Design of Experiments relies on a shared set of terms that define how a process is studied and understood. These terms aren’t just labels. They describe how variables are tested, how data is collected, and how conclusions are drawn.
Once these concepts are clear, it becomes easier to design experiments that reveal how a molding process actually behaves.
Factors are the inputs you control during molding.
These include variables like melt temperature, mold temperature, injection speed, hold pressure, and cooling time. Each one influences how the material flows, packs, and solidifies.
In a DOE, factors form the structure of the experiment. Changing them systematically reveals how sensitive the process is to each input.
Levels are the specific values assigned to each factor.
For example, melt temperature might be tested at two settings, such as 450°F and 475°F. Injection speed might be tested at a low and high value to compare performance.
Choosing levels defines the range of the study. If the range is too narrow, differences may not appear. If it’s too wide, the process may move into unstable or unrealistic conditions.
Responses are the outputs measured during the experiment.
These can include part weight, dimensions, surface quality, or machine data like cavity pressure. Each response reflects how the process reacted to the chosen inputs.
By comparing responses across different runs, patterns begin to emerge that link process settings to part quality.
A main effect shows how one factor influences a response on its own.
For example, increasing melt temperature may reduce viscosity and improve filling consistency. Increasing cooling time may improve dimensional stability but extend cycle time.
These effects highlight which variables have the strongest influence on the process.
Interactions occur when two or more factors influence the outcome together in a way that isn’t obvious when looking at them individually.
A single variable might not cause a defect on its own, but when combined with another variable, the result changes significantly.
This is where DOE becomes especially valuable. It reveals relationships that can’t be seen through one-variable-at-a-time adjustments.
Every molding process includes background variation from sources that aren’t fully controlled.
Material differences, environmental conditions, machine drift, and operator input all contribute to this noise. Without accounting for it, results can be misleading.
DOE separates true cause-and-effect relationships from random variation, making conclusions more reliable.
To ensure results are valid, experiments include repeated runs and varied run order.
Replication confirms that results are consistent and not due to chance. Randomization prevents patterns from forming due to time-based effects, such as temperature drift in the mold.
Together, these methods strengthen the accuracy of the conclusions.
A consistent terminology framework keeps process development structured and traceable.
Each factor, level, and response is clearly defined and documented, allowing results to be analyzed and reproduced with confidence. The result is a process that can be explained, validated, and maintained without relying on assumptions.