Volume 10 issue 3 December 2011
1
from the Chief Editor
The excitement of the Grand Event of ISNT, the NDE2011, to be held in CHENNAI TRADE CENTRE, will bring the ISNT community together for 5 days of science, technology, business, and fun and the Journal of Nondestructive Testing and Evaluation welcomes all delegates. The Editorial Board joins me in congratulating all the ISNT Award Winners for the 2011. This edition of the Journal brings forth the now popular sections BASICS, HORIZONS, PUZZLE, EVENTS, PATENTS, PROBE along with 4 technical articles. The BASICS is focused on Eddy Current Testing Methods that is well compiled by Dr. BPC Rao. The HORIZONS describes the use of an advanced method of harmonic ultrasound that has been shown to be sensitive to micro-structural changes including very early damage in materials. The technical article on indirect methods for inversion of eddy current data employs 2 novel approaches using the ANN for the characterization of buried defects in multi-layered components. The application of nano-structured materials for the magnetostrictive generation of long range guided ultrasonic waves in pipes has been discussed in one of the articles and this novel approach holds great promise for a low profile sensor system for NDE of pipes for corrosion. The technical article on the improved characterization of impact damage in Kevlar composite test coupons using advanced data analysis algorithm using clustering further reiterates the need for applying advanced data handling algorithms in NDE. The use of Wavelet based analysis for impact echo and low frequency ultrasonic inspection of concrete for honeycomb defects further demonstrates the need for signal and image processing tools in NDE. On behalf of the editorial board of the Journal I would like to wish all of the readers a very happy and successful 2012. Dr. Krishnan Balasubramaniam Professor Centre for Non Destructive Evaluation IITMadras, Chennai balas@iitm.ac.in jndte.isnt@gmail.com URL: http://www.cnde-iitm.net/balas
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
2
I S N T - National Governing Council Chapter - Chairman & Secretary President Shri K. Thambithurai President-Elect Shri P. Kalyanasundaram Vice-Presidents Shri V. Pari Swapan Chakraborty Shri D.J.Varde Hon.General Secretary Shri R.J.Pardikar Hon. Treasurer Shri T.V.K.Kidao
Ahmedabad
Kota
Shri D.S. Kushwah, Chairman, NDT Services, 1st Floor, Motilal Estate, Bhairavnath Road, Maninagar, Ahmedabad 380 028. dskushwah@icenet.net Shri Rajeev Vaghmare, Hon. Secretary C/o Modsonic Instruments Mfg. Co. Pvt. Ltd. Plot No.33, Phase-III, GIDC Industrial Estate Naroda, Ahmedabad-382 330 modsonic@modsonic.com
Shri R.C. Sharma, Chairman QAS, RAPS - 5 & 6, PO Anushakti Rawatbhata 323 303 abahl@npcil.co.in Shri S.K. Verma, Hon. Secretary, TQAS, RAPS - 5 & 6, PO Anushakti Rawatbhata 323303. surendrakverma@npcil.co.in
Bangalore Prof.C.R.L.Murthy, Chairman Dept. of Aerospace Engg, Indian Institute of Science, Bangalore 560012 Email : crlmurty@aero.iisc.ernet.in crlmurty@aero.iisc.ernet.in
Hon. Joint Secretaries Shri Rajul R. Parikh Immediate Past President Shri Dilip P. Takbhate Past President Shri S.I.Sanklecha Members Shri Anil V. Jain Shri Dara E. Rupa Shri D.K.Gautam Shri Diwakar D. Joshi Dr. Krishnan Balasubramaniam Shri Mandar A. Vinze Shri B.B.Mate Shri G.V. Prabhugaunkar Shri B.K.Pangare Shri M.V. Rajamani Shri P.V. Sai Suryanarayana Shri Samir K. Choksi Shri B.K. Shah Shri S.V. Subba Rao Shri Sudipta Dasgupta Shri N.V. Wagle Shri R.K. Singh Shri A.K.Singh (Kota) Shri S. Subramanian Shri C. Awasthi Brig. P. Ganesham Shri Prabhat Kumar Shri P. Mohan Shri R. Sampath Shri V. Sathyan Shri B. Prahlad Ex-officio Members Managing Editor, JNDT&E Shri V. Pari Chairman, NCB & Secretary, QUNEST Dr. Baldev Raj Controller of Examination, NCB Dr. B. Venkatraman President, QUNEST Prof. Arcot Ramachandran
& All Past Presidents of ISNT
vol 10 issue 2 December 2011
Shri R.S. Vaghasiya, Chairman, B 4/7, Sri Punit Nagar, Plot 3, SV Road, Borivile West, Mumbai 400 092. ravji.vaghasiya@gmail.com Shri Samir K. Choksi, Hon. Secretary, Director, Choksi Brothers Pvt. Ltd., 4 & 5, Western India House, Sir P.M.Road, Fort, Mumbai 400 001. Choksiindia@yahoo.co.in
Nagpur
Chennai Shri R. Sundar, Chairman Director of Boilers, Tamil Nadu Shri R. Balakrishnan, Hon. Secretary, No.13, 4th Cross Street, Indira Nagar, Adyar, Chennai 600 020. rbalkrishin@yahoo.co.in
Shri Pradeep Choudhari, Chairman Parikshak & Nirikshak, Plot M-9, Laxminagar Nagpur - 440 022 Mr. Jeevan Ghime, Hon. Secretary, Applies NDT & Tech Services, 33, Ingole Nagar, B/s Hotel Pride, Wardha Road, Nagpur 440 005. antstg_ngp@sancharnet.in
Delhi
Pune Shri BK Pangare, Chairman Quality NDT Services, Plot BGA, 1/3 Bhosari, General Block, MIDC, Bhosari, Pune- 411 026 ndtserve@pn3.vsnl.net.in Shri BB Mate, Hon Secretary, Thermax Ltd., D-13, MIDC Ind. Area, RD Aga Road, Chinchwad, Pune- 411 019 bmate@thermaxindia.com
Shri A.K Singhi, Chairman, MD, IRC Engg Services India Pvt. Ltd 612, Chiranjiv Tower 43, New Delhi 110019 ashok.ircengg@gmail.com Shri M.C. Giri, Hon.Secretary, Managing Partner, Duplex Nucleo Enterprise New Delhi 110028 munish.giri@yahoo.com
Sriharikota Shri S.V. Subba Rao, Chairman, General Manager, Range Operations SDSL, SHAR Centre Sriharikota 524124. svsrao@shar.gov.in Shri G. Suryanarayana, Hon. Secretary, Dy. Manager, VAB, VAST, Satish Dhawan Space Centre, Sriharikota-524 124. isnt@shar.gov.in
Hyderabad Shri M. Narayan Rao, Chairman, Chairman & Managing Director, MIDHANI, Kanchanbagh, Hyderabad 500 058. cmd.midhani@ap.nic.in Shri Jaiteerth R. Doshi, Hon.Secretary, Scientist, Project LRSAM DRDL, Hyderabad 500 058. joshidrdl@gmail.com
Tarapur
Jamshedpur Dr N Parida, Chairman, Senior Deputy Director Head, MSTD, NML, Jamshedpur - 831 007 nparida@nmlindia.org Mr. GVS Murthy, Hon. Secretary, MSTD, NML, Jamshedpur gvs@mnlindia.org / gvsmurthy_mnl@yahoo.com
Kalpakkam Shri YC Manjunatha, Chairman Director ESG, IGCAR, Kalpakkam – 603 102 ycm@igcar.gov.in Shri BK Nashine, Hon.Secretary Head, ED &SS, C&IDD, FRTG IGCAR, Kalpakkam – 603 102 bknash@igcar.gov.in
Kochi Shri CK Soman, Chairman, Dy. General Manager (P & U), Bharat Petroleum Corporation Ltd. (Kochi Refinery), PO Ambalamugal 682 302. Kochi somanck@bharatpetroleum.in Shri V. Sathyan, Hon. Secretary, SM (Project), Bharat Petroleum Corporation Ltd. (Kochi Refinery), PO Ambalamugal-682 302. Kochi sathyanv@bharatpetroleum.in
All Chapter Chairmen/Secretaries Permanent Invitees Shri V.A.Chandramouli Prof. S. Rajagopal Shri G. Ramachandran
Mumbai
Kolkata Shri Swapan Chakraborty, Chairman Perfect Metal Testing & Inspection Agency, 46, Incinerator Road, Dum Dum Cantonment, Kolkata 700 028. permeta@hotmail.com Shri Dipankar Gautam, Hon. Secretary, 4D, Eddis Place, Kolkata-700 019. eib1956@gmail.com
Shri PG Bhere, Chairman, AFFF, BARC, Tarapur-401 502. pgbehere1@rediffmail.com Shri Jamal Akftar, Hon.Secretary, TAPS 1 & 2, NPCIL, Tarapur. jakhtar@npcil.co.in
Tiruchirapalli R.J. Pardikar AGM, (NDTL) BHEL Tiruchirapalli 620 014. rjp@bheltry.co.in Shri A.K.Janardhanan, Hon. Secretary, C/o NDTL Building 1, H.P.B.P., BHEL, Tiruchirapalli 620 014. akjn@bheltry.co.in
Vadodara Shri P M Shah, Chairman, Head-(QA) Nuclear Power Corporation Ltd. npcil.bar@gmail.com M S Hemal Thacker, Hon.Secretary, NBCC Plaza, Opp.Utkarsh petrol pump, Kareli Baug, Vadodara-390018. pmetco@gmail.com
Thiruvananthapuram Dr. S. Annamala Pillai, Chairman Group Director, Structural Design & Engg Group, VSSC, ISRO, Thiruvananthapuram 695022 s_annamala@vssc.gov.in Shri. Binu P. Thomas Hon. Secretary, Holography section, EXMD/SDEG, STR Entity, VSSC, Thiruvananthapuram 695 022 binu_thomas@vssc.gov.in
Visakhapatnam Shri Om Prakash, Chairman, MD, Bharat Heavy Plate & Vessels Ltd. Visakhapatnam 530 012. Shri Appa Rao, Hon. Secretary, DGM (Quality), BHPV Ltd., Visakhapatnam 530 012
Journal of Non Destructive Testing & Evaluation
Journal of Non Destructive Testing & Evaluation About the cover page:
Volume 10 issue 3 December 2011
Contents
About the Cover page The Friction stir welding (FSW) is a new and promising welding process which welds the material below its melting temperature and it has shown superior features such as excellent joint performance, mechanical properties and low energy consumption. This process is effective for joining the material which is difficult to weld by conventional fusion welding. Dissimilar alloy material can also be joined with the FSW process The right choice and application of the welding parameter is required to get sound welds which helps to improves the mechanical properties of material. In FSW, both rotational speed and translation speed exert a significant effect on the heat input generation and hence the mechanical properties of the material. Defects like large mass of flash out causes due to excess heat input. Also defects such as cavity or groove like defects are caused due to insufficient heat input and/or abnormal stirring. It was reported the defects like porosity, kissing bong, solid inclusion and linear crack like defects in magnesium alloy. These defects reduce the quality of weld, Hence examination of these defects is very much required to optimize the welding parameter. In the cover page, ultrasonic C-scan image is used for imaging the defects caused due to suboptimal choice of the rotational speed. The difference FSW welds were scanned using a high frequency focused ultrasound beam of 15 MHz in an immersion mode. By appropriate time gating, the image shows varying regions and morphology of damage for difference combinations of translation speed and rotational speed of the FSW tool. This study provides the optimal combination of the two controllable parameters for a defect free FSW process for Aluminum welding. (The results are provided by the Centre for Non-destructive Evaluation at IIT Madras)
5
Chapter News
7
Basics
17
Horizons
27
NDE Events
28
NDT Puzzle Technical Papers Imaging of Impacted Kevlar Composite Armours using Data Clustering
44
Sutanu Samanta and Debasis Datta
Generation of Guided Waves Using Magnetostrictive Nanostructured Sensing Elements for Pipe Inspection
49
A.K. Panda, P.K. Sharan, G.V.S. Murthy, R.K. Roy, S. Palit Sagar and A. Mitra
Detection of honeycomb defects in reinforced concrete structures using acoustic pulse-echo methods and wavelet transforms
53
Krishna Prasad M, Herbert Wiggenhauser, Krishnan Balasubramaniam
Indirect methodologies for inversion of eddy current NDE data
60
S. Shuaib Ahmed, B.P.C. Rao, S. Thirunavukkarasu and T. Jayakumar
Probe
68
Chief Editor Prof. Krishnan Balasubramaniam e-mail: balas@iitm.ac.in
Co-Editor Dr. BPC Rao bpcrao@igcar.gov.in
The Journal is for private circulation to members only. All rights reserved throughout the world. Reproduction in any manner is prohibited. Views expressed in the Journal are those of the authors' alone. Published by Shri RJ Pardikar, General Secretary on behalf of Indian Society for Non Destructive Testing (ISNT) Modules 60 & 61, Readymade Garment Complex, Guindy, Chennai 600032 Phone: (044) 2250 0412 Email: isntheadoffice@gmail.com and Printed at VRK Printing House 3, Potters Street, Saidapet, Chennai 600 015 vrkonline@gmail.com Ph: 09381004771
Managing Editor Sri V Pari e-mail: scaanray@vsnl.com
Topical Editors Dr D K Bhattacharya Electromagnetic Methods
Dr T Jayakumar, Ultrasonic & Acoustic Emission Methods
Sri P Kalyanasundaram Advanced NDE Methods
Sri K Viswanathan Radiation Methods
Editorial Board Dr N N Kishore, Sri Ramesh B Parikh, Dr M V M S Rao, Dr J Lahri, Dr K R Y Simha, Sri K Sreenivasa Rao, Vaidyanathan, Dr K Rajagopal, Sri G Ramachandran, Sri B Ram Prakash
Sri S
Advisory Panel Prof P Rama Rao, Dr Baldev Raj, Dr K N Raju, Sri K Balaramamoorthy, Sri V R Deenadayalu, Prof S Ramaseshan, Sri A Sreenivasulu, Lt Gen Dr V J Sundaram, Prof N Venkatraman
Objectives The Journal of Non-Destructive Testing & Evaluation is published quarterly by the Indian Society for Non-Destructive Testing for promoting NDT Science and Technology. The objective of the Journal is to provide a forum for dissemination of knowledge in NDE and related fields. Papers will be accepted on the basis of their contribution to the growth of NDE Science and Technology.
4
Classifieds Scaanray Metallurgical Services
Transatlantic Systems
(An ISO 9001-2000 Certified Company)
NDE Service Provider Process and Power Industry, Engineering and Fabrication Industries, Concrete Structures, Nuclear Industries, Stress Relieving Call M. Nakkeeran, Chief Operations, Lab: C-12, Industrial Estate, Mogappair (West), Chennai 600037 Phone 044-2625 0651 Email: scaanray@vsnl.com ; www.scaanray.com
Electro-Magfield Controls & Services & LG Inspection Services We manafucture : Magnetic Crack Detectors, Demagnetizers, Magnetic Particles & Accessories, Dye Penetrant Systems etc Super Stockist & Distributors: M/s Spectonics Corporation, USA for their complete NDT range of productrs, Black Lights, Intensity Meters, etc. Plot 165, SIDCO Industrial Estate, (Kattur) Thirumullaivoil, Vellanur Village, Ambattur Taluk Chennai 600062 Phone 044-6515 4664 Email: emcs@vsnl.net
Madras Metallurgical Services (P) Ltd Metallurgists & Engineers
Metallography Strength of Materials Non Destructive Testing Foundry Lab
Serving Industries & Educational Institutes for the past 35 years
24, Lalithapuram street, Royapettah, Chennai 600014 Ph: 044-28133093 / 28133903 Email: mmspl@vsnl.com
OP TECH ASNT Level III Intensive Taining Educational CDs PT, UT, RT, MT, ET, Basic Metallurgy and Mechanical Testing Call 93828 12624 Land 044 - 2446 1159
B Ram Prakash A 114, Deccan Enclave, 72, T M Maistry Street, Thiruvanmiyur, Chennai 600 041
Southern Inspection Services NDT Training & Level III Services in all the following ten NDT Methods
Shri. K. Ravindran, Level III RT, VT, MT, PT, NR, LT, UT, ET, IR, AE
vol 10 issue 3 December 2011
Support for NDT Services NDT Equipments, Chemicals and Accessories Call DN Shankar, Manager 14, Kanniah Street, Anna Colony, Saligramam, Chennai 600093 Phone 044-26250651 Email: scaanray@vsnl.com
Betz Engineering & Technology Zone An ISO 9001 : 2008 Company 49, Vellalar Street, near Mount Rail Station, Chennai 600088 Mobile 98401 75179, Phone 044 65364123 Email: betzzone@vsnl.net / rg_ganesan@yahoo.com
International Training Division 21, Dharakeswari Nagar, Tambaram to Velachery Main Road, Sembakkam, Chennai 600073 www.betzinternational.com / www.welding-certification.com
KIDAO Laboratories NABL Accredited Laboratory carrying out Ultrasonic test, MPL and DP tests, Coating Thickness and Roughness test. We also do Chemical and Mechnical tests
A-3, Mogappair Indl. Area (East) JJ Nagar, Chennai 600037 Phone 044-26564255, 26563370 Email: kidaolab@giasmd01.vsnl.net.in; kidaolabs@vsnl.net www.kidaolabs.com
Dhvani R&D Solutions Pvt. Ltd 01J, First Floor, IITM Research Park, Kanagam Road, Taramani, Chennai 600113 India Phone : +91 44 6646 9880
• Inspection Solutions • Software Products • Training • Services & Consultancy
-
CUPS, TAPS, CRISP, TASS SIMUT, SIMDR Guided Waves, PAUT, TOFD Advanced NDE, Signal Processing C-scans, On-line Monitoring
E-mail: info@dhvani-research.com
www.dhvani-research.com
No.2, 2nd Floor, Govindappa Naicker Complex, Janaki Nagar, Arcot Road, Valasaravakkam, Chennai-600 087 Tamil Nadu, India Phone : 044-2486 8785, 2486 4481 E-mail: sisins@gmail.com and sisins@hotmail.com Website: www.sisndt.com
Journal of Non Destructive Testing & Evaluation
5
CHAPTER NEWS BANGALORE Courses & Exams Conducted : Also organized ASNT Level-III Exam in May 2011, Refresher course were also conducted.
* MT & PT Level-II (ASNT) course was conducted from 5.08.2011-13.08.2011.
Mr. R.K. Singh, Mr. Ashok Agrawal, Mr. Rajpal Singh.
* RT Level-II (ASNT) course was conducted from 19.08.2011 to 28.08.2011.
JAMSHEDPUR Other Activities: Annual General Body Meeting conducted on 29.04.2011 and the new executive committee was elected under the chairmanship of Dr. N. Parida, Scientist from NML . Awards :Dr. Amitava Mitra received the prestigious Materials Research Society of India Medal (MRSI Medal).
Other Activities: Conducted Executive Committee Meeting & the New EC was elected for 2011-2012
* UT Level-II (ISNT) course was conducted from 05.09.2011 to 18.09.2011.
CHENNAI
* PT Level-II (ISNT) course was conducted from 10.10.2011 to 16.10.2011.
Technical Talk: 28.08.2011- Dr. O.Prabhakar, Professor (Retd) – IIT Madras, School of Materials Engineering NTU Singapore delivered a technical talk on “The Importance of NDE in Automobile components”. 25.9.2011- Mr. P.Vijayaraghavan Central Lab, F&F Division, HAL, Bangalore delivered a technical talk on “Role of NDT in Aerospace Industry” .
* UT Level-II (ASNT) course was conducted from 28.10.2011 to 06.11.2011. Other Activities: ISNT DAY celebrated on 21.04.2011. * The Thambithurai Award for Best Technical Paper presented to Dr. Vaidehi Ganesan, Scientist, IGCAR * The PARI award for the Best Member was Mr RG Ganesan, Joint Secretary, ISNT Chennai Chapter
16.10.2011- Dr. T.Jayakumar, Director MMG, IGCAR, Kalpakkam delivered a technical talk on “Role of NDE in Civil Concrete”.
EC Meetings: *01.05.2011 *11.06.2011
30.10.2011 – Dr. R.Sundar, Director of Boilers, Tamilnadu delivered a technical talk on “Residual Life Assessment of Boilers”.
Annual General Body meeting (AGM) was held on 16.07.2011, 07.08.2011, 25.09.2011, 16.10.2011.
6.11.2011 – Dr. M.T.Shyamsunder, Principlal Engineer, NDE Modelling & Imaging Lab, Bangalore delivered a technical talk on “Advances in Electromagnetic NDE –Techniques & Application”. Courses & Exams Conducted : * UT Level-II (ASNT) course from 22.04.11to 30.04.11. * MT – Level - II(ASNT), for M.M.Forgings Ltd from 24th Apr, 08th, 15th ,22nd May 2011 * Surface NDT ( MT & PT) Level - II (ASNT) from 20.05.2011 to 26.05.2011. * UT L- II (ASNT) course from 30.05.2011 to 05.06.2011. * RT L- II (ASNT) course from 10.06.2011 to 16.06.2011. * UT L-II (ASNT) course was conducted from 24.6.2011 to 2.07.2011. * In-House training on MT Level- I & II (ASNT) course for Brakes India Ltd, Chennai was conducted from 24.06.2011 to 28.06.2011. * RT Level- II (ISNT) course was conducted from 11.07.2011 to 24.07.2011. * In house training PT Level –II (ASNT) course at BHEL, Ranipet was conducted from 12.07.2011 to 15.07.2011.
DELHI Technical Talk: A technical talk on Phased Array UT by Mr. Peter Renzel of GE in association Of Kronix was organized at IRC, Noida. Other Activities: One day seminar on NDT was organized on 8th April, 2011 in association with Gurgaon College of Engineering at Gurgaon. AGM conducted on 9th Oct, 2011 in CSOI at KG Marg. New Delhi. Election for the year 2011-2012 was conducted & New Executive committee was formed unanimously as detailed below.: Chairman - Mr. A.K Singhi Vice Chairman - Mr. A.K. Pawar Vice Chairman - Mr. Kuldeep Singh Hon. Gen. Secretary - Mr. M.C. Giri Hon. Joint. Secretary - Mr. Daya Ram Gupta Hon. Treasurer - Mr. R.S. Mahto Members Mr. S.K. Goel, Mr. Dinesh Gupta, Mr. Navita Gupta, Mr. F. Sidique, Mr. D.K. Dua, Mr. T. Kamaraj, Mr. Muneshwar Dayal 1st EC Meeting was conducted on 5th Nov, 2011 at IRC Nehru Place, New Delhi. Following new members were co-opted as EC member Mr. A. Bhatnagar,
Journal of Non Destructive Testing & Evaluation
KALPAKKAM Technical Talk: Conducted a Technical Talk on X-ray endoscopic inspection of T/TS welds in heat exchangers by eminent Prof. Dr Uwe Ewert on Friday, 22nd July, 2011. KOCHI Technical Talk: A technical presentation on “NDT and other technical innovations being implemented in Gas networking Pipelines” was conducted on 17th Aug 2011 by senior executives from M/s GAIL. Courses and Exams Conducted: Certification Programme : Two RT Level II programmes as per SNT TC IA were conducted. One Programme was exclusively for the officials from Dept. Of Factories and Boilers, Kerala which was held during the period 17 /01 /11 to 29 /01 /11. Another ASNT RT level II was conducted from 2 4 /08 / 11 to 03 / 09 / 11, based on the request from nearby Industries. Other Activities: Committee meetings : Two meetings of the executive committee were held during the period (on 28/05/11 and 17/08/11). Membership: 5 new members joined during the period. KOLKATA Courses and Exams Conducted Radiography Testing (RT- II) Training & Certification Course from 23rd May to 29th May, 2011. 1st Magnetic Particle Testing (MPT- II) Training & Certification Course from 15.07.11 to17.07.11 1st Penetrant Testing (PT- II) Training & Certification Course from 22.07.11 to 24.07.11. 8thRadiography Testing (RT- II) Training & Certification Course from 01.08.11 to 08.08.11. 5th Ultrasonic Testing (UT- II) Training & Certification Course from 05.09.11 to 11.09.11.
vol 10 issue 3 December 2011
6
Other Activities: EC Meetings: 1. April’11 , 2.June,’11. Membership: *Life member - 135*Corporate Member – 8 *Life Corporate Member – 7.*Member – 60.*Life Fellow -1. Annual General Meeting held on 21st.October,2011 New EC Committee being constituted : Sri Swapan Chakraborty - Chairman., Sri.D.Gautam - Secretary., Sri Sandeep Agarwal - Treasurer. KOTA Courses and Exams Conducted: Leak Testing Level -II Course during June, 2011. Other Activities: New EC Committee has been constituted. Chairman - Shri A. Bahl Hon.Secretary —- Shri S.K.Verma. MUMBAI Technical Talk: Technical Lecture on 05-09-2011 Speaker: by Mr Roman Fernandez, France Topic: “Simulation in NDE”. Courses and Exams Conducted: Conducted Welding Inspector examination at ITT, on 1st May 2011, Conducted LT Level II Examination for KOTA Chapter on 26- 06- 2011 at Kota. Conducted Welding Inspector examination at ITT, Mahim on 28th August 2011. Conducted Welding Inspector examination for ISNT, Tarapur Chapter, at TARAPUR on 17th September 2011. Conducted General NDT Course for ONGC Engineers from 10th Oct. to 14th Oct. 2011. NDT Level - III & ISNT Level – III Refresher courses & ISNT Level – III Examination from 31st October 2011 to 25th November 2011 at Hotel Atithi. Other Activities: *
APCNDT 2013 committee Meeting was held on 8th April 2011, 19th April 2011. 25th April 2011.13th May 2011. 7th June 2011. 29th July 2011 and 25th August 2011.
*
EC Meeting was held on 22rd June 2011, 17th August 2011 and 20th Oct. 2011.
*
Conducted NCB and NGC Meeting on 21ST August at Hotel Atithi, Vile Parle, Mumbai.
*
NDT Achievement Awards Mumbai meeting held on 25th August at Acres Club, Chembur.
AGM was held on 17th Sept. PUNE Technical Talk: TOFD Technique introduction & Practical Demonstration on Weld”on 14.05.2011 by Shri Ashok Trivedi . Other Activities: EC Meeting held on 29.04.2011. TARAPUR Courses & Exams Conducted The Certified Welding Inspector Level – II was conducted from 07 / 09 / 2011 to 16 / 09 / 2011. Liquid Penetrant Test Level –II conducted from 20/7/2010 to 24/7/2010. 50 chapter is planning training on ultrasonic thickness gauging as per NPCIL requirments.
vol 10 issue 3 December 2011
Other Activities 1. 16th AGM, held on 17.09.2011 in Hotel, Silver Avenue Ostwal Park, Boisar. AGM approved chapter audited account report and annual report for the year 2010-2011. 2. New managing commeetee election was held following members elected for the post indicated against their namesChairman-Shri P.G.Behere, Co-Chairman-Shri R.Murali Secretary-Shri J.Akhtar, Jt. Secretary-Shri D.B.Sathe, Treasurer-Shri V.H.Patil Members : Shri S. Pradhan, Shri E. Mudliyar Shri A.P.Kulkarni, Shri P.M.Bhave, Shri N.K.Roy, Shri Chetan Mali, Smt Nilima Walinjakar. TRICHY Courses and Exams Conducted: a) Radiographers level I-(In association with BARC-Mumbai) from 11.04.2011 to 29.04.2011. b.) MPT-Level –II- From 16.06.2011 to 19.06.2011. c.) LPT – Level – II – from 13.06.2011 to 15.06.2011. d.) Radiography Level – II from 11.07.2011 to 21.07.2011. Other Activities: 1. EC meeting was conducted on June 2011 2. Invited Lecture : “Introduction to SafeRad Radiography System - an unique & innovative ndt technique” BY Mr. Malcolm Wass U.K. on 11th july 2011. TRIVANDRUM Technical Talk: Ms Andreanne Potvin, Product Manager Olympus NDT corporation made a technical Presentation on Latest Advances In Phased Array Ultrasonics on 28th July 2011. Other Activities: 1. Annual General May 2011.
Body Meeting :
was held on 28th
2. Election of office bearers and EC members for the period 2011-2013 was conducted 3. MR Kurup Memorial Lecture 2011 was delivered by Shri V Srinivasan, Deputy Director, PRSO Entity, VSSC, at Hotel Maurya Rajadhani, Trivandrum on 28th May 2011. 4. Four EC meetings were conducted on the months July, August and September and October 2011. 5. One day National Seminar. A one day National seminar on Non-destructive Evaluation of composite structures was conducted on 22nd October 2011 at SP Grand Days, Thiruvananthapuram. The inaugural function was chaired by Dr. A. Jayakrishnan, Vice Chancellor, University of Kerala. There were 8 invited lectures covering various area like UT, RT, Thermography, Acoustic Emission, Holography, Shearography etc. VADODARA Other Activities: Executive Committee Meeting held on 25th March , 2011
Journal of Non Destructive Testing & Evaluation
7
Basics Eddy Current Testing: Basics B.P.C. Rao Non-destructive Evaluation Division, Metallurgy and Materials Group Indira Gandhi Center for Atomic Research, Kalpakkam – 603 102, TN, India e-mail: bpcrao@igcar.gov.in
Eddy current technique is an important electromagnetic nondestructive evaluation technique that is widely used in power, aerospace, petrochemical and other industries for detection of surface cracks and subsurface damage in components made of metallic materials. Besides, it is also used traditionally for assessing the adequacy of heat treatment of alloys, as eddy currents are sensitive to changes in microstructure and stresses, which alter the electrical conductivity and magnetic permeability of the material. This paper gives a brief account of basic principle, features, applications, limitations of the eddy current technique. It also covers instruments and sensors to enable better appreciation of the technique and its capabilities.
Barkhausen emission, micromagnetic, potential drop, microwave, AC field measurement techniques etc. [1]. In these techniques, material under investigation is excited electromagnetically and the manifestation of electromagnetic fields due to material discontinuities affecting electrical conductivity or magnetic permeability or dielectric permittivity are measured using a sensor, with the exception of magnetic particle testing in which magnetic particle are used in place of a sensor [2]. EC technique is the most popular and widely used electromagnetic NDE technique. In industrial scenario,
among other electromagnetic NDE techniques, this technique finds larger number of applications. This technique finds versatile applications in power, aerospace and petrochemical industries. It is not incorrect to say that worldwide almost all the heat exchangers and aircrafts are inspected using this technique. Two main aspects behind this widespread use are excellent sensitivity to surface as well as subsurface defects and testing speed of as high as 10 m/s which no other NDE technique can match. This is especially profitable to industries as it enables rapid examination during manufacturing stages, while it drastically reduces the down time of operating plant components. Many developments are taking place in this existing NDE technique incorporating the rapid progress in the fields of microelectronics, instrumentation, sensors, computers, numerical modelling, digital signal & image processing (Fig.1) [2]. The way EC testing is practised now is different from that it was four decades ago. These concurrent advances in other fields have
INTRODUCTION Early detection and quantification of defects, microstructural variations and stresses is utmost important to ensure high quality manufacturing and safe operation of engineering components. The role played by non-destructive evaluation (NDE) techniques is well known from their wide spread use during manufacture and assembly stages and during service life of components. NDE techniques that use some form of electromagnetic excitation are termed as electromagnetic NDE techniques and some of these include eddy current (EC), magnetic particle, magnetic flux leakage, magnetic
Fig. 1 : Recent advances in eddy current testing that are responsible for its enhanced use by the industry.
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
Basics
8
enhanced the capabilities of the traditional EC technique enabling detection and sizing of incipient surface defects as well as subsurface defects, changes in microstructures and accumulated plastic deformation, stress or damage e.g. prior to crack formation etc. Such possibilities allow efficient preventive actions to be taken avoiding catastrophic failure of components. Possibilities to inspection of large areas with automation, elimination of operator fatigue & uncertainty, inspection of inaccessible as well remote areas, and on-line inspection ensuring high probability of detection and accurate sizing, all have further enhanced the acceptability of the EC technique by the industry. This technique richly benefited by the wisdom and knowledge and contributions from scientists, engineers and other experts from physics, electrical engineering, material technology, microelectronics, computers, automation and robotics domains. For better appreciation of the EC technique and correctly choose it for an application at hand, it is essential to know the basic principles, features, capabilities and limitations of the technique. That forms the objective of this paper.
PRINCIPLE EC testing works on the principles of electromagnetic induction. In this technique, a coil (also called probe or sensor) is excited with sinusoidal alternating current (frequency, f, ~ 50 Hz-5 MHz, ~ 100 mA). Following the Ampere’s law, this current generates primary magnetic field in the vicinity of the coil. When an electrically conducting material is brought close to this coil, eddy currents are induced in the material according to the Faraday’s law (refer Fig. 2) [3]. The eddy currents have very unique and interesting properties such as: Ø
They are induced currents that exist only in electrically conducting materials
Ø
They are always in closed loops, usually parallel to the coil winding (Fig. 2)
Ø
They are distorted by defects such as cracks and corrosion wall loss and by discontinuities such as edgeeffect, end-effect as shown in Fig. 2
Ø
They attenuate with depth (also axially or laterally)
Ø
Their intensity depends on material properties, electromagnetic coupling (lift-off/fill-factor) and excitation frequency, but maximum on the surface
These eddy currents also generate a secondary magnetic field, but in the opposite direction to the primary magnetic field following the Lenz’s law and this field in turn, changes the coil impedance, Z which is a complex quantity with real component, R and imaginary component, X L. Defects such as cracks, voids, inclusions, corrosion wall loss, microstructure degradation, localised stresses alter the local electrical conductivity, σ , and magnetic permeability, µ, of the material and cause distortion of the eddy currents and change the coil impedance. This impedance change, usually of the order of micro ohms, is measured using high-precision bridge circuits, analysed and correlated with defect dimensions. Alternately, the secondary magnetic field can be detected using a separate receiver coil or a solid state field detection sensor. Discontinuities or defects that cause maximum perturbation to eddy currents flow, in other words, distortion produce large change in impedance. The impedance change is also affected by excitation frequency (effect of XL or 2πfL) and electromagnetic coupling.
Fig. 2 : Principle of eddy current testing (left) and distortion of eddy current due to crack, edge-effect, surface crack, and sub-surface void (right). vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
Basics
9
The flow of eddy currents in the test material is not uniform at different depths. The eddy currents are quite denser at the surface as compared to the deep inside, an effect referred to as skin effect [4]. Theoretical standard depth of penetration of eddy currents, δ, that describes the skineffect, can be expressed as (1) where is f is excitation frequency, Hz, µ0 is magnetic permeability of free space, 4π10-7 H/m, µr is relative magnetic permeability, dimensionless, and σ is electrical conductivity, mho/ m. δ is the depth at which the surface eddy current density has fallen to 37%. EC technique capability, applicability, selection of test frequency etc. can be readily understood using equation (1). For example, depth of penetration of eddy currents, in other words, interaction of electromagnetic fields, is very low in highly conducting (e.g. Copper) as shown in Fig. 3, as compared to that of austenitic stainless steel which is less conducting. Due to the skin effect, with EC test one can readily detect the surface-breaking defects as compared to the sub-surface defects or buried defects.
The locus of impedance change during the movement of an EC coil system over the test object is called EC signal. While the amplitude of the EC signal provides information about the defect severity, the phase angle provides information about the defect depth. As depicted in Fig. 4, the impedance change can be displayed in a complex plane (real, X – imaginary, Y) as impedance plane trajectory or as a time-domain signal viz. X(t) or Y(t). In the impedance plane, magnitude and phase can be seen; however, the signal extent or defect length cannot be seen. On the contrary, in timedomain signals, phase angle information that is essential for depth estimation is absent. Selection of test frequency is very important in the EC tests and in general, it is chosen such that a maximum amplitude signal is formed for defects and with a decent phase separation from the lift-off axis. A simpler way to determine the working test frequency range involves assuming value of 1 and 2 for δ in equation (1) and calculating the extreme frequencies upon substituting σ, µ0 and µr values of the test object.
The electrical conductivity is usually expressed as percentage IACS (International Annealed Copper Standard) in which the electrical conductivity of pure copper at 25°C is taken as 5.8 x107 Siemens/meter. For example, the IACS% value of austenitic stainless steel (type 304) is 2.5 with an absolute electrical conductivity of 1.45x106 and that of Admiralty brass is 24% with a conductivity of 1.392x107 [5]. The electromagnetic coupling between coil and test object is very important. For reliable detection of defects, it is always preferable to minimise and maintain uniform liftoff or fill-factor which will be discussed later in this paper. Failing to do so will result in degradation of signal-to-noise ratio (SNR). Instead of continuous A.C. if the exciter is driven with a repetitive broadband pulse, such as a square wave which induces transient eddy currents associated with highly attenuated magnetic pulses propagating through the material, a new technique called pulsed eddy current (PEC) technique is formed. The signals reflected from defects in the object are picked up by a sensor. At each probe location, a
Fig. 3 : Interaction of magnetic fields from a coil at different test conditions. Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
Basics
10
Fig. 4 : Two types of eddy current signals viz. Impedance plane (X-Y) and time-domain (right) from three surface cracks (a, b, c) in a steel plate.
series of voltage-time data pairs are produced as the induced field decays, analogous to ultrasonic A-scan data. Defects close to the surface will affect the eddy current response earlier in time than deep surface defects. PEC technique is useful for detection of hidden corrosion in layered structures such as aircraft lap-splices and corrosion under insulation in insulated components.
treatment inadequacies, coating thickness measurements, and detection of defects in tubes, rods, bars, multi-layer structures, discs, welds, blades and other regular as well as irregular geometries [6]. Some of the attractive features of EC technique include the following: §
FEATURES
EC technique is a preferred technique for material sorting, determination of hardness, heat
§
Ability to distinguish metallic materials from non-metallic ones and sorting of materials based on difference in heat treatment, microstructure and material properties (refer Fig. 5) Ability to easily detect tight hairline cracks which cannot be seen by naked eye
§
Ability to perform tests on regular as well as irregular geometries without the need for using any couplant
§
Ability to carry out tests at more than 10 m/s speed
§
Ability to measure coating thickness as small as 5µ
§
Possibility to carry out high temperature testing, even up to 1000°C
§
Computer based automated testing, data storage, analysis and interpretation without the need for an operator
§
Possibility to perform numerical modelling for optimisation of the technique, probes and test parameters [7]
INSTRUMENTS
Fig. 5 : Response on the impedance plane for different metallic materials, enabling material sorting and determination of conductivity and permeability. vol 10 issue 3 December 2011
In EC technique the alternating current through the coil is kept constant (~ few hundred mA) and the changes in the coil impedance are measured. Since the impedance change is very small (< micro-ohms), high precision A.C. bridge (refer Fig. 6) circuits are employed. The bridge imbalance is correlated with the defect or material attribute responsible. Typical analogue EC instrument consists of an oscillator (excitation frequency, ~ 50 Hz-5 MHz), constant current supply (step down from 230 V AC), a bridge circuit, amplifier, filters, oscilloscope (to display the impedance changes in a 2-D graph or as a vector) or
Journal of Non Destructive Testing & Evaluation
Basics
11
of the test material, different probes such as surface probes (for plates), encircling probes (for rods and tubes) and bobbin probes (for tubes) with coil configurations shown in Fig. 8 are used. Appropriate selection of probe coil is important in eddy current testing, as even an efficient EC instrument cannot achieve much if it doesn’t get the right (desired) information from the coils. Fig. 6 : A.C. bridge circuit used to measure small changes in EC coil impedance.
meter display unit or decision making unit. With the micro-electronic revolution digital EC instruments have replaced the analogue EC instruments. These instruments are smart, high-sensitive, low-cost, automated, modular and efficient (Fig. 7). They are, in many instances, interfaced to personal computers, industrial computers, and laptops with possibility for easy measurements, adjustments, controls, data storage, analysis and management, all performed by suitable software. PROBES EC probe forms the basic link between EC instrument and the test material. Depending on the geometry
EC probes are induction based and are made up of a few turns of copper wire usually wound around a ferrite core with or without shielding. Every probe has an operating frequency range and impedance value matching the bridge circuit of the instrument. It is desirable to operate the probe within that range. It is essential to avoid operating near the resonance frequencies. The probes are operated in absolute (single coil), differential (two coils wound opposite) or send-receive (separate coils for excitation and detection) modes. Their design is dependent on the object geometry viz. tube, plate, bar etc. As shown in Fig. 8 and the expected type and location of discontinuity. Absolute probes are good for detection of cracks (long or short) as well as gradual variations. However, absolute probes are sensitive also to lift-off, probe tilt,
temperature changes etc. Differential probes have two sensing coils wound in opposite direction investigating two different regions of a material. They are good for high sensitive detection of small defects. They are reasonably immune to changes in temperature and the operator-induced probe wobble [5]. The most simple and widely used probe types are: ·
Surface or pancake or pencil probes (with the probe axis normal to the surface), are chosen for testing plates and bolt-holes either as a single sensing element or an array - in both absolute and differential (split-D) modes.
·
Encircling probes for inspection of rods, bars and tubes with outside access and
·
Bobbin probes for pre-and inservice inspection of heat exchanger, steam generator, condenser tubes & others with inside access. Phased array receivers also possible for enhanced detection and sizing.
While pancake or surface probes are used for testing plates and regular geometries, encircling or bobbin type probes are employed for testing tubes, rods, and other cylindrical objects. The EC probes possess directional properties i.e. regions of high and low sensitivity (impedance change). Defects that cause maximum perturbation to eddy currents are detected with high sensitivity. For good sensitivity to small shallow defects, a small probe should be used.
Fig. 7 : Advanced general purpose digital instruments for static as well as dynamic EC tests and for multi-frequency EC inspection of non-ferromagnetic and ferromagnetic heat exchanger tubes (Courtesy: M/s. Technofour, Pune). Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
Basics
12
Fig. 8 : Basic configuration of coils in eddy current probes.
Similarly, in order to detect subsurface and buried defects, large diameter high throughput probes operating at lower frequencies are necessary. As a general rule, the probe diameter should be less than or equal to the expected defect length and also comparable to the thickness of the test object. The sensing area of a probe is the physical diameter of the coil plus an extended area of 4d due to the magnetic field spread. Hence, it is common to use ferrite cores as well as shields with high m and low s, to contain the field without affecting the depth of penetration.
Figure 9 shows some typical EC probes developed for specific applications [3, 8]. Coupling of magnetic field to the material surface is important in EC testing. For surface probes, it is called “lift-off” which is the distance between the probe coil and the material surface. In general, uniform and very small lift-off is preferred for achieving better detection sensitivity to defects. The electromagnetic coupling in the case of tubes/bars/rods is referred to as “fill-factor”. It is the ratio of square
of tube diameter to square of coil diameter for encircling coils and is expressed as percentage (dimensionless) Fill factor = ( D2t / D2p )* 100
(2)
where Dp is the probe inner diameter and Dt is the tube outer diameter [3,5]. Usually, 70-90% “fill-factor” is targeted for reliable inspection. The encircling probes exhibit reduced sensitivity for shallow and localised defects and for such applications, motorised rotating probe coil
Fig. 9 : Different types of EC probes. vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
Basics (MRPC), phased-array, plus-point etc. are used. The inspection data from these probes can be displayed as images which allow easy identification of circumferential location of defects. For inspection of irregular and inaccessible regions, flexible and conformal sensors are employed. The pancake type probes show reduced sensitivity for subsurface and buried defects and for such needs, integrated probes with coils for excitation and solid state sensors for reception are very attractive [2]. Such integrated probes are useful for inspection of rivets and multi-layer structures in aircrafts and for detection of deeply located (> 10 mm) defects in steel components. TEST PROCEDURE
General EC test procedure for detection of defects involves calibration of EC instrument using reference standard defects in a material with similar chemical composition and geometry as that of the actual component. Artificial defects such as saw cuts, flat bottom holes, electro-discharge machining (EDM) notches are used while well characterised natural defects, cracks in failed or withdrawn components are always preferred. Instrument test parameters such as excitation
13
frequency, gain, phase angle etc. are optimised for a desired performance. In general, signal phase is rotated such that it is parallel to lift-off or wobble axis and phase separation between ID and OD defects is nearly 90 degrees. A suitable EC signal parameter, e.g. signal peakto-peak amplitude or phase angle is identified and an appropriate threshold is determined for incorporating accept/reject criterion. When defect sizing is required, a calibration graph between signal parameter and defect size is generated and used [3]. During actual testing, any region that produces EC signals with parameter greater than the threshold is recorded defective, while its equivalent size is determined using the calibration graph. Similar procedure is followed for material sorting, conductivity measurement, microstructure characterisation, and coating thickness measurement. TESTING NONFERROMAGNETIC TUBES For periodic monitoring of corrosion of tubes in heat exchangers, steam generators and condensers in power, petrochemicals, fertiliser and other industries, EC technique is employed because of its ease of operation,
sensitivity, versatility, speed (~ 10 m/s) and repeatability. This technique can detect wall thinning, cracks, pitting, stress corrosion cracking, hydrogen embrittlement, carburization, denting and crude deposits etc. Typical EC signals from an ID defect, OD defect and hole in a heat exchanger tube and a surrounding steel support plate are shown in Fig. 10 for absolute and differential bobbin probes. Using phase discrimination, it is possible to readily distinguish various defects as well as support plates. However, single frequency EC technique is inadequate for detection of defects under support plates, baffle plates and in the presence of probe wobble. Many a time, it is under the support plates corrosion damage takes place. To eliminate signals from support plates, multifrequency technique is employed. This technique involves simultaneous excitation of two or more frequencies in an EC coil and processing of the corresponding signals to suppress the contributions from disturbing sources, similar to solving a system of linear equations [4, 6, 8]. The multifrequency test procedure is described in detail in ASME Section V, Article 8, Appendix 1.
Fig. 10 : Typical absolute (left) and differential probe (right) EC signals for an ID defect (A), OD defect (B) and hole (C) in a heat exchanger tube and for a steel support plate (D). Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
14
TESTING FERROMAGNETIC TUBES Examination of ferromagnetic tubes is difficult using conventional EC test procedures due to high relative magnetic permeability which restricts penetration of eddy currents and produces disturbing signals due to continuously varying permeability. This disturbance can be eliminated by employing bias or saturation direct current (D.C) which saturates the material magnetically and makes the tube material behave nonferromagnetic, thus, allows conventional EC testing after which the tubes are to be demagnetised. Typical test set up is shown in Fig. 11. However, in the case of installed tubes of smaller diameters, saturation units cannot be accommodated due to limited access. For mildly magnetic materials, partial saturation using high-energy permanent magnets such Nd-Fe-B is a possibility. The other technique possible is remote field eddy current (RFEC) technique [2]. This technique uses low frequency excitation and a separate receiver coil kept at two to three tube diameters away from exciter coil. The phase lag of induced voltage in the receiver coil with respect to the exciter is measured using lock-in amplifiers and correlated with wall loss or defect depth. The advantages of RFEC technique include ability to test tubes
with equal sensitivity to internal and external wall loss and linear relationship between wall loss and measured phase lag. RFEC technique can achieve inspection speeds of 1 m/s. It is used for inspection of carbon steel and other ferromagnetic tubes in process industry. One recent application of this technique is inservice inspection of steam generator tubes of sodium cooled prototype fast breeder reactors with 23 m long modified 9Cr-1Mo ferromagnetic steel tubes For this application, a comprehensive RFEC technology comprising of instrumentation, probes, robotic device has been developed as a solution to the problem of smaller diameter, expansion bends, support plates, one-side access and electrically conducting sodium deposits. CHARACTERISATION OF MICROSTRUCTURES
During manufacturing stages of components made of alloys, heat treatment is given to ensure required levels of mechanical and physical properties and desired microstructures. Likewise, during service life of components, it is essential to ensure that there is no undesirable degradation in microstructures. EC technique is useful for assessing these two situations as it exploits measurement of changes in electrical conductivity and magnetic permeability. Changes
Fig. 11 : D.C. Saturation based EC testing of ferromagnetic steel tubes. vol 10 issue 3 December 2011
in microstructure, precipitate size and distribution, cold work, deformation, dislocation pile-up etc. alter the coil impedance or induced voltage in a pick-up coil. The magnitude and phase of induced voltage or impedance change are used for quantitative characterisation of microstructures and to estimate the volume fraction of various phases present. Cold worked and annealed conditions, e.g. in stainless steel 316 or 304 effect electrical conductivity in opposite directions and straininduced martensite, being a magnetic phase, increases the magnetic permeability. This phase can be detected using EC technique. Specimens subjected to heat treatments are used to simulate the service exposed conditions and the expected microstructures. Usually, EC measurements for microstructure characterisation are location-based. In general, absolute probes are used and analysis is based on impedance plane signal interpretation. Reference standards with known electrical conductivity and magnetic permeability are used also for establishing calibration graph, apart from specimens heat treated to different ageing conditions through measurement of conductivity and permeability. EC technique has been used to characterise microstructures in titanium alloy (VT 14 alloy Ti-4.5Al3Mo-1V) subjected to a series of heat treatments consisting of solutionizing for 1 h at selected temperatures in the range of 9231303 K at an interval of 50 K, followed by water quenching. This treatment produces variety of microstructures due to controlled ιβ transformation and formation of various phases. The experimentally measured EC response for various specimens is shown in Fig. 12 along with that of the reference standards viz. Hastelloy-X, Hastelloy-B and Ti6Al-4V. It has been found that, both magnitude and phase angle of
Journal of Non Destructive Testing & Evaluation
Basics15 REFERENCE STANDARDS Reference standards are used for adjusting the eddy current instrument sensitivity to enable detection of desired size of defects and quantification of conductivity, permeability and material thickness etc. They are also used for sizing defects [2]. Some commonly used standards in EC testing by ASME (American Society for Mechanical Engineers), BS (British Standards), ASTM (American Society for Testing of Materials) and IS (Bureau of Indian Standards) are: ·
ASME, Section V, Article 8, Appendix 1 and 2), Electromagnetic (EC) testing of heat exchanger tubes
·
ASTM B 244 Method for measurement of thickness of anodic coatings of aluminum and other nonconductive coatings on nonmagnetic base materials with EC instruments
·
ASTM B 659 Recommended practice for measurement of thickness of metallic coatings on nonmetallic substrates
·
ASTM E 215 Standardising equipment for electromagnetic testing of seamless aluminium alloy tube
·
ASTM E 243 Electromagnetic (EC) testing of seamless copper and copper alloy tubes
·
ASTM E 309 EC examination of steel tubular products using magnetic saturation
·
ASTM E 376 Measuring coating thickness by magnetic field or EC (electromagnetic) test methods
·
ASTM E 426 Electromagnetic (EC) testing of seamless and welded tubular products austenitic stainless steel and similar alloys
·
ASTM E 566 Electromagnetic (EC) sorting of ferrous metals
·
ASTM E 571 Electromagnetic (EC) examination of nickel and nickel alloy tubular products
·
ASTM E 690 In-situ electromagnetic (EC) examination of non-magnetic heatexchanger tubes
·
ASTM E 703 Electromagnetic (EC) sorting of nonferrous metals
·
BS 3889 (part 2A): 1986 (1991) Automatic EC testing of wrought steel tubes
·
BS 3889 (part 213): 1966 (1987) EC testing of non-ferrous tubes
Fig. 12 : Impedance plane response for various heat treated specimens at 150 kHz.
impedance change decrease with increasing solutionizing temperature up to 1123 K and this is attributed to decrease in α phase (reduction in electrical conductivity). Beyond 1123K, formation of α′ martensite dominates the interactions and results in increase in effective electrical conductivity and hence, the impedance change. Beyond 1273K, magnitude and phase angle reach a constant maximum, due to 100% formation of α′ martensite. Comparison of impedance magnitude and phase angle with hardness measurements has established that EC technique can be implemented in production line to quickly assess the adequacy of heat treatment.
·
·
·
·
·
·
Quality assurance of austenitic stainless steel tubes, plates and welds.
·
Inspection of installed heat exchanger/steam generator/ condenser tubes (single and multifrequency)
On-line automated saturation based quality assurance of steel (ferromagnetic) tubes. Location of garter springs in PHWRs and measurement of gap in coolant channels Detection of intergranular corrosion (IGC) in stainless steels (316, 316L and 304 L) Detection of weld centre line in austenetic stainless steel welds at high temperature
·
Measurement of coating thickness of SiC on carbon-carbon composites
·
Sorting of materials based on electrical conductivity and magnetic permeability
APPLICATIONS A few specific practical applications of EC technique are given below for better appreciation of the technique.
Detection of surface as well as subsurface defects in multi-layer aircraft structures (single frequency, multi-frequency & pulsed techniques)
·
·
·
Journal of Non Destructive Testing & Evaluation
Characterisation of heat treated as well as degraded microstructures in alloys Non-contact detection of metallic objects, land mines, security metal detectors Monitoring of liquid levels and for position encoding
vol 10 issue 3 December 2011
16
·
IS 6398:1983 Code of practice for EC testing of ferrous seamless pipes and tubes
·
IS 11612: 2004 Code of practice for EC testing of non-ferrous seamless pipes and tubes
·
IS 13190: 1991 Recommended practice for EC examination by rotating probe method of round steel bars.
·
IS15540:2004 Recommended practice for EC testing of installed non-ferromagnetic heat exchanger tubing using duel frequency method.
LIMITATIONS
Like other NDT technique EC technique has certain limitations too. But interestingly, most of the original limitations of the technique in 60s and 70s have been overcome by the advances in instrumentation, sensors, computer and signal and image processing techniques. Some of the important limitations include: ·
Applicability to only electrically conducting (metallic) materials
·
Inspection of installed ferromagnetic components with the exception of tubes can be inspected by remote field EC technique
·
Difficulty to separate the influence of one desired variable in the combined presence (at the same location beneath the probe) of several other disturbing variables such as stress, microstructure, texture, anisotropy etc. that simultaneously change conductivity and permeability.
·
Inability to identify circumferential location of a defect when encircling or bobbin coils are used.
·
Difficulty in detection of a small defect under a large defect
·
Inability to detect defects at the centre of rods using encircling coils
·
Need for skilled personnel for interpretation of signals and results
vol 10 issue 3 December 2011
SUMMARY
REFERENCES
Working on the principle of electromagnetic induction, eddy current technique is a widely used NDE technique for detection of surface and sub-surface damage. The attractive features of this technique include ease of operation, high sensitivity to tight cracks, versatility, extremely high testing speeds (up to 10 m/s), repeatability and reliability. This technique can detect wall thinning, cracks, pitting, stress corrosion cracking, hydrogen embrittlement, carburization, denting and crud deposits etc. This technique finds a lot of applications in engineering industry including material sorting, determination of hardness, heat treatment adequacy assessment, material property determination, coating thickness measurements, and detection of defects in tubes, rods, bars, multilayer structures, discs, welds, blades and other regular as well as irregular geometries. Successful testing requires selection of proper instrument and probes, optimisation test frequency and use of reference calibration standards. When appropriate standards are used, not only detection of defects but also their sizing is possible using eddy current technique.
1. B.P.C. Rao and T. Jayakumar, Discontinuity characterisation using electromagnetic methods, J of NonDestructive Testing & Evaluation, Vo.2, No.2, 2002, pp 23-29 2. B.P.C. Rao, T. Jayakumar, Baldev Raj, Electromagnetic NDE Techniques for Defect and Microstructural Characterization, in Ultrasonic and advanced methods for nondestructive testing and material characterisation, Ed. C.H. Chen, World Scientific Publishing Co. (Singapore), June 2007, pp. 247-278. 3. B.P.C. Rao, Introduction to Eddy Current Testing, Narosa Publishing, New Delhi, India, February, 2007. 4. H.L. Libby, “Introduction to Electromagnetic Non-destructive Test Methods”, Wiley-Interscience, New York, 1971 5. V.S. Cecco V.S, G. Van Drunnen and F.L. Sharp, “Eddy current manual: test method”, Vol.1, AECL-7523, Chalk River, Ontario, Nov., 1983. 6. D.J. Hagemaier, “Fundamentals of Eddy Current Testing”, ASNT, Columbus, OH, USA, 1990. 7. W. Lord, “Electromagnetic methods of Non-destructive Testing”, Gordon and Breach, New York, 1985. 8. Moore Patrik, Udpa S.S, Nondestructive testing handbook. 3rd edition. Vol.5: Electromagnetic testing, ASNT, Columbus, OH, USA, 2004.
ACKNOWLEDGEMENTS Author expresses special thanks to Dr. T. Jayakumar, Mr. S. Thirunavukkarasu, Mrs. B. Sasi of NDE Division, India Gandhi Center for Atomic Research, Kalpakkam, India.
Journal of Non Destructive Testing & Evaluation
17
Horizon
Nonlinear Ultrasound and Harmonic Imaging for NDE Dr. CV Krishnamurthy
wave number, and ω is the angular frequency. Assuming that the nonlinearity in the solid is small, the solution to the nonlinear wave equation for this time-harmonic wave is obtained by a power series U
U
It appears that there may not be a single mechanism for the nonlinear response that would be common to all classes of materials. Nevertheless, experimentally, the nonlinear response is usually described in terms of a
single parameter (the symbol β is commonly employed) with the understanding that this is a “lumped” parameter or some kind of a macroscopic average. It must be kept in mind that certain studies are specific to observing second harmonic components and the same symbol is used to describe it. The equation governing the propagation of a longitudinal wave (of velocity co) along the x-axis of an isotropic solid (of density ρ0) is:
where U is the displacement and CII=ρ0c20 is a linear combination of the second-order elastic constants, while CIII includes the second- and third-order constants. The nonlinear parameter is defined as
β =−
C III 2C II
The linear wave equation results for β = 0. Depending on the sign of CIII, nonlinear distortion of longitudinal waves in solids can result in a sawtooth-like velocity waveform if C III < 0 or an N-type wave if C III > 0. The majority of solids possess CIII < 0 and hence β > 0, however, for some materials like glass or fused silica, β < 0. If a time-harmonic plane (displacement) wave U=A1cos(kx–ωt), where A1 is the amplitude, k is the
Journal of Non Destructive Testing & Evaluation
A0+A1cos(kx–ωt) +A2cos[2(kx–ωt)]+. . .
This series can be more explicitly written as
Centre for NDE and Department of Physics, IIT Madras
Nonlinear response to ultrasound has been observed in a variety of materials ranging from liquids such as water, soft matter such as biological tissues, polycrystalline metals, amorphous materials such as glasses, composites and porous materials such as stones and rocks. The linear response of any material to ultrasound can be said to arise from Hooke’s law irrespective of the nature of the material and has been well understood in terms of the fundamental interatomic forces. Experiments on nonlinear response indicate that in most cases the nonlinearity is not arising from the higher-order effects of the interatomic forces. In other words, nonlinearity sets in at lower amplitudes than what would be expected on the basis of the strength of the interatomic forces. Unlike the linear response, studies of the observed nonlinear response in these materials suggest several causes for the nonlinear response that depend on mesoscale and macroscale structural features - at the mesoscopic level, some of these feature dislocation networks and grain boundaries while others feature microcracks; at the macroscopic level, cracks are common in most ductile or brittle solids.
=
=
– –81 βk2A12x + A1cos(kx–ωt) + –81 βk2A12x[2(kx–ωt)]+....
It is noted that the amplitude of the second-harmonic displacement is proportional to the acoustic nonlinearity parameter and a subharmonic; that is, the static displacement is induced by the material nonlinearity. The acoustic nonlinearity parameter is determined experimentally by measuring the absolute amplitudes of the fundamental (A1) and the secondharmonic (A2) displacement signals, or
It is to be noted that the discussion neglects the effect of attenuation losses that may be present in the fundamental and second-harmonic. If the difference in attenuation rates at the fundamental and the secondharmonic frequencies is large, then a correction factor must be included in the measurement of β. Experimentally, β is evaluated as the ratio of the amplitude of the higher harmonics to the square of the amplitude of the fundamental (all in the frequency domain). It is normally assumed that the stress or strain associated with an ultrasonic wave as it propagates through a medium is small enough to be within the linear regime of the stress-strain curve of that medium – the regime where Hooke’s law is valid. It is natural to expect that the received vol 10 issue 3 December 2011
18 HORIZON
signals would only undergo structurebased dispersion and amplitude changes due to geometric attenuation and material absorption. For small amplitude vibrations, it is found that the restoring force on each atom is linearly proportional to the displacement of that atom from the equilibrium position. The validity of Hooke’s law on a macroscopic scale arises out of the linearity at the atomic scale. Considerable degree of spatial averaging takes place though when we go from the microscopic scale to the macroscopic scale - as much as to alter an inherently anisotropic crystalline metal to an isotropic polycrystalline metal! Averaging takes place again when we compare a metal that is solidified from melt with a metal that has been forged, rolled and cut into a component for industrial applications. The processes of forging, rolling etc., introduces a complex microstructure involving dislocation networks and grains and yet on a macroscopic scale, regular ultrasonic experiments barely reveal anything!
expressed in a power series of displacements, with the first term describing linear response, the first and second terms describing secondorder effects, the first, second and third terms describing third-order effects etc. These higher-order effects can be expressed through higher-order elastic constants for crystalline materials or through Gruneisen parameters for crystalline and non-crystalline materials. However, since the nonlinearities involved happen to be small, the interatomic displacement amplitudes required are large to be able to manifest in regular ultrasonic experiments. Table 1 gives a comparison of the “inherent” nonlinearity parameter manifest in different crystalline materials.
It is seen that the nonlinearity parameters are strongly ordered for a given propagation direction according to the type of crystal structure. The relative lack of dependence on bonding compared to crystal structure is also apparent from the Table. The dependence of the acoustic nonlinearity parameters on the crystalline structure suggests that the geometry of the local atomic arrangement and shape, rather than the strength, of the interatomic potential are dominant factors in determining the magnitude of β. Several techniques have been developed to measure nonlinearity in different classes of materials through the acoustoelastic effect, harmonic generation, resonant modal vibration, frequency modulation, and
Measurements of sound velocity and attenuation performed on a Ti–6Al– 4V dog-bone sample show similar behaviors observed in other fatigued materials such as Al 2024-T4 and 410Cb stainless steel. These two acoustic quantities do not seem to be representative of the level of fatigue damage. Large amplitude interatomic displacements do lead to nonlinear effects, two of which are well known, thermal expansion and thermal conductivity of solids. Other nonlinear effects known for some time now are (i) the hydrostatic pressure dependence of the elastic moduli, and (ii) acousto-elastic effect - ultrasonic waves travel at different speeds depending on whether they are propagating parallel or perpendicular to the applied stress. These nonlinear effects are characterized through restoring forces vol 10 issue 3 December 2011
Fig. 1 : (Left) Longitudinal velocity of sound – uncorrected for dimensional changes during fatigue loading (Right) Attenuation as a function of fatigue level for Ti–6Al–4V with duplex microstructure at 10 MHz frequency (from Frouin et al, J. Mat. Res., 14 (1998), 1295-1298). Journal of Non Destructive Testing & Evaluation
19 HORIZON
even rupture) of the contact between the surfaces. Figure 3 shows the nonlinear bulk wave reflection amplitudes studied for 20 MHz SV-wave incident at 45° on a glass-glass contact interface as a function of contact pressure.
TABLE 1 Comparison of Crystal Structure (Space Group), Bonding, and Range of Values of Longitudinal Mode Acoustic Nonlinearity Parameters
From Cantrell, J. Appl. Phys., 76 (1994), 3372
wave mixing. Where applicable, the acoustoelastic effect is capable of measuring all three third-order elastic constants but is difficult to employ as it relies on small changes in acoustic velocity. Harmonic generation has received the most research effort and relies on the nonlinear material properties generating harmonics of the input signal. This approach is limited by the difficulty in isolating the sources of the measured nonlinearity. Resonant modal vibration, frequency modulation and wave mixing appear to be attractive for practical use and capable of quantifying nonlinearity in different types of materials.
nonlinear response from flat bottom holes in non-weld and electron beam weld regions of a copper block. Unlike atoms in the interior of solid media, atoms in the vicinity of interfaces experience asymmetric strains. The contact acoustic nonlinearity (CAN) is regarded to be due to the lack of stiffness symmetry for near-surface strain across the interface. The compression elasticity is anticipated to be higher than that for a tensile stress because the latter is accompanied by weakening (and
The amplitudes of the first four reflected harmonics as functions of contact pressure are shown in Figure 3. For numerical evaluation of Contact Acoustic Nonlinearity (CAN) efficiency for bulk waves, normalized harmonics amplitude is introduced: Nn = Unω/Uω. In Table 2 the values of N, for CAN calculated from the data of Figure 3 are compared with the harmonics generation efficiency achievable for conventional (material ) non-linearity (superscript MN). The latter calculations were carried out for longitudinal wave of the same intensity and frequency propagating in the aluminium alloy with known third- and fourth-order elastic constants. It is seen that CAN efficiency for bulk waves (especially
NONLINEAR RESPONSE THROUGH HARMONIC GENERATION Taking cue from medical ultrasonics, where nonlinear effects of water and soft matter such as tissue have been observed and exploited for imaging with the harmonic components of the received signal, immersion studies on samples with side drilled, flat bottom and round bottom holes were carried out. A transducer with spherically concentric elements with the annular shaped outer element as a transmitter at 2.26 MHz and the circular shaped inner element as a receiver at 4.83 MHz and radius of curvature of 210 mm was used in the tone burst mode. The water path was set to 30 mm for both blocks. Figure 2 shows the
Fig. 2 : In the echo from a FBH (non-welded zone) the harmonics of 2nd to the 4th order (top) are present, whereas the echoes from an FBH in the weld zone show the presence of the 2nd and 3rd harmonics only (bottom), possibly due to a larger attenuation in the weld region.(from Wu and Stepinsky, IEEE Ultrasonics Symposium (2000), 801804.)
Journal of Non Destructive Testing & Evaluation
Table 2
(from Solodov, Ultrasonics 36 (1998), 383–390). vol 10 issue 3 December 2011
20 HORIZON
Fig. 3 : Amplitudes of the first four reflected SV-harmonics as functions of contact load on glass-glass interface: (o) ω; (p) 2 ω; (∆) 3 ω; (l) 4 ω. (from Solodov, Ultrasonics 36 (1998), 383–390).
for the third- and fourth-harmonics) is much higher than that for material non-linearity. Metals subject to fatigue undergo significant micro-structural changes leading initially to microcracks and finally failing often dramatically. Considerable amount of work has been done over three decades on different metals and alloys to identify precursors to failure in the context of NDE. Figure 4 is an example of a controlled study on Ti-6Al-4V indicating a dramatic change in the second harmonic signal (180% increase in nonlinear factor). Cantrell developed a model for the nonlinearity in metals fatigued due to cyclic loading based on dislocation
dipoles and monopoles with microstructural data to corroborate. The key features of the evolution of the micro-structure during cyclic loading in metals are (a) continuous generation of large numbers of dislocations throughout the fatigue life, (b) to-and-fro motion of the dislocations during cyclic loading leads to the mutual trapping of dislocations of opposite polarity moving on glide planes, (c) accumulating dislocation dipoles selforganize into the vein structure arrangement, (d) increasing dislocation density of vein structure from continued cycling results in an elastic instability that drives the transformation of vein structure into persistent slip bands (PSBs). Cantrell’s model determines the nonlinearity parameter in terms of the contributions from the volume fraction of dislocation dipoles and the volume fraction of dislocation monopoles leading to the assessment of the nonlinear parameter as a function of the number of fatigue cycles. Figure 5 shows a comparison between his model calculations and experiments on Aluminum and Steel samples.
NONLINEAR RESPONSE THROUGH MODAL VIBRATIONS The vibration response over a certain frequency range encompassing the considered natural frequency is investigated using a stepped sine procedure for increasing levels of external excitation. For each steady - state response, a lock – in virtual instrument determines the acceleration level at the fundamental frequency. The harmonics are stored as well. The apparatus (including a
The agreement of the model calculations with measured data indicates that (a) the role of dislocation monopoles and dipoles in evolution of fatigue-generated structures is validated, (b) suggests that evolutionary time-line of
Fig. 4 : Amplitude of the second harmonic signal versus the fundamental signal for various stages of fatigue in Ti–6Al–4V with duplex microstructure. (from Frouin et al, J. Mat. Res., 14 (1998), 1295-1298). See Figure 1 for the material’s linear response. vol 10 issue 3 December 2011
substructural organization is similar in the two metals (and in agreement with pure metals from which the model is derived), and (c) offers promise that measurements may be used to quantify fatigue damage accumulation in a variety of metals at all levels of fatigue.
Fig. 5 : Graph of nonlinearity parameter of aluminum alloy 2024-T4 fatigued in stress-controlled loading at 276MPa (left) and of 410Cb stainless steel fatigued in stresscontrolled loading at 551MPa (right) plotted as a function of percent full fatigue life (from Cantrell, Rev.of Prog. Quant. Nondestr. Eval., 28 (2009), 19-32)
Journal of Non Destructive Testing & Evaluation
21 HORIZON
16-bit A/D converter) is capable of measuring accelerations down to 10-2m/s2, which typically corresponds to inferred strains in the order of a few nanostrains. In order to monitor the resonance peak shift as a function of the acceleration amplitude, typically 20 resonance sweeps are made at successively increasing drive voltages over the same frequency interval. Examples of the raw data sets obtained for the first bending mode on the intact RC beam and after the first loading step are shown in Figure 6.
mode of a beam, which has a stress concentration in the middle of the sample and displacement nodes at a distance of 0.224 L from both edges, with L the length of the sample (120 mm). Each sample is supported by two nylon wires at the node lines. A 1000 period burst excitation at a given amplitude and with a frequency close to the fundamental flexural resonance frequency is applied by a loudspeaker (dia 32 mm, focused by
a cone to 20 mm) centered in the middle of the sample. The response is measured by a laser vibrometer near one of the edges. The analysis procedure yields the evolution of the frequency (f) and damping characteristic (α) as a function of the amplitude A in the decaying signal. If no nonlinearity is present we should obtain a constant frequency and a constant damping characteristic for all amplitudes.
It may be noted that the frequency span on both figures is markedly different. For the damaged case, we notice an obvious frequency shift as a function of the excitation level. At the same time, the resonance curve becomes increasingly asymmetric. Alternately, samples can be studied by a Single Mode Nonlinear Resonant Ultrasound spectroscopy technique which measures the amplitude dependence of the resonance behaviour of a single mode of the samples. The technique is an extension of the simple ringdown measurement for a resonant mode. When a sample is subjected to a sinusoidal excitation at or near the resonance frequency, its response will be given by Ae-αtsin(2πft + ϕ) an exponentially decreasing sinusoidal signal amplitude when the excitation is turned off. The characteristic decay of the amplitude is a measure of the modal damping ratio and the dominant frequency in the decaying signal corresponds exactly to the resonance frequency of the sample. By increasing the excitation voltage and analyzing the ring-down, one can investigate the nonlinear effect on both resonance frequency and damping ratios. The mode under consideration in this study is the fundamental flexural
Fig. 6 : Measured response for the intact (a) and damaged state (b) of an RC beam after the first loading (from Van Den Abeele and Visscher, Cement and Concrete Research 30 (2000), 1453 – 1464).
Fig. 7 : Ultrasonic amplitude C-scan of unidirectional composite (CF/epoxy laminate) exposed for one hour at three temperatures. Delaminations clearly appear at 290°C whereas no sign of damage is seen at 285°C. The measured value of the interlaminar shear strength for the same type of samples changes from 121 MPa for nonexposed samples to 84 MPa at 285°C and to 43 MPa for samples at 300°C. (from Ref [5]).
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
22
Fig. 8 : Nonlinear Q-factor for two samples at different heating temperatures and exposure times - (a) 250° C for 45' and (b) 300° C for 60'. (from Ref [5])
However, when nonlinearity is present and related to the thermally induced micro-damage, we may expect a stronger dependence of the frequency (f) and damping (α) on the amplitude for increasing microdamage. Figure 8 shows the Q factor (Q = πf/α) indicating that this material parameter is amplitude dependent and that the nonlinearity is higher for samples with increased thermal damage. In order to quantify the degree of nonlinearity, the linear proportionality coefficient g between the relative resonance frequency shift and the strain amplitude (from using strainacceleration conversion) is calculated. It may be noted that g only represents a global quantification
of the nonlinearity, integrated over the whole sample and has no direct information on the localization of the defects. The nonlinear parameter γ in Figure 9 shows an overall increase with increasing exposure time and heat temperature, up to a factor 10 with respect to the reference value. As a comparison with a measure of the linear characteristics of the sample (traditional ultrasound), the values of the linear (or low-amplitude) Q-factor of each of the samples are also displayed on the right of Figure 9. Apparently the linear attenuation for heat-treated samples reduces as a function of exposure time and heat temperature (Q-factor increases). If the linear measure of attenuation
would be connected to damage, the Q-factor would be expected to decrease rather than increase. The effect seen in the linear attenuation is regarded not as a measure of the microdamage, but is believed to be related to the chemical and physical change in linear material parameters due to the thermal loading. NONLINEAR RESPONSE THROUGH INTERACTING WAVES Non-collinear method is based on the fact that material nonlinearities cause interaction between two intersecting ultrasonic waves. Under certain circumstances, this can lead to the generation of a third wave with a frequency and wavevector equal to
Fig. 9 : Nonlinearity g (left) and linear Q-factor (right) for all 21 samples as a function of heating temperature and exposure time. (from Ref [5]). vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
23
Fig. 10 : (left) Schematic of noncollinear inspection system. (right) Received ultrasonic signals at frequency of nonlinear interaction (a) with both input transducers fired simultaneously, (b) with a single input transducer (from Croxford, Drinkwater and Wilcox, AIP Conf. Proc. (2011), 1335, 330-337)
the sum of the incident wave frequencies and wavevectors, respectively. Theoretically, there are several incident wave combinations that can achieve this; however, practical material constraints to the theory lead to the interaction of two shear waves generating a longitudinal wave as the most useful case. This approach is advantageous offering the potential for frequency, modal and spatial separation of the nonlinear signals. The configuration shown in Figure 10 allows single sided measurements to be made and is employed for later results in an immersion tank for scanning measurements. As seen on the right bottom of Figure 10 there is very little signal present at the time of arrival for any wave coming from the volume of interaction. This is due to the lack of interaction as a result of the single incident wave. When both waves are fired a clear signal appears (right top of Figure 10) at this location indicative of the material nonlinearity. If the two signals with only a single transducer are summed
then the magnitude in the expected arrival time for the nonlinear signal gives a measure of the underlying system nonlinearity. NONLINEAR RESPONSE THROUGH FREQUENCY MODULATION Two methods employing the modulation of ultrasound by vibration have been developed, namely, VibroModulation (VM) and ImpactModulation (IM) methods. Vibroacoustic modulation (VM) method employs forced harmonic vibration of a structure being tested while impact modulation (IM) method uses impact excitation of natural modes of vibration of the structure. The impact is produced by an instrumented hammer equipped with changeable tips (plastic or metal) and a sensor. This sensor sent an impactgenerated triggering signal to a data acquisition system. The ultrasonic part of the setup consists of two piezo-ceramic water-coupled transducers or piezo-ceramic disks
Journal of Non Destructive Testing & Evaluation
attached to one side (as shown) or the opposite sides of the sample. One disk is used as a transmitter while the other one is a receiver. Static and dynamic (vibration) stresses applied to the sample are controlled with a strain gage attached to the opposite (from a crack) side of a sample. Figure 11 shows the schematic and typical results for defective steel bars. The VM is based on modulation of the ultrasonic signal by low frequency vibration in the presence of a flaw such as a crack, delamination or poor quality bonding. The vibration varies the contact area within a contacttype interface (or alters interface opening) modulating the phase and amplitude of the higher frequency probe wave passing through this variable contact area. The resulting modulated signal contains new frequency (side-band) components, which are associated only with the flaw and can be easily detected. This feature is especially advantageous for differentiation between integrity reducing defect and other structural vol 10 issue 3 December 2011
24
Fig. 11 : (Top) Schematic of the experimental set up for the IM method. It is similar for the VM method except that the hammer is replaced by a shaker. (Middle) Spectra of the probing ultrasonic signals obtained with the IM method: (a) steel pipe without cracks; (b) steel pipe with stress-corrosion cracks.(Bottom) Spectra of the modulated ultrasonic signal obtained with the VM method in steel bars - sample #2-1 has no crack while sample #2-2 has a crack (Donskoy et al, NDT&E Int. 34 (2001), 231–238).
inhomogeneities. The modulation effect has been observed experimentally for various types of defects: cracks, disbondings, delaminations, and microstructural material defects. HARMONIC IMAGING The nonlinear response of defects has been dramatic in the spectral vol 10 issue 3 December 2011
domain - the intact parts of the material outside the defect vibrate linearly, i.e. with no frequency variation in the output spectrum while the defective region generates responses with rich spectral content. As the pulse-type modulation produced by the excitation (pump) wave contains several frequencies, the output signal acquires the
combination frequencies (nν1 ± ν2) around the ultra-harmonics ní1. If the probing wave is strong enough to induce some nonlinearity in the contact, the modulation spectrum expands to multiple side lobes around both, the pump frequency and the probe frequency (nν1 ± mν2). The magnitudes of the modulation sidelobes are indicators of nonlinearity
Journal of Non Destructive Testing & Evaluation
25
of the defect and used for qualitative NDT of cracked flaws in metal parts, concrete and composites.
Fig. 12 : (left) NSLV-imaging of fatigue crack in AL cylinder: 4th (left), 9th (middle) and 16th harmonic images.(right-top) photo of the laser welded joint measured; (right â&#x20AC;&#x201C; bottom) NACE image of laser welded joint in steel.(from Ref. [6]).
Fig. 13 : C-scan of CFRP sample with impact damages (top - transmission mode); C-scan at 2nd harmonic (middle);C-scan with an excitation frequency of 320 kHz (bottom) (from Pfleiderer et al (2003), NDT.net - February 2003, 8 No.2)
Fig. 14 : Ultra-frequency pair imaging of corner delamination in C/C-SiC-fibre reinforced ceramic: excitation frequency 20 kHz; ultra-frequency pair images at 32.25 kHz (left) and 47.75 kHz (right).(from Ref.[6]) Journal of Non Destructive Testing & Evaluation
Figure 12 shows images of the vibration spectra of defects with a laser scanning vibrometry adapted for nonlinear measurements (nonlinear scanning laser vibrometry (NSLV). A piezostack transducer connected to CW high-power source was used for generation of intense ultrasonic waves (frequencies 20 and 40 kHz, strain amplitudes up to ~10-3). For plate-like specimens studied, it excites flexural waves whose outof-plane particle velocity induces frequency modulation of the laser light reflected from the surface of the specimen. After demodulation, the spectrum of the vibrations (particle velocity output) is obtained over 1 MHz bandwidth by FFT. The nonlinear air-coupled emission (NACE) NDE-application was found to be particularly beneficial in metallic components where low acoustic damping facilitates the formation of standing waves which produce a strong spurious background in the nonlinear scanning laser vibrometry (NSLV). Air-coupled ultrasonic transducer provides an alternative to laser Doppler vibrometry for non-contact scanning and imaging applications. Figure 13 shows images obtained by this technique to detect areas of impact damage in CFRP that had a piezo-actuator bonded to it. The aircoupled ultrasonic transducer at 450 kHz was used to detect the harmonics of a low frequency actuator. The C-scan (transmission mode) displays locations of the impacts and of the actuator (Figure 13 - top). By using the 2nd harmonic for imaging, the impacts can be seen (Figure 13 -middle). The impact directly on top of the actuator cannot be detected clearly because of the higher harmonics of the actuator. When the vol 10 issue 3 December 2011
26
plate is excited at 320 kHz, only noise is detected (Figure 13 - bottom). Figure 14 is an example of frequency-pair generation in a fibre reinforced ceramic that had a corner delamination. Excitation was through a piezo-stack and the detection and imaging was carried out with a scanning laser vibrometer to measure both in-plane and out-of-plane components of vibrations. What makes the nonlinear NDT (NNDT) a unique defect-selective instrument for localising and imaging of nonlinear flaws is that a small cracked defect (transparent in a linear ultrasonic NDT) behaves as an active radiation source of new frequency components rather than a passive scatterer in conventional ultrasonic testing. Since the microcontact (nonlinear) defects are only
vol 10 issue 3 December 2011
the fore-runners of further major damage, the NNDT is capable of early recognition of material degradation and “predicting” the oncoming fracture.
4. J.Y. Kim, A.Baltazar, J.W.Hu, S.I.Rokhlin, Hysteretic linear and nonlinear acoustic responses from pressed interfaces , International Journal of Solids and Structures 43 (2006) 6436–6452
FOR FURTHER READING:
5. K. Van Den Abeele, T. Katkowski, N. Wilkie-Chancellier, and W.Desadeleer, Laboratory Experiments using Nonlinear Elastic Wave Spectroscopy (NEWS): a Precursor to Health Monitoring Applications in Aeronautics, Cultural Heritage, and Civil Engineering , Chapter 24, in Universality of Nonclassical Nonlinearity – Applications to NonDestructive Evaluations and Ultrasonics, ed. P.P. Delsanto, (2007) Springer Science, NY (USA)
1. R. Guyer, P. Johnson, Nonlinear mesoscopic elasticity: evidence for a new class of materials, Phys Today 52 (1999), 30–36. 2. Y. Zheng, Roman Gr Maev; I.Y. Nonlinear acoustic Solodov, applications for material characterization: A review, Canadian Journal of Physics 77 (1999), 927967 3. J.H. Cantrell, Fundamentals and Application of Nonlinear Ultrasonic Nondestructive Evaluation , in Ultrasonic Nondestructive Evaluation: Engineering and Biological Material Characterization, ed. Tribikram Kundu, (2004) CRC Press.
6. K. Pfleiderer, I.Y. Solodov and G. Busse, New opportunities in acoustic NDT using frequency conversion by nonlinear defects , in Emerging Technologies in Non-Destructive Testing – Busse et al. (eds) (2008) Taylor & Francis Group, London
Journal of Non Destructive Testing & Evaluation
27
NDE events We hope that this new feature added to the journal since the last two issues has been useful for the readers in planning their activities in terms of paper submissions, registering for seminars, etc. Please send your feedback, comments and suggestions on this section to mandayam.shyamsunder@gmail.com January 2012 Seminar on state-of-the-art and future challenges in Xray computed tomography for materials mechanical behaviour assessment. January 26, 2012 ; University of Southampton, UK. http://www.bssm.org February 2012 NDT Management Association 2012 Conference (NDTMA 2012) February 12 -14, 2012 ; Las Vegas, USA http://www.ndtma.org
21st ASNT Annual Research Symposium and Spring Conference March 19 - 23, 2012 ; Dallas, Texas, USA http://www.asnt.org/events/conferences/sc12/sc12.htm April 2012 18th World Conference on NDT (WCNDT18) April 16 - 20, 2012 ; Durban, South Africa http://www.wcndt2012.org.za Aerospace Workshop and Symposium. April 3 â&#x20AC;&#x201C; 5, 2011 ; Bristol, UK. http://www.bindt.org
March 2012 International Workshop on Magnetic Particle Imaging (IWMPI2012) March 15-16, 2012 ; Lubeck, Germany http://www.iwmpi.org
National NDT Awards
No.
Award Name
Sponsored by
1.
ISNT - EEC National NDT Award (R&D)
M/s. Electronic & Engineering Co., Mumbai
2.
ISNT - Modsonic National NDT Award (Industry)
M/s. Modsonic Instruments Mfg. Co. (P) Ltd., Ahmedabad
3.
ISNT - Sievert National NDT Award (NDT Systems)
M/s. Sievert India Pvt. Ltd., Navi Mumbai
4.
ISNT - IXAR Best Paper Award in JNDE (R & D)
M/s. Industrial X-Ray & Allied Radiographers Mumbai
5.
ISNT - Eastwest
M/s. Eastwest Engineering & Electronics Co., Best Paper Award in JNDE (Industry) Mumbai
6.
ISNT - Pulsecho Best Chapter Award for the Best Chapter of ISNT
M/s. Pulsecho Systems (Bombay) Pvt. Ltd. Mumbai
7.
ISNT - Ferroflux National NDT Award
M/s. Ferroflux Products Pune
(International recognition)
8.
ISNT - TECHNOFOUR National NDT Award for Young NDT Scientist / Engineer
9.
ISNT - Lifetime Achievement Award
M/s. Technofour Pune
Note-1: The above National awards by ISNT are as a part of its efforts to recognise and motivate excellence in NDT professional enterpreneurs. Nomination form for the above awards can be obtained from ISNT head office at Chennai, or from the chapters. The filled application are to be sent to Chairman, Awards Committee, Indian Society for Nondestructive Testing, Module No. 60 & 61, Readymade Garment Complex, SIDCO Ind. Estate, Guindy, Chennai-600 032. Telefax : 044-2250 0412 Email: isntheadoffice@gmail.com
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
Answ ers for P d3 Answers Prrevious issue - NDT C Crrosswor ossword
NDT CROSSWORD- 3
Human hair, image with 5X lens
Plume analysis with gas filters
Stress analysis at high temperature
High speed tire testing
Laser beam profile
Helicopter thermal signature
FLIR Representative:-
Ms. Kaveri Mukherjee, Specialise Instruments Marketing Company E mail - specmco@fastmail.fm * Images herein are for illustrative purpose only. Specifications are subject to change without notice. Terms & Conditions applied.
Qualtech (Quality Management Consultants, NDT Training & Certification) P.N. Udayasankar Quality Management Consultant & Surveyor
Services Offered 1. Vendor Inspection 2. ASNT NDT Level I / II / III Training & Certification
3. API Implementation 4. ASME Implementation 5. ISO - 9001 - 2000 Certification
Plot No. 53, Choudary Nagar Main Road, Valasarawakkam, Chennai-600 087
Ph. : 0091-44-24866810 Mobile : 0091 9444176830 Email : pnuday@gmail.com qualtech747@gmail.com lalita.udayashankar@gmail.com Web : www.qualtech.org
30
vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
31
ndt puzzle We hope you enjoyed solving the “NDT Crossword Puzzle” which was published in the last issue. We received many entries from the readers and based on the maximum number of correct words identified, the following are the WINNERS
Conceptualized & Created by Dr. M.T. Shyamsunder, GE Global Research, Bangalore
CROSS WORD PUZZLE
- P Selvaraj, CNDE, ISRO-SHAR, Shriharikota - Prasanna Kumar NK, ISRO-SHAR, Shriharikota
Congratulations to all the Winners. They will receive their prizes from the Chief Editor of the journal shortly. The correct answers to the Puzzle are published below. In this issue, we have another Crossword puzzle to continue stimulating your brain cells! We hope you will find this section interesting, educative and fun filled. Please send your feedback, comments and suggestions on this section to mandayam.shyamsunder@gmail.com Introduction The “Crossword Puzzle”, contains more than thirty (30) words related to Eddy Current NDE. These include techniques, terminologies, phenomenon, famous people, etc. These words are hidden in the puzzle and may be present horizontally, vertically, diagonally in a forward or reverse manner but always in a straight line. Instructions - All you have to do is identify these words and mark them on the puzzle with a black pen - Preferably you may take a photocopy of the Puzzle sheet and mark your answers on that (see the marked example) - Once completed please scan your answered puzzle sheet as a PDF file and email the scanned sheet to jndte.isnt@gmail.com with your name, organization, contact number and email address Rules & Regulations - Only one submission per person is allowed - The marked answers should be legible and clear without any scratching or overwriting - The decision of the Editor-in-Chief, Journal of NDT &E is final and binding in all matters The correct answers and the names of the prize winners will be published in the next issue Journal of Non Destructive Testing & Evaluation
NAME
: ___________________________
ORGANIZATION : __________________________ PHONE
: ___________________________
EMAIL ID
: ___________________________
vol 10 issue 3 December 2011
32
vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
33
Want to A dvt in our JJournal ournal ? Advt Want to rreach each oovver 6000 members across India? Subscribe to JNDTE Advertisement tariff Classified advt (Business card)
: Rs. 600/-
BackCover (Multi Colour) -
: Rs. 22,000/
Inside Cover - Front (Multi colour)
: Rs. 11,000/-
Inside Cover - Back (Multi Colour)
: Rs. 11,000/-
Inner Pages Full (Multi Colour)
: Rs. 8,000/US $ 300
Inner Pages Full (Black & White)
: Rs. 6,000/US $ 200
Inner Pages Half (Black & White)
: Rs. 3,000/US $ 125
Panel Display
: Rs. 1,200/US $ 50
Journal subscription (India & SAARC countries) includes ISNT membership & subscription to Journal of Non Destructive Testing & Evaluation
Journal Subscription for R & D establishments and educational institutions : Rs. 700/- per annum and US $ 60
Payments should be made by Demand Draft, drawn favoring “JNDE – ISNT”, payable at Chennai. Mr. V. Pari Managing Editor Journal of Non Destructive Testing & Evaluation Modules 60 & 61, Readymade Garment Complex, SIDCO Industrial Estate, Guindy, Chennai 600032 Phone: (044) 2250 0412 Email: isntheadoffice@gmail.com ; jnde.scaanray@gmail.com
Local Conference Secretariat The Conference Company South Africa Tel: 27 31 303 9852 Fax: 27 31 303 9529 Nina Freysen-Pretorius - nina@confco.co.za Deidre Hancke-Haysom - deidre@confco.co.za www.wcndt2012.org.za
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
ONLINE IR/UT SYSTEMS FROM CZECH REPUBLIC A Solution for • Complete Automation of Inspection • High Hi h Speed, S d High Hi h Reliability R li bilit d detection t ti • A “UNIQUE” UT / IR combination to ensure 100% surface defect detections from 0.3 mm to 3.5 mm together with Internal defect detection.
Applications:Ultrasonic Inspection of Rods, Tubes and Billets Ultrasonic C scan inspection of Rods, Tubes, Billets, composite products p of Rods,, Tubes and Billets Infra Red Inspection Ultrasonic Inspection of Weld Seams in Spirally Welded Pipes
Serving NDT for 15+ years One stop shop for “Global” NDT needs Equipments
We Sell We Train We Serve
NDT
Courses
Inspec!on needs
Our Clientele BHEL, ISRO, NTPC, RIL, L&T, TPL, IGCAR, DRDO, SCR, NFC, Midhani, VSSC, KRL and list goes on..
“KALVA HOUSE”, Plot no 17,Road No. 8B, Mamatha Nagar, Nagole, Hyderabad – 500068, India. Ph No: 040-65216871. Tele Fax No: 040 24222476, Official Email: contact@kalva.net Visit us at www.kalva.net
“KALVA HOUSE”, Plot no 17,Road No. 8B, Mamatha Nagar, Nagole, Hyderabad – 500068, India. Ph No: 040-65216871. Tele Fax No: 040 24222476, Email: contact@kalva.net Visit us at www.kalva.net
Our Advanced Product Range Our advanced Range of Equipments Laser Shearography The LTI 5100HD offers the ultimate in shearography NDT performance. This operates in all modes, including continuous real time phase maps, real time subtraction, phase stepping and phase reversal. Benefits Large Area Inspection, Real-Time Imaging of: -De-laminations, - Dis-bonds, Ultra High Through-put Applications:Exclusively for Composites
Digital Radiography DR 1200 system is based on computed radiography flexible imaging plates and high performance NDT workstation technology. This system replaces conventional film radiography with flexible reusable digital imaging plates. Features:Integrated Scanner and eraser. YTB/Se 75/Ir gamma radiation source From 10KV to 10.8Mev 10 8Mev X-Ray X Ray energy operation Professional scientific work station Carbon composites inspection
Acoustic Emission Testing Uniscope system is based on Acoustic Emission T h l Technology f assessing for i the th integrity i t it off products. d t Benefits Detection of crack initiations and propagation Detection and location of corrosion, leak and high stress level zones. positioning g and size measurement. Defect p Real-time monitoring & Condition Monitoring of various processes and critical large dimension structures Applications Pipelines, Pressure Vessels, Tanks, Boilers, Derricks, Cranes, Bridges and other objects
36
vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
37
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
38
vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
39
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
40
vol 10 issue 3 December 2011
Journal of Non Destructive Testing & Evaluation
41
THE STANDARD IN PHASED ARRAY, REDEFINED This second generation OmniScan flaw detector increases testing efficiencies, ensuring superior manual and advanced AUT application performance with faster setups, test cycles, and reporting, in addition to universal compatibility with all phased array modules. The OmniScan MX2 offers a high acquisition rate and new powerful software features—in a portable, modular instrument—to efficiently perform manual and automated inspections.
OMNISCAN MX2 Be truly in touch with phased array • • • • • • •
Bright, large size screen Fast, intuitive touch screen interface Full screen mode Advanced Weld Overlay High-capacity data storage Fast transfer Designed for IP66
Exclusive Distributor in India: Blue Star Limited Tel: +91 444-244-4000• ndtenquiry@bluestarindia.com Visit www.olympus-ims.com/OmniScan-MX2 • info@olympusNDT.com Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
42
Induction Heating Systems for Pre -Heating & Stress Relieving (PWHT)
We also provide Consumables for Heat Treatment Thermocouple Wire (K Type) Nichrome Heating Wire Nickel 212 Wire Ceramic Beads for Heating Pads
Induction Heat Treatment Equipment on Sales / Rental Basis We also provide:
Portable Hardness Tester
Pen Style Hardness Tester
CERBERUS
Phone:
Gala No. 1, Jalaram Ind Estate Ketkipada Road, Dahisar (E), Mumbai - 400068, India
E-mail: sales@cerberus -india.com
vol 10 issue 3 December 2011
Ultrasonic Thickness Gauge
Coating Thickness Gauge
+91 22 4014 2318 +91 99209 56756 [Rohit] +91 97693 20252 [Jemmy]
Website: www.cerberus -india.com
Journal of Non Destructive Testing & Evaluation
43
Journal of Non Destructive Testing & Evaluation
vol 10 issue 3 December 2011
Technical Paper
44
Imaging of Impacted Kevlar Composite Armours using Data Clustering Sutanu Samanta1 and Debasis Datta2 1
Mechanical Engineering Department, North Eastern Regional Institute of Science and Technology (a Deemed University), Nirjuli-791109, Arunachal Pradesh, India 2 Mechanical Engineering Department, Bengal Engineering and Science University, Shibpur, Howrah-711103, West Bengal, India E-mail: suta_sama@yahoo.co.in, debasis_datta @rediffmail.com
ABSTRACT Experimental investigation has been carried out to study the behaviour of Kevlar epoxy and Kevlar polypropylene composite armours subjected to ballistic impact. Immersion type ultrasonic C-scan is performed on impacted armours and extent of damages of different severity are identified through C-scan imaging of A-scan data. The core damage areas for different impact cases are evaluated from the representative C-scan images and their dependency on the impact parameters are noted. In general the core damage area of damage is found to be more in case of Kevlar polypropylene composite armours compared to that found in Kevlar epoxy. However the core damage area is found to increase significantly when the shot remains lodged inside the armour after impact. The present methodology is able to implement the imaging of the impacted area effectively and without any intervention of the skilled personnel. Keywords: Ballistic impact, C-scan, A-scan, Composite armours and Shot lodge.
1.
INTRODUCTION
The last two decades have observed a revolutionary attempt in modifying and implementing new methodology and design concepts resulting in enhanced and successful implementation of laminated FRP composites in almost all areas, thereby gradually substituting traditional monolithic materials. Still there are several inhibiting factors, which have delayed the widespread use of composites in aircrafts, military armaments and spaceships, where the potential for weight reduction is at a premium. A composite may contain defect/damage in several forms, namely, fibre breakage, matrix cracks, fibre debonds, fibre pull out and the delamination cracks. Out of these delamination cracks running at the interface parallel to the plane of the plies, are quite common. Such cracks occur when laminates fail in flexure mostly under impact loads and are responsible for absorbing a significant amount of fracture energy. So there is a growing need in military and civil application for composite materials that not only have good structural characteristics, but also good penetration resistance and greater strength after impact. Many researchers have worked in this area and a considerable amount of literature has been published. Hagemier et al [1] inspected boron, glass and graphite reinforced polymer matrix composites using ultrasonic C-scan. They observed that throughtransmission technique had the advantage of not being affected by surface roughness, surface contour etc. Mool and Stephenson [2] inspected boron/epoxy laminates with known defects using through transmission ultrasonic Cscan with unfocussed probes. In recent years attention has been given to the development of automated C-scan technique. C-scans of tubular
Journal of Non destructive Testing & Evaluation
graphite/epoxy specimens were performed by Rogovsky [3]. In this investigation, flaws were simulated in different layers of the tube and a method was suggested where the number and amplitude of multiple reflections was considered. Jones [4] developed an automatic ultrasonic scanning system which controls the location of probes in a water squirter system. The received signal is digitized to provide a grey level image. A comprehensive review article on ultrasonic NDE of advanced composites was presented by Henneke [5]. The article covers a wide range of works, applied to composite material, which include different modes of wave propagation, material characterization, attenuation measurement, C-scan techniques etc. Preuss and Clark [6] used the time of flight of ultrasonic Cscanning for detecting, sizing and characterization of defects in carbon-fibre composites. Hosur et al [7] presented the results of experimental work on damage of carbon fibre reinforced plastic laminate due to low velocity impact. Both drop weight and low velocity projectile impacts were carried out. The resulting delamination was determined by immersion type ultrasonic C-scans. It is observed that there has been a steady increase in the usage of ultrasonic non-destructive testing for detection of size, shape and orientation of surface and internal discontinuities. In recent years, with the advent of electronics and high speed computers, the methodology has become more powerful and user friendly. Real time digitisation of the analogue signal has added quantitative flavour to the methodology. In the present investigation, a methodology for nearly real time evaluation of impact damage in composite is proposed. The experimental set up and different steps of data grouping methodology are discussed in the following sections. Vol. 10, Issue 3 December 2011
Technical Paper
2. EXPERIMENTAL Kevlar-Polypropylene (Kevlar-PP) and Kevlar-Epoxy composite laminates of different thickness have been used under the study. The reinforcing material in the composite laminates is plain weave Kevlar fabric of size 300mm x 300mm. Two types of resin matrices, (i) epoxy and (ii) polypropylene have been used. The experimental set up and data acquisition is discussed briefly in the following section. 2.1 Ballistic testing Set-up
The composite laminates were mounted on specially designed holders where two sides were clamped for rigid holding and placed in the line of fire of a 7.62mm caliber military rifle. The impact energy was varied by adjusting the propellant mass in the ammunition. The impact and the exit (residual) velocities were measured by the foil and counter method. Each frame consisted of two thin aluminium foil separated by a thin insulating paper board and were connected to a timer. Two frames were separated by a distance of 3m and placed as close to the impacted panel and the velocity was calculated from the time taken by the projectile to travel between two frames at a measured distance apart. A similar set up was used behind the laminate panel to record the exit velocity in case of perforations. 2.2 C-scan and Data Acquisition
C-Scan of composite laminates was done in an immersion type C-scan set up. The set up is furnished with facilities for automated transducer movement and data acquisition. The set up is comprised of (i) an acrylic glass tank furnished at the top with lead screws in mutually perpendicular directions, (ii) stepper motors and controllers to drive the lead screws which, in turn, hold a common nut to accommodate the probe holder assembly, (iii) the PCUS 11 ultrasonic board [8], its compatible software [9], (iv) an interfacing computer, and (v) the normal beam longitudinal wave transducers. The ultrasonic board, installed in the controlling computer, has necessary functions required for ultrasonic measurement, including high voltage pulser, analogue to digital converter, and signal conditioning elements. This board has two connectors for probe connection. For pulse-echo mode of scanning the single transmitter cum receiver probe is connected to the S/E connector and for through transmission mode of scanning the transmitter probe is connected to the S/E connector while the receiver probe is connected to the E connector. In the present work, impacted Kevlar-epoxy and Kevlarpolypropylene composite specimens were scanned in pulseecho mode. The probe used had a central frequency of 1 MHz and a diameter of 12.5 mm and it remained partially immersed in the coupling liquid. The movement of the probe in x and y directions are controlled by two stepper motors and the distance between two successive probe Vol. 10, Issue 3 December 2011
45 positions was kept at 3mm in the present case. The ultrasonic board seamlessly interacts with the compatible software that has the capability to condition, gate and zooming of the digitised waveform. For each point scanned, a gated portion of the A-scan trace of the RF signal covering the backwall echo was digitized at a sampling rate of 80MHz. and stored as an ASCII file in the interfacing computer. Desired ultrasonic features were then extracted from the digitized waveform for each point and they formed a data set. Such a data set, either for a single feature or for a set of features, is subjected to clustering for a systematic classification.
3. CLUSTERING OF DATASET Clustering is an important technique used in discovering the inherent structure present in any given data set. It is widely used in data analysis and pattern recognition. Datta et al [10] performed immersion type ultrasonic C-scans on through transmission mode. The impact was created by repeated dropping of weights on GFRP panels. C-scan image of the specimen was generated by grouping of signal amplitude data set by UPGMA linkage method based on a pre-selected no of cluster. But this is a sensitive task and may require enough pre-idea regarding nature of distribution of object values in the set. Additionally the performances of such algorithms are also influenced by the choice of the distance measure (such as Euclidean, Mahalanobis, Hamming, Jaccard, Chebychev etc.) and are required to be chosen appropriately according to the underlying shapes of the data. In order to improve these drawbacks an algorithm for data clustering was proposed by Wong et al [11]. This method is based on regulating a similarity measure and replacing movable vectors so that the appropriate number of clusters is determined by the best performance index. At initial stage a data point is chosen as a reference vector (ci) and those vector (cj) that have high similarity with the reference vector has to be found. Such similarities are computed with the help of Eq.1. (1)
where, rij represents the similarity between the reference vector ci and the comparison vector cj, and σ is the width of the exponential function. The higher value of rij indicates the closeness between the two vectors. Then reference vector will be replaced with the average of those vectors with high similarity with the reference vector. In this way one time will reach when all the replaced vectors will tend toward their cluster centres. In the iterative process of the algorithm, the width of the similarity measure function can be changed. The value of this width plays an important role in determining how large or small range of data can be grouped in the same cluster. Each different width value may result in different number of clusters and their centres. In the iterative process, the width (σ) is increased by the increment (dσ), calculated on the basis of similarity measure, and for each σ the classification process is Journal of Non destructive Testing & Evaluation
Technical Paper
46 evaluated through calculation of a performance index given by
(2)
Where the value sdk is calculated by
C-scan images of several impacted regions from four composites laminates are presented and discussed. The extents of the core damage zones as obtained from the images are then compared with the respective impact and residual velocities of the projectile. 4.1 Specimens
(3)
Where, hk = the number of clusters which is equal to the number of different vectors in the final results, mdk = the cluster centre of the dth cluster which is one of the different vectors in the final result, ndk =the number of data which belongs to the dth cluster, x= original data. It is obvious that the large value of sdk = suggests a compact and well isolated cluster. That is, the larger is the PI, the more compact and well isolated is the cluster. In this way the optimum numbers of clusters along with the classification results are obtained that correspond to the maximum performance index. In respect of image generation through clustering, this aspect of Wongâ&#x20AC;&#x2122;s algorithm might be useful. In an FRP composite an impact primarily creates delamination, fibre breakage and matrix cracking. The severity of damage is the maximum at the impact zone and the said zone may be called as the core damage zone. Damages of lesser severity are expected to surround the core damage region. In general, 2-4 distinct regions are expected to be present in an impacted composite and this number may not be judged just by looking at the spread of the data. Thus, the clustering procedure with the ability to decide an optimum number of clusters would not only be helpful to identify number of distinct zones in the impacted region, but also be helpful to identify the extent of such zones in a meaningful manner. One or two outlier points may exist in the classification result but they might not affect the general prediction. In fact the said algorithm is found to be suitable in the image generation process. In the present investigation, the clustering algorithm proposed by Wong et al. and outlined in detail in [11] has been coded in the MATLAB environment and used for classification of ultrasonic data comprising of single or multi-attributes. Before using it for the ultrasonic data, the test data as given in [11] are used and the classification results are found to match verifying the correctness of the code. The images obtained thereof are used for finding the core damage region, the area of which is then related to the impact and residual velocity of the bullet.
Journal of Non destructive Testing & Evaluation
4. RESULTS AND DISCUSSIONS
Two composite laminates viz. 20 mm Kevlar-Poly Propylene and 20 mm Kevlar-Epoxy each subjected to ballistic impacts at several regions have been studied. Three impacted regions from each of 20 mm Kevlar-epoxy and Kevlar-polypropylene laminate have been C-scanned. For striking velocities higher than the ballistic limit, the bullets perforate the laminates and come out with a residual velocity. In some cases, however, shot lodging takes place, i.e., the bullet does not perforate the laminate and remains within it. Such phenomena are expected to happen when the striking velocity is close to the ballistic limit or if the bullet gets diverted from its striking direction while perforating the plate. 4.2 Dependence of core damage areas of impacted regions in the 20 mm thick Kevlar epoxy Composite Plate on impact parameters
To check the dependence of core damage area on impact parameters, three impacted regions in the Kevlar epoxy composite plate has been chosen. Out of the three, two are square regions (zone-1, & 3) of sizes 72mm x 72mm and 50mm x 50mm. Rest one is rectangular (zone-2) in shape , size 72mm x 60mm. Zone 2 had to be made rectangular to avoid overlapping with neighboring impact zones. In Figs. 1 to 3, the C-scan images of the above regions generated by Wong algorithm are shown. During generation of each of the images, area of the core damage region was estimated. The correlation between the damage areas and the impact velocities is discussed in Table 1 where variations of the core damage areas for zone-1, zone-2 and zone-3 with the striking velocity, residual velocity and root of the difference between their squares are shown. Table 1: C-scan result of Kevlar Epoxy composite plate based on Peak Amplitude feature by Wang algorithm Scanned zone
Striking Residual Velocity Velocity (Vs) (Vr) (m/s) (m/s)
(Vs-Vr) (Vs2 - Vr2) Core Damage Area by Wang. (m/s) algorithm
Zone-1
427
159.1
267.9 396.25
190.77
Zone-2
514.7
378.5
136.2 348.78
167.23
Zone-3
521.9
308.5
213.4
251.48
421
It is observed that in general, the core damage area increases with the increase in the root of their squared velocity difference which is proportional to the loss in kinetic energy of the bullet. Vol. 10, Issue 3 December 2011
Technical Paper
47
Fig. 1 : Peak amplitude based C-scan image of scanned zone-1 by Wong algorithm
Fig. 4 : Peak amplitude based C-scan images of zone-1 by Wong algorithm
Fig. 2 : Peak amplitude based C-scan image of scanned zone-2 by Wong algorithm
Fig. 5 : Peak amplitude based C-scan images of zone-2 by Wong algorithm
Fig. 3 : Peak amplitude based C-scan image of scanned zone-3 by Wong algorithm
Fig. 6 : Peak amplitude based C-scan images of zone-3 by Wong algorithm
Vol. 10, Issue 3 December 2011
Journal of Non destructive Testing & Evaluation
Technical Paper
48 4.3 Dependence of core damage areas of impacted regions in the 20 mm thick KevlarPolypropylene Composite Plate on impact parameters
In Figs. 4-6, C-scan images of an impact region of size 72mm by 72mm Kevlar-polypropylene armour for feature peak amplitude are shown. The images clearly bring out the damage region in the central portion marked with white shade, and these are consistent with those in the actual specimen. It is clear that the images in Fig. 4 and 5 are more consistent as it clearly reveals the existence of different distinct regions around the location of impact. Fig. 6 represents the C-scan image of Kevlar polypropylene armours where shot lodging takes place. The area of the core damage region for this case is comparatively larger. Details of impact information with the areas of the core damage regions for the case cited above are summarized in Table 2. It shows that the core damage area is dependent on the root of their squared velocity difference. As the difference increases the core damage area also tends to increase. But the magnitude of striking velocity also seems to play an important role in the core damage area. However, the shot lodging case (zone-3) can be taken as a special case where there is an abrupt large increase in the core damage area. This can be attributed to the embedding of a bullet of diameter 7.62mm and length 39mm. This is expected to create a damage that covers a large part of the scanned region of 72mm x 72mm. The visible area of damage on the plate is also comparatively larger. Table 2 : C-scan result of Kevlar Poly propylene composite plate based on Peak Amplitude feature by Wang algorithm Scanned zone
Striking Residual Velocity Velocity (Vs) (Vr) (m/s) (m/s)
(Vs-Vr)
(Vs2 - Vr2)
(m/s)
Core Damage Area by Wang. algorithm
Zone-1
477.8
62.5
415.3
473.7
505.96
Zone-2
594.8
440.2
154.6
400.01
464.5
Zone-3
423.2
0
423.2
423.2
2977.7
From the data presented in Table 1 & 2, it has been observed that the damage size in case of Kevlar Epoxy panel is less compared to that in case of KevlarPolypropylene indicating the localised nature of damage in the Kevlar-Epoxy armour.
5.
of composite armours. An algorithm suitable for C-scan data clustering has been identified. Performance of this algorithm is independent of predetermined number of clusters. C-scan images generated show that the core damage area in the impacted composite armours is dependent on the matrix, striking velocity and the phenomenon of shot lodging. Damage of a more localized nature has been observed in Kevlar-epoxy composite plates as compared to Kevlar-PolyPropylene. The damage area for both the types of laminate is found to depend directly on the loss of kinetic energy of the bullet. However, the striking velocity alone may influence the core damage area. The core damage area increases sharply in cases of shot lodging.
REFERENCES 1. Hagemaier D.J., Mcfaul H.J. and Moon D., 1971, Nondestructive Testing of Graphite Fibre Composite Structures, Materials Evaluaion, Vol. 29, No. 6. pp. 133-140. 2. Mool D and Stephenson R, 1971, Ultrasonic inspection of a boron/epoxy-Aluminium Composite Panel, Materials Evaluation, Vol. 29, No.7, pp. 159-164. 3. Rogovsky A.J., 1985, Ultrasonic and Thermographic Methods for NDE of Composite Tubular Parts, Materials Evaluation, Vol. 43, pp. 547-555. 4. Jones T.S., 1985, Inspection of Composites Using the Automated Ultrasonic Scanning System (AUSS). Materials Evaluation, Vol. 43, pp. 746-753. 5. Henneke E.G.II, 1990, Ultrasonic Non Destructive Evaluation of Advanced Composites, Non Destructive Testing of Fibre Reinforced Plastics Composites, Vol.2, Editor: John SummerScales. pp. 55-159. 6. Preuss T.E. and Clark G., 1988, Use of Time-of-Flight C-scanning for Assessment of Impact Damage in Composites, Composites, Vol. 19, No. 2, pp.145-148. 7. Hosur M.V., Murthy C.R.L., Ramamurthy T.S. and Shet Anita, 1998, Estimation of Impact-Induced Damage in CFRP Laminates Through Ultrasonic Imaging, NDT&E International, Vol. 31, No. 5, pp. 359-374. 8. PCUS11 Ultrasonic P/R Board Manual, 1999, Doc # EBD0031, Fraunhoffer Institute for Non-Destructive Testing, Saarbruecken, Germany. 9. QUT Ultrasonic testing software Manual, Version 4, 1999, Quality Network (QNET) Pvt. Ltd. 10. Datta D., Samanta S. and Maity P., 2004, Automated Imaging of Composite Specimens, Journal of Non Destructive Testing and Evaluation, Vol. 3, pp. 21-27.
CONCLUSIONS
The ultrasonic C-scan technique has been employed for assessment of damage due to ballistic impact on two types
Journal of Non destructive Testing & Evaluation
11. Wong Ching-Chang, Chen Chia-Chong, Su Mu-Chun, 2001, A Novel Algorithm for Data Clustering, Pattern Recognition, Vol. 34, pp. 425-442.
Vol. 10, Issue 3 December 2011
Technical Paper
49
Generation of Guided Waves Using Magnetostrictive Nanostructured Sensing Elements for Pipe Inspection A.K. Panda, P.K. Sharan*, G.V.S. Murthy, R.K. Roy, S. Palit Sagar and A. Mitra CSIR- National Metallurgical Laboratory, Jamshedpur 831007, India *National Institute of Technology, Tiruchirappalli, 620015, India Email : akpanda@nmlindia.org
ABSTRACT The investigation addresses the generation of guided wave in an aluminium pipe using rapidly quenched magnetostrictive ribbons prepared by melt spinning technique. The experimentation involved development of a sensing device which has the capability of excitation of guided waves as well as sensing the back-wall signals using the magnetostrictive sensor elements. The device comprised of a transmitting coil ‘TC’, a receiving coil ‘RC’ and a biasing coil ‘BC’ placed between TC and RC. The sensor materials used in the study were rapidly solidified nanostructured ribbons of Co36Fe36Si3Al1B20Nb4, Fe78Si8B 14, Fe80Si8B12 and Fe40Ni40B20 alloys. The effect of as-spun and annealed ribbons on the sensor output signal was studied. The Fe80Si8B12 ribbons showed highest back-wall signal after annealing at 400oC for 15 minutes. Keyword: Guided wave, magnetostrictive, rapidly solidified ribbon, pipe.
1. INTRODUCTION The ultrasonic guided wave technique is a promising nondestructive method bearing great potential in the inspection of bounded mediums such as pipes, plates and rods [1,2]. It’s advantages include long range inspection capability, structure health monitoring (SHM) of pipes in relatively inaccessible regions such elevated location and in buried state. The source of the guided waves can be of various origins. It can be of piezoelectric crystals, EMAT or Magnetostrictive sensor (MsS) element. The first commercially available guided wave actuators were based on both piezoelectric [3] and MsS [4] transducers. Array of piezoelectric transducer and electromagnetic acoustic transducer (EMAT) were commonly used for the generation of guided waves. Long range ultrasonic testing (LRUT) technique has the capability of inspecting defects such cracks and corrosion from a single sensor position on a structure over long distances using MsS [5]. In many industries pipe corrosion is one of the major problems for plant maintenance that might be due to water logging or environment effect. Long range guided wave (LRGW) technique bears interesting investigations [6-8] on nondestructive detection and classification of pipe integrity using MsS sensors. The MsS technology is finding wide industrial applications in various industries including oil, gas, chemical, petrochemical, aerospace, electric power and civil engineering where the long range cable inspection and monitoring is beneficial for safety and integrity of the structure. It is the process of inducing ultrasonic waves in a solid material with the help of an electrically driven coil in the presence of magnetic field. In the MsS technique for the generation of guided waves, the sensor materials used were mostly crystalline materials. The materials used sofar mostly belonged to CoFe-based Vol. 10, Issue 3 December 2011
crystalline alloys which are not only expensive but also required high bias magnetizing field [9]. In the present investigation, rapidly quenched ribbons of different materials have been used for the generation of guided waves and the signals from the back-walls have been analysed.
2. ANALYSIS USING SOFTWARE AND EXPERIMENTATION 2.1 Estimation of guided wave frequency
An infinite number of the vibration modes are theoretically possible in the structure due to the mode conversion and reflection from the surface of structure. The propagation velocity of the wave modes is dependent on frequency and the wall thickness of the pipes. Dispersion properties of Lamb waves, which refer to longitudinal guided waves in tube, were investigated theoretically using a generalpurpose software package called DISPERSE version 2 [10]. Exciting Frequency for generation of single wave mode was obtained from dispersion curve (fig-1). For Al tube, density equal to 2.696 g/cm3 was fed into the software. The input parameters for cylindrical structure are; pipe inner radius = 15.5 mm, wall thickness = 8.5 mm. If the ultrasonic pulse is considered of many waves with different frequencies, the guided waves will propagate in different velocities. The vibration of particles is the composite of actions of many frequencies. At least, their phase velocities are the velocity of the wave-before of same frequency, and their group velocities are the ones of wave packets consisted of different frequencies. There are three kinds of modes in the pipes: longitudinal modes, torsional modes and flexural modes, where the longitudinal Journal of Non destructive Testing & Evaluation
Technical Paper
50
Fig. 1 : Dispersion curve showing phase velocity for the tube of OD 48 mm and wall thickness 8.5 mm of length 3.67 m Al tube.
modes are subdivided into L(0,1), L(0,2),..; the torsional ones into T(0,1), T(0,2),…; and flexural modes into F(1,1), F(1,2)… the first two wave modes are axisymmetry and last one is the non-axisymmetry. When frequency is increased to a certain value, a new mode will appear. From the dispersion curves, there are several modes at one frequency. For the inspections, we usually do not want to generate all modes simultaneously. Therefore, mode control needs to be implemented. With the same frequency value, phase velocities are different for different modes. Therefore, only the mode with the “right” phase velocity is generated. The modes and frequencies can be optimized from the dispersion curve, which can predict the propagation characteristics for guided acoustic waves in the structure.
close to each other as shown in (Fig. 2b). Subsequently a compact coil system bounding the magnetostrictive strips was made using SWG 22 with 40 windings for TC, 34 SWG with 200 windings for DC bias and 34 SWG 180 windings for RC (Fig 2a). The windings of receiver were made in opposite direction of transmitter coil. The counterwound coil design provides a spatial filter that maximizes the voltage output when the wave of appropriate frequency is within the aperture of the sensor i.e. when the strain wave generated from the transmitter passes through the periodically spaced coils of the receiver. The magnetic field strengths along the circumference of the electromagnetic coils were measured that generate field equivalent to 1.411 Oe/A for cylindrical ac excitation coil and generate field equivalent to 97.41 Oe/A for DC biasing coil. The magnetic field was measured using a gauss meter. Experiments were carried out with nanostructured alloys Co 36 Fe 36 Si 4 Al 1 B 20 Nb 4, Fe 78 Si 8 B 14, Fe 80 Si 8 B 12 and Fe40Ni40B20 ribbon obtained by melt spinning technique in the form of stripes of thickness and width of 30-60 µm and 12mm, respectively. Hanning signal with three tone burst waveform was used to excite Lamb waves in the tube during inspection. An APLAB 7645 dual DC Power supply was used to provide basing current to the DC bias coil. This DC bias coil gives the desired external magnetizing field to enhance the magnetic anisotropy and thus the magnetostrictive effect in the test structure. The output voltage across the receiver coil was measured by Yokogawa 4456 digital oscilloscope.
3. RESULTS AND DISCUSSION
2.2 Experimental set-up
3.1 Optimization of DC bias magnetising field
Coils were made by wounding enamelled Cu wire on PVC pipes. The sensing device consisting of a transmitter coil (TC) and receiving coil (RC) was placed close to the end of the test pipe as shown in Fig. 2a. A DC biasing coil (BC) was placed between transmitter (TC) and receiver coil (RC). Initially the individual separate coils were placed
A DC bias and ac excitation magnetization along the direction of the axis of the coil was required for a generation of the longitudinal wave mode in the one direction. A longitudinal magnetization on a magnetostrictive stripe can be achieved by applying the biasing to DC bias coil along the longitudinal direction of the tube. A high current along the circumferential direction can generate an axial magnetization. At a constant ac excitation magnetizing field, the DC bias current was varied and noted. The 1st Table 1 : Maximum signal amplitude of the as-spun and annealed ribbons corresponding to the DC bias field Composition of
As-Spun Ribbons
Ribbons Annealed at 400oC
magnetostrictive Ribbons DC bias
Maximum DC bias
Field (Oe) amplitude
Field
of Signal
Fig. 2 : Experimental arrangement for the guided wave generation in an aluminium pipe using rapidly quenched magnetostrictive ribbons (inset Fig. 5b shows separate coils of sensors Journal of Non destructive Testing & Evaluation
Maximum amplitude of Signal
(mV)
(Oe)
(mV)
Co36Fe36Si3Al1B20Nb4
34.12
16
16.58
36
Fe80Si8B12
23.19
24
12.16
94
Fe78Si8B14
40.74
50
18.78
34
Fe40Ni40B20
23.19
78
42.96
12
Vol. 10, Issue 3 December 2011
Technical Paper
51
reflected back wall amplitude signals were obtained for a range of varying DC magnetizing field corresponding to the DC bias current. However, care should be taken as to limit the range of DC current in order to avoid overheating and damage to the bias coil. The maximum signal amplitude of the as-spun ribbons corresponding to the DC bias field is shown in Table 1. Fe40Ni40B20 alloy had maximum reflected amplitude value of 78 mV at 23.19 Oe DC bias magnetizing fields. However, after annealing at 400oC for 15 minutes, the Fe80Si8B12 ribbon showed maximum signal amplitude value of 94 mV at a DC bias field of 12.16 Oe. 3.2 Optimization of AC magnetising field
Keeping the optimized DC bias magnetizing field obtained above fixed for the as-spun ribbon and 400oC/15min annealed ribbons; the effect of time varying magnetizing field (AC excitation field) on signal amplitude was obtained and shown in Table 2. Table 2 : Maximum signal amplitude of the as-spun and annealed ribbons corresponding to the AC magnetizing field Composition of
As-Spun Ribbons
Ribbons Annealed at 400oC
magnetostrictive Ribbons DC bias
Maximum DC bias
Field (Oe) amplitude
Field
of Signal
Fig. 4 shows experimental waveform taken from oscilloscope with the annealed magnetostrictive alloys at 400째C. After annealing at 400oC it was found that the maximum amplitude value for Fe80Si8B12 alloy was 120 mV at 4.77 Oe ac excitation magnetizing fields. On the other hand, the maximum amplitude value 18 mV for Fe 40 Ni 40 B 20 alloy is low at 4.77 Oe ac excitation magnetizing fields as compared to other magnetostrictive materials. Thus, the reflected amplitude signal was maximum for annealed Fe80Si8B12 ribbon instead of the Fe40Ni40B20 alloy in DC bias field and AC excitation field. It is known that in the as-spun state the low attenuation in sensor output voltage of FeNiB compared to FeSiB system is due to its high stress sensitivity while after annealing the FeSiB ribbon had enhanced magnetostriction showing high sensor output. The waveform recorded at 76 kHz shown in Fig. 4 contains the dominant longitudinal wave L (0,2) mode arrival after reflection from the other
Maximum amplitude of Signal
(mV)
(Oe)
(mV)
Co36Fe36Si3Al1B20Nb4
5.53
22
3.43
44
Fe80Si8B12
5.53
32
4.77
120
Fe78Si8B14
4.77
84
4.77
46
Fe40Ni40B20
4.77
112
4.77
18
In order to investigate the effect of an alternating current, the 1st reflected back wall amplitude was considered. The maximum sensor output amplitude value for the as-spun Fe40Ni40B20 ribbon was 112 mV at 4.77 Oe ac magnetizing field. Just as in the case of DC bias field (Table 1), the Fe80Si8B12 ribbons showed maximum signal output of 120mV at 4.44 Oe. Thus, it was observed that melt spun Fe80Si8B12 ribbons revealed higher signal amplitudes after annealing treatment compared to the other three magnetostrictive ribbons. After optimising the DC bias field and the AC magnetising field, guided wave signal from the aluminium pipe was observed using an oscilloscope as shown in Fig. 3. Experimental data was taken from oscilloscope with the as-spun and 400oC/15min annealed magnetostrictive ribbons. The first echo in the Fig. 3 is the initial tone burst pulse applied to the transmitting coil which was electrically linked to receiving coil and the oscilloscope. The second and third signals are the end reflected echoes. The first end reflected signal is the one signal during the return trip of the elastic wave after reflection from the other end of the tube. The second end Vol. 10, Issue 3 December 2011
reflected signal is one detected when the returned wave made another trip after reflection from the sensor end of the tube. The two ends reflected signals, therefore, are separated by the round trip time from the receiving MsS end to the other end of the tube. The span between the 1st and the 2nd back wall was approximately 1.5013 ms at an experimental wave velocity approximately 4835.16 ms-1.
Fig. 3 : Signals taken after the optimization of the time-varying and DC bias magnetizing field of the different composition of as-spun magnetostrictive ribbons
Fig. 4 : Signals taken for using melt spun ribbon annealed at 4000C for 15 minutes. Journal of Non destructive Testing & Evaluation
Technical Paper
52 end of the Al tube. An additional benefit of exciting at lower frequencies is that the L (0,2) mode is primarily excited whereas at higher frequencies both the axisymmetric modes and non-axisymmetric modes are excited. Based on these results, an excitation frequency of 76 kHz was used for the Al tube configuration, and the first L (0,2) mode arrival was used in the tube inspection.
4. CONCLUSION Guided waves could be generated in an aluminium pipe 3.67 m long, 8.5 mm wall thickness and 48 mm outer diameter using melt spun magnetostrictive ribbons of Co 36 Fe 36 Si 3 Al 1 B 20 Nb 4, Fe 78 Si 8 B 14, Fe 80 Si 8 B 12 and Fe40Ni40B20 alloys. These ribbons are in nanostructured state as they are rapidly quenched. A magnetostrictive sensing device was designed for the generation of the guided wave in the pipe. Using disperse software a suitable excitation frequency of 76 kHz for longitudinal L (0,2) wave mode was determined and used for the pipe inspection. DC bias magnetizing and AC excitation field were optimized and found to be 3-5 Oe and 10 - 15 Oe respectively. The Fe80Si8B12 ribbon annealed at 400oC/ 15minutes gave high amplitude signal suitable for generation and sensing the back wall reflections. The technique can be used for the detection of defect in the pipe. Moreover, very low bias magnetizing field ~ Oe was used for the melt spun ribbons compared to magnetizing field ~ kOe reported for crystalline alloys.
ACKNOWLEDGEMENT
REFERENCES 1. ASNT, Non-destructive Testing Handbook, ultrasonic testing, volume, 7 2. Rose, J. L., Ultrasonic Waves in Solid Media, Cambridge University Press, New York (1999). 3. Alleyne, D. N, and Cawley, P., 1996, ‘The Excitation of Lab Waves in Pipes Using Dry- Coupled Piezoelectric Transducers,’ J. Nondestructive Evaluation 15, pp. 11–20 4. N.S.Tzannes, “Joule and Widemann effects – the simultaneous generation of longitudinal and torsional stress pulses in magnetostrictive materials”, IEEE transactions on Sonics and Ultrasonics, Vol. SU-13, NO. 2, July 1966. 5. H. Kwun and C.M. Teller, “Detection of Fractured Wires In Steel Cables Using Magnetostrictive Sensors,” Materials Evaluation, Vol. 52, 1994, pp. 503-507. 6. J. B. Nestleroth, “Pipeline In-line Inspection – Challenges to NDT”, Proceedings, ECNDT 2006, 9th European Conference on NDT, Berlin, Germany, September 25-29, 2006. 7. C. I. Park, S. H. Cho, and Y. Y. Kim, “Z-shaped magnetostrictive patch for efficient transduction of a torsional wave mode in a cylindrical waveguide,” Appl. Phys. Lett., vol. 89,pp. art. no. 174103, Oct. 2006. 8. E. Kannan, B. W. Maxfield, and K. Balasubramaniam, “SHM of pipes using torsional waves generated by in situ magnetostrictive tapes,” Smart Mater. Struct., vol. 16, pp. 2505-2515, Dec. 2007. 9. Y.M.Cheong etal, [12th Asia-Pacific conference on NDT, 2006 10. http://www.me.ic.ac.uk/dynamics/ndt, DISPERSION: DISPERSE version 2.0.
The authors express their sincere gratitude to Director, NML for permitting to carry out the investigation and also permitting to publish the results.
Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
53
Detection of honeycomb defects in reinforced concrete structures using acoustic pulse-echo methods and wavelet transforms Krishna Prasad M1, Herbert Wiggenhauser2, Krishnan Balasubramaniam1 1
Centre for Nondestructive Evaluation, Department of Mechanical Engineering,Indian Institute of Technology, Madras, Chennai, India, 600036. 2 Federal Institute for Materials Research and Testing (BAM), IV.4, Berlin, Germany, 12205.
ABSTRACT Honeycombs/compaction faults occur in the concrete structures due to improper solidification of the concrete, which may reduce the strength of the concrete and also act as a passage for the water/ acids that further corrodes the reinforcements. This paper explores about the acoustic pulse-echo techniques for the detection of honeycomb defects in a laboratory specimen located at the Federal Institute for Materials Research and Testing (BAM), Berlin. Since concrete is an inhomogeneous medium, the defect signals are masked by the material noise due to large amount of scattering/ reflections of acoustic waves. A filtering method using the discrete wavelet transforms is applied on the ultrasonic time signals for the better localization of defects. Keywords: Honeycomb defects, Acoustic pulse-echo methods, Wavelet transforms.
1. INTRODUCTION The non-destructive testing of concrete using acoustic pulse-echo methods operates in the low frequency range of around 2-200 kHz, and uses primarily two methods. (1) Impact-echo method and (2) Ultrasonic pulse-echo method. Impact-echo method involves introducing low frequency stress waves into the test object using a low energy mechanical impact on the surface and monitoring the surface displacements caused by the arrival of reflections/ scattering of the waves from internal defects and external boundaries. These reflected waves (Primary waves) produces surface displacements, which are recorded by a displacement transducer. Here, the analysis of time domain
signal is time consuming and so the analysis is carried out in its frequency domain by calculating the Discrete Fourier Transform of the signal. Since the reflected primary waves are periodic, the depth of a reflecting surface, (d) can be calculated by using the relation d = V/2f [1] where V is the primary wave velocity and f the frequency. The Fig 1.1 shows a schematic illustration of the impact-echo method [2]. Ultrasonic nondestructive testing of concrete structures involves the usage of low frequency ultrasonic waves in the order of 20-100 kHz. Ultrasonic waves are generated using the principle of inverse piezoelectric effect, which travels into the test object and get reflected when encounters a material with different acoustic impedance. The same transducer receives these reflected waves and stores as a time vs. amplitude signal. The ultrasonic signals are analyzed in the time domain and the depth of the reflecting surface can be determined by using the simple relation d = Vt/2, where t is the travel time.
2. BACKGROUND
Fig.1.1 Schematic illustration of impact-echo Vol. 10, Issue 3 December 2011
Wiggenhauser et al [2] automated the Impact-echo and presented the Impact-echo data as 2-D and 3-D Impact echograms. Krause et al [3, 4] carried out non-destructive testing of concrete using Ultrasonic pulse-echo methods. They used a variety of transducers such as single point transducers and array of broadband transducers to detect inclusions, defects in tendon ducts and reinforcements. Jansohn and Scherzer [10] used an ultrasonic shear wavearray transducer for detecting the improperly filled tendon ducts. They used a normal shear wave transducer with point contacts without the need for additional coupling Journal of Non destructive Testing & Evaluation
Technical Paper
54 agent. They also reported that ultrasonic devices utilizing shear waves offer advantages with respect to backscattering in the direction of receiver compared to longitudinal waves. Reinhardt et al [6] applied the Wavelet Transforms for the Acoustic Emission signals emitted from the defects in the concrete structures. They decomposed the signals into different frequency bands using Discrete Wavelet transforms and carried out denoising operations to remove the noise. They also used the Scalegrams to represent the energy over each scale. Grosse et al [9] carried out filtering operations using both conventional Fourier transform based filter techniques and the discrete wavelet transforms and proved wavelet transforms to be a suitable tool for denoising the acoustic emission signals because the conventional filtering has some artifacts. Park et al [7] applied the wavelet transforms for non-destructive evaluation of material degradation of Cr-Mo steels using Ultrasonics. They used Continuous wavelet transforms to evaluate the attenuation factor with the degradation of material and the discrete wavelet transforms for the noise suppression of the signal.
3. WAVELET TRANSFORMS Wavelets are the mathematical functions that decompose the signal into different frequency bands and then study each component with a resolution matched to its scale [8]. This leads to a new concept known as Multi Resolution Analysis (MRA) apart from the fixed resolution of Short Time Fourier Transforms (STFT). Wavelet transforms give good time resolution and poor frequency resolution at high frequencies and vice versa. The Continuous Wavelet Transforms (CWT) are computed by correlating the signal with scaled (dilated) and shifted (translated) versions of the mother (analysing) wavelet. Mathematically it can be expressed as [5]
the time-scale information simultaneously and cannot be exactly used for the localisation of defects. Discrete Wavelet Transforms (DWT) is based on the sub band coding or the pyramidal algorithm. It basically uses a scaling function nothing but a half band low pass filter and a wavelet function nothing but a half band high pass filter. The signal is passed through a series of half band low pass and half band high pass filters respectively. The half band low pass filter removes all the frequencies that are above half of the highest frequency in the signal and the half band high pass filter removes all the frequencies that are below half of the highest frequency. The filtering of a signal corresponds to the convolution of the signal with the impulse response of the filter. The convolution operation in the discrete time case is defined as in the equation 3.3.
(3.3)
Then the signals are downsampled by two to decrease the ambiguity with out losing any information. The coefficients obtained on the high frequency side are known as details and the coefficients obtained on the low frequency side are known as approximations. The approximations are again passed through the complementary filters leaving the details aside. The process of filtering and downsampling is referred to as single level decomposition and the decompositions can be carried out till a single sample is reached. A one level decomposition can be mathematically expressed as [5] (3.4)
(3.5) (3.1)
Where x(t) is the signal itself, t and s are the translation and scale (inverse of frequency) parameters respectively and y(t) is the mother wavelet. The wavelets must satisfy the admissibility and regularity conditions which shows that the wavelets are small waves with finite duration and oscillatory. The information obtained from CWT is redundant and so the time-scale plane is discretized in a dyadic manner. This discretized version is known as Wavelet series and is mathematically represented as [5] (3.2)
The value of s 0 is found to be 2 and the value of
Ď&#x201E;0
to be 1 for the better reconstruction of the signal. This continuous wavelet transforms and the wavelet series gives Journal of Non destructive Testing & Evaluation
Thus a single level decomposition halves the time resolution since only half the number of samples now characterizes the signal. However, the frequency resolution is doubled since the frequency band of the signal now spans only half of the previous frequency band. The signal reconstruction (synthesis) is done by following the same procedure as used for decomposition but in the reverse order. That is the signals at each level are upsampled by two and then passed through synthesis filters gâ&#x20AC;&#x2122;(k) and hâ&#x20AC;&#x2122;(k). The important point here is that the analysis and the synthesis filters are identical to each other and they are known as Quadrature mirror filters (QMF). However, if the filters are not ideal half band, then perfect reconstruction cannot be achieved. Daubechies invented orthonormal, compactly supported family of wavelets where perfect reconstruction is possible. These wavelets are used in the current work. The Daubechies wavelet, db7 is used for the present analysis. Vol. 10, Issue 3 December 2011
Technical Paper
The value 7 represents the number of vanishing moments and the filter length used was 14.
4. RESULTS AND DISCUSSION The layout of the Large Concrete Specimen (LCS) consisting several simulated defects including two regions of honeycombs is shown in the Fig 4.1 which was obtained from BAM, Berlin. The LCS has an area of 10 x 4 square meters and a thickness of 300 mm, which is designed to incorporate most of the common defects that occur in concrete structures. Three honeycombs namely K1, K2 and K3 are located in the right top corner of the specimen and are divided into two test areas of 1000 mm x 500 mm and 500 mm x 500 mm as shown in the Fig 4.2. The honeycombs are of 150 mm diameter and 150 mm length. K1 is laid horizontally, K2 is laid vertically and K3 is laid in inclined direction. K1 and K2 are having a concrete cover of around 110 mm and K3 has 112 mm and 156 mm at the two corners respectively. The impact-echo tests are carried out on the specimen using an Olson instruments impactor unit having an impactor equivalent to a diameter of 3 mm. Unfortunately there is no indication of honeycombs except the back wall echo. Even the shadow of the defects at the back wall is not observed. The B and C-scans of the area where K1 and K2 are lying is shown in the Fig 4.3. The reflection from the back wall can be seen in the B-scan at a frequency of about 7.35 kHz, which gives the thickness as around 280 mm for a velocity of 4100 m/sec.
55 Then the Ultrasonic pulse-echo testing was carried out with a commercially available shear wave â&#x20AC;&#x201C;array transducer as described in the section 2. The centre frequency used is 55 kHz. The measurement grid is taken as 20 mm x 20 mm for both the test areas. The B and C-scans obtained from the analysis of ultrasonic time signals using BAMPEâ software at the defect depth are shown in the Figs 4.4 and 4.6. The Fig 4.5 is a reference figure to analyze the B and C-scans. It is a vertically flipped version of the original test area and is done to have a better visualization. The shear velocity in the concrete structure was taken as 2630 m/sec. The C-scan in the Fig 4.4 shows the K1 clearly at a depth of around 80 mm and the direct reflection is also observed in the corresponding B-scan. The lateral position of the K1 also matches with the reference Fig 4.5. Similarly the Fig 4.6 shows the C-scan at a depth of 105 mm. The C-scan does not show any indication of K2 and there is a rough indication of K2 in the B-scan, which is almost nothing to analyze. The Fig 4.7 shows the Cscan at the back wall whose thickness is obtained as 270 mm. The shadow of the honeycombs and plastic pipes at the back wall can also be observed in the C-scan. The B and C-scans obtained from the same ultrasonic time signals using Synthetic Aperture Focusing Technique (SAFT) are shown in the Figs 4.8, 4.9 and 4.10. These SAFT results are presented for comparison with the normal ultrasonic time signals and as shown in the Figs 4.8-4.10, these did not show any additional information regarding K2 either in the C-scan or as a direct reflection in the Bscan.
Fig.4.1 Large Concrete Specimen (LCS) showing various types of defects. Vol. 10, Issue 3 December 2011
Journal of Non destructive Testing & Evaluation
56
Technical Paper
Fig.4.2 Test areas showing K1, K2 and K3.
Fig.4.3 C and B-scans of K1 and K2 from Impact-echo.
Fig.4.4 C-scan at a depth of 80mm and its corresponding B-scan
Fig.4.5 Reference figure for the analysis of B and C-scans for K1 and K2
Fig. 4.6 C-scan at a depth of 105mm and its corresponding B-scan Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
57 There is no indication of honeycomb K3 from the second test area either in the B-scan or in the C-scan and so the results are not presented. Similar results are obtained from the SAFT also. It is expected that since the honeycomb K3 is inclined in its orientation, the transducer is unable to receive the reflected signal and so no information has been obtained about the honeycomb defect. The discrete wavelet transforms are applied on the ultrasonic time signals as described in the section 3. An ultrasonic time signal obtained from the tests conducted using shear wave-array transducer is shown in the Fig 4.11. The sample size was 1000 and the sampling frequency was 1 MHz. Thus from the Nyquist’s theorem, the folding frequency is 0.5 MHz (500 kHz). Then a seven level decomposition is performed using discrete wavelet transform with db7 wavelet.
Fig.4.7 C-scan at the backwall (depth of 270mm) and its corresponding B-scan
Fig.4.8 SAFT C-scan from between 67-119 mm
It is observed from the reconstructions as shown in the Fig 4.12 that the level 1(250-500 kHz) and level 2 (125250 kHz) reconstructions pertain to high frequencies, which consists of high frequency glitches or noise. So these levels can be successfully eliminated. The level 5 (15.6231.25 kHz) and level 6 (7.81-15.62 kHz) reconstructions pertain to very low frequencies and the analysis of these frequency bands is also not appropriate, as these do not contain any useful information. The level 3 (62.50-125 kHz) and level 4 (31.25-62.50 kHz) coefficients represent mostly the signal characteristics and the analysis of these levels can be done individually as usually like ultrasonic time signals. This is known as wavelet compression in terms of wavelet terminology or simply known as wavelet filtering. The level 4-frequency band mostly matches with the centre frequency of the ultrasonic transducer used whose value is 55 kHz. The whole work is carried out in MATLABâ and the reconstructed signals are transferred to BAMPEâ for analysis. The results obtained in the form of B and Cscans are shown in the Fig 4.13. The Fig 4.13 represents the B and C-scans at a depth of 105 mm. This shows the honeycomb K2 clearly in the Cscan, which was not identified from the analysis of normal ultrasonic time signals or from the SAFT results. The direct reflection from the honeycomb K2 also can be identified very clearly.
Fig.4.9 SAFT B-scan from over the whole test area
Fig.4.10 SAFT C-scan at the backwall (at a depth of 270 mm). Vol. 10, Issue 3 December 2011
The intensity variation along honeycomb K2 is plotted as shown in Fig 4.14 before and after performing wavelet transforms This reveals a clear increase in the intensity where the honeycomb K2 is located and also shows the suppression of noise after performing wavelet filtering when compared with the original signals. Then the mean intensity ratios of the defective region (considering K2 only) to non-defective region are calculated both before and after performing wavelet transforms. This resulted in a 22.20% increase in the intensity ratio after performing wavelet transforms. Journal of Non destructive Testing & Evaluation
Technical Paper
58
Fig.4.11Ultrasonic time signal
Fig.4.12 (a) Level 1 reconstruction.
Fig.4.12 (b) Level 2 reconstruction.
Fig.4.12 (c) Level 3 reconstruction.
Fig.4.12 (d) Level 4 reconstruction.
Fig.4.12 (f) Level 6 reconstruction.
Fig.4.12 (e) Level 5 reconstruction.
Fig.4.12 Reconstruction of wavelet coefficients Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
59 §
Discrete wavelet transforms based on the principles of sub band coding were found to be suitable for the analysis of ultrasonic time signals.
§
Selecting a fourth level frequency band and the further analysis of it revealed the honeycomb defect K2, which did not turn up from the analysis of normal ultrasonic time signals or from the SAFT results.
§
The mean intensity ratio of the defective to nondefective region showed a 22.20% increase after performing wavelet transforms on the ultrasonic time signals.
6. ACKNOWLEDGEMENTS This research work is carried out at the Federal Institute for Materials Research and Testing (BAM), Berlin, Germany in collaboration with IITM, Chennai, India through German Student Exchange Service (DAAD). I am also thankful to IZFP group at Saarbrucken, Germany for their cooperation in providing the SAFT images. Fig.4.13 C-scan at a depth of 105 mm and it’s corresponding B-scan obtained from fourth level wavelet reconstruction.
REFERENCES 1. Sansalone, M., and W. B Streett. Impact-echo: Non-destructive Testing of Concrete and Masonry, Bullbrier Press, Jersey shore, PA, 1997. 2. C. Colla, G. Schneider, J. Wöstmann, H. Wiggenhauser. (1999) Automated Impact-echo: 2-and 3-D Imaging of Concrete Elements, DGZfP Fachtagung Bauwerksdiagnose, 4(5), 307318. 3. Martin Krause, Frank Mielentz, Boris Milmann, Doreen Streicher, Wolfgang Müller. (2003) Ultrasonic Imaging of Concrete Elements: State of the art using 2D synthetic aperture, Proceedings of the International Symposium (NDT-CE 2003), Berlin, Germany, In Print.
Fig.4.14. Intensity plot along ‘K2’ before and after applying wavelet transforms
5. CONCLUSIONS Non-destructive testing of reinforced concrete structures is quite essential to avoid catastrophic failures and the honeycomb defects play an important role, since these reduces the solid strength of the structure thus leading to failure and also acts as a passage for moisture which further corrodes the reinforcements. An attempt is made through this research work for the effective detection of honeycombs in the reinforced concrete structures. The following conclusions can be drawn based on this current research work: §
Effective detection of honeycombs is obtained through ultrasonic shear wave -array transduction where the impact-echo method did not work out.
§
The wavelet transforms, a recent advancement in signal processing is successfully applied in the field of nondestructive testing of concrete.
Vol. 10, Issue 3 December 2011
4. M. Krause, F. Mielentz, B. Milmann, W. Müller, V. Schmitz, H. Wiggenhauser. (2001) Ultrasonic imaging of concrete members using an array system, NDT&E international, 34, 403-408. 5. Robi Polikar. The Wavelet tutorial, Rowan University, http:// users.rowan.edu/~polikar/WAVELETS/WTtutorial.html, 1996. 6. Jochen H. Kurz, Hans-Jürgen Ruck, Florian Flinck, Christian U. Grosse, Hans-Wolf Reinhardt. (2003) Wavelet algorithms for non-destructive testing, Proceedings of the International Symposium (NDT-CE 2003), Berlin, Germany, In Print. 7. Ik Keun Park, Un Su Park, Hyung Keun Ahn. (2000) Experimental Wavelet analysis and Applications to Ultrasonic Non-destructive Evaluation, 15th WCNDT, Roma. 8. Amara Graps. (1995) An Introduction to Wavelets, Institute of Electrical and Electronics Engineers, Inc, 2(2). 9. Christian U. Grosse, Hans W. Reinhardt, Markus Motz, Bernd H. Kröplin. (2002) Signal conditioning in acoustic emission analysis using Wavelets, The e-journal of non-destructive testing, 7(09). 10. Reinhard Jansohn and Jan Scherzer. (2003) Improper filled ducts detected by ultrasound reflection, The e-journal of nondestructive testing, 8(4). Journal of Non destructive Testing & Evaluation
Technical Paper
60
Indirect methodologies for inversion of eddy current NDE data S. Shuaib Ahmed, B.P.C. Rao, S. Thirunavukkarasu and T. Jayakumar Nondestructive Evaluation Division, Metallurgy and Materials Group Indira Gandhi Centre for Atomic Research, Kalpakkam, Tamil Nadu- 603 102 E-mail: bpcrao@igcar.gov.in
ABSTRACT This paper presents two inversion methodologies for inversion of eddy current data. The first methodology uses competitive learning neural network with cosine similarity algorithm to identify the layer to which an EC image of a defect belongs in a two-layer structure. The second methodology aims evaluation of depth location and height of subsurface defects using multidimensional learning based multilabel radial basis function neural network. The performance and robustness of these two methodologies have been validated using entirely different type of dataset used for training. The eddy current data required for training and validation have been generated using CIVA numerical modeling software. Both these methodologies have shown promise for evolving artificial intelligence based automated identification and quantification of defects. Keywords: Inversion, eddy currents, quantification, classification, signal processing
1. INTRODUCTION One of the important aspects during structural integrity assessment of engineering components using NDE techniques is quantitative characterization of defects, especially location and size. Among various NDE techniques, eddy current technique is widely used for detection and sizing of defects in components made of metallic materials. While high-speed couplant-free testing of components for reliable detection of defects is a clear advantage, the technique suffers difficulties for defect detection and sizing, due to the combined influence of several variables at a given location. Diffusion behavior and non-uniform decay of eddy currents in different directions result in formation of identical EC signals or images for defects of same size but located at different depths as well as formation of identical signals or images from different size defects present at same location. These two are expected to influence reliable quantitative characterization of defects and demand the use of multifrequency EC data and inversion approaches. Extracting quantitative information of defects from measured NDE data i.e. amplitude, signal or image is called inversion. Two popular approaches for inversion of EC data are i) model based direct inversion and ii) empirical indirect approach. In the first approach, scatterer size and location are determined following the physics based interactions and
associated mathematical models [1-2]. By this approach, only limited success has been reported, essentially due to complex geometry of defect as well as the component. In the second approach, empirical relationship between measured data (signals or images from a variety of expected defects) and defect dimensions is established as shown in Figure 1. The heart of an indirect approach is the inversion algorithm which simulates a mapping between the extracted features and the defect dimensions and location. This approach is gaining popularity. In this approach, the eddy current signals and images are given as the input to an inversion module which consists of a feature extraction module and an inversion algorithm that learnt the relationship between input and output through supervised or unsupervised learning or training. Signals and images are higher dimensional data. It is expensive, in terms, of processing speed and memory requirement to process and handle higher dimensional data. Hence, instead of providing raw data to the inversion algorithms, it is a good practice to extract important nonredundant features which describe the signals/images crisply. Fourier descriptors, Principal Component Analysis (PCA), features from impedance plane such as phase, amplitude are some of the features useful for inversion of EC data [3]. For indirect inversion, use of expert systems and artificial intelligence (AI) methods such as feed forward artificial neural networks, radial basis function neural
Fig. 1 : Indirect inversion of eddy current data. Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
61
networks, case based reasoning and support vector machines etc. have been reported [4-7].
2.2 Competitive learning neural network with cosine similarity
In this paper, two indirect methodologies have been proposed for inversion of eddy current data. While one methodology uses EC images as input, the other uses EC signals as input. This paper gives a brief description of the two methodologies and discusses the results of inversion.
Competitive learning is an on-line method of unsupervised learning. It is an efficient algorithm in terms of speed and accuracy where cosine similarity is used as a distance metric [10]. The goal of unsupervised learning is to separate the observed data into clusters or to provide understanding about the underlying structure or pattern of the data. The outcome of the input data is not provided to the learning algorithm. The important characteristic of any unsupervised learning algorithm is the similarity measure used to discriminate the clusters.
2. INDIRECT INVERSION METHODOLOGIES The paper presents two inversion methodologies for two diverse problems. The first methodology is based on unsupervised learning using competitive learning with cosine similarity, which uses EC images as input to identify the layer to which a defect belongs to in a two-layered structure. The second methodology is based on multidimensional learning which uses EC signal as input to determine the depth location and height of a sub-surface defect. The EC data necessary for training and validation of the two methodologies have been obtained using CIVA numerical modeling software [8]. This section presents a detailed description of CIVA and inversion methodologies. 2.1 CIVA model for EC data generation
The success of an inversion methodology depends on the amount of training dataset used in the learning process. Generation of experimental signals or images for training purpose is time consuming and cumbersome. Hence, in this study, numerical modeling has been used for generation of dataset for both training and validation of the inversion methodologies. CIVA software has been used for this purpose. CIVA is a benchmarked NDT modeling software based on semi-analytical methods using dyadic Greens functions method [8]. In this method, the interaction between defect and electric field generated by EC probe is described with an integral equation (1), which is derived from Maxwell’s equations and solved numerically using the method of moments. (1)
The unknown fictitious current density, JΩ is defined in the volume Ω containing the defect and depends on the total electric field. The solved current density is used for calculating the probe response or signal for a defect. The term J0 in equation (1) is an excitation term that depends on the total primary electric field E0(r) emitted by the —ee probe in the region Ω containing defect. The dyad GΩ links the fictitious current density to the electric field it creates inside Ω [9]. The contrast function f(r) is defined by (2) σ −σ (r ) f (r ) = 0 σ0 where σ0 is the tube conductivity and σ(r) is the flaw conductivity. CIVA eddy current module has been extensively validated through a series of experiments [8-9]. Vol. 10, Issue 3 December 2011
Competitive learning is an artificial neural network type of learning process with no hidden layers. The first layer is the input layer and the second layer is the output layer or the layer of competitive neurons, as shown in Figure 2. The output neurons compete among themselves to respond to an input, and only a single output neuron which is close to an input, in the sense of cosine similarity, wins to respond. Let X be a set of input vectors, each vector x is normalized to be of unit length ( ||x|| = 1), as magnitude is not relevant feature to learn for directional data. Thus, the learning is on unit hypersphere. Let O = {O1...OK} be a set of weights for output neurons, then learning aims to maximize the average cosine similarity objective function L
Fig. 2 : Schematic of competitive learning neural network.
L = ∑ x X T Ok ( x )
(3)
maxXTO . The weight of output neuron where k(x) = arg k k which closely matches the input neuron is the winner. Once the winner is identified, the weights of winning neuron is updated as ) Ok( new (x) =
Ok ( x )+η X Ok ( x )+η X
(4)
where η is the learning rate parameter. Each group of inputs which are activated by one output neuron form a cluster. Figure 3 gives an illustration for the type of clusters of competitive learning on Euclidean and unit hypersphere. In the Euclidean space, clusters will be in the form of ellipsoids scattered throughout the feature space and the decision boundary will be hyper-planes which vary with each other in magnitude as well as in direction. On the Journal of Non destructive Testing & Evaluation
Technical Paper
62
2.3.2 Multi-dimensional learning through multi-label radial basis function
Given a multi-dimensional training set D ={(xi,ci)|1≤i≤n}, then multi-dimensional learning attempts to learn a function f3 that would assign a vector of class label with d dimension for each unseen m dimensional vector of instance given by x ∈ X. (a)
f 3 : φx1 × φx2 × . . . φxm → φc1 × φc2 × . . . × φcd
(b)
Fig. 3 : Clusters on a) Euclidean space and b) unit hypersphere for two features with M as cluster centers and dotted line as the decision boundary.
contrary, for hyper-sphere, the features will be embedded on the surface of unit hypersphere and the decision boundary will be the hyper-planes which vary with each other only in direction (with magnitude being 1). The data which are represented with unit length and learned with cosine similarity are termed as the ‘directional data’. 2.3 Multi-dimensional multi-label radial basis function neural network
Multi-dimensional learning is more generalized form of supervised learning and relatively a new concept in the literature of machine learning and artificial intelligence where more than one dimension of classes can be simultaneously mapped for an input. The following subsection gives the definition of multi-label learning and multi-dimension learning and relationship between them for solving an eddy current inverse problem of simultaneous quantification of subsurface defect height and depth. 2.3.1 Multi-label learning
Multi-label classification is a variation of supervised learning where the task is to learn a function f1 which assigns a subset of label from a set, e.g. L= {1, 2, …, l} to each unseen instance given by a vector of feature x∈X, the domain of instance and a multi-label training set D = {(xi,ci)|1≤i≤n} is given with ci⊂L.
f1 : φx1 × φx2 × . . . × φxm → 2 L
(5)
(7)
where φc denoting the sample space of ci assumed to be i discrete, for all i =1, 2, …, d. With, |φc|≥2. φx denoting j the sample space of feature variable xj for all j=1,2, …m. A wrapper approach is used for multi-dimensional learning for simultaneous quantification of defect depth and height. This wrapper consists of a problem transformation, a multi-label ranking algorithm and class assigning module for each dimension [12]. Multi-label radial basis function (ML-RBF) is multi-label ranking algorithm derived from popular radial basis function type of neural network. The architecture of ML-RBF is shown in Figure 4 and it consists of input, hidden and output layers [12]. The m dimensional feature vectors form the input. The hidden layer consists of L sets of prototype vectors, where L is the number of class. Each set comprises of kl prototype vectors which are the centroids of the clusters taken by performing e.g., popular K-means algorithm on all instances corresponding to each class. kl is derived from fraction α from the total number of instances in class l. l k . The hidden neurons are re-indexed as 1≤j≤K, where Σi=1 1 In addition, a bias prototype neuron Ω0 is present in the hidden layer which is set to 1. The radial basis function given in equation (8) is used for activation of feature vector and the hidden layer.
dist ( x , c )2 i j Ω j ( xi ) = exp − 2 2σ j
(8)
where σ is the smoothing parameter derived from a fraction ì of average distance between each pair of prototype
where φx denoting the sample space of feature variable xj j for all j=1, 2, …, m. Multi-label classification is closely associated with multilabel ranking and it is a problem of learning a function f2 which produces a vector of real numbers with size equal to |L| as output to each unseen instance of a feature vector x∈X, and a training set D={(xi,ci)|1≤i≤n} is given.
f 2 : φx1 × φx2 × . . . φxm × L → R
(6)
where R is the set of real numbers. The function f2 is expected to have the property of ordering the set of labels L, so that the topmost labels are more related with the new instance. Overview of multi-label learning and feature selection strategy for multi-label ranking is discussed in detail elsewhere [11]. Journal of Non destructive Testing & Evaluation
Fig. 4 : Architecture of multi-label radial basis function neural network. Vol. 10, Issue 3 December 2011
Technical Paper
vectors. The output layer is connected with the hidden layer through weights. The weights are learned by minimizing the error from the actual label E = ÎŁ(hâ&#x20AC;&#x201C;t)2 (9) where h is the actual output and t is the desired output.
3. IMPLEMENTATION OF THE INVERSION METHODOLOGIES Competitive learning with cosine similarity (CLCS) algorithm and ML-RBF algorithm have been developed using MatLab. These two methodologies have been tested and validated using the EC data predicted using CIVA modeling software. 3.1 Methodology for identification of layer with defect
CLCS algorithm has been used to identify the layer to which the defect belongs from the EC images of defects. This methodology consists of the following steps: i) Generation of dataset of EC images at two different frequencies ii) Principle component analysis based extraction of features from EC images iii) Unit length normalization of the features iv) Training a competitive learning neural network with cosine similarity v) Validation and evaluation of the algorithm A cylindrical two-layer system with 0.45 mm thick stainless steel (first layer) and 0.2 mm thick sodium (second layer) as shown in Figure 5 has been considered. Defects are assumed to be present either in stainless steel (SS) or in sodium. Unambiguous identification of layer in which the defect exists has been attempted using the CLCS methodology. Dataset of parallelepiped shape defects are modeled in three locations viz., outer diameter (OD) type, sub-surface or inner diameter (ID) type as illustrated in Figure 6. For each type of defect, three different depths (0.05, 0.10 and 0.15 mm) and lengths (1.0, 1.5 and 2.0 mm) have been modeled. With 3 depths, 3 lengths and three locations in SS (layer1) and sodium (layer2), total 54 defects have been modeled. Based on the thickness of the two layers, 300 kHz and 600 kHz have been chosen as the test frequencies. Thus, with two excitation frequencies, there are 108 images in total.
Fig. 5 : The geometry modeled by CIVA to generate EC images of defects. Vol. 10, Issue 3 December 2011
63 These EC images are subjected to principal component analysis for dimensionality reduction and feature extraction. PCA is a linear transformation method where a computation of covariance of input image is performed and the subspace with larger Eigen values of covariance is retained [13]. In this study, three dominant Eigen values from PCA are determined for each image. With two excitation frequencies, each defect is associated with two images. Hence, a defect is represented by six feature vectors which are normalized to unit length. The normalized data is given as input to the CLCS methodology for unsupervised learning. 3.2 Methodology for quantification of subsurface defects
The objective of this methodology is to simultaneously determine the depth location and height of subsurface defects in SS plates as shown in Figure 7. Multidimensional learning through multi-label radial basis functions (MD-Learn ML-RBF ) has been used. This methodology consists of the following steps: i) Generation of dataset of EC signals
Fig. 6 : Parallelepiped shaped a) OD, b) sub-surface, c) ID defects for which EC images are predicted using CIVA.
ii) Extraction of features from impedance plane EC signals iii) Training of ML-RBF by multidimensional learning iv) Validation and evaluation of the algorithm EC signals of subsurface defects having different depths and heights present in a 5 mm thick SS plate have been predicted using CIVA. Defects of three different lengths (3, 6, 9 mm) have been modelled. The width of all defects has been fixed as 0.5 mm. A total of 192 defects have been modelled with different heights (0.5, 0.6, 0.7, 1.0, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 2.0 mm) and depth locations (1.0, 1.3, 1.4, 1.5, 2, 2.5, 2.6, 2.7, 3.0 mm). For convenience, the defects have been categorized into three classes based on both height and depth. Table 1 gives detail of the classes and total number of training data sets present in each class.
Fig. 7 : Subsurface defects used for testing and validation of MDLearn ML-RBF methodology. Journal of Non destructive Testing & Evaluation
Technical Paper
64 Table 1 : Description of class for quantification of subsurface defects. Depth, mm
Class
Number Height, of training mm data
Class
Number of training data
<1.5
A1
51
<0.7
A2
54
1.5-2.5
B1
66
0.7-1.6
B2
90
>2.5
C1
75
>1.6
C2
48
Ten characteristic features viz. F1 to F10 have been determined from the impedance plane of EC signals as shown in Figure 8. These features for each defect along with their corresponding classes for both height and depth are given to MD-Learn ML-RBF. The outputs of the MDLearn ML-RBF are depth and height of the subsurface defects. Among the features, F9 and F10 have been proposed for the first time. A restricted forward subset selection method has been used to select optimal number of features [14]. This method works by evaluating the learning algorithm with a set of inputs. Features are sorted in combinations i.e. most contributing to the least.
4. RESULTS AND DISCUSSION 4.1 Identification of layer with defect
Figure 9 shows the typical EC images of parallelepiped type defects (length, 1.5 mm, width, 1.5 mm and depth, 0.1 mm) in SS and sodium layers at 300 kHz. As can be seen, it is difficult to identify the layer information from the images as they are nearly identical. Figure 10a shows the dominant Eigen values of 300 kHz and 600 kHz, corresponding to set of defects from SS and sodium layers. The unit length normalized Eigen values for these two frequencies are shown in Figure 10b. As can be seen, there is directionality in the Eigen features of the images of the two layers. Thus, use of cosine similarity is beneficial for this directional dataset. In order to arrive at an optimal number of features, three dominant Eigen values from 300 kHz and 600 kHz have been chosen and ranked based on leave one cross validation
strategy. Table 2 shows the number of occasions of misclassifications i.e. where defects in SS are predicted as in sodium or vice versa. It can be observed from Table 2 that successful classification i.e. correct identification of layer with defect for all trained data is possible only with two dominant Eigen values from 300 kHz and 600 kHz. Table 2 : Cross validation results with CLCS method. Input features
Number of misclassifications out of 54 data
300 kHz - Eigen1 and Eigen2
15
300 kHz - Eigen1, 2, 3
13
600 kHz - Eigen1, 2
16
600 kHz - Eigen1, 2, 3
13
300 kHz - Eigen1; 600 kHz â&#x20AC;&#x201C; Eigen1
0
300 kHz- Eigen1, 2; 600 kHz â&#x20AC;&#x201C; Eigen1, 2
0
In order to validate the performance and robustness of the CLCS methodology, fresh set of images of defects of same size, but of slightly different shape have been fed to the CLCS method. In contrast to the parallelepiped shaped defects (shown in Figure 6), elliptical shaped defects as shown in Figure 11 have been used. In the context of training with parallelepiped defects and evaluating with elliptical type defects, the following three different cases have been studied: Case 1: Train with all lengths and depths and evaluate for all lengths and depths. Case 2: Train with all possible combinations of two out of three lengths and evaluate with all lengths and depths. Case 3: Train with all possible combinations of two out of three depths and evaluate with all lengths and depths. The classification results for these three cases are shown in Table 3. As can be noted, the CLCS methodology is robust enough with > 94% success to identify the layer to which the defect belongs, despite change in the shape of defects used for training.
Fig. 8 : Characteristic features derived from impedance plane EC signals of subsurface defects. Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
65
Fig. 9 : Model predicted images of a defect in a) SS and in b) sodium at 300 kHz.
Fig. 10 : Dominant Eigen values at 300 kHz and 600 kHz in a) two-dimensional plane and b) unit hypersphere.
Table 3 : Performance evaluation of the CLCS methodology.
0.05
1.5
SS
2.0
Sodium
Number of data set
Identification success, %
0.05
Case 1
54
100
0.1
1.0
SS
Case 2
108
99
0.1
1.5
Sodium
Case 3
108
94
0.1
2.0
SS
0.15
1.0
SS
0.15
1.5
SS
0.15
2.0
SS
Test case
A special test case wherein an ID defect in SS (layer 1) filled with sodium has been analyzed to further investigate the robustness of the CLCS methodology. Typical predicted images of this scenario are shown in Figure 12. As can be noted from the grey level intensity bar of the images, sodium filling has increased the EC response. This is attributed to the approximately 25 times higher electrical conductivity of sodium over SS. The results of the CLCS methodology are given in Table 4. It can be observed that the CLCS methodology has shown good performance, despite a few misclassifications. The studies clearly reveal that competitive learning with cosine similarity methodology is reliable and robust and can be used for identification of layer in which the defect has formed. Table 4 : Results of CLCS methodology for ID defects in SS filled with sodium. Defect depth, mm
Defect length, mm
CLCS methodology predicted layer
0.05
1.0
SS
Vol. 10, Issue 3 December 2011
4.2 Quantification of subsurface defects
Figures 13a and 13b show the typical impedance plane EC signals of defects having different heights and depths. As can be seen, there is no direct correlation between the shape of the impedance plane signals and the defect dimensions. Features extracted from these signals have been subjected to MD-Learn ML-RBF and Table 5 shows the results. It is seen from Table 5 that the new feature F9 (ratio of maximum magnitude at two frequencies) has attained rank 1 with maximum contribution for the quantitative performance. In order to find the number of optimal input features for effective quantification of subsurface defects, quantification accuracy has been analyzed as shown in Figure 14. As can be noted, the accuracy is high and nearly constant for the first five features while it drops significantly beyond six features. Thus, an optimal number of first Journal of Non destructive Testing & Evaluation
Technical Paper
66 Table 5 : Performance of MD-Learn ranking.
ML-RBF
Feature
method of Rank
F9-Ratio of maximum magnitude at two frequencies 1
Fig. 11 : Elliptical shaped a) OD, b) sub-surface and c) ID defects used for validation.
five ranked features have been chosen as input to the MD-Learn ML-RBF. The MD-Learn ML-RBF methodology has been trained using 192 datasets (length of 3, 6 and 9 mm, width 0.5 mm) and validated with the following three unique cases: Case 1: Evaluation of lengths 4, 5, 7 and 8 mm (interpolation of length). Case 2: Evaluation of lengths 4, 5, 7 and 8 mm and widths 0.75 and 1 mm (interpolation of length, extrapolation of width). Case 3: Evaluation of lengths 2 and10 mm and widths 0.75 and 1 mm (extrapolation of length and width). Table 6 shows the results of the MD-Learn ML-RBF methodology. In all the validation cases, the MD-Learn ML-RBF has quantified the depth location and height with 100% accuracy. This reveals the fact that the MD-Learn ML-RBF methodology is robust for simultaneous quantification of defect depth location and height.
F2-Phase at maximum magnitude
2
F8-Phase at maximum reactance
2
F6-Phase at maximum resistance
3
F1-Maximum magnitude
4
F5-Maximum resistance
5
F4-Magnitude at maximum phase
6
F3-Maximum phase
7
F7-Maximum reactance
8
F10-Ratio of maximum phase at two frequencies
9
Table 6 : Results of validation of MD-Learn methodology. Test case
Number of data set
ML-RBF
Quantification Accuracy, % Depth
Height
Case 1
8
100%
100%
Case 2
8
100%
100%
Case 3
10
100%
100%
Fig. 12 : EC images of a) ID SS defect (depth 0.1mm, length 1.0 mm) at 300 kHz and 600 kHz and b) when it is filled with sodium. Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011
Technical Paper
67
Fig. 13 : CIVA model predicted EC signals of defects at 300 kHz, a) varying height (depth 1 mm) b) varying depth (height 0.5 mm).
REFERENCES
Fig. 14 : Result of restricted forward feature selection.
CONCLUSION Two different methodologies have been proposed for inversion of eddy current data (images/signals). Competitive learning with cosine similarity method has been able to successfully identify defects between two layers, despite small change in defect shape and size. The method has also been able to handle complex situations such as defect in stainless steel filled with sodium i.e. an untrained dataset. Multi-dimension learning using multi-label radial basis function (MD-Learn ML-RBF) has been able to simultaneously quantify both depth location and height of subsurface defects in stainless steel plates when five ranked features have been used as input. A new parameter called ‘Ratio of maximum magnitude at two frequencies’ identified in this study has ranked 1 by the restricted forward selection algorithm. MD-Learn ML-RBF has shown capability for interpolation and extrapolation of depth locations and heights. The two indirect inversion methodologies proposed in this paper are promising for developing artificial intelligence based automated identification and quantification of defects. ACKNOWLEDGEMENTS Authors thank Shri S.C. Chetal, Director, IGCAR, Kalpakkam for encouragement and support during the course of this work.
Vol. 10, Issue 3 December 2011
1. Udpa L, Udpa S.S. “Solution of inverse problems in eddycurrent nondestructive evaluation (NDE)”. J Nondestructive Evaluation 7(1/2) (1988) 111-120. 2. Bowler, J.R. “Thin-Skin Eddy-Current Inversion for the Determination of Crack Shapes”, Inverse Problems 18 (2002) 281- 291 3. Sung-Jin Song, Young-Kil., Shin, “Eddy current flaw characterization in tubes by neural networks and finite element modeling”, NDT&E International 33 (2000) 233-243 4. Rao, B.P.C., Raj, B., Jayakumar, T. and Kalyanasundaram, P.,”An artificial neural network for eddy current testing of austenitic stainless steel welds” NDT & E International 35:6 (2002) 393398. 5. Thirunavukkarasu S., Rao B.P.C, Jayakumar T, Kalyanasundaram P and Baldev Raj, “Quantitative eddy current testing using radial basis function neural networks”, Materials Evaluation, 62:12, (2004), 1213-1217. 6. Andrea Bernieri, Luigi Ferrigno, Marco Laracca, and Mario Molinara, “Crack Shape Reconstruction in Eddy Current Testing using Machine Learning Systems for Regression”, IEEE Transactions on Instrumentation and Measurement, 57:9 (2008) 1958-1968 7. Jacek Jarmulak, Eugene J.H. Kerckhoffs, Peter-Paul van’t Veen, “Case-based reasoning for interpretation of data from nondestructive testing”, Engineering Applications of Artificial Intelligence 14 (2001) 401–417 8. Gilles-Pascaud, C., Pichenot, G., Premel, D., Reboud, C. and Skarlatos, A., “Modeling of Eddy Current Inspections with CIVA”, Review of Progress in Quantitative Non Destructive Evaluation (2008) 9. Reboud, C., Pichenot, G., Prémel, D. and Raillon, R. Benchmark Results: Modeling with CIVA of 3D Flaws Responses in Planar and Cylindrical Work Pieces. AIP Conf. Proc. 1096, (2009) 1915. 10. Zhong, S.,”Efficient online spherical K-means clustering”, IEEE International Joint Conference on Neural Networks, (2005) 3180 - 3185. 11. Tsoumakas G and Katakis I. “Multi-label classification: An overview”, International Journal of Data Warehousing and Mining, 3:3 (2007) 1-13. 12. Zhang M.L., “ML-RBF: RBF neural networks for multi-label learning”, Neural Processing Letters 29 (2) (2009) 61–74. 13. Rao B.P.C, Thirunavukkarasu S, Nand K.K, Jayakumar T, Kalyanasundaram P and Baldev Raj, “Enhancement of magnetic flux leakage images of defects in carbon steel using Eigen vector based approach”, Journal of Nondestructive Testing and Evaluation (Taylor & Francis), 23:1, (2008), 35-42. 14. Kan Deng, “OMEGA: On-line memory based general purpose system classifier”, PhD Dissertation, The Robotics Institute School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213 (1998) Journal of Non destructive Testing & Evaluation
68
Perspective
Probe The second in the list is Perception (a very tricky thing). It depends upon the mind- the data store house (It is akin to interpretation which is based on experience) and differs from person to person and for the same person from time to time. The teacher explained to the class that the equation which he was about to write is complicated and an un solved mystery. He wanted the students to try and solve it in their homes and come back to the class the next day and proceeded to write the equation. As he was writing the equation a boy entered the class. He had not listened to what the lecturer was saying and thinking that the equation was the home work took it down. Next day when the class commenced no one except the late comer boy had solved the equation. Except this boy all the other boys were conditioned by the teacher that they could not solve the equation and hence they could not. This boy was free of that influence and hence succeeded. We go buy the data that is already available with us (Our mind). As brought out in the earlier issue we are neither the body nor the mind. We are a part of the supreme consciousness (remember the Boson). The base of all material things is that same particle or energy, only the manifestation is different. This fact cannot be illustrated better than this story. A disciple in the monastery was upset and hence angry about how he was treated by fellow disciples and wished to meet the master about the same. He was getting worked up while he was going to meet the master and to let out the steam he kicked the door. The master was watching this from his room. When the disciple reached his presence and complained to the master, the master enquired what the door did to him to receive the kick. He asked the disciple to go and apologize to the door. In the perception of the master everything in the world is same. So we shall treat all them as equal. Why did the disciple behave like that? It is his reaction to the incident. All of us are conditioned in that way because of the past data (our mind). We may think that the response to the stimuli is the right thing as it is instantaneous. But there is a small time gap between the stimuli and the response. As long as we retain the response we retain the control with us. We shall become aware of this fact and take appropriate action and promote harmony in our relationships. The situation can be compared to that of PT. A drop of liquid on any surface is subjected to 3 forces (Amongst the Surface, the Liquid and the Air) which decide the surface tension of the liquid. As the surface or the liquid changes the surface tension also changes. For human beings the changes in the environment brings in various forces into action creating perceived tension, as things do not go as per our wish. (Just think can the world move if it has to satisfy every ones wish). A droplet of water on steel surface spreads thin and tries to cover the whole surface(Wetting ability), whereas on a lotus leaf it assumes the form of a sphere and reflects the whole universe. But unlike the penetrant particles since the human being is an evolved manifestation of the same basic energy, it is bestowed with the chance of a choice. We can choose and act - either spread out and look for defects in others or we can encompass the whole universe and merge with it. Remember the choice lies in that small gap of time. I wish to end this soliloquy with a story of Mullah. A farmer approached the Mullah with a complaint that his house is too small to accommodate his entire family and he does not have privacy even to say his daily prayers. The Mullah asked him whether he has faith in him and will do whatever he says. The farmer said yes. Then the Mullah asked him whether he has a goat? The farmer replied in the affirmative. The Mullah asked the farmer to tie the goat inside his house and come and see him after 10 days. The farmer as agreed did the same and met the Mullah after 10 days. Now the Mullah asks him to shift the goat outside and come and see him the next day. Next day when the Mullah asks him how he is feeling the farmer replies that he is on top of the world and very happy. That is in a nutshell is Perception.
Ram Journal of Non destructive Testing & Evaluation
Vol. 10, Issue 3 December 2011