Machined components

The operations manager was speaking calmly to the marketing manager - ‘I said it usually takes 70 days to make a batch of these components. We have to buy parts and materials, make subassemblies, set up machines, schedule operations, make sure everything is ready to start production - then actually make the components, check them and shift them to the finished goods stores. Actually making the components involves 187 distinct steps taking a total of 20 days. The whole process usually takes 70 days, but there's a lot of variability. This batch you're shouting about is going to take about 95 days because we were busy working on other jobs and couldn't start immediately - and a major production machine broke down and we had to wait for parts to be flown in from Tokyo and that took another 5 days. It's your fault that you heard my estimate and then assumed that I was exaggerating and could promise the customer delivery in 65 days.' The marketing manager looked worried. ‘Why didn't you rush through this important job? Why is there such variation in time? Why did the breakdown of one machine disrupt production by so much? What am I going to say to our customer?' The operations manager's reply was, ‘To answer your questions in order. Because I was rushing through other important jobs. The variation isn't really that much; our estimates are usually within 10 days. It is a central machine that affects the capacity of the whole plant. I can only suggest you apologise and say you will listen to the operations manager more carefully in the future.' Despite his apparent calmness, the operations manager was concerned about the variability in production times. He could see why there was some variability, but the total amount for the component they were considering did seem a lot. As an experiment, he had once tried to match capacity exactly with expected throughput. Then he found that operations near the beginning of the process performed reasonably well, but towards the end of the process the variability was magnified and the throughput times went out of control. At one point he had eight machines in a line, each of which processed a part for 10 minutes before passing it to the next machine. Although this arrangement seemed perfectly balanced, he found that stocks of work in progress built up dramatically. Some people suggested that this was because the actual processing time could vary between 5 and 15 minutes. Whatever the reason, the experiment was stopped.

Question

1. Operations really need a study to see why there is variability, how much is acceptable, what its effects are, how it can be reduced, what benefits this will bring and so on. Such a study needs funding - and your job is to write an initial proposal for this funding, including a detailed proposal for a larger study.