Course:

LOGISTICS ENGINEERING AND SUPPLY

CHAIN DESIGN

International University
Vietnam National University HCMC
 Chapter 6
Aggregation in Supply Chain

Lecturer: Dr. Nguyễn Hằng Giang Anh

Email: [email protected]
International University
Vietnam National University HCMC
 Main Contents

1. Aggregation in Supply Chain

- Aggregating multiple products in single order
- Safety Inventory with Supply Uncertainty
2. Impact of Aggregation on Safety inventory

International University
Vietnam National University HCMC
 1.
AGGREGATION IN SUPPLY
CHAIN

International University
Vietnam National University HCMC
 Aggregation in Supply Chain

Benefit?
International University
Vietnam National University HCMC
 Shipment Aggregation

Shipment aggregation may be considered when partial shipments to the same

customer may be combined to save overall logistics costs. In this example:
- Four separate 2000lb loads are scheduled to ship over the course of a three day
period. Total cost of logistics is $1840
• $1600 in truck cost ($400/shipment for each 2000lb shipment)
• $240 in locating, scheduling, cutting BOL’s, loading and unloading four trucks
($60/truck)
- Four separate 2000lb loads are aggregated into one 8000lb truckload shipment. Total
cost of logistics is $1260
• $1200 in truck cost ($1200 to ship the 8000lb load)
• $60 to locate, schedule, cut a single BOL, load and unload one truck
International University
TOTAL LOGISTICS COST SAVINGS: $580
Vietnam National University HCMC
Aggregating Multiple Products in a Single Order

❖ Aggregate multiple products originating from the same

supplier.
❖ Single delivery coming from multiple suppliers or single
truck delivering to multiple retailers.
❖ Fixed costs for each product can be reduced by savings in
Transportation costs as well as Receiving and loading
costs.
→ Fixed costs are decreased → Reducing Lot size →
Reducing cycle inventory.

International University
Vietnam National University HCMC
Lot Sizing with Multiple Products or Customers

❖ Ordering, transportation, and receiving costs of an order

grow with the variety of products or pickup points.
❖ Is it cheaper or more expensive for WinMart to receive a
truck containing a single product than it is to receive a
truck containing many different products?
❖ A portion of the fixed cost of an order can be related to
• Transportation (depends only on the load and is
independent of product variety on the truck).
• Loading and receiving (increases with variety on the truck).
→ How optimal lot sizes may be determined in such a setting?

International University
Vietnam National University HCMC
 Lot Sizing Determination

❖ Our objective is to arrive at lot sizes and an ordering

policy that minimize the total cost.
❖ We assume the following inputs:
• 𝐷𝑖 : annual demand for product 𝑖
• 𝑆 : order cost independent of the variety of products
included in the order (transportation)
• 𝑠𝑖 : additional order cost incurred if product 𝑖 is included in
the order.

International University
Vietnam National University HCMC
 Approaches to the lot-sizing decision

1. Lots are ordered and delivered independently for

each product.
2. Lots are ordered and delivered jointly for all products.
3. Lots are ordered and delivered jointly for selected
subset of the products.

International University
Vietnam National University HCMC
1. Lots are ordered and delivered independently
for each product

❖ Each product is ordered independently of the others.

❖ This approach does not use any aggregation and results
in high cost.
❖ This scenario is equivalent to applying the EOQ formula
to each product when evaluating lot sizes.

International University
Vietnam National University HCMC
 Example 1 (1/3)
• Best Buy sells three models of computers, the Litepro, the
Medpro, and the Heavypro. Annual demands for the three
products are 𝐷𝐿 = 12,000 units for the Litepro, 𝐷𝑀 = 1,200
units for the Medpro, and 𝐷𝐻 = 120 units for the Heavypro.
• Each model costs Best Buy $500. A fixed transportation cost
of $4,000 is incurred each time an order is delivered. For
each model ordered and delivered on the same truck, an
additional fixed cost of $1,000 per model is incurred for
receiving and storage. Best Buy incurs a holding cost of 20%.
• Evaluate the lot sizes that the Best Buy manager should order
if lots for each product are ordered and delivered
independently. Also evaluate the annual cost of such a policy.

International University
Vietnam National University HCMC
 Example 1 (2/3)

International University
Vietnam National University HCMC
 Example 1 (3/3)

International University
Vietnam National University HCMC
 REVIEW

𝟐𝐱𝑫𝐱𝑺
𝒉𝐱𝑪

International University
Vietnam National University HCMC
 2. Lots are ordered and delivered jointly
for all products (1/2)

❖ This approach aggregates all products in each order →

thus amortizing the order cost 𝑆 across multiple products.
What is the weakness of the approach??
❖ The weakness of the approach is that low-demand
products are aggregated with high-demand products in
every order → resulting in all order costs 𝑠𝑖 to be incurred
in each order.
❖ Complete aggregation → results in high costs if the
product-specific order cost for the low-demand products
is large.

International University
Vietnam National University HCMC
 2. Lots are ordered and delivered jointly
for all products (2/2)

• All products are ordered and delivered on the same truck

each time an order is placed.
• The combined fixed order cost per order:
𝑆 ∗ = 𝑆 + σ𝑖 𝑠𝑖
• Let 𝑛 be the number of orders placed per year
𝐷𝑖 ℎ𝐶𝑖
→ Total annual cost: 𝑇𝐶 = σ𝑖( ) + 𝑛𝑆 ∗
2𝑛
σ𝑖 𝐷𝑖 ℎ𝐶𝑖
• The optimal order frequency: 𝑛∗ =
2𝑆 ∗
• If the capacity of each truck is considered, then:
∗ 𝐷𝑖
𝑛 = Capacity for each product

International University
Vietnam National University HCMC
 Example 2 (1/3)

• W.W. Grainger sources from hundreds of suppliers and is

considering the aggregation of inbound shipments to lower
costs.
• Truckload shipping costs $500 per truck along with $100
per pickup. Average annual demand from each supplier is
10,000 units. Each unit costs $50 and Grainger incurs a
holding cost of 20%.
• What is the optimal order frequency and order size if
Grainger decides to aggregate four suppliers per truck?
What is the optimal order size and frequency if each truck
has a capacity of 2,500 units?

International University
Vietnam National University HCMC
 Example 2 (2/3)

• Demand for each product 𝑖, 𝐷𝑖 = 10,000 units

• Fixed Ordering cost:
• Common order cost, S = $500
• Product-specific order cost, s𝑖 = $100
• Holding cost, h = 0.2
• Unit cost for each product 𝑖, 𝐶𝑖 = $50

→ What is the lot size for each product?

→ What is optimal order frequency?

International University
Vietnam National University HCMC
 Example 2 (3/3)

International University
Vietnam National University HCMC
 3. Lots are ordered and delivered jointly
for selected subset of the products (1/3)

❖ We discuss a procedure that is more selective in

combining products to be ordered jointly. The
procedure does not provide the optimal solution;
however, it yields an ordering policy whose cost is close
to optimal.
❖ Assume that the products are indexed by 𝑖, where 𝑖 varies
from 1 to 𝐼 (assuming a total of 𝐼 products). Each product
𝑖 has an annual demand 𝐷𝑖 , a unit cost C𝑖 and a product-
specific order cost s𝑖 . The common order cost is 𝑆.
❖ We first describe the procedure in general and then apply
it to the specific example.

International University
Vietnam National University HCMC
 3. Lots are ordered and delivered jointly
for selected subset of the products (2/3)

• Step 1: Identify the frequency of the most frequently ordered

product 𝑖 ∗ , assuming each product is ordered independently

ℎ𝐶𝑖 𝐷𝑖
𝑛ത = 𝑛ത 𝑖∗ = 𝑚𝑎𝑥 𝑛ത 𝑖 =
2 𝑆 + 𝑠𝑖

→ The most frequently ordered product is 𝑖 ∗ included each time

an order is placed.
• Step 2: For all products 𝑖 ≠ 𝑖 ∗ , evaluate the ordering frequency:

ℎ𝐶𝑖 𝐷𝑖
𝑛ധ𝑖 =
2𝑠𝑖

International University
Vietnam National University HCMC
 3. Lots are ordered and delivered jointly
for selected subset of the products (3/3)

• Step 3: For all 𝑖 ≠ 𝑖 ∗ , evaluate the frequency of product 𝑖

relative to the most frequently ordered product 𝑖 ∗ to be:
𝑚𝑖 = 𝑛/ത 𝑛ധ𝑖
• Step 4: Having decided the ordering frequency of each
product 𝑖 , recalculate the ordering frequency of the most
frequently ordered product 𝑖 ∗ to be:
σ𝑖 ℎ𝐶𝑖 𝑚𝑖 𝐷𝑖
𝑛=
2 𝑆 + σ𝑖 𝑠𝑖 /𝑚𝑖
• Step 5: Evaluate an order frequency of 𝑛𝑖 = 𝑛/𝑚𝑖 and the
total cost of such an ordering policy
𝐷𝑖
𝑇𝐶 = 𝑛𝑆 + ෍ 𝑛𝑖 𝑠𝑖 + ෍ ℎ𝐶𝑖
2𝑛𝑖
𝑖 𝑖

International University
Vietnam National University HCMC
 Example 3 (1/3)

Consider the Best Buy data (Example 1). Product managers

have decided to order jointly, but to be selective about which
models they include in each order. Evaluate the ordering
policy and costs using the procedure discussed previously.

❖ Demand, DL = 12,000/year, DM = 1,200/year, DH = 120/year

❖ Fixed Ordering cost: a separate truck delivers each model
• Common order cost, S = $4,000
• Product-specific order cost, sL = $1,000, sM = $1,000, sH = $1,000
❖ Holding cost, h = 0.2
❖ Unit cost, CL = $500, CM = $500, CH = $500

International University
Vietnam National University HCMC
 Example 3 (2/3)
Step 1: We obtain

Step 2: We first obtain:

Step 3: We have:

Step 4: Recalculate the ordering frequency of the most frequently ordered model
as:

Step 5: Ordering frequency for each product will be:

International University
Vietnam National University HCMC
 Example 3 (3/3)

International University
Vietnam National University HCMC
Impact of Supply Uncertainty on Safety Inventory
(1/2)
❖ Supply uncertainty arises because of many factors:
• Production delays
• Transportation delays
• Quality problems
❖ Supply uncertainty → lead time uncertainty → Identify the
impact of lead time uncertainty on safety inventories.
❖ Assumption:
• Customer Demand per period → normal distribution
• Replenishment lead time from supplier → normal distribution
❖ Inputs:
D: average demand per period
𝐷 : standard deviation of demand per period
L: average lead time for replenishment
𝑆𝐿 : standard deviation of lead time
International University
Vietnam National University HCMC
Impact of Supply Uncertainty on Safety Inventory
(2/2)
❖ If demand during the lead time > ROP

→ Thus, we need to identify the distribution of customer

demand during the uncertain lead time.
❖ Assume demand during the lead time is normally
distributed with:
𝐷𝐿 = 𝐷 ∗ 𝐿
𝜎𝐿 = 𝐿𝜎𝐷2 + 𝐷2 𝑠𝐿2

❖ With a given desired CSL, we have: Cycle service level: Replenishment cycle

𝑠𝑠 = 𝑁𝑂𝑅𝑀𝑆𝐼𝑁𝑉 𝐶𝑆𝐿 ∗ 𝜎𝐿
International University
Vietnam National University HCMC
 Example 4 (1/3)

• Daily demand for tablets at Amazon is normally distributed,

with a mean of 2,500 and a standard deviation of 500. The
tablet supplier takes an average of L = 7 days to replenish
inventory at Amazon.
• Amazon is targeting a CSL of 90 percent (providing a fill
rate close to 100 percent) for its tablet inventory. Evaluate
the safety inventory of tablets that Amazon must carry if the
standard deviation of the lead time is seven days. Amazon
is working with the supplier to reduce the standard
deviation to zero.
→ Evaluate the reduction in safety inventory that Amazon can
expect as a result of this initiative.

International University
Vietnam National University HCMC
 Example 4 (2/3)

International University
Vietnam National University HCMC
 Example 4 (2/3)

International University
Vietnam National University HCMC
 2.
IMPACT OF AGGREGATION ON
SAFETY INVENTORY

International University
Vietnam National University HCMC
 Aggregation on Safety Inventory

❖ Physical Aggregation:
• Inventory Centralization
❖ Non-physical Aggregations:
• Information Centralization
• Specialization
• Product Substitution
• Component Commonality
• Postponement
International University
Vietnam National University HCMC
 How aggregation affects forecast accuracy

❖ Consider 𝑘 regions, with demand in each region normally

distributed with the inputs:
𝐷𝑖 : mean periodic demand in region 𝑖, 𝑖 = 1, … , 𝑘
𝜎𝑖 : standard deviation of periodic demand in region 𝑖, 𝑖 = 1, … , 𝑘
𝜌𝑖𝑗 : correlation of periodic demand in region 𝑖, 𝑗, 1 ≤ 𝑖 ≠ 𝑗 ≤ 𝑘
L: replenishment lead time
CSL: desired cycle service level
❖ There are two ways to serve demand in the 𝑘 regions.
• Case 1: have local inventories in each region

• Case 2: Aggregate all inventories into one centralized facility.

→ Our goal is to compare safety inventories in the two cases.

International University
Vietnam National University HCMC
 Case 1: Inventory Decentralization

❖ Total safety inventory in decentralized option calculate for the 1st case

𝑠𝑠 = σ𝑘𝑖=1 𝐹𝑆−1 (𝐶𝑆𝐿) x 𝜎𝑖 x 𝐿

International University
Vietnam National University HCMC
 Case 2: Inventory Centralization (1/3)
All regions demand is
Aggregate demand is All regions demand is
independent and
normally distributed identically distributed
identically distributed
𝑘
Mean
𝐷𝐶 = ෍ 𝐷𝑖 𝐷𝐶 = 𝑘𝐷 𝐷𝐶 = 𝑘𝐷
𝑫𝑪
𝑖=1

𝜎𝐷𝐶 = 𝜎𝐷𝐶 = 𝑘𝜎𝐷

Standard 𝜎𝐷𝐶 =
𝑘
deviation
෍ 𝜎𝑖2 + 2 ෍ 𝜌𝑖𝑗 𝜎𝑖 𝜎𝑗 𝑘 𝜎𝐷2 + 𝑘 𝑘 − 1 𝜎𝐷2 𝜌 (𝜌 = 0)
𝝈𝑪𝑫
𝑖=1 𝑖>𝑗

❖ Total safety inventory on aggregation:

𝑠𝑠 = 𝐹𝑆−1 𝐶𝑆𝐿 x 𝜎𝐷𝐶 x 𝐿
❖ Holding cost savings on aggregation per unit sold:
𝑘
𝐹𝑆−1 𝐶𝑆𝐿 𝐻 𝐿 𝐶
× ෍ 𝜎𝑖 − 𝜎𝐷
International University 𝐷𝐶
Vietnam National University HCMC 𝑖=1
 Case 2: Centralized Inventory - Insights (2/3)

❖ Conclusions regarding the value of aggregation:

• The safety inventory savings on aggregation increase with the
desired cycle service level CSL.
• The safety inventory savings on aggregation increase with the
replenishment lead time L.
• The safety inventory savings on aggregation increase with the
holding cost H.
• The safety inventory savings on aggregation increase with the
coefficient of variation (𝝈𝑫 /D) of demand.
• The safety inventory savings on aggregation decrease as the
correlation coefficients 𝝆𝒊𝒋 increase.

International University
Vietnam National University HCMC
 Case 2: Centralized Inventory - Insights (3/3)

❖ There are two major disadvantages of aggregating all

inventories in one location:

• Increase in response time to customer order.

• Increase in transportation cost to customer.

❖ This is because the average distance between the inventory

and the customer increases with aggregation.

• Either the customer must travel father to reach the product

• Or the product must be shipped over longer distances to reach the customer

International University
Vietnam National University HCMC
Example 5 - Impact of Correlation on Aggregation
(1/3)
A BMW dealership has k = 4 retail outlets serving the entire
Chicago area (disaggregate option). Weekly demand at each
outlet is normally distributed, with a mean of D = 25 cars and a
standard deviation of D = 5. The lead time for replenishment
from the manufacturer is L = 2 weeks. Each outlet covers a
separate geographic area, and the correlation of demand across
any pair of areas is 𝜌 . The dealership is considering the
possibility of replacing the four outlets with a single large outlet
(aggregate option). Assume that the demand in the central outlet
is the sum of the demand across all four areas. The dealership is
targeting a CSL of 0.90.

→ Compare the level of safety inventory needed in the two

options as the correlation coefficient 𝜌 varies between 0 and 1.
International University
Vietnam National University HCMC
 Example 5 (2/3)
❖ We have:
• Average demand per period, D = 25 cars
• Standard deviation of weekly demand, D = 5
• Lead time for replenishment, L = 2 weeks
• 𝑘 = 4, 𝐶𝑆𝐿 = 0.9
❖ First, we analyse for the case when demand in each area is
independent (i.e., 𝝆 = 𝟎)
• The required safety inventory in decentralized option:

• The required safety inventory in centralized option:

International University
Vietnam National University HCMC
 Example 5 (3/3)

International University
Vietnam National University HCMC
Example 6 - Trade-Offs of Physical Centralization
(1/3)
An online retailer is debating whether to serve the United States through
four regional distribution centers or one national distribution center.
Weekly demand in each region is normally distributed, with a mean of
1,000 and a standard deviation of 300. Demand experienced in each
region is independent, and supply lead time is four weeks. The online
retailer has a holding cost of 20 percent and the cost of each product is
$1,000. The retailer promises its customers next-day delivery. With four
regional distribution centers, the retailer can provide next-day delivery
using ground transportation at a cost of $10/unit. With a single national
distribution center, the retailer will have to use a more expensive mode
of transport that will cost $13/unit for next-day service. Building and
operating four regional DCs costs $150,000 per year more than building
and operating one national distribution center.
→ What distribution network do you recommend? Assume a desired
CSL of 0.95.
International University
Vietnam National University HCMC
 Example 6 (2/3)
❖ Each region, we have:
• Average demand per period, D = 1,000pcs/week
• Standard deviation of weekly demand, D = 300
• Lead time for replenishment, L = 4 weeks
• 𝑘 = 4, 𝐶𝑆𝐿 = 0.95
❖ The required safety inventory in across all four regional
distribution centers:

❖ Because demand in all four areas is independent, 𝝆 = 𝟎. The

required safety inventory in centralized option:

International University
Vietnam National University HCMC
 Example 6 (3/3)

❖ We can now evaluate the effects of the changes in inventory,

transportation, and facility costs on aggregation as follows:
• Decrease in annual inventory holding cost on aggregation

• Decrease in annual facility costs on aggregation

• Increase in annual transportation costs on aggregation

• The annual costs for the online retailer increase by

→ The online retailer is better off with four regional distribution

centers.
International University
Vietnam National University HCMC
 Aggregation on Safety Inventory

❖ Physical Aggregation:
• Inventory Centralization
❖ Non-physical Aggregations:
• Information Centralization
• Specialization
• Product Substitution
• Component Commonality
• Postponement
International University
Vietnam National University HCMC
 Information Centralization
❖ Retailers such as Gap use information centralization
effectively.
→ If a store does not have the size or color that a customer
wants, store employees can use their information system to
inform the customer of the closest store with the product in
inventory → Customers can then either go to this store or have
the product delivered to their house.
❖ Gap thus uses information centralization to virtually
aggregate inventory across all retail stores even though the
inventory is physically separated.
→ This allows Gap to …..
reduce the amount of safety inventory --> high level of product availability

International University
Vietnam National University HCMC
 Specialization

❖ Specialization of inventory based on Product Type: The higher

the coefficient of variation of an item, the greater is the
reduction in safety inventories as result of centralization
• Items with low demand (slow-moving items) with high coefficient
of variation → stocked at a centralized location → reduce the safety
inventory without hurting customer response time or adding to
transportation costs.
• Items with high demand (fast-moving items) with low coefficient
of variation → stocked at decentralized locations close to the customer.

International University
Vietnam National University HCMC
Example 7 - Impact of Coefficient of Variation on
Aggregation (1/2)
Assume that W.W. Grainger, a supplier of MRO products, has 1,600
stores distributed throughout the United States. Consider two products -
large electric motors and industrial cleaner. Large electric motors are
high-value items with low demand, whereas the industrial cleaner is a
low-value item with high demand. Each motor costs $500 and each can of
cleaner costs $30. Weekly demand for motors at each store is normally
distributed, with a mean of 20 and a standard deviation of 40. Weekly
demand for cleaner at each store is normally distributed, with a mean of
1,000 and a standard deviation of 100. Demand experienced by each
store is independent, and supply lead time for both motors and cleaner is
four weeks. W.W. Grainger has a holding cost of 25 percent.
→ For each of the two products, evaluate the reduction in safety
inventories that will result if they are removed from retail stores and
carried only in a centralized DC. Assume a desired CSL of 0.95.
International University
Vietnam National University HCMC
 Example 7 (2/2)

International University
Vietnam National University HCMC
 Product Substitution

❖ Substitution → the use of one product to satisfy demand for a

different product. Substitution may occur in two situations:
• Manufacturer-driven one-way substitution: The
Manufacturer/Supplier makes the decision to substitute.
Typically, the manufacturer substitutes a higher-value product
for a lower-value product that is not in inventory.
→ Increases profitability for the manufacturer by allowing aggregation
of demand → reduces the inventory requirements for the same level of
availability.
• Customer-driven two-way substitution: Customers make the
decision to substitute.
→ Recognizes of customer-driven substitution and joint management
of inventories across substitutable products → allows to….
reduce the required inventory while ensuring high level of product availability

International University
Vietnam National University HCMC
 Component Commonality
❖ Given the large number of components in each finished
product → demand uncertainty will be high → resulting in
high levels of safety inventory.
❖ When products with common components are designed →
the demand for each component is an aggregation of the
demand for all the finished products using the component.
❖ Component demand is thus more predictable → reduces
the component inventories carried in the supply chain.
❖ With increasing product variety, component commonality is
a key to ….
reducing supply chain inventories required without hurting product availability

→ a key factor for success in the electronics industry.

International University
Vietnam National University HCMC
Example 8 - Value of Component Commonality (1/3)

Assume that Dell is to manufacture 27 servers with three distinct

components: processor, memory, and hard drive. Under the disaggregate
option, Dell designs specific components for each server, resulting in 3 * 27 =
81 distinct components. Under the common-component option, Dell designs
servers such that three distinct processors, three distinct memory units, and
three distinct hard drives can be combined to create 27 servers. Each
component is thus used in nine servers. Monthly demand for each of the 27
servers is independent and normally distributed, with a mean of 5,000 and a
standard deviation of 3,000. The replenishment lead time for each
component is one month. Dell is targeting a CSL of 95 percent for component
inventory.
→ Evaluate the safety inventory requirements with and without the use of
component commonality. Also evaluate the change in safety inventory
requirements as the number of finished products of which a component is a
partInternational
varies from one to nine.
University
Vietnam National University HCMC
 Example 8 (2/3)

International University
Vietnam National University HCMC
 Example 8 (3/3)

International University
Vietnam National University HCMC
 Postponement
❖ Postponement is the ability of a supply chain to delay
product differentiation or customization until closer to the
time the product is sold.
Ex: the final mixing of paint is done at the retail store after the customer
has selected the color he or she wants → Thus, paint variety is produced
only when demand is known with certainty. Postponement allows paint
retailers to carry significantly low safety inventories.
❖ Without component commonality postponement, product
differentiation occurs → most of the supply chain inventories
are disaggregate.
❖ Postponement allows to delay product differentiation → most of
the inventories in the supply chain are aggregate.
❖ Postponement thus allows a supply chain to ….reduce safety inventories
without hurting product
availability

International University
Vietnam National University HCMC
 Example 9 - Value of Postponement (1/2)

Consider a paint retailer that sells 100 different colors of paint. Assume
that weekly demand for each color is independent and is normally
distributed with a mean of 30 and a standard deviation of 10. The
replenishment lead time from the paint factory is two weeks and the
retailer aims for a CSL = 0.95.
→ How much safety stock will the retailer have to hold if paint is mixed
at the factory and held in inventory at the retailer as individual colors?
→ How does the safety stock requirement change if the retailer holds
base paint (supplied by the paint factory) and mixes colors on
demand?

International University
Vietnam National University HCMC
 Example 9 (2/2)

International University
Vietnam National University HCMC