Madeleine Gibescu

Impacts of Wind Power on Thermal Generation Unit Commitment and Dispatch

This paper proposes a new simulation method that can fully assess the impacts of large-scale wind power on system operations from cost, reliability, and environmental perspectives. The method uses a time series of observed and predicted 15-min average wind speeds at foreseen onshore- and offshore-wind farm locations. A Unit Commitment and Economic Dispatch (UC-ED) tool is adapted to allow for frequent revisions of conventional generation unit schedules, using information on current wind energy output and forecasts for the next 36 h. This is deemed the most faithful way of simulating actual operations and short-term planning activities for a system with large wind power penetration. The problem formulation includes ramp-rate constraints for generation schedules and for reserve activation, and minimum up-time and down-time of conventional units. Results are shown for a realistic future scenario of the Dutch power system. It is shown that problems such as insufficient regulating and reserve power-which are typically associated with the variability and limited predictability of wind power-can only be assessed in conjunction with the specifics of the conventional generation system that wind power is integrated into. For the thermal system with a large share of combined heat and power (CHP) investigated here, wind power forecasting does not provide significant benefits for optimal unit commitment and dispatch. Minimum load problems do occur, which result in wasted wind in amounts increasing with the wind power installed

Short-Term Energy Balancing With Increasing Levels of Wind Energy

Increasing levels of wind energy are adding to the uncertainty and variability inherent in electricity grids and are consequently driving changes. Here, some of the possible evolutions in optimal short-term energy balancing to better deal with wind energy uncertainty are investigated. The focus is mainly on managing reserves through changes in scheduling, in particular market structure (more regular and higher resolution scheduling), reserve procurement (dynamic as opposed to static), and improved operational planning (stochastic as opposed to deterministic). Infrastructure changes including flexible plant, increased demand side participation, more interconnection, transmission, larger balancing areas, and critically improved forecasting can also be significant and are dealt with in the discussion. The evolutions are tightly coupled, their impact is system-dependent and so no “best” set is identifiable but experience of system operators will be critical to future developments.

Technical benefits of distributed storage and load management in distribution grids

On the one hand, several developments make the operation of the distribution grids more and more complex. On the other hand developments in energy storage and the availability of loads which are not time critical bring up possibilities to provide more flexibility in the grid and to use the available system more efficiently. In this paper, these and other advantages of as well distributed energy storage as load management of not time critical loads are discussed. An approach is described to analyse the available capacity in the grids, to investigate how this capacity can be used by applying storage or load management and to compare the benefits of storage and load management. This approach is illustrated using a real medium voltage network and measured data. It shows the part of the capacity in the distribution grids which is now unused, but can be made available by applying storage or load management of non-critical loads. The size of the storage and the non-critical loads which are needed to use this capacity are determined. The characteristics of future loads, the need to support integration of distributed generation and the desired level of reliability of supply are factors that determine the size of the storage and how storage or load management can be applied best.

Statistical Wind Speed Interpolation for Simulating Aggregated Wind Energy Production under System Studies

In this paper we present a statistical interpolation method to generate a time series of system-aggregated wind power production values that can be used as an input to system operations planning tools such as unit commitment (UC) and economic dispatch (ED). We use historical wind speed data measured at several locations, in order to estimate average wind patterns and express the covariance between locations as a function of their distance. Then, for a new set of locations where wind parks are planned, we create wind time series for the study period such that the spatial correlation between the sites is taken into account. Depending on the system under study, this may be of specific importance due to the concentration of areas with favorable wind conditions, resulting in strong correlations between wind park outputs. These cross-correlations are essential when evaluating system adequacy and security in planning mode, in the presence of large-scale wind power. The resulting aggregated wind power time series are finally fed into the UC-ED module to help evaluate the amount of total system reserve required to maintain an adequate level of reliability. The method is applied to a simplified version of the Dutch power system

Impact of wind power on the unit commitment, operating reserves, and market design

This article highlights and demonstrates the new requirements variable and partly unpredictable wind power will bring to unit commitment and power system operations. Current practice is described and contrasted against the new requirements. Literature specifically addressing questions about wind power and unit commitment related power system operations is surveyed. The scope includes forecast errors, operating reserves, intra-day markets, and sharing reserves across interconnections. The discussion covers the critical issues arising from the research.

Dynamic models for transient stability analysis of transmission and distribution systems with distributed generation: An overview

Distributed Generation is increasing in nowadays power systems. Small scale systems such as photovoltaic, biomass or small cogeneration plants are connected to the distribution level, while large wind farms will be connected to the transmission level. Both trends lead to a replacement of large synchronous generators as the dominating generation technology. Up to now, transient stability of transmission systems has been analysed to a satisfactory degree of accuracy with a simplified representation of the distribution systems. In future, distributed generation will more and more influence the behaviour of the system. Stiff, inverter-based local generation technologies may improve the system stability; however, increasing electrical distances between large synchronous generators in operation will impede the system stability. These (and other) diverging effects have to be studied in detail. This overview paper summarises the latest findings and reveals future research questions. It is concluded that the accuracy and validity of the currently applied dynamic models for transient stability analysis of power systems with high penetration of DG should be further investigated.

System integration of large-scale wind power in the Netherlands

This paper presents the results of a simulation of system operation in the Netherlands in the presence of future large-scale wind energy production. The study is aimed at identifying bottlenecks in system planning and operation due to wind integration, in particular base-load and ramp rate problems. These may constraint the amount of wind that can be accommodated given a projected production park of dispatchable units and yearly load profile by 2012. Wind data from 2004-2005, interpolated to existing locations for onshore and planned locations for offshore wind parks, were used to create a realistic yearly wind energy output profile. The unit commitment and economic dispatch formulation includes ramp rate constraints for generation schedules and reserve activation as well as minimum up- and down times. Of particular interest in this study are the combined heat & power (CHP) units, which impose additional constraints coupling their heat and energy production. Since no insight was available into the aggregated predictability of wind generation, both a 0-MW prediction, where conventional units are scheduled to meet the total load, and a perfect prediction have been investigated. No forms of electrical or heat storage were considered. The results show no ramp rate problems in the Dutch system by 2012, however base-load problems may arise at high wind penetration levels, only to be prevented by wasting available wind resources

A Market-Based Transmission Planning for HVDC Grid—Case Study of the North Sea

Summary form only given. There is significant interest in building HVDC transmission to carry out transnational power exchange and deliver cheaper electricity from renewable energy sources which are located far from the load centers. This paper presents a market based approach to solve a long-term TEP for meshed VSC-HVDC grids that connect regional markets. This is in general a nonlinear, non-convex large-scale optimization problem with high computational burden, partly due to the many combinations of wind and load that become possible. We developed a two-step iterative algorithm that first selects a subset of operating hours using a clustering technique, and then seeks to maximize the social welfare of all regions and minimize the investment capital of transmission infrastructure subject to technical and economic constraints. The outcome of the optimization is an optimal grid design with a topology and transmission capacities that results in congestion revenue paying off investment by the end the projects economic lifetime. Approximations are made to allow an analytical solution to the problem and demonstrate that an HVDC pricing mechanism can be consistent with an AC counterpart. The model is used to investigate development of the offshore grid in the North Sea. Simulation results are interpreted in economic terms and show the effectiveness of our proposed two-step approach.

Advanced Hybrid Transient Stability and EMT Simulation for VSC-HVDC Systems

This paper deals with advanced hybrid transient stability and electromagnetic-transient (EMT) simulation of combined ac/dc power systems containing large amounts of renewable energy sources interfaced through voltage-source converter-high-voltage direct current (VSC-HVDC). The concerning transient stability studies require the dynamic phenomena of interest to be included with adequate detail and reasonable simulation speed. Hybrid simulation offers this functionality, and this contribution focuses on its application to (multiterminal) VSC-HVDC systems. Existing numerical interfacing methods have been evaluated and improved for averaged VSC modeling. These innovations include: 1) ac system equivalent impedance refactorization after faults; 2) amended interaction protocols for improved Thévenin equivalent source updating inside the EMT-type simulation; and 3) a special new interaction protocol for improved phasor determination during faults. The improvements introduced in this contribution lead to more accurate ac/VSC-HVDC transient stability assessment compared to conventional interfacing techniques.

A topological insight into restricted Boltzmann machines

Restricted Boltzmann Machines (RBMs) and models derived from them have been successfully used as basic building blocks in deep artificial neural networks for automatic features extraction, unsupervised weights initialization, but also as density estimators. Thus, their generative and discriminative capabilities, but also their computational time are instrumental to a wide range of applications. Our main contribution is to look at RBMs from a topological perspective, bringing insights from network science. Firstly, here we show that RBMs and Gaussian RBMs (GRBMs) are bipartite graphs which naturally have a small-world topology. Secondly, we demonstrate both on synthetic and real-world datasets that by constraining RBMs and GRBMs to a scale-free topology (while still considering local neighborhoods and data distribution), we reduce the number of weights that need to be computed by a few orders of magnitude, at virtually no loss in generative performance. Thirdly, we show that, for a fixed number of weights, our proposed sparse models (which by design have a higher number of hidden neurons) achieve better generative capabilities than standard fully connected RBMs and GRBMs (which by design have a smaller number of hidden neurons), at no additional computational costs.