Building Energy Modeling

For decades, researchers and practitioners have developed ever more sophisticated simulations of the behavior of buildings. They allocated significant effort in establishing a variety of highly detailed and specialized modules for a range of different materials, systems, heat transfer phenomena, electrical equipment, and occupant behavior. As a result, simulation programs such as EnergyPlus, eQuest, IES-VE, and many others are now able to make estimations about the performance of a building under a wide range of scenarios among others.

While the science of more accurately simulating the performance of a behavior has developed significantly over the last several decades, the role of these simulations in decision processes has evolved as well. Simulation models are used in the design phase to predict energy consumption and comfort levels of a building, at the delivery phase to guarantee performance, during the use phase to monitor performance or in a control loop, and during retrofit decisions to decide about the potential benefits of different interventions.

However, a truly value-driven approach to the design of buildings is seldom taken while making these decisions. This is largely due to the low level of maturity of value-driven methods, which would be responsible for outlining a framework in which the simulation models should be used in order to ensure effective decision making. However, another key obstacle remains the significant computational cost of evaluating an energy model. As a result of the advancing complexity of tools, individual runs can be extremely time consuming, ranging from seconds for overly-simplified models, to hours or even days for some very large buildings. Because these simulations are so time-consuming, a truly predictive analysis of a building, including uncertainty, is extremely computationally expensive. Further, uncertainty about the accuracy of the increasingly specialized modules in energy simulation tools is seldom quantified, let alone available to the general user base. This makes it difficult to develop accurate and unbiased information about a building alternative.

Understanding Sources of Uncertainty in BEM

Ongoing research performed in participation with researchers from the College of Architecture at Georgia Tech has led to the identification and analysis of several sources of uncertainty for BEM. The sources are organized into 5 scales:

  • Meteorological / Macro-Climate
  • Urban / Meso-Climate
  • Building Construction
  • System Performance
  • Occupant Behavior

Efficient Evaluation of BEM Scenarios

In order to facilitate the use of value-driven methods in the building energy modeling community, a dedicated software workbench is being developed. The Georgia Tech Uncertainty and Risk Analysis Analysis Workbench (GURA-W) provides architecture and engineering practitioners with an automated interface to evaluate uncertainty about the performance and value of a particular building design or retrofit.

GURA-W,  in its current design iteration, contains 5 modules, as well as a repository. By including, omitting, or reorganizing different modules, users are able to change overall functionality to fit their specific context. The 5 base modules relate to the use of a specific energy simulation engine, EnergyPlus. The EnergyPlus energy simulation tool requires two structured input files, one specifying the building geometry, construction, and operation, as well as selection of certain physical phenomena to be included, and the other specifying the meteorological conditions for the immediate vicinity of the building. A single plug-in is used to define the functionality for each of these three modules, namely the Building Module (responsible for parsing and editing the Input Data file (IDF)), the Weather Module (responsible for parsing and editing the Energy Plus Weather file (EPW) ), and the Simulation Engine Module (responsible for executing and verifying the correct termination of the actual executable file). The Post-Processing Module is then responsible for parsing the numerous output files generated by EnergyPlus in order to capture important characteristics as identified within the value model specification process.

While flexibility is required for the tool to be effective, automation is the key towards being able to make the process more efficient. The two modules that deal with this aspect to the largest degree include the Sampling Plug-in and the Building Plug-in. The Building Plug-in, which is responsible for parsing building parameters, does this through the definition of parser objects in Java. After a parser object is specified once, it automatically parses that parameter for any other instance of an IDF. This means that once the initial work is completed in defining parsers, the user does not need to perform any manual script wrapping in order to include uncertainties with respect to the energy simulation portion of the value model. Because of the large number of parameters in which uncertainty exists for building models, this is an extremely valuable characteristic. Previously, the parsing of parameters would have to be performed manually for each specific instance of an IDF, leading to the high likelihood of transcription errors.

The Sampling Module is responsible for automating the extraction of uncertainties from a UQ Repository, which stores the results of previous research in quantifying uncertainty about various simulation parameters and models. Once the uncertainty distributions have been properly extracted, the Sampling Module then propagates uncertainty through the model via Latin-Hypercube Sampling. Each of these plug-ins have been specified using ModelCenter's Java open API, except for the Post-Processing Module, which is written using MATLAB. The Sampling Plug-in then interfaces with the UQ Repository to specify default uncertainty distributions for each of these parameters, where relevant. The UQ Repository is an XML database for storing information about each type of parameter, and can be easily accessed, modified, and updated using an Excel interface.

Case Studies:

Energy Savings Performance Contract (ESPC)

As a direct result of several executive orders and congressional acts, many government agencies are required to retrofit their current building stock in order to meet guidelines on sustainability and energy consumption. However, many of these agencies possess line item funding authority, and therefore cannot necessarily afford the sizable upfront costs associated with large scale building retrofits. A potential solution to the problem are Energy Savings Contractors, who are willing to perform the retrofits with no initial payments, and be repaid through savings in the building's operations and management costs. However, the unique nature and impact of uncertainty in the weather and operation of the building results in risk to both parties about whether the predicted energy savings will actually be realized. As a direct result of this risk, both parties may be reluctant to enter into any contractual agreement.

Net Zero Energy (NZE) Home Design

An ongoing research topic in the field of Building Energy Modeling regards the design and development of homes which consume a net value of zero energy. The design of such homes is technologically and financially risky, since their development often involves the use of unproven building technologies and large financial investments. Even when evaluating a particular home design, current evaluation methods struggle with determining an acceptable probability of thermal discomfort or power failure.

Peak Power Tariff Avoidance Strategies

As aging electrical infrastructures struggle to keep up with demand, some utilities have implemented tiered-based and peak/off-peak electrical billing rates that increase the consumer's price for electricity either based upon total usage or time of use. As a response to these altered rates, it may be possible for a home or business owner to replace electrical appliances with 'smart' versions that can be operator by a central controller, or to add photo-voltaic panels and battery storage to offset consumption during peak hours. However, the investment required for such systems can be quite high, and depends heavily on a number of system characteristics which are context specific. In such circumstances, it can be difficult to determine the optimal path forward.