C
CFong
Does anyone have suggestion on what kind of work sampling (random sampling, stratified random sampling, systematic sampling, rational subgrouping) that I should use for the case below?
I am working on a six sigma project at the melt department for my company (foundry). One of the goals is to reduce the cycle time to 60 minutes or less. I am trying to do time study to determine the cycle time before improvement. There are two melters ran by two operators simultaneously. These two melter will melt raw materials and tap the molten iron into a holder at 2775F. The holder will maintain the iron temperature and supply the iron to two production lines when requested. We run three shift and 5 days a week. Sometimes we are scheduled to run production in the weekends.
The melter that we use is induction type and its capacity is 15.5 ton per heat (we called it heat instead of melt). I had done a time study for a chosen day and came out with the average of 68 and 71 minutes (average of 20 heats) cycle time for each melter. The heats per day that we produce varies from day to day depending on the time that the holder is sitting full waiting for production line to request for iron. But, sometimes the holder runs out of iron due to heavy demands. We are going to exclude these delays since it has nothing to do with the cycle time.
The process is broken down to 6 activities. The activitiy that consumed the most time is melting the metals. Therefore, we are going to work on raw material charging method that will reduce the melting time.
Since there are many factors that affect the cycle time, I want to expand the sample collection to 6 months period to produce a standard cycle time that we can use as baseline before improvement. We have a system that store the activities’ time that I can mine, but it will take forever to convert the entire data.
Therefore, my question is how many random samples is enough to produce a representative cycle time that has 90% confidence level? Do I need to consider things such as weather effects, start up after not working day, etc? How do I plan the sample collection method?
I am working on a six sigma project at the melt department for my company (foundry). One of the goals is to reduce the cycle time to 60 minutes or less. I am trying to do time study to determine the cycle time before improvement. There are two melters ran by two operators simultaneously. These two melter will melt raw materials and tap the molten iron into a holder at 2775F. The holder will maintain the iron temperature and supply the iron to two production lines when requested. We run three shift and 5 days a week. Sometimes we are scheduled to run production in the weekends.
The melter that we use is induction type and its capacity is 15.5 ton per heat (we called it heat instead of melt). I had done a time study for a chosen day and came out with the average of 68 and 71 minutes (average of 20 heats) cycle time for each melter. The heats per day that we produce varies from day to day depending on the time that the holder is sitting full waiting for production line to request for iron. But, sometimes the holder runs out of iron due to heavy demands. We are going to exclude these delays since it has nothing to do with the cycle time.
The process is broken down to 6 activities. The activitiy that consumed the most time is melting the metals. Therefore, we are going to work on raw material charging method that will reduce the melting time.
Since there are many factors that affect the cycle time, I want to expand the sample collection to 6 months period to produce a standard cycle time that we can use as baseline before improvement. We have a system that store the activities’ time that I can mine, but it will take forever to convert the entire data.
Therefore, my question is how many random samples is enough to produce a representative cycle time that has 90% confidence level? Do I need to consider things such as weather effects, start up after not working day, etc? How do I plan the sample collection method?
