1. Name the defect before the sensor
“Check every pack” is an objective, but it is not yet an inspection specification. Start with a short list of outcomes you need to find: an underweight pack, metal contamination, a missing film, an unreadable code or an abnormal sealing pattern. Give each defect an identifier and a clear description.
Prepare examples the quality team already understands. Keep the expected result separate from the proposed detection method. For a questionable seal, for example, record what the physical inspection found before asking whether a camera can distinguish that example from an acceptable pack.
This prevents a common purchasing mistake: accepting an attractive display as proof that the original quality question has been answered. The relevant evidence is the result for known samples under the agreed production conditions.
2. Match the question to the inspection method
| Method | Question it can address | Boundary to verify |
|---|---|---|
| Checkweighing | Is the moving pack within agreed weight limits? | Weight alone does not identify the reason for a deviation or prove the seal is sound. |
| Metal detection | Can the agreed metal test pieces be detected in this product and package? | Product characteristics, packaging and aperture affect the application; it is not a test for every foreign material. |
| Visible-light vision | Are defined visible features present and acceptable, such as film position or a printed code? | The relevant feature must be visible and consistently imaged. Appearance does not establish leak rate. |
| Thermal imaging | Does the observed heat pattern match the agreed normal range for this sealing process? | A thermal result is indirect evidence. It cannot by itself guarantee a leak-free pack or a specified seal strength. |
METTLER TOLEDO's product-inspection overview distinguishes weighing, contamination detection and vision tasks. Its conveyor metal-detector guide also explains why the actual product and packaging belong in the selection review.
For equipment discussions, Sealerweb lists an Online Checkweigher and Vision Inspection System and a Food Metal Detector. Treat each listed inspection function as a scope item to confirm, rather than assuming all functions are supplied together.
3. Separate seal appearance from seal performance
A camera may show that film is present yet leave another question unresolved: whether the bond meets the required physical test. The Seal Vision Inspection and Rejection Machine addresses defined visible anomalies and explicitly keeps leak and strength validation separate.
Thermal inspection adds information about heat distribution. FLIR's food-industry application note describes observing the heated sealing region. Its measurement documentation explains that material emissivity and reflected temperature influence temperature interpretation. A colour image should not be read as a universal temperature or integrity certificate.
The Inline Thermal-Imaging Foil-Seal Integrity Inspection Machine is intended to screen agreed foil-seal anomalies. Establish the relationship between its result and your known acceptable and defective samples. Keep the corresponding physical seal checks in the acceptance plan.
A good thermal image is not a guarantee against leakage.Define what the system must detect, test that claim with representative samples, and identify which separate tests are still needed for the finished pack.
4. Build a sample set that can challenge the claim
Create a register containing normal production variation as well as the agreed defects. Label each sample with its format, material lot and expected result. Include borderline examples for the quality team to classify before the equipment trial. If the team cannot agree on the reference result, resolve that first.
Use one set for setup and a separate, representative set for evaluation where practical. In a blind evaluation, the person running the inspection does not receive the expected answer for each sample beforehand. Agree sample counts and defect coverage with the responsible quality team instead of selecting an impressive percentage in advance.
Keep two error counts separate: defective samples accepted, and acceptable samples rejected. A system can appear successful if the report combines them or contains too few examples of the difficult defect. Request results by defect class and format, together with the operating conditions.
Prepare the inspection sample pack
- Acceptable packs covering the normal formats and appearance variations
- Agreed defect examples, with reference results and unique identifiers
- Product dimensions, packaging materials, orientation and actual spacing
- The proposed inspection position and relevant environmental conditions
- A test record separating missed defects from false rejects
- Written acceptance criteria and responsibility for approving the outcome
5. Verify the reject action, not only the decision
Follow a labelled test pack all the way through the proposed inspection and reject route. Record whether the system identified it, associated the decision with the correct pack, and delivered that pack to the intended location. Confirm how operators know that a reject occurred.
Ask the supplier and your controls team to define tests for the agreed operating situations: normal spacing, closely following packs, stops and restarts, and the specified response when a reject cannot be completed. These should be planned commissioning checks under the site's procedures, not improvised changes during production.
Agree what the operator sees after a fault and how production resumes. Also decide which records are retained: the recipe or inspection version, event time, defect category and any agreed image or measurement. Define how a format change triggers a recipe review and a fresh verification.
Place these requirements into the wider filling and closing line brief. For an existing seal problem, first use the seal troubleshooting record to preserve the evidence. You can then discuss the sample brief with Yijia Packaging at Sealerweb, with a defined inspection question and a practical way to test the answer.
References and scope
Linked manufacturer pages were reviewed on 15 October 2026. METTLER TOLEDO and FLIR references explain inspection principles; they do not demonstrate the performance of Sealerweb equipment. Equipment-specific claims remain limited to the linked Sealerweb descriptions and must be checked on the actual project. This guide proposes an enquiry and trial framework, not a product-release approval.