Macro Operations Case Study
Thursday 24th May 2018

Case Study #5


A factory employing 300 people had been laid out as a flowline with 5 process steps.
Since the time required at each step varied (by up to 15:1) the flowline could never work. It looked like the M25 on a Friday.


Operations modelling on a macro scale

This factory repaired product

It didn't make product, like a conventional factory.
Since the time required at each step varied so much a conventional flowline could never work. Flowlines require predictable cycle times. The clue was staring us in the face - repairs take variable times anyway, and sometimes we discover further faults which then take even longer.
The response had been to put items into store between each step - to break the flowline, and effectively run 5 independent factories.
No downstream process had sufficient confidence to start work until they could 'see, feel and touch' the WIP, so no item cleared 5 steps in less than a week without (manual) queue jumping.
The WIP store had become a bottleneck, limiting the throughput.
Costs and overall process times had spiralled.
As a standalone operation, the factory was bankrupt. Part of a large PLC, it was dragging the group under.

The Brief

This was open ended, since the client knew our capabilities.
"Help, see what you can do. But do it quickly!"
The pressures were to bring costs down and to speed process times to the point where 'queue jumping' would not be necessary.
An open minded brief is a perfect place to start, since this allows us to follow through the unexpected insights which modelling always produces.

Modelling and Analysis

Taking estimated and actual process times for ~3,000 items, we randomised arrivals and ran them through the 5 process steps under a variety of 'rules'. We froze the random sequence until we understood which causes linked to which effects.
We then re-randomised the inputs to simulate 330,000 items.
We ran 102 different scenarios

The Results

Tiny changes in 2 parameters had enormous impact on minimum, average and peak process times. One of the parameters - the extent of cross training - had been dismissed out of hand as 'too expensive'. We showed that 10% cross training and a change in inter-process queue management would solve 90% of the problems.

The Outcome

It would be nice to report total success, but this is the real world!
Our study established a theoretical minimum cost, which was below that being paid to the factory operator.
Showing that he could never make money under any circumstances gave him the ammunition - and, it must be said, confidence - to negotiate a price increase.
The factory and group survived, but continue to process by their old method.
The price increase removed the incentive to improve.


Coincidentally we proved that the amount of queue jumping meant the market price of a fast repair was too low.

We might have a similar problem, please contact us

An expert knows tomorrow why the things he predicted yesterday didn't happen today.
Home | About Us | Showcase | Research | Cases | The Vaults | Tips & Quips | Contact Us …
© 2002 - 2018 Supply Chain Tools Ltd.