Capability Studies in Engineer to Order Small Batches

[Martin_v_v]

Dear all,

Currently I am working for a manufacturer of large cylinders. Our company produces cylinders which are at max. 70 meters long and can weigh as much as 80 tonnes. Demand for any particular size cylinders is limited in this branch, which for us means that a customer need is translated into a drawing in which all critical cylinder sizes can vary. (This is the engineer to order part) To stick to the length example, the maximum cylinder length is 70 meters but the minimum is 2 meters. We literally produce every size between 2 and 70 depending on customer demand. Another characteristic, depending on customer usage our cylinders are designed to operate between five and one-thousand bar.

Dealing with demand in this way has lead to average batch sizes between 2 and 3, also over a longer period the amount of times an exact cylinder design is repeated is low. Last but not least all our machines are only able to process one piece at a time. So even when we produces a batch of 2 cylinders, in our process every machine will only process a single piece of that batch at a time. Now the great challenge, how do we assure quality in these products?

We want to apply something similar to Process Capability Studies, in which we aim at the trigger function of Capability studies. We want to know when the process is in danger of going out of control so that we can timely undertake action so that we prevent greater damage and at the same time lower the costs of our internal defects.

Do you have any thoughts of how to monitor the capability of the process in our situation?

 

[Curtis317]

Martin, Much depends on what you measure for each part. Short Run SPC concepts can help you get where you need to be. If the diameters of the cylinders all have the same or close tolerances you can use coding to standardize the measurements so you can put them on SPC charts and do capability studies. It has been some time since I have used short run SPC but the concept is if you have a 1 ft and a 2 ft dia cylinder with a tolerance of +/- 1 inch you can chart both parts just by charting how they measure and subtracting 1 ft. That way both diameters can be charted. There is much more with Short Run SPC and others may be able to provide more information.

 

[Martin_v_v]

Thank you for your reply Curtis.
Short run SPC has proven to be a rather good lead for us.
However now we have run into the following.

To assure a machine is able to achieve the planned results one can perform a machine capability study. It is basically the equivalent of a process capability study but now the other variables are kept equal as much as possible so the variation of the machine, so the natural variation, can be determined.

In our line of business we would also like to have the same certainty, we would like to know a machine is capable of achieve the planned result. However we cannot afford to perform test runs due to the prices of material. Further more I feel that doing such a test run would not tell us much about the machine because we have to make a lot of changes to it before we can run the next part.

Does anyone here have any thoughts or experience in how to determine whether a machine is capable or not when working with small batches like we do. Should we even aim at this?

 

[Curtis317]

Martin. Look into the thread SPC Analyais and Techniques there is a long thread discussing this subject with files for further information.

 

[Darius]

Maybe this tread can help Z Charts

It doesn't say a thing about Capability index, but if you transform the data to Z values, the target must be 0 and the Lower and Uppper specs +/-3.

 

[Martin_v_v]

Hi Darius,

It helps with the initial question of this thread. Thanks for that.

Basically my current situation is like this.
I have distinguished 4 moments of a machines life with the question we want answered. The first 2 of these moments I have mentioned below, because they are the ones that are giving us the most trouble.

1. Pre delivery, the machine is at the machine supplier.
- Is this machine capable of producing parts to specifications?
- Can it do that repeatedly?
- How much variation does the machine have?
2. Pre production, the machine has been placed in our production facility (also applies after large maintenance)
- Is this machine capable of producing parts to specifications?
- Can it do that repeatedly?
- How much variation does the machine have?

The other 2 moments are when the machine has been taken into use. Once it has been taken into use we have several methods we can use to monitor its performance. The just proposed Zed Chart might be a good addition to that. (However so far that judgement is made purely on one article.)

Most likely we can either run test batches, but we don't want to spend money on so much test material. The other option is to focus on proces parameters, but will that method give good answers to the mentioned questions.