Multiobjective building design optimization using an efficient adaptive Kriging metamodel

Abstract

Multiobjective building design optimization is a challenging problem because it involves finding a set of solutions that simultaneously optimize multiple conflicting objectives. Simulations-based optimization is widely used, but it is a computationally expensive process in terms of time, as it requires a large number of evaluations of the objective functions. Metamodel-based optimization is an alternative to reduce the time-consuming simulations during the optimization process. Metamodels can approximate the building simulation model with analytical expressions. However, the accuracy of metamodels depends on the number of simulations used to train the model and the sampling strategy used to select informative samples over the design space. This study proposes an efficient sequential sampling approach to fit the metamodels toward the regions of the design space where their accuracy is higher and can improve all objectives simultaneously. To demonstrate the effectiveness of this approach, it was applied to optimize the energy and investment costs of a multi-story residential building. The optimization results were compared with those obtained using a non-dominated sorted genetic algorithm II (NSGA-II). The results of this study show that the proposed method reduces the number of building energy simulations required by up to 50% while guaranteeing accurate optimization results. Fifteen energy-efficient buildings designs were proposed, with a wide range of trade-offs between energy and investment costs. This study highlights the potential of the proposed approach to achieve faster and accurate building design optimization and allowing for a larger design space, leading to more creative and innovative solutions.

MEVO: A Metamodel-Based Evolutionary Optimizer for Building Energy Optimization

“…Lahmar et al [18] presented an efficient multi-objective optimization approach called AKSUMO, which utilizes Kriging surrogate models to optimize building designs. The surrogate model replaces time-intensive simulations with approximations, employing an adaptive sampling algorithm to enhance accuracy in the vicinity of Pareto solutions.…”

“…Surrogate models are created using simulation data, approximating the original model. There are two main categories of surrogate models: static surrogate models [9][10][11][12][13], trained using an initial set of sampling points, and iterative surrogate models [14][15][16][17][18], that gradually add samples during optimization. The accuracy of static surrogate models is limited via the initial precision of the trained model, while iterative models increase their accuracy during optimization.…”

Sustainable Road Infrastructure Decision-Making: Custom NSGA-II with Repair Operators for Multi-Objective Optimization

“…Previous research focused on applying single-objective optimization (SOO) strategies to minimize the economic cost and embodied energy in transportation infrastructure solutions [10]. Recently, there has been a shift towards employing MOO in the design of various structural typologies, from reinforced concrete buildings to wind turbine foundations [11,12]. Adopting MOO strategies marks a significant progression in harmonizing technical compliance with sustainability objectives within structural engineering [13].…”

Framework for metamodel-based design optimization considering product performance and assembly process complexity

“…9 However, different disciplines were considered as well, for instance, electrical machines, 10 thermal management systems, 11 or building design. 12 Later, this approach was extended to multi-disciplinary optimization, which considers systems that involve several disciplines and subsystems. 13,14 Consideration of the assembly process during design optimization attracts significantly less attention compared to performance-based optimization, and hardly any widely used software for assembly process analysis exists.…”