Computational methods analyzing interval data from Communicating Thermostats (CTs) have been proposed to remotely evaluate home thermal performance, e.g., to identify priority homes for energy retrofits or characterize homes’ demand response potentials. Such methods usually consider homes with a single CT. Many homes have two CTs and the two streams of CT data could provide greater insight into home thermal performance. However, important contextual information for the CTs, such as the heating system type (boiler, furnace, etc.), floor space, and floor served by the CT are typically lacking. These uncertainties make remote characterization of homes with two CTs challenging.
Building on our work that developed effective algorithms to use CT data from homes with one CT to identify insulation and air-sealing retrofit opportunities, we developed a coarse-grained approach to model a home with two CTs as an equivalent home with one CT. Using a “static” form of a grey-box model (i.e., the model’s differential equations integrated over a period of time), we optimally fuse the data from two CTs to calculate the runtime and indoor temperature of the equivalent one-CT home and to subsequently estimate the whole-home R-value and ACH50. We applied this approach to 74 homes with two CTs with known overall R-values (from home energy assessments) and eight homes with known ACH50 values (blower door tests). Our approach correctly classifies 87% homes with high vs. low whole-home R-value and all homes with high vs. low air leakage.Finally, we report first results of the randomized controlled trial in which we gauge the effectiveness of our algorithms in targeted outreach to increase the customer uptake and implementation of the retrofits.
Read the full article here: Link