Many in our industry have become enamored with Normalized Metered Energy Consumption (NMEC) as a fresh new metric that we can use to measure baselines, determine energy savings and help value energy efficiency. If you work outside of California, you may not have heard of NMEC–it is emerging as a valuable tool in the growing body of analysis methods tied to Advanced Metering Infrastructure (AMI) data, and sometimes is thought of as a part of the Enhanced Measurement and Verification (EM&V/M&V2.0) toolbox. As with anything new, there is always a race from idea to implementation and execution. However, when dealing with an ‘out with the old and in with the new’ paradigm, it is important to carefully consider the approach and implications, prior to full execution. NMEC is no different—we have smart metering technology data that can inform an array of powerful analytical methods, but to date, the capabilities have been notably narrow and largely untested. As NMEC methodology is still in its infancy, we thought we would help provide a bit of context because taking stock now will go a long way to proactively shaping NMEC’s future.
The march of smart meter data.
Essentially NMEC focuses on estimating energy consumption from metered data as a way to tie an energy efficiency program to the grid. Advanced Metering Infrastructure (AMI) data only serves to reinvigorate existing consumption analysis practices, building greater precision and confidence within these models due to the greater observations available through the vast streams of AMI data now available to both implementers and evaluators alike. With this wave of new data and the promise of accompanying program designs, we can innovate new methods to standardize and quantify the measurement of energy consumption and, more specifically, changes in consumption relative to a reference point: an intervention. This dimension of analysis holds tremendous potential for supporting the measurement of energy efficiency gains with quick turnaround on initial savings estimates supported by a more rigorous measurement of achieved savings at the meter as time accumulates post-intervention. It may facilitate the participation of energy efficiency savings alongside other distributed energy resources in markets that match supply, demand, and timing on the grid. While this potential is exciting and encouraging for future applications, it is taking place in a rapidly changing environment regarding policy and software development. The realization of NMEC’s promising future depends on recognizing both the strengths and limitations of existing NMEC methods and the important contextual considerations in using NMEC for different programs.
To automate or not to automate.
The commercial products currently available represent much of the push to be first to market with automated NMEC software tools. Each of these products boils down to an underlying set of modeling algorithms that have been chosen based on the judgment and background of the product developers. Disaggregating the energy savings signal from the background energy consumption noise can be highly complex and not amenable to automated methods in a variety of settings. Changes in building occupancy, weather-sensitive measures, or energy price fluctuations may confound the effort to attribute a specific change, through the overall energy savings signal, to a given energy efficiency measure. Changes in equipment or energy usage patterns that are unrelated to the energy efficiency measure (e.g., non-routine adjustments) also threaten to cloud the energy savings data. Proper NMEC analysis factors in all the pre- and post-consumption data. It’s data with context allowing us to analyze all the variables to view the whole picture of what was in place before, what conditions existed and to which audience it can be applied, that help establish a solid baseline to measure future initiatives.
Data, data everywhere? Not so much.
There is value in data. In this digital age, companies pay for it, guard it fiercely, and leverage it. There is little in our day-to-day that isn’t tracked or monitored so that someone can have access to data to make informed business choices. It’s the same with energy data. But, with the proliferation of AMI data current NMEC development runs into a conundrum: accessibility.
In some cases, the utility companies that implemented and invested in the smart meter system are dependent on the Software as a Service (SaaS) companies they commissioned to collect and disseminate their AMI data. Ideally, this would be viable if guidelines were in place to protect client data, maintain transparency, and confirm that the SaaS is truly returning the results requested.
Consumers (considered the legal owners of their data) and competitive service providers can be subjected to overly complex authorization processes to gain access. Many states that were early adopters of smart meters had initiated state-wide data repositories. Most were difficult, if not impossible, to interface or access. In 2012, the Obama administration started the Green Button Initiative to help grant easy access to energy consumption data in an easy to understand format. Non-profit alliances and national coalitions have taken up championing the ‘data for all’ initiative and often serve as industry watchdogs. However, there still exists an invisible barrier to access and permissions. A recent settlement agreement in Texas between utilities, energy customers, and third-party users addresses the bottlenecks between smart meter data and accessibility which, if approved by the Public Utility Commission of Texas, may pave the way for similar agreements.
Another hurdle is that many utility data systems were built around the core purpose of customer billing and regulated as such. Even in cases where the regulated permissions issue can be resolved, AMI data has historically been stored in systems that make it difficult to transfer. AMI data can be prolific. Such large quantities of data require advanced storage capabilities, and Personally Identifiable Information (PII) approaches.
Data is so much more than just its capture; its provision, analysis, and meaning are imperative. Universal access is just the start. Businesses thrive on data analysis, but what’s interpreted from the data is vital to leveraging that data. Embedding data collection and processing protocols into the design of future energy efficiency programs would provide much-needed transparency to achieving reproducible and replicable results allowing stakeholders to make informed decisions.
“What specific question are we trying to answer?” and “Who is asking the question?” These questions are at the root of statistics and econometrics in virtually all cases. So, who are we trying to answer the fundamental questions for? At present, those most vested are utility program administrators followed by third-party program implementers and external auditors. Current automated commercial products are only capable of answering some of the questions these stakeholders are looking for. But they are not the only audience; regional transmission operators, investors, and consumers have a stake as well.
From a Program Administrator’s Perspective:
One specific and promising case for using NMEC methods is pay-for-performance (P4P). P4P would shift programs away from a flat-rate rebate system to a more market-based approach. For this type of program to be effective, it would factor heavily on NMEC data to show that change in consumption relative to pre-existing conditions is adequate to suit a P4P initiative and that a certain period of post-intervention measurement is adequate to demonstrate savings from the intervention. In this setting, the program implementer may receive information in a timely fashion from NMEC outputs that can inform real-time adjustments to how the program is being implemented. Final P4P payments could be based on a refinement of the overall savings estimate as time accumulates post-intervention. This arrangement works well when the background energy usage is relatively stable, the energy savings signal is relatively large, and very little energy usage change happens at the site level such as significant changes in usage patterns unrelated to the energy efficiency intervention. Program effectiveness can also benefit from early Measurement and Verification (M&V) partnered with NMEC tools and methods. The key criterion for reviewing the effectiveness of early M&V is to gauge whether it converges with ex-post M&V over the course of 9-16 months, where the ex-post EM&V is conducted using a full complement of pre- and post-period data and, in some cases, triangulation across multiple modeling and measurement methods.
From a Consumer’s Perspective:
For the electricity consumer, their interests are two-fold: involvement in the initial intervention and the gains after. However, there are unintended consequences to these new program models that require a single intervention to measure effectiveness. For the sake of accurate pre- and post-measurement of savings, a program administrator may forego allowing customers to participate in more than one energy efficiency program at a time so as not to cloud the signal of changes in energy usage before and after the energy efficiency intervention. When the consumer’s end goal is the maximum possible energy savings, this can feel a bit unfair.
From a Regional Transmission Operator’s Perspective:
One of the powerful advantages of AMI data over monthly billing data is that it is capable of reflecting the daily load shapes of individual sites and therefore estimating the changes in those load shapes as a function of energy efficiency measures. Aggregated energy efficiency measures can potentially deliver the absence of demand on specific distribution grids at particular times over daily and yearly cycles. These traits give energy efficiency (when seen as a grid resource) some of the same load shifting or peak shaving characteristics often associated with demand response resources. If grid operators can observe that a set of aggregated energy efficiency measures “moved the needle” regarding demand at a specific location and time of interest on the grid, then the value of energy efficiency as a commodity in markets is reinforced. These observed savings, if attributed to specific energy efficiency interventions, constitute claimable savings for which the implementer can be paid based on stacked benefits associated with avoided congestion.
From an Investor’s Perspective:
Currently, investing in infrastructure technology revolves around renewables with some turning to transmission investments since they are considered ‘low-risk’ with a low rate of return but with proven stability as utilities provide an essential service with steady demand. This stability has the potential to grow as the advances in AMI and smart grid technologies progress. Future pay-for-performance models may drive more returns and therefore make energy as a tradable commodity more appealing.
Current rates of return are set by state utility commissions, and this regulation makes energy investing attractive. As reporting models and methods advance, implementers and aggregators within the energy market may be pressed to be more transparent with their energy savings reporting or risk policy change that may impact future investment opportunities.
Define your needs.
An important path forward with NMEC tools and methods is to anticipate potential stakeholder uses for the data and articulate what questions the data can answer. Essentially, delineate if the priority is to measure energy savings at the site level, at the efficiency program level, or both. Well-designed NMEC methodology will provide context and insight on whether the methods deployed are, in fact, answering the questions at hand. Thoughtful and informed data-driven approaches that integrate stakeholder perspectives while minimizing risk and maintaining appropriate levels of accuracy are needed to refine the process and the algorithms.
So, which methods stakeholders come to embrace will depend on how NMEC methods are allowed to grow and expand through iterative cycles of refinement. With a stable foundation to start building up from, technology and methodology will be able to advance the process of refining NMEC’s ongoing applications. In the grand scheme, we are early in the process of the growth and implementation of NMEC where context and perspective are vital. NMEC methods will ideally grow to complement other existing tools used for purposes of energy savings evaluation, measurement, and verification. We feel it is important to keep context in the forefront as we move forward. Prioritizing our motivations for keeping NMEC methodology on track comes back to understanding what we are trying to measure and for whom. At Opinion Dynamics, we understand the kinds of testing, experimenting, and verification required to advance NMEC development. It is this type of work that is at the core of what we do for our clients as professional analysts and statisticians.