HEVI-LOAD Tool for MHDEV Grid Analysis

• HEVI-LOAD Tool is a simulation framework that generates minute-level MHDEV charging profiles for evaluating distribution grid impact.
• It integrates projected fleet deployments with spatially explicit charger siting data using queueing models for realistic load profiling.
• The tool maps charging loads onto grid nodes to identify voltage violations, guiding infrastructure upgrades and planning.

The HEVI-LOAD Tool (Medium- and Heavy-Duty Electric Vehicle Infrastructure – Load, Operation, and Deployment) is a software system developed by Lawrence Berkeley National Laboratory (LBNL) for the synthesis of minute-level charging load profiles for medium- and heavy-duty electric vehicle (MHDEV) fleets. By integrating projected fleet deployments and spatially explicit charger siting data at the county scale, HEVI-LOAD produces realistic, time-resolved event logs for individual vehicle charging behavior at megawatt-scale commercial charging stations. Its outputs are utilized to analyze the compatibility of distribution grid infrastructure with large-scale fleet electrification scenarios, as evidenced in recent studies investigating voltage stability and grid reinforcement requirements (Hassan et al., 23 Jul 2025).

1. Input Data and Software Outputs

HEVI-LOAD generates its primary outputs by simulating projected MHDEV fleets and charger deployment strategies, with explicit assignment of vehicles to geographically resolved high-power charging stations (HPCSs). The output data for each HPCS takes the form of a CSV file with columns for:

• Time (in minutes, 0–1440 for a single day)
• Vehicle ID
• Charging Power (kW)
• Charging Duration (minutes).

This enables direct modeling of individual vehicle charging events at the granularity required for grid impact studies. Importantly, the raw event logs may include concurrent charging requests from the same Vehicle ID, outnumbering the station's physical charger capacity. Data cleaning is achieved by imposing an M/M/c queueing model—where Poisson arrivals with rate $\lambda$, exponential service times with mean $1/\mu$ (extracted from event durations), and $c$ parallel charger ports represent the station configuration. Vehicle charging requests are dispatched on a first-come, first-served basis, with excess arrivals queued. The cleaned result is a smoothed, feasible $P_{\text{HPCS}}(t)$ time series per station, aligned with hardware constraints.

2. Integration with Distribution System Models

The filtered HEVI-LOAD power demand series are mapped onto distribution system nodes for grid stability analysis. In the referenced benchmark, the IEEE 33-bus radial network (nominal voltage 12.66 kV) is utilized, with two classes of loads at each bus $i$ at time $t$:

• Baseline residential load, $P_{\text{res},i}(t)$ and $Q_{\text{res},i}(t)$, interpolated to 1-minute resolution from hourly DOE residential data;
• MHDEV charging load, $P_{\text{HPCS},i}(t)$ and $Q_{\text{HPCS},i}(t)$, applied only at buses hosting HPCSs.

AC power-flow solutions are obtained at each time step using the PandaPower framework, utilizing the standard Kron-reduced nodal equations:

$$P_i = V_i \sum_{j=1}^n V_j (G_{ij} \cos \theta_{ij} + B_{ij} \sin \theta_{ij})$$

$$Q_i = V_i \sum_{j=1}^n V_j (G_{ij} \sin \theta_{ij} – B_{ij} \cos \theta_{ij})$$

where $V_i$ and $V_j$ are voltage magnitudes; $\theta_{ij}$ is the voltage angle difference between buses $i$ and $j$; $G_{ij}$ and $B_{ij}$ are real and imaginary parts of the bus-admittance matrix.

3. Simulation Workflow and Scenario Construction

The analytical workflow proceeds through the following chain:

1. Selection of three geographically dispersed HPCS locations and extraction of their $P_{\text{HPCS}}(t)$ charging profiles, post-queue cleaning.
2. Assignment of these stations to specific “host” buses in the IEEE 33-bus feeder model (e.g., Bus 15 in the main scenario, with additional analysis at Buses 1, 10, 20, and 30).
3. Alignment of all load time series to 1-minute granularity and construction of the total bus-wise load as $P_{\text{res},i}(t) + P_{\text{HPCS},i}(t)$. Reactive charger load is assumed to be negligible ($Q_{\text{HPCS}}\approx 0$; unity power factor).
4. For $t=0 \dots 1440$ min, a full AC power-flow is solved, recording the resulting bus voltages $V_i(t)$.
5. Detection of voltage violations, defined as any $V_i(t) < 0.95$ p.u. or $> 1.05$ p.u., characterizing system stability constraints.

4. Grid Impact and Observed Voltage Violations

Analysis of HEVI-LOAD-driven scenarios reveals substantial voltage violations in the radial feeder network during periods of peak MHDEV charging. When the load at Location 1 reaches a maximum of approximately 1.4 MW (at $t \approx 570$ min), buses 15, 20, and 30 exhibit sub-0.95 p.u. voltage events, while buses 1 and 10 remain compliant.

Summary of minimum voltages observed (from Table I):

Bus Number	Minimum Voltage (p.u.)
1	0.9984
10	0.9245
15	0.8734
20	0.9766
30	0.9081

Persistent undervoltage at several nodes signals the inability of existing radial feeders to absorb unconstrained MW-scale MHDEV charging loads without risking downstream voltage collapse. These findings indicate a need for substantial infrastructure upgrades or flexibility mechanisms to ensure stability under ambitious electrification targets. A plausible implication is that without grid reinforcement or intelligent load management, widespread deployment of MHDEV fast charging will be accompanied by system-level reliability risks.

5. Model Assumptions and Methodological Limitations

Key modeling assumptions and simplifications in HEVI-LOAD and its application to grid impact studies are as follows:

• Queue model parameters ($\lambda, c$) are derived from generic HEVI-LOAD defaults; actual charger port counts may differ in situ.
• All charging events are modeled at unity power factor; real-world MW charging may exhibit nontrivial reactive power demand or be mitigated by on-site DERs.
• The analysis is restricted to a single 24-hour snapshot, with no co-optimization over charger siting or application of smart-charging controls; variations in arrival rates by day or season are not explored.
• Only balanced, single-phase power-flow is considered, not unbalanced or phase-diverse loading.

These limitations suggest that results should be interpreted as indicative of voltage stress under plausible but not exhaustive operational regimes.

6. Prospects for Extension and Application

Potential avenues for further development of the HEVI-LOAD analytical framework include:

• Implementation of dynamic smart-charging dispatch algorithms responsive to local voltage and system constraints.
• Integration of co-located photovoltaic or battery energy storage systems to buffer peak grid draws.
• Extension to unbalanced, three-phase distribution modeling for fidelity with real-world feeder topologies.
• Adoption of stochastic fleet models with arrival and departure distributions transcending the Poisson queueing abstraction.
• Coupling with distribution reinforcement and optimization modules (e.g., conductor upsizing, voltage regulatory device placement).

Such enhancements would enable more robust planning for MHDEV electrification, refining actionable guidance for infrastructure upgrades and grid code compliance.

7. Significance for Distribution System Planning

By synthesizing minute-level MHDEV charger demand scenarios, enforcing operational queueing constraints, and embedding these into distribution feeder power-flow models, HEVI-LOAD closes the analytic chain from county-level electrification projections to bus-level voltage stability assessment. This enables quantification of the spatial and temporal incidence of voltage violations, supporting distribution planners in identifying reinforcement needs and the locations most susceptible to stress under projected electrification policy goals (Hassan et al., 23 Jul 2025).