TRENDING


Insider Reports



QUICK Sign-up

New Content Updates
Educational Webcast Alerts
Building Products/Technology Notices
Access Exclusive Member Content


All fields are required.




Building Operating Management
Data PAGE Fault Detection and Diagnostics: How To Find Energy-Wasting Problems Understanding The Human Element of Fault Detection and Diagnostics

Understanding The Human Element of Fault Detection and Diagnostics

Understanding The Human Element of Fault Detection and DiagnosticsFDD doesn't actually fix anything itself. A competent facilities team is still essential for performing corrective actions based on the data the FDD provides.

By Michael Brusic September 2017 - Building Automation   Article Use Policy

The depth of analysis that can be built into a rule-based system is almost limitless and such a system may diagnose a wide range of fault conditions, but the open-ended nature of FDD may also pose its biggest challenges.

“FDD only provides information — it doesn’t fix anything,” says Stephen Samouhos, partner and co-founder at KGS Buildings. “You need a facilities team that wants to improve things and is going to take action.” That means a staff that understands the equipment they operate, the FDD system interface, how to narrow down a list of potential diagnoses, and how to use information from the FDD system to help get management approval to make repairs.

David Unger, CEO of Sentient Buildings, makes the point even more explicitly. “Training is the crux of everything we do,” he says. “It can’t just happen at the end of the project. We’ve built it into our project management plan and start as soon as we deploy. You have to give the operator some time to learn, then do a second or third or fourth training, and so on.”

Failure to properly involve building operators from the start may lead to a lack of trust in FDD results. Most operators are familiar with the concept of alarm fatigue: page upon page of nuisance alarms created because the system is not properly configured. FDD systems may suffer the same problem. By capturing operator knowledge during the design and commissioning phase of an FDD project, such problems may be addressed in ways that have benefits beyond the software. “It’s usually a good experience for the customer as well,” says Samouhos. “They often don’t have an equipment list, don’t have a written sequence of operations. It’s like spring cleaning.”

Technical challenges may also arise. Oftentimes implementing FDD is pursued as a means of adding functionality to an aging BAS and deferring investment in its replacement. Older hardware may nominally support BACnet or LON protocols, but often has quirks in its protocol implementation. Worse still are proprietary protocols; integrating FDD on a proprietary network may involve costly per-point license fees and frustrating, trouble-prone protocol conversion. Newer hardware and software tends to work better but is still far from trouble-free.

Quantification of system failures is an oft-touted feature of FDD platforms that may be difficult to realize in practice. The ramifications of faults (such as the overridden supply fan in our VAV example) have complex consequences. Their associated energy waste has a significant uncertainty attached to it when calculated in a detailed energy model, much less by a pre-set algorithm. Compounding the problem is the issue of differing utility rates and tariffs. It would be impossible for the author of a FDD software platform to include the rates and tariff structures of every utility in every part of the world where the FDD software may be deployed. Failure to properly model such costs may, however, can result in cost savings calculations that are off by 50 percent or more. Unless the system has been tailored to a specific utility tariff, faults may be best prioritized by energy waste instead of cost, and even those results should be taken with a grain of salt.

Given these challenges and the often time-consuming process of setting up and tuning FDD, it’s not surprising that many vendors offer an ongoing service component with their software. A service agreement may help spread the cost and effort of FDD configuration over a longer period of time, while building trust in the system from operators and management. After the first six to 12 months of FDD deployment, the service may scale back or cease entirely, once operators become comfortable interpreting the results of the system. BAS vendors that provide FDD as an optional add-on software and service package may provide far more value than possible under their base BAS service contract. Purchasing consulting and service together with the FDD software adds cost, but may provide cheap insurance against a FDD system that generates too many results, wrong results, and is then ignored after six months.

FDD saves the day

When the right steps are taken, FDD can be a huge success, speeding the diagnosis of subtle problems that might otherwise have taken months or years to discover. Samouhos recalls a university facility at which FDD was deployed shortly after construction: “It was a newer building, and they constantly had problems with spaces overheating. They couldn’t explain what was going on. FDD quickly identified that the reheat valves weren’t holding because the close-in pressure of the valves was too low for the system pressure. In fact, because the valves couldn’t fully close, they were pitting and the high water velocity was destroying them.” With the diagnosis in hand, the facility saved substantial money on heating energy and valve replacements.

As FDD becomes a more regular add-on or feature of BAS, and as the costs of hardware and data storage continue to decrease, FDD may see wider and deeper deployment. We are likely to see the addition of considerably more instrumentation on existing equipment, allowing more sophisticated detection of faults and evaluation of equipment health. A condenser water pump in a commercial building may, for example, be monitored for changes in bearing vibration spectra, bearing and seal temperatures, or noise induced by cavitation. While a standard practice in industrial facilities, that FDD feature is only now becoming cost effective in commercial, institutional, and residential facilities.

Another much anticipated future direction is machine-learning, which uses statistical methods and large volumes of data to build computer models that may adapt and react to changing conditions. “The industry trend that we’re seeing is moving beyond analytics and to machine learning” says Unger. “You can not only evaluate past performance but predict future performance.” Such systems will also incorporate data external to the facility, such as weather forecasts and real-time energy pricing.

As FDD provides increasingly complex automated analysis of increasingly large volumes of data, its core challenges remain the same and may even grow in magnitude. Ultimately, these systems serve human operators: They don’t replace them. Operator training and trust, and the ability of FDD software to provide targeted information with explanatory context, will continue to determine the success or failure of future FDD systems.

Mike Brusic is a senior energy engineer at Bright Power. He provides technical direction to the engineering team as well as consulting on projects across the company’s spectrum of services, including cogeneration, HVAC, building envelope, and controls, and works closely with the firm’s energy management clients. He can be reached at mbrusic@brightpower.com.

Email comments to edward.sullivan@tradepress.com.

 


Continue Reading: Data

Fault Detection and Diagnostics: How To Find Energy-Wasting Problems

Understanding The Human Element of Fault Detection and Diagnostics



Comments

Find us on Google+