Statistical Quality Improvement Action for Small Batch Production

[Nixverstehikus83]

Hello ,

I hope that I am writing in the right topic of the forum. I searching for a statistical quality improvement method for small batch production?
For example: You can´t use SPC for small batch production, or? What other opinions are there? If you produce one until three pieces.

thank you in advance

nixverstehikus

 

[Darius]

You can?t use SPC for small batch production, or? What other opinions are there? If you produce one until three pieces.

Maybe (at least the closest I can think of)..

Short Run or Zed Charts

 

[Steve Prevette]

I would not be so quick to say you "can't do SPC" on small batches. Even as few as five data points can be the basis for an SPC chart.

To give a longer time view of improvement there are several options:

1. Across many batches, plot the difference between a critical specification (or specifications) and actual and trend using SPC. Chi Square / Hoteling type of data transformations may be useful.

2. If a specific item is made sporadically, just have a chart for that product, adding data point(s) whenever the product is made.

 

[TWA - not the airline]

Not sure if I understood your question right, but whenever I hear quality improvement I think of the DMAIC cycle. You can use appropriate statistical methods for all sections of it.

 

[Bill McNeese]

I think Steve has the right approach. You can use deviation from target if you seldom make the "same" thing. For each batch, there is a target value. Then you determine actual - target and plot that over time using an individuals control chart or an Xbar-R control chart if you want to subgroup the data.

If you do make repeat products over time, the z control chart would be a good approach.

Bill

 

[mdurivage]

I agree with Steve and Bill. In the past, I worked with a chemical manufacturing process. There were around 30 different 'formulas'. I charted percent from nominal to monitor the process.

 

[Nixverstehikus83]

Hallo ,

thank you very much for your answers. The background of this question is that I will have an Interview for a Job. And one of the the question in this interview is:
Discuss a statistical method for quality improvement in use for small series production in mechanical processing.

The first critical point is that it has to be a statistical method. So methods like FMEA, QFD are out of question. I know statistical methods that are in use in the Qualitymanagement: SPC, machine capability and process capability.

The next critical point in that question ist that the method should be used in small series production. As much as I know the methods like SPC, machine and process capability are used in large series production.

Are there other statistical methods that can help to improve the quality and can be used in small series production? It is project manufacturing. For example you produce a ball. The first customer want 3 balls with a diameter of 56+-0,1 Inch the next customer want 4 balls of the diameter: 55 +-0,05 Inch. So the specification is from customer to customer a another one.

thanks in advance.

 

[Bill McNeese]

Hello,
How do you know the interview questions ahead of time? Kind of odd.

Anyway, I believe the posters gave you some ideas on the SPC side. So, a customer wants a ball with a diameter of 56. You make a ball with a diameter of 56.4 So, the deviation from targe is 56.4 - 56 = 0.4. Another wants a ball with diameter of 55. You make one with a diameter of 54.8. So, deviation from target is 54.8 - 55 = -0.2. The plot the deviation from target values over time as a control chart.

 

[Steve Prevette]

Again, I don't understand your misconception of SPC. I strongly suggest you read up on Hotelling distributions.

 

[reynald]

Even as few as five data points can be the basis for an SPC chart.

Hi Steve,
Care to expound on this? This is totally new to me. Thanks.

Edit: Basically I was in the impression (from Montgomery) that you need about 25 points to establish the limits.