The case study optimizes the HP DeskJet printer supply chain by redesigning the network using component commonality and risk pooling. The redesign leads to considerable savings to the business.

1. 2
Executive Summary
Hewlett-Packard (HP) is an American multinational information technology company
headquartered in California. HP offers wide range of electronic products like computers, printers,
electronic peripherals etc. The DeskJet printer, one of HP’s most successful products, was
introduced in 1988. The company was facing an issue of alarming inventory levels of DeskJet
printers in Asia while having low service level in Europe. Involvement of multiple divisions within
the supply chain network like the production, material and distribution departments with
conflicting goals made it difficult to find an optimum solution. HP realized that they need to
redesign the supply chain to improve their profit margins as well as service level.
The report analyzes the problems in the current system and redesigns the network
to resolve the issues. In the new network, FAT has been moved to DC thus pushing the Customer
Order Decoupling Point (CODP) towards the later stage, generating savings of around 31.3% in
inventory holding cost. The change of transportation mode to air freight will reduce the lead
time, resulting in additional savings of 44.9%. However, these come at an initial investment in
assembly setup at DCs as well as an increase in logistics cost which are discussed in relevant
sections.
In addition, another approach of supply chain inventory planning (Guaranteed
Service Approach) is discussed. This can be a potential next step which identifies the best nodes
to hold safety stock instead of holding some amount at all nodes. Both the designs can then be
compared to identify the best solution that is in line with HP’s long-term strategy.

2. 3
Contents
Executive Summary......................................................................................................................................2
1 Introduction..........................................................................................................................................4
2 Current Supply Chain Design................................................................................................................5
2.1 Product Portfolio..........................................................................................................................6
2.2 Cost Structure...............................................................................................................................7
2.2.1 Penalty Cost..........................................................................................................................7
2.2.2 Holding Cost..........................................................................................................................8
2.3 Inventory Levels ...........................................................................................................................8
2.3.1 Inventory Level: DC Stage ....................................................................................................8
2.3.2 Inventory Level: IC to FAT Stages.........................................................................................9
2.4 Drawbacks of Current System....................................................................................................10
3 Supply Chain Redesign .......................................................................................................................10
3.1 Assembly and Localization at DC ...............................................................................................11
3.1.1 Risk Pooling.........................................................................................................................12
3.2 Mode of Transportation.............................................................................................................12
3.3 Product Portfolio Reduction ......................................................................................................13
3.4 Inventory Levels .........................................................................................................................13
3.4.1 Revised Cost Structure .......................................................................................................13
3.4.2 Inventory Level: DC Stage ..................................................................................................14
3.4.3 Inventory Level: IC to FAT Stages.......................................................................................15
3.5 Special Case ................................................................................................................................15
4 Conclusion and Results ......................................................................................................................16
5 Future Scope.......................................................................................................................................17

3. 4
1 Introduction
Hewlett-Packard (HP) was founded in 1939 and went on to become a major player in the
consumer electronics industry over the next fifty years. The entire organization was segmented
into six different units primarily based on its product groups ranging from printers and plotters
to network products. The following study deals with the inventory distribution of DeskJet
Printers, one of HP’s most successful products introduced by its Peripheral group in 1988. This
product alone accounted for a large portion of HP’s sales at that time.
The retail printer market was rapidly evolving in the 1980s wherein the dot matrix
technology was going obsolete and the laser technology still being expensive. The introduction
of the inkjet printers, which incorporated the best of both ends, seemed promising. Many players
wanted to take advantage of this budding technology including HP. As a result, the Vancouver
Division (US) was setup in 1979 to cater to the design and production of these inkjet printers.
The DeskJet printers produced at the Vancouver Division were shipped to the
distribution centers (DCs) across the three continents from where they were redistributed to
retailers. While the Vancouver division didn’t stock any finished products, higher inventory levels
were being maintained at Europe and Asia’s DCs owing to the longer shipment times and for
providing high service levels at the same time. Despite having such high levels, shortages were
common for some models while some other products were just getting piled up. Therefore, the
need for a statistically driven inventory system was realized.

4. 5
2 Current Supply Chain Design
The current supply chain of HP DeskJet printer followed a Make-to-Stock policy. The supply chain
followed a simple structure which consisted of suppliers, manufacturing stages (Printed Circuit
Assembly and Test(PCAT) and Final Assembly and Test(FAT)), distribution centers (DC), dealers
and customers as depicted in the figure below.
The raw materials required for manufacturing and assembly of the final DeskJet
printer were either acquired from suppliers based in North America or imported from other
suppliers worldwide. Once the raw materials were sourced, the manufacturing process took
place in two key stages in Vancouver namely PCAT and FAT. The PCAT manufacturing stage
involved the assembly and testing of the electronic components like ASICs, ROM and raw printed
circuit boards required to make logic boards and print head driver boards for the printers. Thus,
it comprised of the complete manufacturing of the printed circuit assemblies. The second stage
i.e. FAT, involved the assembly of printed circuit assemblies with other subassemblies like
motors, cables, keypads, plastic chassis and skins and gears along with the final testing of the
Figure 2.1: Current Supply Chain Design

5. 6
printer. As these products were shipped to Europe and Asia, localization process involving the
assembly of suitable power supply modules, power chords and user manuals were carried out in
this stage. The total time consumed in the above two stages was about a week.
Once the final printer was manufactured, it was shipped to the three different
distribution centers located in North America, Europe and Asia. The time required to ship the
products to the North America DC was one day, whereas the time required to ship the product
to the other two DCs took almost 4-5 weeks. The long shipment time to the DCs in Europe and
Asia was due to the ocean transit and the time it took to clear the customs and duties.
Different inventory levels were maintained at different stages in the supply chain.
There was no finished product inventory at Vancouver mainly because of the HP policy of
operating the factories in a pull mode. The Vancouver management was able to reduce the
uncertainties caused by delivery variabilities of incoming materials due to an effective supplier
management system. The distribution centers, however had to maintain a high level of finished
product inventory to meet customer demands which were highly variable in nature. The demand
of products was assumed to follow normal distribution with no bullwhip effect.
2.1 Product Portfolio
There were eight different models of DeskJet printers sold across three markets. The demand
across products in a specific location varied drastically, thus making the analysis of product
portfolio necessary. The analysis categorized the products into four groups at each of the three
locations.

6. 7
The groups were determined based on the volume and variability of demand (Exhibit 1):
1) High demand and high variability (HH) 2) Low demand and high variability (LH)
3) High demand and low variability (HL) 4) Low demand and low variability (LL)
Country/Categories HH LH HL LL
Europe AB, AQ, AU A, AA AY
Asia AB, AG, AU A, AK
North America AB, AG, AK, AU A
The products in high demand category had to be managed with utmost care since
any loss of demand for these products could incur significant loses. The products with high
variability account for a major share of the safety stock, thus need to be reviewed whether to
continue offering them or not.
2.2 Cost Structure
The cost structure assumed for the analyses based on estimates provided by HP management
are discussed below in detail.
2.2.1 Penalty Cost
The penalty cost was based on demand share of the product and loss of customer good will. As
majority revenue comes from high demand products, a loss of customer goodwill in this category
could incur a huge loss for the company. Thus, high penalty cost should be imposed for stockout
of these products. Based on this rationale, the penalty cost for products in the high and low
demand categories was taken to be 5 times and 3 times the unit cost respectively.
Table 2.1: Product Categorization

7. 8
2.2.2 Holding Cost
The inventory carrying cost to be used in safety stock analyses were estimated by management
in the range of 12% to 60% of unit cost. The various factors considered in the estimate were HP’s
cost of debt, warehouse expenses and ROI expected of new product development. Considering
that ROI of 60% was too ambitious, we assumed three different values of 28%, 48% and 60%
added to a 12% debt and warehouse expense rate and analyzed their impact on the inventory
levels.
2.3 Inventory Levels
Considering the various holding costs discussed before, sensitivity analysis was carried to
understand the effect of different holding costs on inventory level.
Country/Holding Cost 40% 60% 72%
North America 12098 (45.01%) 11324 (41.25%) 10962 (39.3%)
Europe 51092 (32.15%) 48750 (28.89%) 47652 (27.26%)
Asia 5866 (50.6%) 5427 (46.62%) 5222 (44.5%)
It was observed that as holding cost was increased, the cost-minimizing service
level reduced. Thus, the inventory levels and the safety stock percentage decreased
proportionally (Exhibit 2). In the current market scenario, it is reasonable to consider ROI as 48%
which gives a net holding cost of 60%.
2.3.1 Inventory Level: DC Stage
To maintain a 98% service level, which was HP’s target, inventory levels that were to be
maintained at DCs are shown below (Exhibit 3).
Table 2.2: Inventory Levels and Percentage Safety Stock

8. 9
Country/Inventory Total Inventory Level Safety Stock (in %)
North America 14422 54%
Europe 58148 40%
Asia 7251 60%
It is observed that for maintaining 98% service level, a very high safety stock level
must be maintained (around 40-60%). The problems leading to such high levels of safety stock
are discussed in next section.
2.3.2 Inventory Level: IC to FAT Stages
After analyzing the inventory levels at DCs, the optimal inventory levels at the remaining stages
were calculated. The estimates of the holding cost and lead time at each stage of the current
network are depicted below. The lead time for the distribution center in North America is 1 week
compared to that of 6 weeks in Europe and Asia.
Table 2.2: Inventory Levels and Percentage Safety Stock for 98% service level
Figure 2.2: Current Network - Europe and Asia
Figure 2.3: Current Network – North America

9. 10
Using the Shang-Song approach, echelon inventory levels were calculated based
on the lead times and holding cost as specified in the diagram above (Exhibit 4). The following
results were obtained:
Country/Echelon IC Manufacturing PCAT/FAT
North America 133293 (30.15%) 124747 (30.70%)
Europe 150219 (22.93%) 142925 (23.26%)
Asia 12535 (22.97%) 11926 (23.30%)
2.4 Drawbacks of Current System
In the current supply chain, the production of printers including localization is done at the
PCAT/FAT stage. Since the customer order decoupling point is at an early stage in the network,
any possibility of pooling the variability across products is eliminated. Thus, separate safety
stocks were being maintained for different models which in turn increased the total inventory
holding costs.
Another major problem with the current setup is the shipping time for the delivery
of the final products from PCAT/FAT stage to the DCs in Europe and Asia (4-5 weeks). Thus, a high
level of safety stocks must be maintained to account for the high lead time.
3 Supply Chain Redesign
Following the drawbacks discussed in the previous section, the supply chain network was
restructured. The recommended design moved FAT to the DC while changing the mode of
transportation to air freight. While the Application Specific Integrated Circuit (ASIC) which is used
to build the logic boards and print heads are unique to each product, other subassemblies can
Table 2.3: Echelon Inventory Levels

10. 11
mostly be used across products. Thus, to achieve component commonality for subassemblies it
was necessary to move the FAT stage to DCs. The revised structure created after a thorough study
mitigated the issue of long lead times and variance in demand of products. The entire production
stage of the supply chain remained unaltered with components like logic boards and print head
drivers still being produced at Vancouver. Few notable changes in design are discussed below.
3.1 Assembly and Localization at DC
The FAT stage responsible for the final assembly, testing and localization was moved to DC in
respective continents. This change requires HP to redesign components, logic boards and print
heads in a way that they can be plugged with minimal effort and expertise. A one-time
investment of setting up the assembly line along with the hiring of labor would be required at
the DCs. The subassemblies like motors, cables, key pads, plastic chassis and skins could be
procured from the local suppliers. The management needs to identify local vendors who can
provide high quality components at a lower cost.
Figure 3.1: New Supply Chain Design

11. 12
3.1.1 Risk Pooling
One major advantage with this redesign is component commonality. Currently, the components
of printers can be segmented into two groups – (a) Customized parts containing PCBs, ASICs and
print heads, and (b) Common parts like subassemblies, power supplies etc. The subassemblies
like keypads, cables, motors, power supply etc. being common to all the products can be used
across product lines. This results in the pooling of demand variance across products and thus
reduces the safety stock of subassemblies required to maintain the desired service level.
Moreover, DC will have to maintain relatively less inventory of finished products due to reduced
lead time from FAT stage. Thus, there is a significant decrease in inventory holding cost as the
storage of logic boards and print heads at FAT is less costlier than that of finished products.
Moving FAT to DC resulted in shift of CODP further downstream (to the DC). This
enabled pooling of demand variance for a given product across different locations thereby
reducing the safety stocks at the initial stages of the network (IC and PCAT).
Additional savings can be made by designing logic boards and print heads in a way
that they can be used across products. However, this will require investment of resources and
could be a potential area of improvement for HP.
3.2 Mode of Transportation
One of the major reasons for high inventory levels at Europe and Asia’s DC was long lead times.
In the proposed design, air freight is used as the mode of transportation to these markets while
keeping the present mode of transport unchanged for DC in North America. This is expected to
reduce the shipping time from around 5 weeks to 1 week. Thus, the inventory levels at FAT node

12. 13
would be reduced significantly as they need to maintain safety stock of logic boards and print
heads for 1.5 weeks of uncertainty in demand. The procurement of subassemblies from local
suppliers also results in a reduction of lead time, thus reducing the safety stock levels
considerably. However, some portion of this savings will be lost due to an increase in logistics
cost.
3.3 Product Portfolio Reduction
HP can also streamline the range of products offered in various markets. The segmentation of
current portfolio of products based on the volume and variance of demand can be used to govern
the portfolio. The management should first focus on the low volume and high variance products
as they contribute significant safety stocks at the DCs, and then on the low volume and low
variance products. The revenue contribution and potential use of a product as a substitute should
be considered before phasing out the product. Also, the company should focus on promoting
some product pairs as substitutes preferably in the high volume and high variance product
category to reduce safety stock by pooling their demand uncertainty.
3.4 Inventory Levels
After redesigning the supply chain network, the next step was to calculate optimum inventory
levels at each node in the network. This would require the revised cost structure in accordance
with the new design.
3.4.1 Revised Cost Structure
The entire supply chain of HP was approximated as a serial system starting from IC manufacturing
and ending at DC. The new lead times were estimated based on changes in the network. The

13. 14
holding cost of components was assumed to be 20% lower than the holding cost of finished
product. Thus, the local holding cost at FAT was 80% of 0.6 i.e. 0.48. The penalty cost was
assumed to be same as before. The revised lead times and holding costs for Europe and Asia
according to the new design are depicted below.
3.4.2 Inventory Level: DC Stage
There is a need to reassess the management’s initiative to maintain service level of 98%. The
safety stock levels required to maintain this service level for a product with high volume and high
variance will be quite high. The best strategy would be to identify products that drive revenue or
customer loyalty and maintain 98% service level for them and maintain cost minimizing service
level for all other products.
However, for analysis purpose, a 98% service level was considered for all the
products. The analysis shows that the holding cost at DC is reduced by 31.3% if the FAT were
moved to DC. An additional savings worth 44.9% could be attained by changing the transport
mode to air freight (Exhibit 5). These savings are due to pooling of common components across
products and the fact that holding cost rate of components is less than that of the finished
products. The safety stock for finished products needs to account for 3 days of FAT processing
time, thus making the holding cost of finished products negligible.
Figure 3.2: Revised Cost Structure

14. 15
3.4.3 Inventory Level: IC to FAT Stages
The target service level of 98% at DC doesn’t imply that each node in the system should maintain
98% service level. Thus, we calculated optimal inventory levels at different nodes according to
the cost structure using multilocation Shang Song heuristic approach. The effect of pooling led to
significant reduction in the inventory levels at PCAT and IC stages. The total inventory level in
system till FAT stage has come down by u27.23% (Exhibit 4).
Stage
Echelon Inventory Level
% ChangeCurrent
System
Redesigned
System
IC 296,048 215,444 -27.23%
PCAT/FAT 279,600 201,574 -27.91%
3.5 Special Case
There can be few occasions where supply chain network needs to respond to the unexpected
calamities or terrorist activity etc. In the wake of these incidents, inventory planning at various
affected nodes needs to be modified accordingly.
Such incidents might increase the port clearance time thereby increasing the lead
time of FAT/DC stage in our redesigned network for Europe and Asia. However, this would not
affect any of the preceding stages as the port clearance comes into effect after the PCBs and print
heads are being shipped. This would also not affect the DC in North America as no customs are
involved.
The probability of HP’s consignment being caught at customs for additional checks
was assumed to be 20% since HP is a renowned brand and won’t be flagged that often. The lead
time was assumed to increase by 2 weeks due to these checks. This made the lead time uncertain
Table 3.1: Comparison of Inventory Levels

15. 16
and thus additional safety stock needs to be piled up at FAT to safeguard against this uncertainty.
There was a 46.2% increase in safety stock level and associated holding costs due to this
unexpected scenario (Exhibit 6).
4 Conclusion and Results
HP faced the problem of not having the inventory of the right product in the right place at the
right time. The redesigned network addresses the issue of high variance by shifting FAT to DC
thus moving the decoupling point closer to the customer (to the DC) resulting in pooling of
demand variance. The bill of materials of HP printers allowed the business to employ air freight
as a mode of transport. HP could now maintain 98% service level at 62.4% lesser holding cost
than what they would have incurred with the old network. HP could also strategize phasing out
few product lines especially the low demand-high variance products.
However, this redesign needs some initial investment as it requires setting up a
new assembly line in Europe and Asia’s DC for performing the FAT operations. This would also
require some structural changes within the production stages to accommodate the new process
flow. The change of transportation mode should be analyzed for potential savings as some
portion of inventory savings will be lost to increase in logistics cost. These initial investments will
not only mend the current problem but also drive major savings for HP in the long run as
management sees huge potential in Europe and Asia market for HP products.

16. 17
5 Future Scope
Another potential approach for calculating the base stocks could be the Graves & Willems (2000)
guaranteed service model wherein we fix a maximum demand level within which all the demand
will be met within a given maximum service time. The solution to this model is based on the
shortest path algorithm which uses the costs associated with each pair of nodes in the network.
To apply this model, a maximum demand level must be determined which is
usually set as a percentile of the observed demand. This decision rests with the management
who fix the service levels desired for each product based on its volume sales and demand
uncertainties. Coming to the cost components, the fixed (setup) costs aren’t available and cannot
be assumed to be same throughout the supply chain as the costs associated with production
nodes as opposed to the assembly nodes will be different. Additionally, exact holding costs at
different stages need to be known. The actual lead times for each node are required to estimate
the holding costs at that node and to determine the costs across the paths associated with that
node.
Once we gather the required data, this model will identify the suitable nodes for
holding inventory and estimate the base stock levels for those nodes. Having an accurate
forecasting model and better information flow across the supply chain would enhance its
efficiency. Another set of actions must be planned for situations where the demand exceeds the
threshold level set by the management.

17. 18
APPENDIX

18. 19
Glossary of Terms
Monthly demand of product,
Weekly demand of product,
Mean of weekly demand,
Variance of weekly demand,
Standard Deviation of weekly demand,
Mean of monthly demand,
M
D
D
D
M
D
D
V
V



Variance of monthly demand,
Lead Time (weeks),
Lead time demand,
Mean of lead time demand,
Standard deviation of lead time demand,
Cost/unit ($),
Annual
M
LTD
LTD
L
LTD
c
i


carrying cost (%),
Holding cost/unit (= ) ($),
Penalty cost/unit ($);
Total inventory-related cost ($)
Echelon base stock model:
Holding cost at stage j ($),
Ech
j
j
h ic
p
TC
h
H elon holding cost at stage j ($),
Lead time at stage j (weeks),
Echelon lead time at stage j (weeks),
Echelon base stock level at stage j (units).
j
j
j
l
L
S

19. 20
Exhibit 1: Product Portfolio
Country Product Weekly
mean (μ)
Percentage
of total
demand (%)
Weekly
standard
deviation (σ)
Coefficient
of Variation
(%)
Category
Europe
A 10.6 0.18 16.2 153 LH
AA 105.1 1.82 102.0 97 LH
AB 3957.5 68.5 2812.3 71 HH
AQ 575.3 10 584.3 102 HH
AU 1052.0 18.21 1102.3 105 HH
AY 76.7 1.33 51.6 67 LL
Σ 5777.2 100
Asia-Pacific
A 28.7 5.94 140.8 491 LH
AB 82.9 17.17 132.9 160 HH
AG 251.5 52.07 302.3 120 HH
AK 7.8 1.6 11.8 152 LH
AU 112.1 23.22 277.9 248 HH
Σ 482.9 100
North
America
A 6608.0 99.32 3688.6 56 HL
AB 36.6 0.55 65.2 178 LH
AG 3.7 0.06 15.2 412 LH
AK 4.1 0.06 11.8 290 LH
AU 0.7 0.01 2.3 341 LH
Σ 6653.0 100
Note: Cutoff for demand volume is 10% while that for demand variance is 70% (as per coefficient of
variation values).
: consider Product A (Europe):
1. Since the weekly demand is 0.18% (< 10%) of the total weekly demand of Europe,
it is taken as a low-volume product.
2. Its coefficient of variation
Sample Calculation
(i.e. ratio of standard deviation to the mean) is 153% which is
greater than the cutoff value we considered ( 70%), so it has a high variance.
This product can therefore be classified as a (low-volume, high-variance) product (LH).

20. 21
Exhibit 2: Sensitivity Analysis of Holding costs
Country Product Penalty
cost
factor
Carrying rate
i = 40%
Carrying rate
i = 60%
Carrying rate
i = 72%
Service
level
Base
stock (y*)
Service
level
Base
stock (y*)
Service
level
Base
stock (y*)
Europe
A 3 0.88 110 0.83 102 0.8 98
AA 3 0.88 927 0.83 872 0.8 846
AB 5 0.92 33707 0.89 32300 0.87 31640
AQ 5 0.92 5521 0.89 5229 0.87 5092
AU 5 0.92 10217 0.89 9665 0.87 9407
AY 3 0.88 610 0.83 582 0.8 569
Asia-
Pacific
A 3 0.88 581 0.83 506 0.8 470
AB 5 0.92 968 0.89 902 0.87 871
AG 5 0.92 2580 0.89 2428 0.87 2357
AK 3 0.88 81 0.83 74 0.8 71
AU 5 0.92 1657 0.89 1518 0.87 1453
North
America
A 5 0.92 11942 0.89 11189 0.87 10836
AB 3 0.88 114 0.83 100 0.8 93
AG 3 0.88 22 0.83 18 0.8 17
AK 3 0.88 18 0.83 15 0.8 14
AU 3 0.88 3 0.83 3 0.8 3
: consider Product A (Europe):
Lead time for Europe L = 6 weeks (1 week for PCAT/FAT stage + 5 weeks shipping time);
Since it is a low volume product, its penalty cost is taken as p =
Sample Calculation
1
0
3*c,
For carrying cost rate i = 0.4, holding cost h = 0.4*c;
Then, critical fractile = c / (c ) 0.88 and z = F ( ) 1.18;
So, base stock level y* = * * * 6*10.6 + 1.18* 6 *16.2 110.
u u
D D
c
L z L
 
 

  
  

21. 22
Exhibit 3: Base Stock Levels at DCs (98% Service Levels)
Country Product Weekly
Mean (μ)
Weekly
Std. Dev.
Lead Time
(L)
Pipeline
Inventory
Safety
Stock
Base Stock
Levels (y*)
Europe
A 10.6 2 6 63.45 81.49 145
AA 105.1 12.7 6 630.3 512.87 1143
AB 3957.5 351.5 6 23745.15 14147.66 37893
AQ 575.3 73 6 3451.8 2939.15 6391
AU 1052 137.8 6 6312 5545.27 11857
AY 76.7 6.4 6 460.2 259.32 720
Asia-
Pacific
A 28.7 70.4 6 172.05 708.31 880
AB 82.9 66.5 6 497.4 668.57 1166
AG 2515 151.1 6 1508.85 1520.51 3030
AK 7.8 5.9 6 46.5 59.1 106
AU 112.1 138.9 6 672.75 1397.76 2071
North
America
A 6608 1844.3 1 6608.02 7575.35 14183
AB 38.6 32.6 1 36.55 133.8 170
AG 3.7 7.6 1 3,675 31.11 35
AK 4.1 5.9 1 4.05 24.13 28
AU 0.7 1.2 1 0.675 4.72 5
: consider Product A (Europe):
Mean of weekly demand: / 4 10.6,
Standard deviation of monthly demand: / 4 16.2,
Since it is continuous review, L = 6 weeks,
Then, y* = L* *
D M
D M
D
V
z
 


 
 

Sample Calculation
* 144.94 145.DL  

22. 23
Exhibit 4: Echelon Base Stock Policies
Current Structure IC Manufacturing PCAT/FAT DC
hj 0.12 0.3 0.6
Hj 0.12 0.18 0.3
lj 1 12 6 (E,A), 1 (NA)
Lj 19 (E,A), 14 (NA) 18 (E,A), 13 (NA) 6 (E,A), 1 (NA)
Country Product
Sj (Current Structure)
IC Manufacturing PCAT/FAT
Europe
(E)
A 264 251
AA 2622 2495
AB 103046 98042
AQ 14980 14252
AU 27392 26062
AY 1915 1822
Asia-Pacific
(A)
A 716 681
AB 2159 2054
AG 6548 6230
AK 193 184
AU 2920 2778
North America
(NA)
A 132434 123944
AB 699 654
AG 70 66
AK 77 72
AU 13 12
Σ 296048 279600
Redesigned Network IC Manufacturing PCAT FAT DC
hj 0.12 0.3 0.48 0.6
Hj 0.12 0.18 0.18 0.12
lj 1 12 1.5 0.5
Lj 15 14 2 0.5
Sj 215444 201574 34407 -

23. 24
:
: Product A (Europe) - PCAT/FAT stage:
1. Backorder cost b = 3;
Lead time l 12 weeks; Echelon lead time L 18 weeks;
Local holding cost h 0.3, Echelon
J
j j i
i j
j
L

  


Sample Calculation
Current Method
1
1
11
1
2
1
1
1
holding cost H 0.18;
+
2. l ratio = 0.86; z = F ( ratio) = 1.12;
+
Then, S = + z * 28.67*18 1.12* 18*19824.6 240.18,
+
3. u ratio =
+
j j j
j
i
li
jJ
i
i
l l
j j j j
j
i
i
j
i
i
h h
b H
l
b H
L L
b H
b H
 







  

  





1
2
0.94; z = F ( ratio) = 1.6;
Then, S = + z * 262.23,
4. So, Echelon base stock level: S 251.2.
2
: PCAT/FAT stage:
1. Backorder cost b = 4; Lead time l 12 weeks;
u
j
u u
j j j j
u l
j j
j
j
u
L L
S S
 



 

Redesign Network
1
16 16 16
2 2
1 1
Echelon lead time L 14 weeks;
Local holding cost h 0.3, Echelon holding cost H 0.18;
2. Pooled variance across products and locations = ( , )
J
j i
i j
j j j j
T i i j
i i j i
L
h h
Cov   


  
 
   


 
1
1
1
1
1
6
1
13846549.5;
Pooled mean across products 12913.1
=
+
3. l ratio = 0.89; z = F ( ratio) =
+
;T
j
i
li
i
i
jJ
i
i
b H
l
b H
 





 


1
11
1
2
1.23;
Then, S = + z * * 28.67*18 1.12* 18*13846549 ,
+
4. u ratio = 0.95; z = F ( ratio) = 1.65;
+
Then, S = + z * ,
5. So
198286.
,
2
2048
Echel
62.
b
5
on
l l
j T j j T j
j
i
ui
jj
i
i
u u
j T j j j T
L L
b H
u
b H
L L
 
 



  




ase stock level: 20157S .
2
4.4
u l
j j
j
S S
 

24. 25
Exhibit 5: Holding costs for FAT Localization at DCs (98% Service Levels)
Country Product
No Localization
(FAT) at DCs
(Water Transport)
Localization (FAT) at DCs
(Water Transport)
Localization (FAT) at DCs
(Air Transport)
Customized Pooling Customized Pooling
Europe
A 48.87 19.55
3769
9.77
1885
AA 307.72 123 61.5
AB 8488.5 3395 1698
AQ 1763.4 705 353
AU 3327.2 1330 665
AY 155.6 62.23 31.1
Σ $14091 $9406 $4703
Asia
A 424.98 170
453
84.9
226
AB 401.14 160.4 80.2
AG 912.3 365 165
AK 35.46 14.2 7.09
AU 838.65 335 168
Σ $2612.6 $1498 $731
North
America
A 4545.2 1818
1820
1285.5
1286
AB 80.28 32.11 22.7
AG 18.66 7.46 5.28
AK 14.47 5.79 4.09
AU 2.83 1.13 0.8
Σ $4661.4 $3685 $2604
Note: All costs are in terms of the unit cost ‘c’.

25. 26
consider Europe:
:
Base stock levels are calculated using the formula y* = * * * ;
Total holding cost for all products = $14091,
(using
D DL z L 
Sample Calculation :
No Localization at DCs
Localization at DCs the present mode of : L = 6 weeks):
Since the FAT stage is shifted to the DCs, the inventory maintained at each DC is assumed to
constitute 50% of components that are specific to a parti
water transport
cular model (like PCBs) and the
remaining 50% can be used across any model (like cables, keys & motors). Also, the holding cost
is now 80% of that corresponding to holding an entire assembled printer.
For Customized parts: (taking product A for example)
Holding cost = (proportion of the original safety stock)*(new holding cost)
= (0.5*81.49)*(0.8*0.6) = $19.55,
For Pooled parts:
P
5 5 5
2 2
1 1
ooled variance of products across Europe = = ( , )
= 9746884 + 268072.58 = 10014956.58,
Safety stock of pooled componen
T i i j
i i j i
Cov   
  
 
ts = * * 2.06* 6 *3164.64 15705.6,
Holding cost = (0.5*15705.6)*(0.8*0.6) = $3769.34,
So, Total holding cost = $5636.5 (for all products) + $3769.34 = $9405.93.
(using
Tz L   
Localization at DCs air transport: L = 1.5 weeks):
For Customized parts: (taking product A for example)
Safety stock = 0.5*( * * ) 20.39,
Holding cost = (proportion of the original safety stock)*(new holding cost)
Dz L  
5 5 5
2 2
1 1
= (20.39)*(0.8*0.6) = $9.78,
For Pooled parts:
Pooled variance of products across Europe = = ( , )T i i j
i i j i
Cov   
  
 
= 10014956.58,
Safety stock of pooled components = * * 2.06* 1.5 *3164.64 7984.3,
Holding cost = (0.5*7984.3)*(0.8*0.6) = $1884.67,
So, Total holding cost = $2818.29 (for all products) + $1884.67 = $4
Tz L   
702.96.

26. 27
Exhibit 6: Fixed and Variable Lead Times (Air Transport) (98% Service Levels)
Country Product
SS (Fixed LT) SS (Variable LT)
Customized Pooled Customized Pooled
Europe
A 41
7853
49
12970
AA 256 336
AB 7074 10280
AQ 1470 1905
AU 2773 3567
AY 130 193
Σ 11744 16330
Asia-Pacific
A 354
944
401
1326
AB 334 400
AG 760 950
AK 30 36
AU 699 808
Σ 2177 2595
North America
A 5357
5357
5357
5357
AB 95 95
AG 22 22
AK 17 17
AU 3 3

27. 28
: consider the case of Variable lead time:
: consider product A (Europe):
Lead time is usually 1.5 weeks by air (with 80% probability) and can increase by 2 week
Sample Calculation
Customized safety stock
2
s
due to unforeseen delays in customs which adds upto 3.5 weeks (with a probability of 20%).
Then, mean of lead time 0.8*1.5 0.2*3.5 1.9;
Variance of lead time 0.64;
Also, mean of lead t
L
L


  

2 2 2 2
0.98
ime demand = = * 20.14;
Variance of LTD * * 570.47;
So, safety stock level for product A = * 49.
: consider Europe:
Mean of all products
LTD D L
LTD D L D L
LTD
T i
i
z
  
    

 

  


Pooled safety stock
5
1
5 5 5
2 2
1 1
2 2 2 2
5772.7,
Variance of all products ( , ) 10,014,956.58,
Then, mean of lead time demand = = * 10968;
Variance of LTD * * 40,355,819.28;
So,
T i i j
i i j i
LTD T L
LTD T L T L
Cov   
  
    

  

  

  

 
0.98pooled safety stock level for Europe = * 12970.
Note that no customs are required for shipping to the American DC, so no variable
lead times exist in America and hence there's no change with r
LTDz  
espect to safety stocks.