Smart Charging and Load Balancing

By Michael Samuel Naeem · October 27, 2025 · 5 min read

Six 48A chargers on a 200A panel. All six start charging simultaneously. The main breaker trips. The electrician's fix—hard-cap each charger at 32A—means no vehicle ever charges at full speed, even when only one is in use. OCPP smart charging solves this dynamically: the CSMS sends charging profiles that adjust power limits per connector based on how many sessions are active, time-of-use tariffs, and grid demand signals.

Charging profile structure

{
  "connectorId": 1,
  "chargingProfile": {
    "chargingProfileId": 101,
    "stackLevel": 0,
    "chargingProfilePurpose": "TxDefaultProfile",
    "chargingProfileKind": "Absolute",
    "chargingSchedule": {
      "chargingRateUnit": "A",
      "chargingSchedulePeriod": [
        { "startPeriod": 0, "limit": 32.0 },
        { "startPeriod": 3600, "limit": 16.0 },
        { "startPeriod": 7200, "limit": 48.0 }
      ]
    }
  }
}

Profile purposes

ChargePointMaxProfile  → Hardware maximum (never exceed breaker)
    ↓ overridden by
TxDefaultProfile       → Default for new transactions
    ↓ overridden by
TxProfile              → Specific to one active transaction

Set ChargePointMaxProfile at installation to the site breaker rating. Layer TxDefaultProfile for normal operation. Override with TxProfile for priority vehicles or demand response events.

Load balancing algorithm

def allocate_current(
    active_sessions: list[Session],
    site_max_amps: float,
    min_amps_per_session: float = 6.0,
) -> dict[int, float]:
    if not active_sessions:
        return {}

    fair_share = site_max_amps / len(active_sessions)

    allocations = {}
    remaining_amps = site_max_amps

    # Priority: vehicles departing soon get more current
    sorted_sessions = sorted(active_sessions, key=lambda s: s.departure_time)

    for session in sorted_sessions:
        requested = session.max_amps
        allocated = min(requested, fair_share, remaining_amps)
        allocated = max(allocated, min_amps_per_session)
        allocations[session.connector_id] = allocated
        remaining_amps -= allocated

    return allocations

Send updated profiles when sessions start, stop, or every 60 seconds.

OCPP 2.0.1 smart charging

OCPP 2.0.1 adds vehicle-side input:

Vehicle → Charger: ISO 15118 charging needs (target SOC, departure time)
Charger → CSMS: NotifyEVChargingNeeds
CSMS → Charger: SetChargingProfile (optimized for needs + grid)

This enables the charger to balance user needs ("I need 80% by 8 AM") against grid constraints ("site max 100 kW until 6 PM").

Time-of-use integration

def get_tariff_limit(timestamp: datetime, tariff: TariffSchedule) -> float:
    hour = timestamp.hour
    if tariff.is_peak(hour):
        return tariff.peak_max_amps  # e.g., 16A during peak
    return tariff.off_peak_max_amps  # e.g., 48A overnight

def build_profile(connector_id: int, limit: float) -> dict:
    return {
        "connectorId": connector_id,
        "chargingProfile": {
            "chargingProfileId": int(time.time()),
            "stackLevel": 1,
            "chargingProfilePurpose": "TxDefaultProfile",
            "chargingProfileKind": "Relative",
            "chargingSchedule": {
                "chargingRateUnit": "A",
                "chargingSchedulePeriod": [
                    {"startPeriod": 0, "limit": limit}
                ],
            },
        },
    }

Update profiles when tariff periods change—typically at midnight and peak-hour boundaries.

Demand response

When the utility sends a demand response signal:

1. CSMS receives DR event (reduce load by 30% for 2 hours).
2. Calculate new limits: site_max * 0.7.
3. Push updated TxDefaultProfile to all active chargers.
4. Log compliance: actual draw vs target.
5. Restore full limits when DR event ends.

async def handle_demand_response(event: DREvent):
    chargers = await get_active_chargers(event.region)
    reduced_max = event.site_capacity * (1 - event.reduction_pct)

    for charger in chargers:
        profiles = compute_profiles(charger.sessions, reduced_max)
        for profile in profiles:
            await csms.set_charging_profile(charger.id, profile)

Monitoring

Track per site:

• Total power draw vs site capacity (utilization %)
• Per-connector allocated vs actual current
• Profile override events (vehicle requesting more than allocated)
• DR compliance during events

Alert when site utilization exceeds 95% for > 10 minutes—time to add capacity or tighten allocation.

OCPP charging profile stack

Profiles stack with priority — lower stack level wins:

Stack Level 0: TxDefaultProfile    (site-wide limit, always active)
Stack Level 1: TxProfile           (per-transaction limit from CSMS)
Stack Level 2: ChargingStationMaxProfile  (hardware max)
Stack Level 3: ExternalConstraints (ISO 15118 from vehicle)

{
  "connectorId": 1,
  "csChargingProfiles": {
    "chargingProfileId": 42,
    "stackLevel": 0,
    "chargingProfilePurpose": "TxDefaultProfile",
    "chargingProfileKind": "Recurring",
    "recurrencyKind": "Daily",
    "chargingSchedule": {
      "duration": 86400,
      "chargingRateUnit": "W",
      "chargingSchedulePeriod": [
        {"startPeriod": 0,    "limit": 11000},
        {"startPeriod": 28800,"limit": 22000},
        {"startPeriod": 64800,"limit": 11000}
      ]
    }
  }
}

Off-peak (0–8h): 11kW. Peak (8–18h): 22kW. Evening (18–24h): 11kW. Update at tariff period boundaries.

Vehicle-aware load balancing

ISO 15118 enables vehicle to communicate charging needs:

Vehicle → EVSE: "I need 80% SoC by 7 AM, max 11kW acceptable"
CSMS → Charger: SetChargingProfile with vehicle constraints
Charger → Vehicle: Charge at allocated power within vehicle limits

Without ISO 15118, CSMS allocates blindly — vehicle may request more than allocated, causing profile override events. Log override events — frequent overrides indicate allocation algorithm needs tuning.

Capacity planning from utilization data

def capacity_recommendation(site_id: str, months: int = 6) -> dict:
    history = get_utilization_history(site_id, months)
    peak_util = max(h["utilization_pct"] for h in history)
    p95_util = percentile([h["utilization_pct"] for h in history], 95)

    if p95_util > 85:
        return {"action": "add_capacity", "urgency": "high",
                "recommended_kw": site_capacity * 0.3}
    if peak_util > 95:
        return {"action": "tighten_allocation", "urgency": "medium"}
    return {"action": "monitor", "urgency": "low"}

Six months of utilization data informs infrastructure investment decisions — not guesswork.

Failure modes

• Profile not updated at tariff boundary — peak rates applied during off-peak hours
• No DR event handler — utility demand response signal ignored; penalty charges
• Site capacity hardcoded — actual available capacity changes with other loads
• Override events not logged — allocation algorithm bugs undetected
• Single profile for all connectors — one high-priority session starves others

Production checklist

• Charging profile stack configured with correct priority levels
• Profiles updated at tariff period boundaries (automated)
• DR event handler reduces site max and restores after event
• Per-connector allocated vs actual current monitored
• Profile override events logged and alerted
• Capacity planning based on 6-month utilization history

Resources

• OCPP 1.6 Smart Charging — SetChargingProfile specification
• OCPP 2.0.1 Smart Charging use case — NotifyEVChargingNeeds
• ISO 15118 vehicle-to-grid communication — vehicle charging needs protocol
• IEA Global EV Outlook — grid impact projections
• OpenADR demand response standard — utility signal integration