IBP Optimizer inconsistent Net Demand

[adrien_delanoe]

Hello,

I encountered an issue while setting up costs for TS-based Supply Optimizer.

On our product-location combinations, we get inconsistently high Net Demand, thus generating equally high receipts that I don't want:

Edit 2 (with images):

All Key Figures impacting net Demand are either null or much lower than computed Net Demand.

Also, below is the Optimizer cost set up:

Edit 2 (with images):

I have only added costs on a few key Figures such as Inventory Holding Cost for the Child Location to be higher than on the Parent Location to avoid the shipment of unnecessary inventory.

Any thoughts how I can prevent this from happening?

Many thanks in advance for your help on this,

Adrien

[adrien_delanoe · Accepted solution]

Hello Carsten and Ayan,

Many thanks for your investigation and input on this topic.

For the reference of the answer to my topic: The issue is indeed due to the Operator penalty costs that were pushing inventory to the child location. Key Figure costs won't be considered if Penalty costs are enabled.

Disabling the S&OP operator Penalty costs + setting up costs KF gave me the correct expected results.

Thanks a lot.

Adrien

[carsten_schumm]

Hello Adrien,

please note that the penalty costs set all costs in the scenario. Existing entries in any cost key-figure aren't considered when penalty costs are active.

Concerning the issue with the dependent/net demands: these values are computed from the sum of all supplies after the optimizer has provided its result. Please check all supplies (demands / component supplies / transport supplies / inventory corrections / min transport / ...). If that doesn't help please open a support case as suggested by Matheus.

Kind regards, Carsten

[adrien_delanoe]

Hello Carsten,

I indeed figured this out after I asked my question so I disabled Penalty costs. Unfortunately this didn't changed anything, inventory is still "randomly" pushed to child location due to inexplicable demand.

I've also tried purging all Key figures from basically all Per* Planning Levels as per Ayan's suggestion, results are still not good.

For the reference: I also tried with no costs set on KF and the issue remains. Are all parameters affecting this either cost Key Figures, master Data or Optimizer S&OP profile? Couldn't I miss some additional parameters?

I opened the support case right after receiving Mateus's answer to this question and unfortunately still no answer until now.

Thank you and best regards,

Adrien

[AyanBishnu]

Hello adrien_delanoe

Since the optimizer is pushing stocks from parent to child, please check if the parent location is marked as a non stocking location in the LOCPROD MDT. Also check the inventory holding cost rate at both parent and child location. If need make the cost of inventory holding at the child location higher than the parent location.

Please remove the penalty costs and use the above mentioned cost to see if you are getting the desired outcome or not.

[carsten_schumm]

Hello Adrien,

there must be some incentive for the optimizer to push the inventory from your parent to the child locations. Please check again the inventory holding costs and that no adjusted or other constraints (e.g. inventory target) force the pushes.

I found your support case: the attached diagnosis file still used penalty costs. With that enabled all inventory holding cost rates are equal. Thus it doesn't matter at which location a product is stored. Please disable the penalty costs and set suitable costs yourself. Additionally please provide some exact product ID in the case (if the issue still persists). The scenario contains over 10000 products.

Cheers, Carsten

[AyanBishnu]

Hello adrien_delanoe

Although it is very hard to find the root cause based on the limited input and lack of screenshots, but still I would suggest you to please check the following

1) SInce you are getting 55 in the NETDEMAND KF for the child location even without having any type of demands. please check the past buckets if thier are any values for any of the demand KF for the child. Also if you are running optimiser on a different time bucket than the base planning level of the input & output KF, please make sure that you select the current technical week as the start while executing the optimizer. in case you are running optimizer in months and the KFs are at TWK, then while executing optimizer select the current TWK and not anything before that.

2) Also check once if the output and input KF for optimizer are correctly marked as input / output or both in the KF settings.

4) Check if you have maintained any frozen horizon or not. If yes delete all the values int he frozen horizon. Also check if you are using any lot size or rounding that is equal or close to 55 or its multiple.

3) I could see you are most probaby using substituion, so please check if thier is any substitution demand anywhere in the network or not.

4) as a last step initialize all output KF of the optimizer before executing it. Make sure KFs like DEPENDENTDEMAND, Dependent location demand, depdendent production demand, NETDEMAND, TRANSPORT & TRANSPORTDS (along with the MIn and adjusted KF) are all initialized.

Let me know if this helps. Also if possible share screenshots of the S&OP operator profile that you have created.

[adrien_delanoe]

Hello Ayan,

Many thanks for your very thorough answer. I have finally been able to upload the images.

I will answer all your points but unfortunately I tried to keep data as simple as shown on the pictures to easy the understanding.

1) Past buckets have been cleared prior to the latest run in order to make sure they do not impact current & future buckets. I am running Optimizer at base planning level.

2) I kept standard key figures in the Optimizer but will check anyways to make sure none of them have accidentally been changed.

3)This very product doesn't not have any substitute even though I am using product location substitution on this PA.

4) No frozen horizon is maintained, nor lot size or rounding values in order to keep the model as simple as possible and understand the current results.

5) I have the "Initialize Lead-Time Horizon of Transports" checkbox checked if that's what you mean.

This indeed helped check whether I asked the good questions before posting here but this does not solve my problem, thank you so much for your time and input.

Adrien

[AyanBishnu]

Hello Adrien

4) as a last step initialize all output KF of the optimizer before executing it. Make sure KFs like DEPENDENTDEMAND, Dependent location demand, depdendent production demand, NETDEMAND, TRANSPORT & TRANSPORTDS (along with the MIn and adjusted KF) are all initialized.

In the above step what i meant is use the version deletion program from the job template and make sure that all the output Kfs of optimizer is initialized, which means delete any exisitng vaules. Many times this helps in getting rid of this kind of issues.

[adrien_delanoe]

Thank you Ayan for the additional solution.

I've Purged all Per* Planning levels and result is still the same, Net Demand is still much higher than expected. Looks like the Optimizer tries to push inventory from parent to child location for a reason that eludes me.

Thanks and best regards,

Adrien

[AyanBishnu]

Hello adrien_delanoe

Since the optimizer is pushing stocks from parent to child, please check if the parent location is marked as a non stocking location in the LOCPROD MDT. Also check the inventory holding cost rate at both parent and child location. If need make the cost of inventory holding at the child location higher than the parent location.

Please remove the penalty costs and use the above mentioned cost to see if you are getting the desired outcome or not.

[mkorndoerfer]

Hi Adrien,

The best channel to address such complex issues is via incidents. Therefore, I'd recommend you creating an incident under the component SCM-IBP-SUP-OPT and provide the optimizer dump for further investigation. You may follow the instructions of SAP Note 2380705, specifically to section b) Time-series-based supply planning optimizer run (parameter OPT_DIAGNOSIS_LEVEL).

Best regards,
Matheus

[adrien_delanoe]

Thank you Matheus, I will follow your recommendation.