significance of 3 sigma and 6 sigma in manufactured company

Re: significance of 3 sigma and 6 sigma in manufactured comp


The terms come from the definition of a normal distribution. A lot of manufacturing processes produce normal distributions, so you can use the number of "sigmas" away from the specification to discuss process distribution and defect rate.

Example: Say you are trying to make widgets (a very common manufacturing example) with a length of 50 cm. You promise the customer that your widgets will be between 49 and 51 cm. That's the specification. You then do an audit on your process and find that you are making widgets with a length of 50.2 cm and a standard deviation of 0.3 cm. You also do a normality test and see that the distribution is normal. Standard deviation is designated by the Greek lower case letter "sigma". The distance between your mean and your closest specification is 51-50.2 or 0.8 cm. This represents a 2.67 sigma process. (0.8 / 0.3). At that level, you are making some level of failing product and you have to measure every widget you make to ensure the customer of satisfactory product.

Seeing that your process is not centered and that you are making some level of out of specification product, you adjust your manufacturing process to center up the distribution, but you still have the same variability. Now the distance to the nearest specification is 1.0 cm since you have a mean of 50 cm and your sigma is still 0.3 cm, so you are running a 3.3 sigma process. At that level, you fail very little and can reduce your measurement process down to sampling product occasionally to ensure that your process is staying on target. You still might make a failing widget every now and them, but it is down in the parts per thousand level.

Because the competition is ever improving, you make some process improvements and now your process is much more consistent. You are still on target at 50 cm, but now your sigma is only 0.167 cm. You are now running a six sigma process. Your failure rate is down in the parts per million. Your customers are very impressed because when they put your widgets in their process, their process (which is optimized to run with 50 cm widgets) runs very smoothly. Because you took the variability out of your process, they can have more in theirs. You are getting more orders and manufacturing at a lower cost since there is less waste in failed widgets and less measuement costs. Your commercial guys see how well you are doing and propose a tighter specification to your customers. Your competitors scramble, measure more and their costs go up.

Of course if you're the one running a three sigma process when your competitors hit six, you are in trouble. Also, when the specs get tighter, you are no longer at six sigma and you have to go back to improving the process.