This action might not be possible to undo. Are you sure you want to continue?
Section 5 – Taguchi Design Tutorial
Taguchi’s orthogonal arrays provide an alternative to standard factorial designs. Factors and interactions are assigned to the array columns via linear graphs. For example, look at the first of 18 linear graphs for the Taguchi L16 (16 run two-level factorial).
3 2 13
14 15 11 B N
6 5 10 4 12 8
F E K D M H
First Linear Graph for L16 Array The figure at the left displays 15 column numbers available for effect estimation. At the right you see the corresponding factor letters. Starting at the top and going counterclockwise, you can see that factor C is connected to AB, implying confounding of factor C with the AB interaction. Factor F is connected to BD, and so forth. These relationships describe aliasing for a handful of the possible relationships. The complete alias structure for the L16, generated by Design-Ease, is shown below.
[A] [B] [C] [D] [E] [F] [G] [H] [J] [K] [L] [M] [N] [O] [P] = A - BC - DE - FG - HJ -KL - MN - OP = B - AC - DF - EG - HK -JL - MO - NP = C - AB - DG - EF - HL - JK - MP - NO = D - AE - BF - CG - HM - JN - KO - LP = E - AD - BG - CF - HN - JM - KP - LO = F - AG - BD - CE - HO - JP - KM - LN = G - AF - BE - CD - HP - JO - KN - LM = H - AJ - BK - CL - DM - EN - FO - GP = J - AH - BL - CK - DN - EM - FP - GO = K - AL - BH - CJ - DO - EP - FM - GN = L - AK - BJ - CH - DP - EO - FN - GM = M - AN - BO - CP - DH - EJ - FK - GL = N - AM -BP - CO - DJ - EH - FL - GK = O - AP - BM - CN - DK - EL - FH - GJ = P - AO - BN - CM - DL -EK - FJ - GH
Aliasing of Main Effects with Two-Factor Interactions for L16 (First Linear Graph) The underlined effects are the aliases revealed by Taguchi’s first linear graph.
Design-Ease 6 User’s Guide
Taguchi Design Tutorial • 5-1
italicized effects shown below.BL .AP .GL = N .BC .CJ .AG .AK .DM .EF .AJ .EG .DK .HJ .EP .JN .FG .BD .HN .BH . The experimenters identified nine factors (see table below).EL .GN = L .DE .DJ . But.HO .NP = C .AF .AC .EM .AH .EK . 1991).BM .DO .BF .CD .AM .FK .JP .AO .AB .CK .KN .LM = H .KM .DH -EJ .BG .AL .JK .CL . why bother? The complete alias structure is given by Design-Ease software via its Design Evaluation tool.GJ = P .LO = F .DL .HK .MO .KP .EO . 9 J 3 10 13 1 14 7 11 K N C A O G L 6 5 2 12 4 8 15 F E B M D H P Second Linear Graph for L16 This second linear graph reveals the underlined.FN .MN .CF .AE .DF . page 189 (Quality Resources.KO . Case Study To see how Design-Ease software handles Taguchi arrays.NO = D . let’s look at a welding example out of System of Experiment Design. you could build the entire alias structure by going through all 18 linear graphs.JO .DP . 5-2 • Taguchi Design Tutorial Design-Ease 6 User’s Guide .CM .BK .AN .LP = E .GK = O .KL .CP .BJ .HL .FO .GH Aliasing of Main Effects with Two-Factor Interactions for L16 (Second Linear Graph) In theory.GM = M .GO = K .FH .AD .CO .MP .CH .CE .CN .DG .EH .FM .FL .HM .FJ .OP = B .BE .The second of Taguchi’s 18 linear graphs is given below.BN .LN = G .CG . [A] [B] [C] [D] [E] [F] [G] [H] [J] [K] [L] [M] [N] [O] [P] = A .HP .JL .DN .GP = J .EN .JM .FP .BO . Volume 1.BP .
5 none SS41 Factors for Welding Experiment The experimenters wanted estimates of several interactions: AB. we see that column C can be used to estimate the AB interaction. 1/10/01 Factor Brand Current Method Drying Thickness Angle Stand-off Preheat Material Units amps Level 1 J100 150 weaving none Level 2 B17 130 single 1 day 12 60 3. AD and BD. Looking at the first L16 linear graph (reproduced in part below). Taguchi used columns M and N to estimate error.Rev.0 150 deg C SB35 mm degrees mm 8 70 1. Column A B C D E F G H Factor Brand Current AB Method AD BD Drying Thickness Column J K L M N O P Factor Angle Stand-off Preheat AP Material Factor Assignments for L16 on Welding Design-Ease 6 User’s Guide Taguchi Design Tutorial • 5-3 . column F to estimate BD and column O to estimate AP. 1 A 3 2 5 6 14 15 B C O P E F 4 D Subset of First Linear Graph for L16 on Welding The factor assignments are summarized below. column E to estimate AD.
KP . On all other designs you would now be prompted to enter factor names. New Design off the menu bar.CN .MP .CL .BD .BE .EH .EJ .KN .GL [N] = N . Then from the default Factorial tab click Taguchi OA and choose L16(2^15) from the pull down menu.HO .LO [F] = F .BP .DH .FN .CJ .DO .AM .FP .AF .FK .FO .CF .EF .GH Alias Structure for L16 Click on the Continue button.DL .DJ .JL .EK .AD .AL .EN . Design-Ease now shows the response screen.JN .AP .JO .AN .BK . F is aliased with BD and O is aliased with AP.CD . (Note: there is no column labeled “I” in this or other designs because this is reserved for the intercept of the predictive model.GO [K] = K .BC .HM .LP [E] = E . Choose File. after you generate the layout of runs.GJ [P] = P .) Design the Experiment Let’s build this design.CG .CE .DP .BO .BN .NO [D] = D .Columns M and N (blank) will be used to estimate error.JK .GN [L] = L .OP [B] = B .AK .BM . However.AG .KO .DM .AO .GP [J] = J .FG .CK .DK .AE . 5-4 • Taguchi Design Tutorial Design-Ease 6 User’s Guide .MO .EL .NP [C] = C .JM . Notice that C is aliased with AB.DG .BG .AJ .HN .EO .CH .HK .EG .KL .HP . [A] = A .DN .GK [O] = O .GM [M] = M .CO .LN [G] = G .JP .DE .FH .CP .CM .AC .LM [H] = H . Selecting the Taguchi Orthogonal Array Click on the Continue button. The software then presents the alias structure for the chosen design.BF . for Taguchi designs this will be done later.AH .AB .BH .DF .BL .FJ .FM . Enter the 1 response name as “Tensile” and the units as “kg/mm^2”.KM .FL .HL .HJ .EP .BJ .EM . E is aliased with AD.MN .
This will greatly simply the analysis. the letters associated with the remaining factors change. the focus shifts back to the leftmost part of the design. (Warning: when deleting factors. Design-Ease now displays the design in random order. Open Design from the main menu. Recall that several of the columns (C. but to save time read in the response data via File. 1/10/01 Response Entry Click on the Continue button. F and O) are being used to estimate interactions.de6. E. Next. always start at the right and work left. Right-Click Menu for Factor Column in Design Layout Choose Delete Factor and confirm with a Yes. Two others (M and N) are used to estimate error. Using Edit Info Screen The rest of the factors can be entered in the same way. Right click on the header for column O (Factor 14) to bring up a menu as shown below. right click on the heading of column A and choose Edit Info: Enter “Brand” as the name and “J100” and “B17” as level 1 and level 2. We can delete all these columns and fit the interactions directly. Right click on the Std column and choose Sort by Standard Order to see Taguchi’s design order.Rev. so to avoid confusion. Select the file named Taguchi-L16. Also. so you Design-Ease 6 User’s Guide Taguchi Design Tutorial • 5-5 . Be patient – it takes the software a few moments to retabulate the design.
All this work will eventually pay off by putting you in position to take advantage of Design-Ease’s powerful analytical capabilities. In the table below. It will be helpful to make a table for cross-referencing the original assignment of factor letters with the new (condensed) list. and underlined the two-factor interactions of interest. F. Right click on the Std column and Sort by Standard Order.must scroll back each time to select the next column for deletion. only the nine experimental factors should remain. Your design should now look like that pictured below.) Repeat the Delete Factor operation on the columns for factors N. Taguchi L16 After Deleting Columns so Only Factors Remain Let’s look at the alias structure of the remaining columns. Original (9) A: Brand B: Current C: AB D: Method E: AD F: BD G: Drying H: Thickness J: Angle New (9) A: Brand B: Current C: Method D: Drying E: Thickness F: Angle G: Stand-off H: Preheat J: Material Discarded (6) K: Stand-off L: Preheat M: error N: error O: AP P: Material Cross-Reference Table for Factor Assignments Note that some interactions also get re-labeled: AB stays AB. click on the design Evaluation node and for Order select 2FI (two-factor interaction) and click on Results. we ignored interactions of three or more factors. 5-6 • Taguchi Design Tutorial Design-Ease 6 User’s Guide . BD becomes BC and AP becomes AJ. When you finish deleting columns. but AD becomes AC. To review the alias structure. M. E and C.
AF . plus the four interactions of interest. Click on the two largest effects (J and D) on the halfnormal plot of effects. but they too are aliased with two-factor interactions.AE .BG .BH [G] = G . 1/10/01 Factorial Effects Aliases [Est.Rev. The effects now labeled BJ and CE are the two columns used to estimate error.HJ [D] = D .DE [AB] = AB + CD + EH + FG [AC] = AC + BD + GJ [AD] = AD + BC + FJ [AJ] = AJ + CG + DF [BJ] = BJ + CF + DG [CE] = CE + DH Alias Structure After Deleting Columns from L16 Notice that all of the main effects. Terms] Aliased Terms [Intercept] = Intercept [A] = A .) Click on the analysis node labeled Tensile.CJ [J] = J . (If you haven’t already done so. Then. click on the Effects button displayed in the toolbar at the top of the main window.BE [H] = H . Half Normal Plot of Effects Design-Ease 6 User’s Guide Taguchi Design Tutorial • 5-7 . All of these aliased interactions must be negligible for an accurate analysis. are aliased with one or more two-factor interactions.EJ [E] = E .EG . go back and complete this tutorial.EF .AG .AH . Analyze the Results To analyze the results.FH [C] = C .CH . follow the usual procedure for two-level factorials as illustrated earlier in the Factorial Design Tutorials.GH [B] = B .BF .DJ [F] = F . which is found in the tree structure along the left of the main window.
You should now see an annotated ANOVA report by default. Design-Ease now warns you about aliasing and offers a list for you to view. Click on the Model Graphs button. Click No to skip this. 5-8 • Taguchi Design Tutorial Design-Ease 6 User’s Guide . ANOVA Report Click on the Diagnostics button. Click on the upper end of D:Drying. One-Factor Plot of the Effect of Factor D (Drying) Click the Term down arrow and select J (or right-click on the J bar to make it the Xaxis). but they’re acceptable. because you’ve already looked over the aliases. Notice that Design-Expert defaults to the “lower” level of the categorical factors. The ANOVA shows that the effects of D and J are significant.Click on the ANOVA button. select View. A graph of factor D (drying) should now appear. Annotated ANOVA. The graphs don’t look great. If not. which we now know is best for tensile strength.
Make sure that likely interactions are not confounded with main effects.Rev. it would be a good idea to do a followup experiment to confirm the effects of D and J. Due to the fractional nature of Taguchi designs. This provides visual verification that the effect shown on the plot is statistically significant. but as shown in the alias table. and J could be CH and/or DE. only two main effects (D and J) appear to be significant. This concludes our tutorial on Taguchi orthogonal arrays. these are actual design points. You may be wondering about the circular symbols. perhaps D is really EJ. As indicated by the legend. Click on those points that do appear to get a readout of their actual response value. in this case. Note that this LSD does not overlap with the one at the lower right. For example. Therefore. 1/10/01 Plot of Factor J (Material) with Factor D Set at High Level and Best Point Picked The square symbols at either end of the lines show the predicted outcomes. Click on the predicted value at the upper left to get a readout on the LSD (printed to the left of the graph). Design-Ease 6 User’s Guide Taguchi Design Tutorial • 5-9 . you won’t see too many points on the plots of predicted effects. Use these designs with caution! Take advantage of Design-Ease’s design evaluation feature for examining aliases. The bars going up and down from these points represent the least significant difference (LSD) at a 95% confidence level (the default).
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.