Exercise set
Calibration Drift and Interval Decision Exercises
Solved calibration drift and interval decision exercises for check standards, drift rate, out-of-tolerance impact, interval changes and release gates.
These exercises practise calibration drift as a lifecycle decision. A calibration result is not only a point-in-time certificate. It is evidence about whether the measurement system stayed fit for use during the interval, whether past measurements remain valid, and whether the next interval should be shortened, extended or held.
Assume simplified drift models unless an exercise states otherwise. Real interval decisions should also consider instrument history, environmental exposure, transport, overload, repair, use severity, reference uncertainty, check-standard records, process risk and the consequence of accepting bad product or rejecting good product.
How to Use These Exercises
For each problem, identify the as-found error, allowed error, time since last calibration, drift trend, use history and impact window. Do not treat adjustment as cleanup; if the instrument was out of tolerance as found, the measurements made before adjustment require impact review.
Release Evidence Notes
Calibration interval decisions should be based on evidence, not habit. Useful evidence includes as-found error, as-left error, adjustment amount, repair history, check-standard trend, environmental exposure, transport events, overload events, use frequency and the severity of decisions supported by the instrument. A laboratory instrument used gently once a month and a field gauge dropped into hot, wet service should not automatically share the same interval.
Out-of-tolerance impact review should start from the last known acceptable state. If no intermediate check standard exists, the review window may extend to the previous calibration. That does not mean every past result is wrong, but it means the organization must decide which products, tests, releases or commissioning records were exposed to the failed measurement system.
Interval extension needs a higher evidential bar than interval reduction. Consecutive in-tolerance results can support extension only when use conditions, risk, method and measurement function remain stable. A hidden change in fixture, software scaling, environmental exposure or operator method can invalidate a favorable calibration history.
Engineering Boundary Notes
These exercises use simple drift and interval screens. They do not replace a calibration-system procedure, instrument family reliability study, uncertainty budget, measurement system analysis, risk-based impact review or regulatory metrology requirement. A stable trend in one range or environment may not transfer to another range, method or field condition.
Common Release Mistakes
- shortening or extending intervals from calendar habit instead of as-found evidence;
- treating as-left adjustment as proof that exposed past measurements were valid;
- ignoring check-standard data between formal calibrations;
- using average drift while missing abrupt overload, repair or transport events;
- extending intervals after a method, fixture, operator or environment changed;
- closing an out-of-tolerance event without defining affected products, tests or releases.
Scenario Map
| Scenario | Exercises | Primary check | Engineering decision |
|---|---|---|---|
| Drift rate and trend | 1, 2, 3, 4, 5 | drift per month, time to limit, check standard and slope | Decide whether the interval is still credible. |
| Out-of-tolerance impact | 6, 7, 8, 9 | as-found status, guard band, zero/span drift and use window | Decide whether past measurements need review. |
| Interval management | 10, 11, 12, 13, 14 | interval reduction, extension, environment and action limits | Shorten, keep, extend or suspend calibration interval. |
| Release decisions | 15, 16, 17, 18 | standard disagreement, recall scope, risk score and final gate | Release, restrict, recall or quarantine evidence. |
Validation Package Checklist
- as-found and as-left results are recorded separately;
- last known acceptable state and impact window are defined;
- check-standard trend, use severity and environment are reviewed;
- drift model, allowed error and guard band are explicit;
- interval change rationale is tied to risk and evidence history;
- affected measurements, products, reports or commissioning records are dispositioned;
- next calibration interval, restriction or recall action is named.
Exercise 1: Drift Rate per Month
A check standard reading changed from 10.002\ \text{mm} to 10.026\ \text{mm} over 6 months. Estimate drift rate.
Solution
Change:
Drift rate:
Engineering Comment
Drift rate should be compared with tolerance, guard band and calibration interval. A small monthly value can still become large over a long interval.
Plausibility Check
Six months at 0.004\ \text{mm/month} gives 0.024\ \text{mm}.
Exercise 2: Time to Reach a Limit
An instrument is currently 0.03\ \text{mm} from its allowed error limit. Drift rate is 0.005\ \text{mm/month}. Estimate time to the limit.
Solution
Engineering Comment
The next calibration interval should leave margin before the expected limit, not land exactly at it.
Plausibility Check
At five micrometers per month, six months gives thirty micrometers.
Exercise 3: Check-Standard Z Score
A check standard expected value is 50.000\ \text{mm}. Current mean is 50.018\ \text{mm} and historical standard deviation is 0.006\ \text{mm}. Find z-score.
Solution
Engineering Comment
A three-sigma shift deserves investigation even if the instrument is not yet outside tolerance.
Plausibility Check
The shift is 0.018\ \text{mm}, exactly three times 0.006\ \text{mm}.
Exercise 4: Linear Trend Slope
Check-standard errors over four months are:
Estimate average monthly trend from first to last.
Solution
There are three month-to-month intervals:
Engineering Comment
Trend evidence can trigger preventive calibration before an out-of-tolerance event occurs.
Plausibility Check
The sequence rises by roughly 0.006\ \text{mm} each month, matching the estimate.
Exercise 5: Warning Limit Check
A warning limit is \pm 0.020\ \text{mm}. Current check-standard error is 0.017\ \text{mm}. Does it trigger warning?
Solution
Since:
the point does not exceed the warning limit.
Engineering Comment
Passing one warning limit does not cancel a rising trend. Trend and point limits should both be reviewed.
Plausibility Check
The current error is close to the limit but still below it.
Exercise 6: As-Found Out-of-Tolerance
A gauge has allowed error \pm 0.50\ \text{bar}. As-found maximum error is 0.68\ \text{bar}. Classify status.
Solution
The gauge is out of tolerance as found.
Engineering Comment
As-found out-of-tolerance triggers impact review for measurements made since the previous acceptable check.
Plausibility Check
The observed error exceeds the allowed error, so classification is direct.
Exercise 7: Out-of-Tolerance Impact Window
The last acceptable calibration was 365 days ago. A failed check occurred today. There was no intermediate check standard. What is the maximum impact window?
Solution
Without intermediate evidence, the maximum impact window is:
Engineering Comment
Intermediate check standards reduce impact uncertainty. Without them, the review may have to include all use since the last known acceptable state.
Plausibility Check
The only known good point is one year old, so the maximum window is one year.
Exercise 8: Guarded Drift Margin
Allowed error is 0.20\ \text{mm}. Current error is 0.14\ \text{mm} and expanded uncertainty of the check is 0.04\ \text{mm}. Find guarded margin to the allowed error.
Solution
Guarded error:
Margin:
Engineering Comment
The instrument is nominally inside tolerance, but only with a narrow guarded margin.
Plausibility Check
Adding uncertainty moves the result closer to the limit and leaves a small positive margin.
Exercise 9: Zero Drift and Span Drift
At calibration, a transmitter zero error is 0.1\% span and full-scale error is 0.7\% span. Estimate span drift after removing zero drift.
Solution
Engineering Comment
Zero drift and span drift have different causes and corrections. Resetting zero will not remove span drift.
Plausibility Check
The full-scale error includes the zero shift, so subtracting zero leaves a smaller span term.
Exercise 10: Interval Reduction Rule
A procedure halves the calibration interval after any as-found failure. The current interval is 12 months. Find the new interval.
Solution
Engineering Comment
Automatic interval reduction is simple, but the root cause still matters. Damage and stable drift require different actions.
Plausibility Check
Half of one year is six months.
Exercise 11: Interval Extension Rule
An interval may increase by 25\% after three consecutive in-tolerance calibrations with no adjustment. Current interval is 8 months. Find proposed interval.
Solution
Engineering Comment
Extension should also consider usage severity and risk, not only pass history.
Plausibility Check
A quarter of eight months is two months, so the new interval is ten months.
Exercise 12: Temperature-Coefficient Drift
An instrument temperature coefficient is 0.002\ \text{mm}/^\circ\text{C}. It was calibrated at 20^\circ\text{C} and used at 32^\circ\text{C}. Estimate temperature-induced shift.
Solution
Temperature change:
Shift:
Engineering Comment
Environmental use outside calibration conditions can look like drift even if the instrument itself has not aged.
Plausibility Check
Twelve degrees at two micrometers per degree gives twenty-four micrometers.
Exercise 13: Action Limit
A check-standard action limit is \pm 0.030\ \text{mm}. Current error is -0.034\ \text{mm}. Decide action.
Solution
Magnitude:
Since:
the action limit is exceeded.
Engineering Comment
Action should include hold, investigation, recalibration, and review of affected measurements.
Plausibility Check
The sign is negative, but limit comparison uses magnitude.
Exercise 14: Use Severity Factor
A handheld gauge has a normal interval of 12 months. Heavy use applies severity factor 0.5. Find adjusted interval.
Solution
Engineering Comment
Usage severity can be more important than calendar time for portable tools, field equipment and harsh environments.
Plausibility Check
A factor below one should shorten the interval.
Exercise 15: Redundant Standard Disagreement
Two check standards indicate errors of 0.012\ \text{mm} and 0.031\ \text{mm} on the same instrument. Their agreement limit is 0.015\ \text{mm}. Check disagreement.
Solution
Difference:
Since:
the standards disagree beyond the limit.
Engineering Comment
The issue may be the instrument, one standard, the setup or the method. Do not average conflicting standards without investigation.
Plausibility Check
The difference is slightly larger than the agreement limit, so investigation is expected.
Exercise 16: Recall Scope from Use Records
An out-of-tolerance torque wrench was used on 42 assemblies since the last acceptable check. Inspection can verify 30 assemblies. How many remain unresolved?
Solution
Engineering Comment
Unresolved use history may require risk assessment, customer notification, rework, concession or quarantine depending on severity.
Plausibility Check
If most but not all assemblies can be verified, a smaller unresolved count remains.
Exercise 17: Drift Risk Score
A simple risk score multiplies severity, occurrence and detection:
For S=8, O=4, D=5, find risk score.
Solution
Engineering Comment
Risk scores are screening tools. High-severity measurement failure should get attention even when the arithmetic score is not the highest in the list.
Plausibility Check
The product of three moderate-to-high integers should be in the hundreds.
Exercise 18: Drift and Interval Release Gate
A calibrated instrument has:
| Check | Result | Gate |
|---|---|---|
| As-found error | 0.46\ \text{mm} | \le 0.50\ \text{mm} |
| Guarded error | 0.53\ \text{mm} | \le 0.50\ \text{mm} |
| Check-standard trend | 3.2\sigma | \le 3.0\sigma |
| Use records complete | yes | yes |
| Current interval | 12 months | under review |
Decide whether to keep the interval unchanged.
Solution
Nominal as-found error passes:
Guarded error fails:
Trend fails:
The interval should not remain unchanged. Shorten the interval, investigate drift, review affected measurements and consider a check-standard control between calibrations.
Engineering Comment
Nominal pass is not enough when guarded evidence and trend both fail. The system is telling you the next failure is likely.
Plausibility Check
The result is close to the limit and trending, so interval reduction is consistent.