Technical publication · Clean utilities and facility fit

WFI system sizing calculation: generation capacity and storage-tank volume

Abstract. This method note converts a time-resolved point-of-use schedule into a deterministic Water for Injection inventory balance. It calculates effective and minimum nominal generation rates, peak bucket withdrawal, the critical cumulative deficit, required usable storage, operating inventory including reserve, installed tank margin and the minimum simulated level. A synthetic 24-hour demand ledger, interactive calculator and explicit model limits make the result reproducible without presenting a scheduling screen as loop-hydraulic design, water-quality qualification or GMP validation.
Acatian Engineering27 min technical publication
Synthetic WFI generator and storage sizing balance A nominal 2.5 cubic metre per hour WFI generator at 90 percent deterministic availability supplies 2.25 cubic metres per hour. A 24-hour demand schedule totals 43.5 cubic metres and requires 9 cubic metres usable storage plus a 4 cubic metre reserve. A 15 cubic metre operating maximum leaves 2 cubic metres of storage margin. TIME-BUCKETED INVENTORY BALANCE · SYNTHETIC 24 h BASIS WFI GENERATOR2.50 m³/h× 0.90 = 2.25 EFFECTIVE 15.0 m³MAX OPERATING INVENTORY 4.0 m³ RESERVE PEAK BUCKET6.0m³/hNOT PUMP DUTY TOTAL DEMAND43.5 m³ / 24 h USABLE STORAGE NEEDED9.0 m³ INSTALLED MARGIN+2.0 m³ CRITICAL DEFICIT 07:00–13:00 · MINIMUM SIMULATED INVENTORY 6.0 m³
Total demand43.5 m³ Effective generation2.25 m³/h Required operating inventory13.0 m³ Installed storage margin+2.0 m³
Synthetic deterministic screen. Point-of-use timing, usable tank range, generator behavior, reserve policy and time resolution remain project-specific evidence inputs.
01

Scope and decision question

Can one generator–tank combination serve the declared WFI schedule without crossing the reserve?

Question answered

Given withdrawal volume in each time bucket, nominal generation rate, an explicit deterministic derating factor, maximum operating inventory, starting inventory and a protected minimum reserve, what generation and storage capacity does the schedule require—and when does the controlling deficit occur?

Why a time profile matters

Industrial high-purity-water demand is variable and schedule dependent. Riedewald, Byrne and Cronin modelled discrete point-of-use events together with continuous tank level specifically because an aggregate diversity factor can miss dynamic constraints [2].

Regulatory boundary

WHO guidance asks the design basis to consider required quantity and an optimum generator size or variable control that avoids excessive start–stop cycling, while treating generation, storage and distribution as an integrated system [5]. This calculation addresses quantity only; it does not establish WFI quality.

Nomenclature

Separate schedule volume, operating inventory and hydraulic flow

On small screens, swipe horizontally to compare the symbol, SI basis and required evidence.

SymbolDefinition and SI basisRequired evidence
DkTotal WFI withdrawn during bucket k, m³Point-of-use event ledger with start, duration, volume and simultaneous users
ΔtUniform calculation-bucket duration, hResolution short enough to preserve every material demand peak
Gnom, aNominal generator rate, m³ h−1, and deterministic availability fraction, dimensionlessQualified operating range, recovery, sanitation, maintenance and turn-down policy
Vmax,opMaximum permitted operating inventory, m³—not geometric vessel volumeUsable level range after headspace, heel, instruments and control limits
VresProject-defined minimum operating reserve, m³Approved continuity and control policy; not a universal GMP allowance
BreqLargest contiguous cumulative withdrawal deficit, m³Calculated from the declared schedule and effective generation rate
02

Point-of-use demand matrix

Build the demand series from events—not from one daily total

1 · Capture each event

For formulation, media and buffer preparation, final rinse, CIP/SIP support and other authorized uses, retain point of use, water grade, batch or campaign, start window, duration, withdrawal volume, local flow requirement and recurrence. Keep process demand distinct from loop recirculation return.

2 · Resolve simultaneity

Allocate each event across uniform buckets without erasing overlaps. A one-hour bucket containing 6 m³ reports an average 6 m³ h−1 for that interval; a 6 m³ draw completed in 15 minutes has a different local peak. Shorten Δt or retain an event-level hydraulic case before specifying pumps or pipework.

3 · Preserve scenarios

Create an approved base schedule plus credible campaign, maintenance, sanitation and expansion cases. The peer-reviewed comparison found deterministic modelling suitable as a first screen in many applications, then used Monte Carlo only where volume and schedule uncertainty justified it [2].

03 · Interactive engineering calculator

Simulate the WFI inventory before selecting generator and tank capacity

Deterministic time buckets · SI units · local browser calculation
Demand schedule
Generation basis
Tank operating range
Replace every synthetic value with an authorized point-of-use and equipment basis.

Worked-example inputs loaded. Select “Calculate WFI capacity” to reproduce the published result.

Total scheduled demand43.50 m³24 one-hour demand buckets Average demand1.81 m³/hΣDk / (nΔt) Effective generation2.25 m³/haGnom Minimum nominal generator2.01 m³/hMean-demand sustainability only Peak bucket withdrawal6.00 m³/hHour 07:00–08:00; not pump duty Required usable storage9.00 m³Critical deficit 07:00–13:00 Required operating inventory13.00 m³Usable swing + declared reserve Installed storage margin+2.00 m³11.00 m³ usable operating range Minimum simulated inventory6.00 m³At 13:00 Reserve margin+2.00 m³No reserve breach in this schedule Curtailed generation8.50 m³Top-level control or overflow-equivalent Deterministic screenPassMean generation, storage and start level pass

Interpretation. The synthetic schedule withdraws 43.50 m³ over 24 h. A 2.50 m³/h nominal generator at 90% deterministic availability supplies 2.25 m³/h, above the 2.01 m³/h nominal rate required by mean demand. The controlling 07:00–13:00 deficit requires 9.00 m³ usable storage; adding the declared 4.00 m³ reserve gives 13.00 m³ required operating inventory. The installed 15.00 m³ maximum leaves 2.00 m³ usable-storage margin, and the declared start level produces a 6.00 m³ minimum. This pass applies only to the entered deterministic quantity balance.

04

Equations and calculation workflow

Size sustained generation and transient storage as two separate constraints

1 · Effective supply and mean demand

Geff = aGnom. Over n equal buckets, Qd,avg = ΣDk/(nΔt) and Gnom,min = Qd,avg/a. If Geff is below average demand, stored inventory can postpone but cannot remove the long-run shortfall.

2 · Time-step inventory

Vk+1 = min[Vmax,op, max(0, Vk + GeffΔt − Dk)]. The upper clamp records curtailed generation; any V below Vres is a reserve breach. This assumes supply and demand can be represented by their net volume in each bucket.

3 · Required storage swing

Define Cb = Σk=1…b(Dk − GeffΔt). Then Breq = maxb[Cb − min0≤a<bCa], the greatest contiguous cumulative deficit after any refill opportunity within the entered chronological horizon. Required operating inventory is Vop,req = Vres + Breq.

4 · Peak bucket signal

Qbucket,peak = max(Dk)/Δt. Petrides and colleagues distinguish instantaneous withdrawal, interval-average demand and cumulative demand when screening loop, generator and tank capacity [1]. This calculator reports only the bucket-average signal; loop flow, pressure and return hydraulics require a separate model.

05

Reproducible worked example

From 24 point-of-use buckets to a 13 m³ operating-inventory requirement

Declared schedule basis

The synthetic series contains 24 one-hour withdrawals totalling 43.50 m³. Small background uses are interrupted by illustrative overlapping process and cleaning draws of up to 6.00 m³ in one bucket. These values are calculation fixtures, not recommended pharmaceutical-facility demand allowances.

Declared equipment basis

Nominal generation is 2.50 m³ h−1 and the deterministic derating factor is 0.90, giving 2.25 m³ h−1 effective supply. Maximum operating inventory is 15.00 m³, minimum reserve is 4.00 m³ and the simulation starts at 13.00 m³. Geometric tank capacity is deliberately not inferred.

Decision result

The mean-demand constraint requires at least 2.0139 m³ h−1 nominal generation at the declared factor. The 07:00–13:00 cumulative deficit is 9.00 m³, so reserve-inclusive operating inventory is 13.00 m³. The selected range supplies 11.00 m³ usable storage and a +2.00 m³ margin.

Download the synthetic WFI demand and calculation ledger CSV

Calculation ledger

Every capacity statement retains its time and inventory basis

On small screens, swipe horizontally to inspect each substitution.

QuantitySubstitutionResult
Effective generation2.50 × 0.902.25 m³ h−1
Average demand43.50 / 241.8125 m³ h−1
Minimum nominal generation1.8125 / 0.902.0139 m³ h−1
Critical cumulative deficitΣ(Dk − 2.25 × 1 h), 07:00–13:009.00 m³ usable storage
Required operating inventory9.00 + 4.00 reserve13.00 m³
Installed usable range15.00 − 4.0011.00 m³; +2.00 m³ margin
Simulated minimummin(V0…24) from V0 = 13.006.00 m³ at 13:00; +2.00 m³ above reserve
Top-level curtailmentΣmax(0, Vk + 2.25 − Dk − 15.00)8.50 m³ over the horizon

Generator–storage trade-off

The same schedule can move capacity between generation and storage

Each deterministic scenario retains the 24-hour demand series, 0.90 derating factor and 4.00 m³ reserve. It changes nominal generator rate only; none is a vendor recommendation.

On small screens, swipe horizontally to compare generator and storage requirements.

Nominal generatorEffective rateMean-demand statusUsable storage requiredReserve-inclusive inventory
2.00 m³/h1.80 m³/hFail: −0.30 m³/24 h13.10 m³17.10 m³
2.25 m³/h2.025 m³/hPass: +5.10 m³/24 h10.35 m³14.35 m³
2.50 m³/h · worked basis2.25 m³/hPass: +10.50 m³/24 h9.00 m³13.00 m³
3.00 m³/h2.70 m³/hPass: +21.30 m³/24 h6.30 m³10.30 m³

Method rejection and limitations

Reject a quantity pass that hides quality, hydraulics or uncertainty

Failure modeWhy the result failsRequired control
Daily total used as a flat loadIt erases simultaneous point-of-use withdrawals and the controlling cumulative deficit.Build an event ledger and choose Δt below the shortest decision-relevant demand interval.
Peak bucket called pump dutyA bucket average can be lower than the instantaneous local flow and ignores return-loop hydraulics.Model event flows, concurrent branches, loop recirculation, pressure drop and control valves separately.
Availability factor hides an outageA constant derating does not reproduce a consecutive generator shutdown or sanitization window.Build a separate time-varying supply scenario with explicit zero-generation intervals and approved recovery/start-up logic; this calculator accepts one constant effective rate.
Deficit crosses the schedule boundaryThe storage calculation evaluates contiguous deficits only within the entered order; an arbitrary start can split a controlling block between the end and beginning of a repeating cycle.Start at a documented full/refill boundary, or enter two consecutive cycles and assess the second cycle with a justified starting inventory.
Operating inventory called tank volumeHeadspace, heel, vortex protection, sensor span, thermal expansion and level-control bands change geometric capacity.Translate the working range through equipment geometry and supplier design data.
Curtailed volume treated as harmlessFrequent top-level control, recycling or dumping may conflict with generator turn-down, water use and microbial-control strategy.Verify minimum stable rate, recirculation, start–stop sequence, sanitization and disposal basis.
Deterministic pass hides uncertaintyOperator timing, batch slippage and withdrawal volume can shift the critical overlap. Fuzzy and stochastic high-purity-water models were developed for this uncertainty [3].Stress credible shifts first; use Monte Carlo or another justified uncertainty model when decisions remain sensitive.
Capacity pass presented as WFI qualificationThe balance says nothing about generation technology, chemistry, endotoxin, microorganisms, materials, sanitization or sampling.Use the applicable pharmacopoeia, quality system, risk assessment, qualification and ongoing monitoring. FDA inspection guidance likewise treats system design, validation and microbiological control as separate evidence [7].
07

Facility-fit checklist

Minimum evidence before approving the WFI capacity basis

Demand and schedule

Every point of use; correct water grade; event start, duration and volume; local flow; batch and campaign link; simultaneous users; cleaning and sanitation windows; maintenance; expansion cases; time-zone and calendar basis; low/base/high schedule; source and owner.

Generator and storage

Qualified output range; feed-water and recovery basis; ramp, start and stop logic; minimum stable production; sanitation downtime; redundant trains; maximum and minimum operating levels; headspace and heel; vent and overflow strategy; tank turnover; instrumentation and alarm policy.

Distribution and governance

Loop recirculation and return; simultaneous branch flows; pipe and pump curves; pressure at points of use; thermal or chemical control strategy; sampling and monitoring; URS; applicable water-quality specification; scenario approvals; model version; sensitivity; reviewer and equipment-vendor confirmation.

08 · Primary and authoritative sources

References

  1. Petrides, D.; Carmichael, D.; Siletti, C.; Koulouris, A. “Biopharmaceutical Process Optimization with Simulation and Scheduling Tools.” Bioengineering 1(4), 154–187 (2014). doi:10.3390/bioengineering1040154.
  2. Riedewald, F.; Byrne, E.; Cronin, K. “Comparison of Deterministic and Stochastic Simulation for Capacity Extension of High-Purity Water Delivery Systems.” PDA Journal of Pharmaceutical Science and Technology 65(4), 404–424 (2011). doi:10.5731/pdajpst.2011.00751.
  3. Riedewald, F.; Byrne, E.; Cronin, K. “A Fuzzy Logic Model of Deionised and Water for Injection Systems for Sizing and Capacity Assessment Under Uncertainty.” Journal of Pharmaceutical Innovation 6, 125–141 (2011). doi:10.1007/s12247-011-9108-4.
  4. Riedewald, F. Comparison of deterministic, stochastic and fuzzy logic simulation methods for capacity analysis of an industrial-scale water for injection system. University College Cork doctoral thesis. Open repository record and model appendices.
  5. World Health Organization. WHO good manufacturing practices: water for pharmaceutical use, WHO Technical Report Series No. 970, Annex 2 (2012). Official guidance PDF.
  6. European Medicines Agency. Guideline on the quality of water for pharmaceutical use, EMA/CHMP/CVMP/QWP/496873/2018, effective 1 February 2021. Current official guideline record.
  7. U.S. Food and Drug Administration. Guide to Inspections of High Purity Water Systems (7/93). Official inspection reference.
FAQ

Technical FAQ

Questions engineers ask about WFI generation and tank sizing

How do you size WFI generation capacity?

Build a time-resolved point-of-use demand series, then first require effective generation to cover mean demand over the repeating horizon. With a declared deterministic availability fraction a, minimum nominal rate is total scheduled volume divided by a times total horizon. Check recovery, turn-down, sanitation and explicit outage windows separately.

How do you calculate WFI storage-tank volume?

For a selected effective generation rate, calculate the largest contiguous cumulative deficit Σ(Dk − GeffΔt). That is the required usable storage swing. Add the approved minimum reserve to obtain required operating inventory, then convert that working range to geometric vessel capacity using headspace, heel, instruments and control limits.

Can average daily demand size a WFI system?

No. Average demand provides a long-run generator constraint but does not size the transient buffer. Clustered formulation, rinse or cleaning withdrawals can create a much larger short-duration deficit. Keep the point-of-use event schedule and test simultaneity explicitly.

Does peak bucket demand define WFI loop pump flow?

No. The peak bucket value is only average withdrawal over the selected time step. Pump and piping design require event-level instantaneous flow, continuous loop recirculation, concurrent branches, pressure drop, control-valve authority, return conditions and the applicable hygienic design basis.

When should WFI demand be modelled stochastically?

Start with deterministic low, base and credible stressed schedules. If the capacity decision remains sensitive to uncertain event timing or withdrawal volume, use a justified stochastic, Monte Carlo or fuzzy model with documented distributions, correlations, run count and acceptance probability.

Does this WFI sizing calculator validate the water system?

No. It is a deterministic quantity and schedule screen. It does not establish pharmacopoeial quality, select generation technology, design loop hydraulics, control microorganisms or endotoxin, qualify sanitization, set sampling limits or replace URS, risk assessment, commissioning, qualification and continued monitoring.

One commercial application · clearly separated

Carry WFI demand into the complete process, utility and facility schedule.

Acatian Professional connects point-of-use withdrawals, generation and storage constraints, CIP/SIP demand, process equipment, finite-capacity scheduling, utilities, TEA, LCA, source evidence and review history in one model. The current listed price is €590 per named seat per month, excluding applicable VAT. When eligible, checkout grants a seven-day trial; monthly billing starts after day seven unless cancelled beforehand. The order flow confirms eligibility, renewal date and recurring terms before payment; card details are handled by Stripe.