Reverse osmosis systems are typically built according to a standard designs developed by the original equipment manufacturer (OEM). After the order is received, the equipment is shop assembled, mounted on a skid, and shipped to the customer for installation. The final result is a system that is a close fit with the customer’s floor space, water quality and permeate flow requirements.
Standard design RO systems are less costly for the OEM to produce because the designers only have to engineer the system once. This saves time and expense as compared to producing custom-built RO systems that meet the customer’s more-exacting specifications. From the end user’s point of view, however, the standard design model typically requires the purchaser to settle for a system that is a compromise between the OEM’s standard design and the optimum design for that particular application. A fair analogy would be the difference between purchasing a model home built on speculation by a general contractor versus a custom home designed by an architect and built to your specifications. In the first case, you live in a home that the contractor has selected based on his tastes and preferences as compared to a home you designed to meet your specific needs. In a similar way, it is often to your advantage to take a more active role in developing RO design specifications to insure that you get a system that suits your specific needs and not those of the OEM.
Here are some basic considerations that one needs to take into account when developing specifications for a custom-built RO system.
The local water quality greatly influences the design of an RO system. A key water quality parameter used to evaluate the suitability of an RO feedwater source is the Silt Density Index (SDI). This value is not commonly reported on standard water analyses, but can be determined easily using an SDI filter apparatus. Well water sources generally yield lower SDI values than surface water supplies. SDI values greater than 3 suggest the water is prone to fouling the membranes. Since water quality may fluctuate, it is best to take SDI measurements over the course of several days or weeks to arrive at a good working average value.
Water supplies that produce an SDI greater than 3 can be treated by filtration to remove suspended impurities. This is commonly accomplished by multimedia filters. These backwashable filters consist of graded layers of anthracite, gravel and sand, which remove impurities down to 30 microns.
Many municipal water supples contain chlorine, which must be removed prior to the RO to prevent oxidation of the thin film composite (TFC) membranes. One dechlorination method is to pass the water through an activated carbon filter. Carbon has an infinite capacity for chlorine removal. It also has the ability to absorb organic molecules and thus reduce the total organic carbon loading on the RO. The other option is to dechlorinate by chemical injection using sodium bisulfite. Many RO systems include chemical injection ports in the standard design to allow for this. The capital cost for a chemical metering pump and tank is less than for a carbon filter. However, a chemical feed system is higher maintenance and labor intensive as compared to a passive carbon filter bed. My preference leans toward carbon filtration for most projects.
Final filtration of the RO feed includes the use of 5 micron filters ahead of the RO array. These filters are available as either replacement cartridges or bags. This is the last line of defense to protect the membranes from fouling. Both absolute and nominal filters are available. While these terms are somewhat poorly defined, absolute filters are rated to remove 99% of particulates greater than 5 micron in size. Absolute filters are more expensive than nominal filters. For most RO designs, nominal filters are acceptable and sufficient to protect the membranes. The filters are replaced when the pressure differential exceeds 15 psi.
MINERAL SCALE DEPOSITION
The dissolved solids in the feedwater concentrate 4-fold as the water passes through the RO array. This promotes the deposition of sparingly soluble salts on the membrane surface. These impurities include the salts of calcium, magnesium, barium, strontium, boron, silica and iron. If these are not controlled, premature fouling of the membrane surfaces and water passages will occur. A key indicator that this is occurring is a steady increase in the pressure differential between the feedwater and reject streams. When the pressure differential increases by 15%, the RO should be removed from service for chemical cleaning.
The Langelier Scaling Index (LSI) is typically used to predict the scaling tendency of the feedwater for waters less than 10,000 micromhos/cm. For water sources greater than 10,000 micromhos/cm and for sea water, the Stiff Davis Index is used. These index values are easy to compute using the results of a standard water analysis. A positive index value suggests that the water supply is scale-forming and should be pretreated prior to use as RO feed.
Three methods for scale prevention are used: (1) acid injection for pH control, (2) chemical antiscalant injection, and (3) ion exchange softening.
Many standard design RO skids come equipped with chemical tanks, metering pumps, and static inline mixers to allow for pH adjustment and antiscalant injection. Acid neutralizes bicarbonate and carbonate alkalinity to prevent its reaction with calcium and magnesium hardness. By carefully controlling the acid dosage, it is possible to prevent the deposition of these scale-forming salts. Likewise, chemical antiscalants work to increase the solubility of sparingly-soluble mineral salts by sequestration, chelation and dispersion. As mentioned with regard to chemical dechlorination, the success of this approach depends on diligent maintenance of the chemical feed system. And, of course, any chemical(s) that are added to the RO feedwater increase the ionic loading and must be removed by the membranes. An accidental overfeed (or underfeed) of acid or antiscalant can irreversibly foul or damage the membranes.
The other option for scale control is to soften the RO feedwater by ion exchange. A standard water softener removes calcium, magnesium and other cations by sodium ion exchange. This produces a feedwater that is non-scaling. The downside to a water softener is that it must be periodically regenerated with salt, which produces more wastewater than with the acid and antiscalant chemical injection options. The initial cost for a water softener is greater than for a chemical injection system. Overall, all things considered, my preference is to include a water softener in the system design for those cases where the control of mineral scale is a requirement.
Reverse osmosis membrane manufacturers offer software to assist in the design of RO systems. Using this tool, the design engineer can enter the raw feedwater data, calculate the scaling potential, evaluate various design configurations and project the final permeate water quality. This software is readily available for download and is recommended for anyone wishing to custom design/build an RO system or to forecast the performance of a proposed OEM design or existing system.
Some general guidelines are in order. A key design parameter is the RO flux. Flux is the gallons of permeate per square foot of membrane surface area per day (gfd). The flux should be equal to or less than 15 gfd. For example, a 100 gpm RO produces 144,000 gallons of permeate per day. To produce a flux of 15 gpd, the RO array requires 9,600 square feet of membrane surface area. A typical spiral wound, 8-inch membrane module contains 365 sq ft of surface area. Therefore, 27 modules are required. However, newer RO modules are available that contain 400 sq ft of membrane. If these are selected, 24 modules are required. Pushing the flux beyond the 15 gfd guideline by using fewer modules to reduce cost is not a good practice as this shortens the useful life and reliability of the membrane system.
Membrane modules are housed in pressure vessels. Some design configurations use 6 modules per pressure vessel. A standard 8 inch diameter module is 40 inches long. Putting 6 modules in a pressure vessel results in an assembly that is over 20 feet long. This makes for a foot print that is often difficult to fit into the available floor space. A newer trend is to design the RO with 4 modules per pressure vessel. This reduces the length by about 6 feet, which can be significant in some applications.
The other consideration that needs to be made is how to stage the pressure vessels in the array. Let’s say that we require 24 RO modules (each 400 sq ft) to achieve a flux of 15 gfd. We would then need 6 pressure vessels (PV) to house 4 modules per PV. Should the array consist of 4 PV feeding 2 PV (4:2 array) or some other configuration? It’s here that the software is of great value in that the design engineer can work with various combinations to determine which works best based on the feedwater quality, flux and other criteria.
Most RO systems are designed with a recovery of 75%. Recovery refers to the percent of feedwater that is recovered as usable permeate. In our example, a 100 gpm permeate flow operating at 75% recovery requires a feedwater flow of 133 gpm. The balance, 33 gpm, is sent to drain as concentrated waste. 75% recovery is a standard OEM performance expectation. On some feedwater supplies, however, higher recovery rates are achievable. And higher recovery rates produce an improvement in the overall efficiency of the RO. Once again, the design software is of value in determining if this goal is achievable.
Once the design elements have been established, the layout of the RO skid can be determined. Using our example of an RO designed with 6 pressure vessels, several options exist. The layout can be 6 x 1, 3 x 2, 2 x 3, or even 1 x 6. Some designers prefer to keep the height of the RO skid as low as possible to help in the loading and unloading of the membrane modules. It’s difficult to pull 8 inch modules out of a pressure vessel while standing on a ladder, for example. However, in some cases, floor space is limited and it is best to stack the pressure vessels as high as possible. Once again, this is where custom design RO systems shine in that they can be configured to accommodate the specific needs of the owner.
One final point, RO systems require control and monitoring instrumentation and software. Most standard design systems include a Human Machine Interface (HMI) that allows the engineer to operate the system and monitor its performance. A custom design RO allows one to specify the number, type and brand of these devices. It also insures that the control instrumentation is compatible with the plant’s existing network.
The choice between purchasing a standard design versus custom design RO system is one worth considering. Many standard design RO systems are currently in service producing a reliable, continuous supply of RO permeate for a variety of applications. Standard design ROs are generally less costly to manufacture and therefore offer some savings over custom design systems.
That said, adopting the engineering approach of design/bid/build allows the end user to specify the details of the RO system. This includes pretreatment requirements, scale and fouling control, physical configuration and control instrumentation and software. This approach offers the advantage of guaranteeing that the final RO design will meet or exceed the specific needs of the end user.