The final CalTRACK 2.0 working group meeting recaps hourly methods updates including 3-month weighted baselines, reviews the four major tasks completed during CalTRACK 2.0, and previews the CalTRACK 3.0 sandbox on GitHub.
CalTRACK working group explores how hourly savings can be valued using constant, step, avoided cost, and avoided energy valuation methods, and compares outcomes across home performance, lighting, and load shifting scenarios.
CalTRACK working group validates the Time-of-Week and Temperature model for residential buildings, finds evidence of overfitting with monthly regression, and recommends a three-month weighted regression approach for hourly methods.
CalTRACK working group reviews three methods for calculating portfolio-level hourly energy savings — constant marginal pricing, static peak pricing, and real-time pricing — and plans empirical testing for common use cases.
CalTRACK working group examines the challenges of aggregating building-level hourly savings into portfolio load shapes, comparing vertical and horizontal roll-up approaches and their tradeoffs between uncertainty and information granularity.
CalTRACK working group finalizes site-specific hourly methods including temperature bins, data sufficiency by coverage, and LBNL's occupancy algorithm, then begins discussion on aggregating hourly savings into portfolio load shapes.
CalTRACK working group reviews the Time-of-Week and Temperature (TOWT) model from Lawrence Berkeley National Lab as the specification for CalTRACK 2.0 hourly methods, covering its strengths, weaknesses, and application to site-specific hourly savings.
CalTRACK working group discusses hourly methods proposals covering data sufficiency, the TOWT modeling approach with occupancy and temperature covariates, use case uncertainty, and a restructured documentation plan for CalTRACK 2.0.
CalTRACK working group continues discussion of hourly methods, with test results for hourly methods to be analyzed on GitHub and a standing meeting scheduled for May 3rd.
CalTRACK working group begins hourly methods discussion, introducing the Time of Week and Temperature (TOWT) model from LBNL and identifying key topics for empirical testing including data sufficiency, model selection, and portfolio uncertainty.