Machine Capability Analysis in Multi-Dimensional Cases

[isbalanli]

Hi everybody
As a newcomer I have the question about machine capability analysis.

1) We have 4 CNC machines. Then we machining one reference part (reference part means: all parts have the same shape), three times in each machine. This meeans that we have 12 machined parts. [cosider a 12x500 table]

2) Then we measure all 12 parts with CMM. As measured points, we have 500 points in each part. (we have the CAD model. This means that for all points we have the nominal value.)

The question is:
How to calculate the capability of each machine?
How to compare the machine's performances?

Thanks in advance for your help.

 

[tbsiva]

As I dont have experience in CNC machining, I will try to explain the answer with an example from injection moulding. I am not sure whether my answer will help you.

Even in my case, a injection moulding part has 100 dimensions. To calculate the machine capability, we need to produce 50 samples. So, I need to measure 50 X 100 = 5000 dimensions/machine. We have 50 machines. So, to calculate machine capability for 50 machines, it will be 5000X50 machines = 250000 dimensions, which is practically impossible.

What we have done is we have decided to measure the part weight, because consistency in part weight is the parameter which defines the performance of the machine. That means we have converted 250000 dimensions into 1 X 50 X 50 = 2500 measurements of weight, which is a much simpler and effective process

Coming back to your concern, Is it possible to define one or two parameters which is the most vital parameter that will effectively say that my machine is capable or not? If yes, identify those parameters and conduct capability study for those charateristics alone.

 

[isbalanli]

As I dont have experience in CNC machining, I will try to explain the answer with an example from injection moulding. I am not sure whether my answer will help you.

Even in my case, a injection moulding part has 100 dimensions. To calculate the machine capability, we need to produce 50 samples. So, I need to measure 50 X 100 = 5000 dimensions/machine. We have 50 machines. So, to calculate machine capability for 50 machines, it will be 5000X50 machines = 250000 dimensions, which is practically impossible.

What we have done is we have decided to measure the part weight, because consistency in part weight is the parameter which defines the performance of the machine. That means we have converted 250000 dimensions into 1 X 50 X 50 = 2500 measurements of weight, which is a much simpler and effective process

Coming back to your concern, Is it possible to define one or two parameters which is the most vital parameter that will effectively say that my machine is capable or not? If yes, identify those parameters and conduct capability study for those charateristics alone.

Thanks for answer. Good idea. But the problem is that, when I talk about dimension, they are scanned parts. So I have 500 point cloud for each part. Reverse engineering and making CAD model to calculate the weight is so time consuming process. (also sampeled points are not dense enough to lead the accurate answer) .Of course I have calculated the differences between all nominal (CAD) and real (scanned) parts. (bofore I had [x,y,z,i,j,k] for each point and I calculated the differences for obtaining the scalar number instead of torsors with 6 elements).