High‑pressure HRSGs run on water that’s closer to a lab reagent than a river. Operators are targeting ≥10+ MΩ·cm resistivity and near‑zero contaminants, then steering the boiler with an all‑volatile chemistry program built on ammonia and a volatile oxygen scavenger.
High‑pressure HRSGs (heat‑recovery steam generators) live or die by water purity and tight cycle chemistry. The modern standard: feedwater at ≥10+ MΩ·cm resistivity (megaohm‑centimeter, a measure of water’s electrical resistance) and virtually zero contaminants. Completely deionized water clocks in at ≈0.055 µS/cm at 25 °C (~18 MΩ·cm) (microsiemens per centimeter, the inverse of resistivity) (www.power-eng.com), and guidelines call for ≤0.1 µS/cm (≥10 MΩ·cm) in makeup water (www.power-eng.com).
The mineral budget is ruthless: industry targets are Na ≤2 ppb and SiO₂ ≤10 ppb (www.power-eng.com). Anything higher deposits in high‑heat‑flux tubes or on high‑pressure turbine blades. For perspective, ordinary potable water sits at 100–200+ µS/cm with hundreds of ppb of salt — totally unacceptable for 10+ MPa steam.
Ultrapure makeup water targets
Design targets for the polished makeup stream are exceptionally stringent: specific conductivity ≤0.10 µS/cm (≈10 MΩ·cm) (www.power-eng.com), with silica ≪0.01 ppm and typical design at SiO₂ ≤10 ppb (www.power-eng.com). Sodium and alkalinity are held to the single‑ppb range. In practice, operators may measure the polished water at 106–108 Ω·cm (0.01–0.02 µS/cm), well above the 10 MΩ·cm goal, to build in margin.
Final water should meet ASME “ultrapure” criteria for high‑pressure feedwater — essentially no hardness, low TOC (<100–200 ppb), and pH ≈7 (neutral) before conditioning (studylib.net) (www.power-eng.com).
RO‑based demineralization train
The go‑to architecture is a multistage demineralizer centered on RO (reverse osmosis) plus polishing. Raw water — river, lake, or even seawater — is pretreated to protect membranes, then treated with dual‑stage RO followed by ion exchange or EDI (electrodeionization) polishing. Many plants specify integrated membrane systems so RO, NF, and UF are engineered as one train for industrial duty.
Adding ultrafiltration (UF) ahead of RO is now common. One case study found that installing microfiltration (300 gpm) cut RO feed turbidity from ~0.5–1.0 NTU to <0.05 NTU (www.power-eng.com), a ~95% particulate reduction that “led to a dramatic reduction in RO cartridge filter and membrane cleaning frequency” (www.power-eng.com). As pretreatment, compact ultrafiltration units are frequently paired with RO to stabilize turbidity.
Dechlorination and solids control also matter. Plants often include carbon beds and high‑surface‑area filters to shield membranes; for example, upstream activated carbon media can remove oxidants and organics, and upstream cartridge filters can blunt fine particulate spikes that would otherwise foul RO.
For brackish and seawater sources, RO staging is mandatory. Plants using brackish groundwater typically add 2–3 stages and brine reject handling; compact brackish‑water RO skids are sized for maximum TDS of 10,000. Where intake is marine, power plants often turn to sea‑water RO, including containerized systems for industrial duty.
Polishing is either ion exchange or EDI. Many sites prefer ion‑exchange polishers for maximum silica and sodium removal; others opt for chemical‑free EDI stacks to avoid acid/caustic handling. Mixed‑bed trains remain a staple where ultimate purity is needed; high‑grade mixed‑bed polishers routinely deliver less than 20 ppb silica and very low TDS.
Pretreatment and source variability
Pretreatment note: if raw water is hard or saline, ion‑exchange softening or RO desalting is mandatory. Plants drawing on hard groundwater commonly add a dedicated softener ahead of membranes to prevent scale formation on high‑pressure surfaces.
Indonesian plant water sources vary (river, well, or municipal). Whatever the source, treat to the above targets. Water storage tanks and piping should be stainless or lined, as even trace iron causes fouling. Mixed‑bed or EDI polishers are periodically regenerated or replaced to maintain purity. For RO element selection, many plants specify robust membranes — for example, UF/RO elements from established lines such as Toray — to match duty cycles without adding chemicals or new operating modes.
Capacity matters. Raw‑water consumption is often on the order of 10–20 gal/MWh (wide variation), so a 500 MW HRSG might need ~15,000–30,000 gal/hr of estimated demin water at full load. Water‑treatment capacity should be sized so annual demineralizer throughput (minus regenerant waste) equals total make‑up demand. Many plants use off‑site regenerated ion‑exchange canisters or on‑site EDI for polishing, avoiding caustic/acid handling.
Controls and materials are part of the design envelope. Any generator makeup is metered with conductivity and RO permeate monitors. Remote telemetry and alarms for turbidity, chlorine, or organics are installed to prevent upset. All resins, valves and pots must use high‑purity materials (copper alloys avoided) to prevent leaching.
All‑volatile treatment (AVT) chemistry
Once ultrapure water is available, only volatile additions are fed to the boiler and feedtrain. The standard is All‑Volatile Treatment, AVT (volatile alkalization and oxygen scavenging), using ammonia for pH control and a volatile oxygen scavenger to strip dissolved O₂ (www.power-eng.com) (studylib.net). No phosphates or polymers are added to the drum.
pH control with ammonia and amines
Ammonia (NH₃) in water forms ammonium and OH⁻, elevating pH to passivate carbon steel with a protective magnetite film and counteract acidity from CO₂ ingress. The target feedwater pH is typically ~9.6–10.0 (25 °C) (www.power-eng.com). ASME consensus for 100+ bar boilers cites pH ≈8.8–9.6 (studylib.net), and IAPWS guidance calls for feed pH >9.4 to avoid FAC (flow‑accelerated corrosion) in low‑pressure circuits (ebrary.net).
Because adding NH₃ raises total conductivity, cation conductivity after a hydrogen column — CACE (cation conductivity after cation exchange, a contamination detector that removes weak bases) — is used to detect non‑volatile ingress (www.power-eng.com). In addition to ammonia, many plants use one or more volatile amines such as cyclohexylamine, morpholine, or ethanolamine to tune pH distribution; neutralizing blends are commonly delivered via high‑accuracy dosing pumps, with formulations aligned to options like neutralizing amines.
Film‑forming amines (FFA) — very high‑volatility organics that create protective monolayers — are sometimes used in large systems to protect long piping runs, but are less common in simpler HRSGs.
Oxygen scavenging and ORP control
Dissolved oxygen is controlled in two stages: thermal deaeration by heaters to ≈10–20 ppb, then oxygen scavenger addition to drive O₂ <5–10 ppb before the drum. Hydrazine (N₂H₄) historically has been standard, typically dosed to achieve ~0.02–0.1 ppm residual (studylib.net), but newer alternatives (carbohydrazide, octahydro‑1,3,5,7‑tetrazocine (OHT), or sodium sulfite for closed feedheaters) are used due to hydrazine’s toxicity. In fact, “oxygen scavengers were developed to replace hydrazine, as it has suspected carcinogenic properties” (studylib.net). Volatile scavengers for boiler duty are widely supplied in packages akin to oxygen scavengers, and metered with dedicated dosing pumps to maintain tight residuals.
Scavenger dosage is constrained by FAC risk. Over‑dosing can induce FAC, per IAPWS guidance (ebrary.net). The goal is a small reducing reserve — typically –250 to –350 mV ORP (oxidation‑reduction potential) at the feedpump discharge — while keeping O₂ near zero (ebrary.net). In mixed‑metal circuits with copper, even small oxygen levels (6–8 ppb) are tolerated to protect copper alloys (AVT(O) mode), but most all‑ferrous HRSGs run AVT(R) (near‑zero O₂).
Operating chemistry limits
With AVT in place, typical operating targets at the economizer inlet are: pH (25 °C) 9.6–10.0 (often ~8.5–9.0 at drum temperature) (www.power-eng.com); dissolved O₂ <10 ppb (ASME consensus <7 ppb) (studylib.net); total iron as low as measurable, ideally <10 ppb in drum/boiler (www.power-eng.com) (ebrary.net) — and best practice is <2 ppb in feedwater to control FAC (www.power-eng.com); CACE typically <0.2 µS/cm (www.power-eng.com) (ebrary.net); and silica in feedwater <10–20 ppb (ebrary.net). Steam‑pressure‑dependent turbine limits apply; for example, ~0.06 mg/kg SiO₂ in a 60 bar drum.
Non‑condensible gases (air ingress) are vented by a deaerator or condenser vacuum ejector. Feedwater and condensate polishing are used if condensate return exceeds ~20–30% or if impurity ingress is suspected; on HRSG cycles, compact condensate polishers provide mixed‑bed cleanup after heat exchange cooling.
Monitoring and performance outcomes
Instrumentation is non‑negotiable. Online monitors track specific conductivity and CACE, pH (calculated), and dissolved oxygen (ECD probes). Offline lab analyses confirm silica, sodium, iron, copper, ammonia, and amine levels daily or weekly. Trending these values quantifies program success.
Data‑backed targets guide decisions. For instance, “keeping feedwater iron below 2 ppb” is associated with “truly effective FAC control” (www.power-eng.com). In one evaluation, moving from ~5 ppb to <2 ppb total iron in feedwater extended clean‑up cycles by years. Maintaining CACE well under 0.1–0.2 µS/cm can halve corrosion product transport compared to lax chemistry. Conversely, slipping above 0.2 µS/cm is a red flag for contamination ingress (possibly condenser leaks).
Quantifiable outcomes include dramatically lower deposit rates and higher availability. After instituting strict RO/UF pretreatment and AVT chemistry, one plant saw RO membrane cleaning intervals double and economizer tube life extended (no tube leaks) over a 5‑year span. In industry surveys, plants with AVT programs report <2–3 unplanned shutdowns per decade attributed to water chemistry, whereas poorly treated units often succumb every 1–2 years. Published statistics cite corrosion‑related tube failure as a leading cause of HRSG outages.
Key measurable goals and governance
Key measurable goals: maintain make‑up resistivity >10 MΩ·cm and silica <10 ppb; maintain feedwater CACE <0.2 µS/cm, pH ≈9.8 and O₂ <10 ppb; keep total Fe/Cu in drum water <10 ppb (ideally <2 ppb for FAC control) (www.power-eng.com) (www.power-eng.com) (studylib.net) (www.power-eng.com). Regular reports and audits should track these values against limits. Following these data‑backed guidelines (from ASME, EPRI and IAPWS technical guides) ensures that the water/steam cycle operates on the “razor’s edge” safely (www.power-eng.com).
Evidence base and references
Inline citations: design criteria and outcomes are supported by industry sources (www.power-eng.com) (www.power-eng.com) (studylib.net) (www.power-eng.com) (www.power-eng.com), ensuring this water‑treatment program is grounded in proven practice.
Sources: Authoritative guides and case studies from power‑plant chemistry literature and industry standards were used. For example, Buecker & Perryman (Power Eng. 2019) detail RO/UF makeup design (www.power-eng.com) (www.power-eng.com); Buecker & Kuruc (Power Eng. 2022) provide HRSG feedwater pH/O₂ limits and FAC concerns (www.power-eng.com) (www.power-eng.com); Setaro/GE Water (ASME consensus) supply numeric water chemistry targets for 100+ bar boilers (studylib.net) (studylib.net); and IAPWS/EPRI guidelines discuss AVT objectives and risks (ebrary.net) (ebrary.net). These and other industry references form the basis of the recommended scheme.
Availability and risk posture
By meeting the outlined purity and AVT benchmarks, high‑pressure HRSGs in Indonesia or elsewhere can achieve reliable, high‑efficiency operation. Any deviation — for example, makeup conductivity rising to 0.5 µS/cm, or feedwater pH dropping below 9.2 — should trigger immediate corrective action (more polishing, chemical dosing adjustment, etc.). Over time, this preventive approach is far more cost‑effective than boiler tube repairs or forced outages. The industry cases and guidelines show that upgrades to stringent water treatment — albeit requiring investment in membranes, resins, and precise dosing systems — pay off in prolonged unit life and reduced unscheduled maintenance.
