IE3120 Manufacturing Logistics

Inventory Control Part 1:

Economic order quantity models

Department of Industrial and Systems Engineering

National University of Singapore

1
Outline
 Inventory Systems
 Relevant Cost Components
 Basic EOQ Model
 Quantity Discount Models
 Resource Constrained Multi-Product
Systems

2
Inventory Systems
 Types of inventory
◦ Raw Materials
◦ Work-in-Process
◦ Finished Goods
 Why Hold Inventories?
◦ Economies of Scale of Production
◦ Uncertainties
◦ Speculation (Seasonal demands)
◦ Transportation
◦ Production Smoothing

3
Inventory System Characteristics
 Product demand:
◦ Deterministic: constant vs. variable
◦ Uncertain: Stochastic
 Replenishment order quantity
 Replenishment cycle time
 Replenishment lead-time
◦ Time from order placement till arrival
◦ Zero, constant, variable, uncertain…
 Inventory Review Schemes:
◦ Continuous review, periodic review
4
 Defining Inventory Model:
Relevant Performance (Cost) Measures
 Purchase order Cost
◦ Item unit cost c
 Holding Costs
◦ Consists of:
 Space and resource
 Taxes and insurance
 Opportunity
◦ Express as percentage i of item unit cost
◦ Holding cost per item per unit time: h = i * c
 Fixed order Costs
◦ Fixed cost: per shipment: K
 Penalty Costs on demand under-fulfillment
◦ Shortage, backorder, loss-of-goodwill

5
Basic Economic Order Quantity (EOQ) Model

 Demand is constant: λ units per unit time

 No shortages are allowed
 Zero replenishment lead time
 Cost components:
◦ Fixed order cost: K per order placed
◦ Variable order cost: c per unit product
◦ Holding cost: h per unit product per unit time
 Decision problem: how much to replenish?
When to replenish?
6
Basic EOQ System Behavior
 Inventory levels over time

7
Basic EOQ Model: Formulation
 Average Annual Costs G(Q)
 G(Q)=annual ordering cost + annual
purchase cost + annual holding cost
K  cQ hQ K  cQ hQ K  hQ
G(Q )       c 
T 2 Q/ 2 Q 2

8
Annual Cost Function: Graphical View

9
Basic EOQ Model: Algebraic Solution
 We want to find Quantity Q* to minimize G(Q)

 In order to minimize G(Q), consider:

G '(Q )  K  / Q 2  h / 2
and G "(Q )  2K  / Q3  0 for Q  0

 The optimal value of Q:

2K 
G '(Q )  0  Q* 
h

10
Basic EOQ Model: Example
 Pencils are sold at a rate of 60 per week;
 The pencils cost 2 cent each and sell for
15 cent each;
 It costs $12 to initiate an order;
 Holding cost are based on an annual
interest rate of 25 percent.
Question: the order quantity? the
replenishment cycle time? yearly holding
and ordering costs?
11
Basic EOQ Model: Solution
 The annual demand: λ = (60)(52) = 3,120
 The holding cost: h = I * c = (0.25)(0.02) = 0.005
 The optimal order quantity:
2K  2  12  3,120
Q*    3,870
h 0.005
 The replenishment cycle time:
T = Q/λ =3,870/3,120 = 1.24 years;
 The annual holding cost:
= h(Q/2) = 0.005(3,870/2) = $ 9.675
 The annual ordering cost:
=Kλ/Q = 12(3,120)/3,870 = $9.675
12
 Finite Replenishment Lead Time:

 Lead time: τ < Replenishment cycle: T

Suppose in the same example that the pencils take four months to
ship.

Reorder-point R: τ * λ
R = τ * λ = (0.3333) (3,120)= 1,040
13
Lead Time > Replenishment Cycle
 Suppose now the delivery lead time τ is 32 months (2.67 years),
and T = 1.24 years.
 Note that 2.67 = 2*(1.24) + 0.19
 Re-order point = 0.19*λ = 0.19*(3120) = 592.8
 In general, Re-order point = Remainder (τ/T) * λ

14
Finite Production Rate
 Assume now items are internally
produced at production rate P.
 Assume P > λ.
 Replenishment quantity Q = P*T1, where
T1 is the production time in each cycle.

15
Finite Production Rate P
 Assume finite production rate P > λ.

 T: Replenishment Cycle
 T1: Production uptime
 T2: Production downtime.
16
Finite Production Rate: Cost Model

 H: maximum level of on-hand inventory

 H = (P- λ)*T1 , T1 = Q/P ,
 G(Q) = K/T + h*H/2
= K λ / Q + Q/2*h(1- λ/P)
= K λ / Q + Q/2*h’

2K 
 Q* 
h'

17
 Quantity Discount Models:
All-Units Discount
 The Company has the following price schedule for
its large trash can liners:
0.30Q for 0  Q  500

C(Q )  0.29Q for 500  Q  1000
0.28Q for 1000  Q


18
 Average annual cost functions
 The average annual cost functions are
given by

G j (Q)   c j   K / Q  Ic jQ / 2 for j  0,1, and 2.

G 0 (Q) for 0  Q  500,

G (Q)   G1(Q) for 500  Q  1000 ,
G (Q) for 1000  Q
 2

19
All-units Discount average annual cost function

20
All-units Discount: Example and
Optimal Solution
 EOQ values corresponding to each
unit costs:
2K  2(8)(600)
Q0    400
Ic0 0.2(0.3)
2K  2(8)(600)
Q 
1
  406
Ic1 0.2(0.29)
2K  2(8)(600)
Q 
2
  414
Ic2 0.2(0.28)

Which are feasible?

21
 All-units Discount: Example and Optimal
Solution

 An EOQ is feasible if it falls within the

interval that corresponds to the correct unit
cost.

 Q0 = 400  [0,500), Q0 is realizable;

 Q1 = 406  [500,1000), Q1 is not realizable;
 Q2 = 414  [1000,+∞), Q2 is not realizable;

22
There are three candidates for the optimal solution: 400(Q0), 500,
1000.
G(400) = G0(400) = $204.00
G(500) = G1(500) = $198.10
G(1000) = G2(1000) = $200.80
Hence the optimal order quantity is 500 units at an annual cost of
$198.10.

23
Summary of Solution Procedure
1. Determine largest realizable EOQ value.
2. Compare cost under largest realizable
EOQ and at all price-breakpoints larger
than the EOQ.
3. Select Order Quantity yielding minimum
costs.

24
Incremental Discounts:
 Consider previous Example, but assume
incremental quantity discounts:
 Bags cost 30 cents each for quantities of 500 or
fewer;
 For quantities between 500 and 1000,
◦ the first 500 cost 30 cents
◦ and the remaining amount cost 29 cents each;
 For quantities of 1000 and over,
◦ the first 500 cost 30 cents each,
◦ the next 500 cost 29 cents each,
◦ and the remaining amount cost 28 cents each.
25
Incremental Discounts:

26
 Incremental Discounts: Example
 Cost formula for incremental discounts
0.30Q for 0  Q  500

C(Q )  150  0.29(Q  500)  5  0.29Q for 500  Q  1000
295  0.28(Q  1000)  15  0.28Q for 1000  Q

so that
0.30 for 0  Q  500

C(Q ) / Q  0.29  5 / Q for 500  Q  1000
0.28  15 / Q for 1000  Q


27
 Incremental Discounts: Example and
Solution Procedure
The average annual cost function G(Q):

G(Q)  C(Q) / Q  K  / Q  I [C(Q) / Q]Q / 2

 G(Q) has three different algebraic

representations [G0(Q), G1(Q), and G2(Q)]
depending on C(Q)/Q

28
 Incremental Discounts: Example
and Solution Procedure

G0 (Q)  (600)(0.30)  (8)(600) / Q  (0.20)(0.30)Q / 2

G1(Q)  (600)(0.29  5 / Q)  (8)(600) / Q  (0.20)(0.29  5 / Q)(Q / 2)

G2 (Q)  (600)(0.28  15 / Q)  (8)(600) / Q  (0.20)(0.28  15 / Q)(Q / 2)

29
 Incremental Discounts: Average Cost
Function

G(Q)  C(Q) / Q  K  / Q  I [C(Q) / Q] / 2

 G(Q) has three different algebraic G2

representations [G0(Q), G1(Q), and G2(Q)]
G0 G1

30
 Incremental Discounts: Example and
Solution Procedure
1. Computing the three minima of the average annual
cost curves;
G0 (Q )  (600)(0.30)  (8)(600) / Q  (0.20)(0.30)Q / 2
2K  (2)(8)(600)
 Q0    400;
Ic0 (0.20)(0.30)
G1(Q )  (600)(0.29  5 / Q )  (8)(600) / Q  (0.20)(0.29  5 / Q)(Q / 2)
 (0.29)(600)  (13)(600) / Q  (0.20)(0.29)Q / 2  (0.20)(5) / 2
(2)(13)(600)
 Q1   519;
(0.20)(0.29)
G2 (Q )  (600)(0.28  15 / Q )  (8)(600) / Q  (0.20)(0.28  15 / Q)(Q / 2)
 (0.28)(600)  (23)(600) / Q  (0.20)(0.28)Q / 2  (0.20)(15) / 2
(2)(23)(600)
 Q2   702;
(0.20)(0.28)

31
 Incremental Discounts: Example and
Solution Procedure
2. Determining which of these minima falls into the
correct interval;
Both Q0 and Q1 are realizable. Q2 is not realizable
because Q2 < 1000.
3. Finally, comparing the average annual costs at the
realizable values.
min{G0 (Q0 ),G1(Q1 )}  min{$204.00,$204.58}  $204.00
 Q0 is the optimal order quantity.

32
Resource Constrained Multiple Product
Systems
 Constraints that may make EOQ
infeasible, eg: not enough space to store
products, not enough money to purchase
that many at one time…

33
Resource Constrained Multiple Product
Systems: Example
 Three items are produced in a small
fabrication shop. The shop
management has established the
requirement that the shop never
have more than $30,000 invested in
the inventory of these items at one
time.

34
Resource Constrained Multiple
Product Systems: Example
 The management uses a 25 percent annual interest
charge to compute the holding cost. The relevant cost
and demand parameters are given in the following table.
What lot sizes should the shop be producing so that
they do not exceed the budget?

Item

1 2 3

Demand rate λj 1850 1150 800

Variable Cost Cj 50 350 85

Setup cost Kj 100 150 50

35
Resource Constrained Multiple Product
Systems
 Suppose that n items have unit weights of
w1, w2, …, wn, respectively, and the total
budget available for them is W.
w1Q1 + w2Q2 +…+ wnQn ≤ W.
 Let
2K i i
EOQi  for i  1,..., n
hi
where Ki, hi, and λi are the respective cost
and demand parameters.

36
Resource Constrained Multiple Product
Systems
 If n

 w EOQ
i 1
i i W

then the optimal solution is Qi = EOQi

 If
n

 w EOQ
i 1
i i W
What can be
done now??

*
 Let i be the correct optimal solution. Then we
Q
must have: n

 i i W
w Q
i 1
*

37
Resource Constrained Multiple Product Systems

 Such a problem can be solved by formulating the

Lagrangean function. The optimal lot sizes are of the
form:
2K i i
Qi  *
hi  2 w i
where  > 0 and is so that
n

 i i W
w Q
i 1
*

38
Resource Constrained Multiple Product Systems:
Special Case
 Suppose: w1/h1 = w2/h2 = … = wn/hn=w/h

Qi  2K i i  2K i i / hi
(hi  2 w i ) (1  2 w i / hi )
 EOQi 1
 EOQi * m
(1  2 w / h)

39
Resource Constrained Multiple Product Systems:
Special Case

 Substituting in resource-constraint,
n n
 w i Qi   w i m *EOQi  W
i 1 i 1
W
m n
 w i * EOQi
i 1

40
Back to Example

 3 items have unit costs of c1, c2, …, cn,

respectively, and the total budget available
for them is C.
c1Q1 + c2Q2 + c3Q3 ≤ C.
 Let
2K i i
EOQi  for i  1,...,3
hi
where Ki, hi, and λi are the respective cost
and demand parameters.

41
 Example: Solution
1. Compute the EOQ values for all items to determine
whether or not the budget constraints is active.
(2)(100)(1850)
EOQ1   172
(0.25)(50)
(2)(150)(1150)
EOQ2   63
(0.25)(350)
(2)(50)(800)
EOQ3   61
(0.25)(85)
 Total investment:
(172)(50)  (63)(350)  (61)(85)  35835  30000

42
 Example: Solution
 Note that in this case weights wi are the
unit costs ci and holding cost
hi = ci *interest rate
we have h1/w1 = h2/w2 = h3/w3
 Therefore:

Qi*  mEOQi
n 
where m  C /   (ci EOQi )
 i 1 

43
 Example: Solution
n 
m  C /   (ci EOQi )
 i 1 
m  30000 / 35835  0.8372
Q1*  (172)(0.8372)  144; Q2*  (63)(0.8372)  52;
Q3*  (61)(0.8372)  51

• Total budget required for these lot sizes is $29735.

• Allocate remaining budget: increase the lot sizes of
product 1 by 3 units and product 3 by 1 unit, resulting
in a total budget of $29970.

44
 Appendix: Lagrangean Relaxation

 Consider Z = Minimize f(x) s.t. c’x <= C

 For any θ>0, consider the relaxed problem:
LD(θ) = Minimize L(x, θ) ,
where L(x, θ) = f(x) + θ(c’x - C)
 Clearly, LD(θ) <= Z

 Idea is to find θ that maximize LD(θ)

 Maxθ LD(θ) = Maxθ {Minx f(x) + θ(c’x - C)}

45
 Lagrangean Relaxation

 Idea is to find a stationary point (θ*,x*) on L(x, θ)

such that: it is a maxima w.r.t to θ and minima w.r.t. x
 Formally this is known as the saddle-point conditions for
optimality

 First order-conditions:
L
 0 for all i = 1, ... , N
Qi
L
0


46
 Lagrangean Relaxation

L(Q, )   K    c  hQ     cQ W 
N N
i i i i
i i i i 
Qi 1 i
2  i 1 
N K i i (h  2 w i )Qi
  i ci  i  W
i 1 Qi 2

Stationary conditions
L  0 L  0
Qi 
  Ki   (hi  2 w i )  0  N 
   w iQi W 0
i

Qi2 2 
 i 1 
 Qi  2K i i N

(hi  2 w i )   w iQi  W
i 1

47