An enhanced linear regression-based building energy model (LRBEM+) for early design. Issue 2 (3rd March 2016)

• Record Type: Journal Article
• Authors: Al Gharably, Maged; DeCarolis, Joseph F.; Ranjithan, S. Ranji
• Is Part Of: Journal of building performance simulation. Volume 9:Issue 2(2016)
• Page Start: 115
• Page End: 133
• Publication Date: 2016-03-03
• Subjects: EnergyPlus -- Monte Carlo simulation -- multivariate regression -- building geometry; 690.0113
• DOI: 10.1080/19401493.2015.1004108
• Languages: English
• ISSNs: 1940-1493

Abstract

Abstract : The design community lacks simple, data-driven energy assessment tools to explore energy-efficient alternatives during the early stages of building design. A promising option is to utilize a whole building energy simulation engine (e.g. EnergyPlus) within a Monte Carlo simulation framework to develop a linear regression-based building energy model (LRBEM) that can predict idealized heating and cooling loads based on parameters relevant to early design. Previous work was limited to medium-sized US commercial office buildings with rectangular geometries. A key limitation is addressed in this paper by considering complex geometries. A reformulated model, LRBEM+, is developed and tested with a suite of building geometries that represent limiting cases. The resultant relative error between LRBEM+ and EnergyPlus is generally less than 10%. Furthermore, LRBEM+ correctly predicts the direction and magnitude of changes in heating and cooling loads in response to changes in the most influential early design parameters.