Design of Experiments

Design of Experiments

The Design of Experiments is so complex that many researchers simply do not conduct such experiments. AsterWrite provides an intelligent system of step-by-step procedures that ensure a designed experiment cannot be erroneous.

Select Response Type

Using AsterWrite, a researcher can conduct Design of Experiments easily. AsterWrite simplifies a rather long and complex process so that researchers who understand the concept can conduct fairly good experiments.

Current Loss

AsterWrite records the current process performance for comparison with the future process performance.

Quality Loss Function

AsterWrite also evaluates the loss function for a sample of the current performance. The loss function is based on the sample mean and the variation.

Identify Factors

AsterWrite encourages the researcher to identify factors (independent variables, IV) that affect the response (dependent variable, DV).

Select the Control Factors

Control factors are then set at 2 or more levels depending on the experiment. It is important to study Noise factors as well.

Select the experimental design

The experiment is then conducted using an orthogonal array. AsterWrite provides many orthogonal arrays for experimentation, e.g.

  • L4(2^3) - three 2-level factors

  • L8(2^7) - seven 2-level factors

  • L9(3^4) - four 3-level factors

  • L16(2^15) - fifteen 2-level factors

  • L18(3^1x2^7) - one 3-level factor and seven 2-level factors

Trials are conducted randomly with the factor settings as given by the orthogonal array.

A Target Performance Measure (TPM, e.g. mean) and a Noise Performance Measure (NPM, e.g. variance) are studied.

Conduct the analyses

AsterWrite completes the analysis of variance for both the TPM and NPM. Insignificant factors can be pooled and the factor contribution is shown as Rho percent.

Select the optimum condition

From a chart comparing the TPM and NPM, it is possible to determine factors which largely affect

  • Mean

  • Variance

  • Both

  • Neither

The optimum condition is then selected and evaluated.

Compare the Before-After performance

The optimum performance is then compared with the current performance.

Estimate the Cost Savings

The improvement attained is compared in terms of monetary loss reduction.