Escolar Documentos
Profissional Documentos
Cultura Documentos
Visit us at www.cognizant.com
Executive Summary
Pre packing refers to packing of individual Stock Keeping Units (SKUs) of merchandise into bigger cases for
easier handling in the supply chain. The pre packs consist of varying quantities of different SKUs clubbed
together to form the lowest level of packaging hierarchy and are designed to flow through from the vendor
to the retail stores. Handling of these larger pre packs rather than individual SKUs proves to be cheaper
and faster at all touch points in the vendor to retail store supply chain.
Although pre packing has proven benefits in terms of easier and cheaper handling, the development of pre
pack configurations determining different pre pack compositions is a major challenge. Clubbing together of
individual SKUs into larger cases reduces flexibility of the supply chain. If pre packs need to be opened at
any point before they reach stores, additional cost and time is incurred. Inefficient decision making
regarding pre pack configuration may result in extensive opening up of pre packs and reconfiguration at
distribution centers. It may also lead to excess inventory at stores in case whole pre packs containing
superfluous SKUs are shipped to the stores. This can potentially negate the efficiency enhancement
targeted to be achieved as a result of pre packing.
An efficient pre pack decision-making process involves taking multiple decisions throughout the supply
chain management cycle, starting right from the demand forecast to initial planned allocation during
assortment planning, purchase order generation, as well as allocation performed at the distribution centers.
These decisions have to be taken under various degrees of uncertainty regarding the exact SKU Quantity
demand foreseen at the stores. Demand forecasts are typically probabilistic, based upon which assortment
planning is done and purchase orders are generated. At the time of stocks reaching distribution centers,
store demand is much more concrete in nature, with reduced variability. The pre pack decisions taken at
different points in the supply chain have to accommodate the probabilistic demand forecast and be robust
enough to handle variations in the actual allocation to stores from the distribution centers with respect to
planned allocation at the time of assortment plan development. The ultimate objective is to maximize flowthrough in pre pack terms and minimize handling of individual units throughout the supply chain. This
involves managing various trade-offs at each point with the objective of enhancing overall supply chain
efficiency and reducing the total supply chain costs.
We have developed an approach to pre pack optimization that balances the trade-offs at different points in
the supply chain using data intensive models. This approach has its roots in various standard operations
research problems and their mathematical modeling techniques. It is a modularized approach that focuses
on taking the right decisions at the right time in order to arrive at the best solutions that attain targeted
objectives.
With this approach and with a robust demand forecast, there is scope for a significant improvement in the
efficiency of the supply chain using pre packs that are packaged at the vendor shipment point and that flow
through the supply chain with minimal handling, straight to the retail outlet servicing store demands.
Table of Contents
1 INTRODUCTION TO THE PRE PACKING PROCESS
1.1
1.2
4
5
2.1
2.2
2.3
6
7
8
9
10
13
4 CONCLUDING REMARKS
16
17
5.1
5.2
5.3
5.4
17
18
20
22
6 REFERENCES
24
25
1.1
Pre pack sizes are normally decided by the retailer after considering vendor constraints, if any. This
decision on the retailers side is driven by the SKUs sales history and demand patterns. Demand patterns
depend on the sophistication of the data capture/forecasting/planning process followed by the retailer
leading to variations in the granularity and accuracy of the forecasted data. Some forecasting engines
predict weekly sales by store for a category, while others provide daily sales by SKU and store. The
consolidated demand
pattern
Figure 1 Pre Pack Touch Points in the Supply Chain
drives the planning and ordering
of the assortment that needs to
be supplied to an individual store.
Pre pack sizes are normally
decided when a new vendor
comes on board or when a new
SKU is launched by an existing
vendor. It is our opinion that
retailers need to give more
thought
to
the
pre
pack
configuration at this stage in
order to minimize inefficiencies in
the downstream supply chain. At
1.2
Over the supply chain, the two major points impacted by pre packing are:
1. Configuring the purchase orders into pre packs so that most of the orders are placed in terms of pre
packs and ordering of individual SKU units is minimized. This is done on the basis of forecasted
demand months before the SKU reaches the store.
2. Allocating to stores in terms of pre packs so that most of the allocation is in terms of pre packs and
opening of pre packs/overstocking due to extra SKUs sent is minimized. This is done on the basis of
forecasted demand a couple of days before the stock reaches the store.
In an ideal scenario where forecasted demand is the same as actual demand, the pre packing exercise boils
down to finding the best fit configurations that minimize ordering in individual units and opening of pre
packs. In reality, this is not the case. Actual demand can vary a lot from forecasted demand. This is mainly
due to the time difference between ordering and sales. Ordering to vendors may precede store allocations
and sales by months, especially if the sourcing is global. Events, unaccounted for by the forecasting model,
could occur which could affect the accuracy of the forecast.
Forecasting accuracy increases as we come closer to the event. So, to link the two areas impacted by pre
packing, we need to consider demand as probabilistic. Creating pre pack configurations in a probabilistic
demand scenario involves greater complexity in modeling the business scenario for optimization.
Success or failure of the pre pack optimization process, in terms of whether pre packs have to be opened
or not, depends on a lot of decision elements in the supply chain, such as:
a) Demand Forecasting the forecasting accuracy determines up to what degree the actual demand
would vary from the expected demand. With high forecast error, pre packs would have to be opened to
fulfill store orders even with accurate initial pre packing.
b) Assortment Planning the assortment planning determines what quantity of each SKU is planned to
be stocked at different stores. In a scenario where the actual store allocation demand varies greatly
from the planned assortment, opening of pre packs would become unavoidable.
2.1
A pre pack can be defined as a combination of different SKUs and quantities of each SKU contained within
a case of a particular weight/dimension.
Figure 2 Pre Pack Definition
PP-A
QTTY
SKU 1
a
SKU 2
b
SKU 3
c
SKU 4
d
SKU 5
e
In the example, PP-A is a Pre Pack Definition that has SKU 1, 2, 3, 4 and 5 in quantities a, b, c, d and e
respectively.
An SKU is defined as a style/color/size combination. Theoretically, a pre pack can have SKU
combinations of multiple styles/multiple colors/multiple sizes. However, in practice, pre packs have
A pre pack can be as large as possible, within the given constraints of dimensions, weight, total value
(in price terms), and so on. These constraints are defined on the basis of various factors, such as the
maximum weight that can be handled and the maximum value of goods the vendor or retailer is willing
to put in a pre pack.
2.2
Based on the understanding of pre pack touch points in the supply chain, there can be two approaches to
configure pre packs. The first one defines pre packs at the purchase order level and the second one defines
pre packs at the store ordering level. In line with the pre pack configuration information flow in the supply
chain, these approaches can be categorized as being top down and bottom up approaches.
2.2.1
The assortment plan provides planned allocation of SKUs for each store for each period (usually a week). In
modeling terms, the assortment plan output for each period (normally a week) can be denoted in terms of
the matrix shown in figure 3.
Figure 3 Assortment Plan Matrix
This assortment plan drives purchase order generation
that results in the flow of SKUs from the top (vendor)
down to the store. Pre pack configuration in this case is
done on the basis of top down planned allocation for each
store and hence is seen as the top down approach. At the
time of generation of purchase orders, the assortment
plan is configured in terms of pre packs and orders are
placed in terms of quantities of each pre pack definition rather than each individual SKU.
Figure 4 Pre Pack Decision at Ordering Level
Creating
larger
pre
pack
configurations and having greater
quantities and higher number of SKUs
ensures greater cost advantages in
terms of handling. But larger pre
packs
reduce
flexibility
in
redistribution at the Distribution
Centre
(DC)
and
consequently
increase probability of opening or
overstocking. Thus, it is necessary to
balance divergent goals to ensure that
the pre packs created go straight to
stores without opening. In the top down approach, even though configuration is done at an aggregate
planned allocation level, individual store allocation needs to be considered.
In another approach that initiates at the individual store allocation level, each store places its demand in
terms of configured pre pack definitions
Figure 5 Pre Pack Decision at Store Level
rather than SKU quantity terms. In this
approach, actual bottom up store orders
are configured into pre packs and sent
to the ordering level. Hence, this is
seen as the bottom up approach to pre
pack configuration.
Each store decides on its own pre pack
configurations within the overall preset
constraints of weight, dimension or
value. The store orders are then placed in terms of defined pre packs and some individual units remaining.
At the ordering level, orders are received from each store in terms of store level pre pack configurations as
well as individual SKU orders. These are aggregated and purchase orders are generated accordingly.
The stocks received from suppliers at distribution centers in this approach are in terms of store level pre
pack configurations. Thus the probability of opening of pre packs is minimized to a large extent, unless
allocation needs to be drastically different from initial store level demand.
2.3
The top down and bottom up approaches address the same questions in different ways. While the top
down approach handles the question of pre pack configuration on the basis of planned allocation, the
bottom up approach points to the actual store demand. These occur at different times in the supply chain
management cycle and are pieces of a larger picture. In terms of the rationale for taking the pre packing
decision at the procurement/ordering level or at the store level, both approaches have their pros and cons.
The decision to configure pre packs at the store level increases the probability of a pre pack passing
through to the store because it has been specifically designed for fulfilling a particular store demand. But
on the other hand, designing store specific pre packs reduces the flexibility of distribution. Having different
pre pack definitions for each store may mean a huge, almost unmanageable number of configurations at
the aggregate level. Plus, these pre packs may not be suitable for allocation to any other store in case the
store allocation requirement changes. This may also lead to a greater percentage of the overall assortment
being ordered in individual units.
The top down approach to pre pack configuration resolves some of these issues and creates pre packs
based on aggregate demand rather than individual store demand. This may ensure a manageable number
of pre packs and lesser individual SKUs in the ordering stage. It may also create more flexibility in
allocating pre packs to stores. But the major concern at this stage is the increased probability of pre pack
opening at warehouses.
3.1
3.1.1
The process of servicing store demand starts from defining the service levels the level and the degree up
to which store demand would be fulfilled through the supply chain. Service levels are defined based on the
balancing act between the cost of overstocking and the cost of shortage. While excess inventory has its
own adverse impact on ROI, stock outs mean lost sales and may also translate to loss of brand and store
image. A balanced service level attains the right Fill Rate (demand met/expected demand) and increases
the probability of no stock out, keeping in view the criticality of the SKU to the store image, as well as the
speed at which it moves off the shelf. In an ideal scenario, there should be a distinct service level set for
each Store-SKU combination. But in practice, service levels for stores are generally defined for Store-SKU
group combinations, or even more broadly, at the store level.
3.1.1.1 The Service Level Classification Matrix
At the store level, SKUs can be classified
based on the speed of their movement off
the shelf as well as the criticality of their
demand. Once SKUs are classified into
each segment of this matrix, service levels
can be set for each segment. These
service levels and the classification matrix
affect the pre pack optimization process
at every subsequent step. Collectively,
these service levels can be represented by
a service set for each store. This service
level set would have a number of service
levels for each SKU segment for the store.
With pre packing, pre packs rather than
10
The pre pack configuration exercise can be broken into two parts: determining different sizes for pre packs
and determining the composition of each pre pack definition. Both steps are taken in a retailer-supplier
collaborative environment, where mutually agreed fundamental pre pack properties, such as weight and
dimension of the carton, and regulations on pre pack composition are set. In most cases, it is important to
set the carton sizes (weight and permissible number of SKUs) before determining the composition. In this
context, a template is defined as the capacity of a carton that is used to pack in individual units (SKUs) of
mutually agreed style/color/size variations.
The Optimization Model for Pre Pack Template Definition involves:
Figure 8 Pre Pack Template Definition
Based on past data on store demands as well as flow through of individual units into the supply chain,
clubbing together, or lot-sizing of SKUs in different quantity groups can be arrived at, resulting in an
optimal number of cartons and optimal carton capacities. This lot-sizing needs to fulfill the boundary
conditions of maximum permissible weight of a lot packed in a carton and the dimension constraints.
3.1.3
Assortment plans determine the width and depth of the assortment to be carried catering to the cross
section of consumer demand. They also determine planned allocation of SKUs to stores. Assortment plans
are normally based on probabilistic demand scenarios and in cases where actual allocation is widely
11
Development of an optimized assortment plan, along with service levels and pre pack template definitions,
fulfills the pre season activity set for pre pack optimization. These steps form the foundation for the
subsequent in season steps.
12
3.2
In Season Steps
After the pre season steps are developed, the two steps of configuring the purchase order into pre packs
and allocating pre packs from distribution centers to stores are performed.
3.2.1
Purchase orders are raised against the open-to-buy derived from the assortment plan. In pre packing,
these orders consisting of an SKUQuantity assortment plan for each vendor have to be configured in terms
of pre packs rather than individual SKU-Quantities. In a scenario in which the vendor is ready to ship as
soon as the order is received, the purchase order raised needs to specify the pre pack configurations. But in
a scenario where there is a significant time lag between the instant the order is received and the moment
the vendor starts shipment, the initial order can be sent in terms of aggregate SKU-Quantity, and pre pack
definitions can be sent just before the packing stage starts at the suppliers point. This would ensure that
the pre pack decision is taken as close to the actual demand as possible.
As an output of this model, the assortment plan is clubbed into pre pack configurations and after this point,
pre packs become the lowest level of packaging hierarchy. The pre pack configurations map the whole
assortment plan and fit into the Store-SKU-Quantity demands. The key element in this step is to determine
13
Pre packs may contain SKUs of a particular style, size or color. Generally, SKUs are characterized first
by style and then by size and color. Even if there are no constraints due to pre pack template
definition, pre pack configurations may have SKUs that differ in color and size, but belong to the same
style.
Fast-moving SKUs may be clubbed together or in some pre packs, there could be a number of fastmoving SKUs together with a few slow-moving SKUs. Slow-moving non-critical SKUs may not be
clubbed together in pre packs, as this may create excess inventory in stores.
14
When the pre packs reach the distribution centers, demand from stores is closest to the actual consumer
demand. The accuracy of the assortment plan is tested at this instant. If the assortment plan is totally
accurate, allocation is the same as the assortment plan, and this is a straight flow through process. But the
assortment plan being probabilistic, in most cases, actual store allocation from distribution centers differs
from the assortment plan. This variance introduces complexities at two levels.
The overall Store groupSKU group demand at the time of allocation may differ from the demand
forecast. If this deviation is significant, either excess inventory or shortages are inevitable, whether pre
packing is done or not. Inaccurate demand forecast is an input which causes a major portion of excess
inventory or shortages, and thus the pre pack optimization process needs to have as accurate a
demand forecast as possible. Using pre packs, some inaccuracies in forecast can be smoothened, but
the utility of pre pack optimization in a widely inaccurate demand forecast scenario is compromised to a
great extent.
The overall Store groupSKU group demand at the time of allocation may not vary significantly, but
actual SKU-Store allocation may be widely different from the assortment plan output. This is a result of
inaccurate assortment planning something which requires regrouping of SKUs at distribution centers
and revised allocation. Using an un-optimized assortment plan as an input to pre packing may
necessitate the opening of pre packs for reallocation of SKUs, in case the plan is way off target. The
utility of assortment optimization lays in the fact that it goes a long way in reducing variances in the
assortment plan to allocation figures.
With an optimized assortment plan used as input for pre pack configurations, reallocation at the level of
entire pre packs rather than individual SKUs becomes much more feasible. The decision at this level is to
allocate the pre pack stocks received according to the deterministic store allocation demand. The best
situation is when all store demands can be fulfilled in an integral number of pre packs. If this is not the
case, a decision may be taken to hold back entire pre packs at distribution centers and ship lesser SKUs to
stores than demanded, or
Figure 14 Pre Pack Allocation to Stores
to ship entire pre packs to
stores with more SKUs than
demanded. The third option
is to open pre packs and
ship
individual
units.
Stocking/allocation
decisions are based on the
cost of opening pre packs,
handling individual units,
holding
stock
at
a
warehouse, as well as
overstocking/under stocking
at retail stores.
15
4 Concluding Remarks
The intent of pre pack optimization is to eliminate handling of individual SKUs throughout the supply chain.
In the pre packing context, individual units may appear at the time of ordering or at the time of shipment
from vendors to distribution centers, or at the time of opening of pre packs to allocate SKUs for fulfilling
store demands. These touch points in the supply chain require various decisions to be taken so that overall
handling of individual units is minimized. In the modularized, step-by-step approach taken to achieve this
purpose, there is an effort to answer the right questions at the right time in the supply chain, so that the
efficiency of the overall process is enhanced. These modules appear as independent pieces but are actually
part of a unified picture because the fundamental assumptions, inputs and constraints used at each step
are the same. Through simulation of each module and feedback of one module output to another, the
optimization outputs coming out of each module are refined in order to give enhanced benefits.
Furthermore, the underlying objective in developing all these modules is to eliminate individual SKUs from
the supply chain and use pre packs as the lowest level of packaging hierarchy. By a systematic and stepwise approach, this objective is attained.
16
5.1
Problem Statement
A template is defined as the capacity of a carton that is used to pack SKUs. At this level, all SKUs of a
particular style are considered identical. Past data for allocation made to different stores for each style and
for order sizes placed to vendors for each style is available. Individual units of a style have to be grouped
into cartons of different capacities. Thus, most of the units can be packed in the cartons. The objective of
this exercise is to find out the optimum carton sizes for each vendor within which the pre packs to be
ordered can be defined.
Theoretical Reference
This problem is similar to a number of Operations Research problems, such as lot-sizing in a multi-echelon
inventory system, as well as a multidimensional bin-packing problem. In essence, both template definition
and order configuration optimization modules work with similar theoretical references. The overall decision
of determining the size of pre packs and composition is broken up. Template definition focuses on
determining sizes, while order configuration determines composition. Thus, template definition turns out to
be a special and simplified case of a three- dimensional bin-packing problem, where the objective is to
determine a minimum number of bin sizes in which smaller three-dimensional rectangular boxes can be fit.
There are also elements of the multi-echelon lot-sizing problems in this case, as the bins (or pre packs)
have to fulfill the past demand data of each store at a disaggregate level, that is, without opening the bins.
Modeling Constructs
Inputs
1. Supplier/retailer agreement on style/color/size variety permissible for each pre pack. It is assumed that
a template is defined for all styles supplied by the vendor.
Vendor V produces styles Style 1 Style 2 Style n. It is assumed that a single set of pre pack
sizes has to be developed for all styles produced by vendor V.
2. Past data on order sizes and allocation. It is assumed that data is available on the past y number of
years for order sizes placed and allocation to individual stores.
For vendor V, there is data on quantities ordered in each order in the past y years.
17
For all styles manufactured by vendor V, there is data on quantities of each style allocated to each
store for the past y number of years.
Constraints
1. It is assumed that the maximum number of templates that can be defined is fixed.
For a particular vendor V, there can be only p numbers of template sizes defined.
2. There is a cap on the maximum number of units in a particular pre pack template. This is based on the
fact that any pre pack capacity cannot be too large, as handling it would be impossible.
Let us assume that for a particular vendor V, a pre pack template cannot contain more that r
number of individual units.
Objective Function
The number of different pre pack sizes has to be optimized so that the sizes are a best fit with the past
allocation to each store as well as the order sizes for each style.
Define different sizes X (X1, X2, a maximum of p sizes) so that
a. The past allocation demand from stores for the style group shipped by vendor V is fulfilled in the
sizes defined to the maximum level.
b. Past order sizes can be grouped into the sizes defined to the maximum level.
c. In both cases, the total number of individual units remaining after grouping into sizes defined has
to be minimized.
Desired Output
A set of pre pack sizes for each vendor X (X1, X2, a maximum of p sizes)
Assumptions
1. The first level of assumptions is that all the inputs defined above are available and constraints are
provided in quantitative terms.
2. Each individual unit (a pair of shoes in its own box) is of the same dimensions.
3. It is assumed that all styles manufactured by a vendor can be packed into one carton. If that is not the
case, carton sizes specific to those styles which can be packed in one carton size would have to be
developed.
5.2
Assortment Optimization
Problem Statement
There is a weekly demand forecast for a particular style for a particular group of stores. Each style can be
disaggregated in terms of color and size SKUs. The SKU groupStore group demand forecast has to be
disaggregated in terms of probabilistic SKU-Store forecast based on past data available from stores on their
SKU sales as well as their service level requirements.
18
Inputs
1. Weekly Demand Forecast
a. Assumption: Style A can have n number of color/size variations. Thus the forecast would be for
SKU group A that has n number of SKUs. The SKU group A denotes all the color-size
variations (n number) for the Style A.
SKU group n contains all SKUs with Style A colors and sizes may vary within that style,
giving n number of SKUs.
b. Assumption: The forecast is for s number of stores.
s number of stores can be in one channel or area for which the forecast is made.
Forecast for n SKUs for s stores = f (D), where f (D) may be a probabilistic function obeying a
given distribution. It is assumed here that this demand would follow normal distribution with
mean D and Std. Deviation d. This demand can be readily assumed to follow some other
distribution function and the equations derived would change accordingly.
2. Past data on allocation of each SKU/similar SKUs for each store.
a. For each SKU among the n SKUs for which the demand forecast has been made, past sales data
from each store (from the s number of stores) is available.
3. The service levels to be maintained at each store are provided. Based on the demand distribution
(whether it follows normal distribution or some other one), the minimum stock requirement for a
particular service level can be calculated.
Let us assume Qij is the stock of SKU i (which can vary within the n number of SKUs in the
group) for Store j (which can vary within the s number of stores in the group). Thus Qij(sl) is
the minimum stock requirement for the SKU to meet the service levels.
Constraints
1. Total demand for all SKUs for all stores within the group has to be equal to the forecast demand.
19
F(Qij), the planned stock of SKU I for store J is a probabilistic quantity having the same probabilistic
distribution as the demand f(D).
Sum of all f(Qij) distributions would be equal to f(D) distribution.
2. Demand for each SKU-Store combination falls within the range of demand projections done on the
basis of past SKU-Store allocation data.
Time Series analysis of past demand for each SKU-Store combination-analysis of base, trend and
seasonality would provide demand curves demand = f(time) for all SKU-Store combinations.
These demand curves would have a tolerance level of T. The probabilistic demand functions for
future demand for each SKU-Store combination-f(Qij) would have to fall within the tolerance range.
3. Service levels at stores have to be met.
Allocation Qij, the instance of the distribution f(Qij) >= Qij(sl) for all stores and SKUs.
Objective Function
Determine optimal quantity of each SKU to be kept in each store to be able to fulfill the service levels.
Determine the probabilistic functions f(Qij) for all SKUs and stores within the group so that service
levels are met and the sum of all the individual distributions is equal to the demand function f(D).
Mathematical Model
Number of stores in the store group for which the demand forecast is available = s
Number of SKUs in the SKU group for which the demand forecast is available = n
Qij = demand for SKU I for Store j
Inputs
Qij for all past periods
Demand forecast = i j(qij) = D (D is a probabilistic demand and all the equations have to be expressed
in terms of probabilistic demand)
Service levels for each store within the group
Model
Determine qij for all I = 1-n and j = 1-s
So that:
D is fulfilled and service levels are met
5.3
Problem Statement
A number of SKUs of a particular style have been ordered from vendor V. The probabilistic allocation of
each SKU for each store is available. The cost of handling pre packs and the cost of handling individual
units is known. The cost of handling of individual units is very large compared to the cost of handling pre
packs. Pre packs should be defined so that most of the flow of items happens in terms of pre pack units
and handling of individual units is minimized. A pre pack definition would mean selecting a combination of 1
to n number of SKUs for a particular style and selecting quantities for each SKU so that the total number
of units in the definitions equals one of the pre pack sizes defined in the template.
20
Modeling Constructs
Inputs
1. Probabilistic demand for each SKU for each store from the assortment plan.
F(Qij) is known for each SKU-Store combination for SKUs.
2. Total order placed for each SKU aggregated for all stores.
For the vendor V, total orders placed for each SKU shipped is known.
3. Cost of handling of pre packs and handling of individual units.
Let us assume a uniform cost Cp of handling all individual units packed into pre packs, and a
uniform cost Cu of handling all individual units handled individually. These costs are summation of
all costs throughout the process for pre packs as well as individual units. It is assumed that Cu is
much larger than Cp.
4. Service levels to be maintained at each store.
Qij(sl) for all SKU-Store combinations is known.
Constraints
1. Demand for each store has to be fulfilled within the permissible service levels.
The total demand for a particular store, summed over all SKUs and for all vendors, has to be
fulfilled so that Qij(sl) for all SKUs within the store is fulfilled.
2. The sizes of pre packs defined can only be among the ones given in the templates.
If xij is the number of units of SKU I stocked in pre pack j, then for a particular pre pack definition
j, the sum of individual units of all SKUs present in it has to be equal to one of the pre pack sizes
as developed in the template module X (X1, X2, a maximum of p sizes).
3. The sum of quantities of a particular SKU aggregated over all pre pack definitions configured in the
order have to be equal to the total order quantity of that SKU.
If xij is the number of units of SKU I stocked in pre pack j, then for a particular SKU I, the total
number of units packed into all pre packs of different combinations has to be equal to the quantity
ordered for the SKU.
Objective Function
Minimize overall cost of handling inventory by maximizing ordering and shipment in pre packs.
21
If each pre pack definition PPj has Nj number of units in the configured order, the total cost of handling
of all pre packs would be Cp times the total number of individual units packed in pre packs of all
definitions for the order. If is the total number of individual units left over after pre pack
configuration, the total cost of handling individual units would be Cu times . The objective is to
minimize the total cost of handling. While doing so, the lower cost Cp would ensure that most units are
handled in pre packs and opening of pre packs to handle individual units is minimized.
Mathematical Model
Inputs
X, where X is an element of the set of pre pack sizes (capacities available)
P = number of different pre pack definitions (a decision variable)
PPj = pre pack definition j, 0 =< j =<P (total number of pre pack definitions can be P)
Xij = number of units of SKU i in PP j
0 =< Xij =<X (minimum number of items of a particular SKU in a particular pre pack can be zero,
meaning that SKU is not defined within that pre pack. Maximum number can be equal to the pre pack size,
meaning only that SKU is defined for that pre pack)
Nj = number of pre packs of j type in an order (an order can have many units of the same pre pack
definition. Nj signifies the number of each pre pack definition that the order to be configured would
contain)
Cp = cost of handling of pre packs
Cu = cost of handling of individual units (Cu>>Cp)
= total number of individual units handled in the system
Model
Minimize total cost Z = Cp (
(over j=1-p) (nj)) + Cu
5.4
Problem Statement
The supplier has sent stocks in pre pack definitions sent during ordering. The cost of stocking pre packs as
well as the cost of stocking individual units at the store and at the warehouse is known. A pre-determined
service level fixed for each of the stores needs to be maintained. At this stage, the actual deterministic
demand for each SKU for each store is available. The allocation of stocks to each store has to be decided
upon, to maintain service levels while minimizing the handling of individual units.
22
Inputs
1. Total number of different pre packs as well as individual units received from supplier.
There are Nj number of pre packs each of definition PPj that have been received at the distribution
center from supplier. Each PPj contains SKUs I in quantities xij. In addition to them, there are
number of individual units received from supplier which couldnt be packed in pre packs.
2. Demand from each store for each SKU.
Actual allocation demand Qij (deterministic) for each SKU-Store combination at the distribution
center.
3. Service level to be maintained at each store Qij(sl) is known for each SKU-Store combination.
4. Cost of stocking pre packs and individual units at warehouse and retail store points.
Holding cost of inventory, in pre packs as well as individual units, is known for DCs and stores.
Constraints
1. Allocation has to meet service levels.
Allocation Qij >= Qij(sl)
Objective Function
Minimize the overall cost of stocking in the Warehouse-Store part of the chain, while maintaining a
minimum quantity of stocks at the stores to meet service levels.
Total cost of stocking is equal to the cost of stocking at DC plus the cost of stocking at each store.
Stocking can be done either in pre packs or individual units after breaking of pre packs. The total cost
function is calculated. It needs to be minimized in a way that the minimum in-store allocation for each
SKU-Store combination is equal to or greater than that required for meeting service levels.
23
24
25
Contact Details:
Cognizant Technology Solutions
500 Glen Pointe Center West
Teaneck, NJ 07666
Ph: 201-801-0233
Fax: 201-801-0243
Toll Free: 1-888-937-3277
E-mail: saleseast@cognizant.com
United Kingdom
Ph: 44 -(0) 207 321 4888
Fax: 44 - (0) 207 321 4890
E-mail: sales@uk.cognizant.com
Germany
Ph: (069) 66 04 450
Fax: (069) 66 04 542
E-mail: sales@de.cognizant.com
Chennai
Ph: 91-44-8113063, 8116033
Fax: 91-44-8112622, 8112507
E-mail:sdc@chn.cognizant.com
Visit us @ www.cognizant.com
26