Harmonic analysis and power quality studies in time domain simulation

Key Takeaways

• • Time domain simulation shows how harmonic current becomes bus voltage distortion under actual operating states.

  • Model boundaries, feeder impedance, switching detail, and analysis windows shape the credibility of total harmonic distortion results.

  • Bus specific limit checks lead to better mitigation choices than device level screening alone.

Frequency scans help with screening, but they won’t show how converters, cable impedance, transformers, and controls interact over time. Solar and battery storage were set to account for 81% of new U.S. utility scale generating capacity in 2024, so more feeders now include switching equipment that injects non-sinusoidal current. Useful power quality analysis will model the switching source, the feeder path, and the measurement bus used for compliance or troubleshooting. That is how you’ll calculate total harmonic distortion in a model with results that match what connected equipment actually sees.

“Time domain studies show when distorted current becomes distorted voltage at a bus.”

Time domain studies show when harmonic distortion actually forms

Time domain studies show when distorted current becomes distorted voltage at a bus. Switching, controls, and feeder impedance stay inside the same simulation. Cause and effect stays visible. That makes the result easier to trust when several devices share a feeder.

Consider a feeder with two active front ends and one six-pulse rectifier. Each load can look acceptable on its own. Shared source impedance can still turn their current harmonics into a bus voltage problem when all three operate at once. A frequency scan will flag resonant points, but it won’t show how settling controls and capacitor switching shape the waveform over time.

That timing matters because power quality problems are often state-dependent. Startup and light load can produce worse waveforms than final steady operation. A step change in converter output can also shift the spectrum. If you want to know what causes power quality problems on a feeder, you need the sequence as well as the spectrum.

Feeder impedance sets where converter harmonics become voltage distortion

Feeder impedance decides where harmonic current turns into a voltage distortion problem. The same converter can look harmless on a stiff source and troublesome on a weak feeder. Each harmonic current creates a voltage drop through the network path. That is why bus location matters in harmonic analysis.

A short feeder tied to a large transformer usually holds bus voltage shape better than a long cable run with a smaller upstream transformer. A nearby capacitor bank can also amplify a specific harmonic order. Current measured at the converter terminals will not tell you that full story. The important question is what nearby motors, relays, and sensitive loads see at their own bus.

Many models miss this point because the upstream network is reduced to an ideal source with simple inductance. That shortcut hides source strength and local resonance. Transformer leakage, cable capacitance, and correction capacitors all matter here. Harmonic distortion limits are checked at buses, so your network model has to explain bus voltage, not just device current.

Build the model around the bus where limits apply

The right model boundary starts at the bus where you must judge compliance or diagnose equipment stress. Harmonic studies become clearer when that bus is fixed first. Nearby buses then become supporting measurement points. The feeder elements between them define the distortion path.

A plant feeder shows why this works. If complaints come from a motor control centre, measurements taken only at the incoming service will miss distortion created downstream of the transformer. A campus microgrid can show the same problem. The utility interface can look clean while a long secondary feeder shows high voltage total harmonic distortion during heavy converter loading.

• Choose the bus where limits or complaints actually apply.
• Include source strength from the upstream network and transformer data.
• Represent feeder segments that add meaningful impedance between buses.
• Add capacitor banks or filters connected near the monitored location.
• Run the operating states that place the highest stress on that bus

That structure keeps the study focused. You’re not building the whole system for completeness. You’re building enough of the system to explain the waveform at the bus that matters. That discipline also keeps mitigation work tied to the right location.

Model each nonlinear load with switching detail that matters

Nonlinear loads need enough switching detail to reproduce the current shape that drives feeder distortion. You do not need identical detail for every device. You do need each major source represented at the level that affects harmonic current injection. That choice sets the quality of the final spectrum.

Electric motor systems account for about 45% of global electricity consumption, and many now use power electronic drives that alter current waveforms rather than drawing smooth sinusoidal current. A six pulse rectifier for a DC drive should not be reduced to a static power sink if you’re studying current distortion. A PWM inverter tied to a filter and control loop also needs more than an averaged source when feeder interaction is the question. The model has to preserve the waveform features that shape harmonic content.

SPS SOFTWARE fits this step when you need transparent converter and network models that you can inspect and tune. That matters on mixed feeders with several nonlinear loads. One device can be modelled with averaging for power flow, while another needs explicit switching to reproduce the distortion that shapes bus voltage. The useful goal is the right detail where distortion begins.

Choose sample time that captures converter ripple correctly

Sample time must be short enough to resolve the switching ripple and commutation features that create harmonic content. A coarse step will smooth the waveform before analysis starts. That error will carry straight into the spectrum. Total harmonic distortion will then be biased from the start.

A converter switching at 5 kHz cannot be represented well with a step that barely samples each cycle. A line commutated rectifier has a different need because overlap and current notches dominate the distortion. Your time step should match the fastest event that influences the measurement bus. Grid fundamental frequency alone is not enough for this choice.

Good sampling practice saves time later. A wrong step can make every later plot look precise while carrying the same error. Harmonic studies are unforgiving on this point. Solver setup belongs inside the electrical study, not beside it.

Run steady state windows long enough for clean spectra

Steady state analysis windows must exclude startup and include enough settled cycles to produce a clean spectrum. Harmonics are measured from a selected time record. Record length and placement shape the answer directly. A poor window will blur a good model.

A common mistake is capturing the first few cycles after a converter reaches its reference and sending that waveform straight to an FFT. Control loops can still be settling. DC link voltage can still be drifting. A better workflow is to let the system settle fully, then extract an integer number of grid cycles from the bus voltage and device current that matter.

Window length also depends on what you need to see. Short records are useful for quick screening, but they smear closely spaced components and hide slow modulation effects. Longer records give cleaner harmonic bins and more stable results. If the spectrum shifts when you slide the time window, the system was not settled or the record was too short.

Calculate total harmonic distortion from voltage or current waveforms

Total harmonic distortion is calculated from a settled waveform after harmonic RMS components are extracted. The standard calculation uses the root sum square of all harmonic components above the fundamental. That value is divided by the fundamental RMS value. The result is then multiplied by 100%.

Voltage total harmonic distortion and current total harmonic distortion answer different questions. Voltage distortion tells you what connected equipment sees at a bus. Current distortion tells you what a converter or nonlinear load injects into the network. A front-end drive can show high current distortion while service voltage stays acceptable on a stiff system, then push bus voltage past the limit on a weak feeder.

You’ll get better results if you state exactly how the number was produced. Record the measurement bus, the operating state, the record length, and the harmonic range used in the calculation. Some studies also need individual harmonic magnitudes because one troublesome order can excite resonance or heat a filter branch. 

“The number matters, and the waveform behind it matters too.”

Check harmonic limits bus by bus before choosing fixes

Harmonic distortion limits should be checked at each relevant bus before you choose a mitigation method. That approach separates harmless injected current from harmful bus voltage distortion. It also keeps the study tied to the location that matters. Good fixes start with the right bus and the right operating state.

A utility point of common coupling, a plant main bus, and a sensitive secondary bus can all show different results under the same loading condition. One bus can meet voltage limits while another exceeds them because local impedance and capacitor placement are different. That is why harmonic distortion limits and how to check them always come back to location, operating state, and measurement method. You’re judging network response where equipment actually sees it.

Careful time domain work usually leads to a calmer answer than quick screening. You see which source is dominant, which bus matters, and which operating state triggers the problem. SPS SOFTWARE belongs in that closing step because transparent feeder and converter models make it easier to test a filter, move a capacitor bank, or revise control settings with confidence. Good power quality analysis comes from disciplined modelling followed by bus-specific checks.