Not getting the return on investment for your ERP/MRP application?

Most organizations have a certain degree of optimism when starting an implementation of an Enterprise Resource Planning application. They have been assured by the vendor that after an implementation period of 3 to 4 months they will begin to see the improvements in visibility and the savings in resources that would bring the Return on Investment (ROI).

They then may have success in implementing the General Ledger part of the system and also the basics such as Accounts Payable, Accounts Receivable, and Inventory, but then are usually disappointed when implementing the guts of the Supply Chain system to find out that greater visibility is not forthcoming any more than the savings of resources.

Why is it then the Material Requirements Planning (MRP) system does not give the expected return?

An MRP is very simple in concept. There are 5 inputs to an MRP system and 2 outputs. The 5 inputs are Sales/Forecasting, Inventory, Bills of Material, Open Orders (PO’s and work orders), and Planning data. If these are as accurate as possible, then the 2 outputs, Planned PO’s and Work Orders, and the MRP Action Messages will be accurate and give you the visibility and information that will lead to saving your organization’s resources.

What then, goes wrong?

The issue is not in the concept but in the details of how these inputs affect MRP. Let us look at the inputs in detail:

Sales/Forecasting

Most organizations have a method of processing customer purchase orders and converting them to the sales orders and the work orders/PO’s necessary to get the work done. However it is on the forecasting side that the data accuracy is badly affected.

However accurate or otherwise the forecasts are, is usually dependent on the industry, these are the functions that are usually not done well:

• Forecasts are not backed up with information
  • Unusual activity in one period
  • Information relating to future customer promotions or other activities
• Forecasts are not done by the Sales department
  • Leave it to Planners who can only extrapolate historical data
• Forecasts are not monitored and measured for accuracy
  • No input for future forecasts
  • No increase in accuracy as there is no feedback for problem-solving
• Forecasting is done outside of the ERP/MRP system
  • Spreadsheets are passed around without an audit trail or any version control
• Forecasting data is not set up correctly in the implementation
  • Forecasting time fence
  • Are differences between sales and forecasts to be carried on to future months or should they be taken that if it didn’t happen it won’t happen

Inventory

Most organizations have a handle on shelf accuracy (and if they have not they will not get get any visibility from their MRP system), but the issues again are in the details.

• There must be an understanding of inventory status
  • Available and non-available
  • Nettable (MRP plans for it) and non-nettable (MRP does not plan for it
• How inventory statuses are set up
  • MRB (Material Review Board) is usually set as non-available and non-nettable
  • Incoming inspection is usually set as non-available and nettable

Bills of Material (BOM’s)

If you were to ask most organizations about their BOM’s they would say that they would have a high degree of accuracy, and by that they mean that the “quantity pers”, which is the number of units of a component that are needed to make the parent. However, there are two issues relating to BOM’s that are usually not done well.

• Yield losses or scrap factors. If it takes 20 units of B to make 1 A, but there is a yield loss of 10%, then the BOM should be set up as 22 units (actually 20/(1-0.1) which is 22.22, but 22 is sufficient). If this is not done planning will be affected.
  • Sufficient component material will not be available
  • Purchasing then orders more than necessary and there is a build-up of inventory
• The Engineering Change Control (or Engineering Change Notification, ECN) process lacks discipline.
  • This usually occurs because there are many changes needed because of market-driven forces or because of lack of quality in the original product.
  • There is not sufficient discipline in the process that keeps the Engineering BOMS separate from the Production BOM’s and this causes problems in Planning and Purchasing, and causes reworks.

Planned PO’s and Work Orders

This is probably the weakest area of control in the inputs to MRP. There is little understanding of how open orders affect MRP.

• The accuracy of WIP inventory can be as low as 50%. Finance normally does not do a WIP count but relies on the open work orders.
• Work orders are often left open for a variety of reasons:
  • There is a problem with the material and it is waiting action from Quality Control. MRP still sees these as material due to arrive soon, and if this is not the case the material needs to be moved into a location defined as non-nettable.
  • Because of a shortage or an overage the work order has not been closed because the people responsible are not certain how to close these.
• It should be understood that any open work orders are expected to be available on the order due date, and any open work orders with due dates in the past will be expected by MRP to be arriving immediately.

Planning Data

This is another area that is not well understood. The main Planning data elements that MRP uses are lead times (and the elements making up lead time such as manufacturing lead time, shipping lead time, and safety time), safety stock, order minimums, and order multiples.

• Depending on the type of ERP system MRP uses Planning data to plan the due dates for the planned orders, but uses the Routing information to calculate the due date of a work order when it is firm-planned or released.
  • Routing information for a particular part number when the manufacturing elements of lead time (Queue, Set-up, Run, Wait, and Move)are cumulated should agree with the lead time set up in the Planning Data file.
• MRP will expect the material on the due date of the order. This applies to both PO’s and work orders. It is this aspect that is least understood by those people involves in planning.
  • If a vendor or a manufacturing process has indicated that the material will be available later that the due date, then the due date needs to be changed. If it is not MRP will assume that the material will be available on that published due date and plan accordingly.
  • If a due date is in the past then MRP will assume that the material will be available immediately and also plan accordingly.
• MRP will always plan to cover net requirements (gross requirements less that which is already in inventory). Therefore it will plan to have zero inventory once the requirements have been satisfied.
  • Zero inventory is a satisfactory situation for MRP if all requirements have been met.
  • If safety stock is in place, MRP will plan to have that amount in inventory.
  • Safety stock is considered by MRP to be immediate demand, so action messages will be generated immediately if a new safety stock level is set.
  • Fixed order quantities, such as minimums and multiples, will be used by MRP and result in larger quantities planned that may be necessary. These minimums and multiples should be kept as low as possible to avoid excessive inventory.

If these 5 inputs to MRP are understood by everyone involved, and the details and the MRP mechanisms used also understood, then a working MRP system will be in place and the organization can expect the returns on investment that it had hoped for when the MRP/ERP investment was made in the first place.

Most people in the Supply Chain profession understand the concept of cycle counting: the use of Pareto analysis to concentrate activities on those items that are, for cost or other reasons, strategically important to the organization.

Most people also understand the classification of items into A, B, and C items, with a rule of thumb that categorizes the top 20% as A’s, the next 30% as B’s, and the remaining 50% as C’s.

Why is it then, that in my experience in consulting and in full-time employment, that cycle counting is not executed well and does not give the coverage that is advertised? Management is often surprised that cycle counting is not giving beneficial results, and many still retain an annual physical inventory as a “safety net”.

Let us deal with the issues that need to be addressed to ensure that a Cycle Counting program is optimal:

Classifying items into A, B, and Cs

When this is done manually, or even within an ERP system, the main factors considered are the cost of the item and the throughput volume. These are then multiplied to get a cost-volume figure and then Pareto analysis is done to complete the item classification.

This is good in so far as it goes, but this approach ignores certain items that may be of strategic value to the company. There may be items that have a long lead time (and that you protect by having a level of safety stock) that should be counted with the A’s even though their cost-volume amount would have them as a B or a C.

How then could you practically add Lead Time as a factor in the Pareto analysis?

This could be accomplished by having a weighting factor of (say) 10 for Lead Times. Give every item that has a long lead time a weighting towards the top of the scale, i.e. 10, and give those other items with shorter lead times a 4 or a 5, and then those items that you can receive within a day a Factor of 1.

You would now have three columns and a cost-volume-lead time calculated figure that now may be more useful in determining which of your items should be A’s and thus counted more frequently.

Criticality is another criterion that may be regarded as important in the classification, and weighting should be in place that reflects the ease of replacement and whether a substitute is available.

2. ABC classifications should look towards the future not the past

This particularly is relevant if you happen to be in a fast-moving, change-oriented environment. During my time in the DNA Testing arena, basing ABC counts on the history of (say) the last two years would not have given anything even close to an accurate classification.

Basing inventory classifications on forecasts and projections, however imperfect these may turn out to be, is still an improvement on basing them on a history that will not come close to repeating itself.

However, if you are in a mature, stable industry with history repeating itself into the future, you can keep using the history as it is a good predictor of the future.

ABC classifications should be reviewed by Management.

Whatever the results of the Pareto analysis, it should not be used for setting cycle count schedules without the review of Management. Manufacturing, Supply Chain, Finance, along with other departments such as Quality and Marketing should have the opportunity to review the classifications and ensure that items that have special significance or strategic value to the company are included in the A classification regardless of the criteria used.

Sufficient resources need to be allocated to the Cycle Count program

In my experience cycle counting is typically under-resourced, and this becomes one of the primary reasons why cycle-counting often does not produce the expected and advertised results.

Let us assume that a specific item count is accurate, i.e. that the quantity in the perpetual inventory in the ERP system is equal to the physical quantity. The process of counting, checking, and entering in the system will then take 2 minutes – use your own figures if they are different. The problem occurs when the physical is different from the ERP system, then counts have to be checked, transactions have to be reviewed, and an investigation done. Let us say this takes 30 minutes per item.

Many companies use the following for a cycle counting schedule:

Let N be the number of items in inventory. Then:

0.2N is the number of A’s

0.3N is the number of B’s

0.5N is the number of C’s

If A’s are counted 4 times per year, there are 0.2N*4, or 0.8N counts.

If B’s are counted 2 times per year, there are 0.3N*2, or 0.6N counts.

If C’s are counted 1 time per year, there are 0.5N*1, or 0.5N counts.

Therefore the total number of counts is 0.8N+0.6N+0.5N or 1.9N. This means that whatever number of items you have in inventory you have to do 1.9 times that, or almost twice the number of items, to complete the cycle count every year.

Some public companies have A’s counted each month, B’s every 3 months, and C’s every 6 months. If this is the case, then the total number of counts is (0.2N*12) + (0.3N*4) + (0.6N*2) = 4.8N or almost 5 times the number of items in inventory.

The math now is simple. Let us assume that you have 90% inventory accuracy, then, on average, 90% of your items take 2 minutes and 10% take 30 minutes.

90% of 1.9 = 1.71 and 10% of 1.9 = 0.19

Therefore resources needed = 1.71 * 2 minutes + 0.19 * 30 minutes

This amounts to 9.12 minutes per item

Therefore if you have 2,000 items in inventory, the resources needed would be 2,000 * 9.12 = 18,240 minutes or 42 days annually. This would be one person for almost 1 day per week.

If your cycle count schedule is more demanding as it is for many public companies so that your counts completed per year are 4.8N, or if your inventory is larger, or your inventory accuracy lower, then you will need a lot more resource.

If, for example, you had 5,000 items in inventory, and your inventory accuracy is 85%.

85% of 4.8 = 4.08 and 15% of 4.8 = 0.72

Therefore the resources needed = 4.08 * 2 minutes + 0.72 * 30 minutes

This amounts to 29.76 minutes per item

With 5,000 items in inventory, the resources needed would be 5,000 * 29.76 = 148,800 minutes or 344 days annually. This would be beyond the scope of one person and would need 2 people almost continuously cycle counting.

If these resources are not in place, then management will not get the expected results, simply because the schedule cannot be adhered to and the hoped-for inventory accuracy improvement will not occur.

To get the advertised, and hoped-for, results, ensure that your ABC classifications are accurate and up-to-date, that they are reviewed, and that adequate resources are allocated to the program.

One of the reoccurring issues in planning when using a Material Requirements Planning (MRP) system concerns expiration dates of material in inventory.

For example, if today is January 1^st, there is a requirement for 50 of part A on February 15^th, and there are 60 units of Part A in inventory, is the requirement covered? The obvious answer is yes.

However, if 40 of those units have an expiration date of February 28^th, and 20 have an expiration date of January 31^st, is the requirement now covered? Again, the obvious answer is that only 40 units are covered.

However, MRP will not take into account expiration dates and MRP will plan as if the requirement is covered and thus will not plan any orders for Part A. MRP acts as if the requirements are completely covered.

So the question becomes, why does MRP not take the expiration dates into consideration when it plans? It has the ability to figure out when the material becomes obsolete and can act accordingly.

The answer is that expiration dates can mean different things to different customers. For some customers an expiration date is a hard stop. The material has to be removed the day after the expiration date. For other customers it is a guide line, and for yet others, including a DNA testing company that I ran the Supply Chain for, the expiration date can be changed and moved out by a re-validation procedure.

Bearing this in mind, it would then be difficult for some users that had a reasonable expectancy of still be able to use the material, if MRP had already created planned orders based on the fact that some material was going to expire before the requirement existed.

This is in line with other date issues such as a phase out – phase in, when a revision is made to a particular product or material. Again MRP will not take into account the fact that there is material with a previous revision in inventory, and for the same reason that the dates can, and often are changed by the users.

So, how should expiration dates be dealt with in MRP?

It is vital to have, in your ERP system, a type of report that shows you what is in inventory with the expiration dates for each lot. If you can also have in your report a “requirements per day” (or week, or month) amount based on the average requirements for the material for (say) the previous 3-6 months, then you will be even better placed.

You can now review this report on a regular basis and highlight those items that are in danger of passing their expiration date. You can then:

1. Revalidate the material and change the expiration date if possible.
2. Remove the material to a non-nettable (one that MRP ignores) location.
3. Adjust MRP by creating a Firm-Planned work order.

These actions are part of the Planners responsibility and the review of all material expiration dates should be a priority for the Planning department, whichever environment and whichever approach you are going to use.

International Rectifier, Mexico

Excel is an excellent data-analysis tool.  If you need to dissect, or “slice and dice” amounts of data there is probably no better tool available for regular business users.  However, if Excel is used in business processes it has two inherent weaknesses; one is that it has no audit trail (if an amount or a formula is changed in a cell there is no record of what was there previously) and the second is that it does not have version control (as files are downloaded or moved around by email it is difficult to keep track of the versions used.

The following case study illustrates the problems of using Excel for more than it was designed for.  The company relied on data being downloaded to Excel files from various disparate systems and then combined to produce financial and monthly operational results.

Situation The company had a large Maquiladora set up in Tijuana which produced a large portion of the semi-conductors that the company made.  This operation had difficulty controlling its Work-In-Progress (WIP) and could not account for many of its assemblies, and was concerned that it was producing inaccurate monthly financial results. We were called in to 1) find out the extent of any problem and 2) to correct the immediate issues and recommend improvements going forward. A major hurdle that they had was they had five disparate systems covering Purchasing, Planning, Operations, General Ledger, and Accounts Payable.

In order to get the needed data and be able to send the monthly financials to Head Office, they had created an Excel network of downloaded data from all the separate systems, and had an exceptionally knowledgeable IT person maintaining and running the whole data network. Each system would have its data downloaded to an Excel work sheet and then each of these were merged and manipulated so that the General Ledger detail would then be sent off in the correct format.

Approach This data network was of such sophistication that the only person in the operation that understood the mechanics was the IT person that had put it together.  To everyone else in the operation it had become a black box.  The first step was to get a conceptual understanding of the black box. It then became apparent that there was no real connection between the data from these separate areas, and when the WIP was analyzed in detail it became apparent there were real issues.

The local management had assumed that the downloading and merging of files was being done correctly and had not checked by going into the operational system and listing out all open work orders. When the WIP report was extracted from the system there were work orders showing that had start dates of up to three years previously.  With a maximum production cycle time of 2 weeks it was obvious that the closing mechanism was not functioning correctly and the WIP was overstated by almost $3 million.

Result Modifications were made to the month-end process that included management review and cross checking.  Also recommendations were made for the company to implement a comprehensive ERP system across all functions that would give the local management the visibility it had been lacking. Management and staff were also educated in the use of Excel, showing how the software can be used very effectively in data analysis, but if it is used in the processing of data, weaknesses such as the lack of an audit trail and the lack of editing of the reasonability of data can cause real problems.

EPSON

The Sales and Operations process is not well understood in many companies.  The information that is collected by the Sales department is often inaccurate and incomplete, because unfortunately forecasting and gathering data is not the first priority of a sales department, but rather done as a necessary evil and an afterthought.

However, if it is not done well, and the Operations personnel have little confidence in the data, various things will happen that will denigrate the company’s ability to have the right quantity of the right stock at the right time.  The following case study illustrates this point.

Situation

Epson had their distribution warehouse, their production facility, and their Sales/Marketing operation in three different locations.  They were having difficulty keeping the optimum amount of inventory in their distribution warehouse.  This resulted in back orders and low customer fill rates.

The major issues revolved around their forecasting operation and how it provided information to the production facility, and their sales and operations process as a whole.

Analysis found that there was a disconnect between the forecast information and how that information was being used at the production facility.  In addition, forecast error was not being measured correctly and management had a more optimistic view of the forecasting effort that was actually warranted.  There was also “gaming” involved by the Sales operation in that they added a certain percentage to the forecast to encourage the production facility to hit the target rates.

This resulted in a bigger disconnect between Sales and Operations as the separate production entity would not have the required confidence in the forecasts and thus would begin to chase sales rather than adhere to the forecast.

This was compounded by the fact that each entity had their own disparate system, which remained out of balance with the other two despite IT’s efforts to interface them.  There was always a timing difference between the systems, and even if this was only one day, it meant that people were always looking at different figures which also had an adverse effect on confidence.

Approach

The first item achieved was the removal of the forecast bias along with the correct calculation and publication of the forecast error that began to improve the forecasting techniques and give the production facility more confidence putting together their build schedules.

The second item was to set up a formal Sales and Operations process that kept forecasts in line with production schedules and moved the responsibility for the inventory levels from Production to Sales/Marketing.

The third item was a complete analysis of their disparate systems situation so that management understood the cost in lack of visibility of not having a comprehensive Distributed Requirements Planning (DSP) and Enterprise Resource Planning (ERP) system.

Result

In a 4-month time frame the company was able to significantly improve its back order situation and its customer fill rates.  There was also a foundation in place for future improvements, as they now had a good forecasting method and measurement, along with a more cohesive alignment of production and forecasting resulting from the Sales and Operations process.