Refineries are rebuilding their boiler feedwater trains — and the numbers back it up

Engineers are moving to membrane-heavy designs with smart polishing to push steam purity, slash chemicals, and tame blowdown. The data — from turbidity to TDS — point to a clear playbook.

Industry: Oil_and_Gas | Process: Downstream_

When boiler feedwater gets cleaner, refinery balance sheets do too. Effective pretreatment and demineralization can double boiler cycles — for example, from 7× to 14× or more — while typical reverse osmosis (RO) adoption yields 50–60% lower chemical consumption and 4–5% lower blowdown, according to an industry analysis (Water Online) (Water Online). Those gains ride on a simple premise: remove turbidity, drop total dissolved solids (TDS), and let the boiler run clean.

Refinery utility and water treatment engineers are converging on a comprehensive train: clarification and deep-bed filtration to drive turbidity below 1 NTU (nephelometric turbidity units), primary demineralization via ion exchange (IX) or RO, and a final polish — often mixed-bed IX or continuous electrodeionization (CEDI) — before deaeration and chemical conditioning. Each choice carries trade-offs in energy, chemicals, and waste volumes, with credible data spelling out the deltas (Power Engineering), (ResearchGate), (Power Engineering) (Power Engineering).

The market signal is loud: boiler demineralized water equipment was roughly $62.6 billion in 2023 and is projected to grow ~6.4% annually through 2030 (Verified Market Reports).

Pretreatment: clarification and media filtration

Raw makeup water first gets clarified and filtered. Typical coagulation/flocculation (for example, 5–10 mg/L alum or polymer dosing) is followed by a settling clarifier or dissolved air flotation (DAF; air-induced flotation) to remove >90% of suspended solids and organics (Power Engineering). Engineers commonly apply coagulants — a fit for a coagulants program — and meter them with dosing systems such as a dosing pump to keep treatment stable.

Clarifier effluent is then polished by a deep-bed multimedia filter — for example, 0.5–0.7 m anthracite over 0.4–0.5 m sand — to achieve turbidity <1 NTU (Power Engineering). In practice, well-designed clarifier plus media filters deliver 0.3–1.0 NTU, sufficiently low for downstream RO or ion exchange (IX) (Power Engineering). For media beds, refiners typically specify anthracite and sand/silica grades tuned to flux and backwash rates.

Cartridge prefilters (5–25 µm) are also used immediately upstream of RO trains (Power Engineering), where a cartridge filter becomes the final guard. For high-chlorine raw water, an activated carbon filter is advisable to protect membranes and resins — a straightforward fit for activated carbon stages.

Design note: Pretreatment sizing must handle peak loads. Buecker et al. report that storm runoff can raise raw TSS by 10–100×, potentially overwhelming filters (Power Engineering). A design margin — for example, allow 2–3 m/h (meters per hour) hydraulic loading and automatic backwash every few hours — is prudent. Typical clarifier detention (~30–60 min) and media-filter flux (5–10 m/h) are selected so that even high-turbidity water is reduced to <5 NTU entering demineralizers. Pressure gauges and turbidity monitors should be installed to signal resin swap-out or filter backwash.

Hardware choices mirror those demands: a conventional clarifier handles high solids, while a DAF option can target oily loads common around refineries. Filter housings and internals must tolerate frequent backwash and industrial pressures.

Primary demineralization: IX vs. RO

After pretreatment, plants remove dissolved ions and gases to boiler feed standards. Two main approaches dominate: ion exchange demineralization and reverse osmosis (RO).

Ion exchange demineralization (IX). Traditional two-bed ion exchangers — a cation column in H⁺ form, then an anion column in OH⁻ form — often finish with a mixed-bed polisher and can produce essentially 100% demineralized water. One power-plant feed system, for example, used chemical decarbonization (caustic soda), sand filtration, and sequential strong-acid and strong-base exchangers, finishing with a mixed-bed polisher (ResearchGate). This can achieve conductivity on the order of 0.1 µS/cm (≈10⁶ Ω·cm resistivity) and hardness <0.01 mg/L, meeting high-pressure boiler specs. However, the downside is high chemical and waste loads: regeneration of the cation resin with HCl and the anion resin with NaOH produces highly concentrated brine and regenerant effluent requiring neutralization. Lukic et al. note the current IX-based process yields very low dissolved-salt discharge per volume, but consumes “a large amount of chemicals” (ResearchGate). In their case, replacing IX with RO would increase throughput by ~17.5% and blowdown volume by ~150%, but drastically reduce salt loading (by 20–30×) in the waste stream (ResearchGate). This implies that an IX system will generate smaller effluent volumes but with very high salt/chemical loads, whereas an RO-based plant will discharge 2–3× more volume of much lower-salinity brine.

Operational data: Two-bed IX can remove essentially all hardness and silica: for example, cation exchange will reduce hardness from tens of ppm to <0.1 ppm, and anion exchange then reduces TDS to a few ppm (ResearchGate). After polishing with a small mixed-bed column, conductivity in-feed to the boiler can be below 0.2 µS/cm (≈5 MΩ·cm). Thermal-cycle of concentration can be raised significantly (e.g. from 5–10 up to 20+), reducing boiler blowdown frequency. Typical chemical consumption is on the order of 5–50 g NaOH and 3–30 g HCl per m³ treated (depending on hardness) and produces ~5–20 L of spent regenerant per m³. For context, Lukic’s analysis implied ~0.5 m³ more input water per m³ treated with RO vs. IX (ResearchGate), illustrating the larger makeup needs of an RO scheme.

Engineers specifying IX typically look at a packaged demineralizer train backed by a cation/anion exchange system and high-capacity ion‑exchange resin inventories to sustain long runs between regenerations.

Reverse osmosis (RO). Modern refineries increasingly use membrane treatment as the core demineralizer to cut chemical costs and improve water recovery. A multi-stage RO system (brackish-water or high-recovery design) can reject 95–99% of dissolved salts (Water Online). A well-designed RO train typically achieves >98% TDS removal (Water Online), with permeate quality of 50–150 µS/cm (depending on feed TDS), and concentrate containing the bulk of the impurity load. RO also removes silica and hardness to a large extent, often giving 90–95% silica rejection if properly guided at neutral pH. For this duty, refineries often opt for a brackish-water RO configuration.

Pros: Adding RO ahead of IX slashes demineralizer regeneration frequency and chemical use. Sharpe (U.S. Water) reports that integrating RO can cut boiler-treatment chemical usage by ~50–60% (Water Online). Reduced feed alkalinity also lowers neutralizing-amine demand. Less scaling in the boiler yields ~3–4% fuel savings (by higher heat transfer efficiency) (Water Online). RO increases the boiler’s allowable cycles of concentration: for example, the same boiler blowdown rate now contains much lower TDS, permitting 4–5% reduction in blowdown frequency (Water Online). RO concentrate (reject) may be reused: with a softener ahead of the RO, the concentrate quality (low hardness) can even be used for cooling towers or other plant make-up (Water Online).

Cons: RO requires robust pretreatment (as above). It also requires high-pressure pumps (energy ~0.5–1.5 kWh/m³) and generates a waste stream (often 10–15% of feed volume) that must be handled. If concentrates cannot be reused, that 15–25% of feed becomes blowdown. Design-wise, with RO the engineer must size two or three membrane stages and plan for periodic CIP (clean-in-place), as well as consider energy recovery (e.g., isobaric exchangers) if feed TDS is high (Water Online).

RO+IX hybrid. In many designs, RO is combined with a small post-polisher. Because RO permeate alone may still contain 10–50 mg/L ionic content (and trace silica), a mixed-bed polish or continuous EDI is usually added to reach the ultra-pure feedwater needed for high-pressure boilers. Lukic et al. specifically recommend retaining a final mixed-bed after RO to achieve feed specifications (ResearchGate). Empirical data show that RO+polisher can produce the same purity as IX/MB but with far lower overall chemical waste.

Final polishing and conditioning steps

ChatGPT Image Oct 10, 2025, 04_23_27 PM

Mixed-bed ion exchange. A small mixed-bed condensate polisher blends cation and anion resins to continuously scavenge residual ions, driving feedwater conductivity to near-zero (10⁶–10⁷ Ω·cm) and silica to the ppb (parts per billion) range, well below boiler limits. Industry rule-of-thumb: achieve <0.01 mg/L silica for >600 psig boilers. These systems are skid-mounted and automatically regenerated by diversion in batches, reliably polishing an RO or two-bed output to two-bed quality. They still require periodic chemical regeneration (HCl/NaOH) and handling of brine. For this duty, refiners frequently specify a mixed-bed polisher.

Electrodeionization (EDI/CEDI). An alternative is continuous EDI. Recent advances allow “chemical-free deionization” that is now cost-competitive with conventional IX and well-suited for boiler feed polishing (Power Engineering) (Power Engineering). In an RO-CEDI scheme, RO pretreatment lowers TDS so the EDI can produce mixed-bed quality; with RO removing ~98% of salts, a two-stage EDI can raise water resistivity to >1 MΩ·cm (Power Engineering). Because CEDI continuously regenerates via an electrical field (no regeneration chemicals), it eliminates brine waste. It does require very clean feed (hardness <0.01 mg/L, TOC <0.1 mg/L) and carries higher capital cost. Many large refineries requiring the highest purity (<0.1 µS/cm) are beginning to adopt RO+CEDI trains (Power Engineering) (Power Engineering). Packaged options include a standalone EDI/CEDI unit.

Deaeration and chemical conditioning. Polished water should pass through a deaerator (steam-stripping tower) to remove dissolved gases. A steam-heated deaerator typically raises water to ~100–110 °C and reduces O₂ to <7–10 ppb. A small slipstream from the deaerator feed may also serve as condensate polish (removing any silica or ionic carryover from returned condensate), commonly handled by a condensate polisher. Finally, boiler chemicals are added: oxygen scavengers (sodium sulfite or hydrazine at 0.2–0.4 ppm O₂ scavenging levels) and pH adjusters (toluene sulfonate or amine to ~8.5–9.5 pH) to protect metal surfaces. For steam reuse, neutralizing amines or amines + phosphate may be used; selection should follow Indonesian effluent rules if condensate is vented. These tasks map to an oxygen scavenger and a neutralizing amine program.

Performance outcomes and adoption trends

Well-designed pretreatment plus demineralization dramatically improves boiler reliability. Effective RO pretreatment can double boiler cycles — for example, from 7× to 14× or more — because the feedwater TDS is so low (Water Online). Typical RO adoption yields 50–60% lower chemical consumption and 4–5% lower blowdown (Water Online), directly boosting operating profit by reducing utility and supply costs. If a refinery can re-use RO concentrate (e.g., for cooling or process), even more water is conserved (Water Online).

Economically, the equipment market was ~$62.6 billion in 2023 and is projected to grow ~6.4% annually through 2030 (Verified Market Reports). In Indonesia specifically, stringent environmental controls on discharge and the prevalence of seasonal turbidity make RO-based designs attractive (since they sharply reduce high-salt brine effluent) (ResearchGate). Utilities should weigh trade-offs quantitatively: Lukic et al. showed replacing IX with RO would modestly increase raw water demand (by ~17%) but drastically lower salt load in plant effluent (20–30× less) (ResearchGate).

What a modern refinery train includes

The core blueprint is consistent: (1) robust pretreatment — flocculation/clarifier plus multimedia filtration — to <1 NTU (Power Engineering); (2) primary demineralization by either two-bed ion exchange or an RO train, often with a preceding softener to optimize performance (Water Online); and (3) final polishing by a mixed-bed IX or a continuous EDI unit to meet boiler-spec purity (Power Engineering) (Power Engineering). IX gives extreme purity at the expense of chemical OPEX, while RO/RO+CEDI cuts chemicals and blowdown but requires energy and waste management — with the benefit of far less saline waste (ResearchGate).

For pretreatment hardware selection, engineers lean on a clarifier for suspended solids control and media filters using sand/silica, with cartridge filters guarding RO membranes. On the back end, polishing via a mixed-bed or EDI, plus a condensate polisher where required, rounds out the train.

The result — when built to the data above — is stable turbidity control, membrane-friendly feeds, and boiler purity benchmarks that hold under seasonal upsets.

Sources: Power Engineering; ResearchGate (link); Water Online (link); Power Engineering (link); Verified Market Reports.

Chat on WhatsApp