5 Key Steps in Measurement & Verification (M&V) Modelling

Why is Modelling Important in M&V?

Energy consumption of buildings is generally influenced by multiple factors including:

• Operational behaviours
• Weather variations
• Occupancy levels
• Equipment performance to name a few.

To isolate the impact of an ECM, we need a robust model that can account for these variables and ensures an accurate comparison between baseline and post implementation energy usage.

This is where statistical modelling, in particular regression analysis, plays a crucial role in M&V.

Key Steps in M&V Modelling

1. Establishing the Baseline Model

The first step in M&V modelling is to create a baseline energy model. This model predicts what the energy consumption would have been without the ECM in place, providing a reference for comparison. The process involves:

• Collecting historical energy data (typically a year or more).
• Identifying key influencing variables (weather data, occupancy, etc.).
• Applying regression analysis to establish a relationship between energy consumption and the influencing variables.

2. Choosing the Right Modelling Approach

Depending on the complexity of the modelling required and data availability, different modelling approaches can be used, with the two most commonly used as follows:

• Simple Linear Regression: This establishes a direct relationship between energy consumption and single variable, such as the average external temperature.
• Multiple Linear Regression: This goes a step beyond simple linear regression and incorporates multiple different variables to try and improve model accuracy and understand the variability of energy consumption further.

We ultimately end up with an equation of the form

Energy Consumption (kWh) = a + (b x Variable 1) + (c x Variable 2)

Where a, b and c are determined through the modelling. These are called coefficients. They help us understand for every one unit change in the variable, how much the energy consumption varies by.

3. Validating the Model

We need to ensure that the model constructed is statistically robust. There are a lot of statistical measures available, but the most commonly used are the R2 and the adjusted R2.

• R2 – This is a value that ranges from 0 to 1. A value of 1 indicates perfect correlation between predicted and actual energy consumption. A value of 0 means the model created is no better than just fitting an average to the data. Note that R2 should only be used if one variable is used in the model.
• Adjusted R2 – This is used when more than one variable exists in the model. As with R2, it ranges between 0 and 1, but it penalises the accuracy for when non-critical variables are included. It therefore urges you to only include relevant variables within the model.

4. The Adjusted Baseline

Since the external conditions, i.e. the variables we have decided to use to model the energy consumption, can change post implementation of the ECM, we then take these and project out what the energy usage should have been under these new external conditions using the model we have constructed. This ensures a fair assessment of the ECM’s impact.

5. Calculating the Savings

Once the ECM is implemented, the actual energy consumption is compared to the adjusted baseline model to determine the impact.

Energy Impact = Actual Post Implementation Consumption – Adjusted Baseline Consumption

• A negative number indicates a saving
• A positive number indicates an increase

Conclusion

Through using data driven modelling techniques, M&V can provide a structured and transparent approach to verifying the energy savings. This ensures that the reported savings are both accurate and credible, helping businesses make informed decisions, achieve financial savings they can stand over and progress their sustainability goals.