Undertaking a Membrane Filtration Pilot Test

This presentation/paper was given at the American Filtration and Separation Society (AFS) 2017 Fall Conference, September 18 – 20, 2017, sponsored by the American Filtration and Separations Society.

Author: David Pearson* BA (Oxon), Managing Partner, Membrane Specialists LLC

Lewis Pain B Sc, Managing Partner, Membrane Specialists LLC

2 Rowe Court, Hamilton, OH 45015

The performance of membrane filtration separation systems is well understood for brackish and sea water applications with some membrane manufacturers providing free internet downloadable design models to enable, in particular, Reverse Osmosis and Ultrafiltration systems on water to be configured.

Some process applications are also standardized with substantial public domain information from many years of research and development work by industry and academia. Perhaps best resourced are membrane applications in the dairy industry, although process limits continue to be challenged and expanded.

Published data sufficient for a proper design is unlikely to be available and the process ideas being investigated may be unique, novel or confidential.However when companies are developing new products where the use of membrane separation technology is being

Membrane System Design: Standardized Processes

• Water RO
• Water UF
• Dairy RO (Cheese Whey)
• Dairy NF (Desalting)
• Dairy UF (Whey Protein Concentrate)
• Juice Clarification (Apple Juice but with care)

At this point the first Go/No Go decision point is reached with neither party having yet done any testing. The EU has to make a decision as to whether the membrane process, if it did work, is economically viable. A decision may be made there and then not to proceed, however if the parties consider there is a realistic chance of achieving the separation required at an economical enough cost, they continue around the loop moving to the first testing phase.

The EU and MSDS need now to act as a team working closely to ensure that each party obtains enough information from the testing and subsequent evaluation to move onto the next decision point, where it may be determined from the data obtained that the development process should be curtailed.

If the results are acceptable then a revised design is made and costed to reach the Go/No Go decision point again. Assuming the costing is still acceptable more detailed testing will start until after enough turns around the loop a decision to move forward can be made.

The collaboration between the MSDS and EU will need to look at the factors that affect the choice of membrane and membrane configuration. There are a number of key decisions, which are not irrevocable, to be made early in the development cycle. Some of these key decisions may have been already made during the initial feasibility and budget assessment. These Key Decision Points will be reviewed at each major review to ensure that the development process continues to move in the right direction and informed decisions can be made before committing additional resources to the project.

At this point the goals of the two parties to some extent diverge. A test protocol needs to be drawn up which clearly states the objectives of the trial, what tests are conducted, what resources are needed, what samples are to be taken and what analysis is to be made. The protocol may include what is considered a pass mark that is designed to avoid later disappointment or conflict between the parties and to ensure sufficient data is captured for the next key review. It is therefore appropriate to look at the key objectives of each party in the testing phase so that both parties understand the reasons for each other’s needs.

Typical key objectives will center on the quality of the product, generating a need to produce samples early in the development process. In the case of new product development these samples will usually be needed to sell the benefits to potential customers or to ensure the product is meeting internal quality goals. In the case of a food product, these tests might involve taste, flavor and texture. Often the membrane process is only a part of a complete process and samples might be needed for testing the performance of downstream processes.

The use of membranes on wastewater treatment and possible recovery of product or water from wastewaters might be in the increase as water resources become more stretched. In the wastewater field, a third party appears on the scene, the Regulator. Regulations may define goals by law, but regulators such as the EPA, State agencies or the local municipal wastewater plant receiving factory wastes or both. For these representative samples are required to ensure treatment objectives are met, such as the level of suspended solids, Fats, Oils and Greases, Biological Oxygen Demand (BOD) or Chemical Oxygen Demand COD).

Membrane separation processes do not treat anything (other than membrane bioreactors MBR’s). They simply separate a single stream into two separate streams. End users are also going to want to know enough about the unwanted stream so that it can be converted to alternate products or safely and economically disposed of. This too may require sampling for regulators approving receipt of those wastewaters to municipal wastewater treatment plant, surface water discharge or perhaps simply disposal by tanker to landfill.

Ultimately the end user is going to want to know the value of the investment that must be made and the long term operational costs.

The goals of the membrane system designer are a little simpler to define equally challenging to reach.

There are only a few variables that can be tested in a membrane separation process:

1. Operating Transmembrane Pressure
2. Operating Temperature
3. Cross-Flow Velocity
4. pH

Of these variables, temperature and pH may be non-negotiable and defined by the product itself.

From these variables the MSDS is needing to understand the effect on flux (flow rate of filtrate through the membrane per unit area) and what, if any, membrane fouling results from changes to these variables.

All membrane processes will concentrate something be it suspended solids by microfiltration or ultrafiltration, or dissolved proteins, sugars, acids or other organics with any of the four principle separation processes, RO, NF, UF or MF.

Establishing an effective and sustainable membrane cleaning regime is key to the long term success of a membrane system and often forms a substantial part of a piloting operation.

The ultimate goal for the MSDS is shared with the end user and that is to know the value of the investment that must be made and the long term operational costs.

Figure 2 Once Through or Tapered Design

• Simple Pilot Testing Method
• Accurate Scale Up if Full Scal Modules/Vessels used
• Samples are representative of full scale if same taper arrangement is used.
  • Length of Taper
  • Use of Booster Pumps
• Pilot Plant may be large for sufficient feed flow for length of taper

In this concept, feed is added to a batch tank and the feed stream circulated around the batch tank. Permeate is continually bled off so that the contents of the batch tank become more and more concentrated. Fresh material can be added to the batch tank during the process. At the end of the process run, a volume of permeate has been collected and there is the residual volume of concentrate or retentate left in the batch tank.

This process is typical of a juice clarification system where a large batch tank is continuously added to while clear clarified juice is captured for downstream bottling or evaporation.

The end user can obtain plenty of samples of representative product, whether the product is the permeate or retentate. The membrane system designer obtains valuable information with an infinite number of points on a flux concentration design curve. Membrane fouling and cleaning are easily investigated as the membranes are seeing at pilot scale exactly what they would see at full scale. Scale up issues are principally associated with tank sizes, batch volumes and batch residence times. Care must be taken in the design of the pilot test design to ensure that these parameters are close to expectations at a full scale.

This is the simplest membrane design to pilot, design and build. Tests are relatively easy to set up and operate at the pilot scale and the pilot plant itself is not complex. Feed volumes required will match the membrane area piloted. Scale up issues are primarily associated with whether or not a full scale membrane vessel or housing is being tested.

Figure 4 Continuous Feed and Bleed (Multistage Recycle) Design – MSR

• Complex Pilot Plant
• Simple Pilot Testing Method
• Accurate Scale Up if Full Scale Modules/Vessels used and number of stages the same
• May need to run at lower concentrations at each stage
• Short hold-up time
• Retnetate and permeate continuous and possible constant flow
• Permeate flow will vary with time

The testing phase in the iterative loop described above in Figure 1 above might go through a number of different stages, each increasing in complexity and resource demands. These may be summarized, in Table 2.

The first test might be a very simple laboratory scale test on a single sheet tester or with a single hollow fiber membrane. Very small volumes can be used and an idea of whether or not a separation can be achieved. Flux data is unreliable, separation may not be accurate as the surface conditions probably will not relate to those seen on a larger scale module so although the cost of testing is low, the accuracy of a cost estimate derived from such a test is low. If the feed stream has the potential to foul the membrane this may materialize in these tests and some initial cleaning trials can be done.

At a similar cost, a volume of feed material might be sent to a test facility. Some MSDS’s have this testing capability as well as universities and research institutes. Larger sample sizes might be tested and larger membrane housings used in this type of test providing more reliable data. However the material being tested might have changed its nature in transportation and the flux and separation results might be affected. Where high volumetric concentration factors are being considered, large sample volumes may be needed as the final retentate volume will be small and the hold up volume of the pilot plant may become a limiting factor on the achievable concentration factors.

The Test Protocol – Designer Key Needs

• Feed Analysis
• What are the variations in the feed analysis
• Optimized operating parameters
• Flux decline with time
• Cleaning Regime
• Flux VS Concentration
• Flux VS Separation
• Investment Costs
• Operating Costs

The test protocol will define the tests that are to be made. In this area the protocol will be a living document in that as the testing progresses, changes in direction might be needed. The protocol should clearly define the objectives of the current stage of testing, the responsibilities of each party, the data that has to be captured and how to capture it. Ultimately the objective is to produce permeate or retentate to a defined quality and the pass/fail criteria will be stated.

From the MSDS’s perspective the following are of critical importance:

What is the feed Specification and what variations are there over time

What are the best operating conditions to achieve the desired separation at maximum flux

What is membrane flux decline with time

How to keep a membrane clean

How to generate a flux versus concentration graph

How to generate a Separation/Concentration graph

The EU’s key objectives are likely to be the generation of samples of permeate and retentate to establish product quality and product losses. The MSDS will have a need for more sampling to be taken, particularly if the final design is to be an MSR.

Figure 5 Flux Concentration

Setting up in the earlier testing phases a functional analysis, with one variable changed at a time will shorten a testing period establishing optimum operating conditions efficiently. The key variables are listed under the Key Objectives for the MSDS above and would be an early part of the testing protocol.

The level of membrane fouling by the process is of key concern to the MSDS and will often define the length of a test program. The protocol will need to create a baseline, usually water flux, for the brand new membrane and then monitor changes to the water flux over time. Critical for performance is not that the brand new water flux out of the factory is maintained but that the performance of the membrane stabilizes soon after first use.

Generation of a flux concentration curve is challenging for a continuous feed and bleed system where the pilot is not designed as an MSR system. In a three stage system for example, each stage will be operating at permanently at or around a single but different point on the flux/concentration curve (see Figure 5 Flux Concentration – MSR below). To get the flux, separation and fouling potential at that point, may require a single stage or batch pilot plant to operate for a time at that concentration, meaning that concentrate samples for the client are no longer available except when operating as the final stage.

Test Protocol

• Tests
  • Clean water flux
  • Membrane Selection
    • MWCO
    • Configurations tested
  • Functional Analysis
    • Pressure
    • Temperature
    • Cross Flow
    • pH
  • Concentration Tests
• Sample Protocol
  • What needs analyzing
  • What can be done in house
  • How many samples needed
  • Who pays
  • Turnaround time – delays and there affect on testing progress

The development of a membrane process will be most efficient and rewarding if the two parties work closely together to define objectives, establish a clear and encompassing test protocol, clearly define the roles of each party. Communication is key to ensure appropriate and necessary data is captured to allow the development process to move forward with informed decisions made each step of the way.

Risks should be properly planned for and the cost and time of testing reflect the risks being taken.

With a carefully planned and well executed program, the ultimate goal of having a well designed membrane system providing no operating surprises when first put into service will be met.