One of the most prominent applications of optimization in today’s business world is using advanced planning in supply chain management (SCM). In short, SCM refers to coordinating material, information and financial flows in a company’s value chain including business partners such as suppliers, contract manufacturers, distributors, and customers. In the following chapter we give a concise introduction to the SCM and supply chain planning terms and concepts sufficient for the following context of optimization applied to supply chain problems and introduction of the SAP APO software. For a more thorough treatment of SCM and advanced planning see the ample literature on this topic; Stadtler and Kilger (2004, [92]) is an excellent reference.
1.1 Supply Chain Planning – a Brief Introduction The term supply chain management was introduced by the business consul- tants Oliver and Webber in the early 1980’s (see Oliver and Webber, 1992, [75]) and since then a wide variety of definitions depending on individual’s point of view has been created. We will stick to our short and broad defini- tion as it is sufficient for the purpose of this book.
SCM is a business economics term and the involved tasks and processes as well as solution methodologies are classified in a business-oriented way. To someone with a mathematically oriented science background this may appear less exact and precise than desired. Therefore there is an arbitrary large potential for misunderstandings and conflict when dealing with mathematical issues such as optimization in SCM. We see it as a primary target for this book to help build a bridge between business administration and economics on the one hand and the exact sciences on the other hand. The basic advise in this context is to agree on some sort of communication quality standards ensuring that precise definitions are used and that it is checked whether all participants involved in the communication mean and understand the same when using certain terms.
As manifold as the definitions of SCM are the attempts to model processes
and concepts of actually doing supply chain management in a standardized
way. The Supply-Chain Council, a non-profit organization formed as an in-
dependent consortium in 1996, standardizes supply chain terminology and
processes in the widely accepted and adopted Supply-Chain Operations Ref-
erence (SCORR ) model. The Supply-Chain Council focuses on practitioners
rather than academia and comprises several hundred members, the majority
of which are companies and organizations applying SCM and the SCOR prin-
ciples to their business. The SCOR model aims at improving supply chain
processes and structures to serve customers’ needs as well as possible. There-
fore it takes into account processes and transactions from the “supplier’s sup-
plier” to the “customer’s customer” enabling supply chain evaluations from
different aspects – from within and outside the company. The model is di-
vided into four hierarchical levels dealing with process types, process categories,
process elements, and, finally, implementation. None of these four hierarchical
levels touches solution methodologies such as mathematical methods or opti-
mization techniques, however. In each level predefined best practice building
blocks are available which can be used to model supply chain processes in an
easily reconfigurable way. Figure 1.1 shows the SCOR model’s level one with
the five elementary management processes plan, source, make, deliver, and
return. In each of the process types there is potential for optimization such
as in long-term capacity planning, production planning, detailed scheduling,
or vehicle routing. Kallrath (2002, [51]) discusses this in more detail. The
other levels of the SCOR model provide a deeper level of detail; level two, for
instance, distinguishes between 30 process categories covering planning, exe-
cuting, and enabling. One of the biggest benefits of using a standard model
like SCOR is introducing a standard terminology enabling efficient communi-
cation between the parties involved in implementing a supply chain strategy.
The Supply-Chain Council has created a glossary that defines more than 300
terms and metrics allowing standardized performance benchmarking of a given
supply chain.
Diving a bit deeper into the four processes source, make, deliver, and return
we have to distinguish between planning the future events in the supply chain
and dealing with current events and tasks. The earlier is widely called sup-
ply chain planning (SCP), the latter supply chain execution (SCE) or supply
chain operations. Examples of SCP are strategic and tactical planning such as
network design, network or master planning, production planning, transporta-
tion planning and routing, demand forecasting, and so on; examples for SCE
include event tracking (for instance, in transportation), warehouse operations,
transportation load consolidation, shop floor control in manufacturing execu-
tion, etc. From the nature of these tasks it is quite straightforward to see that
on the one hand SCP typically is done once in a while in order to get results
that remain valid between instances of executing the SCP process, and SCE
on the other hand is in some sense “always online” because it has to be ready
to trigger and execute certain actions in response to real-world events. Master
planning as a typical SCP process, for instance, is – depending on the type
of industry and particular business – done once a day, once a week or even
less frequently and results in a rough-cut plan that allocates resources in the
production network to certain activities that work towards fulfilling customer
demands. Due to the fact that these plans typically affect multiple locations
and hence involve not only communication with electronic systems, but also
a certain amount of human interaction before they are actually executed, it
is not feasible to execute them continually.
1
Optimization as a solution technology for supply chain problems, at least on the tactical planning level, is “offline”, too. It takes a snapshot of the business data of interest, optimizes according to a well-defined model, and writes the results back to the business data repository (which usually is some sort of transactional business software system such as SAP R/3 or mySAP ERP). Often, performance, process, and problem localization requirements chase optimization away from SCE tasks: for most companies it is undesirable to re-calculate all delivery routes in the case of one delivery truck out of many dozen being involved in a traffic accident, for example. A more desirable scenario in this case would probably be to apply some local, rules-based algorithm that might suggest to extend the
This, of course, is an oversimplification. There are more facts to be taken into
account such as machine setup times for certain production processes that make product changes expensive. A good master planning algorithm takes this into account.
tour of another truck or to rent an additional vehicle serving the remaining
customers from safety stock. Optimization will be successful in this area if
the model is thoroughly designed to match the specific businesses, but this
usually rules out commercial, preconfigured optimization applications.
1.1.2 Supply Chain Planning and Advanced Planning Systems
A step towards formalizing and defining SCP in a more concise way than we
just did has been taken by Rohde et al. (2000, [79]) who define the supply
chain planning matrix classifying SCP tasks by planning horizon and supply
chain process. In Fig. 1.2 we give a version of the SCP matrix; note the
similarity of the processes along the x-axis with the SCOR process types
source, make, and deliver. The vertical axis in the SCP matrix corresponds
to the time horizon affected by the corresponding planning processes and also
gives an idea about how frequently the planning activities are performed.
Although the SCP matrix is not completely adopted in the literature and has
some structural drawbacks (cf. Tempelmeier, 2001, [96]), we will use it as a
tool displaying SCM functionality “at one glance” – without asking questions
like “Why does MRP belong to procurement?” or “Does demand planning
really belong to supply chain planning?”. With the exception of stand-alone
Material Requirements Planning (MRP) we see the functional modules of the
SCP matrix in software systems called advanced planning systems (APS):
Strategic Network Planning plans and coordinates strategic supply
chain processes creating suggestions for network design, cooperative supplier
contracts, distribution structures, manufacturing programs, etc. Decisions
made based upon this module are strategic and thus long-term in nature
and consequently cannot be undone or changed without considerable finan-
cial impact. The underlying data of such decision processes are mostly not in
the transactional business software but in archives such as data warehouses.
This leads to most companies setting up strategic network planning projects
using in-house or external consultants with customized mathematical software
tools independently of their enterprise business software.
Demand Planning takes a supporting role to the planning processes
by generating forecasted demand figures that are fed into the other planning
modules. Its functionality is based on statistical methods, on “collaboration”
between business partners such as key customers or distributors that can
help estimate future demand, and on data analysis methods such as “what-if
analyses”, aggregation/disaggregation, etc.
Master Planning creates feasible mid-term production plans synchro-
nizing the material flow along the supply chain and ensuring efficient resource
utilization in procurement, production, warehousing, and distribution. Usu-
ally this is a centrally executed process because its outcome affects the whole
supply chain and respects interdependencies of different supply chain parts
such as production facilities being able to manufacture the same product.
Master planning depends on input data obtained from network design, de-
mand planning, and cost data from all parts of the supply chain – these costs
are used to decide between options in procurement, production, and trans-
portation of goods. Depending on the complexity of the supply chain and its
processes master planning is often restricted to consider bottleneck materials
and/or resources or aggregated production processes.
Available- and Capable-to-Promise (ATP/CTP) help in order pro-
mising. When a customer order for a specific product comes in, ATP checks
quantities in stock and planned receipts (from procurement and production)
across the entire supply chain to determine a delivery date for the order. Op-
tionally, CTP can create production orders for the required product, which
involves changing and adapting production plans according to incoming cus-
tomer orders and available resource capacity.
Production Planning and Scheduling creates detailed, short-term
production plans for individual production areas (e.g., plants) based on the
results from master planning. The tasks can be divided into lot sizing, resource
utilization planning and detailed scheduling. Similar to master planning, the
goal is a feasible plan that respects resource and material constraints, but here
we look at only one production area in all detail, i.e., without aggregating or
restricting processes as in master planning. The detailed production plan is
passed on to manufacturing execution / shop floor control systems and hence
leaves the classical domain of APS.
Distribution and Transportation Planning determines which quan-
tities of goods are transported via which routes in the supply chain at what
times. Distribution planning deals with transportation quantities and stock
levels in connection with customer deliveries considering stock and transport
capacities whereas transportation planning performs routing and load plan-
ning determining cost effective and timely deliveries.
1.1.3 Advanced Planning Systems and Optimization
APS supplement the existing optimization programming libraries and pure
optimization engines with “ready-to-use” applications covering certain SCM
processes. Almost all major business software providers offer an APS as part
of their application suite covering the processes described above to a larger or
smaller extent. They typically divide their software into modules covering one
or more of the SCP matrix elements; often enough the quality of this coverage
is dependent on the industry – good functionality for production planning in
the process industry does not necessarily imply that the respective APS is
well-suited for discrete manufacturing such as high tech. In a complete SCM
solution these modules have to work together in an integrated way which sets
high standards for implementing and running those APS solutions. Often it
is most beneficial to use the APS and the ERP system from the same vendor
to take advantage of native system integration technologies.
Optimization techniques are applicable in the areas of Strategic Network
Planning, Master Planning, Production Planning and Scheduling, and Dis-
tribution and Transportation Planning. The remaining areas are typically
tackled with statistics (Demand Planning), rules-based algorithms (sales or-
der promising, ATP/CTP), or transactional and/or rules-based processing
(MRP). Commercially available APS that make use of optimization usually
offer comprehensive, but predefined mathematical models for one or more of
these application areas. We see those commercial APS as an augmentation to
the programming libraries, pure optimization engines, e.g., Xpress-MP TM and CPLEX.
No comments:
Post a Comment