Many-Objective Optimization Design of a Public Building for Energy, Daylighting and Cost Performance Improvement

Cheng Sun,Yunsong Han,Qianqian Liu

doi:10.3390/app10072435

Abstract

The energy performance of buildings especially public buildings needs to be optimized together with environmental, social and cost performance, which can be achieved by the multiobjective optimization method. The traditional building performance simulation (BPS) based multiobjective optimization is time-consuming and inefficient. Practical projects of complex public building design usually involve many-objective optimization problems in which more than three objectives are considered. Using BPS based multiobjective optimization is not sufficient to solve this kind of design problem. This paper aims to propose an artificial neural network (ANN) based many-objective optimization design method, an architect-friendly integrated workflow has been implemented. The proposed method has been applied on a public library building in Changchun city of China to optimize its Energy Use Intensity (EUI), Spatial Daylight Autonomy (sDA), Useful Daylight Illuminance (UDI) and Building Envelope Cost (BEC). The optimization process has obtained 176 non-dominated solutions. By adopting the selected relative optimal solutions, 1.6×105–2.1×105 kWh energy can be saved per year; sDA value and UDI value can be increased by 8.1%–11.0% and 4.3%–4.7% respectively; BEC can be reduced by ¥1.2×105–2.1×105 ($1.7×104–3.0×104). The optimization time has been greatly shortened in this method and the whole process is highly efficient without manual data conversion between different platforms.

Highlights

Cities and energy play a key role in solving the environmental crisis [1], and building energy efficiency has received a widespread interest
20.6% of the country’s total energy consumption, statistical data shows that public buildings and urban residential buildings account for the two highest proportions of building energy consumption, each with a percentage of 38% [3]
artificial neural network (ANN) models for Energy Use Intensity (EUI), Spatial Daylight Autonomy (sDA) and Useful Daylight Illuminance (UDI) prediction were developed through four steps: step 1, Latin Hypercube Sampling (LHS) was used to get the samples of feasible solutions from the design space; step 2, a parametric simulation model was established based on the decision variables of the samples and with the consideration of objectives; step 3, energy and daylight simulations were performed for each sample and step 4, with the samples as the input data and the simulation results as the output data for ANN training, ANN models for EUI, sDA and UDI prediction were developed and trained

Summary

Introduction

Cities and energy play a key role in solving the environmental crisis [1], and building energy efficiency has received a widespread interest. More than 1/3 of the total global energy consumption can be attributed to the building and construction sectors [2]. 20.6% of the country’s total energy consumption, statistical data shows that public buildings and urban residential buildings account for the two highest proportions of building energy consumption, each with a percentage of 38% [3]. The energy consumption per unit floor area in public buildings is more than twice as much as that in residential buildings [4]. Energy efficiency of buildings especially public buildings needs to be optimized. To reach the best possible solutions that will ensure the maximization of the building energy efficiency and satisfy the needs of the building’s final user/occupant/owner simultaneously [5], designers

Objectives

Methods

Results

Conclusion