How new technology is replacing old supervisory controls

A wide gap has grown between (1) what technology is capable of and (2) what our building automation systems are capable of. Simultaneously, building owners are demanding better performance out of building systems—a level of performance that’s extremely rare, perhaps impossible, with the old technology. Consider a few use cases from recent conversations and projects:

• Can the BAS find the optimal amount of ventilation for maximizing coronavirus dilution and minimizing energy consumption? (Umm, no.)
• Can the BAS detect a poorly-programmed or overridden control sequence and correct it automatically? (Hmm… no.)
• Can the BAS predict when the building will set a new peak in electric demand and choose between potential alternative sequences to mitigate it? (Yeah, no.)

Startups are popping up left and right to try and capture the opportunity inherent in closing the technology and performance gap. I’ve seen two startup approaches to closing the gap (so far). The first is what I call the blank sheet of paper approach. They’re crumpling up and throwing the old model in the trash.

Then they’re building the BAS from scratch with only modern technology. Here we have startups like PassiveLogic (covered in depth here) and perhaps 75F (here).

The second is what we might call the overlay approach—building an intelligence layer on top of what is already commonly installed. Both approaches are compelling, but in this deep-dive series we’re focusing on the overlay approach and specifically, overlay version 2.0.

While the overlay approach has been around a long time now, the strategy has largely been based on one-way communication. Analytics overlays pull data from the BAS and produce insights that a human must then act on. This is the sort of smart building solution that I cut my teeth on in my younger days. I know the value it can create; I also know the limitations. And while there are still startups popping up with this one-way overlay approach, most of today’s startups are going one step further. They’re installing two-way overlays in order to perform advanced supervisory control.

I’m seeing overlay vendors using artificial intelligence to learn, predict, simulate, and optimize HVAC loads. I’m seeing overlay vendors pulling in occupancy data from Density sensors or Cisco wifi routers and using that data to reduce AHU static pressure while office employees are out on their lunch break. I’m seeing algorithms designed to manage load flexibility as buildings begin to interact with the electric grid.

These changes are exciting! I see the opportunities—better controls, better analytics, and more powerful platforms—and I like what I see. However, like any good incumbent who has been around the block a few times, I’m also a little worried. Not because I’m afraid of change—we need change!—but because I’m seeing signs that these newcomers to the overlay approach haven’t quite learned from the experiences of the earliest movers.

I’m worried we’re forgetting our industry’s history of silos as we create new and more intelligent silos. I’m worried we’re forgetting our industry’s history of vendor lock-in as we create new and more sophisticated locks and keys. I’m worried it will take 10+ years to get building operators onboard the bus—time we can’t afford to waste—if we ignore the role of change management and oversell these new solutions. But let’s not let my worries get in the way of disruption—that’s not my point in this series. The point is to orient ourselves, ask the right questions, and approach this disruption strategically.

Here’s an outline of where we’re headed with this deep dive series:

• Part One: 
• What’s wrong with traditional supervisory controls?  
• What is advanced supervisory control (ASC) 1.0 and how does it help?
• Part Two:
• What is ASC 2.0?
• Cutting through the hype
• What the hype-busters could be missing

• Part Three:
• A strategic approach to ASC

WHAT’S WRONG WITH TRADITIONAL SUPERVISORY CONTROLS?

Let’s start by backing up just slightly. To discuss advanced supervisory control, we first need to define traditional supervisory control. Supervisory controllers are part of almost every building automation system. They typically sit on the local IP network and support a lower-level network of field devices. The building’s “front end” supervisory computer or server pulls in all these supervisory controllers, looking something like this (image credits: J2):

Together, the supervisory devices perform some set of these functions: Displaying graphics, exposing field level points, hosting system-level control sequences and programs, storing and displaying trends, coordinating time of day and calendar scheduling, and creating and displaying alarms. These functions are vital to the system, but by today’s technology standards they’re really f-ing dumb, right? In fact, the architecture hasn’t changed much since the 1980s when it looked like this:

The only difference is the webserver, the network, and the protocol. That’s 30+ years of progress! Perhaps the most sophisticated advancement from a building performance standpoint was the creation of trim-and-respond style supervisory control sequences allowing the system to “reset” formerly static setpoints based on actual heating and cooling demand. Let that sink in: it’s 2020 and our biggest automation advancement so far has been the addition of some if-then-else logic!

To make matters worse, if you’ve ever clicked around on one of these supervisory devices, you know these basic functions are quite difficult to use. We might be using IP addresses and HTML5, but the user experiences are still straight outta 1987. As Troy Harvey says,

Today, the buildings industry has a user experience crisis on its hands. We essentially hit “Peak UX” decades ago, and by some metrics we are in UX decline. A broad survey of “smart building” products shows that our technologies are actually asking more of us — more of our attention and time — not less. However, our professional and personal lives demand systems that “just work” — our buildings technologies should act as agents on our behalf, not vie for more of our limited resources. But in fact, we are more likely than ever before to be spammed by poor quality notifications, disruptive queries into the state of our comfort, and continuous data streams that we have little time to parse.

This point has been crystal clear to me since I was a commissioning agent for a brand new hospital that opened in 2017. The user experience on the brand new Siemens Desigo CC BAS was terrible—easily the most difficult software I’ve ever used. It was like asking my 85-year-old grandpa, who uses Spotify on his iPhone today, to instead use Napster or LimeWire to burn tracks to a CD like I did circa 1999 (shoutout to the 90s kids!). Sure, he could listen to his Lawrence Welk in both scenarios, but they’re entirely different experiences.

One-way overlay solutions have stepped up to fill some of the void left by this dumbness and poor UX for the last twelve years or so. For example, we’ve seen graphics overlays like DG Logik (founded in 2007) and data analytics overlays from SkySpark (2008) and KGS Buildings (2008) see solid adoption.

Now, in 2020, we’re seeing another wave of overlays. This time they’re designed to not only pull data from the BAS, but also push commands back down. This trend is supported by the fairly recent emergence of several enabling technologies:

• Open protocols like BACnet and integration platforms like Niagara allow standardized commands and overrides. Frankly, control sequences are no longer completely dependent on the old guard of BAS: OEMs, design engineers, and installers
• OT and Cellular infrastructure, such as cheap modems and pricing plans for IoT solutions, allow quick deployment of edge gateways, communication with intelligence in the cloud, and monthly fees that don’t kill the startup’s SaaS pricing model
• Speaking of edge gateways, cheap and customizable hardware like the raspberry pi bring the upfront cost of the overlay down significantly. And not-so-cheap, open-ish supervisory controllers (like a Niagara JACE) have now become prevalent.
• AWS and Azure have democratized cloud storage and computing and made it easy-ish to plug in advanced algorithms for learning, simulating, predicting, and optimizing systems
• Accurate, affordable, and richer Weather APIs support these algorithms with new data like cloud cover and cloud thickness

These technologies are enabling vendors to bring new sorts of intelligence to the marketplace and allowing researchers to dream up new ways to solve age-old energy efficiency problems. New opportunities are now knocking at the door. When I evaluate new technologies in our space, I’m grouping them under a smart building capability called Advanced Supervisory Control. Within that, there are two main categories:

1. ASC 1.0: Doing traditional supervisor functions better
2. ASC 2.0: New algorithms that are improving upon 1.0

For today, let’s dive into the first category: how can new technology help building owners do supervisory control better?

HOW NEW TECH IS IMPROVING TRADITIONAL SUPERVISORY CONTROLS

There is massive room for the performance improvement of existing supervisory functions. Where does the old way fall short and where can advanced supervisory control help? Let us count the ways.

In the traditional BAS model, the three S’s (setpoints, sequences, and schedules) can be overridden and then forgotten about for years. The BAS is often no help at reminding you to release the override. This is even worse across a portfolio of buildings. Different BAS brands, contractors, and O&M teams make it nearly impossible to keep things standardized and optimized across the whole fleet. Oh, and sometimes you have the added challenge of needing to be on-site with a proprietary programming tool to make changes (try that during a pandemic). Overlay software allows the owner to remotely standardize, consolidate, and maintain BAS-agnostic setpoints, sequences, and schedules at the portfolio level.

And as Facilio CEO Prabhu Ramachandran points out in an upcoming podcast episode, these new supervisory commands don’t need to happen automatically:

“It can happen automatically, but it can also be semi-automatic, where we trigger a workflow, somebody looks at it, and approves it. This can happen within a minute. And then they go and make these changes into the buildings, because sometimes, not all use cases customers want the automation.

They want some time for these tickets. So it could be like a supervisor getting a pop up message on their mobile app saying, Hey, this is what the system detected. The system can go and make these changes. Can I go ahead? Then they approve that, it goes and does stuff.”

If you think about it, this goes hand in hand with fault detection and diagnostics. The old method was this: the FDD software detected a supervisory control fault, such as a schedule override or a sequence that needed to be fixed, then the faults went on a prioritized list for the BAS contractor to fix in the field (if we were lucky). But what if there was a way to make the fix automatically or semi-automatically?

This is the approach taken by the developers of VOLTTRON at PNNL with their AIRCx algorithms (image source):

They target common retro-commissioning/retuning measures that have traditionally been the focus of 1-way overlay FDD rules:

These algorithms automatically identify the operational problem or an advanced control opportunity and report them to the building operator; or in some cases the algorithms have the ability to automatically self-correct the problem.

The measures that can be automatically detected and corrected include schedule adjustments, damper minimum flow adjustments, thermostat adjustments, as well as dynamic resets to static pressure, supply-air temperature, condenser chilled- and hot-water temperatures, and chilled- and hot-water differential pressure set points. For this project, fourteen re-tuning measures were selected for automation.

Another downfall of traditional controls is that supervisory functions are implemented in building system silos. Supervisory HVAC controls are separate from supervisory lighting controls. Elevators, access control, and the like are also all in separate silos. Vendors like Facilio and Cohesion IB, another upcoming podcast guest, are integrating supervisory control across previously siloed building systems. This opens up new use cases and makes each individual subsystem smarter.

Picture a world where all building systems are running off of one occupancy schedule, which is updated based on the actual real-time occupancy and the location of occupants. Or, consider a specific supervisory control sequence that is almost always poorly implemented: optimal start/stop. If it’s installed, which in itself is extremely rare, the old paradigm was to use the last 3 or so days of historical zone temperature data to decide how early to start the building up and how late to shut it down. Here’s what that looks like over a day and where the savings are:

But what if the control sequence wasn’t limited to just the historical data stored in the controller? New types of optimized start/stop can also consider months or years of historical data, the local weather forecast, occupancy and event calendars, time of use utility rates, and the output of machine learning algorithms to decide the truly optimal start/stop times. Prescriptive Data is a startup that comes to mind here—their start/stop strategy is covered well in this NYSERDA case study.

Finally, let’s talk about a traditional supervisory control method that happens as part of the commissioning process: functional performance testing. As Altura Associates wrote in the February 2016 edition of the ASHRAE Journal, the old way to do functional testing using supervisory control looked something like this:

(…) testing forces equipment into the test conditions to then observe and record equipment response. Test execution details are documented on written or electronic forms while a data logger or automation system records time-series trend- data for specific data points. These trends are then post- analyzed and performance is compared to the expected response to determine acceptance.

The forced conditions can be achieved through the use of automation system overrides by the controls programmer (equipment schedule adjustment, speed/ position commands, temperature/pressure sensor overrides) or field interventions (occupant thermostat resets, in-hand valve/window/sash position adjustments, etc.). Active testing is generally required to comprehensively test the SOO in a timely manner.

Time-consuming and sometimes cost-prohibitive... and the new way looks something like this:

With (analytics including in the commissioning process), the step-by-step test procedures for each piece of equipment are still included. However, the execution of these tests is transformed from a manually executed, time-intensive process to an automated process that can run a virtually limitless sequence of override steps on multiple equipment at once, without the need for human intervention.

This process unlocks an unparalleled degree of scalability, repeatability, and schedule flexibility in the testing regime that is impossible with a traditional approach.

Since that article was written, ASHRAE Guideline 36 has added provisions throughout the document for including testing/commissioning overrides for all relevant control points, so that commissioning can be performed automatically and verified via FDD.

These sequences and practices might seem like small (and nerdy) performance details, but this is where the rubber hits the road for building performance. In my opinion, BAS vendors have shown themselves inept at making these small details easy for building operators. Advanced supervisory control vendors can help fill this void.

In the next installment of this series, we’ll dive into more advanced types of supervisory control, including automated systems optimization (ASO) and intelligent load control (ILC). We’ll walk through the potential implications on existing solutions and what this means for energy efficiency in the built environment. Finally, we’ll summarize the whole thing with how the right strategy could help our collective results.

Thanks for reading!