Inside Look: a Product Management Perspective on Developing AI-based Building Controls (from an Energy Perspective)
In 2017, PNNL published a 252-page document for the US Department of Energy to estimate the potential energy savings that could be captured through periodic re-tuning of US commercial building stock.
The document is called “Impacts of Commercial Building Controls on Energy Savings and Peak Load Reduction.”
Spoiler: The potential savings is a lot — roughly 5 quads of commercial building energy consumption (about 30%) overall and 10–20% of peak load can be managed or curtailed.
According to EIA 2018 wholesale electricity market prices in the United States, the weighted average cost per MWh was $44.43908. The consumption savings for 5 quadrillion BTUs of energy is equal to 1.465B MWh, or roughly $65B dollars annually.
And, according to EIA’s 2012 CBECS study (Commercial Buildings Energy Consumption Survey), there are 5.6M commercial buildings, covering 87B square feet of floor space.
It’s worth restating without the math — the United States could save $65B a year in energy costs just through periodic re-tuning of commercial building controls.
That’s over $11.5k per building, or $0.75 per square foot, available for capture every year.
That companies and organizations have since made some of these efficiency measure packages mandates and seen real savings from implementation is a testament to the value of the work.
But how much does it cost to earn those savings?
As illustrated in this excerpted table, one of the practical limitations is that not all energy efficiency measures have the same cost, nor do they have the same payoff.
As a building owner or operator, your basic strategy would be to do the stuff at the top left and avoid the stuff at the bottom right.
But what would happen to this chart if we had a technology that reduced the effort level of every energy efficiency measure (blue arrows) and increased their potential savings because it was optimizing continuously 24x7 (green arrows)?
Blue arrows indicate how AI-based controls reduce effort to realize savings from efficiency measures. Green arrows indicate how AI-based controls increase the efficiency of actioning these measure through more anticipatory timing.
As a building owner or operator, your new strategy would be simpler. You’d start with the highest savings opportunities and move down the list.
Thinking in this new paradigm is an instructive framework for developing AI-based building energy management and AI-based automation products.
For example, we can think of EEM23: Advanced RTU (rooftop unit) controls as an outlier unmet opportunity for differentiation. Due to its high effort to implement, this is a service we can assume vendors only offer under strict conditions. It could be that the customer is cost insensitive or weighing other forms of ROI like occupant comfort or operating uptime. It could be that the vendor is selling the capability as part of a package of higher margin services.
The strictness of the condition is Verdigris opportunity to create delight for customers.
Here are some other question this study reveals or allows us to infer that I’m eager to dig into:
* What combination of geographic focus, building type and asset configurations would most benefit from our product today?
* What product capabilities would result in maximal energy savings? Dollar savings?
* What increase in potential return does real time data, predictive analytics, and artificial intelligence offer for customers?
* Which capabilities would lead to collaborative opportunities with the existing marketplace of value added service providers?
As frequently as possible, my goal is to share our roadmap for commercializing Verdigris’ autonomous buildings product — Adaptive Automation.
* To gather input and feedback
* To connect with like-minded problem solvers
* To maximize value for our customers
* To inspire adoption
* And to accelerate the industry segment and further the mission