Process Capability study of CNC Machining Process

Hypothetical Question: Let's say I'm CNC machining a 1000 piece order for a new part and it has 3 key dimensions that require "inspection". How would you go about setting up and conducting a process capability study on that job?

I guess the main thing I need help with is determining how many parts to check when performing a Cpk study and at what frequency to check them?

My current method is listed below:

1) First, we dial in the CNC program, setup, speeds and feeds until we're able to produce a good part. The program is then commented with setup, speed & feed info and saved in the DMR (Device Master Record) for the part/part family! It is also referenced in the appropriate operation on our routing for that part (which is created in our shop management ERP system).

2) Second, we develop a detailed inspection report and we inspect the initial order per Z1.4, level II, AQL = 1.0. If we get a repeat order and we know we're in tight control and would like to reduce our inspection frequency, we'll take our Z1.4 results (from the first job) and analyze them to determine if our Cpk is acceptable.

I'm looking for some assistance in devising a simple procedure for conducting process capability studies on the parts we machine and could use some help because I don't have a firm grasp on how to execute a process capability study from a practical standpoint.

If you are doing precision machining with your CNC process, I recommend reading the following:

SPC for precision machining

If you have the correct sawtooth curve and it falls within your control limits - which means your process is in control, you are capable. If you have a normal distribution, you have incorrect sampling, or you are out of control. Details are in the link.

It seems that you are doing batch production. The best you can do is Ppk. You are not including all of the variables in your samples over time, which would include set-up variation. Your samples do not include enough of the long term variation to calculate a valid Cpk.

George Holz said:

It seems that you are doing batch production. The best you can do is Ppk. You are not including all of the variables in your samples over time, which would include set-up variation. Your samples do not include enough of the long term variation to calculate a valid Cpk.

Click to expand...

Actually, if he is doing precision machining with the CNC operation, neither is applicable. If the process is in control, the results should be non-normal, so those capability indices are not applicable. Also, the use of AQL inspection is not good for production. The sampling rate should be a function of the tool wear slope and its intersection to the control limits. All of this is discussed in the link above.