Despite routine cleaning and recalibration, wastewater online analysers often experience signal degradation under harsh operating conditions, creating a need for reliable automated data quality assurance. This study develops and evaluates a framework for near real-time fault detection and data imputation in wastewater analysers. Established statistical methods and recent deep learning approaches, including sequence-to-sequence imputation, were reviewed, implemented, and tested using both synthetic and real operational data. A synthetic dataset was generated using Benchmark Simulation Model No. 2 (BSM2), and representative faults, including noise, spikes, frozen measurements, and sensor drift, were systematically introduced to enable controlled evaluation. Fault detection performance was assessed using confusion matrix metrics, while imputation methods were evaluated based on reconstruction accuracy. The framework was further validated using operational data from the Hias Water Resource Recovery Facility (Hamar, Norway), providing insight into the challenges of transferring methods from simulation to real systems. No single method consistently outperformed others across all fault types. Wavelet and Kalman filters achieved the highest precision and recall for spike detection, Spearman correlation showed the highest recall for frozen measurements, and CUSUM and ridge-regularized partial correlation performed best for drift detection. Sequence-to-sequence neural networks and k-nearest neighbors methods delivered the most accurate imputations. Although methods calibrated on BSM2 data showed reduced performance on operational data, the results demonstrate the feasibility of a practical framework for automated near real-time monitoring of wastewater analyser data quality. Future work should focus on ensemble fault detection and computationally efficient time-series imputation models.
Maintaining snow- and ice-free surfaces on outdoor facilities such as sports fields and pedestrian walkways during winter is essential for safety and year-round usability, yet difficult to achieve under harsh and highly variable environmental conditions. Underground heating systems (UHS) used for this purpose typically rely on a flow–thermal architecture, where warm fluid circulates through a buried pipe network to supply heat to the soil. Automatic control of such systems is challenging due to the large thermal inertia of the soil, delayed heat propagation in the fluid network, uncertain ground properties, and rapidly changing weather conditions, while most existing installations are still manually operated. This paper presents a dynamic modeling and control framework for the underground flow–thermal system based on a differential–algebraic equation (DAE) formulation that captures the coupled thermal dynamics between the soil, circulating heating fluid, and ambient environment. Model parameters are identified using a limited set of experimental data, explicitly accounting for uncertainties in soil thermal properties and heat transfer characteristics. To improve robustness against parameter variations and measurement noise, an Extended Kalman Filter (EKF) is employed for soil temperature estimation. A constrained Linear Quadratic Gaussian (LQG) tracking controller with integral action and anti-windup compensation is then designed to regulate the inlet fluid temperature in accordance with specified quality-of-service objectives. Simulation results using real operational data demonstrate accurate soil temperature estimation, effective disturbance rejection, and reliable reference tracking under varying outdoor conditions while respecting actuator constraints, indicating that the proposed approach provides a physically consistent and practically applicable solution for the automatic control of UHS in outdoor environments.
The amine-based solvent method for carbon capture is important for reducing carbon dioxide (CO₂) emissions in multiple industrial sectors. Blended amine systems like methyl diethanolamine/piperazine (MDEA/PZ) have the capacity to reduce energy consumption and increase efficiency compared to the traditional monoethanolamine (MEA). This study simulated CO₂ absorption and desorption processes employing 29 wt% MEA and a blended solvent of 30 wt% MDEA and 20 wt% PZ in Aspen HYSYS v14, utilizing absorbers and desorbers with 10 and 6 stages, respectively. The effectiveness of each solvent system was assessed based on an assumed CO₂ removal efficiency of 85 %, with equipment sizing and cost estimation performed using Aspen In-Plant Cost Estimator. The simulation findings demonstrated heat of consumption values of 3.77 MJ/kg CO₂ for MEA and 3.65 MJ/kg CO₂ for the MDEA/PZ blend. The computed CO₂ capture costs were 31 €/tonne CO₂ for MEA and 28 €/tonne CO₂ for MDEA/PZ. In parameter variations, the MDEA/PZ blend consistently demanded less energy than MEA. There is some work in the open literature comparing the energy consumption of the MEA and the MDEA/PZ blend, but little work on comparing total cost. The results indicate that blends like MDEA/PZ can be an alternative to MEA offering lower energy requirements and lower total cost.
Conventional Haber-Bosch (HB) process for NH3 production remains dominated by fossil sources and is associated with almost 2 % of the total final energy consumption. Therefore, sustainable ways for its generation are sought. This study develops an Aspen Plus modelling framework to quantify how alternative systems could shift operating requirements and enable tighter integration with low-grade heat sources compared with HB. A built-in flowsheet is first implemented as a reference case. A chemical looping-based ammonia synthesis configuration, with iron-based oxygen carrier, is then modelled and compared. Finally, a third system small-scale for ammonia production is presented, based on a reactor that treats exhaust gases from an internal combustion engine as a H2 source via iron oxidation. The recovered H2 is subsequently reacted with N2 from the exhaust to form NH3, while iron can be regenerated. Parametric studies over the process conditions map the feasible operating regimes, while comparing NH3 yield and materials/energy consumptions.
In today’s rapidly evolving world, digital services play a pivotal role in fostering a smart circular economy. These services leverage advanced technologies to optimize resource use, minimize waste, and promote sustainability. By facilitating the efficient flow of information and resources, digital platforms enable businesses and consumers to adopt circular practices more easily. In this work, we highlight key digital services that help build a robust and interconnected circular economy, showcasing their potential to transform traditional business models and drive long-term environmental benefits. In addition, we mapped and aligned these services with the circularity strategies, commonly referred to as the 10 Rs, within the production chain. The proposed services-to-strategies mapping provides a framework for promoting circular design and manufacturing practices that aim to extend the lifecycle of resources and prevent waste from affecting the environment.
