Jan Kloppenborg Møller

The District Heating company of Copenhagen owns and maintains operation of the large pipe distribution network under the city. This means replacing pipes as failures occur. As the network age the frequency of failures has increased, and with it the accumulated evidence of pipe lifetime.The question is whether replacement at failure remains the right strategy for the company, or whether planned bulk replacements with lower unit costs should be considered. This standard question of Age Replacement is made curious by the fact that the number of pipes replaced as reaction to a failure varies in Copenhagen. Thus, the lifetime records contain two entangled random variables; time to failure and the scope of replacement. In this paper, we argue for untangling the two consistent with the concept of independent censoring in Survival Analysis.Independence between the mechanisms that govern censoring and failure is what allows unbiased estimation of the optimal replacement time. Using both parametric and nonparametric methods we estimate the survival function of pipes in Copenhagen and predict the optimal replacement time under different cost scenarios.

Solvent-based post-combustion carbon capture (PCC) is a mature but energy-intensive technology for capturing CO2. To reduce the operational costs, flexible operation plays a key role by exploiting time-varying electricity prices. Advanced control strategies, such as economic model predictive control (EMPC) together with accurate and computationally efficient dynamic models, can be used to enable flexible operation, while ensuring that operational constraints are satisfied. In this work, we propose a nonlinear neural network-based grey-box model of the solvent-based PCC process. We train and validate the model using experimental data from the pilot plant at the Danish Technological Institute (DTI). Using the validated model, we design, implement, and demonstrate EMPC on the DTI pilot plant. To the best of the authors' knowledge, this is the first real-world demonstration of EMPC on a solvent-based PCC plant. We compare a non-flexible baseline scenario that minimizes energy consumption to a flexible operation scenario that minimizes the total energy costs, given a time-varying electricity price. We demonstrate that the closed-loop EMPC is able to reduce energy costs by 25%, compared to the baseline scenario, while meeting a minimum capture efficiency requirement of 30%. These results demonstrate the suitability of the proposed NN-based grey-box model for closed-loop EMPC and the potential of flexible operation to reduce the operational costs of solvent-based PCC plants.

Accurate forecasting of indoor temperature is essential for effective climate control in buildings. However, existing models used in Model Predictive Control (MPC)-based demand response applications often do not fully take into account key dynamic factors, particularly those introduced by heat emission system(s), weather variability, and occupancy. One critical limitation is the treatment of thermostatic radiator valve (TRV) dynamics and the discrepancy between the temperature perceived by occupants (i.e. the actual indoor temperature) and that measured by the TRV. This discrepancy is typically approximated by a fixed offset, despite evidence that it varies over time and depends on radiator activity, weather, and occupancy. In this paper, a novel dynamic grey-box model is proposed for room-level temperature forecasting that explicitly accounts for four key contributors, namely outdoor air temperature, solar radiation, TRV-controlled radiator heating and building envelope characteristics. The model introduces separate dynamic states for TRV temperature and actual room temperature, enabling the model to capture the time-varying offset while respecting the causal thermostatic control structure, where radiator actuation depends solely on the TRV temperature. The results demonstrate substantial improvements in forecasting accuracy, highlighting the model’s suitability for advanced control strategies aimed at reducing energy use for space heating while maintaining thermal comfort for occupants.

Solvent-based post-combustion carbon capture (PCC) is a mature and highly exploited carbon capture technology that benefits from dynamic models for design improvement, optimal operations, and model-based control. The current models rely on detailed properties of solvents, have high complexity that limits their use to steady-state dynamics, and typically do not integrate timedependent changes in data from specific capture plants. Here we present a lumped, low-complexity, data-driven, dynamic, and stochastic grey-box model of a PCC system, validated on experimental data from a Danish carbon capture pilot plant. The model accurately predicts the transient responses of CO2 concentrations and temperatures in both the absorber and desorber, where we investigate responses to step changes in various input variables (flue gas flow rate, liquid flow rate, and temperatures). Specifically, the hard-to-quantify interfacial CO2 capture rate is modeled as a linear function of the concentrations and temperatures. The model uses stochastic differential equations for the physical states and maximum likelihood estimation for the parameters. Several models are explored and the best is chosen. We demonstrate that the best model is able to predict dynamic behavior of the pilot plant with high accuracy in most operating scenarios, providing foundation for model predictive control.

A novel data-driven stochastic state space system for modeling and forecasting the sludge blanket height in secondary clarifiers is presented. The model is trained on sensor measurements of the sludge blanket height and uses as inputs (1) the clarifier sludge mass inflow rate, and (2) the clarifier recycle flow rate. The model's prediction accuracy is evaluated on data from two Danish wastewater treatment plants, for a summer and a winter month, by means of root-mean-square errors and compared with a persistence model. The model consistently outperforms the persistence model in the summer, but only one plant performs well in the winter month. The worst performing plant is challenging to model due to data quality issues and problematic (uneven and time-varying) flow distributions to the clarifiers. This led us to conclude that the best performance and stability is seen to require high data quality and well-controlled flow distribution. In summary the model achieves, in almost all cases, prediction error reductions in the order of 30–50% and 0.1–0.4 m in relative and absolute terms when compared with the predictions from a persistence model.

Background Metabolomics data is often complex due to the high number of metabolites, chemical diversity, and dependence on sample preparation. This makes it challenging to detect significant differences between factor levels and to obtain accurate and reliable data. To address these challenges, the use of Design of Experiments (DoE) techniques in the setup of metabolomic experiments is crucial. DoE techniques can be used to optimize the experimental design space, ensuring that the maximum amount of information is obtained from a limited sample space. Aim of review This review aims at providing a baseline workflow for applying DoE when generating metabolomics data. Key scientific concepts of review The review provides insights into the theory of DoE. The review showcases the theory being put into practice by highlighting different examples DoE being applied in metabolomics throughout the literature, considering both targeted and untargeted metabolomic studies in which the data was acquired using both nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry techniques. In addition, the review presents DoE concepts not currently being applied in metabolomics, highlighting these as potential future prospects.