Microgrid Battery Sizing (kW/kWh + PF + Inductive Loads)

Microgrid battery sizing starts from the load: you size kW and kWh to keep the site stable in both grid-parallel and islanded operation, while meeting ride-through and economic goals.

This page gives a practical sizing method for hybrid microgrids (grid + PV + genset + BESS) where power factor (PF) and inductive loads like motors, air compressors, and pumps are common. It also explains why “perfectly matching battery power to average load” is a design trap.

Define the load envelope (not a single number)

Start with a load profile that is granular enough to capture what actually causes interruptions and generator stress.

Critical load vs non-critical load (segment them early; continuity cost is not uniform).
Base load (kW) and peak load (kW), plus step changes (ΔkW) from motor starts and process events.
Power factor (PF): assume 0.9 for planning, but validate whether some operating states drop toward 0.8.
Permissible interruption window: minute-level (black start), ~20 ms class (STS), or 0 ms (online UPS).

Convert the electrical reality: kW vs kVA, PF headroom

If your loads include induction motors, compressors, pumps, or VFD-driven equipment, kW alone is not the full story. Your inverter and switching devices experience kVA demand. A design that “matches kW” may still be underpowered when PF dips or reactive demand spikes.

Rule of thumb: for planning, treat PF=0.9 as typical and test PF=0.8 as a stress case.
kVA ≈ kW / PF. Example: 500 kW at PF 0.9 is ~556 kVA; at PF 0.8 it is 625 kVA.
Battery power (kW) should include redundancy to maintain voltage/frequency during transients—especially for islanded operation.

Size battery power (kW): stability and transients first

In microgrids, battery kW is often sized for stability and disturbance response before it is sized for energy shifting. That’s why “perfect match” tends to fail: it ignores transients and control headroom.

Islanded stability margin: keep enough inverter headroom to regulate bus voltage/frequency under the worst credible step change.
Inductive load starts: support inrush/acceleration events (motors, compressors, pumps).
Genset coordination: battery kW should be able to absorb or supply power to keep the genset within a healthy loading band.
Grid support: battery kW should cover the site’s fast fluctuations so weak-grid events don’t cascade into PV trips or process stops.

Size battery energy (kWh): separate ‘PV-covered hours’ from ‘non-PV hours’

For hybrid microgrids, a practical energy-sizing method is to split the day into: (1) hours when PV can carry a meaningful portion of the load, and (2) hours when PV cannot (night / low-irradiance / curtailment windows).

A simple workflow (aligned with your stated approach):

Start from the load side: define daily energy demand (kWh) for critical + non-critical segments.
Define a PV coverage goal: e.g., maximize PV contribution to load where economically sensible and where the site can absorb it.
Use site irradiance / PV production data to estimate PV energy by hour (typical day, seasonal day, worst month).
Size battery kWh so the microgrid can shift PV energy into the non-PV hours and maintain the SOC reserve policy.

Define the SOC reserve policy (don’t bury it)

SOC reserve is not an afterthought; it is the contract between reliability and economics. Your EMS will enforce reserve to guarantee ride-through and continuity behavior.

Reserve for ride-through / transfer: enough energy to cover the time window required by your continuity strategy and genset start sequence.
Reserve for critical loads: keep a guaranteed minimum SOC so the most sensitive loads remain protected.
Economic SOC band: the remaining SOC window is used for PV shifting, diesel optimization, and grid import limit/peak shaving.

Common sizing mistakes (and how to avoid them)

Sizing kW to average load instead of worst credible transient + PF stress case.
Sizing kWh from ‘days of autonomy’ without separating PV-covered vs non-PV hours.
Ignoring load segmentation (everything treated as critical, driving unnecessary cost).
No explicit reserve policy: the system either under-delivers reliability or leaves money on the table.

Quick checklist for procurement-ready sizing inputs

Load profile: 15-min or 1-min resolution (at minimum), plus a list of major inductive loads and start behavior.
PF range: typical and worst case (0.9 typical, validate 0.8 scenarios).
Continuity target: black startvs STS vs online UPS, and which load segment requires it.
PV profile: hourly production estimate by season and inverter coupling constraints.
Genset data: rated power, minimum loading guidance, fuel curve points if available.
Grid constraints: import limit, transformer capacity relief requirement, voltage/frequency stability issues.

FFD POWER Note

FFD POWER typically sizes hybrid microgrids from the load backward, separating power (kW/kVA) requirements from energy (kWh) requirements.

Example – compressor-driven factory load
A factory operates 10 × 37 kW air compressors from 08:00–18:00 (10 hours). The nameplate active load is:

P ≈ 10 × 37 kW = 370 kW

1) Power sizing (kW / kVA): why 370 kW is not enough

For motor/compressor sites, the battery PCS (inverter) must cover not only kW but also power factor (PF) and transient behavior (starts, step loads, short overloads).

If PF ≈ 8, the apparent power is:

S ≈ 370 / 0.8 = 462.5 kVA
- Adding practical redundancy for motor dynamics and continuity margin (e.g., step changes / start assist / short overload headroom), the system is commonly tuned toward:
~500 kW-class battery PCS power(order-of-magnitude target)

This prevents “paper sizing” that looks fine in kW but fails under real motor conditions (voltage dip, overload trips, or EMS reserve being consumed too quickly).

2) Energy sizing (kWh): why “10 hours = 3700 kWh” is usually too crude

A brute-force backup calculation would be:

E_brutal ≈ 370 kW × 10 h = 3,700 kWh

But in most hybrid microgrids, PV output overlaps strongly with the factory’s operating window (08:00–18:00). That overlap means the battery does not need to supply the full load for 10 hours. Instead, the battery mainly covers:

morning ramp before PV ramps up,
PV fluctuations (cloud transients),
late afternoon drop-off,
short PV deficits vs load,
and the configured SOC reservefor ride-through / continuity targets.

So the required battery energy can be much lower, depending on:

local solar resource (irradiance profile),
PV installed capacity (kWp),
allowable curtailment / load shedding policy,
and the chosen SOC reserve strategy.

In many regions and with reasonable PV sizing, a 2–4 hour battery (at the target discharge power) can already achieve daily energy balance and continuity objectives:

2 h @ 370 kW → ~740 kWh
4 h @ 370 kW → ~1,480 kWh

Then EMS is tuned with SOC reserve bands and priority rules (critical vs non-critical loads, compressor staging logic, generator start thresholds if applicable) to match the site’s continuity requirement while reducing upfront CAPEX versus an oversized “10-hour battery.”

FAQ

How do I choose kW vs kWh for microgrid battery sizing?

Size kW first for stability, PF headroom, and transients (motors/steps). Size kWh next based on the energy you need to shift (PV-covered hours vs non-PV hours) plus the SOC reserve policy.

What PF should I assume in early-stage sizing?

PF 0.9 is a reasonable planning default, but you should stress-test 0.8 if the site has significant inductive loads or known PF issues.

Why is matching battery kW to the load a bad idea?

Because microgrids must handle kVA demand, transient steps, and control headroom. A ‘perfect match’ often fails during motor starts, PF dips, or islanded stability events.

How does PV affect battery kWh sizing?

Battery kWh is largely determined by how much PV energy you want to store and shift into non-PV hours, plus the reserve needed for ride-through and contingencies.

Do I always need a generator in a hybrid microgrid?

Not always, but in weak-grid or high-reliability sites a genset can provide long-duration backup. In many designs the battery enables efficient genset operation rather than replacing it.