Battery pack sizing is the process of translating application requirements — energy, power, voltage, lifetime, mass, volume — into a cell configuration (S×P) and a set of first-order design specifications. It is the first quantitative step in any battery pack development programme, and every subsequent design decision builds on it.
Done well, pack sizing gives the design team a validated starting point: a cell type, a series-parallel count, a first-order thermal load estimate, and a mass and volume prediction. Done poorly, it produces a pack that either cannot meet its requirements at end of life, or is substantially oversized — wasting cost, mass, and volume.
This page works through the sizing process in full, from requirements capture through to a complete first-order specification.
The Sizing Process at a Glance
The steps are sequential — each depends on the outputs of the one before:
- Capture the application requirements
- Select a candidate cell
- Set the usable SoC window
- Calculate the S×P configuration
- Verify peak and continuous power
- Estimate pack mass and volume
- Size the thermal management system
- Account for ageing
- Iterate
→ Full data required to size a pack — comprehensive input list → Inputs to pack sizing → Battery calculators
1. Application Requirements
Before any cell can be selected or any S×P calculated, the application requirements must be defined. These are not just performance targets — they include every constraint the pack must satisfy throughout its lifetime.
The core set:
| Requirement | Typical unit | Notes |
|---|---|---|
| Usable energy | kWh | Energy the application must access — not total stored energy |
| Peak discharge power | kW | With duration (e.g. 150kW for 10s) |
| Continuous discharge power | kW | Sustained indefinitely within thermal limits |
| Peak charge power | kW | Determines charge time and thermal load during charging |
| Pack voltage range | V (min, max) | System-defined; determines S count limits |
| Charge time target | minutes/hours | Implies minimum charge C-rate |
| Operating temperature range | °C | Ambient range the pack must operate within |
| Lifetime — calendar | years | Typically 10–15 years for automotive |
| Lifetime — cycle | cycles or km | e.g. 1,500 full equivalent cycles |
| Mass budget | kg | Strict upper limit in many applications |
| Volume envelope | L or mm dimensions | With any dimensional constraints |
| End of life capacity | % of initial | Typically 80% — defines pack ageing allowance |
| Cost target | $/kWh or absolute | Sets the ceiling on cell selection and architecture |
Mission profiles / duty cycles are very difficult sometimes to get at the start of a pack cycle, but they are incredibly important. Even if there is a collection of cycles it is really useful, eg: minimum Range Test (for usable energy test), Fast Charge, and repeated heavy duty cycle (such as tow, charge, tow, charge, tow) that use for thermal and electrical sizing.
Two points worth emphasising:
Usable energy is not total pack energy. Total energy = S × P × V_nom × Ah. Usable energy is what the BMS allows the application to access, defined by the usable SoC window. Always size to the usable energy requirement, not the total.
Requirements must be met at end of life, not beginning of life. If the application requires 60kWh at end of life and the pack loses 20% capacity over its lifetime, the beginning-of-life pack must provide 60 / 0.8 = 75kWh. Sizing to beginning-of-life requirements and ignoring ageing is one of the most common — and most expensive — sizing errors.
→ Data required to size a pack
2. Cell Selection
Cell selection and pack sizing are iterative — the cell choice affects the S×P count, the thermal load, the BMS design, and the cost. But iteration has to start somewhere. The first pass cell selection is guided by:
Chemistry — does the application prioritise energy density (NMC, NCMA), cycle life and safety (LFP), low cost (LFP, Na-ion), or high power (NMC power cells, LTO)?
Format — cylindrical (robust, cost-effective, proven at scale), prismatic (high packing efficiency, dominant in large automotive and BESS), pouch (highest cell-level energy density, needs module compression), blade (C2P LFP designs).
Cell energy (Wh) — larger cells (e.g. 300–600Wh prismatic, 46xx cylindrical) mean fewer cells in the pack, simpler BMS harness, and lower cell-joining cost, but less flexibility in S×P tuning and higher consequence of a single cell failure.
Voltage — LFP cells at 3.2V nominal require more cells in series for a 400V pack (≈125S) than NMC cells at 3.6V nominal (≈111S). This affects pack geometry and costs.
DCIR — the internal resistance drives peak power capability and I²R heat generation. A cell with 2× the DCIR of an equivalent cell requires either more cells in parallel (cost, mass) or accepts lower peak power. DCIR of a Cell →
Datasheet vs real performance — cell datasheets specify performance at 25°C, typically at C/3 or C/5. At the duty cycle temperature and C-rate of the actual application, available capacity may be 10–30% lower. Always obtain test data across the operating envelope before committing to a cell.
→ Cell selection starts with the pack, not the datasheet → Cell benchmarking database → Power vs energy cells → Key cell metrics
3. Usable SoC Window
The usable SoC window is the fraction of cell capacity the BMS allows the application to access. It is always narrower than 0–100%:
- Setting a maximum SoC below 100% prevents the cell from sitting at full charge, which accelerates SEI growth and calendar ageing
- Setting a minimum SoC above 0% prevents deep discharge, which causes copper dissolution and dendrite formation in lithium-ion cells
- The combination of upper and lower limits defines the usable window, which directly determines cycle life
The SoC window is a design variable — a wider window delivers more energy from the same cells but reduces cycle and calendar life. A narrower window preserves lifetime at the cost of energy density. The optimum trade-off depends entirely on the application.
Typical usable SoC windows by application:
| Application | Typical usable SoC window | Rationale |
|---|---|---|
| Long-range BEV | 90–95% | Range is the primary value; accepted lifetime trade-off |
| PHEV | 65–75% | Balance between EV range and longevity |
| HEV (power buffer) | 25–35% | Narrow window maximises cycle life; energy is secondary |
| Stationary BESS — daily cycle | 80–90% | High utilisation; chemistry-dependent |
| BESS — frequency regulation | 20–30% | Very frequent shallow cycles |
| Industrial vehicle | 80–90% | Opportunity charging; predictable duty cycles |
| Aerospace | 70–80% | Safety and reliability margins dominate |
Note: these are indicative starting points. The optimal window for a specific application must be validated against the cell’s actual ageing data at the operating conditions of that application.
→ Usable SoC window in depth → Usable energy
4. S×P Configuration
With a candidate cell and a voltage requirement, the series count (S) is determined first, then the parallel count (P) is set to meet the energy requirement.
Series count (S)
The pack nominal voltage equals S × V_nom_cell. Given a target nominal pack voltage:
S = V_pack_nominal ÷ V_nom_cell
This must be an integer, so S is rounded to give the closest achievable nominal pack voltage. Note that the actual pack voltage range is:
- Maximum: S × V_cell_max (e.g. 4.2V for NMC, 3.65V for LFP)
- Minimum: S × V_cell_min (e.g. 2.5V for NMC, 2.5V for LFP)
These bounds must fall within the system voltage range requirements.
Example — 400V NMC pack:
- Cell V_nom = 3.6V
- S = 400 ÷ 3.6 = 111.1 → round to 111S
- Nominal pack voltage = 111 × 3.6V = 399.6V
- Maximum pack voltage = 111 × 4.2V = 466.2V
- Minimum pack voltage = 111 × 2.5V = 277.5V
Example — 400V LFP pack:
- Cell V_nom = 3.2V
- S = 400 ÷ 3.2 = 125 → 125S
- Nominal pack voltage = 125 × 3.2V = 400V
- Maximum pack voltage = 125 × 3.65V = 456.3V
- Minimum pack voltage = 125 × 2.5V = 312.5V
Note that an LFP pack requires 13% more cells in series than an equivalent NMC pack at 400V. This increases harness complexity, BMS measurement channels, and inter-cell busbar count.
Parallel count (P)
The parallel count is set to meet the usable energy requirement:
P = E_usable_BOL ÷ (S × V_nom × Ah_cell × SoC_usable/100)
where E_usable_BOL is the beginning-of-life usable energy, already factored up from the end-of-life requirement by the ageing allowance.
Example — 75kWh usable (BOL = 75/0.8 = 93.75kWh), 111S NMC, 50Ah cell, 90% usable window:
P = 93,750Wh ÷ (111 × 3.6V × 50Ah × 0.90) P = 93,750 ÷ 17,982 = 5.21 → round up to 6P
With 6P: BOL usable energy = 111 × 3.6V × 6 × 50Ah × 0.90 = 108kWh EOL usable energy (at 80% SoH) = 108 × 0.80 = 86.4kWh
This exceeds the 75kWh requirement — the design has 15% margin, which may be acceptable or may justify revisiting cell selection with a lower Ah cell.
Granularity matters. Adding or removing one cell from the parallel count changes pack energy by:
- ΔE per parallel cell added = S × V_nom × Ah × SoC_usable = in the example, 111 × 3.6 × 50 × 0.90 / 1000 = 18.0kWh
Adding or removing one cell from the series count changes pack energy by:
- ΔE per series cell added = V_nom × P × Ah × SoC_usable = in the example, 3.6 × 6 × 50 × 0.90 / 1000 = 0.97kWh
This asymmetry is important: the granularity of energy adjustment is much finer via series count than via parallel count. For applications requiring precise energy delivery, tuning S within a small range allows much closer matching to target than adjusting P.
→ Cell capacity and pack size → Variability in battery pack capacity → Series and parallel — fundamentals
5. Power Verification
The S×P configuration must deliver the required peak and continuous power within the cell’s operating limits.
Peak power check
Peak discharge current = P_peak ÷ V_pack_min
This is the worst case — power is demanded when pack voltage is at its minimum (low SoC, low temperature).
Example: 150kW peak, V_pack_min = 277.5V → I_peak = 150,000 / 277.5 = 541A
Current per cell (P = 6): I_cell = 541 / 6 = 90A → 1.8C on a 50Ah cell
Check against cell voltage at this current: V_cell_min_under_load = V_cell_OCV(SoC_min) − I_cell × DCIR_10s
If V_cell_min_under_load > V_cell_cutoff, the power requirement is met. If not, either P must increase, or a lower DCIR cell must be selected.
Continuous power and thermal load
The continuous current determines the steady-state I²R heat generation per cell:
Q_cell = I_cell² × DCIR_cont (W per cell)
Summed across all cells, this is the total thermal load the cooling system must reject to keep cell temperatures within limits. If this load exceeds the capacity of the passive or active cooling system, the continuous power must be derated.
→ Continuous current rating → Peak vs continuous power → Battery power demand solution
6. Mass and Volume Estimation
First-order mass estimate:
Total cell mass = S × P × m_cell
Pack mass ≈ total cell mass × 1.6
The 1.6× multiplier is derived from benchmarking data across more than 160 production packs. It accounts for the mass of the enclosure, BMS, busbars, thermal management system, HV components, and structural elements. It varies with pack architecture:
| Architecture | Pack mass / cell mass ratio (typical) |
|---|---|
| Cell-Module-Pack (CMP) | 1.6–1.8× |
| Cell to Pack (C2P) | 1.3–1.5× |
| Cell to Body / Structural | 1.2–1.4× |
First-order volume estimate:
Total cell volume = S × P × V_cell (litres)
Pack volume ≈ total cell volume ÷ cell-to-pack volume ratio
Typical cell-to-pack volume ratios from benchmarking: 55–65% (CMP), 65–75% (C2P), 75–80% (structural pack).
These are estimates — verified against actual component dimensions as the design matures. At the concept stage they provide a useful sanity check against the volume envelope constraint.
→ Pack mass → Pack benchmarking — real mass and volume data → Key pack metrics
7. Thermal Sizing
The thermal system must keep all cells within their operating temperature window — typically 15–40°C for optimal performance and lifetime, with an absolute maximum of 45–60°C depending on chemistry — and maintain cell-to-cell temperature uniformity within ±2°C.
Passive vs active cooling decision
The first question is whether passive cooling (conduction and natural convection to the environment) is sufficient, or whether active cooling is required. The answer depends on the ratio of heat generated to heat that can be passively rejected.
Heat generated per cycle = total I²R losses + reversible entropy term
For a first estimate at high C-rates where joule heating dominates:
Q_cycle ≈ I_pack² × R_pack × t_discharge (joules)
where R_pack = DCIR_cell / P (series resistance of the pack in the discharge path).
If the pack can absorb this heat within its thermal mass without exceeding maximum temperature, and dissipate it to ambient between cycles, passive cooling may suffice. If not, active cooling is required.
Passive thermal sizing check:
ΔT = Q_cycle / (m_pack × Cp_pack)
where Cp_pack ≈ 900–1100 J/kg·K for a lithium-ion pack.
If ΔT < (T_max_cell − T_ambient), passive cooling may be adequate. If not, specify active cooling.
Cooling architecture selection
The main options, in increasing order of cooling intensity:
| Cooling type | Typical application | Heat transfer coefficient |
|---|---|---|
| Passive (air, no fan) | Low-power BESS, portable | ~5–25 W/m²K |
| Forced air | HEV, light applications | ~25–100 W/m²K |
| Liquid — cooling plate (water-glycol) | EV, BESS | ~500–3,000 W/m²K |
| Direct liquid — immersion dielectric | High-power EV, fast charge | ~1,000–10,000 W/m²K |
| Refrigerant (active) | Fast charge, hot climates | ~3,000–15,000 W/m²K |
→ Battery cooling options → Cooling plate design → Thermal management overview
8. Ageing Margin
Ageing is not a post-design consideration — it is a sizing input. A pack sized to meet its requirements at beginning of life but not at end of life will fail in service.
Battery cells degrade in two ways:
- Capacity fade — available energy decreases
- Resistance growth — available power decreases and charging time increases
Both must be accounted for in the sizing:
Energy ageing margin: E_BOL = E_EOL_requirement ÷ SoH_EOL
If the pack must provide 60kWh at end of life and SoH_EOL is 80%: E_BOL = 60 / 0.80 = 75kWh
Power ageing margin: DCIR increases over lifetime — typically 25–100% growth from BOL to EOL depending on chemistry, operating conditions, and lifetime target. The peak power check in Step 5 must be performed with EOL DCIR, not BOL DCIR.
The key variables that drive ageing rate:
- Operating temperature — higher temperature accelerates SEI growth and electrolyte decomposition
- Usable SoC window — wide windows (high max SoC, low min SoC) accelerate ageing
- Charge and discharge rate — high C-rates increase mechanical stress and lithium plating risk
- Depth of discharge — deeper cycles cause more mechanical fatigue
The SoC window (Step 2) and the thermal management system (Step 7) are the two primary engineering levers for controlling ageing rate.
→ Cell lifetime and degradation → Calendar ageing → Cycle ageing
9. Iteration
The first pass through Steps 1–8 rarely produces an optimal design. The output is a feasibility check — does a pack meeting all requirements exist within the mass, volume, and cost constraints? If not, something has to give, and the sizing process identifies where.
Common iteration drivers:
S×P granularity misses the energy target. The combination of integer S and P counts means achievable pack energies are not continuous. A different cell (different Ah or V_nom) may hit the target more precisely.
Mass or volume budget exceeded. Options: higher energy density cell (NCMA over NMC), C2P architecture (reduce cell-to-pack ratio), or revisit whether the energy requirement can be reduced.
Thermal load exceeds passive cooling capacity. Either add active cooling, reduce the continuous power requirement, increase the P count (reduces per-cell current), or select a lower DCIR cell.
Peak power not met at EOL DCIR. Increase P count, select lower DCIR cell, or accept a derated power capability at cold temperature or end of life.
Cost target not met. Evaluate LFP vs NMC trade-off (lower cell cost but higher S count and pack mass), optimise cell purchasing volume, or revisit the architecture.
Worked Example: 75kWh Long-Range BEV Pack
Requirements:
- Usable energy at EOL: 60kWh
- Peak discharge: 200kW for 10s
- Continuous discharge: 80kW
- Nominal pack voltage: ~400V
- Lifetime: 10 years / 1,500 full equivalent cycles to 80% SoH
- Mass target: <500kg
- Operating temperature: −30°C to +45°C ambient
Step 1 — BOL usable energy with 20% ageing margin: E_BOL = 60 / 0.80 = 75kWh
Step 2 — Cell selection: NMC prismatic, 3.65V nom, 60Ah, DCIR_10s = 0.8mΩ, mass = 1.0kg
Step 3 — SoC window: 90% (long-range BEV, accept moderate calendar ageing trade-off)
Step 4 — S×P: S = 400 / 3.65 ≈ 109S → V_pack_nom = 109 × 3.65 = 398V P = 75,000 / (109 × 3.65 × 60 × 0.90) = 75,000 / 21,452 = 3.50 → 4P
Check: BOL usable energy = 109 × 3.65 × 4 × 60 × 0.90 / 1000 = 85.9kWh EOL: 85.9 × 0.80 = 68.7kWh ✓ (exceeds 60kWh requirement)
Total cells: 109 × 4 = 436 cells
Step 5 — Peak power check at EOL: V_pack_min = 109 × 2.5V = 272.5V I_peak = 200,000 / 272.5 = 734A I_cell = 734 / 4 = 183A = 3.1C on a 60Ah cell V_cell under load = OCV(SoC_min) − 183A × DCIR_EOL DCIR_EOL ≈ 0.8mΩ × 1.5 (50% resistance growth) = 1.2mΩ ΔV = 183 × 0.0012 = 0.22V — acceptable on a 2.5V cutoff
Step 6 — Mass: Cell mass = 436 × 1.0kg = 436kg Pack mass ≈ 436 × 1.55 (C2P architecture) = 675kg ← exceeds 500kg target
Iteration required: Either use a higher energy density cell (reducing cell count) or revisit the mass target. Using NCMA cells at 220Wh/kg (vs ~180Wh/kg NMC) would reduce cell count by ~18%, bringing pack mass into range.
Special Applications
Aircraft and Aerospace
Aircraft pack sizing is driven by gravimetric energy density above all other metrics. Weight is the binding constraint.
Key targets for electrified aviation:
- 400Wh/kg at system (pack) level — required for meaningful range
- High reliability and very low probability of thermal runaway
- Thousands of cycles for commuter and air taxi applications
- Very high peak C-rates during takeoff (typically 5–10C) with lower cruise C-rates
Current NMC/NCMA cells achieve 270–300Wh/kg at cell level; with a typical 50–55% cell-to-pack ratio for aerospace, 150–165Wh/kg at system level is achievable. The 400Wh/kg target requires either solid-state lithium metal cells or lithium-sulfur chemistry — neither of which is available at production scale for aviation today.
→ Electrified aircraft propulsion targets
Hybrid Battery Pack
A hybrid battery pack uses two different cell types — typically a high-energy cell and a high-power cell, or a battery and supercapacitor — to provide a broader set of capabilities than either type alone can offer.
Common use cases:
- Energy + power: NMC energy cells for range, LTO power cells for regen and acceleration peaks
- Hot and cold: a cell chemistry with better cold performance (LTO) in parallel with a higher energy chemistry for ambient conditions
- Fast charge + high energy: supercapacitor for rapid charge acceptance during regeneration, battery for stored energy
The BMS must manage both chemistries simultaneously, which adds significant complexity to balancing, protection limits, and SoC estimation.
Quick Reference: Sizing Formulas
| Quantity | Formula |
|---|---|
| Series count | S = V_pack_nom ÷ V_cell_nom |
| Parallel count | P = E_BOL_usable ÷ (S × V_cell_nom × Ah_cell × SoC_window) |
| Pack nominal voltage | V = S × V_cell_nom |
| Pack usable energy (kWh) | E = S × P × V_cell_nom × Ah_cell × SoC_window / 1000 |
| BOL energy from EOL req. | E_BOL = E_EOL ÷ SoH_EOL |
| Peak discharge current | I = P_peak ÷ V_pack_min |
| Cell current | I_cell = I_pack ÷ P |
| Cell heat generation | Q = I_cell² × DCIR (watts) |
| First-order pack mass | m_pack ≈ S × P × m_cell × 1.6 |
| Cell temperature rise | ΔT = Q_cycle ÷ (m_pack × Cp_pack) |
Related Pages
- Data required to size a pack
- Inputs to pack sizing
- Usable SoC window
- Cell selection starts with the pack, not the datasheet
- Series and parallel
- Cell capacity and pack size
- Variability in battery pack capacity
- Higher voltage packs
- Peak vs continuous power
- Continuous current rating
- Battery power demand solution
- Pack mass
- Pack benchmarking
- Battery cooling options
- Cell lifetime and degradation
- Battery calculators
- Battery Pack overview
- Key pack metrics
- Hybrid battery pack
- Electrified aircraft propulsion targets
- Pack Sizing and Requirements
- Example Pack Sizing using Power Demand
See also: Propane to Electric