais EPA, mais rico será o mapa. •

Construção de cada conceito a partir dos EPA: cada conceito que é construído a partir do EPA está baseado nas ações que o contexto sugere, explicitando-o em verbo infinitivo. Cada conceito deve ser o mais abreviado possível, buscando sempre manter as palavras e frases utilizadas pelo decisor.

•

Construção da hierarquia de conceitos: após definir os conceitos, é importante relacioná-los, por meio de ligações de influência, aos fins aos quais eles se destinam, e aos meios para se chegar a estes conceitos, fazendo as seguintes perguntas: “Por que esse conceito é importante?” e “Como você poderia obter tal conceito?”, respectivamente.

Os mapas cognitivos usados para estruturar um problema de decisão em grupo, são elaborados seguindo regras específicas, considerando os passos necessários descritos por Eden e Ackermann (1998), Ensslin et al. (1998), Montibeller Neto (1996) e Bana e Costa (1992). Após a construção dos mapas cognitivos individuais pelas pessoas envolvidas no problema em questão, esses mapas devem ser agrupados pelo facilitador, mediante a união dos conceitos similares que transmitam as mesmas idéias, ligando os conceitos que sejam relacionáveis, formando assim, um mapa agregado. Por exemplo, a Fig. 1 mostra a representação de quatro mapas cognitivos individuais representados como peças, que são unidos para formar uma só peça. A agregação desses mapas é apresentada ao grupo ou aos representantes dos tomadores de decisão, pelo facilitador. Estas pessoas vão adicionar; modificar ou reduzir os conceitos que considerem não relevantes para o problema em questão, e relacionarão os conceitos por meio das ligações de influência, conduzindo a

um sistema de valores enriquecido, que representa a estrutura cognitiva do grupo, também denominado mapa congregado. Uma vez que o mapa cognitivo congregado foi construído, procede-se à avaliação do mesmo. Segundo Ensslin, Montibeller Neto e Noronha. (2001), para que se possam identificar os PVF do problema, é necessário realizar uma análise da estrutura e do conteúdo do mapa cognitivo, identificando mediante a observação, a forma do mapa, as linhas de argumentação, determinando através da análise de conteúdo destes, os ramos, e assim gerando um eixo de avaliação do problema. A análise do problema conforme Ensslin, Montibeller Neto e Noronha (2001) tem duas etapas: tradicional e avançada. A primeira análise permite a compreensão do mapa cognitivo e a gerenciar sua complexidade, e a segunda possibilita identificar os eixos da avaliação do problema, levando em conta a forma e o conteúdo do mapa cognitivo. Dessa maneira, as duas análises cumprem uma sequência e se complementam entre si. Análise tradicional Na análise tradicional têm-se as seguintes observações: •

Hierarquia de conceitos meio e fim: os mapas são formados por conceitos meio e fim. Tomando como exemplo dois conceitos que se relacionam através de uma flecha, o conceito que recebe a flecha é um conceito fim, e o conceito que origina a flecha é um conceito meio. Por exemplo, na Fig. 2, o conceito C1 é um conceito fim para os conceitos C5 e C6, pois se caracteriza como um objetivo para estes conceitos. Já os conceitos C5 e C6 são conceitos meios, ou seja, são maneiras de se atingir o conceito fim. O conceito C1 por sua vez, é um conceito meio para o conceito C4. Portanto, a análise dos conceitos meios e fins

Indivíduo 2 Indivíduo 1

Representação coletiva Indivíduo 4

Indivíduo 3 Figura 1. Representação de quatro mapas individuais formando um mapa agregado. Fonte: adaptado de Bouzdine-Chameeva, Durrieu e Mandják (2001). J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

217

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

C4 Cluster III

Cluster I

C3 C14

C11

C12 C13

C15

C16 C17

C10

Cluster II

C9 Figura 2. Representação dos clusters. Fonte: Lima (2008).

permite compreender as relações existentes entre os meios disponíveis aos decisores e os fins que eles aspiram alcançar, sendo os meios uma forma de obter os fins respectivos. •

•

218

Conceitos cabeça e cauda: os conceitos cabeças do mapa são aqueles que não têm flechas saindo deles, somente chegando neles, revelando os objetivos, fins, resultados e valores fundamentais dos decisores. Como exemplo, na Fig. 2 só existe um único conceito cabeça: o conceito C4. Em contrapartida, os conceitos caudas são aqueles que têm flechas saindo deles, mas não chegando neles, mostrando os meios, ações, alternativas e opções do mapa, como exemplificado nos conceitos C9, C6, C13, C12 e C17 da Fig. 2. Laços de realimentação: consistem em um conceito meio que influencia um conceito fim, que por sua vez também influencia aquele mesmo conceito meio, ocasionando uma retroalimentação dos conceitos. Não são bem-vindos no mapa, pois ocasionam uma falha na hierarquia dos conceitos. Clusters: para identificá-los visualmente, se agrupam os conceitos que estão relacionados de acordo com a área de interesse do decisor e ao conteúdo dos conceitos, segundo a visão do facilitador. Dessa maneira, cada cluster pode ser analisado como um mapa independente, reduzindo a complexidade do mapa cognitivo original, e permitindo fazer uma análise do conteúdo separadamente, como ilustrado na Fig. 2.

Análise avançada Segundo Ensslin, Montibeller Neto e Noronha (2001), depois de identificar a hierarquia de meios e fins, os conceitos cabeças e caudas, os laços de realimentação e clusters, é importante analisar: •

Linhas de argumentação: constituídas por um conjunto de conceitos que são influenciados e hierarquicamente superiores a um conceito cauda. Uma linha de argumentação começa com um conceito cauda e termina em um conceito cabeça que é um fim para aquele conceito meio. Para a Fig. 2, por exemplo, são identificadas seis linhas de argumentação: (C9– C7–C5–C1–C4), (C9–C10–C8–C5–C1–C4), (C6– C1–C4), (C13–C11–C2–C4), (C12–C2–C4), (C17– C16–C15–C14–C3–C4).

•

Ramos: constituídos por uma ou mais linhas de argumentação que demonstrem preocupações similares sobre o contexto de decisão.

Depois de realizada a análise avançada é necessário encontrar no mapa o conjunto de PVF. Um PVF é uma razão essencial de interesse na situação, sendo um fim em si mesmo. Para que um ponto de vista seja considerado fundamental é indispensável que exista uma vontade consensual entre os atores intervenientes no processo de tomada de decisão, de submeter as ações a uma avaliação parcial segundo os aspectos elementares que formam o PVF; e o desenrolar do processo de estruturação confirme a validade da hipótese de independência que os

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil

atores afirmam existir (Corrêa, 1996). Esses PVF poderão ser usados como critérios nos métodos multicritérios (Escobar-Toledo e López-García, 2003), identificando-se as áreas e setores de maior relevância a serem consideradas na prioridade do contexto situacional. Ensslin, Montibeller Neto e Noronha (2001) explicam que para conhecer os PVF e construir a árvore dos PVF, devese verificar o cumprimento das seguintes propriedades: •

Essencialidade: devem representar consequências fundamentalmente importantes de acordo com os objetivos dos decisores;

•

Controlabilidade: devem ser influenciados apenas pelas ações potenciais em questão;

•

Mensurabilidade: devem permitir a medida do desempenho das ações potencias de acordo com os aspectos que os decisores consideram como fundamentais;

•

Operacionalidade: devem possibilitar a aquisição de informações sobre o desempenho das opções de decisão;

•

Isolabilidade: devem permitir a análise de forma independente com relação aos demais aspectos do conjunto de PVF;

ser a de assegurar a defesa dos interesses vitais da nação contra qualquer ameaça estranha, defendendo com ênfase e determinação a fronteira entre os interesses vitais e os interesses estratégicos. Assim, o setor aeroespacial de defesa aborda o desenho, fabricação, comercialização e manutenção das aeronaves, naves espaciais e foguetes de defesa, sendo uma atividade relacionada ao setor econômico da indústria aeronáutica e espacial, que está estreitamente ligado às atividades de fornecimento de materiais militares. Segundo Silva (2008), o Brasil apresenta uma defasagem tecnológica desses materiais, relacionada aos recursos tecnológicos fundamentais para uma operação de defesa de acordo aos desafios que ao país se impõem. Além disso, o Brasil fica dependente da tecnologia importada em um mundo caracterizado por grandes incertezas e mudanças unilaterais de regras, não parecendo ser uma postura estratégica muito coerente. Cruz (2006) acrescenta que o desenvolvimento da dualidade tecnológica contribui para o fortalecimento e modernização do parque produtivo nacional, constituindo assim, um meio para incrementar o aporte de recursos destinados à pesquisa e desenvolvimento militar, e da mesma forma, contribui para aproximar a sociedade civil e as empresas dos problemas de defesa. Considerando as opiniões de Cambeses Júnior (2009), Silva (2008) e Cruz (2006) verifica-se que existe o interesse em proteger a nação, e que existe uma problemática quanto às decisões que envolvem o investimento das tecnologias utilizadas no setor aeroespacial de defesa brasileiro.

•

Compreensividade: devem ser de fácil entendimento aos decisores de modo que seja possível a geração e comunicação de idéias;

•

Não redundantes: cada PVF deve ser único, ou seja, ter objetivos considerados apenas uma vez;

Descrição do caso

•

Concisos: o conjunto de PVF deve ter o mínimo necessário para representar a visão dos decisores sobre o problema;

O presente trabalho se concentrou no problema de decisão do setor aeroespacial de defesa explicitada por Silva (2008), o qual trata de uma decisão entre comprar ou desenvolver tecnologias para esse setor.

•

Completos: devem incluir todos os objetivos considerados como fundamentais pelos decisores.

Alguns exemplos de autores encontrados na literatura que mencionam as propriedades dos PVF em seus trabalhos são Franco e Montibeller (2010a); Santos (2010); Piratelli e Belderrain (2010); Giffhorn, Ensslin e Vianna (2008); Fitz e Hasenack (2011), dentre outros.

Identificação do contexto decisório •

Atores envolvidos no processo decisório: os atores envolvidos no caso de estudo são os considerados na Tabela 1, os quais são os autores dos depoimentos escolhidos neste trabalho, bem como as funções que exerciam em 2008, ano de realização do trabalho de Silva (2008). Este mesmo autor Silva (2008) atuou como decisor neste trabalho, e os facilitadores são os autores do presente estudo.

•

Escolha dos decisores: como neste caso particular a estruturação do problema foi elaborada através de depoimentos de diversas pessoas de difícil contato,

ESTUDO DE CASO Segundo Cambeses Júnior (2009), o Brasil está interessado em defender, assumir responsabilidades e desempenhar um papel na segurança e defesa do país, em nível hemisférico e mundial. O primeiro objetivo da política de defesa deve

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

219

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

Tabela 1. Autores envolvidos no processo decisório

AUTOR DE DEPOIMENTO Samuel Pinheiro Guimarães Helio Jaguaribe Celso L. N. Amorim General Rui Monarca da Silveira Ozires Silva Roberto Amaral

FUNÇÃO (2008) Secretário-Geral das Relações Exteriores Sociólogo, membro do Instituto de Estudos Políticos e Sociais Ministro das Relações Exteriores General-de-Brigada, Subchefe do Estado-Maior do Exército Ex-ministro e fundador da Embraer Cientista político, Ex-Ministro de Estado da Ciência e Tecnologia

optou-se por utilizar o próprio autor Silva (2008) como decisor para validar o mapa agregado. •

Definição das ações disponíveis: comprar ou desenvolver tecnologias para o setor aeroespacial de defesa do Brasil.

•

Definição da problemática de referência: para este caso, a problemática é a de descrição, pois se restringe a estruturar o problema para descrever as alternativas e suas características.

Estruturação do problema Para melhor entender o problema de decisão, optou-se por utilizar o mapa cognitivo, identificando os objetivos e valores do problema encontrados em depoimentos de diversos profissionais, conhecedores da situação brasileira em termos de tecnologias aeroespaciais, os quais foram extraídos do trabalho de Silva (2008). A partir destes depoimentos, foram construídos os mapas cognitivos individuais, para em seguida, o mapa cognitivo agregado, o qual foi avaliado pelo autor Silva (2008) obtendo-se, desta forma, o mapa cognitivo congregado.

entre os conceitos, representando todas as ideias expostas pelos atores. Cada mapa individual foi ligado com os demais através de flechas indicando relacionamentos entre os conceitos, formando uma hierarquia de conceitos meios e fins. Para identificar as relações de influência entre os conceitos meios e fins, utilizou-se a técnica abordada por Ensslin, Montibeller Neto e Noronha (2001). 3° passo: construção do mapa congregado Após a construção do mapa agregado, o passo seguinte foi obter o mapa congregado. O mapa agregado recebeu algumas modificações e inclusões, devido à análise do decisor, transformando-se no mapa congregado que se encontra no Anexo 1. 4° passo: análise do mapa congregado O mapa congregado foi analisado na forma tradicional e avançada. Na análise tradicional, foram observados os seguintes aspectos: •

Hierarquia de conceitos meio e fim: para interpretar as relações de influência entre os conceitos, foi necessário analisar os conceitos meios e fins. No mapa congregado (Anexo 1), é possível identificar, através dos conceitos meios, as ações potenciais para se atingir os objetivos intermediários e o objetivo geral que está no topo do mapa. Como exemplo de análise de conceitos meios e fins, os conceitos: 19-Incorporar contingentes populacionais, 20-Modernizar setores industriais intensivos em capital e 21-Absorver tecnologias de ponta, são conceitos meios para atingir o conceito fim 18-Garantir estabilidade dos setores de alta tecnologia.

•

Conceitos cabeça e cauda: O único conceito cabeça do mapa congregado, “Empregar tecnologias essenciais para o setor aeroespacial da defesa”, está indicado com o número 1, no topo do mapa. Este conceito foi abordado como o objetivo principal que leva a decidir entre comprar ou desenvolver tecnologias para o setor aeroespacial de defesa do Brasil. Os conceitos caudas estão no mapa congregado do Anexo 1 com

Construção do mapa cognitivo 1° passo: construção de mapas individuais Para a construção do mapa cognitivo congregado foram escolhidos seis depoimentos contidos no artigo de Silva (2008). Estes depoimentos foram feitos sob a ótica de decisão entre comprar ou desenvolver tecnologias aeroespaciais. Para cada depoimento, um mapa cognitivo foi construído. Os conceitos relacionados com o problema de decisão foram retirados das declarações de cada depoimento. 2° passo: construção do mapa agregado O mapa agregado foi obtido com a junção dos seis mapas individuais através de ligações de influência 220

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil

os seguintes números: 7, 9, 10, 11, 13, 19, 22, 24, 27, 29, 34, 36, 42, 43, 46, 50, 52, 56, 57 e 59. Todos esses conceitos caudas também são conceitos meios, pois são maneiras de se atingir os objetivos estratégicos ou fundamentais dos decisores. •

•

Laços de realimentação: não foi encontrado nenhum laço de realimentação no mapa. Caso o mapa congregado apresentasse algum laço de realimentação, este deveria ser analisado e retirado conforme Ensslin e Montibeller Neto (1998) explicam. Apesar do tema do problema ser complexo, os conceitos dos mapas individuais foram analisados e hierarquizados de forma que não se apresentassem laços de realimentação no mapa congregado. Clusters: Foram identificados quatro clusters no mapa: Segurança, Desenvolvimento Tecnológico no Setor de Defesa, Preservação da Soberania Nacional e Parceria Internacional. Todos os clusters estão indicados com cores diferentes no mapa do Anexo 1.

Com esta separação do mapa por áreas, os decisores podem analisar o problema de forma mais organizada sabendo que para quais pontos dedicar maior atenção para resolver o problema. Depois da análise tradicional, foi realizada a análise avançada, verificando-se as linhas de argumentação e ramos identificados no mapa congregado, e são apresentas na Tabela 2. •

Linhas de argumentação: foram encontradas 28 linhas de argumentação, nomeados como: A1, A2, A3,..., A28.

•

Ramos: foram listados 10 ramos diferentes de acordo com o conteúdo dos conceitos, nomeados como: RA, RB, RC, RD, RE, RF, RG, RH, RI e RJ. São mostrados na Tabela 2 e Anexo 1.

Os conceitos seguintes identificados com a letra C seguida da numeração, são os candidatos a PVF encontrados no mapa. Os conceitos podem ser vistos na Fig. 3.

Tabela 2. Separação dos conceitos em clusters, ramos e linhas de argumentação

Cluster Segurança Segurança Segurança Segurança Segurança Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Desenvolvimento tecnológico no setor de defesa Preservação da soberania nacional Preservação da soberania nacional Preservação da soberania nacional Preservação da soberania nacional Parceria internacional Parceria internacional Parceria internacional

Ramo Linha de argumentação Sequência de conceitos (→) RA A1 11–8–6–2–1 RA A2 10–8–6–2–1 RA A3 9–6–2–1 RA A4 7–6–2–1 RB A5 13–12–2–1 RC A6 19–18–17–16–15–14–3–1 RC A7 22–20–18–17–16–15–14–3–1 RC A8 22–21–18–17–16–15–14–3–1 RD A9 24–23–3–1 RD A10 29–26–25–23–3–1 RD A11 27–25–23–3–1 RE A12 29–26–25–32–31–30–3–1 RE A13 27–25–32–31–30–3–1 RD A14 36–28–25–23–3–1 RE A15 36–28–25–32–31–30–3–1 RE A16 36–35–33–32–31–30–3–1 RE A17 34–32–31–30–3–1 RF A18 42–41–39–38–37–3–1 RF A19 42–41–40–38–37–3–1 RF A20 43–41–39–38–37–3–1 RF A21 43–41–40–38–37–3–1 RG A22 50–44–4–1 RH A23 50–49–48–47–4–1 RH A24 46–45–49–48–47–4–1 RI A25 52–51–47–4–1 RJ A26 57–55–54–53–5–1 RJ A27 56–55–54–53–5–1 RJ A28 59–58–54–53–5–1

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

221

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

Empregar tecnologias essenciais para o setor aeroespacial de defesa

Desenvolvimento tecnológico

Segurança

Aumentar segurança Militar

Indústria de equipamento e material militar

Preservação da soberania nacional

Medidas macroeconômicas apropriadas

Desenvolvimento científico tecnogólico - industrial

Parceria Internacional

Intercâmbio doutrinário entre países

Figura 3. Estrutura de árvore de PVF.

Cluster: segurança

C14 – ampliar a indústria de equipamentos e material militar.

Com o conjunto de PVF já definidos, é possível continuar o estudo focando nestes pontos como objetivos mais importantes do problema. Este trabalho se restringiu a estruturar o problema e chegar à estrutura de árvore de PVF, obtendo também uma estrutura hierárquica de um modelo multicritério. É importante considerar também que os PVF são considerados os critérios para a análise da decisão.

C38 – adotar medidas macroeconômicas apropriadas para ampliar a capacidade tecnológica.

CONSIDERAÇÕES FINAIS

C6 – aumentar a segurança militar. Cluster: desenvolvimento tecnológico no setor de defesa

Cluster: preservação da soberania nacional C44 – ter industrial.

desenvolvimento

científico-tecnológico

Cluster: parceria internacional C53 – aproveitar exercícios conjuntos e intercâmbios doutrinários entre países. Todos esses conceitos foram avaliados sobre o cumprimento das propriedades fundamentais para identificar os PVF. O Anexo 2 apresenta a análise de cada PVF sob a ótica de seis das nove propriedades que os definem como PVF. Além dessas propriedades, os PVF foram analisados para saber se eles eram não-redundantes, concisos e completos, completando assim as nove propriedades. Como resultado se obteve que todos eles cumprem as características estabelecidas por Ensslin, Montibeller Neto e Noronha (2001) e portanto são considerados PVF. Este conceito PVF é considerado análogo a critério quando a construção da estrutura hierárquica de um problema de decisão.

222

A aplicação do mapa cognitivo para estruturar um problema complexo através da extração de conceitos contidos em documentos escritos mostrou-se uma prática possível através deste estudo, confirmado pelos trabalhos de Ackermann, Éden e Cropper (2004) e Hart (1977). O uso do mapa cognitivo possibilitou alcançar o objetivo proposto mostrando-se como uma estratégia metodológica eficaz para explicitar a estruturação da problemática de decisão do Setor Aeroespacial de Defesa do Brasil. O mapa cognitivo foi usado como um instrumento de pesquisa baseada na perspectiva cognitivista através de conceitos mentais de forma individual e coletiva. Tais conceitos obtidos dos depoimentos escritos foram fundamentais na identificação das possíveis ações. A aplicação do mapa cognitivo em grupo, por meio da construção dos mapas cognitivos individuais, agregado e finalmente o congregado, permitiu descrever e analisar de forma conjunta os pontos de vista obtidos por meio dos depoimentos considerados. Os PVF levantados podem ser usados como critérios de avaliação de desempenho de ações potenciais para o mesmo problema em estudo futuro, no qual se poderá continuar o processo de apoio à decisão, utilizando uma metodologia multicritério.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil

Ademais, sugere-se também a continuação das aplicações do mapa cognitivo mediante documentações, em distintas áreas como apoio à estruturação de problemas de decisão em grupo.

Edkins, A. J., Kurul, E., Maytorena-Sanchez, E., Rintala, K., 2007, “The application of cognitive mapping methodologies in project management research”. International Journal of Project Management, v. 25, n. 8, pp. 762-772.

AGRADECIMENTOS

Ensslin, L., Montibeller Neto, G., 1998, “Quais critérios deve-se considerar em uma avaliação?”. ENEGEP Encontro Nacional de Engenharia de Produção.

Os autores agradecem a colaboração do professor pesquisador Luiz Maurício de Andrade da Silva, em participar deste trabalho atuando como decisor. Agradece-se também à Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) e ao Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) que são responsáveis pelo subsídio da pesquisa dos autores. REFERÊNCIAS Ackermann, F., Eden, C., Cropper, S., 2004, “Getting started with cognitive mapping”. Disponível em: http:// pkab.wordpress.com/2008/01/19/getting-started-withcognitive-mapping/, Acesso em: 09 de agosto de 2011. Bana e Costa, C. A., 1992, “Structuration construction et exploitation d’un modele multicritére d’aide à la decision”. Tese, Instituto Superior Técnico, Universidade Técnica de Lisboa, Lisboa, Portugal. Bana e Costa, C. A., 1995, “Processo de apoio à decisão: problemáticas, atores e ações”. Apostila do Curso de Metodologias Multicritério em Apoio à Decisão. ENE, UFSC, Florianópolis. Bouzdine-Chameeva T., Durrieu F., Mandják T., 2001, “Cognitive mapping methodology for understanding of business relationship value”. University of Economic Sciences and Public Administration. Cambeses Júnior, M., 2009, “As forças armadas e a política de defesa nacional”. Revista UNIFA, Vol. 22, Nº 24, pp. 152-.154. Corrêa, E. C., 1996, “Construção de um modelo multicritério de apoio ao processo decisório”. Dissertação, Universidade Federal de Santa Catarina, Florianópolis (SC). Cruz, E. L. V., 2006, “Tecnologia militar e indústria bélica no Brasil”, Security and Defense Studies Review, Vol. 6, No 3., pp. 359-416. Eden, C.; Ackermann., 1998, “ Making Strategy”. London: Sage Publications Ltd.

Ensslin, L., Montibeller Neto, G., Noronha, S. M. D., 2001, “Apoio à decisão: metodologia para estruturação de problemas e avaliação multicritério de alternativas”. Florianópolis: Insular, 296 pp. Ensslin, L., Montibeller Neto, G., Zanella, I. J., Noronha, S. M. D., 1998, “Metodologias multicritério em apoio à decisão. Santa Catarina”. LabMCDA. Universidade Federal de Santa Catarina. Escobar-Toledo, C. E., López-García, B., 2003, “Procesos cognitivos para la toma de decisiones a través de métodos multicritérios para la función de mantenimiento en equipos de perforación de pozos petroleros”. Cadernos Românicos em Ciências Cognitivas, Vol.1, Nº 3, pp. 105-119. Fitz, P. R., Hasenack, H., “O processo de tomada de decisão e os sistemas de informação geográfica”. Disponível em: <http://www.mendeley.com/research/o-processo-tomadadeciso-e-os-sistemas-informao-geogrfica/> Acesso em: 05 de junho de 2011. Franco, L. A., Montibeller, G., 2010a, “Problem structuring for multicriteria decision analysis interventions” Wiley Encyclopedia of Operations Research and Management Science. Franco, L. A., Montibeller, G., 2010b, “Facilitated modelling in operational research,” European Journal of Operational Research, Vol. 205, Nº 3. pp. 489-500. Giffhorn, E., Ensslin, L., Vianna, W. B., 2008, “Modelo multicritério para avaliação do desempenho de empresas terceirizadas”. XL SBPO Simpósio Brasileiro de Pesquisa Operacional. Gomes, L. F. A. M., Rangel, L. A. D., Jerônimo, R. L., 2010, “A study of professional mobility in a large corporation through cognitive mapping”. Revista Pesquisa Operacional, Vol.30, Nº. 2, pp.331-344. Hart J. A., 1977, “Cognitive maps of three Latin American policy makers”. Word Politics, Vol. 30, Nº 1, pp. 115-140.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

223

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

Iederan O. C., Curseu P. L., Vermeulen, P. A. M., Geurts, J. L. A., 2011, “Cognitive representations of institutional change. Similarities in the cognitive schema of entrepreneurs”. Journal of Organizational Change Management, Vol. 24, Nº. 1, pp. 9-28. Jardim, S. B., Silva, J. H. S., 2009, “Uma proposta para a estruturação do problema da cobrança pelo uso da água do ambiente”. Revista INGEPRO, Vol.1, Nº 4, 137-146. Kelly, G. A., 1955, “The Psychology of Personal Constructs”. Norton, New York. Lima, A. S., 2008, “Proposta de método para modelagem de critérios de priorização de projetos de pesquisa e desenvolvimento aeroespaciais”. Dissertação, Instituto Tecnológico de Aeronáutica, São José dos Campos (SP). Montibeller Neto, G.,1996, “Mapas Cognitivos: uma ferramenta de apoio à estruturação de problemas”, Dissertação, Universidade Federal de Santa Catarina, Florianópolis (SC). Novak, J. D., Cañas, A. J., 2008, “The theory underlying concept maps and how to construct and use them”, Technical Report IHMC Cmap Tools. Piratelli, C. L., Belderrain, M. C. N., 2010, “Apoio à fase de projeto de um sistema de medição de

224

desempenho com o Strategic Options Development and Analysis (SODA)”. XIII SIMPOI Simpósio de Administração da Produção, Logística e Operações Internacionais. Rieg, D. L., Araujo Filho, T., 2003, “O uso das metodologias Planejamento Estratégico Situacional e Mapeamento Cognitivo em uma situação concreta: O caso da pró-reitoria de extensão da UFSCar”. Revista Gestão e Produção, Vol..9, Nº. 2, pp.163-179. Roy, B., 1981, “The optimization problem formulation: criticism and overstepping”. Journal of Operational Research Society, Vol. 32, Nº 6, pp. 427- 436. Roy, B., 1996, “Multicritério methodology for decision aiding”. Dordrecht: Kluver Academic Publishers. Santos, J. L. C., 2010, “Estruturação de um modelo de avaliação multicritério para a seleção de medidas de gerenciamento da mobilidade voltadas aos pólos geradores de viagens” 4° Congresso Luso-Brasileiro para o Planejamento Urbano, Regional, Integrado, Sustentável, PLURIS. Silva, L. M. A., 2008, “Análise da decisão “comprar ou fazer” em tecnologias essenciais para o setor aeroespacial de defesa”. X SIGE Simpósio de Aplicações Operacionais em Áreas de Defesa.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

preservação da soberania nacional;

parceria internacional

segurança;

desenvolvimento tecnológico no setor de defesa;

Clusters:

Anexo 1. Mapa cognitivo congregado

Aplicação do mapa cognitivo a um problema de decisão do setor aeroespacial de defesa do Brasil

225

226

É essencial? Sim, porque a motivação de esperar aumentar a segurança militar leva a busca por empregar novas tecnologias no setor aeroespacial de defesa. Sim, pois investir em indústria de equipamentos e material militar é uma maneira de criar novas tecnologias para o setor aeroespacial.

É controlável? Sim, pois a forma como se conseguirá o aumento da segurança militar depende da decisão de comprar ou desenvolver tecnologias para o setor aeroespacieal de defesa. Indústria de Sim, porque a criação equipamentos e de indústrias de equipamentos e material material militar militar vai ser influenciado pela decisão de comprar ou desenvolver tecnologias para o setor aeroespacial de defesa. Sim, porque considerar as Medidas Sim, porque o medidas macroeconômicas macroeconômicas emprego de novas apropriadas, dentro do tecnologias vai setor aeroespacial de requerer de medidas macroeconoômicas defesa, depende só da apropriadas. decisão de comprar ou desenvolver tecnologias. Desenvolvimento Sim, porque o emprego Sim, pois a forma de novas tecnologias científico – como se conseguirá vai necessitar do tecnológico – o desenvolvimento desenvolvimento industrial científico – tecnológico – científico tecnológico industrial dentro do setor – tecnológico – aeroespacial de defesam industrial. depende só da decisão de comprar ou desenvolver tecnologias. Intercâmbio Sim, porque Sim, porque o intercâmbio doutrinado entre o intercâmbio doutrinário entre países países doutrinário entre vai ver-se influenciado países vai incentivar só pela decisão de a compra ou comprar ou desenvolver desenvolvimento das tecnologias para o setor tecnologias. aeroespacial de defesa.

PVF Aumentar segurança militar

Anexo 2. Verificação das propriedades dos pontos de vista fundamentais

Sim, porque é viável obter informações para quantificar essas medidas macroeconômicas.

Sim, porque pode-se obter informações para quantificar os intercâmbios doutrinários entre países.

Sim, porque se pode especificar, sem ambiguidade, uma quantidade por ano do desenvolvimento científico – tecnológico – industrial. Sim, porque se pode especificar, sem ambiguidade, uma quantidade para os intercâmbios, doutrinários entre países, frente ã desisão.

Sim, porque permite uma análise de forma independente com relação a outros PVF

Sim, porque permite uma análise de forma independente com relação a outros PVF, pelo fato que considera-se os aspecto científico.

Sim porque permite uma análise de forma independente com relação a outros PVF

Sim, pois a sua avaliação náo interfere nos outros PVF.

Sim, se dispoe de informações para quantificar o desenvolvimento de indústrias de equipamentos e material militar.

Sim, pois os investimentos para criar ou desenvolver a indústria de equipamentos e material militar é possível de ser medido e conhecido. Sim, pois pode ser feito um estudo que quantifique essas medidas macroeconômicas. Sim, porque é viável obter informações para quantificar essas medidas macroeconômicas.

É isolável? Sim, pois podese interferir no nível de segurança militar sem alterar necessariamente nos outros PVF.

É operacional? Sim, porque é possível coletar informações que permitam medir o aumento da segurança militar.

É mensuravel? Sim, é possível criar uma forma de medir o nível de segurança militar a partir da atuação militar.

Sim, porque este PVF transmite uma ideia de fácil entendimento.

Sim, porque este PVF transmite uma ideia de fácil entendmento.

Sim, pois é entendível e claro o seu significado.

É compreensível? Sim, pois é possível entender claramente o significado do PVF.

Santos, P.R.D., Curo, R.S.G., Belderrain, M.C.N.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 215-226, May-Aug., 2011

Thesis abstracts This section presents the abstract of most recent Master or PhD thesis related to aerospace technology and management

Environmental conditioning effects on the shear and damping properties of fiber metal hybrid composites

aggressive conditioning, due to the formation of corrosion pits on the metal layers of the laminate.

César Augusto Damato Faculdade de Engenharia do Campus de Guaratinguetá da Universidade Estadual Paulista (UNESP) cesardamato@uol.com.b

Study of curing and water sorption on glass-epoxy composites by luminescence and infrared spectroscopy

Thesis submitted for Master in Engineering at UNESP, Guaratinguetá, São Paulo State, Brazil, 2011. Advisors: Edson Cocchieri Botelho and Mirabel Cerqueira Rezende

Rita de Cássia Mendonça Sales Instituto Tecnológico de Aeronáutica rita.sales@fatecsjc.edu.br

Keywords: Fiber metal laminates, GLARE®, CARALL®, Environmental conditioning, Shear properties, Viscoelastic properties.

Thesis submitted for PhD degree in Materials Science at Instituto Tecnológico de Aeronáutica, São José dos Campos, São Paulo State, Brazil, 2010.

Abstract: The technological development of hybrid composites combining long-fiber reinforced polymeric laminates with metallic sheets, usually aluminum alloys, denominated FML, aims to obtain a range of materials with high mechanical strength values, stiffness, and low weight. This combination of properties becomes the FML attractive as replacements of both metal alloys and conventional thermosetting composites in aerospace applications. The objective of this paper was to evaluate the effects of different environmental conditionings (high moisture and temperature, saline atmosphere, and sudden changing of temperature) in GLARE® (glass fiber-metal laminates) and CARALL® (carbon fiber-metal laminates) FML samples. Interlaminar and losipescu shear and free vibration tests were used to evaluate the studied environmental conditioning effects. The study shows that the exposure to high-temperature cycles significantly influences the shear properties of the laminates, mainly due to the thermal expansion coefficient differences of their constituents. It is also observed that the loss and storage modules are affected, due to the possible formation of microcracks in the polymer matrix, and the metal/composite interface degradation. On the other hand, the presence of low moisture content does not affect so significantly the shear and viscoelastic properties. Finally, the saline atmosphere exposure showed the most

Advisor: Deborah Dibbern Brunelli Keywords: Composites, Curing of polymers, Photophysics, Luminescence, FT-IR, Water sorption. Abstract: High-performance composite materials have been widely used in the manufacture of primary and secondary structural aircraft parts, rockets, and several types of aerospace artifacts. This wide use is related to excellent material mechanical properties, beside a lower density compared to the one for alloys commonly used by the aerospace industry. The mechanical properties of laminates in composite materials are strongly dependent on several factors, such as: structure-building, fiber volume fraction, number of voids, interaction between the fiber and thermosetting resins, curing degree of the resin, and water content. The goal of this paper was to study the curing process and moisture influence in prepreg materials and glass fiber and epoxy resin laminates, using luminescence spectroscopy in the stationary mode, and intrinsic and extrinsic luminescence methods. The first method is based on the matrix polymer emission and the second one is related to the nine-anthroic acid emission (9-AA). The studies were monitored by gravimetric analysis, absorption spectroscopy in the infrared with Fourier transform (FT-IR), and the scanning electron microscopy

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 227-229, May-Aug., 2011

227

Thesis abstracts

(SEM). The study of the curing process was conducted by comparing results from the intrinsic luminescence and absorption spectroscopy in the near infrared with Fourier transform (FT-NIR). The extrinsic luminescence of 9-AA was used to evaluate the reduction of free volume during heat treatment. The FT-IR technique was also used to characterize the epoxy resin contained in prepregs F-155 and F-161. Prepregs, with or without 9-AA, were treated at 121 °C (F-155) and 177 °C (F-161) for 1,100 minutes at a heating rate of 2 °C/minutes. The results indicated that the curing reaction can be monitored by analyzing emission and excitation bands spectral changes of the prepreg and the 9-AA. For the analysis of the moisture influence, the samples were cured and subjected to environments with humidity controlled at 60 and 80 °C for time periods of 1, 7, 15, 30, 60 and 90 days. It was verified that the decrease in the maximum emission of the samples is directly related to the moisture content of the material.

typical ballistic atmospheric reentry of the SARA vehicle is selected to illustrate the potential for using the developed methodologies in an real design situation.

Surface hardening of a VC131 tool steel using a fiber laser Flávia Aline Goia Instituto Tecnológico de Aeronáutica milton@ieav.cta.br Thesis submitted for Master in Engineering at Instituto Tecnológico de Aeronáutica, São José dos Campos, São Paulo State, Brazil, 2010. Advisor: Milton Sérgio Fernandes de Lima

Hybrid solutions for heat transfer on ablative thermal protection system Daniel Fraga Sias Universidade Federal do Rio de Janeiro danielfs_7@hotmail.comr Thesis submitted for PhD degree in Mechanical Engineering at Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro, Rio de Janeiro State, Brazil, 2009. Advisors: Renato Machado Cotta and Nerbe José Ruperti Júnior Keywords: Integral transforms, Hybrid Ablation, Pyrolysis, Thermal protection.

methods,

Abstract: In this paper, hybrid numerical-analytical solutions have been developed for heat transfer with ablation in thermal protections of space vehicles under atmospheric reentry, it aimed at contributing with the development of a design and optimization methodology of ablative thermal protection systems, including materials that undergo pyrolysis. Four theoretical models of increasing complexity have been proposed and implemented within the symbolic computation platform Mathematica, and they were solved through the generalized integral transform technique (GITT) and the integral balance approach for convergence acceleration of the eigenfunction expansions. Hybrid solutions are verified with literature results and covalidated among themselves, and a

228

Keywords: Surface treatments, Tool steels, Fiber lasers, Hardening (materials), Heat treatment, Mechanical properties, Materials testing, Tribology, Materials engineering, Metallurgy. Abstract: Surface treatments have shown great importance on the enhancement of materials properties, allowing cost savings and gathering new methods and fabrication processes. Laser remelting and hardening involves rapid heating and cooling of metallic surfaces, resulting in microstructural transformations, which improve the mechanical properties without volumetric changes. This paper used a high-power fiber laser in order to produce a hardened layer at the surface of VC131 cold work steel. The hardening parameters studied were: scanning speed, beam defocusing, and laser power. The choice of best conditions for treating the entire surface was carried out after single-shots and line scans. The best conditions were those when the case depth was at maximum, with little remelting and absence of surface defects. The temperature variation during line scans was below 13 ºC; therefore, the piece remained cold. The overall laser absorptivity was calculated as approximately 37%. The microstructural and X-ray diffraction analyses revealed that hardened surfaces showed significant amount of martensite and retained austenite, which confers the increase on the surface hardness. The remelted zone presented hardness between 400 and 500 HV, although the hardness arose to 800 HV at the interface between the remelted and the heataffect zones. The maximum case depth for surface treated samples was between 1 and 2 mm. The wear resistance of laser-treated samples was higher than the unlasered ones, as could be seen in the tribological tests.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 227-229, May-Aug., 2011

Thesis abstracts

Manufacture and characterization of carbon fiber/PEKK composites with aeronautical applications Rogério Lago Mazur Faculdade de Engenharia do Campus de Guaratinguetá da Universidade Estadual Paulista (UNESP). rogermaz@uol.com.br Thesis submitted for Master in Engineering at UNESP, Guaratinguetá, São Paulo State, Brazil, 2011. Advisors: Edson Cocchieri Botelho and Mirabel Cerqueira Rezende Keywords: PEKK, Processing, Environmental Conditioning, Mechanical Properties, Viscoelastic Properties. Abstract: Advanced polymeric composites have been introduced as structural materials for high-performance aerospace applications over the last 40 years. However, these materials can present problems when exposed to moisture, temperature, ultraviolet (UV) radiation and thermal cycling. Besides, it may occur synergy among the main degradation mechanisms. Among the thermoplastic composite materials, carbon fiber reinforced PEKK (poly(ether-ketone-ketone)) laminates have shown excellent balance of properties, including high glass transition temperature, high strength, stiffness and fracture toughness values, low moisture absorption, and good environmental resistance. The aim of the present paper was to evaluate the influence of processing parameters on the carbon fiber/PEKK composites manufacturing, by hot compression molding, and also the influence of different environmental conditionings as hydrothermal, UV radiation, and thermal cycling effects on viscoelastic and mechanical behaviors. In this paper, the effects of environmental degradation on mechanical and viscoelastic properties were studied by compression, interlaminar shear strengths, dynamic mechanical analysis, and free vibration tests. Hot compression molding seemed well to produce the thermoplastic composite, but it needs to be optimised. Viscoelastic tests of the samples submitted to the UV radiation effects showed that glass transition temperatures of nonconditioning and conditioning specimens are near (~157 to 162 °C), when they are compared. On the other hand, the storage modulus presents a significant degradation emphasising the deleterious effect of UV conditioning on the studied composite stiffness.

Law airborne enforcement helicopters technological updating Márcio Luiz Ramos Pereira Instituto Tecnológico de Aeronáutica marciolrp@hotmail.com Thesis submitted for Master in Engineering in the Professional Master in Aviation Safety and Continued Airworthiness Program at Instituto Tecnológico de Aeronáutica, São José dos Campos, São Paulo State, Brazil, 2011. Advisor: Marcio Cardoso Machado Keywords: Air safety, Technological Helicopters, Law airborne enforcement.

updating,

Abstract: This study proposes the presentation of some possibilities to reduce the inherent risks of accident to aeronautic activities and qualitative increase to air safety, through technological updating of helicopters in operation by law airborne enforcement operators. The Esquilo HB 350/AS, an aircraft of French conception and assembled in Brazil, the main equipment of the segment, presents the possibilities of revitalization and integration of technological resources that enhance the use of helicopters in airborne law enforcement. These assets are not restricted to this model, but they are also applicable to the rest of the aircrafts with the same role. The case study was the methodology used to achieve this purpose. It investigated the technological updating of the US Navy in upgrade of the reliable naval helicopter Kaman SH-2F Seasprite to SH-2G Super Seasprite program and the Military Police of Federal District, in which the operator dared himself in converting his only helicopter HB 350 Esquilo B model in an AS 50 B2 model. The place, where the viability of such procedures, is identified and the positive impact in the institutions that use them. Its anticipate to this case study a pertinent theoretical review of airborne law enforcement in time with its historical origin in Brazil; it contemporaneously analysis the operational aspect of the environment as it develops and its elements adapted from survivability of combat aviation for the Brazilian reality. To give rise to the technology upgrade approach is consider the importance of maintenance in air safety and the experience of the use of HUMS in civil and military aviation offshore. The results are a list of recommendation to the operators of the technological updating measures setup of external accessories suitable for police mission, and the probable consequences to the air safety that these can adapt according to each operational peculiarities and realities.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 227-229, May-Aug., 2011

229

Instructions to the Authors

Scope and editorial policy The Journal of Aerospace Technology and Management is the official publication of Institute of Aeronautics and Space (IAE) of the Department of Aerospace Science and Technology (DCTA), São José dos Campos, São Paulo State, Brazil. The journal is published three times a year (April, August and December) and is devoted to research and management on different aspects of aerospace technologies. The authors are solely responsible for the contents of their contribution. It is assumed that they have the necessary authority for publication. When submitting the contribution the author should classify it according to the area selected from the topics: • Acoustics • Aerodynamics • Aerospace Systems • Applied Computation • Automation • Chemistry • Defense • Electronics

• Management Systems • Materials • Mechanical Engineering • Meteorology • Propulsion • Structures • Vibration

The journal uses the “double blind peer review process” for evaluation of the manuscript. The submissions, except thesis and book reviews, will be evaluated by three Editorial board members or ad hoc referees, and may be selected for publication according to the editorial policy of the journal.

Mandatory requirements All papers must include: type of contribution (review article, original paper, short communication, case report, book reviews or theses), title, authors’ names, abstract and key words (three to six items that should be based on NASA Thesaurus volume 2 – Access Vocabulary). All authors should be identified with full name, e-mail, institution to which they are related, city and country. One of them should be indicated as the author for correspondence.

Contents • Editorial Any researcher may write the editorial on the invitation of the Editor-in-Chief. • Review articles They should cover subjects falling within the scope of the journal. These contributions should be presented in the same format as a full paper, except that they should not be divided into sections such as introduction, methods, results and discussion. However, they must include a 150 to 200-word abstract, key words, concluding remarks, acknowledgment and references. The article should not exceed 20 pages. • Technical papers These articles should report the results of original research and must include: a 150 to 200-word abstract, key words, introduction, methods, results and discussion, acknowledgment, references, tables and/or figures. The article should not exceed 16 pages. •Communications These articles should report previous results of ongoing research. They should include a 150 to 200-word abstract, key words, tables and/or figures, acknowledgment and references. The communication should not exceed eight pages. • Thesis abstracts The journal welcomes Masters and PhD thesis abstracts for publication.

Paper submission Manuscript should be written in English or Portuguese and submitted electronically. The manuscripts written in Portuguese must present the title and the abstract translated into English, with the exact same content. If there is any conflict of interest with regard to the evaluation of the manuscript, the author must send a declaration indicating the reasons, for the review process occur fairly. See the instructions on www.jatm.com.br/papersubmission. 230

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 230-231, May-Aug., 2011

After submitting the manuscript, the corresponding author will receive an e-mail with the Term of Copyright Transfer, in which the author agrees to transfer copyright to the Institute of Aeronautics and Space (IAE), in case of acceptance for publication, thus being forbidden any means of reproduction (printed or electronic) without previous authorization of the Editor-in-Chief. If the reproduction is allowed, it is mandatory to mention the Journal of Aerospace Technology and Management. The author also declares that the manuscript is an original paper and that its content is not being considered for publication in other periodicals and that all co-authors participated satisfactorily in the paper elaboration as to make public the responsibility for its content. The declaration must be printed, signed by the main author and sent back by mailing to the following address: Instituto de Aeronáutica e Espaço (IAE)/ATTN: Helena Prado/ Praça Mal. Eduardo Gomes, 50 – Vila das Acácias/ CEP 12228-901/São José dos Campos/ São Paulo/Brazil. References References should be cited in the text by giving the last name of the author(s) and the year of publication. Either use “Recent work (Smith and Farias, 1997)” or “Recently Smith and Farias (1997)”. With four or more names, use the form “Smith et al. (1997)”. If two or more references would have the same identification, distinguish them by appending “a”, “b” etc., to the year of publication. Acceptable references include journal articles, numbered papers, books and submitted articles, if the journal is identified. References from private communications, dissertations, thesis, published conference proceedings and preprints from conferences should be avoided. Self citation should be limited to a minimum. It is recommended that each reference contains the digital object identifier number (DOI). References retrieved from the internet should be cited by the last name of the author(s) and the year of publication, or n.d. if not available, followed by the date of access. Standards should be cited in text by the acronym of entity followed by their number, and doesn’t need to appear in the reference list. References should be listed in alphabetical order, according to the last name of the first author, at the end of the article. Some sample references follow: Alves, M. B., Morais, A. M. F., 2009, “The management of Knowledge and Technologies in a Space Program”, Journal of Aerospace Technology and Management, Vol. 1, No 2, pp. 265-272. doi:10.5028/jatm.2009.0102265272 Bordalo, S. N., Ferziger, J. H. and Kline, S. J., 1989, “The Development of Zonal Models for Turbulence”, Proceedings of the 10th Brazilian Congress of Mechanical Engineering, Vol. 1, Rio de Janeiro, Brazil, pp.41-44. Coimbra, A. L., 1978, “Lessons of Continuum Mechanics”, Ed. Edgard Blücher, São Paulo, Brazil, 428p. Clark, J. A., 1986, Private Communication, University of Michigan, Ann Harbor. Silva, L. H. M., 1988, “New Integral Formulation for Problems in Mechanics” (In Portuguese), Ph.D. Thesis, Federal University of Santa Catarina, Florianópolis, S.C., Brazil, 223p. EMBRAPA, 1999, “Polítics of R&D”, Retrieved in May 8, 2010, from http://www.embrapa.br/publicacoes / institucionais/polPD.pdf,. Sparrow, E. M., 1980a, “Forced Convection Heat Transfer in a Duct Having Spanwise-Periodic Rectangular Protuberances”, Numerical Heat Transfer, Vol. 3, pp. 149-167. Sparrow, E. M., 1980b, “Fluid-to-Fluid Conjugate Heat Transfer for a Vertical Pipe-Internal and External Natural Convection”, ASME Journal of Heat Transfer, Vol.102, pp. 402-407. Illustrations All illustrations, line drawings, photographs and graphs should be referred as “Figure” and submitted with good definition (1 to 2 mega pixels). References should be made in the text to each illustration using the abbreviated form “Fig.”, except in the beginning of the phrase. Explanations should be given in the figure legends, so that illustrations are kept clean Tables Authors should take notice of the limitations set by the size and layout of the journal. Therefore, large tables should be avoided. All tables must be numbered and mentioned in the text as “Table”. Equations Equations should be typed on individual lines, identified by numbers enclosed in parenthesis. References should be made in the text to each equation using the abbreviated form “Eq.”, except in the beginning of the phrase, where the form “Equation” should be used. Acknowledgments The financial support received for the elaboration of the manuscript must be declared in this item.

J. Aerosp.Technol. Manag., São José dos Campos, Vol.3, No.2, pp. 230-231, May-Aug., 2011

231

General Information

Correspondence

Journal of Aerospace Technology and Management

Published and distributed by: Institute of Aeronautics and Space

Vol. 3, n.2 (may/aug. 2011) – São José dos Campos: Zeppelini Editorial, 2011

Desktop publishing and printing: Zeppelini Editorial

Four monthly issue

Edition: 750 São José dos Campos, SP, Brazil

1. Aerospace sciences

ISSN 1984-9648

2. Technologies

Note: This publication is sponsored by the Institute of Aeronautics and Space (IAE) to whom the copyright on

3. Aerospace engineering CDU:629.73

all published material belongs. Permission must be requested prior to use

Vol. 3 N. 2 May/Aug. 2011

ISSN 1984-9648 ISSN 2175-9146 (online) www.jatm.com.br

Journal of Aerospace Technology and Management V.3, n. 2, May/Aug., 2011