In California, energy efficiency programs will soon be contracted out to third parties in hopes of increasing the impact of energy efficiency programs and lowering resource costs. In our previous post we discussed why energy savings based on normalized metered energy consumption (NMEC) has the potential to streamline estimations of savings and reduce overall energy efficiency procurement costs by standardizing the analysis and data requirements. Monitoring and analyzing energy use over time will be the key to success of NMEC programs, but only if energy engineers have the necessary analytical skills to address a new set of issues.
Since our post, two distinct ‘flavors’ of NMEC have emerged: site-specific and aggregated project analysis, both of which require a strong background in analytics and statistics.
Site-Specific NMEC
Site-specific NMEC relies on many observations at a single site to understand building operations and reliably develop valid models. It requires continuous monitoring of energy use in facilities and enables the detection of unexpected use or degradation of savings. In essence, it provides a means to assure savings are maintained over time.
This analysis requires engineers to understand data analytics to detect and investigate energy use anomalies, quantify impacts of non-routine events, track on-going performance, and accurately normalize annual savings. Engineers must not only become ‘data analytic building whisperers,’ they must be able to understand the ‘whispers’ from other engineers and their buildings.
Program-Level NMEC
Program-level, or site-aggregated NMEC relies on a larger number of sites to show statistical validity at the portfolio level. Using this approach, savings are determined and reported for the entire population of projects in a program. It requires systems to collect the massive amount of project data, enable dynamic project updates, and enable access for program evaluation. These are not small challenges either. This approach also requires unique analytical and statistical approaches, as well as an understanding of the technologies behind the savings.
Why does NMEC analysis need different skills?
Both types of NMEC demonstrate the need for engineers familiar with big data management methods and information systems, practical knowledge of advanced data analytic methods, and methods for informative visualization of results for our decision makers. Engineers cannot rely on ordinary least squares regressions anymore.
Needed skill set
To date, spreadsheet analysis and energy models have made up the majority of approaches to energy engineering. Now we need to add data analytics, and specifically the analysis of time-series data and statistically-valid models.
Recently when we advertised for an engineer with data analytics experience we got resumes loaded with acronyms. Most candidates seemed good on data with no engineering background, or strong on engineering, with no real experience with statistical modeling. Having both might be the “unicorn” of energy efficiency.
We’ve been passing this message to educators. We need a ready workforce that understands real buildings and their systems AND have the analytical skills and tools to make use of the massive amounts of data that are becoming available. Skills like coding in R, Python and SQL are a huge leg up for making sense of data and finding recommendations in the noise that one can find when you have access to tons of trend data. If you don’t know how a VAV box works, how are you going to interpret all the BAS data you have on cooling calls, reheat valve position and discharge air temperatures.
Why software alone can’t do the job
Many software tools are available on the market address some of these challenges but, as yet, we’re far from a complete solution. So far it still takes a human to interpret the data from one-off system designs to determine a root cause for a problem. It’s not helpful, for instance, to know we have simultaneous heating and cooling in the building if we can’t identify precisely what’s causing it. This is where many of the automated systems fail to deliver, though as we see more modular fault detection and diagnostics (FDD such as Skyspark’s modular rules) you can imagine that it won’t take long for algorithms that catch common problems will be ubiquitous.
Calling all future energy engineers
NMEC and other analytical approaches to energy efficiency have enormous potential and we want to see these methods succeed. Our hope is that by bringing light to industry needs and inspiring training options for energy engineers, we can help to bring the full potential of these methods to solve the big energy efficiency problems of the day.
Like this post? Share it on LinkedIn.