The Threshold Of Uncertainty
October 30th, 2013
Our threshold for uncertainty is too low.
Early in projects, even before the first prototype is up and running, you know what the product must do, what it will cost, and, most problematic, when you’ll be done. Independent of work content, level of newness, and workloads, there’s no uncertainty in your launch date. It’s etched in stone and the consequences are devastating.
A zero tolerance policy on uncertainty forces irrational behavior. As soon as possible, engineering gets something running in the lab, and then doesn’t want to change it because there’s no time. The prototype is almost impossible to build and is hypersensitive to normal process variation, but these issues are not addressed because there’s no time. Everyone agrees it’s important to fix it, and agrees to fix it after launch, but that never happens because the next project is already late before it starts. And the death cycle repeats project after project.
The root cause of this mess is the mistaken porting of manufacturing’s zero uncertainly mindset into design. The thinking goes like this – lean and Six Sigma have achieved magical success in manufacturing by eliminating uncertainty, so let’s do it in product design and achieve similar results. This is a fundamental mistake as the domains are fundamentally different.
In manufacturing the same product is made day-in and day-out – no uncertainty; in product design no two product development efforts are the same and there’s lots of stuff that’s done for the first time – uncertainty by definition. In manufacturing there’s a revision controlled engineering drawing that defines the right answer (the geometry and the material) – make it like the picture and it’s all good; in product design the material is chosen from many candidates and the geometry is created from scratch – the picture is created from nothing. By definition there’s more inherent uncertainty in product design, and to tighten the screws and fix the launch date at the start is inappropriate.
Design engineers must feel like there’s enough time to try new things because new products that provide new functionality require new technologies, new materials, and new geometries. With new comes inherent uncertainty, but there are ways to manage it.
To hold the timeline, give on the specification and cost. Design as fast as you can until you run out of time then launch. The product won’t work as well as you’d like and it will cost more than you’d like, but you’ll hit the schedule. A good way to do this is to de-feature a subassembly to reduce design time, and possibly reduce cost. Or, reuse a proven subassembly to reduce design time – take a hit in cost, but hit the timeline. The general idea – hold schedule but flex on performance and cost.
It feels like sacrilege to admit that something’s got to give, but it’s the truth. You’ve seen how it goes when you edict (in no uncertain terms) that the timeline will be met and there’ll be no give on performance and cost. It hasn’t worked, and it won’t – the inherent uncertainty of product design won’t let it.
Accept the uncertainty; be one with it; and manage it. It’s the only way.
Posted in 
Tags: 2 Responses to “The Threshold Of Uncertainty”
Leave a Reply
Mike Shipulski
Prototype shops are obviously going to be a different case. Or it wouldn’t be called a prototype shop.
Perhaps Mike’s statement doesn’t apply to your case, but it certainly does to many others’ manufacturing situations. In our case, our manufacturing facilities may run multiple products down the same lines, but the statement in general holds true. Unless there is a major refresh or new launch, the product is pretty stable. Volumes of one product vs another another may shift, but that’s part of managing the uncertainty.
Resource management, now that’s another story.
I believe Mike’s statement to be most likely based on what he sees at the company he works for. My background is quite a bit different than Mike’s because I have spent most of my life working for machining job shops. Things are very different in modern machining job shops today than they were even a few years ago. The way parts are machined is very different and very short runs of a few parts are the norm not the exception.
The demands of a modern, high tech machining job shop, that does very short runs of varied parts, can be even more demanding than a prototype machine shop because you have to manage so many different parts and part programs! Change is constant because the way you manufacture parts is always changing.
Here is an example that should illustrate exactly what I mean by constant change occurs in modern manufacturing done in a machining job shop:
Today I have an order for one part. I program the part for a Haas Toolroom Mill. Four weeks later the customer decides to go into limited production with that part. The same part now needs to be made in very different way on a very different machine. The part will need to be completely reprogrammed because it is now going to be made on a Mori-Seiki horizontal milling machine with a linear pallet pool. The reason for this is I can radically cut the cycle time and the labor needed to make the part by doing things like using coolant through drills which the shop can now afford to buy for this part. In addition, I am now going to use different speeds and feeds for all the milling that needs to be done on this part. I may even use different types of toolpaths because getting the cycle time down is now much more critical and the Mori-Seiki can do things that Haas Toolroom mill can’t.
Hopefully you can now see that change is constant in modern machining job shops and is the norm and not the exception.
Jon Banquer
CADCAM Technology Leaders group on LinkedIn