Creating Factorial Reliability Designs

To create a factorial reliability design, follow the steps outlined next. For more general information on building designs, see Building a New Design.

- General Settings
- - Replicates allows you to specify the number of test samples you want to observe at each combination. In other words, the number of replicates is the number of times you want to perform the complete set of runs required by the design. Using replicates helps to separate error caused by a poor model fit from error caused by natural variation (i.e., experimental error).
  - Block on Replicates allows you to choose whether you will use each replicate of the design as a block. By blocking on replicates, you can account for "nuisance factors" (i.e., effects that are not of interest to the investigator) due to conditions that differ from one replicate of the design to another. For example, if the first replicate of the design is performed on Day 1, and the second is performed on Day 2, you could account for variation in the response due to the day. This option is available only if the number of replicates in the design is greater than 1.
- Block Settings
- - Number of Blocks allows you to specify the total number of blocks that will be used in the design. Blocks per Replicate (shown when you choose to block on replicates) allows you to specify the number of blocks that will be used per replicate. The more blocks you use in the design, the more potential nuisance factors you can account for. Note, however, that using blocks requires aliasing them with one or more effects.
    
    Select a number of blocks from the drop-down list; the options available in this list depend on the type of design and the number of factors. Using 1 block in the entire design is equivalent to not using blocks.
  - Center Points per Block allows you to specify the number of center points that will be used per block. Center points are used primarily to help you determine if there is curvature in the relationship between the factors and the response. In addition, using several center points provides an estimate of pure error.
  - Block Name allows you to enter a name for each block.
  - Design Generators will appear when you have multiple blocks. This field is used to specify the factorial interaction effects that will be aliased with that of the nuisance factor for which you are blocking the design. (See Specifying Generators.)
- Factorial Settings
- - Fraction allows you to select the number of base runs you want to use in the experiment, based on a fraction of the total number of runs that would be required for a full factorial design. For example, if your design includes five factors with two levels each, a full factorial design would require 25 = 32 base runs. For a fractional factorial design in this case, you can select to use a one-half fraction (16 runs) or a one-fourth fraction (8 runs). The total number of runs required for the experiment may be greater than the number of base runs (e.g., if you use replicates and/or center points).
  - Design Generators allows you to specify the factorial interaction effects that will be aliased with the main effects of particular factors. (See Specifying Generators.)

For more information on this design type, see Two Level Factorial Designs.

- General Settings
- - Replicates allows you to specify the number of test samples you want to observe at each combination. In other words, the number of replicates is the number of times you want to perform the complete set of runs required by the design. Using replicates helps to separate error caused by a poor model fit from error caused by natural variation (i.e., experimental error).
  - Block on Replicates allows you to choose whether you will use each replicate of the design as a block. By blocking on replicates, you can account for "nuisance factors" (i.e., effects that are not of interest to the investigator) due to conditions that differ from one replicate of the design to another. For example, if the first replicate of the design is performed on Day 1, and the second is performed on Day 2, you could account for variation in the response due to the day. This option is available only if the number of replicates in the design is greater than 1.
- Block Settings
- - Number of Blocks allows you to specify the total number of blocks that will be used in the design. Blocks per Replicate (shown when you choose to block on replicates) allows you to specify the number of blocks that will be used per replicate. The more blocks you use in the design, the more potential nuisance factors you can account for. Note, however, that using blocks requires aliasing them with one or more effects.
    
    Select a number of blocks from the drop-down list; the options available in this list depend on the type of design and the number of factors. Using 1 block in the entire design is equivalent to not using blocks.
  - Center Points per Block allows you to specify the number of center points that will be used per block. Center points are used primarily to help you determine if there is curvature in the relationship between the factors and the response. In addition, using several center points provides an estimate of pure error.
  - Block Name allows you to enter a name for each block.
- Placket-Burman Settings
- - Base # of Runs allows you to specify the number of runs you want to use in the experiment. The available choices are determined by the number of factors you specify and the design matrix developed by Robin L. Plackett and J. P. Burman.

For more information on this design type, see Plackett-Burman Designs.

- General Settings
- - Replicates allows you to specify the number of test samples you want to observe at each combination. In other words, the number of replicates is the number of times you want to perform the complete set of runs required by the design. Using replicates helps to separate error caused by a poor model fit from error caused by natural variation (i.e., experimental error).
  - Block on Replicates allows you to choose whether you will use each replicate of the design as a block. By blocking on replicates, you can account for "nuisance factors" (i.e., effects that are not of interest to the investigator) due to conditions that differ from one replicate of the design to another. For example, if the first replicate of the design is performed on Day 1, and the second is performed on Day 2, you could account for variation in the response due to the day. This option is available only if the number of replicates in the design is greater than 1.
- Block Settings
- - Number of Blocks allows you to specify the total number of blocks that will be used in the design. Blocks per Replicate (shown when you choose to block on replicates) allows you to specify the number of blocks that will be used per replicate. The more blocks you use in the design, the more potential nuisance factors you can account for. Note, however, that using blocks requires aliasing them with one or more effects.
    
    Select a number of blocks from the drop-down list; the options available in this list depend on the type of design and the number of factors. Using 1 block in the entire design is equivalent to not using blocks.
  - Center Points per Block allows you to specify the number of center points that will be used per block. Center points are used primarily to help you determine if there is curvature in the relationship between the factors and the response. In addition, using several center points provides an estimate of pure error.
  - Block Name allows you to enter a name for each block.

For more information on this design type, see General Full Factorial Designs.

- General Settings
- - Replicates allows you to specify the number of test samples you want to observe at each combination. In other words, the number of replicates is the number of times you want to perform the complete set of runs required by the design. Using replicates helps to separate error caused by a poor model fit from error caused by natural variation (i.e., experimental error).
  - Block on Replicates allows you to choose whether you will use each replicate of the design as a block. By blocking on replicates, you can account for "nuisance factors" (i.e., effects that are not of interest to the investigator) due to conditions that differ from one replicate of the design to another. For example, if the first replicate of the design is performed on Day 1, and the second is performed on Day 2, you could account for variation in the response due to the day. This option is available only if the number of replicates in the design is greater than 1.

- Block Settings
- - Number of Blocks allows you to specify the total number of blocks that will be used in the design. Blocks per Replicate (shown when you choose to block on replicates) allows you to specify the number of blocks that will be used per replicate. The more blocks you use in the design, the more potential nuisance factors you can account for. Note, however, that using blocks requires aliasing them with one or more effects.
    
    Select a number of blocks from the drop-down list; the options available in this list depend on the type of design and the number of factors. Using 1 block in the entire design is equivalent to not using blocks.
    
    With Taguchi OA designs, you must block on replicates to use blocks, and you can use at most 1 block per replicate.
Taguchi OA Settings

The options available in the Taguchi Design Type drop-down list will depend on the numbers of levels for each factor. For example, if all your factors have three levels, then you could choose the L9 (3^4) design, which uses nine runs to investigate up to 4 three-level factors. If some factors have two levels and the rest have four, then you could choose the L16 (2^9*4^2) design, which uses 16 runs to investigate up to 9 two-level factors and up to 2 four-level factors.

Click the View Alias Table icon to see the alias table for the selected Taguchi design type. The table shows what aliases result from assigning factors to particular columns in the orthogonal array.

See Taguchi OA Alias Table.

Taguchi Column Indices allows you assign factors to columns of the selected orthogonal array. No two factors can have the same index.

The Specify Interaction Terms link opens a window that helps you assign factors to columns without aliasing any main effects or any interaction terms of interest. (See Specify Interaction Terms Window.)

For more information on this design type, see Taguchi Robust Designs.

	Related Topics and Links:
	Factorial Reliability Designs