The Department of Energy's Small Business Innovation Research program was recently reauthorized, and with it came exciting news for CEL: we received SBIR Phase II funding! This award will help us test, replicate, and scale our technology.
The SBIR Phase II grant will allow us to build on our successful Phase I award by validating our scalable MPC technology. We'll also use this award to establish our technology's replicability across our target marget. This target market includes 13,720 schools, community colleges, and universities in California, Oregon, and Washington who will all be subject to building codes that in some cases will levy penalties by 2026 for non-compliance.
With Phase I funding, we were able to put the incredible research of Joshua New, Ph.D., C.E.M., PMP, CMVP, CSM, IREE to work. We drilled down on building energy modeling parameters to determine which ones relate most strongly to energy use in our target building types. Then, we worked with 11 west coast school districts (ten K-12 and one university) to test the time, cost, and complexity to gather the necessary model inputs using our pared-down parameters. We then sped up the process using a series of isolated and then combined workflow alternatives.
Our results showed significant time and cost savings for data collection. We found that combining alternative collection methods could reduce data collection time by 17-38% (5-32 hours) compared to a baseline, depending on the building size. Early tests with the data collected showed that the resulting data—when used in a model predictive control framework developed at Berkeley Lab—produced accurate indoor air temperature predictions. We also found that our platform has the potential to reduce customer onboarding cost by 50% without impacting the accuracy of the resulting MPC. Preliminary results using time-series data from thermostats promise to eliminate manual calibration and model tuning. This further reduces the amount of expert time necessary to make MPC a reality for SMSCBs beyond Phase I.
Now, we're off to Phase II to make the tool faster and more affordable. We'll also be able to test the data collected with even more backend model predictive control algorithms. In Phase II, we have four primary objectives:
We will develop a commercial prototype with user interfaces, workflows, and data streams and incorporate it into CEL’s software stack. We will incorporate best-in-class coding practices such as user experience design, unit and functional testing, and internal code reviews as a means to improve usability, code quality, and reliability.
This objective ensures that when we integrate the application’s models into MPC controllers, they take less time to set up and calibrate while still producing physically feasible, scalable, and financially attractive outcomes compared to conventional PID and rule-based control.
Field demonstration will ensure that the results achieved in simulation can be replicated in the field and are attractive to customers. We will work with 11 existing demonstration partners to deploy EDI in conjunction with MPC in at least one school per district. We'll do this using a ‘randomized block’ testing scheme to alternate between the EDI+MPC control and the customer’s own baseline control policy over pre-defined testing periods.
In addition to technical product integration, we will produce specifications, installation processes, and marketing materials. We will use these in direct sales to contractors, resellers, and channel partners. By 2027, these partners will make up the majority of our non-recurring installation sales.
We're excited to share our progress with you. Stay tuned for more updates!
