Quality Engineering トピックス 英語版

Quality engineering(QE) 13

投稿日:

I show you the comparicent of two cases in which transferability of molding is applied to quality engineering(QE).

Click here → QE13

p.1: For Cases A and B, the control factors and their levels are shown in the table. The signal factors are the thickness, width and overall length of the dumbbell shaped part. As noise factors, PMA and PMB (tentative name) with different resin flowability (MFR:Melt flow rate) were used. We compare the level values ​​in the pink frame of Case A and the blue frame of Case B. In case A, a skilled molding staff decided the level. I think probably that this person empirically thought that the VP change position should be around 6.25 mm. Under the unconsciousness, he narrowed the level width so that the shape of the molded product can be obtained in all 18 rows of the orthogonal table. This judgement will have an unfavorable effect later. In Case B, I instructed a new trainee to make changes that broaden the level.

p.2: See the SN ratios of cases A and B. MIN of the SN ratio is in the pink frame and MAX is in the blue frame. Calculating the difference reveals that Case B is wider.

p.3: Let's compare Cases A and B for the 18 graphs that plot raw data. Pink frame(Min SN ratio) and blue one(Max SN ratio) are added to graphs obtained in the caseA and caseB. Comparing L1 of Case A and L9 of Case B, it is clear that Case B has a worse SN ratio. I mentioned in a previous blog that "50 points problems are more effective", but this case B gives exactly better results. In Case B, there were some “short” that did not have the shape of dumbbells and only half of the resin entered, and some moldings called “burrs” that protruded. Thus, it is better to have imperfect results appearing in some rows of L18. 'Good results' mean that we get the correct ' the graph of factorial effects'. In addition to widening the level range, it is also effective to adjust the toughness with an noise factor.

p.4: Line up 'the graph of factorial effects' for Cases A and B. Please see the gain of the confirmation experiment. The gain of Case A is 3.9, and the gain of Case B is 22.6, showing that Case B is more reliable in 'the graph of factorial effects' . Control factor C is the VP change position and factor E is the holding time, but the magnitude relationship of the SN ratio to the level is different. According to the person in charge of molding, among the control factors A to H, the VP change position has the greatest effect on the quality of the moldability. The optimum level for the VP change position is 6.25 mm for Level 2 in Case A and 6 mm for Level 1 in Case B. It is almost the same value. With regard to quality engineering(QE), it was possible to show results equivalent to empirical knowledge in 'the graph of factorial effects' , even if not experienced in molding. This is a case in which the wonder of quality engineering was recognized again.

-Quality Engineering, トピックス, 英語版

Copyright© 進化するガラクタ , 2020 All Rights Reserved Powered by STINGER.