The unvalidated metal detector and the pesto changeover
Hillcrest Fresh makes ready-to-eat salad bowls under ISO 22000:2018: 130 employees, two high-care lines.
Line 2 was recommissioned in April for 250 g bowls at 30 m/min. The metal detector after sealing is CCP-2, with critical limits Fe 1.5 mm and SUS 2.5 mm. The validation file contains a 2021 study on 100 g bowls at 18 m/min. The hourly check uses Fe and SUS test pieces and is recorded.
On 12 June the 09:00 check failed on the Fe test piece. Product packed between 08:00 and 09:00 was re-run through the same detector and released. The record reads re-tested, released - no quantity, no batch numbers, no authorisation, no root cause.
The line also changes between a pesto bowl (pine nut, milk) and a plain bowl. Changeover is a wet clean with a documented checklist, no verification step and no allergen test; the allergen risk assessment states shared equipment, cleaning validated 2019.
The pine nut paste supplier was approved in 2020 and last reviewed in 2021; no certificate is on file, incoming checks record temperature only, and supplier performance is not reviewed.
1. Is the metal detector validation adequate for line 2? Identify the clause and grade the finding, explaining what the sensitivity depends on.
Read the worked answer
Requirement. The hazard control plan rests on validated critical limits: the organization must demonstrate, with evidence, that the chosen limits detect the hazard at the required sensitivity for the product and process actually run. Validation for one configuration does not transfer to another.
What drives sensitivity. Aperture size, product weight and density, product effect (moisture and salt), orientation of the test piece, and belt speed. The only study is 100 g at 18 m/min. Line 2 runs 250 g at 30 m/min - a 67% speed increase on a heavier, wetter product. At a fixed aperture, both changes reduce the signal-to-noise ratio and raise the smallest detectable size. The stated limits may not be achievable without a smaller aperture, a lower speed or a different detector position.
Clause. Nonconformity against the hazard control plan requirements of ISO 22000:2018 clause 8.4 (critical limits must be validated), with clause 8.3 because the change to line 2 was not reflected in the hazard analysis, and 8.8 for verification.
Grade. Major: product may be released with undetected metal and there is no data showing whether it has been.
Containment and next steps. Validate on line 2 with test pieces in the actual product matrix at production speed - ideally three runs including worst-case product and orientation - then adjust limits or settings. Review all positive metal findings and complaints since April, and decide what to do about released product.
2. Assess the handling of the 12 June detector failure. Which requirements apply, and what records and tests would you use to establish whether potentially unsafe product was controlled?
Read the worked answer
Requirement. When a CCP fails, product produced since the last successful verification is potentially unsafe and must be handled under the nonconforming product requirements of ISO 22000:2018 (clause 8.9), with corrective action under 10.1. The record must identify the product, quantity, evaluation, disposition and authorisation.
Analysis of the record. 'Re-tested, released' is not a disposition. The product was passed through the same detector that had just failed. That only works if the failure lay in the test piece rather than the detector - and the organization never established which. If the detector had drifted, re-testing through it proves nothing. There is no quantity, no batch identification, no reconciliation against packing output, no root cause, and no verification that the corrective action was effective.
What I would test. Trace batch records and quantities for the 08:00–09:00 window against packing output; check whether the detector was re-verified with a second test piece and a different operator; review the detector fault log and belt speed record; check whether other shifts were affected; check whether customers were informed. Then examine the corrective action: was the cause identified - test piece condition, detector drift, speed change - and was the hourly frequency justified?
Judgement. Nonconformity against 8.9 and 10.1, likely major because the organization cannot demonstrate that potentially unsafe product was evaluated. The correction is a documented risk assessment of the affected lots, not a note in a log.
3. The allergen changeover between the pesto bowl and the plain bowl relies on a wet clean and a checklist. What is your judgement, and what evidence decides the grade?
Read the worked answer
Requirement. Allergens are identified in the hazard analysis (8.3) and controlled through PRPs and, where needed, an OPRP or CCP within the hazard control plan (8.4), with verification that the controls work (8.8). Cleaning is a PRP; its effectiveness against allergens is not established by a visual check.
Analysis. The organization relies on a wet clean, a checklist and a 2019 validation. Since then the range has changed to include pine nut pesto, and a 2019 validation cannot be assumed to cover a different soil, detergent, temperature and contact time. There is no verification at all - no specific protein test, no ELISA swab, not even ATP as a proxy. Control rests entirely on an operator completing a form. The outcome of failure is an undeclared pine nut allergen in a plain bowl, which is a recall-class event, and the plain label presumably declares no pine nut.
Judgement. Nonconformity against 8.3 and 8.4 (the hazard is identified but the control is not validated for current products) and 8.8 (no verification). Major while product is on sale and the organization cannot show that cleaning removes pine nut protein.
Evidence that decides. The validation report - soil, chemistry, contact time, test method; any swab results; whether line scheduling can place the allergen product last; and whether precautionary labelling is being used to cover an unverified process rather than a genuine risk assessment.