membrane-economics-edi

What this skill does

# 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

1. Electricity unit price
2. Membrane lifespan & replacement rate
3. Recovery rate (R)
4. Plant capacity factor (uptime)
5. 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