Sensor Reduction in Multi-Cavity Injection Molding

Sensor Reduction in Multi-Cavity Injection Molding

This is some text inside of a div block.


In injection molding, molds equipped with multiple cavities for resin flow are used to produce several molded products from a single mold. This process is known as multi-cavity molding. The resin filled into the mold from the injection molding machine passes through the sprue (A), branches off in the runner (B), and then reaches each cavity.


Figure1. Sprue and Runner

Traditionally, in multi-cavity molding, internal pressure sensors were installed in each cavity to detect any molding defects. However, equipping all cavities with internal pressure sensors presented challenges, including the need for mold modifications and the costs associated with purchasing sensors. These factors contributed to a high barrier to entry when starting to detect molding defects.

Experimental Method

In multi-cavity molding, the structural design where resin, branched off in the runner, fills each cavity can lead to changes in the resin state of one cavity affecting others. Consequently, it is considered that the resin state within each cavity is interrelated. Focusing on this characteristic, a research team explored the feasibility of detecting molding defects by installing internal pressure sensors only in selected cavities among the multiple ones.

Figure2. Installation image of internal pressure sensor

The mold used in the experiment featured four cavities, with internal pressure sensors installed only under the gate and at the end of one cavity. No sensors were placed in the other cavities. Using molding conditions for both good and defective products, internal pressure data for 20 shots were collected and analyzed to ascertain the distinguishability of the molded products.

Figure3.Experiment overview


From the internal pressure sensor data acquired, feature extraction and dimension reduction were performed, followed by calculating the centroid of the good quality data set for 20 shots. Subsequently, the distance from this centroid to the dimension-reduced features was plotted for each shot, as shown in Figure 4 below.


Figure4.Distance from feature vectors to the centroid of good product data group

In the figure, blue points up to the 20th shot represent good quality products, while the orange points indicate shots where molding defects occurred. A trend was observed where shots with defects exhibited a relatively greater distance from the centroid of the good quality data set. By setting a threshold (indicated by the dotted line in the figure), shots exceeding this threshold can be identified as defective.


This study successfully utilized data from an internal pressure sensor installed in only one cavity to detect molding defects across the entire multi-cavity mold. The results suggest that analyzing resin flow in a single cavity can reveal changes in flow patterns across all interconnected cavities via the runner system. This technique eliminates the need to install sensors in every cavity of a multi-cavity mold, significantly reducing both the complexity of mold modifications and the costs associated with sensor installation. Consequently, this lowers the barrier to implementing defect detection, greatly enhancing efficiency and quality control in the injection molding process.