Design-Expert 7.1

Introduction

Design Expert

Design-Expert offers an impressive array of design options. Version 7.1 includes dozens of new features that increase the ease-of-use, functionality, power and appeal of an already great product.

After you build your design in Design-Expert, generate worksheets with your experiments laid out for you in randomised run-order.  Add, delete or duplicate runs in any design with the handy design editor. With annotated statistical analysis and an extensive context-sensitive help system, you can easily interpret the outputs.  Interactive 2-D graphics support use of your mouse to drag contours or set flags that display coordinates and predicted responses.  Rotatable 3-D plots make response visualisation easy.

With the powerful optimisation features in Design-Expert, you can maximise desirability for dozens of responses simultaneously.  There are also unique tools for generating and graphing propagation of error (POE), thus allowing you to achieve six-sigma objectives for reducing variation.  Maximise, minimise or hit targets with factor levels set to give you robust results.

A Tremendous Variety of Designs Meet All Your Experimental Needs

  • Standard two-level full and fractional factorials (up to 512 runs) for testing up to 21 factors simultaneously, now also with minimum-aberration blocking choices.
  • General (multilevel) factorial designs (up to 32,000 runs) using factors with mixed levels.
  • Taguchi orthogonal arrays.
  • High-resolution irregular fractions, such as 4 factors in 12 runs.
  • Placket-Burman designs for 11, 19, 23, 27 or 31 factors in 12, 20, 24, 28 or 32 runs respectively.
  • Response Surface Method (RSM) designs, including central composite (small, face-centered, etc.), Box- Behnken (3-level), hybrid and D-Optimal.
  • Mixture designs, such as simplex-lattice, simplex-centroid screening (for up to 24 components) and D-optimal.
  • Combined mixture and process designs (mix your cake and bake it, too!).
  • Ability to graph any two columns of data on the XY graph (this is a great way to view a blocked effect).
  • Easy-to-use automatic or manual model reduction.
  • Ability to easily analyse designs with botched or missing data.

Enjoy Incredible Flexibility in Design Modification

  • Define your own generators for fractional factorial designs.
  • Impose linear multivariable constraints on RSM or mixture design.
  • Add categorical factors to RSM, mixture or combined designs.
  • Create a factorial candidate set for RSM designs when only specific factor levels are available.
  • Ignore a row of data while preserving the numbers.

Build Confidence with Statistical Analysis of Data

  • If your model is aliased, a warning will pop up prior to viewing the ANOVA for two-level fractional factorials, allowing you to make substitutions for aliased effects.
  • Select optional annotated views for assistance interpreting the ANOVA.
  • Inspect F-test values on individual model terms and confidence intervals on coefficients.
  • Automatically select effects using Lenth's criteria or probability values.
  • Take advantage of new user preferences, for example, make a global change in the significance threshold (0.05 by default vs. 0.01 and 0.1).

Take Advantage of Powerful Tools for Response Modeling

  • Change models from RSM to factorial and back and from Scheffe (mixture) to slack (during design building and at model selection).
  • Add integer power terms to the model, for example, quartic.
  • Select terms for model, error, or to be ignored (allows analysis of split-plot and nested designs).

Simplify Interpretation with Terrific Graphics

  • A quick summary of the design type as well as factor, response and model information is available by clicking on the design status node.
  • Discover significant effects at a glance with half-normal or normal probability plots, made easier by including points representing estimates of pure error (if available from your design).
  • See the Box-Cox plot for advice on the best response transformation.
  • View a complete array of diagnostic graphs to check statistical assumptions and detect possible outliers (bonus feature: predicted-versus-actual graphs with a 45º line).
  • Graph alternative aliased interactions.
  • See the effects plot in the original scale after transforming the response.
  • Observe variation in predictions by viewing the least significant difference (LSD) bars on the model graphs.
  • Poorly predicted regions on contour maps are shaded to give you confidence in your predictions.
  • Slice your contour plots using a simple slide bar (and see actual design points when they're on a slice!)
  • Set flags to reveal the predicted response at any location.
  • Drag 2-D contours using your mouse.
  • Rotate 3-D graphics and see projected 2-D contours.
  • Edit colors, text and more to produce professional reports.
  • See all effects on one graph with trace and perturbation plots.
  • Plot the standard error of your design on any graph type (contour, 3-D, etc.)

Locate Your Sweet Spot with Multiple Response Optimisation

  • Maximise, minimise or target specific levels for both responses and factors.
  • Set weight and importance levels to prioritise responses for desirability.
  • Choose 2-D contour, 3-D surface, histogram or ramp desirability graphs.
  • Include categorical factors.
  • Set factors at constant levels.
  • Add equation-only responses, such as cost, to the optimisation process.
  • Look at the overlay plot to view constraints on your process or formulation.
  • Predict responses at any set of conditions (including confidence levels).
  • Discover optimal process conditions or formulations.
Rate this page
Comment