Leanne Viera
IBM GBS Public Sector Advanced Analytics Service Area Lead
Sung Seo
IBM GBS Public Sector Strategy and Analytics Consultant
2 May 2014
Disaster Response Planning and Logistics
A Unique Challenge
© 2014 IBM Corporation
The requirements for response to disasters is growing
200
Total Disasters
Major Disasters
100
2013
2010
2007
2004
2001
1998
1995
1992
1989
1986
1983
1980
1977
1974
1971
1968
1965
1962
1959
1956
1953
0
§  Most don’t realize the number of national disasters
each year and the full requirements of a response
§  What is the timeframe that drive the response
timeline?
§  What impact does policy have?
§  To what extent should survivors be considered self
sufficient?
2
© 2014 IBM Corporation
Pandemic
capability to respond to multiple
Katrina
disasters regardless of their severity
Synchronized
9/11
Disaster Density
§  Density: Scaling responses to
account for variances in populations
and infrastructure impacted by the
Andrew
Collaborative
Ike
So Cal Fires
Integrated
Maturity Arcs
§  Multiplicity: Maintaining the
High
Examining a framework for looking at Disaster Surge Requirements
Gustav
disaster
Oklahoma City
Functional
§  Intensity: Adapting responses to
of varying intensities
3
Tornadoes
Static
Low
provide adequate support to disasters
Hannah
Disaster Intensity
Low
High
© 2014 IBM Corporation
The objective of supply chain disaster response is theoretically
simple…but operationally and organizationally complex
Survival Needs
coupled with a hoarding mentality
Transportation
with damaged infrastructure
and constrained fuel
Supplier Capacity
if number of functioning
suppliers is reduced
Inventory Planning
Safety Stock for multiple sequential
or simultaneous disasters
Distribution
with law enforcement
restrictions and constraints
due to damage or looting
Uncertain disaster density,
intensity, and multiplicity
These challenges defy blanket application private sector best practice and defy
translation of requirements and complexities into a traditional model
4
© 2014 IBM Corporation
Intense public scrutiny only intensifies the pressure
“Tons of Food Spoiled As FEMA Ran Out Of Storage Space”
(Washington Post, 4/12/07)
“Food stored by FEMA spoils”
(United Press International, 4/13/07)
“Stockpiled Food Goes Bad”
(CNN: The Situation Room, 4/13/07)
“FEMA let millions of meals rot on Gulf Coast”
(Washington Post, 4/13/07)
“FEMA Wastes $40 Million in Food for Katrina”
(CBS News, 4/13/07)
“Another FEMA goof”
(Bradenton Herald, 4/26/07)
“Pre-Prepared Meals Wasted”
(CNN: American Morning, 4/26/07)
“FEMA To Melt $24 Million In Unused
Ice”
(Congressional Quarterly 7/16/07)
“FEMA Ice Goes To Waste”
(WVUE-TV-New Orleans 7/4/07)
“FEMA Dumps Ice In Storage”
(FOX Report, 7/4/07)
“Doesn’t FEMA Know There Are Starving Children? $40 million in Food Thrown Away”
(Associated Content, 4/13/07)
5
© 2014 IBM Corporation
No matter the preparation, there will always be the unknown and
unexpected
6
© 2014 IBM Corporation
Competitive Advantage
Applying analytics to the landscape of disaster response
Stochastic Optimization
How can we achieve the best outcome
including the effects of variability?
Optimization
How can we achieve the best outcome?
Predictive modeling
What will happen next if ?
Forecasting
What if these trends continue?
Simulation
What could happen…. ?
Alerts
What actions are needed?
Query/drill down
What exactly is the problem?
Ad hoc reporting
How many, how often, where?
Standard Reporting
What happened?
Degree of Complexity
7
Prescriptive
Predictive
Descriptive
Based on: Competing on Analytics, Davenport and Harris, 2007
© 2014 IBM Corporation
Disaster Supply Chain Response Simulation
© 2014 IBM Corporation
Defining a model to address the supply chain challenge
Challenge
Distribute emergency supplies such as water,
meal, blankets, generators and tarps to disaster
victims in the event of various natural and
man-made disasters in a timely and effective
manner to save and sustain lives, minimize
suffering, and protect property
Technical Solution
Agent-based simulation
•  Simulate the flow of emergency supplies and
optimize the selection of shipment destinations
and cross-shipping between distribution points
Benefits
•  Assess the effectiveness of disaster response
plan for distributing emergency supplies to
disaster victims
•  Evaluate the impact of transformation/
improvement initiatives
•  Identify bottlenecks, risks and improvement
opportunities
•  Provide decision support for developing an
effective disaster response plan
Large-scale logistics optimization algorithms
•  Optimized cargo or personnel movements to
maximize time-definite delivery at lowest overall
logistics costs
Web-enabled dashboard
•  Geographic information system (GIS) for
capturing, managing, analyzing, and displaying
geographically referenced information.
9
© 2014 IBM Corporation
New Madrid Scenario
Intra-plate earthquake in the southern and mid-western United States
•  Earthquake occurs at 2:00 AM on February 7th •  Event represents rupture of three New Madrid Fault segments simultaneously with magnitude, Mw = 7.7 •  Modeled as sequenEal rupture of individual segments similar to 1811-­‐1812 series of events over several months •  Severe shaking occurs throughout western Kentucky, Tennessee, southeastern Missouri, northeastern Arkansas and southern Illinois with localized amplificaEon of ground shaking due to variaEons in soil condiEons •  Significant ground deformaEons likely in soO soils, parEcularly near riverbeds •  AOershocks around magnitude 6, ~Mw 6, are likely in the days and weeks aOer the main shock FEMA Region
State
Alabama
Kentucky
Region IV Mississippi
Tennessee
Total RIV
Illinois
Region V Indiana
Total RV
Arkansas
Region VI
Total RVI
Missouri
Region VII
Total RVIi
Total
Water
Liters Truckloads
1,823,169
2,641,557
2,225,166
6,765,444
13,455,336
2,044,269
1,755,087
3,799,356
3,045,516
3,045,516
2,706,450
2,706,450
102
147
124
376
749
114
98
212
169
169
150
150
3,500 Truckloads of CommodiPes 23 M liters of water 15.3 M MREs Commodities (First 72 hours)
MREs
Ice
Number Truckloads
Pounds
Truckloads
1,215,446
1,761,038
1,483,444
4,510,296
8,970,224
1,362,846
1,170,058
2,532,904
2,030,344
2,030,344
1,804,300
1,804,300
56
81
68
207
412
63
54
117
93
93
83
83
4,861,784
7,044,152
5,933,776
18,041,184
35,880,896
5,451,384
4,680,232
10,131,616
8,121,376
8,121,376
7,217,200
7,217,200
122
176
148
451
897
136
117
253
203
203
180
180
23,006,658 1,280 15,337,772 705 61,351,088 1,533
Impacted Counties (IC)
Alabama
Arkansas
Illinois
Indiana
Kentucky
Mississippi
Missouri*
Tennessee
Total IC
Water
MREs
Liters Truckloads Number Truckloads
-­‐
1,646,337
1,304,613
521,913
1,073,664
1,162,497
2,227,620
4,713,111
12,649,755
-­‐
92
68
27
92
65
119
262
723
-­‐
1,097,558
869,742
347,942
715,776
774,998
1,485,080
3,142,074
8,433,170
-­‐
68
34
12
33
36
68
145
396
Ice
Pounds Truckloads
-­‐
4,390,232
3,478,968
1,391,768
2,863,104
3,099,992
5,940,320
12,568,296
33,732,680
-­‐
148
81
30
72
77
148
314
872
Source: New Madrid Seismic Zone Catastrophic Event Planning Project. Mid America Earthquake Center, Virginia Tech, George Washington University (2010) 10
© 2014 IBM Corporation
Modeling Challenges for Disaster Response
Goals
–  Best possible coverage
–  Allocation/Adjudication
–  Transportation planning and scheduling
–  Sensitivity analysis
Challenging Factors
–  Non-stationary demands and unstable supply chains
–  Uncertain/unknown demands and capacities
–  Multiple sourcing from dynamic supply points
–  Joint sourcing and transportation decisions with
disruptions
–  Adjudication/Allocation for fairly balancing coverage
11
© 2014 IBM Corporation
New Madrid Scenario
Coverage simulation: Sample Snapshot
12
© 2014 IBM Corporation
Key Takeaways
§  There is no perfect solution
–  Mitigation of damage – the magnitude and timing
of the disaster prevents perfect coverage
–  Application of sophisticated analytic techniques
could help drive coverage improvements of up to
9.6% leading to an estimated 57.1% coverage for
the given scenario constraints.
–  Seemingly modest improvement would still result
an additional 1.5M days of water rations with
improved distribution
§  Location flexibility and additional scalability of
potential distribution of staging areas can drive
substantial improvements
§  Scalability of supply capacity becomes both a critical
success factor and challenge
13
© 2014 IBM Corporation
Mobile adoption continues to explode
41%
Compound
Annual Growth Rate
(CAGR)
Wearable
Wireless
Devices
1 Trillion
Connected
Sensors
B2C
B2B
B2E
5.6
Billion
Personal
Devices
Sold
2013
2014
M2M
2015
Source: ABI Research, Cisco ‘The Internet of Things: How the Next Evolution of the Internet is Changing Everything’, Cisco Visual Networking
Index (VNI) Global Mobile Data Traffic Forecast for 2010-2015, IBM Analysis
14
© 2014 IBM Corporation
The onset of social and mobile presents a wide range of new
opportunities
§  Mobile represents a shift from a hardened
infrastructure to one more individual and flexible
–  Provides an avenue to push and receive real time
information immediately
–  Can be used to identify and locate victims in need
§  Hurricane Sandy showed the power utilizing mobile
and social media to generate and transmit data
–  Agencies pushed information through twitter and
Facebook and manage information
–  FEMA and the Red Cross mined tweets and posts
for actionable information that were out to
response teams
–  Disaster reporter apps receive structured disaster
reports
15
© 2014 IBM Corporation
Applying analytics to growing amounts of unstructured data to glean
insights?
§  How much do we leave untapped if we
don’t go after the unstructured data?
§  What innovative applications can we
apply to drive more effective disaster
response?
§  What are the challenges that come with
the data?
16
© 2014 IBM Corporation
Thank you!
§  Leanne Viera, IBM Partner, viera@us.ibm.com
§  Sung Seo, IBM Senior Managing Consultant, sungseo@us.ibm.com
© 2014 IBM Corporation