Formula sheet

Battery Energy Storage System Sizing Formula Sheet

Battery energy storage formulas for service definition, power rating, usable energy, SOC evolution, duration, reserve, C-rate, inverter kVA, thermal loss, degradation, availability, and validation.

Branch: Energy Engineering
Content: Formula sheet
Updated: Jun 20, 2026
Revision: v1.1.0 · reviewed

This formula sheet collects first-pass calculations for battery energy storage system sizing. Use it for screening, concept design, design-review checks, commissioning tests, and operating review. Detailed design still requires cell and system data, thermal limits, battery management system constraints, degradation models, protection review, fire-safety review, inverter capability, grid-code requirements, site integration, and vendor-specific warranty limits.

State the service before sizing. A BESS designed for frequency response, backup power, solar shifting, peak shaving, black-start support, microgrid resilience, grid-forming operation, or demand response can have a different power rating, energy duration, reserve policy, ramp-rate requirement, cycling profile, and degradation budget.

Sign Convention and Boundary

Define the sign convention before using any formula. A common convention is:

$P_{dis}>0$ when the battery discharges to the load or grid;
$P_{ch}>0$ when the battery charges from the grid, renewable source, or local bus;
$E$ is stored energy inside the usable battery boundary;
AC metered energy includes inverter and transformer effects if the boundary is at the point of interconnection.

The boundary matters. Cell, rack, DC block, inverter AC terminals, transformer secondary, and point of interconnection can all produce different efficiencies, power ratings, and validation data.

Power and Energy

Power is the rate of energy transfer:

\displaystyle P=\frac{dE}{dt}

For approximately constant power:

E=P\Delta t

Approximate discharge duration:

\displaystyle t_{dis}=\frac{E_{usable}}{P_{dis}}

If discharge efficiency is included explicitly:

\displaystyle t_{load}=\frac{E_{usable}\eta_{dis}}{P_{load}}

where $P_{load}$ is the delivered load or grid power. Use usable energy, not nameplate energy, when evaluating a real service.

For sampled time-series data:

E=\sum_{k=1}^{N}P_k\Delta t_k

Use consistent units. If $P$ is in kW and $t$ is in hours, $E$ is in kWh.

Nameplate, Usable, and Contracted Energy

Nameplate energy is not the same as deliverable service energy. A practical usable-energy estimate is:

E_{usable}=E_{nameplate}(SOC_{max}-SOC_{min})f_{reserve}f_{age}f_Tf_{avail}

where:

$SOC_{max}-SOC_{min}$ is the allowed operating window;
$f_{reserve}$ accounts for protected emergency or operational reserve;
$f_{age}$ accounts for end-of-life capacity fade;
$f_T$ accounts for temperature or cooling derating;
$f_{avail}$ accounts for blocks unavailable due to maintenance or fault assumptions.

If a contracted service requires $E_{service}$ at end of life:

\displaystyle E_{nameplate}\geq \frac{E_{service}}{(SOC_{max}-SOC_{min})f_{reserve}f_{age}f_Tf_{avail}}

This is a screening expression. Real systems use manufacturer operating windows, warranty envelopes, battery management limits, and site-specific derating.

State of Charge

State of charge:

\displaystyle SOC=\frac{E_{stored}}{E_{rated}}

Percentage form:

\displaystyle SOC_{\%}=100\frac{E_{stored}}{E_{rated}}

SOC is usually estimated rather than measured directly. Estimation error matters because dispatch, reserve, safety functions, warranty compliance, and operator decisions depend on it.

Discrete SOC update for one time step:

\displaystyle E_{k+1}=E_k+\eta_{ch}P_{ch,k}\Delta t-\frac{P_{dis,k}\Delta t}{\eta_{dis}}

The corresponding SOC update is:

\displaystyle SOC_{k+1}=\frac{E_{k+1}}{E_{rated}}

Enforce operating limits:

SOC_{min}\leq SOC_k\leq SOC_{max}

If a simulation violates these limits, the battery is undersized, the dispatch policy is infeasible, or the reserve assumption is inconsistent.

Depth of Discharge

Depth of discharge:

DOD=1-SOC

For a cycle from high SOC to low SOC:

DOD_{cycle}=SOC_{high}-SOC_{low}

Deeper cycling usually increases energy delivered per cycle but can accelerate degradation. The best operating window depends on chemistry, temperature, C-rate, dwell time at high SOC, calendar aging, cycle count, and warranty constraints.

Power Rating

Required discharge power:

P_{dis,rated}\geq P_{service,max}

Required charge power:

P_{ch,rated}\geq P_{charge,max}

Include headroom for auxiliary loads, inverter derating, thermal derating, low or high SOC limits, aging, and reactive power requirements. A battery that has enough energy may still fail a service if it cannot deliver the required power at the required SOC and temperature.

For a power-limited service with ramping:

\displaystyle R=\frac{\Delta P}{\Delta t}

where $R$ is ramp rate. Frequency response and grid-support services may be constrained by ramp rate as much as by steady-state power.

Backup and Ride-Through Sizing

For a backup service:

\displaystyle E_{required}=\frac{P_{critical}t_{backup}}{\eta_{dis}}

where $P_{critical}$ is the supported critical load and $t_{backup}$ is required support duration.

If reserve must remain after the event:

\displaystyle E_{installed}\geq \frac{E_{required}}{(SOC_{start}-SOC_{min}-SOC_{reserve})f_{age}f_Tf_{avail}}

The start SOC must be enforceable operationally. A backup requirement is not satisfied if the battery is normally allowed to sit below the SOC needed for the event.

Peak Shaving

Energy required to shave load above a demand limit:

E_{shave}=\int_{t_1}^{t_2}\max(P_{load}(t)-P_{limit},0)\,dt

For sampled data:

E_{shave}\approx \sum_{k=1}^{N}\max(P_{load,k}-P_{limit},0)\Delta t_k

Required discharge power:

P_{BESS}\geq \max(P_{load}(t)-P_{limit})

Recovery charging should be checked:

P_{net}(t)=P_{load}(t)+P_{ch}(t)-P_{dis}(t)

Charging too aggressively after a peak event can create a second peak or violate an interconnection limit.

Renewable Energy Shifting

Energy available for shifting behind an export limit:

E_{shift}=\int_{t_1}^{t_2}\max(P_{renew}(t)-P_{local}(t)-P_{export,limit},0)\,dt

Delivered energy after round-trip losses:

E_{delivered}\approx E_{shift}\eta_{rt}

This estimate should be checked with time-series data. Renewable surplus, load demand, export constraints, curtailment rules, and battery SOC may not align in time.

Round-Trip Efficiency

Round-trip efficiency:

\displaystyle \eta_{rt}=\frac{E_{out}}{E_{in}}

If charge and discharge efficiencies are separated:

\eta_{rt}\approx \eta_{ch}\eta_{dis}

Only compare efficiencies with the same boundary. Cell-only efficiency, DC-block efficiency, inverter AC efficiency, transformer efficiency, HVAC auxiliary energy, and standby losses can produce different values.

Efficiency also depends on operating point. A system may have high efficiency near rated power but lower efficiency at low load, high temperature, long standby duration, or degraded cooling.

C-Rate

For current-based battery data:

\displaystyle C_{rate}=\frac{I}{Q_{Ah}}

where $I$ is current and $Q_{Ah}$ is ampere-hour capacity.

For energy-based screening:

\displaystyle C_{rate}\approx \frac{P}{E_{rated}}

where $P$ is in kW and $E_{rated}$ is in kWh, giving inverse hours. A 1C discharge approximately empties the rated battery in one hour. A 0.25C discharge takes about four hours.

High C-rate can increase heat generation, voltage drop, degradation, and protection stress. Final design should use manufacturer current, voltage, SOC, temperature, and duration limits rather than a single C-rate number.

Inverter Apparent Power and Reactive Support

For AC-connected systems:

\displaystyle S=\frac{P}{PF}

where $S$ is apparent power and $PF$ is power factor.

If active and reactive power are specified:

S=\sqrt{P^2+Q^2}

The inverter must be rated for the required active and reactive power combination. Providing voltage support can reduce active-power headroom at the same apparent-power rating:

P_{available}=\sqrt{S_{rated}^2-Q_{required}^2}

Grid-forming or voltage-support services also need checks for overload duration, fault current behavior, control mode, stability limits, protection coordination, harmonics, and grid-code compliance.

Thermal Loss Estimate

Simple loss estimate on an input-power basis:

P_{out}=\eta P_{in}

P_{loss}=P_{in}-P_{out}=P_{in}(1-\eta)

On an output-power basis:

\displaystyle P_{loss}=P_{out}\left(\frac{1}{\eta}-1\right)

If auxiliaries are included:

P_{site,loss}=P_{conversion,loss}+P_{HVAC}+P_{controls}+P_{standby}

Thermal management should be sized for realistic duty cycles, ambient conditions, degraded cooling, fire-zone constraints, emergency operation, and aging. Heat generation is not constant across SOC, current, temperature, resistance growth, and balancing behavior.

Degradation Allowance

End-of-life energy:

E_{EOL}=E_{BOL}(1-f_{fade})

where $E_{BOL}$ is beginning-of-life energy and $f_{fade}$ is fractional capacity fade.

If the service requires $E_{service}$ at end of life:

\displaystyle E_{BOL}\geq \frac{E_{service}}{1-f_{fade}}

Equivalent full cycles can be approximated from discharged energy:

\displaystyle N_{eq}\approx \frac{\sum E_{discharged}}{E_{rated}}

This simplified metric does not replace a degradation model. Calendar aging, cycle count, depth of discharge, C-rate, temperature, dwell SOC, rest time, and warranty definitions all affect capacity and resistance.

Availability and Redundancy

If $N$ identical blocks are installed and $k$ blocks are required for the service, the service is available only when at least $k$ blocks can operate. Simple spare capacity is:

P_{spare}=P_{installed}-P_{required}

E_{spare}=E_{installed}-E_{required}

Spare capacity is useful only if it remains accessible in the relevant failure state. Common cooling systems, shared controls, common transformers, fire-zone separation, communication dependencies, and protection trips can remove more capacity than the failed block alone.

For an $N+1$ energy check:

E_{installed}-E_{largest\ block}\geq E_{required}

For an $N+1$ power check:

P_{installed}-P_{largest\ block}\geq P_{required}

Measurement Reconciliation

Energy balance over an event:

E_{in}-E_{out}-E_{loss}-\Delta E_{stored}=0

For AC validation, reconcile metered AC energy, estimated SOC change, auxiliary energy, transformer loss, and standby loss within measurement uncertainty.

SOC-based energy estimate:

\Delta E_{stored}=E_{rated}(SOC_{end}-SOC_{start})

This estimate is only as good as the SOC estimator and the current capacity estimate. Commissioning should compare SOC-based energy with calibrated meters over defined charge and discharge events.

Screening Example

Suppose a site must support a 500 kW critical load for 2 hours, with discharge-path efficiency of 0.92. The required battery-side energy is:

\displaystyle E_{required}=\frac{500\times2}{0.92}=1087\ \text{kWh}

If the system must deliver this at end of life with a usable SOC window of 80 percent, 10 percent protected reserve, 85 percent end-of-life capacity factor, and no availability derating:

\displaystyle E_{nameplate}\geq \frac{1087}{0.80\times0.90\times0.85}=1776\ \text{kWh}

This does not yet include site-specific thermal derating, block redundancy, inverter limitations, future load growth, or fire-safety separation. The result is a screening size, not a procurement specification.

Validation Checklist

Useful validation checks include:

delivered active power versus dispatch command;
reactive power capability at the requested active power;
energy delivered versus SOC change;
round-trip efficiency over the defined boundary;
reserve preserved after normal dispatch;
ramp-rate response and control stability;
thermal response during high-power events;
protection and interlock behavior;
harmonic and power-quality performance;
degradation trend versus duty cycle;
availability under planned maintenance or single-block outage;
measurement uncertainty and meter reconciliation.

Common Mistakes

Common mistakes include sizing only by energy and ignoring power rating, using nameplate energy instead of usable energy, ignoring SOC reserve, assuming one battery can provide every service at once, and forgetting that reactive power can consume inverter apparent-power headroom.

Another frequent mistake is treating round-trip efficiency as a fixed number. Boundary, load level, temperature, auxiliary loads, standby duration, SOC window, and degradation all affect measured efficiency.

Battery sizing is credible only when the required service, operating envelope, control policy, protection limits, and validation method are stated together.

REF

Disciplines