Reformer tube simulation

B. CFD Simulation of Reformer Tube Heat Transfer with... 

This study was carried out to simulate the 3D temperature field in and around the large steam reforming catalyst particles at the wall of a reformer tube, under various conditions (Dixon et al., 2003). We wanted to use this study with spherical catalyst particles to find an approach to incorporate thermal effects into the pellets, within reasonable constraints of computational effort and realism. This was our first look at the problem of bringing together CFD and heterogeneously catalyzed reactions. To have included species transport in the particles would have required a 3D diffusion-reaction model for each particle to be included in the flow simulation. The computational burden of this approach would have been very large. For the purposes of this first study, therefore, species transport was not incorporated in the model, and diffusion and mass transfer limitations were not directly represented. 

The reformer tube operation was simulated on the basis of a set of continuity-, energy- and momentum equations using one and two dimensional heterogeneous models. Intraparticle gradients in the rings were accounted for by the use of the generalized modulus concept. 

Finally, the question rises whether an accurate simulation of the reformer tubes does not have to include consideration of radial gradients. The two dimensional model developed for this purpose in this work neglects interfacial gradients, for reasons explained already above, but maintains the mass transfer limitations inside the catalyst, of course. 

It is of interest and practical importance to show the situation where different feeds are introduced to the model with different steam partial pressures while the feed partial pressures of the other components are kept constant, which means of course, a change of the total pressure. The steam reformer tube chosen for simulation is 5 m long to illustrate the kinetic effects rather than the thermodynamic equilibrium effect since the assumption of constant temperature along the tube causes a fast approach to thermodynamic equilibrium of the mixture. 

The catalyst (0.15 g) was loaded into a quartz tube reactor (internal diameter = 4 mm). The catalyst was pretreated in nitrogen at 400°C. Simulated gasoline reformate was used for the activity test of the catalyst. The composition of the simulated reformate was 36 wt% H2, 17 wt% CO2, 28 wt% N2, 17 wt% H2O, 1 wt% CO, and air was added additionally as the oxidant. The total flow rate was maintained at 100 ml/min. The test was performed over the temperature range of 120 280°C at various flow rates of inlet air. 

The overall effect of catalyst pellet geometry on heat transfer and reformer performance is shown in the simulation results presented in Table 1. The performance of the traditional Raschig ring (now infrequently used) and a modern 4-hole geometry is compared. The benefits of improved catalyst design in terms of tube wall temperature, methane conversion and pressure drop are self-evident. 

A tube reactor with dimensions given in Table 11.9 was simulated. A high ratio between C02-acceptor and catalyst was used because the reforming kinetics are fast compared to the sorption rate. The reactor was filled with steam (97 mole%) and a small amount of hydrogen at the desired temperature, 848 K, at startup. The input to the reactor was methane and steam, in which the steam to methane ratio is set to 6. A high steam to carbon ratio is necesarry... 

The nuclear heat source was again simulated by an electric heater. The new EVA-II reformer constructed by the Steinmuller company has used the concentric tube design. It... 

The simulations can be used to calculate the approach to chemical equilibrium for the reaction in the tube. Figure 3.18 shows the measured temperature approach to equilibrium at the very bottom of the reformer as a function of the calculated local heat flux. 

Simulation of tubular steam reformers and a comparison with industrial data are shown in many references, such as [250], In most cases the simulations are based on measured outer tube-wall temperatures. In [181] a basic furnace model is used, whereas in [525] a radiation model similar to the one in Section 3.3.6 is used. In both cases catalyst effectiveness factor profiles are shown. Similar simulations using the combined two-dimensional fixed-bed reactor, and the furnace and catalyst particle models described in the previous chapters are shown below using the operating conditions and geometry for the simple steam reforming furnace in the hydrogen plant. Examples 1.3, 2.1 and 3.2. Similar to [181] and [525], the intrinsic kinetic expressions used are the Xu and Froment expressions [525] from Section 3.5.2, but with the parameters from [541]. 

Marigliano et al. compared the performance of two different types of tubular methane steam reforming membrane reactors by numerical simulations [413]. In the first, the catalyst was packed into the palladium/silver membrane tube, and in the second it was positioned in the annular region surrounding the membrane tube. Both configurations were heated from the outside. Owing to the indirect heat transfer to the catalyst bed inside the membrane tube, methane conversion was lower in this instance, and under certain conditions it was even inferior to a conventional fixed bed... 

Figure 6.2.32 Evolution of methane conversion ( Ht). mean radial process gas temperature and external and internal tube skin temperatures (Tprocessi and Tint), and total pressure in a tube of a steam reformer simulations by a one-dimensional reactor model (Section 4.10.7.3), internal/external tube diameter 10.2/13.2cm, heated tube length 11.1m, ring-shaped catalyst (height 1 cm, diameters 0.8 and 1.7cm), molar steam to methane ratio 3.4, average flue gas temperature 1100°C [data from Xu and Froment (1989a, b) Plehiers and Froment (1989)].

A very similar combined process scheme integrating OCM and steam reforming of methane (SRM) was suggested at the same time by Tiemersma et al [44]. OCM is performed in membrane reactor for distributed O2 feeding. The membrane reactor tubes are immersed into a fluidized-bed reforming reactor for optimal heat transfer. Simulation of such process based on kinetic data obtained... 

Both the naphtha and butane feed cases show the methane profile rising from very low inlet values to a maximum, and falling to the outlet composition. The hydrocarbon species compositions fell quickly, and are essentially zero at about 4 to 6 meters from the tube inlet. These profiles are in close agreement with the profiles shown in References 4 and 13. For more active catalyst, the hydrocarbon species disappear closer to the tube inlet. The simulated temperature profiles are also in good agreement with profiles in those references, but more importantly, they agree precisely with the measured profiles. Reference 23 shows that the temperature profile in a top fired reformer is significantly different than in a wall-fired furnace. 

A fired tube reactor was configured to match the dimensions and catalyst loading of an existing oxo-alcohol synthesis gas steam reformer. Simulation results at observed conditions (feed gas composition, outlet temperature, steam to earbon ratio, ete.) agree very well with observed results. Catalyst activity is first determined by matching key effluent eomposition. 

In the operating reformer, sulphur poisoning is a dynamic phenomenon, and the pore diflusion restrictions in the catalyst particles have a complex influence on the gradually developing sulphur profiles through the tube. This can be simulate in a computer model (Christiansen et al, 1980) giving axial profiles in the tube as well as radial gradients in the pellets as function of time. 

---