Tuesday, 22 September 2009

Supply Chain Performance Management

This post is an extract of a project related to implementing BI at a Automotive Parts Manufacturing.

In order to effectively manage the Supply Chain, it is critical to obtain quality KPI in order to ensure that business objectives are being achieved, but more important required improvement’s can be identified and executed. Avoid the ‘so what KPI’.

Therefore SCM BI are fundamental to monitor and continuously optimize the Supply Chain, quality data allows for effective and rapid decision´s to carry out change.

In a manufacturing organization, just looking a inventory level and customer service on its own will not ensure that the specific weakness are identified so that relevant improvements can be initiated. In this scenario; taking factory capacity, manufacturing change-over rate, inventory levels, and customer service will provide a clear picture. Trends in high capacity utilization, high inventory and low customer service indicates that either the wrong product is being produced or the batch size produced is too big (low change over, because of large batch size not enough capacity available to satisfy customer requirements). With this information a more objective decision can be made for improvements which could result in improvements in planning methodology(IT and process problem) or modify factory tooling (physical problem) to allow quicker change over, or reduce batch size. Any one of these actions can actually worsen a specific KPI, example factory capacity, why ? Lower factory utilization capacity (capacity KPI on its own is own provide a poor KPI) means that manufacturing equipment are not churning out at its maximum, this could be due to higher product cut-over due to shorter batches, but the trade off would be lower inventory and higher service level. This means that one is getting higher cash throughput (capacity KPI, linked to inventory and service level provides a more realistic/value KPI). This is what business is about (The Goal). When this happens certain KPI become secondary by achieving total SCM benefit and not individual benefit like factory capacity.

If one had just Inventory and Customer service, one would be limited in trying to optimize stock levels without understanding the causes.

In order to define optimum SCM Performance Management KPI for a Business Intelligence System (BW) , the following cardinal rules apply:

• Define
Important to define the KPI that will measure performance in the Supply Chain, critical to understand performance versus plan and KPI’s that can provide meaningful data for rapid improvements.

The scope of KPI must provide required details to relevant BI Customer (Financial controller, Factory Manager , Supply Chain Manager ect..)

The scope of KPI relate to two key aspects:
 Width, how many KPI’s needed for BI customer, example for Factory Manager would require different KPI to SCM Manager, the SCM manager would look at the total integrated KPI of plants while the factory Manager a more limited picture, Factory Manager might be indifferent to Customer Service , but more on Factory Capacity
 Depth entails the meta data available to BI customer, a SCM Manager with respect to Inventory levels might require classification based on slow moving, expired, quality (certain products might be out of specification but can still be sold). This level of detail allows SCM Manager to initiate relevant actions to reduce inventory. A Marketing Manager might require classification based on brand, product introduction ect…

• Measure
In order to satisfy the KPI for the relevant Business Customer, data may be sourced from multiple system’s and applications. It is critical for the technical efficiency of a BI system to avoid duplication of extraction and update to the relevant info cube and multi cube’s. Obviously an organizing that has achieved homogenous (IT systems and process) ways of working will simplify data extraction. Knowing the bigger picture of scope of the KPI allows simplified and efficient extractors as well as the BW design (info cube, info sources ect..).

Dimension and time buckets must satisfy the BI customer. A Factory Manager is interested in a hourly view of critical bottleneck resources, while a SCM Manager interest might be limited to daily view (More SOP orientated)

• Analyze
The KPI must be simple and provide quality, better 10 quality KPI rather than 50 KPI. They must provide trends, graphs are critical, they must flag abnormalities based on specific baseline predefined by BI customer. The scope of KPI must enable decision that can improve efficiency and reduce costs. Graphical information is fundamental to provide view of data that enables better understanding of the situation. A graphical view to SCM Manager in a single graph containing Factory Capacity, Inventory Levels and Customer service provide an ideal snap-shot. A single view of Customer Service level provides the ‘so what KPI’

• Improve
The KPI provided can then be the stepping stone to initiate change and improvement by the relevant BI customer. Important that the BI tool allows for flagging the start an improvement/change/optimization so that benefits can be measured.

The difficulty of the above is to extract data considering that source can be R3 data or SCM APO platform consisting of BW Cube for demand planning, planning areas in SNP, order based data in gATP and PPDS versus simple tables in SAP R3.

Consideration must be give to multi-cubes to mesh data coming from SAP R3 and APO data sources (info cube) considering the interrelationship between the two systems.

Building a BI platform using APO PPDS as base will be discussed in a later post.

Wednesday, 16 September 2009

Advanced Inventory Optimization with SNP

The optimum inventory balance would be managed via combination of the following key factors and elements driving the level of inventory and supply chain costs:
- buffer stock (safety stock) needed within the market to avoid potential Stock-Out caused by adverse effects of forecast errors and supply chain underperformance
- stock availability or customer service performance required to ensure customer satisfaction with their order fulfillment
- DRP planning parameters driven by manufacturing or 3rd party suppliers costs associated with the inventory (inventory value, storage costs, shipping costs, financial interest of holding inventory etc)
Safety stock is a function of all these factors. To make things more complex, all these factors are interrelated.
At the item level, there are some immediate obvious sources of error and uncertainty:

► Forecast error – the worse it is, the more safety stock you need to cover the uncertainty.
► Production batch size – the larger it is, the less often the stock will be low.
► Production reaction time – the longer this is, the greater the time over which the uncertainty has to be managed.
► Customer service level – the higher this is, the greater the safety stock needs to be.
The APO SNP provides a number of techniques to optimize inventory.
The current offering consists of:

     ► Standard Safety Stock Planning
     ► Extended Safety Stock Profile

Should the above not be suitable, the custom approach can be adopted using advance marco functionality available within the SNP Planning Book.
A particular example wrt to inventory optimization is where SNP planning book was enhanced to manage a complex safety stock formula based on Robert G.Brown instead of using what is available in standard SNP.
The above formula shows safety stock is calculated based on:

Safety stock formula:
LT = Supply Chain Lead time (Total = Production + Transit , and expressed in months)
SR = Std. Deviation over production reaction time
BS = Batch size (Expressed in units as Production Minimum Order Quantity)
CS = Customer Service Level (shown as percentage)

Weighted forecast error calculation

F1 to Fn = Forecasts for weeks 1 to n (13 periods will do), for selected forecast types. The current average forecast over the next three months is.

 AF = Average Monthly Forecast = (Sum (F1 to F13)) / 3

An ‘average’ forecast error determined by SKU.

 WE = Weighted error

The previous 2 years Forecast accuracy percentages by SKU will be exported as 24 discrete monthly values.

The optimal inventory calculations require a single value of forecast error per SKU. The forecast error is defined as follows:

Forecast Error = 100 – Forecast accuracy Percentage.

That is an accuracy of 85% equates to a forecast error of 15%.

The functionality provided allows the error to be calculated on the last six months accuracy figures. The weighting is configurable. A higher weight is likely to be given to the more recent forecast error.


The possible approach would be to use would be:

1. Copy standard SNP supplied Planning Books into Custom Planning Book

2. Add the required custom key fields, custom fields depends on how the macro calculation are carried. If standard macro functions are used then more custom key figures will be used. In the SNP planning, limit the additional field to Safety Stock only and use macro to calculate desired stock levels, other key figures like service level and forecast accuracy can be included for reporting purposes and for macro function to facilitate user mgt. Also handy for alerts. These key figures will be loaded from custom DP planning book which contains the guts of Safety Stock Calculation using Process Chain.

3. Create DP planning book with custom fields (key figures) to contain the fundamental fields needed for SS calculation. The number of Key Figures will depend on using SAP supplied macro’s functions or reducing Key Figures using custom built macro function (see http://sapscminfo.blogspot.com/2008/10/custom-macro-functions.html) The advantage of creating custom DP planning book is that it provides maximum flexibility and does not have the SNP constraints and simplifies the import from BW to the relevant key figures. Note it is possible to load data from DP planning book to custom SNP using the Process Chain mechanism to Load data from DP Planning Book to SNP planning on a regular basis. This custom planning book will import from DP planning books data such as forecast accuracy.

Concluding remarks: If implementing SNP with release SCM5.1 onwards, then priority would be to exploit the standard functionalities for advanced safety stock (ASS) planning. The ASS allows for the creation of profiles containing rules such as:

• Forecast Error : Percentage to correct demand forecast

• Determination of Replenishment Lead time rules

• Demand type, sporadic or regular demand

• Source determination

• BADI’s (very important to enhance logic) BADI for custom formula’s, replenishment lead time and forecast error.