The primary objective of this paper is to present the findings of a survey conducted among shipping companies in Norway during spring 2025. In the survey, the status and plans of the fleet concerning the green transition are monitored. Shipowners’ perceptions of current and future real fuel alternatives are mapped and discussed in relation to the authorities’ emission targets and incentive schemes.

Rising geopolitical instability, increasing labor costs, and sustainability demands are exposing the fragility of globalized manufacturing systems and supply chains. Automotive wire harness production, currently outsourced to best-cost regions due to its manual complexity, exemplifies this vulnerability, as it depends on long, rigid and carbon-intensive supply chains. Nearly all wire harnesses are unique, customer-specific systems, tailored to each vehicle configuration, making stocking or substitution difficult and thus increasing supply chain vulnerability. These conditions necessitate a shift towards more resilient and localized manufacturing models, which can only be achieved by advanced automation. Therefore this paper introduces a system architecture pipeline for the fully automated robotic assembly of automotive wire harnesses. Implemented within a robotic cell, the system features a multifunctional end-effector for automated crimp insertion, wire routing and bundling, embedded in a modular process sequence. A series of experiments was conducted across multiple wire types and connector families, demonstrating high process stability, successful end-to-end harness assembly, and robust insertion performance of 88,9\% accuracy. The results substantiate the feasibility of substituting manual harness production with scalable automation. The proposed approach offers a viable pathway towards localized manufacturing and enhanced supply chain resilience in the automotive sector.

Industrial facilities operating under indexed electricity pricing are increasingly exposed to intra-day price volatility, creating opportunities for cost reduction through coordinated use of demand-side flexibility. This paper proposes a structured day-ahead decision framework that schedules flexible electrical loads based on price signals and forecasts of baseline consumption and on-site photovoltaic generation. The approach formulates the scheduling problem as a constrained optimization model and solves it using evolutionary algorithms, requiring only load measurement and no additional physical assets such as storage systems. The framework is evaluated using real industrial consumption data, simulated photovoltaic production, and wholesale market prices under both time-of-use and indexed tariff structures. Results show measurable cost reductions across scenarios, with higher savings observed under indexed pricing due to greater price dispersion. A sensitivity analysis of photovoltaic forecast deviations indicates that the optimized schedules retain positive economic benefit within realistic uncertainty ranges. The findings demonstrate that structured scheduling of existing demand flexibility provides a practical mechanism for improving industrial energy cost control in increasingly dynamic electricity markets.

Accurate day-ahead residential energy load forecasting at individual household level is essential for utility operations including Demand Response programs, dynamic pricing, and distribution network management. Existing Mixture of Experts (MoE) frameworks for energy forecasting exclusively employ Deep Learning architecture despite their computational overhead and data requirements. This study demonstrates that gradient boosting models (XGBoost and LightGBM) can serve effectively as MoE expert components. A hybrid two-level architecture combining K-Means clustering for daily regime identification is validated on seven residential households using only historical energy consumption data. Results show that MoE with XGBoost achieves MAE improvements up to 6.4% in comparison to standalone forecasting models, with particularly substantial benefits for households with volatile, occupancy driven consumption profiles. Ex-ante analysis reveals that MoE framework could significantly increase forecasting accuracy for households with electricity independent heating systems (i.e. natural gas), while electrified heating systems (i.e. heat pumps) showed unremarkable improvements. Feature importance analysis further confirms that historical consumption lags and temperature dominate prediction drivers. These findings demonstrate the practical viability of MoE frameworks using gradient boosting components for higher accuracy in day- ahead forecasting needs using only anonymized energy timeseries.