暖通空调
建筑围护结构
建筑能耗模拟
经验模型
能源消耗
可靠性(半导体)
可靠性工程
均方误差
组分(热力学)
计算机科学
工程类
模拟
能源性能
空调
统计
机械工程
气象学
数学
功率(物理)
物理
电气工程
量子力学
热的
热力学
作者
Yong Tae Yoon,Sungmoon Jung,Piljae Im,Mikael Salonvaara,Mahabir Bhandari,Niraj Kunwar
标识
DOI:10.1016/j.rser.2023.113889
摘要
This paper presents a critical advancement in Building Energy Modeling (BEM) through an empirical validation approach using a high-quality dataset from a multizone commercial office building in Oak Ridge, TN, USA. BEM is widely utilized in diverse construction applications, but its effectiveness relies on the accuracy of its predictions. The study focuses on empirical validation of input parameters in BEM, including building envelope data, infiltration modeling, and rooftop unit system performance curves. The validation of simulation input parameters leads to substantial improvements in the accuracy of simulation results. Notable both NMBE and cv (RMSE) values are reduced by 0.5 % for indoor air temperature and 17 % for indoor air relative humidity compared to the previous model. At the system level, both NMBE and cv (RMSE) values are reduced by 2 % for fan energy consumption and 4 % for cooling energy consumption, compared to the previous model. A literature review highlights a significant gap in empirical validation studies, which predominantly concentrate on either component-level or whole building validation. Furthermore, many studies employ simplified setups that may not faithfully represent the complexities of multizone commercial buildings. This paper distinguishes itself by emphasizing the critical importance of component-level input parameter validation. It underlines the need to validate data related to building envelope components and HVAC system performance curves, resulting in more accurate simulation outcomes. The utilization of actual multizone commercial building data enhances the study's practical relevance. In summary, this research underscores the pivotal role of input parameter validation in enhancing the accuracy and reliability of BEM.
科研通智能强力驱动
Strongly Powered by AbleSci AI