If the measured sensor output deviates from the model output beyond a threshold, what should you do?

• A. Ignore the deviation to avoid alarms.
• B. Increase loop gain until the outputs match.
• C. Trigger fault detection when the discrepancy exceeds a threshold.
• D. Recalibrate all sensors immediately and revalidate.

Answer: (C)

When a measured sensor output diverges from what the model predicts by more than a set threshold, the system should trigger fault detection. This approach uses the difference between actual readings and expected behavior (the residual) to identify potential sensor faults, drifts, or model mismatches. If the discrepancy crosses the threshold, it flags a fault so the control system can respond safely: raise alarms, isolate the faulty sensor, switch to a safe or degraded mode, or initiate diagnostics and maintenance. This keeps the system from blindly trusting bad data or pushing the loop to compensate in unsafe ways.

Ignoring the deviation is risky because it allows an undetected fault to influence control decisions. Simply cranking up loop gain to force outputs to match can cause instability and mask the underlying problem instead of solving it. Recalibrating all sensors immediately is usually unnecessary and disruptive; fault detection focuses the response on diagnosing and handling the fault rather than blanket recalibration.