IEDA 4100

Integrated Production Systems

Inventory Control Subject to

Known Demand
Allen Wu
2025 Spring
Outline
} Introduction

} Deterministic inventory models

} Basic EOQ model
} EOQ with finite production rate
} EOQ with quantity discounts
} Power-of-two policies
} Lot sizing with multiple products
} EOQ models for production planning

Page 2
Why Hold Inventory?
} Economies of scale
} Uncertainties
} Uncertainty of external demand, lead time, and supply
} Speculation
} Purchase large quantities in advance of the price increase
} Transportation (pipeline inventory)
} Smoothing
} Producing/storing inventory in anticipation of peak demand

Page 3
Characteristics in Inventory Systems
} Demand
} Constant versus variable
} Known versus random
} Lead time
} Amount of time elapsed from order placement to arrival
} Review time
} Continuous review, periodic review
} Excess demand
} Backordered, lost

Page 4
Costs Related to Inventory
} Holding cost - Costs proportional to the quantity of inventory
held
} Storage cost
} Expenses related to storage space, insurance, taxes,
obsolescence etc.
} Inventory carrying cost
} Interest charges related to the investment tied up in inventory
} Depends on the interest rate and unit price of the item
} h = Ic
} I: annual interest rate
} c: dollar value of one unit of inventory
} h: holding cost in terms of dollars per unit per year

Page 5
Costs Related to Inventory (cont’d)
} Order cost (or production cost)
} A fixed component : order processing, delivery, production setup
} A variable component : proportional to order (production) quantity

C(x) = K + cx

} Shortage (stock-out, penalty) cost – charged on a per unit basis

} Lost sales – including lost profit
} Back orders – including whatever bookkeeping or delay costs

Page 6
Inventory Models
} Deterministic models
} Continuous model : constant, known demand rate
} Economic Ordering Quantity (EOQ)
} Discrete model : d1, d2,…,dt,…,dT
} Dynamic Lot Size Model

} Stochastic Models:
} Single period problem (Newsvendor problem)
} Multi-period problem

Page 7
Economic Order Quantity (EOQ)
} Basic modelling assumptions
} A single product
} Demand rate is known and constant over time
} No order or production lead time
} Shortage is not allowed

} Cost incurred: setup cost, proportional order cost and

holding cost

} Trade-off between fixed order cost and holding cost

Page 8
The EOQ Model
} Given parameters
} l: Demand rate (in units per year)
} c: Unit production/purchase cost (in dollars per unit)
} K: Fixed setup cost to produce/order one lot (in dollars)
} h: Holding cost of one unit for a year (in dollars per unit per
year)
} Decision variable:
} Q: Lot size/order quantity (in units)

Problem:
When to order and how much quantity for each order?
Page 9
The EOQ Model (cont’d)
} h = Ic, where I is annual interest rate or holding cost per dollar
} n = l /Q, production/order frequency, i.e., number of lots per
year (times per year)
} T = Q/ l, cycle length of a lot (in years). T and Q are
dependent variables.
} Observations:
} Order only if inventory is zero (instantaneous replenishment)
} Order the same amount (Q) every time
Inventory level
Q

0 T 2T 3T Time
Page 10
Total Average Annual Cost
} Total fixed plus proportional order cost:
} C(Q) = K+cQ
} Average inventory level: Q/2
} Average annual cost:
K + cQ hQ
G (Q) = +
T 2
K + cQ hQ
= +
Q l 2
Kl hQ
= + lc +
Q 2
annual setup cost + annual purchase cost + annual holding cost
Page 11
Graph of G(Q)

G(Q) = K(l/Q) + h(Q/2)

Annual Cost

Holding Costs

Setup Costs
EOQ (optimal order quantity) Order Quantity (Q)

Page 12
Minimize G(Q)
} Minimize G(Q) = K + cQ + hQ
T 2
dG Kl h
=- 2 +
dQ Q 2

d 2 G 2 Kl
2
= 3 >0 for Q > 0
d Q Q

} First order condition

dG Kl h 2 Kl
=- 2 + =0 Q =
*

dQ Q 2 h

Page 13
Properties of EOQ Solution
2 Kl
Q =
*

h
} Q is increasing with K and λ and decreasing with h
} Q changes as the square root of these quantities
} Q is independent of the proportional order cost c

Page 14
EOQ Formula
}Optimal order quantity :
2 lK 2 ( Annual demand ) ( Setup cost )
Q* = =
h Annual unit holding cost

Order frequency and order interval :

*
l Q 1
n = *
*
T =
*
= *
Q l n
}Total annual cost (setup and holding) at optimality

G (Q* ) = 2 Klh

Page 15
Cycle Inventory
} The average inventory
Q *
cycle inventory = lot size =
2 2

} Average flow time (Little’s Law)

cycle inventory Q *
Average flow time = =
demand 2l

Page 16
Example
} Demand: 20 units of demand per week
} Unit cost : $4
} Fixed cost: $20/order
} Annual interest rate: 0.25 (per $ per year)
} Findthe optimal lot size, cycle time, average flow
time, and total annual holding and setup costs

Page 17
Example (cont’d)
l = 20*52 = 1040 units per year

h = Ic = 0.25*4=1, K=20
Q* = (2Kl/h)1/2 = (2*20*1040/1)1/2 = 203.96
Round Q* down or up to check the total cost:
G(203)= (20*1040)/203 + 1*(203/2) = 203.96305
G(204)= (20*1040)/204 + 1*(204/2) = 203.96078
Then, Q* = 204 units

Page 18
Example (cont’d)
n* = l/Q* = 1040/204 = 5.10 times per year
T* = Q*/l = 204/1040 = 0.196 year = 10.2 week
G(Q*) = (2Klh) 1/2 = $204
Average flow time = Q/2l=204/(2*1040)=5.1 week

If order in batches of 100 rather than 204, what additional

cost is incurred by using a suboptimal solution?

Page 19
Sensitivity to Q
} For any Q>0,

G (Q ) Kl Q + hQ 2 Kl hQ
= = +
( )
GQ *
2 Kl h Q 2 Kl h 2 2 Kl h
1 Kl Q h
= +
Q 2 h 2 2 Kl
1 2 Kl Q h
= +
2Q h 2 2 Kl
1 æ Q* Q ö
= çç + * ÷÷
2è Q Q ø

Page 20
Sensitivity to Q (cont’d)
} Sensitivity of setup and holding cost G(Q) with
respect to lot size Q
G(Q)
G(Q*)
1.3

1.2

1.1

Q
1.0
Q*
0.5 1 1.5 2

Page 21
Sensitivity to T
} Instead of lot size Q, take cycle length T as the
decision variable
G(T) 1 æ T T * ö
= ç + ÷
G(T) G(T*) 2 è T * T ø
G(T*)
1.3

1.2

1.1

1.0 T
0.5 1 1.5 2 T*

Page 22
Inclusion of Order Lead Time
} Relax the assumption: there is no time lag between production
and availability to satisfy demand
} Let the order lead time to be equal to t. Order t units of time
before inventory position is zero. Trigger an order when
inventory drop to the reorder point.
Q Demand
rate

Quantity
on hand

Reorder
point

Time
Receive Place order Receive order Place Receive
order order order
Lead time
Page 23
 Inclusion of Order Lead Time (cont’d)
} Does it matter if τ < T or τ > T ?
} Reorder point R=λ*τ, if τ < T
} When τ > T, R = λ*T*MOD(τ/T)

Example:
λ=500, Q=25, τ=6 weeks
T=Q/ λ = 0.05 year=2.6
weeks

τ/T = 2.31
MOD(τ/T) = 0.31
R = 500*0.05*0.31 =
7.75
Page 24
Inventory Levels for Finite Production Rate
Model

Page 25
Extension to a Finite Production Rate
} Additional parameter: Production rate, P (in units per year)
} Analysis
Lot size Q is still the decision and inventory holding cost needs to
}
be modified
} The uptime (when production happens) T1 = Q/P
} The downtime (when production does not happen) T2 = Q/l- Q/P
} Maximum inventory level in a cycle = (Q/P) (P-l) =Q(1- l/P)
} Average inventory level in a cycle = ½ Q(1- l/P)

} Total cost per year = hQ (1 - l P ) + Kl + cl

2 Q
} Optimal lot size
2Kl 2Kl l
Q* = = ¢
, where h = h(1- ).
l h¢ P
h(1- )
P
Page 26
Example
} Demand: 2,500 units of demand per year
} Production rate: 10,000 units of demand per year
} $2 production price for one unit
} $50 for each setup
} 30% annual holding cost per dollar
}h = (0.3)(2) = 0.6, h’ = 0.6(1-2,500/10,000) = 0.45.
} Q* = 745. T* = 745/2,500 = 0.298 year.
} T1 = Q/P = 745/10,000= 0.0745year. T2 = T- T1 = 0.2235 year.
} The maximum inventory level = Q* (1- l /P) = 559 units.

Page 27
Quantity Discount Models
} One of the most severe assumptions: the unit variable
cost c did not depend on the replenishment quantity
} In practice: quantity discounts exist based on the
purchased price or transportation costs
} All Units Discounts: the discount is applied to ALL of
the units in the order.
} Incremental Discounts: the discount is applied only to
the number of units above the breakpoint.

Page 28
All Units Discount Order Cost Function

ì0.30Q for 0 £ Q < 500

ï
C (Q ) = í0.29Q for 500 £ Q < 1,000
ï0.28Q for 1,000 £ Q
î

Page 29
EOQ with All-Unit Quantity Discounts
G j (Q) = lc j + lK / Q + Ic j Q / 2 for j = 0,1, and 2
ìG0 (Q) for 0 £ Q < q1 ,
ï
G (Q) = íG1 (Q) for q1 £ Q < q2 ,
ïG (Q) for q2 £ Q
î 2
G(Q)
G0(Q)
G1(Q)
G2(Q)

Q
500 1,000 Page 30
Solution Technique for All-Unit Quantity
Discounts
} Two-phase optimization

} Determine the largest realizable EOQ value.

} Compare the value of the average annual cost at the

largest realizable EOQ and at all the price break
points that are greater than the largest realizable
EOQ. The optimal Q is the point at which the
average annual cost is a minimum.

Page 31
All-Unit Quantity Discounts: Example
c0 = $0.30, c1 = $0.29, c2 = $0.28
l = 600 units/year, K = $8/lot, I = 0.2

2 Kl ( 2)(8)(600)
Q( 0) = = = 400,
I c0 (0.2)(0.3)
(1) 2 Kl ( 2)(8)(600)
Q = = = 406,
I c1 (0.2)(0.29)
2 Kl ( 2)(8)(600)
Q( 2) = = = 414.
I c2 (0.2)(0.28)
G(400)=G0(400) = (600)(0.30) +(600)(8)/400+(0.2)(0.30)(400)/2 = $204.00
G(500)=G1(500) = (600)(0.29) +(600)(8)/500+(0.2)(0.29)(500)/2 = $198.10
G(1,000)=G2(1,000) = (600)(0.28) +(600)(8)/1000+(0.2)(0.28)(1000)/2 = $200.80

Page 32
Incremental Quantity Discount

Page 33
EOQ with Incremental Quantity Discounts

ì0.30Q for 0 £ Q < 500,

ï
C (Q ) = í150 + 0.29(Q - 500) = 5 + 0.29Q for 500 £ Q < 1000,
ï295 + 0.28(Q - 1000) = 15 + 0.28Q for 1000 £ Q
î

} Hence,
ì0.30 for 0 £ Q < 500,
ï
C (Q ) / Q = í0.29 + 5 / Q for 500 £ Q < 1000,
ï0.28 + 15 / Q for 1000 £ Q
î

Page 34
EOQ with Incremental Quantity Discounts
(cont’d)
} Average annual cost function:
G(Q)= lC(Q)/Q+ Kl/Q+ I[C(Q)/Q]Q/2

G0(Q)= (600)(0.3) + (8)(600)/Q+ (0.20)(0.30)Q/2

(0) 2 Kl ( 2)(8)(600)
Q = = = 400;
I c0 (0.20)(0.30)

G1(Q)= (600)(0.29+5/Q) + (8)(600)/Q+ (0.20)(0.29+5/Q)Q/2

=(600)(0.29) +(13)(600)/Q + (0.20)(0.29)Q/2 +(0.20)(5)/2

(2)(13)(600)
(1)
Q = = 519;
(0.20)(0.29)
Page 35
EOQ with Incremental Quantity Discounts
(cont’d)
G2(Q)= (600)(0.28+15/Q) + (8)(600)/Q+ (0.20)(0.28+15/Q)Q/2
=(600)(0.28) +(23)(600)/Q + (0.20)(0.28)Q/2 +(0.20)(15)/2.

(2) (2)(23)(600)
Q = = 702.
(0.20)(0.28)

Both Q(0) and Q(1) are realized, yet Q(2) is not.

The optimal solution is obtained by comparing
G0(Q(0)) and G1(Q(1)).
G0(Q(0)) = $204.00
G1(Q(1)) = $204.58.

Page 36
Average Annual Cost Function

Page 37
Solution Technique for Incremental Discounts
} Find C(Q) and C(Q)/Q

} Find the minimum value of Q corresponding to each

price interval separately of G(Q).

} Determine which minima are realizable. Compare the

values of the average annual costs at the realizable
EOQ values and pick the lowest.

Page 38
Power-of-2 Policy: Idea
} The power-of-2 policy means that we produce quantity at
intervals given by power of 2
} This policy is, of course, a heuristic approach
} This is easy to use and can coordinate shipments of different
products
Interval Week
0 1 2 3 4 5 6 7 8

1= 20

2 = 21

4 = 22

8 = 23
Page 39
Power-of-2 Policy: Implementation
Assume that
1) the optimal interval obtained from the EOQ model is T*
2) T* is between 2m and 2m+1
3) we use power-of-2 policy
Thus, produce at a cycle length T = 2m if 2m £ T* £ 2m 2
produce at a cycle length T = 2m+1 if 2m 2 £ T* £ 2m+1

1 T 1 T*
£ £ 2 and £ £ 2
2 T* 2 T

The error of using the above policy can at most result in an

error of 6% with regard to the inventory holding cost plus the
setup cost

Page 40
Lot Sizing with Multiple Products
} Transportation is a significant contributor to the fixed
cost per order.
} In practice, the fixed ordering cost consists of two parts:
} A portion of the cost is related to transportation (independent of
variety)
} A portion of the cost is related to loading and receiving
(dependent on variety)
} Delivery options:
} No aggregation – Each product ordered separately
} Complete aggregation – All products ordered each time
} Tailored aggregation – Selected subsets of products ordered
jointly
Page 41
Example
} Suppose there are three computer models: Litepro,
Medpro, and Heavpro
} Demand per year
} lL = 12,000; lM = 1,200; lH = 120
} Fixed ordering cost, K0 = $4,000
} Product specific order cost
} KL = $1,000; KM = $1,000; KH = $1,000
} Holding cost, I = 0.2
} Unit cost
} cL = $500; cM = $500; cH = $500
} Inventory holding cost h=Ici ($/unit/year)
Page 42
No Aggregation
} Order each product independently

2(K 0 + K L )l 2 ´ (4000 + 1000 )´12000

Q = *
L = = 1095
IcL 0.2 ´ 500
2(K 0 + K M )l 2 ´ (4000 + 1000 )´1200
Q = *
M = = 346
IcM 0.2 ´ 500

2(K 0 + K H )l 2 ´ (4000 + 1000 )´120

Q = *
H = = 110
IcH 0.2 ´ 500

Page 43
No Aggregation (cont’d)
Litepro Medpro Heavypro
Demand per year 12,000 1,200 120
Fixed order cost $5,000 $5,000 $5,000
($/order)
Order frequency 11/year 3.5/year 1.1/year

Optimal order size 1095 346 110

} Total setup and holding cost = $ 155,140

Page 44
Complete Aggregation

} Total ordering cost K=K0+KL+KM+KH=$7,000

} Common order frequency: n
} Annual ordering cost = nK
} Annual holding cost: lL IcL + lM IcM + lH IcH
2n 2n 2n
} Total cost: Kn + lL IcL + lM IcM + lH IcH
2n 2n 2n
I (lL cL + lM cM + lH cH )
n =
*

2K
Page 45
Complete Aggregation (cont’d)
} In the example:
0.2(12000 ´ 500 + 1200 ´ 500 + 120 ´ 500 )
n =
*
= 9.75
2 ´ 7000

lL
12000
QL* = *
= = 1230
n 9.75
l 1200
QM* = M* = = 123
n 9.75
lG
120
Q = * =
*
H = 12.3
n 9.75

Page 46
Complete Aggregation (cont’d)
Litepro Medpro Heavypro
Demand per year 12,000 1,200 120
Fixed order cost $5,000 $5,000 $5,000
($/order)
Order frequency 9.75/year 9.75/year 9.75/year

Optimal order size 1230 123 12.3

} Annual setup cost = 9.75 × $7,000 = $68,250

} Total setup and holding cost = $136,557.69 (155,140)

Page 47
Tailored Aggregation
} Which product is ordered most frequently?
lL IcL 12000 ´ 0.2 ´ 500
n =
*
= = 11
2(K 0 + K L )
L
2 ´ 5000
nM* = 3.5 n*H = 1.1

} Most frequently ordered product L

} We can associate the fixed order cost K0 with L as it is ordered
every time there is an order
} Products other than L are associated only with their
incremental order costs (KM, KH)

Page 48
Tailored Aggregation (cont’d)
} i is the generic index for items, i is L, M or H.
} Step 1: Find the most frequently ordered item:
li Ici
ni =
2(K 0 + K i )
n = max ni { }
The frequency of the most frequently ordered item will be
modified later. This is an approximate computation.
} Step 2: Relative order frequency of other items, mi

li Ici énù
ni = mi = ê ú
2Ki êê ni úú
mi are relative order frequencies, they must be integers.
They do not change in the remainder.
Page 49
Tailored Aggregation (cont’d)
} Step 3: Recompute the frequency of the most frequently
ordered item. This item is ordered in every order whereas
others are ordered in every mi orders. The average fixed
ordering cost is: Ki
K = K0 + å
i mi
} Annual ordering cost = nK
li Ici
} Annual holding cost = å 2n m
i i

å l m Ic
i i i
n= i
Reflect the fact that product i
æ Ki ö
2çç K 0 + å ÷÷ is ordered with frequency mi
è i mi ø
Page 50
Tailored Aggregation (cont’d)
Step 4: Evaluate the ordering frequency ni of other products:
n
ni =
mi
and calculate the total cost of such an ordering policy.

Total ordering cost: nK 0 + ån K

i
i i

li Ici
Total holding cost: å
i 2ni

Page 51
Tailored Aggregation (cont’d)
} In the example:
} Step 1: nL = 11 nM = 3.5 nH = 1.1 { }
n = max ni = 11
} Step 2:
lM IcM 1200 ´ 0.2 ´ 500
nM = = = 7.7 nH = 2.4
2K M 2 ´1000

é n ù é n ù
mM = ê ú=2 mH = ê ú = 5
êê nM úú êê nH úú

Item L is ordered most frequently.

Every other L order contains one M order.
Every 5 L orders contain one H order.
Page 52
Tailored Aggregation (cont’d)
} Step 3:
n=
(12000 + 1200 ´ 2 + 120 ´ 5)´ 0.2 ´ 500 = 11.47
2(4000 + 1000 + 1000 2 + 1000 5)
n
} Step 4: nL = = 11.47 nM = 11.47 = 5.74 nH = 2.29
mL 2
} Total ordering cost: nK0+nLKL+nMKM+nHKH
} 11.47(4000)+11.47(1000)+5.74(1000)+2.29(1000)=$65380
} Order size:
l 12000 l 1200
QL = L = = 1046 QM = M = = 209
nL 11.47 nM 5.74
QH = 52
Page 53
Tailored Aggregation (cont’d)
Litepro Medpro Heavypro
Demand per year 12,000 1,200 120
Fixed order cost $5,000 $5,000 $5,000
($/order)
Order frequency 11.47/y 5.74/year 2.29/year
ear
Optimal order size 1046 209 52

}Total setup and holding cost = $ 130,763.42

Compare with $136K of total aggregation and with $155K of
no aggregation

Page 54
EOQ Models for Production Planning
} This model is an extension of the EPQ model
} Consider the problem of producing many products in a single
facility. The facility may produce only one product at a time.
} For example, if there are three products A, B and C, then a
production sequence under the rotation cycle policy is A, B,
C, A, B, C, ….
} Determine the optimal procedure for producing n products on
the machine to minimize the cost of holding and setup, and to
guarantee that no stock-outs occur during the production cycle.
} For the problem to be feasible we must have that
l
n
å j £ 1.
j =1 Pj
Page 55
EOQ Models for Production Planning (cont’d)
} Rotation cycle policy: In each cycle there is exactly
one setup for each product, and products are produced
in the same sequence in each production cycle.
} Finding optimal production quantity of each product
separately using the EPQ formula
2K j l j
Q =
*
j
h 'j where hj¢= hj (1- lj/Pj)
} It is likely that some of the lot sizes Qj will not be
large enough to meet the demand between production
runs for product j, thus resulting in stock-out.

Page 56
EOQ Models for Production Planning (cont’d)
} Assume that exactly one lot of each product is
produced between time T (cycle time)
} The lot for product j must be large enough to meet the
demand during time T: Q j = l jT
} The average annual cost for all products:
n n é l ù
å G (Q j ) = å ê K j ( Q ) + h¢ jQ j / 2 ú.
j
j =1 j =1 ë j û

Page 57
EOQ Models for Production Planning (cont’d)
} Substituting T = Qj lj

G (T ) = å é K j + h¢ j l jT / 2 ù.
n

j =1 ê
ë T úû
¶G (T )
} The necessary condition for an optimal T is: =0
¶T
} Then,
2å K j
j
T = *

å jl j
h
j
'

Page 58
EOQ Models for Production Planning (cont’d)
} If setup time is a factor, let sj be the setup time for
product j, the following constraint should be satisfied:
n

å (s + Q / P ) £ T .
j =1
j j j

} Using the fact Q j = l jT and rearrange,

n
åsj
j =1
Thus T ³ n = T min .
1 - å (l j / P j )
j =1

} Therefore, the cycle time T is: T = max {Tmin ,T * }

Page 59
Example
} Bali produces several styles of men’s and women’s shoes at a
single facility near Bergamo, Italy.
} Rotation policy used to schedule a single cut machine
} Setup costs $110 per hour, holding costs are based on 22%
annual interest rate, and the relevant data are as follows:
Style Annual Production rate Setup time Variable cost
Demand (units/year) (hours) $/unit
Women’s pump 4520 35800 3.2 40
Women’s loafer 6600 62600 2.5 26
Women’s boot 2340 41000 4.4 52
Women’s sandal 2600 71000 1.8 18
Men’s wingtip 8800 46800 5.1 38
Men’s loafer 6200 71200 3.1 28
Men’s oxford 5200 56000 4.4 31
Page 60
} Verify the problem’s feasibility
} Compute the value of optimal cycle time
} Obtain setup costs and modified holding costs, and then get
T*=0.1529.
Setup cost Kj Modified holding cost hj’ Optimal lot size for each
production run
352 7.69 691
275 5.12 1009
484 10.79 358
198 3.81 398
561 6.79 1346
341 5.62 948
484 6.19 795
Page 61
} Assume 8 working hours per day and 250 working
days per year
} Tmin=0.04. thus T* =0.1529> Tmin is feasible and,
hence, optimal.
} The total average annual cost of holding and setups at
an optimal policy is G(T*)=$35,244.44.

} The relatively simple solution is the result of two

assumptions:
} The setup cost is not sequence dependent.
} The plant uses a rotation cycle policy.
Page 62