The main goal of the E-Cloud, as with every microgrid, is to maximize the consumption of energy produced locally. To reach this goal, based on consumption profiles of customers willing to participate in the E-cloud and given some local restrictions (e.g. wind turbines cannot be put everywhere), an optimal mix of green generation sources (in kW) and local storage (in kWh) needs to be computed.
Then according to this computation, the required generating units and storage device are installed. A repartition mechanism grants the customer a share of the generated electricity and storage capacity. These shares are either computed offline, or dynamically adapted on line. The project will test two models: either the DSO or a producer owns and operates the storage device.
Two flows of information (real-time for operation of the storage facility and ex-post for its settlement) are needed to correctly manage the E-Cloud and to ensure correct information exchange with the wholesale market. These information flows are completed thanks to a forecast that provides members of the E-Cloud the full capability to anticipate and obtain the maximum benefits of the local generation.
The expected benefits for the customer are a reduction of their electricity bill by a minimum of 10%. Societal benefits should also arise: 1) easing the technical integration of renewables’ generation embedded in the distribution network, and 2) avoids extra investment on the DSO network. The E-Cloud may also ensure new revenue for the DSO thanks to new services provided to the E-Cloud community.
This paper addresses the problem of an aggregator controlling residential heat pumps to offer a direct control flexibility service. The service consists of a power modulation, upward or downward, that is activated at a given time period over a fixed number of periods. The service modulation is relative to an optimized baseline that minimizes the energy costs. The load modulation is directly followed by a constrained rebound effect, consisting of a delay time with no deviations from the baseline consumption and a payback time to return to the baseline state. The potential amount of modulation and the constrained rebound effect are computed by solving mixed integer linear problems. Within these problems, the thermal behavior of the building is modeled by an equivalent thermal network made of resistances and lumped capacitances. Simulations are performed for different sets of buildings typical of the Belgian residential building stock and are presented in terms of achievable modulation amplitude, deviations from the baseline and associated costs. A cluster of one hundred ideal buildings, corresponding to retrofitted freestanding houses, is then chosen to investigate the influence of each parameter defined within the service. Results show that with a set of one hundred heat pumps, a load aggregator could expect to harvest mean modulation amplitudes of up to 138 kW for an upward modulation and up to 51 kW for a downward modulation. The obtained values strongly depend on the proposed flexibility service. For example, they can decrease down to 2.6 kW and 0.4 kW, respectively, if no rebound effect is allowed.
The major challenges in power systems are driven by the energy shortage and environmental concerns, namely facilitating the penetration of renewable energy and improving the efficiency of the renewable powers. Due to the variable nature of renewables, the generated power profile may not be able to match the load requirement. Accordingly, much attention has been focused on the development of energy storage technologies to guarantee renewable power penetrations. Recently, advances in the supercapacitor (SC) have made the SC and battery hybrid energy storage systems (HESS) technically attractive. Compared with other energy storage technologies the principal advantages of SC are: the high power density, high cycling life, and high peak current handling capacities. However, SC is also deficient in low energy density. The battery is characterised by large energy density but low in power capacity. In the microgrid systems, high-frequency power fluctuations will cause a significant degree of battery power cycling. This, in turn, has been shown to lead to a significant reduction in battery service life. Therefore, the concept of the SC and battery hybrid scheme is proposed. A case study of the HESS based on a microgrid is introduced in this paper. A simplified microgrid system is established to test the performance of the proposed design.
There is a need to clearly state an interaction model that formalizes interactions between actors of the distribution system exchanging flexibility. In previous works we quantitatively evaluated the performance of five interaction models devised with industrial partners using the agent-based testbed DSIMA. Simulation results showed that these interaction models relying on active network management suffer from a lack of coordination between the distribution and the transmission system operator, activating flexibility simultaneously in opposite directions. This paper introduces a new interaction model fixing this issue based on dynamic access bounds to the network changing throughout the day and preventing the activation of flexibility leading to congestions. This new interaction model is implemented in DSIMA and compared to a model restricting the grid users to a very restrictive but safe access range. Results show that this new model allows to safely increase by 55% the amount of distributed generation in the network.
With the increasing share of renewable and distributed generation in electrical distribution systems, active network management (ANM) becomes a valuable option for a distribution system operator to operate his system in a secure and cost-effective way without relying solely on network reinforcement. ANM strategies are short-term policies that control the power injected by generators and/or taken off by loads in order to avoid congestion or voltage issues. While simple ANM strategies consist in curtailing temporary excess generation, more advanced strategies rather attempt to move the consumption of loads to anticipated periods of high renewable generation. However, such advanced strategies imply that the system operator has to solve large-scale optimal sequential decision-making problems under uncertainty. The problems are sequential for several reasons. For example, decisions taken at a given moment constrain the future decisions that can be taken, and decisions should be communicated to the actors of the system sufficiently in advance to grant them enough time for implementation. Uncertainty must be explicitly accounted for because neither demand nor generation can be accurately forecasted. We first formulate the ANM problem, which in addition to be sequential and uncertain, has a nonlinear nature stemming from the power flow equations and a discrete nature arising from the activation of power modulation signals. This ANM problem is then cast as a stochastic mixed-integer nonlinear program, as well as second-order cone and linear counterparts, for which we provide quantitative results using state of the art solvers and perform a sensitivity analysis over the size of the system, the amount of available flexibility, and the number of scenarios considered in the deterministic equivalent of the stochastic program. To foster further research on this problem, we make available at http://www.montefiore.ulg.ac.be/~anm/ three test beds based on distribution networks of 5, 33, and 77 buses. These test beds contain a simulator of the distribution system, with stochastic models for the generation and consumption devices, and callbacks to implement and test various ANM strategies.
Probabilistic methods are emerging for operating electrical networks, driven by the integration of renewable generation. We present an algorithm that models a stochastic process as a Markov process using a multivariate Gaussian Mixture Model, as well as a model selection technique to search for the adequate Markov order and number of components. The main motivation is to sample future trajectories of these processes from their last available observations (i.e. measurements). An accurate model that can generate these synthetic trajectories is critical for applications such as security analysis or decision making based on lookahead models. The proposed approach is evaluated in a lookahead security analysis framework, i.e. by estimating the probability of future system states to respect operational constraints. The evaluation is performed using a 33-bus distribution test system, for power consumption and wind speed processes. Empirical results show that the GMM approach slightly outperforms an ARMA approach.
This paper addresses the problem of an aggregator controlling residential heat pumps to offer a direct control flexibility service. The service is defined by a 15 minute power modulation, upward or downward, followed by a payback of one hour and 15 minutes. The service modulation is relative to an optimized baseline that minimizes the energy costs. The potential amount of modulable power and the payback effect are computed by solving mixed integer linear problems.
Within these problems, the building thermal behavior is modeled by an equivalent thermal network made of resistances and lumped capacitances whose parameters are identified from validated models. Simulations are performed on 100 freestanding houses. For an average 4.3 kW heat pump, results show a potential of 1.2 kW upward modulation with a payback of 600 Wh and 150 Wh of overconsumption. A downward modulation of 500 W per house can be achieved with a payback of 420 Wh and 120 Wh of overconsumption.
This paper extends the Global Capacity ANnouncement
procedure proposed in  along two directions. First,
two new stopping criteria are considered. Second, annual
losses are evaluated using representative days to
approximate the injection duration curve. The extensions
are validated on an updated model of a real-life system.
The emphasis is on the situation in the Walloon region of
Belgium considered in the GREDOR project . A way
for a DSO to publish Global Capacity ANnouncement
computation results is shortly discussed.
This paper addresses the problem of an aggregator controlling residential heat pumps to offer a direct control ﬂexibility service. The service is deﬁned by a 15 minute power modulation, upward or downward, followed by a payback of one hour and 15 minutes. The service modulation is relative to an optimized baseline that minimizes the energy costs. The potential amount of modulable power and the payback effect are computed by solving mixed integer linear problems. Within these problems, the building thermal behavior is modeled by an equivalent thermal network made of resistances and lumped capacitances whose parameters are identiﬁed from validated models. Simulations are performed on 100 freestanding houses. For an average 4.3 kW heat pump, results show a potential of 1.2 kW upward modulation with a payback of 600 Wh and 150 Wh of overconsumption. A downward modulation of 500 W per house can be achieved with a payback of 420 Wh and 120 Wh of overconsumption.
This article proposes an open-source testbed to simulate interaction models governing the exchange of flexibility services located within a distribution network. The testbed is an agent-based system in which the distribution system operator, the transmission system operator, producers and retailers make their decisions based on mixed-integer linear programs. This testbed helps to highlight the characteristics of an interaction model, the benefits for the agents and eases the detection of unwanted or abusive behaviors which decreases the welfare. The testbed is implemented in Python and the optimization problems are encoded in the modeling language ZIMPL. A web interface is coupled with an instance generator using macroscopic parameters of the system such as the total power production. This testbed is, therefore, well suited to test the implemented interaction models on various scenarios and to extend the implementation to other models. Five interaction models developed with industrial partners are simulated over a year on a 75-bus test system. Simulations show that interaction models relying on active network management, as they have been developed, lead to substantial welfare even though they suffer from a lack of coordination between the DSO and the TSO. A conservative interaction model restricting grid users to an access range that is computed ahead of time to prevent any congestion, avoids shedding distributed generation but considerably restrains the amount of distributed production.
We propose a pragmatic procedure to facilitate the connection process of Distributed Generation (DG) with reference to the European regulatory framework where Distribution System Operators (DSOs) are, except in specific cases, not allowed to own their generation. The procedure is termed Global Capacity ANnouncement (GCAN) and is intended to compute the estimates of maximum generation connection amount at appropriate substations in a distribution system, to help generation connection decisions. The pragmatism of the proposed procedure stems from its reliance on the tools that are routinely used in distribution systems planning and operation, and their use such that the possibilities of network sterilization are avoided. The tools involved include: long-term load forecasting, long-term planning of network extension/reinforcement, network reconfiguration, and power flow. Network sterilizing substations are identified through repeated power flow computations. The proposed procedure is supported by results using an artificially created 5-bus test system, the IEEE 33-bus test system, and a part of real-life distribution system of ORES (a Belgian DSO serving a large portion of the Walloon region in Belgium).
This article deals with the problem of automatically es- tablishing a correspondence between two databases popu- lated independently over the years by a distribution com- pany, for instance a SCADA system and a geographical information system. This problem is abstracted as a graph matching problem, well known in the combinatorial op- timisation community. It is then casted as an integer quadratic program. An idea of achievable results on a real system is provided, and needs for approximation or decom- position algorithms are discussed.
This paper presents a general process set in the GREDOR (French acronym for “Gestion des Réseaux Electriques de Distribution Ouverts aux Renouvelables”) project to address the challenges in distribution systems posed by the integration of renewable generation, changing load patterns, and the changes in the electricity market sector. A use case describing interactions among different players that fits the process is also presented. A pseudo-dynamic approach to Global Capacity Announcement as a way to increase penetration of Renewable Energy Sources in a distribution system is elaborated in more details.
To ease the transition towards the future of distribution grid management, regulators must revise the current interaction model, that is, the set of rules guiding the interactions between all the parties of the system. Five interaction models are proposed, three of them considering active network management. This paper evaluates the economic efficiency of each model using macroscopic representation of the system, by opposition to more techniques requiring a complete picture of the system. The interaction models are simulated on the horizon 2015-2030. Results show that for the first five years all the models provide similar economic efficiency. For the remaining ten years, interaction models implementing active network management provide up to a 10% higher economic efficiency.
We consider a class of optimal power flow (OPF) applications where some loads offer a modulation service in exchange for an activation fee. These applications can be modeled as multi-period formulations of the OPF with discrete variables that define mixed-integer non-convex mathematical programs. We focus on the optimization of this Mixed-Integer Non-Linear Programming (MINLP) problem through a separation into a non-linear programming (NLP) and a mixed-integer programming (MIP) component. The NLP is a feasibility problem involving the power flow equations and the flexible loads needs. The MIP is used to choose which flexible loads to activate and in which order. In several papers, the MIP is based on a linearization of the non-linear power flow equations. We compare several variants of the linearization. We propose new formulations based on prior knowledge of the network to improve the decision-making process when the relaxation is inappropriate. We show computationally with many real-world instances that they help find feasible solutions faster than standard MINLP techniques.
We consider a class of optimal power flow (OPF) applications where some loads offer a modulation service in exchange for an activation fee. These applications can be modeled as multi-period formulations of the OPF with discrete variables that define mixed-integer non-convex mathematical programs. We propose two types of relaxations to tackle these problems. One is based on a Lagrangian relaxation and the other is based on a network flow relaxation. Both relaxations are tested on several benchmarks and, although they provide a comparable dual bound, it appears that the constraints in the solutions derived from the network flow relaxation are significantly less violated.
We propose and analyze a day-ahead reserve market model that handles bids from flexible loads.
This pool market model takes into account the fact that a load modulation in one direction must usually be compensated later by a modulation of the same magnitude in the opposite direction.
Our analysis takes into account the gaming possibilities of producers and retailers, controlling load flexibility, in the day-ahead energy and reserve markets, and in imbalance settlement.
This analysis is carried out by an agent-based approach where, for every round, each actor uses linear programs to maximize its profit according to forecasts of the prices.
The procurement of a reserve is assumed to be determined, for each period, as a fixed percentage of the total consumption cleared in the energy market for the same period.
The results show that the provision of reserves by flexible loads has a negligible impact on the energy market prices but markedly decreases the cost of reserve procurement.
However, as the rate of flexible loads increases, the system operator has to rely more and more on non-contracted reserves, which may cancel out the benefits made in the procurement of reserves.
Afin d’opérer un réseau de distribution d’électricité de manière fiable et efficace, c’est-à-dire de respecter les contraintes physiques tout en évitant des coûts de renforcement prohibitifs, il devient nécessaire de recourir à des stratégies de gestion active du réseau. Ces stratégies, rendues nécessaires notamment par l’essor de la production distribuée, reposent sur des politiques de contrôle à court-terme du niveau de puissance des dispositifs producteurs ou consommateurs d’électricité. Alors qu’une solution simple consisterait à moduler à la baisse la production des générateurs, il paraît néan- moins plus intéressant de déplacer la consommation aux moments adéquats afin d’exploiter au mieux les sources d’énergie renouvelables sur lesquelles reposent généralement ces générateurs. Un tel moyen de contrôle introduit néanmoins un couplage temporel au problème, menant à un problème d’optimisation non-linéaire, séquentiel sous incertitude et à variables mixtes. Afin de favoriser la recherche dans ce domaine très complexe, nous proposons une formalisation générique du problème de ges- tion active d’un réseau de distribution moyenne tension (MT). Plus spécifiquement, cette formalisa- tion se présente sous la forme d’un processus de décision markovien. Dans cette article, nous pré- sentons également une spécification de ce modèle décisionnel à un réseau de 75 noeuds et pour un ensemble de services de modulation donnés. L’instance de test qui en résulte est disponible à l’adresse http://www.montefiore.ulg.ac.be/~anm/ et a pour objectif de mesurer et de comparer les performances des techniques de résolution qui seront développées.
This paper addresses the problem faced by a distribution system operator (DSO) when planning the operation of a network in the short-term. The problem is formulated in the context of high penetration of renewable energy sources (RES) and distributed generation (DG), and when flexible demand is available. The problem is expressed as a sequential decision-making problem under uncertainty, where, in the first stage, the DSO has to decide whether or not to reserve the availability of flexible demand, and, in the subsequent stages, can curtail the generation and modulate the available flexible loads. We analyze the relevance of this formulation on a small test system, discuss the assumptions made, compare our approach to related work, and indicate further research directions.
Our work is driven by a class of practical problems of sequential decision making in the context of electric power generation under uncertainties. These problems are usually treated as receding horizon deterministic optimization problems, and/or as scenario-based stochastic programs. Stochastic programming allows to compute a first stage decision that is hedged against the possible futures and -- if a possibility of recourse exists -- this decision can then be particularized to possible future scenarios thanks to the information gathered until the recourse opportunity.
Although many decomposition techniques exist, stochastic programming is currently not tractable in the context of day-ahead electric power generation and furthermore does not provide an explicit recourse strategy. The latter observation also makes this approach cumbersome when one wants to evaluate its value on independent scenarios.
We propose a supervised learning methodology to learn an explicit recourse strategy for a given generation schedule, from optimal adjustments of the system under simulated perturbed conditions. This methodology may thus be complementary to a stochastic programming based approach. With respect to a receding horizon optimization, it has the advantages of transferring the heavy computation offline, while providing the ability to quickly infer decisions during online exploitation of the generation system. Furthermore the learned strategy can be validated offline on an independent set of scenarios.
On a realistic instance of the intra-day electricity generation rescheduling problem, we explain how to generate disturbance scenarios, how to compute adjusted schedules, how to formulate the supervised learning problem to obtain a recourse strategy, how to restore feasibility of the predicted adjustments and how to evaluate the recourse strategy on independent scenarios. We analyze different settings, namely either to predict the detailed adjustment of all the generation units, or to predict more qualitative variables that allow to speed up the adjustment computation procedure by facilitating the ``classical'' optimization problem. Our approach is intrinsically scalable to large-scale generation management problems, and may in principle handle all kinds of uncertainties and practical constraints. Our results show the feasibility of the approach and are also promising in terms of economic efficiency of the resulting strategies.
The solutions of the optimization problem of generation (re)scheduling must satisfy many constraints. However, a classical learning algorithm that is (by nature) unaware of the constraints the data is subject to may indeed successfully capture the sensitivity of the solution to the model parameters. This has nevertheless raised our attention on one particular aspect of the relation between machine learning algorithms and optimization algorithms. When we apply a supervised learning algorithm to search in a hypothesis space based on data that satisfies a known set of constraints, can we guarantee that the hypothesis that we select will make predictions that satisfy the constraints? Can we at least benefit from our knowledge of the constraints to eliminate some hypotheses while learning and thus hope that the selected hypothesis has a better generalization error?
In the second part of this thesis, where we try to answer these questions, we propose a generic extension of tree-based ensemble methods that allows incorporating incomplete data but also prior knowledge about the problem. The framework is based on a convex optimization problem allowing to regularize a tree-based ensemble model by adjusting either (or both) the labels attached to the leaves of an ensemble of regression trees or the outputs of the observations of the training sample. It allows to incorporate weak additional information in the form of partial information about output labels (like in censored data or semi-supervised learning) or -- more generally -- to cope with observations of varying degree of precision, or strong priors in the form of structural knowledge about the sought model.
In addition to enhancing the precision by exploiting information that cannot be used by classical supervised learning algorithms, the proposed approach may be used to produce models which naturally comply with feasibility constraints that must be satisfied in many practical decision making problems, especially in contexts where the output space is of high-dimension and/or structured by invariances, symmetries and other kinds of constraints.
Tree based ensemble methods can be seen as a way to learn a kernel from a sample of input-output pairs. This paper proposes a regularization framework to incorporate non-standard information not used in the kernel learning algorithm, so as to take advantage of incomplete information about output values and/or of some prior information about the problem at hand. To this end a generic convex optimization problem is formulated which is first customized into a manifold regularization approach for semi-supervised learning, then as a way to exploit censored output values, and finally as a generic way to exploit prior information about the problem.
The aim of this work is to design intra-daily recourse strategies which may be used by operators to decide in real-time the modifications to bring to planned generation schedules of a set of units in order to respond to deviations from the forecasted operating scenario. Our aim is to design strategies that are interpretable by human operators, that comply with real-time constraints and that cover the major disturbances that may appear during the next day. To this end we propose a new framework using supervised learning to infer such recourse strategies from simulations of the system under a sample of conditions representing possible deviations from the forecast. This framework is validated on a realistic generation system of medium size.
In this paper we propose a methodology based on supervised automatic learning in order to classify the behaviour of generators in terms of their performance in providing primary frequency control ancillary services. The problem is posed as a time-series classification problem, and handled by using state-of- the-art supervised learning methods such as ensembles of decision trees and support-vector machines combined with several preprocessing techniques. The method was designed in the context of the Belgian system and is validated on real-life data composed of more than 600 time-series recorded on this system.