In our Quality Control lab materials are visually inspected for defects & then manually inspected using vernier calipers, a micrometer and a comparator based on specified measurements. An additional process of functional & IR testing is completed on 2 materials. Functional sees if the parts work & IR checks if coating is correct. These measurements are the CTQ items.
The goal of this project is to reduce the cycle time and variability for the inspection process of our materials 1, 2, 3, 4 and 5. These are manually inspected in our Quality Control lab. Inspection time varies from material to material.
Current times per material and proposed future test times are shown in Fig 1 below. The project will save €1,000 per month across the 5 materials.
The goal of this project is to reduce the cycle time and variability for the inspection process of our materials 1, 2, 3, 4 and 5. These are manually inspected in our Quality Control lab. Inspection time varies from material to material.
Current times per material and proposed future test times are shown in Fig 1 below. The project will save €1,000 per month across the 5 materials.
Fig 1
A SIPOC was carried out on 20th and 28th January 2016, see Fig 2, I facilitated with the inputs from two QC lab personnel. The findings for me from the SIPOC were that the technicians fill out two logbooks and that functional and IR (infra-red) was needed on some materials. Numerous forms and handovers were needed for the process. Documentation & cert checks take a considerable length of time too. A high level SIPOC process flow is shown below in Fig 4.
A VOC was sent to QC lab and our production team, see Fig 3 above. 3 questions were asked, (What is going well?, What is not good/ and What is ideal state?). There was over-processing of inspection report information between departments. The waste of over-processing to sign out the materials before testing and also re-work by production if original orders were not released on time by the lab. Ideally the lab would want a better schedule to prioritize their inspection work.
The 5 DMAIC phases of the project and timelines are shown in the Gantt chart Fig 5 above.
Six Sigma Project: Measure phase Feb 2016
Introduction: In the measure phase of my project we wanted to see exactly what the current state is. Only from this can we know, how bad or good the current system is after initially defining our problem.
Firstly we got a report called a BI (Business intelligence) report which links with SAP and shows us if the different component tests have met their target release dates. This information was from 2015 and 2016 and there were 59 data points shown in the report for the various materials. For these type of components the test release time is 28 days. This date is internally set.
A graphical summary was ran with this data on Minitab with the results of mean of 48 days, median of 41 days and standard deviation of 44 days. P-value is extremely small (<0.005) and therefore this tells us that the data does not follow a normal distribution. The data is very spread.
Medians are usually a better indicator of central tendency for skewed data.
Fig 6 shows us the histogram of the data and we can see that the data is skewed to the right and shows us that the data is not normal. 28 days release time is shown also. Fig 7 shows us a Minitab boxplot of the data. We can see very clearly from this graph that >75% of the data is above the target of 28 days. Also outliers (1.5 X IQR) are shown at approx 150 days plus.
Fig 8 shows us an individual data plot for the 59 data points. UCL is calculated by minitab but a lower limit (LB) of zero days was inserted. We can see the 4 outliers based on these control limits and the target of 28 days is also shown. The data mean of 48 days is above this figure. Fig 9 shows us a time and motion study completed by QC lab which gave us a total time of 9 hrs 17 mins. Visual inspection is a very large section of this test. It currently is 62% of total test time.
Right first time data for 2015 was viewed at 95% so it might be tough to improve this percentage just through this project as SAP and other tests come into account in the calculations.
A graphical summary was ran with this data on Minitab with the results of mean of 48 days, median of 41 days and standard deviation of 44 days. P-value is extremely small (<0.005) and therefore this tells us that the data does not follow a normal distribution. The data is very spread.
Medians are usually a better indicator of central tendency for skewed data.
Fig 10
Fig 10 above shows the current state map. I facilitated a working session with four QC personnel and we went through the various steps in the process. For each step we listed the numerous problems that were occurring which increased the cycle time. We saw how paperwork is left in documentation pigeon holes and can be left for days until the next person reviews. There can be numerous sheets in each section and the forms are reviewed on a FIFO basis. This waste of waiting time is significant and is not captured in our time and motion study. Over-processing of batch details throughout different forms was also listed. I now know where the end time is for our 28 days i.e. where the clock stops for this component release process.
Right first time data for 2015 was viewed at 95% so it might be tough to improve this percentage just through this project as SAP and other tests come into account in the calculations.
Conclusion: We can see that the release time for components exceeds the target of 28 days. Such wastes are over-processing results on paperwork will have to be removed. Also component inspection is only one part of this process where as other parts will not be covered by this green belt project. We now know in relation to our measurements, "where we are now".
Six Sigma Project: Analyse phase Mar 2016
Introduction: In the analyse phase of my project we wanted to see exactly what the causes are which affect key inputs to the current process. These factors cause delays in our test cycle times.
The first items that we wanted to analyse was the test itself and the movement and people involved in it. To do this I created an operator balance chart which shows the test steps and the personnel movement for the process. In the left of the sheet you can see the necessary steps conducted through the process by operator and the 2nd person. In the middle of the sheet it shows a time for that step & this builds up to the total cycle time at the bottom of the sheet, 11 hrs 30 mins. For this test the manual times are shown in yellow while walking time is shown in green. A spagetti diagram is shown on the right of the sheet to show movement in the lab.See Fig 11 below.
The other major improvement is to introduce reduced visual inspection. We can significantly reduce the inspection numbers if a certain amount of tests have been past historically. Other improvements to be done are revise 5S in the area, review the documents themselves and stagger calibrations of tools.I facilitated a future state mapping session and we listed the improvements at the various process steps. Some ideas could not be implemented. See Fig 18. As you can see the steps are still the same bar one but there are numerous improvements at each stage. Through a side paperless project in QC, which I am involved with, a documentation step is eliminated (marked with a X on map). We will also reduce our documentation pigeon holes from 27 down to only 9 through this paperless project.
An Operator balance chart was created for the new process, see Fig 22. You will see that the overall cycle time has come down to 11 hrs 29 mins from a before of 28 hours (to be shown in final report). This is a total saving of 61% and will save the QC lab 15 working hours.
The first items that we wanted to analyse was the test itself and the movement and people involved in it. To do this I created an operator balance chart which shows the test steps and the personnel movement for the process. In the left of the sheet you can see the necessary steps conducted through the process by operator and the 2nd person. In the middle of the sheet it shows a time for that step & this builds up to the total cycle time at the bottom of the sheet, 11 hrs 30 mins. For this test the manual times are shown in yellow while walking time is shown in green. A spagetti diagram is shown on the right of the sheet to show movement in the lab.See Fig 11 below.
Fig 11
The next step in the process was to see which section of the test took the longest time. For this I completed a pareto diagram through minitab. Through this we can see that the 2 biggest time consumers are visual inspection at 42,5% and dimensional inspection at 18.8%. Combined these two sections are 61.3% of the test time. Eye breaks is 5.1% which is due to the large number of parts been visually inspected. Every 30 minutes the tester has to take a 5 minute eye break. See Fig 12 below for the pareto diagram.
Fig 12 
Our next step in analyse was to find out what the root cause(s) were which caused the delays in the process. I facilitated a RCA (root cause analyse) session. In the left of the sheet we defined the problem. It is the what, where, when and who of the problem. To the right of this we showed the high level process map of the testing procedure on post-its. To the right of this we have the fishbone diagram. The legs are manpower, method, machine, environment, measurement & materials. From the flow chart, potential causes were placed in the relevant fishbone leg. Fig 13 below (left) shows the RCA sheet. Possible causes were marked with an X. Items considered as "noise" marked as N and items in "control" marked with C. Some causes listed were numerous handovers, excessive sample sizes and over-processing of written data. Overall looking at the fishbone we can see that there are numerous causes to delays in the testing process.
Fig 13 Fig 14
We picked one X from the method part of the fishbone diagram and completed a 5 whys on it. You can see from Fig 14 above (right) that there are two potential root causes for this fault. The 5 whys helped us to drill down further on the issue. 5 whys were not applicable for other X's.
Conclusion: The overall view of this was that there were numerous root causes to the problem of long test time. The analyse phase brought out these wastes. We can see from the pareto that potentially if we could eliminate some of the visual inspection time then it would save us considerable time. Solutions to the problems will come in the next phase, "improve".
Six Sigma Project: Improve phase Apr 2016
Introduction: In the analyse phase we have found the root causes to the various problems which are causing delays in the cycle time for component testing. In the improve phase we will now brainstorm to see what the possible solutions are, so we can improve our process. We will see then if some solutions are feasible or not and gauge our improvements.
The first step in "improve" is to view the various problems that we have gathered from our current state map and our fishbone diagram. I facilitated a session with QC personnel and we created an affinity diagram. To create the affinity diagram layout we brainstormed and wrote all possible solutions on post-its. Every idea was considered. When these were all wrote out we grouped the ideas into relevant themes. The themes we decided on were, people, documentation, QC lab area, tools and systems/processes. Fig 15 below shows the affinity diagram that we worked on.
The first step in "improve" is to view the various problems that we have gathered from our current state map and our fishbone diagram. I facilitated a session with QC personnel and we created an affinity diagram. To create the affinity diagram layout we brainstormed and wrote all possible solutions on post-its. Every idea was considered. When these were all wrote out we grouped the ideas into relevant themes. The themes we decided on were, people, documentation, QC lab area, tools and systems/processes. Fig 15 below shows the affinity diagram that we worked on.
Fig 15 
The next step was to create an action list with the various actions from the affinity diagram that are feasible and will improve the process. Actions are assigned to the relevant people with proposed close out dates and status blocks on the right to view progress over their implementation. See Fig 16.
Fig 16
One of the major actions out of this plan was to install go-no go gauges. For a number of the components we designed these gauges with a vendor, see Fig 17 (left) below. When testing, the part should "go" in through the top section of slot i.e. upper spec limit and then it should "not go" into the lower section of slot i.e. lower spec limit. Some of the components are too soft therefore parts would be distorted if they were forced into the gauge. The 2nd action was to get a stamp which we could use to put the batch number on each page of the inspection report. See Fig 17 (centre). Before we had to write the number on 25 pages. We also bought overlays for our comparator. The part is magnified and then we can see from the overlay on the screen if the part is within the spec limits. See Fig 17 (right).
Fig 18
Conclusion: After finding our root causes from our analyse phase we have now decided on improvements to our process. These will significantly reduce cycle time for our inspection process. In the control phase we will proceed to quantify and embed these solutions through the systems that we have on site.
Six Sigma Project: Control phase Apr 2016
Introduction: In the improve phase the solutions had been generated. In the control phase we now want to embed the solutions and maintain these gains. We will quantify the improvements which we have gained through this project.
A part of our action plan we listed 5S as a solution to maintaining a good work area. A new 5S owner was put in place in the area. He was brought through the 5 stages, sort, set, shine, standardise and sustain. While an area looks great when it is neat and tidy, we have found that without regular audits 5S cannot be maintained. Fig 19 below shows example of "set" (putting materials where they are needed) and also our 5S board with room layout, room pic and room rules, 5S owner and audit sheet.
A pareto diagram below, Fig 21, shows the before and after times so it clearly can be seen that visual has been reduced considerably. Dimensional times and eye breaks have now been brought to more manageable levels too. Other areas of the testing have only gained small increment reductions.
A part of our action plan we listed 5S as a solution to maintaining a good work area. A new 5S owner was put in place in the area. He was brought through the 5 stages, sort, set, shine, standardise and sustain. While an area looks great when it is neat and tidy, we have found that without regular audits 5S cannot be maintained. Fig 19 below shows example of "set" (putting materials where they are needed) and also our 5S board with room layout, room pic and room rules, 5S owner and audit sheet.
Fig 19
With installing the solutions of a numerical stamp, reduced visual inspection and reduced dimensional inspection (based on positive test history and then using ANSI tables) and introduction of go-no go gauges we have developed a new operator balance chart. Also as an additional gain, because visual and dimensional times are reduced we have reduced the number of required eye breaks. Note: this component is different than the component used in the analyse phase. The bar chart, Fig 20 below shows the reductions in the various steps throughout the process. Before and after times are listed in the table to the right.
Fig 22
Conclusion: The project has been a success in the reduction of testing cycle times. We will work through outstanding actions over the next few weeks. A capacity analysis of the testing in the QC lab will be done now so that we can see how we can level load our testing demand based on the production schedule. Lessons learned were that weekly meetings were a must and sponsorship which I got from the QC manager was a brilliant advantage.
No comments:
Post a Comment