The EOQ Model

To a pessimist, the glass is half empty.

to an optimist, it is half full.

– Anonymous

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 1

 EOQ History

• Introduced in 1913 by Ford W. Harris, “How Many

Parts to Make at Once”

• Interest on capital tied up in wages, material and

overhead sets a maximum limit to the quantity of parts
which can be profitably manufactured at one time;
“set-up” costs on the job fix the minimum. Experience
has shown one manager a way to determine the
economical size of lots.

• Early application of mathematical modeling to

Scientific Management

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 2

 MedEquip Example

• Small manufacturer of medical diagnostic equipment.

• Purchases standard steel “racks” into which components are
mounted.
• Metal working shop can produce (and sell) racks more cheaply if
they are produced in batches due to wasted time setting up shop.
• MedEquip doesn’t want to tie up too much precious capital in
inventory.

• Question: how many racks should MedEquip order at once?

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 3

 EOQ Modeling Assumptions
1. Production is instantaneous – there is no capacity constraint and
the entire lot is produced simultaneously.
2. Delivery is immediate – there is no time lag between production
and availability to satisfy demand.
3. Demand is deterministic – there is no uncertainty about the
quantity or timing of demand.
4. Demand is constant over time – in fact, it can be represented as a
straight line, so that if annual demand is 365 units this translates
into a daily demand of one unit.
5. A production run incurs a fixed setup cost – regardless of the size
of the lot or the status of the factory, the setup cost is constant.
6. Products can be analyzed singly – either there is only a single
product or conditions exist that ensure separability of products.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 4
 Notation

D demand rate (units per year)

c unit production cost, not counting setup or inventory costs

(dollars per unit)

A fixed or setup cost to place an order (dollars)

h holding cost (dollars per year); if the holding cost is consists

entirely of interest on money tied up in inventory, then h = ic
where i is an annual interest rate.

Q the unknown size of the order or lot decision variable

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 5
 Inventory vs Time in EOQ Model

Inventory

Time
Q/D 2Q/D 3Q/D 4Q/D

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 6

 Costs

Holding Cost:
Q
average inventory =
2
hQ
annual holding cost =
2
hQ
unit holding cost =
2D
Setup Costs: A per lot, so
A
unit setup cost =
Q

Production Cost: c per unit

hQ A
Cost Function: Y (Q) = + +c
2D Q
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 7
 MedEquip Example Costs

D = 1000 racks per year

c = $250

A = $500 (estimated from supplier’s pricing)

h = (0.1)($250) + $10 = $35 per unit per year

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 8

 Costs in EOQ Model
20.00

18.00

16.00

14.00
Cost ($/unit)

12.00

10.00 Y(Q)
Q* =169
8.00

6.00 hQ/2D

4.00

2.00 c A/Q

0.00
0 100 200 300 400 500

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 Order Quantity (Q)

http://www.factory-physics.com 9
 Economic Order Quantity

dY (Q) h A 2 AD 2 AD
0= = − 2 ⇒ Q = 2
⇒ Q=
dQ 2D Q h h
d 2Y (Q) 2 A
∀Q ≥ 0, 2
= 3 ≥0 ⇒ Q* is a minimum.
dQ Q
2 AD EOQ Square
Q =*

h Root Formula

2(500)(1000)
Q = *
= 169 MedEquip Solution
35
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 10
 EOQ Modeling Assumptions
1. Production is instantaneous – there is no capacity relax via
constraint and the entire lot is produced simultaneously. EPL model
2. Delivery is immediate – there is no time lag between production and
availability to satisfy demand.
3. Demand is deterministic – there is no uncertainty about the quantity
or timing of demand.
4. Demand is constant over time – in fact, it can be represented as a
straight line, so that if annual demand is 365 units this translates into a
daily demand of one unit.
5. A production run incurs a fixed setup cost – regardless of the size of
the lot or the status of the factory, the setup cost is constant.
6. Products can be analyzed singly – either there is only a single
product or conditions exist that ensure separability of products.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 11
 Notation – EPL Model
D demand rate (units per year)

P production rate (units per year), where P>D

c unit production cost, not counting setup or inventory costs

(dollars per unit)

A fixed or setup cost to place an order (dollars)

h holding cost (dollars per year); if the holding cost is consists

entirely of interest on money tied up in inventory, then h = ic
where i is an annual interest rate.

Q the unknown size of the production lot decision variable

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 12
 Inventory vs Time in EPL Model

Inventory
Production run of Q takes Q/P time units

(P – D)(Q/P)
P–D
(P – D)(Q/P)/2
–D
Time

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 13

 Solution to EPL Model

Annual Cost Function:

 D
h 1 −  Q
Y (Q) =
AD
+  P
+ Dc
Q 2
setup holding production

Solution (by taking derivative and setting equal to zero):

• tends to EOQ as P → ∞
2 AD
Q* =
 D • otherwise larger than EOQ
h 1 −  because replenishment
 P
takes longer
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 14
What If We Allow Up To B Backorders in EOQ Model?

Inventory

Q−B B
D D

Q –B
–D

0
–B Time
Q/D 2Q/D 3Q/D 4Q/D

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 15

 Notation for EOQ with Backorders
D demand rate (units per year)
c unit production cost (dollars per unit)
A fixed or setup cost to place an order (dollars)
h holding cost (dollars per year)
b backorder cost (dollars per year)
Q the unknown size of the order or lot decision variable
B the unknown maximum number of backorders decision variable

Annual Cost Function:

(Q − B )
2
D B2
Y (Q, B ) = A + h +b + Dc
Q 2Q 2Q
setup holding shortage production

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 16

 Solution to EOQ with Backorders Model
Solution (by taking derivatives and setting both equal to zero):
∂ ∂
0= Y (Q, B) 0= Y (Q, B)
∂Q ∂B
h  B2  AD bB 2 h B  bB
= 1 − 2 − 2 − =  2 − 2 +
2 Q  Q 2Q 2 2 Q  Q
h 1  2 h+b  h+b
= − 2 B + AD  =B −h
2 Q  2  Q
h + b 2 AD h
Q2 = B2 + B=Q
h h h+b
2
 h  h + b 2 AD
= Q  + 1 1 As b → ∞,
 h+b h h Q* = 2 AD  + 
h 2 AD h b Q* tends to EOQ
=Q 2
+ h and
h+b h B* = Q*
2 AD 1 1 h+b B* tends to 0
= = 2 AD  + 
 h  h b
h 1 − 
 h+b http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
17
 The Key Insight of EOQ

There is a trade-off between lot size and inventory

D
Order Frequency: F=
Q

cQ cD
Inventory Investment: I = =
2 2F

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 18

 EOQ Tradeoff Curve
50
45
Inventory Investment

40
35
30
25
20
15
10
5
0
0 20 40 60 80 100
Order/Year
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 19
 Sensitivity of EOQ Model to Quantity

Optimal Unit Cost:

* We neglect
hQ A
Y = Y (Q ) =
*
+ *
*
unit cost c,
2D Q since it does
h 2 AD h A not affect Q*
= +
2D 2 AD h
2 Ah
=
D

Optimal Annual Cost: Multiply Y* by D and simplify,

Annual Cost = 2 ADh

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 20
 Sensitivity of EOQ Model to Quantity (cont.)

Annual Cost from Using Q':

hQ′ AD
Y (Q′) = +
2 Q′

Ratio:
Cost (Q′) Y (Q′) hQ′ 2 + AD Q′ 1  Q′ Q* 
= = =  *+ 
* *
Cost (Q ) Y (Q ) 2 ADh 2 Q Q′ 

Example: If Q' = 2Q*, then the ratio of the actual to optimal cost is

(1/2)[2 + (1/2)] = 1.25

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 21

 Sensitivity of EOQ Model to Order Interval

Order Interval: Let T represent time (in years) between orders

(production runs)
Q
T=
D
Optimal Order Interval:

2 AD
*
T* =
Q
= h = 2A
D D hD

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 22

Sensitivity of EOQ Model to Order Interval (cont.)

Ratio of Actual to Optimal Costs: If we use T' instead of T*

annual cost under T ′ 1  T ′ T * 

=  *+ 
annual cost under T *
2 T T′ 

Powers-of-Two Order Intervals: The optimal order interval,

T* must lie within a multiplicative factor of √2 of a
“power-of-two.” Hence, the maximum error from using
the best power-of-two is
1 1 
 2 +  ≈ 1.06
2 2

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 23

 The “Root-Two” Interval

2m T1* 2 m
2 T2* 2 m +1
divide by multiply by
less than less than
√2 to get √2 to get
to 2m to 2m+1

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 24

 MedEquip Example

Optimum: Q*=169, so T*=Q*/D =169/1000 = 0.169 years = 62 days

hQ * AD 35(169) 500(1000)
Y (Q*) = + = + = $5,916
2 Q* 2 169

Round to Nearest Power-of-Two: 62 is between 32 and 64, but since

32√2=45.25, it is “closest” to 64. So, round to T’=64 days or Q’=
T’D=(64/365)1000=175.
hQ ' AD 35(175) 500(1000)
Y (Q ' ) = + = + = $5,920
2 Q' 2 175
Only 0.07% error because we were lucky and happened to
be close to a power of two. But we can’t do worse than 6%.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 25
 Powers-of-Two Order Intervals

Order Interval Week

0 1 2 3 4 5 6 7 8
1 = 20

2 = 21

4 = 22

8 = 23

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 26

 Powers-of-Two Computation Process

2 AD
1. Compute the optimal order quantity (number of units): Q = *
.
h
Q* 2A
2. Compute the optimal order interval (years between orders): T = = . *

D hD
3. Express T * in the smallest possible distinct delivery time units (typically days).
4. Compute µ = log 2 (T * ) .
5. Round µ to the nearest integer m.
6. Compute the power-of-two order interval (days between orders): T ′ = 2m.
7. Express T ′ in years.
8. Compute the power-of-two order quantity (number of units): Q′ = T ′D.
9. Round Q′ up to an integer.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 27

 Example
Given the data below for three products (with a carrying charge
of 0.2 $/unit/yr), and assuming that a delivery is possible every
day (assume 365 days per year), what would be the order
quantities for each product using the power-of-two policy?

In addition to the individual setup cost, there is also a $2,000

truck cost that can be shared between products if they are
ordered at the same time.

Product Annual demand Individual setup cost Unit cost

X 200,000 $1,200 $20

Y 1,000,000 $1,100 $30
Z 180,000 $1,120 $50
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 28
 EOQ Takeaways

• Batching causes inventory (i.e., larger lot sizes translate

into more stock).

• Under specific modeling assumptions the lot size that

optimally balances holding and setup costs is given by the
square root formula:
2 AD
Q =
*

h
• Total cost is relatively insensitive to lot size (so rounding
for other reasons, like coordinating shipping, may be
attractive).

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 29

 The Wagner-Whitin Model

Change is not made without inconvenience,

even from worse to better.
– Robert Hooker

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 30

 EOQ Assumptions

1. Instantaneous production.

2. Immediate delivery.

3. Deterministic demand.

4. Constant demand. WW model relaxes this one

5. Known fixed setup costs.

6. Single product or separable products.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 31
 Dynamic Lot Sizing Notation

t a period (e.g., day, week, month); we will consider t = 1, … ,T,

where T represents the planning horizon

Dt demand in period t (in units)

ct unit production cost (in dollars per unit), not counting setup or
inventory costs in period t

At fixed setup cost (in dollars) to place an order in period t

ht holding cost (in dollars) to carry a unit of inventory from period t to

period t +1
decision
Qt the unknown size of the http://www.factory-physics.com
order or lot size in period t
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
variables
32
 Wagner-Whitin Example
Data
t 1 2 3 4 5 6 7 8 9 10
Dt 20 50 10 50 50 10 20 40 20 30
ct 10 10 10 10 10 10 10 10 10 10
At 100 100 100 100 100 100 100 100 100 100
ht 1 1 1 1 1 1 1 1 1 1
Lot-for-Lot Solution
t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 20 50 10 50 50 10 20 40 20 30 300
It 0 0 0 0 0 0 0 0 0 0 0
Setup cost 100 100 100 100 100 100 100 100 100 100 1000
Holding cost 0 0 0 0 0 0 0 0 0 0 0
Total cost 100 100 100 100 100 100 100 100 100 100 1000
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 33
 Wagner-Whitin Example (cont.)

Fixed Order Quantity Solution

t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 100 0 0 100 0 0 100 0 0 0 300
It 80 30 20 70 20 10 90 50 30 0 0
Setup cost 100 0 0 100 0 0 100 0 0 0 300
Holding cost 80 30 20 70 20 10 90 50 30 0 400
Total cost 180 30 20 170 20 10 190 50 30 0 700

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 34

 Wagner-Whitin Property

Under an optimal lot-sizing policy

either the inventory carried to period t
from a previous period will be zero
or the production quantity in period t
will be zero.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 35

 Basic Idea of Wagner-Whitin Algorithm

By WW Property I, either Qt=0 or Qt=Dt+…+Dk for some k ≥ t.

If jk* = last period of production in a k-period problem

then we will produce exactly Djk*+…+Dk in period jk*.

We can then consider periods 1, … , jk*–1

as if they form an independent problem with jk*–1 periods.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 36

 Wagner-Whitin Example

Step 1: Obviously, just satisfy D1 (note we are neglecting

production cost, since it is fixed).
Z1* = A1 = 100
j1* = 1
Step 2: Two choices, either j2* = 1 or j2* = 2.
 A1 + h1 D2 , produce in 1
Z = min  *
*

Z1 + A2 , produce in 2
2

100 + 1(50) = 150

= min 
 100 + 100 = 200
= 150

j2* = 1
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 37
 Wagner-Whitin Example (cont.)

Step 3: Three choices, j3* = 1, 2, 3.

 A1 + h1 D2 + (h1 + h2 ) D3 , produce in 1

Z 3* = min Z1* + A2 + h2 D3 , produce in 2
 Z*2 + A3 , produce in 3
100 + 1(50) + (1 + 1)10 = 170

= min 100 + 100 + (1)10 = 210
150 + 100 = 250
= 170

j3* = 1
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 38
 Wagner-Whitin Example (cont.)
Step 4: Four choices, j4* = 1, 2, 3, 4.
 A1 + h1 D2 + (h1 + h2 ) D3 + (h1 + h2 + h3 ) D4 , produce in 1
Z* + A + h D + (h + h ) D ,
 1 produce in 2
Z 4 = min  *
* 2 2 3 2 3 4

 Z 2 + A3 + h3 D4 , produce in 3
 Z*3 + A4 , produce in 4
100 + 1(50) + (1 + 1)10 + (1 + 1 + 1)50 = 320
100 + 100 + (1)10 + (1 + 1)50 = 310

= min 
150 + 100 + (1)50 = 300
170 + 100 = 270
= 270

j4* = 4
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 39
 Planning Horizon Property

If jt*= k, then the last period in which production occurs in an

optimal t+1 period policy must be in the set k, k+1,…t+1.

In the Example:

• We produce in period 4 for period 4 of a 4-period

problem.
• Thus, we would never produce in period 3 for period 5
in a 5-period problem.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 40

 Wagner-Whitin Example (cont.)
Step 5: Only two choices, j5* = 4, 5.

 Z 3 + A4 + h4 D5 , produce in 4
*
Z 5 = min  *
*

 Z 4 + A5 , produce in 5
170 + 100 + 1(50) = 320
= min 
270 + 100 = 370
= 320

j5* = 4

Step 6: Three choices, j6* = 4, 5, 6.

And so on.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 41
 Wagner-Whitin Example Solution
Last Period Planning Horizon (t)
with Production 1 2 3 4 5 6 7 8 9 10
1 100 150 170 320
2 200 210 310
3 250 300
4 270 320 340 400 560
5 370 380 420 540
6 420 440 520
7 440 480 520 610
8 500 520 580
9 580 610
10 620
Zt 100 150 170 270 320 340 400 480 520 580
jt 1 1 1 4 4 4 4 7 7 or 8 8
Produce in period Produce in period Produce in period
1 for 1, 2, 3 4 for 4, 5, 6, 7 8 for 8, 9, 10
(20 + 50 + 10 (50 + 50 + 10 + 20 (40 + 20 + 30
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com
= 80 units) = 130 units) = 90 units) 42
 Wagner-Whitin Example Solution (cont.)
Optimal Policy:
• Produce in period 8 for 8, 9, 10 (40 + 20 + 30 = 90 units)
• Produce in period 4 for 4, 5, 6, 7 (50 + 50 + 10 + 20 = 130 units)
• Produce in period 1 for 1, 2, 3 (20 + 50 + 10 = 80 units)
t 1 2 3 4 5 6 7 8 9 10 Total
Dt 20 50 10 50 50 10 20 40 20 30 300
Qt 80 0 0 130 0 0 0 90 0 0 300
It 60 10 0 80 30 20 0 50 30 0 0
Setup cost 100 0 0 100 0 0 0 100 0 0 300
Holding cost 60 10 0 80 30 20 0 50 30 0 280
Total cost 160 10 0 180 30 20 0 150 30 0 580

Note: we produce in 7 for an 8-period problem, but this

never comes into play in optimal solution.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 43
 Problems with Wagner-Whitin

1. Fixed setup costs.

2. Deterministic demand and production (no uncertainty)

3. Never produce when there is inventory (WW Property I).

• safety stock (don't let inventory fall to zero)
• random yields (can’t produce for exact number of
periods)

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 44

 Statistical Reorder Point Models

When your pills get down to four,

Order more.

– Anonymous, from Hadley & Whitin

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 45

 EOQ Assumptions

1. Instantaneous production. EPL model relaxes this one

2. Immediate delivery. lags can be added to EOQ or other models

3. Deterministic demand. newsvendor and (Q, r) relax this one

4. Constant demand. WW model relaxes this one

5. Known fixed setup costs. can use constraint approach

6. Single product or separable products. Chapter 17 extends (Q, r)

to multiple product cases
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 46
 Modeling Philosophies for Handling Uncertainty

1. Use deterministic model, then adjust solution

- EOQ to compute order quantity, then add safety stock
- deterministic scheduling algorithm, then add safety lead time

2. Use stochastic model

- news vendor model
- base stock and (Q,r) models
- variance-constrained investment models

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 47

 The Newsvendor Approach

Assumptions:
1. single period
2. random demand with known distribution
3. linear overage/shortage costs
4. minimum expected cost criterion

Examples:
• newspapers or other items with rapid obsolescence
• Christmas trees or other seasonal items
• capacity for short-life products

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 48

 Newsvendor Model Notation

X = demand (in units), a random variable.

G ( x) = P( X ≤ x)
= cumulative distribution function of demand
(assumed continuous.)
d
g (x) = G (x)
dx
= probability density function of demand.
co = cost (in dollars) per unit left over after demand is realized.
cs = cost (in dollars) per unit of shortage.
Q = production or order quantity (in units), the decision variable.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 49

 Newsvendor Model

Cost Function:
Y (Q) = expected overage + expected shortage cost

= co E [ units over ] + cs E [ units short ]

∞ ∞
= co ∫ max {Q − x, 0} g ( x)dx + cs ∫ max { x − Q, 0} g ( x)dx
0 0

Q ∞
= co ∫ (Q − x) g ( x)dx + cs ∫ ( x − Q) g ( x)dx
0 Q

Note: for any given day, we will

be either over or short, not both.
But in expectation, overage and
shortage can both be positive.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 50
 Newsvendor Model (cont.)
Optimal Solution: taking the derivative of Y(Q) with respect to Q,
setting it equal to zero, and solving yields:
d Leibnitz's Rule:
0= Y (Q)
dQ  d f2 (Q )
 ∫ f ( Q, x ) dx
d  Q  dQ 1 f ( Q )
= co ∫ (Q − x) g ( x)dx
dQ  0 
f2 (Q )  ∂ 

 =∫ f ( Q, x )  dx
∞ 
+ cs ∫ ( x − Q) g ( x)dx   f1 (Q )  ∂Q 
Q  
d
= co  ∫ (1) g ( x)dx ± 0 
Q  − f ( Q, f1 ( Q ) ) f1 ( Q )
 0   dQ

d
+ f ( Q, f 2 ( Q ) )
∞

+cs ∫ ( −1) g ( x)dx ± 0   f2 (Q )
 Q   dQ
= co G (Q) − cs [1 − G (Q) ]
G (Q ) = P { X ≤ Q } =
* * cs
= ( co + cs ) G (Q) − cs co + cs
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 51
 Newsvendor Model (cont.)

Notes: Critical Ratio is

* *
{
G (Q ) = P X ≤ Q =
cs
co + c s
} probability stock
covers demand

G(Q)
Q ↓ co
*

cs
1
Q * ↑ cs co + c s

Q* Q

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 52

 Newsvendor Example – T Shirts

Scenario:
• Demand for T-shirts is exponential with mean 1000,
so G(x) = P(X ≤ x) = 1– e–x/1000.
(Note: this is an odd demand distribution;
Poisson or Normal would probably be better modeling choices.)
• Cost of shirts is $10.
• Selling price is $15.
• Unsold shirts can be sold off at $8.

Model Parameters:
cs = 15 – 10 = $5
co = 10 – 8 = $2

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 53

 Newsvendor Example – T Shirts (cont.)

Solution:
Q
− cs 5 5
G (Q) = 1 − e 1000
= = = = 0.714
co + cs 2 + 5 7
 5
Q = −1000 ln 1 −  ≈ 1, 253
*

 7
Sensitivity: If co = $10 (i.e., shirts must be discarded) then

Q
− cs 5
G (Q) = 1 − e 1000
= = = 0.333
co + cs 10 + 5
Q* ≈ 405

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 54

 Newsvendor Model with Normal Demand

Suppose demand is normally distributed with mean µ and

standard deviation σ. Then the critical ratio formula
reduces to: 3.00

 Q * −µ  cs
Φ(z)
G (Q * ) = Φ =
 σ  co + c s

Q * −µ cs
= z where Φ ( z ) =
0.00
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121 127 133 139 145 151 157

0 z
σ co + c s

Q* = µ + zσ Note: Q* increases in both

µ and σ if z is positive (i.e.,
if ratio is greater than 0.5).

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 55

 Multiple Period Problems

Difficulty: Technically, the Newsvendor model is for a single period.

Extensions: However, the Newsvendor model can be applied to

multiple period situations, provided:
• demand during each period is iid, distributed according to G(x)
• there is no setup cost associated with placing an order
• stockouts are either lost or backordered

Key: make sure co and cs appropriately represent

overage and shortage cost.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 56

 Example

Scenario:
• GAP orders a particular clothing item every Friday
• mean weekly demand is 100, standard deviation is 25
• wholesale cost is $10, retail is $25
• holding cost has been set at $0.5 per week (to reflect
obsolescence, damage, etc.)

Problem: how should they set order amounts?

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 57

 Example (cont.)

Newsvendor Parameters:
co = $0.5
cs = $15
Solution:
15
G (Q ) = *
= 0.9677
0.5 + 15
Q* − 100
Φ ( z ) = 0.9677 ⇒ z = 1.85 =
25 Every Friday, they
Q* = 100 + 1.85(25) = 146.25 ≈ 146 should order up to
146, that is, if there
are x on hand, then
order 146–x.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 58
 Newsvendor Takeaways

• Inventory is a hedge against demand uncertainty.

• Amount of protection depends on “overage” and

“shortage” costs, as well as distribution of demand.

• If shortage cost exceeds overage cost, optimal order

quantity generally increases in both the mean and
standard deviation of demand.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 59

 The (Q,r) Approach

Assumptions:
1. Continuous review of inventory.
2. Demands occur one at a time.
3. Unfilled demand is backordered.
4. Replenishment lead times are fixed and known.

Decision Variables:
• Reorder Point: r – affects likelihood of stockout (safety stock).
• Order Quantity: Q – affects order frequency (cycle inventory).

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 60

 Inventory vs Time in (Q,r) Model

Inventory

r
l
Time

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 61

 Base Stock Model Assumptions

1. There is no fixed cost associated with placing an order.

2. There is no constraint on the number of orders that can

be placed per year.

That is, we can replenish

one at a time (Q = 1).

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 62

 Base Stock Notation

Q = 1, order quantity (fixed at one)

r = reorder point
R = r +1, base stock level
l = delivery lead time
θ = mean demand during lead time l
σ = standard deviation of demand during lead time l
p(x) = Prob{demand X during lead time l is equal to x}
G(x) = Prob{demand X during lead time l is less than or equal to x}
h = unit holding cost
b = unit backorder cost
S(R) = average fill rate (service level)
B(R) = average backorder level
I(R) = average on-hand inventory level
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 63
 Inventory Balance Equations

Balance Equation:

(inventory position) = (on-hand inventory) – (backorders) + (orders)

Under Base Stock Policy

(inventory position) = R

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 64

 Inventory Profile for Base Stock System (R = 5)

R 5

r 4 l On Hand Inventory
Backorders
Orders
3
Inventory Position

0
0 5 10 15 20 25 30 35
Time

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 65

 Service Level (Fill Rate)

Let:
X = (random) demand during lead time l

so E[X] = θ. Consider a specific replenishment order.

Since inventory position is always R,
the only way this item can stock out is if X ≥ R.

Expected Service Level:

S ( R) = P( X < R)
G ( R − 1) = G (r ) if X is discrete
=
G ( R ) if X is continuous

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 66

 Backorder Level
Note: At any point in time, number of orders equals number demands
during last l time units (X) so from our previous balance equation:
R = on-hand inventory – backorders + orders
on-hand inventory – backorders = R – X
Note: on-hand inventory and backorders are never positive at the same
time, so if X=x, then
0 if x < R
backorders = 
 x − R if x ≥ R
Expected Backorder Level (also called First-Order Loss Function):
∞
∑ ( x − R ) p ( x) if X is discrete
 x= R
B ( R ) = E  max { x − R, 0} =  ∞
 ( x − R ) g ( x)dx if X is continuous
 ∫R
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 67
 Backorder Level (Cont.)
If X ~ Poisson(θ ), then If X ~ Normal (θ , σ 2 ) , then
∞ ∞ Note that
B( R) = ∑ ( x − R) p( x)
x = R +1
B( R ) = ∫ ( x − R ) g ( x)dx
zφ ( z ) = −φ ′ ( z )
R
∞ ∞
∞
= ∑ xp ( x) − ∑ Rp ( x) = ∫ ( x − R)
1  x −θ
φ

 dx
x = R +1 x = R +1
σ  σ 
e −θ θ x
∞ ∞ R
∞
= ∑ x − R ∑ p( x)
x = R +1 x! x = R +1
= ∫θ (θ + σ z − R)φ ( z ) dz
R−
∞
e −θ θ x −1
=θ ∑ − R [1 − G ( R ) ]
σ
∞ ∞
x = R +1 ( x − 1) ! =σ ∫θ zφ ( z ) dz + (θ − R) ∫θ φ ( z ) dz
∞
e −θ θ x R− R−
=θ∑ − R [1 − G ( R) ] σ σ
x=R x! ∞ ∞

∫θ φ ′ ( z ) dz + (θ − R) ∫θ φ ( z ) dz
∞
= −σ
= θ ∑ p ( x) − R [1 − G ( R ) ] R− R−
x=R
σ σ
= θ [ p ( R) + 1 − G ( R ) ] − R [1 − G ( R ) ]
 R −θ    R − θ 
= θ p ( R ) + (θ − R ) [1 − G ( R ) ] = σφ   + (θ − R ) 1 − Φ  
 σ    σ 
= Var [ x ] p ( R ) + (θ − R ) [1 − G ( R) ]
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
= σ 2 g ( R ) + (θ − R) 1 − G ( R )  68
http://www.factory-physics.com
 Inventory Level

Observe:
• on-hand inventory = R – X + backorders
• E[X] = θ from data
• E[backorders] = B(R) from previous slide

Result:

I ( R) = R − θ + B( R)
θ p( R) + ( R − θ )G ( R) if X ∼ Poisson
= 2
σ g ( R ) + ( R − θ )G ( R) if X ∼ Normal

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 69

 Base Stock Example

l = one month

θ = 10 units (per month)

Assume Poisson demand, so

x
 10 k e −10 
x
G ( x) = ∑ p (k ) = ∑   Note: Poisson
k =0 k =0  k!  demand is a good
choice when no
variability data
is available.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 70

 Base Stock Example Calculations
x p(x) G(x) B(x) x p(x) G(x) B(x)
0 0.000 0.000 10.000 12 0.095 0.792 0.531
1 0.000 0.000 9.000 13 0.073 0.864 0.322
2 0.002 0.003 8.001 14 0.052 0.917 0.187
3 0.008 0.010 7.003 15 0.035 0.951 0.103
4 0.019 0.029 6.014 16 0.022 0.973 0.055
5 0.038 0.067 5.043 17 0.013 0.986 0.028
6 0.063 0.130 4.110 18 0.007 0.993 0.013
7 0.090 0.220 3.240 19 0.004 0.997 0.006
8 0.113 0.333 2.460 20 0.002 0.998 0.003
9 0.125 0.458 1.793 21 0.001 0.999 0.001
10 0.125 0.583 1.251 22 0.000 0.999 0.000
11 0.114 0.697 0.834 23 0.000 1.000 0.000
This table is correct only for a Poisson distribution with mean θ = 10.
If θ has some other value, a whole new table needs to be generated.
It is usually better to approximate the Poisson with the normal
distribution, and scale it to the standard normal CDF table.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 71
 Base Stock Example Results

Service Level: For fill rate of 90%, we must set R – 1 = r = 14,

so R = 15 and safety stock s = r – θ = 4. Resulting service
is 91.7%.

Backorder Level:

B(R) = B(15) = 0.103

Inventory Level:

I(R) = R – θ + B(R) = 15 – 10 + 0.103 = 5.103

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 72

 “Optimal” Base Stock Levels
Objective Function: Solution: if we assume X is continuous,
Y ( R) = hI ( R ) + bB ( R ) 0=
d
Y (R)
dR
= h [ R − θ + B ( R) ] + bB ( R ) d
=  h ( R − θ ) + ( h + b ) B ( R ) 
= h ( R −θ ) + (h + b) B ( R) dR
d
= h + ( h + b) B ( R)
holding plus backorder cost dR
∞
d
= h + ( h + b) ∫ ( x − R) g ( x)dx
dR R
 ∞ 
Implication: set base stock = h + ( h + b )  − ∫ g ( x)dx 
level so fill rate is b/(h+b).  R 
= h + ( h + b ) G ( R ) − 1

⇒ G(R ) =
Note: R* increases in b * b
and decreases in h. h+b
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 73
 Base Stock Normal Approximation

If X ~ Normal(θ,σ), then

 R * −θ  b
Φ(z) = Φ  = G ( R*) =
 σ  h+b

So R* = θ + z σ
Note: R* increases in θ
and also increases in σ
provided z > 0.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 74

 “Optimal” Base Stock Example

Data: Approximate Poisson with mean 10 by normal with mean

10 units/month and standard deviation √10 ≈ 3.16 units/month.
Set h = $15, b = $25.
b 25
Calculations: = = 0.625
h + b 15 + 25
since Φ(0.32) = 0.625, z = 0.32 and hence

R* = θ + zσ = 10 + 0.32(3.16) = 11.01 ≈ 11

Observation: from previous table fill rate is G(10) = 0.583, so

maybe backorder cost is too low.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 75

 Inventory Pooling
Situation:
• n different parts with lead time demand Normal(θ, σ)
• z = 2 for all parts (i.e., fill rate is around 97.5%)
cycle stock safety stock
Specialized Inventory:
• base stock level for each item = θ + 2σ
• total safety stock = 2nσ

Pooled Inventory: suppose parts are substitutes for one another

• lead time demand is normal (nθ,√nσ)
• base stock level (for same service) = nθ + 2√nσ
• ratio of safety stock to specialized safety stock = 1/√n

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 76

 Effect of Pooling on Safety Stock

Conclusion: cycle stock is

not affected by pooling, but
safety stock falls dramatically.
So, for systems with high safety
stock, pooling (through product
design, late customization, etc.)
can be an attractive strategy.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 77

 Pooling Example

• PCs consist of 6 components (CPU, HD, CD ROM,

RAM, removable storage device, keyboard)
• 3 choices of each component: 36 = 729 different PC’s
• Each component costs $150 ($900 material cost per
PC)
• Demand for all models is Poisson distributed with
mean 100 per year
• Replenishment lead time is 3 months (0.25 years)
• Use base stock policy with fill rate of 99%

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 78

 Pooling Example - Stock PCs

Base Stock Level for Each PC: θ = 100 × 0.25 = 25, so using
Poisson formulas,

G(R – 1) ≥ 0.99 R = 38 units

Average On-Hand Inventory for Each PC:

I(R) = R – θ + B(R) = 38 – 25 + 0.0138 = 13.0138 units

Value of Total On-Hand Inventory:

13.0138 × 729 × $900 = $8,538,358

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 79

 Pooling Example - Stock Components

Necessary Service for Each Component:

(assuming independence)
1/6
729 models of PC.
S = (0.99) = 0.9983 3 types of each component.

Base Stock Level for Components: θ = (100 × 729/3) × 0.25 = 6075,

so
G(R – 1) ≥ 0.9983 R = 6306

Average On-Hand Inventory Level for Each Component:

I(R) = R – θ + B(R) = 6306 – 6075 + 0.0363 = 231.0363 units

Value of Total On-Hand Inventory:

231.0363 × 18 × $150 = $623,798 93% reduction!
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 80
 Base Stock Insights
1. Reorder points control probability of stockouts by
establishing safety stock.
2. To achieve a given fill rate, the required base stock level (and
hence safety stock) is an increasing function of the mean and
(provided backorder cost exceeds shortage cost) standard
deviation of demand during replenishment lead time.
3. The “optimal” fill rate is increasing in the backorder cost and
a decreasing in the holding cost. We can use either a service
constraint or a backorder cost to determine the appropriate
base stock level.
4. Base stock levels in multi-stage production systems are very
similar to kanban systems and therefore the above insights
apply.
5. Base stock model allows us to quantify benefits of inventory
pooling. http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
81
 The Single Product (Q,r) Model
Motivation: Either
1. Fixed cost associated with replenishment
orders and cost per backorder.
2. Constraint on number of replenishment orders
per year and service constraint.

As in EOQ, this makes

batch production attractive.
Objective:

 fixed setup costs 

 + 
min  holding costs 
Q,r
 + 
backorder or stockout costs 

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 82

 Summary of (Q,r) Model Assumptions

1. One-at-a-time demands.
2. Demand is uncertain, but stationary over time and
distribution is known.
3. Continuous review of inventory level.
4. Fixed replenishment lead time.
5. Constant replenishment batch sizes.
6. Stockouts are backordered.

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 83

 (Q,r) Notation
D = expected demand per year
ℓ = replenishment lead time (assumed constant)
X = (random) demand during replenishment lead time
θ = E[X ] = expected demand during replenishment lead time
σ = standard deviation of demand during replenishment lead time
G (x) = P(X ≤ x) = cdf of demand during lead time
d
g (x) = G (x) = pdf of demand.
dx
p ( x) = P( X = x) = pmf of demand during lead time
A = fixed cost per order
c = unit cost of an item
h = annual unit holding cost per year
k = cost per stockout
b = annual unit backorder cost per year
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 84
 (Q,r) Notation (cont.)

Decision Variables:
Q = order quantity
r = reorder point
s = r −θ
= safety stock implied by r

Performance Measures:
F (Q) = average order frequency
S (Q, r ) = average service level (fill rate)
B(Q, r ) = average backorder level
I (Q, r ) = average inventory level

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 85

 Inventory and Inventory Position for Q = 4, r = 4
9

5
Quantity

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
-1

-2

Time

Inventory Position Net Inventory

http://www.factory-physics.com
Inventory Position is uniformly distributed between r + 1 = 5 and r + Q
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
86 = 8.
 Annual Costs in (Q,r) Model

Fixed Setup Cost: A×F(Q)

Stockout Cost: k×D×[1 – S(Q,r)], where k is cost per stockout

Backorder Cost: b×B(Q,r)

Inventory Carrying Costs: h×I(Q,r)

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 87

 Fixed Setup Cost in (Q,r) Model

Observation: since the number of orders per year is D/Q,

D
F (Q) =
Q

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 88

 Stockout Cost in (Q,r) Model

Key Observation: inventory position is uniformly

distributed between r+1 and r+Q.
Thus, service in (Q,r) model is weighted sum of
service in base stock model.

Result:
 1 r +Q −1
 Q ∑ G ( x) if X is discrete
 x=r
S (Q, r ) =  r +Q
1
Q ∫
G ( x)dx if X is continuous
 r

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 89

 Stockout Cost in (Q,r) Model (Cont.)
If X is Discrete, then
∞
1 r +Q −1
B(r ) − B(r + 1) = ∑ ( x − r ) p ( x) S (Q, r ) = ∑ G ( x)
x=r Q x=r
∞ 1 r +Q −1
− ∑ ( x − r − 1) p ( x) = ∑ [1 − B( x) + B( x + 1)]
x = r +1
Q x=r
r + Q −1 r + Q −1 r + Q −1

∑ 1− ∑ ∑
∞
B( x) + B ( x + 1)
= 0 p(r ) + ∑
x = r +1
( x − r ) p ( x)
= x =r x=r x=r

Q
∞ ∞
r + Q −1 r +Q
− ∑
x = r +1
( x − r ) p ( x) + ∑
x = r +1
p ( x) Q− ∑ B( x) + ∑ B ( x)
= x=r x = r +1

= 1− G (r ) Q
Thus, G ( r ) = 1 − B (r ) + B (r + 1) B(r ) − B (r + Q)
= 1−
∞ ∞
Q
∑ 1 − G ( y ) = ∑ [ B( y) − B( y + 1)]
y=x y=x
∞ ∞ ∞ ∞
= ∑ B( y ) − ∑ B ( y + 1) = ∑ B ( y ) − ∑ B ( y ) = B ( x)
http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
90
y=x y=x y=x y = x +1
 Stockout Cost in (Q,r) Model (Cont.)
If X is Continuous, then
r +Q
d 1
B′ ( r ) = B (r )
dr
S (Q, r ) =
Q r ∫ G ( x)dx
∞
d r +Q
= ∫ ( x − r ) g ( x)dx 1  d 
dr r = ∫ 1 + B ( x )  dx
∞
Q r  dx 
= − ∫ g ( x)dx ( r + Q ) − r  +  B ( r + Q ) − B ( r ) 
r =
Q
= G ( r ) −1
B(r ) − B (r + Q)
d = 1−
Thus, G ( r ) = 1 + B (r ) Q
dr
∞ ∞

∫ 1 − G ( y ) dy = ∫ − B′ ( y ) dy

x x
Note: this form, which is the same for
both Discrete and Continuous demand,
= B ( x ) − B (∞ ) is easier to use in spreadsheets because
= B( x) it does not involve a sum.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 91
 Service Level Approximations

Type I (base stock):

Note: computes number
S (Q, r ) ≈ G ( r ) of stockouts per cycle,
underestimates S(Q,r)
Type II:
B(r ) Note: neglects B(r+Q) term,
S (Q, r ) ≈ 1 −
Q underestimates S(Q,r)

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 92

 Backorder Costs in (Q,r) Model

Key Observation: B(Q,r) can also be computed by averaging

base stock backorder level function over the range (r, r + Q].

Result:
 1 r +Q
 Q ∑ B( x) if X is discrete
 x = r +1
B (Q, r ) =  r +Q
1
Q
 r
∫ B( x)dx if X is continuous

Notes:
1. B(Q,r)≈ B(r) is a base stock approximation for backorder level.

2. If we can compute B(x) (base stock backorder level function),

then we can compute stockout and backorder costs in (Q,r) model.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 93
Second-Order Loss Function (Discrete Distribution)
1
Let β ( x) ≡ E  max {0, X − x} max {0, X − x − 1}
2
1 ∞
= ∑ (k − x)(k − x − 1) p (k )
2 k = x +1
1 ∞ 1 ∞
β ( x − 1) − β ( x) = ∑ (k − x + 1)(k − x) p( k ) − ∑ (k − x)(k − x − 1) p(k )
2 k =x 2 k = x +1
1 1 ∞
= ( x − x + 1)( x − x) p (k ) + ∑ ( k − x + 1)(k − x) p (k )
2 2 k = x +1
1 ∞ 1 ∞
− ∑ (k − x)(k − x + 1) p (k ) − ∑ (k − x)(−2) p (k )
2 k = x +1 2 k = x +1
∞
= 0+ ∑ (k − x) p(k )
k = x +1
= B( x)
∞ ∞ ∞ ∞

∑ B( y) = ∑ [ β ( y − 1) − β ( y)] = ∑ β ( y − 1) − ∑ β ( y) = β ( x)
y = x +1 y = x +1 y = x +1 y = x +1

1 r +Q 1 r +Q β (r ) − β (r + Q)
∑ [
B(Q, r ) = ∑ B( x) = http://www.factory-physics.com
β ( x − 1) − β ( x ) ] =
Q x = r +1 Q x = r +1
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
Q 94
 Second-Order Loss Function (Continuous Distribution)
∞
1 1
Let β ( x) ≡ E  max {0, X − x}  = ∫ ( y − x ) g ( y ) dy
2 2

2   2
x
∞
d 1 d
∫ ( ) g ( y ) dy
2
β ′( x) = β ( x) = y − x
dx 2 dx x
∞
1  d ∞   dx  d  
 ( ∞ − x ) g ( ∞ ) −   ( x − x ) g ( x ) + ∫ ( y − x ) g ( y )  dy 

2 2 2
= 
2  dx   dx  x
dx 
∞
1 
= 0 − 0 + ∫  −2 ( y − x ) g ( y )  dy 
2 x 
∞
= − ∫ ( y − x ) g ( y ) dy
x
Note: this form, which is the
= − B( x) same for both Discrete and
∞ ∞

∫x B( y)dy = ∫x [ −β ′( x)] dy = β ( x) Continuous demand, is also

easier to use in spreadsheets,
1
r +Q
β (r ) − β (r + Q) because it does not involve a
Q ∫r
B(Q, r ) = B ( x ) dx =
Q sum.
http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
95
 Second-Order Loss Function (Poisson Distribution)
1 ∞ 1 ∞ θ k e−θ
β ( x ) = ∑ (k − x)(k − x − 1) p(k ) = ∑ ( k − k − 2 xk + x + x )
2 2

2 k = x +1 2 k = x +1 k!
e −θ  ∞ θ k ∞
θ k ∞
θ k

= ∑(
2  k = x +1
k 2
− k ) k! ∑ k! (
− 2 x k + x 2
+ x ) ∑ k!
k = x +1 k = x +1 
e −θ  2 ∞ θ k − 2 ∞
θ k −1 ∞
θk 
= θ ∑ − 2θ x ∑ + ( x + x) ∑
2

2  k = x +1 ( k − 2 ) ! k = x +1 ( k − 1 ) ! k = x +1 k ! 
e −θ  2 ∞ θ k ∞
θk ∞
θk 
= θ ∑ − 2θ x ∑ + ( x + x ) ∑ 2

2  k = x −1 k ! k =x k ! k = x +1 k ! 
1  2  θ x −1e −θ θ x e −θ  θ x e−θ ∞
θ k e−θ 
= θ  +  − 2θ x + (θ − 2θ x + x + x ) ∑
2 2

2   ( x − 1) ! x!  x! k = x +1 k! 
1 2 θ x e−θ ∞
θ k e−θ 
= (θ − θ x ) + (θ − 2θ x + x + x ) ∑
2 2

2 x! k = x +1 k ! 
θ (θ − x ) p ( x ) + (θ − x ) + x  1 − G ( x ) 
2

=  
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 96
2
Second-Order Loss Function (Geometric Distribution)
∞ ∞
B( x) = ∑ 1 − G ( y )  β ( x) = ∑ B( y )
y = x +1
y=x
∞
1
{ }
∞
∑ (1 − p )
y
= ∑ 1 − 1 − (1 − p ) 
y =
y=x
  y = x +1 p
(1− p)
x +1
∞
∞
∑ (1 − p )
y − x −1
= ∑ (1 − p )
y =
p y = x +1
y=x
(1− p)
x +1
∞
∞
∑ (1 − p )
k
=
= (1 − p ) ∑ (1 − p )
x y−x

y=x
p k =0

(1− p)
x +1
∞ 1
=
= (1 − p ) ∑ (1 − p )
x k
p p
k =0
(1− p)
x +1
1
= (1 − p ) =
x

p p2
x +1
 1
x  1
= θ 1 −  = θ 2 1 − 
 θ  θ
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 97
 Second-Order Loss Function (Normal Distribution)

∞
1
β ( x) = ∫ ( y − x ) g ( y ) dy
2

2x
∞
1 1
= ∫ (θ + σ z ) − 2 x (θ + σ z ) + x 2  φ ( z ) σ dz
2

2 x −θ  σ
σ
∞
1
= ∫ (θ 2 + 2θσ z + σ 2 z 2 ) − 2 x (θ + σ z ) + x 2  φ ( z ) dz
2 x −θ
σ
 ∞ ∞ ∞

1 2 
= (θ − 2 xθ + x 2 ) ∫ φ ( z ) dz + 2σ (θ − x ) ∫ zφ ( z ) dz + σ 2 ∫ z 2φ ( z ) dz 
2 x −θ x −θ x −θ 
 σ σ σ 
 ∞ ∞

1   x − θ  
= (θ 2 − 2 xθ + x 2 ) 1 − Φ   − 2σ (θ − x ) ∫ φ ′ ( z ) dz − σ 2
∫ zφ ′ ( z ) dz 
2   σ  x −θ x −θ 
 σ σ 
⋮ http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
98
Second-Order Loss Function (Normal Distribution, Cont.)
1 2 2   x −θ    x −θ 
= (θ − 2 xθ + x ) 1 − Φ    − 2σ (θ − x ) 0 − φ  
2   σ    σ 
 ∞

2  ∞ 
− σ  zφ ( z ) x −θ − ∫ φ ( z ) dz  
 σ x −θ
σ
 
1 2   x − θ   x −θ 
= (θ − 2 xθ + x 2 ) 1 − Φ   + 2σ ( θ − x ) 
φ 
2   σ   σ 
2  x −θ  x −θ   x − θ  
−σ 0 − φ  −1+ Φ   
 σ  σ   σ  
1   x − θ   x − θ 
= (θ 2 − 2 xθ + x 2 + σ 2 ) 1 − Φ   + σ (θ − x ) 
φ 
2   σ   σ 
1   x − θ   x − θ 
=  ( x − θ ) + σ 2  1 − Φ  ( ) 
2
 − σ x − θ φ 
2     σ   σ 

=σ 2 ( z 2
+ 1) 1 − Φ ( z )  − zφ ( z )
where z =
x −θ
http://www.factory-physics.com
σ
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000
99
2
Second-Order Loss Function (Exponential Distribution)

∞ ∞

B( x) = ∫ 1 − G ( y )  dy β ( x) = ∫ B( y )dy
x x
∞ y ∞ y
− −
= ∫ e θ dy = ∫ θ e θ dy
x x
∞ y ∞ y
1 − 1 −
= θ ∫ e θ dy =θ ∫θ e θ dy
2

x
θ x
∞ ∞

= θ ∫ g ( y ) dy = θ 2 ∫ g ( y ) dy
x x

= θ 1 − G ( x )  = θ 2 1 − G ( x ) 
x x
− −
= θe θ =θ e 2 θ

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 100

 Inventory Costs in (Q,r) Model
Approximate Analysis: on average, inventory declines from Q+s
to s+1 so
(Q + s ) + ( s + 1) Q + 1 Q +1
I (Q, r ) ≈ = +s= + r −θ
2 2 2
Exact Analysis: this neglects backorders, which add to average
inventory since on-hand inventory can never go below zero. The
corrected version turns out to be
Q +1
I (Q, r ) = + r − θ + B (Q, r ) (*)
2
Note that for a continuous demand distribution, the correct
expression for inventory costs would be
Q
I (Q, r ) = + r − θ + B (Q, r )
2
However, if the continuous distribution is just used to approximate
the actual discrete distribution, the discrete version (*) is better.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 101
 Inventory vs Time in (Q,r) Model

Expected Inventory Actual Inventory

Exact I(Q,r)
s+Q = Approx I(Q,r)
Inventory

+ B(Q,r)

r Approx I(Q,r)
θ+1
s+1=r–θ

Time

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 102

 Expected Inventory Level for Q = 4, r = 4, θ = 2

s+Q 6

5
Inventory Level

s 2

0
0 5 10 15 20 25 30 35
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 Time
http://www.factory-physics.com 103
 Optimizing (Q,r) Model with Backorder Cost

Objective Function:
D
Y (Q, r ) = A + bB(Q, r ) + hI (Q, r )
Q
AD  Q +1 
= + h + r − θ  + ( b + h ) B(Q, r )
Q  2 

Approximation: B(Q,r) makes optimization complicated

because it depends on both Q and r. To simplify,
approximate with base stock backorder formula, B(r):
Y (Q, r ) ≈ Yɶ (Q, r )
AD  Q +1 
= + h + r − θ  + ( b + h ) B(r )
Q  2 
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 104
 Results of Approximate Optimization
Assumptions:
∂ ɶ
0= Y (Q, r ) • Q,r can be treated as
∂Q continuous variables
∂  AD  Q +1  
=  + h + r − θ  + ( b + h ) B(r )  • G(x) is a continuous cdf
∂Q  Q  2  
AD h
Results: Note:
=− 2 + this is
Q 2 2 AD
∂ ɶ Q* = just the
0 = Y (Q, r ) h EOQ
∂r
∂  AD  Q +1   formula
=  + h + r − θ  + ( b + h ) B(r ) 
∂r  Q  2   G (r*) = b Note:
d
= h + ( b + h ) B(r )
h+b this is
dr ⇒ r* = θ + zσ just the
= h − ( b + h ) 1 − G ( r )  if G is Normal(θ,σ2), base
where Φ(z)=b/(h+b) stock
http://www.factory-physics.com
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 formula
105
 (Q,r) Example

Stocking Repair Parts:

D = 14 units per year

c = $150 per unit
h = 0.1 × 150 + 10 = $25 per unit
l = 45 days
θ = 14×45/365 = 1.726 units during replenishment lead time
b = $40
A = $10
Demand during lead time is Poisson

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 106

 Values for Poisson(θ) Distribution

r p(r) G(r) B(r)

0 0.178 0.178 1.726
1 0.307 0.485 0.904
2 0.265 0.750 0.389
3 0.153 0.903 0.140
4 0.066 0.969 0.042
5 0.023 0.991 0.011
6 0.007 0.998 0.003
7 0.002 1.000 0.001
8 0.000 1.000 0.000
9 0.000 1.000 0.000
10 0.000 1.000 0.000 107
 Calculations for Example

2AD 2(10)(14)
Q* = = = 3.3466, so Q* = 3 or 4
h 25

b 40
= = 0.615
h + b 25 + 40

Using Poisson:
G (1) = 0.485 < 0.615 < 0.750 = G ( 2) , so r* = 1 or 2

Or using Normal:
Φ(0.29) = 0.615, so z = 0.29
r* = θ + zσ = 1.726 + 0.29(1.314) = 2.107, so r* = 2 or 3
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 108
 Performance Measures for Example
Choose Q* = 4 and r * = 2

F (Q ) = *
* D
Q
14
=
4
= 3.5 Orders placed at rate of 3.5 per year.
B ( r * ) − B ( r * + Q* )
S ( Q* , r * ) = 1 −
Q*
B(2) − B (2 + 4)
= 1−
4
0.389 − 0.003
= 1−
4
= 0.904 Fill rate fairly high (90.4%).
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 109
 Performance Measures for Example (Cont.)
r * + Q*
1
B (Q , r ) = *
* *
∑ B( x)
Q x = r * +1

B(3) + B(4) + B(5) + B(6)

=
4
0.140 + 0.042 + 0.011 + 0.003
=
4
= 0.049 Very few outstanding backorders.

*
+1 *
I (Q , r ) = + r − θ + B ( Q* , r * )
* * Q
2
4 +1
= + 2 − 1.726 + 0.049
2
= 2.823 Average on-hand inventory just below 3.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 110
 Varying the Example
Change: suppose we order twice as often so F = 7 per year,
then Q = 2 and:
1 1
S (Q, r ) = 1 − [ B(r ) − B (r + Q)] = 1 − [0.389 − 0.042] = 0.826
Q 2
which may be too low, so increase r from 2 to 3:
1 1
S (Q, r ) = 1 − [ B(r ) − B(r + Q)] = 1 − [0.140 − 0.011] = 0.936
Q 2
This is better. For this policy (Q = 2, r = 3) we can compute
B(2,3) = 0.026,
I(Q,r) = 2.80.

Conclusion: this has higher service and lower inventory than

the original policy (Q = 4, r = 2). But the cost of achieving
this is an extra 3.5 replenishment orders per year.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 111
 Optimizing (Q,r) Model with Stockout Cost

Objective Function:
D
Y (Q, r ) = A + kD [1 − S (Q, r ) ] + hI (Q, r )
Q

Approximation: Assume we can still use EOQ to compute

Q* but replace S(Q,r) by Type II approximation and
B(Q,r) by base stock approximation:

ɶ D B( r ) Q + 1 
Y (Q, r ) ≈ Y (Q, r ) = A + kD +h + r − θ + B( r ) 
Q Q  2 

© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 112

 Results of Approximate Optimization

Assumptions:
• Q and r can be treated as continuous variables
• G(x) is a continuous cdf

Results:

2 AD
Q* = Note: this is just the EOQ formula
h

kD Note: another version of the base

G (r*) =
kD + hQ stock formula (only z is different)

⇒ r* = θ + zσ if G is Normal(θ, σ2),
where Φ(z) = kD/(kD+hQ)
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 113
 Backorder vs. Stockout Model
Backorder Model
• when real concern is about stockout time
• because B(Q,r) is proportional to time orders wait for
backorders
• useful in multi-level systems

Stockout Model
• when concern is about fill rate
• better approximation of lost sales situations (e.g., retail)

Note:
• We can use either model to generate frontier of solutions
• Keep track of all performance measures regardless of model
• B-model will work best for backorders, S-model for stockouts
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 114
 Lead Time Variability

Problem: replenishment lead times may be variable,

which increases variability of lead time demand.

Notation:
L = replenishment lead time (days), a random variable
l = E[L] = expected replenishment lead time (days)
σL = standard deviation of replenishment lead time (days)
Dt = demand on day t, a random variable,
(assumed independent and identically distributed)
d = E[Dt] = expected daily demand
σD = standard deviation of daily demand (units)
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 115
 Including Lead Time Variability in Formulas

Standard Deviation of Lead Time Demand:

if demand is Poisson

σ = ℓσ D2 + d 2σ L2 = θ + d 2σ L2
Inflation term due to
lead time variability
Modified Base Stock Formula (Poisson demand case):
R = θ + zσ
= θ + z θ + d 2σ L2
Note: σ can be used in any base stock or (Q,r) formula
as before. In general, it will inflate safety stock.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 116
 Single Product (Q,r) Insights

Basic Insights:
• Safety stock provides a buffer against stockouts.
• Cycle stock is an alternative to setups/orders.

Other Insights:
1. Increasing D tends to increase optimal order quantity Q.
2. Increasing θ tends to increase the optimal reorder point.
(Note: either increasing D or l increases θ.)
3. Increasing the variability of the demand process tends to
increase the optimal reorder point (provided z > 0).
4. Increasing the holding cost tends to decrease the optimal
order quantity and reorder point.
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 117
 Main Models Discussed So Far
Features: Model: EOQ EPL WW NV BS (Q, r)
Time: Continuous (C) or Discrete (D) C C D D C C
Products: Single (S) or Multiple (M) S S S S S S
Periods: Single (S) or Multiple (M) - - M S - -
Backordering (B) or Lost Sales (L) - - - L B B
Setup or Order Cost: Yes (Y) or No (N) Y Y Y N N Y
Demand: Deterministic (D) or Random
D D D R R R
(R)
Production: Deterministic (D) or
D D D D D D
Random (R)
Demand Rate: Constant (C) or Dynamic
C C D - C C
(D)
Production Rate: Finite (F) or Infinite (I) I F I - I I
Time Horizon: Finite (F) or Infinite (I) I I F F I I
Echelons: Single (S) or Multiple (M) S S S S S S
© Wallace J. Hopp, Mark L. Spearman, 1996, 2000 http://www.factory-physics.com 118