The 80% rule and why it produces wrong answers for fleet electrification
The 80% rule — reserve 20% of transformer capacity as permanent headroom — is a long-standing design guideline in low-voltage distribution engineering. It exists to account for transformer thermal derating under sustained load, load growth, and the protection relay settings that typically trip at 110–115% of rated current rather than at exactly 100%. For static, slowly-changing loads like HVAC systems and industrial production equipment, the 80% rule provides a reasonable safety margin.
Applied to fleet electrification planning, the 80% rule is a starting point that quickly becomes misleading. The issue is temporal diversity — or, in the fleet case, the systematic absence of it. A factory's load profile is roughly steady across a shift, with predictable peaks tied to production cycles. A fleet depot's EV charging load is sharply pulse-shaped: vehicles return in a wave after the day's shifts, create a large demand spike during the first 60–90 minutes as batteries recover from travel depletion, and then the demand drops as vehicles reach higher SoC states and battery management systems reduce AC current draw.
If you apply the 80% rule statically — "our transformer is 630 kVA, so we can load it to 504 kVA for EV charging" — you might install 22 AC charge points at 22 kW each (484 kW nameplate capacity) and conclude you're within headroom. But if 18 of those 22 chargers have vehicles connected within 10 minutes of shift end, all drawing maximum current simultaneously, the instantaneous load is 396 kW — well within the 80% figure. The problem is that this calculation ignores the building's baseline load. Add 80 kW of HVAC, lighting, and other building services still running at shift handover, and the total incomer load is 476 kW against the 504 kVA / 454 kW active power limit. That's a 5% overshoot that will eventually trip the transformer's overcurrent relay.
What transformer capacity actually needs to account for
A proper transformer capacity assessment for fleet electrification needs to model several factors that the static 80% rule ignores:
Baseline load profile by time-of-day
The EV charging load is additive on top of the existing site load. The worst-case hour is not the hour of highest EV demand in isolation — it's the hour when EV demand and baseline site load peak simultaneously. For most logistics depots, that coincidence occurs at 17:00–19:00: the day shift ends, vehicles return, and the site is still running office HVAC, workshop equipment, and security lighting at near-peak levels. A transformer capacity model that uses average baseline load rather than time-of-day baseline load understates the peak by 20–40% at these coincident hours.
Simultaneous connection factor
Not all vehicles connect simultaneously even when they return in a similar time window. Traffic patterns, dock allocation time, driver debriefing, and charger location all create natural staggering. A depot with 50 vehicles might see 35 connected within 30 minutes and the remaining 15 over the following 90 minutes. The peak charging demand occurs in the window when the early 35 are all drawing near-maximum current — before their SoC has risen enough to trigger current tapering. Understanding the actual simultaneous connection factor for a specific depot's operational pattern requires either historical EVSE data from an existing partially-electrified site, or conservative assumptions (assume 70–80% of vehicles connect within the first 20 minutes of the peak arrival window).
Battery thermal effects in winter
Battery management systems accept significantly higher AC charge current at temperatures above 15°C than at 0–5°C. A 150 kWh battery pack in a 3°C January depot may accept only 60–70% of the rated maximum charge current during the first 20 minutes of a session while the battery thermal management system warms the pack. This means winter peak demand from EV charging is often 15–25% lower than the summer equivalent for the same fleet. Paradoxically, this makes winter a worse season for transformer upgrade planning — because assessments done in summer show the true peak demand, while winter assessments may give false comfort.
A worked example: 800 kVA transformer, 55-vehicle BEV fleet transition
A cold-chain logistics depot in the Rhine-Ruhr region is transitioning 55 of its 80-vehicle mixed fleet to BEV, replacing diesel 18-tonne rigids with 400 kWh capacity BEV equivalents. The site has an 800 kVA distribution transformer serving both the depot operations and EV charging. At 0.95 power factor (corrected), this delivers 760 kW of active power. The existing depot baseline load peaks at 190 kW during afternoon shift (compressors, cold storage, loading dock heating in winter).
Available charging budget: 760 kW − 190 kW = 570 kW. At 80% utilisation: 456 kW available for EV charging with the conventional headroom rule. Fifty-five 400 kWh vehicles at 22 kW AC charging simultaneously: 1,210 kW nameplate demand — way over capacity. But the vehicles don't all charge simultaneously at full rate: those with higher arrival SoC (say, 60%+ for local routes) will require less intensive charging and may not be on maximum current at all. A conservative 60% simultaneous draw factor with 75% current acceptance (accounting for BMS tapering) gives: 55 × 22 kW × 0.60 × 0.75 = 544 kW. That's 95% of the 80%-rule limit — well within the overload zone when combined with baseline load at peak coincidence.
The static headroom calculation says: upgrade the transformer. The dynamic scheduling analysis says: with smart charging that enforces a 350 kW EV ceiling (190 kW baseline + 350 kW EV = 540 kW, comfortably within 760 kW), the full 55-vehicle fleet can charge overnight from 18:00 to 06:00 without a transformer upgrade. The available 350 kW budget across a 12-hour window delivers 4,200 kWh of fleet capacity — sufficient for all 55 vehicles to charge from an average 30% arrival SoC to 95% target (65% × 400 kWh × 55 vehicles = 14,300 kWh required; the math only works if charging starts spreading the load over 12 hours, not all at 22 kW simultaneously). This is precisely where the scheduling engine's value is demonstrated — it makes a non-trivial calculation that the static headroom rule cannot make.
Dynamic monitoring as the foundation of accurate planning
The worked example above depends on the scheduler's ability to enforce the 350 kW EV ceiling in real time. That enforcement requires continuous monitoring of the actual incomer load — not just the EVSE sub-panel — with sufficient measurement accuracy to detect incipient overloads before they happen. The specification for incomer monitoring in a smart charging context is more demanding than typical energy monitoring installations:
- Measurement sampling interval: ≤5 seconds for the control loop input signal
- Current transformer accuracy class: Class 0.5 or better (IEC 61869-2)
- Three-phase measurement: all three phases independently, not just apparent single-phase aggregate (phase imbalance from large 1-phase EV chargers can cause one phase to trip while aggregate power appears within limit)
- Integration with the scheduling engine: the incomer measurement must be available to the MILP solver within the solver's re-optimisation cycle, not just logged for post-hoc analysis
We're not saying that every depot needs a new energy monitoring installation before fleet electrification. Many sites with existing industrial energy management systems (BEMS, SCADA) already have appropriate incomer monitoring that can be integrated via Modbus TCP or BACnet. The gap is usually the integration layer between the BEMS measurement and the EV charging management software — they are often purchased from different vendors with no native data exchange protocol.
When a transformer upgrade is genuinely necessary
Not every depot can avoid infrastructure upgrades through scheduling alone. The cases where an upgrade is genuinely necessary share a common characteristic: the ratio of total theoretical EV charging demand to available transformer headroom is so high that no amount of load shifting within the available operational window can deliver sufficient energy to the fleet.
The calculation is straightforward: if a fleet requires X kWh per night on average, and the transformer has Y kW of EV charging budget available for Z hours, then Y × Z must exceed X with meaningful margin (typically 25–30% above the calculated requirement to handle high-demand days, battery variability, and unexpected arrivals). If Y × Z < X, no scheduling optimisation changes the arithmetic — additional transformer or grid capacity is needed.
For depots with highly utilised fleets (vehicles operating 2 or 3 shifts with short turnaround times), the operational window Z may be only 3–4 hours, making the energy delivery constraint very tight. Similarly, depots in urban locations with older low-voltage distribution networks may find that the grid connection level — the Netzanschlussleistung agreed in the grid connection contract — is lower than the site's existing transformer capacity, effectively capping available power below what the on-site transformer could theoretically deliver. In these cases, engaging the DSO for a grid connection upgrade is the right path, and the scheduling system's role shifts to operating optimally within the expanded but still finite limit.