Skip to content

Home / DOE / Additional Information / Screening vs. Modeling

Screening Designs vs. Modeling Designs

While there are many different types of designs, the two main categories are screening and modeling.

Screening designs do not have interactions and are typically used identify which, of a large set of factors, are the most important or critical factors. Modeling designs do include interactions and are typically used to build a model or transfer function. Screening designs have the advantage of being smaller, or less expensive to run, while modeling designs provide interactions and are therefore suitable for modeling.

As a general rule, the number of runs (and expense) to execute a modeling design increases exponentially with the number of factors.

When the number of factors is relatively small, Quantum XL will recommend a modeling design. As the number of runs increases, Quantum XL will recommend a screening design. If your experiments are not expensive, there is nothing wrong with bypassing the recommended screening design and executing a modeling design.

Note

Some DOE practitioners, in particular those following the Taguchi methodology, use screening designs more often than others.

Screening Design Example: You have 8 candidate factors (inputs) which all might affect the fuel efficiency of a car (the "output"). The inputs are fuel octane, total weight, driving speed, tire size, air conditioner (on/off), fuel additive used (yes/no), acceleration (fast/slow), and windows (up/down). Executing a screening design would not only identify which of the 8 factors is most significant, it would also provide a rank order of all 8 from most to least significant.

Modeling Design Example: In the aforementioned screening design, the weight of the car, driving speed, and tire size were the top three contributors to fuel efficiency. Executing a modeling design will provide interactions between these variables and a transfer function relating these inputs to fuel efficiency. The transfer function comes in the form as Y=f(x) or in this case Fuel Efficiency = f(weight, driving speed, tire size). A hypothetical example for fuel efficiency would be Fuel Efficiency = 23 - 4*weight - 3*driving speed - 1*tire size - .5*weight*driving speed.

Supported Screening Designs

Supported Modeling Designs