COMPARISON OF ENERGYPLUS AND DOE-2 DETAILED WINDOW HEAT TRANSFER MODELS

Two commonly used building energy simulation engines, EnergyPlus and DOE-2, can differ significantly in their calculations of window heating load. In this paper, the authors identify the issues in the window heat transfer algorithms used in detailed window models, and provide suggested changes to make the calculations more accurate. We estimate that up to 82% of the observed differences can be resolved through the suggested changes. The methodology employed for this investigation involved programming the algorithms from the respective simulation engines in the Engineering Equation Solver (EES) so that the impact of each proposed change could be investigated and quantified independently. INTRODUCTION Window heat transfer represents a significant portion of the overall heating load in buildings. This is especially true for the tens of millions of older homes with single pane windows. When assessing these buildings for energy savings potential through retrofits, it is important to be able to accurately predict the heat transfer through the windows. For single pane windows, the predicted heat transfer is more sensitive to the convection and radiation boundary conditions than it is for multiple-pane, less-conductive window types. Two commonly used building energy simulation engines, EnergyPlus (version 6.0.0.023) (U.S. Department of Energy, 2010) and DOE-2 (version 2.2-47h2) (James J. Hirsch & Associates, 2010), both offer a number of ways to model window heat transfer. A comparison of each window model is presented in Table 1. The inputs for many of the more detailed models are not often available in most applications of energy modeling. Therefore, it is desirable to have a model – such as the EnergyPlus detailed model with simple inputs – that can provide a detailed level of analysis given a limited, but readily-available set of inputs [e.g., the information provided on an NRFC (National Fenestration Rating Council) energy performance label]. Although the EnergyPlus detailed model with simple inputs is not explicitly available in DOE-2, it is possible to use the same methodology to create a near-equivalent model using the WINDOW software (LBNL, 2012) input method for the DOE-2 detailed model. This simple input methodology is thoroughly described in Arasteh et al. (2009). Kruis et al. (2012) compared results between EnergyPlus and DOE-2 [each using the detailed model with the simple input methodology described by Arasteh et al. (2009)] and found significant differences in the calculation of the window boundary conditions (i.e., exterior convection, interior convection, and interior radiation) causing up to 41% difference in the predicted net window heating load (including impacts of solar gains). The differences found in the boundary condition algorithms are described and quantified in this paper. INTERIOR CONVECTION Interior convection is modeled as natural convection for both EnergyPlus and DOE-2. EnergyPlus EnergyPlus uses the ISO (2003) correlation to calculate the interior convection coefficient: hc,int = Nu ·λ H (1)