Selecting a Machine for Reverse Osmosis

January 2017
by Clark Smith

Reverse osmosis machine sales are suddenly going crazy—up an order of magnitude from past years. Three decades after the introduction of reverse osmosis (RO), wineries are now prosperous enough to afford their own machines, and they are at last getting hip to the technology’s many quality-enhancing powers.

An onsite RO machine with the proper ancillary equipment can adjust juice Brix (an unusual practice in California) or wine alcohol either up or down, thus uncoupling the harvest decision from Brix and freeing up winegrowing decisions to focus on desired flavor, color, tannin and fruit character. RO is the core technology in systems that remove a variety of taints: acetic acid, ethyl acetate, Brett, smoke taint and ladybug taint, and it can be pressed into service to reclaim tartrates and other valuable byproducts as well as purify water. Properly configured machines also can perform some ultrafiltration tasks.


Applications that drive RO purchase are alcohol adjustment (in dry climates) and juice concentration (in wet zones). Occasional applications such as VA, Brett and smoke taint don’t drive purchases for the same reason that you don’t buy a hospital in case you have a heart attack. Since my former company Vinovation in the early 1990s pioneered many of the membrane applications now in use in California, it’s a good time for me to share my advice to wineries considering a purchase. (Don’t worry, I have had no financial connection to reverse osmosis equipment sales or production services in nine years.)


Before I dive in, I want to clarify my intention. I find no villains in the winemaking game. Most have healthy egos, and many are struggling. I love our industry because everyone in it “gets” the spiritual appeal of wine, often sacrificing more lucrative career options in its service. While I intend to acquaint you with some suppliers’ positions I find misguided, I do so with respect and in the hope that better informed customers will encourage them to shift their strategies.


I also have no monopoly on the truth. In sharing my personal perspective, I hope to benefit both wineries and suppliers by illuminating the purchase process.


My intention here is to spell out every useful distinction in purchasing and operating an RO machine. I include every technical nuance and trade secret I have found valuable in a quarter-century of experience with tens of thousands of processing jobs. I will share my mistakes, and I’ve made plenty that I hope you won’t repeat.


Each winery has its own unique needs and will have different applications it considers most important when selecting equipment. The winemaker must consider filter porosity, pump configuration, dead volume, operating pressure, automatic operation and resin sourcing. Besides system cost, the selection process should weigh tangential flow rates and operating pressure against limitations such as watt draw, heat generation, noise and refrigeration requirements.


It is easy to buy the wrong machine. As with any emerging technology, buyers have a very poor understanding of high performance. Imagine trying to buy a car without knowing the difference between a Yugo and a Lamborghini. When it comes to RO, your typical winemaker understands the desired result but not the practice—why you want one, but not how it works or what system suits your specific situation.


RO purchasers face several hindrances to smart shopping. Just as with cars, commercial suppliers can’t be expected to provide an unbiased comparison with their competitors. They also may, in the same way as printer manufacturers, guard as proprietary some features of the systems they are selling in order to retain purchasers for after-market replaceables.


For winemakers, a second barrier to technical learning is self-doubt. The anti-manipulation views of the natural wine movement, in the faith that wine makes itself, have created an atmosphere in which it’s challenging to joyfully embrace new technologies.


For many of us, our younger selves are the most insidious of enemies. Many flinch at reminders of our embarrassing high school chemistry experiences. If you doubt that you may be your own worst enemy, here’s a simple test. Can you define what reverse osmosis actually means?

If not, don’t despair. Even the federal government has had its difficulties, as we shall see.

High-pressure sales pitch
The most important thing to understand about RO is filter porosity. This is as critical as knowing your brake from your gas pedal.

Membrane porosities are expressed in a number of peculiar ways. Unlike sterile filters measured in microns, the porosity of tangential-flow membranes is typically expressed as the molecular weight (in daltons) of a molecule that will be rejected by the membrane half the time. For example, a 180-dalton membrane working on a 20% glucose solution (molecular weight, or MW=180) would produce a 10% glucose filtrate.

Another way to rate porosity is salt rejection. The U.S. Navy developed RO in the 1960s so sailors could drink seawater. The industry’s main focus remains on water applications. A single 8-inch x 40-inch 80-dalton membrane will produce drinkable water (300 ppm sodium chloride) from seawater (30,000 ppm sodium chloride), so the salt rejection is said to be 99%.


While salt rejection can be related to dalton ratings, it isn’t a very good way for you to shop for your RO membranes, because it doesn’t discriminate between porosity and craftsmanship.


An 8-inch system with 16 elements in series (membrane surface area of 4,800 square feet), which costs about $250,000, can produce 50 gpm of drinkable water (enough to keep a battleship’s crew hydrated and happy).


Water flows readily through these membranes, so tiny imperfections in sealing get diluted and don’t affect the permeate salt content very much. A cheap, poorly made membrane can still perform acceptably. But alcohol inhibits flow, so on wine, that same system might yield only 4 gpm of permeate. Bleed -through from tiny perforations or gluing defects is instantly apparent, especially in red wine.


Unearthing a supplier of high-integrity membranes for our tiny niche market was a major research focus for Vinovation in the 1990s. Manufacturing costs are higher for fine wine filtration than for water applications. There are only a small number of reliable manufacturers that make dependable membranes for wine. Today, most RO manufacturers use the membranes we championed (see below).


Osmose invers was in experimental use in France in the 1980s for juice concentration, essential to remove rainwater by employing a membrane that retained all flavor constituents, while H2O (MW=18) passed readily. Wine applications are still forbidden in Europe outside an experimental permit basis. This means that European equipment manufacturers have three decades of experience with RO on juice but originally almost none on wine, though some have acquired good experience in New World markets in recent years.


Water, juice and dry wine behave entirely differently. When the feed stream is a 13% alcohol wine, the alcohol seems to form a gel on the surface, decreasing the flow about 15-fold. That not only means you need a lot more machine to get anything done, but its optimal design is very different.

Juice membranes have similarly slow flow, but for a different reason. Each degree Brix adds about 2 bars to the juice’s osmotic pressure, so a 21° Brix juice pushes back with about 40 bars. Only a high-pressure machine, preferably with a 70-80 bar upper limit, has the differential pressure to concentrate juice. A 40-bar (600 psi) machine won’t work at all.


As the juice is concentrated and the Brix climbs, the flow declines accordingly. The practical limit for a 72-bar machine (1,000 psi) is about 35° Brix. In these circumstances, a machine you can occasionally push to 80 bars can be a godsend.


The other problem with juice is pectin and pulp, which will clog a membrane and require cleaning with warm water and pectinolytic enzymes. (Important tip: It’s critical to remove pectin before raising pH for phenolic cleaning, lest precipitated pectin permanently seal your membrane surfaces.)


If pulp is present, membrane elements need wider spacers to prevent plugging. This decreases the amount of surface area that can be included in the spiral wind by 30%. The increased open cross-section also increases the pump flow necessary to achieve turbulence.


The nice thing about juice is that its aromatic elements are largely tied up as glycones. That means they are much bigger than they will be after fermentation liberates them. Juice membranes can be more porous. That’s why many European manufacturers install highly porous membranes, with unfortunate consequences for wine applications.


In practice, the winery should decide whether the machine is to be used primarily for juice or wine. If both are important, pump speed adjustment should be built in. Separate banks of membranes may be an advisable purchase, in which case designs that accommodate easy switching of multiple banks would be preferred.


You don’t need wider spacers if you clarify juice prior to processing to less than 0.5% solids via settling, centrifugation or flotation. Since clarifying juice is an unwelcome distraction during crush, it’s great to work on a small volume. If you desire a Brix bump from, say, 21° to 23° Brix, you can clarify 20% of your juice and squeeze it to 31° Brix, then recombine it. The higher pressure your machine, the smaller the percentage you can work with. Tartrate fouling is important too, especially at high concentrations like 32° Brix.


How do we get the most out of nice, tight RO membranes? With dry wine, pump pressure has to overcome an osmotic differential of only about 175 psi (12 bars), so a 70-bar system gives twice the flow of a 40-bar system. I am a high-pressure guy. Opponents talk about shear forces associated with 1,000 psi systems, but in tens of thousands of trials, I have never seen any validity to this concern.


Wine systems are much different from designs for water. Because permeate flow is a small percentage of feed flow, you can string a large number of membranes in series without losing much crossflow scrubbing in the tail-end membrane. Consequently, small ROs with two to four elements are not very efficient. High-pressure systems fitted out with tight membranes can handle 12 to 16 membranes in series with little drop in efficiency. Same pump and controls. Why not?


Adding filters is an affordable way to get more for your buck. You should also rig for overnight, unattended operation using a properly designed fail-safe loss-prevention system that can double your throughput.


Sometimes you just want to work on one barrel of your $200-per-bottle Reserve from grapes that the birds liked as much as you did. That means you need a shunt so you can re-plumb for one or two membranes when you rehabilitate your little darling.


My life in reverse
Since my divestiture of the service business in 2008, I have had no financial connection to RO equipment sales or production services. But ever since the Benzigers at Glen Ellen Winery chained me quite reluctantly to an RO machine in 1990, I have been pioneering its application in winemaking.


Bruno Benziger had a dream to market a good-tasting, non-alcoholic wine, a project I explored for three years in the early 1990s. Although eventually aborted, the project gave me a window on wine’s mysteries such as colloidal structure, aromatic integration and sweet spots. It also led to the development of VA-reduction and alcohol-adjustment patents and the 1992 founding of the wine technology service firm, Vinovation Inc. to exploit these inventions.


Vinovation processed about 4,000 volatile acidity jobs in its 17-year history. More than 100 of these were subsequently named in Wine Spectator’s Top 100 Wines of the World. We achieved this level of quality by using exclusively tight membranes with a n MW cutoff of 80 daltons. It seems that today’s expectations are lower (a wine getting a VA haircut is expected to take a quality hit). I believe the primary factor in the process’s sliding reputation is the use of looser filters to make cheaper, faster machines.


That said, 80-dalton membranes are not the right choice for applications where passage of a larger MW taint is required. In these applications, quality loss is a necessary evil as in fining. Porosities of more than 150 daltons may be useful in juice concentration, but they are of questionable utility and legality in wine.

Getting the membrane right

The Benzigers’ machine was built by Millipore and contained membranes made for producing drinking water from seawater. Like most RO membranes, these were spiral-wound cylinders containing a very high degree of surface area (see “RO is a Cross-Flow Technology”). These were spiral-wound seawater membranes that worked fine for salt rejection but turned out to have a low manufacturing standard.


When (so-called) nano-filtration membranes came on the market in 1995, Vinovation eagerly stocked up on the new, improved membranes that were indeed better built and gave great flow rates.


Early in 1996, I got a panicked phone call, jumped in my car and drove six hours south only to confirm that we had completely ruined a batch of Central Coast estate Chardonnay with these overly porous membranes that had stripped its flavor and body. We went on an exhaustive hunt for a replacement membrane, ultimately pairing up with Osmonics, a division of G.E., which developed a hand-rolled polyamide membrane with the acid and amino sides reversed that became known as the Vinocon RO5 and remains the industry standard. (See “GE Vinocon R05 Characteristics.”)


Ions such as acetate and lactate do not pass because they are hydrated by water and have functional molecular weights of several hundred. For this reason, SO2, which is almost entirely ionized, is largely conserved as well.


GE went on to develop two other membranes with looser porosities. The Vinocon RO1 is used for removal of Brett character and smoke taint, which don’t pass at acceptable rates in a tight RO membrane. Unlike the Vinocon RO5, the deployment of a Vinocon RO1 entails a loss of flavor and body similar in severity to a fining agent.

GE also includes its Vinopro nanofilters in its wine offerings. This is not an RO membrane but instead graduates up in porosity to the nanofiltration range. The purpose of a nanofiltration membrane is to separate mono- and divalent ions, so it needs to be loose enough to deplete acid ions such as bimalate and bitartrate. These membranes strip acidity. I have seen dozens of jobs run on these membranes where titratable acidity (TA) is cut in half and pH shifted up a whole point to more than 4.5.


High-porosity membranes improve permeate flow but also strip flavor. The Vinopro nanofilter series is, in my opinion, too porous for use on wine, though it may be a reasonable choice for juice concentration and flavored non-alcoholic wine where flavor retention is not an issue.


Reverse osmosis equipment suppliers do not make membranes. They purchase and install them in the machines they sell, often repackaged under proprietary names that conceal their identities. Yet at least one RO manufacturer refuses to reveal the identity and characteristics of the membranes it installs in its machines. A spokesman explained to me that without this protection of proprietary expertise, customers would be able to purchase replacement membranes directly from suppliers, eliminating the middle man who provided the know-how, and also enabling competitors to copy their designs.


Beyond pore size, membrane specificity can get pretty exotic and include features such as hydrophobicity and ion selectivity. Knowledge of porosity and other membrane characteristics is essential to the winemaker. Better yet, buy the machine you want, then independently purchase the membranes that suit your needs. It’s easy to install them yourself, and you should learn how in any case.

In any complex new technology, it is natural for suppliers to adapt a “Father Knows Best” role, and we are happy for their advice. But you can’t farm this particular decision out. In my opinion, membrane selection is properly placed with the proprietor of the bonded premise, in whose hands lies the responsibility for compliance, care, custody and control—not to mention the economic impact of the resulting wine quality.


Given the current turmoil of federal definitions (see below), it is not enough for the supplier to provide assurances of legal compliance. Until the federal government clarifies what is compliant, buyer beware.

Nor is a list of happy customers much help when researching unless their applications are the same as yours. For instance, a large winery with blending options may be content with a level of quality that is unacceptable to a niche estate.


An easy way to tell whether your membranes have acceptable porosity is to look at pH and TA changes during processing. Tight RO membranes don’t alter pH, and TA is only lowered by the differential in volatile acidity achieved. Loose membranes can cut TA in half and raise pH by as much as an entire pH point (from 3.6 to 4.6, for example).


Some suppliers supply cation resins charged with mineral acids to correct acidity, but in my experience they only mask the problem and can’t mitigate flavor stripping.

Legal foundations
As governmental agencies go, the U.S. Treasury Department’s Bureau of Alcohol, Tobacco and Firearms (BATF) and its successor, the Tax and Trade Bureau (TTB), deserve enormous credit for their forward thinking and industry-friendly pursuit of the technical complexities of winemaking technology. The openness to experimentation of American regulators is the envy of the wine world. In 1980, BATF even went so far as to send one of its most brilliant and personable agents, Richard Gahagan, to study enology at California State University, Fresno. Since he retired in 2002, his successor, Mari Kirrane, has carried on as a ready source of information and education. Winery compliance seminars were conducted for two decades by the aptly named Perky Ramroth. This is how a federal agency is supposed to work.


Federal regulations of cross-flow technologies, however, were put in place long before fine wine applications existed, and they are not designed to keep you out of wine quality trouble.


In 1984, Barry Gnekow began experimenting at J. Lohr Vineyards & Wines, first with ultrafiltration for removal of protein instability, and then with the production of non-alcoholic wine via reverse osmosis and addback of water from the tap, a process he patented. Gnekow and Lohr’s attorney organized a historic meeting in Washington, D.C., in 1985 that established the taxable background for both technologies. Present were many industry notables including BATF director Stephen Higgins, Wine Institute attorney Jim Seff and compliance specialist Raymond Williams.


“The focus was on clarifying taxable status,” Gnekow says. “The filtrate of ultrafiltration had already been established as taxable as wine, then enjoying a special tax status of 17 cents per gallon, and everybody was excited at the idea of classifying RO permeate as a distillate so they could tax it at $13.50 per proof gallon.”


The outcome, finally published in the Federal Register in 1990, established definitions dividing the micro-membrane world in two: the UF zone above 500 molecular weight and the RO zone below 500 MW. Thus, the 500 dalton line was laid down for tax reasons rather than scientific reasons. This was a decade before anybody envisioned RO as a standard wine practice.

The nanofiltration zone, scientifically considered as 150 to 1,000 MW, was completely ignored. In 2004, TTB, at the request of VA Filtration, corrected this oversight, but in a strange way, by defining nanofiltration as below 150 MW.


Other types of membranes also were vetted. Ion-selective membranes (electrodialysis) and hydrophobic membranes that pass only ethanol (osmotic transport) were added to the approved list in 2004.


In effect, nanofiltration is permitted as long as it isn’t really nanofiltration from a scientific standpoint. Since VinoPro claims in its literature to be an NF membrane with a porosity of 150-300 daltons, it would seem to be illegal for wine (but OK for juice?). But the federal definition of RO goes up well into the NF range (500 daltons), a situation that needs reform. In my opinion, 80 daltons is ideal for wine, and anything above 150 daltons should be forbidden.


The current situation only makes sense when viewed historically. The 500 MW boundary wasn’t a scientific definition of RO. It was a demarcation of filtrate taxable as wine vs. taxable as distillate. That’s fair, because permeates in that range certainly lack vinous character. But the result now is that anybody can sell you highly porous membranes that produce great flow at the expense of the depletion of soluble compounds with potential wine quality impacts. Now that we know more about the applications, it’s time to call upon TTB to clean up the resulting confusion. I recommend banning membranes above 150 daltons for use on wine.


TTB has no jurisdiction over suppliers and does not regulate the equipment they provide. It is the sole responsibility of the licensed winery to ensure that the membranes, operating conditions, resin columns, distillation and other procedures employed are both legal and advisable.

Try before you buy

The specialized know-how and high capital cost involved have, for the past 25 years, confined RO almost entirely to a service business. We are currently experiencing an explosion of U.S. wineries producing less than 5,000 cases (there are more than 7,000 such wineries, according to Wines Vines Analytics’ December 2016 numbers). These mom-and-pop shoestring operations throughout the United States can ill afford to lose a batch to volatile acidity, but they also have to watch their pennies. Besides the cost of a visit by a service provider, they often are equipped with low power and refrigeration capabilities that don’t accommodate large mobile systems.


These challenges have been cleverly overcome by VA Filtration, whose Sweetspotter is a small VA system for rent to small wineries. They will ship it anywhere, it rents at reasonable rates, and its low-energy operation allows it to be plugged into a 20-amp 110-volt wall outlet and to operate without refrigeration. It also has a terrific instruction manual.


At present, it’s the only alternative for the micro-winery outside California. It’s wonderfully ingenious and a lot better than nothing.

It’s a shame that it is unkind to wine. VA Filtration’s Sweetspotter employs NF membranes that can de-acidulate wine substantially with concomitant losses in flavor, color and texture. While tight membranes could be substituted, the resulting permeate flow is impractically slow.

Building a better mousetrap for rental work is a tricky challenge whose time has come. Given the demand, we are likely to see more systems in the near future. In the meantime, kudos to VA Filtration for an imaginative first step.

Choosing an RO dealer
European suppliers like Bucher-Vaslin, Della Toffola and Vason all build high-pressure machines that are as solidly reliable as the companies themselves. They have a lot of experience with reverse osmosis, dating back to the introduction of juice concentration in the 1980s, though not as long experience with California applications as the domestic companies that are smaller and less financially gold-plated.


The big companies offer machines as standard models. Obtaining custom design options is more challenging than with a small domestic manufacturer.


Mavrik, VA Filtration, WineTech and WineSecrets are all service companies that have a wealth of experience in California applications and sell machines. Domestic companies are generally more willing to customize equipment to your winery’s requirements. You might covet interchangeable membrane banks for red and white wine, or a separate programming for UF operations in spiral and tubular formats. They are more likely to let you “have it your way.”

Membrane cost is not as big a consideration as you might think. RO membranes can last for years with proper care. They are easy to replace and fittings are universally standardized. A well-built RO machine also requires little maintenance. That said, care in operation, cleaning and storage is a specialized skill. Membranes can be destroyed by high solids, elevated temperature, freezing, high pH, mold growth, bleach and well water among many other perils. Store membranes in a cool environment with 25% alcohol or 1,000 ppm SO2/citric solution. Cleaning should always be done with purified, unchlorinated soft water.

The biggest challenge in owning an RO machine is turnover of your trained staff. You will probably need regular retraining by manufacturer’s staff trained in your applications. One solution is to purchase from a local service provider, but several European firms have local experts, too.


Once you get your new RO, be sure to carefully document every detail of your winery’s procedural specifics.


Your new RO machine gets you a ticket to a wide variety of applications. There is much more to discuss concerning specific applications. Such topics as resin column selection and operation, permeate distillation, juice concentration logistics and loss prevention will be addressed in subsequent articles.

Posted on 07.08.2018 - 09:52:08 PST
Excellent information thank you Clark.
Paul Gardner