I can finally answer back here. The RollupDate still updates even if the cost doesn’t change. I think my issue was that I was letting the system use unapproved revisions in the roll and apparently that was messing with some things. I changed it and now I am reliably getting about 98% of my active, approved revision parts to show a rollup date and be included in the roll. Still trying to figure out how accurate these costs are. I mentioned in my other thread that I am finding that sometimes a rolled cost does not match what I believe the rolled cost should be. The ones that don’t roll appear to mostly be related to the thoughts posed above such as unapproved revision, site mismatch, etc. The only thing I’m still troubleshooting is why a rolled part cost might be wrong.
I did this a while back and spent a long time doing what you’re doing. I had other users telling me it did weird things and didn’t work all the time, but after spending time with it, reviewing logs, etc. it all came together.
I have a really good example of one that doesn’t seem to want to roll. Is there anything I can paste in here to illustrate the problem and or things you’d want to check/confirm?
This Top-Level BOM is in our Cleveland site. However, I am running the Cost Group with the South2 site selected. From what I can see, it normally doesn’t care what site I have selected and it will normally roll all sites (I assume because the Cost ID is the same for all sites). But this one doesn’t. If I switch it to Cleveland, it will roll properly. But I was hoping not to have to roll everything by site (because we have 5 production sites). Also of note, the system is “rolling” it in South2 but it calculates a different cost. Of all the ones I found like this before, it was explainable as one specific subassembly that had some goofy setting on it. But this one all looks good. It’s fairly flat. Only 2 subassemblies inside and the rest just materials.
Long buildup to a short question: Am I supposed to rollup costs by site?
Dan,
Am I understanding that you want the Cleveland cost to be the cost that shows on the part?
Yes - exactly.
It wouldn’t surprise me if there are some site implications here. Operations are site specific, so while you can sometimes look at things from other sites, you aren’t supposed to be able to use them at different sites. You get an error if you get details for a part with operations from the wrong site for example. So it could be some logic it’s using to roll up is getting caught with current site check (when it shouldn’t) because code is either re-used or copied from another place where that would make sense.
This is where my head is at as well, maybe a labor rate or burden rate different for that site? And to @Banderson 's point @dr_dan, if you truly want separate costs in the different sites and this is intentional, you’d then use two cost site IDs and you would roll them independently I believe… Our other site is distribution only, no MFG so I can’t say how this works. I do have a consultant who may be able to help, haven’t worked with her in a while, but she was pretty knowledgeable on the costing workbench, which for some reason doesn’t have a ton of documentation to it (nor an edit list), but look at that, the idea says it’s coming soon in 2025.2!: Make a Group Edit List for Costing Workbench | Epicor Ideas Portal
Thanks guys!
We don’t intend to have two different costs lol. But that’s how the engine is rolling them.
Yes - I believe this could be a contributor. We opened a new building last year and 2 production lines moved from one site into that new site. To accomplish this, I reused many operations but added an additional scheduling resource on the operation (using the new resource in the new site). It appeared that it only used the resources valid in the site the operation was in and I thought this was intentional by Epicor to allow you to use the same method across multiple sites. Maybe, maybe not.
That’s probably it then Dan, again, I feel like they should have an entire guide on the costing workbench because it plays such an important role (no pun intended) in finance. We can’t be left guessing whether it worked or not or having to spend days trying to understand it every time. A guide would be extremely helpful.
Don’t be a coward! Intend your puns! (Right @Mark_Wonsil ?)
One more breadcrumb and item of note:
Suppose I have a part with 4 revisions. The only approved revision is in Site B. But the other 3 unapproved revisions are in Site A. When I load the cost details, it will fetch the most current revision (approved or not) in the site that I selected in my Cost Group. This then sets the stage for a BOM Roll to the wrong revision… even if I mark the “Only consider Approved Revisions”. It appears to only apply to the manufactured parts under the top level rolled part.
Okay, thank you for mentioning that, I was on a thread last week about the “logic” of multiple revs that I found interesting as well.
In troubleshooting ‘The Bench’ (becoming a curse word around here), have you encountered any MFG parts with a Burden Unit Cost which appears to be about double what it should be? We are seeing this and cannot seem to find the root cause or what is contributing to the approximately 2x of the expected value. Just wondering if you have encountered anything like this during your costs validation efforts?
I don’t have anything specific to the burden cost that I’ve trouble shot. Have you checked your resource / resource groups? Could there be two scheduling requirements on the operation?
That was the first thought but none of the cases have more than a single resource or resource group per operation. We experimented in several ways. In a pilot instance, we’ve gone so far as defining a resource group and moved a single resource into it to isolate it and then systematically make adjustments to using resource values or group values, a simpler MOM to try to determine where the doubling is coming from. As a Resource Group of a single resource, we’ve tried with the burden as flat and percentage of labor rates numbers, toggled Burden=Labor to perhaps jog its memory (or cause it to forget), and same with Split Burden and about every button, nob or slider we could find. We are Std Cost, public cloud hosted. But hey, manufacturing would be boring if we didn’t have gremlins.