Additive manufacturing (AM) process is used to elaborate deposits of alloy 625 on a low-alloy ASTM A387 steel substrate. The effect of NaCl deposit on the alloy corrosion behavior is investigated at 800 degrees C. The presence of NaCl deposit conduct to an important corrosion of the alloy with the formation of a thick corrosion scale and numerous voids in the metal beneath the scale. The presence of gaseous metal chlorides lead to the establishment of a self-sustaining active corrosion mechanism. Thermodynamic calculations allow to assess chlorine influence on alloying elements and to understand their contribution in oxide layers and voids formation.
While visual foundation models aim to provide all-purpose features that generalize across tasks, their out-of-the-box performance can be suboptimal in specialized domains with significant distribution shifts. This paper investigates how to best adapt them in such scenarios. We propose a simple yet powerful two-stage adaptation strategy: we first continue the model’s self-supervised pre-training on in-domain images, then perform standard supervised fine-tuning. We validate this framework by adapting the DINOv2 model to the task of sewer defect classification. Our results demonstrate that this adaptation strategy is effective, achieving state-of-the-art performance on the public Sewer-ML benchmark by matching the leading method’s F_2^CIW score (73.15) while attaining a superior F_1 score for the normal class (93.67), the two metrics established for the official benchmark. Most significantly, this result is achieved using a standard ViT-Small model with only 22 million parameters—an order of magnitude smaller than the previous 227M-parameter SOTA model—and without any task-specific architectural modifications or hyperparameter tuning. Furthermore, we show our method is highly competitive with leading semi-supervised techniques in low-data regimes. These findings establish a simple, efficient, and general framework for deploying foundation models in specialized industrial domains.
Two moving bed bioreactors (MBBRs) with AnoxK™K5 carriers were operated for 156 days to compare methanol and ethanol as carbon sources in a single-stage partial denitrification-anammox (PdNA) process treating real municipal secondary effluent. The ethanol-fed reactor demonstrated superior performance with nitrogen removal rates (NRR) of 0.54 ± 0.1 gN.m−2.d−1 and nitrogen removal efficiencies (NRE) of 76 ± 7% at 20 °C, compared to 0.43 ± 0.1 gN.m−2.d−1 and 61 ± 6% for the methanol-fed reactor. Anammox contribution in the ethanol reactor remained consistently high (80 ± 7%) compared to the methanol reactor (57 ± 9%), with microbial analysis revealing selective enrichment of methylotrophic denitrifiers (Hyphomicrobium_A) under methanol conditions, promoting complete denitrification at the expense of anammox. Both systems demonstrated robustness across three different wastewater compositions and rapid recovery from operational disturbances. At 15 °C, the ethanol reactor maintained 82 ± 4% NRE, although requiring ∼50% lower nitrogen loading rates. COD consumption ratios of 2.2 ± 0.3 gCOD/gN-NO3 for ethanol achieved 53 ± 7% carbon savings compared to full denitrification requirements. Thermodynamically derived PdN and anammox coupled stoichiometries accurately predicted performance (R2 = 0.99 for NRR), identifying optimal feeding ratios of 1.00 gN-NO3/gN-NH4 and 2.06 gCOD/gN-NO3 for ethanol systems. This long-term evaluation of ethanol-based PdNA in an MBBR system treating real municipal wastewater, a combination not previously assessed - demonstrates the superiority of ethanol over methanol through higher anammox contribution, lower carbon requirements, and robust performance across varying wastewater compositions and temperatures.
Urban flooding: Constructing climate-resilient infrastructure Àlex de la Cruz Coronas and Patricia Molina López discuss the escalating risks of urban flooding in Europe, highlighting the limitations of traditional flood hazard maps and emphasizing the need for more comprehensive data on the impacts of flooding. Urban flooding is rapidly emerging as one of the most disruptive climate risks faced by Europe’s metropolitan and coastal regions. Intensifying rainfall, sea-level rise, and continued urbanisation are increasing both the frequency and severity of flood events. In densely populated areas, impacts extend far beyond water depths on streets: transport systems stall, energy and water services are disrupted, businesses close, and public safety is compromised. Flood hazard maps alone are no longer sufficient; adaptation requires evidence on who and what is at risk and how severe impacts may be.
Return sludge sidestream (RSS) enhanced biological phosphorus removal (EBPR) processes currently lack clear design guidelines for sidestream tank (SST) hydraulic retention time (HRT) and return activated sludge (RAS) diversion ratio, which have often been evaluated as interdependent parameters. This study evaluates a pilot-scale Modified Ludzack-Ettinger (MLE) reactor with and without an RSS treating real domestic wastewater, independently assessing sidestream tank HRT and RAS diversion ratio. Nutrient removal performance, nitrous oxide (N₂O) emissions, sludge production, and microbial community dynamics were investigated. A sidestream tank HRT of 24 h led to improved phosphorus (P) removal performance and stability (93.3 ± 7.0% versus 75.7 ± 17.3% at 16 h), though anaerobic mass fraction was controlled by either sidestream tank HRT or RAS diversion ratio. The optimised RSS phase eliminated chemical P precipitation requirements and reduced total sludge yield by ∼31-47% via in situ fermentation as compared to MLE operation, comparable to reductions previously reported at sidestream tank HRTs >48 h, with substantially larger anaerobic tank volumes. Coupling RSS with intermittent aeration enabled stable nitrite accumulation while maintaining low N₂O emissions, with an average emission factor of 1.1 ± 0.7%). Microbial community analysis revealed enrichment of polyphosphate-accumulating organisms (PAOs) (notably Dechloromonas) and suppression of glycogen-accumulating organisms (e.g., Competibacter <1 %), consistent with sustained EBPR stability. Collectively, these findings provide practical design guidance for RSS retrofits, including selection of sidestream tank HRT, RAS diversion ratio, and anaerobic mass fraction, enabling stable EBPR performance, reduced sludge production, and minimal environmental footprint, thereby supporting cost-effective full-scale implementation.
