March 2020

Water Management

Demineralized water system design: Considerations for the petrochemical industry—Part 1

Site utility leads, engineers and other professionals in the hydrocarbon/chemical processing industries (HPI/CPI) are facing a perfect storm of increasing demineralized water demand, end of life of existing demineralized water plant equipment, changing source water quality, corporate directives to diversify water sources, and pressure from regulators and community stakeholders to minimize the volume of waste generated from water treatment.

Fan, J., Ananthanarayan, S., Hodgkinson, A., Advisian, a Worley Company

Site utility leads, engineers and other professionals in the hydrocarbon/chemical processing industries (HPI/CPI) are facing a perfect storm of increasing demineralized water demand, end of life of existing demineralized water plant equipment, changing source water quality, corporate directives to diversify water sources, and pressure from regulators and community stakeholders to minimize the volume of waste generated from water treatment. The last decade has seen large-scale investment in ethane crackers, and a second wave of crackers are being designed and engineered that will begin service in the next few years. These investments have resulted in a keen focus on raw water treatment and demineralization for steam production, which, along with ethane, is a basic input to the cracking process. Many refineries, especially those along the U.S. Gulf Coast, have also seen increased demineralized water demands as they upgrade their operations to produce higher-value products. The main steam uses in refineries include steam cracking, stripping, steam distillation and vacuum distillation. Steam is also used for process heating, pumping and electric power generation.

Over the last 10 yr to 15 yr, these developments have coincided with increased water stress from both droughts and floods in various watersheds tapped by HPI/CPI for their source water. Key design considerations and approaches utilized in demineralized water treatment system design for HPI/CPI are presented in this article. Alternatives for demineralization systems are compared and analyzed, and key decision-making criteria are discussed to introduce utility engineers and professionals to the thought process behind demineralized system design. The choice of demineralized water treatment technology boils down to several factors, including the type of raw water, temperature and associated chemistry, nature of demineralized water demand and quality of steam required (mainly the pressure). Constraints related to the footprint (especially in brownfield developments) and any wastewater volume limitations at the site should also be considered in the design process.

Source water

The following are types of source water that can be used in processing plants.

Surface water. Surface water sources for demineralized water treatment include rivers, lakes and oceans. Overall, 72% of all water used by refineries in the U.S. is drawn from rivers and lakes.1 Surface water is most often pumped directly from a water body or drawn through a canal by a quasi-state body, which then supplies the water to several industrial customers. However, the total volume allocated to users might be regulated by state or local authorities to ensure that supply is available for all users. Seawater is also used for once-through cooling in some coastal refineries. The use of seawater is limited to coastal refineries and chemical facilities due to high conveyance cost and associated corrosion-resistance material requirements. In addition, high ionic strength, compositional variability in estuarine locations, high suspended solids and macrofouling organisms all contribute to high maintenance cost and operational complexity in seawater systems.

Groundwater. It has been reported that 10% of the water used by refineries is drawn from groundwater.2 Groundwater can be very hard, with high mineral content, making it more expensive than lake water to treat into demineralized water. Some brackish groundwater is also available as a water source, but its high cost of treatment limits its use by most refineries. Furthermore, there are concerns related to the settlement of ground due to long-term groundwater extraction. Groundwater also typically has higher concentrations of reduced ions—such as iron and manganese—which, when brought to the surface, will oxidize and require specialized pretreatment prior to handling in a demineralized system. Combined, these factors mean that groundwater rarely provides a significant portion of the demineralized water supply at most industrial facilities—although there are some exceptions in the absence of alternate water sources.

Municipal wastewater reuse. Municipal wastewater generally consists of grey water and black water. Grey water includes water from bathing, hand washing and clothes washing, while black water includes water from kitchens, sinks and toilets. The combination of grey water, black water and storm water is treated in a municipal treatment plant and the effluent could be used in a refinery for boiler feedwater with further polishing, particularly for trace organics. The cost of conveyance from municipal wastewater treatment plants (WWTPs) into the refinery, along with extensive pretreatment to remove soluble organic carbon, should be taken into consideration and will result in increased treatment complexity and higher capital and operating costs than in instances when a typical city water supply is used. A more common strategy is to reuse treated municipal wastewater for cooling, which, in turn, frees up surface water for use as feed for demineralized water treatment.

Industrial wastewater reuse. Wastewater generated from oil refinery and petrochemical plants could potentially be reused after further treatment, as cooling water offers the prospect of freeing up higher-quality raw water for demineralized water usage. Cooling system blowdown water is also a potential candidate for demineralized system source water, after further treatment. However, in contemplating cooling water for use as demineralized supply, the water chemistry and process design must be thoroughly reviewed, with special consideration given to the capability of the demineralized plant to handle condenser leak events in the cooling system. The benefits of reusing refinery wastewater include consistent water quantity and quality; reduced demand on other water sources; and reduced discharge to water bodies. Total dissolved solids (TDS) concentration is the primary concern of industrial wastewater reuse. A TDS concentration of 1,000 mg/l or less is generally considered acceptable for reuse as cooling water, fire water and wash water. Typically, reusing any type of wastewater, including from industry, requires evaluation on a site-specific basis regarding wastewater availability, reliability, quality, energy consumption and process design.

Reusing all industrial wastewater with no liquid waste stream discharged into the environment is termed zero liquid discharge (ZLD). This water reuse concept has the potential to produce a portion of its recovered water as a high-quality distillate, which is very suitable for supplementing demineralized water production. It should be noted that ZLD usually entails a high treatment energy requirement, high capital cost (due to the need for exotic metallurgy) and high operational complexity. A decision to pursue a ZLD outcome is typically driven by either severe water scarcity and/or regulations and is rarely considered as part of site demineralized water strategy.

Water quality characterization

It is critical that appropriate water quality characterization studies are undertaken at the outset of any demineralized plant design project. For a relatively small sum of money (low thousands of dollars vs. hundreds of millions of dollars required for the demineralized plant construction and the billions of dollars for the overall facility), a good water quality data set can be collected to characterize and select source water and decide treatment process configuration. Often, data is not collected or inadequately detailed (e.g., some parameters are not collected, or no data is collected during times when the river is in flood or at minimum flow). To prepare a robust system design, data covering all typical quality and flow scenarios giving a complete picture of the anticipated water quality in the watershed is required. Understanding trends in suspended and colloidal solids, including colloidal silica, can be very valuable in the design of the treatment system. Assessing the potential for algal blooms and any anecdotal data related to past bloom events (e.g., duration and frequency) is also helpful in designing an effective pretreatment process. At a minimum, seasonal data over a period of 5 yr–10 yr is typically used as a starting point for a water chemistry basis of design. If available, longer periods of data—from site records or publicly available databases—can also be helpful to understand longer-term trends.


Prior to demineralized water treatment, most raw water sources must first undergo pretreatment to remove suspended solids and reduce scale forming hardness and/or compounds that have a propensity to generate corrosion. Depending on the source of raw water, water chemistry varies significantly. Relative ranges for key parameters are shown in TABLE 1. Main contaminants of concern for boiler feedwater include:

  • Suspended solids. Suspended solids will settle out on equipment surfaces and cause deposition.
  • Total hardness. In most waters, nearly all hardness is due to calcium and magnesium. Hardness can cause scale in heat exchangers, pipe and vessel surfaces, resulting in unnecessary downtime and reduced process performance (e.g., in heat exchangers).
  • Silica. Hard scales resulting from silica are called silica-based deposits. These result from either amorphous silica and/or magnesium silicate. Silica entering a boiler can also be carried with the saturated steam as silicic acid, which can cause precipitates on metal surfaces. Hence, silica is considered one of the most critical parameters in any demineralized water treatment system and it is essential to closely monitor and control silica in the demineralized water makeup supply and, in turn, the demineralized water.
  • Iron. Soluble and insoluble iron will combine with phosphates and hydroxides to form scale and cause corrosion and overheating problems.
  • Dissolved solids. High ranges of dissolved solids can cause process interference and foaming in the boiler.
  • Total organic carbon. Under high temperature and pressure, organic compounds can break down and form carbonic acid in the steam and condensate. This results in an increase in the conductivity of the steam and reduced pH in the condensate, increasing the propensity for corrosion around the steam system.
  • Sodium. Sodium can build up on critical components as steam condenses and causes embrittlement, leaks and cracks.

In demineralized treatment processes, it is critical that adequate pretreatment is used to remove colloidal materials and suspended solids. Otherwise, it can lead to the failure of the demineralized treatment system. The degree and complexity of the pretreatment equipment are determined by the raw water quality. In a small flow facility, a low level of suspended solids (< 30 mg/l) might be managed with just cartridge filtration. However, it is more common in the case of surface water for pretreatment to include disinfection, solids clarification and filtration. Two common pretreatment systems for suspended solids removal are media filtration and membrane filtration. The following will introduce these systems, along with their advantages and limitations.

Media filtration captures pollutants through physical filtration and adsorption. Typical media-based filtration systems are composed of sand, anthracite and other media. Media filtration can be utilized in either a fixed-bed or moving-bed configuration (FIG. 1). After a period in service (typically once per day or when differential pressure setpoint is triggered), fixed beds of sand and anthracite and/or other media combinations (housed in either pressure vessels or in concrete basins) are taken offline for backwashing (cleaning) to remove accumulated solids. This requires additional infrastructure (backwash tanks, associated pumps, sludge treatment, solids handling, etc.) and footprint. In a moving-bed (or continuous backwash sand filtration) configuration, the media is continuously backwashed and a separate backwash tank is not required. The backwash wastewater from both continuous backwash filtration and fixed-bed filtration are handled in similar fashion before disposal.

FIG. 1. Comparison between fixed-bed filtration and continuous backwash filtration.

Sometimes, media filters are used to treat soluble components in the water, usually following the primary particle filtration step. The most commonly used media to remove soluble components include activated carbon (for organics) and zeolite (for removing hardness). A comparison of three types of media commonly used in media filtration is shown in TABLE 2.

Membrane-based filtration, using ultrafiltration (UF) membranes that have a pore size of around 0.05 µm, is another way to remove suspended solids. The thin, permeable, hollow-fiber membrane layer physically excludes particles in the water, allowing only soluble components (and the water itself) to pass through into the ultrafiltrate. Two main formats of membrane filtration systems exist: pressure and vacuum-driven. Pressure-driven membrane systems use membrane elements installed in pressure vessels, and the membrane separation process is driven at 20 kPa–250 kPa (3 psi–36 psi). Vacuum-driven systems immerse the membrane modules in a tank, and the filtrate is drawn through the membranes via suction. A comparison of media filtration and UF is provided in TABLE 3.

The selection and performance of pretreatment technologies largely depends on feedwater quality. UF usually shows better removal efficiency for total organic carbon (TOC) and turbidity as compared with media filters for river water.7 UF is also less susceptible to particle breakthrough and offers a more reliable barrier than media filtration. However, UF is more sensitive to algae blooms, high natural organic loading, reactive species such as manganese, certain types of colloidal contaminants and water temperature variation. Increased biofouling has been observed on UF, compared to media filters for seawater pretreatment, in some instances.8 The selection and deployment of the most effective filtration solution should be based on site-specific water quality parameters. Other factors, such as footprint, construction cost and lifecycle costs, should also be considered in the final selection. If the filtered water is used as cooling tower makeup, the value of performance benefits from better-quality water (e.g., increased cycling of cooling towers and reduction in fouling of heat exchanger surfaces) should also be considered in the selection of filtration technology.

After pretreatment, the feedwater—free of suspended solids—passes through the demineralized process to remove undesirable ions. These include heavy metals and, in most refineries, the entire dissolved solids content of the water. The target quality of demineralized water treatment depends on the pressure at which the boiler is operated. The general principle is that the higher the pressure, the higher the quality of water required. A low-pressure boiler can usually tolerate some feed-water hardness, while almost all impurities must be removed from water used for high-pressure boilers. Boiler feedwater quality requirements issued by the American Society of Mechanical Engineers (ASME) are shown in FIG. 2.

FIG. 2. Boiler feedwater quality requirements at various pressures. The actual target silicon dioxide (SiO2) values in the demineralized water are set based on the site-specific considerations, including boiler cycles and types of metallurgy involved. Usually, the concentration will be in the range of 10 μg/l–20 μg/l for greater than 6,000 kPa (approximately 900-psi) steam.

To ensure the purity of boiler feedwater, key parameters, such as TDS and silica, must be monitored. Conductivity is one of the most commonly used methods to obtain a rapid and reasonably accurate measurement of water/steam purity. However, some gases (especially carbon dioxide) can ionize in water solution and interfere with measurement of dissolved solids by increasing conductivity. The measurement of silica can use a colorimetric technique, while the reliability of this method is low under high temperature/pressure (form of silica can change from non-reactive colloidal form into reactive ionic form). Online measurement of sodium using specialized glass electrodes can provide a very sensitive indication of contaminants in water and/or steam condensates. Sodium measurement can give reliable measurements down to a concentration of 0.1 µg/l, which is well below the purity at which electrical conductivity can be measured reliably. However, sodium electrodes can be prone to drift and must be regularly calibrated. Each of these three methods can provide an indication of boiler feedwater purity, and the combination of any two generally provides reliable confirmation of boiler feedwater purity.

The common demineralizing technologies include ion exchange, reverse osmosis and electro-deionization. Frequently, a combination of these technologies is used to achieve the target water quality specification. Thermal methods to demineralize water, once quite common, especially in the Middle East, are now usually only considered in instances where a low-cost source of thermal energy is available.

Part 2

Part 2 of this article, to be published in the April issue, will discuss the common technologies and approaches considered for water demineralization and introduce the reader to demineralized water treatment design variables that must be considered for an effective design. HP


          The authors would like to thank Karen Leber for her contribution to this article.


  1. Blieszner, J., R. Henderson and L. Weaver, “Potential vulnerability of U.S. petroleum refineries to increasing water temperature and/or reduced water availability,” U.S. Department of Energy, January 2016.
  2. Sun, P., et. al, “Estimation of U.S. refinery water consumption and allocation to refinery products,” Fuel, June 2018.
  3. Nasrabadi, T., et. al., “Bulk metal concentrations versus total suspended solids in rivers: Time-invariant and catchment-specific relationships,” PloS one, January 2018.
  4. “Iron and water: reaction mechanisms, environmental impact and health effects,” Lenntech,
  5. “Hardness in groundwater,” Water Stewardship Information Series, February 2007,
  6. Houston Public Works water quality report, 2017,
  7. Abbasi-Garravand, E., et. al., “Using ultrafiltration and sand filters as two pretreatment methods for improvement of the osmotic power (salinity gradient energy) generation process,” 4th Climate Change Technology Conference, Montreal, Canada, 2015.
  8. Badruzzaman, M., et. al., “Selection of pretreatment technologies for seawater reverse osmosis plants: A review,” Desalination, January 2019.

The Authors

Related Articles

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}