IEEE

2015 Stochastic Optimization for Smart Grid Operation (Panel)

The transition toward sustainable and environmental friendly energy supply with massive integration of renewable energy sources and higher demand side participation entails devising smart grid oriented strategies to optimally address operational concerns within a context of increasing uncertainties and significant structural changes. This panel focuses on

i)   An overview and discussion on the formulation complexities, scalability, and solution challenges of different stochastic optimization problems related to technical and economic aspects of power system operation.

ii)  Case studies, performance comparison, and experience with the application of classical, heuristic, and hybrid optimization algorithms.

 

 

Panelists:

 

Zita Vale (Polytechnic of Porto): “Stochastic optimization of distributed energy resources in smart grids”

Renewable energy, Energy Storage Systems (ESS) and Electric vehicles (EVs) are key players in the future smart grid. In this context, energy management systems must be able to efficiently deal with the uncertainties of wind and photovoltaic installations and demand, including EVs charging.

The proposed model for the resource scheduling in smart grids considers a stochastic approach to deal with these uncertainties and meta-heuristic optimization. Demand Response (DR) and the Vehicle-to-Grid (V2G) ability are considered as available resources, together with distributed generation and storage. Energy can be bought from several suppliers offering different commercial options and market based energy transactions enable additional energy buying and selling opportunities.

The application of the proposed model is illustrated with a case study concerning a real distribution network in Portugal. The model performance is compared for the stochastic and the deterministic approaches.

 

Germán Morales-España (Delft University of Technology), Andres Ramos  (Pontifical University of Comillas): “Accelerating the Convergence of Stochastic Unit-Commitment Problems by Using Tight and Compact MIP Formulations”

Stochastic Unit Commitment (SUC) problems under the Mixed-Integer Programming approach are computationally intensive. Despite the significant improvements in MIP solving, the time required to solve SUC problems continues to be a critical limitation their size and scope. Contributions have been done regarding computer power (eg., clusters) and solving algorithms (eg., solvers, decomposition techniques). However, very few attention has been paid to the quality of the MIP formulation, which actually defines its computational complexity.  Creating tight or compact computationally efficient MIP formulations is a non-trivial task because the obvious formulations are very weak or very large, and trying to improve the tightness (compactness) usually means harming the compactness (tightness). We propose an SUC that is simultaneously tight and compact. Consequently, the computational burden is dramatically reduced in comparison with common SUC formulations. Numerical results show that the proposed SUC achieved speedups above one order of magnitude over traditional SUC formulations.

 

Hiroyuki Mori (Meiji University): “Development of Tabu Search with the stochastically reduced neighborhood”

This talk proposes a new technique for combinatorial optimization problems in Smart Grid. The proposed method makes of Tabu Search to carry out the efficient search process in the combinatorial optimization problems. Tabu Search is one of metaheuristics that uses simple rules or heuristics to evaluate better solutions within a given period of time. It introduces adaptive memory called tabu list into the hill-climbing method so that it easily escapes from a local minimum. The applications of Tabu Search are widely spread in the engineering fields because of the efficiency of the algorithm. However, it has a limitation that it is inclined to be time-consuming as the size of problems to be solved becomes much larger.  That is because Tabu Search creates a large-scale neighborhood near a solution and searches better solution candidates in it. To overcome the drawback, an efficient method is developed to reduce the neighborhood by the stochastic selection that comes from Ordinal Optimization. The effectiveness of the proposed method is demonstrated in sample distribution systems.

 

Leontina Pinto (Engenho): “Smart grids and smart consumers: Joining forces towards a sustainable and reliable operation”

Smart grid targets more than simply energy delivery. As consumers transform into clients, the successful supplier will capture customer’s needs and requirements, translating them into tailored products offered through specific pricing schemes. For instance, the consumer of the (near) future will be able to specify his/her degree of sustainability and/or security – and pay for it, or alternatively take the associated risks.

This paper presents a comprehensive framework for the optimal operation under uncertainties (from equipment failures to climate variability), under a multi-objective function (combining costs, emissions and risks). Under this platform, the smart grid will evaluate actual and expected operation conditions, send consumers associated pricing signals reflecting their consumption consequences (expected costs, emissions and risks) and, according to each specific plan, perform corresponding actions (such as automatically load reduction or price increase).

A case study illustrates the symbiosis between network and load, absorbing variability risks while achieving maximum sustainability level.

 

Hsiao-Dong Chiang, Robert J. Thomas  (Cornell University): Stochastic Security-constrained AC Optimal Power Flow Solver for large Power Networks with Renewable

The stochastic contingency-based security constrained AC Optimal Power Flow (SCSC AC OPF) formulation be-hind the SuperOPF makes it applicable to a variety of problems arising in power system planning and operation. The ultimate goal of our work is to develop a robust solver in the context of co-optimization framework that correctly accounts for contingencies, ancillary services, and static and dynamic constraints in determining both dispatch and price. In this talk, we will focus on the following: (i) a robust solver with the capability to deal with a large set of scenarios of different types (uncertainties and contingencies), (ii) how to adjusting  both real and reactive power control variables for the application functions needed in a production ISO/RTO Energy Management System (EMS),  (iii) a solution methodology which can handle large-scale, continuous and discrete, power system optimization models with AC power flow constraints and static security constraints, (iv)  extensive evaluation on an ISO’s OPF models of 6354-bus system with renewable.

 

Vladimiro Miranda (University of Porto), Hrvoje Keko (University of Porto), Leontina Pinto (Engenho): “Electric vehicles in smart grids: a hybrid Benders/EPSO solver for stochastic reservoir optimization”

The feasible connection of electric vehicles to a power grid in large scale demands a smart grid context: the vehicles must hold V2G capacity. This transforms the collection of vehicle batteries in a distributed stochastic reservoir, similar to a water reservoir but with a probabilistic definition of size and storage in each moment. The paper presents a model to solve the generator scheduling problem in such a context. The uncertainties in wind generation, hydro-power availability and distributed battery storage are all taken into account, using scenario reduction techniques. The solving strategy is to develop a Benders decomposition approach where the cuts generated are transformed into penalties that modify the optimizing landscape. The master problem is solved with an Evolutionary Particle Swarm (EPSO) algorithm. The paper will include clarifying examples of the technique.