The L9 Experiment Array

What is an 'L9' Experiment Array?

An L9 array can be used to covers all combinations of two parameters at three levels each. Alternatively, if we are willing to ignore the interactions between parameters (or assume that they are small), an L9 can be used as a 'saturated' array to measure four parameters at three levels each.

For this experiment, we did two saturated L9 arrays. The first covered the levels:

  • Azimuth Angle of Attack (AoA): 0, 90, and 180 degrees
  • Elevation (El): 30, 45, and 60 degrees
  • Upwind Vent Gate (UVG): Open, Half and Closed
  • Downwind Vent Gate (DVG): Open, Half and Closed
and the second covered:
  • Azimuth Angle of Attack (AoA): 45, 90, and 135 degrees
  • Elevation (El): 30, 45, and 75 degrees
  • Upwind Vent Gate (UVG): Open, Half and Closed
  • Downwind Vent Gate (DVG): Open, Half and Closed

Why did we do it?

The main purpose of the L9 tests was to expand our coverage of the parameter space. While the approach was very efficient compared to the full factorial approach (each of which would have required 81 tests), the penalty was a loss of any information on the interactions between factors, and a loss of the 'free' noise estimate.

Results

The L9 arrays have been analyzed using several metrics, and are available on the analysis page of this CD.

For More Information on DOE