PAPER making! FROM THE PUBLISHERS OF PAPER TECHNOLOGY INTERNATIONAL ® Volume 12, Number 2, 2026
screening process. Through systematic testing, we characterize the relationship between 3D-printed mesh geometry and pulp composition, identifying three primary mesh-fiber interactions. Using a Research through Design approach, we demonstrate the versatility of Pulpform through a range of artifacts inspired by real-world scenarios — including masks for performing arts, air filters, and sensing origami. We then discuss how the role of 3D printing shifts from a tool to an integral material component within this process, reflecting on how Pulpform composites lead to “paperness” and “frugality” as two emergent and desirable materialities. ENERGY “ A Three-Stage Reinforcement Learning Pipeline for Autonomous Process Control in the Papermaking Industry ” , Minu Baek & Sangkeum Lee, IEEE Transactions on Industrial Informatics , published online 22 April 2026. The optimization of complex continuous manufacturing processes, such as papermaking, involves multiobjective tradeoffs among production throughput, product quality, and energy-related operating variables under nonstationary and noisy industrial data. This article presents a digital twin- based reinforcement learning (RL) framework for studying simulated control policies in papermaking processes without direct plant interaction. A hybrid digital twin, combining physics-based relations and a data-driven moisture prediction model based on XGBoost, is employed to enable safe and sample-efficient policy learning. To address the time-varying operational priorities within a production lot, a phase-aware reward formulation with dynamic weighting and Gaussian quality terms is introduced to provide stable learning signals across different production stages. Policy training follows a three-stage pipeline consisting of model-based expert data generation, behavioral cloning for policy initialization, and proximal policy optimization for on-policy refinement within the simulated environment. Simulation- based evaluations indicate that the learned policy achieves higher cumulative reward and increased production throughput while maintaining quality-related variables within historical operating ranges, along with a reduction in an energy-related pressure proxy. Additional analyses examine robustness with respect to action noise, reward-weight perturbations, and out-of-distribution states. The proposed framework provides a systematic simulation study illustrating the application of phase-aware RL to energy-intensive continuous manufacturing processes. ENVIRONMENT “ Evaluating adoption strategies for green innovations to decarbonize the United States pulp and paper industry ” , Jingwei Chen & Mingzhou Jin, Communications Earth & Environment, Vol.7, Article number: 489 (2026). The pulp and paper industry is a major contributor to greenhouse gas emissions and energy consumption, underscoring the urgent need for effective decarbonization strategies. Here, we estimate the current lifecycle emissions intensity and energy cost of the United States industry at 3.12 short tons of equivalent carbon dioxide and 289.79 United States dollars per short ton of finished product, respectively. We provide projections through 2050 by combining lifecycle assessment, scenario-based modeling, and learning curve model. The analysis shows the importance of improving waste management, boosting recycling rates, and reducing purchased energy. Achieving net zero was demonstrated to be challenging, as the maximum achievable reduction in lifecycle emissions was only 15% across all simulated scenarios. Simulations reveal that a strategy centered on carbon capture, utilization, and storage offers long-term sustainability and superior resilience to systemic uncertainties compared to broad-spectrum aggressive adoption. These findings emphasize rapidly adopting green innovations.
Page 3 of 13
Technical Abstracts
Made with FlippingBook Online newsletter maker