(702c) A Robust Data-Driven Reduced-Order Model for Real-Time Optimization of Steam-Methane Reformers Under Distributed Sensing

Large quantities of hydrogen are consumed in refineries and for production of important chemicals such as ammonia and methanol. Declining crude-oil quality and increased fertilizer demands, among others, have led to further increase in hydrogen demand. A significant portion (~80 %) of industrial hydrogen consumption is met via natural-gas steam methane reforming. This process takes place in a large scale, high-temperature, and highly energy-intensive unit called a steam methane reformer (SMR), where endothermic reforming reactions are carried out in hundreds of catalyst-filled tubes placed in a gas-fired furnace. A typical modern hydrogen production plant consumes a substantial amount (10⁵ GJ) of natural gas per day.  The overall productivity (energy consumed per unit H₂ produced) of the plant is strongly dependent on how efficiently the SMR is operated, which further depends on the spatial temperature distribution inside the furnace, where a more uniform distribution paves the way for reduced plant-wide energy use.

Rigorous SMR models, such as computational fluid dynamics (CFD) models, can provide accurate temperature distribution information. However, such models are not suitable for real-time, on-line use. In our previous work(Kumar et al., 2015), we demonstrated the feasibility of using data-driven empirical models to optimize fuel distribution to the furnace burners, and achieve a uniform furnace temperature profile. A reduced-order linear model was used to modulate the fuel distribution among the burners in a representative smaller-scale SMR system. It was shown that a reduced-order empirical model with much lower computational requirements, when developed using sufficiently rich data, can be a viable substitute to the detailed modeling of the complex thermal and flow interactions in the furnace.

In this work, the reduced-order modeling approach is extended to the full-scale SMR system. Specifically, we propose a novel data-driven reduced-order model, termed as egg-crate SMR (EC-SMR) model, for the SMR furnace. The EC-SMR model predicts the variations in the furnace temperature profile upon adjustments in the openings of the fuel-stream valves. The resultant redistribution of fuel among the burners upon valve adjustments is explicitly considered. The localized impact due to change in fuel-throughput of an individual burner is modeled as exponentially decaying function and the overall perturbation in the temperature distribution is approximated as a combined effect of local impacts from all the burners. We show that the model is robust to noise in data and overcomes several other shortcomings of previously published models. In particular, the fuel-redistribution based formulation leads to spatially ‘smooth’ predictions, unlike the PLS model predictions, of temperature variations, and allows for easy incorporation of nonlinear interactions, if desired. Being computationally parsimonious, the model is amenable for online optimization of the furnace temperature distribution. Comparisons of model predictions with actual plant measurements (obtained from an array of infrared cameras) show the excellent performance of the EC-SMR model in capturing the redistribution effects, which is crucial for elimination of temperature variability in the furnace at the level of valve manipulations. Subsequently, a case study is presented, whereby the EC-SMR model is employed in an optimization framework with the goal of minimizing temperature variability within the furnace.

References:

Kumar, A., Baldea, M., Edgar, T. F., & Ezekoye, O. A. (2015). Smart Manufacturing Approach for Efficient Operation of Industrial Steam-Methane Reformers. Industrial & Engineering Chemistry Research, 54 (16), 4360-4370.