Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Argonne National Laboratory is a U.S. Department of Energy laboratory managed by UChicago Argonne, LLC.
KARTHIKEYAN BALASUBRAMANIAMNING KANG, SANGIL YIM, SIBY PLATHOTTAM, ROJAN BHATTARAI ENERGY SYSTEMS DIVISIONARGONNE NATIONAL LABORATORY
SPIDERWG MEETINGFOLSOM, CAAPRIL 11, 2019
T&D CO-SIMULATION TOOL (TDCOSIM) –TECHNIQUES AND DEVELOPMENT
Work supported by Ali Ghassemian and Dan Ton, U.S. DOE Office of Electricity, Advanced Grid Research and Development
OUTLINE
USE CASES
HOW?
2
OVERVIEW
FUTURE WORK
Functional requirements Steady-state and dynamic (for transient stability and disturbance ride-thru studies) simulations Scalability to model realistic interconnections and distribution networks Flexibility to implement DER interconnection standards Flexibility to implement advanced DER control functions
Software components PSS®E for T-Simulator OpenDSS for D-Simulator Python-based T&D interface
Benefits for transmission system entities Increase the awareness of the reliability impacts from high DER penetration Identify remedial control strategies that address future reliability concerns Ensure reliable and secure operation of the national grid Offer operational efficiency and economic benefits to various stakeholders across T&D
OVERVIEW OF ARGONNE’S T&D CO-SIMULATION TOOL (TDCOSIM)
DYNAMIC DER MODEL PV dynamic modeling (stand-alone Python-based code)
Phasor models of three-phase and single-phase distributed solar PV system with inverter, controls, and protection Volt-VAR and Volt-WATT implemented Voltage/frequency-deviation based protection compliant IEEE Std 1547 implemented
Integrated dynamic DER models into TDcoSim
USE CASE 1: A QSTS STUDY Transmission system: IEEE 14-bus system Distribution system: 1 IEEE 123-node system connected to transmission
load bus #5 PV: 1 PV system connected to the distribution system
Modeled as static generator operating at unity power factor Rated at 24 KW Representing 7% penetration in the feeder
Fig: 24-hour solar irradiance profile Fig: 24-hour active power dispatch profile at T&D boundary bus
USE CASE 2: IMPACT OF DER TRIP AND RIDE-THROUGH SETTINGS ON BPS STABILITY
6
Transmission: IEEE 118-bus test system Distribution: IEEE 123-node feeder
connected to buses 54, 59, 80, 90, and 116 10% and 20% PV penetration level
implemented for the feeders above Four study scenarios Base case without PV penetration 10% PV penetration with instant tripping 10% PV penetration with ride-through
setting 20% PV penetration with instant tripping
A 10-cycle temporary three-phase fault is applied at bus #59 on transmission system at 0.5 sec.
Increased PV penetration results in more severe voltage sag during the fault and higher swing after the fault
• Increased PV penetration results in more frequency deviation when they trip offline
• PV ride-through helps maintain system frequency
COMPONENTS OF CO-SIMULATION Message passing
– Single computing node– Multiple computing nodes
Data structure manipulation– Use the received message to make necessary changes to the data structure of the
simulator.– Challenges:
• Each simulator has a different data structure• For commercial software, we do not have direct access to all the relevant data structure
but are rather limited by the API provided by the vendor.– Restrictive
Time management– When to synchronize between simulators
Coupling protocol– Loosely coupled– Tightly coupled
7
User Interface Layer
Procedure Layer Model Layer Third-Party Layer
CLASS DIAGRAM
MainInterface
DefaultProcedure
DefaultStatic
Procedure
DynamicStatic
Procedure
PSSEModel
OpenDSSModel
PVDERModel
OpenDSSServer
OpenDSSClient
PSSE
OpenDSS
COMPUTATION VS COMMUNICATION
9
CONTEXT SWITCH, PERFORMANCE, SCALABILITY AND MORE How many number of threads can be run concurrently by a processor? What happens if there are more processes than that?
– Context switches How does TDcoSim handle such a scenario?
– We don’t. The OS does it! What is the maximum number of buses that TDcoSim can handle?
– Simulator bound– Memory bound and not CPU bound!
Performance?– CPU bound– Computation/Communication ratio
• Based on the system data and type of simulation – dynamic simulation or QSTS
10
CONTEXT SWITCH, PERFORMANCE, SCALABILITY AND MORE Performance?
– PVDER• Different solvers – scipy odeint and PETSc
Total computation time– PSSE time + OpenDSS time + TDcoSim overhead– TDcoSim overhead
• Improvements– We overlap computation and communication to improve
computation/communication ratio Scalability
– Multiple computing nodes• Currently exploring container based solution. Will use TCP/IP sockets.
Portability– OpenDSS based APIs developed are cross-platform. Works seamlessly under both
Windows and Linux.– Allows for multiple scalable architectures
11
USER INTERACTION, AUTOMATION AND DATA ANALYTICS User Interaction
– Lots of configurable options. This is both good and bad. Good:
– The user has full control over the simulation and can fine tune the scenarios. Bad:
– The user has full control over the simulation and can fine tune the scenarios. How can we address this issue?
– Templates/boilerplates for different analysis– Manual and automatic configuration– Helper class for configuration options
12
USER INTERACTION, AUTOMATION AND DATA ANALYTICS
Automation– Helper class for configuration options allows to programmatically create different
scenarios/study cases – batch processing! What about results?
– How large is the generated data.• We are currently in the order of million variables. At this scale a 20 second
simulation at half a cycle time step results in ~20GB of data.– How can an hypothesis be validated under such a scenario?
• Data Analytics (work in progress)– Do I need a big memory machine in order to use TDcoSim?
• No.– The user has the option to specify what variables to monitor – granularity.– TDcosim will write chunks of data from memory to disk when it senses that the
available memory becomes small.
13
WORK PLAN – 2019 AND BEYOND
14
T&D Co-simulation Tool• Release the first version of the tool with
user manual by June 2019
• Carry out commercial-grade testing
• Provide user support and training
• Develop additional features and models
• Incorporate real-world network data
• Develop related tools for transmission and distribution grid modeling and planning
Initiatives and Applications• Support SPIDERWG, LMTF• Applications in T&D planning coordination
(DER tripping/ride-thru settings, DER_A model parameterization and performance verification)
• Support NERC Long-Term Reliability Assessment and Probability Assessment
• Applications in T&D operation coordination (in conjuction with DERMS to manage DERs to provide T&D coordinated congestion management, system balancing, frequency and voltage control, and flexibility)
• Defining the operating challenges, dynamics over time
• Applications in T&D resilience enhancement
QUESTIONS?
15
T&d co-simulation tool (TDcosim) – TECHNIQUES AND DEVELOPMENTOutlineOverview of Argonne’s T&D CO-SIMULATION TOOL (TDCOSIM)Dynamic der modelUse case 1: a qsts studyUSE CASE 2: Impact of der trip and ride-through settings on bps STABILITYcomponents of co-simulationClass DiagramComputation Vs CommunicationContext switch, Performance, scalability and moreContext switch, Performance, scalability and moreUSER Interaction, automation and data analytics USER Interaction, automation and data analytics Work plan – 2019 and beyondQuestions?