The normalization feature relates to the rolled up breakdown on the output material. With normalization on all material breakdowns on your inputs that don't equal 100% will be normalized to equal 100%. For example, if your input breakdown has 2 components (45%/45%) totaling 90% the system will normalize it to 100% making each component 50%. The rolled up breakdown would equal 100%. With normalization turned off then the rolled up breakdown would continue to equal 90%.
I think in your example the breakdowns that equal 0% (ones only with ranges defined or no formulation % defined) may be throwing your numbers off. I'm not sure we accounted for regulatory breakdowns that equal 0. Our lower limit should probably be 1% to be more accurate. Is this just test data or is there a business use case for 0% breakdowns?
We can get on a call if needed to discuss, just let me know. We are definitely interested in learning as much as we can around how customers are using the breakdown control.
Thanks, its great to talk about our latest release so quickly!
The zero percent breakdown is a real business use case. We have a formulation of 1000 lbs of a material output. One of the inputs is 1 lb (or 0.10000%) of "Flavor" that contain 3-5 ingredients in the material breakdown from the supplier. The formulation of the "Flavor" is Intellectual Property of the supplier. Supplier is willing to tell us what is in "Flavor" but not in any percentages. Therefore we have the ingredient breakdown but not formulation or any percentage ranges. It was decided that as long as we can search for the ingredient (in case of regulatory need), that we really don't need to know the precise breakdown.
That particular use case is a bit of a side issue to how to get formulation calcuations with or without breakdown normalization.
I was expecting to see that the "Testing" formulation calcuations change between normalization on and normalization off. I can remove the material with zero breakdown total and re-test / re-calc. I was under the impression that the zero material would be ignored but that the remainder of the breakdowns would normalize/not normalize depending on our config setting. I'm fairly certain that I'm missing something in the testing methodology and I'm trying to figure out what. We want to see in v6113 (DEV) what we are experiencing in v6111 (PROD) and v6112 (QA) ... where the output breakdown is A) at 100% and B) not 100%. depending on how the configuration is set. So far, re-calc'ing the formula has not changed the output results so I am still designing my test casess. Removing the zero breakdown material may be the thing that I need.
Thanks for your advice. I will let you know how it is going. Best, Beckie
When we roll up breakdowns we go to the bottom level component so if you have your breakdowns at 0% the flavor in the output itself would be at marked at 0% and you would have an incorrect/misleading composition. Items that are 0% would also not be included in your ingredient statements, or appear not to be present at all.
I would recommend doing one of the following:
1. Capturing the components in a breakdown but not tagging it as regulatory. This way you can still trace their existence if needed but flavor will be accurately displayed in your rolled up breakdowns and subsequent ingredient statements on your labels. A report can easily go deeper and see the existence of these elements. We can discuss offering this in a future report pack and screening tools. We can also help you create that report in the interim if needed.
2. Adding the items where you don't know their composition % but still need to trace their presence for compliance reasons to the compliance grids (additives, allergens, intolerances). These would be rolled up with the output.
3. Putting in a target formulation % of equal parts if you don't know the percentage, the system will normalize it. This is still inaccurate but flavor will at least now appear on your ingredient statement. Since you don't label further into the flavor components it should be ok even though you will have to always manually intervene in LIO. I would tag the components using a formulation tag with "presence only" " % unknown" or something similar so if you are ever performing compliance checks you know that percentages are there only for inclusion. You could also add a report visibility helper in LIO to see the tags at the time of editing.
In general unknown amounts are a little worrisome if you ever have a sub component in a formula more than once. For example: Input A is 1% of your formula and contains additiveXYZ at an unknown percentage, which you know is such a low amount you aren't concerned about it. This is ok until additive XYZ is a known percentage in another input greater than your tolerance threshold. For example you add Input B as 2% of your formula which contains 80% of additive XYZ, the addition of input A and B could bring you over your threshold.
You could have a routine that checks for components more than X amount with the tag of "% unknown" and add these elements as additives. Then you can defer to lab test results for the per 100g amounts. You may also want a notify users if a component is listed more than once and any of them contain the unknown % tag. That way they can manually intervene if needed.
Sorry for such a long post, this topic is a pretty complex one and you have a lot of options. Let me know if you want to dig into it further and we can set up a web conference next week.
Kelly, I am actually grateful for your long post. We have business units that are using formulation in different ways and have different concerns. I'm sure this is true for most companies using PLM4P. It seems that we need to decide what level of granularity is important for the entire organization and do some user documentation around this. I am willing to bet real money that Bruce will want to take you up on the offer for web conference next week. I will leave this post open for a bit longer but schedule with you directly in other communication channel. Thanks again. Best, Beckie
I have been revisiting formulation calculations in our DEV environment under v6113. What I am finding is that your reply assumes we have LIO, which we don't. The expectation from R&D seems that formualtion should recognize %BD and if none, hold % value of material. I am still investigating upper and lower boundaries in v6113 in DEV. We are not yet seeing a predictable behavior. Best, BeckieC