Why Hydrostatic Test Water Is a Special RO Feed
To size an RO for hydrostatic test water, first characterize the feed—typically 200–5,000 mg/L TDS with corrosion inhibitors, iron oxide, and intermittent slug flow—then design pretreatment to SDI <5, select 12–20 LMH flux and 70–85% recovery, and calculate membrane area as permeate flow divided by flux. Confirm discharge limits or reuse goals before locking vessel count. The difficulty is not the salt; it is the way the feed arrives.
Hydrostatic test water is the discharge from pressure tests on vessels, piping, boilers, and skid packages. Operators dose the test water with corrosion inhibitors (nitrite/borate/amine blends, sometimes 500–2,000 mg/L), oxygen scavengers (sulfite at 200–1,000 mg/L), biocides, and occasionally dye penetrants or glycol. The result is a feed whose chemistry, suspended solids, and temperature bear no resemblance to a steady process wastewater stream. In practice, the drains arrive in 2–8 hour slug events at 5–80 m³ per event, separated by days or weeks of zero flow (Zhongsheng field data, 2026). That intermittency breaks the standard "24/7 RO skid" assumption and forces equalization upstream of any membrane system.
Transport through the membrane is pressure-driven, not concentration-gradient-driven, and depends on membrane pore size, the size of the permeating molecules, and feed viscosity (Yale Engineering, 2025). High-viscosity glycol mixes (winter hydrostatic tests on chilled piping can drop feed to 5–10 °C, raising viscosity ~30–50% above 25 °C) reduce flux proportionally and can force a larger membrane area than the TDS alone would suggest. The full envelope a designer must handle: TDS 200–5,000 mg/L, hardness 50–800 mg/L as CaCO₃, Fe 0.5–20 mg/L, occasional oil from pump seals, and occasional dye penetrants (fluorescent or dye-pen residues). A useful cross-reference for intermittent industrial feeds is the broader foundry wastewater treatment guide.
Feedwater Characterization: The Parameters That Drive RO Sizing
Four numbers from the lab report control everything downstream: TDS (or conductivity), hardness, iron, and SDI. Translate them correctly and the rest of the design follows; miss one and the membrane warranty is at risk.
TDS and conductivity. Osmotic pressure scales roughly linearly with dissolved salts: each 1,000 mg/L NaCl adds about 0.7 bar. Brackish hydrostatic water at 3,000 mg/L needs 25–30 bar feed pressure, versus seawater at 35,000 mg/L needing 55–70 bar. This is why an industrial RO system for test water is normally a brackish unit, not a seawater unit—misdimensioning here triples energy cost (Zhongsheng field data, 2026).
Hardness, silica, barium, strontium. Scaling indices (Langelier Saturation Index, Stiff–Davis Stability Index) dictate anti-scalant dose and recovery ceiling. The standard ceiling is 75% recovery for brackish RO without softening; 85% requires ion-exchange softening ahead of the membranes.
Iron and manganese. Fe above 0.1 mg/L fouls thin-film composite RO membranes irreversibly. Hydrostatic test water from rusty carbon-steel piping routinely violates this, which is why iron removal is the single most common pretreatment gap on these systems.
SDI and turbidity. SDI below 5 is the universal RO membrane warranty limit; below 3 is preferred for fouling-prone feeds. The 15-minute SDI test at 30 psi through a 0.45 µm pad is the gate: any reading above 5 means the pretreatment train is not finished.
Temperature and viscosity. Flux drops about 3% per °C below 25 °C. A winter test at 10 °C on a 17 LMH design at 25 °C gives an effective flux closer to 13 LMH, which adds roughly 25–30% to the required membrane area.
Free oil, grease, and dye penetrants. Even trace oil below 1 mg/L can blind a polyamide membrane. Activated carbon or dissolved-air flotation (DAF) pre-polish is required when pump seals or dye penetrants are in the chemistry list.
| Parameter | Typical range in hydrostatic test water | RO design target |
|---|---|---|
| TDS | 200–5,000 mg/L | ≤500 mg/L in permeate |
| Hardness (as CaCO₃) | 50–800 mg/L | LSI <0 at chosen recovery |
| Iron (Fe) | 0.5–20 mg/L | <0.1 mg/L pre-RO |
| SDI₁₅ | 10–25 raw | <3 (max 5 warranty) |
| Turbidity | 5–200 NTU raw | <1 NTU pre-RO |
| Temperature | 5–35 °C | Design at 15–20 °C |
| Free oil & grease | 0–10 mg/L | <0.1 mg/L pre-RO |
A multi-media RO pretreatment filter is the workhorse that gets the feed from raw SDI >10 to SDI <3.
Pretreatment Design: Protecting the RO Membrane
On intermittent, inhibitor-laden test water, pretreatment is not optional polish—it is the unit operation that decides whether the RO survives its first quarter. Six stages, in order.
Equalization tank. Size for 1.5–2× the largest single test event. A 6-hour drain at 12 m³/h means at least 80 m³ of working volume, with mechanical mixing and aeration to oxidize Fe²⁺ to Fe³⁺ for downstream removal. Without equalization, slug flow overruns the clarifier and slugs the high-pressure pump.
Coagulation, flocculation, sedimentation. A lamella clarifier for pretreatment at 20–40 m/h surface loading drops TSS and most suspended iron before filtration. Polymer dose typically 1–5 mg/L, paired with an automatic anti-scalant dosing skid ahead of RO.
Multi-media filter (MMF). Sand–anthracite–garnet media with automatic backwash. Target effluent turbidity below 1 NTU and SDI below 5. Backwash triggered on differential pressure or timer, whichever first.
Iron removal. Aeration plus filtration through greensand or manganese-dioxide media, designed to drop Fe below 0.1 mg/L before the RO feed. This step is non-negotiable when feed Fe exceeds 0.5 mg/L; skipping it is the most common cause of irreversible flux loss in field data (Zhongsheng, 2026).
Activated carbon for inhibitor and biocide polishing. Corrosion-inhibitor amines and several biocides pass through MMF but adsorb on carbon. Empty bed contact time of 10–15 minutes handles 1,000 mg/L inhibitor swings typical of one-off test campaigns.
5 µm cartridge guard filter. Final guard immediately upstream of the high-pressure pump. Differential-pressure alarm at 1 bar across the element triggers change-out.
| Stage | Equipment | Target effluent | Design basis |
|---|---|---|---|
| Equalization | Mixed, aerated tank | Fe²⁺ oxidized, slug absorbed | 1.5–2× largest event |
| Clarification | Lamella / DAF | TSS <20 mg/L | 20–40 m/h loading |
| MMF | Sand–anthracite–garnet | Turbidity <1 NTU, SDI <5 | 10–15 m/h |
| Iron removal | Aeration + greensand | Fe <0.1 mg/L | MnO₂ media, 10–12 m/h |
| Carbon | GAC contactor | Inhibitor/ biocide adsorbed | 10–15 min EBCT |
| Cartridge | 5 µm polypropylene | Particulate guard | ΔP alarm at 1 bar |
RO Sizing Calculation: A Step-by-Step Procedure
The math itself is short; the discipline is to apply it in the right order, because each step locks a parameter the next step depends on.
- Define permeate demand Qp. From the reuse target (boiler make-up, cooling make-up) or the discharge dilution requirement. Add 10–15% design margin for ramp-up, cleaning, and off-spec disposal.
- Choose recovery Y. 70–80% is standard for brackish test water without softening; 85% only with weak-acid cation exchange ahead of the RO. Above 80%, the LSI climbs and calcium carbonate scaling on the tail elements becomes the failure mode.
- Choose flux J. 12–20 LMH for brackish RO on industrial wastewater. Use the low end (12–14 LMH) when feed has high fouling potential (high Fe, high SDI, oil traces) and the high end (18–20 LMH) only for clean, well-softened feeds.
- Calculate membrane area A = Qp / J. Example: Qp = 10 m³/h, J = 17 LMH → A ≈ 590 m². Standard 8-inch 400 ft² (≈37 m²) elements give 590 / 37 ≈ 16 elements, typically arranged as two parallel trains of 6 elements each in a 2:1 array, or one train of 8:4 (12 elements) for a single skid.
- Feed and concentrate flow. Qf = Qp / Y = 10 / 0.75 = 13.3 m³/h; concentrate Qc = 3.3 m³/h.
- Feed pressure Pfeed. Pfeed = ΔP across membrane + osmotic pressure of concentrate + safety margin. For 3,000 mg/L feed at 75% recovery on brackish elements, Pfeed ≈ 12–18 bar. Seawater elements on the same feed need 25–30 bar—over-specified and wasteful.
- Pump and energy. 4–6 kWh/m³ permeate is typical for brackish RO at 75% recovery. Seawater elements on brackish feed inflate this to 10–14 kWh/m³. A 1–2 bar ΔP margin for the unavoidable pressure drop across the membrane and piping belongs in the high-pressure pump head calculation (Yale Engineering, 2025).
| Step | Input / formula | Worked value (10 m³/h, 3,000 mg/L feed) |
|---|---|---|
| Permeate demand Qp | Reuse/discharge target + 10–15% | 10 m³/h |
| Recovery Y | 75% (brackish, no softener) | 0.75 |
| Flux J | 17 LMH (mid-range brackish) | 17 L/m²·h |
| Membrane area A | Qp / J | ≈ 590 m² (~16 × 8″ 400 ft² elements) |
| Feed flow Qf | Qp / Y | 13.3 m³/h |
| Concentrate flow Qc | Qf − Qp | 3.3 m³/h |
| Feed pressure | ΔP + π(concentrate) + margin | 12–18 bar |
| Specific energy | Pump / Qp | 4–6 kWh/m³ permeate |
The full industrial RO system specification drops directly out of this table: 16 elements, two vessels per train, 6:3 array, 13.3 m³/h feed pump at 16 bar, 4–6 kWh/m³.
Discharge vs Reuse: The Recovery Decision Framework
The same RO can be sized for either goal, but the operating point and OPEX profile change. Decide first, size second.
Discharge case. Aim for 70–75% recovery. Send concentrate to a small brine evaporator (high OPEX) or a licensed waste hauler. CAPEX is lower because the permeate does not need polishing; OPEX is dominated by concentrate disposal at USD 5–25 per m³ of concentrate.
Reuse case. Aim for 75–85% recovery. Polish permeate through a mixed-bed deionizer for boiler feed, or send directly to cooling tower. CAPEX is higher (extra polishing, larger equalization, sometimes softening); OPEX is dominated by membrane replacement (3–5 year life) and energy at 4–6 kWh/m³.
Energy comparison: 70% recovery RO at about 4 kWh/m³ permeate, evaporation of concentrate at 25–40 kWh/m³. When a make-up water demand already exists on site, the reuse case almost always wins on 10-year lifecycle cost. ZLD concentrate management options are the right next read when the concentrate cannot go to sewer.
| Decision input | If discharge | If reuse |
|---|---|---|
| Target recovery | 70–75% | 75–85% |
| Permeate polish | None (discharge only) | Mixed-bed DI or UV |
| Concentrate path | Evaporator or hauler | Brine reuse, ZLD, or sewer (if allowed) |
| CAPEX driver | Equalization, RO skid | Equalization, RO skid, polish, softener |
| OPEX driver | Concentrate disposal | Membranes, energy |
| Specific energy | ~4 kWh/m³ | 5–6 kWh/m³ + polish |
Operation, Monitoring, and Common Failure Modes
Intermittent feeds produce three failure modes far more often than steady feeds. Specify the instrumentation and alarm setpoints up front; the field data show they prevent 80% of warranty disputes (Zhongsheng, 2026).
Instrumentation. Inline conductivity on permeate and concentrate (separate probes, not shared), pressure gauges on each vessel, flow meters on permeate and reject lines, ORP and pH probes for CIP-trigger logic. Online water-quality monitoring on the equalization tank adds shock-chlorination interlocks for biofouling control.
Alarm setpoints. Permeate conductivity rise above 15% of baseline; inter-stage differential pressure rise above 10%; normalized flux drop above 10%. Any one of these triggers an automatic shutdown and a CIP decision.
Three common failure modes. (1) Iron fouling from rusty carbon-steel test piping—prevented with Fe below 0.1 mg/L pretreatment; (2) biofouling from stagnant equalization during the weeks between test events—prevented with shock chlorination followed by dechlorination (SBS dosing) ahead of the RO; (3) scaling from hardness in the concentrate—prevented with anti-scalant at 2–5 mg/L and a recovery ceiling of 75–80% unless softening is installed.
Cleaning frequency. Quarterly CIP is typical: low-pH detergent (citric or sulfamic at pH 2–3) for inorganic scale, high-pH detergent (NaOH + EDTA at pH 11–12) for organic and biofouling, with each cycle followed by permeate rinse until conductivity returns to baseline.
Frequently Asked Questions
What flux rate should I use for RO on hydrostatic test water?
12–20 LMH for brackish RO on industrial wastewater. Use the low end (12–14 LMH) when feed has high fouling potential—iron above 0.5 mg/L, SDI above 5, or any oil trace. The high end (18–20 LMH) is reserved for clean, softened feeds. A winter feed at 10 °C drops effective flux roughly 3% per °C below 25 °C, so size at the actual minimum operating temperature, not the annual average.
Do I need a softener before the RO?
Yes, if hardness exceeds 200 mg/L as CaCO₃ and target recovery is above 75%. Without softening, the LSI climbs past zero in the tail elements and calcium carbonate scale will shorten membrane life from 5 years to 18–24 months. A weak-acid cation exchanger or a sodium-cycle ion-exchange softener ahead of the RO is the standard answer. Below 200 mg/L hardness and at 75% recovery, anti-scalant alone is usually enough.
Can I send the RO concentrate to sewer?
Depends on local limits. The concentrate stream at 75% recovery from a 3,000 mg/L feed will run 8,000–12,000 mg/L TDS—not sewer-dischargeable in most jurisdictions. If the feed is softer (under 500 mg/L), concentrate TDS typically lands in the 1,500–3,000 mg/L range, which may pass under a 5,000 mg/L sewer cap but will fail any metals limit because iron, inhibitor amines, and any heavy metals concentrate proportionally. Always check the local pretreatment ordinance and metals limits before specifying sewer discharge.
How much does a 10 m³/h RO cost for hydrostatic test water?
CAPEX for the full train—equalization, pretreatment, RO skid, permeate polish—typically runs USD 180,000–320,000 (Zhongsheng, 2026). The RO skid itself is USD 80,000–150,000 of that total. OPEX lands at USD 0.6–1.2 per m³ of permeate, dominated by membrane replacement (3–5 year life), energy at 4–6 kWh/m³, and concentrate disposal where applicable.
How do I handle intermittent flow to the RO?
Equalize first. Size the equalization tank at 1.5–2× the largest single test event (the 80 m³ drained over 6 hours in the opening scenario becomes a 120–160 m³ tank), with mixing and aeration to oxidize iron and prevent anaerobic fouling. RO skids are typically rated for turndown to 50–70% of design flow; below that, recycle or shut down. Frequent start-stop is more harmful than continuous low-flow operation, so over-size the equalization rather than cycle the RO. The foundry wastewater treatment guide covers equalization sizing in more detail for related intermittent feeds.
