Warpage optimization on primary recycled polypropylene composites: investigating the effect of cooling time and injection temperature with machine learning and in-mold sensors

This study investigates the warpage problem observed in dishwasher bottom cover manufactured via plastic injection molding, considering primary recycled material characteristics, key processing parameters (injection set temperature and in-cycle cooling time), and real-time in-mold sensor data. The material used was a polypropylene homopolymer filled with 40 wt% calcium carbonate (CaCO₃) (I1TPC40), which was mechanically reprocessed (ground) and subsequently reintroduced into injection molding. The effects of material properties, injection set temperature, and cooling time on warpage behavior were systematically evaluated. Material characterization was performed using melt flow index (MFI), differential scanning calorimetry (DSC), and Fourier transform infrared spectroscopy (FTIR). The DSC results indicated no statistically significant differences in thermal behavior between the recycled I1TPC40 and the original PURPC40. MFI analysis showed increased flowability of the recycled material, while FTIR spectra confirmed the absence of noticeable chemical degradation. These findings demonstrate that the material retains its integrity and can be safely reused in further production processes. Samples were molded using three injection temperatures (200, 210, and 225 °C) and six cooling times (23–40 s). Warpage was quantified based on dimensional and mass measurements, and in-mold temperature profiles were monitored using embedded sensors. The results revealed that cooling time and the in-mold temperature distribution play a decisive role in warpage formation. To predict warpage, multiple linear regression (MLR) and classification and regression trees (CART) were employed, and both models achieved prediction accuracies exceeding 90%. Overall, the findings indicate that combining primary recycled materials with appropriately selected processing conditions and sensor-driven data-based models offers substantial potential for reducing warpage and enabling more sustainable injection molding practices.