Blog / Engineering guide
Pressure UF Cleaning Trends | Diagnose Loss of Permeability
Use pressure UF baseline data, cycle comparisons and recovery records to investigate declining output before changing cleaning conditions.
HOSON WATER editorial · 8 min read
Published · Updated

Begin with comparable operating data
A useful UF cleaning decision starts with a verified change in membrane performance under comparable conditions. Read flow, transmembrane pressure, water temperature and cycle state together. Then check instruments and upstream changes before selecting a cleaning response. A rising pressure reading can have several explanations, and each points to a different investigation.
This guide helps plant teams prepare evidence for a membrane supplier and organize routine trend review. HOSON LH and LW families need their own approved settings and cleaning procedures. The examples below describe hypothetical readings and calculations, not cleaning setpoints or observed performance at a HOSON project. Concentration, exposure time, pH and temperature for any proposed cleaning should be approved for the complete installed assembly.
Build a baseline that can be recreated
During commissioning, save the model and serial inventory, operating flow, temperature, relevant pressures and the programmed sequence. Record the source quality and the pretreatment settings. Choose a repeatable point in the filtration cycle for routine comparison, such as a defined interval after backwash and stabilization. Document how that interval is identified in the historian.
Preserve the raw values alongside calculated indicators. If a pressure transmitter is replaced or a flowmeter scaling changes, annotate the event and compare the old and new readings. Record any deliberate change in production target or cleaning recipe. Without that history, a chart can show an apparent trend that actually begins with a changed calculation.
Use separate tags for individual trains where possible. A combined plant flow can hide a weak train because the controller compensates by increasing production elsewhere. The baseline should explain what the instruments physically measure.

Calculate flux and permeability consistently
Flux is filtrate flow divided by the active membrane area. With flow expressed in cubic meters per hour, multiply by 1,000 to obtain liters before dividing by area. Permeability is flux divided by the defined transmembrane pressure. State the pressure unit because a value reported per bar differs numerically from the same result reported per kilopascal.
In a hypothetical train with 800 m² of active area and a filtrate rate of 48 m³/h, flux is 60 L/m²/h. At a measured TMP of 0.40 bar, the observed permeability is 150 L/m²/h/bar. If the same flux later requires 0.50 bar at a comparable temperature and cycle point, observed permeability becomes 120 L/m²/h/bar, a 20% decline from that baseline.
Check the definition of TMP used by the installed system. Confirm pressure tap locations and any correction for elevation or pipe losses before comparing separate racks.
| Hypothetical condition | Flux | TMP | Observed permeability |
|---|---|---|---|
| Baseline | 60 L/m²/h | 0.40 bar | 150 L/m²/h/bar |
| Later reading at comparable temperature | 60 L/m²/h | 0.50 bar | 120 L/m²/h/bar |
Apply the approved temperature correction
Water temperature changes the relationship between flow and pressure. DuPont's January 2026 MEV manual explains temperature normalization as a way to avoid mistaking temperature-driven changes for membrane deterioration. Its correction method belongs to the membrane system it describes. For an LH or LW installation, request the correction method and reference temperature approved by the supplier.
Keep three fields in the report if normalization is used. Retain measured temperature, observed permeability and the corrected value. Store the formula version or controller setting so another engineer can reproduce the calculation. Compare trends at the same reference temperature and confirm that the input values are within the method's valid range.
When a site lacks a validated correction, compare periods with similar temperatures and label the remaining uncertainty. That is more informative than applying a convenient coefficient taken from another manufacturer's module.
Check the instrument chain before changing chemistry
Compare pressure transmitters with a suitable reference under the site's isolation procedure. Check whether a blocked pressure tap, trapped air, incorrect zero or changed signal scaling could explain the shift. Review flowmeter readings against an independent balance where feasible. Confirm that the active-area calculation matches the modules actually in service.
Inspect the sequence records around the apparent deterioration. A valve that closes late, a shortened backwash or an unavailable air supply can change recovery between cycles. Check the programmed state against feedback from the equipment. A controller can display a commanded position while a mechanical fault leaves the actual path partly obstructed.
Annotate the diagnosis instead of erasing suspect readings. Keep the original data with a quality flag and record the corrective action. Future reviewers then know why a section of the trend was excluded from analysis.
Read within-cycle and between-cycle behavior
Plot the pressure or corrected permeability through several full cycles. Note the starting point after backwash, the slope during filtration and the point before the next backwash. Compare these shapes at a similar flux and feed condition. A steep rise within one cycle suggests a different investigation from a progressively poorer starting point across many cycles.
For example, consider two hypothetical trains that reach the same final TMP. The first returns near its baseline after every routine backwash. The second starts each cycle at a higher TMP and has little recovery. Record these patterns separately. They provide evidence about the persistence of the resistance and help the supplier select inspections or tests.
Do not assign a foulant solely from the chart shape. Compare feed analyses, solids events, chemical residuals and inspection findings. The chart is a guide for the investigation, with material compatibility and laboratory evidence informing any cleaning change.
Measure recovery with the same reference
Record performance immediately before cleaning and again after the approved flush, stabilization and return to a comparable duty. Include the clean start time, solution preparation, actual temperature and any deviations. The post-clean reading should have the same active area, reference temperature and cycle position as the chosen baseline.
Suppose hypothetical corrected permeability was 160 at commissioning, declined to 112 and recovered to 144 after cleaning, all in the same units. Recovery to the original baseline is 144 divided by 160, or 90%. Recovery of the lost permeability is the increase of 32 divided by the loss of 48, or about 67%. These percentages answer different questions. State the formula beside the result.
Use consistent definitions in monthly reports. Otherwise a change from one recovery metric to another can look like an improvement even when the measured membrane condition is unchanged.
Link the trend to upstream and operational events
Review source-water changes, pretreatment dosing, filter bypasses, storage residence time and production changes on the same timeline. Mark storms, source switching and maintenance events with actual times. When laboratory results arrive later, connect them to the sample collection time. This is essential for matching the chemistry to the operating event it describes.
Request additional testing to answer a specific question. For instance, if the trend began after a change in coagulant operation, review dose records and residual-related measurements with the process engineer. If the change followed a new industrial feed batch, investigate that batch's composition and its duration. Choose an investigation that can confirm or reject the suspected cause.
Preserve records for unaffected trains as a comparison. A shared change across all trains may suggest a common input, while a single-train change directs attention toward its equipment or recent work.

Control a cleaning trial as an engineering change
Before changing a procedure, obtain approval for the chemical, concentration, time, temperature and operating mode. Confirm compatibility across fibers, seals, potting, piping and the receiving waste system. Define how the team will check solution preparation and what condition requires stopping the trial. Use the site's approved chemical handling and isolation procedures.
Change one meaningful variable at a time where practical. Record the reason, expected observation and post-trial measurement. If the supplier proposes a staged procedure, preserve the readings between stages so its effect can be assessed. Include the final rinse and product-release criteria in the plan.
Review the outcome in terms of recovered production, chemical consumption, downtime and waste. A favorable short-term permeability result should be considered alongside the approved exposure budget and whether the underlying feed problem has been addressed.
Choose a reporting interval that preserves the event
Set historian sampling and storage around the cycle you need to understand. A daily average can describe broad production but can hide a brief pressure spike or a backwash that failed to complete. Keep sufficient resolution to identify the cycle-start condition, the rise during filtration and the recovery step. Confirm that the retained data includes valve states and alarms on the same clock.
Produce a short routine report with a few reproducible measures. These may include corrected cycle-start permeability, cycle-end TMP, net daily output, duration out of service and the approved cleaning events. Give each measure a definition and a data-quality rule. An operator should know whether a missing result means the equipment was stopped, the instrument failed or the calculation had insufficient valid inputs.
Use operating cost to prioritize investigation without making it the only trigger. For a hypothetical decline that reduces net delivery by 4 m³/h during a 20-hour duty day, the shortfall is 80 m³/day. The owner can compare the cost of replacement water or reduced production with the planned maintenance window. This arithmetic describes an operating consequence and does not determine a safe cleaning threshold.
Review the report with operators who observed the event. Add their specific observations, such as an unusual valve sound or a feed change, while separating those observations from confirmed causes. The combined record gives the technical team a more complete basis for its next diagnostic step.
Prepare an evidence package for escalation
Send the support team raw trend data covering a stable baseline, the deterioration and any maintenance response. Include units, tag definitions, timestamps, the operating sequence and the exact models in service. Add the feed analysis and a concise event log. Supply photographs of relevant connections or damage when available, with each image identified by location and time.
Ask the supplier to distinguish its proposed checks from its proposed settings. Establish the next measurement that would decide whether to inspect a valve, sample a deposit, adjust a cycle or undertake approved cleaning. Record which decision the owner must make and what evidence is still missing.
After resolution, save the approved procedure and the verified new baseline with the module inventory. This gives the next operating shift a reproducible record and makes future deterioration easier to recognize at the same defined conditions.