You are on page 1of 22


Introduction to Advanced Planner & Optimizer (APO)

APS Advanced Planning & Scheduling originated with Advanced Manufacturing Research
(AMR) in the early 1990s. It was created because a breed of software was becoming available
that failed the traditional ERP.

APS system hallmarks include


Few internal users than zERP systems.

The use of memory resident models in addition to traditional DBMS.
Ability to do rapid what-if simulation
Planning at a finer time increment
Advanced problem notification
Advance optimization algorithms such as linear programming and heuristics
Advanced planning and scheduling functionality

APS software solution providers


I2 technologies

SAP APO is bundled with several other advanced planning applications into mySAP SCM 5.0

Forecasting and replenishment Retail
Inventory Collaboration Hub
Event Manager
Extended Warehouse Mgmt.

Some functionalities of SAP APO: Designing, Transportation Planning, Order Promising and
Delivery, Supply Chain Performance Mgmt.


OLTP Online Transaction Processing System

SAP APO Demand Planning module Supply network planning
SAP BW data warehouse

Historical data
Key performance indicators (KPIs)

The transaction data and master data from R3 are stored in LC Live Cache. The data
flows through an interface called as the Core Interface (CIF).
APO need to protect itself from any live cache failures. So it stores in traditional RDB
relational database used for backup purposes.
Data flows between R3 to APO and back.
It is used to create a forecast of market demand for your companys products. It allows you to
consider many different causal factors that affect demand.
It is a large library of statistical forecasting models. Customers can create unique forecasting
models using a powerful macro tool. Forecasts created by APO DP may be released to APO
Supply Network Planning or passed to R/3 for MRP planning.
It is a very good medium term (neither short term nor long term) to get a rough cut plan for
fulfilling the estimated sales volumes. It integrates purchasing, manufacturing, distribution and
transportation so that a comprehensive tactical planning and sourcing decisions can be simulated
and implemented on the basis of a single, global consistent model. Uses advanced optimization

PP/DS is intended to e short-term (1-6 weeks) detailed planning and scheduling tool.
The PP portion of PP/DS is capable of creating finite supply chains taking capacities
into consideration.
The DS portion of PP/DS is capable of creating optimized scheduling sequences.
PP/DS accomplishes its primary mission by using various advanced heuristic and linear
programming algorithms.
APO Global Available to Promise (GATP)
advanced order promising tool.
1. We can plan availability of materials across multiple different plants.
2. Simulation of orders (Capable-to-Promise)
3. Rule-based order promising

APO Transportation Planning/ Vehicle Scheduling (TP/VS) intended to optimize the

planning and schedule inbound and outbound freight.
The Transportation Planning portion of TP/VS enables you to make optimal use of the
available capacity of trucks, trains, ships and airplanes.
The Vehicle Scheduling component of TP/VS will optimize the delivery routes.
TP/VS utilizes advanced linear programming algorithms to accomplish its mission.
Vehicle scheduling comes up with lower transport routes.

APO Alert Monitor


It is an advanced monitoring system to detect supply chain problems at the earliest

possible time.
The Alert Monitor is capable of operating within all the APO sub-modules (DP, SNP,
Alert conditions may be customized by company or individual user.

The portion of the APO architecture where the computer model resides is called Live Cache.
The APO modules where forecasting is done is Demand Planning.
APO module used for advanced order promising is called Global ATP.
The APO module used for advanced problem notification is called Alert Monitor.


CIF the APO CIF is used for exchanging data between SAP APO and SAP R/3 systems.
CIF will transfer both master data and transaction data from SAP R/3 to SAP APO and also from
Data passing may be real-time or batched.
The IT technology used for the interface is the Remote Function Call (RFC).
The passing of the data is enabled by the creation and maintenance of Integration models.
All Integration Models are defined in SAP R/3 (host system).

Master Data Transfer through CIF

The master data is unidirectional - moves from R/3 to SAP APO.

APO Master data Location Master, Product Master, Resource Master, PPM or RTO master etc.

Transaction Data Transfer through CIF

Most transaction data can flow in both directions Planned orders can be created in R/3 and
can be used as input in SAP APO for production planning and detailed scheduling.
Planned orders could also be created within APO Supply network planning to R/3 for further
SAP R/3 planning and execution but SAP APO can only be used for planning.
SAP APO cannot be used for execution purposes.
Integration Models
Multiple integration models will be used in connecting SAP R/3 to APO. Each model will
transfer data.
Integration models are always defined in the SAP R/3 system.

APO-relevant master data/transaction data is selected in the active integration models of the CIF.
All the material masters of a plant are passed through the integration models to SAP APO. The
integration model must be active.


Any master data originated in SAP R/3 is maintained in SAP R/3 and any master data unique in
APO will be maintained in APO.

Selecting APO Planned Materials

Any material master whose MRP type is X0 is planned in the APO system.
Models in APO
All Master Data transferred using the CIF is automatically assigned to model 000 (Active model)
in live cache.
Therefore models in APO contain master data. Since APO may be used for simulation or what-if
planning, other models may be created that contain master data different from the model 000
(active model).

Versions in APO
All transaction data is transferred to APO through the CIF is stored in Version 000 (Active
version). Examples of transaction data would be forecasts, sales orders etc.
All master data is stored in Models and all transaction data is stored in Versions.
All transaction data transferring from APO to R/3 must be stored in Version 000.

UNIT 13 MASTER DATA (10 questions finals)

The master data objects in SAP APO have different names from their SAP R/3 counterparts. For
example, the Material Master in R/3 is named Product Master in APO.
APO has many other master data fields that do not have counterparts in R/3. These master data
fields are needed to perform the APS functionality that does not exist in R/3.

Plant, Customer and Supplier from R/3 becomes Location Master in APO.
Material master from R/3 is called as the Product Master in APO.
Work centers become resource masters.
Routing/BOM master becomes PPM or RTO Production Process Marks
Transportation lanes in APO do not have a counterpart in R/3.
Purchasing Info Record and Scheduling Agreement becomes Procurement Relationship in
Crucial for SCM because it is the building block for so many other relationships.
a. Plant, Customer, Supplier and many others from R/3 are housed in APO Location Master.
Location type will distinguish between the different location masters. In APO we can have
several calendars but in R/3 we can have only one calendar.

Transportation Lane
It does not have an R/3 counterpart

Defines all valid transportation method between two locations.

TLs may be defined for all products moved between two locations or may be specific to a
If no TL exists, then one location cant be a source of supply to the other. TL may define
multiple valid transportation means (truck, rail, air) along with transportation time and cost
APO sourcing algorithms will examine the TL for cost and time data to determine the
optimum method.
Quota Arrangement
Determines the source and quantity demanded when several possible suppliers are
It exists in R/3 but are not used in APO.
They are defined for all products sourced from a location or it may be product specific.
Product splits may be defined as
a. Fixed split
b. Determined by a heuristic algorithm
c. Determined optimally and then remain fixed.
If there is only one source of supply, then no quota arrangement is needed.
Product Master
The APO product master is the direct equivalent of the R/3 Material master.
The product master will also contain data elements that do not exist in R/3. They are
A. Penalty costs these reflect the relative cost of missing an order due date.
B. Storage costs the relative cost of storing a product
C. Good receipt/issue costs
These costs are used to influence the Optimizer program in Supply Network Planning.

Production Data Structure

Standard Solution (as of SAP APO 4.1) for
a. Variant configuration
b. Engineering Change Management (Date effectivities)
PPM works for Master Recipes and Co-products but not for Engineering Change
Mgmt. and variant config.

Basis for integrated product and Process Engineering (iPPE) as the manufacturing master data


Two types of PDS

1. PDS for SNP will contain critical materials and resources.
2. PDS for PP/DS will contain all materials and resources.
1. A purchasing info record in R/3 that passes through the core interface will automatically
create what APO Master data object.
Answer Procurement relationship and Transportation lane.
2. The APO master data object that combines the R/3 BOM and routing together is
Answer Production Data Structure (PDS)
3. The APO master data object that defined a valid sourcing relationship is
Answer Procurement relationship
UNIT 14 DEMAND PLANNING BASICS (8 questions finals)
Forecast It is a prediction of a variable of interest for a future time period.
Three absolutes of forecasts are:
a. They are always wrong
b. More accurate in groups rather than individuals.
c. More accurate in the near term.
Forecasting techniques in SAP are:

Sales and operations planning (SOP)
Flexible Planning (R/3)
APO Demand Planning

A monthly forecast update is the most common.

The overall horizon of the forecast is dependent upon several factors including:
a. Purpose of forecast
b. Cumulative product lead time
The forecast cycle may terminate with a consensus forecast as part of S&OP meeting.
When the automatic forecast is completed, any alert conditions are noted and sent to customers.
Adjustments to automatic forecast are usually performed manually.


Update History Automatically generate forecast-Review/Correct forecast Reach Consensus
forecast Approve forecast.
Elements of a good forecast

Easy to use

Approaches to Forecasting
a. Qualitative method subjective inputs (JUDGEMENTAL consumer surveys, sales
staff/managers/executives/expert panels)
Subjective inputs soft info (human factors, person opinion, hunch)
It is hard to quantify
b. Quantitative Method objective or hard data (TIME SERIES uses historical data
assuming the future will be like the past)
Projection of historical data. Associative models utilizing explanatory variables
Time series forecast timed ordered sequence of operations

Trend long term upward/downward movement in data.

Seasonality - short-term regular variations in data
Cycle wave-like pattern of more than one years duration
Irregular Variations caused by unusual variations
Random variations caused by chance

Techniques for Averaging

Historical data has white noise or variation
Averages smooth variations
3 techniques
a. Moving average
b. Weighted moving average
c. Exponential smoothing
Moving Averages

Technique that averages a number of recent actual values, updated as new values become
available. The more time period we average, the less responsive the moving average is to the
actuals. It may be good or bad depends on the product and the profit margins. Sometimes it gives
you a protection during uncertainties.
Weighted Moving Average:
More recent values in a series are given more weight in computing the forecast.
Exponential Smoothing:
Based on previous forecast plus a percentage of the forecast error.
Alpha smoothing factor is greater than or equal to zero but less than or equal to 1.
Low Alpha stable average
High Alpha Changing average

BIAS it tells us whether we have a tendency to over or under-forecast. If our forecasts are in
the middle of the data, then the errors should be equally positive and negative, and should sum to
MAD Mean Absolute Deviation is the average error, ignoring whether the error is positive or

MPE When the numbers in a data set are larger in magnitude, then the error measures are likely
to be large as well, even though the fit might not be as good.
Mean percentage error (MPE) and Mean Absolute Percentage Error (MAPE) are relative forms
of the bias and MAD respectively.
MPE and MAPE can be used to compare forecasts for different sets of data.

UNIT 15 APO DEMAND PLANNING (11 questions finals)

Flexible planning in R/3 uses LIS Logistics Information System where we store our
historical data.
Forecasting in APO Demand planning would be using Business Warehouse (BW) as a
source to store the historical data.
BW infrastructure is integral with APO DP.
Exception handling is integrated and you can design your own alerts.
Planning is based on the main memory. Flexible navigation in the planning table, variable drilldown. Enables collaborative planning and evaluation, like modelling. Wide range of forecasting
Demand Planning BOM.

What influences the Forecast?

Sales, Price and Ad Manufacturer

Promotions, Price and Sales Distributor

Season, Weather Customer

In APO DP, a characteristic is an organizational, or master data field. For ex, material
(product), plant, customer, or sales organization.
Characteristics are used to determine the level at which you are forecasting. APO DP has a
built-in library containing many commonly used characteristics. Customer unique characteristics
can also be created.
CVC - Characteristic Value Combinations: CVCs are stored in Planning Object Structure
(POS). CVCs represent the master data for APO DP.
Historical data is stored in the APO BW info cube and is organized by dimensions
(characteristics). Once historical data is stored, it may be analyzed to create historical facts of
characteristic combinations.

APO Business Warehouse (BW)

Info cube contains two major elements info cube is a multi-dimensional mini data warehouse.
a. Dimensional tables these tables contain location, product hierarchy, region etc.
b. Historical data incoming order, quantities and values of invoices etc.
InfoCubes are used for storingg actual data from OLTP systems in APO.

Forecasting Methods available

a. Univariate (statistical) only one independent variable available and forecast is done.
b. Multiple Linear Regression more than one independent variable. (MLR)
c. Composite

It always contain numeric or quantitative data that is useful for forecasting.
Eg: Sales history of a product, promotional data.
Usually appear as rows in spreadsheet looking screens. APO DP has a built-in library containing
many commonly used key figures. However, customer unique key figures may also be created.
Planning book it is like a spreadsheet. What key figures and characteristics to be displayed?

It is used to perform a particular business function. It also defines how we organize the columns
by week, month etc.
Planning Areas Characteristics and Key figures are stored in Planning Areas.
The Info Cubes are not a part of live cache but a part of APO BW. We are storing the
information into a spreadsheet looking screen and the forecast is stored in an area of live cache
called as TIME SERIES. We may have multiple forecasts and we can distinguish between them
by assigning different version numbers. The best forecast can be passed on from TIME SERIES
to another area of live cache called Orders.
Before creating the forecast, we need to define SELECTION ID
Selection ID it is going to have for each individual user how they wish to forecast?
For ex: a sales individual may wish to forecast on the basis of a Product and customer. Selections
may be used for re-use.
Lower level of details can be seen using the drill-down feature.
Aggregation is an automatic function by which key figure values on the lowest level of detail are
summed at run time and displayed or planned on a high level.
Disaggregation is the automatic function by which a key figure value on a high level is broken
down to the detailed level.
Disaggregation are done by using Proportional Factors.
They are derived from historical data and represent the ratios or percentage.

Forecasting Prerequisites
a. Master data in the form of CVC
b. A selection ID what to forecast
c. A forecast profile How to forecast
Univariate Forecasting technique
It is a statistical forecast.
Examples include


Moving avg.
Constant models, trend models.
Exponential smoothing
Crostons method sporadic demand
Automatic selection
Seasonal linear regression
The Holt-Winters method.

Master Forecast Profile It tells demand planning how to forecast. Definition of the key
figure to be forecast. Definition of the past and forecast horizons. Procedural specification for
the following forecasting types: Univariate, Multiple Linear regression, Composite. It indicates
what statistical methods to use.
Multiple Linear Regression (MLR)
APO DP supports Multiple Linear Regression as a forecasting technique.
MLR is used to determine how a dependent variable such as sales, is connected with independent
variables called casual variables, such as prices, advertising and seasonal factors.

Collaborative Forecasting customers will view your statistical forecast and can give
feedback on whether it is too high or too low. The result can be improved forecast accuracy.
Collaborative planning requires a strong business relationship between trading partners and most
Consensus Based Forecasting
It may be performed various ways including:

Different planning books for different forecasting organizations

Consensus forecasting business meetings where all parties participate in arriving at
the best possible forecast.

Alert Monitor
It may be used to send alerts regarding important abnormal situations.
For Demand Planning, the important abnormality is a forecast error that exceeds a pre-defined
threshold level.
Alerts may be delivered via several media including email.
Forecasting the demand for new products is difficult due to lack of historical data.
Forecasting the end-of-life products can also be done.
Promotional Planning It can be created to apply patterns to the demand forecast. The
patterns can be stored in the promotion pattern library and used, as required (multiple times).

The function is also available to detect promotion patterns in historical data and to create
promotion patterns based on them.

A planning object structure in APO DP will store Characteristics and CVC.
Characteristics are made up from Organizational data elements and Master Data elements.
Forecast profile in APO DP indicates what statistical forecast method to use.


After the DP has been approved, it must be released for operations planning.
After the release process, the forecast requirements will be classified as planned independent
requirements (PIR).
The forecast release will almost always contain 4 basic parameters:


MRP in R/3 or SNP in APO may act on PIRs to create replenishment planned orders.
APO release architecture

Time series (key figures) to Order objects (Category FA).

Time bucket profiles are used to create planned independent requirements.
The location shipping calendar is used to determine workdays.
Location split and Product split.
Daily bucket profile when the DP storage buckets profile does not contain days.

UNIT 17 SUPPLY NETWORK PLANNING (11 questions in final exam)

Part A
Introduction to Supply Network Planning
It is an intermediate time frame planning function. Its primary purpose is to create a good
rough-cut supply plan across the entire supply chain. It can used for either finite or infinite
capacity planning.
Three separate and unique replenishment planning engines are bundled into SNP:
They are

a. Heuristic
b. Capable-to-Match (CTM)
c. Optimizer
In addition to supply and demand matching, SNP is capable of performing two additional supply
management functions.
a. Deployment
b. Transportation Load Builder
Production Horizon

It is a given number of days relative to todays date and any date that arises greater
than this horizon is taken care by Supply Network Planning.
Anything inside of production horizon is taken care by PP/ DS.

SNP Process Flow

Release the demand plan - perform an SNP heuristic run, optimization run or CTM run Review
plan/ solve problems Finalize the SNP plan (make available to PP/DS) Release the feasible
SNP plan to DP Run Deployment Run TLB Manually process unconverted deployment
stock transfers.
Medium to Long-term Planning Strategies

Infinite planning Heuristic: Resources and material availability are not checked when
creating planned orders and stock transfers.
Finite Planning Capable-to-match (CTM) and optimizer: Simultaneous quantity and
finite capacity scheduling: SNP resources and material availability are checked during
creation of orders.

Short-term Replenishment Planning (Deployment)


Heuristic and Optimizer: Adjust stock transfers according to the current demand and
receipt situation.
Transport Load Builder (TLB): Groups together stock transfers.

SNP architecture
Planning Area holds Characteristics and Key figures.
Master planning object structure holds SNP characteristics and DP characteristics, Aggregates.
The APO DP stores its forecast in Time Series Live Cache.
Data view is a subset of a workbook.
Order based live cache All transaction data is stored as orders in the order live cache.
ATP categories are used to differentiate orders.

ATP categories can be grouped together into ATP category groups.

PART B APO Supply Network Planning

The role of heuristic planning is to plan the supply to meet demand throughout the entire supply
Heuristic planning is a quantity-based planning. It will create a supply quantity for a specific
time period regardless of the actual order quantities.
SNP heuristic plans in a level-level planning method similar to MRP in SAP R/3.
Heuristic planning assumes infinite capacity when planning.
Sourcing decisions (that is, what locations should be used as supplies) is driven primarily
by quota arrangements.
SNP creates planned orders and stock transfers in the network.
Quota Arrangements in Heuristic
Inbound quota arrangements control distribution of demands.
Outbound quota arrangements control distribution of receipts.
There are quota arrangements for individual products and selections.
Demands can be split (min quantity) and grouped together for time periods.
The following can be included in a quota arrangement:
a. External procurement
b. Stock transfer
c. In-house production
Other APO Master data used in Heuristic
Transportation lanes valid movements in the supply chain.
Quota arrangements percentage assignment of demands to sourcing locations
Lot Sizing lot for lot, fixed, target days supply, profiles, rounding values.
Calendar, Safety stock, scrap and PPM.
If transportations lanes are equal and there are no quota arrangements, then we compare
the product cost through PPM and choose the lowest price.
IF quota arrangement is defined, heuristic will not use transportation lane or PPM.

It is based on the assumption that resources have an infinite capacity.

After the SNP Heuristic run is complete, the planner can make a capacity check, which allows
the planner to see the impact that planned orders will have on resources and to quickly determine
whether or not the plan is feasible.
If capacity overload, an alert is displayed.
The planner can decide how to modify the production plan to meet demand before actually going
into production.

Increase capacity through overtime or additional resources.

Move work backward in time where capacity is available (subject to material availability)
Move work forward in time where capacity is available (subject to material availability)
Move work to an alternate resource
Split the order (i.e. Sub-divide) into quantities that will meet available capacity.

Capacity level = Consumption / Available Capacity.

Part C
Optimizer and CTM
Optimizer finite capacity planning
Capable to Match infinite or finite capacity planning.
LINEAR PROGRAMMING invented by George Dantzig and John von Neumann
LINEAR PROGRAMMING is a special case of mathematical problems in which a linear
function (called the objective function) is either minimized (like costs) or maximized (like
The other two components of a LP problem besides the objective function are:
a. Decision variables
b. Constraints

Decision variables are used to model the various decisions that may be required in supply
chain planning. Examples are:

What products should be produced?

How much of each product should be produced?
Where the products should be produced?
How should be products be transported?

Constraints represent realities that essentially bound the solution options to that the
result is feasible.
Examples include:
a. Demands must be met
b. Cannot exceed capacity
c. Storage capacity is limited
Non-Linear Programming When the objective function is a non-linear function.
Mixed-Integer Linear programming (MILP)
Mixed-Integer non-linear programming (MINLP)
SAP APO supports LP and MILP problems.


The Optimizer or solver in APO SNP uses LP to consider all relevant factors simultaneously.
SAP has an embedded 3rd party solver ILOG CPLEX into the SAP system. The optimizer
compares alternative solutions using costs that would be incurred.
Most cost-effective solutions chosen based on constraints and objective function defined in the
Penalty costs are used to prioritize demands. If a product brings high sales revenues, you
set high penalty costs.
Result due dates are violated or that safety stocks are not replenished.
Cost may include: production cost (PPM), storage costs (product master) and penalty costs
(product master).

Capable to Match (CTM)

It is an order-based planning method. Every single sales order or planned PIR is planned

CTM uses demand priorities as the primary basis performing supply network planning.
CTM doesnt perform optimization. It terminates when it finds the first feasible solution to
the problem.
Demand priorities may be defined using multiple methods consistent with corporate policy or
CTM process includes Demand Prioritization and Supply Categorization.


Safety stock planning
The uncertainties occur during planning
a. Demand uncertainty (forecast)
b. Replenishment lead time
Safety stock can be used to safeguard against these uncertainties.
a. Maintaining the safety stock manually in the product master
b. Calculate the time-independent safety stock in an SNP key figure in the Interactive
Planning Table.
c. Create a model independent safety stock to achieve a certain customer service level.

APO SNP provides more sophisticated safety stock planning than R/3.

Model-dependent Safety stock

a. The service level (SL) of the product master is used to determine the safety factor.
b. Information from transportation lanes and PPMs is used to calculate the replenishment
lead time (RLT).


Fair-share rule.
Actual deployment orders are created in R/3.

Transportation Load Builder (TLB)

It will optimize the used of the transportation fleet of vehicles. TLB only creates full loads.
Actual transportation orders are created in R/3.
TLB groups by

Products (transportation planner)

Time (pull-in-horizon)
Transportation zones

Ensures that it meets the minimum requirements for

a. Volume
b. Weight
c. Number of pallets.

SNP will perform supply planning Outside the SNP horizon

The heuristic in APO SNP is an Infinite capacity planner
THE CTM planner is Either finite or infinite capacity planner
The CTM program will cease its planning when the first feasible solution is reached.
The SNP supply network planning module is used primarily due to Intermediate Supply


APO SCE is a convenient graphical tool that allows customers to graphically view and edit
their APO master data objects.
The SCE allows a customer to create a filter containing those master data objects that are under
his responsibility. The resulting filtered master data objects is called as Work Area.
Master data is stored in Model 000.
Work Area provide a convenient way to filter the master data.
Access to frequently used objects.

Used for requests

Serve as filters
Configured for users


The supply chain engineer is used primarily to view and manage Master data
BOM cannot be viewed and managed through the SCE.
Work Areas in the SCE are used to Filter Master data in Models.
The supply chain engineer is a graphical tool True.


Factors influencing the success of a software implementation:


Project manager
Top management support
Technical issue
End user issues
End user training

Software implementation methodologies reduce the risk and cost of a software

A methodology is a structured, proven process of accomplishing a task.
It has existed for many years. They are usually defined by many phases.
The phases are


Project preparation
Business blueprint
Final preparation
Go live & support

Project Preparation
General conditions for implementing the project successfully.
It will include:

Defining the goals and objectives of the project.

Establishing the project organization
Creating the project plan
Determining the project standard procedures
Training the project team
Setting up the SAP 3 system landscape
Creating a communication plan for the project
Take certain benchmark measurements

Business blueprint
It is a critical phase of methodology. Failure to take adequate time to lack of fwd. thinking in this
phase will increase risk of overall project.
Realization It will begin the implementation of the functional requirements defined in the
blueprint phase.

Configuring the SAP system

Setting up the test environment
Setting up the security/admin
Setting up any workflow processes
Writing test scripts

Go live and support

This phase will launch the new application and provide the necessary support for a period of time
required to achieve institutionalization
Activities include:
a. End user re-training
b. Software trouble call support
c. Software de-bug support
d. Configuration re-setting