Case Study: The Ink Delamination Experiment1 CUink, Inc. is a company specialized on the printed graphics business, particularly for plastics and other similar materials. One of CUink’s largest customers reported that the ink on the grid of one of their printed products was delaminating and that the customer’s production line was stopped. The newly certified Six Sigma Black Belt at CUink immediately began examining the problem and determined that ink from a new supplier was used on the last batch of parts shipped to the customer. A total of 25,000 parts were shipped using the ink from the new supplier and approximately 20% of them were delaminating to a degree not acceptable for the customer. No failures of this kind were reported using the ink from the previous supplier. A cross-functional team was formed to determine the root cause of the problem and implement a permanent solution. Apparently, the decision to change ink supplier was taken assuming that the material would perform the same and without doing much testing prior to its use in the production process. Due to contractual obligations, CUink was now committed to continue buying from the new supplier. Thus, the team (DOE team) decided to design and conduct an experiment to determine which factors significantly contributed to delamination and then to develop effective process control parameters to prevent delamination defects in the future with the new material. A brain storming session was held to identify possible factors that could create ink delamination. Six key process factors were isolated as being potential causes based upon the experience of the team members. These key input factors are as follows: • Belt Speed – Speed at which the line is run to apply the ink. • Jet Oven Temperature – Temperature setting for the oven that initiates the drying process by applying jets of hot air. • Pot Life – Time until the ink is used after it is mixed and prepared for application. The mixed ink has only a certain usable shelf life and eventually starts to dry-up and harden. The manufacturer would prefer pot life to be as long as possible without affecting delamination in order to decrease scrap costs. • Oven Temperature – Temperature of drying oven after ink application • Oven Time – Time in oven to fully dry the parts. • Mesh – Fabric used in the ink transfer process. 1 Guzman, L., Hammett, P. and Frescoln, K., (2003). Modified by Kim, H. (2006). This Six Sigma case study has been prepared as an instructional tool. Although the data and problem description are based on an actual Six Sigma project, the authors have modified the project data and results for instructional purposes. Thus, no inferences should be drawn about a particular company or its quality levels based on the information in this case study. The team decided to examine samples and scored on a scale of 1 –10 with 1 representing extremely bad delamination and 10 representing a perfect part with no evidence of delamination as illustrated in the examples in Figure 1 below. 1 0 8 4 1 Figure 1. Examples of Delamination and their Rating Scale (see Excel file for full scale) The six-sigma team decided to conduct two different studies as described below: Study 1 In order to evaluate the samples as objectively as possible, master samples were made with various degrees of delamination that were judged and scored by the DOE team. The rating scale described above was created and used to train team members. Three members evaluated thirty samples and rated them on a scale of one to ten. Using your six-sigma knowledge, evaluate the ability of these three members to rate delamination on the subject part. Keep in mind the following: 1. The same 30 sample parts were rated by each of the three team members. Each sample was rated and judged by the DOE team as well and given a master rating. The data is attached in the file Case4Data2017.xls 2. Using an appropriate analysis technique, determine whether the 3 different operators are capable of rating the parts according to the ratings established by the DOE team (Master Rating). Justify your answer based on a statistical significance level of 0.05 (Hint: the 30 parts were run under different conditions and present various degrees of delamination.) The focus is on determining if any - or all - of the operators rate the parts significantly different from the master ratings or from each other. Study 2 After assessing the delamination rating scale (measurement system), the DOE team wants to design and conduct a 2-level factorial experiment to determine the most appropriate parameters to achieve a mean delamination target rating of 9 or above using the ink from the new supplier (assume that all parts will be rated by the same operator). The team can only afford 3 replicates of a Resolution IV design to study the 6 different factors. The settings for the 6 different factors are as shown in Table 1. Table 1. Factor Settings for Delamination Experiment Low High Factor Variable -1 1 A Belt Speed Low High B Jet Oven Temp Low High C Pot Life (hrs) 2 4 D Oven Temp (°F) 140 160 E Oven Time (hrs) 2 3 F Mesh Narrow Wide 1. Define the DOE in Minitab to specify the appropriate design. Keep in mind that the data should be an experiment involving 16 combinations (26-2) with 3 replicates each to study the 6 different factors (i.e., 48 total samples). 2. Using the ink simulator (InkSimulator2017OddID.xls for odd UIN’s and InkSimulator2017EvenID.xls for Even UIN’s), collect the data for your DOE based on the Minitab design. 3. Analyze the data to determine the significant main effects and interaction effects. Re-run your analysis using only those deemed significant. Use a significance level of 0.05. 4. Recommend the settings for the above factors that can help this company achieve the target of delamination for their product of 9.0, using the longest possible Pot Life time. Note1: You may need to round up the response optimizer result to achieve a ~9 rating. Note2: Interpret the coded settings and translate them into actual units for those settings with numerical values, for non-numerical, just interpret your findings as a categorical level between the Low/High levels. (Remember that the maximum allowable rating is 10). 5. Use the ink simulator to simulate a batch of 20-30 parts and use this information to evaluate the expected overall performance of the process based on your recommended settings. Assume that CUink has agreed with the customer to set a rating of 7.0 as the lower specification limit for this product and use a subgroup size of 1. (Hint: use the process settings identified in your answer above). Prepare your assignment report following DMAIC and use the results of your analyses on the previous studies to justify your recommendations for CUink. Sample questions: Define: 1. (5pt) Include a paragraph to briefly define the problem. Measure: 2. (5pt) Explain the key output and the current state. Describe the two studies. Analyze: Study 1: 3. (10pt) Are all the inspectors and the master samples giving consistent ratings? Create an ANOVA table to see the difference between master and inspectors. 4. (5pt) Re-do the ANOVA without the inconsistent inspector(s) and draw the conclusion. Study 2: 5. (10pt) Create a fractional factorial design. 6. (10pt) Run the design and show the significant terms. 7. (5pt) Re-fit the design only with significant terms. What is the conclusion? 8. (15pt) Using the response optimizer option in Minitab, what is the best configuration to obtain a rating of 9? How would the design be changed to minimize scrap costs (large Pot life)? Improve 9. (10pt) Based on this analysis, what general recommendations would you give? 10. (15pt) Evaluate the expected overall performance of the process based on your recommended settings. Assume that CUink has agreed with the customer to set a rating of 7.0 as the lower specification limit for this product and use a subgroup size of 1. (Hint: use the process settings identified in your answer above) Control 11. (10pt) What are the control recommendations for CUink?