Cooperative energy management of a community of smart-buildings: A Blockchain approach

Demand Response (DR) is progressively moving from a centralized, unidirectional structure to a set of advanced decentralized mechanisms that better balance distributed supply and demand. This paper presents a decentralized cooperative DR framework to manage the daily energy exchanges within a community of Smart-Buildings, in the presence of local Renewable Energy Sources (RES). The proposed algorithm taps into the flexibility of the participants to let them decide of a day-ahead community power profile, and subsequently ensures the forecast tracking during the next day. In practice, the algorithm is fully decentralized by the Blockchain technology, that enables a trusted communication medium among the participants and enforces autonomous monitoring and billing via Smart-Contracts. With such an energy management framework, participating Smart-Buildings can together aim at a common objective, such as carbon-free resources usage or aggregated grid services, without depending on a centralized aggregator/utility. Simulations on realistic Swiss building models demonstrate that nearly all the renewable production resources could be harnessed locally through the presented framework, compared to selfish individual optimization. Under a quadratic cost of grid electricity, the considered community profile could dramatically be flattened, hence avoiding costly peaks at the grid interface. A scalability analysis shows that, considering the current public Ethereum Blockchain, the framework could handle a community size of up to 100 Smart-Buildings.

Exact Convex Modeling of the Optimal Power Flow for the Operation and Planning of Active Distribution Networks with Energy Storage Systems

Mostafa Nick

The distribution networks are experiencing important changes driven by the massive integration of renewable energy conversion systems. However, the lack of direct controllability of the Distributed Generations (DGs) supplying Active Distribution Networks (ADNs) represents a major obstacle to the increase of the penetration of renewable energy resources characterized by a non-negligible volatility. The successful development of ADNs depends on the combination of i) specific control tools and ii) availability of new technologies and controllable resources. Within this context, this thesis focuses on developing practical and scalable methodologies for the ADN planning and operation with particular reference to the integration of Energy Storage Systems (ESSs) owned, and directly controlled, by the Distribution Network Operators (DNOs). In this respect, an exact convex formulation of Optimal Power Flow (OPF), called AR-OPF, is first proposed for the case of radial power networks. The proposed formulation takes into account the correct model of the lines and the security constraints related to the nodal voltage magnitudes, as well as, the lines ampacity limits. Sufficient conditions are provided to guarantee that the solution of the AR-OPF is feasible and optimal (i.e., the relaxation used is exact). Moreover, by analyzing the exactness conditions, it is revealed that they are mild and hold for real distribution networks. The AR-OPF is further augmented by suitably incorporating radiality constraints in order to develop an optimization model for optimal reconfiguration of ADNs. Then, a two-stage optimization problem for day-ahead resource scheduling in ADNs, accounting for the uncertainties of nodal injections, is proposed. The Adaptive Robust Optimization (ARO) and stochastic optimization techniques are successfully adapted to solve this optimization problem. The solutions of ARO and stochastic optimization reveal that the ARO provides a feasible solution for any realization of the uncertain parameters even if its solution is optimal only for the worst case realization. On the other hand, the stochastic optimization provides a solution taking into account the probability of the considered scenarios. Finally, the problem of optimal resource planning in ADNs is investigated with particular reference to the ESSs. In this respect, the AR-OPF and the proposed ADN reconfiguration model, are employed to develop optimization models for the optimal siting and sizing of ESSs in ADNs. The objective function aims at finding the optimal trade-off between technical and economical goals. In particular, the proposed procedures accounts for (i) network voltage deviations, (ii) feeders/lines congestions, (iii) network losses, (iv) cost of supplying loads (from external grid or local producers) together with the cost of ESS investment/maintenance, (v) load curtailment and (vi) stochasticity of loads and renewables production. The use of decomposition methods for solving the targeted optimization problems with discrete variables and probable large size is investigated. More specifically, Benders decomposition and Alternative Direction Method of Multipliers (ADMM) techniques are successfully applied to the targeted problems. Using real and standard networks, it is shown that the ESSs could possibly prevent load and generation curtailment, reduce the voltage deviations and lines congestions, and do the peak shaving.

EPFL2016

Multiport Energy Gateway

Yan-Kim Tran

The global problem of energy supply and the reduction of the carbon footprint can only be solved by a massive replacement of fossil fuels or atomic energy by renewable resources. As the latter are mostly weather-dependent, they are uncontrollable and unpredictable in the long term. Their relative volatility can be mitigated by integrating a large number of these resources, by their geographically decentralized distribution, and by the use of storage elements. These ones being mostly DC, their integration would be eased considerably by the use of DC grids which allow the synchronization of resources to be avoided and eliminate line impedance effects, thus allowing energy to be transfered over longer distances. In the past, voltage step-up and step-down transformations were only possible with the help of low-frequency (50-60Hz) and bulky transformers, which are still used nowadays in most AC grids. But the progress of the last two decades in the field of power semiconductors now allows the realization of high-performance DC-DC converters with increasingly high voltages. A promising concept is the solid-state-transformer (SST), which uses a multi-modular structure integrating multiple medium-frequency transformers and offers an efficient and compact solution for voltage adaptation. While several researches on this particular subject have been done or are in progress for bi-port converters, the multiport version with the objective of integrating storage elements on additional ports has been studied only very few. Therefore, this thesis proposes a new converter structure combining both the SST principle for medium voltage and the multiport resonant converter principle to limit switching losses and achieve higher switching frequencies, necessary for an increased power density. After a brief overview of existing storage technologies and multiport converters topologies, a description of the proposed structure and its operating principle are presented. It results in a relatively simple operation based on operating modes dependent on the direction of the energy flow and allowing the use of one or more storage elements independently. Then, the mathematical models required to describe the converter are developed for each of these operating modes and make it possible to highlight the operating areas benefiting from soft switching and having a higher efficiency. They are also used to define sizing criteria that are applied to the design of a prototype converter for the medium-voltage grid and which parameters are used as the basis for the simulation. Finally, a description of the converter dynamics is presented using simplified models from which the control structure is derived and the controller parameters are calculated. In order to illustrate each of the elements developed in this work, a low-voltage prototype is designed to support certain points with experimental results.

EPFL2019

Exact Convex Modeling of the Optimal Power Flow for the Operation and Planning of Active Distribution Networks with Energy Storage Systems

Mostafa Nick

EPFL2016

Multiport Energy Gateway

Yan-Kim Tran

EPFL2019