NPS MDP Study
Outbrief Schedule, 1 JUN 2005
0800-0815
0815-0915
0930-1015
1030-1130
Introductions
Background/Results
Cargo Inspection System (Land)
Cargo Inspection System (Sea)
1130-1230 LUNCH
1230-1330 Sensor System
1345-1445 C3I System
1500-1600 Response Force System
NPS MDP Study
Sea Inspection Group
LT MATT WIENS, USN
MDP System
Operational Architecture
Land Inspection
System
[Detect?]
External
Intel?
Manifest
Info?
Sensor System
Y
[Track COI?]
Sea Inspection
System
[Detect?]
Anomaly?
C3I System
Force System
Y
[Action Taken?]
N
GOOD
OUTCOME
Y
[Action Success?]
N
N
BAD
OUTCOME
Sea Inspection Agenda
•
•
•
•
•
•
•
•
•
•
Bottom Line
Objectives/Requirements
Functional Decomposition
Alternatives Generation
Model Overview
Model Assumptions
Model Factors
Results
Conclusions/Insights
Land Systems Integration
NPS MDP Study System Insights
Land Cargo Inspection
• Effective Cargo Inspection requires
industry cooperation
Sea Cargo Inspection
• Enroute at-sea cargo inspections can be
effective using current sensor technology,
but effective C3I is required
Sea System Group Objectives
• Characterize at-sea ship inspection system
process
• Identify methods to improve container inspection
process and minimize shipping delay
• Develop model for sea inspection system
• Determine driving factors for sea inspection
system
• Recommend system alternatives to implement an
at-sea inspection system
Sea Inspection Requirements
•
•
•
•
•
•
•
Search 80% of ship in 6 hours
9 hr power source
Portable sensor packages
24 hour availability of teams
Compatible with maritime environment
Communications b/w team members
Components exist or viable within five
years
Sea Inspection Functional
Decomposition
“As-Is” SYSTEM
• NO AT SEA INSPECTION OCCURS
•SHIPPING COMPANY RESPONSIBILITY
• ANTI-TAMPER DEVICES ARE OPTIONAL
• CARGO TRACKED BY MANIFEST
FOR OFFICIAL USE ONLY
ALTERNATIVES GENERATION
DETECT
SEARCH
Interior
Exterior
Secret agents
UAV
NEUTRON
Cargo
GAMMA
ARAM (NAI
Detector)
Adaptable Radiation Area
Monitor)
Honor System
Mammals with
sensors
Dog Team
External Portable
Monitor
Manifest
Comparison
check point platform
with multiple sensors
Robotic Inspection
Team with sensors
Robotic Inspection
Team with sensors
LOCATE
Human Inspection
Teams with sensors
AUTOMATED
CHEM/BIO
EXPLOSIVES
APDS (hard
install)
Dog team
Backpack Sensor
APDS (portable)
(HPGe)
High purity Ge
Airborne Particulate Det
Sys
Neutron
Detector
Imaging Device
Swipes
Dogs
Radiation
Sickness
GM Detector
Canaries
Mammals
Imaging
Device?
Robotic Inspection
Team with sensors
TLD (film)
IDENTIFY
Human Inspection
Human Inspection
HUMAN
Teams
with sensors
Teams CLASSIFY
with sensors
INTERPRETATION
Thermo-luminescent
Dector)
TLD (film)
RECOGNIZE Electronic
Dosimeter
Swipes
Symptom
Channel Mounted
analysisCOMMUNICATE
Sonar
INTERNAL
Sensor Mapping
Human Search (smell
Sight)
Data Analysis
Imaging/Display
ARAM (NAI
Detector)
Sodium Iodide
ARAM (NAI Detector)
Radio (UHF, Walkie-Talkie)
Animal Indication
Lab Analysis
Lab Analysis
Encoded Laser
Sickness Locality
Mass Spectrometer
Mass Spectrometer
Cable
Intelligence
FOREIGN
OBJECTS
IR
Ultra Wide Band Radar (LLNL)
VHF Radio
Mammals
Manifest
Anomalies
Wide Band
Radar
Sea Inspection System
Alternatives Overview
ALT 1: “Boarding Team”
• 24 hour rule (manifest data)
• Ships screened by algorithm
• Human Inspection Teams
ALT 2: “Honor System”
• Required level of security standards
• SMART devices on containers
• Human Inspection Teams
ALTERNATIVE 1
BOARDING TEAM INSPECTION SYSTEM
• 24 HOUR RULE FOR MANIFEST DATA
• SHIPS SCREENED BY ALGORITHM
• MANIFEST DISCREPANCIES
• COURSE DEVIATIONS
• SHIP HISTORY
• INTELLIGENCE
• OTHER ANOMALIES
FOR OFFICIAL USE ONLY
ALTERNATIVE 1: BOARDING TEAM INSPECTION SYSTEM
• Teams inspect ship for explosives, contraband and manifest anomalies.
• Review shipping documents for manifest anomalies
• Communications maintained with portable radios
• Portable Gamma Imager
• Spectrometer
• Portable radiation detectors
• Swabs for explosives
• M-8/M-9 paper chemical
• M245A1 nerve agents
• Bio detection device
FOR OFFICIAL USE ONLY
ALTERNATIVE 2
HONOR INSPECTION SYSTEM
● Shipping companies must meet following guidelines to bypass boarding
team inspection.
- SMART container devices working on all containers
- Accurate/Timely manifest data (24 hr limit)
- No SMART container alarms
- “Clean” ship history, including crew
- Port of call accuracy
FOR OFFICIAL USE ONLY
ALTERNATIVE 2: HONOR INSPECTION SYSTEM
• Teams inspect ship for explosives, contraband and manifest anomalies.
• Review shipping documents for manifest anomalies
• Communications maintained with portable radios
• Smart container devices
• Anomaly Detection
• Threat Localization
• Portable Gamma Imager
• Spectrometer
• Portable radiation detectors
• Swabs for explosives
• M-8/M-9 paper chemical
• M245A1 nerve agents
• Bio detection device
FOR OFFICIAL USE ONLY
Overarching Modeling Plan
Pr(detect)
Pr(Defeat)
MOE1
Performance
FY05$
MOE2
Risk (Attack
Damage)
Land Insp
Sea Insp
Delay
time
FY05$
Attack
Damage
Model
Force
Sensors
C3I
Cost Models
Land Insp
Sea Insp
Force
C3I
Sensors
Performance Models
FY05$
Shipping
Delay Cost
Model
FY05$
M1
Commercial
Impact
M2
MDP System
Cost
Sea Inspection Model Assumptions
•
•
•
•
•
•
•
•
Used EXTENDtm for modeling
~10,000 ships/year
Ships detected 300-250 nm from shore
Ships needing inspection wait for teams to be
available
1 hour turn-around time for inspection teams
Teams available around the clock
Boarding team detection = 36 hr detention
Avg US container value was used for delay cost
($ 25K)
Sea Inspection Model Parameters
Factors Considered
• # Inspection Teams/Shift
• Inspection Time
• Soak Time
• P(Rand Inspection)
• P(d) & P(FA) Smart Container
• P(d) & P(FA) Boarding Team
• Team Turnaround Time
• Number of ships/year
RED : Factors varied
Sea Inspection Model Overview
Input Variables
Outputs
• Shipping Data (size, #)
• # Inspection Teams/Shift
• Inspection Time
• Soak Time
• P(random Inspection)
• P(d) & P(fa) Smart Container
• P(d) & P(fa) Boarding Team
Sea Inspection
Performance
Model
• System P(d)
• System P(fa)
• % Ship Inspected
• Delay Time
Sea Inspection Alternative 1
Boarding Team Model Results
Using more than 3 teams/ship has
little effect on delay cost
5% P( Rand Insp) allowed
most inspection without
adding significant delay cost
Sensor false alarm rates higher than 12%
cause a significant delay cost
Sea Inspection Alternative 2
Honor System Model Results
Using 9 teams/shift
significantly reduced
delay costs
1 minute soak time and
1% P(Rand Insp) resulted
in least delay cost
Sea Inspection Model Parameters
Factors
Number of Teams
Inspection Time
Soak Time
P(Random Inspection)
Shipwide P(FA) Smart Containers
Shipwide P(FA) Boarding Team
Neutron
Gamma
Pd - Smart Containers
Chem/Bio
Explosive
Neutron
Gamma
Pd - Boarding Team
Chem/Bio
Explosive
Values
1,3,5,7,9
3,5,6,7,8
1,2,4
1%, 2,5%, 5% , 7.5%, 10%
0.155, 0.198, 0.241, 0.284
0.073, 0.088, 0.103, 0.107
0.5, 0.8
0.6, 0.7
0.3, 0.4
0.1, 0.2
0.1, 0.25
0.1, 0.25
0.2, 0.3
0.4, 0.5
Values
Alt 2 (HONOR)
Alt 1 (BT)
9
3
8
7
1
1
1%
5%
16%
N/A
7%
12%
80%
N/A
70%
N/A
40%
N/A
20%
N/A
25%
25%
25%
25%
30%
30%
50%
50%
Sea Inspection Model Results
for Single Ship
MOE / Metrics
‘As-Is’
ALT 1
ALT 2
Percent Cargo Inspected
0%
100%
100%
P(Detect | Inspect)
0%
25%
25%
P(Detect) WMD on Ship
0%
25%
25%
Comm. Delay Cost ($M)
0
.01
.02
Comm. Cost ($M)
0
0
10
MDP System Cost ($M)
0
17
100
Total System Cost ($M)
0
17
110
* All costs cover 10-year time period
Sea Inspection Model Results
for Sea Inspection System
MOE / Metrics
* All costs cover 10-year time period
‘As-Is’ ALT 1
ALT 2
% Inbound Ships
Inspected
0%
5%
16.5%
P(Detect) WMD on Ship
0%
25%
25%
Overall Pd WMD
Inbound
0%
1.2%
4.1%
Comm. Delay Cost ($M)
0
130
330
Comm. Cost ($M)
0
0
12,680
MDP System Cost ($M)
0
17
100
Total System Cost ($M)
0
147
13,110
Sea Inspection Model Results
Pd on Single Ship vs. System Cost
Pd on Ship
50%
Alt 1
40%
Alt 2
30%
20%
As-Is
10%
0%
$0
$20
$40
$60
$80
$100
$120
10-year Total System Cost (FY05$M)
Overall Pd vs. System Cost
Overall Pd
10%
8%
Alt 2
Alt 1
6%
4%
As-Is
2%
0%
$0
$2,000
$4,000
$6,000
$8,000
$10,000 $12,000 $14,000
10-year Total System Cost (FY05$M)
Pareto Chart of Effects
(Delay – Alternative 1)
Pareto Chart of the Standardized Effects
(response is Delay Cost (FY05$M), Alpha = .05)
1.96
F actor
A
B
C
D
E
A
AD
BD
AB
D
N ame
P (Random Inspection)
Inspection Time
S oak time
N um Teams
P (F A ) BT
Term
B
E
P(Random Inspection) had
greatest effect on
Alternative 1 Delay cost
DE
BE
AE
C
AC
BC
CE
CD
0
10
20
30
Standardized Effect
40
50
Pareto Chart of Effects
(Delay – Alternative 2)
Pareto Chart of the Standardized Effects
(response is Delay Cost (FY05$M), Alpha = .05)
Term
2.0
F actor
A
B
C
D
E
F
C
B
E
CE
AC
BC
A
D
BE
CD
BD
AB
AE
DE
AD
F
EF
AF
CF
DF
BF
Number of Inspection Teams
per shift had greatest effect
on Alternative 2 delay cost
0
50
100
150
Standardized Effect
200
N ame
Inspection Time
S oak time
N um Teams
P (Random Inspection)
SC F A
BT F A
Sea Inspection
Conclusions/Insights
• Enroute at-sea cargo inspections can be
effective using current sensor technology, but
effective C3I is required
• Smart Container benefits possible, but at great
expense
• # Teams/shift, P(Random Inspection) and Soak
Time had greatest effect on system performance
and delay cost
• Deterrence factor of boarding teams may justify
cost of boarding teams
RECOMMENDATIONS
• Continue development of sensor technologies to
increase P(d)
• Money would be best spent on increasing
#Teams/shift to minimize delay cost
• Develop algorithm for AIS data to augment the
inspection selection process
• Develop sensor mapping feature of SMART
container devices to localize anomalies more
accurately
TDSI: Land Systems Integration
• Objective
• Assumptions
• Search Model
• Results
• Conclusions
Objective
• To determine search parameters for input
to the Sea Inspection Model for the
efficient search of container ships using
portable detection systems.
Assumptions
• Focus on search strategies
– Ship-board human search
– Ship-board robotic system as an alternative
– Search strategies matched with detector capabilities
• Performance of inspection entities and detectors
– Not an “As-Is” systems; capability not proven
– Search capability of robotic platforms assumed
• Real-life search times may vary
– Lack of precise performance specs of detectors
Search Model
• Search environment
– Large container ships
• ~ 4000 to 8000 TEUs
– Containers: 20 ft x 8.5 ft x 8 ft
• Search entity
– Human with portable equipment
• Deck search 6 containers/min
• Stack search 1.8 containers/min
• Detector
– Passive Gamma Ray detector (portable germanium detector)
– Excel model to calculate detection distance for U235 @ 186keV
and Pu239 @ 414keV
– Can detect 25kg HEU with ¼ “ lead shielding at 4.2m
– 1, 2 or 4 min soak time
Search Model
• Search parameters
– Detector soak time
– Detection range
– Number of search entities
– Container vessel size
– Container stacking configuration
• MOP: Search time for container stack(s)
Results: Variation of stack search time with
team size
Stack Search Time
(4 min soak time, 10x13 TEU)
4.5
4.20
Stack Search Time (hr)
4.0
3.5
3.0
Detection range
2.5
4m
Optimum team size 8-15 inspectors
2.0
6m
2.00
8m
1.5
1.04
1.0
0.85
0.5
0.41
0.22
0.39
0.20
0.11
0.0
0
5
10
15
20
25
Team Size
0.20
0.10
0.06
30
35
40
45
50
Results: Variation of search time with
detection range and soak time
Search Time vs Detection Range
(4 teams of 10 men, 4680 TEUs)
9
8
7.89
Increasing detector range has more
significant impact on performance
than reducing soak time.
Search Time (hrs)
7
1.7 fold
6
5
4.53
4.45
4
3.83
3
2.85
2.65
2
2.33
1.75
1.58
1
1.97
1.97
1.25
0.89
1.25
0.89
G
O
4 fold O
D
N
E
S
S
0
3
4
5
6
Detection Range (m)
7
8
9
Sensor Soak Time
4 min soak time
2 min soak time
1 min soak time
Autonomous systems: Basic
Robotic Behaviors
¾ In order of decision priority
levels:
¾ Obstacle Avoidance
¾ Dispersion
¾ When density of robots
scanning target sources gets
large
¾ When density of robots in
vicinity gets large
¾ Sense & Scan
¾ Trail Marking Monitoring
Dispersion
¾ Turn away when trail marking
counter for route ahead is
higher or when current
counter is high
¾ ELSE proceed as normal if
current trail marking counter
is much higher than the one
ahead
Sense & Scan
Trail
Marking
Model results
Comparison of Human vs. Autonomous Search Times
(for 3 rows of 6 containers)
900
Optimum achieved with 4
human search entities
818
800
Search Time (seconds)
700
Convergence towards human
search time with greater than 7
robotic search entities
600
527
500
Search Entities
human
autonomous
400
340
300
233
200
187
170
83
100
72
68
58
0
0
209
2
4
58
58
6
59
67
66
57
64
55
63
58
8
10
Number of Search Entities
12
14
16
Conclusions
• Optimum team size is 8-15
• Parameters of significance:
– Soak Time
– Detection Range
• Performance
– Coordinated human search strategy
outperforms autonomous search algorithm
modeled
Questions?
Sea Inspection Group
CPT Ali Serdar Sari
LT Matt Wiens
LT Shannon Hoffmann
LT Tracey Crider
TDSI Land Inspection Group
MAJ Wee Tuan Khoo
CPT Joel Lim
CPT Alan Yeoh
Back-Up Slides
Back-Up
½ inch Pb shielding
.635cm pb shielding
Field strength at 100 m
1.64x10-4 counts/m^2-sec
Field strength at 3 m
.3621 counts/m^2-sec
Assume .635cm steel hull
-25 Kg Wg U-235
-ρ steel 7.85 g/cm^3
-ρ lead = 11.34 g/cm^3
-µ/ρ steel .138cm^2/g
-µ/ρ Pb = .896 cm^2/g
Gamma .19Mev from U-235
Specific Activity U-235 2.162 X10^-11 Ci/g
Build-up factor approx. 1.35
Sea Inspection Model Overview
Alternative 1
Boarding Team Results
Interaction Plot (data means) for Delay Cost (FY05$M)
3
5
6
7
8
1
2
4
1
3
5
7
73
88
03
17
00
0.0 0.0 0. 1 0. 1 0. 4
300
P(Random
Inspection)
0. 010
150
P(Random Inspection)
0. 025
0. 050
0. 075
0
300
0. 100
Inspection
Time
3
150
Inspection Time
5
6
7
0
300
8
Soak
time
1
150
Soak time
0
300
2
4
Num
Teams
1
150
Num Teams
3
5
7
0
P(FA) BT
Alternative 2
Honor System Results
Interaction Plot (data means) for Delay Cost (FY 05$M)
1
2
4
3
5
7
9
0.
0
5
0
5
5
0
15
01 . 0 2 .0 5 .0 7 .1 0
0
0
0.
0
0
0.
8
19
0.
1
24
0.
4
28
0
0.
73
0
0.
88
1
0.
03
1
0.
17
I nspection
Time
3
4000
Ins pection Tim e
5
6
2000
7
8
0
Soak
4000
Soak tim e
2000
time
1
2
4
0
N um
4000
Num Team s
Teams
3
5
7
2000
9
0
P( Random
P (Random Ins pection)
4000
I nspection)
0.010
2000
0.025
0.050
0.075
0.100
0
SC FA
SC FA
4000
0.155
0.198
2000
0.241
0.284
0
BT FA
Pareto Chart of Effects
(Performance)
Pareto Chart of the Standardized Effects
(response is Percentage Completed, Alpha = .05)
Term
2.0
F actor
A
B
C
D
E
F
F
BF
E
D
B
EF
DE
DF
BE
BD
A
AB
AE
AD
AF
CE
C
BC
CD
CF
AC
0
20
40
60
80
100 120
Standardized Effect
140
160
180
N ame
Ty pe
A lternativ e
P (Random Inspection)
Inspection Time
S oak Time
N um of Teams
Alternative 1
Delay Cost vs. Inspection Time & # Teams
Alternative 2
Delay Cost vs. Inspection Time & # Teams
Alternative 1
Boarding Team Model Results
• Performance
– Pd - 25%
– Average Delay Time
• 0.21 hrs/ship
– Ave Percent Inspected
• 100%
• Cost
– System Cost
• $16.93M (3 Teams/Shift)
– Commercial Cost
• $0
– Ave Delay Cost
• $13.17M
Alternative 1
Boarding Team Model Results
System Costs
95% Confidence Interval of 10 Year System Cost V.
Total Number of Inspection Teams
10 Yr Cost (In Millions)
$50.00
$40.00
$30.00
$20.00
Low
Expected
High
$10.00
$0.00
9
1 teams/ shift
27
3 teams/ shift
45
5 teams/ shift
Total Num ber of Inspection Team s
ALL COSTS ARE IN FY05$
63
7 teams/ shift
Alternative 2
Honor System Model Results
• Performance
– Pd - 66.41% (SC)
– Pd - 24.18% (BT)
– Average Delay Time
• 0.53 hrs/ship
– Ave Percent Inspected
• 100%
• Cost
– System Cost
• $100M (9 Teams/Shift)
– Commercial Cost
• $12.68B
– Average Delay Cost
• $32.69M/ship-year
Alternative 2
Honor System Model Results
System Costs
10 Yr Cost (In Millions)
95% Confidence Interval of 10 Year System Cost V.
Total Number of Inspection Teams
$120.00
$100.00
$80.00
$60.00
$40.00
Low
Expected
High
$20.00
$0.00
27
3 teams/ shift
45
5 teams/ shift
63
7 teams/ shift
Total Number of Inspection Teams
ALL COSTS ARE IN FY05$
81
9 teams/ shift
Model Flow Chart
Honor System Model Results
Commercial Costs
95 % Confidence Interval of Commercial Cost for Alternative
2
10 Year Cost (In Billions)
ALL COSTS ARE IN FY05$
$25.00
$20.00
$15.00
$10.00
$5.00
$0.00
Low
Expected
High
Range of Values for Commercial Cost
Note: Commercial Costs will remain the same for all team sizes. Cost is a factor of how many
SMART containers are implemented each year. For the model, it was assumed that all shipping
containers would have the SMART container technology within the first year and a 5% replacement
each year thereafter.
95% Confidence Interval obtained using Triangular Distribution
System Cost vs.
WMD Pd & Delay Cost
3 teams/shift and 9 teams/shift
were selected during design
analysis as best value
3 teams/shift and 9 teams/shift
were most cost efficient factors
Percentage of ships inspected
100%
80%
60%
40%
20%
0%
As-Is
Alt 1
Alt 2
Honor System (Alt 2)
Inspects 3 times as
many ships as Alt 1