Membrane Economics, Energy & EDI Deep Dive

Two-part reference. Part 1 covers SEC/LCOW/lifecycle. Part 2 covers EDI process, KPIs, failure modes, and regulatory context for produced water and pharmaceutical UPW.

Convention markers:

⚠️ — likely bug or incorrect assumption in calling code
💡 — enhancement opportunity (new alert, new field, new module)

PART 1 — Energy & Economics

§1 — Specific Energy Consumption (SEC)

Operator-facing formula:

SEC = (P_HPP − P_ERD_recovered) / Q_permeate          [kWh/m³]

where P_HPP = high-pressure-pump draw (kW), P_ERD_recovered = power recovered by ERD (kW), Q_permeate in m³/h. For brackish/tertiary systems without ERD the second term is zero.

Engineering-grade form:

SEC = ΔP_feed / (3600 · η_pump · R) − (1 − R) · ΔP_brine · η_ERD / (3600 · R)

with R = recovery, ΔP in bar.

1.1 Benchmarks by source water (verified 2025–2026)

Feed type	TDS	Typical SEC	Best-in-class	Source
SWRO (open ocean)	~35 000 mg/L	2.5–4.0 kWh/m³	1.794 kWh/m³ Danfoss DESALRO 2.0 record, Pozo Izquierdo, Gran Canaria, Feb 2025	Danfoss press release / Guinness
Modern large SWRO	35 000 mg/L	2.0–2.8 kWh/m³	Previous record 2.27 kWh/m³ (SWCC, KSA)	Danfoss
BWRO industrial	1 000–10 000 mg/L	0.5–1.5 kWh/m³	0.5–0.8 with ERD + optimal recovery	ScienceDirect S004896972402919X
Tertiary reuse RO	< 2 000 mg/L	0.5–1.5 kWh/m³	—	EPA WaterReuse

1.2 Thermodynamic minimum

≈ 1.06 kWh/m³ for SWRO at R = 0.5, T = 25 °C, 35 g/L TDS
Derived from Gibbs free energy of separation with Spiegler-Kedem reversible-process framework
Theoretical minimum at infinite dilution: ≈ 0.78 kWh/m³

Sources: Journal of Chemical Education 2021 (doi:10.1021/acs.jchemed.0c01194); Wang et al., Desalination 2016 (S001191641630087X "Thermodynamic perspective for SEC of seawater desalination"); US-DOE Seawater Desalination Bandwidth Study 2017.

⚠️ The "Stenzel" attribution sometimes seen on the open web for the 1.06 figure is not confirmed; cite the Spiegler-Kedem / Gibbs derivation instead.

1.3 ERD comparison

ERD type	Peak η	Field η	Notes
Pelton wheel	80–87 %	78–85 %	Older centrifugal recovery; hydraulic → mechanical → hydraulic
Calder DWEER (isobaric dual-work)	98 %	93–96 %	Licensed by Calder AG
Energy Recovery PX (rotary pressure exchanger)	95–98 %	93–96.4 %	Industry standard for large SWRO

ERD payback rule of thumb: < 2 years for plants > 1 000 m³/d at industrial electricity tariffs.

§2 — LCOW (Levelized Cost of Water)

LCOW = (CAPEX_annualized + OPEX_annual) / Q_produced_annual
CRF  = i·(1+i)^n / ((1+i)^n − 1)            # capital recovery factor

2.1 Component breakdown — SWRO

Component	Share
CAPEX amortized	40–50 %
Energy	30–40 % (grid-cost dependent)
Chemicals + cleaning	5–10 %
Membrane replacement	5–10 %
Labor + O&M	10–15 %

2.2 LCOW ranges (verified 2024–2026)

Plant class	LCOW (USD/m³)	Source
Large SWRO (> 50 000 m³/d, excl. intake/outfall)	$0.53–$1.58	Advisian
Utility-scale SWRO with high-solar 2026 forecast	$0.45–$0.70	Energy-solutions.co
Small-island SWRO	$1.50–$3.00	IDA Yearbook / Advisian
BWRO industrial	$0.20–$0.50	Foreverpureplace

2.3 Sensitivity ranking

Electricity unit price
Membrane lifespan & replacement rate
Recovery rate (R)
Plant capacity factor (uptime)
Discount rate / WACC for CAPEX amortization

Sources: Advisian Cost of Desalination; ScienceDirect S2213138823000577.

§3 — Operational KPIs (beyond instantaneous calculations)

KPI	Healthy	Stressed	Action
Membrane life SWRO	5–7 yr (8–10 best)	3–5 yr	Review pretreatment
Membrane life BWRO	7–10 yr	2–3 yr	Antiscalant + CIP review
CIP frequency	1–2/yr	4–6/yr	Diagnose foulant
CIP frequency critical	> 6/yr	—	Replace lead vessel / re-engineer
Availability	≥ 95 %	90–95 %	Train rotation / redundancy
NPF baseline drift (lifetime)	5–10 %/yr	15 %+/yr	Investigate compaction vs fouling
NSP baseline drift (lifetime)	10–20 %	> 25 %	Approach replacement

Sources: Aqualitek SWRO maintenance guide; FilmTec/DuPont Form 45-D01911 Exceptional Membrane Life; Pumps & Systems Life of an RO Membrane.

§4 — Lifecycle Decision Matrix

Signature → action mapping that turns trends into operational decisions:

Observation	Likely cause	Reversibility	Action	Rationale
NPF −15 % in 30 d, ΔPn +20 %	Particulate / biofouling	Reversible	CIP (alkaline + biocide)	Standard recovery
NPF −20 % over 3 yr, ΔPn stable	Compaction	Irreversible	Tolerate or schedule replacement	No CIP benefit
NSP +50 % sudden	Mechanical integrity loss	Localised	Probe sweep; replace lead element	Acute defect
NSP +30 % over 2 yr	Cumulative oxidation	Irreversible	Replace lead vessel set; audit RedOx	Wear pattern
ΔPn rapid stage 2 only	Stage-2 scaling	Treatable	Antiscalant review + acid CIP	Concentration-polarization location
NPF + NSP both down	Severe organic fouling	Often reversible	Hot CIP, then biocide	Combined permeability + selectivity hit

💡 A lifecycle_advisor module consuming this table + cumulative-exposure counters + CIP history → enum recommendation {TOLERATE, SCHEDULE_CIP, IMMEDIATE_CIP, REPLACE_LEAD, REPLACE_VESSEL_SET} is more valuable than a single-snapshot alarm.

PART 2 — EDI Deep Dive

§5 — EDI Process Fundamentals

5.1 Cell geometry

Repeating "cell pair" = one dilute (D) compartment + one concentrate (C) compartment, bounded by alternating cation- and anion-exchange membranes. Electrode compartments at stack ends apply DC field.

Ions in D migrate through membranes into C; dilute stream is product.

5.2 In-situ regeneration

At elevated DC voltage, water splits at the bipolar interface of the resin bed in D:

H₂O → H⁺ + OH⁻

H⁺ regenerates the cation resin
OH⁻ regenerates the anion resin

No NaOH/HCl is needed → no hazardous regenerant waste. This is the central advantage over Mixed-Bed DI.

5.3 Why EDI lives downstream of RO

Feed must be ≤ ~40 µS/cm FCE
Weak acids (CO₂, SiO₂) load the stack when ionized internally
Hardness > 1 ppm CaCO₃ causes concentrate scaling
Free chlorine > 0.02 ppm destroys resin permanently

5.4 Verified feed envelope (Ionpure LX series)

Parameter	Limit	Notes
Feed conductivity (FCE basis)	< 40 µS/cm	Stack-dependent 25–43 µS/cm
Total hardness (as CaCO₃)	< 1.0 ppm	Hard limit; risk = concentrate scaling
TOC	< 0.5 ppm	Anion-resin fouling above
Fe / Mn	< 0.01 ppm each	Irreversible colloidal fouling
Total CO₂	< 5 ppm typical	Loads via FCE; degas if higher
SiO₂	< 1.0 ppm Ionpure LX; < 0.5 ppm preferred	Polymerization risk
Total chlorine	< 0.02 ppm (target 0)	Destroys resin
Temperature	5–45 °C standard; 60 °C HWS modules	HWS = hot-water-sanitizable
pH	4–11	Outside → membrane integrity loss
Inlet pressure	≤ 7 bar (100 psi)	Mechanical limit

5.5 Recommended `EDIDataPoint` fields

Required for computing FCE + full diagnostics:

feed_conductivity_uS_cm: float
feed_co2_ppm: float
feed_sio2_ppm: float
feed_temperature_C: float
product_conductivity_uS_cm: float (or product_resistivity_MOhm_cm)
concentrate_conductivity_uS_cm: float
feed_flow_m3h: float
product_flow_m3h: float
concentrate_flow_m3h: float
stack_voltage_V: float
stack_current_A: float
cell_pair_count: int (stack-config metadata)
Optional: feed_hardness_ppm, feed_chlorine_ppm

Sources: SnowPure Electropure OEM Manual v3.5.0; DuPont EDI-310 Module Manual (45-D01915); Suez E-Cell.

§6 — EDI KPIs and Formulas

6.1 Conductivity removal

η_κ = (κ_in − κ_out) / κ_in · 100        [%]

6.2 Product resistivity

ρ = 1 / κ

Theoretical max 18.18 MΩ·cm @ 25 °C (pure-water self-ionization). Industrial-grade EDI: ≥ 16 MΩ·cm; pharma UPW polish: 17–18.2 MΩ·cm.

6.3 Water utilization (recovery)

R = Q_product / Q_feed · 100

Typical 90–95 % for industrial EDI.

6.4 Reject ratio

RR = Q_concentrate / Q_feed

Typically 5–10 %.

6.5 FCE — Feed Conductivity Equivalent

The manufacturer-standard load metric for EDI sizing:

FCE [µS/cm] = κ + 2.79 · CO₂[ppm] + 1.94 · SiO₂[ppm]

κ = measured conductivity (µS/cm)
CO₂ = total carbonate (CO₂ + HCO₃⁻ + CO₃²⁻) as ppm CO₂
SiO₂ = total silica (reactive + colloidal)

The 2.79 / 1.94 coefficients reflect equivalent ionic conductivity contributed when each weak acid ionizes inside the stack (HCO₃⁻/CO₃²⁻ for CO₂; H₃SiO₄⁻/H₂SiO₄²⁻ for silica).

Target: FCE 1–5 µS/cm at EDI inlet for two-pass RO + EDI to UPW. Hard limit: FCE ≤ 40 µS/cm (stack-dependent 25–43).

⚠️ Some methodologies use 2.66 for CO₂ instead of 2.79; 2.79 is the SnowPure / E-Cell convention. Pick one and apply consistently across alerts and reporting.

⚠️ FCE requires total CO₂ (CO₂ + HCO₃⁻ + CO₃²⁻), not free CO₂ at sample pH. If the data pipeline ingests only "free CO₂", FCE is systematically under-reported → silent off-spec risk.

💡 EDI alerts that compare against raw feed_conductivity miss CO₂/SiO₂ load. Always compute FCE.

6.6 Specific power

kWh per m³ product, or kWh per kg salt removed. Industrial EDI: 0.1–0.3 kWh/m³ product typical.

6.7 Stack voltage drift (ΔV)

Baseline-tracked. Rising ΔV at constant current → resin/membrane resistance ↑ → scaling, fouling, or resin damage.

6.8 Faraday current efficiency

η_F = (n_ions_transported · F) / (I · t) · 100        [%]

F = 96 485 C/mol (Faraday constant); I = stack current (A); t in s; n in equivalents.

Reflects fraction of current moving target ions vs parasitic water-splitting. Industrial EDI: 60–90 % depending on load.

§7 — EDI Failure Modes (8 scenarios)

#	Failure mode	Detection signature	Root cause	Prevention	Recovery
1	Scaling in concentrate compartment	ΔP↑, ΔV↑, removal↓	Hardness leak > 1 ppm; CO₂ + Ca → CaCO₃ at high local pH from water-splitting	Hardness < 1 ppm; lower recovery	Acid clean (citric pH 2); replace if irreversible
2	Organic anion fouling	ΔV↑, anion removal↓ (higher product conductivity), TOC carryover	TOC > 0.5 ppm; humics adsorb on anion resin	RO pretreatment, GAC, low-fouling RO	Caustic + brine CIP; sometimes irreversible
3	Chloride breakthrough	Product Cl⁻ off-spec while bulk conductivity OK	High feed Cl⁻ + weak anion exchange at low voltage	Raise current density; reduce feed Cl⁻ via 2-pass RO	Increase voltage; verify polarity
4	CO₂ overload	Product conductivity↑ but ion-specific analyzers OK	RO permeate pH not raised; degasser bypassed/failed	Membrane degasser before EDI; raise RO permeate pH 8–9 with NaOH	Restore degasser; tune pH
5	Silica gel formation	ΔP↑, removal↓, irreversible	SiO₂ > 1 ppm + low local pH near anode → polymerized SiO₂	Feed SiO₂ < 0.5 ppm; warmer feed (> 15 °C); lower recovery	Generally not recoverable — replace module
6	Fe / Mn colloidal fouling	Visible brown deposit; ΔP↑; ΔV↑	Fe / Mn > 0.01 ppm; oxidation in feed	ORP control upstream; iron filter; < 0.01 ppm	Irreversible; replace
7	Polarization / over-limiting current	Random spikes in product conductivity; high ΔV; pH excursions in concentrate	Current density > limiting current for ion supply in D; excessive water splitting	Tune V below limiting current density; segment stacks	Reduce current; relax flow
8	Free-chlorine breakthrough	Sudden ΔV↑; anion resin → "oatmeal"; quality collapse; irreversible	Cl₂ > 0.02 ppm reaches stack (carbon-filter exhausted; SBS failure)	ORP sensor with hard interlock; redundant chlorine scavenger	Damage is permanent — replace module

Sources: DuPont EDI-310 Module Manual; SnowPure Electropure OEM v3.5.0; WC&P 2007 Fundamentals of EDI Technology; SnowPure Operating EDI in High-Silica Feedwater; ScienceDirect S0011916419308276.

§8 — EDI vs Mixed-Bed DI — Decision Economics

Criterion	EDI	Mixed-Bed DI
Chemical regeneration	No	Yes (NaOH + HCl)
Hazardous regen effluent	None	Yes (acid + caustic, neutralized)
Operation	Continuous	Cyclical (regen downtime)
CAPEX	Higher	Lower
OPEX	Lower (energy + small replacement)	Higher (chemicals + waste handling)
Footprint	Larger	Smaller
Onsite chemical storage	Minimal	Significant (HSE risk)
Best for	≥ 50 m³/d continuous; pharma; semi	< 20 m³/d intermittent; labs

Crossover capacity: ~30–50 m³/d depending on feed (Stark Water sizing guide).

EDI breakeven shifts lower (favors EDI) when:

Chemical procurement is constrained
Hazardous-waste disposal is costly
Site requires continuous UPW
Regulatory regime prefers chemical-free

§9 — Pharmaceutical UPW / WFI Context

9.1 USP <645> Water Conductivity

Three-stage test:

Stage 1 — in-line / at-line, uncompensated. Lookup table keyed to measured T. At 25 °C the limit is 1.3 µS/cm.
Stage 2 — sample equilibrated to 25 °C in container with stirring; limit 2.1 µS/cm. CO₂ ingress may cause failure.
Stage 3 — sample pH-adjusted to 4.0 with saturated KCl; pH-keyed limit (e.g. pH 5.0 → 4.7 µS/cm).

Sources: USP General Chapter <645>; Mettler-Toledo USP 645: 3 Steps.

⚠️ Stage 1 limits are uncompensated. Most online conductivity meters default to 25 °C-temperature-compensated readings. Mixing the two breaks compliance reporting. The compliance signal must be the raw, uncompensated value.

9.2 USP <643> Total Organic Carbon

TOC limit: 500 ppb (0.50 mg C/L) for both PW and WFI
System suitability: response to sucrose vs 1,4-benzoquinone CRS must be 85–115 %
Instrument LoD: ≤ 50 ppb

Sources: Beckman Changes to USP <643>; Veolia/Sievers; Mettler-Toledo.

9.3 Ph. Eur. monographs

Ph. Eur. 0008 (Purified Water): same TOC + conductivity envelope as USP for bulk PW
Ph. Eur. 0169 (Water for Injection): effective 1 April 2017 (Supplement 9.1) — allows non-distillation WFI: "Reverse Osmosis, single- or double-pass, coupled with electrodeionisation, ultrafiltration or nanofiltration, is suitable." Aligned EP with USP/JP

9.4 Ph. Eur. chapter 2.2.44 — effective 1 July 2026

The genuine 2026 normative change for pharma water:

TOC limit re-expressed 0.50 mg/L (from "0.5 mg/L") for significant-figure alignment
New Method B for Sterilized Water for Injection (replaces oxidisable-substances test)
Sucrose R and 1,4-benzoquinone R replaced by CRS reference standards (USP harmonization)
Term "highly purified water" replaced by "water" with κ ≤ 1.0 µS/cm @ 25 °C, TOC ≤ 0.1 mg/L

Sources: FDCELL Revision of Ph. Eur. Water for Pharmaceutical Use to Take Effect on 1 July 2026; EDQM Ph. Eur. 12.x updates.

9.5 Sanitization compatibility

Method	Compatibility
Hot water 80–85 °C	OK for HWS modules; weekly typical; verify TFC RO temperature spec
Steam (> 121 °C)	Limited; only fully steam-sanitizable PVDF assemblies; avoid for typical EDI/RO
Ozone (0.1–0.5 ppm)	Limited; destroys polyamide; OK for stainless distribution loops only
Peracetic acid (0.1–0.3 %)	OK for most modern membranes (consult vendor)
Formaldehyde (0.5–1 %)	Legacy; phased out for GMP/HSE reasons

9.6 Data integrity (21 CFR Part 11 / EU Annex 11)

21 CFR Part 11 (US FDA): audit trails, signature integrity, validated computerized systems, retention. Applies to instruments logging USP <645>/<643> data
EU Annex 11 (EudraLex Vol 4, Computerised Systems): risk-based validation; ALCOA+ principles; functional + data-integrity analog to Part 11

§10 — Regulatory Context (Produced Water)

10.1 WHO GDWQ 4th ed. (2022 + 1st & 2nd Addenda)

Selected drinking-water limits:

Nitrate (NO₃⁻): 50 mg/L
Fluoride: 1.5 mg/L
Arsenic: 10 µg/L
Lead: 10 µg/L
TDS: no health-based guideline; ~1 000 mg/L taste threshold

10.2 ISO 16075-2:2020 — Treated Wastewater for Irrigation

Class	Use	BOD avg/max (mg/L)	TSS avg/max (mg/L)	Faecal coliforms (95 %ile, /100 mL)
A	Unrestricted food crops	≤ 5 / ≤ 10	≤ 5 / ≤ 10	≤ 10
B	Restricted food (drip)	≤ 10 / ≤ 20	≤ 10 / ≤ 25	≤ 200 / ≤ 1 000
C	Industrial crops, fodder	≤ 20 / ≤ 35	≤ 30 / ≤ 50	≤ 1 000 / ≤ 10 000
D	No public access	≤ 60 / ≤ 100	≤ 90 / ≤ 140	Nematodes ≤ 1 egg/L; coliforms n/a

10.3 EN 12952-12 — Boiler-feed and Boiler-water Quality

Tiered limits keyed to drum pressure (general guidance — verify full table in standard before encoding limits):

Drum pressure	Feed κ (µS/cm)	SiO₂ (mg/L)	Hardness
< 20 bar	< 100	< 1	< 0.02 mmol/L
20–60 bar	< 30	< 0.3	< 0.01 mmol/L
> 100 bar (subcritical)	< 0.2	< 0.02	Fe < 0.02 mg/L; Cu < 0.003

⚠️ Do not hard-code EN 12952-12 limits as fixed thresholds. Pattern: drum-pressure-keyed lookup table cited to the standard ID.

10.4 SEMI F63:2021 — Semiconductor UPW

Resistivity: ≥ 18.2 MΩ·cm @ 25 °C
TOC: < 1 ppb (≤ 32 nm nodes); < 0.5 ppb for 3 nm / 2 nm advanced
Silica: < 0.3 ppb (some sources < 0.1 ppb leading-edge)
Particles: < 1 particle/mL above threshold (size class node-dependent)
Bacteria: < 1 CFU / 100 mL

§11 — Cross-cutting Code Recommendations

Tags: 💡 enhancement, ⚠️ potential bug / data-quality risk.

💡 EDIDataPoint should require: stack_voltage_V, stack_current_A, cell_pair_count, feed_co2_ppm, feed_sio2_ppm, feed_temperature_C. Without these, FCE and Faraday efficiency cannot be computed.
💡 EDIKPIs should expose: fce_uS_cm, faraday_current_efficiency_pct, product_resistivity_MOhm_cm, water_utilization_pct, stack_voltage_drift_pct.
💡 EDI_THRESHOLDS must be application-specific. Implement as EDIThresholdProfile enum: PHARMA_WFI, PHARMA_PW, POWER_BOILER_HP, POWER_BOILER_LP, SEMI_LEADING_NODE, SEMI_LEGACY_NODE, INDUSTRIAL_GENERIC.
💡 edi_calculator.py should use FCE as the load metric, not raw conductivity. The 40 µS/cm limit applies to FCE.
💡 New lifecycle_advisor.py module: input = cumulative chlorine exposure, CIP history, NPF/NSP baseline drift, ΔP trend, hours-on-stream; output = enum {TOLERATE, SCHEDULE_CIP, IMMEDIATE_CIP, REPLACE_LEAD, REPLACE_VESSEL_SET} with justification string.
⚠️ Concentrate flow derived as Q_feed − Q_product without sensor reconciliation hides leaks / cross-contamination. Add a closure check with warning at |Q_in − Q_out| / Q_in > 2 %.
⚠️ FCE depends on total CO₂ (CO₂ + HCO₃⁻ + CO₃²⁻), not free CO₂ at sample pH. Free-CO₂-only feeds under-report FCE silently.
⚠️ USP <645> Stage 1 limits are uncompensated. Compliance reporting must use raw uncompensated conductivity, not 25 °C-compensated readings.

Sources

Energy / SEC — Danfoss DESALRO 2.0 case story (Feb 2025); Danfoss Guinness press release; J. Chem. Ed. 2021 doi:10.1021/acs.jchemed.0c01194; Wang et al. Desalination 2016 (S001191641630087X); US-DOE Seawater Desalination Bandwidth Study (2017); Energy Recovery Inc. PX product page; Wikipedia DWEER.

Economics — Advisian Cost of Desalination; Energy-Solutions.co SWRO Cost Trends 2026; Foreverpureplace BWRO costs; ScienceDirect S2213138823000577 LCOW sensitivity; ScienceDirect S004896972402919X BWRO SEC optimization.

EDI — SnowPure Electropure OEM Manual v3.5.0; DuPont EDI-310 Module Manual (45-D01915); Suez E-Cell EDI documentation; SnowPure FCE technical article; Stark Water FCE EDI blog; Ionpure LX-HI datasheet DB-LXHI; WC&P 2007 Fundamentals of EDI Technology; SnowPure Operating EDI in High-Silica Feedwater.

Pharma — USP General Chapters <645> and <643>; Mettler-Toledo USP guides; Beckman / Veolia / Sievers TOC references; PharmOut Ph. Eur. allows non-distillation WFI from April 2017; A3P membrane vs distillation comparison; EMA Q&A on non-distillation WFI; FDCELL Ph. Eur. Water Revision 1 July 2026; EDQM Ph. Eur. 12.x updates.

Produced-water regulatory — WHO GDWQ 4th ed. (2022 + addenda); ISO 16075-1..4 (2020–2021); EN 12952-12; SEMI F63:2021; ASME/EPRI boiler-feed guidance; EPA WaterReuse.

Membrane life — FilmTec/DuPont 45-D01911 Exceptional Membrane Life; Aqualitek SWRO maintenance guide; Pumps & Systems Life of an RO Membrane multipart.

membrane-economics-edi