cyber physical modeling of distributed resources for ... ?· 1 cyber physical modeling of...

Download Cyber physical modeling of distributed resources for ... ?· 1 Cyber physical modeling of distributed…

Post on 29-Jul-2018

214 views

Category:

Documents

0 download

Embed Size (px)

TRANSCRIPT

  • 1

    Cyber physical modeling of distributed resourcesfor distribution system operations

    Spyros Chatzivasileiadis, Member, IEEE, Marco Bonvini, Javier Matanza, Rongxin Yin, Zhenhua Liu,Thierry Nouidui, Emre C. Kara, Rajiv Parmar, David Lorenzetti, Michael Wetter,

    and Sila Kiliccote, Member, IEEE

    AbstractCo-simulation platforms are necessary to study theinteractions of complex systems integrated in future smart grids.The Virtual Grid Integration Laboratory (VirGIL) is a modularco-simulation platform designed to study interactions betweendemand response strategies, building comfort, communicationnetworks, and power system operation. This paper presentsthe coupling of power systems, buildings, communications andcontrol under a master algorithm. There are two objectives.First, to use a modular architecture for VirGIL, based on theFunctional Mock-up Interface (FMI), where several differentmodules can be added, exchanged, and tested. Second, to usea commercial power system simulation platform, familiar topower system operators, such as DIgSILENT Powerfactory. Thiswill help reduce the barriers to the industry for adopting suchplatforms, investigate and subsequently deploy demand responsestrategies in their daily operation. VirGIL further introduces theintegration of the Quantized State System (QSS) methods forsimulation in this co-simulation platform. Results on how thesesystems interact using a real network and consumption data arealso presented.

    Index TermsCo-Simulation, Functional Mock-up Interface,Modelica, Demand Response, Load Flow, DigSILENT Powerfac-tory, OMNET++

    I. INTRODUCTION

    Moving towards smarter grids, power systems complexityincreases through the embedding of communication networks,demand side management, electric vehicles, and the stochasticnature of several renewable energy sources (RES). Simula-tion platforms specialized in power systems can no longerhandle in an adequate way the increasing interdependencieswith systems such as communications, buildings, and electricvehicles. More detailed simulation tools are necessary to studythe system interdependencies and determine the appropriatecontrol strategies for optimizing power system operation. Anoption is to extend the existing power system simulation toolsby incorporating the dynamics of such networks inside thesame simulation platform. On the other hand, research in therespective fields has developed highly detailed and reliabletools, which can simulate the behavior and control of suchsystems. This paper adopts the co-simulation approach, wherehighly developed and reliable simulation tools, specialized inthe respective fields, are merged in a common co-simulationplatform to study the interdependencies between systems andidentify appropriate control strategies.

    The authors are with the Lawrence Berkeley National Laboratory, Califor-nia, USA. E-mail: {initial.lastname}@lbl.gov

    Although co-simulation has found a lot of applications ine.g. the automotive industry or building controls (e.g. BCVTB[1]), in power systems it is a relatively recent field whichhas seen some development during the last 8 years. Theapproach followed in this paper is to couple a commercialpower system simulation platform, widely used by powersystem operators, with advanced modeling tools for buidingsand communication networks. The goal is to determine theimpact the demand response strategies have on the networkand determined optimal algorithms to utilize flexible loads forpower system operation.

    Two are the main objectives. First, we aim at reducing thebarriers for adoption of novel demand response and other con-trol strategies in the daily power system operation. Couplinga trusted power system simulator, with which several powersystem operators are familiar, with other advanced modelingtools will help towards a wider adoption of such tools. Testingand becoming familiar with the impact of different strategieson the power system will allow the wider deployment andutilization of the energy reserves stored in buildings, e.g. inthe form of thermal inertia. Second, we need a modular co-simulation architecture, which will allow the easy exchangeand test of different simulation modules, as well as the easyextension with e.g. electric vehicle simulators, optimizationtools, hardware-in-the-loop, etc. For this reason, we use theFunctional Mock-up Interface standard, which provides astandardized interface for the coupling of several differenttools.

    This paper describes the Virtual Grid Integration Laboratory(VirGIL), which couples a commercial power system simulatorwith models for buildings and communication networks. Thegoal is to estimate the impact of demand response strategieson the grid, and to determine optimal algorithms for exploitingflexible loads (for example, the thermal energy stored inbuildings). To this end, we describe a modular co-simulationarchitecture that allows the easy exchange of different simula-tion models, as well as the easy extension with, e.g., electricvehicle simulators, optimization tools, hardware-in-the-loop,and so on.

    VirGILs architecture is based on the Functional MockupInterface, which defines a standard interface for exposingthe capabilities of a simulation tool [2]. FMI provides for amodular structure that allows the simple exchange and testingof different co-simulation tools. VirGIL is implemented inthe Ptolemy II framework, which combines continuous anddiscrete-event simulation [3].

    arX

    iv:1

    505.

    0007

    8v1

    [cs

    .CE

    ] 1

    May

    201

    5

  • 2

    This paper is organized as follows. Section II reviewsexisting co-simulation methods in power systems. Section IIIprovides the overview of the proposed co-simulation archi-tecture in VirGIL, while Section IV describes the FunctionalMock-up Interface (FMI). Sections V, VI, and VII, VIIIpresent respectively the development of the Power Systems,Buildings, Communications, and Control Functional Mock-upUnits (FMUs). Sections IX and X describe the simulation ofthe model exchange FMUs based on the QSS algorithm andthe operation of the master algorithm respectively. Section XIdescribes simulation results based on real data for the LBNLnetwork.Finally, Section XII concludes this paper and providesan outlook for future extensions of this work.

    II. CO-SIMULATION IN POWER SYSTEMS

    Over the last years, several cosimulation approaches forpower systems have been developed and documented in theliterature. One of the first documented efforts is [4], whichco-simulates power and communications systems. The authorsadvocate the use of already existing simulation tools that excelin their respective fields instead of creating new simulationplatforms (federated approach). In that work, power systemsare simulated with fixed step through PSCAD/EMTDC andPSLF while communications simulations are carried out on thediscrete event simulator ns-2. The two tools are synchronizedat specific synchronization points without an implementationfor a rollback function, which results in accumulation ofsynchronization induced inacurracies over time. The authorsimproved this approach in [5] where they use a masteralgorithm with a common timeline for both modules. Thereare no synchronization points, instead both simulators evolvesynchronously in time.

    Most of the co-simulation approaches for power systemscombine power system with communication network simula-tion (examples for distribution networks are [6], [7]). Ref. [8]reports a co-simulation approach for power systems and EVcharging and control, where they also use FMI for the couplingof one of the simulation tools to the master algorithm. A surveyof the latest simulation tools that are used for co-simulationin power systems is reported, among others, in [9].

    This work focuses on the interactions of building modelsfor energy consumption with distribution system models forpower system operation.

    Among the tools used for co-simulation, Gridlab-D is prob-ably one of the most widespread [10]. It has a flexible envi-ronment, which incorporates advanced modeling techniques,efficient simulation algorithms, but most importantly providea simulation environment not only for power systems, but alsoincorporating detailed load modeling, rate structure analysis,distributed generator and distribution automation.

    In this paper, a commercial power system software, DigSi-lent Powerfactory, is used for power systems simulation.Building a co-simulation platform incorporating Powerfactory,a tool that several utilities trust and use in their daily operation,decreases the barriers for wider adoption of co-simulation toolsfrom the industry. Power system operators can incorporatetheir version of Powerfactory with the co-simulation platform

    to investigate in more detail the effect of demand responsesignals, decide and subsequently deploy the most appropriatein real-time operation. Powerfactory has the additional advan-tage of being capable to model both AC and DC systems.A co-simulation approaches incorporating Powerfactory hasalso been documented in [11]. However, this is the first timethat a modular co-simulation architecture, based on FMI, isimplemented for coupling Powerfactory with the rest of thesimulation tools.

    Besides the development of the appropriate models andcontrols within each simulation tool, the focus in this paper ison the development of the wrapper functions which will makethe modules compatible to the FMI standard for co-simulation.FMI provides for a modular structure of the co-simulationplatform which allows the simple exchange and testing of dif-ferent co-simulation tools. VirGILs master algorithm will bePtolemy II, which can combine both continuous and discrete-event simulation. At the same novel simulation algorithmsare implemented in Ptolemy II, such as QSS (Quantized-State-Simulation) which allow for hig

Recommended

View more >