UPS Runtime Estimator.
— Estimator · UPS runtime
Runtime is not linear — model it.
Critical load, target runtime, ambient and chemistry in — recommended VA tier, battery string count and the Peukert-corrected runtime curve out. A planning figure, not a procurement spec.
Recommended UPS
30 kVA
online double-conversion · N+1 (one spare module)
Battery strings
1
48 kWh per string
Runtime @ 100%
108 min
vs 15 min target
Effective battery life
4.5 yr
at 28 °C ambient
A planning link that reopens this exact configuration — not a quote.
Verification status: An aspect is under engineering review · reviewed 2026-08-30
- tier
- Modular cabinet — 60 kVA modular (Vertiv APM / Schneider Galaxy VM)
- runtime @ 75%
- 144 min
- runtime @ 50%
- 217 min
- topology
- Online double-conversion (VFI)
- va headroom
- 25%
- discharge
- Peukert, chemistry exponent, nameplate read as the 1-hour rate
- rate limit
- Not modelled — check the cell's maximum continuous discharge
- inverter eff
- 94% DC→AC (double conversion)
- max dod
- 80%
- cell voltage
- 2 V
- cells / string
- 240
- module ah
- 100 Ah / string (scale by string count)
- life rule
- Halves per 8 °C above the 30 °C hot-zone
+ Model assumptions (9)− Model assumptions
Indicative — kVA, string count and runtime are concept-level planning figures, not a procurement BOQ. Two limits matter before you use the runtime figure. The discharge model treats the cell nameplate as its one-hour rate, where nameplates are conventionally quoted at a much slower rate, so runtime is optimistic at high discharge rates. And because strings are whole units, a small change in load can push the design onto an extra string and lift the quoted runtime well above the target you asked for. Production sizing uses the manufacturer’s published constant-power discharge tables, which is how IEEE 485 sizing is actually done, and takes cell ageing, charging-current limits, maximum continuous discharge rate and string-balancing into account.
What changes this estimate
- Measured load profile — peaks, ramps, harmonics
- Real battery-room ambient across summer
- Manufacturer runtime tables for the shortlisted unit
- Generator-start coordination window
Engineering notes
VRLA is the default for sub-5-minute critical-load UPS — well-understood, compact, replaced every 4–5 years.
Common enterprise tier — one extra UPS module so the bank survives one module fault without dropping load.
A planning link that reopens this exact configuration — not a quote.
Grid reality · Assam
How long the grid actually goes down here — from APDCL’s own numbers
APDCL publishes circle-wise reliability indices (SAIFI: interruptions per consumer per year; SAIDI: hours lost per consumer per year). For FY 2023-24, APDCL as a whole reported SAIFI 64.63 and SAIDI 119.43 hours — but the spread between circles is what UPS sizing should actually respond to. Nobody plans backup for an average.
Core urban Guwahati (GEC-I)
11.56 interruptions and 20.72 hours lost per year — roughly one sustained event a month, averaging about 1.8 hours. Bridging UPS with a 2-hour design event covers the published pattern; longer outages ride on DG.
District towns (Nagaon / Tezpur class)
Roughly 46–76 interruptions and ~86–126 hours per year — one to one-and-a-half events a week at 1.3–1.9 hours each. Plan at least 2-hour bridging, and treat March pre-monsoon (2–10× the January outage minutes in most circles) as the design season, not the average month.
Hard circles (Jorhat / Cachar class)
212–264 hours lost per year with mean events of 3.4–4.3 hours — Cachar reported 263.73 hours across 61.82 interruptions. Size for 4-hour events, and design the UPS as a DG-bridging stage, not the endurance layer.
Source: APDCL circle-wise reliability report FY 2023-24 and AERC-format quarterly standard-of-performance returns, published at apdcl.org, read 2026-08-30. Indices are consumer-weighted and discom-reported; LT-side and premises outages are excluded, so a single site can experience worse than its circle’s figure.
Power · UPS · IEC 62040-3
Critical load in. UPS VA, battery string count and the Peukert-corrected runtime curve out. The model accounts for power factor, ambient, chemistry and redundancy — the four variables that decide whether the bank holds for the promised minutes.
- Chemistries
- VRLA · LFP
- Topology
- Online VFI
- Redundancy
- N · N+1 · 2N
- Headroom
- 25%
Method
How this is calculated
Recommended UPS (kVA) = critical load (kW) ÷ power factor × 1.25 headroom, rounded up to the next 5 kVA step.
Runtime (min) = usable battery energy (kWh) ÷ load (kW) × 0.94 inverter efficiency × 60, then multiplied by a Peukert derate of (reference rate ÷ load)^(n−1) capped at 1 — the derate only ever removes runtime, never adds it.
Usable energy = string energy × strings per bank × 0.8 depth-of-discharge. Redundant banks add fault tolerance, never minutes: every runtime figure is one bank.
Worked example — Data hall · 100 kW · 15-minute hold · VRLA N+1
- Critical load:
- 100 kW at 0.9 PF
- Chemistry:
- VRLA (sealed lead-acid)
- Target runtime:
- 15 min at full load
- Ambient / redundancy:
- 30 °C · N+1
- 1 · Apparent power and headroom100 ÷ 0.9 × 1.25 = 138.9 kVA → 140 kVA standard step
- 2 · Energy in one string2 × 240 × 100 ÷ 1,000 = 48 kWh
- 3 · Energy-ceiling runtime48 × 0.8 ÷ 100 × 0.94 × 60 = 21.7 min
- 4 · Peukert derate applied× (38.4 ÷ 100)^(1.25 − 1) = 17 min at 100% load
Recommended UPS: 140 kVA · Strings per bank: 1 · Runtime 100 / 75 / 50%: 17 / 24 / 41 min
Load this example in the calculator: Data hall · 100 kW · 15-minute hold · VRLA N+1. Every figure above was computed by the same engine the UPS Runtime Estimator runs — the example cannot drift from the tool.
Constants and assumptions
- Reference-rate limitation: Overstates at high discharge ratesThe model treats the battery nameplate as the 1-hour rate. Cell nameplates are conventionally quoted at the 8- or 20-hour rate, so the Peukert reference is optimistic and runtime is overstated the harder the bank is discharged. This tool is an indicative planning estimate; production sizing uses the manufacturer's constant-power discharge table per IEEE 485.
- Peukert exponent (VRLA): 1.25chemistry constant; LFP is 1.05, which is why LFP holds its rating far better at high C-rate
- Depth of discharge: 80%recommended maximum for runtime sizing on VRLA
- Inverter efficiency: 0.94online double-conversion DC→AC, applied to every delivered-runtime figure
- Ambient: 30 °CVRLA hot-zone is 30 °C — at or below it no capacity derate applies; above it life halves every 8 °C
Field notes · why the model is shaped this way
Peukert's law isn't linear
A VRLA string rated 100 Ah at the 10-hour rate delivers only 60–70 Ah at a 1-hour rate. The model corrects for this; specifying a string at its nameplate rate is the most common UPS sizing mistake.
Heat is the silent killer
Battery life halves for every 8 °C above the chemistry hot-zone (30 °C for VRLA, 40 °C for LFP). A 35 °C summer battery room turns a 5-year VRLA bank into a 3-year bank.
LFP is not always right
For runtimes under 5 minutes, LFP is over-engineered — VRLA gives the same outcome at a smaller footprint and lower lifecycle complexity. The chemistry crossover is around 15 min of runtime.
N+1 over N for ≥ 100 kW
Above 100 kW critical load, a single module fault drops too much load to be acceptable. N+1 modular adds one extra UPS module to the bus — survives one fault without dropping load.
· Starting configurations
Typology presets — pick a scenario, see the calculator.
Each preset opens a curated configuration page with the engineering reasoning behind the numbers. Then the calculator loads with the same inputs — change them and the URL stays shareable.
· Engineering advisory · UPS Runtime
What this answer means for the deployment.
The runtime number is the start of the engineering conversation, not the end. The four notes below frame what the answer predicts about the building beyond the calculated minutes.
01 · Deployment observations
Deployment observations
The observations3 notes — engineering detail behind this section
- A 15-minute target runtime is a different deployment than a 30-minute target — the first is a graceful-shutdown window for IT loads, the second is a generator-start ride-through for mission-critical scope. Specify against the operational requirement, not the catalogue convenience.
- VRLA vs LFP is decided by the eight-year service-life picture, not the day-one hardware tier. Above 20 kVA new-install and above 15-minute runtime, LFP is the disciplined choice; below 10 kVA single-rack scope, VRLA remains defensible on a 5-year refresh cadence.
- The Peukert-corrected runtime curve is the answer the operator actually inherits — the nameplate runtime is the answer the catalogue advertises. The two diverge most at high C-rate / short-runtime points, which is where the design risk concentrates.
02 · Redundancy posture
Redundancy posture
Read the redundancy posture2 notes — engineering detail behind this section
- N+1 protects against a single module fault; 2N protects against a single distribution-path fault. The two answer different failure modes and the cost premium is not the same — N+1 sits at ~1.34× of N, 2N sits at 2.00×+.
- Battery redundancy is independent of UPS-module redundancy. A 2N UPS with a single battery bank shares a failure domain with the battery; 2N at the UPS layer requires twin battery strings on twin paths for the redundancy claim to be honest.
03 · Environmental considerations
Environmental considerations
Read the environmental considerations3 notes — engineering detail behind this section
- Ambient temperature is the single biggest derating factor for VRLA — every 10 °C above the rated 25 °C halves the battery's calendar life. An electrical room running at 35 °C in summer needs the runtime sized against that ambient, not against a 25 °C nameplate.
- Humidity and particulate load drive the UPS-room maintenance interval — high-RH or dusty environments need quarterly filter and connection inspections; standard offices can hold to half-yearly.
- LFP indoor installations are NFPA 855-compliant without bespoke fire suppression for LFP chemistry; NMC chemistry requires dedicated suppression for indoor banks. Chemistry choice carries a fire-and-life-safety implication, not only a cost one.
04 · Commissioning discipline
Commissioning discipline
Read the commissioning discipline3 notes — engineering detail behind this section
- Battery internal-resistance and capacity test at commissioning — the documented baseline lets every subsequent AMC visit catch degradation against the original measurement, not against the catalogue spec.
- Generator-start coordination with the UPS is a commissioning-stage exercise — the UPS rides the load for the generator-start window (typically 10–30 seconds), then transfers to generator. The transfer behaviour is rehearsed under load, not assumed from documentation.
- Configuration baseline (UPS settings, BMS integration map, alarm thresholds) is exported offline at commissioning and after every configuration change — recovery from a UPS controller fault is a same-day exercise against the saved baseline.
05 · Lifecycle implications
Lifecycle implications
Read the lifecycle implications2 notes — engineering detail behind this section
- VRLA banks need a planned full replacement every 4–5 years; LFP banks hold for 7–10 years. The AMC calendar carries the replacement window, not the failure event — reactive replacement is multiple times the cost of preventive.
- UPS module capacitors drift on a 10–12 year horizon — the annual ESR measurement on the AMC catches drift before it shows on the inverter's THD measurement.
06 · Expansion readiness
Expansion readiness
Read the expansion readiness2 notes — engineering detail behind this section
- Sizing with a 25% headroom on the calculated VA gives the deployment foreseeable expansion against an IT-load growth curve — the headroom is the difference between a 5-year and an 8-year refresh window.
- A modular UPS topology lets the deployment add capacity without re-cabling — single-module additions live within the existing distribution path; rack-and-paralleling is a configuration exercise, not a re-installation.
Verification status: An aspect is under engineering review · reviewed 2026-08-30
· Engineering notes
How to read this tool’s output
Stage 2 (survey) or stage 3 (design). Use during UPS sizing to validate chemistry choice and string count before procurement.
The assumptions, limits and reference architectures behind this
What an engineer should know
The assumptions behind the numbers this tool produces.
- Sizes online double-conversion UPS from critical-load kW, power factor, ambient and chemistry — Peukert-corrected runtime curve with VRLA and LFP.
- Applies 25% VA headroom; ambient temperature de-rates effective life via the Arrhenius rule (life halves every 8 °C above chemistry hot-zone).
When this tool is the right one
The project moments and room types this is built for.
- Mid-size IT room UPS sizing (20–60 kVA) for an office or hospitality property.
- Mission-critical UPS sizing for ICU, command operations and broadcast.
- Battery-chemistry crossover decision for an existing VRLA bank approaching end-of-life.
What changes the answer in practice
Field conditions that move the result away from the planning figure.
- Ambient above the chemistry hot-zone (30 °C VRLA, 40 °C LFP) halves life every 8 °C — battery rooms in the NE need dedicated cooling.
- Above 100 kW critical load, N+1 modular is typically the minimum — a single module fault drops too much load to be acceptable.
- Power factor below 0.85 indicates reactive load (motors, transformers) — UPS oversizing may be needed.
Defensible starting architectures
Vendor-neutral reference points, not a recommendation to buy — the right answer is the one that survives your site survey.
- VRLA below 15 min runtime, LFP above — crossover is around 15 min on lifetime cost.
- N+1 modular above 100 kW critical load; 2N redundant rooms for mission-critical contexts such as ICU, command operations or broadcast.
- Dedicated cooled battery room at 22 °C setpoint for tropical / NE deployments.
Common mistakes
Failure patterns we see on real projects.
- Specifying battery at nameplate Ah ignoring Peukert — VRLA loses 30–40% of nameplate capacity at 1-hour rate.
- Sharing ambient with the broader plant HVAC zone — NE / tropical sites drop battery life by ~30% if not actively cooled.
- Picking LFP at < 5 min runtime — VRLA delivers the same outcome at a materially lower hardware tier below the crossover.
How this lands against adjacent systems
What else has to be agreed before this output is safe to build to.
- Pairs with the BESS Sizer when BESS-augmented UPS is on the table.
- Pairs with the DG → Solar+BESS Crossover tool for the broader energy-strategy picture.
What this tool does not do
Deliberate limits — where the estimate stops and design begins.
- Indicative for brief-stage budgeting only — production sizing always uses manufacturer curves which account for cell ageing and BMS firmware.
- Refresh batteries on the conservative half of calculated life when ambient regularly exceeds the chemistry hot-zone.
Where this tool fits
The building types this output is calibrated for — and, where we have said so, the ones it is not.
Best suited for
Not the tool to reach for
The standards and technologies this touches
Reference pages for the protocols, standards and systems behind this tool’s output.
· Why the curve matters
A 100 kW critical load at 0.9 power factor, asked for 15 minutes on VRLA at 30 °C, sizes to 140 kVA on a single string per bank — and delivers 17 minutes at full load, not 15. That overshoot is the lesson: batteries come in whole strings, so the delivered autonomy lands where granularity puts it rather than on the number you asked for. The same bank runs 24 minutes at 75% load and 41 at 50%, which is why the honest design question is the load you will actually carry, not the nameplate. Redundancy here buys fault tolerance, not minutes: N+1 adds a spare UPS module rather than a second string, so the installed count stays at 1, and every runtime figure above is one bank on its own. Treat all of these as an upper bound — this model reads the battery nameplate as the one-hour rate, so it overstates at high discharge rates until the IEEE 485 constant-power table lands.
How to use the UPS Runtime
Step 1
Enter the critical load
Critical load in kW and its power factor — the tool sizes apparent power from these, not from a nameplate.
Step 2
Set the runtime target
Target minutes at full load. The tool iterates string count upward until the model actually meets the target rather than rounding to it.
Step 3
Pick chemistry and ambient
VRLA or LFP, and the battery-room ambient in °C. Ambient above the chemistry hot-zone derates capacity and halves life every 8 °C.
Step 4
Choose redundancy
N, N+1 or 2N. Redundant banks add fault tolerance, never minutes — every runtime figure quoted is a single bank.
Step 5
Read the runtime, then read its limit
Runtime at 100/75/50% load, with the method and its reference-rate limitation stated beside it. Use it for planning, not for a production sizing certificate.
· Frequently asked
UPS Runtime —
what people ask first.
Why online double-conversion only?
Line-interactive and offline UPS topologies transfer load to the inverter only when mains fail — the transfer takes 4–10 ms, which is enough to glitch sensitive IT loads. Online double-conversion always runs the load through the inverter, so battery-supported runtime starts from a stable bus. For critical-load sizing the difference matters; for less critical loads, the sizing logic still applies but the topology can be relaxed.
Why is VRLA so penalised at high discharge rates?
Lead-acid chemistry's Peukert exponent sits around 1.25 — a 100 Ah string rated at the 10-hour rate delivers maybe 60–70 Ah at a 1-hour rate. LFP sits at ~1.05, which is why it stays close to the spec sheet even at high C-rates. Specifying a battery at its nameplate capacity ignoring Peukert is the most common UPS sizing mistake; this tool corrects for it explicitly.
When does LFP overtake VRLA on lifetime cost?
Around 15 minutes of runtime and a 5-year horizon. An LFP bank carries a higher up-front tier than VRLA, but VRLA needs one full replacement at year 4.5 and the LFP bank runs ~10 years. Add the labour of a battery swap (which is non-trivial in an occupied building) and LFP wins on service life above 15 min runtime in most cases. Below 5 min runtime, VRLA wins comfortably.
What does N+1 actually mean?
N is the minimum number of UPS modules required to support the critical load. N+1 adds one extra module so the bus survives a single module fault without dropping load. 2N is two completely independent UPS rooms with twin distribution paths — used in mission-critical contexts like ICUs, command centres and Tier-III data centres. The cost premium of N+1 over N is ~34%; the premium of 2N over N is 100%+.
How accurate are the numbers?
Accurate enough for brief-stage budgeting, not for final procurement. Production sizing always uses the manufacturer's runtime curves, which account for cell ageing, charging-current limits, string balancing and BMS firmware. The tool's job is to get the budget conversation and the rough specification right; the BOM sizing is a different exercise.
Engineering toolkit
Next in this sequence
Calculators whose inputs or outputs connect to this one.
- BMS · BESS
BESS Sizer
Peak demand + solar + storage strategy in, indicative LFP battery size, inverter rating, daily throughput and cell life out. Vertiv, Delta and Fuji-class deployments.
LFP · kWh · cell lifeOpen - BMS · Energy
Energy & Efficiency Estimator
Adjust building variables and see indicative energy and CO₂ savings (kWh) for a BMS-driven energy upgrade plus daylight-harvesting LED retrofit. Conservative; defensible in a first conversation.
kWh · CO₂ savedOpen - Power · BESS · Solar
DG → Solar+BESS Crossover
DG capacity and monthly run-hours in — the kWp solar + kWh LFP BESS that displaces the genset's run-hours out, plus annual diesel litres displaced and CO₂ avoided.
kWp · kWh · yearsOpen
· Begin
Take the brief to a
UPS workshop.
A site walkthrough lets us load-profile the critical bus (peaks, ramps, harmonics) and validate the chemistry choice against your actual ambient — not just summer-mean.
Save or share this planning tool
Copy the link if you need a colleague to check this — broadcast sharing is withheld while the model is under review. Verification status: An aspect is under engineering review, reviewed 2026-08-30.