Monday, July 18, 2011

Prevention beats detection, but what does it save?

Whenever I see a vision system being used for inspection it’s almost always with the goal of separating out the nonconforming product. Customers like that, because no one wants to receive crap. And management likes it because customer complaints are a real pain-in-the-ass to deal with. But the savings from detection can range from small – the cost of a displaced human inspector – to negative, when scrap rates go up because more bad stuff is being found.

The answer, as our Lean friends will tell us, is to avoid making bad product, and that’s the crux of the application case study published in the June Control Engineering under “Machine Vision ROI at Polaris Industries”.

This article describes a machine vision error-proofing application, but what really caught my eye was the financial discussion. Costs and benefits are quantified and the conclusion is that the payback period was between two weeks and a month.

Impressive numbers, to be sure, but this reads more like an engineer’s justification than one that would pass financial scrutiny.

The first issue I see is that much of the benefit is increased revenue because capacity is no longer being wasted making bad product. I accept the capacity point, but has revenue really increased? Are the sales guys selling more units because a vision system has been installed? Or is the reality that less overtime is being worked and on-time delivery performance is up?

The second issue I see is the quoted scrap savings. Now clearly, if fewer bad parts are made less material and labor will be wasted, but in my experience the financial wizards have some funny rules for how they quantify these things. Yes, common sense says there are savings but the Plant Controller may not see it the same way the engineers do.

Don’t misunderstand me; error and defect prevention is exactly the way machine vision should be used. I’m just advocating a little fiscal prudence in calculating and presenting the cost savings. Why claim a two week payback when the project would have been just as good with two months and the accountants eyebrows would have remained in place rather than climbing up over his scalp?

No comments: