Further to July 3 (ADR-First)

If you have not read my July 3 post, please do so before reading this entry.

64% defect removal doesn't seem like much. Of course, you can argue, "I get better". But no one I have seen gets more than 85% on any sorter or ADR; if you were able to maintain that level, the system removal would be only 72%. At 20% incoming defects, that means your product still contains 5.5% defect after the sorters and ADR are done. Right on the cusp of acceptable grade. So, you might blend in some better product from a cleaner load and move on. If you have it.

But if your incoming defect load exceeds 20% by much, you have some difficult choices to make. And as the incoming defect rate gets higher, you start to encounter another problem besides trying to make grade: Your ADR can tolerate only so much flow. You have designed your line around 20% incoming defect, and at your nominal line flow, your ADR is at capacity. As incoming defect levels get higher (especially in late storage season), you send more and more product to the ADR. Sure, it brushes off the excess, but that doesn't help in the long run- that excess comes back around again, further adding to the load. Your system starts making more white cubes than it should, cutting (literally!) into your yield. Eventually, belts spread apart, cutters get jammed, knives bend and break, and you are looking at line downtime. These days, line downtime costs you a bundle.

Or maybe you slow your line down when you run high incoming defect levels. Ever look at what line slowdown costs you? If you have, it is shocking!

A couple of other areas to comment on, regarding the status quo line; both areas are related to missed opportunities for yield and product quality enhancement:

Modern ADRs are capable of cutting good fries in multiple pieces, using one knife throw per cut. Why might you want to cut good strips? Couple of reasons: First, some products have a length specification where too many short pieces ( less than 25mm) will cause you to be under grade on length, and that happens often with bins of short potatoes, especially in Europe. However, if the strips that are extra-long (say, greater than 150mm) can be cut in two pieces, the percentage of "long" pieces (e.g., over 75mm) goes up, even though the average length is reduced. Such a capability can make the difference between making grade or not.

The other mode can be even more significant. If you cut every strip that is longer than a maximum, product flows better in many ways. Every place there is a drop or gate, less segregation occurs, and you end up with more uniform product. Everywhere bridging has been an issue, is will be less of a problem, with no long fries to get hung up. Best of all, bagger jaws will close on product much less frequently, greatly reducing rework cost, film cost and improving bagger productivity. One could imagine shortening the actual bag length for any one quantity of product, again saving film and perhaps pasteboard. And shipping space. And....

Sure, you can send long strips to the ADR if you are using sorters with length sorting capability. But that is a lot more product to be sent to the ADRs, and they will likely not have the capacity to handle it when incoming defect levels get high.

You can also halve your whole potatoes prior to cutting, as an alternative to using an ADR to control length. But anyone who has modeled that approach knows that it increases the short pieces and slivers, compared to cutting whole potatoes. More yield and quality hits.

The other missed opportunity is in limiting the application of the intelligence available on modern machines. Modern ADRs can make very smart decisions about which strip to cut and which to pass. The classic example is a strip with a minor defect that is positioned such that removing the defect will result in an additional nubbin or short piece. If the grade of the moment can tolerate it, that defect should be ignored, and by doing so, yield and length profile will be enhanced.

Problem is, if you are only sending 80% of the defective strips to the ADRs, they often must remove every defect they identify, in order to make grade. If they were to be able to see 100% of the defects, often it will be possible to ignore a fair percentage of minor defects and still stay in grade. How much better that the specific minors ignored also reduce short piece count to maintain length profiles?

OK, I think I am finished (for now) with the issues of the status-quo system.

Next time, we will discuss the solution.

Check this entry again early next week- I will add some pictures to make things more clear.

Tim

Stumble It!

Bookmark on del.icio.us