Package defect and material integrity failure detection using dynamic scanning infrared thermography (DSIRT)

Pouches and trays for shelf-stable foods are replacing traditional packages as production capacity is expanded or replaced. Unfortunately, the fused material structures of these packages lack a comprehensive, fast, non-destructive method for inspection which creates a continuing production safety issue. Additionally, there are similar manufacturing problems with defects in non-food products ranging from electronics assemblies to medical devices. This provides an opportunity for the development of simple, robust inspection methods as a safety check in the fabrication processes of food packaging as well as medical and other devices. It was hypothesized that the obverse surface temperature of a material subjected to a careful "sweep™ of heat input caused by relative motion of heat input to material would show subtle variation that could be captured quickly before it was obscured by thermal diffusion. The Dynamic Scanning Infrared Thermography (DSIRT) system developed to test this at the UIUC Packaging Lab exhibits high repeatability for defect detection, very high potential speeds of material throughput, and is capable of doing so in a non-laboratory environment. Trials using retort pouch film and other packaging materials have shown that automatic processing of the data derived from such a system can rapidly detect anomalies in the material or seal, and point to an automated inspection system. Although not a perfect solution for all applications, the initial trials with a low-resolution camera system was easily able to distinguish between images containing a failed seal and those without, and represents an improvement over simple heat transmission systems. It was also observed that because the system relied on transient, moving heat conduction, resolution increased with scanning speed. The preliminary defect detection system was adequate for determining discontinuities in the thermal signature that was produced by a bolometer-based imaging system. Calculation time was fast enough to suggest that a dedicated graphics processing unit could keep up with a real-time frame rate. DSIRT resolution is at the optical and frame rate resolution of the current equipment, however new imaging equipment incorporated into the DSIRT study achieves much higher frame rate and much higher resolution. The efficacy of the method with metallic and adhesive bonds, the limits of the improved equipment and the implications for image processing are discussed.