EV Battery Swap Station Inventory Planner

Key inputs and what to enter

Demand profile

• Expected swaps per day – Average completed swaps on a typical day. Use historical data from comparable sites, pilot results, or a demand forecast. For a mid-size urban site, values from 100–300 swaps/day are common in early deployments.
• Station operating hours per day – Hours per day the station is open for swaps. Many commuter stations operate 16–20 hours; highway or fleet depots may run 24/7.
• Peak demand multiplier – How intense the peak hour is relative to the daily average hourly swap rate. A value of 2.0 means peak hours are about twice as busy as the average hour.
• Duration of peak window – Number of hours per day at elevated demand. You can combine multiple rushes (morning and evening) into a single equivalent peak window if needed.

Charging system

• Number of chargers in service – Count of charge points simultaneously available for swap packs. Exclude chargers reserved for other uses or frequently down for maintenance.
• Battery charge time – Average time, in hours, to bring a depleted swap pack back to the target state of charge. Use manufacturer data but adjust upward if your site throttles power during grid constraints.
• Post-charge inspection & cooldown – Time, in minutes, for safety checks, BMS diagnostics, and thermal stabilization before a pack can re-enter the ready inventory.

Service targets

• Target ready-pack service level – The percentage of customer arrivals that should find an immediately available charged pack. For example, 95% means at most 1 in 20 customers should experience a stockout under typical conditions.

Core formulas behind the planner

Average demand and cycle time

The first step is to convert daily demand into an average arrival rate per hour:

Formula: λ = (Expected swaps per day) / (Station operating hours per day)

$\lambda =\frac{\text{Expected swaps per day}}{\text{Station operating hours per day}}$

Each pack that enters the charging system spends time charging plus time cooling or under inspection before returning to ready inventory. The average cycle time is:

Formula: W = T_charge + T_cool

$W=T_{\text{charge}}+T_{\text{cool}}$

where $T_{\text{charge}}$ is in hours and $T_{\text{cool}}$ is cooldown time converted from minutes to hours.

Little’s Law for base circulating inventory

Little’s Law links the long-run average number of items in a system to arrival rate and cycle time. In MathML form:

For a swap station:

• L = average number of packs cycling through charging and cooldown
• λ = average swaps per hour
• W = average turnaround time per pack (hours)

The planner uses this as a baseline for how many packs must be in the system just to keep up with typical demand, not yet accounting for peaks or variability.

Safety stock for variability and peaks

Real arrivals are not perfectly smooth. The tool uses a normal approximation to variability in swap arrivals over the cycle time window. If the expected demand over one cycle is $\mu =\lambda W$, the standard deviation is approximately $\sigma =\sqrt{\mu }$ under a Poisson assumption.

For a target service level $SL$, the corresponding normal quantile $z$ is used to set safety stock:

Formula: Safety stock ≈ z ⋅ σ = z sqrt(λ W)

$\text{Safety stock}\approx z\cdot \sigma =z\sqrt{\lambda W}$

The final recommended ready-pack inventory is then roughly:

Formula: Recommended packs ≈ L + Safety stock

$\text{Recommended packs}\approx L+\text{Safety stock}$

Additionally, the peak demand multiplier and peak window length help stress-test the station against concentrated demand, highlighting if your chargers are likely to fall behind during busy periods even when the daily total looks manageable.

Interpreting the results

After running the planner, focus on three aspects:

• Recommended ready-pack inventory – The minimum number of charged packs you should plan to have on-site to hit your chosen service level, given the assumptions above.
• Implied charger utilization – High utilization (close to 100%) suggests that even small demand spikes or outages could cause backlogs. Lower utilization provides resilience but requires more chargers.
• Peak-hour stress – If peak demand during the specified window exceeds what your chargers can recover outside the peak, you may need either more chargers, more inventory, or demand-shaping tactics.

Treat the output as a planning recommendation, not a hard guarantee. In particular, increasing the target service level from 90% to 99% can substantially increase the suggested safety stock, which might or might not be economically justified.

Worked example: commuter swap station

Consider an urban commuter station with these inputs:

• Expected swaps per day: 180
• Station operating hours: 18
• Peak demand multiplier: 1.8
• Peak window: 3 hours
• Chargers: 16
• Charge time: 1.6 hours
• Cooldown: 20 minutes (0.33 hours)
• Target service level: 95%

Average arrivals per hour are $\lambda =\frac{180}{18}=10$ swaps/hour. The cycle time is $W=1.6+0.33\approx 1.93$ hours. Little’s Law gives a base circulating inventory of:

$L=10\times 1.93\approx 19.3$ packs.

Over one cycle, expected demand is $\mu =19.3$, with $\sigma \approx \sqrt{19.3}\approx 4.4$. For a 95% service level, $z\approx 1.65$, so safety stock is around $1.65\times 4.4\approx 7.3$ packs.

That suggests a total of roughly 27 packs in the system to meet the target on a typical day. The actual calculator may adjust this further based on the peak window and charger count to make sure the station can recover from the 3-hour peak and still replenish inventory before the next rush.

As a planner, you might round this to 28–30 packs to allow for maintenance holds and minor modeling error, then check whether 16 chargers can practically support that level of throughput under your grid constraints.

Example comparison: commuter vs. highway corridor

Even when two stations have similar daily swap counts, the shape of demand and chosen service level can lead to very different recommended numbers of packs and chargers. Use the tool to compare such scenarios quickly by adjusting inputs and noting how inventory recommendations change.

Assumptions and limitations

• Steady-state averages – The underlying math assumes demand and processing rates are roughly stable over time. Very early-stage pilots or highly seasonal sites may violate this assumption.
• Poisson/normal arrival model – Arrivals are treated as random but with variability similar to a Poisson process, then approximated by a normal distribution. Real fleets may have more structured behavior (e.g., shift changes) that increases clustering.
• Homogeneous packs and chargers – The planner assumes all packs are interchangeable and all chargers have similar performance. Mixed chemistries, different max C-rates, or prioritized fleets are not modeled explicitly.
• No explicit grid constraints – It does not enforce feeder limits, dynamic tariffs, or curtailment events. If your site must frequently throttle power, adjust the effective charge time input upward.
• Uptime and maintenance handled indirectly – Downtime for chargers or packs is not modeled in detail. You can approximate it by reducing the charger count or adding a few extra packs of buffer inventory.
• Planning, not real-time control – The tool is intended for long-term sizing and what-if analysis, not second-by-second dispatch or scheduling of swaps.

Because of these simplifications, always cross-check the output against real operating data and engineering judgment. A good practice is to re-run the planner periodically with updated demand and turnaround measurements to refine inventory and capacity decisions over time.