Harmonic Impact of DC Fast-Charging Clusters and Mitigation

The problem

Fast-charging hubs place several high-power converters on one connection. Diode or thyristor front ends inject large low-order harmonics, and even active front ends inject switching-frequency harmonics that can excite network resonances. Harmonics heat transformers and cables and distort voltage for other customers. Planners need to know when a cluster breaches limits and which mitigation works best.

Objectives

• Model fast chargers with diode-bridge, 12-pulse and active front ends.
• Model a medium-voltage feeder with transformer, cable capacitance and background distortion.
• Quantify current and voltage THD and individual harmonics at the point of common coupling for varying cluster size.
• Study harmonic cancellation and resonance with cable capacitance.
• Design and compare passive tuned filters and a shunt active power filter.

What you need

Key concepts

Characteristic harmonics

A six-pulse rectifier injects harmonics of order 6k±1 (5th, 7th, 11th, 13th and so on). A 12-pulse arrangement cancels the 5th and 7th, leaving 12k±1.

Point of common coupling

The point in the network where the charging hub connects and other customers are also supplied. Harmonic limits are assessed there.

Parallel resonance

Cable capacitance and power-factor capacitors form a resonance with network inductance. Harmonic currents near that frequency produce large voltage distortion.

Harmonic cancellation

Converters running at different operating points produce harmonics with different phase angles, so the cluster total can be less than the arithmetic sum.

Step-by-step guide

1. 1

   Frame the study

   Define an example hub: four to ten 150 kW chargers on an 11 kV feeder through a 2 MVA transformer, with a short-circuit level chosen as an example. Choose a harmonic standard and write its limits for current distortion and voltage THD at the point of common coupling. State the research questions: how distortion scales with cluster size and topology, the role of resonance and which mitigation is most effective.

   Checkpoint: A study plan with network parameters, limits table and research questions.

2. 2

   Build the network

   In Simscape Electrical, model an 11 kV Thevenin source with the chosen short-circuit impedance, a cable as PI sections with realistic capacitance, the 11 kV/415 V transformer and a point-of-common-coupling bus with background distortion added as small harmonic voltage sources. Run a frequency scan with the powergui impedance measurement tool to find network resonances. Validate the source impedance by applying a three-phase fault in simulation and checking the fault current matches the chosen short-circuit level.

   Checkpoint: An impedance-versus-frequency plot at the point of common coupling showing a resonance peak, and its frequency stated.

3. 3

   Model charger front ends

   Build three front-end variants as switched models: a six-pulse diode bridge with DC choke; a 12-pulse arrangement with a phase-shifting transformer; and a two-level active front end with LCL filter and dq current control at around 10 kHz. Load each DC link with a constant-power DC–DC stage represented by a controlled current sink. Use a discrete solver with a step of 1–2 µs.

   Checkpoint: Each front end alone shows its expected spectrum: 5th and 7th dominant for six-pulse, reduced for 12-pulse and switching-band harmonics for the active front end.

4. 4

   Verify single-charger spectra

   For each topology at 25, 50 and 100 per cent load, run to steady state and compute current harmonics to the 50th order with the powergui FFT tool or a MATLAB FFT over an integer number of cycles. Compare individual harmonics against published characteristic values or theory. Record current THD and total demand distortion. Use a window of at least ten fundamental cycles after the DC link has settled, and keep the same window length for every case so results are directly comparable.

   Checkpoint: A single-charger harmonic table per topology and load, consistent with theory within the expected range.

5. 5

   Cluster studies

   Combine N chargers (N = 1, 2, 4, 6, 10) at the point of common coupling, each with an independently chosen operating point drawn from a usage distribution. Use Monte Carlo with many draws per N. Compute voltage THD and individual voltage harmonics at the bus. Compare the cluster total against N times the single-charger value to quantify cancellation. Draw each charger's load from a realistic distribution of charging power, since a cluster at full power on every unit is rare. Report both the mean and a high percentile such as the 95th, as harmonic standards often assess percentiles rather than single values.

   Checkpoint: Plots of voltage THD against N for each topology, showing sub-linear growth due to cancellation, with spread from Monte Carlo.

6. 6

   Resonance interaction

   Vary cable length and add a power-factor correction capacitor bank to move the network resonance. Identify cases where the resonance aligns with a dominant charger harmonic and quantify the amplification. For active front ends, check whether switching harmonics near the LCL resonance interact with the network. For each case, mark the resonance frequency from the impedance scan on the harmonic spectrum plot, so the link between the scan and the amplification is clear. Record how far the resonance moves for each change in cable length or capacitor size.

   Checkpoint: At least one case showing marked amplification of a harmonic near the resonance, explained with the impedance scan.

7. 7

   Design passive filters

   Design single-tuned filters at the 5th and 7th, and a high-pass filter above, sized for the six-pulse cluster's harmonic currents and reactive power. Choose a quality factor and slight detuning to avoid creating a new resonance at a lower frequency. Recheck the impedance scan with the filters installed. Check the filter capacitor voltage and current ratings including harmonics, and check the reactive power at fundamental frequency, which can cause overvoltage at light load. State the detuning used and why it was chosen.

   Checkpoint: Voltage THD reduced below the limit for the six-pulse cluster, and a new impedance scan showing no harmful resonance.

8. 8

   Design an active filter

   Model a shunt active power filter: a voltage-source inverter with a DC capacitor, current reference from instantaneous power theory or synchronous-frame harmonic extraction, and a hysteresis or PI current controller. Size it from the harmonic current to compensate. Test with the six-pulse cluster and with mixed topologies. Check the filter's own switching ripple at the point of common coupling and add a small output filter if needed. Test its response to a sudden change in charger load to confirm it tracks the changing harmonic currents.

   Checkpoint: Supply current THD reduced to below the limit, with filter rating and losses stated.

9. 9

   Compare mitigation options

   Tabulate, for each topology and cluster size, voltage THD, worst individual harmonic, transformer harmonic loss factor and mitigation needed. Compare 12-pulse transformers, passive filters, an active filter and all-active-front-end chargers. Use proxies such as filter kVA or component count rather than prices. Include losses and the reactive power each option adds or absorbs at fundamental frequency. Note practical factors such as footprint and whether each option still works when the number of chargers on site grows, and mark the cases where no single option is enough on its own.

   Checkpoint: A comparison table showing which mitigation meets limits for each case and its relative size.

10. 10

    Sensitivity to network strength

    Repeat key cases with a weaker and stronger supply (lower and higher short-circuit ratio). Report how voltage distortion and filter effectiveness change. Draw conclusions about which sites need mitigation regardless of charger topology. Use at least three short-circuit ratios that span weak rural feeders to strong urban ones. For each, rerun the impedance scan, because the resonance frequency also shifts as the supply impedance changes, which can move it closer to or further from a dominant charger harmonic.

    Checkpoint: A plot of voltage THD against short-circuit ratio for the main cases.

11. 11

    Write up and validate

    Document the network and converter models, the Monte Carlo method and all results. Where possible compare a single-charger spectrum against a measured or published spectrum to validate the model. State limitations: idealised loads behind the DC link, no interharmonics from DC–DC stages, background distortion assumed. Note the study is simulation only. Present results with confidence ranges from the Monte Carlo runs. Make clear which conclusions depend on the assumed background distortion and network data.

    Checkpoint: A chapter with validated single-unit spectra, cluster results and a mitigation recommendation, all reproducible from scripts.

Testing & validation

• Single-charger spectra consistent with theory or published measurements.
• Impedance scans explain observed resonance amplification.
• Filtered cases meet the chosen limits with no new harmful resonance.
• Monte Carlo spread reported for cluster results.

Expected results

A harmonic assessment of fast-charging clusters by topology and size, including cancellation, resonance effects and a validated comparison of passive and active mitigation against chosen limits.

Take it further

• Add interharmonics from DC–DC stages switching at variable frequency.
• Study supraharmonics (2–150 kHz) from active front ends.
• Coordinate active front ends to provide harmonic compensation for the hub.
• Use measured spectra from installed chargers to calibrate the models.

Viva and review questions

Learn it at DIYguru

These DIYguru courses and programmes teach what this project needs.