An expert system is developed using the science of heuristics to better model energy usage in existing commercial buildings and to predict future improvements more accurately. The software performs an initial audit analysis of all the major building systems including building envelope, HVAC, lighting, office equipment and appliances, water and hot water, and waste handling. A novel feature of the expert system is that it analyzes energy flow within the building more interactively and cohesively, as opposed to looking at each system individually as do most energy analysis tools on the current market. Both forward and backward chaining strategies are used to accomplish this.
During the auditing process, the software queries user habits and system controls to understand occupant behavior, which can have a significant effect on actual energy usage. Responses are analyzed using Bayesian functions to develop heuristic factors, which are then applied to the results of the audit analysis. This ensures that energy usage is modeled as it is used and operated, as opposed to how it was designed, which can differ significantly.
Once the heuristic factors are applied to audit results, the expert system performs a synchronization step with a forcing function to converge the calculated energy usage with actual consumption from the utility bills, so that energy efficiency may be optimized in the target building. The software then generates a list of recommended upgrades that are prioritized by cost, ease of implementation, and projected energy savings. Sustainable and resilient strategies are also recommended by the system, since it is becoming increasingly important that a building not only be “green” but also be resilient in the face of a disaster, natural or otherwise.
The expert system is validated and calibrated with ten schools selected from the Newark Public Schools District in New Jersey. The test group of K-12 buildings proved ideal in that they all had similar usage but also represented a wide range of building age, size, and construction type. They were also subject to the temperature extremes of the Northeast climate. Although the expert system is calibrated for Newark school system, the data libraries are easily modified to model any number of building types and climates.
In general, the model shows very good convergence with actual energy consumption for the ten schools as evidenced by an average synchronization adjustment of -0.9% for electric usage and 0.0% for natural gas. A key finding for the Newark study was the wide range of the heuristic index, which measures how occupant behavior and system controls affect the energy usage within a target building. The heuristic index for the “best” test case is 29%, while for the “worst” test case is 54%, or nearly double. Detail model results show that a well-trained staff and good building management are the most influential factors in reducing the heuristic index and thus energy consumption for a given school. The impacts of factors such as HVAC system type and construction materials on energy efficiency are found to be less significant for this test group. The overall model results suggest that a 17% average reduction in energy usage is achievable by improving building management and custodial staff training, and savings of 10% or more can be realized by implementing modest cost upgrades with rapid payback, such as replacing weather stripping, appliance timers, and filter maintenance.
|
![[NJIT logo]](http://archives.njit.edu/vhlib/images/njit_logo.gif){kind=logo}
![[Van Houten Library logo]](images/vhl-top-image.jpg)
