Skip to main content

Notifications

Announcements

No record found.

Microsoft Dynamics AX (Archived)

Planned purchase orders are not optimized after nightly MRP (increasing stock level)

Posted on by 75

Hi,

We are running AX 2012 R3 and we have some problems with our planned purchase orders. We think we are sometimes indicated to purchase too early than needed.

Since we started using the system we have seen our inventory levels on purchased items (components) increasing. We  have noticed many examples where the planned purchase orders from the nightly MRP run are far from optimized.

Just for information I want to mention that the demand on components are coming from demand forecast of the top level products which are exploded downwards. Some of our purchased components have a long lead time which makes us placing the orders quite early compared to when the component is needed in production.

The work process for purchasing is that our planners / purchasers are looking at the planned order list and are handling orders where order date equals todays date.

Unfortuently we find  planned purchase orders that we need to handle today that are completely unnecessary.

Example.

The screen shot contain a planned purchase order  where the order date equals todays date (2019-05-08).

If the planner just firm the order we will end up with 55  523 pieces 2019-07-09.

It is not necessary to send the purchase order today since it will result in a too high stock level compared to what we need.

purchase-orders-1.PNG

I think I found out why this happened. If I look at the pegging of the planned purchase order I will see that many of the BOM lines had a requested date near 2019-07-09. The requirement dates of the BOM lines were in most much later then the requested dates.

We are running our production with finite capacity on some resources. My theory is that during the MRP run the BOM lines were first needed in July, but then moved by future messages into the future. The conclusion is that the BOM lines got new requirement date but the planned purchase order is still on the same place as it was from the beginning.

Temporary solution

If I do a full regeneration of the net requirements on item level, I will end up with a much more optimized plan. So generating the net requirements of the single item is giving me a better situation.

After regeneration, I will find the first planned purchase order one month later than in the first screen shot of the net requirements. The first planned order now has an order date of 2019-06-11 and is needed first in the middle of August. The on hand quantities are also on a better level than in the first version.

With this result, we would avoid purchasing too early.

purchase-orders-2.PNG

But there must be a way to make AX generate this kind of optimized plan during the nightly MRP?

We can’t be dependent of the planner / purchaser looking at the requirement profile of each purchased item and doing an manual investigation if to purchase the item. It is also time consuming to do manual regenerations of the net requirements.

Someone with  knowledge or experience about situations like these, please reply.

 

 

*This post is locked for comments

  • Suggested answer
    guk1964 Profile Picture
    guk1964 10,877 on at
    RE: Planned purchase orders are not optimized after nightly MRP (increasing stock level)

    As a simplistic approach is to  firm production orders first in date order and then rerun mrp and then firm Purchase orders to meet those planned start dates.

    Then rerun and firm transfer orders.

    That ideal world however  assumes your forecasts, safety stocks, CATP promises, Lead times etc, ensure that capacity and materials are adequate.  

    There is an added sophistication in Ax with regard to production order scheduling policy. Ascending item level and Descending item level use the BOM level to determine the scheduling order

    1. Ascending item level, delivery date

    2. Descending item level, delivery date

    3. Priority – delivery date

    4. Status – scheduled start

  • Mr W Profile Picture
    Mr W 75 on at
    RE: Planned purchase orders are not optimized after nightly MRP (increasing stock level)

    Thank you for your effort in giving me feedback.

    Please describe how the level by level master scheduling could be implemented? I can’t really see how this could be setup...

    ”Unless you have very stable plans protected  by fences then level by level master scheduling is the norm.”

  • guk1964 Profile Picture
    guk1964 10,877 on at
    RE: Planned purchase orders are not optimized after nightly MRP (increasing stock level)

    Unless you have very stable plans protected  by fences then level by level master scheduling is the norm.

    I suppose then the question why is there a shortage of capacity for something already planned ie. was the plan, and the efficiency/downtime/changeover allowances realistic - is there flexibility in e.g.. overtime working. If the plan is changed often  inside lead times then you are bound to have  nervous schedule unless buffered. No harm in that if its profitable enough to make the administration of changes worthwhile /profitable, otherwise S@OP planning has to agree on what is realistic

    Forecasting is by definition wrong (either the dat, or the quantity) otherwise it might just as well be a master scheduled order. They are often self fulfilling . If you forecast only 50% then you probably will not be able to satisfy demand. Your forecast will look 100% right but next year you will probably forecast under again based on last year. if you over forecast and buy materials then you will eventually find a way to sell - so again the forecast accuracy will look good a year later. Price elasticity is a bigger factor. You don't need to forecast 100% or for all items nor offer the same service levels etc. ,  

    Apart from any error of data, or assumptions, there are inherent difficulties in forecasting. How does an order entry clerk know whether an order is, or is not part of the forecast?.     If you sell the entire forecast in the first week, then maybe it was a  specific call off by a customer against long standing commitment. but generally that does not mean you will not continue to sell the rest of the month

    What happens to the difference between actual and forecast-  should sale over forecast reduce future months i.e. the customer ordered early  or should it be considered as a good month, and next month's forecast still stands. Similarly, if I undersell this month does that mean an expected order will still come next month, or does it mean that next month may also be  worse than forecast? does that mean policies and parameters need to change. Who decides what part of the order book are part  of next period forecast or additive to it?

    If sales forecast a sale of 100,000 red pens and they then sell 100,000 blue ones have they made forecast? Or are they on the hook for 100,000 blue pens because that is the agreed plan, and you don't overcommit capacity and  materials to blue until the planned red ones are made? The answers are probably different in every company and maybe also change every period or differ by customer.   

    MRP believes what we  tell it and does its best to reconcile the differences between the rules we give and the real world. Its not practical to satisfy everyone and to constantly juggle parameters, Lead times, calendars.   Ultimately S@OP planning should agree the strategy for the  period e.g. maximse service and build stocks, or to cut out overtime, rundown stocks, put up prices,  and accept some lost sales, run anew promotion. Whatever is agreed ought to be a strategy that lasts for the lead times.  otherwise focus on reducing lead times to give mrp a chance to help stabiles operations rather than automate chaos..  The master schedule then  has to decide whether to expedite to satisfy extra demand, or turn down the order, or delay - no-one else. Configure mrp fences and coverage rules etc on that basis .Without those two foundations mrp has to be micro managed. or be used  simplistically (which is usually what happens in practise).

    Changes in short term capacity don't usually change short term demand/forecast and should therefore logically not impact future planned purchases.

    Forecasts should not change within lead times - what is the frozen window?.

    If you forecast in monthly /weekly buckets then I don't see that a dynamic window has any advantage over a reduction key.

    https://community.dynamics.com/ax/f/33/t/159718

  • Mr W Profile Picture
    Mr W 75 on at
    RE: Planned purchase orders are not optimized after nightly MRP (increasing stock level)

    Thank you for you reply. :)

    "You may have to consider level by level planning i.e. master schedule the first level, then re run mrp then of the next level."

    We are running intercompany master scheduling. Are you suggesting that we schedule on separatly master scheduling for each BOM level ?

    "The questions if you are forecasting why is your manufacturing plan changing enough to disrupt your purchases"

    The manufacturing orders are not changed by human interaction. During the nightly MRP some of our planned production orders are postponed because of limited capacity or shortage of other components. But the planned purchase orders seems to still exist where they first were neeeded before the future messages moved the BOM line into the future.

    "With a  40 day lead time and a 30 day window to buy from the first day that mrp sees a supply is needed.  you are potentially planning purchase supply for demand  70 days ahead."

    Good point.

    We have aldready started discussion to decrease the coverage period for the items with long lead time. This will probably decrease the size of the purchase orders. On the other hand our minimum quantities will be used anyway but we will give this a try.

    "Maybe you can phase/smooth your forecast differently e,g, weekly instead of monthly or increas,e your negative days."

    We are already by using weekly divided forecast. My experience is that it instead are causing us problems this the forecast reduction (Dynamic period) is not working perfect.

    Let's say that we have a forecast of 100 pcs each week, four weeks gives us 400 pcs in total. If a sales order has delivery date during this month with 400 in quantity we are reducing one of the weeks but this gives a total demand of 400 + 3 * 100 = 700 pcs. Theoretical we have sold all of the forecasted quantity but are forcing AX to fullfill the rest of the forecasted quantity.

    If we had a monthly forecast the reduction would have worked better. Internally at my company we do not agree in this conclusion.

  • guk1964 Profile Picture
    guk1964 10,877 on at
    RE: Planned purchase orders are not optimized after nightly MRP (increasing stock level)

    You may have to consider level by level planning i.e. master schedule the first level, then re run mrp then of the next level.

    If your purchases are planned and then you change the production dates then the  planned purchase orders need review again.

    The questions if you are forecasting why is your manufacturing plan changing enough to disrupt your purchases. With a 40 day lead time you might well have firmed an order a day earlier - then what? Or what if a production order is brought forward?

    With a  40 day lead time and a 30 day window to buy from the first day that mrp sees a supply is needed.  you are potentially planning purchase supply for demand  70 days ahead.  and the forecast is driving production and purchases - so changing demand short term production means there will be inevitable noise in the system. Maybe you can phase/smooth your forecast differently e,g, weekly instead of monthly or increas,e your negative days.  

Under review

Thank you for your reply! To ensure a great experience for everyone, your content is awaiting approval by our Community Managers. Please check back later.

Helpful resources

Quick Links

December Spotlight Star - Muhammad Affan

Congratulations to a top community star!

Top 10 leaders for November!

Congratulations to our November super stars!

Tips for Writing Effective Suggested Answers

Best practices for providing successful forum answers ✍️

Leaderboard

#1
André Arnaud de Calavon Profile Picture

André Arnaud de Cal... 291,280 Super User 2024 Season 2

#2
Martin Dráb Profile Picture

Martin Dráb 230,214 Most Valuable Professional

#3
nmaenpaa Profile Picture

nmaenpaa 101,156

Leaderboard

Featured topics

Product updates

Dynamics 365 release plans