Software-as-a-Service Optimal Scheduling of New Mexico Buildings
The University of New Mexico (UNM) and Lawrence Berkeley National Laboratory (LBNL) are demonstrating building and microgrid remote optimisation using a software as-a-service (SaaS) configuration for the complex and highly efficient UNM Mechanical Engineering Building (MEblg), which uses only about 40% as much energy as typical. The SaaS approach lowers weekly electricity bills by a third under the most favourable conditions, and over an 8 week test during summer 2014, an 11% cost reduction was achieved overall. The building's heating, ventilation, and air conditioning (HVAC) system incorporates cooling assisted by a 232 m2 solar thermal array providing heat to a 70 kWth absorption chiller, as well as wintertime heating. A 30 m3 hot thermal storage tank makes heat available at night for both heating and absorption cooling. Additionally, 350 m3 of chilled water (CHW) storage shifts the considerable cooling electrical load of this high desert location off-peak. The big energy and cost savings come from more efficient use of storage compared to the baseline strategy of fully charging the CHW tanks whenever solar production allows. Parasitic loads become significant at these very low energy consumption levels. A MEblg model has been built on LBNL's Distributed Energy Resources Customer Adoption Model (DER-CAM) platform, and a direct MySQL interface delivers daily week-ahead scheduling based on weather forecasts, loads, tariffs, etc. The approach is complex, involving multiple vulnerable interfaces and execution steps. Failure analysis shows that despite the notable potential cost and energy savings, DER-CAM schedules are reliably delivered less than half the time, and implementation accuracy for storage charging-discharging and absorption chiller operation is poor, suggesting better methods and device modelling could improve results. The goal of this work is to find low cost methods for achieving bill and energy savings, and not to achieve outstanding technical performance, which might be cost prohibitive given the small absolute energy savings. MEblg results are reported and set-up described for similar control of a large nearby office building, One Sun Plaza.