AN ADAPTIVE CYBERINFRASTRUCTURE FOR THREAT MANAGEMENT IN URBAN WATER DISTRIBUTION SYSTEMS Kumar Mahinthakumar North Carolina State University DDDAS Workshop, ICCS 2006, Reading, UK May , 2006

2 Our Team North Carolina State University  Mahinthakumar, Brill, Ranji (PI’s)  Sreepathi, Liu, Vankalaya (Grad Students)  Zechman (Post-Doc) University of Chicago  Von Laszewski (PI)  Haetgen (Post-Doc) University of Cincinnati  Uber (PI)  Feng (Post-Doc) University of South Carolina  Harrison (PI) Greater Cincinnati Water Works

3 Outline Overall Scope of Project Test problem Preliminary Results

4 Many security threat problems in civil infrastructure systems

5 Water Distribution Security Problem

6 Water Distribution Problem

7 Why is this an important problem? Potentially lethal and public health hazard Cause short term chaos and long term issues Diversionary action to cause service outage  Reduction in fire fighting capacity  Distract public & system managers

8 What needs to be done? Determine  Location of the contaminant source(s)  Contamination release history Identify threat management options  Sections of the network to be shut down  Flow controls to Limit spread of contamination Flush contamination

9 DDDAS Aspects Dynamic  Data  Optimization  Simulation  Workflow  Computer Resources Data Driven and Vice Versa  Water Demand Data  Water Quality Data

10 Grid Resource Broker and Scheduler Adaptive Simulation Controller Adaptive Optimization Controller Sensors & Data Our Cyberinfrastructure Mobile RF AMR Sensors Static RF AMR Sensor Network Static Water Quality Sensor Network Resource Needs Resource Availability Bayesian Monte- Carlo Engine Optimization Engine Simulation Model Grid Computing Resources Adaptive Wireless Data Receptor and Controller Deci sions Dat a Adaptive Workflow Portal Algorithms & Models Middleware & Resources Model Param eters Model Outputs

11 Key DDDAS Developments Algorithm and Model Development  Dynamic Optimization  Bayesian Data Sampling and Probabilistic Assessment  Model Auto Calibration  Model Skeletonization  Network Assessment using Back Tracking Middleware Development  Adaptive Workflow Engine  Adaptive Resource Management  Controller Designs Cincinnati Application Scenario Development  Source Identification  Sensor Network Design  Flow control design

12 Water Distribution Network Modeling Solve for network hydraulics (i.e., pressure, flow)  Depends on Water demand/usage Properties of network components  Uncertainty/variability  Dynamic system Solve for contamination transport  Depends on existing hydraulic conditions  Spatial/temporal variation time series of contamination concentration

13 Source Identification Problem Find: L(x,y), {M t }, T 0 Minimize Prediction Error  ∑ i,t || C i t (obs) – C i t (L(x,y), {M t }, T 0 ) || where L(x,y) – contamination source location (x,y) M t – contaminant mass loading at time t T 0 – contamination start time C i t (obs) – observed concentration at sensors C i t (L(x,y), {M t }, T 0 ) – concentration from system simulation model i – observation (sensor) location t – time of observation unsteady nonlinear uncertainty/error

14 Interesting challenges Non-unique solutions  Due to limited observations (in space & time) Resolve non-uniqueness Incrementally adaptive search  Due to dynamically updated information stream Optimization under dynamic environments Search under noisy conditions  Due to data errors & model uncertainty Optimization under uncertain environments

15 Resolving non-uniqueness Underlying premise  In addition to the “optimal” solution, identify other “good” solutions that fit the observations  Are there different solutions with similar performance in objective space? Search for alternative solutions

16 Resolving non-uniqueness Search for different solutions that are far apart in decision space x f(x)

17 Resolving non-uniqueness… x f(x) Effects of uncertainty

18 Resolving non-uniqueness… x f(x) Search for solutions that are far apart in decision space and are within an objective threshold of best solution

19 Resolving non-uniqueness… EAs for Generating Alternatives (EAGA) Create n sub populations Sub Pop 1 Evaluate obj function values Best solution (X*, Z*) Evaluate pop centroid (C 1 ) in decision space Selection (obj fn values) & EA operators STOP? Best Solutions Sub Pop 2 Evaluate obj function values Feasible/Infeasible? Evaluate distance in decision space to other populations Selection (feasibility, dist) & EA operators STOP? NYN Y

20 Non-uniqueness

21 Simulation Code EPANET from USEPA Originally for Windows environments Ported to Unix platforms including TeraGrid architectures

22 Current Search Algorithms Classical Random Search (CRS)  Easy to implement  Massively parallel  Tight efficient coupling with EPANET (C) Evolutionary Algorithms (EA)  Stand alone optimization toolkit (Java) Loose coupling with EPANET through files  Specific Algorithms Standard GA Dynamic Optimization (ADOPT)

23 Dynamic optimization Minimize Prediction Error  ∑ i,t || C i t (obs) – C i t (L(x,y), {M t }, T 0 ) || C i t (obs) – streaming data Objective function is dynamically updated Dynamically update estimate of source characteristics

24 Dynamic optimization… Underlying premise  Predict solutions using available information at any time step  Search for a diverse set of solutions (EAGA)  Current solutions are good starting points for search in the next time step

25 Preliminary Architecture MPI EPANET Optimization Toolkit Sensor Data Karigen Workflow Engine Grid Resources

26 Example Test Problem Single source contamination Unknowns  Source Location  Start Time  Release History Algorithms  Classical Random Search  Evolutionary Computation

27 Single Source Contamination Problem

28 Single source contamination problem

29 Single source contamination problem Explain the contamination issues Show animation

30 Solution Comparison ERROR CRS = 6.5%, ADOPT = 5.7%, GA = 10.0 %

31 Preliminary Architecture Parallel EPANET(MPI) EPANET-Driver Optimization Toolkit Sensor Data Karigen Workflow Engine Grid Resources EPANET

32 Speedups on local cluster

33 Java Toolkit Overhead

34 Java Toolkit Overhead

35 Speedups on local cluster

36 Summary A cyberinfrastructure is being developed for water distribution threat management Adaptive dynamic optimization algorithm shows promise for detecting contamination sources

37 What’s Next? Dynamic optimization for determining optimal location of sensors and optimal sampling frequency True integration of workflow engine into the cyberinfrastructure Backtracking to improve source identification search efficiency

38 Questions?