Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 94
Edited by: B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru and M.L. Romero
Paper 103

Computer-Aided Constrained Optimization of Automated Floor Plan Design

A.A. Stamos and I.E. Tzouvadakis

School of Civil Engineering, National Technical University of Athens, Greece

Full Bibliographic Reference for this paper
A.A. Stamos, I.E. Tzouvadakis, "Computer-Aided Constrained Optimization of Automated Floor Plan Design", in B.H.V. Topping, J.M. Adam, F.J. Pallarés, R. Bru, M.L. Romero, (Editors), "Proceedings of the Seventh International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 103, 2010. doi:10.4203/ccp.94.103
Keywords: optimization, floor plan design, computer-aided design, automation, simulated annealing.

The design of architectural floor plans is one of the most fundamental and creative aspects of architectural engineering. Traditionally architects use their experience to make a functional and aesthetic design. Often, there are multiple and/or contradictory constraints which lead to unavoidable compromises.

In this work a new method is presented, which automatically computes a good first approximation of the design and meets the constraints in an optimized way. The floor plan design is approached with simulated annealing, which is a stochastic optimization method suitable for large scale multi-optimization problems, especially problems where the desired global extremum is hidden among many, poorer, local extrema. It is insensitive to the initial approximation and can easily take into account diverse constraints.

The goal is to split a rectangular floor to an arbitrary number of rooms. The dimensions of each room must be larger than a threshold value and the threshold may be different for each room. Similar constraints must be met for the area of each room. If a floor plan design does not comply to the constraints, this non-compliance is translated to penalties, or energy, using suitable functions. A design with a different number of rooms than the one specified, is also translated to energy (penalties). The relative energy of the various constraints is chosen to reflect the relative importance of the constraints.

The energy (non-compliance) is minimized by generating random and appropriate changes to an original design; a large room may be split to two (unequal) smaller rooms, or two smaller rooms may be merged into a bigger one. The split line may be vertical or horizontal and its position is random. The changes are accepted or rejected according to a probabilistic criterion, which sometimes accepts changes even if they increase the energy. Gradually, the changes are becoming less and less agile, through the decrease of a control parameter which simulates the temperature. When, after some time, all generated changes are rejected, the method has converged to the optimum solution.

If the procedure is repeated, multiple plans may be produced, each one having a different geometry, but about the same overall penalty, with slightly different compromises.

The method was implemented into an open source CAD system and has been tested with various dimensions and constraints. The method gives promising results, and encourages further study.

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

go to the previous paper
go to the next paper
return to the table of contents
return to the book description