Professional Documents
Culture Documents
management and
risk pooling
Lu Chen
Dept. of Industrial Engineering &
Logistics Management
Contents
q Introduction
q Economic order quantity model
q Inventory control techniques
q Service Level Optimization
q Risk pooling
q Managing inventory in the supply chain
2
Why Is Inventory Required?
q Delivery lead times
q Incentives for larger shipments --- economies of scale
q Uncertainty in customer demand
q Shorter product lifecycles
q More competing products
q Uncertainty in supplies
q Quality/Quantity/Costs/Delivery Times
3
Cycle inventory
q Cycle inventory is the average inventory in a supply
chain due to either production or purchases in lot
sizes that are larger than the demand (D) from the
customer.
q Lot or batch size (Q) is the quantity that a stage of a
supply chain either produces or purchases at a time.
lot size Q
Cycle inventory = =
2 2
cycle inventory Q
Average flow time = =
demand 2D
4
Supply chain costs influenced by lot
size
q Fixed ordering cost: all costs that do not vary with
the size of the order but are incurred each time an
order is placed.
q Holding cost: the cost of carrying one unit in
inventory for a specified period of time.
q The cost of capital
q The cost of physically storing the inventory
q The cost that results from the product becoming obsolete
5
Contents
q Introduction
q Economic order quantity model
q Inventory control techniques
q Service Level Optimization
q Risk pooling
q Managing inventory in the supply chain
6
Economic order quantity
q Introduced by Ford W. Harris in 1915.
q A simple model that illustrates the trade-offs
between ordering and storage costs.
q Consider a warehouse facing constant demand for a
single item. The warehouse orders from the supplier,
who is assumed to have an unlimited quantity of the
product.
q The model assumes the following:
7
Assumptions
q D items per year, constant demand rate
q Q items per order
q K, fixed setup cost, incurred every time the
warehouse places an order.
q h, inventory holding cost per unit per year
q Lead time = 0
q Initial inventory = 0
8
Economic order quantity
10
Economic order quantity (EOQ)
2 KD
Q =
*
11
EOQ example
Consider a computer ordering case:
Annual demand, D = 1,000 x 12 = 12,000 units
Order cost per lot, K = $4,000
Unit cost per computer, C = $500
Holding cost per year as a fraction of unit cost, h = 0.2
12
EOQ example - answer
2 ×12, 000 × 4, 000
Optimal order size = Q* = = 980
0.2 × 500
Q * 980
Cycle inventory = = = 490
2 2
D
Number of orders per year = = 12.24
Q*
D ⎛Q *⎞
Annual ordering and holding cost = K +⎜ ⎟ hC = 97, 980
Q* ⎝ 2 ⎠
Q* 490
Average flow time = = = 0.041 = 0.49 month
2D 12, 000
13
EOQ example - discussion
q If demand grows to 2,000 per month, how would you
expect the order to change? And how does the total
cost change?
14
Sensitivity Analysis
Total inventory cost relatively insensitive to order quantities
TC (Q) 1 Q Q *
= ( + )
TC (Q*) 2 Q * Q
15
Key observations
q Total cost is relatively insensitive to changes
q Very robust with respect to changes in:
n Q – rounding of order quantities
n D – errors in forecasting
16
EOQ--- in production
q p: production rate
q d: demand rate
q K: setup time for production
q What is the optimal production lot size Q?
inventory
Q1
time
17
EOQ with quantity discounts
q Volume based discount: the discount is based on the
total quantity purchased over a given period,
regardless of the number of lots purchased over that
period.
q Lot size based discount: the pricing schedule offers
discounts based on the quantity ordered in a single
lot.
q All unit quantity discounts
q Marginal unit quantity discount
Questions:
1. How does this decision affect the supply chain in terms
of lot sizes ?
2. Under what conditions should a supplier offer quantity
discounts? 18
All-Unit Quantity Discounts
19
Calculate the lot size to minimize the
total cost
q Step 1: Evaluate the optimal lot size for each price Ci
using EOQ model
q Step 2: Select the order quantity Q*i for each price Ci
1. qi ≤ Qi < qi+1 Qi* = Qi
2. Qi < qi Qi* = qi
3. Qi ≥ qi+1
21
Example solution
Step 1:
2DS 2DS 2DS
Q0 = = 6,324; Q1 = = 6,367; Q2 = = 6,410
hC0 hC1 hC2
Step 2:
Q1* = Q1 = 6,367; Q2* = q2 =10, 000
Step 3:
TC1 = $358, 969
TC2 = $354, 520
Step 4:
Order 10,000 bottles per lot at $2.92 per bottle
22
Discussion
q What is the impact of the quantity discount on the
average inventory and flow time in a supply chain?
23
If there is no discount
D = 120,000 bottles/year, SR = $100, hR = 0.2, CR = $3
SM = $250, hM = 0.2, CM = $2
!D$ !Q $
Annual cost for DO = # & S R + # R & hRC R = $3,795
" QR % " 2 %
!D$ !Q $
Annual cost for manufacturer = # & S M + # R & hM C M = $6,009
" QR % " 2 %
25
Optimal lot size
Annual cost for DO ! D$ !Q $ ! D$ !Q $
= # & S R + # & hRC R + # & S M + # & hM C M
and manufacturer "Q% "2% "Q% "2%
2D(S R + S M )
Q* = = 9,165
hRC R + hM C M
! D$ !Q *$
Annual cost for DO = # & SR + # & hRC R = $4,059
"Q *% " 2 %
! D$ !Q *$
Annual cost for manufacturer = # & SM + # & hM C M = $5,106
"Q *% " 2 %
27
The effect of demand uncertainty
q The EOQ model ignores demand uncertainty and
forecasting.
q Many companies treat the world as if it were
predictable, and design the planning processes as if
the initial forecast was an accurate representation of
reality.
q Principles of forecasts
q The forecast is always wrong.
q The longer the forecast horizon, the worse the forecast.
q Aggregate forecasts are more accurate.
28
Assumptions: Basic EOQ Model
q Demand q Discounts
Constant vs Variable None
Known vs Random All Units or Incremental
Continuous vs Discrete q Excess Demand
q Lead time None
Instantaneous All orders are backordered
Constant or Variable Lost orders
(deterministic/stochastic) Substitution
q Dependence of items q Perishability
Independent None
Correlated Uniform with time
Indentured q Planning Horizon
q Review Time Single Period
Continuous vs Periodic Finite Period
q Number of Echelons Infinite
One vs Many q Number of Items
q Capacity / Resources One
Unlimited vs Limited Many
29
Contents
q Introduction
q Economic order quantity model
q Inventory control techniques
q Service Level Optimization
q Risk pooling
q Managing inventory in the supply chain
30
Measuring product availability
q Service level (SL): the probability of not having a
stockout in a replenishment cycle.
31
A single period model
q Consider a company that designs, produces, and sells
swimsuits.
q About 6 months before summer, the company must
commit itself to specific production quantities for its
products.
q Order quantity to be decided before demand occurs.
q Order Quantity > Demand => Dispose excess
inventory
q Order Quantity < Demand => Lose sales/profits
32
Probabilistic forecast
33
Additional Information
q Fixed production cost F = $100,000
q Variable production cost per unit c = $80.
q During the summer season, selling price p = $125 per
unit.
q Salvage value: Any swimsuit not sold during the
summer season is sold to a discount store s = $20.
34
Two Scenarios
q Manufacturer produces 12,000 units while demand
ends at 10,000 swimsuits
Profit
= 125(10,000) + 20(2,000) - 80(12,000) - 100,000
= $230,000
35
If the production quantity is 12,000,
what is the total profit?
Probability Demand Profit
0.11 8,000 Q>D
0.11 10,000 Q>D
0.28 12,000 D=Q
0.22 14,000 D>Q
0.18 16,000 D>Q
0.10 18,000 D>Q
Average weighted profit
SL* = (p – c) / (p – s)
37
Optimal service level
q SL* = 43%
38
Production Quantity that Maximizes
Expected Profit
40
Risk-Reward Tradeoffs
q Optimal production quantity maximizes average
profit is about 12,000
q Producing 9,000 units or producing 16,000 units will
lead to about the same average profit of $294,000.
q If we had to choose between producing 9,000 units
and 16,000 units, which one should we choose?
41
Profit @ 9,000 v.s. profit @ 16,000
42
Profit @ 9,000 v.s. profit @ 16,000
9,000
Probability Demand Profit
0.11 8,000 200,000
0.11 10,000 305,000
0.28 12,000 305,000
0.22 14,000 305,000
0.18 16,000 305,000
0.10 18,000 305,000
Average 293,500 16,000
Probability Demand Profit
0.11 8,000 -220,000
0.11 10,000 -10,000
0.28 12,000 200,000
0.22 14,000 410,000
0.18 16,000 620,000
0.10 18,000 620,000
Average 294,500
43
Risk-Reward Tradeoffs
44
Observations
q The optimal order quantity is not necessarily equal
to average demand.
q Risk/Reward trade-off: As we increase the
production quantity, both risk and reward increases.
q As the order quantity increases, average profit
typically increases until a certain value, then
decreases.
45
Newsvendor model
The density function of the daily demand (r) is p(r).
46
Daily expected profit
n: daily order quantity
G(n): daily expected profit
n ¥
G(n) = ò [(a - b)r - (b - c)(n - r )] p(r )dr + ò (a - b)np (r )dr
0 n
n
dG
=0
ò0
p (r )dr
=
a -b
¥
dn
ò p (r )dr b - c
n
47
Observation
n ¥
ò p(r )dr = P , ò p(r )dr = P
0 1 n 2
P1 a - b
=
P2 b - c
Marginal profit/piece: a-b
Marginal cost/piece: b-c
(a - b) Þ n , (b - c) Þ n ¯
48
Discussion
q What If the Manufacturer Has an Initial Inventory?
q Trade-off between:
q Using on-hand inventory to meet demand and avoid paying
fixed production cost: need sufficient inventory stock.
q Not have as much inventory, need paying the fixed cost of
production.
49
Profit and the impact of initial
inventory
50
If the initial inventory is 7,000, and no
production
500000
400000
300000
Profit
200000
100000
0
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
15000
16000
Production Quantity
51
If the initial inventory is 7,000, we
produce to increase inventory to 12,000
500000
400000
300000
Profit
200000
100000
0
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
15000
16000
Production Quantity
52
If the initial inventory is 8,500, we
should not produce anything
500000
400000
300000
Profit
200000
100000
0
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
15000
16000
Production Quantity
53
Multiple order opportunities
q What are the objectives?
54
Two policies
q Continuous review policy
q inventory is reviewed continuously
q an order is placed when the inventory reaches a particular
level or reorder point.
q inventory is continuously reviewed
55
Continuous review policy
q D = Average daily demand faced by the distributor
q ! = Standard deviation of the daily demand
q L = Replenishment lead time from the supplier (days)
q h = Unit holding cost
q α = service level
Decisions to make:
q Whenever inventory level falls to a reorder level R, place an
order for Q units.
q What are the values of R and Q?
56
Inventory level over time
57
Demand in probability
58
Evaluating safety inventory given a
desired SL
Desired service level = α
Mean demand during lead time = DL
Standard deviation of demand during lead time = σL
safety stock = + ,- α, /0 , 10 − /0
59
Service Level & Safety Factor, z
Service 90% 91% 92% 93% 94% 95% 96% 97% 98% 99% 99.9%
Level
z 1.29 1.34 1.41 1.48 1.56 1.65 1.75 1.88 2.05 2.33 3.08
60
Continuous Review Policy
q Average demand during lead time: !"
(
+#$ %$&
2
61
Continuous Review Policy
Example
q Wal-Mart orders Lego from a DC, weekly demand for Lego at
the store is normally distributed with a mean of 2,500 boxes and
a standard deviation of 500
q Fixed ordering cost = $5,000
q Cost of a box of Lego = $250
q Annual inventory holding cost = 10% of product cost
q Replenishment lead time = 2 weeks
q Expected service level = 90% (z=1.29)
62
Continuous Review Policy
Example
2 × 5000 × 2500
Optimal order quantity Q = = 7212
25 / 52
63
Evaluate fill rate
q Expected shortage per replenishment cycle (ESC) is
the average units of demand that are not satisfied
from inventory in stock per replenishment cycle
q Product fill rate
f = 1 – ESC/Q
' ! ss $* ! ss $
ESC = –ss )1– Fs # &, + σ L f s # &
)( " σ L %,+ "σ L %
ESC = 30
f = 1 – 30/7212 = 0.996
64
Base-Stock Level Policy
q Determine a target inventory level, the base-stock level.
q Each review period, review the inventory position is
reviewed and order enough to raise the inventory
position to the base-stock level.
q Assume:
r = length of the review period
L = lead time
D = average daily demand
! = standard deviation of this daily demand
65
Base-Stock Level Policy
66
Base-Stock Level Policy
q Average demand during an r + L days = !)*+ = ! " ' + #
!"#
+&" '+#"(
2
67
Example – continued
q Assume:
q Review interval r = 4 weeks
q Lead time is 2 weeks
q Base-stock level needs to cover 6 weeks
q Calculate the base-stock level and average inventory
level
69
Exercise
q Weekly demand for HP printers at a retail store is normally
distributed, with a mean of 250 and a standard deviation of 150.
The store manager uses continuous inventory control policy
and orders 1000 printers each time the inventory drops to 600
printers. HP takes two weeks to fill an order.
q Question:
(1) How much safety inventory does the retail store carry?
(2) What service level does the store have?
(3) What is the fill rate?
70
Contents
q Introduction
q Economic order quantity model
q Inventory control techniques
q Service Level Optimization
q Risk pooling
q Managing inventory in the supply chain
71
Optimal SL with one-time order
Expected benefit of producing extra unit =
Expected cost of producing extra unit
(1 – SL*) (p – c) = SL*(c – s)
SL* = (p – c) / (p – s)
72
Two extreme scenarios
q All demand arising when the product is out of stock
is lost.
q All demand that arises when the product is out stock
is backlogged and filled later, when inventories are
replenished.
73
Desired service level
q Q: replenishment lot size
q D: average demand per unit time
q SL: service level
q Cu: profit for selling one unit of product/cost of
under stocking by one unit
q H: unit holding cost for one unit of time
74
Demand during stockout is lost
q Increased cost per replenishment cycle of additional
safety inventory of 1 unit
Q
SL * H
D
q Benefit per replenishment cycle of additional safety
inventory of 1 unit
(1− SL*)Cu
HQ
SL* = 1−
HQ + DCu
75
Demand during stockout is
backlogged
q Increased cost per replenishment cycle of additional
safety inventory of 1 unit
Q
H
D
q Benefit per replenishment cycle of additional safety
inventory of 1 unit
(1− SL*)Cu
HQ
SL* = 1−
DCu
76
Discussion
q What are the strategies to decrease the
understocking cost?
77
Service level optimization by SKU
78
Contents
q Introduction
q Economic order quantity model
q Inventory control techniques
q Service Level Optimization
q Risk pooling
q Managing inventory in the supply chain
79
Risk Pooling
q Demand variability is reduced if one aggregates
demand across locations.
q More likely that high demand from one customer
will be offset by low demand from another.
q Reduction in variability allows a decrease in safety
stock and therefore reduces average inventory.
80
Acme Risk Pooling Case
q Electronic equipment manufacturer and distributor.
q 2 warehouses for distribution in New York and New
Jersey (partitioning the northeast market into two
regions).
q Each retailer is assigned to a warehouse.
q Warehouses receive material from Chicago, lead time
for delivery to each of the warehouses is one week.
q Service level: 97%
81
Historical Data
PRODUCT A
Week 1 2 3 4 5 6 7 8
New York 33 45 37 38 55 30 18 58
New Jersey 46 35 41 40 26 48 18 55
Total 79 80 78 78 81 78 36 113
PRODUCT B
Week 1 2 3 4 5 6 7 8
New York 0 3 3 0 0 1 3 0
New Jersey 2 4 3 0 3 1 0 0
Total 2 6 3 0 3 2 3 0
82
Summary of Historical Data
Statistics Product Average Standard Coefficient of
Demand Deviation of Variation
Demand
New York A 39.3 13.2 0.34
83
Inventory Levels
Product Average Safety Reorder Q Average
Demand Stock Point inventory
2 warehouses
B 2.375 3.61 6 33 20
84
Critical Points
q The higher the coefficient of variation, the greater the
benefit from risk pooling.
q The higher the variability, the higher the safety stocks kept by
the warehouses. The variability of the demand aggregated by
the single warehouse is lower.
q The benefits from risk pooling depend on the
behavior of the demand from one market relative to
demand from another.
q risk pooling benefits are higher in situations where demands
observed at warehouses are negatively correlated.
85
Impact of aggregation on safety
inventory
Di: Mean weekly demand in region i, i = 1,…, k
si: Standard deviation of weekly demand in region
i, i = 1,…, k
&
Total safety inventory
= " '× )×*#
in decentralized option
#$%
86
Distribution of aggregated demand
rij: Correlation of weekly demand for regions i and j,
1≤i≠j≤k
( ( (
DC = kD σ DC = k σ D
87
An example
A BMW dealership has four retail outlets serving the
entire Chicago area.
Mean value of weekly demand, D = 25
Standard deviation of weekly demand, sD = 5
Replenishment, L = 2 weeks
Decentralized SL = 0.9.
88
Solution
Total required
safety inventory = 4×$%&' 0.9 × 2×5 = 36.24
Disaggregate Aggregate
r Safety Inventory Safety Inventory
0 36.24 18.12
0.2 36.24 22.92
0.4 36.24 26.88
0.6 36.24 30.32
0.8 36.24 33.41
1.0 36.24 36.24
89
Square root law
Total safety
Inventory
Number of independent
Stocking locations
91
Contents
q Introduction
q Single stage inventory control
q Supply contract
q Risk pooling
q Managing inventory in the supply chain
92
Managing inventory in the supply
chain
q Inventory decisions are given by a single decision
maker whose objective is to minimize the system-
wide cost.
q The decision maker has access to inventory
information at each of the retailers and at the
warehouse.
q Echelons and echelon inventory:
q Echelon inventory at any stage or level of the system equals
the inventory on hand at the echelon, plus all downstream
inventory (downstream means closer to the customer)
93
Echelon Inventory
94
Reorder point with echelon inventory
q Le = echelon lead time,
q lead time between the retailer and the distributor plus the
lead time between the distributor and its supplier, the
wholesaler.
q D = average demand at the retailer
q s = standard deviation of demand at the retailer
q Reorder point
95
More than one facility at each stage
96
Inventory positioning
q Consider a two-tier supply chain
q Items shipped from manufacturing facilities to primary
warehouses
q From there, they are shipped to secondary warehouses and
finally to retail outlets
q How to optimally position inventory in the supply
chain?
q Should every product be positioned both at the primary and
secondary warehouses?, OR
q Some product be positioned only at the primary while
others only at the secondary?
97
Integrating Inventory Positioning and
Network Design
98
Supply chain strategy for different
categories
q High variability low volume products
q Inventory risk is the main challenge
q Position them mainly at the primary warehouses
n demand from many retail outlets can be aggregated reducing
inventory costs.
q Low variability high volume products
q Position close to the retail outlets at the secondary
warehouses
q Ship fully loaded tracks as close as possible to the customers
reducing transportation costs.
q Low variability low volume products
q Require more analysis since other characteristics are
important, such as profit margins, etc.
99
Single product single facility
q Assume -
q SI: amount of time between when an order is placed until
the facility receives a shipment (Incoming Service Time)
q S: Committed Service Time made by the facility to its own
customers.
q T: Processing Time at the facility.
q SI + T > S
q Net Lead Time = SI + T - S
q Safety stock at the facility: ! "# + % − "'
100
2-stage system
SI2 S2=SI1 S1
Facility 2 Facility 1
101
ElecComp case
q Large contract manufacturer of circuit boards and
other high tech parts.
q About 27,000 high value products with short life
cycles
q Fierce competition =>
q Customer promise times < Manufacturing Lead Times
q High inventory of SKUs based on long-term forecasts
q High shortages
q Huge risk
102
New supply chain strategy
q Objectives:
q Reduce inventory and financial risks
q Provide customers with competitive response times.
q Achieve the following:
q Determining the optimal location of inventory across the various
stages
q Calculating the optimal quantity of safety stock for each
component at each stage
q Hybrid strategy
q Some stages produce to stock where the company keeps safety
stock
q Some stages keep no stock at all
103
Current safety stock location
104
Optimized safety stock location
105
Optimized supply chain with
more complex product structure
106
Key Points
q Identifying the appropriate strategy for each facility
q Taking advantage of the risk pooling concept
q Demand for components used by a number of finished
products has smaller variability and uncertainty than that
of the finished goods.
q Replacing traditional supply chain strategies that
are typically referred to as sequential, or local,
optimization by a globally optimized supply chain
strategy.
107