Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Computational Science, Engineering & Technology Series
ISSN 1759-3158
Edited by: B.H.V. Topping
Chapter 3

Evolutionary Computation and Visualisation as Decision Support Tools for Conceptual Building Design

Y. Rafiq, M. Beck, I. Packham and S. Denhan

School of Engineering, University of Plymouth, United Kingdom

Full Bibliographic Reference for this chapter
Y. Rafiq, M. Beck, I. Packham, S. Denhan, "Evolutionary Computation and Visualisation as Decision Support Tools for Conceptual Building Design", in B.H.V. Topping, (Editor), "Innovation in Civil and Structural Engineering Computing", Saxe-Coburg Publications, Stirlingshire, UK, Chapter 3, pp 49-74, 2005. doi:10.4203/csets.13.3
Keywords: evolutionary computation, visualisation, conceptual design, genetic algorithms, evolutionary strategies.

The conceptual stage of the design process is undoubtedly the most innovative stage of the design process during which most decisions that determine the future of the project are taken, and the majority of resources are committed. It is noted that with ill-defined requirements during conceptual design, the designer explores a range of possible scenarios and potential solutions to those scenarios. This activity is one of exploration, rather than search.

This paper highlights the difficulties associated with the use of computers at the conceptual stage due to the non-algorithmic nature of the activities involved at this stage. While various KBESs tried to model some of the activities of conceptual design, it is the capturing of expert knowledge and the crisp rule based characteristic of these systems which makes them very difficult for implementation at the conceptual stage of the design process.

The use of evolutionary computation (EC) techniques, in particular the genetic algorithm, on the other hand does not rely on predefined rules and the evolution process is normally steered by the fitness function. This characteristic of these algorithms make them a strong candidate for modelling conceptual design problems. The only problem with these systems is the 'block-box' nature of the process when finding an optimum solution.

The paper reviews existing literature on the application of EC techniques to building design systems, while all stages of the design will be reviewed, the focus will be on the conceptual stage.

Experience has shown that the decision process of the conceptual stage of the design is human-led which is based on human intuition, past experiences and heuristics. Therefore, any system that tries to model activities of the conceptual design must also be human-led.

This paper initially examines how interactive and visualization techniques have played a role in EC, before discussing how an interactive visualization system can enhance design capability. The paper then discusses and demonstrates that the introduction of interactive tools such as the interactive visualisation and clustering GA (IVCGA) has shown to be promising for modelling conceptual design activities because the flexibility offered by the system allows the human expert to steer the search in the desired direction. This system becomes trusted by the human expert. The IVCGA generates a diverse number of alternative solutions that could be assessed by the designer against a set of predefined requirements.

IVCGA was formulated to offer the following features:

  • The fast generation of diverse data by running a simple GA for a low number of generations to reduce the number of function calls with normal crossover and mutation rates. The diversity of solutions is maintained by applying a high mutation rate when duplicate solutions are identified.
  • An easy to use interface that allows direct manipulation of the data and views. Various high dimensional visualization techniques are supplied to enable understanding of the data from different viewpoints and combination of parameters.
  • An automatic clustering procedure is built into the system that identifies clusters relevant to the problem in hand. Colour is used to highlight important clusters, enhancing perceptual understanding of the data.
  • Extensive interaction is supported allowing the generation of further data by the GA, inside or outside regions identified by the user or clustering algorithm. The definition of clusters can be modified by the user or even created manually, ensuring complete freedom of search and human-led exploration of the search space.

A major outcome of the application of the IVCGA in engineering problems was its power of knowledge discovery, which is a by product of this visualisation system. By conducting a concentrated and focused search in the regions of high quality designs, new solutions were discovered that would not have been possible by other means.

purchase the full-text of this chapter (price £20)

go to the previous chapter
go to the next chapter
return to the table of contents
return to the book description
purchase this book (price £80 +P&P)