Professional Documents
Culture Documents
Jerry Banks
Forms of inventory
• Inventory appears in the supply chain as
– Raw materials
– WIP
– FGI
• Must take into account the interactions at
various levels in the supply chain
Customers,
Field demand
Sources: Regional Warehouses: centers
plants Warehouses: stocking sinks
vendors stocking points
ports points
Supply
Inventory &
warehousing
costs
Production/
purchase Transportation Transportation
costs costs costs
Inventory &
warehousing
© Copyright 1999 D. Simchi-Levi, P. Kaminshy & E. Simchi-Levi costs
Why hold inventory?
• To protect from variations in demand
– Uncertainty has increased recently
• Short life cycles and an increasing number of
products implies that historical data about demand
may not be available
• One product group may have many different forms
Why hold inventory?
• Supplier unreliability
– Quantity delivered
– Quality of those delivered
– Lead time is variable
Why hold inventory?
• Economies of scale
– Shipping costs are not linear
– Quantity discounts
Managing inventory can be difficult
Avg Inv
Time
T
Costs for one cycle
• Ordering cost
–K
• Holding cost
– hT [Q/2]
• Total cost
– K + hT[Q/2]
Total cost per day
• TC = K/T + h [Q/2]
• But, T = Q/D
• So, TC = [KD]/Q + h[Q/2]
Finding the optimal
• Take the derivative of TC wrt Q and set it to
zero
– d(TC)/dQ = -[KD]/Q2 + h/2 = 0
• Solve for Q
– [KD]/Q2 = h/2
– Q2 = 2KD/h
– Q* = SQRT(2KD/h)
Example
• Mug sales
• D = 20/week
• K = $12
• Lead time = 0
• h = .25 of inventory value
• Mugs cost $1, sell for $5
• Find Q*
Solution
• Q* = SQRT{2(12)(20)]/[(.25)/52]}
• Q* = 316
Relative Insensitivity of Q
Q TC/week
250 1.560962
275 1.533785
300 1.521154
316 1.519109
325 1.519712
350 1.52706
375 1.541442
Even when Q is far away
• Q = 200
– TC = $1.68
• Q = 400
– TC = $1.56
Shaped like the letter U
1.7
1.65
1.6
1.55 TC
1.5
1.45
1.4
200 250 300 350 400
But, demand is uncertain
• Even though many firms treat it as
predictable
– Production decisions are made on the basis of
forecasts made far in advance
• Forecasts are always wrong
• The longer the horizon, the worse the forecast
• Aggregate forecasts are more accurate
Case: Swimsuit production
• Six months in advance, production
quantities are set
• Overestimating results in unsold units
• Underestimating demand leads to shortage
costs
• Historical data is available for the past five
years
Demand forecast
30% 28%
25%
22%
20% 18%
15%
11% 11% 10%
10%
5%
0%
8000 10000 12000 14000 16000 18000
Unit sales
Additional information
• Fixed production cost of $100,000
• Variable production cost of $80/unit
• Sales price of $125/unit
• Salvage value of $20/unit
Profit model
• D = Demand, Q = Production Quantity
– Q > D, 125D - 80Q + 20 (Q - D) - 100000
– = 105D - 60Q - 100000
– D > Q, 125Q - 80Q - 100000
– = 45Q - 100000
For example
• D = 12000, Q = 10000
– Profit = 45(10000) - 100000 = 350000
• D = 8000, Q = 10000
– Profit = 105(8000) - 60(10000) -100000 =
140000
When Q = 12000
300000
250000
Exp Profit
200000
150000
100000
8000 10000 12000 14000 16000 18000
Production Quantity
Insight
• Since there is a marginal profit of $45 for each
unit short, but a marginal loss of $60 for each unit
of excess production, the production should be
less than the average demand
– Average demand is .11(8000) + …+ .10(18000)
– Or, 12008
– Profit when Q = 10000 is $326900
– Profit when Q = 12000 is $326700
Questions
• Profit for Q = 8000 is about the same as
profit for Q = 15000
• Which one would you choose?
• Profit for Q = 16000 has a .11 probability of
a loss of $220,000 (if D = 8000)
• What about that?
More insights
• Q* is not necessarily equal to average
demand
– Depends on the relationship between marginal
profit and marginal cost
• As Q increases, average profit typically
increases until Q reaches a maximum then
begins to increase
More insights
• As Q increases, the probability of large
losses increases
– At the same time, the probability of large gains
also increases
More interesting with shortage cost
Inventory position
S
s
0 Time
Lead time
Example
• Continue the previous example and assume
that there is a fixed ordering cost of $4500
– Independent of the order size
• Cost of a TV set is $250
• Annual inventory holding cost is 18% of
$250, or, on a weekly basis
– [.18*250]/52 = .87
Example
• Order quantity, Q
– Q = SQRT{[2*4500*44.46]/.87} = 679
• Order up-to-level, S
– S = 679 + 86 = 765
• The distributor should place an order to
raise the inventory position to 765 TV sets
whenever the inventory is below 175 units
Which system is better and why?
Market One
Supplier Warehouse
Market Two
Risk pooling
• Which of the systems will require more
inventory?
– Why?
• With the same total inventory level, which
of the systems provides better service?
– Why?
Compare the two systems shown
in the earlier figure with:
– Two products
– Maintain a 97% service level
– Order cost is US$60
– Holding cost is US$.27/unit/week
– Transportation cost (averages)
• Centralized system US$1.05/unit
• Decentralized system US$1.10/unit
– Lead time is 1 week
Data
Week 1 2 3 4 5 6 7 8
Prod A 33 45 37 38 55 30 18 58
Mkt 1
Prod A 46 35 41 40 26 48 18 55
Mkt 2
Prod B 0 2 3 0 0 1 3 0
Mkt 1
Prod B 2 4 0 0 3 1 0 0
Mkt 2
Prod A 79 80 78 78 81 78 36 113
Cntrl
Prod B 2 6 3 0 3 2 3 0
Cntrl
Descriptive statistics
Supplier
Warehouse
echelon
inventory
Warehouse
echelon lead
time Warehouse
Retailers
How much inventory at each location?