How to model three-phase inverters for grid-connected applications

by Ines Eberena

April 22, 2026

Key Takeaways

Model fidelity should follow the study question, time window, and waveform you need to trust.
Grid impedance, filter design, digital delay, and dc-link dynamics usually shape results more than model size alone.
Disturbance testing is the clearest way to verify inverter control logic before hardware work starts.

A credible three-phase inverter simulation starts with the study objective, not the switching block.

Renewable capacity additions reached almost 560 GW in 2023, and solar PV supplied about 75% of that total. That scale puts more three-phase inverters onto feeders, plant buses, and campus grids, so model quality now affects routine engineering work rather than niche studies. You will get better answers faster when model fidelity follows the grid question you need to resolve.

You are not choosing between a simple model and a detailed model in the abstract. You are choosing the minimum detail that still preserves the behaviour that matters at the point of common coupling, inside the control loops, and across the dc link. That stance keeps inverter simulation useful, readable, and easier to validate before you commit to hardware or protection settings.

“A three-phase inverter model is useful only when its detail matches the question you need answered.”

A useful three-phase inverter simulation matches the study objective

A three-phase inverter model is useful only when its detail matches the question you need answered. Grid current control, filter tuning, fault response, and feeder studies do not need the same inverter simulation, and the wrong level of detail will either waste runtime or hide the failure you need to see.

Use a switching model when PWM ripple or dead time matters.
Use an average model when grid trends matter more than ripple.
Keep the filter explicit when you care about PCC current quality.
Keep the grid source explicit when feeder strength affects stability.
Keep digital delays explicit when control tuning feels too easy.

A 500 kW solar inverter tied to a short industrial feeder gives a clear example. If you need to verify current ripple, semiconductor gating logic, or desaturation of the current loop, a switching model is the right tool. If you need to see feeder voltage response during a 10 s irradiance drop, an average model will answer faster and with less numerical burden.

You will get more value from your inverter simulator once you write the study question as a measurable output. That usually means naming the waveform, event, and time window before you place any block. A model built that way stays focused, and it is much easier to validate when results start to look suspicious.

Switching models fit control validation with waveform detail

Switching models are the right choice when the study depends on instantaneous phase voltage, PWM ripple, dead time, sampling effects, or semiconductor commutation timing. They preserve the behaviour that average models smooth out, so they are the safest option for validating current controllers, protection logic, and filter resonance near the switching band.

A 50 kW inverter with a 10 kHz carrier and an LCL filter shows why this matters. Once you inject one grid voltage sag and inspect phase current at the point of common coupling, you can see ripple growth, saturation of the current regulator, and asymmetry from dead time. Those effects shape harmonic content and controller stress, yet they disappear if the bridge is replaced with a controlled voltage source.

You pay for that fidelity with smaller time steps and longer runs. That cost is worth it when you are testing logic transitions, overcurrent handling, or the link between modulation index and phase current. It is not worth it for a 30 s feeder disturbance where switching ripple contributes very little to the engineering answer you need.

Average models fit system studies with longer time spans

Average models are the right choice when you need correct power exchange, current loop response, dc-link energy balance, and grid interaction over longer windows. They remove switching detail and keep the dynamics that matter for system studies, which makes them far more practical for long disturbances, parameter sweeps, and feeder-level work.

Utility planning needs that efficiency because study scope keeps growing. Solar and battery storage were expected to account for 81% of new U.S. utility-scale generating capacity added in 2024. A feeder with several inverter-based resources can’t be studied effectively if every bridge is resolved at the carrier level for every scenario.

An average model is still only good when its control paths stay honest. You still need the current controller, phase-locked loop, dc-link dynamics, and current limits. If you collapse those into an ideal power source, the model becomes easy to run but hard to trust. That is where many grid studies drift away from physical behaviour, even though the waveforms look clean.

Study question	Model choice that usually fits	What must stay explicit
You need phase current ripple and harmonic content at the point of common coupling.	A switching model will preserve carrier effects and timing detail.	The bridge, PWM method, dead time, and LCL filter should remain explicit.
You need current loop tuning during grid voltage sags or step commands.	A switching model will show how sampling and saturation alter the response.	Controller delays, limits, and measurement filtering should remain explicit.
You need feeder voltage and power flow over several seconds.	An average model will run faster while preserving useful inverter dynamics.	The current controller, phase-locked loop, and dc-link energy balance should remain explicit.
You need many parameter sweeps across line impedance or plant dispatch points.	An average model will support broader scenario coverage within practical runtime.	Grid impedance, current limits, and plant setpoints should remain explicit.
You need to validate protection trips caused by modulation or gating behaviour.	A switching model will expose events hidden by averaged voltage sources.	Bridge states, thresholds, and fault logic should remain explicit.

LCL filter values set current quality at the PCC

LCL filter values determine how much switching ripple reaches the grid and where resonance appears, so they directly shape current quality at the point of common coupling. A credible model must include inverter-side inductance, grid-side inductance, filter capacitance, and damping, because each term changes the closed-loop response.

A 400 V converter tied to a 50 Hz bus makes the tradeoff obvious. If the filter capacitor is oversized, reactive current rises and the controller works harder near nominal operation. If grid-side inductance is too small, switching ripple leaks into the feeder. If damping is ignored, a neat sinusoid in simulation can turn into oscillatory current once the controller excites the resonant mode.

You should place the resonance high enough to separate it from the control bandwidth and low enough to avoid poor attenuation near the carrier. That balance matters more than any single textbook ratio. Good inverter simulation keeps filter losses and damping visible, because current quality problems are often filter problems wearing a control-system disguise.

Grid impedance assumptions set stability margins in simulation

Grid impedance sets the inverter’s effective operating condition, so a model with an ideal stiff source will overstate stability margin on weak feeders. Accurate studies need the source Thevenin equivalent, feeder impedance, transformer leakage, and local capacitance, because each part shifts resonance, controller gain, and phase margin.

A campus microgrid and a rural feeder will not stress the same inverter in the same way. The campus case might look stiff enough that a wide current-loop bandwidth seems harmless. The rural feeder can add enough inductive impedance that the same tuning produces oscillation near the phase-locked loop bandwidth. A simple impedance sweep often reveals the problem faster than another round of controller retuning.

SPS SOFTWARE fits this step well because you can inspect source, line, transformer, and control assumptions directly instead of accepting a sealed-inverter simulator. That transparency matters when results shift after one feeder parameter changes. You’re then checking physics and implementation at the same time, which is exactly where many grid-tied models fail quietly.

Control bandwidth must respect digital timing limits

Control bandwidth must be set with sampling, computation, and PWM update delays included, because digital timing removes phase margin that continuous-time tuning will hide. A model that ignores those delays will look stable on paper and then ring, overshoot, or saturate once discrete control is placed in the loop.

A common mistake appears in a current controller tuned near one-tenth of the switching frequency. The gain margin can still look comfortable until you add one sample of current measurement delay and one sample of modulation delay. That same tuning then produces noisy current, poor disturbance rejection, and a phase-locked loop that interacts badly during voltage dips.

You should model the controller exactly as it will execute, with sampling order, zero-order hold, filtering, and limit handling all included. That does not make the model harder to understand. It makes the result honest. Once those delays are visible, you’ll usually lower the target bandwidth a little and gain far better behaviour across weak-grid conditions.

Solar input models must reflect DC link behaviour

Solar input models must capture the DC-link behaviour because the inverter does not see irradiance directly. It sees source impedance, power limits, control action from maximum power point tracking, and capacitor energy. A fixed DC source can support rough control checks, but it will miss voltage sag, current limiting, and recovery behaviour during solar transients.

A grid-tied PV system during a fast cloud edge is a good test case. Panel power drops, the dc-link capacitor supplies the deficit for a short interval, and the inverter controller adjusts modulation to keep ac current within limits. If your model uses an ideal stiff dc source, none of that energy exchange appears, so the current controller looks calmer than it really is.

You do not need a full cell-level solar model for every study. You do need enough source dynamics to preserve dc-link excursions during the events you care about. That usually means a controlled dc source with realistic source resistance, power limits, capacitor value, and tracking dynamics. Once those are present, grid integration studies stop masking power-balance errors.

“Disturbance tests are the fastest way to prove that a three-phase inverter model is trustworthy.”

Disturbance tests reveal model errors before hardware work

Disturbance tests are the fastest way to prove that a three-phase inverter model is trustworthy. One model that survives step changes, voltage sags, phase jumps, current limits, and impedance variation will tell you far more than a dozen steady-state plots, because weak assumptions usually fail when the system is forced away from nominal operation.

A disciplined test set might start with a current reference step, then move to a 20% voltage sag, then repeat the same event with higher feeder impedance and a lower dc-link voltage. Those cases expose hidden couplings between the phase-locked loop, current regulator, and filter. When a model passes only under ideal grid strength, you are looking at a model that is still unfinished.

SPS SOFTWARE is most useful here when every block stays open to inspection, because good engineering judgment depends on assumptions you can trace and revise. Over time, the strongest grid-connected models are not the ones with the most detail. They are the ones tested against the right disturbances until their limits are clear and their behaviour stays consistent.

Power Electronics|Power Systems

Thermal modeling for power electronics and why switching losses matter

by Ines Eberena

April 20, 2026

Key Takeaways

Switching losses come from voltage and current overlap during finite transitions, and high frequency turns small event energies into significant heat.
Datasheet energies, thermal impedance, and junction temperature feedback belong in the same model if you want reliable converter thermal results.
Gate resistance, layout parasitics, and transient thermal swings often set the safe operating limit before heatsink size does.

Switching losses decide junction temperature sooner than most heatsink calculations admit.

A field failure survey summarized in the IEEE reliability literature found that power semiconductor devices accounted for 31% of reported failures in power electronic systems. That matters because thermal stress is rarely created by conduction loss alone in modern converters. Once your switching frequency climbs, each turn on and turn off event adds a small burst of energy that turns straight into heat. If you only size copper, silicon area, and heatsinks around average current, you’ll miss the part of the loss budget that often sets the safe operating limit.

“That overlap creates energy loss in every cycle.”

Switching losses decide junction temperature sooner than most heatsink calculations admit.

Switching loss starts during finite voltage current overlap

Switching loss begins when drain to source voltage and drain current exist at the same time during turn on and turn off. A MOSFET is not an ideal switch that jumps from fully blocking to fully conducting. Gate charge, parasitic capacitances, and circuit inductance stretch the transition. That overlap creates energy loss in every cycle.

A hard switched half bridge makes this easy to picture. During turn on, the current rises while the device still supports much of the bus voltage. During turn off, the current is still flowing while voltage climbs again. The product of voltage and current during those short intervals creates switching losses in MOSFET devices, even if the on state resistance is low and the conduction interval looks efficient.

You can’t treat those intervals as rounding errors once frequency rises. A converter running at 20 kHz may tolerate a rough estimate early in design, but a design at 100 kHz or 250 kHz will turn a few microjoules per edge into watts of heat. That’s why accurate thermal modelling starts with the overlap event, not with the heatsink.

A simple switching loss formula works only for screening

The common screening formula estimates switching power from the overlap triangle during turn on and turn off. You multiply bus voltage, load current, and transition time, then scale that event energy with switching frequency. It gives a quick first pass. It will not capture the full behaviour of an actual converter.

You’ll often see the estimate written as Psw ≈ 0.5 × V × I × (tr + tf) × fs. That form is useful when you’re comparing candidate devices for the same bus voltage and current. A 400 V converter switching 20 A with combined rise and fall time of 80 ns at 100 kHz produces a rough estimate near 32 W. That number is helpful for screening, but it hides reverse recovery, output capacitance loss, gate loop effects, and load current variation.

The formula also assumes linear transitions and constant current. Actual waveforms rarely behave that cleanly. Parasitic inductance can slow one edge and sharpen the other. A clamped inductive load will produce a different switching shape than a resonant leg. Use the simple formula to reject weak options early, then move to measured or simulated energy per event before you trust a thermal result.

Datasheet curves account for voltage current temperature dependence

Datasheet switching energy curves are more useful than the simple overlap formula because they include how the device behaves under tested voltage, current, gate resistance, and temperature conditions. Those curves convert switching losses in MOSFET parts from guesswork into a parameterized estimate. They still need correction for your exact circuit.

A typical datasheet gives turn on energy and turn off energy at a stated bus voltage, current, and gate resistance. If your converter runs at half the tested current, you can’t assume the energy will scale perfectly in half. The output capacitance discharge, reverse recovery of the companion diode, and Miller plateau behaviour distort that scaling. Junction temperature also matters because carrier mobility, threshold shift, and parasitic behaviour all change with heat.

When you read those plots, treat test conditions as part of the number. A curve measured at 25°C with a 10 Ω gate resistor will understate loss for a converter that actually runs near 100°C with a 22 Ω resistor. This is where you stop thinking about one MOSFET value and start thinking about a switching system.

Average power follows event energy times switching frequency

Average switching power comes from the sum of turn-on and turn-off energy per event multiplied by switching frequency. That relationship is the most reliable bridge between waveform detail and thermal design. Once you know event energy under your conditions, the thermal model has a meaningful heat source to solve.

The practical form is Psw = (Eon + Eoff) × fs. If one device dissipates 120 µJ at turn-on and 90 µJ at turn-off, a 100 kHz operating point gives 21 W of switching power. Double the frequency and that term doubles too, even when load current and duty ratio stay the same. That linear link is why high-frequency designs often become thermal problems before they become current problems.

The checkpoint below helps separate the inputs that deserve attention first when you calculate MOSFET switching losses for simulation and thermal sizing.

Input or check	What it tells you
Bus voltage under worst operating condition	The highest applied voltage will stretch the switching energy and usually sets the harder thermal case.
Load current at the instant of switching	The current during each edge matters more than average output current when you estimate event energy.
Turn on and turn off energy from matched test conditions	Using energies measured near your gate resistance and temperature avoids a large error in average power.
Switching frequency across the operating range	A modest increase in frequency raises switching power in direct proportion and often moves the thermal limit first.
Conduction loss calculated at hot resistance	Hot on state resistance keeps the total loss budget honest once switching heat has already raised junction temperature.
Dead time and diode recovery behaviour	These details often explain why measured loss is higher than a clean energy sum from a datasheet curve.

Electrothermal simulation links switching events to junction temperature

Electrothermal simulation turns electrical loss into junction temperature by coupling a loss model with a thermal network. That link matters because device temperature shifts the same parameters that created the loss. You’re solving a loop, not a one way calculation. A static estimate will miss that feedback.

A useful converter model starts with electrical waveforms or event energies, then feeds those losses into a thermal impedance path from junction to case, case to sink, and sink to ambient. The updated junction temperature then adjusts on state resistance, threshold behaviour, and switching energy for the next step. That is how you move from a spreadsheet number to a believable operating point. SPS SOFTWARE fits this workflow when you need transparent electrothermal blocks that you can inspect and adjust instead of accepting a hidden thermal assumption.

The value of this approach shows up when operating points shift. A converter that looks safe at nominal load may cross a thermal limit during light load high-frequency operation, where conduction loss falls but switching loss still stays high. Once you model that loop, you’ll see why thermal effects belong inside converter simulation rather than after it.

“You’re not only tracking the average hot spot. You’re tracking how far and how often the junction moves.”

Transient impedance shapes temperature rise more than steady averages

Transient thermal impedance tells you how quickly a device heats during pulsed loss, and that matters more than steady thermal resistance when switching power is uneven over time. Junction temperature follows pulses, bursts, and duty cycles with delay. Average dissipation alone will hide those peaks. Short overloads can still push silicon past a safe temperature.

A motor drive shows this clearly during acceleration. Current rises for a few hundred milliseconds, switching energy increases, and the junction responds much faster than the heatsink. The case may still look cool while the die has already reached a dangerous peak. A commonly used power cycling data set showed lifetime dropping from about 10 million cycles at a 60 K junction swing to about 1 million cycles at 100 K, which shows why transient temperature swing matters so much.

That is why thermal modelling improves power converter reliability. You’re not only tracking the average hot spot. You’re tracking how far and how often the junction moves. Packaging fatigue, solder stress, and bond wire wear respond to those swings, so transient impedance belongs in the model from the start.

Gate resistance tuning sets the first switching loss tradeoff

Gate resistance is often the first knob you turn because it directly alters switching speed, voltage overshoot, ringing, and electromagnetic noise. Lower resistance reduces overlap time and cuts switching loss. Higher resistance softens edges and can protect against overshoot. You won’t get the best result from either extreme.

A synchronous buck converter with a very small gate resistor will switch quickly and run cooler in the silicon, yet the drain waveform can overshoot enough to stress the device and raise noise. A much larger resistor will calm the edge, but transition time will lengthen and switching power will climb. The right value depends on package inductance, gate driver strength, and layout quality as much as the MOSFET itself.

Use a smaller gate resistor when overlap loss is the main thermal limit.
Use a larger gate resistor when overshoot or ringing threatens device margin.
Check turn on and turn off separately because the best values often differ.
Measure at hot conditions because edge speed shifts with junction temperature.
Retune after layout changes because parasitic inductance changes the result.

That tradeoff is why reducing switching losses in MOSFET-based converters is rarely a single part choice. Gate drive settings, loop inductance, and thermal margin all move as a group. You’ll get a better answer from measured waveforms and a coupled model than from a nominal resistor value copied from a reference design.

Heatsink sizing fails when switching loss is undercounted

A heatsink calculation fails when the loss number feeding it ignores switching energy, temperature feedback, or transient peaks. The sink can be perfectly sized for the wrong power input and still produce an overheated converter. Good thermal design starts with disciplined loss modelling, then uses the heatsink as the last step rather than the first guess.

A common failure path looks harmless on paper. You choose a low resistance device, estimate conduction loss at room temperature, and pick a sink that seems to hold the case comfortably below its limit. Bench tests then show the junction climbing during high-frequency operation because switching losses in MOSFET devices were understated. That missing heat raises junction temperature, which raises on-state resistance, which pushes total loss higher again. The error compounds rather than staying fixed.

SPS SOFTWARE is most useful at this stage when you want the electrical and thermal assumptions kept visible enough to challenge. That habit will give you better converter margins than any oversized heatsink alone. Careful modelling won’t remove tradeoffs, but it will show you which ones are worth paying for and which ones are just hidden loss.

Grid

How to set up a microgrid model from scratch using simulation software

by Ines Eberena

April 16, 2026

Key Takeaways

A useful microgrid simulation starts with a narrow study question that sets scope, fidelity, and outputs before any modelling begins.
Accurate component ratings, source definitions, and control roles matter more than model size when you build a first-pass microgrid simulator.
Steady state validation will decide if your disturbance results deserve trust, especially for islanded and grid-connected transitions.

The best microgrid simulation starts with a study question and a model scope you can defend.

Good results come from disciplined setup, not from piling every possible component into your microgrid simulator. Solar and battery storage account for 81% of planned United States utility scale generating capacity additions for 2024, which shows how much new power system work now centres on inverter-based assets that need careful control models. You will get farther, faster, when the model starts with a clear operating question, consistent ratings, and controls matched to the study. That approach gives beginners a workable path and gives experienced engineers a model they can trust.

“You should write one sentence that defines success before you model anything.”

Choose the study question before picking a microgrid simulator

Start with the study question. A microgrid simulator only helps when the model answers a specific operating problem such as voltage support, protection response, fuel use, or islanding stability. That choice sets the needed components, control detail, time step, and output signals before you place a single block.

A campus microgrid used for peak shaving needs a different setup than a remote mining site that must carry load after a utility outage. The first case will focus on dispatch logic, tariff windows, and the point of common coupling. The second will focus on source sharing, frequency control, and black start order. Both are microgrids, but the simulation work is not the same.

You should write one sentence that defines success before you model anything. A good version sounds like this: you need to verify that battery storage and one diesel unit will hold frequency inside limits after feeder separation. That sentence cuts out noise, keeps the model small, and tells you what outputs will matter when you review results.

Match model detail to the behaviour you need

Model detail should match the behaviour you need to see. Steady power sharing, fault current, converter switching, and resynchronization do not belong at the same fidelity level in one first pass model. A simpler model with the right states will give you better answers than a detailed model with the wrong focus.

If your goal is feeder loading and energy balance over an hour, average converter models will work well and will run quickly. If you need switching ripple, semiconductor stress, or fast current loop response, you will need a much smaller time step and more internal states. Many beginner projects stall because the model starts at the most detailed level before the basic control logic has even been checked.

Study focus	Model detail that usually fits
Daily energy scheduling across solar storage and diesel units	An average value model is usually enough because the main question is power balance over minutes or hours.
Voltage and frequency recovery after islanding	A dynamic control model with source governors or inverter loops is needed because the transient response sets stability.
Protection pickup and fault current contribution	A short-circuit-capable network model is needed because relay timing depends on current magnitude and source impedance.
Converter switching stress and waveform quality	A detailed electromagnetic transient model is needed because switching states affect current ripple and harmonics.
Resynchronization before reclosing to the utility	A control-focused model is needed because phase angle, slip, and breaker conditions matter more than internal device physics.

You do not need one perfect model that answers every question. You need the smallest credible model for the first question, then you refine only where the next study needs more detail. That sequencing keeps the work clear and stops the simulator from turning into a large drawing that explains very little.

Build the electrical network from rated component data

Build the network from rated data and a single base set. Feeder voltage, transformer ratios, source impedance, cable lengths, and load power must agree before any controller can behave sensibly. When these values line up, the first power flow check will expose wiring or unit errors early.

A clean starting network often includes one utility source, one feeder, one transformer, several aggregated loads, and each local source tied at the correct bus. A common beginner mistake shows up when a 480 V inverter is connected directly to a 13.8 kV feeder with only a nominal ratio entered somewhere else. The simulation will still run, but every current, voltage, and fault level will be misleading.

This is also where transparent modelling matters. SPS SOFTWARE fits well when you want to inspect each electrical parameter and see how buses, sources, and control ports connect before tuning begins. That kind of visibility helps you catch base mismatches early, which is far more useful than trying to explain odd plots later.

Represent distributed resources with the right control detail

Distributed energy resources should be modelled at the control layer that affects the study. A photovoltaic inverter used for ride through needs different internal detail than a diesel genset used only for dispatch and droop sharing. You will get cleaner results when each resource carries only the states that matter.

A battery unit usually needs a state of charge calculation, active power limits, reactive power control, and one clear operating mode. A diesel generator needs governor response, exciter action, and minimum loading logic. A photovoltaic source often needs irradiance input, dc link behaviour at the right abstraction, and voltage or power factor control. Lumping all three into generic controlled power sources strips away the behaviour that makes microgrids hard.

System planners added 14.3 GW of battery storage to the United States grid in 2024, which underlines why storage control assumptions now shape many distributed resource studies. That matters in practice because storage can switch from energy shifting to frequency support in seconds. If your control model cannot represent that role, the microgrid simulation will miss the asset that often keeps the system stable.

Define the grid connection at the point of common coupling

The grid connection should behave like a defined electrical source, not a vague infinite bus icon. Set short circuit strength, X/R ratio, nominal voltage, breaker logic, and export limits at the point of common coupling. Those settings decide how your microgrid will respond to faults, power swings, and reconnection checks.

A weak feeder and a stiff feeder produce very different voltage behaviour when a battery inverter ramps from 0 to rated power. The same difference appears when a motor load starts or when a fault clears near the site. If the point of common coupling is left as an ideal source with no meaningful impedance, you will hide the exact interactions that make grid-connected studies useful.

You should also define who controls active and reactive power while the utility is present. Some microgrids import a fixed amount and let local generation fill the rest. Others hold zero export or run a voltage schedule at the connection point. Those rules shape controller targets and stop confusion when you compare grid-connected results with islanded results later.

Set islanded control before simulating mode transitions

Islanded operation needs its own control design before you test any transfer event. Voltage and frequency support must shift from the utility side to local grid-forming sources, storage, or generator governors as soon as the breaker opens. If that hierarchy is missing, the simulator will report a crisis you actually created in the setup.

A small industrial microgrid offers a good example. While connected to the utility, a battery inverter can run in power control and simply track a dispatch setpoint. Once the tie breaker opens, that same unit must switch to voltage and frequency regulation, or a diesel unit must take that role immediately. If neither source is assigned that duty, the bus frequency will drift and loads will trip for reasons that have nothing to do with equipment ratings.

Transfer studies also need practical timing. Breaker open delay, controller mode change, load shedding thresholds, and resynchronization checks all matter more than a neat single step event. You’re testing a sequence, not a symbol change, so the model should reflect the sequence the plant will actually use.

Fix scaling errors before tuning any controller

Fix units, bases, and sign conventions before you tune controllers. Most unstable beginner models suffer from kilowatts entered as watts, line-to-line values used as phase values, reversed current polarity, or mismatched per-unit bases. A tuned controller will not repair arithmetic that is already wrong.

The easiest way to catch these issues is to run a short steady state case and inspect every source and load measurement before any disturbance is applied. A battery that appears to charge when your dispatch says discharge is a sign error. A current that looks three times too large often points to a line-to-line and phase voltage mix up. You can save hours if you stop here and correct the scaling first.

Check that every source rating uses the same apparent power base.
Confirm voltage entries use the same phase reference across the network.
Verify positive power flow points in one agreed direction.
Match controller limits to equipment ratings rather than default values.
Review initial conditions so storage and generators start from sensible states.

Controller tuning only has value after those checks pass. If you skip them, you will tune compensators around bad data and lock the mistake deeper into the model. That is why experienced engineers spend so much time on setup discipline before they touch gains.

“Microgrid models become useful when you treat them like test benches, build them in a disciplined order, and refuse to trust a plot that the steady state case has not earned.”

Validate power balance before trusting dynamic results

Trust dynamic results only after the microgrid balances power in steady state. If sources, storage, and loads do not settle to sensible active and reactive power values before a disturbance, every later waveform will mislead you. Validation starts with plain checks, and that discipline saves the most time.

A sound validation pass looks ordinary. You check total generation against total load plus losses, confirm transformer taps and bus voltages, review reactive power sharing, and make sure source current stays within ratings before the event starts. If a campus feeder shows a battery exporting reactive power with no control request, you stop and fix that issue before testing islanding or faults.

This is also where engineering judgment matters more than software confidence. SPS SOFTWARE supports clear, physics-based modelling, but the result still depends on your willingness to verify boring numbers before admiring dramatic waveforms. Microgrid models become useful when you treat them like test benches, build them in a disciplined order, and refuse to trust a plot that the steady state case has not earned.

Power Systems

8 Common mistakes engineers make when modeling power systems

by Ines Eberena

April 14, 2026

Key Takeaways

Wrong study scope and wrong model detail create errors long before solver output appears.
Base quantities, source data, load behaviour, and control limits shape result accuracy more than most teams expect.
Model trust comes from repeatable checks against known conditions, not from tidy plots or complex schematics.

Most wrong power system simulation results come from setup errors, not math errors.

Engineers trust a power system simulator when the model reflects the study question, the data, and the operating limits that shape system behaviour. Trouble starts when a convenient template replaces a verified network model or when a stable waveform hides a bad assumption. You’re usually not dealing with a software failure. You’re dealing with a model that answered a different question than the one you meant to ask.

The 8 mistakes that distort power system simulation results

A power system model loses accuracy when its structure, data, or numerical settings do not match the study objective. Each mistake below creates a specific kind of error, and each one can be checked early before you spend hours trusting results that won’t hold up.

“Engineers trust a power system simulator when the model reflects the study question, the data, and the operating limits that shape system behaviour.”

1. Using a study model that does not match the question

A model must match the time scale and physics of the question you’re asking. A steady-state load flow will show bus voltages and line loading, but it won’t tell you how a relay timer responds or how converter current peaks in the first milliseconds of a fault. A common miss appears when an averaged inverter model is used to judge sub-cycle current stress during a breaker operation. That result will look clean, yet it hides the switching and control detail that actually matters. If the study scope is vague, the model becomes a compromise and your answers lose value.

2. Mixing per unit bases across the network model

Per unit errors quietly distort almost every calculated quantity in a network study. Trouble often starts around transformers, where engineers carry a 100 MVA base through one section and a different base through another without converting impedances. A 13.8 kV to 69 kV transformer is a common place for this slip, because the voltage base shifts and the impedance looks reasonable even when it is not. The model still runs, which makes the mistake easy to miss. Short-circuit levels, voltage drops, and machine currents then look believable while every downstream result is biased.

3. Reusing default load models without checking behaviour

Default load blocks are useful for setup speed, but they often hide the wrong electrical behaviour. A constant power load can be acceptable for a planning snapshot, yet it will misrepresent voltage recovery if the actual site has induction motors, heating loads, or mixed feeder demand. A motor-heavy industrial bus will pull current very differently after a sag than a static constant power block suggests. That difference affects fault recovery, motor stalling, and protection pickup. If you don’t check how the load model reacts to voltage and frequency changes, the study will tell a neat story about a system that doesn’t exist.

4. Estimating source strength without verified grid data

Source strength shapes fault current, voltage stiffness, and control interaction, so guessed values will corrupt the whole model. Engineers often plug in a short-circuit level from memory or reuse data from a nearby substation and assume the upstream grid is close enough. A weak connection point for a wind plant, for instance, will behave very differently from a strong urban feeder with the same nominal voltage. Converter stability, flicker response, and fault current all shift when the Thevenin equivalent is wrong. If you haven’t verified source impedance and X/R ratio, you haven’t verified the study.

5. Picking a solver step that misses fast events

Numerical settings matter as much as network data when the study includes fast transients. A solver step that works for a slow voltage profile won’t capture capacitor energization, converter commutation, or a breaker restrike. You’re likely to miss the very spike or oscillation you set out to inspect if the time step smooths it away. That problem shows up when current peaks look modest and switching waveforms appear unusually clean. The model is not calm in that case. The solver is simply averaging out behaviour that occurs between samples, and your protection or insulation assessment will be wrong.

6. Starting dynamic studies from an invalid operating point

Dynamic results are only credible when the starting point is physically consistent. A common error appears when generator dispatch, tap positions, or control references are entered manually and the model begins from a state that could never exist in normal operation. A synchronous machine might start with an exciter output beyond its limit or with terminal voltage that doesn’t match the solved network condition. Once the disturbance is applied, you can’t tell which oscillation came from the event and which came from the bad initialization. The waveform looks busy, but it reflects startup correction rather than system response.

7. Leaving control limits outside the simulation model

Control systems need their limits inside the model or the results will overstate stability and recovery. Engineers sometimes model the main controller and skip current clamps, saturation, deadbands, rate limits, or protection interlocks because the core loop seems more important. A grid-forming inverter, for example, will appear heroic during a voltage dip if its current ceiling is missing. The same happens with exciters and governors when minimum and maximum outputs are left out. The controller then produces elegant responses that no physical device can sustain. If a control action looks perfect, check the limits first because something important often isn’t there.

8. Trusting results before any independent model check

A model should earn trust through simple checks before it is used for deeper studies. Engineers skip this step when the one-line diagram is complete and the waveforms look tidy, but appearance is a poor test. A feeder model should reproduce known voltages, losses, and fault levels before you use it for contingency work. A transparent workflow matters here, and SPS SOFTWARE is useful in that context because you can inspect assumptions, parameters, and equations instead of treating the power system simulator as a sealed box. If the base case fails a basic check, every later scenario will carry the same error.

“If the base case fails a basic check, every later scenario will carry the same error.”

Model issue	What the result is really telling you
1. Using a study model that does not match the question	The output reflects the wrong time scale or device detail, so the answer does not fit the study goal.
2. Mixing per unit bases across the network model	Reasonable-looking values can still be wrong when base conversions are inconsistent across voltage levels.
3. Reusing default load models without checking behaviour	Static defaults can hide how actual site loads react during sags, recovery, and frequency shifts.
4. Estimating source strength without verified grid data	Guessed grid impedance shifts fault current and voltage stiffness enough to distort the whole study.
5. Picking a solver step that misses fast events	Clean plots can come from numerical smoothing rather than from a physically quiet system response.
6. Starting dynamic studies from an invalid operating point	Early oscillations often come from bad initialization rather than from the event you intended to test.
7. Leaving control limits outside the simulation model	Controllers look stronger than they are when current, voltage, and rate limits are missing.
8. Trusting results before any independent model check	Base-case checks catch bad assumptions long before scenario studies make them harder to spot.

How to check model credibility before you trust results

A credible model reproduces known operating conditions, respects device limits, and gives stable answers under simple cross-checks. You should be able to explain every major assumption in plain language. If you can’t trace a result back to verified data and model structure, more detail won’t rescue it.

Match the model type to the study time scale.
Recheck every base quantity across transformers.
Compare load response against site knowledge.
Validate source impedance with utility data.
Confirm the base case before any disturbance study.

That review habit is what separates a useful engineering model from a polished diagram. Teams that keep assumptions visible, test simple cases first, and question clean-looking waveforms will catch more errors before they become report material. SPS SOFTWARE fits that practice when you need open, physics-based models that you can inspect and revise with care. Good modelling isn’t about making the power system simulator look busy. It’s about making every result stand up to scrutiny.

Industry Application

A practical guide to load flow analysis for distribution networks

by Ines Eberena

April 10, 2026

Key Takeaways

Load flow analysis is most useful when feeder data, device states, and study assumptions are checked before solver choice becomes the main focus.
Radial distribution feeders usually need methods and models that reflect high resistance, phase imbalance, and local voltage control rather than transmission habits.
Voltage results only become actionable when you read them beside branch loading, losses, and operating scenarios such as light load and reverse power flow.

Disciplined load flow analysis will show where a distribution feeder will hit voltage and loading limits before field changes create trouble.

Load flow analysis in power systems works best when you treat it as a feeder modelling task first and a solver task second. Average electricity transmission and distribution losses in the United States stayed near 5% of electricity transmitted from 2017 through 2021, which shows how much value sits inside ordinary network studies. You’re looking for a dependable steady-state picture of voltage, current, and losses under a specific operating snapshot. If the network data is clean and the study sequence is repeatable, the results will hold up under engineering review.

Load flow analysis estimates steady-state voltages across networks

Load flow analysis calculates the steady-state electrical condition of a network. It estimates bus voltages, branch currents, source injections, and losses. It assumes transients have settled and system frequency is fixed. That makes it the starting study for feeder planning, switching review, and normal operating checks.

A simple 13.8 kV feeder case shows the point clearly. You set a source bus, add line impedances, place loads at buses, and define any capacitor banks or distributed generation. The solver then reports voltage magnitude at each node and current on each line section. You can immediately see if the far end of the feeder sits at 0.94 per unit while the substation remains close to nominal.

This is why load flow analysis sits near the front of most study sequences. Fault studies, protection checks, and hosting assessments all depend on a believable operating point. If the steady-state case is weak, later studies won’t carry much weight. You’re not asking the model to tell you everything. You’re asking it to describe one operating snapshot with enough accuracy to act on it.

Distribution networks need different power flow assumptions than transmission

Distribution feeders need a different modelling approach because their electrical characteristics are different. Resistance matters more, phase balance is often poor, and radial structure is common. Voltage control devices sit close to the load. Embedded generation also pushes power both away from and back toward the source.

A long rural feeder with single-phase laterals will not behave like a high-voltage transmission corridor. Voltage drop on a high resistance line section can dominate the result, and unequal single-phase loading can pull one phase far lower than the others. Small-scale solar photovoltaic systems produced about 73 billion kWh of electricity in the United States in 2023, which is enough feeder-level generation to make midday reverse power flow a normal study case instead of a special case.

That shift matters because transmission-style simplifications can hide the very issues you need to find. Balanced models will miss single-phase voltage sag. Low resistance assumptions will distort losses and voltage drop. If you’re studying radial distribution feeders, you need solver settings and network representations that match feeder physics rather than transmission habits.

Start with a feeder model before choosing any solver

A good feeder model matters more than solver brand or solver speed. The network topology, phase labels, impedance data, and operating states must match the case you want to study. Load allocation also needs to reflect how the feeder is actually used. If those inputs are weak, the result won’t be worth much.

Confirm the feeder topology matches the current switching state.
Match each line section to the correct phase set and impedance.
Place loads at the right buses with consistent kW and kVAr values.
Set regulator taps and capacitor states for the study case.
Add distributed generation with its control mode and operating point.

A feeder with missing open points will produce currents along paths that don’t exist in service. A regulator left at the wrong tap will shift every downstream voltage and make you chase a false problem. Load placement creates the same risk. If a 500 kW commercial load is lumped at the substation instead of its lateral, your losses and end-of-line voltages will both be wrong.

You’ll get better results from a modest solver fed with careful data than from an advanced solver fed with old records. That’s why utilities usually spend more time cleaning models than running the final case. The solver can only process the feeder you give it. It can’t repair missing phase information or guessed control settings.

A stepwise workflow keeps power flow studies repeatable

A repeatable workflow keeps load flow studies consistent across engineers and study dates. Start with a validated base case. Adjust one operating condition at a time. Record the assumptions that changed. Then compare results against field expectations before the case is filed or shared.

A practical sequence starts with the normal feeder state at peak load. You check source voltage, confirm regulator settings, and run the case. Next, you test light load, capacitor switching states, and distributed generation output levels. A final pass checks that losses, voltage profile, and branch loading look physically believable. This routine keeps small modelling errors from hiding inside a large batch of cases.

Study checkpoint	What it confirms before you trust the result
Source bus and base values	The feeder voltage base and slack source match utility records so every per unit value has clear meaning.
Topology and phase labels	Open points, lateral phases, and missing switches are corrected before current paths are calculated.
Load allocation	Spot loads and distributed load are placed where field data says they belong so losses and voltage drop stay believable.
Voltage control settings	Regulator taps and capacitor states reflect the operating case instead of a stale saved condition.
Output review	Low voltage buses, thermal overloads, and unusual reverse power are checked before the study is accepted.

Forward-backward sweep suits most radial feeder studies

Forward-backward sweep is usually the most practical load flow method for radial distribution feeders. It works with the source-to-load structure of a feeder and handles higher resistance values well. It also fits unbalanced three-phase feeder models. That combination makes it dependable for everyday utility studies.

A 200-node radial feeder with several laterals is a good fit. The backward pass sums load current from the end nodes toward the source. The forward pass updates bus voltages from the source toward each downstream node. Forward-backward sweep works well because radial feeders have a clear source-to-load order. You’ll usually see steady convergence without forcing transmission-oriented assumptions into the case.

Closed loops and heavily controlled networks need more care. A weakly meshed urban system can require compensation techniques or a full three-phase solver that handles loop currents directly. Newton-based methods still have value, especially when the network is meshed or when controls interact strongly. The right question is not which method sounds more advanced. The right question is which method matches the feeder structure you’re modelling.

“Forward backward sweep works well because radial feeders have a clear source-to-load order.”

Voltage results show where feeder limits are being reached

Voltage results tell you where a feeder is close to service limits and where control equipment is already working too hard. The lowest bus voltage is only part of the picture. Phase imbalance, regulator position, and reverse power also matter. Good interpretation focuses on the pattern, not a single number.

A suburban feeder with rooftop solar can look healthy at the substation and still carry overvoltage risk at the far end near noon. Later that day, the same feeder can show low voltage on one phase when vehicle charging and air conditioning rise together. Those two operating points call for different fixes. One case may need regulator deadband review, while the other may point to conductor upgrade or load transfer.

You should also read voltage results beside current and loss results. A feeder that stays inside voltage limits can still run too hot on one branch. Another feeder can show acceptable current loading while one single-phase lateral drops below service targets. You’re looking for the location, operating condition, and control response that line up as one coherent story.

Software choice should match the study scope

Software choice should follow the scope of the study you need to complete. A simple teaching case needs clarity and transparency. A utility planning case needs detailed three-phase modelling and repeatable scenario control. Large study sets also need clean case management. The right tool is the one that supports the feeder detail you must preserve.

A spreadsheet or small script can work for a short radial feeder with balanced loading and one study condition. That same setup will struggle once you add phase-specific loads, regulator logic, switched capacitors, and embedded generation. Utility engineers usually need a platform that keeps every device visible and editable. SPS SOFTWARE fits teams that want transparent, physics-based feeder models they can inspect, adjust, and reuse without hiding assumptions.

You should test software against the cases that matter most to your work. A teaching lab often needs readable models that students can follow line by line. A planning group needs study templates and consistent data import. A research team needs model access for custom controls and altered component equations. Software becomes useful when it preserves the network detail your study depends on.

Weak assumptions cause most distribution load flow mistakes

Most bad distribution studies fail long before a solver misses convergence. They fail when feeder maps are stale, load allocation is guessed, or regulator settings are copied from old files. You can’t repair weak assumptions with a stronger algorithm. Careful inputs and honest validation will decide how useful the result is.

“You can’t repair weak assumptions with a stronger algorithm.”

A common mistake appears when engineers trust a solved case because every bus has a number beside it. Convergence only means the mathematics settled. It does not mean the feeder matches service conditions. Another mistake comes from checking only one operating point. Peak winter load, light summer load, and midday solar export can produce three very different voltage profiles on the same feeder.

Good load flow analysis builds confidence through disciplined modelling, repeatable cases, and plain engineering judgment. That is where teams get lasting value from tools such as SPS SOFTWARE, especially when assumptions remain visible and open to review. You’ll make better calls when the model shows its logic clearly. The study then becomes a dependable basis for feeder planning instead of a file that only the original author trusts.

Power Systems

How EMT and RMS modelling serve different power system studies

by Ines Eberena

April 7, 2026

Key Takeaways

EMT and RMS serve different study purposes because they solve different physics at different time scales.
Protection detail, converter controls, and subcycle effects are strong signals that EMT is the better fit.
Model quality depends as much on validated parameters and scope control as it does on simulation detail.

Choose EMT when the study depends on waveform detail, and choose RMS when the study depends on slower electromechanical behaviour.

That split matters more now because converter-based generation keeps adding fast controls to systems that were once dominated by synchronous machines. Wind and solar supplied 13.9% of global electricity in 2023, which means more studies now sit closer to inverter controls, fault response, and switching effects. You will get better answers when your model matches the physics that decides the result. You will get misleading confidence when it does not.

“An electromagnetic transient simulation is built for events where the waveform shape changes the outcome.”

EMT tracks waveforms while RMS tracks phasor behaviour

EMT and RMS differ mainly in what they solve and what they ignore. EMT follows instantaneous voltages and currents at very small time steps. RMS replaces fast waveforms with phasors and averaged quantities. You get waveform fidelity from EMT and study speed from RMS.

A feeder fault illustrates the split clearly. EMT will show the exact fault inception angle, the dc offset in current, and the way a breaker or converter responds over microseconds and milliseconds. RMS will show the same event as a balanced or unbalanced phasor disturbance with a much smoother response. That is often enough when you care about voltage recovery, power flow redistribution, or rotor angle movement.

The important point is not model sophistication. It is model relevance. Electromagnetic transient simulation is built for events where the waveform shape changes the outcome. RMS modelling is built for cases where the averaged sinusoidal state carries the answer. If your result depends on what happens within a cycle, the phasor abstraction will hide too much.

RMS models fit stability studies with slower dynamics

RMS models are the right fit when the study question sits on a slower time scale than the power frequency waveform. They capture electromechanical swings, voltage regulation, and frequency response efficiently. They also support large networks and many contingencies without excessive run time. That makes them a practical choice for stability work.

A generator trip study shows why. You usually want to know how frequency dips, how governors respond, how automatic voltage regulators support voltage, and whether rotor angles remain bounded. None of those answers depends on individual switching pulses or travelling wave effects. An RMS model lets you screen many disturbances across a transmission network and compare credible operating cases quickly.

You should still be disciplined about model scope. RMS will not rescue a poor representation of controls, load recovery, or protection logic. It simply gives you a strong fit for slower behaviour. When the pass or fail measure is damping, settling, frequency nadir, or post-fault voltage recovery, RMS will usually give the answer you need with less modelling burden.

EMT models fit studies with subcycle switching behaviour

EMT models fit studies where subcycle detail decides the result. They resolve switching events, fast control loops, saturation effects, and non-sinusoidal waveforms directly. That makes them the correct tool for converter commutation, transformer inrush, and many detailed fault studies. RMS models smooth away those mechanisms.

A transformer energization case is a simple illustration. The inrush current peak depends on residual flux, point-on-wave closing, and core saturation, all of which unfold within fractions of a cycle. An RMS model can approximate the event, but it will not reproduce the actual waveform that a relay, filter, or converter controller sees. The same limit appears with pulse-width-modulated converters and dc-link control interactions.

EMT is not just about getting a prettier waveform. It is about representing the mechanism that causes a trip, an overvoltage, or a control instability. If that mechanism lives inside the cycle, your model needs to live there too. That is why electromagnetic transients matter most when switching detail and non-linear effects are part of the study question.

Study time scale should set your model choice

Time scale is the quickest and most reliable screen for model choice. A study dominated by seconds and electromechanical motion belongs in RMS. A study dominated by microseconds, milliseconds, or point-on-wave effects belongs in EMT. Mixed cases need you to decide which time band actually determines the pass or fail outcome.

Protection and control sequences often look mixed at first glance. A fault may start in microseconds, provoke relay logic over milliseconds, and reshape system frequency over several seconds. Your model choice should follow the decision point, not the event duration. If you only need to know system recovery after a cleared fault, RMS is enough. If you need to know why the relay operated late or why the converter blocked, EMT is the safer choice.

That is also where transparent workflows matter. SPS SOFTWARE gives you a way to keep models inspectable and editable, so you can choose detail level with intent instead of treating the simulator as a black box. Teams work faster when they can see which equations and assumptions are carrying the answer.

Study focus	What the model choice usually means
A frequency dip after a generator trip is mainly a slower system response.	RMS usually fits because waveform shape does not decide the result.
A converter control issue appears within a few milliseconds after a fault.	EMT usually fits because fast control interaction is hidden in phasor form.
A relay operation depends on fault inception angle or transient distortion.	EMT gives the quantities the relay will actually see during the event.
A planning team must screen many contingencies across a large network.	RMS gives broader coverage because the models run faster and scale better.
A weak grid study depends on inverter current limits and controller timing.	EMT is usually the safer choice because the deciding physics is too fast for RMS averaging.

Protection studies often need detail beyond RMS models

Protection studies often need more detail than RMS can provide because relays respond to quantities that change within a cycle. Fault inception angle, current dc offset, current transformer saturation, and voltage transformer transients can alter what the relay measures. EMT will represent those effects directly. RMS will often smooth them into a cleaner event than the relay actually sees.

A distance relay on a long line is a good case. The apparent impedance during the first cycles after a fault can shift because of cvt transients, fault resistance, and waveform distortion. A differential relay can also react badly when current transformer saturation distorts one side more than the other. Those are not minor details when your study asks why a trip happened or why it failed to happen.

RMS still has a place in protection work. It is useful for broad coordination checks, grading margins, and large fault sweeps where the relay measurement process itself is not under test. Once the study moves from settings review to relay behaviour under stress, EMT becomes much more than a refinement. It becomes the model class that matches the protection physics.

Systems with many converters push studies toward EMT

Systems with many converters push modelling toward EMT because converter controls react on time scales that phasor models often compress too aggressively. Grid-following controls, current limits, phase-locked loops, and dc-link dynamics can interact within milliseconds. Those interactions can decide stability, protection response, or equipment stress. RMS can miss them even when the wider network looks slow.

A weak-grid solar plant is a familiar example. Voltage dips, current limiting, and phase tracking can create behaviour that looks stable in an averaged RMS representation but becomes oscillatory or blocked in EMT. That matters more as converter penetration rises. Solar photovoltaic generation rose by 25% in 2023, so you will face more studies where inverter detail is part of the main question.

You do not need EMT for every converter case. A well-validated average-value representation can still support many planning studies. The warning sign appears when control limits, harmonics, dc coupling, or weak-grid interaction sit near the event you care about. Once those features are close to the boundary of acceptable performance, waveform-level modelling stops being optional.

Accuracy gains come with heavier model cost

EMT gives more physical detail, but it also asks for more data, more computation, and more care in model building. RMS asks less from you and often returns answers faster. The better choice is the one that captures the deciding mechanism with the least unnecessary burden. More detail will not help if the extra detail is poorly known.

A plant-level study can illustrate the tradeoff. An RMS network with validated machine and controller models might let you test dozens of contingencies in the time one EMT case takes to set up and run. That speed matters when you are screening operating points, seasonal conditions, or protection settings. EMT becomes costly when switching devices, control blocks, and non-linear elements all require careful parameterization.

False precision is the main risk. An EMT model with guessed controller gains or missing transformer saturation data can look authoritative while answering the wrong question. RMS has its own limits, but it often forces clearer simplification. You will make better choices when you treat model fidelity as a targeted tool rather than a badge of seriousness.

“False precision is the main risk.”

A practical screen for choosing EMT or RMS

You should choose the simplest model that still captures the physics that decides the result. RMS is the right mode when averaged quantities answer the study question. EMT is the right mode when switching, control interaction, fault inception, or relay measurement set the outcome. Clear model purpose will save time and avoid false confidence.

Use this screen before you build or refine a model:

Choose RMS when your pass or fail metric is frequency, rotor angle, or slower voltage recovery.
Choose EMT when the result depends on subcycle waveform shape or switching events.
Choose EMT when relay behaviour depends on saturation, distortion, or point-on-wave effects.
Choose RMS first when you need broad contingency screening across a large system.
Choose the model with the best validated parameters when both modes seem plausible.

That judgment gets better with practice, and it improves further when the models stay open enough for you to inspect the assumptions. SPS SOFTWARE fits that kind of work because clear, physics-based modelling helps teams explain results instead of just presenting them. Good studies come from disciplined scope, validated parameters, and the willingness to use less detail when less detail gives the right answer.

Electrical Engineering

Fault analysis methods every protection engineer should know

by Ines Eberena

April 3, 2026

Key Takeaways

Short circuit analysis works best when you choose the method from the protection question instead of starting with the fullest model available.
Three phase faults, sequence networks, and zone based case selection each answer different protection questions, so none of them should be treated as optional shortcuts.
Credible settings come from disciplined validation of data, models, and fault results against plant evidence.

Accurate short circuit analysis keeps relay settings credible and equipment duties honest.

Protection work goes wrong when engineers treat fault analysis in power systems as a one-step calculation instead of a checked chain of assumptions. U.S. electricity customers were without power for an average of 5.5 hours in 2022, which shows how much system performance matters when a fault is cleared poorly or studied badly. You need a method that fits the duty under review, the network detail you trust, and the relay function you’re checking. Short circuit analysis in power systems works best when you start with the protection question, then pick the simplest method that still captures the fault behaviour that matters.

Study scope determines the right short-circuit method

The right short-circuit method depends on what the study must prove. A breaker duty check needs maximum available current. A relay sensitivity check needs the weakest fault that still must trip. Scope comes first because one network can require different assumptions for each task.

A plant expansion shows the difference quickly. A new 15 kV motor bus can need one study for switchgear interrupting duty, another for feeder ground relay pickup, and a third for incident energy. You can’t use the same fault set for all three jobs and expect useful answers. The method is only right when its assumptions line up with the setting or rating you have to approve, so the first step in fault analysis is always defining the protection decision that rests on the result.

“Scope comes first because one network can require different assumptions for each task.”

Network reduction keeps hand calculations useful for first checks

Network reduction still has value because it gives you a fast truth check. A Thevenin equivalent at the fault point shows source strength. It also shows X/R ratio and likely fault level. You don’t need the full model to test first assumptions.

A feeder relay review often starts with the utility source, one transformer, one cable run, and the equivalent motor contribution behind the bus. That stripped network will tell you if expected fault current is closer to 2 kA or 20 kA, and that gap matters before you trust any detailed case file. A reduced model also shows when a result doesn’t make physical sense. Once the order of magnitude looks right, you can move to fuller models for protection coordination and equipment checks with much more confidence.

Three-phase faults set the upper bound for duty

Three-phase faults matter because they usually produce the highest current. They set the largest mechanical stress on equipment. They also set the main thermal limit for interruption. That makes them the standard starting point for breaker duty and bus checks.

A 27.6 kV industrial substation makes the point clearly. A fault placed at the main bus can show the strongest symmetrical current the source and motors can supply, while a ground fault on a remote feeder will often be much lower. The larger case governs breaker interrupting rating and bus bracing. Symmetrical fault analysis is simple compared with asymmetrical studies, yet it answers the first hardware question protection engineers face: can the equipment interrupt the strongest fault the system will deliver?

When you need this answer	Start with this method
A switchgear duty review needs the highest current a bus can see.	A balanced three phase bus fault gives the first current limit for interrupting checks.
A ground relay pickup review needs the weakest fault that still must trip.	A single line to ground study with sequence networks shows the zero sequence path that controls sensitivity.
A distance relay reach review needs apparent impedance along one protected line.	Fault cases placed at several points on that line show how source split alters the relay view.
A coordination review needs current over a practical range of source conditions.	RMS fault studies at minimum and maximum source strength show timing margins that survive operating changes.
A feeder with several converters needs current shape and control response.	An EMT model shows current limiting and first cycle effects that RMS tools smooth out.

Sequence networks remain essential for unbalanced fault studies

Sequence networks remain the clearest way to study unbalanced faults. They separate positive, negative, and zero sequence paths. That split shows why ground fault current rises or collapses for the case under study. Asymmetrical fault analysis becomes useful only when those paths are modelled correctly.

A grounded wye to delta transformer between a utility source and a plant feeder makes this visible. A single line to ground fault on the delta side won’t pass zero sequence current back to the source the same way a grounded wye to grounded wye bank will. Negative sequence current still matters for machine heating and phase unbalance, but zero sequence current will decide how ground elements behave. Engineers who skip sequence networks often end up with ground relays that look generous on paper and blind on the actual feeder.

Data quality errors usually outweigh calculation method errors

Bad data will distort fault results more than the difference between sound methods. Wrong transformer impedance shifts calculated current. Missing motor contribution can change minimum fault values. Protection settings sit on small margins, so data quality has to come first.

Protection system misoperations were reported at a 6.5% rate on the bulk power system in 2023, which is a reminder that settings and models still fail under routine operation. A common plant study error comes from using transformer nameplate impedance on the wrong MVA base, which distorts both maximum and minimum fault levels. Another comes from leaving out local motor contribution after a site expansion. Those errors deserve attention before you refine relay curves.

Source short circuit level and X/R ratio match the latest utility data.
Transformer impedance is converted to the study base correctly.
Grounding method is modelled at every source and transformer.
Motor and converter contribution is included where it matters.
Instrument transformer ratios match the relay inputs and settings.

RMS tools suit steady fault levels better than EMT

RMS tools are best for steady fault levels and most coordination work. EMT tools are better when wave shape and control action matter. The time scale of the protection question should pick the method. That keeps the model focused and the result usable.

A feeder with several converters shows the split clearly. An RMS study can estimate current magnitude seen by time overcurrent elements across many contingencies, which keeps coordination work efficient. An EMT study becomes important when inverter current limiting, control delays, or current reversal can affect protection logic during the first cycle. SPS SOFTWARE is useful in that stage because transparent models let you inspect the assumptions behind source impedance, converter limits, and relay inputs instead of treating the result as a sealed output. You’ll get better answers when you reserve EMT detail for cases where transient behaviour actually changes the protection outcome.

Protection checks should start from zone-based fault cases

Protection checks work best when fault cases follow protection zones. Each zone needs internal and external faults. Each zone also needs strong and weak source conditions. That structure ties short circuit analysis directly to what the relay has to judge.

A distance relay on a transmission line needs faults placed at several points on the protected line, with source strength varied at each end. A feeder overcurrent element needs near faults for speed and remote faults for sensitivity. Differential protection needs internal faults plus through faults that stress restraint and current transformer performance. When you organize cases by zone, gaps show up quickly, and you won’t mistake a complete bus fault report for a complete protection study.

“Matching study results to field evidence turns fault analysis into dependable protection practice.”

Settings are credible only after results match plant data

Settings become credible only when calculated faults agree with plant evidence over time. Relay event files should support the study. Commissioning tests should support it too. Matching study results to field evidence turns fault analysis into dependable protection practice.

A mismatch always means something needs attention. It’s often a grounding connection modelled incorrectly, a motor block omitted from the study, or a relay using different current transformer ratios than the file says. Engineers who keep closing that loop build settings that stay stable through outages, expansions, and audits. SPS SOFTWARE fits that discipline well because transparent models make it easier to trace a result back to the parameter or assumption that created it. Credible protection work comes from checked models, checked data, and checked results, repeated until the network and the relay tell the same story.

Electrical Engineering

Evaluating electrical simulation tools for teaching and engineering

by Ines Eberena

March 28, 2026

Key Takeaways

Define the study question first, then match tool fidelity and outputs to that goal so results stay explainable and defensible.
Choose EMT or RMS based on the time scales and physics you must capture, since the wrong modelling approach will produce confident-looking but wrong answers.
Prioritize transparent models, solver stability, and repeatable workflows over feature count so teams and students can rerun, review, and trust the same cases.

Pick your simulation tool by matching study goals to model fidelity, solver behaviour, and workflow fit.

“Tool selection goes wrong when you start with a feature checklist instead of the question you need answered, the time scales you must resolve, and the outputs you must trust.”

Teaching needs transparency so students can see why waveforms change, not just that they change. Engineering needs repeatable results that stay stable across parameter sweeps, model updates, and handoffs. A Nature survey reported 70% of researchers tried and failed to reproduce another scientist’s experiments, which is a reminder that repeatability is a technical requirement, not a nice-to-have.

A useful electrical simulation tools comparison treats accuracy, usability, and governance as a single package. You’re choosing assumptions, numerical methods, and model transparency, not just a user interface. You also need a plan for adoption in a teaching lab or an engineering team, since licensing, version control, and model review habits will shape results over time. The best power system simulation software is the one that makes your modelling assumptions visible and controllable, so you can explain results and defend them.

Start with study goals and required simulation fidelity

Your first evaluation step is writing down the study question, the events you must represent, and the outputs you will judge as correct. Fidelity is not “high” or “low”; it is a match between time scale and physics. If you cannot state what must be captured, you will overbuild models or miss key behaviours.

Start with three decisions you can document in a few lines: what phenomena matter, what you will ignore, and what error you can accept. Teaching and engineering differ most in what “good” means. A teaching lab often prioritizes clarity, inspectable component equations, and fast setup so students spend time learning, not wrestling with tool friction. Engineering work prioritizes traceability, model review, and stable runs across many cases, because a single unstable run can invalidate a whole set of conclusions.

A concrete way to lock this down is to define a “reference run” and a “stress run” before you install anything. A protection course might set a reference run as a 12.47 kV feeder fault with a grid-following inverter and a simple relay logic check, then use a stress run that tweaks fault resistance and inverter current limits to see if the results stay consistent. Once those two runs are written, every tool trial becomes measurable rather than impression-based.

Compare EMT and RMS approaches for power system modelling

The main difference between EMT and RMS simulation is what the solver treats as an electrical state versus an averaged approximation. EMT modelling resolves fast electromagnetic transients and switching effects with small time steps. RMS modelling focuses on slower electromechanical dynamics and phasor quantities, so it runs longer time horizons with less computational load.

EMT is the right lens when your question depends on waveform shape, fast controls, converter switching behaviour, protection interactions tied to instantaneous values, or harmonics. RMS is the right lens when your question depends on longer-duration voltage and frequency behaviour, stability margins, or operating-point changes where waveform detail does not change the answer. Neither approach is “better” in general, and both can produce misleading confidence if used outside their valid assumptions.

During tool evaluation, look past marketing terms and ask what the platform actually solves, how it initializes states, and what it assumes about network frequency and balance. A tool can offer both approaches, but you still need to check how models transition between time scales and what signals are available for verification. A practical selection habit is to decide EMT or RMS first, then shortlist tools that do that job cleanly, because forcing a tool into the wrong study type is a common source of wasted modelling time.

Check libraries for converters, protection, feeders, and control logic

Library coverage matters when it reduces custom modelling effort without hiding physics behind locked blocks. You want component models that match your study goals, expose parameters that affect behaviour, and provide enough documentation to review equations and assumptions. Library breadth also matters only if the models are consistent and easy to audit.

Converter-heavy grids raise the stakes for this check. A global electricity review reported renewables produced 30% of global electricity in 2023, which means many studies now depend on inverter controls, limits, and protection coordination rather than only synchronous machine dynamics. If the library models hide current limiting, phase-locked loop behaviour, or control saturation, you will get clean-looking plots that do not match field behaviour.

For teaching, model transparency is part of the curriculum. Students learn faster when they can inspect a control loop, change a filter value, and connect that change to waveform effects without guessing what a block does. For engineering, transparency supports peer review and reduces handoff risk between teams. You should also check how protection and control logic is represented, since the tool’s modelling style will shape how you validate timing, thresholds, and state transitions.

Assess solver settings, numerical stability, and reproducible results

“Solver quality shows up as stable runs, clear diagnostics, and repeatable results across small parameter changes.”

You should be able to control time step or tolerances, understand convergence limits, and reproduce a run from saved settings and model versions. If the platform cannot explain why a run failed, you will spend more time debugging than studying.

Numerical stability is not only a “solver problem”; it is a modelling discipline problem you need tool support for. Stiff networks, tight control loops, discontinuities, and ideal switches all push solvers into edge cases. Good platforms help you manage this with clear event handling, sensible defaults you can override, and warnings that point to the underlying cause. Reproducibility also includes governance basics: storing solver settings with the model, tracking library versions, and keeping run metadata so two engineers can confirm they ran the same case.

What you test during a trial	What good behaviour looks like	What breaks if you skip it
You run the same case twice with identical settings.	The results match within a stated tolerance and the tool records key settings.	You cannot tell tool variance from system behaviour changes.
You vary time step or tolerances across a small range.	Trends stay consistent and any differences are explainable and bounded.	Plots look plausible but depend on numerical artefacts.
You test initialization from a steady operating point.	Start-up transients are controlled and initial conditions are inspectable.	Early transient behaviour contaminates protection and control results.
You force a hard event like a fault or breaker action.	The solver reports events clearly and recovers without silent instability.	Hidden discontinuities create non-physical oscillations or solver failure.
You inspect diagnostics after a failed or slow run.	Error messages point to elements, time ranges, or limits you can adjust.	Debug time grows and model trust drops across the team.

Evaluate MATLAB Simulink links, collaboration, and lab deployment

Workflow fit is the difference between a tool that gets used and a tool that sits idle after procurement. You should check how the platform exchanges data with MATLAB and Simulink, how it supports parameter sweeps, and how it packages models for sharing. Lab deployment also needs predictable installs, licensing clarity, and version consistency across machines.

Integration checks should focus on what you will actually do day to day: import and export of parameters, scripted runs, and clean interfaces for controls work that lives outside the power network model. Collaboration checks should focus on model review and change tracking, since simulation credibility depends on being able to explain what changed and why results moved. Teaching labs add another constraint: students need to get running quickly with minimal configuration drift between workstations, or the course becomes an IT exercise.

SPS SOFTWARE is often evaluated in this step because teams want open, editable component models paired with a workflow that fits MATLAB and Simulink based control design. That practical combination matters when you need both transparency for learning and consistent execution for engineering studies. Tool trials should include a short “handoff test” where one person creates a case and another person reruns it from scratch using only the shared package, since that exposes hidden dependencies early.

Build a scoring rubric for electrical simulation tools comparison

A scoring rubric turns tool selection into a repeatable choice you can defend to a lab director or engineering manager. Start with a few non-negotiables tied to your study goals, then score the rest with weights that reflect how often you will use each capability. A good rubric also forces you to document tradeoffs instead of debating preferences.

Keep the rubric short enough that you will actually use it after the first meeting. These five categories cover most selection work without losing technical detail:

Study fidelity fit based on EMT or RMS needs
Model transparency and inspectable equations and parameters
Library coverage aligned to your network and control scope
Numerical robustness and reproducibility across reruns
Workflow and deployment fit for labs and teams

Judgment comes from how the scores behave under pressure, not from a perfect spreadsheet. If a tool wins only when you give it generous weights on minor features, it will fail you later when schedules tighten and you need dependable runs. When you apply this rubric consistently, SPS SOFTWARE tends to show its value where transparent modelling and reproducible execution matter most, which is the part of tool choice that determines long-term trust in results. The goal is not a tool with the longest feature list; it is a tool you can explain, rerun, and defend.

Electrical Engineering

Understanding EMT simulation for electrical system analysis

by Ines Eberena

March 24, 2026

Key Takeaways

Use EMT simulation when sub-cycle waveform detail sets equipment stress limits, and keep RMS studies for slower phasor questions.
Trustworthy EMT results depend on consistent time step, network detail, and solver choices, backed by convergence and initial-condition checks.
Run EMT studies against clear acceptance criteria, then keep the model as simple as possible while still answering that limit-focused question.

EMT simulation tells you what your system does between cycles.

A single cloud-to-ground lightning discharge can reach about 30,000 A, and that kind of impulse is measured in microseconds, not seconds. RMS studies can still be correct for many planning questions, but they will hide the stress that fast events place on insulation, breakers, converters, and protection logic. EMT gives you the instant-by-instant voltages and currents you need when “how high” and “how fast” matters.

The practical stance is simple: treat EMT as a precision instrument, not the default. You’ll get better outcomes when you pick EMT for questions that truly depend on waveform detail, and keep RMS modelling for questions that depend on slower phasor behaviour. That selection step is not academic, since model detail and simulation time rise quickly once you move into microsecond steps. Clear intent up front keeps EMT studies focused, credible, and easier to defend with technical leaders.

“Engineers reach for electromagnetic transient simulation when peaks, wave shape, and timing will set design limits.”

Define EMT simulation and the problems it is built for

EMT simulation is a time-domain method that solves instantaneous voltages and currents in an electrical network at small time steps. It keeps the full waveform instead of compressing it into a single RMS magnitude and phase. That lets you represent switching, saturation, arcing, and control actions as they occur. You use it when those details control equipment stress or system response.

Outputs typically look like sampled waveforms for each phase and conductor, so you can see steep dv/dt, high di/dt, and the exact moment a device changes state. Nonlinear elements such as transformers, surge arresters, and power electronic switches can be modelled with their physical equations instead of simplified steady-state equivalents. EMT also lets you capture unbalanced and zero-sequence effects without leaning on assumptions about sinusoidal behaviour. The trade is that you must manage many more state variables and much smaller numerical steps.

EMT problems are usually defined by “fast” physics. Travelling waves on lines, capacitor and reactor switching, converter gating, and fault inception angle all produce behaviour that does not average out cleanly over a cycle. That matters because protection and insulation coordination are often set by peaks, not averages. A good EMT study starts from an acceptance criterion, such as maximum overvoltage at a terminal or maximum current through a device. Once you name the limit you care about, the needed model detail becomes easier to justify.

Know when EMT is required and when RMS is enough

EMT is required when the decision you need to make depends on waveform shape, sub-cycle timing, or nonlinear switching behaviour. RMS modelling is enough when the question depends on slower electromechanical dynamics and balanced, near-sinusoidal assumptions hold. EMT also becomes the safer choice when protection logic depends on high-frequency content or DC offset. The goal is not to run EMT everywhere, but to use it where RMS will give you false confidence.

You need peak voltage or current, not just RMS magnitude.
You must represent converter switching, gating, or fast control loops.
You are studying breaker operation, prestrike, restrike, or fault inception angle.
You are assessing harmonics, subharmonics, or high-frequency resonance.
You need accurate behaviour for saturation, arcing, or nonlinear surge devices.

Power systems now include many more inverter-connected devices at the distribution and transmission edge, and those devices bring fast controls and switching artefacts into system studies. Solar accounted for 53% of new U.S. utility-scale generating capacity added in 2023, and a large share of that capacity connects through inverters that behave very differently from synchronous machines during transients. A disciplined workflow uses RMS studies to screen cases and narrow the study set, then uses EMT to verify the short list where waveform detail will change the engineering call. That sequencing also keeps compute and model QA effort in check.

How EMT modelling differs from RMS phasor-based studies

The main difference between EMT and RMS modelling is what gets preserved from the waveform. RMS studies solve phasors that represent a sinusoid over a cycle, so fast changes are averaged out. EMT solves instantaneous values, so switching, harmonics, and nonlinearities appear directly in the results. That makes EMT better for transient stress questions, while RMS stays efficient for slower system-level dynamics.

Study checkpoint	RMS phasor modelling	EMT time-domain modelling
What the state variables represent	Voltages and currents are represented as magnitudes and angles of sinusoids.	Voltages and currents are represented as instantaneous waveforms over time.
What time resolution means for results	Changes within a cycle are smoothed, so peaks and steep edges are lost.	Sub-cycle timing is explicit, so peaks and steep edges are visible.
How nonlinear device behaviour shows up	Nonlinearities are often linearized or represented with simplified equivalents.	Nonlinearities can be modelled directly, so saturation and clamping are captured.
How switching events are handled	Switching is often approximated as a change between steady states.	Switching is modelled at the instant it occurs, including transient ringing.
What questions the model answers best	Voltage stability, power flow sensitivity, and slower dynamics are answered efficiently.	Insulation stress, resonance risk, and protection response to fast events are answered directly.

RMS modelling can still include fault currents, relay elements, and control blocks, but it will always assume a smooth sinusoidal backbone for the electrical quantities. EMT breaks that assumption and forces you to pay attention to stray RLC, line representation, and converter switching detail. That extra effort is justified only when the decision hinges on what happens within a few milliseconds or less. Teams get the best value when they treat RMS and EMT as complementary, not competing, study types. Matching the method to the question keeps your results defensible.

“Careful execution will always matter more than the most sophisticated network you can draw.”

Key electrical transients EMT captures that RMS studies can miss

EMT captures transients where the waveform is distorted, asymmetric, or rich in high-frequency content. That includes capacitor bank energization, transformer inrush, fault inception with DC offset, and resonance triggered by switching. It also covers the interaction between converter controls and network impedance at frequencies far above the fundamental. RMS studies will often show the right trend but miss the peak stress and timing that sets equipment limits.

Waveform detail matters because many limits are instantaneous. Surge arresters clamp based on voltage, not RMS, and insulation coordination is based on peak overvoltage and front time. Protection elements that depend on high-frequency components, such as travelling-wave concepts or fast directional logic, also depend on signals that RMS models do not preserve. Converter current limiters and phase-locked loops respond to sub-cycle distortion, which can shift the system response even when RMS voltage looks acceptable. EMT gives you those signals directly, which removes guesswork when you’re validating a protection or equipment limit.

Scope control is still important. Not every harmonic or oscillation matters, and not every part of the network must be modelled at full detail to answer a focused question. The practical approach is to tie each transient type to one measurable outcome, such as arrester energy, breaker TRV stress, or relay pickup time. That keeps interpretation anchored in engineering criteria, not pretty waveforms. When the outcome is clear, you can trim the network to what materially shapes that outcome. EMT then becomes a tool for engineering judgement, not an exercise in complexity.

Choosing time step, network detail, and solver settings for EMT

Time step selection in EMT must be tied to the fastest phenomenon you need to resolve, not the nominal system frequency. Network detail must also match the transient type, since line modelling and stray capacitance can dominate high-frequency behaviour. Solver settings then become a stability and accuracy choice, especially when stiff nonlinearities are present. You will get credible results only when these three choices are consistent with each other.

Time steps that are too large will damp peaks and can shift the frequency of resonances, which looks like “better” behaviour while being numerically wrong. Excessively small time steps can also be a problem, since they can amplify noise and make parameter errors harder to spot. Line representation is a common inflection point: lumped models can be fine for some low-frequency events, while distributed or frequency-dependent models are needed when travelling waves or steep fronts matter. A practical check is to run a short sensitivity sweep on time step and key parasitics and confirm the result converges toward a stable waveform shape.

Model transparency helps when you’re tuning these choices. SPS SOFTWARE is often used in teaching and engineering teams because component equations and parameters are open to inspection, which makes it easier to see what each modelling assumption is doing to your results. That matters when a result changes after you refine a line model or adjust a switch representation, since you can trace the change back to model physics instead of treating it as a tool quirk. Solver choices still require judgement, especially for power electronics with discontinuous switching. Consistency checks, convergence testing, and parameter audits will do more for credibility than any single “recommended” setting.

Typical EMT study workflow from model setup to results

A typical EMT workflow starts with a single question tied to a limit, then builds only the model detail needed to answer it. You’ll define the switching or fault event, set initial conditions, and choose monitoring points that map to the limit. Then you’ll run a baseline, refine time step and network detail until results converge, and only then run variations. The workflow is repeatable when every run is linked to a named acceptance criterion.

A common transient study starts when a utility needs to energize a long distribution feeder with a large capacitor bank and an inverter-based plant connected near the end of the line. The EMT model is set up to close a breaker at controlled points on the voltage wave, then record the peak phase-to-ground voltage at the plant terminals and the current through the capacitor switch. A small set of runs varies breaker closing angle and source strength, since those two inputs drive the worst peaks. Results are accepted only when overvoltage stays under the equipment’s specified withstand and the switch current stays under its rating.

Post-processing is where the study becomes usable. Peaks should be captured with adequate sampling, and plots should be paired with numeric extraction so that teams can compare cases quickly. Initial-condition handling deserves special care, since pre-charge on capacitors or remanent flux in transformers can shift peaks more than a small parameter tweak. Model version control also matters, because the hardest EMT questions usually require iterative refinement across weeks, not a single run. A workflow that records assumptions will save you time when stakeholders ask why a specific case was selected.

Common EMT modelling mistakes and checks for credible findings

Most EMT errors come from mismatched intent, detail, and validation. Models fail when key parasitics are missing, when nonlinear device limits are oversimplified, or when initial conditions are not physically consistent. Time step and solver choices can also create numerical damping that hides the very stress you’re trying to measure. Credible findings come from a small set of disciplined checks, repeated every time the model changes.

Start with a sanity pass on steady-state values before applying any transient event, since an incorrect operating point can poison everything downstream. Confirm that energy storage elements have realistic values, and check that their initial voltages and currents match the pre-event conditions you intended. Run a convergence check on time step, and verify that peak values and ringing frequency do not shift materially as you refine resolution. Then challenge the result by removing one modelling refinement at a time and confirming you understand why the waveform changes.

Good EMT practice also includes a clear stopping rule. When the answer you need is “peak overvoltage at this terminal,” additional model detail that does not move that peak is extra complexity with little value. Teams that build that discipline end up with EMT models that stay usable across multiple studies, because the model is structured around limits and checks, not around maximum detail. SPS SOFTWARE fits well into that mindset because its open modelling style supports inspection and peer review, which is what keeps transient studies defensible over time. Careful execution will always matter more than the most sophisticated network you can draw.