How can we design ad-hoc supply chains that account for the overwhelming uncertainties in the early phases of disaster response? This presentation outlines a scenario-based approach that combines methods from Operations Research, Multi-Attribute Decision Making and Scenario Planning.
Presentation at the 22nd International Conference on Multiple Criteria Decision Making in Malaga, Spain: http://www.mcdm2013.decytec.ccee.uma.es/
Decision Support for Robust Humanitarian Relief Supply Chains
1. Scenario-based Multi-Criteria
Decision Support for Robust
Humanitarian Relief Supply
Chains
Tina Comes
Centre for Integrated Emergency Management
University of Agder
tina.comes@uia.no
Frank Schätter
Karlsruhe Institute of Technology
2. CRISIS AND EMERGENCY
MANAGEMENT
An Era of Change
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
3. What is the difference?
Standard Operations Emergency management
Lead Time Sudden onset
Deliberate and pro-active Reactive
Comprehensive information Partial and heterogeneous information
“No measure” an option “No measure” not an option
Events and developments
foreseeable
Events unforeseen, cause-effect chains can only
be discovered in hindsight
Locations known Unpredictable location
Duration of projects planned Uncertain duration
Forecasts relatively reliable Information dynamically evolving and
heterogeneous; impact can hardly be assessed
Efficiency Effectiveness
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
4. Decisions in emergency management
• Pressure, limited time to make a decision, bounded availability of experts
• Various actors and organisations with different aims, values and perceptions
• Risk of information overload
à prone to cognitive biases
San Bruno Pipeline Explosion, 2010
San Francisco Chronicle
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
5. New opportunities…
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
Sutton, J., Johnson, B., Spiro, E., and Butts, C. (2013).
“Tweeting What Matters: Information, Advisories, and Alerts
Following the Boston Marathon Events.” Online Research
Highlight. http://heroicproject.org
6. Trends: Digitalised Societies
Smartphones are everywhere!
• Provide information to citizens
• Improved situational awareness
and basis for decision support
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
New Challenges
• Heterogeneous Information
• Privacy, trust, liability
• Information overload
• Continuous update
7. Keys to decision-making in complex environments
Understand the context as dynamically evolving
complex system
• focus on relations and developments
(not on individual variables and states)
• identify key drivers of systems’ change
• understand key weaknesses and tipping points
Risk management as continuous processes
à listen to and work with the system to determine where and how to
intervene (feed forward)
à learn from the new information about the system (feedback)
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
8. DISASTER RELIEF SUPPLY
CHAIN MANAGEMENT
Saving the world!
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
9. Decision support in SCM
Business processes and standard operations
• Optimization
• Focus on efficiency, e.g. profit maximisation
… and the risks?
Crisis management
• Precautionary principle
• Focus on effectiveness, e.g. service levels
… at which cost?
Trade-off between effectiveness and efficiency
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
10. Better decisions:
what is a good supply network in emergencies?
The challenges
- How to create ad-hoc networks
of heterogeneous
organisations, groups and
individuals?
- How to align goals and
preferences?
- How to design a matching
network for efficient
information collection,
processing and sharing?
- How to establish flexible and
agile supply networks to
manage and complexity?
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
Strategic: What? Establish objectives
and policies! network design
Tactical: How much?
Deploy resources!
Forecasts, logistics and inventory plans
Operational: When? Where?
Schedule, monitor and adjust!
Scheduling, tracking
Execution: Do!
The decisions
11. Two approaches
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
Bertsimas & Thiele (2004): A Robust Optimization Approach to
Supply Chain Management.
Integer Programming and Combinatorial Optimization, Lecture
Notes in Computer Science Volume 3064, pp 86-100
Chile - Earthquake March 2010, International
Federation of Red Cross Societies, http://
www.ifrc.org/en/news-and-media/photo-galleries/
2010/chileearthquake-march-2010/
12. Humanitarian relief logistics
Aims
Distribute the right goods to the right destinations in time
Complexity and uncertainty
• Critical infrastructure failures
• Lacking and uncertain information
• Respect the context!
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
13. Warehouse Location for Haiti Earthquake
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
14. Facility location for the effective and efficient supply of
disaster relief goods
How good is a
decision?What could happen?
Best Locations
for the situation?
Scenarios Select!
Optimise!
Effective: supply those in need
Efficient: no waste of resources
What to do?
Structured re-design of
alternatives
Robustness and
Flexibilty
What could go wrong?
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
15. Optimisation
Minimise the transportation times and fixed costs of warehouses
How?
Use of quick heuristics to
• Facilitate updates
• Explore more scenarios
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
Dijkstra
algorithm
(shortest
paths)
ADD-
heuristic
(solving FLP)
Optimal
allocation
16. How to design robust options?
• Combinatorial explosion
• Information overload
• Risk averseness
Aim: filtering of options and most relevant scenarios for decision
Measured by stability and quality indicators
1. Maximum number of location changes required
2. Relative loss
3. Regret
Basis for next iterations!
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains
19. Thank you!
Iterative approach for
decision support in the
design of humanitarian
relief supply networks
• Integrate effectiveness and
efficiency by using stability and
quality measures
• Combination of an optimisation
model, scenario-based
techniques and MAVT
• Scenario construction targeted
at risks and vulnerabilities
Future work
• Integration of information from
local sources and ‘zooming in’
• Intervention points
• Number of warehouses
Contact
tina.comes@uia.no
27/06/2013Comes & Schätter: Decision Support for Robust Humanitarian Relief Supply Chains