Search

The training of power grid operators took a major leap forward with the introduction of operator training simulators (OTSs) in the late 1970s, but the demands on OTS-based training have increased with the ongoing transformation of the grid. Today, blackouts occur as a result of outages that typically impact multiple transmission operators, multiple balancing authorities (BAs), and multiple reliability coordinators (RCs). Therefore, this article focuses on training that involves teams of system operators who work cooperatively in neighboring control centers. With more generation resources being placed in the distribution systems, these coordinating operators to be trained are in the distribution control centers. Blackouts used to be associated mainly with instability, uncontrolled separation, and cascading outages. This situation is changing, with blackouts now also occurring due to systemic emergencies, such as firestorms, extreme weather, hurricanes, and cyberattacks.

It is of growing concern to ensure the resilience in electricity infrastructure systems to extreme weather events with the help of appropriate hardening measures and new operational procedures. An effective mitigation strategy requires a quantitative metric for resilience that can not only model the impacts of the unseen catastrophic events for complex electric power distribution networks, but also evaluate the potential improvements offered by different planning measures. In this article, we propose probabilistic metrics to quantify the operational resilience of the electric power distribution systems to high-impact low-probability (HILP) events. Specifically, we define two risk-based measures: value-at-risk ($VaR_\alpha$) and conditional value-at-risk ($CVaR_\alpha$) that measure resilience as the maximum loss of energy and conditional expectation of a loss of energy, respectively, for the events beyond a prespecified risk threshold, $\alpha$. Next, we present a simulation-based framework to evaluate the proposed resilience metrics for different weather scenarios with the help of modified IEEE 37-bus and IEEE 123-bus system. The simulation approach is also extended to evaluate the impacts of different planning measures on the proposed resilience metrics.

The present communication architecture supporting control of the electric power grid makes it difficult to use the wealth of data collected at high rates in substations, retarding their use in new applications for controlling the grid. A flexible, real-time data network would make it possible to use these data for many more control and protection applications, potentially increasing the grid's reliability and increasing its operating efficiency. Applications that could use these data include: decentralised load frequency control; closed-loop voltage control; transient and small-signal stabilisation; and special protection schemes using data gathered over a wide area. Such applications and the flexibility of the underlying communication network imply greater data sharing between utilities, leading to new performance, availability and reliability requirements. This paper examines the security, trust and Quality of Service (QoS) requirements imposed by these applications and shows how they are met by mechanisms included in the GridStat middleware framework that we are developing.

•Propose a new retail electricity market for the distribution grid.•Evaluate the proposed market on test network with real demand and price data.•Compare market performance against existing retail practices.•Discuss regulatory and policy impact on resource participation in market.

The dynamic behaviour of a large interconnected electric power system is characterized by a simultaneous set of nonlinear algebraic and ordinary differential equations. The solution is obtained by numerical methods and the simulation of the transient behaviour for a few seconds after a fault is the standard analytical procedure used in planning and operational studies of the system. The need for on-line simulation in near real time for more efficient operation has encouraged the search for faster solution methods and the use of parallel computers for this purpose has attracted the attention of many researchers. The success of parallelization depends on three factors: the problem structure, the computer architecture, and the algorithm that takes maximum advantage of both. In this problem, the generator equations are only coupled through the electrical network providing some parallelization in (variable) space, and a solution is needed at each time step leading to some parallelization in time (waveform relaxation). However, since the problem formulation is not completely decoupled, parallel algorithms can only be developed by trading off any relaxation with a degradation in convergence. The fastest sequential algorithm used today is the combination of implicit trapezoidal integration with a dishonest Newton solution. The Newton algorithm is not parallel at all but has the fastest convergence while a Gauss-Jacobi algorithm is completely parallel but converges very slowly. A relaxation of the Newton algorithm appears to be a good compromise. As for the parallel hardware, the coupling seems to require significant communication between processors thus favouring a data-sharing architecture over a message-passing hypercube. Special architectures to match the problem structure have also been an area of investigation. This paper elaborates on the above issues and assesses the present state-of-the-art.

Although on-line static security analysis (SSA) of power systems is a standard function in most control centers dynamic security analysis (DSA) is still in its experimental stages. IT SSA detects post-contingency steady state violations, the operator has the option of taking preventive control right away or decide to take corrective control if and when the contingency actually happens; since the former course is conservative and expensive, the latter is more common. IT DSA detects instability for a contingency, the latter course of corrective action is usually not an option because the time required to take such action is not available; thus preventive control is the only course available. The methods to determine such preventive control, however, need to be developed and very little research has been conducted in this area. This paper compares the two approaches that have been suggested so far: an energy margin based approach that determins control action from the sensitivities of the stability energy margin, and a coherency based approach that determins control action by maximizing coherency of the generator. Given that these methods are far from fully developed, the paper also discusses the research and development needed to determine the practical feasibility of preventive control for dynamic security of power systems.

The power system simulation in an operator training simulator (OTS) is made up of a periodic solution of the steady state power network synchronized with a solution of the slow dynamic behavior due to the power balance of the system. The periodicity of the static solution (the OTS cycle time) has to coincide with the data scan rate of a control center so that changes in the variables of the simulated system accurately mimics the actual measurement data acquisition system. The design of such a computationally demanding application requires many considerations almost all of which distill down to a compromise between accuracy and computation. A companion paper describes the models developed to obtain the required accuracy for such a simulation. This paper describes the algorithms developed to meet the computation times required for real time simulation. In a control center, the data acquisition system is usually designed to scan different classes of data at different scan rates. Although every system is different in detail, the AGC data (generator powers and tie-line flows) scan is the most frequent, typically around 4 seconds. This implies that the OTS cycle should be at least as fast as the AGC scan cycle. To achieve this on the size of computers normally used in control centers (it can obviously be done on faster processors) has been difficult for system sizes of say, 1000 buses. The main hurdle has been the simulation just following a large disturbance.

It is now commonly accepted that for security assessment of a power system, the most efficient and practical strategy is to deal with the problem in two stages. First in contingency selection, those potentially critical contingency cases are ranked by the severity of their impacts on a system. Then in contingency analysis, detailed AC power flows are applied only to the most dangerous cases appearing on top of the ranked list. In the past decade, the problem of ranking outage cases in the order of decreasing severity has attracted a lot of intensive studies, and many efficient and reliable algorithms were developed. But most of these techniques can only be applied to MW limit security problems. On the other hand, voltage problems were also found to be a very important aspect of security assessment. It has become the target of many research projects. Because the MVAR-voltage problem involves a much more complicated model than the MW-angle problem, significant obstacles were experienced in developing ranking algorithms for voltage security monitoring, making it an area where further study is still required. Almost all contingency ranking algorithms employ a scalar performance index to represent a global severity of voltage disturbance after a contingency event. The scalar performance index can be viewed as a weighted distance in voltage space measuring the post contingency voltage profile against specified voltage limits.

A major difficulty in on-line security analysis of power systems is the representation of the external but interconnected network from which very little or no real time data are available. One method used for the determination of the external system state is to extrapolate or assume real time data for a large external network representation, which is then solved by a power flow in such a way that the mismatches at the boundary between this external system solution and the state estimated internal network is minimized. This paper presents a method that extends the state estimation techniques, instead of using the power flow for the external network solution. The results indicate improved solution flexibility, sensitivity, error detection and location abilities, and significant reduction in boundary errors.

Models for power plants have been developed over several decades for use in long term dynamic simulation of large power systems. These models are derived from, but are significantly simpler than, models used for studying power plants. This is because in system studies, as opposed to power plant studies, only the trajectories of the output variables of many plants are of interest as opposed to the many variables internal to the plant. The first use of these models was in the simulation and study of automatic generation control. This was done by interconnecting the power plant models through a uniform frequency dynamic model of the power system. Later, the addition of the network model and models of other system components provided the capability to simulate all aspects of power system long term dynamics. The power system simulation required in an operator training simulator is very similar to that needed for long term dynamic studies with the added requirement that it must be run in real time. These available power plant models have been used in operator training simulators and the suitability for such application of these models has also been shown. The fast solution requirement of the operator training simulator, however, has often led to the use of simple low order models whose accuracy, though suspect, has never been reported. In this paper, an attempt is made to provide models of varying complexity.