Membrane Troubleshooting

RO/NF Troubleshooting & Diagnostic Methodology

Scope: spiral-wound thin-film-composite (TFC) polyamide RO/NF elements (8" standard + 4" variants). Numeric thresholds carry vendor + year of source.

Convention markers:

• ⚠️ — likely bug or incorrect assumption in calling code; highest priority
• 💡 — enhancement opportunity (new alert, new state, refined threshold)

§0 — Diagnostic Frame of Reference

0.1 Five orthogonal failure modes

Every observed deviation must be classified into one (or more) of:

1. Fouling — reversible / partially reversible accumulation (colloidal, biological, organic, particulate)
2. Scaling — supersaturation-driven mineral precipitation, mostly in tail/concentrate stages
3. Integrity loss — physical breach of active polyamide layer or seal (O-ring, glue line, telescoping, pinhole)
4. Mechanical compaction / chemical hydrolysis — pressure-driven densification (largely irreversible); polyamide hydrolysis outside pH 2–11
5. Instrumental / operational error — out-of-cal flow meter, conductivity probe drift, dosing pump miscount, baseline never re-established after element swap

0.2 ASTM D4516 — the three observables

0.3 Canonical signature table

Combines DuPont 45-D01650 troubleshooting matrix + Hydranautics TSB 107 logic + WaterTechOnline normalization article:

0.4 Mass balance sanity check

Before any diagnosis:

|Qf − (Qp + Qc)| / Qf  <  0.02      # within 2 %

A 5 % error in recovery shifts scale-potential calculations dramatically.

💡 No mass-balance closure check is the #1 cause of phantom alerts from instrument drift. Add this check before any rule evaluation; halt diagnosis if it fails.

§1 — Fouling Taxonomy

1.1 Colloidal fouling

• Definition — sub-µm inorganic / mixed particles (silt, clays, Fe/Al hydroxides) too small to settle
• Predictor — SDI₁₅ (ASTM D4189-23): < 3 acceptable; 3–5 frequent CIP; > 5 needs MF/UF upstream. Vendors recommend SDI₁₅ < 4
• Signature — NPF↓ first stage, ΔP↑ lead element, NSP stable
• Cleaning — alkaline (pH 11–12, NaOH + EDTA + anionic surfactant) first; acid (HCl/citric, pH 2–3) only if Fe/Al hydroxide confirmed

1.2 Biofouling

• Definition — biofilm (EPS matrix + bacteria) on membrane and feed spacer
• Predictors — AOC (assimilable organic carbon): biofilm-free risk threshold ~1 µg/L; conventional treatment achieves 50–100 µg/L (Pereira et al. 2014, PMC4021920). ATP, BFR, rising SDI without particulate cause
• Signature — NPF↓, ΔP↑ over weeks, slight NSP↑; autopsy shows slime mat on lead-element feed face
• Cleaning — alkaline NaOH (pH 11–12) + EDTA + surfactant, then non-oxidizing biocide soak. Never free chlorine for in-service biocide on PA membranes
• DBNPA dosing (industry consensus, IWA AQUA 2022; Feedwater datasheet):
  • Online maintenance: 10–50 mg/L for ≥ 1 h, weekly
  • CIP additive: 50–200 mg/L circulated 1–3 h
  • Continuous preventive: ~1 mg/L (where allowed)
  • Bisulfite upstream consumes DBNPA → increase dose
• Other accepted non-oxidizers: isothiazolinone (CMIT/MIT), glutaraldehyde for storage. See Hydranautics TSB 110

1.3 Organic fouling (NOM)

• Definition — natural organic matter (humics, fulvics), hydrophobic adsorption on PA surface
• Signature — NPF↓, NSP stable or slight ↑, ΔP → (no spacer plugging until late)
• Cleaning — alkaline NaOH pH 11–12 + EDTA chelant + surfactant (DuPont 45-D01504 default rule for unknown / combined fouling)

1.4 Particulate fouling

• Definition — > 1 µm material from failed pretreatment (cartridge bypass, blown media filter)
• Diff vs colloidal — visible deposit on cartridge filter; rapid ΔP rise on lead element only
• Cleaning — flush forward at high crossflow; alkaline CIP if needed

1.5 Combined fouling (most common in practice)

DuPont 45-D01504 rule: alkaline cleaning as the first step for unknown / combined; acid only when CaCO₃ or Fe/Mn hydroxide is confirmed.

Sequence:

1. Alkaline (NaOH + EDTA + surfactant), pH 11–12, T ≤ 35 °C above pH 10
2. Flush to neutral pH with permeate-quality water
3. Acid (HCl, citric, or sulfamic), pH 2–3, T ≤ 35 °C
4. Optional biocide soak (DBNPA 50–200 mg/L, 1–3 h)

⚠️ Never mix caustic and acid in the same loop (Hydranautics TSB 107).

§2 — Scaling by Mineral Species

Solubility products at 25 °C cross-checked against Hydranautics TAB-111, DuPont 45-D01552, and standard chemistry tables.

2.1 Saturation calculations

LSI (low TDS): LSI = pH − pH_s; pH_s depends on Ca²⁺, alkalinity, TDS, T. LSI > 0 ⇒ CaCO₃ supersaturated.

S&DSI (TDS > 10 000 ppm): same logic, ionic-strength corrected.

Sulfate scales / silica: use ion-product / Ksp ratio: S = IP/Ksp. Vendor projection tools (ROSA, IMSdesign, WAVE, ROProMax) compute these per stage.

⚠️ Compute always at the concentrate (last element). LSI at feed is meaningless for scaling prediction.

§3 — Membrane Integrity Loss

3.1 O-ring failure

• Location — interconnector between elements; permeate-tube to vessel end-adapter
• Signature — sudden NSP↑ localized to one vessel, NPF and ΔP unchanged
• Detection — conductivity profiling (Toray "Probing" white-paper); Rhodamine WT dye test
• Repair — replace O-ring; rebuild dry, light glycerin to seat. Do not use petroleum grease (attacks EPDM and PA)

3.2 Breach / pinhole

• Detection methods:
  • VDT (Vacuum Decay Test, ASTM D6908) — primary for RO/NF
  • PDT (Pressure Decay Test) — for MF/UF
  • Bubble test at 3–5 psig with submerged element
  • Rhodamine WT dye under pressure: diffuse pattern = chemical attack; localized = mechanical
• Threshold for autopsy — industry rule of thumb: salt rejection drop > 10 % vs baseline, or sharp ΔP rise. Always autopsy at least one representative element before discarding

3.3 Chlorine oxidation of polyamide

Cumulative dose metric: ppm·h of free chlorine exposure.

Literature consensus (Gohil & Suresh 2019, Desalination 457:133; vendor literature):

• Onset of measurable degradation: ~200 ppm·h
• Severe degradation: ~1 000 ppm·h
• SWRO membranes (denser, higher MPD crosslink) generally more tolerant than BWRO
• At very high cumulative exposure (thousands of ppm·h) PA RO can transform toward NF-like behavior (rejection drop, flux rise)

⚠️ Specific values like "2 640 / 6 200 ppm·h" sometimes seen in process literature are not directly confirmed by the primary Gohil paper; use the 200 onset / ~1 000 severe range as the defensible cited band, and note SWRO/BWRO/PA-density dependence.

Vendor rule: all manufacturers (DuPont, Hydranautics, Toray) specify "no detectable free chlorine in feed" (< 0.1 mg/L) for warranty.

Mechanism: N-chlorination of PA amide → Orton rearrangement → ring chlorination → loss of crosslink → flux ↑ and rejection ↓.

💡 Track cumulative chlorine exposure as a running ppm·h counter in a cumulative_exposure table. This is essential for lifecycle decisions (replace vs continue) and is invisible to instantaneous KPI checks.

3.4 Mechanical damage

• Telescoping — axial spacer/membrane migration. Causes: pressure shock (water hammer); excessive ΔP per element (> 15 psi normalized; > 50 psi destructive); missing/failed ATD (Anti-Telescoping Device). Pressurization rate limit: ≤ 10 psi/s (≈ 0.69 bar/s) at start-up
• Glue-line crack — same shock causes; dye test shows linear pattern at glue line
• Abrasion — feed sand/grit; visible signs on lead-element feed end at autopsy

3.5 Autopsy decision matrix

Trigger when any one of:

• Salt rejection drop > 10 % vs baseline persisting after 2 CIPs
• ΔPn rise > 25 % not recovered by CIP
• NPF decline > 25 % not recovered by CIP
• Unexplained instrument disagreement after calibration
• End-of-warranty diagnostic to claim manufacturing defect

§4 — Chemical / Operational Failures

4.1 Free chlorine breakthrough

• Causes — SBS pump failure, GAC bed exhausted, chloramine breakthrough not detected by ORP (chloramines have low ORP signal)
• Monitoring — ORP < 200 mV after dechlorination (some target < 150 mV); redundant DPD colorimetric
• SBS stoichiometry — 1.46 mg NaHSO₃ per mg Cl₂ (theoretical); practical 1.5–2.5× theoretical = 1.8–3.0 mg/mg
• Side effect — chronic SBS overdose with dissolved Cu/Co catalyses oxidant generation under O₂ → unintended PA attack

4.2 pH excursion

PA operating window: pH 2–11 continuous; 1–12 short CIP (most vendor datasheets; DuPont 45-D04358).

Out-of-range hydrolysis → permanent rejection loss.

4.3 Temperature excursion

• Max continuous: 45 °C / 113 °F at pH ≤ 10
• Above pH 10: 35 °C / 95 °F max
• Above 45 °C: accelerated hydrolysis + compaction

4.4 Pressure shock / water hammer

• Slow pressurization (≤ 10 psi/s); install soft-start VFD or air-loaded accumulator; check valve on permeate
• Per-element ΔP cap: normalized ΔP per 6-element vessel ≈ 15 psi typical; > 50 psi induces telescoping

§5 — Diagnostic Decision Logic

Pseudo-code consuming an RODataPoint (current readings) + a Baseline (commissioning-normalized values) + cumulative exposure counters:

@dataclass
class RODataPoint:
    Qf: float; Qp: float; Qc: float          # m3/h
    Cf: float; Cp: float                     # µS/cm or ppm
    Tf: float                                # °C
    P_feed: float; P_perm: float; P_conc: float  # bar
    timestamp: datetime
    stage: int = 1

@dataclass
class Baseline:
    NPF0: float; NSP0: float; dPn0: float    # normalized at t0
    R0: float; Tref: float = 25.0

def diagnose(now: RODataPoint, base: Baseline,
             cumulative_Cl_ppmh: float = 0.0,
             cumulative_T_above_45_h: float = 0.0) -> list[Diagnosis]:
    findings = []

    # Step 0 — instrumentation sanity
    if abs(now.Qf - (now.Qp + now.Qc)) / now.Qf > 0.02:
        findings.append(Diagnosis("instrument", conf=0.9,
            note="Mass balance fails > 2 %; calibrate before diagnosing membrane."))
        return findings  # halt: trust nothing else until fixed

    # Step 1 — normalize per ASTM D4516
    NPF = normalize_permeate_flow(now, base.Tref)
    NSP = normalize_salt_passage(now, base.Tref)
    dPn = normalize_dp(now, base.Tref)

    dNPF = (NPF - base.NPF0) / base.NPF0
    dNSP = (NSP - base.NSP0) / base.NSP0
    ddP  = (dPn - base.dPn0) / base.dPn0

    # Step 2 — cumulative-exposure overrides
    if cumulative_Cl_ppmh > 200:
        findings.append(Diagnosis("chlorine_oxidation",
            conf=min(1.0, cumulative_Cl_ppmh / 1000),
            note=f"PA degradation likely; {cumulative_Cl_ppmh:.0f} ppm·h ≥ 200 onset"))

    # Step 3 — signature matching (ASTM D4516 triple)
    if dNPF < -0.15 and ddP > 0.15 and abs(dNSP) < 0.05:
        if first_stage_dominates(now):
            findings.append(Diagnosis("colloidal_or_particulate", conf=0.8))
        else:
            findings.append(Diagnosis("biofouling", conf=0.6))

    if dNPF < -0.10 and dNSP > 0.10 and ddP > 0.15 and tail_stage_dominates(now):
        findings.append(Diagnosis("scaling_advanced", conf=0.85,
            note="Run LSI / S&DSI / SO4·Ba / Sr Ksp on concentrate."))

    if abs(dNPF) < 0.05 and dNSP > 0.15 and abs(ddP) < 0.05:
        findings.append(Diagnosis("integrity_loss", conf=0.8,
            note="Conductivity-profile vessels; expect O-ring or seal."))

    if dNPF > 0.05 and dNSP > 0.15 and abs(ddP) < 0.10:
        findings.append(Diagnosis("oxidative_degradation", conf=0.85))

    if dNPF < -0.10 and abs(dNSP) < 0.05 and abs(ddP) < 0.05:
        if uniform_across_stages(now):
            findings.append(Diagnosis("compaction_or_NOM_adsorption", conf=0.6))

    # Step 4 — timescale modifier
    if dNSP > 0.10 and is_sudden(now, base, hours=24):
        upgrade_confidence(findings, "integrity_loss", +0.15)
    if dNSP > 0.05 and is_gradual(now, base, days=30):
        upgrade_confidence(findings, "oxidation_or_fouling", +0.1)

    return sorted(findings, key=lambda d: -d.conf)

Branching rules encoded above:

• Mass balance fails → instrumentation first; halt
• Sudden change (< 24 h) of rejection → leak / integrity
• Gradual change (weeks) of rejection → cumulative oxidation or fouling
• First-stage flux loss only → upstream fouling (colloid, NOM, bio)
• Tail-stage loss with rising salt → scaling
• Uniform across stages → compaction or systemic oxidation

§6 — CIP Decision Matrix

6.1 Trigger criteria (industry consensus)

From WaterTechOnline 14171304; Hydranautics TSB 107; DuPont 45-D01696:

• NPF decline ≥ 10–15 % below baseline
• NSP increase ≥ 5–15 % above baseline
• ΔPn increase ≥ 15 % above baseline (some sources 15–25 %; don't wait past 25 % — foulant turns irreversible)

6.2 Chemistry selection

6.3 Per-element CIP flow rate

ΔP during CIP must stay below ~10 psi per element; exceeding telescopes.

6.4 When CIP does NOT help (autopsy / replace)

• Chlorine oxidation: damage cumulative; no chemistry restores PA
• Mineral scale cemented (typically Ba/Sr sulfate): EDTA partial at best
• Mechanical telescoping or glue-line crack
• Hydrolysis from pH excursion

§7 — Data-Driven Trending

7.1 Baseline establishment (ASTM D4516)

• Record after 24–48 h stable operation post-startup or post-CIP
• Hold feed T, P, recovery, conductivity constant during baseline window
• NPF, NSP, ΔPn become the t0 anchor; persist to DB
• Re-baseline after every element swap

7.2 Signal-to-noise

7.3 Why snapshots mislead

• Feed-T drift: each +1 °C ≈ +3 % permeate flow → masks fouling
• Recovery drift: 5 % recovery change shifts LSI/S&DSI by tenths
• Dosing-pump cycles (antiscalant ± 10 % short term) affect tail-stage rejection

7.4 Suggested SQLite schema

CREATE TABLE baselines (
  id INTEGER PRIMARY KEY,
  train_id TEXT, stage INTEGER,
  ts_utc TEXT,
  NPF0 REAL, NSP0 REAL, dPn0 REAL,
  R0 REAL, Tref REAL DEFAULT 25.0
);

CREATE TABLE readings (
  id INTEGER PRIMARY KEY,
  train_id TEXT, stage INTEGER, ts_utc TEXT,
  Qf REAL, Qp REAL, Qc REAL,
  Cf REAL, Cp REAL,
  Tf REAL, P_feed REAL, P_perm REAL, P_conc REAL,
  NPF REAL, NSP REAL, dPn REAL   -- pre-computed
);

CREATE TABLE cumulative_exposure (
  train_id TEXT PRIMARY KEY,
  cl_ppmh REAL DEFAULT 0,          -- running ppm·h free chlorine
  t_above_45c_h REAL DEFAULT 0,    -- hours above 45 °C
  pressure_cycles INTEGER DEFAULT 0,
  ph_excursions_below_2 INTEGER DEFAULT 0,
  ph_excursions_above_11 INTEGER DEFAULT 0,
  last_update TEXT
);

CREATE TABLE cip_events (
  id INTEGER PRIMARY KEY, train_id TEXT,
  ts_start TEXT, ts_end TEXT,
  chemistry TEXT, pH REAL, T REAL,
  trigger_metric TEXT,
  pre_NPF REAL, post_NPF REAL,
  pre_NSP REAL, post_NSP REAL,
  pre_dPn REAL, post_dPn REAL,
  notes TEXT
);

💡 The cumulative_exposure and cip_events tables are essential for any lifecycle advisor that recommends CIP vs replace based on long-horizon history.

§8 — Bilingual Glossary (Troubleshooting Terms)

Sources

• ASTM D4516-19a Standardizing RO Performance Data
• ASTM D6908 Integrity Testing of Water Filtration Membrane Systems
• ASTM D3923 Detecting Leaks in RO and NF Devices
• ASTM D4189-23 Silt Density Index
• Hydranautics TSB 107.28 (Apr 2025) — foulants & cleaning, TCF
• Hydranautics TSB 108 — chlorination/dechlorination
• Hydranautics TSB 110 — biocides and storage
• Hydranautics TSB 111 — biofouling
• Hydranautics TSB 124 — VDT and integrity
• Hydranautics TAB-111 — chemical pretreatment for RO/NF
• DuPont FilmTec 45-D01504 Tech Manual; 45-D01650 Symptoms of Trouble; 45-D01696 Rev 13 Cleaning Procedures (Feb 2026); 45-D04358 Temperature & pH Best Practices; 45-D01616 Plant Performance Normalization; 45-D01552 CaCO₃ Scale Prevention; 45-D01569 Chlorination/Dechlorination
• Gohil & Suresh 2019, Desalination 457:133 — free chlorine exposure tolerance
• Pereira et al. 2014, PMC4021920 — biofouling review
• AWWA RF — Guidance Manual for Disposal of Chlorinated Water (SBS stoichiometry)
• WaterTechOnline article 14171304 — Membrane Cleaning Fundamentals
• Toray Probing white-paper — vessel conductivity profiling
• Avista / American Water Chemicals autopsy leaflets — Rhodamine WT dye, integrity thresholds
• IWA Publishing 2022 — non-oxidizing biocides in PA RO