EASI: Energy autonomous smart micro-grids

Research project
Funding STW Perspectief
Duration 2012-09-10 ~ 2016-09-10
Contacts Johann Hurink and Gerard Smit
Staff Gerwin Hoogsteen and Thijs van der Klauw
Collaborations STW and Alliander
Website

Description

In recent years, more and more distributed generation has been installed on a neighborhood level. When enough (renewable) generation like PV panels, biomass installations and wind-turbines and storage are installed, it is possible to create a self-supplying neighborhood in a so-called energy autonomous smart micro-grid. This work focuses on such an autonomous smart micro-grid.

Problem and goal

In recent years, more and more distributed generation has been installed on a neighborhood level. When enough (renewable) generation like PV panels, biomass installations and wind-turbines and storage are installed, it is possible to create a self-supplying neighborhood in a so-called energy autonomous smart micro-grid. A neighborhood can be a collection of residential buildings, but also small business parks, a university campus, a pop festival or a village in a remote area. Since (renewable) distributed generation is not (always) producing the energy when it is needed, and consumption is also not always predictable, a micro-grid behaves highly stochastic. An energy autonomous neighborhood should always be in balance, i.e. at all times the energy consumption should be approximately the same as the production, where up to a certain level, mismatches between consumption and generation can be bridged by energy storage. In a less extreme scenario, when there is temporary shortage of energy the main electricity grid could be used for back up and similarly the neighborhood might provide excess energy to the grid when energy is expensive. In case of emergency in the main grid, the micro-grid can become autonomous temporarily. This means that the autonomous grid must be able to switch to/from the main grid, without interruption of the power supply. To reach this goal, an ICT based control system needs to be developed.

Challenges

  • Stochastic modeling of an energy neutral neighborhood

    The type of generators that can be used in a neighborhood is very much dependent on the environment where it is used. For example, in a suburban neighborhood with a lot of farms energy can be obtained from bio wastage of animals and crops, in a neighborhood close to the sea wind energy might be a good option and neighborhoods close to a river might use water energy. Unfortunately, energy from the environment is not always available and thus generation is inherently unreliable. An important challenge is to stochastically model and predict how much energy can be generated, which depends on the used generation method and may be depending amongst others on time of day and weather conditions.

    In order to control the micro-grid, it is also important to develop a systematic and flexible model of the elements of the grid (various types of energy generators, various static or dynamic loads, batteries and other energy-storage devices, transmission lines, switches, etc.) as well as the network structure. The stochastic model of the uncertainties can then be superimposed on this physical layer. As the demand may be depending on human behavior, the demand has also a stochastic behavior. As mentioned before, to cover mismatch between generation and demand energy storage is needed. Based on the stochastic demand and supply models possible options for the energy infrastructure can be derived. Therefore, the ICT system has to cope with various forms of uncertainties.

  • Matching demand and supply

    Within an autonomous micro-grid, supply and demand must be in balance. This balancing must be done on different levels. When the micro-grid is cut off from the main grid (operating autonomously or islanded), the grid stability must be maintained on a (micro) second level, such that supply and demand within the micro-grid is balanced. When this is not done properly high peak voltages may appear in the micro-grid and connected appliances may be damaged.
    To design control strategies, which prevent these high peak voltages, accurate models of the grid components are essential. In an autonomous micro-grid, the buffers also take care of grid balancing. When no conventional generation is available (i.e. the micro-grid is disconnected from the main grid), there is also no spinning reserve available and very fast changes in demand should be taken care of with batteries.

    Spinning reserve in conventional power plants is a mechanical way of keeping up with fast changes, but for an autonomous (autarchic) micro-grid a fast control algorithm is required to maintain the fine grain stability supplied by the batteries. This algorithm should be a fast, as it has to deal with the real-time fluctuations in demand and supply. On a course grain level (minutes), the supply and demand can be balanced with different methods, for example by shifting demand in time or temporarily switching off appliances (e.g. a freezer); the remaining mismatch should be solved by the energy buffer. To be able to achieve this, we investigate how much buffering is required, which type of buffering is needed and where in the grid it should be placed Next, the developed control algorithms must be capable of using this buffering capacity in a near optimal way. A first version for such an algorithm (TRIANA) is developed at the University of Twente, but this algorithm needs to be tailored for this autonomous situation. To guarantee the good behavior of the demand and supply balancing mechanism, information must flow within the micro-grids. When the micro-grid is connected to the main grid, another higher level control algorithm is required to match the import/export between the micro-grids and information must flow among the micro-grids. Algorithms, which control or provide accurate information to the physical infrastructure, constitute the cyber component of the system.