DEVELOPMENT OF AN INTEGRATED CATALYTIC METHANOL REFORMER

ETSU F/02/00060/REP

Contractor C J B Developments Ltd

Prepared by R A J Dams S C Moore P Hayter M Verhaak

The work described in this report was carried out under contract as part of the New and Renewable Energy Programme, managed bythe Energy Technology Support Unit (ETSU) on behalf of the Department of Trade and Industry. The views and judgements expressed in this report are those of the contractor and do not necessarily reflect those of ETSU or the Department of Trade and Industry.

First published 1996 CONTENTS

1. EXECUTIVE SUMMARY 1

2. BACKGROUND 2

3. OBJECTIVES 4

4. CATALYST DEVELOPMENT 4

5. BENCH-SCALE REFORMER CONSTRUCTION AND OPERATION 6

6. EVALUATION OF RESULTS 10

7. CONCLUSIONS 10

8. RECOMMENDATIONS 11

Figure 1 Figure 2 Table 1 1. EXECUTIVE SUMMARY

As part of a successful collaboration with Vickers Shipbuilding and Engineering Limited (VSEL), CJB Developments Limited (CJBD) developed a methanol reformer to supply -rich gas for a Solid Polymer (SPFC) power system with an output of 10kW.

The methanol reformer used a packed bed pelleted proprietary catalyst. The endothermic heat of reaction was supplied by combusting the fuel cell off-gases. Although the reformer performed successfully in terms of the objectives of the breadboard power system, it was clear that for an on-board fuel cell vehicle application the start-up time and response to transient changes in load was not adequate.

Consequently, CJBD investigated different forms of construction of reformer and as a result formed a collaboration with ECN of the Netherlands and VSEL to develop, in the longer term, an all metal methanol reformer using metal substrates coated with both reforming and combustion catalysts. ECN’s Applied department had previous experience in this area. This report describes the development and evaluation of a bench-scale methanol reformer system, incorporating novel catalysts.

In this project ECN developed methanol reforming catalysts and ways of applying them to metal substrates. These catalysts and coating techniques were used to produce a series of aluminium tubes which CJBD evaluated on a specifically designed test rig.

After initial problems with their coating techniques, ECN produced a set of tubes which demonstrated a performance similar to that achieved by CJBD with pelleted catalyst. As these tubes were not optimised, there is scope for further improvement.

In 1995, CJBD and ECN, as part of an expanded team, applied successfully for funding from the JOULE and DTI Advanced Fuel Cell Programmes to develop a compact methanol reformer and gas clean-up system for a mobile SPFC vehicle. The work undertaken during this project was a significant factor in the success of the joint application.

1 2. BACKGROUND

In 1990, CJB Developments Limited (CJBD) and Vickers Shipbuilding and Engineering (VSEL) developed, built and tested a 10kW breadboard Solid Polymer Fuel Cell (SPFC) power system using methanol as a source of hydrogen-rich fuel. This was a generic demonstrator which showed that such a system could be assembled and operated successfully.

CJBD’s contribution to this programme was the development of the methanol reformer and gas clean-up system. Methanol was chosen as the fuel source because of its high energy density and because it can easily be reformed to hydrogen at low temperatures. Furthermore, methanol can be manufactured from coal, natural gas and other feedstocks including bio-mass. It also has some advantages over other fuels in term of fuel storage and distribution. The method of storing methanol on a vehicle is simple and comparable to conventional vehicles. In addition, a fuel cell vehicle using methanol would be able to attain a similar range as an internal combustion engine vehicle and drivers are comfortable with a liquid fuel.

Steam reforming was chosen because methanol is easily processed in this manner to form a hydrogen-rich mixture. The typical hydrogen content of the reformed gas is 75% with the balance being 24.8% and 0.2% carbon monoxide. The process is a catalytic endothermic reaction which occurs at about 225°C. The endothermic heat is provided by burning the fuel cell off-gases.

At the start of this work, other technologies such as , autothermal reforming and pyrolysis were less well developed and had a number of potential system problems. Partial oxidation requires air introduction (a potential parasitic loss), produces a reformed gas diluted with nitrogen (typical hydrogen content is 45%) and contains higher levels of carbon monoxide (requires a shift reactor prior to final carbon monoxide removal). Additionally, partial oxidation is an exothermic reaction at a higher temperature than and hence produces heat which has to be used within the complete fuel cell system.

Autothermal reforming, which is a combination of partial oxidation and steam reforming resulting in a thermally neutral situation, was in the early stage of development and operates at temperature of about 500°C using natural gas. Pyrolysis

2 was considered to require too high an operating temperature (800°C /1000°C). Both autothermal reforming and pyrolysis would appear to have more difficult start-up problems because of high operating temperatures.

The methanol reformer developed by CJBD was a packed bed device using a commercially available steam reforming catalyst. The reformer performed satisfactorily for the purposes of the breadboard demonstration. However, it was clear that the speed of response to transients and time to start from cold would not meet the requirements for a reformer to be used on-board an SPFC driven vehicle.

To overcome these problems, the use of catalysts coated onto metal substrates was considered to be the best step forward. An organisation with experience of coating techniques and catalyst development was sought. Contact was made with the Applied Catalysis department of ECN in the Netherlands. It was agreed that ECN would lead a group, with CJBD and VSEL as partners, to develop a compact, responsive, methanol reformer for use on board a fuel cell driven vehicle. ECN was to develop an improved methanol reforming catalyst (lower operating temperature, lower carbon monoxide levels in exit gases) and a combustion catalyst for burning fuel cell off­ gases. These catalysts were to be applied to metal substrates which would be used to construct an all metal reformer. The enhanced heat transfer characteristics were expected to improve the start-up time and transient response.

CJBD was to test this reformer and VSEL was to undertake system modelling. This work programme formed the basis of an unsuccessful submission to the JOULE II programme.

However, in order to gain an understanding of the potential for the development of such a reformer, the UK DTI Advanced Fuel Cell Programme and the Netherlands government agreed to fund a preliminary investigation. The aim was to focus on the key development issues, concentrating on developing and evaluating an improved methanol reforming catalyst and its application to a metal substrate. In addition, the project was seen as a step to building a suitable collaboration that would be capable of taking forward the technology in any future phase.

3 3. OBJECTIVES

The objectives of the programme were:

• To develop and evaluate stable and highly active copper based catalysts to produce reformate with low levels of carbon monoxide and apply these catalysts to metal substrates.

• To test and evaluate these catalytic metal substrates to determine their suitability as the basis of an integrated catalytic methanol steam reformer for the production of a hydrogen-rich fuel for SPFCs in transport applications.

• To evaluate different design options for a reformer using catalytically coated metal substrates and assess the likely future cost of volume production of the reformer.

The development and evaluation of the copper based catalysts and support systems was carried out by ECN. The construction of a bench scale reformer and evaluation of the catalyst coated substrates was carried out by CJBD. The use of metal substrates was designed to improve the heat characteristics and temperature profiles in the reformer. Improving these characteristics is believed to lead to improved performances (reduced start-up time and faster dynamic response), reduced costs and a compact construction.

ECN supplied the metal substrates inside a tube which was installed in the bench- scale reformer test rig built by CJBD.

4. CATALYST DEVELOPMENT

ECN has developed a copper based catalyst for methanol steam reforming. Several promoted and unpromoted copper-on-alumina catalysts were prepared by wet impregnation of a gamma alumina support with the objective of improving the activity and selectivity of the copper with respect to the desired reactions. Impregnations were carried out with solutions containing either copper nitrate, the promoter precursor salt(s) or a combination using a one-step procedure. In most preparations the promoters were added to the support by co-impregnation with copper nitrate. After impregnation, the heated powders were homogenised and air dried at 80°C for 15 hours. After drying, the catalyst precursors were calcined in air at

4 different temperatures to form a metal oxide and release nitrogen dioxide. The concentration of copper nitrate in the impregnating solution was adjusted to obtain the desired metal loading.

The gas-phase reformation of methanol to carbon dioxide (CO2) , carbon monoxide (CO) and hydrogen (H2) was studied in a fully automated micro-flow apparatus, operating at atmospheric pressure. Prior to the catalytic experiments, the precursors were dried in situ at 100°C for 30 minutes in a flowing hydrogen-nitrogen (N2) mixture (25% hydrogen). Following the drying step, reduction was performed by raising the reactor temperature to 250°C at a rate of 1°C/min and maintaining this temperature for 4 hours. After reduction, the catalyst sample was cooled to 150°C in flowing nitrogen. The nitrogen flow was subsequently saturated with methanol and water by bubbling the gas stream through two saturators, held at the required temperature either by means of a cooled or a heated circulator, resulting in concentrations of water and methanol in the gas stream of 5-9 vol.% and 4.5-5%, respectively. The gas mixture was passed down through the fixed catalyst bed that initially contained 0.4g of the dried catalyst precursor (i.e. before reduction). The catalyst was contained in a Pyrex reactor with an internal diameter of 10mm. The temperature was measured just below the catalyst bed with a chromel-alumel thermocouple.

In Arrhenius-type experiments the reactor temperature was first set at 150°C (where the catalytic activity of most catalysts developed) and then the temperature of the catalyst bed was stabilised for 30 minutes. Following stabilisation, the bed temperature was raised in steps of 5°C to 250°C with a preset stabilisation time between two measurements. The same procedure was followed with decreasing bed temperature. In isothermal experiments, the reactor temperature was first set at the desired temperature level, the bed temperature stabilised for 30 minutes and, subsequently, the performance of the catalyst was recorded as a function of time on stream. Sampling of the product gas was performed with a pneumatically controlled six-way valve and the gas sample was analysed using a gas chromatograph equipped with a methanizer and a flame ionisation detector in series. Data points were collected at regular intervals.

The general test conditions were as follows:

5 Precursor weight (before reduction) 0.40 g Grain size 0.50 - 0.71 mm Gas flow rate 25 - 150mlmin"1 Space velocity 1500 - 9000 hr-1 Feed 5.4 - 9.4 vol.% water/methanol 4.5 - 4.8 vol.% make-up nitrogen Steam/methanol ratio 1.15 - 2.05 Total pressure atmospheric Temperature range for Arrhenius 150 - 250°C measurements Analysis Gas chromatograph Detector: FID (carrier N2, 6.5psi)

Based on the results of the catalyst development phase, techniques were developed for applying these catalysts to an aluminium substrate. A suitable alumina washcoat was applied to an aluminium foam inside an aluminium tube with the intention of reproducing performance similar to the initial development.

Using these techniques, ECN produced four tubes in January 1995. (see figure 1). All four tubes contained an aluminium oxide washcoat layer. Tube 1 contained copper and potassium with a total catalyst weight of 71g. Tube 2 contained copper and lanthanum with a total catalyst weight of 86g. Tube 3 was almost identical to tube 1 except the method of preparation was different. Tube 4 was similar to tube 3 but contained additional copper and potassium with a total weight of 98g.

5. BENCH-SCALE REFORMER CONSTRUCTION AND OPERATION

Each tube was installed on a bench scale reformer test rig designed and built by CJBD. (see figure 2). A premixed methanol/water solution was pumped into a vaporiser followed by a superheater. Both were electrically heated and filled with bulk conductive material to promote heat transfer. The hydrogen/nitrogen mixture used to reduce the catalyst in the ECN reformer tube was initially introduced before the vaporiser and superheater. After tests using the first two tubes, the rig was modified to introduce the hydrogen/nitrogen mixture after the vaporiser and superheater. This was to avoid reducing the bulk material in these vessels which was predominantly copper. The bulk filling in the superheater was also changed from copper turnings to stainless steel balls to remove the possibility of reforming taking

6 place in the superheater. A high temperature filter was fitted prior to the ECN bed to catch any debri leaving the vaporiser and superheater being trapped in the metal substrate in the reformer. A water cooled condenser reduced the water content of the reformate. The endothermic heat of reaction (i.e. simulated combustion catalyst) was supplied by an electrical heating tape. For each tube a number of parameters were recorded including temperatures, pressures and flowrates. Temperatures were measured on the outer surface of the tube and in the inlet and outlet plenum chambers. Pressures were monitored before and after the reformer tube. A mass flowmeter was positioned downstream of the condenser/catchpot system to monitor the reformate flowrate. The reformate composition was determined by specific CO2, H2 and CO analysers. The vaporiser and superheater were electrically heated. Numerous temperature measurements were made in these vessels. Methanol: water mixtures were premixed and flowrates recorded. All monitored parameters were recorded in a computerised recording system which also controlled the bed temperatures.

TUBE NO. 1

The catalyst was reduced using a N2 flowrate of 6 lmin"1 , and a H2 flowrate of 2 lmin" 1 . Catalyst reduction was undertaken in accordance with ECN’s instructions and generally in line with CJBD’s technique for ICI pelleted catalyst. The initial step, raising the temperature to 100°C, was a drying stage followed by reduction at 250°C, which was reached at a rate of 1°C/minute.

The gas mixture was passed through the tube with the wall temperature set at 50°C. After 24 minutes the wall temperature was increased to 75°C. This wall temperature was slowly increased to 87°C after 37 minutes and then to 100°C after a further 21 minutes. After 71 minutes at 100°C the wall temperature was increased by 5°C every 5 minutes until 250°C was reached. This temperature was maintained for 5 hours and fifteen minutes, at the end of which the inlet hydrogen concentration did not equal the outlet concentration which is the criteria used to determine whether reduction is complete. This was probably due to the copper turnings in the vaporiser and/or the superheater being reduced. Therefore, the gas flow was stopped and the heaters turned off. The following day the bed was heated to 250°C and the same flow was put through it for 5 hours and 41 minutes.

7 The wall temperature was set at 250°C to commence reforming. The methanol and water stoichiometry was 1:1.5 and the feedrate was 2.74 lh-1 . There was some performance initially, probably due to reforming in the vaporiser/superheater, of about 5-6 lmin-1 of reformate which had a carbon monoxide concentration of 0.65%. The pressure slowly dropped as the flowrate fell and the carbon monoxide concentration had increased to over 8.2% at the end of testing, which had lasted for 2 hours and 1 minute.

Reforming was tried again with a lower feedrate of 1.33 1hr -1 maintaining the stoichiometry at the same 1:1.5. The wall temperature set point was 250°C. There was no evidence of reforming occurring.

The copper turnings in the superheater were replaced with stainless steel balls to remove the possibility of reforming onthe surface of the copper.

TUBE NO. 2

The catalyst was reduced using 25% H2 in nitrogen at a total flowrate of 8 lmin-1 . The wall temperature was held at 50°C for 14 minutes, increased to 75°C for 13 minutes then increased to 100°C for 1 hour 42 minutes. The wall temperature was then increased by 5°C every 5 minutes, up to 250°C. The bed was kept at 250°C for 5 hours 30 minutes.

Reforming was attempted using the following parameters:

Feed stoichiometry 1.5 water : 1 methanol

Feedrate 2.7 1hr -1 .

Wall temperature 225°C.

There was a small amount of reforming (<2 lmin-1 ) due to some decomposition in the vaporiser. This decreased with time and after 2 hours 21 minutes reforming has ceased.

Reforming was tried at a lower feedrate (1.33 1hr -1 ) but this was increased to 2.7 1hr -1 when there was no noticeable improvement. The wall temperature was 250°C. The flowrate was initially recorded as 5 lmin-1 , but dropped to 1 lmin-1 after 1 hour of

8 reforming. The carbon monoxide concentration increased from 0.45% to 0.6% over this period.

TUBE NO. 3

Before evaluation of tube 3 the pipework on the rig was changed so that the hydrogen and nitrogen flow did not go through the vaporiser and the superheater.

The catalyst was reduced using 25% H2 in nitrogen at a total flowrate of 8 lmin-1 . The bed was heated from ambient to 100°C, then held for 1 hour at 100°C.

The wall temperature was then increased by 5°C every 5 minutes until 250°C was reached. The bed was kept at 250°C for 5 hours and 6 minutes.

There was no difference in hydrogen concentration between the inlet and the outlet at any point in the “reduction”.

No testing was attempted. The bed was left under nitrogen.

TUBE NO. 4

No reduction was attempted and hence no testing was undertaken on this tube.

It was concluded that the tubes were faulty and after consultation they were returned to ECN for examination at the end of March 1995. ECN concluded that the copper loading of the washcoat was low (less than 1%). Subsequent tests in their microflow apparatus revealed only 3/4% methanol conversion. This necessitated ECN having to revise their coating technique.

TESTING: PHASE II

New tubes were received at CJBD at the beginning of September 1995, although the coating technique was still not optimised. A new tube was reduced in a similar manner to the first batch with the exception that the slow heat-up was eliminated. Reduction was noted and the reforming performance shown in Table 1 observed.

9 6. EVALUATION OF RESULTS

The results obtained demonstrate that methanol steam reforming does take place across the metal substrate coated with methanol reforming catalyst albeit at temperatures higher than predicted and with a higher level of carbon monoxide in the reformate. The performance is in fact similar to pelleted commercial catalysts although strict comparison is difficult because of differences in catalyst copper loading and bed volumes. Considering that the catalyst composition and method of coating have yet to be optimised, the results are encouraging. Further testing of a tube was continued to investigate the stability of the catalyst. This was determined by noting any reduction of the reformate flow. A further forty hours operation suggested that there had been a deactivation of around 30%.

7. CONCLUSIONS

The two major objectives of the project have been achieved in that ECN has developed a copper based catalyst which can be applied to a metal substrate and that this catalytic substrate can be used as the basis of a compact methanol steam reformer. However, although the performance of the catalyst is comparable to that of commercial pelleted catalysts, in terms of temperature of operation and carbon monoxide content of the reformed gas, an improvement was anticipated. The catalysts and application techniques are not yet optimised and there is still scope for the anticipated improvement.

Because of delays at ECN in the early part of the contract and the failure of the first four tubes, the third major objective, to evaluate different design options for the reformers, was not undertaken.

The reasons for the failure of the first tubes was attributed in part to ECN attempting to move from microflow to breadboard size without an intermediate scale stage. It is believed that if an intermediate stage had been used, the faults in coating technique would have been detected before the tubes were manufactured. ECN has now remedied this by building a miniflow apparatus.

10 8. RECOMMENDATIONS

Based on the results obtained so far, ECN need to continue their efforts to optimise the methanol reforming catalyst composition and metal substrate configuration. The means of applying the methanol reforming catalyst to the substrate also needs to be refined. It is also necessary to identify a suitable combustion catalyst to burn fuel cell off-gases. Similar experimental methods used for the reforming catalyst could be applied.

CJBD together with ECN and others, in an expanded programme to include high temperature selective oxidation, successfully applied for EC funding as part of the FRAMEWORK IV programme.

As part of this programme, ECN will be developing the catalyst formulation and coating techniques further and CJBD will test methanol reforming tubes, combustion catalyst tubes and a combined unit in their existing test rig system.

In addition, the project (MERCATOX) will build and test a high temperature selective oxidation reactor developed by Loughborough University. It is intended that CJBD will then design and test a combined methanol reformer and gas clean-up unit suitable for use in a mobile SPFC vehicle to meet a target specification compiled by the Rover Car company. This programme will last for 36 months and started January 1996.

11 Hydrogen

Methanol & Water Nitrogen For the Reduction of Tubes 1&2

Hydrogen

Nitrogen For the Reduction of Tube 3

Filter

Vaporiser

ECN Reformer Tube

Superheater

Condenser

Sample Valve to Analysers

Sample Valve Outlet Valve

Separator

Notes

Automatic Drain Valve Copper Turnings Drain Valve Originally Copper Turnings & Alumina Replaced by Stainless Steel Balls During Testing of Tube 1

Figure 2 : Bench Scale Reformer Test Rig 20.00"

19.50" End View

2.5" OD x 0.125" wall 6061-T6 Aluminium Alloy

0.25" -

Foam Metal Catalyst Substrate 6061-T6 Aluminium Alloy

Figure 1 : Schematic of ECN Tubes TABLE 1 Methanol Feed Catalyst Wall Temperature °C Rate gm/hr

240 250 260 270 280 290 300

241 2.1 3.0 4.0 6.6 Reformate Flow l/min

.51 .58 .63 .77 % CO

628 2.8 4.5 8.0 9.9 118 15.8 Reformate Flow l/min

2.9 .332 .41 .51 .60 .80 % CO

1000 1.5 3.8 5.4 7.0 11.1 Reformate Flow l/min

.16 .22 .24 .29 .36 % CO

1000 2.4 3.4 5.5 8.2 12.2 16.7 18.4 Reformate Flow l/min

.18 .22 .26 .31 .41 .52 .66 % CO

Table 1 : Reforming Performance Results