Normalized differential pressure in RO: how to calculate it and what it tells you

Differential pressure is the earliest warning most reverse osmosis plants have of fouling in the feed channels. It is also the most misread number on the screen, because raw differential pressure moves with flow and temperature as much as with fouling. Normalized differential pressure removes those effects, so a rise means something has changed inside the membranes.

What differential pressure is

Differential pressure (ΔP, also called pressure drop) is the loss of pressure as water flows along the feed side of the membranes:

ΔP = feed pressure − concentrate pressure

It is measured across each stage, and sometimes across each pressure vessel. The pressure is lost to friction in the narrow feed channels and the feed spacer between the membrane leaves. When those channels start to block, ΔP rises.

A note on the abbreviation: NDP usually means net driving pressure, the pressure that pushes water through the membrane. Some plants and software also write NDP or nDP for normalized differential pressure. They are different quantities, so check which one a report means.

Why raw ΔP misleads

Two things change ΔP without any fouling at all:

• Flow. ΔP depends on how much water travels along the feed channels. Pressure drop rises faster than flow, so running the plant harder raises ΔP and turning it down lowers it.
• Temperature. Cold water is more viscous, so the same flow loses more pressure. A cold season raises ΔP; a warm one can hide fouling.

A plant that reduces output while fouling builds up can show a flat raw ΔP for weeks. Normalization removes the flow and temperature effects so the trend shows the condition of the feed channels.

How to normalize differential pressure

The usual approach corrects the measured ΔP back to the flow (and, where used, the temperature) of a reference period, normally start-up or the last membrane replacement:

Normalized ΔP = measured ΔP × (reference feed-channel flow ÷ actual feed-channel flow)ⁿ

Where:

• Feed-channel flow is the average flow along the stage: (feed flow + concentrate flow) ÷ 2.
• n is an exponent, typically about 1.5. Membrane manufacturers and software packages use values between roughly 1.4 and 2, so use one value consistently.
• A temperature correction (a viscosity ratio) is added by some methods. Others leave it out and keep the comparison within similar temperatures. Either works if it is applied consistently.

ASTM D4516, the standard practice for normalizing RO performance data, covers permeate flow and salt passage. Plants and membrane manufacturers normalize ΔP in the same spirit, correcting it to a reference flow.

A worked example

At start-up, stage 1 of a brackish water RO train ran at:

• feed flow 100 m³/h, concentrate flow 60 m³/h: feed-channel flow 80 m³/h
• ΔP 1.6 bar

Six months later, after the plant was turned down:

• feed flow 90 m³/h, concentrate flow 50 m³/h: feed-channel flow 70 m³/h
• ΔP 1.9 bar, at a similar temperature

Raw ΔP has risen by 19%. Normalized to the start-up flow, with n = 1.5:

1.9 × (80 ÷ 70)^1.5 = 1.9 × 1.22 = 2.32 bar

That is 45% above the 1.6 bar baseline. The raw figure hid more than half of the real increase, because the lower flow was masking it. This stage needs cleaning.

When to clean

A common guideline is to clean when normalized ΔP has risen by 10–15% above its baseline, alongside the same trigger for a fall in normalized permeate flow or a rise in salt passage. Cleaning earlier is usually more effective: fouling that is left to build up becomes harder to remove.

Membrane manufacturers also set an absolute limit on pressure drop, per element and per pressure vessel. Above it, the force on the elements can damage them (telescoping). Check the datasheet for your membranes and treat it as a hard limit, not a target.

What a rising ΔP tells you

Where ΔP rises, and what moves with it, points to the cause:

• First stage, lead elements, with falling permeate flow: biofouling or particulate and colloidal fouling. Biofouling often shows as a fast rise over days rather than weeks. Check pretreatment, SDI and biocide dosing.
• Last stage, with rising salt passage: scaling, where the concentrate is most concentrated. Check the antiscalant, recovery and the scaling potential of the concentrate.
• ΔP rising across all stages at a similar rate: a pretreatment upset or cartridge filter breakthrough carrying particles through.
• A sudden step: an event, such as a change of feed source, a dosing failure or a damaged cartridge filter. Look at what changed that day.

See Membrane fouling in RO for the full pattern of each fouling type, and how to measure CIP recovery for checking that a clean actually brought ΔP back down.

Common mistakes

• Subtracting two gauges. ΔP is a small difference between two large pressures. A 1% error on each gauge can be a large error in ΔP. A differential pressure transmitter across each stage is far more reliable.
• Watching the whole train only. Fouling in stage 1 and scaling in the last stage can cancel out in a train-level figure. Normalize each stage.
• Resetting the baseline after each clean. It hides the permanent rise that builds up over the life of the membranes.

From spreadsheet to software

Normalizing ΔP for each stage, every hour, is easy to set up in a spreadsheet and hard to keep up. RO Insight™ normalizes differential pressure, permeate flow and salt passage automatically for each stage, and flags when a cleaning trigger is approaching. If you would rather have an engineer look at your data first, a membrane performance assessment does the same from your plant history. For the wider method, see RO data normalization explained.