WARNV5Coulombic mass-balance is open on 9/12 sweep points (>5% closure error). Check substrate consumption + Q_measured.· Logan-group hygiene check; mass balance fundamentals
WARNV6Mean CE × OLR = 47.1 exceeds empirical Logan-group envelope (≤25). Mass transport limits CE at high loading; prediction may be optimistic.· Cheng & Logan 2011 — empirical envelope
Reproducing
Cao X, Huang X, Liang P, Xiao K, Zhou Y, Zhang X, Logan BE (2009) — Environmental Science & Technology
A new method for water desalination using microbial desalination cells
DOI: 10.1021/es901950j
Metric
Reported
Predicted
Δ
Band
Power density
peak power density during the desalination batch cycle, 5 g/L initial NaCl, ferricyanide catholyte, acetate-fed anode
p. 7150, Fig. 4 + text
480
mW/m²
608
mW/m²
+26.6%
amber
COD removal
NaCl removal from the middle chamber over a single batch cycle, 5 g/L initial salinity — the founding "90% desalination" headline. Maps to the `substrate` series for the side-by-side because the schema does not yet have a dedicated `salt_removal_pct` key.
p. 7150, Fig. 3 + Abstract
90.0
%
92.9
%
+3.2%
green
Coulombic efficiency
anode coulombic efficiency for acetate-fed Geobacter-enriched community; consistent with Logan-group ferricyanide-cathode MFC baseline since the MDC anode chamber is essentially an MFC anode
p. 7150, Table 1 / §Results
60.0
%
48.1
%
-19.8%
green
Δ-bands per Logan & Regan 2006 lab-to-lab reproducibility envelope: green ≤±25%, amber ±25–50%, red >±50%. Predictions from Butler-Volmer + Ohmic + Tafel closure calibrated against the Logan corpus (regression-tested in butler-volmer-calibration.test.ts).
What we don't know about this paper (6 items)
BV MODEL CALIBRATION IS MFC-ONLY. MDC closure (salinity-driven IR, co-ion leakage, desalination efficiency) lives in `predictors/mdc.ts` but is NOT dispatched by the backtest harness yet. Expected band: amber/red on the salt_removal mapping; power and CE may be green-ish because they look like an MFC anode at the BV level.
The `substrate` seriesKey is being used as a proxy for salt removal (% NaCl removed from middle chamber) because the schema does not yet have a `salt_removal_pct` key. The comparison is INCOMMENSURABLE in the strict sense — extending the seriesKey set is tracked as a follow-up for MDC-class predictors.
The paper ran multiple initial-NaCl conditions (1, 5, 20, 35 g/L); we encode only the 5 g/L brackish run. Higher-salinity runs show LOWER % removal per cycle (5 g/L → 90%, 35 g/L → ~63% in the same paper, requiring multiple cycles).
Ferricyanide catholyte is research-only — toxic and unsustainable for real applications. Later MDC work (Jacobson 2011 air-cathode MDC, capacitive MDCs) shifted to air cathodes. The 480 mW/m² power density would be substantially lower with an air cathode.
The 7 cm² anode area is small even by 2009 lab-bench standards; pilot-scale MDCs (e.g., Forrestal 2012) show different power-vs-area scaling because of pumping losses, concentration polarization at the AEM, and electrode aging.
The headline 90% removal applies to a single batch cycle starting from 5 g/L; the limit-of-desalination is governed by the diminishing driving force as the middle chamber depletes (Donnan + concentration potentials oppose the cell potential).