Impact of EV Charging on a Residential Distribution Feeder

The problem

As EV adoption grows, many homes on the same street may charge in the evening when household demand already peaks. Distribution transformers and long feeders were not sized for this, so voltage at the far end can fall below limits and transformers can overload. This project quantifies those effects on an example feeder and tests simple mitigation.

Objectives

• Build a radial LV feeder model with a distribution transformer, line sections and household loads.
• Add EV chargers at chosen buses with realistic charging profiles.
• Measure end-of-feeder voltage, transformer loading and losses at increasing EV penetration.
• Compare uncontrolled, staggered and off-peak charging.
• Identify the penetration level at which limits are first violated.

What you need

Key concepts

Phasor simulation

Instead of solving every 50 Hz cycle, the solver computes voltage and current magnitudes and angles at the fundamental frequency. It is far faster and adequate for steady-state loading and voltage studies.

Voltage drop on a radial feeder

Approximately ΔV ≈ (P·R + Q·X)/V for each section. Because loads accumulate toward the transformer, the far end sees the largest drop when many loads are on.

Penetration level

The share of households on the feeder that own and charge an EV. Studies sweep this from zero upwards to find the point where limits are first breached.

Coincidence of demand

Household peaks and EV arrival times overlap in the evening. The probability that many chargers run at once sets the real peak, not the sum of ratings.

Step-by-step guide

1. 1

   Define the example feeder

   Use an example network: a 250 kVA, 11 kV/433 V transformer feeding a 400 m three-phase radial feeder with ten buses, each supplying five houses split across phases. Choose a conductor and write resistance and reactance per km. Set household peak demand around 2 kW at 0.95 power factor and EV chargers at 3.3 kW single-phase. Adopt a voltage limit, for example ±6 per cent of nominal, and a transformer loading limit of 100 per cent.

   Checkpoint: A parameter script and a one-line diagram showing buses, phases and line lengths.

2. 2

   Build the network in phasor mode

   In Simscape Electrical (Specialized Power Systems), place a powergui block set to Phasor at 50 Hz. Add a Three-Phase Source for the 11 kV grid, a Three-Phase Transformer (Two Windings) block and Three-Phase PI Section Line or series RLC branches for each section. Add single-phase loads per bus as Dynamic Load blocks or controlled current sources, so that P and Q can follow a time profile.

   Checkpoint: With all loads off the 433 V bus voltages sit at nominal, and a single 10 kW load gives a voltage drop matching a hand calculation within a few per cent.

3. 3

   Create household load profiles

   Build a 24-hour, 15-minute household profile with a morning and an evening peak, then add random variation per house so not all houses peak at once. In MATLAB, generate fifty profiles with a fixed random seed. Feed each load from a From Workspace block. Run the base case without EVs and record transformer loading and the lowest bus voltage across the day.

   Checkpoint: Base case transformer loading peaks in the evening below the limit, and the far-end voltage stays within ±6 per cent.

4. 4

   Model EV charging profiles

   For each EV, sample an arrival time from a distribution centred around 19:00 (for example a normal distribution with one-hour spread), an arrival SOC and a battery size such as 30 kWh. The charger draws 3.3 kW until the energy need is met. Implement this in MATLAB to produce a power profile per EV, then add it to the household profile at that house. Keep the random seed so runs are repeatable.

   Checkpoint: A plot of the aggregate EV demand showing an evening peak whose height grows with the number of EVs.

5. 5

   Sweep penetration levels

   Use a script with sim() and Simulink.SimulationInput to run penetration levels of 0, 20, 40, 60, 80 and 100 per cent, choosing which houses get EVs at random. For each, run several random seeds (Monte Carlo) to capture variation in arrival times. Log the minimum bus voltage, maximum transformer loading, maximum line current and total daily losses. Use at least ten seeds per level and store every run's results in a table so averages and worst cases can both be reported later.

   Checkpoint: A results table where minimum voltage falls and transformer loading rises monotonically with penetration, averaged over seeds.

6. 6

   Check phase imbalance

   Because home chargers are single-phase, EVs can cluster on one phase. Compute the voltage unbalance factor at the far bus and the neutral current for a few unlucky allocations, for example all EVs on phase A. Compare with an even allocation. Note that phasor mode handles unbalanced loads directly, so no change to the solver is needed. Compute the unbalance factor from the negative- and positive-sequence voltages, which the Sequence Analyzer block or a short MATLAB calculation provides.

   Checkpoint: The worst-case single-phase allocation shows a clearly lower far-end voltage on the loaded phase than the balanced case.

7. 7

   Test mitigation strategies

   Implement two simple strategies in the profile generator: staggered start, where each EV delays its start by a random 0–3 hours, and off-peak charging, where charging begins after 23:00. Rerun the worst penetration levels. Compare minimum voltage, transformer peak and whether each EV still finishes charging by 07:00. Report any EVs that do not reach their target energy. Keep the same random arrivals and house allocations as the uncontrolled case, so any difference comes from the strategy alone and not from a different random draw.

   Checkpoint: Off-peak and staggered cases lower the transformer peak at high penetration, with all EVs still fully charged by morning.

8. 8

   Find the hosting capacity

   From the sweep results, find the highest penetration level at which no limit is breached in, for example, 95 per cent of Monte Carlo runs. Do this for uncontrolled and for each mitigation strategy. Plot the probability of a voltage violation against penetration. This is the feeder's EV hosting capacity under each strategy. Report the threshold with its Monte Carlo spread, and state which limit is breached first, voltage or transformer loading, because that decides which upgrade a utility would consider.

   Checkpoint: A hosting-capacity plot with one curve per strategy, showing mitigation moving the violation threshold to higher penetration.

9. 9

   Report

   Document the feeder, the load and EV models, the Monte Carlo method and the results. Include the bus-voltage profile along the feeder at peak, the transformer loading curve for each case, the imbalance study and the hosting-capacity plot. State the simplifications: one example feeder, synthetic profiles, constant-power loads and no on-load tap changer. Note what a utility would need to apply this to a real feeder.

   Checkpoint: A report with figures generated from saved results, and a clear statement of hosting capacity for each strategy.

Testing & validation

• Single-load voltage drop matches a hand calculation.
• Energy balance: transformer energy equals load energy plus line losses.
• Monotonic trends in voltage and loading with penetration across Monte Carlo runs.
• All EVs reach target energy under each mitigation strategy, or shortfalls are reported.

Expected results

A phasor-mode feeder model with stochastic household and EV profiles, giving the voltage and loading impact of EV penetration and the hosting capacity with uncontrolled, staggered and off-peak charging.

Take it further

• Add rooftop PV at some houses and study midday reverse flow.
• Model an on-load tap changer or a line voltage regulator.
• Add a controlled charging algorithm that responds to local voltage.
• Use a published test feeder topology for comparison with literature.

Viva and review questions

Learn it at DIYguru

These DIYguru courses and programmes teach what this project needs.