Engineer Food Processing Safety study materials
Engineer Food Processing Safety study resources

Allergen Changeover Failure on a 'Dairy-Free' Chocolate Line

Engineer Food Processing Safety · Put your knowledge to work in context.

All case studies

Allergen Changeover Failure on a 'Dairy-Free' Chocolate Line

A chocolate plant runs milk chocolate and a 'dairy-free' 70% dark chocolate on the same moulding line, enrober and cooling tunnel. The changeover procedure is dry cleaning only: scrapers, compressed air and a wipe-down with 70% ethanol; no water is used because the line is not wet-cleanable. The dairy-free recipe uses a 'natural flavor' purchased from a broker; the specification states only 'natural flavor' with no allergen statement. After the changeover, the plant ran 12,000 kg of the dairy-free product. Protein swabs (casein ELISA) taken from the enrober curtain, the depositor head and the cooling tunnel belt before start-up returned <1.5 µg/100 cm² at the first two sites but 6.8 µg/100 cm² on the belt. A finished-product composite from the first hour tested 15 ppm milk protein by ELISA. The label carries a 'dairy-free' claim and an allergen statement listing soy only. The customer specification for this product is <5 ppm milk.

1. Can the 12,000 kg be released? Justify your answer using the cleaning results, the finished-product data and the labelling rules.

Read the worked answer

No - hold the entire 12,000 kg and do not release against the 'dairy-free' claim.

Cleaning logic

  • The belt swab at 6.8 µg/100 cm² is above the internal limit of 2.5 and above the other sites; a single site failure means the changeover is not validated, because the line is only as clean as its dirtiest food-contact surface.
  • The belt is downstream of the enrober, so it re-contaminates product after the enrober curtain has been cleaned - a classic dry-cleaning failure mode.
  • Compressed air can aerosolise milk powder and redistribute it onto surfaces that were clean before the blow-down; ethanol wipes sanitise but do not remove soil. ATP was not used here, but it would not have detected this either.

Product data 15 ppm exceeds the customer's <5 ppm limit by a factor of three. The first-hour composite being out of specification means the start of the run was contaminated; the rest of the run cannot be assumed clean, and re-contamination from the belt is continuing.

Labelling A 'dairy-free' claim is an absolute claim. It cannot be qualified with 'may contain milk', and a product containing 15 ppm milk cannot carry the claim; the label would be false and misleading. The missing milk in the allergen statement compounds the problem.

Disposition Segregate the lot; test start, middle and end to define the extent; rework into a milk-containing product only if the formulation and allergen declaration permit and the rework is validated; otherwise destroy. Investigate whether the line can support a dairy-free claim at all under dry cleaning.

2. How do you handle the 'natural flavor' with no allergen statement? What supplier verification and risk assessment apply?

Read the worked answer

Absence of a statement is not evidence of absence. In most jurisdictions flavourings are not exempt from allergen declaration (for example, US law requires milk, egg, fish, crustacean shellfish, tree nuts, wheat, peanuts, soybeans and sesame to be declared even when present in a flavour). A specification saying only 'natural flavor' is an information gap, and for a dairy-free claim an information gap is a hazard.

Actions

  • Request a written allergen statement and a completed supplier allergen questionnaire covering the flavour's carriers and processing aids (dairy, soy, gluten, sesame, and cross-contact from shared lines).
  • Ask whether the flavour is dairy-derived or processed on a line with dairy; request a certificate of analysis and, where risk-based, a milk ELISA result on the flavour itself.
  • Review the broker's status: a broker that cannot obtain the manufacturer's declaration is not an acceptable source for an allergen-sensitive claim.
  • If the declaration cannot be obtained, reject the ingredient or remove the claim from the finished product. Do not accept a letter that says 'no allergens' without identifying the flavour's composition and the basis for the statement.
  • Add the ingredient to the supplier approval and change-notification program so that a reformulation triggers reassessment.

Risk assessment Score likelihood (dairy-derived carrier, shared-line risk, unknown composition) against severity (a dairy-free claim to milk-allergic consumers). With unknown composition and an absolute claim, the risk is not acceptable until verified.

3. Design the verification program for the changeover: what to measure, at which sites, with what limits and frequency.

Read the worked answer

Limits first Derive the swab limit from the finished-product specification. With a 5 ppm (5 mg/kg) limit and a 12,000 kg batch, the total milk protein allowed in the batch is 60 g. Back-calculate the allowable residue per food-contact surface area using the equipment's surface area and the fraction that transfers to product; document the assumptions rather than copying a number from a supplier.

Sites Worst-case locations: cooling tunnel belt, enrober curtain and bottom trough, depositor head and valve seats, scrapers, transfer points, and any dead corners or textured surfaces. Mark them on a map so sampling is reproducible.

Methods

  • Specific ELISA (casein or total milk protein) for allergen residue - this is the verification method.
  • A general protein test as a screening tool; ATP only as a cleaning-hygiene indicator, never as allergen verification.
  • Visual inspection as a prerequisite, not as the release criterion.

Validation Run the worst case (highest milk load, longest run, hardest-to-clean product) at least three consecutive times; require all swab sites and finished product (first and last unit of the run) to meet limits. Include a hold-time study for the cleaned line before start-up.

Routine verification

  • Every changeover: rapid allergen test at the defined sites plus first-pack finished-product test.
  • Periodic confirmatory ELISA and full site panel, trended.
  • Revalidation after any equipment, chemical, or recipe change.

The verification program is only credible if the belt is cleaned to the limit, not cleaned and then wiped to pass.