Before we get started, I have a quick announcement:
Training is one of the major bottlenecks for our industry and I have a hunch there’s a role for Nexus. I’m looking for an initial cohort of ten students to join a new introductory course on smart building technology.
If you know someone that is new to the world of smart building technology—whether they’re a new employee or a client or someone looking for a job in smart buildings—please have them email me (firstname.lastname@example.org) or hit reply.
And please move quickly… I’m only looking for 10 of the most engaged learners.
Oh, and these first 10 students will receive a heavy discount on tuition and lifetime access to all future cohorts.
Here’s an outline of this week’s newsletter:
- 🤔 M&V in the age of coronavirus
- 📚 What I’m reading
- 📈 A new framework for automated point tagging
- 💡 New from Nexus
Oh, and by the way: if you missed last week’s edition, you can find it here.
Disclaimer: James is a researcher at the National Renewable Energy Laboratory (NREL). All opinions expressed via Nexus emails, podcasts, or on the website belong solely to James. No resources from NREL are used to support Nexus. NREL does not endorse or support any aspect of Nexus.
1. 🤔 M&V in the age of coronavirus
Energy savings are rarely measured… they’re calculated. That’s because once you change a building, there's nothing left to measure against. COVID-19 has presented a massive change to building operations—the biggest I’ve seen in my career. On one hand, M&V (measurement and verification), and specifically the practice called “Non-routine adjustments”, was made for a time like this.
On the other hand, I’m not sure our standard M&V practices extend to pandemics. As just one simple example, let’s say you have an energy savings performance contract with guaranteed HVAC savings. During quarantine, the HVAC on three floors got shut off at the breaker. Were the energy savings due to the performance contract? No.
In addition, we have an inherent bias in our industry: Non-routine adjustments don’t tend to happen unless they benefit the efficiency provider. So I’d bet there are a lot of baselines out there that need to be adjusted but haven’t been.
What do you think? How are you handling this on your projects?
A few other bits of news related to M&V:
Someone finally exposed Cenergistics. I *highly recommend* this article. It has all the best things: Shoddy M&V, a connection to Enron, bribery, and the waste of taxpayer money in the name of greenwashing. I don’t know whether to laugh or cry. The question is: how are they still in business?
Vacant structures shed light on energy efficiency failings—according to Hatch Data, we haven’t done a great job at shutting buildings down during WFH and the reduction in energy consumption shows.
For reference, looking at CBECS data, simply turning off the non-emergency lights would get 10-15%.
During this time of social distancing, it is tempting to view society’s enhanced environmental performance as a success. However, quite a different story has emerged. If anything, we should be saving significantly more energy. Rather than celebrate our reduced climate impact, we should use the revelation of a systemic failure to improve building design and operations in the future.
👉 If you enjoy reading NEXUS, feel free to share it with colleagues!
3. 📚 Reads
- Can COVID-19 Be Spread Through HVAC Systems? (Engineered Systems)—Good interview, but I’m still astonished by the lack of discussion of the energy/ventilation tradeoff.
- Why Every Environmentalist Should Be Anti-Racist (Vogue)—“The time is now to examine the ways the Black Lives Matter movement and environmentalism are linked.”
- Why white households pay less for utilities (City Lab)— “Poorer households could modify their habits all they want and would still have high EUI, because for these households, EUI is often a function of having larger families or more people living within a relatively small unit, like an apartment, with inefficient heating and lighting infrastructure. Although this demographic is often told to change their behavior, much about their EUI is out of their control.”
- As NYC reopens, Hudson Yards takes high tech precautions (CNBC)
3. 📈 A new framework for automated point tagging
Warning… this gets nerdy. Several NREL colleagues and a few BrainBox AI friends just published an important research paper on automating part of the semantic modeling process. If you listened to episode 8 of the Nexus podcast, this is the paper that Jean-Simon mentioned.
If you’re new to the industry, I think the authors did a great job describing why automated tagging is such a big deal. In short, removing the manual labor required to add context to a building’s data drops the upfront costs of deploying and integrating smart building solutions, therefore increasing deployment.
This opportunity is being widely pursued outside of the laboratory as it’s one of the best use cases for machine learning in buildings. We discussed it on LinkedIn a few months back. It’s one of my favorite topics because of the wide range of opinions and claims. Some say it’s impossible. To others, it’s solved and buildings can be tagged in a few hours.
To provide some context to this paper in light of past discussions, the authors took the following approach:
Haystack over Brick—They provide a good discussion of Haystack and Brick, including the pros and cons and why they chose Haystack, which is more popular and flexible to new/custom situations, whereas Brick is not.
Bottom-up over Top-down—The full semantic modeling process includes point, equipment, and system modeling. The authors chose to start with point modeling. Future work will expand to equipment and systems in a bottom-up fashion. It’s yet to be determined whether this approach is better than a top-down fashion.
Expert rules + several machine learning techniques—To minimize the time required for a human to mess around in Excel, the authors combine domain-specific expert rules (e.g. if a point name is “DAT” it should probably be tagged
discharge air temperature) with supervised and unsupervised machine learning techniques, including:
Supervised: using a model that is trained with correctly pre-tagged points, match Haystack tags to the trend data of each untagged data point (Random Forest and Support Vector Machine).
Unsupervised: Cluster similar raw point names together to identify naming and syntax conventions that are consistent within the individual building (k-mer Analysis).
This is worth the read if you work with data in buildings. Let me know what you think!
This installment of NEXUS is free for everyone. If you would like to get full access to all content, join the NEXUS Pro community. Members get exclusive access to the Nexus Vendor Landscape, monthly events, weekly deep dives, and all past deep dives.
4. 💡 NEW FROM NEXUS
DEEP DIVE—Last week, I sent out part 2 of my series on advanced supervisory control on cutting through the hype and building it back up (Pro members only)
EVENT—June’s member gathering is tomorrow. Pro members received a calendar invite. Here’s the plan:
We’ll do two breakout rooms so you can meet likeminded industry leaders
Dennis Krieger, Director of Engineering at Willow, will present on “Demystifying the Digital Twin: Connecting BIM with IoT”
If you’d like to attend the event, sign up for Nexus Pro.
DISCUSSION—there was an interesting discussion on LinkedIn last week about why vacant buildings aren’t saving more energy.
OK, that’s all for this week—thanks for reading Nexus!
If you have thoughts on this week’s edition, let us know in the comments.
P.S. Next week, I’m taking a break to read, reflect, and roam through the mountains of Colorado. See you 7/7. In the meantime, could you share Nexus with a friend?