courtesy of Kuprevich -
Dreamstime Photo Kuprevich Dreamstime Com

Can AI or automation fix your building inefficiencies?

March 14, 2023
A new tool from the Lawrence Berkeley National Laboratory can help automate fault detection and diagnostics software, minimizing the need for human interaction and resulting in greater efficiencies and carbon reduction.

Building automation and energy management systems are becoming increasingly necessary in facility management, enabling owners and operators to realize greater efficiencies, flexibility, and even resilience in the face of climate change, which has a direct impact on building operations. But along with these sophisticated tools come increased complexity and the introduction of errors—often at the expense of the efficiencies these technologies afford.

As a result, building fault detection and diagnostics (FDD) technologies have grown in popularity, saving owners millions of dollars annually in their buildings, often with less than two-year paybacks. FDD tools automate the process of detecting faults and suboptimal performance of HVAC systems to help diagnose potential causes. They are typically layered on top of an existing building automation system (BAS), according to a February 2022 report by the Lawrence Berkeley National Laboratory (LBNL), in Berkeley, Calif.

However, while commercial FDD tools might appear to be a panacea for attaining energy efficiencies and, by extension, reducing carbon emissions, one hiccup remains: the need for human resolution. The LBNL report noted that “the need for human intervention to fix faults once they are identified often results in delay or inaction, causing additional operations and maintenance costs or deteriorating comfort conditions.”

In other words, a building’s efficiency, energy savings, and carbon emissions are still very much dependent upon people.

Automated fault correction for commercial FDD applications is showing tremendous promise in terms of closing the loop between passive diagnostics and active controls, according to LBNL. These tools can integrate artificial intelligence (AI) in some cases to conduct predictive maintenance, giving facility managers greater flexibility and freedom than ever before.

The problem: Controls are error-prone

Buildings use 70% of electricity in the U.S. and account for nearly 33% of carbon emissions from fuel combustion globally, representing roughly 20% of total greenhouse gas emissions. As such, buildings must become increasingly efficient and anticipate problems with their systems before problems occur.

However, FDD tools aren’t foolproof—not by a long shot. In fact, studies estimate that traditional equipment faults and control problems significantly increase greenhouse gas emissions and energy bills, to the tune of $17 billion and 90 million metric tons of CO2 equivalent annually according to LBNL and the U.S. Department of Energy (DOE).

“It turns out that the opportunities that we encounter most frequently that have some of the biggest energy impacts can be addressed through automated fault correction and control optimization,” says Jessica Granderson, staff scientist and interim director of the Building Technology and Urban Systems Division at LBNL.

These opportunities for improving energy performance include:

  • Optimized economizer high lockout temperature setpoint.
  • Correction of incorrectly programmed HVAC schedules.
  • Release of unnecessary control overrides.
  • Correction of biased temperature sensors.
  • Automated loop tuning.
  • Implementation of best practice reset strategies.
  • Mitigation of rogue zones.
  • Optimized zone temperature setpoint setback.

“We are now working to expand our suite of optimal fault-free control solutions to a wider set of FDD partners, and [include] additional strategies such as automated commissioning/functional testing and demand flexibility,” Granderson says.

The solution: How automation can improve FDD outcomes

In 2016, LBNL partnered with the DOE and various industry partners to launch the Smart Energy Analytics Campaign, a public–private partnership that has resulted in the largest data set on building analytics, costs, benefits, and uses. In the years since, LBNL has also worked with the nation’s leading market providers of FDD technology to extend the state of the art beyond what was previously available. Granderson says her team has developed and implemented additional programming capability to correct faults automaticallyonce they are identified by existing FDD software.

In a 2020 field study with two end-user partners, LBNL developed and deployed a set of seven fault-correction algorithms for HVAC systems that were tested in real buildings using existing BAS vendor platforms. The variables corrected by the algorithms span schedules, setpoints, sensor readings, commands, heating/cooling requests, and proportional, integral, derivative (PID) parameters.

Historically, FDD technologies have integrated with building automation systems, ingesting operational data for system and equipment operations in a “read only” format. “The first thing we did was enhance that interface to enable the FDD system to also ‘write’ back commands to the BAS,” Granderson explained.

The team then developed a library of engineering logic that defines how various controls related problems can be fixed by modifying control system parameters that are commonly accessible through the BACnet protocol.

Finally, the team integrated the corrective logic into the FDD platform and operator-facing user interface. Now, once the FDD system detects and diagnoses a fault, an operator is notified of the problem, as well as a recommended corrective action. Upon operator approval, the corrective action is executed and the fault is resolved.

Granderson offers the following example: A zone temperature setpoint that is too aggressive may be flagged for operator attention and correction with the message, “The cooling setpoint for this zone is 66 degrees, lower than recommended. Would you like to return the setpoint to the recommended 68 degrees?” With operator approval, the FDD system is able to write the corrected 68-degree F setpoint back to the zone controller through its interface with the BAS. Once this action is complete, the fault is resolved, and the FDD system returns to problem detection and diagnosis.

In addition to fault correction, LBNL has extended FDD system capabilities into control optimization. For starters, it has developed and tested methods to implement best practice trim and respond reset strategies for air handling unit static pressure and supply air temperature, based on ASHRAE Guideline 36: High-Performance Sequences of Operation for HVAC Systems. Within those solutions, LBNL’s technology is suppressing “rogue” zones that drive up energy use with unsatisfiable heating or cooling requests.

Though LBNL is currently not using AI in the fault correction methods it has developed, Granderson notes, some FDD providers are using AI in certain portions of their technology stack.

Building IQ, based in Sydney and Fargo, N.D., has introduced what it calls an Outcome-based Fault Detection (OFD) service that combines AI, energy analytics, and human expertise to overcome the shortcomings of many FDD services. “Outcome-based Fault Detection is a comprehensive solution that takes fault detection in a better and broader direction,” said the company's then-president and CEO Michael Nark in a June 2018 press release. “It does this by embracing the key role played by facility experts and augments it with machine learning and cutting-edge artificial intelligence. OFD works whether data is good or spotty and leverages machine learning to take the burden of data analytics into the cloud. The result is building operators do not have to waste valuable time and resources scouring through tables of hundreds of daily faults. Instead, with OFD, operators can focus on the things that truly need fixing, their tenants, and the bottom line.”

Benefits of automated FDD systems

“There is a surprising level of hidden inefficiency in our buildings,” Granderson says. “Automated control systems maintain temperatures and humidity levels and keep systems running for occupant comfort. But they routinely fall out of tune, may not turn off after hours, or may use settings that otherwise waste energy and drive up costs and greenhouse gas emissions.”

Automated FDD technologies can continuously analyze operational data to surface problems for building operators and energy managers, she says, noting that “the benefits are tremendous. Our work has shown that organizations that use FDD systems in their portfolios save 9% on average, with two-year paybacks.” Adding in automated fault correction capabilities extends the benefits even further, she continues. Rather than waiting weeks or months for problems to be fixed, they can be resolved in a matter of hours, and valuable staff expertise can be directed to the hardest problems.

“Moreover, the ability to write control commands back to the BAS also allows us to implement supervisory control optimization,” she says. “Provision of supervisory optimized control through the FDD systems enables scalable implementation across diverse vintages and makes of BAS without expensive upgrades that might be needed with more traditional approaches that directly modify the BAS programming.”

Automated- and AI-based BAS and BEMS solutions are already being adopted across the commercial building sector globally. For example, ABB’s Ability BE Sustainable with Efficiency AI currently manages more than 275 buildings, totaling more than 100 million square feet. Collectively, these installations save more than 1 million metric tons of CO2 each year, all by leveraging building automation investments already made.

The future of smart buildings is continual improvement

Good data is the foundation of building automation and management systems, and the better data that’s fed into your energy management and information systems, the better. As FDD tools and automated software evolve, the implementation, scalability, and reliability of smart buildings will continue to improve—and building owners and facility managers who are looking to start the journey will have tools at their disposal.

In October 2020, LBNL published an application showcase that helps stakeholders understand how to get started, highlights best practices among Smart Energy Analytics Campaign participants, and provides examples of innovation happening in the industry.

“We have tested these new capabilities across a number of buildings and BAS offerings," Granderson says. "Results to date suggest that they can be scaled across diverse controllers, with modest additional development and implementation lift from the FDD provider. As these emerging technology capabilities are made available by its partners through our product features or modules, LBNL will be able to track the incremental costs relative to traditional FDD systems.

“This is all quite new and still maturing, but what’s so exciting about this work it that it shows us the future of smart buildings. We are asking more and more of our buildings—that they be net-zero greenhouse gas emitters, integrate across a growing number of distributed energy resources, and provide a healthy and comfortable indoor environment, all while coordinating with a renewable grid.

“The only way to accomplish that at scale is to harness the modern software-based infrastructure that FDD and other smart building software offers. It gives us the pipeline to continuously ‘push’ improved control and analytics solutions into our buildings.”

For more news, projects, and profiles in the smart buildings ecosystem, subscribe to the SBT newsletter and follow us on LinkedInTwitter, and Facebook.

About the Author

Robert Nieminen | Chief Content Director

Robert Nieminen is the Chief Content Director of Architectural Products, BUILDINGS and i+s, sister publications of Smart Buildings Technology. He is an award-winning writer with more than 20 years of experience reporting on the architecture and design industry.

Voice your opinion!

To join the conversation, and become an exclusive member of Buildings, create an account today!

Sponsored Recommendations

Handwashing Survey Research Drives Industry Solutions

Sharing well researched data on consumer likes and dislikes in public bathrooms, this guide presents handwashing public restroom hygiene trends and survey insights for facility...

Building Security & Technology Series: Webinar 2 - Building Safety Design & Technologies

Date: May 15, 2024Time: 1:00 PM EDT / 12:00 PM CDT / 10:00 AM PDT / 5:00 PM GMTDuration: 1 Hour eachGold Sponsors: Genetec, ISSSilver Sponsors: EVOLV, OpenEye ...

Building Security & Technology Series: Webinar 1 - Emergency Preparedness

Date: May 8, 2024Time: 1:00 PM EDT / 12:00 PM CDT / 10:00 AM PDT / 5:00 PM GMTDuration: 1 Hour eachGold Sponsors: Genetec, ISSSilver Sponsors: EVOLV, OpenEye Register...