Skip to main content

Full text of "DTIC ADA218255: Scoop TRIDENT Repair Part Support Package"

See other formats


AD-A218 25 


* FILE COPY ' 



SCOOP TRIDENT REPAIR 
PART SUPPORT PACKAGE 



OPERATIONS ANALYSIS DEPARTMENT 

NAVY FLEET MATERIAL SUPPORT OFFICE 
Mechanicsbyrg, Pennsylvania 17055 


Report 171 











SCOOP TRIDENT REPAIR 


PART SUPPORT PACKAGE 


PROJECT NUMBER 
Z9321-E69-9011 


REPORT 171 


Submitted by: yv ' ;t '7 A — _ 

>a. M. KLACZAK 
Operations Research Analyst 







L. FOGLE 
3perations Research Analyst 



K. T. ADAMS, LCDR, SC, USN 
Director, Operations Analysis 
Department 



H. M. HARMS, CAPT, SC, USN 

Commanding Officer 

Navy Fleet Material Support Office 


DATE: 


FEB 1 


1990 




ABSTRACT 


Adverse conditions at the TRIDENT Refit Facilities (TRFs) may force 
submarines to obtain their replenishment at a non-Navy port. Replenishment 
requirements (Pull Package) are required, without knowledge of what was used 
in the current patrol, to provide the submarine with sufficient replenishment 
support to complete another patrol. The first part of this study evaluates 
alternative methods for computing Pull Packages. These methods include 
generic versus hull-tailored, demand-based versus the Best Replacement Factor 
(BRF) based, excluding items with sufficient On-Board Replacement Assets 
(OBRA), and using Military Essentiality Codes (MECs). The alternatives are 
evaluated in terms of effectiveness, size of package and cost. The second 
part of this study examines frequency of update and Pull Package refinements. 
We recommend deleting the OBRA items from consideration and annually computing 
a generic, demand-based Pull Package with range based on MEC and depth based 
on average demand quantity for those patrols experiencing demand. 



Accession For 


NT1S GRAA1 

wr 

DTIC TAB 

□ 

Unannounced 

□ 

Justification- 

v ._ 


JL_ _— 

m attribution/ 

Availability 

Codes 


lAvail and/or 
Special 





TABLE OF CONTENTS 


Page 


EXECUTIVE SUMMARY i 

I. INTRODUCTION I 

II. BASIC PULL PACKAGE APPROACH 1 

A. ALTERNATIVES 1 

B. DATA 2 

C. METHODS OF EVALUATION 4 

III. BASIC PULL PACKAGE FINDINGS 5 

A. HULL-TAILORED VS GENERIC 5 

B. DEMAND-BASED VS BRF 7 

C. DEMAND-BASED MINUS OBRA 13 

D. USE OF MECs 16 

IV. PULL PACKAGE REFINEMENTS 19 

A. FREQUENCY OF UPDATING 19 

B. IMPROVEMENT OF PERFORMANCE FOR HIGH MEC ITEMS 23 

C. DEPTH COMPUTATION 26 

D. OBRA COMPUTATION 29 

V. SUMMARY AND CONCLUSIONS 32 

VI. RECOMMENDATIONS 35 

APPENDIX A: REFERENCES A-l 





EXECUTIVE SUMMARY 


1. Problem and Background . Adverse conditions may force the TRIDENT Refit 
Facilities (TRFs) to be unavailable for submarine replenishment support, and 
therefore the submarines may need to obtain their replenishment at a non-Navy 
port. In this situation, the repair part replenishment requirements (Pull 
Package) need to be determined without prior knowledge of the submarine's 
requirements. This Pull Package must provide the submarine with enough 
replenishment support to allow the submarine to complete another 70 day 
patrol. 

2. Objective . To determine the "optimum” Pull Package to be delivered to a 
submarine, in a non-Navy port, without prior knowledge of the submarine's 
requirements. 

3. Approach . The TRF demand history file covering the period from October 
1986 to January 1989 for eight TRIDENT submarines was used in this study. We 
extracted replenishment and corrective maintenance requisition data while 
deleting office supplies, equipage, medical and dental requisitions. For each 
submarine, we gathered statistics for the demands experienced during its pa¬ 
trol plus the number of demands experienced during the refit, but before the 
next patrol. Various options for building a basic Pull Package were evaluat¬ 
ed. They are listed below in the order of evaluation. 

• Tailoring Pull Packages to each hull versus a generic Pull Package to 
be used for all Unit Identification Codes (UICs). 

• Using only observed demand data to create a Pull Packge versus 
supplementing demand with Best Replacement Factors (BRFs). 


i 





• Removing items from the Pull Package candidates list which had 
sufficient On-Board Replacement Assets (OBRA) to satisfy the demand 
from two normal patrols. 

• Using Military Essentiality Codes (MECs) to allow more support for 
higher essential items than less essential items. 

To evaluate the alternatives, we first set aside replenishment demands 
from the last complete patrol (including refit) for use as an evaluation 
patrol. Then a Pull Package was built based on the demand history from the 
four previous patrols. The Pull Package was then evaluated based on: 

• Size (number of National Item Identification Numbers (NIINs)) of the 
package. 

• Cost (from the end-user's perspective) of the package. 

• Process NUN effectiveness (what percentage of the items requested 
during the submarine's evaluation patrol are satisfied). 

• Pull Package effectiveness (what percentage of the items in the 
package are actually needed by the submarine in the evaluation 
patrol). 

After developing a basic Pull Package, we then addressed how frequently 
the Pull Package should be updated and examined ways of improving the 
performance of the Pull Package. 

Findings . Based on the statistics presented in this report, we concluded 
that the best way to create a Pull Package was to use a generic, demand-based 
approach with separate range cutoffs for the different MECs and deleting items 
with sufficient OBRA. The range cutoffs we used satisfied 89-90% of the most 
critical NIINs demanded during the patrol, 73% of all the NIINs demanded, and 
kept the package size within practical limits. Depth should be based on the 


ii 








average demand quantity for the patrols that experienced demand. Finally the 
Pull Package should be computed once a year. As part of this study, we de¬ 
veloped Pull Packages for both FY89 and FY90. The FY90 Pull Package has 2,777 
NIINs and costs $761K. Process NUN and Pull Package effectiveness are pro¬ 
jected to be 73% and 13%, respectively, similar to the FY89 Pull Package. The 
SCOOP Logistics Planning team conducted an actual SCOOP test of the FY89 Pull 
Package. The Pull Package performed better than expected, resulting in a 
Process NUN effectiveness of 78% for the USS GEORGIA and 83% for the USS 
ALABAMA. 


iii 








I. INTRODUCTION 


The Navy SSBN Continuity of Operations (SCOOP) program is designed for the 
replenishment, repair, and crew changes of TRIDENT submarines during wartime 
conditions. Under normal conditions, submarines return to the TRIDENT Refit 
Facility (TRF) for replenishment, but during wartime this may not be feasible. 
Access to the TRF might be cut off or not convenient depending on the location 
of the submarine. The submarines may need to obtain their spare repair parts 
at a remote port, a non-Navy port, or while underway on the ocean. Due to 
tactical considerations the submarine may choose not to communicate her re¬ 
quirements during her current patrol; therefore, a method is needed to deter¬ 
mine the repair part requirements without prior knowledge. These repair parts 
are then pushed to a pre-determined site and the submarine then pulls her 
requirements from the positioned material. Reference 1 of APPENDIX A ini¬ 
tiated a study to develop a method to determine which spare parts to provide 
the submarine (Pull Package) in order to allow the submarine to continue 
operations for a second patrol without returning to the TRF. The Pull Package 
should maximize the replenishment rate for National Item Identification Num¬ 
bers (NIINs) demanded during the current patrol while keeping the package size 
within practical limits. 

II. BASIC PULL PACKAGE APPROACH 


The Pull Package alternatives, data sources and methods of evaluation are 
described in the following paragraphs. 

A. ALTERNATIVES . We evaluated various options for building a Pull 
Package. First, we looked at building hull-tailored Pull Packages versus 
building one generic Pull Package to be used for all Unit Identification Codes 
(UICs). Second, we looked at a strictly demand-based Pull Package versus 




considering items with high demand potential based on the item's Best Re¬ 
placement Factor (BRF). Third, we removed items as Pull Package candidates 
when they had sufficient On-Board Replacement Assets (OBRA) allowed by the 
Coordinated Shipboard Allowance List (COSAL) to satisfy the demand from two 
normal patrols. Finally, we used Military Essentiality Codes (MECs) to 
describe essentiality and then included more of the higher MEC items in the 
package. 

B. DATA . We used the TRF Bangor demand history file covering the period 
from October 1986 to January 1989 for seven UICs (21036, 21037, 21038, 21039, 
21040, 21041, and 21042). TABLE I cross references UIC with hull number and 
name. We extracted requisitions with project codes of either "XE_" (replen¬ 
ishment) or "XK_" (corrective maintenance), then deleted those requisitions 
having fund code equal to "ZC" (office supplies), "ZE" (equipage), or "Z7" 
(medical/dental). TABLE II shows the number of demands experienced during 
each patrol as well as the number of demands experienced during the refit but 
before the next patrol. 

TABLE I 
Test Ships 


UIC 

SSBN 

SHIP NAME 

21036 

726 

OHIO 

21037 

727 

MICHIGAN 

21038 

728 

FLORIDA 

21039 

729 

GEORGIA 

21040 

730 

HENRY M. JACKSON 

21041 

731 

ALABAMA 

21042 

732 

ALASKA 

21043 

733 

NEVADA 


2 








TABLE II 

Demand Frequency by Patrol 






PATROLS 





UIC 

PI 

P2 

P3 

P4 

P5 

P6 

P7 

P8 

P9 

21036 

696 

628 

609 

531 

645 

451 

424 

430 

470 


472 

602 

460 

378 

355 

421 

496 

400 

127 


650 808 570 418 399 512 293 492 

21037 

448 739 531 495 473 483 282 431 


626 460 452 388 566 523 530 561 

21038 

496 723 751 703 430 465 531 342 

313 662 435 393 518 307 409 582 623 

21039 

503 411 681 593 523 577 535 563 284 

425 604 624 282 342 605 362 123 

21040 

532 425 592 445 496 376 635 323 

508 256 401 468 533 493 349 320 588 

21041 

531 390 366 370 421 677 591 390 392 

767 667 391 419 704 421 472 467 

21042 

473 661 426 421 309 341 493 


NOTE: Top # - Demand during patrol dropped the first day back. 
Bottom # - Demand during refit but before ner" patrol. 


We obtained the COSAL allowance file from VITRO, which contains computed 
allowances as oi January 1989. This file was used to identify OBRAs. For 
NIINs that had different COSAL quantities on different submarines, we used the 
smallest quantity. The Navy Ships Parts Control Center (SPCC) provided us 
with MEC data as of January 1989. Finally, we used the BRFs contained in the 
SPCC BRF file as of January 1989. 


3 








































































C. METHODS OF EVALUATION . While we built the Pull Package using both 
corrective maintenance and replenishment demands, we only used the actual 
replenishment demands for evaluating the Pull Package. In order to evaluate 
the alternatives, we extracted replenishment demands from the last complete 
patrol (including refit) for use as an evaluation patrol. We then built our 
alternative Pull Packages using the four previous patrols as the demand 
history or "candidates". Finally, we evaluated our alternative Pull Packages 
based on their size, cost and effectiveness. By size of the package, we 
simply mean the number of different NIINs in the package. During reference 
(2) of Appendix A, we were asked to price the package from the perspective of 
the end-user. Thus, Appropriation Purchase Acbount (APA) items (even Cogs) 
were considered free (unit price equals zero). Also, Depot Level Repairable 
(DLR) items (7 Cog) were priced at their net price (which assumes a carcass is 
turned in). For all 1 and 9 Cog items, we used the standard (or unit) price. 
We look at the effectiveness from three different perspectives: NUN effec¬ 
tiveness, Process NUN effectiveness and Pull Package effectiveness. NUN 
effectiveness tells us what percent of the NIINs requested during the eval¬ 
uation patrol are satisfied by the Pull Package. Process NUN effectiveness 
considers the contribution of the OBRA items and shows the percent of the 
NIINs requested during evaluation patrol that can be satisfied by either the 
Pull Package or OBRA. Both NUN and Process NUN effectiveness are from the 
perspective of the submariner and tell you how happy he is with the Pull 
Package. Meanwhile, the Pull Package effectiveness measures how "good" the 
package is from the perspective of the person who is putting the package 
together who wants to know how many NIINs in the package the submarine 
actually needs (i.e., what the "turnover" rate is). The following equations 
explain the computation of the different types of effectiveness. 


9 






NUN EFFECTIVENESS - 


# OF NIINs DEMANDED THAT ARE IN PULL PACKAGE 


TOTAL # NIINS DEMANDED IN EVALUATION PERIOD 


PROCESS NUN EFFECTIVENESS - 


# OF NIINs DEMANDED THAT ARE IN PKG OR ARE OBRA 
TOTAL # NIINS DEMANDED IN EVALUATION PERIOD 


PULL PACKAGE EFFECTIVENESS = 


# OF NIINs DEMANDED THAT ARE IN PULL PACKAGE 
# PULL PACKAGE NIINs 


III. BASIC PULL PACKAGE FINDINGS 

A. HULL-TAILORED VS. GENERIC . To determine whether to use a generic or a 
hull-tailored approach, we looked at first time demand patterns. First time 
demand is demand occurring in the evaluation patrol which did not occur pre¬ 
viously. First time demand is impossible to predict from history, since we 
have no history. We looked at first time demand using a hull-tailored 
approach (looking at one UIC by itself) and using a generic approach (looking 
at all UICs together). TABLE III shows the percentage of demand that occurred 
the first time for each UIC when using a hull-tailored approach. That is, 
demand that occurred in the evaluation patrol of a given UIC, that had not 
occurred in that particular UIC's previous six patrols demand history. Using 
a hull-tailored approach, an average of 25% of evaluation period demands were 
first time demands. Thus, if we build a hull- tailored Pull Package for UIC 
21036 and included all items with at least one demand in the past six patrols 
we would only achieve a NUN effectiveness of 74%, since 26% of the demands in 
the evaluation period were first time demands. 


5 






TABLE III 


First Time Demands bv UIC Using a Hull-Tailored Approach 



UIC 

21036 

21037 

21038 

21039 

21040 

21041 

21042 

AVG 

% FIRST TIME DEMAND 

26 

32 

29 

16 

25 

22 

24 

25 


Using a generic approach allows us to look at the demand history from all 
submarines, not just the submarine for which we are building the Pull Package. 
If the demands for different UICs are similar, looking at historical demand 
from all UICs instead of just one UIC should lower the percentage of first 
time demand for each individual UIC. This is because a particular UIC's first 
time demand will not be first time demand across all UICs if the demand al¬ 
ready occurred on a different UIC. TABLE IV shows the percentage of first 
time demand for each UIC when using a generic approach. That is, demand that 
occurred in the evaluation patrol of a given UIC which had not occurred in any 
UIC's previous patrols. Using a generic Pull Package for UIC 21036 and 
including all items with any demands over the past six patrols yields a NUN 
effectiveness of 84%, a 10 percentage point gain over the hull-tailored 
package. 

TABLE IV 

First Time Demands bv UIC Using a Generic Approach 



UIC 

21036 

21037 

21038 

21039 

21040 

21041 

21042 

AVG 

% FIRST TIME DEMAND 

16 

18 

19 

16 

18 

18 

16 

17 


We found that when using a generic approach, an average of 17% of demands 
were first time demands versus the 25% average when using the hull- tailored 
approach. The lower percentage of first time demand, in addition to the 
similar results across all UICs, shows us that the different submarines have 


6 





























similar demands. We, therefore, concluded that the best way to build a Pull 
Package is to develop one generic Pull Package for use on any submarine based 
on historical data from all submarines. 

B. DEMAND-BASED VS BRF . We built a generic, demand-based Pull Package 
and then looked at supplementing it with high BRF items. To build a generic, 
demand-based Pull Package, we looked at each NUN demanded in terms of (1) how 
many UICs demanded that NUN and (2) on how many patrols was it demanded. 

Since we used four patrols of history and seven UICs, a NUN could have been 
demanded on up to 28 patrols. TABLE V shows the distribution of NlINs de¬ 
manded during the history period. The numbers across the top, the "x" values, 
represent the number of UICs demanding a particular NIIN. The numbers down 
the side, the "y" values, represent the number of patrols on which a particu¬ 
lar NIIN was demanded. The entries or "cells" in the table represent the 
number of NIINs that were demanded on "y" patrols for "x" UICs. For example, 
cell 3,2 contains the number 232. This means that 232 NIINs were demanded on 
exactly three patrols by exactly two UICs. So, for a given NIIN in that cell, 
one UIC demanded that NIIN twice and a second UIC demanded that same NIIN 
once. TABLE V also shows that there were a total of 7,688 unique NIINs de¬ 
manded across all UICs during the four patrol history period. 

TABLE VI displays the data from TABLE V in a cumulative fashion. In TABLE 
VI, a cell represents the number of NIINs demanded in "y" or more patrols by 
"x" or more UICs. For example, the number 2,033 in cell 3,2 means that 2,033 
NIINs were used on three or more patrols by two or more UICs. Cell 1,1 con¬ 
tains the total number of unique NIINs demanded across all UICs during the 
four patrol history period (7,688). To build a generic, demand-based Pull 
Package, you would select a package size and use the cell values associated 
with that number of NIINs for the range cutoff. For example, if you wanted a 


7 











package size of 1,000, you would see that cell 5,4 contains 1,009 NIINs, and 
therefore, the range cutoff values would be five or more patrols and four or 
more UICs. 

TABLE V 



PATROLS 

1 

2 

3 

4 

5 

6 

7 

TOTALS 

1 

4352 

- 

- 

- 

- 

_ 

_ 

4352 

2 

282 

995 

- 

- 

- 

- 

_ 

1277 

3 

22 

2S2 

330 

- 

- 

- 

_ 

584 

4 

4 

45 

167 

121 

- 

- 

_ 

337 

5 

- 

9 

77 

116 

34 

- 

_ 

236 

6 

- 

1 

27 

65 

42 

9 

- 

144 

7 

- 

1 

11 

40 

42 

22 

1 

117 

8 


0 

1 

28 

41 

25 

5 

100 

9 

- 

- 

1 

14 

30 

24 

12 

81 

10 

- 

- 

1 

3 

22 

25 

6 

57 

11 

- 

- 

0 

0 

16 

31 

17 

64 

12 


- 

0 

0 

7 

25 

10 

42 

13 


- 

- 

0 

6 

17 

15 

38 

14 


- 

- 

0 

6 

17 

15 

38 

15 


- 

- 

0 

2 

12 

24 

38 

16 


- 

- 

0 

0 

9 

16 

25 

17 


- 

- 

- 

0 

5 

21 

26 

18 

- 

- 

- 

- 

0 

4 

18 

22 

19 

- 

- 

- 

- 

0 

3 

25 

28 

20 

- 

- 

- 

- 

0 

3 

11 

14 

21 


- 

- 

- 

- 

1 

16 

17 

f 2 

* 


- 

- 

- 

0 

9 

9 

23 

- 

- 

- 

- 

- 

0 

10 

10 

24 

- 

- 

- 

- 

- 

0 

7 

7 

25 

- 

- 

- 

* 

- 

- 

10 

10 

26 

- 

- 

- 

- 

- 


7 

7 

27 

- 

- 

- 

- 


. 

6 

6 

28 

- 

- 

- 

- 

_ 

_ 

2 

2 

TOTALS 

4660 

1283 

615 

387 

248 

232 

263 

7688 


8 









TABLE VI 


Cumulative Distribution of NIINs Demanded 


» UICs 


PATROLS 

1 

2 

3 

4 

5 

6 

7 

1 

Hi 

- 

- 

- 

- 

- 

- 

2 

3336 

3028 

- 

- 

- 

- 

- 

3 

2059 

mm 

1745 

- 

- 

- 

- 

4 

1475 

1471 

1415 

1130 

- 

- 

- 

5 

- 

1138 

1127 

1009 

743 

- 

- 

6 

- 

902 

900 

859 

709 

495 

- 

7 

- 

758 

757 

743 

658 

486 

263 

8 

- 

641 

641 

638 

593 

463 

262 

9 

- 

- 

541 

539 

522 

433 

257 

10 

- 

- 

460 

459 

456 

397 

245 

11 

- 

- 

403 

403 

403 

366 

239 

12 

- 

- 

339 

339 

339 

318 

222 

13 

- 

- 

- 

297 

297 

283 

212 

14 

- 

- 

- 

259 

259 

251 

197 

15 

- 

- 

- 

221 

221 

219 

182 

16 

- 

- 

- 

183 

183 

183 

158 

17 

- 

- 

- 

- 

158 

158 

142 

18 

- 

- 

- 

- 

132 

132 

121 

19 

- 

- 

- 

- 

110 

110 

103 

20 

- 

- 

- 

- 

82 

82 

78 

21 

- 

- 

- 

- 

- 

68 

67 

22 

- 

- 

- 

- 

- 

51 

51 

23 

- 

- 

- 

- 

- 

42 

42 

24 

- 

- 

- 

- 

- 

32 

32 

25 

- 

- 

- 

- 

- 

- 

25 

26 

- 

- 

- 

- 

- 

- 

15 

27 

- 

- 

- 

- 

- 

- 

8 

28 

- 

- 

- 

- 

- 

- 

2 


TABLE VII shows the cumulative NUN effectiveness values for UIC 21036 
associated with the NIINs in the corresponding cells of TABLE VI. For 
example, the package in cell 5,4 which had 1,009 NIINs yields 48% NUN 
effectiveness for UIC 21036. We also developed cumulative tables for cost and 
Pull Package effectiveness. Since these tables are similar in structure to 
TABLEs VI and VII, we have not included them in the report. However, selected 
values from these tables are included in summary tables shown later. 


9 













TABLE VII 


Cumulative Distribution of UIC 21036 NUN Effectiveness for a Generic Package 

# UICs 



1 

2 

3 

4 

5 

6 

7 

PATROLS 








1 

84 

- 

- 

- 

- 

- 

- 

2 

72 

70 

- 

- 

- 

- 

- 

3 

65 

64 

60 

- 

- 

- 

- 

4 

57 

56 

56 

50 

- 

- 

- 

5 

- 

51 

51 

is 

41 

- 

- 

6 

- 

47 

47 

4 6 

41 

33 

- 

7 

- 

44 

44 

43 

39 

33 

21 

8 

- 

39 

39 

39 

37 

32 

21 

9 

- 

- 

35 

35 

34 

30 

21 

10 

- 

- 

32 

32 

32 

29 

20 

11 

- 

- 

29 

29 

29 

27 

20 

12 

- 

- 

26 

26 

26 

25 

19 

13 

- 

- 

- 

24 

24 

23 

19 

14 

- 

. 

- 

22 

22 

21 

18 

15 

- 

- 

- 

19 

19 

19 

16 

16 

- 

- 

- 

17 

17 

17 

14 

17 

- 

- 

- 

- 

15 

15 

14 

18 

- 

- 

- 

- 

13 

13 

12 

19 

- 

- 

- 

- 

12 

12 

11 

20 

- 

- 

• 

- 

9 

9 

9 

21 

- 

- 

- 

- 

- 

8 

8 

22 

- 

- 

- 

- 

- 

7 

7 

23 

- 

- 

- 

- 

- 

6 

6 

24 

- 

- 

- 

- 

- 

4 

4 

25 

- 

- 

- 

- 

- 

- 

4 

26 

- 

- 

- 

- 

- 

- 

2 

27 

- 

- 

- 

- 

- 

- 

1 

28 

- 

- 

- 

- 

- 

- 

0 


To evaluate a Pull Package, we matched the evaluation patrol demand for 
each UIC against the items in that Pull Package (NIINs passing the range 
cutoffs). This means we developed a table similar to TABLE VII for each UIC. 
TABLE VIII shows the NUN effectiveness (percentage of evaluation NUN demands 
filled by the package) and Pull Package effectiveness (percentage of NIINs in 
the package actually demanded during the evaluation period) values for each 
UIC for three Pull Package sizes and the average values across all UICs. 

TABLE VIII shows that if we could put every item demanded in the Pull Package 


10 








(7,688 items corresponding with cell 1,1 of TABLE VI), we would reach an 
average of 83% NIIN effectiveness and 7% Pull Package effectiveness. The NTTN 
effectiveness is limited to 83% because 17% of the demands are first time 
demands (as previously shown in TABLE IV), so they would not have shown up in 
our data base used to create the Pull Package. The Pull Package effectiveness 
is low because the package is so large and we only need a small percentage of 
the items. When we limit the size of the Pull Package to 541, TABLE VIII 
show'- that we would get a much higher Pull Package effectiveness (45%) , but a 
lower NUN effectiveness (36%). The Pull Package of 1,009 falls between the 
other two. 

TABLE VIII 

Summary of Effectiveness by UIC 
NIIN Effectiveness/Pull Package Effectiveness 



ALTERNATIVE PACKAGES 

7688 ITEMS @ $1.47M 

541 

ITEMS @ $93K 

1009 ITEMS @ $163K 

UIC 

NIIN 

PULL PKG 

NUN 

PULL PKG 

NIIN 

PULL PKG 

21036 

84 

7 

35 

43 

48 

31 

21037 

82 

5 

36 

31 

47 

22 

21038 

81 

8 

35 

50 

48 

36 

21039 

84 

9 

32 

51 

46 

40 

21040 

82 

8 

37 

50 

50 

36 

21041 

82 

6 

37 

39 

48 

27 

21042 

84 

8 

40 

52 

54 

38 

AVG 

83 

7 

36 

45 

49 

33 


When using demand history to build the Pull Package, certain items with no 
previous demand limit the effectiveness we can reach. We looked at using BRFs 


11 














as an indicator of items which have a high probability of experiencing demand 
and therefore belong in the Pull Package. We took the COSAL file and found 
the BRF cut-off values that would give us various Pull Package sizes. For 
example, 575 NIINs had BRFs of .99 or greater. So, to build a package of 
approximately 550 NIINs, we would include all COSAL items having a BRF of .99 
and greater. We then evaluated the various Pull Packages by determining the 
effectiveness values and comparing them with the values we would get using the 
demand-based method. TABLE IX shows the NUN effectiveness comparison between 
the BRF and demand-based methods for various size Pull Packages. For the same 
size package, the demand-based package performed significantly better than the 
BRF based package. As we decreased the BRF cutoff value, the size of the 
package increased dramatically. We found that even if we used a BRF cutoff 
value of 0, that is, create a Pull Package using the entire COSAL file (17,632 
NIINs), the maximum NUN effectiveness we could obtain was 75%. We concluded 
that it is better to build a Pull Package based solely on demand than one 
based solely on the BRF. 

TABLE IX 

NUN Effectiveness 



PULL PACKAGE 
STRICTLY BASED ON 

APPROX. 



PACKAGE 



SIZE 

BRF 

DEMAND 

550 

7% 

35% 

860 

12% 

46% 

1150 

16% 

50% 

1500 

20% 

57% 

3350 

36% 

72% 

7700 

55% 

84% 


12 





We also looked at using BRFs to supplement our demand-based Pull Package 
in an effort to limit the Pull Package size while picking up those items which 
had not yet experienced demand but would be likely to in the future. We 
attempted to reach 90% NUN effectiveness. When we supplemented the Pull 
Package of 7,688 items (which yielded 84% NUN effectiveness for UIC 21036) 
with high BRF items, we found that we needed a package of 12.000 NIINs to 
reach 90% NUN effectiveness for that UIC. We concluded that, although we can 
achieve 90% NUN effectiveness by combining the demand-based method with the 
BRF method, the size of this Pull Package is not practical. It is better to 
base the Pull Package on demand alone. 

C. DEMAND-BASED MINUS OBRA . To help keep the Pull Package size within 
practical limits, we removed from consideration as Pull Package candidates 
those items with "sufficient" OBRA. We defined "sufficient" to mean that (1) 
the COSAL quantity equals three or more and (2) the COSAL quantity exceeds 
twice the average patrol demand. We did not remove from consideration any 
Strategic Weapons System (SWS) or Nuclear Reactor Plant (NRP) items. We found 
that 1,459 of the 7,688 candidates have allowances which appeared to be 
sufficient to satisfy the demand for two normal patrols. TABLE X shows the 
cumulative distribution of NIINs demanded after deleting the OBRA items. 

TABLE XI shows the cumulative distribution of Process NUN effectiveness for 
UIC 21036, when the OBRA items are not in the package, but since we are now 
looking at Process NUN effectiveness their contribution is included in the 
effectiveness numbers. TABLE XII shows the Process NUN effectiveness and 
Pull Package effectiveness for the three different package sizes shown in 
TABLE VIII. 


13 




TABLE X 


Cumulative Distribution of NIINs Demanded Not Including OBRA Items 

* UICs 


PATROLS 

1 

2 

3 

4 

5 

6 

7 

1 

6229 

- 

- 

- 

- 

- 

. 

2 

2414 

2170 

- 

- 

- 

- 

- 

3 

1399 

1381 

1168 

- 

- 

- 

- 

4 

960 

957 

920 

723 

- 

- 

- 

5 

- 

728 

719 

639 

473 

- 

- 

6 

- 

565 

564 

539 

453 

328 

- 

7 

- 

473 

473 

463 

420 

323 

191 

8 

- 

406 

406 

405 

384 

309 

190 

9 

- 

- 

348 

348 

339 

293 

187 

10 

- 

- 

298 

298 

297 

271 

179 

11 

- 

- 

267 

267 

267 

251 

176 

12 

- 

- 

233 

233 

233 

223 

166 

13 

- 

- 

- 

206 

206 

199 

158 

14 

- 

- 

- 

189 

189 

184 

151 

15 

- 

- 

- 

161 

161 

159 

139 

16 

- 

- 

- 

138 

138 

138 

123 

17 

- 

- 

- 

- 

124 

124 

113 

18 

- 

- 

- 

- 

104 

104 

97 

19 

- 

- 

- 

- 

91 

91 

86 

20 

- 

- 

- 

- 

71 

71 

68 

21 

- 

- 

- 

- 

- 

61 

60 

22 

- 

- 

- 

- 

- 

49 

49 

23 

- 

- 

- 

- 

- 

40 

40 

24 

- 

- 

- 

- 

- 

31 

31 

25 

- 

- 

- 

- 

- 

- 

24 

26 

- 

- 

- 

- 

- 

- 

14 

27 

- 

- 

- 

- 

_ 

- 

8 

28 

- 

- 

- 

- 

- 

- 

2 


14 







TABLE XI 


Cumulative Distribution of UIC 21036 Process NUN Effectiveness 

for a Generic Package 
# UICs 



1 

2 

3 

4 

5 

6 

7 

# PATROLS 








1 

84 

- 

- 

- 

- 

- 

- 

2 

75 

74 

- 

- 

- 

- 

- 

3 

70 

70 

67 

- 

- 

- 

- 

4 

65 

65 

64 

62 

- 

- 

- 

5 

- 

63 

63 

61 

57 

- 

- 

6 

- 

61 

61 

60 

57 

54 

- 

7 

- 

58 

58 

58 

56 

53 

47 

8 

- 

56 

56 

56 

55 

53 

47 

9 

- 

- 

54 

54 

53 

52 

47 

10 

- 

- 

52 

52 

52 

51 

47 

11 

- 

- 

51 

51 

51 

50 

47 

12 

- 

- 

49 

49 

49 

49 

46 

13 

- 

- 

- 

48 

48 

48 

46 

14 

- 

- 

- 

48 

48 

47 

46 

15 

- 

- 

- 

46 

46 

46 

45 

16 

- 

- 

- 

45 

45 

45 

43 

17 

- 

- 

- 

- 

44 

44 

43 

18 

- 

- 

- 

- 

43 

43 

42 

19 

- 

- 

- 

- 

42 

42 

42 

20 

- 

- 

- 

- 

40 

40 

40 

21 

- 

- 

- 

- 

- 

39 

39 

22 

- 

- 

- 

- 

- 

39 

39 

23 

- 

- 

- 

- 

- 

37 

37 

24 

- 

- 

- 

- 

- 

36 

36 

25 

- 

- 

- 

- 

- 

- 

35 

26 

- 

- 

- 

- 

- 

- 

34 

27 

- 

- 

- 

- 

- 

- 

33 

28 

' 

■ 

■ 

■ 


' 

32 

Comparing TABLE XII to TABLE 

VIII, 

we see 

that for similarly sized and 

priced packages (e.g. 

, 540 NIINs 

at $94K) the 

Process 

NUN effectiveness 

improves by 

22 percentage points 

(from 

36% to 

58%). 

We concluded 

that de- 

leting OBRA 

items as 

candidates 

improves the 

Process 

NUN effectiveness of 

package and 

allows us 

to keep the size 

of the package 

within more 

manageabl 


limits. 


15 







TABLE XII 


Summary of Effectiveness 

by VIP 

Process NIIN Effectiveness/Pull Package Effectiveness 



ALTERNATIVE PACKAGES 

6229 ITEMS @ $1.3M 

539 ITEMS @ $94K 

920 ITEMS @ $174K 

UIC 

PROCESS 

NUN 

PULL PKG 

PROCESS 

NUN 

PULL PKG 

PROCESS 

NUN 

PULL PKG 

21036 

84 

5 

60 

34 

64 

23 

21037 

82 

4 

53 

25 

61 

19 

21038 

81 

6 

57 

39 

63 

28 

21039 

84 

7 

57 

42 

62 

29 

21040 

82 

6 

57 

41 

63 

29 

21041 

82 

5 

58 

31 

63 

21 

21042 

83 

6 

61 

43 

66 

29 

AVG 

83 

6 

58 

36 

63 

25 


D. USE OF MECs . To tailor the package to more critical items, we 
considered the MEC in addition to the demand history. TABLE XIII defines the 
MECs and provides a distribution of the candidates by MEC. 

TABLE XIII 


MEC Definitions and Distribution 

MEC DEFINITION NON-OBRA ITEMS W/DMD 


95 

98 

101 

104 

107 

no ; 

116* 


Negligible Degradation 
Partial Degradation 

Totally Degrade Missile Launch Capability 
Mission Abortion 


3962 

351 

407 

24 

84 

532 

869 


*MEC - 116 Assumed for NRP Items 


6,229 


16 




























TABLE XIV shows the distribution by MEC of the number of NIINs and the 
cost for five different size Pull Packages. TABLE XV shows the corresponding 
Process NUN and Pull Package effectiveness. These tables allow us to mix and 
match the range cutoffs between MEC groupings in ways that will maximize NUN 
effectiveness while keeping the Pull Package size within practical limits. 
During reference (c), the SCOOP Logistics Planning team decided to maximize 
NUN effectiveness for MEC 116 and 110 items by using the range cutoff of cell 
1,1. In other words, MEC 116 and 110 items having demand on at least one 
patrol from at least one UIC will be added to the Pull Package. MEC 98 
through MEC 107 items having demand on at least two patrols (cell 2,1) will be 
added to the Pull Package. Finally, MEC 95 items having demand on at least 
three patrols from at least two UICs (cell 3,2) will be added to the Pull 
Package. (The selected values are highlighted in TABLES XIV and XV.) TABLE 
XVI shows the incremental and total size, cost, Process NUN effectiveness, 
and Pull Package effectiveness for the Pull Package using these range cutoffs. 

TABLE XIV 

Distribution of # NIINs/Cost bv MEC 





ALTERNATIVE PULL 
# PATROLS - # UICs 

PACKAGES AS DEFINED BY 

NEEDED TO BE PART OF PKG 


Bi 

■ 

3 

3-2 

2-1 

1-1 

MEC 

NIINs 

COST 

NIINs 

COST 

NIINs 

COST 

NIINs 

COST 

NIINs 

COST 

95 

309 

44 

547 

68 

838 

iH 

1,472 

221 

3,962 

642 

98-107 

68 

24 

113 

35 

167 

50 

314 

85 

866 

245 

110 

41 

8 

93 

47 

143 

68 

242 

121 

532 

215 

116 

121 

19 

167 

25 

233 

51 

386 

84 

869 

245 

TOTAL* 

539 

$94K 

920 

$174K 

1,381 

$293K 

2,414 

$511K 

6,229 

$1.35M 


* NOTE: Cost totals may not add due to rounding. 


17 














































TABLE XV 


Distribution of Process 


Effectiveness 


ALTERNATIVE PULL PACKAGES AS DEFINED BY 
# PATROLS - # UICs NEEDED TO BE PART OF PKG 


MEC NIINs PULL PKGiNIINs PULL PKG INI INs PULL PKGINIINs PULL PKGINIINs PULL PKG 


98-107 57 

110 67 


35 


116 


TOTAL 58 


67 31 

65 40 


NIINs 

PULL PKG 

59 

25 

62 

26 

74 

21 

69 

32 

63 

25 


3-2 

NIINs 

PULL PKG 

E9IK91 

64 

19 

77 

16 

73 

25 

68 

19 



NIINs 

PULL PKG 

81 

5 

81 

6 

89 

7 

\wm 

8 

83 

6 


TABLE XVI 
FY89 Pull Package 


PROCESS PULL 

, .NUN PACKAGE 

# NIINs COST EFFECTIVENESS EFFECTIVENESS 


MEC 116 ITEMS WITH DEMAND ON AT 
LEAST ONE PATROL 


MEC 110 ITEMS WITH DEMAND ON AT 
LEAST ONE PATROL 


MEC 98-107 ITEMS WITH DEMAND ON AT 
LEAST TWO PATROLS 


MEC 95 ITEMS WITH DEMAND ON AT 
LEAST THREE PATROLS FROM TWO 
DIFFERENT SSBNs 


869 $245K 


532 $215K 


314 $ 85K 


838 $126K 


2553 $671K 



18 








































































We concluded that the best method for building a Pull Package is to use a 
generic, demand-based approach with separate range cutoffs for the different 
MECs. The range cutoffs we used satisfied 84-89% of the most critical NIINs 
demanded during the patrol, 73% of all the NIINs demanded, and kept the 
package size within practical limits. 

IV. PULL PACKAGE REFINEMENTS 


In the first part of this study, we evaluated different alternatives for 
developing a basic Pull Package. TABLE XVI showed the results of combining 
the best of these alternatives to compute the FY89 Pull Package. An actual 
SCOOP test was conducted using the FY89 Pull Package with the USS GEORGIA and 
USS ALABAMA as the test submarines. This actual test resulted in a Process 
NUN effectiveness of 78% and 83% for the USS GEORGIA and USS ALABAMA, re¬ 
spectively. These test results exceeded our forecasted Process NUN ef¬ 
fectiveness (73%) and validated our approach to developing basic Pull Pack¬ 
ages. Our next step was to examine how often we should update our Pull 
Package in order to reflect current demand patterns and to remove items no 
longer demanded Once we determined frequency of updating, we then examined 
additional ways to improve the performance of the high essentiality (MEC 110/ 
116) items, depth computations and OBRA computations. The following para¬ 
graphs describe our approach and findings for each of these topics. 

A. FREQUENCY OF UPDATING . 

1. APPROACH . We investigated three alternative procedures for 
determining when to update the Pull Package. A Pull Package could be computed 
after each submarine completes her patrol, quarterly (using data from the 
submarines which completed their patrol since the previous quarter's update) 


19 







or once a year (using data from all patrols completed since the last update). 
There are drawbacks in using any one of these possibilities. Computing a Pull 
Package after each individual submarine patrol can be costly in terms of Navy 
Stock Fund (NSF) dollars and manpower. Given that we have eight submarines, a 
Pull Package would be computed every two or three weeks. Churn would become a 
major issue. Items may be deleted after one patrol and then added again after 
the next patrol. Consequently, we limited our data analysis to quarterly and 
annual updates. Any recalculation of a Pull Package lends itself to a timing 
problem. Patrols (including refit) normally last 90 to 100 days. With that 
in mind, a given number of submarines will be on patrol whenever the package 
is recomputed. At the start of a patrol, a submarine receives a list of Pull 
Package items in case she has a demand for a critical item not on the list. 

If this happens, she could submit a message requesting that the critical item 
be included on their Pull Package. If the package changes while the submarine 
is underway, confusion may result. The submarine may need an item on the 
"old" package and think that she will receive that item as part of the Pull 
Package. However, if by chance that particular item is no longer on the "new" 
Pull Package, then the submariner will not get that item unless the people 
sending the Pull Package out keep two Pull Packages on hand so they can send 
the old one to those submarines which are still operating under the old list 
and the new package to those who have switched over to the new list. In order 
to answer the question how often to update, we will consider the trade-off 
between stability and responsiveness to new trends by comparing quarterly to 
annual updates. 

To determine how often to update, we used the same range rules used to 
construct the FY89 Pull Package. Using the annual update concept, we built a 
Pull Package using patrols one through four (our four oldest patrols). Then 


20 






we determined Process NIIN/Pull Package effectiveness for each of the next 
three patrols (patrols five, six and seven). Using the quarterly update 
concept, we recomputed the Pull Package based on patrols two through five and 
used the sixth patrol to determine Process NIIN/Pull Package effectiveness. 

We then recomputed the package again using patrols three to six to develop the 
Pull Package and the seventh patrol to determine Process NIIN/Pull Package 
effectiveness. We developed statistics comparing effectiveness based on an 
annual Pull Package evaluated over three patrols and a Pull Package updated 
quarterly. 

2. FINDINGS . TABLE XVII displays the average Process NIIN/Pull 
Package effectiveness across all UICs by MEC category. The first part of the 
table was developed by computing an annual Pull Package, using patrols one 
through four and evaluated this constant Pull Package against the demands that 
occurred in patrols five, six and seven. Focusing our attention on the ALL 
category, we can see that there is only a one to two percentage point differ¬ 
ence in the overall effectiveness between the patrols. Next we compared the 
recomputed quarterly Pull Packages to the demands in the subsequent patrol. 
Comparing the annual Pull Package to the quarterly updated Pull Package for 
the same evaluation patrol (the columns labeled patrol 6 and patrol 7), we 
observe that the quarterly update of the Pull Package provided a maximum of 
two percentage points increase in effectiveness over the constant Pull Package 
for the same evaluation patrol. 


21 





TABLE XVII 


Average - Process NIIN/Pull Package Effectiveness 



DEVELOPED FROM PATROLS 




1 

-4 



2-5 

3-6 


PATROL 5 

PATROL 6 

PATROL 7 

PATROL 6 

PATROL 7 

MEC 

NUN 

PULL PKG 

NUN 

PULL PKG 

NUN 

PULL PKG 

NUN 

PULL PKG 

NUN 

PULL PKG 

95 

63 

20 

63 

18 

64 


65 

19 

65 

17 

98-107 

69 

13 

69 

11 

67 



11 

71 

11 

110 

85 

7 

91 

6 

86 

5 


6 

89 

9 

116 

86 

10 

88 

10 

85 

8 


9 

85 

8 

ALL 

71 

13 

72 

12 

71 

11 

_ 

74 

12 

73 

_ 

11 


Since the effectiveness between time of update methods is similar, we 
evaluated the impact on churn between the Pull Packages. TABLE XVIII displays 
churn statistics (number of adds and deletes) between the different Pull 
Packages. Observing the first row we see that the Pull Package developed 
using patrols one through four has 2,729 NIINs. At the end of the next 
quarter (a quarterly update), we developed a Pull Package of 2,733 NIINs using 
patrols two through five. This results in 437 NIINs being added to the new 
Pull Package which weren't on the old Pull Package. Also there were 433 NIINs 
which were on the old Pull Package and no longer qualified for the new Pull 
Package. The second row shows churn results for the next quarterly update 
(351 adds and 495 deletes). When consecutive quarterly updates were done, 116 
NIINs were added/deleted on the first Pull Package but deleted/added on the 
third update. When we skip a quarter (semi-annual updates) and then compute a 
Pull Package (third row), the number of adds and deletes increase as compared 
to each quarter's churn. However, the adds (672) and deletes (812) is less 













































than the total 788 adds (437 + 351) and 928 deletes (433 + 495) generated by 
the two updates over the same time period. So it appears to be better to 
compute a Pull Package once a year. An annual computation minimizes needless 
churn does not negatively impact effectiveness, and keeps the process simple. 

TABLE XVIII 
Pull Package Churn 


UPDATE 

FREQUENCY 

DEVELOPED FROM 
PATROLS 

OLD 

# NIINs 

ADDS 

DELETES 

NEW 

# NIINs 

QUARTERLY 


=> 

2-5 

2729 

437 

433 

2733 

QUARTERLY 

2-5 


3-6 

2733 

351 

495 

2589 

SEMI-ANNUAL 


=> 

3-6 

2729 

672 

812 

_ 

2589 


NOTE: 116 NIINs Added/Deleted in Consecutive Quarterly Updates 


B. IMPROVEMENT OF PERFORMANCE FOR HIGH MEC ITEMS . 

1. APPROACH. Under our initial recommended Pull Package, MECs 110 
and 116 had a NUN effectiveness of 89% and 84%, respectively. During 
reference (3) of Appendix A, we were asked what can be done to further improve 
MEC 110 and 116 performance (to at least 90% NUN effectiveness). In our 
initial Pull Package we included all MEC 110 and 116 items with at least one 
demand from any submarine during the four most recent patrols. In an effort 
to increase performance of MEC 110 and 116, we expanded the range of MEC 110 
and 116 items. We accomplished this by adding to the Pull Package all MEC 110 
and 116 items having at least one demand over a longer time interval (i.e., 
our entire demand history instead of four patrols). We then compared Process 
NUN effectiveness for this expanded range (called entire history below) to 
the Process NUN effectiveness under the FY89 Pull Package. 

Currently there are eight TRIDENT submarines patrolling the Pacific. When 
our initial Pull Package was developed, we used data from only seven of the 


23 





























UICs, since one of the UICs didn't seem to have enough valid data to be 
included in the development of the Pull Package. After obtaining more recent 
demand history data, we then included the additional eighth UIC to develop a 
new Pull Package. Instead of using 28 patrols (seven UICs times four 
patrols), we now used 32 patrols (eight UICs times four patrols). Therefore, 
we also developed a Pull Package using 32 patrols vice 28 patrols. We com¬ 
puted Process NUN effectiveness and compared it to both the FY89 and entire 
history alternative Process NUN effectiveness to determine if including the 
eighth UIC increased effectiveness. 

2. FINDINGS . TABLE XIX shows the Process NUN effectiveness for the 
MEC 110 and 116 items under the FY89 approach (one demand in four patrols) and 
the alternative of using a longer time interval (entire demand history). The 
table displays the range, cost. Process NUN effectiveness and Pull Package 
effectiveness by MEC across the seven UICs. An average is computed for NUN 
and Pull Package effectiveness by MEC and UIC. Using a longer time horizon 
for MEC 110 and 116 items achieves an average of 89% Process NUN effective¬ 
ness (a three percentage point increase). It also adds 456 NIINs to the Pull 
Package, increasing its total size from 2,553 NIINs to 3,009 NIINs. The cost 
of the total Pull Package increases by $127K (from $671K to $798K). Since 
NUN effectiveness increased, it appears reasonable to use eight patrol's 
worth of demand for high MEC items. 


24 





TABLE XIX 


Process NIIN/Pull Package Effectiveness 
(More patrols considered in Pull Package 1 ) 



ONE 

DEMAND IN FOUR PATROLS 


ENTIRE DEMAND HISTORY 

MEC 

110 

116 

BOTH 

110 

116 

BOTH 

RANGE 

/COST 

532/$215K 

869/$2' 

5K 

1401/$460K 

683/$258K 

1174/$329K 

1857/$587K 

UIC 

PROCESS 

NUN 

PULL 

PKG 

PROCESS 

NUN 

PULL 

PKG 

PROCESS 

NUN 

PULL 

PKG 

PROCESS 

NUN 

PULL 

PKG 

PROCESS 

NUN 

PULL 

PKG 

PROCESS 

NUN 

PULL 

PKG 

21036 

93 

7 

87 

8 

90 

7 

96 

B 

88 

6 

91 

B 

21037 

87 

4 

83 

6 

85 

5 

87 

B 

89 

5 

87 

B 

21038 

90 

6 

78 

9 

84 

8 

91 

fl 

82 

B 

85 

B 

21039 

87 

11 

87 

9 

87 

9 

92 

B 

92 


92 

8 

j 21040 

88 

17 

89 


88 

9 

91 

6 

91 

8 

91 

B 

21041 

88 

16 

83 

5 

85 

5 

95 

5 

88 

B 

90 

B 

21042 

90 

6 

83 

11 

86 

9 

92 

5 

86 


88 

8 

AVG 

89 

B 

84 

8 

86 

B 

92 

5 

88 

B 

89 

6 


However, when we added the eighth UIC (32 vice 28 patrols), Process NUN 
effectiveness also increased for MEC 110 and 116 items. TABLE XX displays the 
results of using 32 patrols relative to the two alternatives described above. 
Considering 32 vice 28 patrols resulted in the MEC 110 Process NUN effective¬ 
ness remaining the same (89%) and MEC 116 Process NUN effectiveness increas¬ 
ing by six percentage points. Comparing the 32 patrols to the entire history 
shows that they both achieve an average for the 110 and 116 of 89%. Even 
though the both achieve the same average, the 32 patrol approach achieves this 
through higher Process NUN effectiveness for the more critical group (116 MEC 
items). In addition, the 32 patrol approach has a better Pull Package 


23 



















































effectiveness (9% versus 6%), adds fewer items to the package (65 compared to 
456) and results in a smaller additional cost ($45K versus $127K). Therefore, 
we conclude that it is better to add the eighth UIC to the data base, then 
using more history for the 110 and 116 MEC items. 

TABLE XX 

Alternative Method for Improving Performance for 
MEC 110 and 116 Items 



AVERAGE PROCESS 
NUN EFFECTIVENESS 

AVERAGE PULL 1 

PKG 

EFFECTIVENESS 

89 

7 

84 

8 

86 

7 

89 

7 

90 

10 

89 

9 

92 

5 

88 

7 

89 

6 



1. APPROACH . In developing our Pull Package, we initially focused 

attention on the range and type of items the Pull Package should contain, not 

the depth of an item. Initially depth was computed using the simple average 

of units demanded over the 28 patrols (shown below). 

_ , # of units demanded f na 

Depth - _ . / 28 

over 28 patrols / 


26 






















An alternative approach to computing depth is to consider the average 
number of units demanded for the patrols that experienced demand (shown be¬ 


low) . 

, # of units demanded /# of patrols 

" over 28 patrols / that experienced demand 

This method more accurately predicts the average patrol demand 
requirements when a demand occurs (a conditional probability). We gathered 
the following statistics for the two different depth rules: The percent of 
NIINs with same depth quantities, the differences in quantities, the impact on 


partial fills, and unit effectiveness, where 


Unit effectiveness 


funits satisfied 


units satisfied! /Total 


[from Pull Package from OBRA 


isfied! /Tot 
Y Uni 

7 Dem 


Units 

Demanded 


To evaluate the Pull Package in a real world environment, TRIDENT subma¬ 
rine personnel performed an actual SCOOP using the FY89 Pull Package. During 
a SCOOP, the UIC requisitions material from the Pull Package instead of the 
TRF. Statistics were gathered on how many requisitions were filled or par¬ 
tially filled. The test UICs involved were the USS GEORGIA and the USS 
ALABAMA. Pacific Fleet Polaris Material Office (PMOPAC) sent us a list of 
partially filled stock replenishment requisitions from this test for com¬ 
parison under the alternative depth computation to see how many partial fills 
could have been avoided and how many more units would be satisfied. 

2. FINDINGS . TABLE XXI compares the two different methods of 
computing depth by MEC. Recall our initial method involves computing the 
average patrol demand over all patrols but our alternative computes the 
average patrol demand when a demand occurred . The first column displays the 
percentage of items whose depth was equal using the two different depth rules. 
For example, 41% of the MEC 95 items had equal quantities. For MEC 110 and 
116 items, it was 69% and 57%, respectively. The next column shows the 


27 






percentage of items where the quantity differed by one. The 85th percentile 
column shows that 85% of the time, the quantities differ by three or less 
units for MEC 95, and for MEC 98-107, by one or less for MEC 110, and by two 
or less for MEC 116. Results are also shown for the 95th percentile. For MEC 
95 items, for example, the quantities differ by 11 or less 95% of the time. 


TABLE XXI 

Quantity Comparison between the Two Depth Rules 


MEC 

% QTY EQUAL 

% DIFF BY 1 

85TH PERCENTILE 

95TH PERCENTILE 

95 

41 

28 

3 

11 

98-107 

48 

26 

3 

12 

110 

69 

17 

1 

4 

116 

57 

21 

2 

8 


TABLE XXII displays the unit effectiveness and cost of our FY89 and 
alternative depth rul«s. Across all items, unit effectiveness increased eight 
percentage points and cost increased by $258K when using the alternative 
method. MEC 116 unit effectiveness increased by seven percentage points. 

TABLE XXII 

Unit Effectiveness/Cost Comparison 



UNIT EFFECTIVENESS 

COST 

MEC 

FY89 

ALTERNATIVE 

FY89 

ALTERNATIVE 

95 

54 

64 

$126K 

$192K 

98-107 

80 

85 

$ 85K 

$107K 

110 

81 

84 

$215K 

$286K 

116 

77 

84 

$245K 

$344K 

TOTAL 

66 

74 

$671K 

$929K 


28 

















The next question is, how much better would the alternative method have 
done during the SCOOP test? Under the current method the USS GEORGIA had 64 
partials during the test. It satisfied 656 units out of 1,178 units demanded, 
for a unit effectiveness of 56%. The alternative method reduced the number of 
partially filled requisitions from 64 to 25. It satisfied 988 units out of 
1,178, for a unit effectiveness of 84%, an increase of 28 percentage points. 
For the USS ALABAMA the current method resulted in 38 partially filled requi¬ 
sitions and satisfied 122 out of 307 units demanded, for a unit effectiveness 
of 40%. The alternative method reduced the number of partially filled requi¬ 
sitions from 38 to 17 and satisfied 189 of 307 units demanded, for a unit 
effectiveness of 62%, an increase of 22 percentage points. Thus, we concluded 
that the alternative depth computation greatly reduces partially filled requi¬ 
sitions, greatly increases unit effectiveness, and should be used in future 
Pull Package computations. 


D. OBRA COMPUTATION . 

1. APPROACH . Under our initial procedure for determining items with 
sufficient OBRA, the following must be true. 

COSAL Quantity > 3 


_ ... . (total demand * 2) 

J # of patrols 

As with the Pull Package depth computations, the question arose as to 
whether we should be comparing the COSAL quantity to the average quantity per 
patrol or the average quantity for the patrols that experienced demands. 
Dividing by the number of patrols which experience demand may result in a more 
realistic estimate of sufficient OBRA items. We evaluated three methods for 
determining items with sufficient OBRA. In all cases the smallest COSAL 
quantity across all submarines must be greater than two for an item to be 
considered. 


29 






The three methods are: 

• Based on the average patrol demand (used for FY89 Pull Package). 

If COSAL Quantity > (tptal de^napd * 2 1 
J # of patrols 

• Based on the average patrol demand for patrols that experienced demand. 


If COSAL Quantity > 


(total demand * 2) 

# of patrols with demand 


• Based on 85th percentile. 


If COSAL Quantity > MINIMUM 


' (total demand * 2) + x 
# of patrols 

(total demand * 2) 

4 of patrols with demand 


where X - 3 for MEC - 95-107 

1 for MEC - 110 

2 for MEC - 116 


The third method uses the average demand for patrols that experienced demand 
except for the NIINs where there are large differences between this approach 
and the average patrol demand (first method). The variable X restricts the 
third method to be no higher than the 85th percentile of differences between 
the first two methods. Using the 85th percentile method minimizes the effect 
of extreme demand observations (outliers). 

A Pull Package was computed for each of these three methods to determine 
which method provides the best unit effectiveness. Unit effectiveness is 
units satisfied divided by units demanded. For Pull Package items demanded, 
the units satisfied was computed as follows: 

Units Satisfied - Demand Quantity if Demand Quantity less than or 
equal to Pull Package Quantity 


otherwise, 

Units Satisfied - Pull Package Quantity 


30 







For all Non-Pull Package items with demand, we computed units satisfied as 


shown below. 


Units Satisfied - Demand Quantity if Demand Quantity is less 

than or equal to half the COSAL Quantity 


otherwise, 

Units Satisfied - half the COSAL Quantity 
2. FINDINGS . TABLE XXIII displays the unit effectiveness results of 
computing sufficient OBRA items based on the number of patrols undertaken 
(current method), the number of patrols with demand, and on the 85th percen¬ 
tile method. Based on the total average unit effectiveness, there is a 
maximum of two percentage points between the methods. But for critical MECs 
(110 and 116) the OBRA based on the total number of patrols provided the best 
average unit effectiveness of the three methods. Therefore, we concluded, 
since neither of the ''ther two methods improved the unit effectiveness, the 
method of determining sufficient OBRA should remain based on the total number 
of patrols. 

TABLE XXIII 

Average Unit Effectiveness 



-- 

OBRA 

MEC 

BASED ON # OF PATROLS 

BASED OR # OF PATROLS U/DEMANO 

BASED ON 85TH PERCENTILE METHOO 

95 

47 

46 

49 

98-107 

84 

80 

82 

110 

86 

79 

80 

116 

83 

81 

81 

TOTAL 

60 

58 

59 


31 












V. SUMMARY AND CONCLUSIONS 


The SCOOP Logistics Planning team asked us to assist them in determining 
which spare repair parts to provide for the submarine, without knowing which 
parts are required, so the submarine may continue operations without returning 
to the TRF. The Pull Package must maximize the replenishment rate for NIINs 
demanded during the current patrol (especially critical NIINs), and keep the 
package size within practical limits. 

The basic development of the Pull Package consisted of first determining 
the demand history based on four patrols from seven UICs, developing a Pull 
Package with items demanded on "y" number of patrols and "x" number of UICs, 
and then using the patrol period after the last history period to evaluate the 
Pull Package. We gathered statistics in terms of size and cost of the Pull 
Package, Process NUN effectiveness and Pull Package effectiveness. 

When the Pull Package was initially developed, we determined that a 90% 
Process NUN effectiveness goal was not achievable based solely on demand. 

Due to first time demand, the best Process NUN effectiveness was 83%. BRFs 
were introduced to supplement or replace a demand-based Pull Package. We 
showed that 90% Process NUN effectiveness was not achievable with a Pull 
Package based solely on high BRF items. The best we could do was 75% Process 
NUN effectiveness. Using high BRF items as a supplement to a demand-based 
Pull Package provided 90% NUN effectiveness, but the size of the Pull Package 
(12,000 NIINs) was not practical. We determined that it's better to stick 
with only demand-based criteria rather than including high BRF, nondemand 
items. 

During our evaluation, we determined that there were 1,459 NIINs with 
allowances which appear to be able to satisfy the demand from two normal 


32 






patrols. These NIINs, called OBRA, could be deleted as Pull Package can¬ 
didates while increasing the effectiveness for similarly sized packages. 

After reference (2) of Appendix A, we tailored the Pull Package to more 
critical items. Here we mixed and matched the range cutoffs between MEC 
groupings and the number of demands required, so that we could maximize 
Process NUN effectiveness. We could meet the 90% effectiveness goal for 
critical NIINs when we tailored the Pull Package by MEC. TABLE XXIV displays 
the FY89 Pull Package with projections to satisfy 84-89% of the most critical 
NIINs and 73% of all NIINs. (Subsequent actual tests realized higher 
effectiveness rates than we projected.) 

TABLE XXIV 
FY89 Pull Package 





PROCESS 

NUN 

EFFECTIVENESS 

PULL 

PACKAGE 

EFFECTIVENESS 


# NIINs 

COST 

MEC 116 ITEMS WITH DEMAND ON AT 
LEAST ONE PATROL 

869 

$245K 

84 

8 

MEC 110 ITEMS WITH DEMAND ON AT 
LEAST ONE PATROL 

532 

$215K 

89 

7 

MEC 98-107 ITEMS WITH DEMAND ON AT 
LEAST TWO PATROLS 

314 

$ 85K 

70 

12 

MEC 95 ITEMS WITH DEMAND ON AT 

LEAST THREE PATROLS FROM TWO 
DIFFERENT SSBNs 

838 

$126K 

65 

19 

TOTAL 

2553 

$671K 

73 

12 


We then determined how often to update this package. Two possibilities 
were evaluated. They were quarterly (after each boat completes her next 
patrol), and annually. We developed a Pull Package under each possibility and 
compared Process NUN effectiveness. The Process NUN effectiveness changes 


33 





















were no more than two percentage points. We also evaluated churn. With a 
quarterly update, more recent demand is used to build the Pull Package; 
however, churn increases. Thus changing the update frequency creates a trade¬ 
off between stability and responsiveness to new trends of the Pull Package. 

Our results suggest that the Pull Package be updated once a year. This is 
because an annual update causes no significant drop-offs in performance, 
decreases churn, and keeps the process simple, as compared to the quarterly 
update. 

We also evaluated methods to increase Process NUN effectiveness for MEC 
110 and 116 items. We showed that using a longer time horizon for MEC 110 and 
116 items increases the average Process NUN effectiveness from 86% to 89%. 

But this method added 456 NIINs to the package and increased cost by $127K. 

At this point additional data became available, making it possible to add the 
eighth submarine (32 vice 28 patrols) to our evaluation. By doing this, 
Process NUN effectiveness was also raised to about 89%, but fewer NIINS (65) 
were added to the Pull Package, and at a smaller increase in costs ($45K). 
Thus, it appears more reasonable to use all eight submarines with four patrols 
worth of demand data for all MECs rather than trying to use longer demand 
horizons for the higher MEC items. 

In developing the Pull Package initially, we concentrated on which NIINs 
to include without an in-depth analysis of the depth calculations. Under our 
initial method, depth was computed using the simple average of units demanded 
over the 28 patrols. Our alternative method computed the average number of 
units demanded for the patrols that experienced demand. Comparing the two 
methods, we observed that 85% of the quantities differ by three or less units. 
The alternative method had an eight percentage points increase in overall unit 


34 





effectiveness. It also increased MEC 116 unit effectiveness by seven per¬ 


centage points. An actual SCOOP test was conducted using our Pull Package 
with depth computed using the current method. We evaluated a list of par¬ 
tially filled requisitions from two test submarines, to determine how the 
alternative method would fare. The alternative method reduced partially 
filled requisitions 55-61% and increased units satisfied 51-55%. Cost 
increased by 38%. With this increase of requisitions and units satisfied, it 
appears cost effective to use the alternative method to compute depth. 

In terms of OBRA items, there were concerns about our method of computing 
sufficient OBRA items. The concern was that we may have overstate' 1 o tr 
sufficient OBRA items by using the simple average of units demanded over the 
28 patrols. Alternative methods were to compute the average number of units 
demanded for the patrols that experienced demand, and take the minimum of the 
current method plus the 85th percentile difference. Each method was compared 
to an item's COSAL quantity when it was three or greater to determine suffi¬ 
cient OBRA items. A Pull Package was computed using each of these methods. 
The bottom line was that compared to the current method, neither of the other 
two methods made any improvement in unit effectiveness. We concluded that we 
should not change the method of determining sufficient OBRA items. 

VI. RECOMMENDATIONS 


We recommend computing SCOOP Pull Packages by: 

• Using generic data (four patrols and eight UICs). 

• Using demand-based items only. 


35 













• Deleting items from consideration with sufficient on-board 
allowances to satisfy two patrols. 

• Selecting range cuts by MEC. 

• Basing depth on average demand quantity for the patrols that 
experienced demand. 

• Computing Pull Packages once a year. 


36 






APPENDIX A: REFERENCES 


1. Meeting between representatives of FMSO and the SCOOP Logistics Planning 
Team of 11 Jan 1989. 

2. Meeting between representatives of FMSO and the SCOOP Logistics Planning 
Team of 28 Mar 1989. 

3. Meeting between representatives of FMSO and the SCOOP Logistics Planning 
Team of 25 Apr 1989. 


A-l 







'''■vuruv 1*1.»-* I«« .»* t**rt 


DOCUMENT CONTROL DATA • R L D 


' > ■ tint* i l.i' . f /«» .tlfitt i'( t'tlv. !•«•«/% ft . • tr.tr i % m 


t OHiwis* fiiyC *C IiViTt (L'ucpt>'<tffe *im tti of i _ - . 

Navy Fleet Material Support Office 
Operations Analysis Department (Code 93) 
Mechanicsburg, PA 17055-0787 


t R L P O « r TITLE 

SCOOP TRIDENT Repair Part Support Package 




RtPO« T SECuRlTf C L * SSi P i C * T iO» 

_ Unclassified 

2b. CROUP " ~ 




* DESCRIPTIVE NOTES (Type o ( report and inclusive dates) 


S *U TmOHUI (First name, middle initial, last name) 

Brenda M. Klaczak . . 

James L. Fogle 


6 REPOR T OaT£ 

1 February 1990 

?«. TOTAL nO. OP PAGES 

46 

76. no. or urn 

1 

%S. CONTRACT or GRANT no. 

ORIGINATOR’S REPORT N U M O E R(S) 

6. project NO. Z9321-E69-9011 

* 

171 


c. 

9b. OTHER REPORT NOISI (Any other numbera that may be masioned 


this report) 


Li_ 




10 distribution statement 

Distribution of this document is unlimited 



12. SPONSORING MILITARY ACTIVITY 


Adverse conditions at the TRIDENT Refit Facilities (TRFs) may force submarines to obtain 
their replenishment at a non-Navy port. Replenishment requirements (Pull Package) are re¬ 
quired, without knowledge of what was used in the current patrol, to provide the submarine 
with sufficient replenishment support to complete another patrol. The first part of this 
study evaluates alternative methods for computing P-ull Packages. These methods include 
generic versus hull-tailored, demand-based versus the Best Replacement Factor (BRF) based, 
excluding items with sufficient On-Board Replacement Assets (OBRA), and using Military 
Essentiality Codes (MECs). The alternative; are evaluated in terms of effectiveness, size 
of package and cost. The second part of this study examines frequency of update and Pull 
Package refinements. We recommend deleting the OBRA items from consideration and annually 
computing a generic, demand-based Pull Package with range based on MEC and depth based on 
average demand quantity for those patrols experiencing demand. 


DD ,',"“.,1473 (PAGE I) 

5/N OtO 1.807.680 I 


Security Classification 













Distribution List 
Afloat Studies 



Commanding Officer 

Navy Aviation Supply Office 

Attn: SDB4 

700 Robbins Avenue 

Philadelphia, PA 

Commanding Officer 

Navy Ships Parts Control Center 

Attn: Code 041 

5450 Carlisle Pike 

Mechanicsburg, PA 17055 

Commanding Officer 
Naval Supply Center 
Jacksonville, FL 32212 

Commanding Officer 
Naval Supply Center 
Attn: Code 41.3 
Oakland, CA 94625 

Commanding Officer 
Naval Supply Center 
Attn: Code 100 
Norfolk, VA 23512 

Commanding Officer 

U. S. Naval Supply Depot (Code 51) 

FPO San Francisco 96630 (Guam) 

Commanding Officer 

U. S. Naval Supply Depot 

FPO San Francisco 96651 (Subic) 

Chief of Naval Operations 
Navy Department (OP-914) 
Washington, DC 20350 

Chief of Naval Operations 
Navy Department (OP 412) 
Washington, DC 20350 

Commander Submarine Force 
U. S. Atlantic Fleet 
Attn: N5 
Norfolk, VA 23511 


Commander Submarine Force 
U. S. Pacific Fleet 
Attn: Code 41 
Pearl Harbor, HI 96860 

Commander 

Naval Surface Forces 

Naval Amphibious Base, Coronado 

San Diego, CA 92155 

Commander 
Naval Air Force 
U. S. Atlantic Fleet 
Attn: Code 42 
Norfolk, VA 23511 

Commander 
Naval Air Force 
U. S. Pacific Fleet 
Attn: Code 44 

Naval Air Station, North Island 
San Diego, CA 92135 

Commander 

Naval Logistics Command 

U. S. Pacific Fleet (Code 4121) 

Pearl Harbor, HI 96860 

Center for Naval Analyses 
4401 Ford Avenue, Box 16268 
Alexandria, VA 22302-0268 

Naval Postgraduate School 
Attn: Alan McMasters (Code 54MG) 
Monterey, CA 93940 (3 cys) 

Naval Postgraduate School 
Attn: Thomas Moore (Code 54MR) 
Monterey, CA 93940 

Superintendent (Code 1424) 

Naval Postgraduate School 
Monterey, CA 93940 

Naval Postgraduate School 
Attn: Code 55 (MI) 

Monterey, CA 93940 






Afloat Studies 



Commander 

Naval Surface Forces 
U. S. Atlantic Fleet 
Attn: N72 
Norfolk, VA 23511 

Commanding Officer 
Naval Supply Corps School 
Attn: Code 43 
Athens, GA 30605 

U. S. Army Logistics Management Center 
Defense Logistics Studies Information Exchange 
Fort Lee, VA 23801 

Defense Documentation Center 

Cameron Station 

Alexandria, VA 22314 (2 cys) 

Defense Technical Information Center 
Cameron Station 
Alexandria, VA 22314