Define the separation task
Filtration separates material by passing water through a medium. In a granular bed, particles can attach within the depth through several transport and attachment mechanisms. A membrane provides a selective barrier whose performance depends on material, structure, operating conditions and integrity. Neither process can be selected from the appearance of the water alone.
Start with the contaminants and the required downstream quality. A filter intended to protect disinfection has a different performance question from a membrane intended to remove dissolved salts. Describe the raw-water envelope, upstream conditioning, solids loading, required output and available residuals-handling route. The plant needs to work during cleaning and maintenance as well as with every unit online.
Choose between process families
| Process | Useful role | Important limitations |
|---|---|---|
| Slow sand filtration | Particle removal with biological activity in and near the upper bed. | Requires suitable source quality, area and management of maturation and cleaning; not a simple substitute for rapid filtration. |
| Rapid granular filtration | Removal of suitably conditioned particles, often after clarification. | Performance depends on coagulation, media condition, hydraulic loading and backwash quality. |
| Microfiltration and ultrafiltration | Physical removal of particles and selected microorganisms. | Dissolved salts and many dissolved organics pass; pathogen credit depends on validated performance and integrity monitoring. |
| Nanofiltration and reverse osmosis | Removal of selected dissolved substances, with different selectivity. | Pressure, pretreatment, recovery, concentrate management and chemical compatibility become central design issues. |
Microfiltration and ultrafiltration are not interchangeable with desalination membranes. Nominal pore size is not, by itself, an assurance of virus removal. Manufacturers’ ratings, challenge data and the monitoring method must be interpreted within an agreed validation framework.
Calculate the loading with a unit unavailable
Use the total effective area of units actually filtering, with consistent flow and time units.
Consider four rapid filters, each with 40 m² of area. At 480 m³/hour total flow, the all-online rate is 3 m/hour. With one unit isolated and the same total flow, it rises to 4 m/hour. This is an arithmetic example, not a recommended design rate. The permissible rate must come from process evidence and the adopted criteria.
If a 24-hour period produces 10 ML of filtrate but consumes 0.4 ML in washwater and sends 0.1 ML to waste during restart, the net output is 9.5 ML. Do not promise 10 ML/day to customers from that operating cycle. Where backwash is recycled, account for its quality, treatment burden and timing instead of treating the recycled flow as free additional production.
Understand a filter run
During a run, retained material changes the bed or membrane resistance. Headloss or transmembrane pressure can rise, while filtrate quality may eventually deteriorate. A run can end because of a quality limit, hydraulic resistance, maximum duration or another validated operating constraint. The first limit reached governs.
A granular filter needs an effective backwash sequence that expands or agitates the bed sufficiently without unacceptable media loss. Air scour, water wash and surface arrangements are process-specific. After cleaning, a ripening period may require filtration to waste. A low headloss immediately after backwashing does not prove acceptable pathogen removal.
For membranes, track permeability normalised appropriately for temperature, transmembrane pressure and recovery. A falling raw flux can reflect colder water as well as fouling. Separate reversible accumulation, chemically removable fouling and irreversible deterioration before prescribing more aggressive cleaning.
Integrity determines the meaning of a membrane barrier
A small breach in a module, seal or connection can undermine removal even when most membrane area performs well. Direct integrity tests interrogate the barrier itself, for example through a validated pressure-based method. Indirect measurements such as filtrate turbidity or particle counts provide ongoing evidence but have their own detection limits.
Link a credited log reduction to what the integrity test can actually verify. Test sensitivity, frequency, isolation arrangements and response time matter. If a failure is detected after several hours of operation, determine how the potentially affected water is identified and managed. A process flow diagram containing a membrane symbol is not evidence of continuous barrier performance.
Diagnose before adding units
| Observation | Questions to investigate |
|---|---|
| Rapid headloss development | Has raw-water solids loading increased? Is coagulation producing suitable floc? Is washing restoring the bed? |
| Quality spike after backwash | Is the restart sequence appropriate? Is filtration to waste operating? Is flow ramping too quickly? |
| High membrane pressure | Has temperature changed? Are scaling, organic fouling or solids accumulation involved? |
| Persistent integrity failure | Which module or seal is affected? Is the test itself functioning? Can the train be reliably isolated? |
A proposed extra filter may hide a weakness in pretreatment or washwater supply. Conversely, excellent average filtrate quality can hide inadequate standby capacity. Assess the event conditions and maintenance state that actually govern service.
Residuals, resources and operating evidence
Filters concentrate captured material into backwash water, sludge, spent media or membrane cleaning streams. Membranes do not eliminate residuals. Include chemical storage, washwater equalisation, waste treatment, disposal and replacement consumables when comparing options. Energy and whole-of-life cost depend on the specific process and source; broad claims that one technology always uses less are unreliable.
Retain representative pilot results, loading assumptions, integrity-test evidence, backwash recovery, unit availability and the net water balance. Explain how the system responds to an adverse raw-water event. Then connect the verified filtered-water envelope to disinfection design and the capacity assumptions to supply reliability.
Sources & further reading
- Australian Drinking Water Guidelines ↗NHMRC and NRMMC · Living guidelines; publication and draft status checked 25 September 2026
External source · Checked 25 September 2026 - WaterVal treatment validation framework and protocols ↗Water Research Australia · Living technical resource
External source · Checked 25 September 2026
Source findings are distinguished from editorial interpretation. Apply current local criteria and project evidence when making engineering decisions.