Back to Blog

CalTRACK 1.0

This year, PG&E will launch a residential Pay-for-Performance (P4P) pilot program that lays the groundwork for an entirely new approach to energy efficiency.

Under P4P, instead of subsidizing energy efficiency projects based on a rigid set of pre-approved measures and modeled results, PG&E will pay nothing up front, but instead will pay program implementers for actual energy savings that are generated through their efficiency retrofits.

While pay-for-performance efficiency has been tried in the past, the PG&E pilot represents the first of new breed of programs that are trying to animate markets by giving program implementers (aggregators in the parlance of the program) the ability to make the same savings calculation as the utility.

Part of the problem with earlier attempts to create pay-for-performance systems was getting everyone to agree on what actually constitutes energy savings—a task that’s harder than it sounds, given the varying assumptions that different experts bring to efficiency measurement.

To solve the problem of weights and measures, PG&E and a working group of energy experts have spent the past four years developing a new set of open methods to calculate payable energy savings savings (full disclosure, my company OpenEE was the lead consultant on the project). The first version of this set of methods was released last week under the name CalTRACK.  

Alongside the methods release, as part of a broader effort to bring consistency and transparency to energy efficiency savings calculations, OpenEE is releasing a full suite of tools and resources in the hopes that we can reduce barriers to investment in energy efficiency and spark a new wave of innovation into the industry.

Specifically, OpenEE is offering a desktop version of the OpenEEmeter, the open source platform that includes a reference implementation of CalTRACK methods. With the desktop version of the OpenEEmeter, any analyst who has access to basic energy consumption data will be able to plug in numbers and get a quick and simple output of site-level energy savings. This tool is supported by extensive documentation that breaks down the mysteries of how to perform site-based weather normalized savings calculations.

Open Calculation Makes Pay-for-Performance Possible

The success of PG&E’s P4P  experiment will depend on several factors, but one critical element  is the ability of aggregators to track and improve their own performance. Lack of performance data has long been a problem for contractors who provide home upgrade services.. Contractors literally have no idea how much energy is saved in the homes they retrofit. With the release of CalTRACK methods, and the deployment of the OpenEEmeter, contractors can now track the results of their retrofits on an ongoing basis. This is good for both contractors and consumers.

Because the OpenEEmeter is an open source platform and MIT licensed, anyone is free to download, use, and build on top of the source code. This makes third-party verification of savings much easier—which in turn helps to increase confidence in results and ensures that all parties are working under the same set of assumptions.

The PG&E P4P pilot is novel in another regard as well. Rather than paying based on calculations of “net” energy savings, the way that most programs are currently evaluated, PG&E will pay for savings as they are calculated using the CalTRACK method. Whereas program savings calculations usually include some kind of post-evaluation adjustment to try to ensure that savings are specifically attributed to particular measures (rather than to exogenous variables such as free-ridership or spillover benefits), the CalTRACK method measures the grid impact of energy savings.

The difference is stark. Program evaluations that determine “net” savings create massive uncertainty because the results are based on evaluators’ judgements and therefore can’t be reproduced. CalTRACK takes this uncertainty off the table and puts the payment risk solely on the performance of the projects themselves. The combination of CalTRACK and the OpenEEmeter means that aggregators can now determine the likelihood that a portfolio of projects will achieve a certain amount of energy savings.

This is the game changer. Pay-for-Performance unlocks the potential for billions of dollars of new investment into energy efficiency and a wholesale market transformation, as aggregators develop new business models and deploy new technologies to create unprecedented value for consumers.

For PG&E, allowing flexibility amongst program implementers and paying only on savings that actually materialize promises to reduce administrative costs and increase the value being generated for ratepayers. And while this pilot is relatively small in scope and limited in reach, the lessons learned from this initial experience will inform future procurements of energy savings.

Insights from California

Developing the CalTRACK implementation of the OpenEEmeter in California over the past year has provided us with a number of insights into what a new approach to measurement can mean for energy efficiency.

Just a few examples of what’s now possible include:

  • Simple load disaggregation.  By running actual consumption figures against weather-normalized degree heating and cooling days, the OpenEEmeter algorithm can break out consumption in a particular building, showing what part is related to heating versus cooling versus “always on” baseload consumption. This allows analysts to guide contractors to certain kinds of homes that provide the “biggest bang for your buck” for specific kinds of interventions.
  • Providing detailed insights for certain kinds of buildings. What are the most effective efficiency interventions for coastal California homes that were built in the 1950s and 60s? How would older or newer homes built inland be different? We’ve been able to illustrate those differences to a level of detail that was never before possible. While similar past efforts have based their calculations on engineering models, the OpenEEmeter, calculates efficiency based on actual, real-world data from customer utility bills.
  • Using machine-learning to aggregate similar days for particular types of buildings. What is the typical energy consumption in a given area of a Monday in July versus a weekend in September? How does that profile change for a certain kind of commercial building where time of year may be less important than whether it’s a weekend or a holiday? Given that information, what kind of procured intervention would give the utility the targeted savings it needs?
  • Managing Peak Load and Outlier Events. Utility managers know that the hardest part of grid management isn’t planning for average days—it’s making sure they’ve prepared for peak times that stretch the system to its limits. We’re using weather-normalized data to predict what usage patterns might look like in an extreme case—such as the hottest day of an unusually hot summer—to allow utilities to procure “bundles” of efficiency that could help them avoid the need to build new peaker power plants.
  • Understanding and managing the resource curve. Because energy data has traditionally included only monthly consumption figures, efficiency programs have targeted overall energy savings as a goal. But in a carbon-constrained world where distributed, intermittent renewable energy must take the lead in production, figuring out how to get the right savings at the right time and place will be key to making the whole system work. With the advent of advanced meter infrastructure (AMI, or “smart meters”), utilities can now access consumption patterns down to the hour. Using this data, the OpenEEmeter can show not just how much savings has occurred, but where and when it matters the most—allowing utilities to pay for the kinds of energy efficiency measures that help balance the grid and accelerate the move toward a 100 percent renewable future.