The Hidden Power of Default Setpoints: How Stuck Sensors Quietly Wreck Your BAS

Most technicians look for big, obvious failures. A fan that will not start. A valve that will not move. A chiller that keeps short cycling. Those are loud problems. You can hear them and you can see them.

The most expensive problems, though, are often the quiet ones. The ones that never alarm. The ones that look exactly like “normal” every time you open the graphic. If you have ever stared at a supply air temperature that sits at a believable number all day, or a humidity reading that never seems to budge, you know the feeling. Something looks fine, yet the building is not behaving.

A common reason is simple and sneaky. A sensor failed or dropped offline, the controller substituted a default setpoint or a default input value, and your system kept running with a number that looks right but is not real. If your site does not use fault detection and diagnostics, or the FDD rules are not implemented well, there is a good chance no one will tell you. The default value hides in plain sight and the building pays for it, day after day.

A quick story from the field

I learned this lesson in a lab building where morning humidity kept drifting and nobody could explain why. The space humidity on the BAS was a clean, respectable 45 percent all day, every day. The graph was perfectly flat. It looked like a marketing brochure. The occupants, however, kept complaining of sticky rooms when occupancy ramped up.

We checked coils and valves. We checked schedules. We tuned loops. Nothing moved the needle.

Then I looked at the trend more carefully. Not the value itself, the shape of the value. Zero wiggle. No noise at all. No tiny rises when the doors opened. No slow drifts with load changes. The number was not just stable. It was frozen. We dug into the controller and found the smoking gun. The humidity sensor had dropped offline months earlier and the program had been substituting the BAS default setpoint value every second since. It was set to a believable number, so no one noticed. The sequence never alarmed, the operators trusted the “good” number, and the building ran with bad assumptions that cost comfort and energy every single day.

Once we replaced the sensor and added a few simple checks, the “perfect” line turned into a living signal again. You could see the room breathe with people, doors, and weather. Comfort improved, reheat dropped, and the help desk went quiet.

That building taught me something I never forgot. The most dangerous number in your BAS is often the one that looks normal.

What “default” really means inside a BAS

It helps to separate two things people lump together.

1. BAS Default setpoint

   A controller often ships with a starting setpoint. Space temperature might default to 72 F, discharge air to 55 F, static pressure to 1.5 in. w.c. Installers change these during commissioning, but many sites keep factory defaults for entire buildings because “it worked fine during startup.” These defaults are not evil, yet they are rarely optimal for your climate, your loads, or your equipment.

   
   

2. BAS Default input value

   When a sensor fails, some controllers substitute a hardcoded value so downstream logic does not crash. A return air sensor might fail to a fixed 72 F. A pressure sensor might fail to 0.0 or to a nominal mid-range. That substitution can be useful for safety, yet it is dangerous if the value looks realistic. Operators trust the number, loops behave as if the world is normal, and you never get a clear alarm.

In practice, both defaults can hide inside your site at the same time. A VAV box can be controlling to a default discharge air setpoint while its space sensor is flatlined at a default input value. The loop “works,” the graphic looks calm, and energy disappears.

Why sites miss stuck sensors and default values

• The number looks right

  A failed sensor that reports 72 F in a space that is often near 72 F will not look suspicious. You need motion, not just position.

  
  

• Alarms focus on limits, not behavior

  Typical alarms look for high or low thresholds. A flatline at 72 F will never cross either threshold. Without a rule for “no change over time,” you never get a page.

  
  

• Trend windows are too short

  If you only trend 24 hours, a flat signal can still look fine. The trick is catching the lack of natural variation across days, not just minutes.

  
  

• Graphics that hide relationships

  If your graphic shows only the sensor value and not the actuator that should affect it, you miss the “this valve is moving but the temperature never moves” clue.

  
  

• No FDD, or FDD without the right rules

  Out-of-the-box analytics often look for simple faults. They do not always include no-change, no-variance, or “implausible relationship” rules unless you build them.

How to spot a stuck sensor fast

You do not need a data science degree, just a few habits and a little curiosity.

1. Look for natural noise

   Real signals wiggle. Space temperature should show little bumps when doors open. Return temperature should nudge upward as people fill a room. Static pressure should dance when VAVs throttle. If a value is ruler-flat for hours during occupied time, investigate.

2. Compare against something that should move it

   If the chilled water valve travels from 10 percent to 90 percent and the discharge temperature does not move at all, either the valve is not connected to anything or the temperature sensor is lying. The relationship is the point, not the absolute number.

3. Use rolling change tests

   Pull a seven-day trend. Scan for the smallest visible step. If you cannot find one, that is your red flag. You can bake this into the BAS: “If this value does not change at least X units in Y minutes during occupied time, set a fault.”

4. Compare peer signals

   Two adjacent VAV zones should not have identical space temperatures for hours. Two nearly identical air handlers should not show perfect static pressure twins all day. Nature creates small differences. Perfect twins rarely exist in the field.

5. Use the weather and the clock

   Outside air temperature cycles daily. If your return air signal ignores the day and night pattern, start asking why. The clock is the easiest model you own. Your building should respond to it.

The physics cross-check you can run anywhere

Technicians often forget the cheap tests. Here are simple comparisons that expose lies.

• Mixed air sanity test

  Mixed air temperature sits between return and outside air temperature. If MAT is outside that range for more than a minute, a sensor is wrong or a damper is lying.

  
  

• Enthalpy and humidity pairing

  If space humidity stays flat while outside humidity swings and coils change state, your humidity sensor may be stuck. Humidity is a noisy signal in occupied spaces. Flat is not normal.

  
  

• Fan command versus static pressure

  If the fan cycles or ramps and static pressure stays flat, the transducer might be dead or held at a default.

  
  

• Reheat versus space temperature

  If reheat opens and space temperature ignores it for long periods, the space sensor may be stale.

Use these like a field sobriety test. You are not proving guilt. You are deciding what to check next.

Where defaults hide most often

• Space temperature sensors in offices with variable occupancy

• Return air temperature in big open spaces

• Mixed air and coil sensors in units with economizers

• Humidity sensors in labs, museums, health care, and data rooms

• Static pressure transmitters on long, leaky duct runs

• Flow meters that revert to a nominal value when signal is noisy

If your site has any of these and no active “no-change” rules, start there.

The Stuck Sensor Detection Playbook

Use this as a repeatable routine. It works even without analytics software.

Step 1: Pull longer trends

Look at a full week. If you can, look at two. Use occupied windows as your baseline. You want to see small, natural variations and daily rhythm.

Step 2: Add the right companions to the chart

Never trend a value alone. Pair it with the actuator that should influence it and a nearby reference. For a discharge temperature, include valve position and supply water temperature. For a room temperature, include reheat command and occupancy.

Step 3: Scan for variance

Your eyes are good at this. If the line is too clean, treat it as suspicious. Mark the flattest signals and the ones that seem to ignore actuation.

Step 4: Run the physics checks

Mixed air between return and outside. Space humidity that follows occupancy and coil activity. Static pressure that responds to fan speed. Anything that violates these deserves a meter on the wall and a second look.

Step 5: Go touch it

Measure with a handheld. Verify wiring. Confirm the sensor model and range. Do not assume the right part was installed or that it is still healthy.

Step 6: Fix and protect

Replace or rewire the bad part, then add rules and guardrails so you do not get fooled again. The next section shows how.

Program guardrails that catch defaults and flatlines

Add these patterns to your sequences. Most platforms support them, from Niagara and ALC to older Johnson, Siemens, Delta, and even legacy Staefa.

1. Heartbeat timers on critical sensors

   Add a timer that resets every time the value changes by a small amount. If the timer exceeds a window during occupied hours, raise a fault, tag the value as unreliable, and switch to a safe fallback mode that is obviously a fallback. Do not hide it.

Example logic idea:

If change in space temperature is less than 0.1 F for 45 minutes while reheat or cooling is active, flag “Space Temp Stuck.”

2. No-change rules tied to modes

   Use larger time windows at night and tighter ones during occupancy. A flat signal during unoccupied time may be normal. A flat signal during active control is not.

3. Plausibility windows

   Set believable ranges for each sensor. A return air sensor that reports colder than outside air for long periods needs attention. A mixed air sensor outside the return to outside range is suspect by definition.

4. Quality flags and reliability

   BACnet points include a reliability property and status flags. Make use of them. If the controller loses the sensor and substitutes a default, mark the point bad and surface that status on the graphic, the alarm list, and your trend. Never let a substituted value look identical to a healthy value.

5. Last-value hold with a timer

   If you must hold the last good value when a sensor goes down, limit the hold time. After the timer expires, switch to a safe mode that changes the operator’s screen and raises a clear alarm.

6. Voting sensors for critical signals

   For high-risk areas, use two sensors and flag disagreement. You do not need expensive triplex systems to gain confidence. Two is often enough to avoid being fooled by a single flatline.

7. Automatic cross-checks

   Create a light analytics layer with simple rules, even without a full FDD engine. Examples:

   "If fan starts, static pressure must rise within 20 seconds.”

   “If reheat valve opens past 50 percent for 10 minutes, space temperature must rise at least 0.5 F.”

   “If economizer opens beyond 60 percent with cooler outside air, mixed air must move toward outside air within two minutes.”

These are not complex. They catch a huge fraction of bad data quickly.

FDD and analytics help, yet setup is everything

Fault detection tools can watch hundreds of points for you, yet they are only as good as the rules you implement and the tags you provide. Two tips make them powerful for stuck sensors:

• Include no-change over time rules, not just high and low limits.

• Feed the engine the relationships it needs, not just raw points. Rules work best when they know which actuator moves which temperature.

If you do not have an FDD platform, you can mimic the most valuable rules right in the BAS using the guardrails above.

Where default setpoints themselves drain money

Not every default problem involves a failed sensor. Many buildings run for years on factory setpoints that were never revisited after startup. Easy places to look:

• Static pressure setpoints that are higher than needed. Lowering static pressure a few tenths with a trim and respond strategy often saves more than fancy capital upgrades.

• Supply air temperature setpoints that are too cold in mild weather. A reset schedule that floats warmer during low load cuts reheat and chiller runtime.

• Chilled water or hot water setpoints that ignore outdoor air or building load. Seasonal or load-based resets are simple and effective.

When you audit a site, ask one question. “Which of these setpoints are still on the day-one defaults?” If no one knows, assume the answer is “most of them.”

Prioritize like a pro

There are endless sensors and setpoints. Start where impact is largest and risk is smallest.

1. Air handlers

   Space comfort, ventilation, and energy converge here. Focus on return, mixed, discharge temperatures, humidity where present, damper positions, and static pressure.

   
   

2. Critical rooms

   Labs, data rooms, health care spaces, archival areas. Humidity and pressure sensors in these spaces must never lie. Add voting or frequent verification.

   
   

3. Hydronic plants

   Supply and return temperatures and flow meters. Flow devices that fail to a default value can mask heat exchanger or pump issues for months.

   
   

4. Representative VAVs

   Pick one per riser or per floor and verify that its space sensor and reheat response show life during occupied periods.

Training the team to see the story

You can catch a stuck sensor with math or with good eyes. Teach your team the same three habits.

• Always trend with cause and effect on the same chart. Sensor plus actuator.

• Always look for noise and rhythm. Real data breathes with people and weather.

• Always chase a signal that is too perfect. It is easier to fake perfection than reality.

Once your team sees data as a living thing, flatlines pop off the screen.

The one-hour weekly routine

You can keep buildings honest with a small, consistent ritual.

• Pick one air handler and one critical room per week.

• Pull the last seven days of data.

• Add the actuator pair to each sensor.

• Scan for variance and relationship.

• If anything looks perfect for too long, schedule a meter check.

• After any fix, add a guardrail so you never get fooled the same way twice.

This costs less than a coffee break and pays like a major project.

Frequently asked questions

Is a flat overnight line always bad?

No. During unoccupied periods, signals can settle. Your watch window is when the building is breathing, usually the first two to four hours of occupancy.

Should I alarm every flatline?

Alarm behavior, not just value. Tie no-change rules to modes. Occupied with active control plus no change is suspicious. Unoccupied plus no change is usually fine.

What about noisy sensors?

Add minimum change thresholds so electrical noise does not reset heartbeat timers. The goal is to see real motion, not chatter.

Is “last value hold” ever acceptable?

Yes, for safety or short outages, yet use a timer and a bright operator cue. After a set time, switch to a safe mode that is clearly visible and raises a fault.

Do I need two sensors everywhere?

No. Use voting for high-risk spaces and the physics checks elsewhere. Good rules catch most problems without extra hardware.

The bigger lesson

Default setpoints and default input values are not villains. They are tools. The danger appears when they hide. You avoid that by building graphics that show relationships, by trending for motion and rhythm, by adding small pieces of logic that prefer truth over calm numbers, and by training everyone to be suspicious of perfection.

If you start this week with one air handler and one critical room, you will find at least one number that has been lying to you. Fix it, protect it, then repeat. The savings compound. So does the trust.

The best part is simple. You do not need to rip anything out. You only need to see what is already there and refuse to be fooled by a beautiful, believable lie.

Book Recommendation

Control Systems for Heating, Ventilating, and Air Conditioning

Written by: Roger W. Haines & Douglas C. l-Iittle

Buy it: Hardcover | Paperback

Best for: Translating HVAC control theory into sequences that actually work in real buildings, great for commissioning and troubleshooting.

What you’ll get: Clear explanations of control strategies, sensor behavior, loop tuning, economizer logic, and reset schedules

How to use it: This week (45–60 min): Read the chapters on sensors, loops, and sequences; audit one AHU against its written sequence and note mismatches.

Field drill:

Draw a simple state diagram for supply fan/heat/cool; verify each sensor placement and unit/scale; check reset logic vs trend data.

Pro tip:

Convert your most-used sequences into a one-page checklist and tape it inside the panel door.

Gear Recommendation

Fluke 971 Temperature Humidity Meter

Buy it: Fluke 971 Meter

Best for: Fast room spot-checks to verify what your BAS says, perfect for comfort complaints and sensor validation.

What it is: A handheld temperature/RH meter with quick response and backlit display; ideal for comparing live room conditions against BAS points.

How to use it: Let it equilibrate 30–60 seconds; hold at chest height away from your body; sample center of room and near the return; avoid exhaled breath; log readings.

Field drill: Do a 5-room audit, record BAS vs meter. Flag >±1.5°F or >±5% RH and build a correction/replace list.

Pro tip: Keep fresh batteries and store in a case; annual check with a 75% salt solution keeps RH accuracy honest.