We’ve tried to make Carbon Signal as accessible as possible when it comes to terminology, but there are a few terms that we use that have specific meanings.
Calibration
The process of aligning modeled energy use with measured data for existing buildings. Typically, this is a manual process whereby an energy modeler will adjust the inputs of the energy model until the outputs (energy use) match measured values. Carbon Signal uses ML/AI methods to automate the calibration process.
Characterization
The Characterization of a building represents the unique combination of model input parameters (performance characteristics of specific building components or systems) that can explain the observed pattern in monthly energy use. Since Carbon Signal uses ensemble modeling techniques, the performance Characterization of each building is represented as a range of possible values for each input parameter. Think of Characterization as a set of hypotheses about your building.
Design Space
On Carbon Signal, a Design Space is a collection of model input parameters representing a wide range of possible values. The Design Space for wall R-values, for example, might extend from R-5 to R-30. The full Design Space for a Carbon Signal model is an n-dimensional range, where n represents the number of possible model input parameters.
Disaggregation
Disaggregation is the process of taking aggregate utility data (e.g. electricity), and breaking it down into its component end-uses (e.g. cooling, lighting). On Carbon Signal, we use energy models in the disaggregation process, meaning the relationships between different end-uses are based on established thermodynamic equations.
District Energy
Carbon Signal has two energy inputs for District Energy: District Heating and District Cooling. We use these terms specifically to mean thermal energy that is delivered to the building, typically from an off-site plant. When adding District Energy data, you should only count the thermal energy delivered to the building and not the fuels used to generate that energy. For example, if you have a central utility plant in your portfolio, and that plant delivers heating hot water (HHW) and chilled water (CHW) to a building, when you upload the utility data for that building it should only include the HHW and CHW energy and not the energy consumed for HHW and CHW production (commonly a combination of natural gas and electricity). Adding both would be double counting the energy needed for heating and cooling your building.
Emissions Factor
An Emissions Factor is a multiplier applied to energy data to calculate the carbon emissions associated with energy use. Emissions factors are typically measured in units like kg/kWh, which represents the amount of carbon dioxide equivalent emissions (measured in kilograms) associated with using one kilowatt of electricity. Emissions factors vary by fuel type and by region.
Energy Model
We use Energy Model as shorthand for the more specific term Building Energy Model. A Building Energy Model is a tool for simulating the energy performance of buildings under varying conditions. Carbon Signal relies on the open source standard EnergyPlus for all of its energy models.
Ensemble
In machine learning, an ensemble is a collection of diverse base models that are used to predict a specific outcome. In Carbon Signal, the base models (also known as base estimators or first-level models) are EnergyPlus energy models with a diverse set of assumptions that are all capable of explaining a specific pattern of energy use. The probability distributions that are associated with Carbon Signal’s analysis are the result of diversity in the ensemble.
Intervention
An intervention is a practice or technology that reduces energy or carbon. In terms of the energy model, an intervention is a change to an assumption in the model. Changing the R-value of a wall assembly in the energy model, for example, would be an intervention. On Carbon Signal, interventions are typically defined by multiple changes to the energy model. The intervention labeled “Opaque Envelopes”, for example, is based on changing the R-value for walls to R-30, the R-value for roofs to R-40, and the infiltration rates to 0.02 cfm/ft². In practice, this may be achieved through multiple actions, such as adding insulating material, reducing thermal bridging, or air sealing.
Scenario
A Scenario is a combination of interventions applied to one or more buildings.