by Lyn Gomes, PE, CCP, LEED AP
Senior Commissioning/MEP Coordinator, DPR Construction

This is distilled from my own experience as well as input from the authors of the Building Commissioning Handbook, 3rd edition (Bjornskov & Stum), a steam control system reference (Spirax Sarco, a classic that’s now available online) and two former coworkers who analyze control systems (Raphael Vitti and Lincoln Harmer).

What is an untuned loop or hunting?

  • Untuned loop:
    • A system output that exhibits an underdamped response
    • Crosses above/below the setpoint more than 3-5+ times in 5-15+ minutes (these parameters may differ depending on the speed of the system or the needs of the process), or
    • Takes longer than three cycles to settle within setpoint deadband
    • Overshoots/undershoots the setpoint by 5% (depends on the system and process needs)
    • Changes so slowly that the output never reaches setpoint (strictly speaking, this is tuned too much and is considered an overdamped response)
  • Hunting:
    • Inputs that cause continuous adjustments over hours/days, but never results in the output reaching setpoint
    • Also: never maintains setpoint within deadband or tolerance

Additional resource: Figure 5.4.4 of Spirax Sarco’s basic control theory Module 4.

It’s important to remember that for systems with nested control loops, the output that you’re trying to tune could be:

  • The primary loop: the thing you’re trying to control (e.g. space temperature in the picture at right)
  • An upstream/secondary/nested loop: an intermediate output (e.g. discharge air temperature or reheat valve position)

Note: There may be 2, 3, 4, or more nested control loops.

How do you detect a poorly tuned loop or hunting?

Trend with sufficient intervals to detect hunting. The Building Commissioning Handbook (page 144) recommends 1-2 minute intervals, but notes BAS storage or network traffic speed capabilities might prevent this.  In that case, 5 minute intervals may be sufficient or trending for shorter intervals (a few hours at smaller intervals rather than days or weeks). Sometimes COV trends can be appropriate. Trending at too long of an interval will mask poor system control.
The best times for review are after programming and before/during/after functional testing. A detailed review (either manual or automated) of short interval trend data after functional testing (and the system has gone through a first round of tuning) is the gold standard.

How do you fix it?

Best practice is to tune the primary loop first, then tune the loops upstream. Failure to tune the upstream loops will result in the system becoming unstable in other modes of operation (i.e., under load vs. lightly loaded; unoccupied vs occupied). Basic Control Theory Module 5 provides good troubleshooting advice for diagnosing and adjusting PID tuning parameters.

Keep in mind that loop tuning is an art. Automated loop tuning algorithms may provide a great head start, but they are rarely foolproof and systems usually require manual tuning by someone with experience. It may take several rounds of manual adjustment to make sure all the loops (primary and secondary) are tuned and stay tuned under all operating scenarios. 

In the big picture, untuned or hunting loops lead to equipment failure, chiefly from fatigue (e.g., valve cycling), and poor control. We tune them to keep the occupants/processes happy and have an efficient, stable system. A well-tuned loop is a thing of beauty (truly!) and a statement of the talent and professionalism of the folks that did the work.