Search

Demand forecasting is and has been for years a topic of great interest in the electricity sector, being the temperature one of its major drivers. Indeed, one of the challenges when modelling the load is to choose the right weather station, or set of stations, for a given load time series. However, only a few research papers have been devoted to this topic. This paper reviews the most relevant methods that were applied during the Global Energy Forecasting Competition of 2014 (GEFCom2014) and presents a new approach to weather station selection, based on Genetic Algorithms (GA), which allows finding the best set of stations for any demand forecasting model, and outperforms the results of existing methods. Furthermore its performance has also been tested using GEFCom2012 data, providing significant error improvements. Finally, the possibility of combining the weather stations selected by the proposed GA using the BFGS algorithm is briefly tested, providing promising results.

Background: The COVID-19 pandemic has had global effects, cases have been counted in the tens of millions, and there have been over two million deaths throughout the world. Health systems have been stressed in trying to provide a response to the increasing demand for hospital beds during the different waves. This paper analyzes the dynamic response of the hospitals of the Community of Madrid (CoM) during the first wave of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic in the period between 18 March and 31 May 2020. The aim was to model the response of the CoM’s health system in terms of the number of available beds. Methods: A research design based on a case study of the CoM was developed. To model this response, we use two concepts: “bed margin” (available beds minus occupied beds, expressed as a percentage) and “flexibility” (which describes the ability to adapt to the growing demand for beds). The Linear Hinges Model allowed a robust estimation of the key performance indicators for capturing the flexibility of the available beds in hospitals. Three new flexibility indicators were defined: the Average Ramp Rate Until the Peak (ARRUP), the Ramp Duration Until the Peak (RDUP), and the Ramp Growth Until the Peak (RGUP). Results: The public and private hospitals of the CoM were able to increase the number of available beds from 18,692 on 18 March 2020 to 23,623 on 2 April 2020. At the peak of the wave, the number of available beds increased by 160 in 48 h, with an occupancy of 90.3%. Within that fifteen-day period, the number of COVID-19 inpatients increased by 200% in non-intensive care unit (non-ICU) wards and by 155% in intensive care unit (ICU) wards. The estimated ARRUP for non-ICU beds in the CoM hospital network during the first pandemic wave was 305.56 beds/day, the RDUP was 15 days, and the RGUP was 4598 beds. For the ICU beds, the ARRUP was 36.73 beds/day, the RDUP was 20 days, and the RGUP was 735 beds. This paper includes a further analysis of the response estimated for each hospital. Conclusions:This research provides insights not only for academia, but also for hospital management and practitioners. The results show that not all of the hospitals dealt with the sudden increase in bed demand in the same way, nor did they provide the same flexibility in order to increase their bed capabilities. The bed margin and the proposed indicators of flexibility summarize the dynamic response and can be included as part of a hospital’s management dashboard for monitoring its behavior during pandemic waves or other health crises as a complement to other, more steady-state indicators.

In liberalized electricity markets, aggregated stepwise supply and demand curves are at the core of many relevant processes. Efficient and meaningful representations of the offer curves is an essential procedure for agents participating in those markets. However, there is not a formal framework that allows operating with those offer curves using basic arithmetic operations. In this paper we first formalize the concept of stepwise offer curve by explicitly defining the standard True Offer Curve (TOC). To overcome the inherit difficulties of this non-continuous TOC, we propose the Encoded Offer Curve (EOC), a continuous piecewise version that approximates the steps of the TOC with high accuracy. We present fast and simple specialized algorithms to obtain both TOC and EOC models, as well as a formal framework to deal with elementary mathematical operations involving TOCs and EOCs. The proposed framework has been tested in the Italian electricity market, computing the residual demand curves of the producers in a particular Market Zone; and in the Iberian electricity market, quantifying the differences in the bidding behavior of market agents in different stages of the COVID-19 pandemic.

When facing any forecasting problem not only is accuracy on the predictions sought. Also, useful information about the underlying physics of the process and about the relevance of the forecasting variables is very much appreciated. In this paper, it is presented an automatic specification procedure for models that are based on additivity assumptions and piecewise linear regression. This procedure allows the analyst to gain insight about the problem by examining the automatically selected model, thus easily checking the validity of the forecast. Monte Carlo simulations have been run to ensure that the model selection procedure behaves correctly under weakly dependent data. Moreover, comparison over other well-known methodologies has been done to evaluate its accuracy performance, both in simulated data and in the context of short-term natural gas demand forecasting. Empirical results show that the accuracy of the proposed model is competitive against more complex methods such as neural networks.

Residual demand curves (RDCs) can be used to represent the strategic interaction of participants in electricity markets. RDCs relate the energy that an agent can buy or sell in one hour with the clearing market price that would be obtained in such hour, assuming the market is organized as simple bid independent auctions. Despite the fact that they have been widely used in the literature, the existence of time and/or spatial constraints in the market clearing algorithm makes the RDCs not directly applicable. This paper tries to overcome these difficulties by extending the concept of RDCs to zonal pricing markets where complex offering conditions and transmission constraints are taken into account. Therefore, the RDCs are redefined in order to capture such effects, which are usually neglected or oversimplified. A new method for computing the redefined RDCs is established and its application to the Iberian electricity market is presented. The results show that modeling complex conditions and transmission constraints in RDCs can have a significant effect when compared to the standard approach found in the literature. Therefore, the method presented in this paper modeling the effect of firm's decisions on market prices in a more accurate way.

In order to mitigate the excessive computational cost of atrium fire simulations, a novel methodology based on the use of the Fractional Factorial Design technique to obtain an experimental validated tool, in the form of a surface response model, capable to predict fire induced conditions is proposed. This methodology is supported by results from a Design of Experiments benchmark, which consists of a set of FDS simulations in the present work. Specifically, a $$2^{6-2}_{IV}$$ approach has been considered and applied to a 20 m cubic atrium. Thus, six factors have been considered, namely the fire Heat Release Rate (HRR) and location, the exhaust flow rate, the exhaust location and activation time, and the inlet vents area. Furthermore, the smoke temperature at the roof and 15 m high and the smoke layer height have been considered the variables of interest. Subsequently, a multiple linear regression analysis has been performed to predict and compare the steady and non-steady temperature profiles and the smoke layer drop with six novel full-scale atrium fire tests, and also with specific adjusted FDS models. In addition, this methodology has been extended successfully to predict the non-steady behaviour of the fire tests. At the steady state, the HRR and the exhaust flow rate have been found to be the most relevant factors. The results obtained with the proposed methodology show a good fit both with the fire tests and with the adjusted FDS models, with discrepancies mostly below 14%. For non-steady conditions, a time analysis of the influence of the six factors has been carried out. Again, remarkable good agreement with the time-dependent experimental results is achieved, with average discrepancies below 12%, being the larger differences found in the prediction of local effects, such as the smoke ceiling jet, for high HRR or when the make-up air influence is significant. The results turn this methodology into a powerful and useful tool for fire safety designs.

Electricity markets are based on several sequential energy and reserve trading mechanisms to constantly maintain the balance between generation and demand. During the last years, reserve markets are getting much importance all around the world with the increasing social awareness of the renewable energy benefits. Additional reserve quantities and larger remunerations are being implemented by the regulators since this renewable energy is highly dependent on weather conditions uncertainties. Utilities are therefore demanding more and more powerful models to better optimize their reserve bidding curves to be sent to the system operators. This paper describes a new methodology to obtain an optimized real bidding curve for the secondary (spinning) reserve market. The approach is based on the maximization of a company profit, assuming a set of residual demand curves scenarios for its competitors' representation, and considering the day-ahead market opportunity cost to build the reserve cost curve. The case study validates the main features of the proposed methodology by its application to the Spanish secondary reserve market. This methodology is being daily used by one of the most important electricity companies in Spain.

This paper studies an approach to identify representative operating and contingency (OC) scenarios for the design of underfrequency load-shedding (UFLS) schemes. In small isolated power systems, contingency scenarios are outages of generating units. Usually, only N-1 outages are considered. In this paper, simultaneous outages of several units are also taken into account. Data mining techniques such as K-Means and Fuzzy C-Means algorithms are used to group scenarios in terms of system frequency and to identify representative OC scenarios. The approach has been applied to the design of UFLS schemes of two of the Spanish isolated power systems. The results have also been compared to the common practice of scenario selection. Clustering techniques yielded to satisfactory results, i.e., representative OC scenarios can be identified. Furthermore, these representative OC scenarios cover a wider range of possible system responses than the scenarios selected following the common practice.

PurposeThe purpose of this paper is to analyze medium‐term risks faced by electrical generation companies in competitive environments. Market risks faced by generation companies are caused by several variables subject to uncertainty. Hydro conditions, fuel (coal and natural gas) prices, system demand, and CO2 emission price are the risk factors considered in the paper. Taking into account these risk factors, generation companies have to take decisions that would affect their economic results and their risk exposure.Design/methodology/approachThis paper proposes a methodology to support the risk‐analysis decision‐making process. Firstly, different scenarios of risk factors are generated. Then, a market equilibrium model is used in order to assess the impact of the different sources of uncertainty. Finally, decision trees are used in order to analyze the variables subject to interest, such as electricity prices or companies' profits.Research limitations/implicationsThe proposed methodology can be enhanced to take into account scenarios of more risk factors, such as equipment failure or agents' behavior. Another future enhancement could be a detailed study of correlation between different risk factors.FindingsA realistic case study is presented, showing the advantages of these techniques for medium‐term risk‐analysis and decision‐making processes. Several decision trees have been generated to assess the impact of the different risk factors in electricity prices and companies' profits. These decision trees provide valuable information for companies when facing their risk‐management process.Originality/valueThe approach presented here constitutes a valuable support to gain useful information for wise decision making and to hedge against risk.

Forecasting industrial end-use natural gas consumption is an important prerequisite for efficient system operation and a basis for planning decisions. This paper presents a novel prediction model that provides forecasting in a medium-term horizon (1–3 years) with a very high resolution (days) based on a decomposition approach. The forecast is obtained by the combination of three different components: one that captures the trend of the time series, a seasonal component based on the Linear Hinges Model, and a transitory component to estimate daily variations using explanatory variables. The flexibility of the model allows describing demand patterns in a very wide range of historical profiles. Furthermore, the proposed method combines a very simple representation of the forecasting model, which allows the expert to integrate judgmental analysis and adjustment of the statistical forecast, with accuracy and high computational efficiency. Realistic case studies are provided.

In deregulated electrical systems, production schedule for power plants is the result of an auction process. In the Spanish case, this schedule includes two main concepts: energy production (to be actually produced) and secondary reserve (to maintain available). The generation company faces the problem of converting energy schedule into a power schedule, respecting the reserve schedule as well as technical constraints, and trying to accomplish different goals: to minimise the production costs, to obtain smooth shapes for the power schedules and to optimise eventual compensation in schedules. A weighted goal mixed integer programming model with a real-size application to deal with this problem is presented.

With the increasing penetration of intermittent technologies, such as renewable energy sources, electricity secondary (spinning) reserve markets are getting much relevance all around the world to constantly maintain the balance between generation and demand. In this new context, generation utilities are increasingly demanding powerful tools to better optimize their bidding reserve curves, trying to maximize their benefits while satisfying the secondary system necessities published by the System Operators. This paper describes three new integer optimization models to obtain an optimized real bidding secondary reserve curve of an only generation company, assuming a set of residual demand curves for its competitors' representation. The case study shows the main features of the optimization models by their application to the Spanish secondary reserve market.

Flexibility of distributed energy resources is increasingly needed in the electricity markets and grids. But this dispersed flexibility can be used efficiently only if it is aggregated to wholesale market products. A set of software tools was developed for an actor or a company who aggregates flexible household demand for the electricity market and network operators. The tools comprise 1) short term forecasting of electricity market prices, 2) forecasting loads and their responses to control signals, 3) optimal selection of the control signals and consequently also the responses in each situation, 4) relevant external interfaces, and 5) a database and coordinating logic that integrate the tools into toolbox software. In addition to the operational time horizon tools also tools were considered for longer term decision making such as support for development of a portfolio of retail and wholesale market contracts in order to manage risks and maximize the value of flexibility.

In this paper, the three main forecasting topics that are currently getting the most attention in electric power systems are addressed: load, wind power and electricity prices. Each of these time series exhibits its own stylized features and is therefore forecasted in a very different manner. The complete set of forecasting models and techniques included in this revision constitute a guided tour in power systems forecasting.

This paper studies an approach to identify representative operating and contingency scenarios for the design of underfrequency load-shedding (UFLS) schemes. In small isolated power systems, contingency scenarios are outages of generating units. Not only N-1 outages are considered, but simultaneous outages of several units must be considered. Cluster analysis and classification trees are used to group scenarios in terms of system frequency and power-system parameters and to identify representative scenarios. The approach has been applied to the design of the UFLS scheme of one of the Spanish isolated power systems. Clustering techniques yielded to satisfactory results, i.e. representative operating and contingency scenarios can be identified. In addition, classification trees are able to classify new operating and contingency scenarios according to the clusters obtained.

We propose in this paper a model for the description of electricity spot prices, which we use to describe the dynamics of forward curves. The spot price model is based on a long-term/short-term decomposition, where the price is thought of as made up of two factors: A long-term equilibrium level and short-term movements around the equilibrium. We use a non-parametric approach to model the equilibrium level of power prices, and a mean-reverting process with GARCH volatility to describe the dynamics of the short-term component. Then, the model is used to derive the expression of the short-term dynamics of the forward curve implicit in spot prices. The rationale for the approach is that information concerning forward prices is not available in most of power markets, and the direct modeling of the forward curve is a difficult task. Moreover, power derivatives are typically written on forward contracts, and usually based on average prices of forward contracts. Then, it is difficult to obtain analytical expressions for the forward curves. The model of forward prices allows for the valuation of power derivatives, as well as the calculation of the volatilities and correlations required in risk management activities. Finally, the methodology is proven in the context of the Spanish wholesale market

Huge amounts of data are available in many disciplines of Science and Industry. In order to extract useful information from these data, a large number of apparently very different learning approaches have been created during the last decades. Each domain uses its own terminology (often incomprehensible to outsiders), even though all approaches basically attempt to solve the same generic learning tasks. The aim of this paper is to present a new perspective on the main existing automatic learning strategies, by providing a general framework to handle and unify many of the existing supervised learning models. The proposed taxonomy allows highlighting the similarity of some models whose original motivation comes from different fields, like engineering, statistics or mathematics. Common supervised models are classified into two main different groups: structured and unstructured models. Multidimensional models are shown as a composition of onedimensional models, using the latter as elementary building blocks. In order to clarify ideas, examples are provided.

This paper describes a procedure for medium- and long-term risk analysis by using decision trees. A market equilibrium model is presented in order to assess the impact of the different sources of uncertainty. Decision trees are defined and applied in a study case, showing the advantages of these techniques for medium-term operation and planning. The paper analyzes different scenarios of five main risk factors: Hydro-inflows, fuel (coal and gas) costs, system demand, and CO2 emission price.

Due to the large size of electric power systems, there is a very high computational burden when obtaining the optimum network by using classical optimization techniques. Several authors have used heuristics and/or sensitivities in order to guide the search of optimal network investments. This paper proposes an automatic learning approach in order to decide whether a network change will improve the overall costs or not. More specifically, decision trees methods are used to identify a set of simple and reliable rules which combine criteria based on both heuristics and sensitivities. These decision trees are integrated in a subtransmission planning tool, improving dramatically both the "optimality" of the resultant network and the computational time.

This paper proposes a novel methodology to identify congestion problems under both "traditional" and "new" uncertainties such as generation costs, location and size of new generators, retirement of old ones, generation patterns, etc. The methodology allows not only identifying the transmission paths and corridors which will have congestion problems, but also the scenarios producing these critical situations. Thus, it can be used not only to simplify the study of new investments (reinforcement of existing lines), but also to facilitate the evaluation of hedging strategies and the design of proactive policies to avoid the detected congestion.

In this paper, we present and compare two approaches to model supply and demand curves of a sealed-bid auction market. Both the linear hinges model and the sigmo model are able to extract the relevant information from these bidding curves, without losing significant market information. We discuss their main similarities and important differences using a unified framework, to highlight their main strengths and weakness. A practical comparative study based on real supply functions from the Californian electricity market has been included to derive practical conclusions.

This paper describes SGO, a management information system for bidding in deregulated electricity markets, developed for the Spanish case. SGO has a client-server architecture and consists of a set of integrated cooperative and flexible software tools for assisting the users during the whole bidding process: resources identification, bids generation, market performance characterisation, bidding strategy analysis and optimisation, generation of markets results reports and automatic performance monitoring for suggesting on-line corrective actions.