- Filter cartridge flow rate is not simply the LPM number on the manufacturer's catalog — that figure is the ideal value for "water, 20 °C, zero load"
- Real-world flow rate follows Darcy's law: Q = ΔP × A / (R × η). Double the viscosity and the flow rate is cut in half
- 70% ethanol has 4× the viscosity of water; syrup can be 1000× or more — never size a system using water data directly
- Five-step practical sizing workflow: calculate flow rate → look up viscosity → set acceptable ΔP → match against manufacturer flow curve → add 50–100% safety margin
- This article walks through three real cases (pharmaceutical 100 LPM, semiconductor 50 LPM, food syrup 200 LPM)
- "Insufficient flow" is the most common filter cartridge sizing failure
- Darcy's law in brief: flow rate depends on four factors
- The hidden viscosity trap: 70% ethanol is 4× slower than water
- Reading the pressure differential curve: when to change the cartridge
- Five-step sizing calculation workflow (with worked examples)
- Three real-world case comparisons
- Safety margin: why over-size by 50–100%
- Common mistake: just divide by the nominal flow rate?
- Frequently Asked Questions
- References
"Insufficient flow" is the most common filter cartridge sizing failure
The phrase every plant manager dreads on commissioning day: "Pressure won't build downstream of the cartridges, throughput is down 30%." Nine out of ten engineers immediately blame the cartridge — wrong pore size, bad brand, defective unit. In reality, most of the time the cartridge is fine; the original sizing was wrong.
The catalog says "10-inch PES 0.22 µm, clean water flow 30 LPM @ 0.1 bar" — and the engineer assumes the cartridge can deliver 30 LPM all day long. On the actual line running 70% ethanol, a 0.5 mPa·s buffer, or feed liquid containing 5 ppm of particles, the flow rate may immediately drop to 12 LPM, and within two hours pressure differential climbs to 1 bar requiring a cartridge change.
This article covers filter cartridge sizing from the theory (Darcy's law) through to practice (how to read manufacturer flow curves) in one go. At the end you'll see the calculation process for three real cases — pharmaceutical aqueous solution, semiconductor SC1, and food-grade syrup — each with completely different sizing logic.
Darcy's law in brief: flow rate depends on four factors
A filter membrane is essentially a porous medium covered with microscopic pores. Fluid flow through a porous medium obeys Darcy's law:
From this equation, four variables are within your control:
- Pressure differential ΔP: higher ΔP means higher flow, but the usable range is limited. Recommended initial ΔP is 0.1–0.2 bar, with terminal operating ΔP no higher than 0.7–1.0 bar. Anything higher compacts the filter cake and flow rate collapses.
- Filtration area A: switching from a 10-inch to a 20-inch cartridge (roughly 2× the area) almost doubles the flow; running an additional cartridge in parallel does the same. Adjusting A is the heart of sizing.
- Resistance R: includes membrane resistance (fixed) + cake and fouling resistance (rises over time). The more particulates and the more complex the feed, the faster R rises.
- Viscosity η: a property of the fluid itself, also strongly affected by temperature. Water is 1.00 cP at 20 °C, drops to 0.65 cP at 40 °C, and rises to 1.52 cP at 5 °C.
Of these four variables, viscosity and area are the two engineers most often underestimate — and viscosity is exactly the trap covered next.
The hidden viscosity trap: 70% ethanol is 4× slower than water
Take the same 10-inch PES cartridge, swap the fluid for ethanol, buffer, or syrup, and the flow rate can differ several fold. The table below lists the viscosities of common process fluids (20 °C):
See the pattern? 70% ethanol is 4× the viscosity of pure water. Same cartridge, same 0.1 bar pressure differential — running 70% ethanol delivers only about 1/4 of the water flow rate. If you use the water flow rate table to size an ethanol process, the line is guaranteed to choke.
Reading the pressure differential curve: when to change the cartridge
The cartridge "pressure differential curve" is the other key to sizing. From start-up to retirement, a normally operating cartridge passes through three phases:
- Phase 1 (stable): from start-up through small accumulated filtrate volume. ΔP holds around the initial value of 0.05–0.1 bar; the curve is nearly flat.
- Phase 2 (linear rise): filter cake gradually accumulates and ΔP rises slowly and linearly to 0.3–0.5 bar. This is where the cartridge is genuinely "earning its keep".
- Phase 3 (rapid blockage): when ΔP reaches initial + 0.7 bar (about 10 psi), the pores are heavily blocked and the curve climbs exponentially. This is the industry-consensus change-out point.
In practice there are two change-out signals:
Five-step sizing calculation workflow
Step 1 — Confirm process flow rate
Get the target flow rate from the process designer and standardize the units. Common conversions: 1 m³/h ≈ 16.67 LPM ≈ 4.40 GPM. Distinguish "peak" from "average" flow — sizing is generally based on peak flow rate.
Step 2 — Look up viscosity (at process operating temperature)
Not the standard 25 °C value, but viscosity at the actual process operating temperature. For mixed fluids (containing solvents, sugars, salts, biomass), measure directly or request data from the feed supplier. When data is unavailable, you can approximate aqueous solutions with a "weight-percent weighted average".
Step 3 — Set acceptable initial ΔP
Most liquid filter cartridge sizing sets initial ΔP at 0.1–0.2 bar (1.5–3 psi). Below 0.1 bar the flow is too slow; above 0.2 bar means insufficient capacity and very rapid blockage.
Step 4 — Match against the manufacturer's flow vs ΔP curve
Every cartridge manufacturer publishes a "water flow rate vs pressure differential curve" (in the manufacturer's data sheet on their website). Read off how much water a single 10-inch cartridge can deliver at your chosen ΔP, then convert per Darcy's law:
Step 5 — Add safety margin and decide cartridge count
Divide required flow by "actual flow per cartridge" to get the minimum cartridge count, then multiply by a 1.5–2.0 safety margin. Example: requirement 50 LPM, 7.5 LPM per cartridge → 50 / 7.5 = 6.67 → round up to 7 → add 50% margin = 11 cartridges.
Three real-world case comparisons
Case A — Pharmaceutical WFI sterile filtration, 100 LPM
| Item | Value |
|---|---|
| Process flow rate | 100 LPM peak |
| Fluid | WFI (water for injection) |
| Operating temperature | 20–25 °C |
| Viscosity | 1.0 cP |
| Cartridge selection | 10-inch PES 0.22 µm (hydrophilic, ~30 LPM single-cartridge water flow @ 0.1 bar) |
| Actual flow per cartridge | 30 LPM × (1/1) = 30 LPM |
| Cartridge count needed | 100 / 30 = 3.33 → 4 cartridges |
| With 50% safety margin | 4 × 1.5 = 6 cartridges (in practice, 4 + 1 standby) |
| Recommended configuration | 4 × 10-inch PES 0.22 µm in parallel housing |
Case B — Semiconductor SC1 (NH₄OH/H₂O₂) chemical filtration, 50 LPM
| Item | Value |
|---|---|
| Process flow rate | 50 LPM |
| Fluid | SC1 (NH₄OH:H₂O₂:H₂O = 1:1:5) |
| Operating temperature | 70–80 °C |
| Viscosity (70 °C) | ~0.5 cP (high temperature lowers viscosity) |
| Cartridge selection | 20-inch PFA-housed PTFE 0.05 µm (alkali- and heat-resistant) |
| Single-cartridge water flow | ~25 LPM @ 0.1 bar (smaller pore size, lower flow velocity) |
| Actual flow per cartridge | 25 × (1/0.5) = 50 LPM (lower-than-water viscosity actually speeds it up) |
| Cartridge count needed | 50 / 50 = 1 cartridge |
| With 100% safety margin | 2 cartridges (high-purity semiconductor applications justify a larger margin) |
| Recommended configuration | 2 × 20-inch PTFE 0.05 µm in parallel PFA housing |
Case C — Food 50% syrup filtration, 200 LPM
| Item | Value |
|---|---|
| Process flow rate | 200 LPM |
| Fluid | 50% syrup (fructose/sucrose) |
| Operating temperature | 50 °C (process heating to reduce viscosity) |
| Viscosity (50 °C) | ~6 cP |
| Cartridge selection | 30-inch PP depth filter cartridge 1.0 µm |
| Single-cartridge water flow | ~70 LPM @ 0.2 bar |
| Actual flow per cartridge | 70 × (1/6) = 11.7 LPM |
| Cartridge count needed | 200 / 11.7 = 17.1 → 18 cartridges |
| With 50% safety margin | 18 × 1.5 = 27 cartridges |
| Recommended configuration | 27 × 30-inch PP depth filter cartridges; raising temperature to 60 °C lowers viscosity and saves ~25% of the count |
Safety margin: why over-size by 50–100%
Even with the first four steps perfectly executed, a real production line still runs into many factors that "weren't in the spreadsheet":
Industry-consensus over-size ratios:
- Clean aqueous solutions (WFI, buffers): a 50% margin is enough
- Feeds with trace particulates (API solutions, pharmaceutical pre-treatment): 75–100% margin
- Feeds with suspended solids / high-fouling feeds (syrups, fermentation broths, oils): 100–200% margin, plus an upstream coarse pre-filter
- High-purity semiconductor chemicals: 100%+ margin + N+1 redundancy + endurance test required
Common mistake: just divide by the nominal flow rate?
Frequently Asked Questions
How do I look up the "resistance R" in Darcy's equation?
You don't need to compute R directly in practice. The manufacturer integrates R and A into the "flow vs ΔP curve" for you. You simply: (1) find the curve for the matching pore size and specification, (2) read off the flow at your target ΔP (typically 0.1 bar), (3) scale by the viscosity ratio for actual flow. For more rigorous academic calculations, see Coulson & Richardson, Chemical Engineering Vol. 2, Chapter 7 on filtration.
Why is the cartridge "terminal pressure differential" 0.7 bar rather than 1.0 or 2.0 bar?
It is an industry rule of thumb: when ΔP reaches initial + 0.7 bar (about 10 psi), the filter cake is compacted and the pores are nearly fully blocked. Pushing further only wastes electricity (the pump has to work harder) and risks cake breakthrough contaminating downstream. 0.7 bar is the optimal balance between flow, energy, and product quality. Some high-pressure designs allow up to 2 bar, but end-of-life flow is so low it's not worth it.
Can I just raise pump pressure to increase flow?
Within limits, but not recommended. Raising ΔP does initially boost flow (linear proportionality), but it accelerates cake compaction and shortens cartridge life. Excessive ΔP can also push deformable particles (protein aggregates, colloids) through the membrane and contaminate downstream. The correct answer is to add cartridges (increase area A), not to push ΔP harder.
Does swapping a 10-inch for a 20-inch cartridge exactly double the flow?
Almost. A 20-inch cartridge has roughly 2× the effective filtration area of a 10-inch (two stacked sections), so flow at the same ΔP is about 1.9–2.0×. A 30-inch is ~3×; 40-inch ~4×. Larger sizes save floor space and simplify piping, but each unit costs more to dispose of and is heavier to handle.
Is the flow vs ΔP curve linear?
The low-pressure region (ΔP < 0.3 bar) is nearly linear, consistent with Darcy's Q ∝ ΔP. The high-pressure region deviates from linearity — the membrane itself flexes slightly and pores are compacted by the feed. This is why sizing always operates in the linear region (0.05–0.2 bar): predictions stay accurate.
Do trace bubbles in the feed affect sizing?
Yes. Bubbles block part of the membrane area (especially on hydrophilic membranes that trap them), and effective flow can drop 20–40%. Install an inline degasser or degassing tank upstream of the cartridge; high-purity chemical lines typically have one as standard.
When multiple cartridges run in parallel, does flow split evenly?
Not necessarily. If the piping is asymmetric (different lengths, different fitting losses), each cartridge sees a different ΔP and some block early while others stay clean. Housings should use reverse-return or symmetrical manifold piping to equalize ΔP across cartridges; otherwise even a correct sizing calculation will under-perform in practice.
References
- Pall Corporation — Filter Selection and Sizing Guide
- Cytiva — Process Filtration Sizing & Application Notes
- Sartorius — Process Filtration Technical Documents
- Cobetter Filtration — Flow Rate & Pressure Drop Charts (semiconductor / pharmaceutical liquid cartridge sizing)
- Merck Millipore — Filtration Sizing Tools (Filter-V Plus)
- 3M Purification — Liquid Filtration Cartridge Selection Guide
- Coulson & Richardson — Chemical Engineering Vol. 2, Chapter 7: Filtration
- Engineering Toolbox — Dynamic Viscosity of Common Fluids
- SEMI Standards — F58 Liquid Filter Test Methods (semiconductor process cartridge testing standard)
- PMDA / FDA Guidance — Sterile Filtration of Liquid Pharmaceuticals (aligned with PIC/S Annex 1)