Saturday, 15 March 2014

Improve Phase


The Improve phase of the project kicked off on Feb 24th. The team gathered all the information from the various tests that were conducted and decisions were made on the implementation of the fixes. We found that the tools in the Improve Phase were not as abundant as they were in the other 3 phases. Our issues did not suggest we conduct  a DOE so we opted for the following tools:


1. Chart of Condition - to graph the results for "channels in the Hub"
2. Bar Graph - to graph the differences between the flaring mandrel and the tapered mandrel
3. Box Plot - to show there was no difference in mandrels using tensile at 10 N

Channel in the Hub

We started working on the biggest issue on the line "channels in the hub". The PD Engineer worked with the team to solve this issue. A series of tests were conducted on the station with the Product Builders and the testing proved to be quite a long process.  The confidence graph below shows that when the units were not put through the plasma step there was on average 75%  failure rate whereas when the plasma was applied to the units we had a 100% pass rate. Following implementation of the plasma process on the line the scrap rate for "channel in the hub" dropped from 75% to 0%. (sample of one test conducted below)

















Glue in the flow path 

In parallel  with the channel issue we also worked on the issue of "glue in the flow path". Below is a production inspection form filled out by the Product Builder conducting one of the trials on the line. These inspection forms are used in my company to record information. It is a controlled document which is printed with a date stamp and reviewed by quality. (omitted names for this purpose)

The bar chart below shows the difference between the flared mandrel and the tapered mandrel for a sample of 15 units. As shown below the flared mandrel had a scrap rate of 60% for glue in the flow path and the tapered mandrel had a scrap rate of 0%.




















The box plot below also shows that there is not a significant difference between the flared mandrel (balloon tubing mandrels) and tapered mandrel for tensile hub performance.   
The samples built with both mandrels were tensile tested to ensure they were not having an adverse effect on the tensile performance of the product. All units passed tensile testing and had a higher tensile force then the product hub tensile specification of 10N. 


















Kinks

In conjunction with the other two yield issues on the line we looked at the kink issue. A number of fixes were put in place for this issue. After conducting a Process Flow of station 6 we were able to high light area's for improvement. Below are  sample photos of before and after to help with this issue. We also in produced two improvements to the process,removal of balloon tubing and hooping units at station 5 and securing the hoop unit at station 6. 


Before - Old Rack heads. Hubs sitting on top of rack




After - New rack heads. Units fitting snugly on racks



All the tools we used in this phase were quite easy to use and they gave us the results we required. The improvement in yield has been significant and the line has gone from being at a RTY of 22%  to a RTY of 79%. ( there is only 1 type of product now running on line,the CE product) Further improvements are being implemented  in the form of a new plasma machine which will give an additional improvement of 3-4%. (currently the plasma machine is on another line) 
Our overall experience with the DMAIC process has been enlightening and has taught us that the tools we used in this process enabled us to reach the improvements we have seen to date. 
All improvements have been updated in the relevant procedures.

The final step in the process is the Control Phase. 


Daily Yield Pareto – monitor the yield on a daily basis to ensure issues are not re-occurring. Action as appropriate.
Process Map - if the line yield dips below the 85% yield 
Team meetings - weekly updates on yield to Product Builders to keep the focus








No comments:

Post a Comment