Realization of bi-level optimization of adaptive building envelope with a finite-difference model featuring short execution time and versatility

Abstract

The application of adaptive building envelope (ABE) is an important approach to energy conservation and carbon emissions reduction in buildings. In order to achieve high performance, both the design and control of an ABE system require optimization. Bi-level optimization is proposed for this type of problem. The greatest challenge to the application of bi-level optimization is the tremendous amount of computational resources required to solve the problem. This paper realizes bi-level optimization of ABE with a finite-difference (FD) model written in MATLAB. This FD model, compared to packaged building energy simulation (BES) programs like EnergyPlus, can potentially shorten the execution time by more than 84% and model various ABE systems whose modelling is not supported by packaged BES programs. The design and control of a ventilated façade is optimized using this model as an example. It is found that performance of the explicit constraint and the implicit constraint is similar, but their applicability is different. Moreover, design parameters have a great impact on the model predictive control (MPC) sequence. MPC, on the other hand, has the ability to reduce the ratio of the energy cost of the worst design to that of the best design from 7.41 to 3.63.