Past and future of microfiltration
Memcor Australia's technology centre general manager Andrew Layson investigates the inception of MF, its development through the years and expectations for its future use
Microfiltration (MF), like all other technologies, has evolved from a research and development phase with relatively high-cost initial applications through to widespread use. It is now a proven, validated and cost-effective treatment option for a wide variety of feed streams.
Developments and improvements continue as the technology gains recognition for being the process of choice for a range of applications. Practitioners of MF technology have learned a great deal about its limitations and applications.
A great deal is of this knowledge is distributed amongst the hundreds of membrane plant owners around the world. The experiences and collective understanding of these users is
presented in this article.
MF systems were first installed in the mid-1980s. The first commercial MF systems were built to treat high-value, niche feed streams including wine, juices and water for pharmaceuticals.
The next stage of development saw the technology applied to larger feed streams where site-specific requirements justified the additional cost of the developing technology. For example, a Melbourne-based manufacturer installed a MF system in 1992 after off-spec performance of its conventional treatment plant resulted in thousands of dollars of rejected products.
Further development and cost improvements have seen MF plants become viable on low-value feed streams such as potable water and secondary effluent in the mid-1990s. Product developments and improvements continue today as the technology continues to define its ‘sweet spot’ among the large number of filtration technologies available. Figure 1 shows the growth in MF system installations. Today there are more than 1,000 microfilter plants in operation worldwide with a net capacity of more than 2,000Ml/d. In addition, there are around 200 MF plants under detailed design or construction with a net capacity of around 1,000Ml/day.
Throughout MF’s development, practitioners of the technology have learned a great deal about its optimisation, limitations and applications.
Microfilters and traditional treatment systems both remove suspended solids, although the mechanisms of removal differ. MF units remove solids at their surface to form a filter cake, irrespective of the nature of the solids.
The MF membrane is a positive barrier to contaminants. Traditional processes (media filtration and clarification) rely on attractive forces, chemical addition and gravitational settling to achieve removal and exhibit variable performance. As the mechanism of removal in MF differs from traditional treatments, the sensitivity of MF to contaminants in feedwater also
differs from traditional experience. Some factors critical to traditional treatment performance are irrelevant to MF, while other factors generally ignored in traditional treatment design are extremely important to MF. Contaminants having an impact on MF can be split into three categories shown in Table 1.
The presence of consistent low-level contaminants is usually addressed by good cleaning system design and/or feed conditioning. For example, pH correction or chlorination in the case of soluble iron presence. The impact of spikes and seasonal variations are more difficult to solve. MF systems produce consistent water quality irrespective of a spike in feed conditions.
However, a drop in plant flow, recovery or cleaning or backwash interval (or any combination of these) usually occurs. This poses two key questions for end-users and MF suppliers ‘how long does a spike have to be before it is considered a base condition?’ and ‘what reduction in plant performance can be accepted during a spike?’
There is no simple answer, but a good knowledge of spike frequency, duration and extent is essential to achieve the correct balance between membrane inventory and plant flexibility. Some end-users are have optimised membrane inventory and operating cost by recognising different modes of feed conditions. This approach was used recently on a 33Ml/day MF plant installed in New Zealand, shown in Table 2. A number of other feed water factors have an influence on MF plant performance, including:
- weed and debris – large debris and particularly weeds can block feed strainers, pumps and membranes, reducing performance. Suitably designed pre-screens are essential,
- polymer – organic polymers have a tendency to foul MF membranes. Polymers should be avoided in processes feeding MF systems. If unavoidable, a robust cleaning system must be included to control polymer fouling,
- chlorination – pre-chlorination has proven extremely effective in improving MF performance and extending cleaning intervals, especially in secondary sewage filtration applications (even for polypropylene membranes, which perform better with long membrane life in chlorinated sewage),
- coagulant addition – in some cases a small level of coagulant addition (1-2mg/l) can improve the filterability of a feed by conditioning the filter cake formed on the membrane surface,
- coagulant selection – individual coagulants behave
differently in contact with membranes. Some are easier to backwash off the membrane than others.
MF systems produce three waste streams:
The treatment of these wastes has varied, resulting in a number of operational problems:
In each case, treated water quality and/or plant capacity was compromised. Recommended waste treatment for two of the largest MF applications are shown in Table 3 in order of preference. Trials have become a common feature of MF system design and have many benefits. Unfortunately, some trials have been carried out in a manner that has failed to achieve the expected outcomes.
Inadequate trialling (in some cases) has led to treatment systems that failed to operate as expected. It is useful to differentiate between two levels of trialling:
- demonstrations – are short on-site tests that demonstrate the technology and indicates the quality of treated water that can be achieved. They should not be used as a basis of final plant equipment sizing,
- trials – are well-planned programmes designed to investigate the expected operating conditions of full-scale plants.
Characteristics of each are shown in Table 4. Perhaps the most important measure of a successful trial is the demonstration of repeatable performance through two or more chemical cleaning cycles.
Figure 2 is an excerpt of the data taken from a recent eightmonth cold water surface water trial. There are two key features to note:
- the repeatable slope of the curve in each cycle,
- the long period of time needed to achieve two cleaning cycle (approximately two months). Even the most thorough trial has limitations. It is important to recognise trialling is unlikely to demonstrate the presence (if any) of long-term fouling issues.
Many early MF installations were based on short-term trials. Issues affecting long-term stability were not thoroughly understood, so design fluxes were higher than would be applied today. As a result, several installations were originally rated at fluxes that were not sustainable. Many of these plants have been upgraded through the addition of membrane area (usually at the MF vendors’ expense) and operate at flux rates 30% lower than originally designed.
Today’s MF market includes a number of MF vendors. End-users should ensure installations are sized on the basis of extensive operating experience. New entrants into the market are prone to making ambitious flux selections, which do not take heed of the lessons learned by more established vendors. Successful equipment sizing and selection today relies on:
- benchmarking proposed design flux versus similar applications,
- ensure extensive trials are conducted if vendors with limited proven applications experience are considered,
- ensuring trial operating parameters match the full-scale plant selections.
The last point sounds obvious but it is surprising how often this basic principle is not followed. Large-scale MF plants have been operating for approximately ten years.
The increased rigour and attention characteristic of large treatment plant operations has resulted in greater scrutiny of long-term operating conditions. Around 1997, long-term fouling effects in some installations were recognised.
The first significant site where this was observed was San Jose water company’s Saratoga plant in California. The SJWC surface water filter plant, rated at 19Ml/d, was installed in 1994. The plant operated extremely well but while plant capacity continued to remain at the design value, membrane resistance steadily increased over a period of three-four years.
After detailed investigation of the problem, Silica was revealed as the prime cause of the change. This was addressed by the once-off use of ammonia bifluoride cleaner.
A similar change in membrane condition was observed at Eraring power station. The MF plant was installed in 1995, which filters secondary effluent as part of an MF/RO reuse system. During the first two years of operation, cleaning intervals slowly reduced from around four weeks to a little more than seven days.
Detailed examination of the membrane surface and foulants showed manganese was the cause. A single citric acid clean restored membrane performance to almost as good as new condition. Not all MF installations have shown long-term fouling.
Applications where long-term fouling is potentially an issue now use a combination of design and operations practices to maintain stable, long-term performance:
l limit the number of times cleaning chemicals are re-used,
- ensure cleaning system is automatic and is flexible enough to allow maintenance cleans,
- carry out pre-peak season flow trials,
- initiate chemical cleaning on throughput or resistance change rather than time or operating pressure,
- use RO permeate or softened water for cleaning solution makeup (particularly important for feedwaters high in salinity, hardness or reactive silica).
Like any process, reliable long-term MF operation and optimisation relies on measurement of key parameters. These parameters need to provide an accurate indication of membrane condition.
Early measurement techniques involved measurement of differential pressure, flow and turbidity. These measures have been refined to give accurate measures that filter out the effects of changing plant conditions:
- resistance – total resistance to flow caused by the membrane and filtercake on the membrane and equals
P/Flux at a reference temperature,
- backwash and cleaning efficiency – a measure of change in resistance through each cycle (crude measures of efficiency use driving pressure, however, these are inherently inaccurate,
- integrity testing – pressure decay testing (PDT) is the primary method of membrane integrity measurement. Particle counting is used at some plants. However, PDT is more reliable, requires less maintenance and is both more precise and accurate (by two orders of magnitude). PDT has also been shown to correlate with microbial removal.
Real unit area membrane costs have dropped by more than 80% since 1990 and continue to decline. At the same time, membranes in numerous plants are demonstrating extended life. This has significantly reduced MF plant operating costs. At the same time, intelligent control systems, efficient plant designs and practical plant operation strategies are enhancing membrane cost savings by:
- running plants at lower fluxes or pressures at low load periods, for example, during winter,
- taking plants off-line for short periods when feed conditions spike to extremes, for example, a two-hour shutdown during the first flush of a river after rainfall can eliminate the need for early cleaning,
- use of cost-effective cleaning chemicals, for example, sulphuric acid in lieu of phosphoric or citric acid,
- flexible control system design that allows operators to fine-tune variables or set-point to achieve cost-savings, for example, cleaning chemical batch strength.
MF system costs have dropped dramatically over the last decade due to increased competition, improved equipment design and lower production costs. As a result, MF installations are increasing rapidly. Several ‘signature’ applications have emerged for the technology:
- filtration of secondary sewage for re-use (either direct to application or via reverse osmosis),
- pre-treatment for RO by MF stabilises RO performance, extends RO membrane life and reduces cleaning requirements, filtration for potable water supply to remove biological pathogens such as Giardia and Cryptosporidium. Drinking water standards around the world are tightening to recognise the health impacts of these contaminants and the effectiveness of MF in guaranteeing safe drinking water.
Other developing applications include:
- membrane bioreactors (MBR’s) that combine membrane filtration with secondary effluent treatment without the need for settling,
- combination with ozone and biologically active carbon filters (BAC’s) to effectively treat surface water without the need for coagulant dosing.
MF systems are available in pressurised and submerged configurations. Lower operating and capital costs are increasing the dominance of submerged systems. Many plants installed as pressurised systems two-three years ago would today be built as submerged systems at lower costs. MF membrane configurations and materials are also being refined. Improvements include chlorine resistant membranes, higher area modules, higher permeability and lower particle size cut-offs
References: Craig G, Recycling effluent using safe environmental practices – University lecture 1998. Gere AR, Climbing the microfiltration learning curve – Proceedings of NWRI microfiltration II international conference, November 1998. Leslie G et al, Pilot testing of microfiltration and ultrafiltration of reverse osmosis during reclamation of municipal wastewater – Proceedings of ADA biennial conference, 1996. Hong SK et al, Removal of micro-organisms by MF process: Correlation between integrity test results and microbial removal efficiency, proceedings of AWWA membrane technology conference, April 1999.