FY 2001 RESEARCH AND DEVELOPMENT PLAN FOR CERTS COMPLEX SYSTEMS APPROACH TO CASCADING FAILURES

Objective

The objective of this work is to build on recent ORNL/PSerc advances in complex dynamics modeling techniques in an efficient and cost-effective manner to help characterize the nation’s large, complex nonlinear electric power transmission system, focusing on complexity, uncertainty effects, and system criticality behavior. The rapidly increasing complexity of today’s energy system presents significant challenges for maintaining its operational stability and reliability. The U.S. energy supply and delivery system is experiencing unprecedented changes and will continue to be operated closer to a stressed condition in which there is substantial risk of cascading outages. The dynamical models developed by ORNL/PSerc provide a means that will support simulation studies to be done on desktop machines in relatively short run times. This advantage allows for numerous evaluation studies to be done in a fast and cost-efficient manner.

The deliverables will be analytical tools (i.e., software for network enabled complex system analysis programs). This project will develop the basic tools needed to represent demand-driven complex dynamical systems for electrical power transmission systems. The project will investigate the nature of the SOC-like and other complex dynamics concepts in cascading failures so that prospects for controlling these global dynamics to mitigate catastrophic failures can be assessed.

Vision

The overall vision is that management of the electrical power network to avoid catastrophic blackouts should account for the global dynamics of series of these blackouts and interdependencies between blackouts. Analysis of historical NERC show power tails in the distribution of blackout sizes; that is, there are many more large blackouts than expected. Simulation results to date also show underlying nontrivial complex dynamics and interdependencies. These complex dynamics between blackouts are of great significance if one wants to operate the system to avoid blackouts. For example, in other complex systems showing self-organized criticality, well-intentioned policies to avoid small "blackouts" can inadvertently lead to an increased frequency of large "blackouts". The project is designed to establish models and tools for a new global approach to cascading failures in stressed power systems, which will be complementary to the valuable and traditional efforts to analyze cascading events on a more individual cause-and-effect basis.

Overall project plan

The first year of the project will concentrate on improving, implementing and understanding models capturing the complex dynamics of series of blackouts. The second year of the project will improve the realism of the models and data, try to reproduce qualitative features of historical NERC data on blackouts, and will assess the prospects for controlling the complex model dynamics to mitigate or avoid large blackouts.

First year tasks and lead investigators

For planning purposes we divide the first year project activities into the following tasks. The tasks build on the promising initial work at ORNL, Wisconsin and Cornell that is documented in the January 2001 HICSS conference session on self-organized criticality.

1.  Improve global models of series of cascading failures (lead Dobson)

2.  Implement global models in software (lead Carreras)

3.  Develop test networks (lead Dobson)

4.  Study models by running software on test networks; develop diagnostics and analyze and interpret results (lead Carreras)

5.  Project first year report in November 2001 (lead Dobson)

Critical steps, task detail, milestones and coordination needed

Task 1: Improve global models of series of cascading failures.

(a) Clarify and develop representations of outages, overloads, and system memory between blackouts (Dobson, Carreras, Newman). Coordination between Wisconsin and ORNL and advice from David Newman of the University of Alaska is needed. Milestone: description of representations May 2001.

(b) Search for heuristics, theory and simplifications that can lead to better understanding (Carreras, Dobson, Thorp). Any successful approaches will be documented in the first year report in November 2001.

Task 2: Implement global models in software

(a) Assess need for new LP solver to be able to simulate networks of more realistic size (Carreras, Dobson). The main coordination needed is a meeting planned at ORNL in April. Milestone: decision in May 2001.

(b) Incorporate model improvements from Task 1(a) OR implement new LP solver from Task 2(a) (Carreras) Milestones: improvements incorporated August 31, 2001; testing completed September 30, 2001.

Task 3: Develop test networks

(a) Finish data set for IEEE 118 bus network (Dobson). Milestone: finish by May 31, 2001.

(b) Set up WSCC system for input to ORNL code (Dobson). Wisconsin needs to coordinate with Cornell to get the raw data. Milestone: finish in July 31, 2001.

Task 4: Study models by running software on test networks; develop diagnostics and analyze and interpret results

(a) Scaling studies of the blackout dynamics for different network sizes using current ORNL model (Carreras) Milestone: report by July 31, 2001.

(b) Parameter sensitivity studies on 176 bus WSCC system (Thorp) Milestone: report by November 1, 2001.

(c) Investigate criticality as a function of loading in models without network improvements (Carreras, Dobson, Thorp). Wisconsin/ORNL and Wisconsin/Cornell coordination is needed. Milestone: report by July 31, 2001.

Budget

B. A. Carreras and B. J. Kirby (ORNL)

$60,000 per year for 2 years, beginning Jan 1, 2001

Ian Dobson, (PSERC Wisconsin)

$40,000 per year for 2 years, beginning Jan 1, 2001

Jim Thorp (PSERC Cornell)

$10,000 per year for 2 years, beginning Jan 1, 2001