What might the crystal ball say about fault detection diagnostics?

Alarms are a common part of nearly any BAS. Building operators are expected to monitor and respond to alarms. Although responding to alarms is an important part of building operations, it is generally a reactive process. An alarm event indicates that a problem has already occurred. For example, an alarm may be triggered if the static pressure across a filter within an AHU reached the threshold. By the time this occurred, it is quite likely that indoor environmental quality has been negatively affected for the last several weeks or months, not to mention any increase in fan energy consumption.


As building operations practices transition from reactive to proactive, it is necessary to consider if responding to alarms is really the most efficient way to operate buildings, especially high-performance buildings. What if predictive algorithms could be embedded within BAS to help operators proactively detect when an alarm event may be triggered in the future? This could allow the operator to take action to prevent the alarm event from occurring to begin with.

The concept of predictive algorithms is not that futuristic. Predictive control algorithms, also called fault detection diagnostics (FDD), have been an area of research for many years and have been implemented within some packaged equipment, as well as within some standalone software programs. The concept of predictive algorithms is not that futuristic. Packaged equipment with FDD algorithms ranges from local chiller controllers to packaged rooftop controllers. FDD algorithms within controllers could include, but are not limited, to:
  • Identifying when a sensor has failed

  • Determining when a control loop is not properly tuned

  • Determining when a valve, damper or actuator is broken, stuck, or leaking

  • Troubleshooting of equipment when improperly installed
Uses of FDD algorithms within standalone software programs include, but are not limited to, assisting with commissioning of HVAC systems and chiller plant optimization.

Although the use of FDD in rooftop units are estimated to help reduce energy consumption by 10% or more, the largest barrier to implementation is that FDD applications often require fault thresholds and statistical parameters for each unit to be determined on a case-by-case basis. This can make implementation time consuming and expensive.


Ideally, fault detection algorithms could easily be incorporated as an integrated part of BAS. The use of open protocols, factory integrated controls, and new software applications will help to make this a reality. As FDD algorithms occasionally found today in packaged controllers are incorporated into BAS, more advanced predictive control strategies may be able to be successfully incorporated as well. These strategies could include:
  • Supervisory control using active and passive thermal storage

  • Weather forecast driven control strategies

  • Electric power load forecasting
As more building owners and facility managers seek to have high-performance buildings and transition to proactive management practices, FDD and predictive control algorithms will become of higher interest. However, in order for FDD and predictive control to become more than research concepts the market (engineers, building owners, facility managers, and building operators) will need to ask for it.

This month’s column is courtesy of our associate Angela Lewis. She can be contacted atangela@buildingintelligengegroup.com.ES