Diving into the interesting world of energy efficiency data
Updated: Jun 2
Part 1 in a series
In the universe of data, information about energy efficiency has some idiosyncrasies that make working with the data interesting. Those characteristics are largely due to the nature of the energy efficiency process. When we reduce the electricity and natural gas consumption in our homes and businesses, by turning off our furnace when we are away, installing solar panels on our roofs, or replacing appliances with more efficient alternatives, we increase our energy efficiency. That is, we consume less energy than we did previously to achieve the same results.
Why does energy efficiency matter?
Generally speaking, people consume more energy each year than we did the year before. New technologies emerge that we adopt and add to our lives. And each year we have more humans than we did the year before. So we have an ever growing population multiplying against an ever growing per-capita appetite for energy. That means that in the absence of energy efficiency improvements that reduce our per-capita consumption, our population-level demand is increasing at a significant pace every year.
When I was a kid, my dad said to me there are two ways to be rich; one is to make more money, the other is to need fewer things. It's a simple idea that translates here - in the same way we as individuals align how much money we make and how much money we spend, as a community we need to align energy demand and energy production, and we can achieve balance by producing more or by consuming less.
But why not just increase supply to meet increasing demand? We can build more power plants and produce more energy. Well, power plants have direct and indirect financial and environmental costs, whether we are talking about nuclear power plants, or coal, or one of the other options. We also have finite resources that can be converted to energy, and those resources are valuable for other things besides producing energy.
And who says we have to align energy production and energy consumption? Well, pretty much everybody. When demand exceeds supply, prices go up and some people may not get the energy they need. And that doesn't go well, for lots of reasons. So I'm going to presume we agree that 1) trying to meet the moment by increasing production isn't our best option and 2) we want to keep demand and production in-line. Let's get back to it.
Can energy efficiency really offset increasing technology use and growing populations?
Yes, but don't take my word for it. Energy efficiency programs has been around long enough that we have data to prove it. Check this out.
The black vertical line is when California started investing in energy efficiency programs to reduce demand. Before that line we see California and the US increasing at about the same trajectory. And after the line, the rest of the US continues its upward trajectory and the California line flattens out. Wow, right? That’s pretty amazing.
How did California do that?
California was already struggling to align energy production and consumption in the early 1970s, when an energy crisis made the situation even more difficult. The state faced hugely unpopular rolling blackouts that were forecast to only get worse. So the state started investing in energy efficiency to reduce demand. Today California invests nearly US$1B per year in energy efficiency. The California Public Utility Commission authorizes utilities, regional energy networks, and community choice aggregators to administer energy efficiency programs that are paid for by the people of California. That is, if you pay an energy bill in California, a small fraction of the amount you pay each month goes to fund energy efficiency programs. To ensure those funds will be well spent, program administrators have to forecast program savings and cost data and demonstrate that they will be taking cost effective actions before their program budgets are authorized. And at the end of each program year, the administrators have to report full detailed data on every action taken to document the work that was done.
That's interesting. Can we talk about the data now?
OMG yes! So energy efficiency is characterized by:
Energy consumption in the absence of efficiency
An action taken to increase efficiency
Reduced energy consumption after the efficiency action
We need data about those things to understand the impact of the efficiency action. Data describing an energy efficiency action are straightforward; we want to track who/what/when/where/how the efficiency was done. But how do we know the before and after energy consumption?
Ideal data model
In a perfect data world we would meter energy consumption at the place where the efficiency action will be taken for a year ahead of time, take the action to improve efficiency, and then continue to meter the site for another year. We’d then calculate the energy consumption before and after the intervention, perhaps control for weather or other relevant outside factors that differed between the before and after years (e.g. maybe a global pandemic made everyone stay home for a year), and produce a reasonable estimate of the energy savings due to the efficiency action. Here's a simple illustration of the before (baseline) and after consumption data and savings calculation.
Let's now consider how the real life intersects our ideal data model. The perfect data model requires that the people in the place the intervention is done are good study subjects and change nothing else about their energy consumption during those two years to confound the calculation of the delta between before and after the intervention. Yeah, right.
Plus, the perfect data approach takes more than two years to produce an answer, and means we'd have to tell customers who want to take efficiency action they have to wait a year before they can take the action so we can measure their baseline consumption. Then they have to wait another year to find out if it worked. No way.
Beyond that, trying to do all of that work to calculate our best approximation of the actual difference between what was previously consumed and the new efficient consumption for every incentivized light bulb, thermostat, and water heater that will be changed isn’t realistic and would be enormously expensive.
So how do we estimate energy savings in the real world?
Since perfect data aren't a realistic option, we use building models to produce a reasonable estimate of energy savings of a typical installation under various conditions. To estimate the baseline energy consumption, we turn to our building codes and appliance standards, and assume that in the absence of the energy efficiency action, the building would have met the current standard or code. So codes and standards become our baseline that we can calculate typically expected energy savings against.
We can use the same building models we use for baseline to estimate our efficient consumption. Then we figure out the difference between the two, and we have our estimated savings.
In California, the program administrators create what are called "work papers" to document ahead of time the estimated savings for the efficiency actions they want to offer in their programs. Each work paper is a text document full of detailed calculations, engineering assumptions, and detail backing up the assumptions, methodology for the modeling, and estimated savings calculations. Those work papers have to be approved by the commission before the utilities can offer the described actions in their programs. That process results in a library of work paper documents that hold the estimated savings numbers for actions that are approved by the commission.
Energy efficiency program data reporting
Program administrators use work papers like giant look up tables of energy savings for each possible action to create their forecast data when they propose their next year's programs to the commission. As the oversight body, the commission is wants to ensure that the program administrators have every number right in their forecast data. So the commission takes the forecast data and the work papers and reconciles them to check that the program administrators correctly assembled the data.
While that seems like a slow and manual, but reasonable thing to do at a modest scale, when California needs to verify hundreds of thousands of of program forecast records each year against hundreds of technical work paper documents, that becomes a Herculean task. Add to this that there is no consensus data system that all program administrators use, and the reconciliation job gets bigger. The scale of the data grows even more when we get to reporting the actual program accomplishments after they happen, when we are then looking at millions of records that need to be verified.
Data experts reading this will already be recognizing the potential data through-put and quality control problems of a system like this. And efficiency improvement programs dealing with this potential inefficiency is staggeringly ironic.
This is a rich topic, so we'll pause here for now, and pick up looking in more detail at the challenges and opportunities in energy efficiency data reporting in part 2.