This study concerns the dynamic operation of methanol steam reforming in fixed bed reactors for shipboard hydrogen production. The work is motivated by the need to reduce greenhouse gas emissions and harmful exhaust pollutants from the maritime sector, while also addressing the practical difficulty of storing hydrogen directly on board ships. Methanol is considered a promising alternative fuel and hydrogen carrier because it is liquid under ambient conditions, has relatively high energy density, is already handled at industrial scale, and can be used both in combustion-based transition technologies and in future fuel cell systems. In this context, methanol reforming offers a route to produce hydrogen on demand, avoiding some of the storage challenges associated with compressed or cryogenic hydrogen.

The dissertation focuses specifically on methanol steam reforming for polymer electrolyte membrane fuel cell systems. This reforming route is attractive because it provides a high hydrogen yield and comparatively low carbon monoxide selectivity. This is important because PEM fuel cells require pure hydrogen feed. The selected reactor type is a fixed randomly packed catalytic bed. This a configuration is widely used and commercially relevant for reforming applications.

The shipboard operation introduces requirements that are not prioritized in conventional land-based chemical plants. A ship power system must respond to changing loads, and therefore the reformer must be understood in transient operation, in addition to steady state operation. The central question of this dissertation is how fixed-bed methanol reformers behave dynamically, what limits their response time, and what modelling or experimental tools are needed for their reliable analysis.

The second chapter examines the broader problem of dynamic operation in fixed bed reactors. It shows that dynamic behaviour can arise from several motivations, including process flexibility, periodic operation, and on-demand production. These different cases require different performance metrics, modelling approaches, and measurement strategies. The dissertation points how dynamic reactor analysis cannot simply reuse steady state design priorities. Moreover, the objective of the transient analysis must first be clearly defined in each case. This chapter also reviews available numerical modelling approaches and analytical techniques for measuring changing gas compositions. A key point is that models of higher fidelity than 1D are often necessary for dynamic analysis, but not always readily available or computationally convenient. Likewise, transient experiments require instruments with suitable sampling rates, and often a combination of analytical techniques is more likely than using a single method.

To demonstrate the importance of model fidelity, this study develops and applies a dynamic 2D fixed bed reactor model for methanol steam reforming. A comparison with a 1D plug flow model shows that models may agree reasonably well at steady state but differ during transients, especially when predicting local temperature fields and species selectivity. This is important because transient CO formation and hydrogen production are directly relevant to operation and control in fuel cell systems. The 2D model therefore becomes the central tool for the later chapters, which focus on origins of reformer response time.

The second major part of the dissertation, starting with chapter 3, investigates the internal heat transfer dynamics of the fixed bed reactor. Methanol steam reforming is endothermic, so reactor performance depends strongly on heat transport from the reactor wall into the catalyst bed. For this work, the model only deals with the internal packed bed behaviour by imposing changes directly in wall temperature and reactant throughput. Thus the limitations of the external heating system influence is excluded. This allows for an estimation of a theoretical lower limit for the reactor transient time. The results show that, for typical tube diameters of 2-3 cm, internal heat and mass transfer lead to transient times of up to about one minute. This suggests that internal radial heat transfer alone is unlikely to explain the much slower response times often associated with practical methanol reformer systems. The packed bed may only amplify slow external heat supply changes, but it is probably not the dominant source of long load-change durations.

Chapter 4 combines experiments and modelling to study external heat transfer effects and validate the dynamic reactor model. A laboratory fixed bed methanol steam reformer is used for gathering steady state and transient data, mainly consisting of internal bed temperatures and reformate carbon monoxide content. The model is adapted to include all detected experimental non-uniformities and heat transfer imperfections. This enables a complete comparison between simulated and measured reactor behaviour. The results show that the model can capture the main dynamic trends, but they also reveal a strong influence of practical heating details during transient operation. Important external factors include the heat capacity of the reactor and supporting structure, axial heat loss profiles, non-uniform axial heat supply, and instability in inlet reactant temperature. Among these, the heat capacity of the reactor assembly is especially influential because it stores and releases heat during load changes, thereby prolonging the transient response even in a relatively small and well-insulated laboratory reactor.

Overall, the conclusions of this study is that dynamic shipboard methanol reformer design requires a broader approach than conventional steady state reactor sizing. A comprehensive analysis includes definition of the cause and purpose of dynamic operation, selection of a model with sufficient spatial and temporal fidelity, and finally the use analytical instruments fast enough to validate transient predictions. From this study it is also concluded that the intrinsic internal dynamics of typical fixed bed reformer tubes are relatively fast compared with the load change times reported for practical systems. Therefore, future improvements in transient performance should focus strongly on the external heating arrangement, reactor thermal mass, heat integration strategy, and system level control. These findings are relevant not only for methanol reforming but also for other shipboard chemical reactors that may become part of future low emission maritime energy systems. ... Read more

Recent research in chemical plant operation shows increasing interest in dynamic process operation as part of designed operating strategy for reasons such as increased dependency on renewable energy, and process intensification. Conventional analyses of fixed bed reactors are developed for steady state optimization and may not be adequate for dynamic operation. In fact, the important metrics and targets in dynamic process design are not entirely clear. The first objective of this article is to provide a state-of-the-art survey categorize types of dynamic operation, and rank the available common modelling and analytical tools suitable for quantification of dynamic process variables. The article then examines a case study of 1D and 2D model differences in a methanol steam reforming reactor. The case study shows model prediction differences of up to 15% for conversion, and up to 50% for CO concentration at the outlet during extreme load changes. The study concludes that the complexity of analytical and numerical techniques for dynamic processes is notably higher compared to steady state analyses, but appropriate tools and procedures are currently lacking. ... Read more

The growing adoption of renewable power in the chemical processing industry shifts many processes into the dynamic operating regime. The common 1D reactor models used for process design are no longer necessarily adequate for representing transient operation. In this study, a 2D pseudo-homogeneous dynamic model of a fixed bed reactor is developed and validated using methanol steam reforming as focused case. The model is experimentally validated with combined measurements of reactor bed temperature and outlet carbon monoxide concentration in time. The study found that the 2D model matches well in values and trends with experimental data despite numerous parameter uncertainties and use of standard equations from the literature. The largest influence on model prediction was found in external practical heating non-uniformities, such as heat losses and added heat capacity of the reactor support structure. The non-uniformities can cause up to 20% change in CO concentration predicted at the outlet, and double the transient time in certain cases. The model captures well the reactor transient characteristics well, and is regarded fit for design of reactors in the unsteady operating regime. ... Read more

Hydrogen economy is spreading across the maritime sector in response to increasingly stringent regulations for shipping emissions. The challenging on-board hydrogen logistics are often mitigated with hydrogen carriers such as methanol. Research on methanol reforming to hydrogen for fuel cell feed is conducted mostly in steady state, overlooking dynamic reactor operation and its effects on the power production system. Forced reactor operations induce fluctuations of CO content in the reformate potentially harmful to the PEM fuel cell, and drops in methanol conversion causing inefficient operation. In present research, simulations with a physical 2D unsteady model of a packed bed methanol steam reforming reactor resulted in methanol conversion drop durations of up to a minute. Additionally, temporary increases of CO content up to 112% were observed. Throughput ramp ups most notably impact the conversion, while ramp downs negatively affect selectivity. The investigation on reactor geometry concludes that larger tube diameters increase transient time and CO spikes, while they decrease with reactor length. Amplified unsteady effects are also observed with larger changes in input process variables. The results imply that heat transfer rate to the reactor are most often the detrimental factor for transient effects and durations in practice. Following this work, inclusion of realistic heating methods is recommended, instead of uniform tube temperatures used in present simulations. Heating system characteristics are necessary for realistic evaluation of the methanol reformer constraint on fuel cell feed demand in fully integrated systems. ... Read more

A growing concern is associated to the greenhouse gases emissions of superyachts, consequently alternative fuels are introduced to the market. For the yachting decarbonisation, this work focuses on hydrotreated vegetable oil (HVO) and methanol. Nevertheless, the uncertain global availability of these fuels can undermine the operations of ocean-crossing superyachts. Thus, a multi-fuel system is installed allowing for fuels switchover and built-in flexibility. Moreover, non-dedicated tanks are installed for the alternative storage of HVO and methanol to make optimal use of the tanks’ capacity. However, the alternative storage of HVO and methanol causes mutual fuels’ contamination. The lack of standards and research on accepted fuels impurity makes full fuels’ separation relevant to be explored. In this work, to avoid degradation of dual-fuel engines or fuel cells, gravity-settling tanks and disc-bowl centrifuges were studied to separate HVO-methanol mixtures. Shake tests were conducted on HVO-methanol mixtures to quantify the separation time and relative concentrations to obtain complete gravity separation. The gravity tests revealed methanol traces in HVO for all the tested mixtures within the 1 hour-3 days observation time, due to the low-density difference between the fuels. This makes the use of gravity-settling tanks impractical onboard for quasi-instantaneous fuels supply to the converters. A mathematical model was developed for disc-bowl centrifuges to assess the separator performance and separation time. Furthermore, the centrifuge was sized by providing the separator working conditions for varying engine modes. Moreover, spin tests were conducted to validate the mathematical model. The model showed that full separation is achievable with a larger centrifuge compared to existing designs. The larger design is due to the low-density difference between the fuels. The maximum separation time ranges from 5-10 minutes. Nevertheless, all the tested mixtures with the spin tests failed at achieving a state of full separation due to the dilution of a certain residual volume in the continuous liquid. The discrepancy between the mathematical model and the spin test results can lie in the neglected diluted phase of the dispersed fuel in the continuous liquid in the mathematical model. However, the mathematical model is a good tool to simulate the dynamic behaviour of the dispersed droplets. Consequently, the onboard use of a centrifuge for separating HVO-methanol mixtures should be evaluated by quantifying the concentration of the fuels’ mixture entering the separator tailored per yacht. Furthermore, tests on dual-fuel engines or fuel cells are recommended to establish tolerable limits of fuel’s contamination. ... Read more