Last updated on July 7th, 2023
USE CASE
Predicting injection molding defects with AI
Objectives
- Reduce defect and scrap rates for injection molded automotive parts
- Mimimize injection molding machine stoppages
Challenge
- Traditionally, it has been difficult to monitor for anomalies across all signals at once
- Some injection molding process malfunctions occur even when individual signals are within control limits
Key Results
Reduced scrap
Increased production efficiency
Background on injection molding
Injection molding is a precision manufacturing method that requires a high degree of technical expertise to execute successfully. Various measures of temperature, pressure, humidity, and tonnage must be finely tuned to ensure that molds are filled completely, cooled, and ejected properly.
A variety of deviations in the process can produce defective parts that must be scrapped. Since defective injection molded parts cannot be repaired or reworked, it is essential to reduce the scrap rate as much as possible.
Early detection of process deviations and defects
LinePulse uses machine learning and artificial intelligence algorithms on manufacturing data to provide advanced warning of potential process issues and defects, so operators and engineers can intervene proactively to prevent scrap.
LinePulse detects anomalous behaviour in the injection molding process much earlier than traditional methods. Your screw RPM, barrel zone temperatures, pellet humidity, injection pressure, ejection force, moving platen close speed, tonnage at start of ejection, final tonnage and more can all be analyzed for early issue detection.
Let’s explore an example using hold pressure. During the injection molding process, melted plastic material is pushed through the barrel by helical screw to fill the molds. A specific amount of pressure is used to inject the correct amount of plastic, and this pressure needs to be held for a certain period of time to ensure the plastic fills the mold.
Various injection molding defects can occur.
- If this pressure isn’t controlled and held long enough, the mold might not be filled completely, resulting in a defective part.
- A partially filled mold can become difficult to eject. This requires the machine to be stopped and cleaned manually, which is a time consuming and unsafe procedure.
In some cases, the pressure signal will move in a non-linear pattern towards the control limits, making it difficult to predict exactly when these problems will occur. With LinePulse, however, historical patterns in signal pressure are tracked. When the signal begins to execute a pattern that previously led to a failure, LinePulse will issue an alert. This results in a much earlier warning when the hold pressure becomes anomalous. With these early alerts, operators and engineers can intervene before set points are exceeded, which will prevent scrap and minimize unexpected downtime.
Detection of multi-variate anomalies causing defects
LinePulse can also combine and monitor groups of machine signals to detect anomalous relationships between them. This is especially valuable in injection molding as many different signals are generated simultaneously and need to synchronize in behaviour.
For example, an injection molding barrel might be comprised of ten temperature zone measurements and all signals might be within individual control limits when viewed individually. This would give the appearance of a stable temperature throughout the barrel.
However, LinePulse would detect immediately if all temperature measurements together revealed that the overall heat profile of the barrel was inconsistent and changing, even when individual signals were not anomalous. By alerting operators when there is an abnormal pattern developing between signals, greater insight into the overall process can be gained.
Results
By displaying all signals and custom alerts in one centralized dashboard, LinePulse gives your team insights about your injection molding processes. The advanced data analysis provides advanced warning of process deviations, allowing you to intervene proactively and avoid potential defects and scrap.