Exercise set

Process Analyzer Calibration, Sampling, and Measurement Release Exercises

Solved analyzer exercises for lab comparison, bias, sample age, drift, lag, calibration error, duplicate checks, availability and measurement release gates.

Branch: Chemical Engineering
Content: Exercise set
Updated: Jul 03, 2026
Revision: v1.1.0 · reviewed

These exercises focus on process analyzer and measurement release evidence. They cover lab comparison, analyzer bias, sample age, dead time, drift, calibration span, duplicate precision, grab-sample mismatch, analyzer availability, uncertainty guard band, stale data and measurement release gates.

Assume simplified screening calculations unless an exercise states otherwise. Real analyzer release requires sample-system design, calibration standards, lab QA, response time, validation state, data historian status, maintenance bypasses, alarm use, environmental reporting basis and responsible sign-off.

How to use these exercises

Use the set as a measurement-release walk-through. Exercises 1 to 6 check whether the analyzer is calibrated, biased or drifting. Exercises 7 to 12 test whether the reported value is timely and complete enough for a live process decision. Exercises 13 to 17 connect availability, bypass exposure, uncertainty and risk scoring to release evidence. Exercise 18 then combines the controls into an all-of gate.

For every calculation, keep the sample location, stream condition, lab timestamp, analyzer timestamp, calibration range and decision limit visible. Analyzer numbers are only defensible when the sample and the release decision refer to the same physical stream and time window.

Release Evidence Notes

Analyzer evidence should state what the analyzer measures, where the sample is taken, how old the sample is, how it compares with lab data and whether its uncertainty is acceptable for the release decision.

A good analyzer reading can still be unusable if the sample is stale, the sample line is plugged, the analyzer is in maintenance, the stream is outside calibration range or the release limit is too close to uncertainty.

The release package should also show how the analyzer is used by operations. State whether the value drives an alarm, interlock, product release, emissions record, batch disposition or operator advisory. Then connect that role to required data completeness, validation frequency, bypass controls, historian quality flags, independent lab confirmation and escalation rules when the analyzer is unavailable.

Engineering Boundary Notes

These calculations do not replace analyzer validation protocols, laboratory QA, regulatory monitoring rules, sample-system engineering or calibration management. They are screening exercises for measurement release.

Real analyzer performance depends on sample conditioning, dead volume, line plugging, phase changes, pressure and temperature correction, calibration gas or liquid traceability, lab method bias, instrument maintenance state and historian configuration. Treat the exercises as a way to identify the controlling evidence gap: calibration, sampling, timing, completeness, uncertainty, bypass control or decision authority.

Common Release Mistakes

comparing online and lab values without accounting for sample delay;
accepting a reading near a limit without a guard band;
using an analyzer outside calibration span;
treating availability as valid data when the analyzer is in maintenance mode;
releasing operation from one good grab sample while online trend is biased;
matching analyzer and lab samples by date instead of by actual stream residence time;
hiding held-last-value periods inside apparently continuous historian trends;
approving a bypass without compensating monitoring or production limits;
using a calibrated instrument as release evidence after the sample system has changed.

Scenario Map

Scenario	Exercises	Primary check	Engineering decision
Bias and comparison	1, 2, 3, 4, 5, 6	lab comparison, percent bias, span and drift	Decide whether analyzer data are valid.
Timing and uncertainty	7, 8, 9, 10, 11, 12	sample age, lag, duplicate precision, guard band and stale data	Decide whether the measurement can support release.
Availability and evidence	13, 14, 15, 16, 17	valid hours, maintenance bypass, data completeness, RPN and package completion	Decide whether analyzer evidence can close.
Release gate	18	all-of measurement release	Decide whether analyzer data can be used for operation.

Exercise 1: Analyzer Validation Against Lab Result

An online analyzer reads $4.8\%$ component concentration. A validated lab sample from the same stream reads $5.1\%$ . Compute absolute bias.

Solution

b=4.8-5.1=-0.3\ \text{percentage points}

Engineering Comment

The analyzer reads low. A low bias can falsely release product or emissions if the limit is an upper concentration.

Plausibility Check

The online value is $0.3$ points below the lab value.

Exercise 2: Relative Analyzer Bias

Using analyzer $4.8\%$ and lab $5.1\%$ , compute relative bias to the lab result.

Solution

b_r=\dfrac{4.8-5.1}{5.1}=-0.0588=-5.9\%

Engineering Comment

Relative bias helps compare analyzer error across ranges, but release limits usually need absolute guard bands too.

Plausibility Check

An error of $0.3$ on about $5$ is about six percent.

Exercise 3: Analyzer Limit Guard

The release limit is $5.5\%$ . Analyzer reading is $4.8\%$ and absolute bias allowance is $0.4$ points. Compute guarded value and margin.

Solution

Guarded value:

C_g=4.8+0.4=5.2\%

Margin:

M=5.5-5.2=0.3\ \text{points}

Engineering Comment

The guarded margin is narrow. It should not be treated like the nominal $0.7$ point margin.

Plausibility Check

Adding uncertainty reduces the apparent margin.

Exercise 4: Calibration Span Error

Calibration span is $0$ to $10\%$ . A validation standard is $6.00\%$ and the analyzer reads $5.82\%$ . Compute percent of span error.

Solution

Absolute error:

e=5.82-6.00=-0.18\%

Percent of span:

\dfrac{0.18}{10}=1.8\%

Engineering Comment

Percent-span error is useful for calibration criteria; release still depends on the process limit.

Plausibility Check

An error below two tenths over a ten-point span is below two percent of span.

Exercise 5: Drift Rate

Analyzer bias changed from $-0.10$ to $-0.30$ percentage points over $14$ days. Compute drift rate.

Solution

r=\dfrac{-0.30-(-0.10)}{14}=-0.0143\ \text{points/day}

Engineering Comment

Drift direction matters. A growing low bias may hide off-spec product.

Plausibility Check

Two tenths of a point over two weeks gives about one hundredth point per day.

Exercise 6: Calibration Interval Screen

Maximum allowed bias change is $0.25$ percentage points. Drift rate magnitude is $0.0143$ points/day. Estimate time to reach the allowed change.

Solution

t=\dfrac{0.25}{0.0143}=17.5\ \text{d}

Engineering Comment

The calibration interval should be shorter than the drift-to-limit time, with margin for operating changes.

Plausibility Check

At about $0.014$ points per day, one quarter point takes a little over two weeks.

Exercise 7: Analyzer Sample Age

Analyzer sample transport delay is $9$ minutes and analyzer cycle time is $4$ minutes. Estimate sample age at reported value.

Solution

t_{age}=9+4=13\ \text{min}

Engineering Comment

Sample age matters during startup, transients and diversion decisions.

Plausibility Check

The reported value must lag by the transport plus measurement cycle time.

Exercise 8: Process Change During Sample Age

Concentration is changing at $0.06$ percentage points per minute. Sample age is $13$ minutes. Estimate possible mismatch from current stream condition.

Solution

\Delta C=0.06(13)=0.78\ \text{points}

Engineering Comment

During fast transients, a delayed analyzer may not be valid for release even if calibrated.

Plausibility Check

Several hundredths per minute over more than ten minutes gives nearly one percentage point.

Exercise 9: Analyzer Response Lag

Analyzer first-order time constant is $3$ minutes. Estimate time to reach $95\%$ of a step change using $t_{95}\approx3\tau$ .

Solution

t_{95}=3(3)=9\ \text{min}

Engineering Comment

Response lag should be included in alarm timing and release decisions during transitions.

Plausibility Check

A first-order system needs several time constants to settle.

Exercise 10: Duplicate Sample Precision

Two lab duplicate results are $5.10\%$ and $5.24\%$ . Compute relative percent difference.

Solution

Average:

\bar{C}=\dfrac{5.10+5.24}{2}=5.17\%

RPD:

RPD=\dfrac{|5.24-5.10|}{5.17}=2.71\%

Engineering Comment

Poor duplicate precision can make analyzer-lab comparison inconclusive.

Plausibility Check

The duplicate difference is small relative to a five-percent concentration.

Exercise 11: Data Completeness

An analyzer should report every $5$ minutes during a $12$ hour shift. It recorded $132$ valid results. Compute completeness.

Solution

Expected results:

N=\dfrac{12(60)}{5}=144

Completeness:

C=\dfrac{132}{144}=91.7\%

Engineering Comment

Missing results during rate changes or alarms matter more than missing results during stable operation.

Plausibility Check

The analyzer missed twelve of one hundred forty-four expected values.

Exercise 12: Stale Data Duration

The historian holds the last analyzer value for $22$ minutes during communication loss. The stale-data alarm threshold is $15$ minutes. Compute exceedance.

Solution

\Delta t=22-15=7\ \text{min}

Engineering Comment

Held values should not be used as fresh evidence for product or emissions release.

Plausibility Check

The stale value persisted several minutes beyond the threshold.

Exercise 13: Analyzer Availability

Analyzer valid operating time is $640$ hours in a $720$ hour month. Compute availability.

Solution

A=\dfrac{640}{720}=88.9\%

Engineering Comment

Availability is not the same as accuracy, but low availability can invalidate continuous-release claims.

Plausibility Check

Eighty hours unavailable out of seven hundred twenty is a little over eleven percent downtime.

Exercise 14: Maintenance Bypass Exposure

Analyzer maintenance bypass is active for $4.5$ hours. The approved bypass limit is $3$ hours. Compute exceedance.

Solution

\Delta t=4.5-3=1.5\ \text{h}

Engineering Comment

Bypass exposure should trigger compensating monitoring or production restriction.

Plausibility Check

The bypass lasted one and a half hours too long.

Exercise 15: Measurement Error Budget

Analyzer calibration uncertainty is $0.18$ points, sampling uncertainty is $0.22$ points and lab comparison uncertainty is $0.15$ points. Combine by root-sum-square.

Solution

U=\sqrt{0.18^2+0.22^2+0.15^2}=0.321\ \text{points}

Engineering Comment

The combined uncertainty should be used for guard-banding near release limits.

Plausibility Check

RSS combination is larger than any single component but smaller than their sum.

Exercise 16: Analyzer Evidence Completion

The measurement release package requires calibration, validation standard, lab comparison, sample age, response lag, duplicate precision, uncertainty budget, data completeness, bypass status and historian status. Eight of ten records are complete. Compute completion.

Solution

C=\dfrac{8}{10}=80\%

Engineering Comment

The package is not release-ready if missing records include bypass status or uncertainty budget.

Plausibility Check

Eight of ten is exactly eighty percent.

Exercise 17: Bad Analyzer RPN

A biased analyzer release failure has severity $8$ , occurrence $4$ and detection $5$ . Compute RPN.

Solution

RPN=8(4)(5)=160

Engineering Comment

The score supports independent lab checks, stale-data alarms and guarded release limits.

Plausibility Check

Moderate-high ratings multiply to a three-digit score.

Exercise 18: Analyzer Release Gate

A release gate requires guarded margin positive, sample age below $10$ minutes, data completeness above $95\%$ , no active bypass and evidence completion above $90\%$ . Current values are guarded margin $0.3$ points, sample age $13$ minutes, completeness $91.7\%$ , bypass exceeded and completion $80\%$ . Decide release status.

Solution

Guarded margin passes. Sample age, completeness, bypass and evidence completion fail:

13>10

91.7\%<95\%

Release status:

\text{hold}

Engineering Comment

The analyzer reading should not be used for release until timing, completeness, bypass and evidence gaps are corrected.

Plausibility Check

An all-of measurement gate fails when several validity controls fail.

Validation Package Checklist

Online analyzer values are compared with lab results on a time-aligned sample basis.
Bias, span error, drift, response lag and sample age are quantified.
Data completeness, stale values and maintenance bypasses are visible.
Measurement release uses guard bands, not nominal readings alone.
Calibration standards, lab QA records and analyzer range are traceable to the release limit.
Historian quality flags distinguish valid, stale, bypassed, maintenance and communication-loss states.
Operating procedures define what happens when analyzer evidence is unavailable or conflicting.
Final sign-off states whether the analyzer supports control, alarm, compliance or product release.

The final acceptance question is practical: would an operator, process engineer, quality reviewer and environmental reviewer reach the same release decision from the same evidence record? If the answer depends on undocumented judgment, the analyzer may be useful for monitoring but is not ready as a formal release instrument.

REF