An optimum design of low-cost housing offers low-income urban inhabitants great opportunities to obtain a shelter at an affordable price and acceptable indoor thermal conditions. In this paper, the design and operation of a low-cost dwelling were numerically optimized using a simulation-based approach. Three multi-objective cost functions including construction cost, thermal comfort performance and 50-year operating cost were applied for naturally ventilated and air-conditioned buildings. Thermal environment inside the house was controlled and assessed by two thermal comfort models. Optimization problems which consist of 18 design parameters and 6 ventilation strategies were examined by two population-based probabilistic optimization algorithms (particle swarm optimization and hybrid algorithm). Optimum designs corresponding to each objective function, differences in optimal solutions, energy saving by the adaptive comfort approach and optimization effectiveness were outlined. The optimization method used in this paper shows a considerable potential of comfort improvement, energy saving and operating cost reduction.

The applications of simulation-based optimization have been considered since the year 80s and 90s based on the rapid growth of computational science and mathematical optimization methods. However, most research in building engineering which combined a building energy simulation tool with an optimization ‘engine’ have been published in the late 2000s although the first efforts were found much earlier. A pioneering study in the optimization of building engineering systems was presented by J.A. Wright in 1986 when he applied the direct search method in optimizing HVAC systems (Wright 1986). Genetic algorithms were then introduced and applied in the optimization of building envelopes, HVAC systems, and controls (Wright 1994; Wright et al. 2002). In 2001, Wetter (Wetter 2001) first introduced the optimization program GenOpt with different optimization algorithms that significantly contributed to optimization solutions in building engineering. GenOpt was originally targeted to the building performance simulation (BPS) community hence it offers architects and engineers many advantages in their simulation work. Another optimization toolkit that has similar optimization capabilities to GenOpt is Dakota (Adams et al. 2009). Dakota provides a framework for single, multi-objective, or surrogate-based optimization, parameter estimation, uncertainty quantification, and sensitivity analysis to the simulation-based community, but its usage requires advanced programming knowledge. Some other optimization programs, e.g. BE opt, Top Light, MATLAB, Go SUM, and LION solver has also been developed, providing many more appropriate methodological frameworks to the simulation-based optimization community. Consequently, numerous optimization researches have been carried out, aiming to optimize building designs, passive strategies, energy consumption, HVAC controls, construction costs, life cycle costs, environmental impacts… Nevertheless, optimization research related to low-cost housing (LCH), which are actually essential in most developing countries, has rarely been mentioned.