Hochschule Niederrhein

Côte d’Ivoire faces rapidly increasing electricity demand and persistent access inequalities while pursuing low-carbon development objectives. This paper presents a spatially explicit, scenario-based electricity system analysis using PyPSA-Earth, an open-source optimisation framework. The national power system is modelled with 14 spatial clusters and hourly resolution to assess long-term generation expansion, dispatch, and system costs for 2030, 2040, and 2050 under three internally consistent scenarios: Baseline (current policies), Policy (progressively tightening electricity-sector emission limits), and Renewables (near-zero emissions). Electricity demand increases from 12.27 TWh in 2030 to 26.74 TWh in 2050 across all scenarios. Solar photovoltaics emerge as the dominant expansion technology, while existing hydropower provides stable system flexibility. In the Baseline pathway, electricity-sector CO₂ emissions rise from 1.10 Mt in 2030 to 2.66 Mt in 2050. The Policy scenario limits CO₂ emissions to 1.05 Mt by 2050, whereas the Renewables scenario reaches zero annual emissions within the model boundary. Annualised system costs increase with demand, reaching 1,380 M EUR (Baseline), 1,291 M EUR (Policy), and 1,493 M EUR (Renewables) in 2050. The analysis demonstrates a transparent and reproducible framework for long-term electricity planning in data-constrained contexts.

Computer-aided modeling and mathematical optimization of energy systems are essential for improving operational efficiency and achieving emission reductions, particularly for steam turbine systems with part-load-dependent efficiency characteristics. Mixed-Integer Linear Programming (MILP) is the state of the art, due to its short computational times and reliable convergence. However, its simplifications often reduce model accuracy. Mixed-Integer Nonlinear Programming (MINLP) offers high accuracy but faces long computational times and potential convergence issues. Recent advancements in Mixed-Integer Quadratically Constrained Programming (MIQCP) offer a promising approach for more accurate energy system modeling by enabling quadratic and bilinear representations while avoiding the full complexity of nonlinear programs. This study compares the optimization methods MILP, MINLP and MIQCP for the operational optimization of a steam turbine system. The parameterization of the models is based on hourly measurement data of two real-world steam turbines. Key evaluation criteria include accuracy, computational time, implementation complexity and the deviation in the calculated optimum. The results show that MIQCP improves accuracy compared with MILP while requiring lower computational time than MINLP. Overall, the results demonstrate that MIQCP provides a suitable compromise between model accuracy and computational efficiency for the operational optimization of steam turbine systems.

<title>Abstract</title> Among other fluoroquinolones, difloxacin, norfloxacin, ofloxacin, and pefloxacin are regularly detected in various bodies of water around the world. Advanced oxidation processes such as the electrochemical oxidation are employed to eliminate these micropollutants. Using a microflow cell and a synthesis cell, the four fluoroquinolones were submitted to electrochemical oxidation through a boron-doped electrode. They were subsequently analyzed using high-performance liquid chromatography coupled with high-resolution mass spectrometry. Electrochemical oxidation occurred above 800 mV. Products were formed according to common mechanisms: hydrogen and ethylene elimination, methylene/methyl cleavage, hydroxylation in combination with methyl cleavage. To facilitate future identification of electrochemical oxidation products of the fluoroquinolone class, fragmentation patterns in higher-order mass spectrometry were investigated deriving typical motives. A total of 17 products were identified, 13 of which were novel. Quantitative structure-activity relationship analysis suggested that the electrochemical oxidation products would be less ecotoxic than the initial drugs.

Abstract The DICER1 gene is mutated in cancer, including DICER1 syndrome, a rare tumour predisposition syndrome. Cancer-associated hotspot mutations have been reported in both catalytic domains of DICER1 and are predicted to disrupt miRNA biogenesis. To understand how these hotspot mutations contribute to cancer development, we have generated cell lines harbouring single amino acid substitutions within the catalytic RNase IIIa (S1344L) or RNase IIIb (D1709N) domains of the endogenous DICER1 gene. We show that both mutations result in a widespread loss of 5p miRNAs, and, unexpectedly, an increase in 3p passenger strands loading into AGO2. The shared similarities between both mutants can be attributed to the structural proximity of the S1344 residue to the RNase IIIb catalytic centre. Functionally, we found that changes in the repertoire of miRNAs loaded into AGO2 result in altered gene expression, impacting critical pathways for cancer development, including metastatic potential. Additionally, our results indicate that inactivating the processing activity of DICER1 does not result in genomic instability. Instead, mutations cause specific upregulation of human endogenous retrovirus H (HERVH) through miRNA-independent mechanisms, suggesting that both canonical and non-canonical DICER1 functions are important to understand DICER1 syndrome.

Thanks to tremendous progress in generative AI, a wide range of applications has become possible. In this paper, we present a web application that uses artificial intelligence to generate book covers. This application demonstrates the potential of generative AI in the publishing industry. Self-publishers, who generally cannot afford a commercial illustrator, can use the application to enhance their books for better visibility and sales. For this project, we examined the DALL-E, Stable Diffusion, and MidJourney AI models, selecting the DALL-E model to generate the book covers. We discuss the challenges of prompt generation and how we optimized it to produce quality results. The application generates several background images based on user input, such as genre selection and a short description. These images can be customized further by using a visual editor and downloaded as book covers. We discovered that short descriptions resulted in images with a better focus on one key visual element. Our research revealed that DALL-E’s "painted" style is more suitable for book covers than Stable Diffusion’s photorealism.

In model predictive control (MPC), data-driven models are increasingly used to predict system dynamics, particularly for complex systems. Gaussian processes (GPs) are a popular choice as prediction model due to their ability to model nonlinear systems and quantify predictive uncertainty. However, the performance of GP-based MPC strongly depends on the model structure and the quality of the training data. The computational burden of GPs scales poorly with the number of training data points. Therefore, satisfactory model and control performance is desirable with minimal data to achieve faster control using GP-based MPC. This work investigates the impact of training data and GP model structure on the closed-loop control performance for varying numbers of training data. Key aspects of the training data are varied, including dataset generation, scaling, and selection strategies for dataset reduction. This study focuses on investigating different GP model structures, comparing full system state versus state change predictions, and zero versus non-zero prior mean functions. All methods are evaluated in simulation on a crane pendulum and an inverted pendulum system. Control performance is assessed using the MPC objective function. The results highlight the importance of data preprocessing and show that using state-change predictions or non-zero prior mean functions significantly improve closed-loop control performance.

The m-out-of-n bootstrap is a possible workaround to compute confidence intervals for bootstrap inconsistent estimators, because it works under weaker conditions than the n-out-of-n bootstrap. It has the disadvantage, however, that it requires knowledge of an appropriate scaling factor tau(n) and that the coverage probability for finite n depends on the choice of m. This article presents an R package moonboot which implements the computation of m-out-of-n bootstrap confidence intervals and provides functions for estimating the parameters tau(n) and m. By means of Monte Carlo simulations, we evaluate the different methods and compare them for different estimators.

A green image is becoming increasingly important for companies in the textile industry and recycling textiles is a problem because fibers are difficult to identify and therefore difficult to recycle. Especially the cross-linking of DNA and textile materials can achieve a significant benefit within the textile industry. In addition to DNA markers, alternatives like cryptographic marking, RFID technology, and other innovative approaches will be discussed. Markers and signature methods on textile materials are important for efficient recycling processes, but this topic has not yet been sufficiently researched. This review gives an overview of ongoing activities on signature methods and recycling approaches for textiles and focuses on the recycling of garments to address the challenges and solutions within the textile- and fashion industry.