SIMULATIONS FOR COUPLED SYSTEMS

As we progress toward GridWise transformation, the coupling of engineered systems and markets will impact broader and broader areas of the electric power industry. Energy trading products will cover shorter time periods and demand response programs will move toward real-time pricing. Financial market-based activity will impact ever more directly the physical operation of the system. The boundaries of these coupled systems will extend beyond the traditional perimeter of our energy systems. Progress is being made at PNNL to address the gaps in our simulation capabilities so that we may better understand the issues posed by questions of market and physical systems interactions.

A simulation of the transformed energy systems that models how the technologies and markets would function is required in order to establish the potential benefits of GridWise. It is also needed to determine appropriate system control and operational strategies, direct technology development in the most beneficial directions, study the impacts of open, evolving energy markets and develop public policy and regulation in ways that equitably enhance the benefits of the transformation. Hence, a key part of this simulation is that it incorporates dynamic, evolving markets, strategies and financial instruments.

Progress is being made at PNNL to address the gaps in our simulation capabilities so that we may better understand the issues posed by questions of market and physical systems interactions.

The complexity of the market and physical operation of our electric power system is reflected in the complexity of the computational framework needed for its simulation. This simulation environment must integrate a great number of components from different sources using different technologies and multiple computing platforms. This sophisticated integration illustrates the challenges involved in achieving an effective simulation framework.

Simulating the Combined Market and Physical Operations of our Electric Power System

The vision for the simulation framework is to create an open source environment where independently developed software components can be shared by other people and organizations, and a variety of simulation environments can be configured to address analysis needs. This unique set of simulation tools will span energy systems currently analyzed in isolation—the transmission grid, distribution systems and customer systems (equipment and appliances)—and link their physical performance and control with the economic markets that will drive them in the future. Today's tools evaluate specific aspects of the system, and since they are not interconnected, it is very difficult to model the interactions of behaviors that are exogenous to a given tool. Yet, this very capability is required to understand the effect of GridWise solutions on our energy systems.

The simulation tools must meet several technical challenges:

Spanning levels of physical scale. We are constructing the simulation environment so that it can operate at two levels: the distribution-to-customer level and the generation/ transmission-to-distribution bulk power level. At the bulk power level, only a few substation-level distribution systems will be simulated in detail; other substation loads will be modeled as "smart boundary condition elements" that capture their essential response to cost and control signals.
Simulating engineering and market behaviors that occur on time scales that cross several orders of magnitude. These orders of magnitude range from sub-second intervals for dynamic stability of the electric grid to years for planning capacity expansion for fuel and electric systems. The second challenge involves analyzing the relatively smooth dynamics of the interactions between engineering and market behaviors. Load profiles and generation dynamics are injected into a power flow solution engine that provides quiescent, steady-state results in incremental steps as the system marches forward in time. Higher order and complex effects associated with detailed protection schemes, equipment behavior, and control models associated with transient or dynamic stability will be added as the modeling work progresses.

Industry-standard grid tools are used as a core of the engineering solution (the MATLAB toolkit for the distribution level, and AREVA Dispatcher Training Simulator at the bulk level). This builds technical credibility and reduces duplication of effort, allowing us to focus on loads and markets as key new ingredients. In the initial component-oriented simulation environment under construction at PNNL, these components are integrated together using an open source J2EE framework, which supports several services enabling the components to exchange simulation data and control.

At the distribution level we have added populations of individual customer loads, with end-use appliance and equipment cycles of operation simulated with unprecedented resolution, including curtailment options. Non-cycling loads match arbitrary utility totals for end-use load control; these totals represent probability distributions. Among other refinements to this model, we are adding distributed generation. Reduced-order models of the distribution system are being developed to study transmission-level impacts.

We have adopted an agent-based computational economics (ACE) modeling strategy as a way of incorporating market mechanisms that allow the system to evolve over time in response to market forces. The ACE modeling approach blends concepts from experimental economics and evolutionary economics, and both tools and concepts from artificial life studies. The agents in an ACE model interact with other agents and their environment on the basis of internalized data and behavioral rules. These agents are different from each other; they are usually modeled to have considerably more autonomy than conventionally modeled economic agents and a great deal more internal cognitive structure. The agents can act cooperatively or in a predatory manner; they can collude to exploit advantages that together they have but individually would not have, and they can evolve over time.

These models are implemented as virtual economic worlds that follow a time line; the virtual worlds are implemented on computers much as a culture grows in a petri dish. In principle, once the initial conditions of the agents are specified, this computer world evolves without intervention on the part of the modeler.

Project Lead: Steve Widergren

Related GridWise Sites

SIMULATIONS FOR COUPLED SYSTEMS

Simulating the Combined Market and Physical Operations of our Electric Power System