Can you provide any general guidance on how to perform an EMT simulation?
Refer to the detailed answer.
Several things are generally worth to keep in mind when executing an EMT simulation with PowerFactory:
1. Necessity of accurate representations of line models. PowerFactory offers lumped as well as distributed parameters transmission line models (either constant or frequency dependent parameters) for EMT type transient simulations. It is the user who should make a decision on whether a distributed parameters model is required in a particular case. The decision should be made considering (a) the length of the line (smaller line lengths imply a higher frequency up to which a lumped model is still suitable for dynamic simulation, and vice-versa for longer lines), (b) the frequency domain of the transient under study (transformer inrush simulations have characteristic transients of double the fundamental grid frequency, while line switching actions have transients in the kHz range). As a good verification tool, the Frequency sweep analysis can be employed to compare the line impedance frequency characteristic (over a pre-defined frequency interval) in the case of a lumped or distributed parameters model (either constant or frequency dependent). This is an easy to use visual tool that can plot both characteristics on the same diagram. While at low frequency the impedance between the models is negligible at higher frequencies the error will gradually increase. As a consequence, one can decide for a certain frequency domain whether a lumped parameters model is still adequate for EMT simulation or they should turn to a constant parameters or even a frequency dependent parameters model. Part 4 of IEC60071 (Computational guide to insulation co-ordination and modelling of electrical networks) gives recommendations for the modelling of power system components for EMT type transient simulations as well.
2. Keeping the maximum observable simulation frequency at one, maximum two frequency decades above the characteristic transient dominant frequency will generally ensure reliable results while keeping to a minimum the time required for performing the simulation.
3. Making sure that line models are reasonably set up when the "Distributed Parameters" model is selected. Try to reasonably set the "Frequency for travel time estimation" in order to be in the same range with the expected transient. Make sure that the frequency for travel time estimation can be practically represented using the given simulation settings/ simulation time steps (if automatic step size adaptation is used then the relevant time step is the maximum value) . If frequency dependent parameters model is selected, try to reasonably set the minimum and maximum frequency for parameter estimation (e.g. 3 decades below for minimum, 2-3 decades above the expected transient).
4. Make sure data is correctly/reasonably set in all models for all positive, negative as well as zero sequence parameters since with EMT simulation uses almost all data.
5. There is no theoretical limit in terms of minimum simulation time step, hence it should be possible to run a simulation with smaller time steps, for example 0.1 us. It is more a matter of the underlying model if it has been correctly configured, while changing the simulation time step will result in a slightly different model behavior. This is much more pronounced when increasing the time step since most dynamics of causal systems tend to happen in the lower frequency range, while in the high frequency domain physical systems tend to naturally dampen their response while approaching infinity. As an important exception, when resonance frequencies already existing in the simulated network are in the high frequency domain, dynamic simulations with large time steps will not capture these high frequency dynamics and will give the impression of a stable/damped high frequency behavior. It is only when lowering the time step enough that the characteristic frequency can be represented in a time domain simulation that these dynamics are observed.