Department of Defense (DoD) energy success is measured against mandated goals for energy reduction and sustainable facility management. In order to make consistent and well-informed decisions across its entire portfolio of buildings, DoD has a critical need for a consistent, scalable approach to evaluating energy consumption of existing facilities, to compare tradeoffs between energy conservation measures, and to identify facilities that are in greatest need of improvement.
In the last several years, it has become increasingly evident that existing methods of simulating and estimating energy use in buildings require highly trained engineers to spend significant time constructing energy analysis simulations. Shortcomings of past approaches included labor-intensive data inputs, the need for subject matter experts to operate the modeling systems, and the inability to model the DoD building inventory in a timely or cost-effective way. Autodesk began looking at ways to combine various data collection methods, best practices, and software tools to address this problem, and the idea of Rapid Energy Modeling (REM) was conceived.
The objective of this demonstration was to evaluate REM workflows and performance by comparing simulated to actual building energy consumption and investigate the scalability of REM workflows for DoD. This project demonstrated that the REM workflow quickly captures and utilizes information on operations, geometry, orientation, weather, and materials, generating three-dimensional (3D) Building Information Models (BIMs) guided by satellite views of building footprints and simulating energy use patterns. The project also demonstrated the application of simulated Energy Conservation Measures (ECMs) on a subset population of buildings to understand effective ways to reduce their energy consumption.
The REM technology, including the ECM capabilities, uses whole-building energy simulation algorithms driven by the Department of Energy (DOE) 2.2 engine for energy analysis. In this project, REM was applied to a sample of 23 DoD buildings across eight locations and representing seven building types. The simulated and actual building energy data was analyzed by energy type (electricity and natural gas) and energy use intensity (EUI) and further segregated by building type.
The results show that the models for offices and specialty use buildings performed better than models for barracks, where variable occupancy did not match model assumptions.
Quantitatively, a primary performance objective was to have REM electric and natural gas estimates come within < 10% of actual utility information (90% average accuracy). Aggregate results indicate average accuracy of 81.88% for predicting electric consumption with a mean absolute percentage error of 18.12%, considered to be a good forecast according to published criteria. Natural gas and combined EUI predictions were on average 58.20% accurate and 77.56% accurate respectively, considered reasonably accurate. The demonstration produced margins that, while outside the target range, were still within the range of useful forecasting values, with strong correlations in energy use curves for many buildings.
Qualitatively, the training completed indicates that the project meets the performance objectives showing that DoD participants can learn the workflow and begin creating and analyzing using REM in less than 1 day. Participants also indicate a high level of satisfaction with the REM ES-2 workflow. Preliminary results indicate that energy models can be completed in less than 3 hours after the process is learned (the performance objective was 2 days).
A significant number of considerations were uncovered that will help guide the refinement of the REM process in the future, including data gathering and model sensitivity. Additionally, the quality of the DoD building meter data was not as high as expected before the start of this project, and as a result, there may be discrepancies in comparison of simulations to the meter data.
While the REM process and reports do not mirror traditional audits, the workflow has potential benefits in that it can be implemented by DoD personnel directly. It is difficult to directly compare cost or time savings with traditional audits as there is not complete overlap in capabilities, but results indicate that REM can yield >90% savings in time and cost compared to traditional American Society of Heating, Refrigerating and Air Conditioning Engineers (ASHRAE) Level 2 auditing approaches, with the added benefits of the computer simulation characteristic of Level 3 audits. REM analysis can be completed in less than 1 hour, with up to 2 additional hours that may be required for data collection. The modeling process requires minimal training or expertise and has been taught to DoD staff in less than 1 day during this demonstration project.
The results of this study indicate that REM can meet the need of the Energy Independence Security Act (EISA) 2007 data reporting requirements as well as support government policy, including Executive Order 13423. REM provides DoD with a way to quickly establish building geometry, scale energy analysis of the existing building portfolio, visualize end-use breakdowns of energy consumption, compare tradeoffs and potential energy savings between energy conservation measures automatically, identify facilities that are in greatest need of improvement, and enhance scalability of energy evaluations and retrofits.