Advanced
Planning and Scheduling: A Critical Part of Customer Fulfillment
S. McVey - December 10th, 1999
Drivers
from E-Commerce
The
face of retail is rapidly evolving from traditional brick-and-mortar facades
to electronic ones. While web businesses spend considerable effort in polishing
Internet front ends with sophisticated graphics and animation, they must also
give attention to back end fulfillment operations. Invisible to the web consumer,
these operations encompass networks of manufacturers, warehouses, and distributors
that shoulder the burden of filling orders and delivering products. Advanced
Planning and Scheduling (APS) offers powerful tools for addressing the challenges
presented to these networks by e-commerce.
For
all its promise, the Internet presents web businesses with significant challenges,
including:
Global
marketplace potential: Geographic marketplace boundaries have been all but
obliterated by the Internet. Web portals appear with the same clarity on laptops
in Boston or Bangkok and data transmission technologies continually improve
the speed at which on-line ordering activities take place. Though web businesses
can choose to limit delivery areas, in doing so they ignore a significant avenue
for growth. All that is needed is a means for making expanded distribution cost
effective.
Large
order volumes: Free access from every corner of the globe implies the potential
for massive order volumes. Unfortunately, an increase in order volume is usually
accompanied by a disproportionately large increase in the complexity faced by
back end fulfillment resources. These complexities arise because companies must
balance customer demand fulfillment with constraints of time, distance, resources,
and cost.
Easy
access to alternatives: Because the Internet dissolves physical distances,
companies are faced with the sobering fact that competitive sites lie a mere
mouse click away from their own. Web businesses that fail to satisfy customer
expectations regarding product quality or delivery risk losing customers to
companies that offer superior alternatives. In addition, the Internet serves
to level the competitive landscape, allowing small unknown companies to take
on established giants.
What
is APS?
Broadly
speaking, APS is a set of techniques that facilitate and/or automate human decision-making.
More specifically, APS comprises methods for solving complex problems that arise
in the operational infrastructure used to support customer fulfillment. The
solution to these problems is usually a sequence, strategy, or configuration
that results in the optimum material, time, or cost savings subject to a set
of constraints. An example from everyday experience is the problem of getting
dressed in the morning. If one specifies the goal of this problem as "a fully
dressed person", then many possible solutions exist, however, constraints imposed
on the dresser quickly reduce the number of acceptable outcomes. For instance,
shoes are not put on before socks (sequence), the activity cannot take more
than twenty minutes (strategy), fashion dictates that no one wear two ties (configuration),
and so on. This problem is a simple example of linear programming, a mathematical
technique for finding solutions of linear equations subject to simultaneous
constraints. It is just one example of the vast array of problems that are addressed
by APS.
In
manufacturing and distribution environments, constraints are far more difficult
to juggle because the problems are many times more complex. A few constraints
come easily to mind, such as raw material availability, resource availability,
and order due date compliance, but others arise from many sources, including:
Constrained
market conditions
-
Impossible
to support all market demands
-
Orders
must be prioritized by key customers, region, technology, etc.
Fixed
capacity requirements
Geographically
dispersed manufacturing and distribution operations
-
Complicated
supply chains
-
Operations
can extend to multiple plants, warehouses, distribution centers and cross
international borders
Discrete
manufacturing characteristics
-
Short
product, component life cycles place pressure on throughput
-
Component
and sub-assemblies can also be sold as end products
Process
manufacturing characteristics
-
Product
grading and inverted BOMs (Bills of Materials) introduce complexity
-
Routings
may visit operations iteratively
-
Setup
procedures vary significantly by product
Product
mix variability within manufacturing and transportation lead times
Unlike
the dressing experiment, problems in manufacturing and distribution that face
these types of constraints rarely lend themselves to a perfect or optimum solution,
since finding one would involve an inordinate amount of time and processing
power. Instead, reasonably good solutions are made to suffice. APS routines
usually employ a "back door" that enables them to stop searching for the optimum
configuration, strategy, or sequence and offer a merely "good" one - or at least
one that is better than could be found without APS.
Shortcomings
of older planning and scheduling techniques, such as MRP, MRP-II, and CRP, used
by many Enterprise Resource Planning (ERP) systems provide ideal opportunities
for APS. While these techniques represented a significant improvement over older
methods when they were introduced in the 1960s, many companies find they can
no longer match increasing demands on manufacturing and distribution networks.
The fundamental disadvantage of these approaches is that all fail to address
real world constraints sufficiently. In general, these approaches:
-
Fail
to respond quickly to changes in supply and demand
-
Do
not enable management of priorities across products and channels
-
Rely
on fixed lead times to calculate delivery dates, failing to take into account
all relevant delivery constraints
-
Do
not search exhaustively through BOMs or recipes to check the availability
of components, sub-assemblies and alternate parts when quoting availability
of a finished product
-
Do
not allow for Available-to-Promise (ATP) visibility across multiple sites,
resources, business units, and warehouses
APS
addresses these issues much more effectively than other techniques. Properly
implemented, APS can help back end fulfillment networks achieve reduced inventory
levels, better utilization of resources, shorter order cycle times, and lowered
operation and delivery costs
APS
Territory
APS
can address a variety of problems in different areas of the fulfillment network.
Four key areas are described below (see Figure 1).
|
Figure1
- Application Areas in Customer Fulfillment
|
|
|
-
Demand
management
Successful customer fulfillment begins, in many cases, with the ability
to make accurate predictions of customer demand. Forecasting involves making
assumptions about the marketplace and attempting to predict customer acceptance
for a company's products. By superimposing the effects of seasonal factors,
promotions, new product introductions, economic cycles, competitive actions,
and other events, companies seek to build a demand profile that can be acted
on by procurement, manufacturing, and distribution operations. Forecasting
algorithms, ranging from basic extrapolation to neural networks, which take
into account business events, are usually classified under demand management.
Demand management is an important venue for APS more because it provides
a point from which the rest of fulfillment activities follow. .
-
Production
planning
Production planning determines the blueprint for meeting customer demands
over long periods of time. Plans specify the product to be ordered or manufactured,
its required quantity, any planned raw material sourcing or procurement
activities, and other information. Usually, companies decide to allocate
available materials and resources to those customers' orders that will maximize
their profitability, fulfill previous inventory agreements, or aid the development
of preferred business partners. Production plans usually take into account
these allocations, or material and capacity set aside with particular business
objectives in mind. Industry frequently determines the scope of production
planning. For discrete manufacturing, production plans typically deal with
material availability only. For process manufacturing, resource availability
plays a larger role. Though most companies selling products over the Internet
do not manufacture widgets, their suppliers or their supplier's suppliers
most definitely do. Visibility across all links in the supply chain is becoming
increasingly necessary for web businesses to respond to customer orders
quickly.
-
Production scheduling
Production scheduling differs from production planning primarily in regard
to time horizon and granularity. Where production plans map out expected
material flows by week, month or even year, production schedules focus on
smaller intervals of time: day, hour, and even minute. Production scheduling
frequently involves horizons of less than two weeks, although industry is
the determining factor. Generally, volatile businesses tend to have shorter
scheduling horizons as longer ones are usually wasted effort because of
order cancellations. A good example of algorithms used for scheduling is
capacity balancing. Capacity balancing seeks to ensure that no resource
(production line, mixing vessel, kiln, packaging station) is loaded beyond
its available capacity. This is not as straightforward as it may at first
appear, since there may be more than one alternative resource for offloading
tasks, movement of tasks scheduled on an overloaded resource may be restricted
by upstream tasks, or due dates can be violated if tasks are postponed.
Since these problems involve a high degree of complexity, algorithms usually
employ heuristics, or "back doors", that allow them to return a feasible
answer within a reasonable time.
-
Logistics
Logistics involves planning the flow and storage of finished goods from
warehouses to distribution centers and end customers. Transportation time,
methods and costs are the primary trade-offs that APS seeks to balance in
order to arrive at the optimum distribution plan. One famous problem in
logistics concerns the plight of the travelling salesman who must visit
a number of cities scattered across the country while minimizing the total
distance traveled. Like scheduling problems, the logistics problems seek
to find a good solution quickly given the constraints. As in the case of
production planning, web businesses, while not involved in the actual shipping
of goods, can be held responsible by disgruntled customers when their order
doesn't arrive on time, making logistics a priority consideration.
-
Available-To-Promise
(ATP)
No discussion of APS is complete without mentioning ATP. ATP is not a specific
functional area like demand planning or production scheduling but is a feature
that provides visibility into the aforementioned areas. ATP refers to a
quantity of finished goods, work in process (WIP), raw materials, or some
combination of the three that is available for promising to customer orders.
ATP figures also incorporate the results of allocation routines that limit
supply to particular customers. Global ATP extends the reach of this allocation
to facilities and warehouses around the globe. Although vendors profess
to offer these capabilities, advances in data sharing technologies and security
are just beginning to make global ATP a reality.
Shortcomings
of APS
Though APS can
help web businesses satisfy high customer expectations, there are serious consequences
associated with packages that prospective users should understand. Two of the
most insidious are level of representation and algorithm visibility.
-
Level
of Representation
APS packages require that users create an electronic model of their fulfillment
network that represents how the business operates at a functional level.
Though the software provides the basic development tools for constructing
these models, it is often difficult to know just how detailed they should
be. A single model designed for a large network is often unable to sufficiently
represent important business process details. On the other hand, highly
detailed models that attempt to replicate every nuance of an organization's
workflows usually fail because of data integration and usability issues.
Success of APS hinges on striking a balance between the two extremes.
-
Algorithm
Visibility
Most vendors do not allow users to "peek behind the curtain" and observe
how their APS algorithms work. While some, like SAP and i2 Technologies,
license solver technologies from third party vendors such as ILOG, they
usually augment the acquired techniques with others developed internally.
Many subtleties of planning algorithms do not become evident until late
in the implementation phase when detailed testing takes place. For instance,
few users realize that i2's Rhythm Factory Planner will automatically schedule
low priority orders in advance of more critical orders if they are significantly
overdue.
Summary
Advanced
Planning and Scheduling (APS) has received new relevance with the advent of
the Internet marketplace. E-tailers like Amazon.com are successful not so much
because of their glittery graphics, but because of their back-end fulfillment
and customer service capabilities that enable them to satisfy customer demands.
As heavier "brick and mortar" manufacturers move towards e-commerce, APS will
continue to play an important role in the invisible back end network of manufacturing
and distribution. According to Ray Hood, CEO of EXE Technologies, "as companies
attempt to transform their traditional stores into e-businesses, they often
overlook the complexity and cost associated with having an efficient back-end
order fulfillment system." Armed with APS, web businesses can respond well to
challenges resulting from these complexities and prosper in the new global marketplace.
Glossary
MRP
(Materials Requirements Planning): A planning method that schedules on-hand
materials and/or procurements to external demand.
MRP-II
(Manufacturing Resources Planning): Successor to MRP, this method attempts
to factor in capacity constraints.
CRP
(Capacity Requirements Planning): A detailed procedure that checks capacity
against production plans produced from MRP.