Effective energy management programs require a substantial investment of time and money, which is why executive sponsors expect an attractive and dependable return on that investment. So how do you measure the return (savings) in a way that makes sense to your diverse set of stakeholders?
Linear regression modeling of energy consumption data is about establishing a mathematical relationship that determines how to account for changes in production, weather and other factors when evaluating actual returns from the program. The process of modeling energy data reveals the true impact of your energy efficiency projects, enabling collaborative decision-making.
Is Your Program Effective?
The actual impact of an efficiency program, in terms of energy savings or cost reduction, should be an impartial measurement without bias so the same result is obtained no matter who makes the calculation. In many situations, this is easier said than done. When energy data is not adjusted with a model, the result is a frustrating lack of agreement on program effectiveness, and often a loss of confidence—and thus investment—in energy efficiency.
Consider the example below (Figure 1). Actual Energy Consumption represents a finance team’s view for evaluating program impact and returns. In this case, it appears energy conservation measures have had no significant impact on energy performance.
This is a common situation in manufacturing operations where production levels and product mix can change frequently, influencing energy expenditures and misleading perceptions of energy management program performance. So how do you calculate actual energy savings and determine the true impact of your program when so many factors are in flux?
That’s where energy data modeling comes into play, as illustrated in our next example (Figure 2). Here, Modeled Energy Consumption (orange line) has been added to reflect how much energy the company would have used if energy efficiency measures were not in place. In this case, true or actual energy savings would be the gap between ‘blue’ (actual consumption) and ‘orange’ (modeled usage). This figure shows the difference between actual and modeled usage (what would have occurred without implementation of energy projects) so it is apparent to all stakeholders that program returns have grown consistently over time.
Agreeing on a Model
Without a way to adjust energy usage data to strip away the influence of external variables, program leaders are challenged with proving the true impact of investments and making the business case to fund program growth.
To do this, analyze your energy usage against variables that drive it, such as weather and production activity. Develop and agree on a valid model so the impact of cost reduction efforts can be objectively measured and agreed on by all stakeholders.
When it comes to energy data, there’s either too much, too little or it’s not all in one place. Hear how companies today are maximizing value of their enterprise data.
For commercial buildings, weather conditions (mostly ambient temperature) and occupancy tend to have the greatest influence on energy usage. Weather data can be used for comparison to other months, or perhaps the same month last year. Once data is adjusted, discussion and analysis can focus on true building performance without the distractions of variable weather conditions or occupancy changes.
Similarly, it is not uncommon for production levels and product mix to change frequently in a manufacturing operation. Modeling energy data in manufacturing plants can often require additional effort due to independent variables that must be considered. Examples of this include adding a third shift, adding new or additional equipment, or changing the product mix being manufactured. Historical data correlation will reveal variables that are statistically significant determinants, i.e. the “greatest influencers” of energy usage.
When you see energy efficiency devices with a “performance guarantee”, approach with caution. Unwitting customers claiming refunds based on this guarantee often find themselves in an argument over savings. The vendor may claim expected savings were not realized because the weather was unusually hot/cold, or production levels were not typical and occupancy was extraordinary – the list goes on. Avoid this situation by addressing energy usage modeling as part of the procurement process.
Need help with modeling energy data for your company? Contact us to speak with an energy expert.
Contributed by: Paul Stiller, Principal Engineer, Schneider Electric with assistance from Anand Varahala and Jacob Freeman