{"id":"c736df97-1b6a-402c-9788-e139ec23b00e","arxiv_id":"2509.08394","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A CFD model of heat and mass transfer in wet cooling towers, implemented in code_saturne, reproduces MISTRAL loop exit temperatures within 10% and evaporation rates within 15%.","lead":"This paper develops a computational fluid dynamics model for wet cooling towers and tests it against data from a large experimental loop at a French nuclear plant. It reports temperature and evaporation predictions within 10 to 15 percent of measurements, a step toward forecasting water consumption and plume dispersion in power plants.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"2D vs 3D validation gap: the claimed agreement with MISTRAL is load-bearing but the paper gives no evidence a 2D slice captures the 3D side-inlet/recirculation/rain-air flow; a 3D run is needed.","rationale":"Good-faith reading: the paper is a short engineering-validation paper, and it is transparent about missing optimization and about systematic bias. The most important thing that must be true for the headline claim is that the simulation reproduces the actual MISTRAL facility. The text explicitly says the simulation is 2D, but the facility is 3D and has a side inlet. Without any 2D/3D comparison the validity of comparing a 2D field to 3D measurements is unsupported. The reader's weakest assumption is exactly this, and I agree. I considered the adjusted coefficients A=0.8, n=0.6 as an alternative concern; their provenance may affect the interpretation of the agreement, but the 2D geometry is upstream and more clearly load-bearing because even a perfectly calibrated closure set could not rescue a wrong flow topology. My proposed test is a targeted 3D rerun on a subset, which would directly settle whether the 2D assumption is benign. Since this is a call for missing evidence rather than a demonstrated contradiction, the appropriate disposition remains the reader's CONDITIONAL (no verdict change).","tokens_in":3675,"tokens_out":4046,"duration_ms":44132,"concrete_test":"Run a full 3D simulation of the MISTRAL geometry (7×7 m section, 5×10 m east inlet, packing/rain zones, top fan) for a representative subset of the 55 cases spanning the tested air/water flow-rate ranges, using the same turbulence model, closure laws, boundary conditions, and numerical settings as the 2D runs. Compare 3D exit temperatures, evaporated mass flow, and exchanged power with both the experimental data and the published 2D results. If the 3D values fall outside the claimed 10%/15% error band or deviate from 2D by more than those bands, the 2D validation cannot support the central claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that the proposed code_saturne model reproduces MISTRAL measurements to within 10% (temperatures) and 15% (evaporation) over 55 cases. The only simulation evidence shown is 'a typical result of a 2D MISTRAL bench simulation' (§3.2). But MISTRAL is fundamentally 3D: a 7×7 m square section, a fan at 18.7 m, a 5 m wide × 10 m high air inlet on the east side located 25.5 m from the packing centre, and a 10 m rain zone. A 2D vertical-plane model cannot represent the lateral spread of the incoming air jet into the square section, the spanwise distribution of rain, or 3D recirculation loops. The two recirculation regions reported in Fig. 1 (west of the building and bottom of the rain zone) are plane vortices; their 3D counterparts need not have the same intensity or location. No symmetry argument, no 3D-vs-2D comparison, and no quantified spanwise variation is given. If the 2D geometry artificially channels the flow and enhances/cancels heat and mass transfer, the claimed agreement could be a numerical artifact rather than evidence that the Poppe evaporation, Ranz-Marshall, and Dreyer closures are adequate. This is the weakest load-bearing assumption for the validation claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a CFD model in code_saturne for heat and mass transfer in wet cooling towers, with focus on evaporation in the fill-pack and rain zones. The model closes the two-phase exchanges through Poppe evaporation, Bosnjakovic Lewis-factor convection, Ranz-Marshall rain-zone Nusselt correlation, and Dreyer rain-air friction. The authors validate the model against the MISTRAL reduced-scale experimental loop at Bugey (55 cases) and report temperature predictions within a 10% error margin and evaporation within 15%. They conclude that the modeling is a first step for predicting water consumption and humid-air plume dispersion.","tokens_in":4022,"tokens_out":4030,"duration_ms":47288,"significance":"If the reported agreement holds, the model would provide a useful, open-source-based CFD capability for wet-cooling-tower design and plume-impact studies. The paper is honest about several limitations: it explicitly states that no optimization of the packing evaporation correlation was performed, that the rain injection zone is not yet simulated, and that the simulations are currently 2D. The strength of the work is the comparison against an independent experimental dataset. However, the validation evidence is currently insufficient to support the central quantitative claim: the 2D-vs-3D gap, the lack of a defined error metric, missing mesh-convergence and uncertainty information, and the unstated provenance of the adjusted coefficients A and n all leave room for the reported 10%/15% agreement to be partly an artifact. The paper has value as a progress report, but the validation needs to be made robust before it can support the claimed predictive capability.","major_comments":[{"comment":"The MISTRAL bench is described in §3.1 as a 7×7 m square-section loop with a side air inlet (5 m wide, 10 m high, 25.5 m from the packing center), which is inherently three-dimensional. The only simulation shown is a 'typical result of a 2D MISTRAL bench simulation' (Fig. 1). No symmetry justification, no 2D-vs-3D comparison, and no quantified spanwise variation are provided. The two recirculation zones in Fig. 1 are plane vortices; their 3D counterparts need not have the same location or intensity. Because the claimed 10%/15% agreement is the core validation evidence, the possibility that 2D channeling artificially changes heat/mass transfer is load-bearing and must be addressed by a 3D run or a convincing reduced-dimensionality argument.","section":"§3.2, Fig. 1"},{"comment":"Equation (4) uses the packing evaporation coefficients A≈0.8 and n≈0.6. The text calls them 'adjusted coefficients' but does not state whether these values were taken from the literature, chosen from prior calibration, or fitted to the same MISTRAL dataset used in §3.2. If they were fitted to these 55 cases, the validation is partly circular. The statement in §3.2 that 'we did not conduct an optimization of the packing evaporation correlation' mitigates, but does not resolve, the ambiguity. Please state the provenance of A and n and whether any parameters were adjusted after comparison with the experimental data.","section":"§2.2, Eq. (4)"},{"comment":"The reported '10% error margin and 15% for the evaporation rate' is not defined. Is it a mean relative error, a maximum per-case error, or a band enclosing most points? No experimental uncertainties are given, and no mesh-convergence study is reported. Without these, the reader cannot assess whether the agreement is within the noise of the measurements or the discretization. Add a precise error metric, report measurement uncertainties, and include at least one mesh-convergence test for the MISTRAL configuration.","section":"§3.2"},{"comment":"The authors state that the rain injection zone (the spray zone above the packing) is not simulated, and that this omission likely explains the stable underestimation of evaporation and the corresponding systematic bias in outlet temperatures. Since this zone is an integral part of a real cooling tower and contributes to evaporation and plume formation, the validation does not yet cover the full configuration relevant to the paper's stated goals. This limitation is acknowledged in good faith, but the conclusions should be reframed as partial validation: the model captures the fill-pack and cold-rain-zone behavior, but the missing spray zone could affect water-consumption and plume predictions. Please quantify the expected contribution of the spray zone or explicitly limit the scope to the modeled zones.","section":"§3.2, §4"}],"minor_comments":[{"comment":"The sentence 'A 0.8 and n 0.6 are adjusted coefficients that ≈ ≈ represent the evaporative capacity' has garbled typography; it should read 'A≈0.8 and n≈0.6 are adjusted coefficients that represent...'.","section":"§2.2, Eq. (4)"},{"comment":"In the version provided, Equations (1)–(4) are referenced but not displayed. Please ensure all equations appear in the final manuscript, with definitions of every symbol.","section":"§2.1, Eqs. (1)–(3)"},{"comment":"Figure 2 does not specify what quantity is shown on each axis beyond the text, nor whether points represent individual cases, means, or ranges. Add axis labels, units, and a legend, and indicate whether the comparison is per-case or averaged.","section":"Fig. 2"},{"comment":"The phrase 'typical result of a 2D MISTRAL bench simulation' should be clarified: is the 2D slice aligned with the east-west inlet direction? What is its thickness/span? Adding a schematic of the computational domain would help.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"This manuscript reads as a condensed conference paper. Even after revising the major technical points, the authors may need to expand the description of the numerical setup (boundary conditions, mesh, solver settings, turbulence closure details) to meet the standards of a full archival journal. The 2D-vs-3D issue and the provenance of the packing coefficients are the key factors that currently prevent acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a legitimate 'first steps' engineering paper, not a fundamental advance. It implements known closure laws (Poppe, Bosnjakovic, Ranz-Marshall, Dreyer) in code_saturne and validates against the MISTRAL loop. The 10% temperature / 15% evaporation agreement over 55 cases is a solid engineering result, and the authors are honest that they did not tune the packing correlation to these data and that the rain injection zone is missing. I'd give credit for that honesty and for validating against an independent experimental facility.\n\nWhere it gets soft: the only simulation shown is a 2D slice of a fundamentally 3D rig. MISTRAL has a 7x7 m square section and a 5 m side inlet; a 2D plane cannot represent lateral spreading or 3D recirculation. The paper gives no symmetry argument or 2D-vs-3D comparison, so the agreement could be partly a byproduct of geometry channeling the flow. That's my main concern, and it's not minor. The adjusted coefficients A and n have no stated provenance; the authors say they didn't optimize them for this dataset, but we don't know where they came from. If they came from a prior fit to MISTRAL-like data, that's a small circularity burden. There are also no error bars, no mesh-convergence study, and the error metric isn't defined (relative to what? mean of measured range?).\n\nStill, the central argument—the model captures the dominant heat and mass transfer—probably holds. The 2D issue would be addressed by one 3D run; the coefficient origin is a one-line fix. Don't treat this as a takedown; it's a conference paper that does what it promises, with clear prose and reasonable claims.\n\nWho should read it: engineers working on cooling tower CFD or plant water consumption, and anyone developing multiphase heat/mass transfer closures in code_saturne. For a journal submission, I'd send it to peer review with the request for a 3D validation case and better reproducibility info. For this proceedings version, it's fine as-is but shouldn't be cited as strong evidence of 3D predictive capability.","headline":"Modest engineering validation of a standard wet-cooling-tower CFD model in code_saturne; the 10%/15% agreement is believable but the 2D-vs-3D gap is the load-bearing weakness.","tokens_in":4486,"tokens_out":2184,"would_cite":false,"duration_ms":22020,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proposes a CFD model in the code_saturne solver for heat and mass transfer in wet cooling towers and validates it against the MISTRAL experimental loop, claiming exit temperatures within 10% error and evaporation rates within 15%","keywords":["wet cooling towers","CFD","heat and mass transfer","evaporation","plume formation","code_saturne","MISTRAL","validation"],"falsifier":"Run the same 55 MISTRAL cases in a 3D simulation of the square-section loop with identical closure laws and check whether exit temperatures and evaporated masses stay within the claimed 10% and 15% margins; if they do not, the 2D representation is carrying the agreement.","tokens_in":3611,"feed_emoji":"💧","tokens_out":5915,"duration_ms":57148,"temperature":0.7,"pith_summary":"The paper is trying to establish that a dedicated CFD model, implemented in the code_saturne solver, can reproduce the main heat and mass transfer physics inside wet cooling towers. It validates the model against the MISTRAL experimental loop, a reduced-scale facility at a nuclear power plant, and claims agreement within 10% for air and water exit temperatures and within 15% for evaporated water mass flow across 55 cases. This matters because wet cooling towers are central to power-plant cooling and are responsible for water consumption and visible humid plumes. The authors position the model as a first step toward detailed CFD predictions of water consumption and plume dispersion.","feed_headline":"CFD model predicts cooling-tower evaporation within 15%","feed_subtitle":"Validated CFD reproduces cooling-tower temperatures and evaporation, a step toward plume prediction.","key_machinery":"The load-bearing machinery is the closure-law stack: Poppe's evaporation flux in the fill packing with adjusted coefficients A ≈ 0.8 and n ≈ 0.6; the Bosnjakovic Lewis-factor relation tying convective heat transfer to evaporation; the Ranz-Marshall Nusselt correlation for rain-zone interfacial heat transfer; and Dreyer's interfacial friction model between air and rain. These closures turn a two-phase CFD calculation into a locally computable heat-and-mass-transfer model, rather than prescribing a global tower characteristic. The drift-velocity transport of injected liquid water is also central because it lets the model track water through the packing and rainfall.","core_discovery":"The paper's central claim is that a CFD model built in code_saturne, combining humid-air Navier-Stokes equations with scalar transport of water vapour and injected liquid water, reproduces the MISTRAL cooling-tower experiment well enough to be useful. The model computes evaporation and convection locally: Poppe's evaporation formulation is used in the fill packing, the Bosnjakovic Lewis factor couples convective and evaporative heat transfer, Ranz-Marshall correlation handles rain-zone heat transfer, and Dreyer's approach models air-rain friction. The validation shows a satisfying agreement on exit temperatures, evaporated mass flow, and total exchanged thermal power, with a stable underesti","pith_inferences":["If the reported accuracy transfers to full-scale natural-draft towers, cooling-tower water consumption could be estimated from CFD rather than calibrated one-dimensional correlations, altering how plant water balance and plume impact are assessed.","Because the packing coefficients are adjusted rather than derived, a useful test is to replace them with geometry-resolved fill-pack hydrodynamics; the MISTRAL dataset would then serve as a benchmark for whether resolved packings outperform calibrated ones.","The 2D representation is a specific assumption tied to this facility's geometry; a 3D simulation under crosswind conditions is a natural next validation before relying on the model for real-world plume dispersion.","The identified missing rain-injection zone implies a testable prediction: adding that zone should simultaneously reduce the evaporation error and the temperature bias."],"forward_implications":["For the 55 MISTRAL cases, the model reproduces exit water and air temperatures to within 10% and evaporated water mass flow to within 15%.","The model provides a CFD-based route to estimating cooling-tower water consumption, since the evaporated mass flow is a computed output rather than a prescribed global value.","Because exhaust air temperature and humidity are solved fields, the same simulation yields inlet conditions for modelling plume dispersion in the atmosphere.","The systematic underprediction of evaporation, linked to the missing pre-packing rain injection zone, identifies a concrete next modelling step rather than a fundamental flaw.","The separation of fill-pack and rain-zone closures allows each zone to be refined independently with new experimental data."],"supporting_citations":[{"why":"Supplies the finite-volume CFD solver in which the wet-cooling-tower modelling is implemented.","marker":"[8]"},{"why":"Provides the linearised-production k-epsilon turbulence closure used for the air flow.","marker":"[9]"},{"why":"Supplies the drift-velocity model for transporting injected liquid water relative to the humid air.","marker":"[10]"},{"why":"Gives the Poppe evaporation flux formulation used in the fill-pack zone.","marker":"[11]"},{"why":"Supplies the Bosnjakovic Lewis-factor relation that couples convective heat transfer to evaporation.","marker":"[12]"},{"why":"Provides the Ranz-Marshall Nusselt correlation for interfacial heat transfer in the rain zone.","marker":"[13]"},{"why":"Gives the Dreyer approach for air-rain interfacial friction that shapes recirculation and rainfall flow.","marker":"[14]"},{"why":"Cited as a previous CFD model that optimized the packing evaporation correlation; the paper compares its non-optimized approach against this baseline.","marker":"[3]"}],"fun_headline_variants":["Validated CFD captures cooling-tower evaporation and heat transfer","Cooling-tower plume prediction advances with validated CFD","Code_saturne CFD reproduces cooling-tower plume dynamics","Heat and mass transfer in wet cooling towers now CFD-validated"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The claimed agreement rests on the untested premise that a 2D representation of a 3D square-section experimental loop with a side air inlet captures the heat and mass transfer that governs the measured exit quantities.","fun_headline_variants_meta":{"raw":{"variants":["Validated CFD captures cooling-tower evaporation and heat transfer","Cooling-tower plume prediction advances with validated CFD","Code_saturne CFD reproduces cooling-tower plume dynamics","Heat and mass transfer in wet cooling towers now CFD-validated"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00037,"raw_usage":{"total_tokens":1784,"prompt_tokens":673,"completion_tokens":1111,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":417,"completion_tokens_details":{"reasoning_tokens":1042}},"tokens_in":417,"tokens_out":1111,"duration_ms":11558,"temperature":1.0,"reasoning_tokens":1042,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T20:37:55.941630+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same 55 MISTRAL cases in a 3D simulation of the square-section loop with identical closure laws and check whether exit temperatures and evaporated masses stay within the claimed 10% and 15% margins; if they do not, the 2D representation is carrying the agreement.","supporting_citations":[],"review_version":1}