High-purity water is the quiet engine of HRSG (heat recovery steam generator) reliability. The plants that win treat the demineralization train like a 24/7 asset—backed by rigorous preventive maintenance, N+1 design, and a stocked spares room.
Downtime is a costly habit. Siemens data show industrial outages can exceed $2M/hour, a ~50% jump since 2020 (watertechnologies.com). One site cut water-issue downtime from ~≈30 days/year by ~66 hours in 18 months after upgrading treatment controls (watertechnologies.com).
The throughline: the demineralization (DM) plant—often RO (reverse osmosis) plus ion exchange—has to deliver ultra-pure makeup water continuously or the HRSG loses its margin for error. Combined‑cycle plants demand that quality because chemistry upsets drive corrosion, scale, and unscheduled outages (mcilvainecompany.com) (power-eng.com).
Delgado‑Padilla et al. put it bluntly: “Production of high-purity makeup water is crucial for the reliable operation of high-pressure steam generators,” and chemistry upsets “cause corrosion and scaling…leading to failures and shutdowns” (powergen.com). One O&M note warns that staffing cuts are false economy: “Even just one evaporator tube failure from a chemistry upset can offset” those savings (power-eng.com).
Ultra‑pure water and demand assurance
High‑level guidance (IAPWS, EPRI, ASME) points to sub‑µS/cm conductivity and stringent purity targets to prevent even trace fouling or corrosion (mcilvainecompany.com). On process design, many HRSG sites standardize on RO/mixed‑bed trains; a typical RO unit for industrial makeup is a brackish‑water RO, with polishing handled by a mixed‑bed stage capable of very low silica and very low TDS.
Quantity matters alongside quality. Many plants carry a 24‑hour reservoir of demin water at normal usage rate—an industry “standard” buffer allowing one train to be serviced without tripping the boiler (mcilvainecompany.com).
Where pretreatment is required from surface or variable sources, ultrafiltration (UF, a membrane step for fine solids removal) is commonly paired ahead of RO—aligning with PowerGen’s discussion of microfiltration, ultrafiltration, and ion exchange in modern makeup treatment (powergen.com). System integrators package such trains as complete membrane systems for industrial water.
Redundancy in parallel trains
Design that relies on “a single or unspared component” is a dominant reliability risk (mcilvainecompany.com). Best practice is N+1: two RO/mixed‑bed trains per boiler, each at ~50–75% of peak load, so one can be out of service while the other meets demand (with storage support). That can be achieved with parallel ion‑exchange trains and a dedicated demineralizer configuration for each HRSG block.
The case for duplication is quantitative. One analysis found duplicating a critical component improved the overall reliability index by +10.6% (mdpi.com). Translated to the DM plant, having a spare RO pump or IX train on hot standby pushes availability toward five‑ or six‑nines.
Operationally, the absence of redundancy shows up fast. In one audit, Stage‑2 trains intended for resin regeneration every 2 months were regenerated weekly due to poor design—making the train unavailable 8 additional hours per week beyond expectations (mcilvainecompany.com).
Preventive‑maintenance program structure
Consensus guidance is clear: planned, documented maintenance reduces unscheduled outages, stabilizes processes, and extends installed life (grupoep.com.br) (grupoep.com.br). Plants with strong PM programs remain “ready to produce water according to required specifications,” meet safety and environmental standards, and minimize downtime, lowering repair expenses (carewater.solutions).
Core documentation matters. Addison & Weir emphasize detailed schedules and logs for pumps, valves, filters, membranes, exchangers, and instrumentation—listing each unit’s design throughput, chemical feed rates, regeneration intervals, and action limits (e.g., ion‑conductivity or differential pressure thresholds) (mcilvainecompany.com). Calibration routines for sensors and checklists for mechanical/electrical inspections must be written down, with immediate availability for breakdown repairs ensured (mcilvainecompany.com). For dosing and monitoring hardware, many operators include supporting equipment in the same PM plan.
Performance monitoring and analyzers
Continuous monitoring—flow, pressure, conductivity, silica, TOC (total organic carbon), pH—enables early problem detection. A site‑wide Performance Monitoring Program (PMP) “simplifies fault‑finding” and allows corrections before failures, maximizing availability (mcilvainecompany.com).
One plant installed TOC and sodium analyzers on train outlets so rising contamination was spotted immediately; this led to smarter regeneration cleaning and stabilized throughput (mcilvainecompany.com) (mcilvainecompany.com). Using this information, the site doubled full‑output service cycles (from 2 up to 6 runs) by optimizing the brine cleaning schedule (mcilvainecompany.com). Plants adopting reliability‑centered maintenance with online analyzers “reduced the number of problems and troubleshooting activities” (mcilvainecompany.com).
Routine service and predictive triggers
Regular actions on operating‑hour intervals are essential: lubricating pump bearings, replacing seals, cleaning strainers and prefilters, verifying valve actuations, and servicing high‑pressure RO pumps. For prefiltration, operators often standardize on a cartridge filter upstream with appropriate housing, such as a steel filter housing for industrial pressure service.
In ion exchange, routine checks of resin exhaustion (spot‑resin testing or effluent conductivity) trigger scheduled regenerations or resin replacement. The PM plan schedules major consumables: membranes cleaned or replaced per manufacturer guidance, resins regraded, degasifier internals serviced, UV lamps changed—typical for a plant using ultraviolet disinfection in the train. Addison & Weir note a PMP can identify when problems will occur and suggest the most cost‑efficient time to replace major cost items like membranes, resin, pumps, and vessels (mcilvainecompany.com). Rising RO feed pressure, for example, can justify a pre‑emptive membrane clean to avoid trips. Stocking ion‑exchange resin for planned changeouts helps keep service windows short.
Training and staffing requirements
PM works only when personnel understand it. Owners designate trained operators/technicians—at least one per shift—to perform DM‑plant checks and minor maintenance. Addison & Weir underscore that a technician per shift trained by a chemist “to conduct minor but meaningful tests,” and empowered to act on alarms, improves reliability (mcilvainecompany.com). These roles log resin volumes, regeneration dates, and water analyses for continuous improvement. A packaged demineralizer system with accessible sampling ports supports this operator‑led routine.
Redundancy configuration and storage
Even with perfect maintenance, failures occur. Redundancy ensures graceful degradation. Two trains allow annual inspections or upgrades on one while the other carries the load; oversized demin storage—24 hours of water—is often used with dual trains to provide a buffer (mcilvainecompany.com).
Design rules of thumb used in practice include: (a) at least one complete spare train (or equivalent capacity) for any critical stage (RO, cation/anion/trim), (b) shared utilities where appropriate but with attention to single points of failure, and (c) pump and valve configurations that allow isolation of any train without bleeding the system. Critical auxiliaries—high‑pressure booster pumps or control PLCs (programmable logic controllers)—should be dual‑rated or paired with spares. As CWET notes, “reliable operation…depends not only on advanced equipment but also on…the availability of high‑quality critical spares” (including duplicate hardware) (cwet.in). For polishing of condensate return streams in certain plants, a dedicated ion‑exchange line simplifies train isolation.
Critical spares strategy and inventory
No PM plan is complete without parts on hand. The lack of critical spare parts is a top cause of WTP outages in Addison & Weir’s findings (mcilvainecompany.com). Typical spares include high‑pressure pump impellers and seals, actuator and control valves, pressure vessels or housings for quick swap‑out, RO membrane elements, mixed‑bed resin, density/resin controllers or PLC modules, and calibration standards. CWET’s field guidance includes pumps, valves, UV lamps, conductivity/pH sensors, and filter cartridges among recommended kits (cwet.in). Stocking water‑treatment parts and consumables shortens repair windows.
Planning favors at least one of each long‑lead or critical item—membranes and resins (which can take weeks to ship), specialized valves, proprietary control cards—along with “soft” spares like O‑rings, gaskets, and cable ties. A PMP should “identify” these and ensure “immediate availability for breakdown repairs” (mcilvainecompany.com). In practice, that means a spare‑parts analysis listing predicted failures (e.g., a yearly pump rebuild or a 5‑year membrane replacement) and stocking for one worst‑case event. Even a failed conductivity transmitter can be swapped in minutes if a spare is on the shelf; days may be lost waiting otherwise. Similarly, mixing one case of new resin and a spool of rooting bits may enable on‑site resin replacement without delay. For membrane trains, carrying spare membrane systems components aligns with this philosophy.
The payoff is direct. Even “minor component failures can lead to costly downtime” without spares (cwet.in). One plant avoided an urgent new pump purchase—and half a week of outage—by installing an idle spare motor already in storage.
Economic impact and KPIs
The economics favor discipline. “The more reliance on preventive maintenance, the lower the maintenance costs…as well as lower the production losses from continuous stops” (carewater.solutions). Unplanned downtime could quickly run into tens of thousands of dollars—even for small outages (carewater.solutions), dwarfed by multi‑million‑dollar risks in power generation (watertechnologies.com).
Measured improvements are tractable. Veolia documented a 66‑hour drop in downtime over 18 months (a 35% reduction) after comprehensive water‑treatment fixes (watertechnologies.com). In the HRSG context, preventing even a 4–8 hour forced outage in the DM circuit can save tens of millions of dollars per year in delivered power.
Targets include Annual Availability for the DM plant (>99%), Number of Unplanned Regenerations (aim ≪1/month), MTBF (mean time between failures) of major pumps, and completion rate of scheduled tasks. One analysis found each dollar on PM can save up to $3–$20 in avoided corrective maintenance (equipment‑dependent) (carewater.solutions). Formal reliability models show the benefit of redundancy, with Liu et al. increasing a system’s reliability index from 0.891 to above 0.991 by adding a parallel component, making it “more resistant to changes” and meeting utility reliability targets (mdpi.com).
Bottom line for HRSG operators
Treat the DM plant as critical, 24/7 infrastructure. Reliability is built on (1) a rigorous preventive‑maintenance program that keeps all components in top condition (mcilvainecompany.com) (grupoep.com.br), (2) a fail‑safe design with redundant trains so no single failure interrupts supply (mcilvainecompany.com) (mdpi.com), and (3) a stocked inventory of critical spares for immediate repairs (mcilvainecompany.com) (mcilvainecompany.com). Plants following this approach run at design output and avoid the forced outages that cascade from water‑system failures (watertechnologies.com) (watertechnologies.com). For operators specifying or upgrading systems, the equipment mix—RO trains, mixed‑bed polishers, and ion‑exchange units—should be matched with spares, consumables, and maintenance practices from day one.
