Development of 10-kWe preferential oxidation system for fuel cell vehicles

A preferential oxidation (PROX) reactor for a 10-kWe polymer electrolyte membrane fuel cell (PEMFC) system is developed. Pt-Ru/Al₂O₃ catalyst powder, with a size of 300–600 μm is applied for the PROX reaction. To minimize pressure drop and to avoid hot spots in the catalyst bed, the reactor is designed as a dual-staged, multi-tube system. The performance of the 10-kWe PROX unit is evaluated by feeding simulated gasoline reformate which contains 1.2 wt.% carbon monoxide (CO). The CO concentration of the treated reformate is lower than 20 ppm in the steady-state and is under 30 ppm at 65% load change. Hydrogen loss in the steady-state is about 1.5% and the pressure drop across the reactor is 4 psi. Start-up characteristics of the 10-kWe PROX system are also investigated. It takes 3 min to reduce the CO concentration to below 20 ppm. Several controllable factors are found to shorten the start-up time.     

Improving temperature uniformity and performance of CO preferential oxidation for hydrogen-rich reformate with a heat pipe

Preferential oxidation (PROX) is an effective, but highly temperature-sensitive, method of CO removal for hydrogen-rich reformates. In a packed-bed catalytic reactor, oxidation is strongest at the inlet side and the local catalyst pellets become over-heated with poor heat conduction. As a result, the enhanced parasitic H₂ oxidation consumes oxygen and suppresses CO conversion. This study applies a heat pipe to improve the temperature uniformity in a tubular one-stage packed-bed reactor by transporting heat downstream and thereby improve CO removal. In the experiments, the fuel mixture containing 2% of CO, 75% of H₂, and 23% of CO₂, further mixed with air at O₂/CO = 0.75, 1.0 or 1.25, is supplied with stepwise increase of feeding rate under a fixed environmental temperature of 99 ± 1 °C. The proposed simple method is found to significantly improve temperature uniformity and CO removal for the present test conditions with O₂/CO = 1.0 and 1.25.     

Performance of a poly(2,5-benzimidazole) membrane based high temperature PEM fuel cell in the presence of carbon monoxide

The performance of a poly(2,5-benzimidazole) (ABPBI) membrane based high temperature PEM fuel cell in presence of carbon monoxide, at various temperatures is reported here. The ABPBI was synthesized by polymerization of 3,4-diaminobenzoic acid in a polymerization medium containing methanesulfonic acid (CH₃SO₃H) and phosphorous pentoxide (P₂O₅). The ABPBI membranes were characterized by fourier transform infrared spectroscopy (FT-IR) and scanning electron microscopy (SEM). A maximum conductivity of 0.026 S cm⁻¹ at 180 °C was obtained for the membrane doped with 1.2 molecules of phosphoric acid (H₃PO₄) per polymer repeat unit. Fuel cell performance was evaluated using dry hydrogen/oxygen gases and was comparable with that reported in the literature. Performance of a single cell at different temperatures was studied with 0.48 and 1.0 vol.% of CO in the hydrogen fuel. The studies lead to the conclusion that CO poisoning is not a serious problem above 170 °C. Performance of the fuel cell operating at 210 °C is not at all affected by 1.0 vol.% of CO in the hydrogen feed.     

Catalytic performance of a copper–promoted CeO2 catalyst in the CO oxidation: Influence of the operating variables and kinetic study

A study of the CO oxidation reaction was conducted on a CuO–CeO₂ catalyst, with a copper content of 20 at%. The stability, activity and selectivity of this sample were evaluated in the absence of H₂ and also under PROX conditions (i.e. in great H₂ excess). The influence of the reaction temperature and the feed composition was also analyzed. It was found that water and CO₂ have a negative effect on the catalytic activity. Except for the undesired oxidation of hydrogen, the occurrence of other lateral reactions that could be present in the PROX reactor can be discarded. A kinetic study was carried out in order to fit a classical power-law expression for CO oxidation rate. Although this mathematical expression gave a good fit in limited concentration ranges, it was found that the partial reaction orders depend on the reactant molar fractions.     

Experimental optimization analysis on operating conditions of CO removal process from hydrogen-rich reformate

The catalytic effects of CO preferential oxidation and methanation catalysts for deep CO removal under different operating conditions (temperature, space velocity, water content, etc.) are systematically studied from the aspects of CO content, CO selectivity, and hydrogen loss index. Results indicate that the 3 wt% Ru/Al₂O₃ preferential oxidation catalysts reduce CO content to below 10 ppm with a high hydrogen consumption of 11.6–15.7%. And methanation catalysts with 0.7 wt% Ru/Al₂O₃ also exhibit excellent CO removal performance at 220–240 °C without hydrogen loss. Besides, NiCl_x/CeO₂ methanation catalysts possess the characteristics of high space velocity, high activity, and high water-gas resistance, and can maintain the CO content at close to 20 ppm. Based on these experimental results, the coupling scheme of combining NiCl_x/CeO₂ methanation catalysts (low cost and high reaction space velocity) with 0.7 wt% Ru/Al₂O₃ methanation catalysts (high activity) to reduce CO content to below10 ppm is proposed.     

Study of Ce-Pt/γ-Al2O3 for the selective oxidation of CO in H2 for application to PEFCs: Effect of gases

In order to supply pure hydrogen to proton exchange membrane (PEM) fuel cells and avoid CO poisoning, selective CO oxidation in H₂ was studied over Ce-Pt/γ-Al₂O₃. Adding the Ce promoted the CO conversion and selectivity of Pt/γ-Al₂O₃ with changing loading weights of Pt and Ce, oxygen concentration, residence time, and the composition of gases (H₂O, CO₂, and N₂). At 250 °C, adding H₂O to the feed gas enhanced the CO conversion due to the water–gas shift reaction. While, adding CO₂ to the feed gas suppressed the CO conversion due to the reversible water–gas shift reaction. In situ BET and XRD tests showed that well-dispersed metallic Pt particles (−2 nm) existed on the Ce oxide over the alumina support, which helps to supply oxygen to the Pt for a high activity of CO oxidation and selectivity.     

Kinetics,simulation and insights for CO selective oxidation in fuel cell applications

The kinetics of CO preferential oxidation (PROX) was studied to evaluate various rate expressions and to simulate the performance the CO oxidation step of a methanol fuel processor for fuel cell applications. The reaction was carried out in a micro reactor testing unit using a commercial Engelhard Selectoxo (Pt–Fe/γ-alumina) catalyst and three self-prepared catalysts. Temperature was varied between 100 and 300 °C, and a of range feed rates and compositions were tested. A reaction model in which three reactions (CO oxidation, H₂ oxidation and the water gas shift reaction) occur simultaneously was chosen to predict the reactor performance. Using non-linear least squares, empirical power-law type rate expressions were found to fit the experimental data. It was critical to include all three reactions to determine good fitting results. In particular, the reverse water gas shift reaction had an important role when fitting the experimental data precisely and explained the selectivity decrease at higher reaction temperatures. Using this three reaction model, several simulation studies for a commercial PROX reactor were performed. In these simulations, the effect of O₂/CO ratio, the effect of water addition, and various non-isothermal modes of operation were evaluated. The results of the simulation were compared with corresponding experimental data and shows good agreement.     

Negative pressure dependence of mass burning rates of H2/CO/O2/diluent flames at low flame temperatures

Experimental measurements of burning rates, analysis of the key reactions and kinetic pathways, and modeling studies were performed for H₂/CO/O₂/diluent flames spanning a wide range of conditions: equivalence ratios from 0.85 to 2.5, flame temperatures from 1500 to 1800 K, pressures from 1 to 25 atm, CO fuel fractions from 0 to 0.9, and dilution concentrations of He up to 0.8, Ar up to 0.6, and CO₂ up to 0.4. The experimental data show negative pressure dependence of burning rate at high pressure, low flame temperature conditions for all equivalence ratios and CO fractions as high as 0.5. Dilution with CO₂ was observed to strengthen the pressure and temperature dependence compared to Ar-diluted flames of the same flame temperature. Simulations were performed to extend the experimentally studied conditions to conditions typical of gas turbine combustion in Integrated Gasification Combined Cycle processes, including preheated mixtures and other diluents such as N₂ and H₂O.Substantial differences are observed between literature model predictions and the experimental data as well as among model predictions themselves – up to a factor of three at high pressures. The present findings suggest the need for several rate constant modifications of reactions in the current hydrogen models and raise questions about the sufficiency of the set of hydrogen reactions in most recent hydrogen models to predict high pressure flame conditions relevant to controlling NO_x emissions in gas turbine combustion. For example, the reaction O + OH + M = HO₂ + M is not included in most hydrogen models but is demonstrated here to significantly impact predictions of lean high pressure flames using rates within its uncertainty limits. Further studies are required to reduce uncertainties in third body collision efficiencies for and fall-off behavior of H + O₂(+M) = HO₂(+M) in both pure and mixed bath gases, in rate constants for HO₂ reactions with other radical species at higher temperatures, and in rate constants for reactions such as O + OH + M that become important under the present conditions in order to properly characterize the kinetics and predict global behavior of high-pressure H₂ or H₂/CO flames.     

Effects of dilution with carbon dioxide on the laminar burning velocity and flame stability of H2–CO mixtures at atmospheric condition

The objective of this investigation was to study the effect of dilution with CO₂ on the laminar burning velocity and flame stability of syngas fuel (50% H₂–50% CO by volume). Constant pressure spherically expanding flames generated in a 40 l chamber were used for determining unstretched burning velocity. Experimental and numerical studies were carried out at 0.1 MPa, 302 ± 3 K and ? = 0.6–3.0 using fuel-diluent and mixture-diluent approaches. For H₂–CO–CO₂–O₂–N₂ mixtures, the peak burning velocity shifts from ? = 2.0 for 0% CO₂ in fuel to ? = 1.6 for 30% CO₂ in fuel. For H₂–CO–O₂–CO₂ mixtures, the peak burning velocity occurred at ? = 1.0 unaffected by proportion of CO₂ in the mixture. If the mole fraction of combustibles in H₂–CO–O₂–CO₂ mixtures is less than 32%, then such mixtures are supporting unstable flames with respect to preferential diffusion. The analysis of measured unstretched laminar burning velocities of H₂–CO–O₂–CO₂ and H₂–CO–O₂–N₂ mixtures suggested that CO₂ has a stronger inhibiting effect on the laminar burning velocity than nitrogen. The enhanced dilution effect of CO₂ could be due to the active participation of CO₂ in the chemical reactions through the following intermediate reaction CO + OH ? CO₂ + H.     

Ultrasonic-assisted synthesis of highly active catalyst Au/MnOx–CeO2 used for the preferential oxidation of CO in H2-rich stream

A series of nano-gold catalysts supported on binary oxides MO_x–CeO₂ (atomic ratio M/Ce = 1:1, M = Mn, Fe, Co, Ni) are prepared by deposition–precipitation (DP). An innovative and rather convenient ultrasonic pretreatment of the support is employed for Au/MnO_x–CeO₂ preparation. It is found that for preferential CO oxidation Au/MnO_x–CeO₂ is more active than Au/CeO₂. Ultrasonic pretreatment of MnO_x–CeO₂ further promotes the performance of Au/MnO_x–CeO₂, with CO conversion increased by 24 % at 120 °C. Meanwhile, the selectivity of oxygen to CO₂ is promoted in the whole temperature range, especially in 80–120 °C, the selectivity is increased by 15–21%. HR-TEM and XRD results indicate that ultrasonic pretreatment is favorable to the formation of much smaller gold nanoparticles (<5 nm). The characterization of XPS, UV–vis DRS, H₂-TPR and CO-TPR confirms that the strong interaction between Au and the support effectively inhibits the dissociation and oxidation of H₂ over the ultrasonically pretreated catalyst Au/MnO_x–CeO₂, making it highly selective to CO oxidation.     

Selective CO oxidation over a commercial PROX monolith catalyst for hydrogen fuel cell applications

Preferential oxidation (PROX) of CO over noble-metal-containing monolith catalysts is one of the most promising approaches for removing CO to generate low temperature fuel cell quality H₂. The monolith-supported washcoated catalyst comprising Cu and Fe promoted with Pt is highly effective in reducing the CO in practical reformates to less than 10 ppm over a broad range of feed compositions, inlet temperatures and turn down ratios. It is speculated that Pt dissociates the H₂ which then reduces the CuO to its active state. Pt may also act as a cocatalyst for CO adsorption with metal oxides supplying oxygen for PROX reaction. The catalytic system is operated adiabatically with an inlet temperature between roughly 65–120 °C reaching an exit temperature close to 150 °C with no evidence of reverse water gas shift or methanation. The goal was to find the proper operating conditions to achieve <10 ppm CO. Turn down ratios (varying space velocities) at a factor of 4–5 are routinely achieved up to at least 34,000 h⁻¹ with high steam levels of up to 45%. The wide operating window simplifies the control of the PROX reactor and improves the fuel processor’s performance for fast startup and shutdown and responses to transient loads. The catalyst also retains its performance after multiple start and stops modes of operation in reformate.     

High-temperature sulfuric acid decomposition over complex metal oxide catalysts

Activity and stability of FeTiO₃, MnTiO₃, NiFe₂O₄, CuFe₂O₄, NiCr₂O₄, 2CuO·Cr₂O₃, CuO and Fe₂O₃ for the atmospheric decomposition of concentrated sulfuric acid in sulfur-based thermochemical water splitting cycles are presented. Catalyst activity was determined at temperatures from 725 to 900 °C. Catalytic stability was examined at 850 °C for up to 1 week of continuous operation. The results were compared to a 1.0 wt% Pt/TiO₂ catalyst. Surface area by nitrogen physisorption, X-ray diffraction analyses, and temperature programmed desorption and oxidation were used to characterize fresh and spent catalyst samples.Over the temperature range, the catalyst activity of the complex oxides followed the general trend: 2CuO·Cr₂O₃ > CuFe₂O₄ > NiCr₂O₄ ≈ NiFe₂O₄ > MnTiO₃ ≈ FeTiO₃. At temperatures less than 800 °C, the 1.0 wt% Pt/TiO₂ catalyst had higher activity than the complex oxides, but at temperatures above 850 °C, the 2CuO·Cr₂O₃ and CuFe₂O₄ samples had the highest activity.Surface area was found to decrease for all of the metal oxides after exposure to reaction conditions. In addition, the two complex metal oxides that contained chromium were not stable in the reaction environment; both leached chromium into the acid stream and decomposed into their individual oxides. The FeTiO₃ sample also produced a discoloration of the reactor due to minor leaching and converted to Fe₂TiO₅. Fe₂O₃, MnTiO₃ and NiFe₂O₄ were relatively stable in the reaction environment. In addition, CuFe₂O₄ catalyst appeared relatively promising due to its high activity and lack of any leaching issues; however it deactivated in week-long stability experiments.Complex metal oxides may provide an attractive alternative to platinum-based catalyst for the decomposition of sulfuric acid; however, the materials examined in this study all displayed shortcomings including material sintering, phase changes, low activity at moderated temperatures due to sulfate formation, and decomposition to their individual oxides. More effort is needed in this area to discover metal oxide materials that are less expensive, more active and more stable than platinum catalysts.     

The kinetics of oxidation of Diesel soots by NO2

The kinetics of oxidation of three soots, from a Diesel engine fuelled by either Ultra-Low Sulphur Diesel (ULSD) or biodiesel, by NO₂ have been measured in a packed bed at various temperatures (300–550 °C) and [NO₂] (20–880 ppm) relevant to regenerating a Diesel Particulate Filter. Adsorbed hydrocarbons and oxygen accounted for a significant fraction (～20% by mass) of the otherwise carbonaceous material. After pre-treatment (heating up to 550 °C in a flow of pure Ar and holding the temperature at 550 °C for 1 h) to ensure consistency between samples, they were subsequently burned at a fixed temperature in a flow of NO₂ + Ar. For this, a balance on oxygen atoms entering and leaving the packed bed showed that during oxidation in NO₂ any oxygen remaining in a soot after pre-treatment was not rapidly liberated as CO or CO₂. A mass balance on the element nitrogen demonstrated that no N₂ or N₂O was formed below 550 °C; mass balances on carbon and oxygen demonstrated that all the carbon ended up as CO or CO₂ and below 550 °C the nitrogen yielded only NO. The oxidation of soot in NO₂ was found to be first-order with respect to NO₂. Also, the soot derived from biodiesel was more reactive than soot from ULSD; nevertheless, the apparent activation energies for oxidation by NO₂ were the same (70 ± 18 kJ mol^?1) for each carbon. When the distribution of diameters of the individual spherules of soot was taken into account, it was not possible to tell whether there was internal burning of porous spherules or, on the other hand, non-porous, solid spherules were burning on their exteriors.     

Hydrogen production for fuel cells through methane reforming at low temperatures

Hydrogen production for fuel cells through methane (CH₄) reforming at low temperatures has been investigated both thermodynamically and experimentally. From the thermodynamic equilibrium analysis, it is concluded that steam reforming of CH₄ (SRM) at low pressure and a high steam-to-CH₄ ratio can be achieved without significant loss of hydrogen yield at a low temperature such as 550 °C. A scheme for the production of hydrogen for fuel cells at low temperatures by burning the unconverted CH₄ to supply the heat for SRM is proposed and the calculated value of the heat-balanced temperature is 548 °C. SRM with and/or without the presence of oxygen at low temperatures is experimentally investigated over a Ni/Ce–ZrO₂/θ-Al₂O₃ catalyst. The catalyst shows high activity and stability towards SRM at temperatures from 400 to 650 °C. The effects of O₂:CH₄ and H₂O:CH₄ ratios on the conversion of CH₄, the hydrogen yield, the selectivity for carbon monoxide, and the H₂:CO ratio are investigated at 650 °C with a constant CH₄ space velocity. Results indicate that CH₄ conversion increases significantly with increasing O₂:CH₄ or H₂O:CH₄ ratio, and the hydrogen content in dry tail gas increases with the H₂O:CH₄ ratio.     

Char oxidation study of sugar cane bagasse,cotton stalk and Pakistani coal under 1% and 3% oxygen concentrations

Chars of Sugar cane bagasse (1 & 2), Cotton stalk and low rank Pakistani coal have been studied by TGA under low oxidative environments with O₂ concentrations of 1% and 3%. The maximum reactivity of the chars was found to be greater by a factor of 2 under 3% oxygen compared to 1% O₂ conditions. Overall conversion levels at 3% O₂ for Sugar cane bagasse-2 increased from 63% to 100%, Sugar cane bagasse-1; 54% to 97%, Cotton stalk; 45% to 100% and Pakistani coal; 63% to 90% in comparison to 1% O₂. The maximum average rate of weight loss was found in Region III compared to Region I and II supported by CO/CO₂ FTIR Chemigram analysis. On the other hand, % conversion was maximum in Region II under 1% and 3% O₂ concentration. Overall average rates of weight losses were dependant on O₂ concentration and temperature ranges, however for all the regions % conversion and average weight loss were twice in 3% compared to 1% O₂ concentration. Biomass chars were found to be more reactive than the coal studied here during each region of the oxidation process. Evaluated apparent energy of activations for biomass chars was found within range of 41.2–105.8 kJ mole^?1 under 1%, 46.9–125.6 kJ mole^?1 under 3% compared to coal; 70.3–183.9 kJ mole^?1 under 1% and 83.1–167.4 kJ mole^?1 in 3% O₂ concentration for order of reaction (n) varying between 0.5 ≤ n ≤ 2. From the tests carried under O₂ levels of 1% and 3%, it is possible to give the following sequence to the apparent activation energies under any of the fixed value of n, obtained for the biomasses and coal; Pakistani coal > Cotton stalk > Sugar cane bagasse-2 > Sugar cane bagasse-1.     

Operation of ceria-electrolyte solid oxide fuel cells on iso-octane–air fuel mixtures

Reduced-temperature solid oxide fuel cells (SOFCs) – with thin Ce_0.85Sm_0.15O_1.925 (SDC) electrolytes, thick Ni–SDC anode supports, and composite cathodes containing La_0.6Sr_0.4Co_0.2Fe_0.8O₃ (LSCF) and SDC – were fabricated and tested with iso-octane/air fuel mixtures. An additional supported catalyst layer, placed between the fuel stream and the anode, was needed to obtain a stable output power density (e.g. 0.6 W cm⁻² at 590 °C) without anode coking. The Ru-CeO₂ catalyst produced CO₂ and H₂ at temperatures <350 °C, while H₂ and CO became predominant above 500 °C. Power densities were substantially less than for the same cells with H₂ fuel (e.g. 1.0 W cm⁻² at 600 °C), due to the dilute (≈20%) hydrogen in the fuel mixture produced by iso-octane partial oxidation. Electrochemical impedance analysis showed a main arc that represented ≈60% of the total resistance, and that increased substantially upon switching from hydrogen to iso-octane/air.     

A shock tube study of the rate constants of HO2 and CH3 reactions

HO₂ and CH₃ are major intermediate species presented during the oxidation of natural gas at intermediate temperatures and high pressures. Previous theoretical calculations have identified several product channels for HO₂ and CH₃ reactions, with CH₃ + HO₂ → CH₃O + OH and CH₃ + HO₂ → CH₄ + O₂ being the dominant reaction pathways. Both reaction pathways play an important role in the kinetics of CH₄ oxidation as CH₃ + HO₂ → CH₃O + OH is a chain-branching reaction whereas CH₃ + HO₂ → CH₄ + O₂ a chain termination reaction.H₂O₂/CH₄/Ar mixtures were shock-heated to a temperature between 1054 and 1249 K near 3.5 atm to initiate the reaction. OH radicals yielded from H₂O₂ thermal decomposition react with H₂O₂ and CH₄ respectively to produce HO₂ and CH₃ in the reacting system. Using laser absorption spectroscopy, time-histories of H₂O, OH and HO₂ were measured behind reflected shock waves. The rate constant of reaction CH₃ + HO₂ → CH₃O + OH was determined to be 6.8 × 10¹² cm³ mol^?1 s^?1 with an uncertainty factor of 1.4. The rate constant of the competing CH₃ + HO₂ → CH₄ + O₂ reaction was determined to be 4.4 × 10¹² cm³ mol^?1 s^?1, with an uncertainty factor of 2.1. In addition, the rate constants of two other major reactions of the reacting system, H₂O₂ (+M) → 2OH (+M) and OH + CH₄ → CH₃O + OH, were found to have excellent agreement with values recommended in literature.     

Syngas production by gasification of aquatic biomass with CO2/O2 and simultaneous removal of H2S and COS using char obtained in the gasification

Applicability of gulfweed as feedstock for a biomass-to-liquid (BTL) process was studied for both production of gas with high syngas (CO + H₂) content via gasification of gulfweed and removal of gaseous impurities using char obtained in the gasification. Gulfweed as aqueous biomass was gasified with He/CO₂/O₂ using a downdraft fixed-bed gasifier at ambient pressure and 900 °C at equivalence ratios (ER) of 0.1–0.3. The syngas content increased while the conversion to gas on a carbon basis decreased with decreasing ER. At an ER of 0.1 and He/CO₂/O₂ = 0/85/15%, the syngas content was maximized at 67.6% and conversion to gas on a carbon basis was 94.2%. The behavior of the desulfurization using char obtained during the gasification process at ER = 0.1 and He/CO₂/O₂ = 0/85/15% was investigated using a downdraft fixed-bed reactor at 250–550 °C under 3 atmospheres (H₂S/N₂, COS/N₂, and a mixture of gases composed of CO, CO₂, H₂, N₂, CH₄, H₂S, COS, and steam). The char had a higher COS removal capacity at 350 °C than commercial activated carbon because (Ca,Mg)S crystals were formed during desulfurization. The char simultaneously removed H₂S and COS from the mixture of gases at 450 °C more efficiently than did activated carbon. These results support this novel BTL process consisting of gasification of gulfweed with CO₂/O₂ and dry gas cleaning using self-supplied bed material.     

100 t/a-Scale demonstration of direct dimethyl ether synthesis from corncob-derived syngas

Direct conversion of biomass-derived syngas (bio-syngas) to dimethyl ether (DME) at pilot-scale (100 t/a) was carried out via pyrolysis/gasification of corncob. The yield rate of raw bio-syngas was 40–45 Nm³/h with less than 20 mg/Nm³ of tar content when the feedrate of dried corncob was 45–50 kg/h. After absorption of O₂, S, Cl by a series of absorbers and partial removal of CO₂ by the pressure-swing adsorption (PSA) unit sequentially, the obtained bio-syngas (H₂/CO≈1) was directly synthesized to DME over Cu/Zn/Al/HZSM-5 catalyst in the fixed-bed tubular reactor. CO conversion and DME space-time yield (STY) were 67.7% and 281.2 kg/m_cat³/h respectively at 260 °C, 4.3 MPa and 3000 h^?1(GHSV, syngas hourly space velocity). Synthesis performance would be increased if the tail gas (H₂/CO > 2) was recycled to the reactor when GHSV was 650–3000 h^?1.