c i f i c a DIgSILENT Pacific P Power system engineering and - PowerPoint PPT Presentation

c i f i c a DIgSILENT Pacific P Power system engineering and software T N What future for RMS simulation? E Scott Hagaman L DIgSILENT Pacific Seminar 28 November 2019 I S g I D

c i f i c a Presentation Overview P • A brief history of power system simulation T • Different types of simulation N • Time domain: RMS vs EMT simulation • Hybrid RMS/EMT approach E • Stability of grids dominated by inverter based generation L I S g I 2 D

c i f i c a P T N E L A brief history of power system simulation I S g I 3 D

c i f i c a Transient Network Analysers P • Electromagnetic transient studies T • Developed late 1930s • Dominant simulation technique for many years N • Series of inductors wound on magnetic cores to represent transformers, lines, source impedances etc. • Generators represented as ideal voltage sources behind reactance E • Advantages • Simulation in real time – short run time for multiple transient simulations (ms per run)! L • Computationally stable! • Disadvantages I • Relatively inflexible S • Time consuming setup • Limited system size g I 4 D

c i f i c a The Gettysburg Times P Wednesday, February 4, 1953 T N E $400k USD in 1953 L = $3.86M USD I in 2019 S g I 5 D

c i f i c a Westinghouse AC Network Analyser - Victoria P • Prior to the 1950’s, several independent electricity systems T serviced Victorian load centres • Supplied by Westinghouse to State N Electricity Commission of Victoria (SECV) in 1950 E • Used for development of Victorian power system including the L connection between the Latrobe Valley (Yallourn, Morwell and I Hazelwood brown coal power stations) S and Melbourne, and from Melbourne to NSW (Snowy Mountains Scheme) g I 6 https://collections.museumvictoria.com.au/articles/10181 D

c i f i c a Analogue computer simulation P • Advantages T • Simultaneous integration of differential equations (digital computers require sequential calculation) N • Disadvantages E • Was slower than transient network analysers • High set up time L • Algebraic equations requiring loops cannot be solved easily I S g I 7 D

c i f i c a Digital computer simulation P • Rapid increase in power of T digital computers N • Mathematical models of E components programmed into a computer L I • Models and simulation S methods developed and matured over time g I 8 https://ourworldindata.org/grapher/transistors-per-microprocessor?time=1971..2017 D

c i f i c a Different types of simulation P • Balanced and unbalanced load flow T • Short circuit • Contingency analysis N • Protection system analysis • Harmonics E • Time domain simulation (RMS and EMT) • Eigen analysis L • Quasi-dynamic simulation I • Network optimisation: S • optimal tie open points • optimal capacitor placement g • NPV analysis I 9 D

c i f i c a Model data P • Detail of data required commensurate with analysis sophistication T • Unbalanced network analysis requires unbalanced network representation N • Harmonic analysis requires frequency dependent parameters • Protection analysis requires device models and settings E • Time domain simulation requires controller models • NPV analysis requires capital and operational costs L I S g I 10 D

c i f i c a P T N E L Time domain: RMS vs EMT simulation I S g I 11 D

c i f i c a Types of time domain analysis P • RMS (root mean squared) simulations primarily investigate the interaction between the electrical and mechanical systems and controllers. T • Fundamental frequency response only (voltage and current phasors). • Typical time frame of interest 0.5-60 seconds. N • Uses include transient stability, controller design and controller optimisation. E • EMT (electro-magnetic transients) also consider the interaction between electrical and magnetic phenomena. • Full sinusoidal representation of three phase voltage and current. L • Historically: I • typical time frame of interest is 20 ms – 1 second. S • uses include insulation co-ordination, capacitor inrush, transformer inrush and wind- farm fault ride through. • Recently EMT simulation used to examine transient stability over longer g durations (0.5-60 seconds) due to the proliferation of inverter based generation I 12 D

c i f i c a Time domain: RMS vs EMT simulation P • The choice of simulation method depends on the phenomena that is T being investigated and/or the level of detail required in the analysis. N 400.00 2.00 350.00 1.00 E 300.00 0.00 250.00 L 200.00 -1.00 150.00 -2.00 I -100.00 0.00 100.00 [ms] 200.00 T1: Phase Voltage A/LV-Side in p.u. 100.00 0.00 2.50 5.00 7.50 [s] 10.00 T1: Phase Voltage B/LV-Side in p.u. G1: Active Power in MW T1: Phase Voltage C/LV-Side in p.u. SubPlot Date: 4/6/2009 Transformer_flux(1) Date: 4/6/2009 S Annex: /5 Annex: /5 RMS simulation EMT simulation g I 13 D

c i f i c a Time domain: RMS vs EMT simulation P • Iterative procedure to solve AC and DC load flows at any given time point T (algebraic equations), along with the solution for dynamic model state variables (differential equations) N • RMS simulation E • Steady-state, symmetrical (balanced) or three-phase (unbalanced) representation of the passive electrical network L • EMT simulation I • Dynamic behaviour of passive network elements is also taken into account S • The integration step size has to be significantly smaller than in the case of a steady-state representation and as a result, the calculation time increases g I 14 D

c i f i c a RMS-EMT co-simulation P • The objective of RMS-EMT co-simulation is to bring the best of the two T worlds (EMT, RMS) together: • RMS analysis works on fundamental frequency (e.g. 50 Hz), faster N and main focus is on electro-mechanical dynamics only • EMT analysis works on instantaneous values, slower (for smaller time E steps) and can examine the electro-magnetic transient effects L • The general implementation of RMS-EMT co-simulation is to let each I computation task carried out independently in each time step and S feedback the other task with new results at the end of each time step g I 16 D

c i f i c a Hybrid RMS/EMT approach P T N E L I S g I 17 D

c i f i c a Hybrid RMS/EMT approach P • Definition of regions allows for calculation parallelisation T N • Takes advantage of modern multicore processing E • Efficiency benefit in RMS simulation of slow moving phenomena with more detailed EMT simulation of faster transient phenomena L I S g I 18 D

c i f i c a P T N E L Demonstration – RMS/EMT hybrid I S g I 19 D

c i f i c a Accuracy vs precision P • Both EMT and RMS models are T only models that will always respond slightly different from the N reality. • Both EMT and RMS models should E be validated. It cannot be assumed that an EMT model is less or more accurate than an RMS model. L • Because EMT models consider I higher frequency phenomena, it is S generally more difficult to validate other than in a laboratory environment g I 20 https://circuitglobe.com/accuracy-and-precision.html D

c i f i c a Unbalanced RMS and EMT models vs P measurement T N E L I S g Symmetrical faults I 21 D

c i f i c a Unbalanced RMS and EMT models vs P measurement T N E L I S g Asymmetrical faults I 22 D

c i f i c a Discussion P • Unbalanced RMS model can accurately represent the main behaviours of non- T synchronous generators at the system fundamental frequency. N • EMT model is not necessarily more accurate. It can show more detailed responses and can possibly be more accurate but at the expenses of computing time and resource, higher development and maintenance cost, E higher modelling complexity and dependencies. It is often encrypted/blackboxed due to IP concerns and hence more difficult to use and debug. L • Models are subjected to the level of details, modelling quality and accuracy as I well as mistakes. S • Models should be validated against measurement to confirm the accuracy of the g models for a specific study objective. I 23 D

c i f i c a Automation bias P T N E L I S g I 24 https://abcnews.go.com/blogs/headlines/2012/03/gps-tracking-disaster-japanese-tourists-drive-straight-into-the-pacific D

c i f i c a P T N E Stability of grids dominated by inverter based L generation I S g I 25 D

c i f i c a Inverter dominated grids P • Increasing levels of renewables resulting in system dynamics more and more T dominated by inverter based generation • Renewable technologies currently deployed will shape the dynamic grid N behaviour for the next 5-10 years • Most commercially available solar and wind inverters utilise grid-following control strategies E • Grid-following controls • Current injection based on PI characteristic L • Works as long as the grid can absorb the current • Unstable under weak grid conditions I S • Growing demand for grid-forming controls • Basically generators that can operate in a stable fashion within an integrated grid g • Synchronous condensers exhibit grid-forming characteristics I 26 D