A review on simulation-based optimization methods applied to building performance analysis

TLDR: The review indicates that future researches should be oriented towards improving the efficiency of search techniques and approximation methods for large-scale building optimization problems; and reducing time and effort for such activities.

About: This article is published in Applied Energy.The article was published on 2014-01-01 and is currently open access. It has received 1009 citations till now. The article focuses on the topics: Engineering optimization & Simulation-based optimization.

Figures

Table 2: Classification of optimization problems – adopted and modified from [35; 14]

Figure 1: The increased trend of number of optimization studies in building science

Table 3: Classification of mostly-used algorithms applied to building performance optimization

Figure 11: Sensitive and robust optimal solutions of a single-variable function (adapted from [100])

Table 4: An overview of optimization programs applied to building performance optimization

Figure 6: Use frequency of different optimization algorithms, the result was derived from more than 200 building optimization studies given by SciVerse Scopus of Elsevier

Frequently asked questions

Q1. What are the contributions in "A review on simulation-based optimization methods applied to building performance analysis" ?

This paper provides an overview on this subject, aiming at clarifying recent advances and outlining potential challenges and obstacles in building design optimization. This paper also gives bibliographic information on the issues of simulation programs, optimization tools, efficiency of optimization methods, and trends in optimization studies. Further effort is also required to quantify the robustness in optimal solutions so as to improve building performance stability. 

Q2. What is the common reason for the sensitivity analysis of building simulation programs?

Due to the complexity of detailed building simulation programs, simulation outputs are generally nonlinear, multi-modal, discontinuous [32; 5], nonmonotonic [87] and may contain both continuous and discrete variables, global sensitivity analysis rather than local one should be used. 

Q3. What is the important task of analysts in the optimization phase?

In the optimization phase, the most important task of analysts is to monitorconvergence of the optimization and to detect errors which may occur during the whole process. 

Q4. What are the main reasons why different optimization algorithms are not trivial?

Convergence behaviors of different optimization algorithms are not trivial and are acrucial research area of computational mathematics. 

Q5. How can evolutionary algorithms be surprisingly robust to high failure rates?

By simply rejecting the solutions having a failed simulation run, evolutionary algorithms can be surprisingly robust to high failure rates (p.117 in [15]). 

Q6. How many extra function evaluations are needed to evaluate the robustness of candidate solutions?

To accurately evaluate the robustness of candidate solutions with respect to uncertainties, a significant amount of extra function evaluations is needed [104]. 

Q7. What are some of the tools and algorithms that are able to handle constraints?

For dependent variables’ constraints, most optimization algorithms force users to define constraints by using penalty or barrier functions, but some optimization tools and algorithms are able to handle constraints separately and automatically (e.g. Matlab optimization toolbox, MOBO [43], CONLIN method [85]…). 

Q8. What is the strength and weakness of various surrogate methods?

The strength and weakness of various surrogate methods is a great research field ofcomputational and statistical science and well beyond the scope of the building simulation community. 

Q9. How much additional investment could the optimization method add to the energy consumption of the building?

They found that the optimization method could further reduce up to 10% of the annual energy consumption, accompanied by an additional investment of about 0.6 million Euros. 

Q10. What are the two computer programs with user-friendly interfaces?

These two computer programs with user-friendly interfaces can be considered as fully-functional simulationoptimization tools that can be used in building design practice. 

Q11. What are some examples of hybrid algorithms?

The hybrid algorithms have been implemented in some computer programs (e.g. GenOpt, Matlab optimization toolbox…) that can be applied to building performance analysis. 

Q12. What is the common explanation for the use of building optimization tools?

Possible explanations are likely to be the textbased format of inputs and outputs which facilitates the coupling with optimization algorithms and, of course, their strong capabilities as well.