Documentation Home Page HYPERSIM Home Page
Pour la documentation en FRANÇAIS, utilisez l'outil de traduction de votre navigateur Chrome, Edge ou Safari. Voir un exemple.

Switches for EMT Studies

Types of Switch Models for EMT Studies

There are several types of switch models available to model the power electronic (PE) interfaces of distributed energy resources (DER) to perform studies related to DER integration to the power grid, and they all have their pros and cons. Thus proper selection requires knowledge of the test conditions and requirements, where tradeoffs can be made between the speed of simulation and the representation of details and achievable resolution, etc. A classification of the types of models available is shown in the figure below:

Knowing How/When to Reconcile Bandwidth (Computational Effort) with Resolution (Model Accuracy) for Best Simulation Results

The tradeoff between resolution (accuracy, coverage) and real-time performance (no skipped timesteps, throughput) is important to the strengths and limitations of simulation options. Users should know what they need from their simulation and what is best at providing it under the various circumstances under which they'll simulate. The figure below positions the available switch models on the model accuracy vs computational ease trade-off spectrum. 

See the second appearance of this graphic at the bottom of this topic to explore this very important strategic tradeoff in more detail.

Detailed Semiconductor

1 Source:

2 Source: A. Sokolov, “Variable-Speed Power Switch Gate Driver for Switching Loss Reduction in Automotive Inverters.”

Modeled Features

  • Instantaneous turn on/off time representation
  • Conduction and switching losses (Requires good tuning of parameters)
  • Thermal model simulation with high accuracy
  • Ripple representation with high accuracy
  • Device transient characteristics (e.g. MOSFET, IGBT, etc.) can be modeled.
  • Highest accuracy in the representation of the power electronic converters
  • Computationally intensive, since the switches are modeled with their details
  • Requires a very low timestep (~10s of ns) for accurate solution of the discretized non-linear switch models
  • Not suitable for real-time simulations or Hardware-in-the-Loop testing

Ideal Switch

Modeled Features

  • Instantaneous turn on/off time representation
  • Conduction and switching losses (Requires tuning of parameters)
  • Ripple representation with high accuracy
  • The classic EMT-type software model
  • A straightforward model which does not require handling particular cases
  • Good accuracy for most power electronics CHIL tests with small enough time-step
  • A 10 times smaller timestep than the switching time period T_(s_max )=1/(10×f_sw ) to get a 10% resolution accuracy on the PWM (may result in numerical oscillations)
  • Computationally intensive and requires matrix pre-calculation or system decoupling for real-time simulation or larger systems
  • Requires larger memory if pre-calculation of matrices is used
  • Requires tuning snubbers with respect to time-step and surrounding model eigenvalues

Constant Conductance

Also known as Pejovic Method | Associate Discrete Circuit.

Modeled Features

  • Instantaneous turn on/off time representation
  • Ripple representation with higher accuracy
  • Good accuracy and allows fast simulation for CHIL testing of faster power electronics controls
  • Low computational burden allowing very low time-steps when implemented on FPGA
  • Creates virtual power loss (compensated for in the eFPGASIM implementation for standard converter topologies)
  • Requires tuning of the Gs parameters (eFPGASIM provides a Gs calculation tool to help tune the Gs parameter)

Switching Functions

Also known as Time Stamped Bridge (TSB) | Virtual FPGA Switching.

Modeled Features

  • Suitable for voltage-source converters modeling
  • Compensates for the adverse effects of pulsing from controllers (CHIL) occurring in between discrete-time steps
  • Accurately represents the voltage harmonic spectrum near the fundamental frequency of operation
  • Allows effective modeling of switch dead-times
  • Good accuracy for system-level and converter level studies
  • Fast execution time
  • Requires T_(s_max )=1/(~4×f_sw ) for an accuracy of ~2% on the duty cycle
  • Allows study of converters in larger systems without requiring as much decoupling as an ideal switch to achieve real-time execution
  • With computation technology (ex. FPGA) which is very fast, but not enough to simulate very fast power electronics (ex. fsw=100 kHz), the switching function remain a very good solution for real-time simulation
  • Certain particular cases may not be possible to simulate (ex. internal faults)

Average Models

Modeled Features

  • Models the average signal produced by the converters
  • Models the near fundamental dynamics of the system
  • Effects of switching are neglected
  • Very fast execution
  • Good for large system studies and controller interactions
  • Switching frequency and its related phenomena are neglected
  • Does not include low-frequency phenomenon due to switching such as non-linearity due to dead-time

Types of Average Models

Voltage Source

  • Implemented with the output filter
  • Can include DC side dynamics
  • Models the filter related dynamics of the system

Current Source

  • Usually does not include DC side dynamics
  • Filter dynamics are also neglected
  • System-level control dynamics can be modeled

Model Suitability for Microgrid/DER studies

This graphic depicts two continua, both along the X-axis:

  • Computational ease, i.e., the requirement of higher processing power and bandwidth, going from lowest required at the right to highest at the left.
  • Model Exactitude, i.e., resolution or detail required, going from lowest requirements at the right to highest at the left.
    • Also included on this graphic is real-time simulation to the right: real-time simulation makes very particular demands in terms of the processing power, such as no skipped time steps and higher speed of simulation.
    • Conversely, offline simulation may be faster or slower than real-time, but it removes some factors from consideration: if one has more time, one can request more fidelity to models and less importance be assigned to the speed of simulation. 

This graphic details the typical frequencies of interest that should concern users simulating various circumstances, from switching harmonics to controller interactions.

Similarly, it lays out the range of time step lengths, from nanoseconds to full seconds at which users may execute their models. 

Example Available in HYPERSIM: 2 MW Connected PV Array

The following example is available for exploration and further understanding in HYPERSIM. Using the information outlined above, make changes and see how this affects the accuracy/bandwidth of your simulation.


S. Rosado, R. Burgos, S. Ahmed, F. Wang, and D. Boroyevich, “Modeling of Power Electronics for Simulation-Based Analysis of Power Systems,” p. 8, 2007.

D. Maksimovic, A. M. Stankovic, V. J. Thottuvelil, and G. C. Verghese, “Modeling and simulation of power electronic converters,” Proc. IEEE, vol. 89, no. 6, pp. 898–912, Jun. 2001, doi: 10.1109/5.931486.

C. Dufour, “ArtEvents, a simplified and reliable event-based block set for Simulink,” p. 7.

G. De Carne et al., “Which Deepness Class Is Suited for Modeling Power Electronics?: A Guide for Choosing the Right Model for Grid-Integration Studies,” EEE Ind. Electron. Mag., vol. 13, no. 2, pp. 41–55, Jun. 2019, doi: 10.1109/MIE.2019.2909799.

A. M. . A. Amin, M. I. El-Korfolly, and S. A. Mohammed, “Exploring aliasing distortion effects on regularly-sampled PWM signals,” in 2008 3rd IEEE Conference on Industrial Electronics and Applications, Singapore, Jun. 2008, pp. 2036–2041, doi: 10.1109/ICIEA.2008.4582878.

F. Gao, “Real-Time Simulation Methods of Power Electronic Systems”, IEEE Power Electronics Society and Transportation Electrification Committee joint webinar, 2020. [Online]. Available: [Accessed: 04 – Jun – 2020]

OPAL-RT TECHNOLOGIES, Inc. | 1751, rue Richardson, bureau 1060 | Montréal, Québec Canada H3K 1G6 | | +1 514-935-2323