Sorbent-enhanced/membrane-assisted steam-methane reforming

Abstract

Thermodynamic equilibrium and kinetic reactor models are used to simulate a fluidized bed membrane reactor with in situ or ex situ hydrogen and/or CO₂ removal for production of pure hydrogen by steam methane reforming. In the equilibrium model, the membranes and CO₂ removal are located in separate vessels downstream of the reformer. As the recycle ratio increases, the overall performance approaches that where membranes are located inside the reactor. Whether located in situ or ex situ, hydrogen removal by membranes and CO₂ capture by sorbents both enhance hydrogen production. In the kinetic reactor model, a circulating fluidized bed membrane reformer is coupled with a catalyst/sorbent regenerator. Sorbent enhancement combined with membranes could provide very high hydrogen yields. In addition, since carbonation is exothermic, with its heat of reaction similar in magnitude to the endothermic heat of reaction of the net reforming reactions, sorbent enhancement can provide much of the heat needed in the reformer. The overall heat needed for the process would then be provided in a separate calciner, acting as a sorbent regenerator. While the technology is promising, several practical issues need to be examined.

Introduction

Hydrogen is one of the most important industrial chemicals. In addition, it is the subject of considerable attention as a major future energy carrier (e.g. Goltsov and Veziroglu, 2002, Ohi, 2002). Annual world production is currently $\sim 5\times {10}^{11}{\mathrm{Nm}}^3$ (Ewan and Allen, 2005), and demand is increasing rapidly. High-purity hydrogen (typically $<$10 ppm CO) is required for fuel cells (mainly proton exchange fuel cells) to produce electricity with an efficiency of 45–55% and with reduced acid gas emissions (Borroni-Bird, 1996, Brown, 2001).

There are a number of catalytic and noncatalytic approaches for the production of hydrogen in industry: e.g., steam reforming of hydrocarbons, partial oxidation of higher hydrocarbons (heavy oils and coal), and electrolysis of water. Steam methane reforming (SMR), developed in the 1930s, is the most important large-scale process for producing hydrogen (Twigg, 1989, Scholz, 1993). The principal reactions involved in catalytic SMR are

Steam reforming:

$${\text{CH}}_4+{\mathrm{H}}_2\mathrm{O}\iff \mathrm{CO}+3{\mathrm{H}}_2,\mathrm{\Delta }{H}_{298}^{\circ }=206.2\mathrm{kJ}/\mathrm{mol}\text{,}$$

$${\text{CH}}_4+2{\mathrm{H}}_2\mathrm{O}\iff {\mathrm{CO}}_2+4{\mathrm{H}}_2,\mathrm{\Delta }{H}_{298}^{\circ }=165.0\mathrm{kJ}/\mathrm{mol}\text{.}$$

Water–gas shift (WGS):

$$\mathrm{CO}+{\mathrm{H}}_2\mathrm{O}\iff {\mathrm{CO}}_2+{\mathrm{H}}_2,\mathrm{\Delta }{H}_{298}^{\circ }=-41.2\mathrm{kJ}/\mathrm{mol}\text{.}$$

The first two reactions are strongly endothermic, and both lead to significant increases in molar flow rates as the reaction proceeds. Equilibrium conversions of both reforming reactions therefore benefit from high temperatures and low pressures, whereas the water–gas shift (WGS) reaction (3), being exothermic and having no change in the number of moles, benefits thermodynamically from lower temperatures and is independent of pressure. The overall endothermic and reversible process is usually carried out in parallel vertical fixed bed catalyst tubes, suspended within huge high-temperature $(\text{typically}>850{}^{\circ }\mathrm{C})$ furnaces, coupled with pressure swing adsorption (PSA) for hydrogen purification.

Despite its industrial importance, this conventional SMR process suffers from a number of significant disadvantages:

• Thermodynamic equilibrium constraint: The reversibility of the reforming reactions and their rapidity under normal operation conditions constrain hydrogen production to the thermodynamic equilibrium values (Twigg, 1989; Grace et al., 2005).

• Internal diffusion resistance: To minimize pressure drops, fixed bed catalyst particles must be large, e.g., $16\times 6\times 16\mathrm{mm}$ Raschig rings (Elnashaie and Elshishini, 1993) or $\sim 6$–17 mm thick-walled rings (Twigg, 1989), causing the effectiveness factor of the nickel reforming catalyst to be extremely low, typically $\sim {10}^{-3}$–${10}^{-2}$ (Rostrup-Nielsen, 1977, De Deken et al., 1982; Soliman et al., 1988).

• Carbon formation and catalyst deactivation: Carbon formation can deactivate reforming catalysts leading to low efficiency, possibly even blocking the reformers and increasing the pressure drops through the beds (Bartholomew, 2001, Ren et al., 2002, Sehested, 2006).

• Heat transfer, temperature gradients and tube materials: To maintain high reactor temperatures $(\text{typically}>850{}^{\circ }\mathrm{C})$, heat must be transferred to and through the walls of the tubes. Only $\sim 50\%$ of the heat of combustion is used directly for the steam reforming reactions (Armor, 1999). Extensive efforts have been spent on waste heat recovery by exporting steam and preheating the feeds (Tindall and King, 1994). In addition, high temperature gradients require expensive metal alloys for the tubes.

• Environmental emissions: The furnaces emit ${\mathrm{NO}}_X$ during the burning of fuels in the furnaces (IPCC, 2001). In addition, CO₂ is produced both in the furnace and due to the reforming process, limiting the environmental advantages of hydrogen as a clean fuel.

A number of alternative reforming configurations have been investigated in efforts to improve the process, e.g., fixed beds with permselective membranes (Dyer and Chen, 2000, Sammels et al., 2000, Shah and Drnevich, 2000), microchannel reformers (Makel, 1999), catalytic oxidative steam reforming (Theron et al., 1997), bubbling fluidized bed membrane reformers (FBMR) (Elnashaie and Adris, 1989, Adris et al., 1991; Adris et al., 1994a, Adris et al., 1994b, Adris et al., 1996, Adris et al., 2002) and circulating fluidized bed membrane reformers (CFBMR) with continuous catalyst regeneration (Chen et al., 2002; Chen and Elnashaie, 2005a, Chen and Elnashaie, 2005b; Prasad and Elnashaie, 2004). Oxygen has been successfully introduced into fluidized bed membrane reactors (Roy et al., 1999, Roy et al., 2001, Roy et al., 2002, Chen et al., 2003; Rakib and Alhumaizi, 2005) to generate the heat needed for the endothermic reforming reactions, without serious penalty in terms of hydrogen production. Recent pilot plant tests with two commercial catalysts, SMR and ATR (autothermal reforming) catalysts (Chen et al., 2007, Mahecha-Botero et al., 2008), showed improved overall membrane performance and hydrogen permeation fluxes compared with previous tests. A major review has recently been published on membrane-assisted fluidized bed reactors (Deshmukh et al., 2007).

The fluidized bed membrane reactor process, represented schematically in Fig. 1, offers several potential benefits:

• By extracting hydrogen in situ, the equilibria of reversible reactions (1)–(3) are shifted forward, improving methane conversion and hydrogen yield.

• High-purity hydrogen, free of CO, can be produced by the membranes, so that a separate downstream purifier is no longer needed.

• Lower temperatures can be used, reducing heat losses and permitting less expensive low-temperature metal alloys, e.g. $-316$ stainless steel, to be employed for the vessel walls.

• In situ withdrawal of hydrogen allows the adverse effect of pressure on reactions (1) and (2) to be largely neutralized, in accordance with the well-known LeChatelier's principle.

• The fluidized bed facilitates greatly improved heat transfer relative to fixed beds. With membranes present, there are similar improvements in bed-to-surface mass transfer.

• Fluidized beds require greatly reduced pressure drops than fixed beds.

• The much smaller catalyst particles $(\sim 100\mathrm{\mu }\mathrm{m})$ in the fluidized bed eliminate internal diffusion resistances, leading to catalyst effectiveness factors close to unity.

• Combining the reformer, WGS shift converter and purifier in a single vessel results in process intensification, providing potential for reduction of capital cost.

• Adding air or oxygen to the reformer reduces the need for external heating, minimizes emissions of CO₂ and NO${}_x$, and reduces coking of the catalyst.

• Fluidization provides the possibility of replacing deactivated catalyst in a continuous or periodic manner, without having to shut down the reactor.

An alternative or complementary way of shifting the equilibrium forward is by injecting a CO₂ sorbent, e.g. calcined limestone, into a reformer. The resulting carbonation reaction can also supply heat to the reformer and allow nearly pure CO₂ to be released in the sorbent regenerator (calciner) as a key component in greenhouse gas sequestration. Recently Johnsen et al., 2006a, Johnsen et al., 2006b studied sorption-enhanced SMR in a bubbling fluidized bed reactor with dolomite as CO₂-acceptor. The outlet H₂-concentration exceeded 98% (dry basis) at 600 °C and 1 atm, whereas after the sorbent's carbonation capacity had been exhausted, the hydrogen concentration dropped to $\sim 73\%$.

In this paper, an ASPEN equilibrium model and a kinetic reactor model are used to simulate sorbent-enhanced and/or membrane-assisted steam methane reforming for pure hydrogen production. The objective is to explore the possible combination of sorbent-enhanced reforming with the membrane reactor to see whether membranes and calcium sorbents, located in situ or ex situ, could provide synergistic benefits to improve methane conversions, hydrogen yields and provision of heat.