8/13/2019 4 Distribution Inventories
1/46
8/13/2019 4 Distribution Inventories
2/46
8/13/2019 4 Distribution Inventories
3/46
8/13/2019 4 Distribution Inventories
4/46
8/13/2019 4 Distribution Inventories
5/46
8/13/2019 4 Distribution Inventories
6/46
8/13/2019 4 Distribution Inventories
7/46
8/13/2019 4 Distribution Inventories
8/46
8/13/2019 4 Distribution Inventories
9/46
8/13/2019 4 Distribution Inventories
10/46
8/13/2019 4 Distribution Inventories
11/46
8/13/2019 4 Distribution Inventories
12/46
8/13/2019 4 Distribution Inventories
13/46
Inventory 13
Calculations with the EOQ costconstant
Let unit price = $10, annual demand = 100 units
Low investment, high workload policyK = 1Q = K (demand in $) 1/2
= 1 ($10 * 100) 1/2 = $31.62
Avg. investment = order qty./2
= $31.62/2 = $15.81
Workload = demand/order qty.= $1000/$31.62 = 31.62 orders
8/13/2019 4 Distribution Inventories
14/46
Inventory 14
Calculations with the EOQ costconstant (cont.)
High investment, low workload policyK = 6Q = K (demand in $) 1/2
= 6 ($10 * 100)1/2
= $189.74
Avg. investment = order qty./2= $189.74 / 2 = $94.87
Workload = demand/order qty.= $1000 / $189.74 = 5.3 orders
8/13/2019 4 Distribution Inventories
15/46
Inventory 15
Tradeoffs between investment andworkload
Avg. invest. = Workload =K order qty./2 demand/order qty.1 $15.81 31.62 orders2 31.62 15.8
3 47.43 10.54 63.24 7.95 79.04 6.36 94.86 5.3
6.32 100.00 5.08 126.48 4.010 158.10 3.520 316.20 1.6
8/13/2019 4 Distribution Inventories
16/46
Inventory 16
Tradeoffs between investment andworkload (cont.)
$
300
250 The optimalpolicy curve
200
150
100
50
0 5 10 15 20 25 30
Workload
A v g .
i n v e s
t m e n
t
8/13/2019 4 Distribution Inventories
17/46
Inventory 17
Optimal policies for multi-iteminventories
Given only the sum of square roots of demand in $, you cancompute aggregate workload and investment.
Read as the sum of:
Investment formula
Single-item Multi-item
Q$ = K (demand in $) 1/2 Q$ = K (demand in $) 1/2 Q$ = K (demand in $) 1/2
Q$ / 2 = (K/2) * (demand in $) 1/2 Q$ / 2 = K/2 (demand in $) 1/2
8/13/2019 4 Distribution Inventories
18/46
Inventory 18
Optimal policies for multi-iteminventories (cont.)
Workload (F) formulas
Single-item Multi-item
F = (demand in $) / Q $ F = (demand in $) / Q $
F = (1/K) * (demand in $) 1/2 F = 1/K (demand in $) 1/2
8/13/2019 4 Distribution Inventories
19/46
Inventory 19
Multi-item example
5,000 line-item inventory
(demand in dollars) 1/2 = $250,000
K K/2 Investment 1/K Workload1 0.5 $125,000 1.0 250,000 orders2 1.0 250,000 0.5 125,0005 2.5 625,000 0.2 50,00010 5.0 1,250,000 0.1 25,000
8/13/2019 4 Distribution Inventories
20/46
Inventory 20
Multi-item example (cont.)
For K = 5:
avg. investment = Q$ /2
Q$ /2 = (K/2)
(demand in $)
1/2
= 2.5 * 250,000= 625,000
workload = 1/K (demand in $) 1/2
= 0.2 * $250,000= 50,000 orders
8/13/2019 4 Distribution Inventories
21/46
Inventory 21
Achieving management goals forinvestment
Goal = Q $ /2
Goal = (K/2) * (demand in $) 1/2
Solving for K yields:
K = (2 * goal) / (demand in $) 1/2
This value of K meets the investment goal exactly.
The workload for that K is: F = (1/K) * (demand in $) 1/2
= ( (demand in $) 1/2 )2 / (2 * goal)
8/13/2019 4 Distribution Inventories
22/46
Inventory 22
Average inventory behavior withuncertain demand
S t o c k
Day12 160 4 8 20 24 28
On hand
Avg. inv.
ROP
SS
01
23
45
6
789
10
11
12
8/13/2019 4 Distribution Inventories
23/46
Inventory 23
Average inventory behavior withuncertain demand (cont.)Demand = 1 unit per dayLeadtime = 2 daysLeadtime demand (LTD) = 2 units
Safety stock (SS) = 2 unitsReorder point (ROP) = LTD + SS = 4 unitsOrder quantity (Q) = 10 unitsMaximum inventory = Q + SS = 12 units
Avg. investment = Q/2 + SS = 7 units
8/13/2019 4 Distribution Inventories
24/46
Inventory 24
Reorder point with uncertaindemand
AssumptionLength of leadtime is constant
ConceptsReorder point = mean demand + safety
during leadtime stock
standard deviation
Safety stock = safety factor * of forecast errorsduring leadtime
8/13/2019 4 Distribution Inventories
25/46
Inventory 25
Reorder point with uncertaindemand (cont.)
The standard deviation is a measure of variability of theforecast errors.
With a perfect forecast, the standard deviation is zero.
As forecast accuracy gets worse, the standard deviation getslarger.
The larger the safety factor, the smaller the risk of running outof stock.
8/13/2019 4 Distribution Inventories
26/46
Inventory 26
Safety factor and probability of shortagez P(z) z P(z)
0.00 0.50000 2.30 0.010720.10 0.46017 2.40 0.008200.20 0.42074 2.50 0.006210.30 0.38209 2.60 0.004660.40 0.34458 2.70 0.003470.50 0.30854 2.80 0.002560.60 0.27425 2.90 0.001870.70 0.24196 3.00 0.001350.80 0.21186 3.10 0.000970.90 0.18406 3.20 0.000691.00 0.15866 3.30 0.000481.10 0.13567 3.40 0.000341.20 0.11507 3.50 0.000231.30 0.09680 3.60 0.000161.40 0.08076 3.70 0.000111.50 0.06681 3.80 0.000071.60 0.05480 3.90 0.00005
1.70 0.04457 4.00 0.000031.80 0.03593 4.10 0.000021.90 0.02872 4.20 0.000012.00 0.02275 4.30 0.000012.10 0.01786 4.40 0.000012.20 0.01390 4.50 0.00000
8/13/2019 4 Distribution Inventories
27/46
Inventory 27
Probability of shortage on one order cycle
P(Z) = Probability demand will exceed Z standard deviationsof safety stock on one order cycle
Example:
Mean demand during LT = 100 unitsStd. dev. = 20 unitsSafety factor (Z) = 1.5Reorder point = mean demand + Z (std. dev.)
during LT= 100 + 1.5 (20)= 130 units
From table, P(Z) = .06681ROP.xls
8/13/2019 4 Distribution Inventories
28/46
Inventory 28
Probability of shortage on one ordercycle (cont.)
From table, P(Z) = .06681
.50 .43319
Z0 1.5
X100 130
8/13/2019 4 Distribution Inventories
29/46
Inventory 29
Number of annual shortageoccurrences (SO)
The probability of shortage on one order cycle is misleadingsince the ordering rate can vary widely across theinventory. A better measure of customer service is thenumber of annual shortage occurrences.
Probability of Number ofSO = shortage on one * annual
order cycle order cycles
8/13/2019 4 Distribution Inventories
30/46
Inventory 30
Number of annual shortageoccurrences (SO) (cont.)
Example:
Safety factor = 1.5
Probability = .06681
Annual demand = 1000 units
Order quantity = 50 units
Number of annual order cycles = 1000/50 = 20
SO = .06681 * 20 = 1.34
8/13/2019 4 Distribution Inventories
31/46
Inventory 31
Units or dollars backordered as aservice measure
E(Z) = Expected units backordered for a distributionwith mean = 0 and standard deviation = 1 onone order cycle
E(Z) = Expected units backordered for a distributionwith mean = X and standard deviation = on one order cycle
8/13/2019 4 Distribution Inventories
32/46
Inventory 32
Units or dollars backordered as aservice measure (cont.)
Example:
Annual demand = 1000 unitsOrder quantity = 50 unitsX = 25 units = 5Z = 1.2From table, E(Z) = .05610
Reorder point = 25 + 5 (1.2) = 31
Units short per cycle = .05610 (5) = .2805
Annual order cycles = 1000/50 = 20
Units short per year = 20 (.2805) = 5.61
8/13/2019 4 Distribution Inventories
33/46
Inventory 33
Quiz #1: Computing customerservice measures
Given: Annual demand = $2,000Order quantity = $100Mean demand during leadtime = $80Standard deviation = $60
Suppose we want the probability of shortage on one order cycle to be.09680. Compute the following:
Safety stockReorder point
Number of annual shortage occurrencesDollars backordered during one order cycleDollars backordered per year
Average cycle stock investment Average inventory investment
8/13/2019 4 Distribution Inventories
34/46
Inventory 34
Quiz #2: Computing customerservice measures
For the same data as the previous problem, what reorderpoint will yield:
a. 3 shortage occurrences per year?
b. 5% of annual sales backordered?
8/13/2019 4 Distribution Inventories
35/46
Inventory 35
Inventory tradeoff curves
A variety of different workload and investment combinationsyield exactly the same customer service.
To develop a tradeoff curve for dollars backordered:
1. Compute and plot the optimal policy curve showing tradeoffsbetween cycle stock investment and workload.
2. Select a percentage goal for dollars backordered.
3. For each workload, compute the safety stock needed to meet thegoal.
4. Add cycle stock to safety stock to obtain total investment.
8/13/2019 4 Distribution Inventories
36/46
Inventory 36
Inventory tradeoff curves (cont.)
$ Isoservice curve -- eachpoint yields the same dollars backordered
Safetystock
Optimalpolicy orcycle stock curve
Workload
I n v e s t m e n
t
8/13/2019 4 Distribution Inventories
37/46
Inventory 37
U.S. Navy application of tradeoffcurves
Inventory system8 Naval supply centers
Average inventory statistics at each center80,000 line items$25 million investment
Budget constraint Average investment limited to 2.5 months ofstock
8/13/2019 4 Distribution Inventories
38/46
Inventory 38
U.S. Navy application of tradeoffcurves (cont.)
Investment allocation strategies
Old NewSafety stock 1.5 months 1.0
monthsCycle stock 1.0 months 1.5months
Total 2.5 2.5
Results of new allocationReordering workload cut from 840,000 to 670,000 per year
$2 million annual savings in manpower
8/13/2019 4 Distribution Inventories
39/46
Inventory 39
Workload/service tradeoffs
C u s t o m e r s e r v
i c e
New policy Previous policy
90%
85%
80%
75%
0.0 0.5 1.0 1.5 2.0
Safety stock (months)
112 121 184 240 289
Workload (thousands of orders)
(Total investment fixed at 2.5 months)
8/13/2019 4 Distribution Inventories
40/46
Inventory 40
Strategic problems in managingdistribution inventories
1. Controlling inventory growth as sales increase
2. Controlling inventory growth as new locations areadded
3. Push vs. pull decision rules
4. Continuous review of stock balances vs. periodic review
5. Choosing a customer service measure
8/13/2019 4 Distribution Inventories
41/46
Inventory 41
Inventory growthInventories should grow at slower rate than sales.
Why? Order quantities are proportional to the square root of sales.
Example:One inventory itemK = 1Q$ = K (demand in $) 1/2
Sales Average Investment
Sales Growth Q$ Investment Growth100 --- 10 10/2 = 5 ---200 100% 14 14/2 = 7 40%
8/13/2019 4 Distribution Inventories
42/46
Inventory 42
Effects of adding inventory locationsInventories must increase as new locations are added.
One reason is that forecasting is easier when customer demands areconsolidated. Thus forecast errors are smaller and less safety stockis required.
Another reason stems from the EOQ:One inventory itemSales of $100K = 1
With one location:Q$ = K (demand in $) 1/2 Q$ = 1 (100) 1/2 = 10
Average investment = 10/2 = 5
8/13/2019 4 Distribution Inventories
43/46
Inventory 43
Effects of adding inventory locations(cont.)
With two locations:Location 1: Q $ = 1 (50) 1/2 = 7.07Location 2: Q $ = 1 (50) 1/2 = 7.07
Average investment = (7.07 + 7.07) / 2 = 7.07
Investment increase = 7.05 5 = 2.05Percentage increase = 2.05 / 5 = 41%
8/13/2019 4 Distribution Inventories
44/46
Inventory 44
Continuous review vs. periodic reviewsystemsContinuous review
Check stock balance after each transaction
If stock on hand below reorder point, place new order in a fixedamount
Periodic reviewCheck stock balance on a periodic schedule
If stock on hand below reorder point, place new order:in a fixed amount, or
in a variable amount (maximum level on hand)
Investment requirementsPeriodic review always requires more investment than continuousreview to meet any customer service goal.
8/13/2019 4 Distribution Inventories
45/46
Inventory 45
Push vs. pull control systems
Push or centralized systemCentral authority forecasts demand, sets stock levels,and pushes stock to each location.
Pull or decentralized systemEach location forecasts its own demand and sets its ownstock levels.
Investment requirements A pull system always requires more investment than apush system to meet any customer service goal.
8/13/2019 4 Distribution Inventories
46/46
Inventory 46
Comparison of shortage values
Inventory statistics5,790 line items$45 million annual sales
Shortage values at investment constraint of $5 millionShortage value Shortage Dollarsminimized occurrences backordered
shortage occurrences 1,120 $3.46 milliondollars backordered 2,812 1.48