2
3
4
CONTENT
1 SOUND PERCEPTION SO UN D IN TR O DUC TIO N 0 0 6 ' L o- Fi ' SO UN DSC A P E 0 1 4
2 SOUND & SPACE STUDY SP A C E EFFEC T O N SO UN D 0 2 0 SO UN D EFFEC T O N SP A C E 0 3 0
3 SITE ANALYSIS SITE IN TR O DUC TIO N 0 3 8 SO UN D R EFL EC TIO N SIMUL A TIO N 0 4 8 DATA A N A L Y SIS / MA C H IN E L EA R N IN G 0 6 2
4 URBAN EXPLORATION P R O G R A M DISTR IB UTIO N 0 7 8 FO R M G EN ER A TIO N 0 8 6 DETA IL O P TIMIZA TIO N 1 0 4 ' L o- Fi ' C ITY 1 3 0
5
6
1 SOUND PERCEPTION
Sound is the energy that elements produce when they vibrate. Hearing is one of the most important senses that affect the human behaviour and orientation within the urban context. Urban sounds have multiple effects on the architecture of the cities. Our objective is to use sound as a tool to experiment and design new urban environments.
1.1 S O U N D I N T R O D U C T I O N There are two different aspects of sound: there's a physical process that produces sound energy which propagates through the air, and there's a psychological process that occurs within the human ears and brai ns, which converts the incoming sound energy in to sensations that humans interpret as noise, music, speech, etc. Sound always needs a transmission medium to travel through such as air, water, metal, or glass. All sound waves travel through a medium by making atoms or molecules shake back and forth. Concurrently, all sound waves are different. There are loud sounds and quiet sounds, high-pitched sounds and low-pitched sounds, and even two instruments playing exactly the same musical note will produce sound waves that are quite different.
W h a t is so und ? Sound is a vibration that propagates as an audible wave
Sound Propagation_ Sound Wave
of pressure, through a transmission medium that can be solid, liquid or gas. For the human ear, sound is the reception of such waves and their perception by the brain. The human ear can hear sound waves when the frequency lies between 20Hz and 20kHz and can distinguish around 1400 different pitches. Sound above 20 kHz is ultrasound and is not perceptible by humans. Sound waves below 20 Hz are known as infrasound. The frequencies of typical sounds produced by human speech lie between 100 and 1,000Hz.
H o w d o e s so und p ro p ag ate? Sound moves through a medium from the point of generation to the listener. When an object vibrates, it sets the particles of the medium around it vibrating. The particles do not travel all the way from the vibrating o bject to the ear. A particle
Distance Short
Long
of the medium in contact with the vibrating objects is first displaced from its equilibrium position. It then exerts a force on the adjacent particle. As a result of which the adjacent particle gets displaced from its position of rest. After displacing the adjacent particle the first particle comes back to its original position. This process continues in the medium till the sound reaches the ear. The disturbance created by a source of sound in the medium travels through the medium. A wave is a disturbance that moves through a medium when the particles of the medium set neighbouring particles into motion. They in turn produce similar motion in others. The particles of the medium do not move forward themselves, but the disturbance is carried forward. This is what happens during propagation of sound in a medium, hence sound can be visua lized as a wave.
W h a t is a sp e c tro g ram? A sound spectrogram is a visual representation of an acoustic
signal.
A
spectrogram
provides
information
on a number of sound properties that may be subject to quantitative analysis. The frequency of a sound is displayed in the v ertical axis of a spectrogram and the time scale over which a sound was recorded is represented by the horizontal axis. In addition, degrees of amplitude of sound are represented with colour (light-to-dark, as in white=no energy, black=lots of energy).
007
Sound Reflection Times
S o u n d Pro p e rtie s ·Frequency/Pitch
refers
to
the
number
of
periodic,
compression-rare-faction cycles that occur each second as a sound wave moves through a medium. It is measured in Hertz (H z). The term pitch is used to describe our perception of frequencies within the range of human hearing.
·Amplitude/Loudness refers to how loud or soft the sound is.
·Timbre refers to the characteristic sound or tone color of an instrument.
·Duration refers to how long a sound lasts.
·Envelope refers to the shape or contour of the sound as it evolves over time. A simple envelope consists of three parts: attack, sustain, and decay. An acoustic guitar has a sharp attack, little sustain and rapid decay. A piano has a sharp attack, medium sustain and medium decay. Voice, wind, and string instruments can shape the individual attack, sustain, and decay portions of the sound.
·Location refers to the sound placement relative to our listening position. Sound is perceived in three-dimensional space based on the time difference it reaches our left and right eardrums.
Pulse As Plomp forwarded in 1964, the decay of auditory sensation is a gradual one such that give n two pulses separated by
Diagram of the stimulus level (upper) and the
silence, the decay of the sensation of the first pulse is not
sensation level (lower) in regard to time of two
immediate and lingers within the silence between pulses.
pulses. First proposed by Plomp (1964).
008
R e v e r be ratio n Reverberation is a persistence of sound after the sound is produced. It is created when a sound or signal is reflected causing a large number of reflections to build up and then decay as the sound is absorbed by the surfaces of objects in the space. This is most noticeable when the sound source stops but the reflections continue, decreasing in amplitude, until they reach zero amplitude.
Reverberation occurs naturally when a person talks, sings, or plays an instrument in a hall or performance space with sound-reflective surfaces.
Typical reverberation times for various rooms. The metric T60 is a measure of the amount of time required for the level of a steady sound to decay by 60dB after the sound has stopped.
Reverberation is frequency dependent. The length of the decay, or reverberation time, receives special consideration in the architectural design of spaces which need to have specific reverberation ti mes to achieve optimum performance for their intended activity.
R e fl e c tio n When a longitudinal sound wave strikes a flat surface, sound is reflected in a coherent manner provided that the dimension of the reflective surface is large compared to the wavelength of the sound. Note that audible sound has a very wide frequency range (from 20 to about 17000 Hz), and thus a very wide range of wavelengths (from about 20 mm to 17 m). As a result, the overall nature of the reflection varies according to the texture and structure of the surface. For example, porous materials will absorb some energy, and rough materials (where rough is relative to the wavelength) tend to reflect in many directions—to scatter the energy, rather than to reflect it coherently. This leads into the field of architectural acoustics, because the nature of these reflections is critical to the auditory feel of a space. In the theory of exterior noise mitigation, reflective surface size mildly detracts from the concept of a noise barrier by reflecting some of the sound into the opposite direction. Sound reflection can affect the acoustic space.
Diagram of typical behaviors of sounds regarding various geometries 009
D iffu s io n Diffusion is the method of spreading out sound energy in order to improve the interior sound experience. A perfectly diffusive sound space is one that has certain acoustic properties which are the same anywhere in the space.
Diffusors (or diffusers) are used to treat sound aberrations, such as echoes, in rooms. They are an excellent alternative or complement to sound absorption because they do not remove sound energy, but can be used to effectively reduce distinct echoes and reflections while still leaving a live sounding space. Compared to a reflective surface, which will cause most of the energy to be reflected off at an angle equal to the angle of incidence, a diffusor will cause the sound energy to be radiated in many directions, hence leading to a more diffusive acoustic space.
Diffraction of sound waves around objects allows sound to be perceived both behind an obstacle (shadow aone diffraction)
T y p e s o f D iffuso rs
as well as emerging from a hole (spherical diffraction).
Maximum length sequence diffusors (MLS) Maximum length sequence diffusors are made of stripes of material with two different depths. The width of the stripes is smaller than or equal to half the wavelength of the frequency where the maximum scattering effect is desired. Ideally, small vertical walls are placed between lower stripes, improving the scattering sound effect. The disadvantages of these diffusors is that their bandwidth is quite limited; when one octave above the design frequency occurs, the diffusors obtain the features of a flat surface.
Quadratic-residue diffusors (QRD) The Quadratic-residue diffusors can be designed to diffuse sound in either one or two direct ions. QRD function is based on shifting the phase of the reflected sound to spread the energy. Pyramid and cube-shaped reflectors are commonly used as diffusion elements, but they work better at a farther distance from listeners than curved or spherical diffusors due to their flat reflective surfaces.
The sections of cylindrical and spherical shapes are better geometry options because they evenly spread sound energy within a space. The fixed radius of cylindrical and spherical shapes spreads the sound consistently regardless of the direction of incoming sound waves. Spherical geometries diffuse in two dimensions (heights and width), while cylindrical ones diffuse in either width or height, depending on orientation.
Polar Responce of a 2D QRD 010
Sound Propagation Study The folded-horn speaker takes advantage of the use of an acoustic horn to increase the overall efficiency of the driving element. The interior of a folded-horn speaker is divided by partitions to form a zigzag flaring duct which functions as a horn. The horn can improve the coupling efficiency between the speaker driver and the air. The folded-horn speaker can have as result greater acoustic output power. The construction of a folded-horn speaker followed a series of experimentations to test the effect of space on sound propagation. Firstly, the acoustic function of the speaker was tested digitally. Six receivers were placed in the interior of the speaker and one receiver was placed outside the speaker.
Through
digital
simulation,
we
recorded
three attributes (definition-D50, sound pressure level, reverberation
time-T60)
of
sound
collected valid comparative data.
011
separately,
and
012
1.2 ' L o - F i ' S O U N D S C A P E The urban soundscape is constantly adapting and changing, as humans and cities interact with each other and create new spatial and acoustic patterns. Modern humans inhabit cities with acoustic environments significantly different from the previous. These new post-industrial sounds, which differ in quality and intensity from the previous ones, have been split from their original source and given an independent and amplified existence.
A soundscape refers to acoustic environments that can be perceived by humans, from natural sounds to music and voice. People orientate themselves in their surrounding environment with the help of sensorial stimuli and the information gathered and processed through them. Even though vision is considered to be the main source of spatial perception, hearing holds a crucial role in spatial experience because it evaluates information from a variety of sources and helps the person conceive the space and her/his relationship with it. The soundscape and the auditory perception can influence people’s orientation in a city and the choice to use an urban space.
Globalisation, economic relations of production and consumption and new digital technologies compose the character of the modern digital cities. Noise is a natural product of economic growth. Modern roads are noisier than ever before, as in some cases traffic noise has reached levels where it is dangerous to people’s health. In the United Kingdom the government’s environmental p rotection watchdog found recently that around one in seven residents of the British cities of London, Manchester and Southampton were exposed constantly to traffic noise above 65dB. Increasing noise levels and poor acoustics are related to a variety of health issues, from high blood pressure to anxiety, irritability, insomnia and even heart disease.
Concept of "Lo-fi City" According to Schafer, the transition from the rural to the urban introduced two different acoustic systems: the high fidelity (hi-fi) and the low fidelity (lo-fi). In a hifi environment discrete sounds can be heard clearly due to the low ambient noise level. Even the slightest sound might be considered disturbance that can communicate important information and the listener is able to hear farther into the distance.
The lo-fi soundscape originates from the Industrial and Electric Revolution. In this case, there is an overdose of new sounds that can be perceived as unpleasant and perspective is lost. Acoustic signals need to be amplified in order to be heard. The sound of nature tends to lose its character under the abundant jamming of technology, machinery and traffic. We aim to take advantage of this plurality of sounds and create an urban intervention
Photo of the London Olympic Games
that redistributes the sounds in the city.
Representing the Industrial Revolution
015
The modern urban centres are characterised by road traffic, acoustic signals deriving from people’s voices and activities, and the industrial sounds associated with buildings, urban elements, construction etc.
There is the potential to redefine the character of our cities by forming new soundscapes in accordance to our desires. Our design principle is to include sound as a resource and redistribute the sounds in the city according to their character and form new soundscapes that resonate with people. 016
2 SOUND & SPACE STUDY
Sonic and spacial elements are highly correlated. The auditory perception influences people’s orientation in a city and the choice to use an urban space. The acoustic perception of sonic environments is related with two different approaches: the objective and the subjective. First come the environmental factors that affect perception, such as spacial configuration, built and unbuilt space or weather conditions, and then the identification of sound sources and individual evaluation which is influenced by subjective factors such as socio-economical background, culture, emotion, etc.
2.1 S P A C E E F F E C T O N S O U N D Since the urban fabric changes rapidly, so does the sonic environment of the contemporary city. New sounds make the familiar sound different. These sounds form a character for an area if they are uniquely related to this area, such as people’s voices in a residential area or children voices in a school area. Sound then appears to depict reality and acquires an informative function. Without sound, the sense of visibility is perceived completely different- it is less informative and more attention-demanding. In the same way, acoustic perception is different in various spaces. Space and acoustics interact and support each other in a way that they form a united entity.
Based on the urban context of London, five prototypical forms of open spaces are chosen - rectangular, triangular, circular, semi-circular, and linear. Then elements of the initial prototypes are combined and mixed. We create 10 combinations and we conduct in each one of them two different simulations, one with a point sound source and one with a linear sound source. The third experimentation takes place in transformed complex space, where prototypical elements are combined and transformed. The same point sound source can have a completely different distribution and reflection in various prototypical forms of spaces.
Single Prototypes of Urban Open Space
Open Space Forms
Enclosed Space [Park / Square / Playground]
Semi-circular
021
Rectangular
Triangular
Linear Space [Street]
Circular
Linear
Prototypes of Urban Open Space Sound Simulations in Single Prototypes
Top View
Perspective View
Linear Open Space
Circular Open Space
Triangular Open Space
Rectangular Open Space
022
Combination of Single Prototypes
Single Prototypes
Combination
Two types of Sound Source
Point Sound Source
Linear Sound Source
023
Sound Simulations in Combined Prototypes Point Sound Source
Linear Sound Source
024
Sound Particle Movement in Enclosed Spaces
Sound Particle Movement between Facades
025
Transformation of Prototypes
026
Sound Simulation in Transformed Prototypes
027
Single Sound Line Simulation
Single Sound Line Simulation
Multiple Sound Lines Simulation
Initial Energy:
150
Initial Energy:
600
Initial Energy:
249
Bounces:
1, 0, 0
Bounces:
1, 1, 1
Bounces:
4, 3, 1
Average Energy (%):
27.1656, 0, 0
Average Energy (%):
81.7914, 70.656, 45.1811
Average Energy (%):
68.651, 34.821, 12.424
Bounces:
495, 104, 47
Bounces:
425, 203, 96
Bounces:
375, 145, 57
Bounces:
343, 111, 30
Average Energy (%):
95.8, 93.3, 88.4
Average Energy (%):
92.3, 85.7, 76.9
Average Energy (%):
89.8, 77.7, 66.2
Average Energy (%):
83.4, 71.1, 54.9
Bounces:
442, 123, 44
Bounces:
407, 176, 59
Bounces:
379, 140, 33
Bounces:
345, 120, 26
Average Energy (%):
95.3, 91.8, 87.5
Average Energy (%):
92.4, 86.3, 80.8
Average Energy (%):
90.1, 81.7, 72.1
Average Energy (%):
87.7, 79.0, 73.7
Bounces:
449, 186, 118
Bounces:
412, 165, 70
Bounces:
378, 145, 41
Bounces:
349, 120, 27
Average Energy (%):
93.9, 90.1, 86.9
Average Energy (%):
91.8, 85.4, 79.5
Average Energy (%):
89.1, 81.7, 73.7
Average Energy (%):
87.1, 74.9, 67.2
Bounces:
460, 219, 129
Bounces:
417, 197, 96
Bounces:
382, 156, 62
Bounces:
353, 129, 41
Average Energy (%):
93.1, 89.4, 86.4
Average Energy (%):
90.7, 85.5, 80.3
Average Energy (%):
88.3, 78.9, 67.9
Average Energy (%):
86.1, 73.8, 61.9
028
32
2.2 S O U N D E F F E C T O N S P A C E From the data produced and collected from the digital simulations, we move forward to the rewriting process, in an attempt to transform and manipulate the urban environment of the area, bas ed on its acoustic character. We select an area within the site to deconstruct the building environment and re-simulate the sound reflection in the deconstructed space. Based on the digital simulation, we replace elements of the built environment with elements of different shape and colour, in order to form new shapes and recreate the built environment according to its sonic character. Each element represents different sound reflection line.
Data Rewriting from Sound Simulation [Selected Area of the Site]
Sound Reflection Simulation
Axonmetric View
031
Data Rewriting from Sound Simulation
Sound Reflection Lines Energy & Direction Rewriting Low
High
Energy Level
Perspective View
032
033
Sound Particle Simulation in Rewriting Model
034
3 SITE ANALYSIS
The site analysis contains extensive research and collection of data related to sound distribution and the factors that affect it. Features, such as the noise distribution and the built environment in London (house density, building height) are analysed. The collection of site data involves field studies and recordings and digital simulations. Firstly, by visualizing the recorded data, a better observation of the sound distribution and change on site can be succeeded. Secondly, in the digital sound simulation, a classification of the types of sound is determined, so as to increase the accuracy of the simulation results. Later on, with the help of machine learning algorithms, we pointed out areas with common characteristics that face similar problems or benefits and require the implementation of common strategies and interventions.
3.1 S I T E I N T R O D U C T I O N
By comparing the noise distribution in different areas, it is concluded that in the areas with higher buildings and higher building density the sound is mostly concentrated in the streets, while in the areas with lower buildings and lower building density, the sound is spread faster and further than the built environment. From this initial analysis we select a diverse and mixed-use area to run fur ther analysis and experimentations: the King’s Cross area. King's Cross is served by London King's Cross Railway Station, one of the main railway routes in the UK. Since the mid-1990s, the region has been constantly rebuilt. In 2007, the terminal of the Eurostar Railway Service Station, which connects London with European cities, and the reconstruction of King's Cross Station, a major reconstruction project in the northern part of the region, have been completed. King's Cross still exhibits reminders of its rich industrial heritage, the most important clusters of which are King’s Cross and St Pancras stations. Additionally, there is a mixture of offices, residencies, shops, hotels, galleries, bars and restaurants and also an arts college, part of the University of the Arts London. There’s also diversity of people, including studen ts, professionals, visitors, tourists and residents, a combination of built and open spaces, of the old and the new. 40% of King’s Cross is dedicated as open space – streets, parks and squares. All these features make King’s Cross a diverse, dynamic and always changing environment – a vibrant area with particular sound qualities. The traffic of King's Cross is composed of underground, train and car traffic. In addition, the King's Cross, as an important hub of daily activities in London, carries a large number of people. Therefore, the contradiction between people and traffic noise is prominent, which is our key solution direction.
Urban Soundscape
Road Traffic Noise Level [A 24 Hour Annual Average Noise Level in Decibels ] >75.0 70.0-74.9 65.0-69.9 60.0-64.9 55.0-59.9
>75.0 70.0-74.9 65.0-69.9 60.0-64.9 55.0-59.9
040
Building Morphology Analysis
Building Density [Greater London]
Building Heights [Greater London] Building Height High
Low
Building Heights in Specific Area [Central Area in Greater London]
041
Built Envrionment Effect on Sound Distribution
Road Noise Distribution
Building Heights
Suburban Area - Lower Building Heights
Urban Context
Road Noise Distribution
Building Heights
Central Area - Higher Building Heights
Urban Context 042
Site Selection After comparing the effects of different building heights on sound in two areas in London – one suburban and one central, King's Cross station and its s urrounding square are selected as the site of our proposal. There is large quantity of traffic and people’s flow passing through the area every day. Yet, the selected area demonstrates diversity in terms of its sonic environment, with the northern part to be mostly occupied by human sounds and the southern part to be mainly characterised by traffic sounds.
1
043
2
3
4
5
Visualization of Recorded Data (Decibel) of the Site
Decibel (dB) High
Low
044
Maximum Decibel Visualization [Southern Part of the Site]
Minimum Decibel Visualization [Southern Part of the Site]
045
Positive Sound
Negative Sound
046
3.2 S O U N D R E F L E C T I O N S I M U L A T I O N As designers, we must find a way to make acoustics a positive study program. Which sounds do we want to preserve, encourage or amplify? Which sounds are boring or destructive and we wish to eliminate or dampen? Only a total appreciation of the acoustic environment can give us the resources for improving the urban soundscape. A classification of sounds can reveal similarities, contradictions and motifs. Sounds can be classified according to their physical characteristics, according to their function, according to the way that they are perceived or according to the emotions they cause.
In the part of computing data visualization, sound is firstly classified into four categories: railway noise, road noise, people's voice, and special sound. The special sound consists of music, nature sounds, such as water or wind and other sounds which are produced at a specific time or in a particular place. In order to realize the digital simulation, we later on determine the location of different types of sound sources. As far as the sound sources of railway and road are concerned, there is no fixed position since the vehicles along the railway and the streets are constantly moving. To solve this problem, we divide the noise distribution of railway and streets into five levels according to the annual average decibel level, and select a specific number of sound sources within the corresponding geographical range of each level. These sound source points represent the distribution of the annual average sound source points. For people's voice, we use GPS traces to select the area with the highest human flow density as the main distribution area of people's voice sources. For special sound, taking into account our field investigation and the analysis of the activities on the area, we select several typical points in which special sounds are produced as the sound sources.
Railway Noise Sources Distribution Average Noise Level (dB)
Number of Sound Sources
>75.0
13
70.0-74.9
20
65.0-69.9
15
60.0-64.9
17
55.0-59.9
9
049
Railway Noise Reflection Simulation
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
[N: 300]
Road Noise Sources Distribution Average Noise Level (dB)
Number of Sound Sources
>75.0
13
70.0-74.9
20
65.0-69.9
15
60.0-64.9
17
55.0-59.9
9
051
Road Noise Reflection Simulation
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
[N: 300]
Poeple's Sound Sources Distribution
Distribution of people's sound sources
Determine the center point of each lattice as the sound source point
Select lattices with the highest density of people's GPS traces
People's GPS Traces
053
People's Sound Reflection Simulation
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
[N: 300]
Special Sound Sources Distribution
Playground
Square
Fountain
Music
Fountain
Distributioin of other sound sources
Lattices with the highest density of people's GPS traces
People activity points Subway / bus station Shop Market Restaurant Landscape Hotel Kindergarden Greenspace
055
Special Sound Reflection Simulation
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
[N: 300]
All Sound Sources Distribution
Distribution of railway noise sources
Distribution of road noise sources
Distribution of people's sound sources
Distribution of special sound sources
Distribution of all sound sources
057
All Sound Reflection Simulation
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
[N: 300]
All Sound Reflection Simulation [Perspective]
Original Emitted Sound Lines
[N: 1200]
First Reflected Sound Lines
[N: 900]
Second Reflected Sound Lines [N: 600] Third Reflected Sound Lines
059
[N: 300]
060
3.3 D A T A A N A L Y S I S / M A C H I N E L E A R N I N G Urban environments operate as dynamic networks of spacial practices and evolve through circulation, repetition and interaction. The city and the urban elements modify and adjust themselves to the logic of their surrounding environment. In an attempt to understand the city as a multi-layered organism and to explore its dimensionality, we introduce machine learning technics to collect and analyse data related to the urban environment and sound.
Two machine learning algorithms are used: Principal Component Analysis (PCA) and K-Means clustering algorithm. These algorithms can help us determine the degree of data interaction and cluster analysis. By using machine learning, we tried to analyse the collected data in an extensive way, in order to identify spatial patterns in the site that need different kinds of intervention. Finally, the results of PCA and K-Means algorithm are projected on the geographic space of the site. Based on this, we have a preliminary location of the scope of design intervention.
Maps of Attributing Data
Runner Flow
Cyclist Flow
Visual Connectivity
Choice Radius 3200
Choice Radius 1600
Choice Radius 800
Intergration Radius 3200
Intergration Radius 1600
Rail Noise Class
Building Heights
Intergration Radius 800 063
Road Noise Class
3-Dimensional Scatter Plots
064
Relationship betweeen Attributing Data
065
Principal Component Analysis and K-Means Clustering
Principal Component Analysis and Heatmap As a further step, we used machine learning algorithms to define the interaction between our datasets. Our datasets contain Runner’s Flow, Cyclist Flow, Total Rail Noise and Road Noise in the area, Buildings’ Heights- all factors that affect the distribution of sound in the area. Furthermore, in order to identify the human factor in the site, we run integration and choice analysis with different radii and also a visual connectivity analysis.
First, we used the Principal Component Analysis [PCA]. Through mathematical procedures PCA converts a set of variables that are possibly related into a set of values of linearly unrelated variables (principal components). The dimensionality reduction is a useful way to visualise and process high-dimensional datasets, while still retaining as much of the variance in the dataset as possible. The data are categorized based on 4 noise levels. PCA can give us as output areas that demonstrate the highest variation of data.
The scatterplots produced through machine learning, demonstrate the correlation between the different imported datasets.
066
PCA [Principal Component Analysis] - Scatter Plots
2-Dimensional
3-Dimensional
All Noise Class
0 067
1
2
3
K-Means Clustering - Scatter Plots
2-Dimensional
3-Dimensional
Cluster
Cluster
Cluster
Cluster
Cluster
Cluster
068
Principal Component Analysis [1-dimensional Result Projected on Site]
In order to identify regions of interest, we projected the 1-D result of PCA algorithm into the site. Principal component analysis results are divided into negative and positive values, both of which have the same importance. In this map we use the absolute value to project the results of PCA on the site. The higher the absolute value, the brighter the colour on the map. The areas covered by bright colours show their
distinctive
characteristics,
which
means the changes of principal component variables are more likely to cause changes in the sonic environment in these areas. We regard them as sensitive areas, which are also the reference for determining the design boundary. On the contrary, low values indicate that these areas do not have significant common characteristics. Therefore, we will focus our design intervention on the areas with brighter colours.
Data Values of PCA High [Absolute Value]
Low [Absolute Value]
069
Overlapping Analysis
PCA Values Projected on Site
Height of Lines Representing the Level of Data Values PCA Value High
Low
Urban Context
070
K-Means Clustering [1-dimensional Result Projected on Site]
In the K-Means algorithm, we set up 4 clusters. As shown in the map, the original attribute data show obvious differences about similarity after clustering. Through K-Means analysis, we can further determine the spatial similarity in the key intervention areas from PCA analysis.
Since our datasets mainly contain data about road and rail noise and people’s flow and circulation, we conclude that the areas with high interaction of data are mostly concentrated around the streets and the stations.
K-Means Clustering Value 0 1 2 3
071
Overlapping Analysis
K-Means Clustering Values Projected on Site
Spatial Distribution of K-Means
Clustering Values
Clustering Value 0 1 2 3
Urban Context
072
Selected Areas from PCA
In order to select the area of design intervention, combined with PCA map, some areas with higher values are selected. In this region, area A and B are highlighted as areas with highest PCA values. From the right map, we can see that most of the areas with higher values are distributed along the road. However, in Area A, areas representing higher values are not completely distributed on the road, showing more varied spatial distribution. On the contrary, in Area B, the points with higher values are almost completely distributed on the road, which is in accordance with the busy traffic in Area B, where the sonic environment is diverse and multi-layered.
A
B
Selected High Values from PCA Result 5 4 3 2 1
073
Selected Areas from K-Means Clustering
The results of K-Means algorithm are projected on the selected areas from PCA, as shown in the right map. The region is divided into 4 types of areas with different colours
that
demonstrate
4
different
clusters extracted from K-Means algorithm. Area A and B belong to the same cluster, thus, they have high similarity. In addition, area C and D have similar characteristics with area A and B and they will also be included in the design scope as secondary areas of intervention.
C A
D
K-Means Clustering Values
B
0 1 2 3
074
78
4 URBAN EXPLORATION
Our fundamental steps as designers towards providing solutions to the acoustic problems of contemporary cities and taking advantage of the correlation between vision and sound should be two-sided: firstly to reduce and control noise and secondly to enhance the informative character of each sound. These steps to new sonic experiences within the urban context would not only improve daily city life by overcoming health issues related to bad acoustic quality, such as stress and anxiety, but would also contribute to sensory awareness of city habitants and would provide human-designed environments.
79
80
4.1 P R O G R A M D I S T R I B U T I O N Different urban functions require different acoustic environments. According to our acoustic and topographic analysis we define the types and locations of interventions. Firstly, we define two different sound environments that need different design approach: the first design approach is to enhance sound, and the second is to decrease it. To define a sound environment that requires dampening of sound, we take into consideration building functions, such as schools, libraries, hospitals, etc. that suffer from urban noise and require a quieter sound environment.
Our initial studies focus on the relationship between the sound envi ronment and the building function. Based on the building function, we extracted buildings in the area that are sensitive to noise, such as hospital, hotel or school. These buildings often require a quieter acoustic environment. From this analysis, we created a Noise Sensitivity Heatmap that represents areas that suffer from noise and need dampening of sound.
Consequently, following the same logic, we extract ed areas that are insensitive to sound and could stand for a more diverse acoustic envi ronment. Then, by overlapping the sound sensitivity information with the sound decibel distribution of the site, we identified the areas that need to amplify sound and the areas that need to dampen sound.
81
Sound Sensitivity Heatmap Noise Sensitivity Point "class" = 'communitiy_centre' or 'doctors' or 'embassy' or 'garden_centre' or 'hospital' or 'hostel' or 'hotel' or 'kindergarten' or 'library' or 'memorial' or 'monument' or 'museum' or 'police' or 'post_office' or 'school' or 'shelter' or 'university'
Sound Sensitivity Level 6
0
079
4-6
High Sensitivity
0-2
Low Sensitivity
Decibel Map
Decibel Level (dB) High
Low
Decibel Level (dB) 90 >65.0
High Sensitivity
50.0-55.0
Low Sensitivity
0
080
Map Overlapping
Decibel Map
Sound Sensitivity Map
Overlapped Map
Logic of Design Area Selection
081
Area for Dampening Sound
Area for Amplifying Sound
Activity Space Distribution
Gathering Square
Bridge Music Plaza
Fountain Plaza
Office Building Noise Barrier
Amphitheater
Cafe & Bar
Restaurant Gathering & Sitting Area
Shops
Main Entrance
083
Passing Area
084
4.2 F O R M G E N E R A T I O N The purpose of the urban structure is to influence and manipulate the surrounding sonic environment. Sound reflection is used as the primary tool to design with the aim to form structures that can reflect sound and recreate the urban environment. However, the methods to simulate the reflection of sound provide countless possibilities for shaping forms due to the complexity and diversity of the reflected sound structure. Machine learning methods are used to create suitable functional structures to achieve the design goals for amplification and dampening of sound. In this chapter, we will demonstrate the structure generation process to illustrate our design philosophy.
Space Division - Horizontal Layout [People Activity Point Selection]
087
Space Division - Horizontal Layout [Connection Generation]
Strength: -50
Strength: -50
Cutoff: 10
Cutoff: 10
Strength: -50
Strength: -50
Cutoff: 10
Cutoff: 10
Strength: -50
Strength: -50
Cutoff: 10
Cutoff: 10
People Activity Points & Links
Simplified Link
088
Optimized Space Division By
identifying
the
main
crowd
activity points, we simulated the crowd
activity
route
within
the
design area. Then, based on these activity
routes,
we
defined
the
design areas which we can apply the sound reflection structure.
Additionally,
we
simulated
the
annual sunshine data of this area as the reference data, so that the sunshine out
analysis
for
the
entire
was
carried
design
area.
According to the analysis results, we define the vertical layout of the design areas, which comprises of 4 different levels of 4m, 8m, 12m and 16m respectively.
Elementary Volume Generation Then, we further subdivided each design area. Firstly, a number of control points are defined by the equal range, and control lines are obtained by selecting some of these control
points.
Finally,
thro ugh
these control lines we create the sound reflection structure. The
blue
shapes
represent
the
dampening zone and the pink shapes represent the amplifying zone. The algorithm
controls
the
reflective
surface, so as to gradually reduce the number of sound lines reflected to the dampening zone, and at the same time provide the amplifying zone with the most reflected sound lines possible.
089
Space Division – Three Dimensional Layout
Volume Height Definition Duration of Sunlight: 22nd Dec. [Avg. 0.00-24.00]
Design Volume Projected on Site
Sunlight Analysis
090
Initial Sound Reflection Structure
Elementary Volumes
Zones for Amplifying and Dampening Sound
091
Volume Deconstruction
Defining Centroids
Selecting Centroids
Creating Polylines
Creating Mesh
Form Optimization
Initial Sound Reflection Structure
Influence Value(AN-DN): 68 Iteration Times: 653 Dampening Zone's Soundlines Number(DN): 35 Amplifying Zone's Soundlines Number(AN): 103
Sound Reflection Model Applied to All Volumes
092
Iteration of Initial Sound Reflection Structure
Influence Value: 820
Influence Value: 939
Influence Value: 1088
Influence Value: 879
Influence Value: 965
Influence Value: 1139
Influence Value: 912
Influence Value: 1007
Influence Value: 1178
Influence Value: 374
Influence Value: 1040
Influence Value: 1216
Times of Sound Reflection 1 2 3 4
Iteration Times: 417 093
Influence Value: 1077
5
Iteration of Initial Sound Reflection Structure
Influence Value: 1245
Influence Value: 1365
Influence Value: 1528
Influence Value: 1269
Influence Value: 1423
Influence Value: 1568
Influence Value: 1298
Influence Value: 1490
Influence Value: 1592
Influence Value: 1327
Influence Value: 1513
Influence Value: 1602
Times of Sound Reflection 1 2 3 4
Iteration Times: 1250
Influence Value: 1605
5
094
Iteration of Initial Sound Reflection Structure
Influence Value: 1609
Influence Value: 1645
Influence Value: 2639
Influence Value: 1616
Influence Value: 2435
Influence Value: 2655
Influence Value: 1623
Influence Value: 2599
Influence Value: 2683
Influence Value: 1631
Influence Value: 2615
Influence Value: 2712
Times of Sound Reflection 1 2 3 4
Iteration Times: 1668 095
Influence Value: 2619
5
Iteration of Initial Sound Reflection Structure
Influence Value: 2732
Influence Value: 2791
Influence Value: 2885
Influence Value: 2747
Influence Value: 2856
Influence Value: 2894
Influence Value: 2756
Influence Value: 2870
Influence Value: 2901
Influence Value: 2767
Influence Value: 2878
Influence Value: 2960
Times of Sound Reflection 1 2 3 4
Iteration Times: 2506
Influence Value: 2960
5
096
Optimization Initial Final Result Identification of Result Initialof ofSound Morphology Main Structure Reflection Structure Iteration Times: 2506 Influence Value: 2960
Iteration Time: 2506 Influence Value: 2960
重画或者重新 P
Times of Sound Reflection
1 2 3 4 5
Form Iterative Optimization
Original Vertical Surfaces
Trimmed Vertical Surfaces & Sound Amplifying & Dampening Zones
Structure Integration for Testing
Sound Reflection Simulation
Sound Lines Density Calculation Counting Number of Intersections between Sound Lines and Blue Areas [for Amplifying Sound] Counting Number of Intersections between Sound Lines and Pink Areas [for Dampening Sound] 099
Number of Intersection of Sound Lines 2084 3002
2202 2948
2602 2728
100
Circulation and Activity Space
Circulation Activity Space
Platform / Square Entrance / Exit Sitting Area Amphitheater
101
Masterplan
1. Bridge 2. Amphitheater 3. Viewing Platform 4. Acoustic Corridor 5. Commercial Complex 6. Railway Noise Barrier 7. Viewing Platform 8. Open Square 9. Gallery 10. Sitting Area 11. Scenic Pathway 12. Side Entrance 13. Main Entrance 14. Bridge 15. Waterfront Terrace
106
4.3 D E T A I L O P T I M I Z A T I O N The reflective urban structure requires further optimization in order to upgrade the design results. The initial design is optimised through two main methods. The first method is optimizing the main sound reflection structure by creating a more detailed structure, which enhances the sound effect of the reflective structure. We studied the effects of different structures on sound reflection and performed acoustic studies on each structure, such as reverberation time detection. The second method is to take advantage of the natural illumination of the site.
107
The first detailed design is based on the study and effects of sound reflection. The original reflective surfaces are divided into multiple layers. Through the acoustic effect of reflective panels, scattering and diffusing effects
can
be
achieved.
Consequently,
concert hall, amphitheatre, acoustic corridor and
other
public
spaces
are
proposed,
and a variety of activities related to sonic experiences can take place. In parts of the overall structure, the layers serve as sitting spots and amphitheatres or they supplement the existing landscape, where people can walk on, while in other parts they are multiplied in order to form larger interventions such as bridges or building.
105
The second detailed design focuses on the north-eastern part of the intervention, in the exterior facade of the main commercial buildings, which is adjacent to the King’s Cross railway. In order to control the noise influence, the second detail will take advantage of diffusion. Some parts of the layers are highlighted with the help of the digital sound simulation. Following the rewriting method that was introduced in chapter 2, triangulated elements are extracted from the walls in the spots, where the sound lines originating from the station and the trains, reach them.
106
Detail Morphology Design_1 Variation of Panel Prototype
107
Detail Morphology Design_1 Panel Acoustic Study
A series of diffusing panels with regular
geometric
patterns
were
designed and tested. Diffusion is a complement to sound absorption that reduces echoes and reflections while still leaving a live sounding space. Compared to a reflective surface, which will cause most of the energy to be reflected, a diffusor will cause the sound energy to be spread in many directions. Several sound properties, such as Sound Pressure Level, Reverberation time
and
early
decay
time,
were
tested in every panel. Depending on the pattern each panel has a different
Test Parameter diffusing behaviour. We also tested variations of panels with irregular Pressure Level the ones patternsSound in order to choose with
Reverberation Time the most effective
diffusing
Early Decay Energy function.
Square Room [10*10*3 m]
Cube Array
Quadratic Residue Diffusor
3m
10
m
10
Right Triangle Periodic Corrugated Surfaces
m Original Plywood Wall
108
Detail Morphology Design_1 Panel Acoustic Study
Acoustic Energy Time Curve Percentage of energy lost: 69.6114%
Quadratic Residue Diffusor
Receivers' Reflection Path
Reflection Path
Sound Reflection Lines
Sound Pressure Level Heatmap (db)
(bounce times: 8, initial lines: 1000)
Reverberation Time (s)
8000 Hz
4000 Hz
2000 Hz
1000 Hz
500 Hz
250 Hz
125 Hz
63 Hz
109
Early Decay Time Time (s)
Detail Morphology Design_1 Panel Acoustic Study
Acoustic Energy Time Curve Percentage of energy lost: 72.8138%
Triangle Periodic Corrugated surfaces
Receivers' Reflection Path
Reflection Path
Sound Reflection Lines
Sound Pressure Level Heatmap (db)
(bounce times: 8, initial lines: 1000)
Reverberation Time (s)
Early Decay Time Time (s)
8000 Hz
4000 Hz
2000 Hz
1000 Hz
500 Hz
250 Hz
125 Hz
63 Hz
110
Detail Morphology Design_1 Panel Acoustic Study
Acoustic Energy Time Curve Percentage of energy lost: 69.6880%
Cube Array
Receivers' Reflection Path
Reflection Path
Sound Reflection Lines
Sound Pressure Level Heatmap (db)
(bounce times: 8, initial lines: 1000)
Reverberation Time (s)
8000 Hz
4000 Hz
2000 Hz
1000 Hz
500 Hz
250 Hz
125 Hz
63 Hz
111
Early Decay Time Time (s)
Detail Morphology Design_1 Panel Acoustic Study
Acoustic Energy Time Curve Percentage of energy lost: 73.7295%
Original Plywood Wall
Receivers' Reflection Path
Reflection Path
Sound Reflection Lines
Sound Pressure Level Heatmap (db)
(bounce times: 8, initial lines: 1000)
Reverberation Time (s)
Early Decay Time Time (s)
8000 Hz
4000 Hz
2000 Hz
1000 Hz
500 Hz
250 Hz
125 Hz
63 Hz
112
Detail Morphology Design_1 Panel Structure Optimization
Average Fluctuation
Multiple Fluctuation
Vertical Fluctuation
Horizontal Fluctuation
113
Detail Morphology Design_1 Panel Structure Optimization
Sunlight Duration Projected on Whole Design Mesh
Sunlight Duration on Panel Structure
Duration of Sunlight: 22nd Dec. [Avg. 0.00-24.00]
In order to take into consideration the illumination of the site, the second methodology we used is related to sunlight analysis. From the sunlight analysis, we determine the surfaces of the main building that are exposed to sunlight more than 5h per day. These surfaces are subtracted from the overall structure. This is the methodology we used to shape the interior façade of the main building. The exterior wall of the
Sunlight Duration Result on One of the Panel Structure
main building, which borders on the King’s Cross Station serves as noise barrier from the railway.
Areas which are exposed to sunlight for more than 5 hours are removed and the remaining holes are used as windows to guide sunlight through the structure.
Excavated Facade 114
115
116
Detail Morphology Design_2 Structure Generation
Sound Reflection Intersections
Facade Basis
Patterns Generation
Structure Details
Sound Reflection Simulation
Railway Noise Barrier
117
Detail Morphology Design_2 The second detail design is to apply the
Structure Application
diffusor structure and function in order to dampen sound. The northeast corner of the design structure is adjacent to the railway and it suffers from noise. The new facade serves as a noise barrier.
118
Pattern Generation Different materials are applied in various
spots
according
to
their
acoustic function. Metal, for instance, due
to
its
reflective
features,
is
used in areas where sound needs to be amplified. Wood or concrete are used in locations where sound needs to be diffused or dampened. They can reduce echoes and reflections while still preserving a vivid acoustic function. Compared to a reflective surface,
such
as
metal,
diffusors
will cause the sound energy to be radiated
in
many
directions,
and,
thus, resulting to a diffusive acoustic space, which can serve as a music hall or performance space.
119
Material Application
Number of Soundline Intersection 165
0
120
AREA 1 Sound Amplification
121
Existing Acoustic Level
Material Application
[Number of Sound Line Intersection]
[Amplification Level]
0-1
[Wood]
1-2
[Concrete]
2-6
[Metal]
6-12
12-20
20-165
[Glass]
122
AREA 2 Sound Amplification
AREA 3 Sound Dampening
AREA 4 Sound Amplification
123
Existing Acoustic Level
Material Application
[Number of Sound Line Intersection]
[Sound Amplification Level]
0-2
2-6
6-12
12-76
2-6
6-30 [Sound Dampening Level]
0-2
2-12
124
Area Perimeter
125
126
130
131
4.4 'L o - F i ' C I T Y The Lo-Fi City is an urban intervention that takes advantage of the plethora of sounds and redistributes them in the city. Acoustic signals need to be amplified in order to be heard and other sounds need to be dampened in order to reveal their pleasant features .
Currently, our urban centres are defined by road traffic and the industrial sounds associated with buildings and urban elements. However, there is the prospective to redefine the character of our cities by sculpting new soundscapes in accordance to how we wish them to sound like. We can include sound as a resource, instead of a detractor, to form soundscapes for urban spaces that resonate with people.
In this chapter, we will present our design results in detail, by demonstrating
technical
plans,
sections,
perspective
views
and
physical models of our proposal.
133
131
132
133
134
Shopping Mall University
Granary Square
Office Building
Railway
Kings Cross Station
135
136
137
138
Section A-A
Section B-B
140
Section C-C
1
Bridge
Viewing Platform
Acoustic Corridor
Commercial Complex Amphitheater
142
Section D-D
Viewing Platform
Open Square
Acoustic Corridor
Restaurants & Shops
Gallery
Railway
144
148
149
Amphitheater Acoustic Corridor
Platform Sitting Area Entrance Plaza
147
148
Commercial Complex Acoustic Corridor Gallery
149
150
154
155