Calibration to Measured Energy Usage
The Glimpse APIs always create a baseline energy model to quantify the energy usage of a building in its current state before using that model to test different EIM scenarios. In some use cases, the actual current energy usage is available to override this information. For this reason, monthly energy usage by fuel for an average year is available to be specified in both Tier-1 and Tier-2 Glimpse API calls. This is used to refine the calculations according to the methodology below.
End Use Disaggregation
The energy model generates estimates of energy usage in the building at every hour, for every end use category, for every fuel. Downstream processes require that this information be rolled up to the monthly level, so that is how it is returned for most API use cases. In order to incorporate the actual usage information given, the Glimpse Calculator first needs to update this baseline information. That requires disaggregating the given energy usage (per month, per fuel) into all the end uses present in the model.
The steps of the energy end-use disaggregation in the Glimpse Calculator are as follows.
Determine which of the fuels given are used for the end uses that typically use combustible fuels.
- If usage for two fuels are given in addition to electricity, then it is assumed that one of those fuels is used for heating and one for water heating. The fuel with the highest usage in January is chosen as the space heating fuel, and the other as the water heating fuel.
- If usage for one fuel is given in addition to electricity, then it is assumed that fuel is used for both space heating and also water heating (if water heating is present in the baseline model).
- If only electricity is passed, electricity is assigned as the fuel for each end use present in the baseline model.
Determine based on building type, if cooking and laundry end uses are expected. If the building is a hotel or residential building, it is assumed they do, otherwise, it is assumed they do not.
Next, the modelled energy usage (as modelled at the meter) by end use per month is aggregated into four categories: baseload, heating-dependent (heating, fans, and pumps associated with heating), cooling-dependent (heating, fans, and pumps associated with cooling), and solar energy production. This aggregation is completed for each fuel, using the fuel assignments for each end use determined in previous steps.
Next, the zip code of the building is used to get weather information by month. TMY (typical meteorological year) data for the associated weather station (see Data Sources section). Heating degree days (HDD), cooling degree days (CDD), and global horizontal radiation totals per month are used. HDDs and CDDs are calculated using an assumed building balance point depending on the type of building.
- Buildings with larger surface area to volume ratios and higher internal heat gains have lower balance points.
- Buildings with lower balance points have higher cooling loads at lower ambient temperatures and vice versa.
The measured energy usage per fuel is disaggregated into the same four end-use categories. This step utilizes a least-squares machine learning algorithm that minimizes the error between each given fuel’s measured energy usage and the disaggregation constructed by correlating the following using scaler multipliers. The disaggregation with scalar multipliers that produce the lowest squared monthly error with respect to the given monthly energy usage per fuel is used.
- Monthly heating-dependent energy usage and monthly HDDs
- Monthly cooling-dependent energy usage and monthly CDDs
- PV production with monthly global horizontal irradiation
- Baseload (no correlation)
Using the percent makeup of each end-use to its end-use category, the measured energy usage is further disaggregated into each energy end-use present in the energy model.
A final smoothing step applies any remaining monthly energy usage differences between the disaggregation and the given usage to the baseload (specifically, the generic interior equipment end-use).
EIM Savings Updates
The baseline energy end-use disaggregation exercise provided the Glimpse Calculator with more information about how the building uses energy. That information can be extrapolated upon to increase the accuracy of the EIM savings calculations.
The starting point for the EIM savings calibration is the savings returned without calibration. Those metrics are in the form of energy usage difference between the EIM and the baseline model, per end use per fuel. To incorporate the building insights gained in the baseline energy end use disaggregation, percentage factors of measured to modeled energy usage at the end use level are calculated. Depending on the type of EIM, these factors are used to refine the EIM savings in one of two ways.
Fuel Switch EIMs (Electrification)
Some EIMs use a technology to electrify a certain end use (ie. heat pumps, induction stoves). In these cases, the following process is used:
- A ratio of the energy usage of the end-use in question from the EIM model to the baseline model is calculated, representing the overall efficiency effect gained.
- The non-electric energy usage in the measured usage disaggregation for that end use is used as the savings for the EIM.
- The electric energy usage increase for the EIM is calculated by applying the overall efficiency effect ratio to the total of the non-electric energy usage savings.
Energy Reduction EIMs (Efficiency)
Most EIMs simply reduce the energy usage required for a certain end use (ie. VFDs reduce fan energy, high efficiency AC units reduce cooling energy). Measured energy usage is used to calibrate the effect of these EIMs by increasing or decreasing the modelled savings by the ratio of the modelled to the measured energy usage of the end use(s) affected by the EIM.