Open the file titled “Full Factorial 2^3 DOE”, and once opened, ensure that the tab “Design” is selected.

Figure 28 – DOE Section 1 – Set-up

The example that is pre-set in the sheet is the data used for the “DOE Helicopter” exercise demonstrated during formal training. The yellow highlighted, cells require data to be entered.

The Main Data Entry cells, require the description of the 3 factors that are being assessed (in the example case above, Wing Length, Body Length, Body Width) The High Value column requires the dimensional data for the factors, ie The length of the Wing length Option being tested. This means what setting will you be testing in the study. And the opposite for the Low Value column. The difference between the Low and High values is the experiment inference space.

The Max and Min columns are then asking for what the maximum and minimum possible values those factors can be set at (so that it knows the boundaries of the operating range).

For example the cup height might go from 200mm to 300mm but you will be testing a high setting of 280 and a low setting of 220mm. But the MAX/MIN is still 300/200.

Once Section 1 has been completed and the experiment is ready to run, scroll down to Step 3.

Figure 29 – DOE Section 2 – Data Entry

Firstly, select a Random Order Selection, changing this simply changes the run order column in step 4 to various orders so that the experiment run order can be randomised between trials.

The grid in Step 4 will now be partially completed. Each row shows the Model number for the DOE, the order in which it will be tested (to create a random element) followed by the dimensions of the 3 factors to be tested. The experiments are run in a randomised order as per the run order column.

In the example above, Sample 1 (Helicopter 1) will be the 4th to be tested, when this is tested, the helicopter settings will be set to:

  • Wing Length = 74
  • Body Length = 78
  • Body Width = 31

The first sample to be tested will actually be Sample 8 (as run order is marked as 1).

T1, T2, T3 etc are the Trial Numbers – In the example above, only 2 trials are carried out (this is the minimum).

As each test is carried out, enter the result in to the appropriate yellow highlighted cell, in the example above, time in seconds.

Once the results of the tests have been added, go to to Effects Analysis. This tab will give initial results on the effect that each factor has on the response.

Figure 30 – DOE Section 3 – Standardised Effects

The charts will show the standardised effects that influence the result and will highlight Key Deductions and guidelines on the right. This uses Wedge G methodology and displays figures in absolute values, and does not show direction of change (positive or negative). It also includes all interactions.

Scroll down to Factors / Interaction Analysis.

Figure 31 – DOE Section 4 – Interaction Plots

This shows how the different factors interact with each other. Full explanation of these interactions will be provided by your Capella trainer, but in summary, lines that cross indicate a very strong interaction between the 2 factors, lines that run parallel to each other indicate no interaction at all.

 

 Scrolling down further will show the Main Effects Analysis charts.

Figure 32 – DOE Section 5 – Main Effects and Optimisation

This shows the graphical representation of the Main Effects on the result. Effects are demonstrated by the distance between the high and low point and the steepness of the angle between the 2 points.

This will always be a linear measure as the tool does not check non-linear relationships (such as Cubic or Quadratic etc.)

Once reviewed, go to the Optimisation tab.

Figure 32A – DOE Section 6 – Optimisation

In this section, figure 32A is showing the response optimiser, which are the settings required to optimise the output measure to a specified value.

Use the dropdown box in Step 1 highlighted in yellow to stipulate whether you want to maximise or minimise the output (aim for largest or smallest length). You can also stipulate a target approach, if selected, 3 additional boxes will appear requesting target value to aim for, as well as USL and LSL as optional.

You will then be presented with 5 different possible optimal settings. Each one will show in light green, what to set that factor at to achieve the “Fit” (fit is known as the predicted output). Options 1 to 4 are the top 4 most desirable predicted outputs based on your request and assumed the settings are variable.

The fixed option simply gives an optimised response assuming that the factors can only be set at the values tested and there are no intermittent values. This is usually if your process is held by setting constraints.

Figure 32B – DOE Section 7 – Optimisation

 

In the last section, figure 32B allows you to fit the factors into a regression model. Yellow boxes can be changed to any value and the green box will show the predicted output based on the model.

The references below are coefficients of each factor and the regression equation used to calculate the predicted output, it is for reference if needed only.