Building something better: How EnergyX’s Retrofit Artificial Intelligence engine understands buildings and eliminates data collection
WRITTEN BY team ENERGYX ∙ TORONTO, ONTARIO
In a matter of months, COVID-19 turned the world upside down. Life as we know it changed in record time. Most industries have had to shift the way they work overnight, and with that employees have had to figure out how to convert their homes into working spaces. Retail shops shut their doors or began working at limited capacity, bars and restaurants moved to take-out only and some public transport systems have vacated entirely.
The world didn’t stop, we just shifted the way we function. Social distancing is saving lives but it has forced utilities to think out of the box to continue delivering energy efficiency programs.
How can a utility move ahead with its programs if they can’t actually enter a customers home or business?
This is a problem that needed solving far before COVID-19, and it will continue to be important after the pandemic is over. EnergyX’s Retrofit Artificial Intelligence (AI) engine is more relevant than ever as the industry has pivoted from the status quo and toward the idea of “virtual auditing”.
Program Managers are often mandated to achieve the energy savings they put forth in their DSM plans. How does this Retrofit AI engine solve the problem?
Our solution solves an immediate problem in the COVID-reality, but it also offers an opportunity to do things better even when we’ve returned to normal. Our Retrofit AI engine uses building analysis machine learning models to produce evaluation-grade assessments with minimal data inputs. This analysis requires only 5% of the data inputs required during walk-through assessments, and they are automatically captured via publicly available data. This means we can produce evaluation-grade audits at scale, accurately predicting a building’s energy performance and energy savings potential. Most importantly, this is all done without the need for video conferencing or other time-consuming experiences for the customer, “virtual’ or otherwise. Online energy audits have been successful at engaging customers and providing them with Home Energy Reports, these tools often require extensive customer inputs and struggle to propel them to action as a result.
EnergyX’s Retrofit AI engine can instantly target and engage customers, provide them with recommendations based on their unique needs while simultaneously quantifying whole-home and measure level savings. When paired with our Program Manager tool, we can automate the entire process from assessment through to rebate, including vendor management. Our Retrofit AI engine provides assessments instantaneously and accurately, allowing Program Managers to measure and validate the energy savings of program participants and guide future program design.
What makes EnergyX’s Retrofit AI engine different?
Our building analysis machine learning models are trained on existing home audit data and public records, and our results are benchmarked to be statistically equivalent with on-site energy audits in terms of both evaluation results and consumption estimation.
Our Retrofit AI engine is an emerging energy efficiency tool that allows utilities to achieve the following goals:
Identify homes with the most energy saving potential at scale -- without the need for customer consumption data
Consumption data is not the only option to understand buildings and drive energy savings in homes. Energy assessments can be completed -- even with limited information -- through the use of models that utilize patterns analyzed from building science data. With 3 simple data inputs our Retrofit AI engine can achieve 80% accuracy in comparison to a certified on-site energy audit. With more detailed information regarding the technical features of the home, such as Insulation, it is more than 90% accurate. The Retrofit AI engine replicates space heating energy consumptions with 5% of the data inputs compared to what’s typically required during an in-home energy audit. This means a utility can instantaneously assess every home in the service territory, and identify the buildings with the highest propensity for cost savings or deep energy retrofits.
Instantly generate accurate energy saving potential and personalized recommendations
Two things make this possible! Firstly, our machine learning models have been trained to replicate industry-standard energy simulation software as well as the retrofit recommendations made by certified energy auditors. Secondly, we leverage past on-site audits and publicly available data sources to correlate input features. This means we can infer technical information about future homes from a few simple inputs that most homeowners would know. Hence, even with 3 simple inputs, our Retrofit AI engine is able to provide accurate and personalized retrofit and measure recommendations as well as the associated cost savings.
Run smart marketing campaigns based on customers’ saving potential and propensity to engage
Data management and analysis is key to audience targeting. Because of the data already captured by EnergyX’s Retrofit AI engine, program managers hold the key to success before the program even begins. Using this knowledge, they have the opportunity to understand which homes have the highest potential savings and the ability to target them with timely and relevant opportunities. This allows Program Managers to spend marketing dollars where they will have the largest impact.
Energy advisors can generate audit-grade reports virtually with 10X fewer inputs from the customer compared to a traditional in-home audit
By using publicly available building data, basic customer inputs and integration with the utilities database, our Retrofit AI engine can generate results that are statistically equivalent to onsite audits. The tool is able to identify potential energy savings as accurately as an ASHRAE Level 2 audit, and assign measure-level savings on par with the ASHRAE Level 2 computer simulated modelled savings. This has many important implications, including:
Enormous time savings when compared to alternative virtual data collection solutions, let alone walk through assessments.
High customer satisfaction. With minimal need for direct inputs, customers can avoid the hassle of trying to find difficult-to-access information and spending time entering it.
Accuracy. By focusing on publicly available building data instead of customer inputs, the accuracy of the results remain high.
Inform program design and evaluation by maximizing energy savings per program dollar spent
The advantage of the Retrofit AI engine is that it is highly scalable, with a competency to audit thousands of households at the same time. By being able to virtually audit homes across states, provinces and countries, we can identify high potential customers and aggregate results to get a bird’s-eye view of regional savings. This ultimately enables utilities to match their energy efficiency programs with households and regions that will benefit most from them, ensuring the energy savings per program dollar spent are maximized.
All of these goals have one thing in common: they focus on delivering energy efficiency programs more effectively. As we navigate our current challenges as an industry, we should do so with the intention of emerging better than we were before. Instead of just returning to a more digital version of the status quo, we can carry forward some new, innovative solutions to improve program design and evaluation.
We have an unprecedented opportunity to rethink what is possible. We can reduce the cost to identify energy savings opportunities and achieve climate change targets. We can optimize on-site visits and minimize truck-rolls. We can segment customers and deliver exactly what they need, when they need it. Most importantly, we can make people happy and make the energy efficiency sector stronger than ever before.
