Kamol Keatruangkamala and Krung Sinapiromsaran

pp. 673 - 686

Abstract

Various techniques have been used to solve a challenging architectural layout design problem for more than a decade, such as an expert systems, an evolutionary algorithm, a simulated annealing and a mathematical  programming methods. This paper concentrates on the mathematical  programming technique that formulates an architectural  layout design optimization as the mixed  integer programming model using state-of-the-art optimization solver  to determine the optimal solution. All non-linear relationships among design components  are captured using the corresponding linear equalities and linear inequivalence, Due to the combinatorial nature of the MIP solution, the MIP can be solved for small problems sizes . 2-6 rooms, within a reasonable  time limit. To remedy this situation, the valid inequality of non-circular connections has been adopted that reduces the computational time significantly. Moreover, the guided constrains based on the architect's preferences of a specific room have been embraced. This helps abandon some alternative solutions and  reduce the search space considerably. The computational time and iterations gain of more than  80% is now archievable for the architectural layout design for  7- 10 rooms.

Keywords

• architectural layout design
• optimization
• mixed integer programming
• valid inequalities
• automated design tool