Professional Documents
Culture Documents
net/publication/291011936
CITATIONS READS
8 998
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Key techniques of high speed railway noise and vibration control View project
All content following this page was uploaded by Shuai Lu on 10 February 2017.
Research Article
Shuai Lu1 (), Xiang Yan1,2, Weiguo Xu1,2, Yuxiao Chen1, Jie Liu1
Abstract Keywords
Auditoriums require both acoustic performance and aesthetic values, while the efficiency of their auditorium design,
current design process should be improved for architects to better realize these two goals. This room acoustics,
paper first proposes a new design process for auditoriums by integrating parametric models and parametric models,
acoustic simulation, aiming to provide architects with rapid architectural and acoustic feedback. acoustic simulation,
The new process is then implemented by: (1) programming a parametric model of auditoriums in rapid feedback
Rhinoceros using Component-Based Method, which can generate more varied designs and better
suit architects’ design practice than existing methods; (2) developing an interface connecting
Article History
Received: 15 September 2015
Rhinoceros and CATT, which adds acoustic simulation functionality on its input, manipulation and
Revised: 23 November 2015
output to facilitate architects’ utilization in auditorium designs. Finally, a demonstration is
Accepted: 25 November 2015
conducted to verify the effectiveness of the new design process as well as the functionality of
rapid feedback in its current implementation.
© Tsinghua University Press and
Springer-Verlag Berlin Heidelberg
2015
can generate could be more varied and reliable, and the sightlines, circulations, etc. Some existing attempts have
simulation tools could be more friendly to architects who already shown the difficulties. For example, a parametric
are unfamiliar with acoustics (more details in Sections 1.2 model of vineyard concert hall was built by integrating
and 1.3). This paper explores the possibility and imple- polyhedrons of different shapes as shown in Fig. 1 (Spaeth
mentation of a new design process of auditoriums, which and Menges 2010), but a large proportion of designs
integrates two digital methods, parametric models and generated by it is invalid, because this parametric model
acoustic simulation. This process aims to provide rapid only covers shape characteristics of auditoriums and omits
feedback on architectural design and acoustic performance others. Another example is a parametric model of horse-
for architects, thereby helping them pursue better auditorium shoe opera house, which was created by defining the topology
designs. of the building as shown in Fig. 2 (Xu 2011a), and the
problem of this parametric model is that it can only generate
1.2 Parametric model very similar designs, because the topology is completely fixed
and only simple modifications like changing dimensions
The parametric model is a widely used digital design are possible.
method and its idea is to model buildings by defining their
essential characters and rules using computer language, 1.3 Acoustic simulation
instead of describing the information of actual geometries
or components (Janssen 2004; Xu 2011b). When architects The acoustic simulation itself is a mature technology with
change parameters or input modification commands, several commercial software packages available that could
parametric models can automatically provide them with a predict acoustic parameters of auditoriums with deviations
family of designs that vary significantly while all meet the smaller than twice of the subjective limens (Vorländer 1995;
basic requirements in a short time. As a result, architects’ time Bork 2005a,b). However, existing commercial software as
and efforts spent in accomplishing or modifying designs well as simplified algorithms are designed for acousticians
can be largely reduced, thus they can attempt more design rather than for acoustically unschooled architects, because:
possibilities and increase the chance of acquiring satisfactory (1) they need very detailed information about forms and
designs. materials, while architects can only provide form information,
Although parametric models have reasonable theories and and even form information needs format conversions;
successful application examples on other types of buildings, (2) the manipulation process is complicated itself, while it
to build a sufficient parametric model of auditoriums is needs to be redone at every design modification; (3) simulation
very difficult. This is because that auditoriums are complex results contain large amounts of data that shows the values
systems that have many components correlated with each of different acoustic parameters, which makes it difficult for
other and are constrained by functional requirements of architects to understand the overall acoustic performance
2.2 Proposed design process of auditoriums In this new design process, architects can be more
concentrated on the creative part of designs and have more
Parametric models and acoustic simulation could be capable time to test novel ideas, thus outstanding designs are more
to improve these two deficiencies of the current design likely to be achieved. For acousticians, it needs to be
process of auditoriums. Parametric models could provide emphasized that their importance is not reduced at all.
architectural feedback, which means that once architects Now they are released from constantly testing different
have a design idea, the design they want can be automatically designs superficially, and can focus on pre-selected ones by
generated by a computer very quickly, thus the time spent computers to make more detailed examinations and provide
on accomplishing or modifying designs is reduced and more specific suggestions.
architects can have more attempts following continuous Although the new design process is promising, there
design thinking. Acoustic simulation with proper supplements are critical difficulties in implementing it, because to build
for architects could provide acoustic feedback, which means a sufficient parametric model of auditoriums is very hard
that once architects are satisfied by the design, they can and acoustic simulation software has to be adjusted and
employ acoustic simulation for a primary acoustic judgment improved properly to suit architects’ demands (as indicated
and then they can continue to design according to the in Sections 1.2, 1.3). Here we try to overcome these difficulties
judgment instantly. and implement the new design process by developing a
Taking parametric models and acoustic simulation into software package, which includes a parametric model of
account, theoretically the design process of auditoriums concert halls based on Rhinoceros, and an acoustic simulation
could be optimized to a three-party activity: architects, interface for architects bridging Rhinoceros and CATT.
acousticians, and computers (Fig. 5). For most of the time
architects can work with the computer. They can convey 3 Implementation
their design ideas by drawing drafts, changing parameters or
inputting modification commands, and then the parametric 3.1 Parametric model
model can synthesize this information and provide the
design they want very quickly. Then the architects just need 1. Methodology
to evaluate whether the design provided by the computer
is satisfactory or not, and proceed to send the design to As indicated in Section 1.2, researchers have attempted to
simulation or modify their design idea accordingly. The build parametric models of auditoriums by integrating unit
simulation part is employed to test the design acoustically. elements (Spaeth and Menges 2010) or describing the
Even though not perfectly precise, its quick judgment can topology (Xu 2011a); however, these methods could lead to
help architects make more reasonable design decisions: deficiencies including generating a large number of invalid
when the acoustic judgment is undesirable, architects can designs or restraining design possibilities. Herein a new
immediately start to rethink their design idea; otherwise methodology to build parametric model is raised, which
the acousticians should take over the work and verify the is called Component-Based Method. A significant feature
acoustic quality of the design more carefully, because there of auditoriums is that they have several fixed types of
might be some issues that cannot be fully recognized by components (e.g. wall, stage, aisle, ceiling) and the types of
architects. If the acousticians think there is still some potential components will not change for most design possibilities
to improve the acoustic performance, they can provide of a certain type of auditoriums. Moreover, there are
architects with suggestions for further design modifications. definitive inter-connections and constrains between every
two components in one type or in different types (e.g. every
two adjacent walls need to be continuous; every door needs
to locate on a wall and at the end of an aisle, every seat
needs to have proper altitude according to the position of
the stage and other seats to ensure good sightline), and these
inter-connections and constrains could be used to define the
essential characters and rules of a certain type of auditoriums.
Therefore, we might be able to build a parametric model of
auditoriums by defining the types of components and the
inter-connections and constrains between the components.
This is the Component-Based Method.
Fig. 5 Proposed design process involving parametric model and The Component-Based Method can be seen as an extension
acoustic simulation of the Object-Oriented Method in programming (Booch and
Lu et al. / Building Simulation / Vol. 9, No. 3 239
Maksimchuk 2007; Ambler 2001). The concept Component walls of it should rotate and scale in order to keep the
in Component-Based Method corresponds to the concept walls’ continuity; doors on it should move in order
Object in Object-Oriented Method, Type of Components to remain on the wall, etc. By defining these inter-
corresponds to Class, while Inter-Connection and Constrain connections and constrains that include all possible
corresponds to Attribute, Method and Event. dependencies of two components, the essential characters
and rules of shoe-box concert halls are sketched out.
2. Programming
c) Defining Initialization, which is how the initial design is
Following this method, a parametric model of shoe-box generated. Architects can draw their drafts in Rhinoceros,
concert halls (the reference auditorium is the Boston and a corresponding design of shoe-box concert hall can
Symphony Hall) has been programmed based on Rhinoceros be generated automatically as the initial design for further
(V.5 SR2) with its development kit, Rhino Python Editor, modifications.
following several steps (Fig. 6): d) Defining Response Mechanism. When one component is
a) Defining Types of Components. Here we selected six modified by architects, other components have to react
types of components, which are wall, stage, door, aisle, not only following the preset inter-connections and
balcony and seat. Each type has its own attributes, methods constrains in step (b), but also in proper orders, so that a
and events. Ceiling is not included here because for new design can be generated without any error. Therefore
shoe-box concert halls it can be determined by walls. we need to define a Response Mechanism to control this
b) Defining Inter-Connections and Constrains, which here process. Here a “chain reaction” prototype is employed
refer to the dependencies of different components, or how (Fig. 7). The component modified by architects is the
other components should react when one component is stimulator of the chain reaction (Wall 1 in Fig. 7), and it
modified. For example, if one wall is moved, adjacent will inform its related components to change (Walls
2&4, Stage 1, Doors 1&2, Balcony 1, Chair 1 in Fig. 7). After 3. Manipulation process and user interface
these components finish changing, they will also inform
Besides the core part of the parametric model, we also design
their related components to change. As a result, this “chain
the manipulation process and the user interface carefully to
reaction” process will continue until no more components
allow architects to communicate with computers easily,
need to change, and this is the time when a new design
because this is very important to realize the rapid feedback
is generated (more details available in Appendix A).
loop. The current manipulation process and interface are
e) Defining User Interface, which will be introduced later in
designed as follows (Fig. 9):
this section.
a) Input drafts. Drafts are the start of designs and express
After these programming processes, a parametric model of
architects original ideas. Here architects can draw the
shoe-box concert halls is achieved, and it has the following
draft of an auditorium in their own way using commands
features: in Rhinoceros (including Line, Polyline, Curve, etc.) and
a) This parametric model can generate a large variety of the only requirement is that they put the drafts into
designs, which are all valid and meet the basic principles right layers (as different colors indicated in Fig. 9(a)).
of auditoriums, thus architects’ design ideas can be Naturally, it could be better for architects if they are able
turned into designs simply by drawing different drafts to draw the drafts in a more intuitive way (e.g., drawing
or modifying different components. Specifically, any on a sketch broad with pencils and then the sketch being
concert hall that has vertical walls and no curvy com- recognized by computer), but this could affect the
ponents (standard shoe-box, wedged-shape, center-stage, robustness of draft recognition, so is not included here
etc.) can be generated by the parametric model (Fig. 8, yet (Wei et al. 2012).
the ceiling is hidden for visual clarity), and this gives b) Drafts recognition. Once architects start the program we
designers much more design freedom and possibilities developed, the draft drawn by them will be recognized
than the existing ones. automatically and be converted into the information
b) Every component of this parametric model can be modified that the parametric model needs to generate a design.
directly, which means that architects can modify the c) Choosing type. Architects can input information to the
design by rotating a wall, moving a door or editing stage computer on the required type of auditoriums by simply
shape, while in the existing parametric models only click one button. At the current stage only shoe-box
changing parameters (length, width, height, etc.) and concert hall is available (Fig. 9(b)).
editing control lines are available. Therefore, the new d) Input parameters. There are several architectural parameters
parametric model is more compatible with architects’ that cannot be expressed by drafts, thus architects need
design custom and can be used more efficiently. to input them in a window (Fig. 9(c)).
Lu et al. / Building Simulation / Vol. 9, No. 3 241
Fig. 8 Designs that can be generated by the parametric model (star indicates changed component)
understanding of the acoustic performance of the design. recommended ranges for architects according to the
Therefore, the data needs to be simplified and visualized in requirements of a certain project.
order to be more helpful for architects to make design Finally, the simplification and visualization of results. In
decisions. order to help architects interpret the results better, a radar
First, the number of acoustic parameters needs to be chart and a distribution map are provided for each design
reduced for more convenient design decisions. The retained (Fig. 15). The radar chart shows the mean values (red line)
parameters need to be able to reflect the overall acoustic
performance of designs, while parameters that are dependent
on others or are not related to form designs should be
omitted. Note that different kinds of auditoriums can have
divergent selections of acoustic parameters because the
acoustic requirements are different. Here we take concert hall
as an example in order to keep consistent with the parametric
model we developed in Section 3.1. Beranek (2004) has
summarized the parameters that are important for concert
halls, including Reverberation Time (RT), Early Decay Time
(EDT), Binaural Quality Index (BQI), Strength of the Sound
(G), Bass Ratio (BR), Initial-Time-Delay Gap (ITDG), Lateral
Fraction (LF), Surface Diffusivity Index (SDI) and Clarity
(C80). Among these parameters, BQI and LF are highly
correlated with each other while BQI is more stable for
simulation (Bork 2005a,b), thus only BQI is included here.
Fig. 14 Modification interface of recommended ranges
BR and SDI are entirely determined by detailed acoustic
designs by acousticians and have little relationship with
form designs by architects, which are thus excluded here.
As a result, we reduce the number of acoustic parameters
to six for concert halls: RT, EDT, BQI, G, ITDG, C80, and
they respectively represent the fullness of tone, the sequence
of early sound, spaciousness, loudness, intimacy and clarity.
Therefore, they can reflect the overall acoustic performance
of concert hall designs effectively.
Second, the recommended ranges for each acoustic
parameter need to be determined to provide architects
with reliable references for judgments. Here we use the
recommended ranges suggested by Beranek (2004) for
typical symphonic concert halls with more than 1400 seats
(Fig. 13). Considering that even auditoriums of the same
type could have different acoustic requirements due to
differences in performances, sizes, client preferences, etc.,
modifying the recommended ranges is also possible in this
interface (Fig. 14), thus acousticians can set the specific
and standard deviations (pink shadow) of all test points in qualities of a particular design and then make design decisions
two simulations for every acoustic parameter, respectively accordingly.
representing the average and uniformity of the acoustic
performance of a design. It also shows the acoustic parameters 4 Verification
of the acoustically “worst point” (yellow line), in order to
warn architects about potential extremely poor conditions. The new design process of auditoriums raised in Section 2
The green shadow in the radar chart shows the recommended has been implemented in Section 3, and this part mainly
ranges as a reference for architects to judge the acoustic aims to verify their effectiveness, especially the functionality
quality of the current design. The distribution map, on the of rapid feedback. Because the software we developed is not
other hand, shows the acoustic qualities of different positions mature enough to be used by architects in actual projects,
within the auditorium by colors of balls. Greener color instead we chose a typical shoe-box auditorium (imitating
indicates better acoustic qualities of a certain position, while the Boston Symphony Hall) as a demonstration and attempted
redder color indicates worse acoustic qualities. The position some design modifications on it in order to test: (1) whether
of the “worst point” is also shown in the distribution map the software can perform as expected; (2) how long does the
by the biggest ball (more details available in Appendix B.4). software need to provide architectural and acoustic feedback.
By the help of the radar chart and the distribution map, The demonstrating results are shown in Fig. 16. The
architects can have a general understanding of the acoustic results are drawn from a common computer (CPU: i7-3720;
Memory: 8 GB; Graphic: GT650; System: Win7-64ibt) and test in the manual of CATT (Dalenbäck 2012). In other
could change in other situations. In all of these six cases, words, reducing the delays of acoustic feedback by using
the software performs as expected with no bugs, thus its fewer rays seems to be not very promising. Note that this
reliability can be proved primarily. The calculation time of conclusion is valid only when using similar computers,
parametric model (the time delay of architectural feedback) while the proper number of rays as well as the delays of
is approximately 10 s, which is very short and means that acoustic feedback could change when different hardware is
architects can see the design result almost immediately after employed.
they have a design idea. The calculation time of acoustic To simplify geometries is another possible way to reduce
simulation is much longer, and makes the delay of acoustic the delays of acoustic feedback, and it has been widely used
feedback to be about 12–15 minutes. Although it is already by acousticians to make quick acoustic tests. Although it is
much better than the current design workflow, in which not difficult for experienced acousticians to do geometry
negotiations of architects and acousticians usually take several reduction properly, to automate this process is of technical
days, the delays of acoustic feedback could be further reduced difficulties, because the simplified geometric models have
to enhance its functionality of rapid feedback. to retain acoustically important information. As a result,
simplifying geometries is not yet included in this research,
5 Discussion but will be explored in the future based on some existing
successful researches (Siltanen et al. 2008; Andujar and
5.1 Methods to reduce the delays of acoustic feedback Brunet 1999; Andujar et al. 2002).
As indicated in Section 4, the delays of acoustic feedback 5.2 Limitation of the current acoustic evaluations
are not short enough and should be further reduced. Two
possible ways are discussed here: reducing number of rays The current acoustic evaluation system only includes objective
and simplifying geometries. parameters that are widely used and have well-recognized
For simulation software using geometrical acoustic recommended ranges, because these quantitative parameters
methods, reducing the number of rays can make the can be easily judged by architects (e.g., by comparing the
simulation faster, but too few rays could affect the precision parameters of designs and the recommended ranges) and
of simulation results (Rindel 2000). In this research, where are not likely to be misleading. These objective parameters
both precision and speed are treasured, the proper number are effective to reflect major aspects of the acoustic per-
of rays should be a balance of these two factors. formance of concert hall designs; however, it has to be
In order to explore this issue, 5 cases with the same admitted that there are still some important acoustic features
scenario (as same as the first case in Section 4) but different which cannot be captured by any of the existing room
numbers of rays are conducted for comparison. The simulation acoustic parameters. For example, the temporal envelope of
results are shown in Fig. 17. In terms of precision, it can be early reflections and the distribution uniformity of sound
seen that the mean values and standard deviations of Cases energy are important in concert halls (Lokki et al. 2011), but
1–4 are very similar, while Case 5 shows larger standard there are no quantitative parameters and recommended
deviations. The acoustic spatial distributions of Cases 1–3 ranges that have been proved to be able to reflect these
are similar and all indicate that the back parts of the main characteristics sufficiently. If these characteristics are included
audience areas are the relatively poor parts, while Case 4 in the acoustic evaluation system, they could be difficult for
and Case 5 are not consistent with Cases 1–3 and show that architects to make judgments, or could mislead architects
the side parts of the first balconies are the poor parts. The to abandon acoustically acceptable designs, thus they are
acoustic parameters and positions of the “worst point” are decided to be excluded at the current stage. Although this
the most unstable results and are different in every case, could lead to the possibilities that designs with acoustic
while Cases 1–3 show similar positions (in the back parts of defects might be accepted by architects, these unsatisfactory
the main audience areas). In terms of time, it can be seen designs can be recognized by acousticians, since every
that the time reduces significantly with the number of rays design approved by architects has to be further verified by
in Cases 1–3, but for Case 4 and Case 5, the time reduction acousticians as indicated in Section 2.2. Therefore, this
is very limited. Considering both precision and speed, situation is considered to be better than misleading architects
10 000 rays (Case 3) can be regarded as the best choice to abandon promising designs.
based on these 5 cases, while further reducing the number There have been some researches trying to quantify these
of rays can affect the precision and is not effective to characteristics that cannot be captured by existing room
shorten the time delays. This could also be part of the acoustic parameters, for example, using the autocorrelation
reasons that using 10 000 rays is recommended for a quick function to represent the distribution uniformity of sound
246 Lu et al. / Building Simulation / Vol. 9, No. 3
Fig. 17 Simulation results using different numbers of rays (*the time between CATT is launched and closed in the first simulation; † the
time between CATT is launched in the first simulation and the result graphs are shown)
energy (Robinson et al. 2014). The acoustic evaluation system 5.3 Role of computers in the proposed design process
here could be more comprehensive if these promising
methods are included. However, these attempts need to be In the proposed design process of this research, the role of
considered very carefully to avoid potential misleading. computers is generally regarded as design support medium,
Even if new acoustic indices are successfully added in the which means that computers are employed to eliminate
evaluation system, verifications of acousticians are still previously human activities with less cost and higher quality,
indispensable, because room acoustics is still be regarded as and assist designers in areas where they do not have sufficient
a mixture of art and science at the current stage (Robinson knowledge or competence themselves (Janssen 2004). However,
et al. 2014). Design ideas and major design decisions are still determined
Lu et al. / Building Simulation / Vol. 9, No. 3 247
Apfel RE (1992). Deaf Architects & Blind Acousticians: Challenges to Mahalingam G (1999). A new algorithm for the simulation of sound
Sound Design. Madison, CT, USA: Printworks. propagation in spatial enclosures. In: Proceedings of 6th IBPSA
Bassuet A, Rife D, Dellatorre L (2014). Computational and optimization International Conference (BS1999), Kyoto, Japan.
design in geometric acoustics. Building Acoustics, 21: 75–86. Mahalingam G (2003). Representing architectural design using a
Beranek L (1979). Acoustics of concert halls. Acta Acustica, 1979(4): connections based paradigm, In: Proceedings of Conference of
251–259. (in Chinese) the Association for Computer Aided Design in Architecture
Beranek L (2004). Concert Halls and Opera Houses. New York: Springer. (ACADIA 2003), Indianapolis, USA, pp. 270–277.
Beranek L (2011). Concert Hall Acoustics. Architectural Science Review, Microsoft (2015a). SendKeys Class Help. Available at
54: 5–14. https://msdn.microsoft.com/us-en/library/system.windows.forms.
Booch G, Maksimchuk RA (2007). Object-oriented analysis and design sendkeys.aspx. Accessed 15 July 2015.
with applications, 3rd edn. Boston: Addison-Wesley. Microsoft (2015b). System. Diagnostics Namespaces Help. Available at
Bork I (2005a). Report on the 3rd round robin on room acoustical https://msdn.microsoft.com/us-en/library/gg145030(v=vs.110).
computer simulation—Part I: Measurements. Acta Acustica united aspx. Accessed 15 July 2015.
with Acustica, 91: 740–752. Monks M, Oh B, Dorsey J (2000). Audioptimization: Goal-based acoustic
Bork I (2005b). Report on the 3rd round robin on room acoustical design. IEEE Computer Graphics and Applications, 20(3): 76–90.
computer simulation—Part II: Calculations. Acta Acustica united Pehle M, Marek T (2013). SU2CATT Help. Available at http:// www.
with Acustica, 91: 753–763. rahe-kraft.de/rk/en/software/su2catt. Accessed 15 July 2015.
Burry M (2011). Scripting Culture: Architectural Design and Programming. Peters B (2010). Acoustic performance as a design driver: Sound
Hoboken, NJ, USA: John Wiley & Sons. simulation and parametric modeling using smart geometry.
Cheng Y (2013). Contemporary Buildings for Performing Arts in International Journal of Architectural Computing, 8: 337–358.
Multiple Views. Beijing: China Architecture & Building Press. (in Rindel JH (2000). The use of computer modeling in room acoustics.
Chinese) Journal of Vibroengineering, 3(4): 220–224.
Dalenbäck BI (2012). Introduction Manual of CATT-Acoustic v9.0. Robinson P, Siltanen S, Lokki T, Savioja L (2014). Concert hall geometry
Available at http://www.catt.se/CATT-Acoustic.htm. Accessed optimization with parametric modeling tools and wave-based
15 July 2015. acoustic simulations. Building Acoustics, 21: 55–64.
Dritsas D, Rafailaki E (2007). A computational framework for theater Saksela K, Botts J, Savioja L (2015). Optimization of absorption placement
design. In: Proceedings of 3rd International Conference of the Arab using geometrical acoustic models and least squares. Journal of
Society for Computer Aided Architectural Design (ASCAAD the Acoustical Society of America, 137: 274–280.
2007), Alexandria, Egypt, pp. 165–182. Siltanen S, Lokki T, Savioja L, Lynge C (2008). Geometry reduction in
Foged IW, Pasold A, Jensen MB, Poulsen ES (2012). Acoustic room acoustics modeling. Acta Acustica united with Acustica, 94:
environments: Applying evolutionary algorithms for sound based 410–418.
morphogenesis. In: Proceedings of 30th International Conference Spaeth B, Menges A (2010). Performative design for spatial acoustic:
on Education and Research in Computer Aided Architectural Concept for an evolutionary design algorithm based on acoustics
Design in Europe (eCAADe 2012), Prague, Czech Republic, pp. as design driver, In: Proceedings of 29th International Conference
347–353. on Education and Research in Computer Aided Architectural
Janssen P (2004). Design method and computing architecture for Design in Europe (eCAADe 2010), Ljubljana, Slovenia, pp.
generating and evolving building designs. PhD Thesis, Hong Kong 461–468.
Polytech University, Hong Kong, China. Szalapaj P (2005). Contemporary Architecture and the Digital Design
Kouzeleas S, Semidor C (2002). Architectural translation of the acoustical Process. London: Architectural Press.
results. In: Proceedings of International Congress of the Institute Vorländer M (1995). International round robin on room acoustical
of Acoustics 2002, London. computer simulations. In: Proceedings of 15th International
Lawson B (2005). How Designers Think: The Design Process , 4th edn. Congress on Acoustics, Trondheim, Norway, pp. 689–692.
London: Routledge. Wei L, Zhang Q, Xu Z, Zhang X (2012). Coming out of the
Lokki T, Pätynen J, Tervo S, Siltanen S, Savioja L (2011). Engaging concert misunderstandings of digital design technologies. Architectural
hall acoustics is made up of temporal envelope preserving reflections. Journal, 2012(9): 1–6. (in Chinese)
Journal of the Acoustical Society of America, 129: EL223–EL228. Xu S (2011a). A design study of auditoriums based on CAAD. Master
Mahalingam G (1998). The algorithmic auditorium: A computational Thesis, Tsinghua University, China. (in Chinese)
model for auditorium design. In: Proceedings of 3rd Conference Xu W (2011b). Parametric design and form generating with algorithms.
on Computer Aided Architectural Design Research in Asia World Architecture, 2011(6): 110–111. (in Chinese)
(CAADRIA 1998), Osaka, Japan, pp. 143–152.
Lu et al. / Building Simulation / Vol. 9, No. 3 249
In order to explain the technical details of the parametric keep the walls’ continuity; (2) if one wall is changed, doors
model (especially its response mechanism) more clearly, a on it will move in order to remain on the wall; (3) if one
simplified example is provided here, which contains only door is changed (moved), related aisles (which means that
three types of components: four walls, two doors and an the door is at one endpoint of the aisle) will move the
aisle (Fig. A1(a)). related endpoint to keep it occupied by the door; (4) if one
Its Inter-Connections and Constrains contain: (1) if one aisle is changed (endpoint moved), related doors (which
wall is changed (including moved, rotated, scaled, split, and means that the door is at one endpoint of the aisle) will
deleted), adjacent walls will rotate and scale in order to move to keep being at the endpoint of the aisle.
Now if architects move Wall 1 (Fig. A1(b)), the response other words, the length of their translation vectors is 0). In
mechanism will be stimulated as Fig. A2(a). Wall 2, Wall 4 order to avoid infinite loops, these invalid changes are
and Door 1 will be informed to change according to the detected so that they will not cause other components to
preset Inter-Connections and Constrains (as Fig. A2(b)) change again (as they are shown in grey in Fig. A2(c)).
and the design will become Fig. A1(c). The change of Aisle will again make the chain reaction
The changes of Wall 2, Wall 4 and Door 1 will make the continue as Fig. A2(d). However, there is no valid change
chain reaction continue as Fig. A2(c) and the design will now, which means that the chain reaction has finished and
become Fig. A1(d). However, only the change of Aisle is the new design has been generated (as Fig. A1(d)).
valid, while other changes don’t actually happen (or in
250 Lu et al. / Building Simulation / Vol. 9, No. 3
Front wall Side walls Rear wall Ceiling Audience floor Stage floor Balcony fronts
Boston Symphony Hall Plaster Plaster Plaster Plaster Audience Wood for stagefloor Plaster
Vienna Grosser Musikvereinssaal Wood for walls Plaster Plaster Plaster Audience Wood for stagefloor Plaster
Amsterdam Concertgebouw Plaster Plaster Plaster Plaster Audience Wood for stagefloor Plaster
Dallas Meyerson Symphony
Wood for walls Wood for walls Wood for walls Wood for ceiling Audience Wood for stagefloor Plaster
Center
National Concert Hall of China Gypsum Gypsum Gypsum Gypsum Audience Wood for stagefloor Gypsum
Porto Case Da Musica Fabric Wood for walls Fabric Wood for ceiling Audience Wood for stagefloor —
Two software packages are employed here to realize the automatically by scripting. In this way we can manipulate
manipulation of CATT in the background: Windows. Forms. CATT to run an acoustic simulation without any manual
Sendkeys package of .NET Framework 4.5 (Microsoft 2015a) operations. Once a simulation is finished, the CPU utilization
and System. Diagnostics. Process package of .NET Framework rate of CATT will decrease. System. Diagnostics. Process
4.5 (Microsoft 2015b). Because CATT is a capsulated software package of. NET Framework 4.5 is employed to detect this,
and cannot be invoked inside, Windows. Forms. Sendkeys thus the end of a simulation can be recognized, and then
package of .NET Framework 4.5 is employed here as a virtual further operations including exporting simulation results
keyboard, which can simulate keyboard manipulations and post-processing can start after a simulation ends.
The color of a certain point is determined by the number green and red will be used in other situations.
of acoustic parameters of that certain point within the The worst point here refers to the point with the least
recommended ranges. For example, if all six parameters of parameters within the recommended ranges. If more than
a point satisfy the recommended ranges, the color of that one point has the same number of unsatisfactory parameters,
point will be pure green (RGB: 0, 255, 0), while if no the one that deviates more from the recommended ranges
parameters satisfy the recommended ranges, the color will will be picked as the worst point.
be pure red (RGB: 255, 0, 0). A gradient color between