Hydrogen production using ethylene glycol steam reforming in a micro-reformer: Experimental analysis, multivariate polynomial regression and genetic programming modeling approaches
Introduction
Nowadays, many researches in energy field have been focused on hydrogen production from biomass-derived sources due to rising demand for clean energy, fossil fuel reserves reduction and increasing greenhouse gas emissions [1,2]. Fuel cells are the efficient devices that generate electricity from hydrogen as a sustainable and environmentally friendly energy source [3,4]. The main processes for hydrogen production are auto-thermal reforming (ATR) [5], steam reforming (SR) [6], sorption-enhanced steam reforming (SESR) [7], aqueous phase reforming (APR) [8], partial oxidation (POX) [9], and supercritical water gasification (SWG) [10]. Recently, attentions focused on biomass derived oxygenated source for hydrogen generation due to their carbon neutral, renewable and accessible properties [11]. Ethylene glycol (EG) is the most plentiful product of catalytic conversion of cellulose (the main component of biomass) which can be considered as a non-volatile hydrogen source through two main approaches; SR [12] and APR [13] processes. The APR reaction occurs at low temperatures (150°C to 270°C) and pressures in the range of 1500 to 5000 kPa. However, the APR reaction rate and H2 selectivity are low and in order to inhibit catalyst deactivation, the EG concentration in the feedstock should be low. Compared to the APR, the SR process takes place at higher temperatures and at atmospheric pressure which high energy cost of the SR reaction can be balanced by the use of inexpensive catalysts and considering higher H2 selectivity [14]. In ethylene glycol steam reforming (EGSR) process, several chemical reactions should be considered, which include the overall SR (Eq. (1)), EG decomposition (Eq. (2)), methanation (Eq. (3)) and water gas shift (WGS) (Eq. (4)) reactions, as follows [15]:
The elements in group VIII e.g. rhodium [16], ruthenium [17], platinum [18], and palladium [19] have been introduced as promising catalyst active phases for hydrogen generation from hydrocarbon reforming processes. However, noble metallic particles have high selectivity towards hydrogen but their cost is high. Another alternative for reforming reaction are nickel based catalysts due to their low price, high availability and excellent activity for carbon-carbon bonds [20,21]. As a result, the use of Ni-noble metal bimetallic catalysts in various reactions including reforming, hydrogenation and dehydrogenation has been the subject of many studies compared to single-metal catalyst [20] due to their unique properties such as excellent resistance to coke formation and high catalytic activity (H2 selectivity and feed conversion) [22,23].
In most previous research studies in the EGSR field, the process was performed in a conventional macro scale reactor (fixed bed reactor) with mass and heat transfer limitations which are critical issues for fuel cell applications. Micro-reformer (including sub-millimeter channels) deposited by the catalyst is an attractive light-weight device for portable fuel processing as hydrogen source [24,25]. Due to the endothermicity of steam reforming process, accurate monitoring and control of the reaction temperature in these compact fuel processors resulted in reduction of hot spot regions and inhibition of catalyst deactivation [26]. Izquierdo et al. [27] performed the EGSR process over Rh-based catalysts loaded on La2O3 and CeO2 modified α-Al2O3 in a micro-reactor.
Mathematical modeling is a useful tool for the design and control of chemical processes, in particular for steam reforming in a micro-reformer [28,29]. It should be noted that, very few studies have been performed on the modeling of the EG reforming process, and all of them are limited to kinetic and micro-kinetic modeling over Pt catalyst [30]. In recent years, considerable attention has shifted to modeling of chemical processes with multivariate polynomial regression (MPR) and artificial intelligence methods including artificial neural networks (ANNs), genetic algorithm (GA) and genetic programming (GP) approaches [31], [32], [33], [34]. These models (MPR, ANN and GP) do not use any transport equation and only require experimental input (independent variables) and output (dependent variables) data of process. Omata et al. [35] used an experimental design procedure coupled with ANNs to investigate the influence of preparation parameters of Co–MgO catalysts in methane dry reforming process. Ayodele and Cheng [36] applied the Box–Behnken design and ANNs together to model and optimize the input variables (CH4 partial pressure, temperature of reaction and reactant feed ratio) to achieve desired conversion of CH4 and CO2 and syngas ratio. Hossain et al. [37]. used ANNs for modeling of H2 and CO yields as well as CH4 and CO2 conversions for methane dry reforming over Ni/CaFe2O4 catalysts. Azarhoosh et al. [5,33,38] used GA for optimization and simulation of a horizontal ammonia synthesis reactor and hydrogen generation by low-pressure ATR of natural gas. Optimization of hydrogen production from toluene steam reforming by ANN coupled GA (ANN-GA) and response surface methodology (RSM) models studied by Yahya et al.[39]. Recently, Lotfi et al. [40] applied 2-dimentinal heterogeneous model and a modified ANN and non-dominated sorting genetic algorithm-II (NSGA-II) to maximize methane conversion in the industrial combined dry and steam reforming (CDSR) process.
Notably, the GP model has not been used for the reforming process, so far. For other processes, Azarhoosh et al. [41] used GA and GP approaches for presenting a new kinetic model for methanol-to-light olefins (MTO) process in the presence of the hierarchical SAPO-34 catalyst. A comparison between the experimental and predicted results of the new kinetic model showed that the proposed model has good accuracy. In another work, they modeled the performance of hierarchical SAPO-34 catalyst in the MTO process using multiple linear regression (MLR) and two intelligent methods, i.e., GP and ANN [34]. According to their finding, the models achieved using the GP method had the highest accuracy for training and test data compared to other models.
In previous work, we studied the EGSR process over Ni, Pt and NiPt catalysts loaded on γ-alumina support in a micro-reformer, experimentally. The γ-alumina was chosen as the catalyst support because of its perfect properties such as good thermal stability, high specific surface area, adjustable pore structure and good adsorption performance [31,42,43].
To the best of our knowledge, the MPR and GP approaches have not yet been used to model the efficiency of a micro-reformer, in particular for the EGSR reaction. Therefore, in this research work, the main objective is the development of MPR (linear, quadratic and cubic) and GP models which can predict the behavior of EGSR process in a micro-reformer for hydrogen production in the presence of three types catalysts (Pt/γ-alumina (PGA), Ni/γ-alumina (NGA) and Ni-Pt/γ-alumina (NPGA)), followed by validation with experimental data. Finally, by selecting the optimal model from the MPR or GP models, a model was used to study the effects of operating parameters (temperature and WHSV) on the target functions i.e. EG conversion, H2 selectivity, H2 yield and CO selectivity with the best accuracy and prediction power.
Section snippets
Catalyst synthesis and characterization
The synthesis method and characterization results of γ-alumina supported Pt or/and Ni catalysts were reported in our previous work [15]. Notably, in this work, 3%Pt/γ-alumina, 12%Ni/γ-alumina and 12%Ni 3%Pt/γ-alumina catalysts designated as PGA, NGA and NPGA, respectively. The γ-alumina support was synthesized with the sol-gel route. Then, PGA, NGA and NPGA catalysts were prepared via incipient wetness impregnation (IWI) method. According to BET, XRD, H2-TPR and FESEM analyses, the PGA, NGA and
Modeling results
Predicted results (by MPR and GP models) and experimental data of EG conversion, H2 selectivity and CO selectivity for EGSR experiments over NGA, PGA and NPGA catalysts in the micro-reformer are listed in Tables S1-S9. To evaluate the accuracy of the models, the experimental data for each catalyst were randomly divided into two training and test data sets. For this purpose, for each catalyst, the data related to 16 experiments were used as training data and remaining data (the data related to 6
Conclusion
In this work, the performance of the EGSR process by the NGA, PGA and NPGA catalysts in the micro-reformer was modeled using the MPR and GP methods. The catalysts were prepared by the IWI route and were evaluated at different WHSVs (80.8, 121.3 and 161.7 h−1) and temperatures (300–450°C). The main findings can be summarized as follows:
- •
For all catalysts, increasing the reaction temperature improved the H2 yield and the EG conversion while elevating WHSV decreased EG conversion.
- •
While the NPGA
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
References (48)
- et al.
Fabrication, characterization, and application of palladium composite membrane on porous stainless steel substrate with NaY zeolite as an intermediate layer for hydrogen purification
Int J Hydrogen Energy
(2019) - et al.
Lab-scale and pilot-scale two-stage gasification of biomass using active carbon for production of hydrogen-rich and low-tar producer gas
Fuel Process Technol
(2020) - et al.
Renewable hydrogen production by steam reforming of butanol over multiwalled carbon nanotube-supported catalysts
Int J Hydrogen Energy
(2019) - et al.
Aqueous-phase reforming of bio-oil aqueous fraction over nickel-based catalysts
Int J Hydrogen Energy
(2019) - et al.
Modeling interactions between chemistry and turbulence for simulations of partial oxidation processes
Fuel Process Technol
(2015) - et al.
Hydrogen production through biomass gasification in supercritical water: a review from exergy aspect
Int J Hydrogen Energy
(2019) - et al.
Renewable hydrogen production by ethylene glycol steam reforming over Al2O3 supported Ni–Pt bimetallic nano-catalysts
Renew Energy
(2018) - et al.
Aqueous phase reforming of ethylene glycol over bimetallic platinum-cobalt on ceria–zirconia mixed oxide
Int J Hydrogen Energy
(2017) - et al.
Evaluation of a micro-channel reactor for steam reforming of ethylene glycol: a comparative study of catalytic activity of Pt or/and Ni supported γ-alumina catalysts
Int J Hydrogen Energy
(2018) - et al.
On the dynamics and reversibility of the deactivation of a Rh/CeO2ZrO2 catalyst in raw bio-oil steam reforming
Int J Hydrogen Energy
(2019)
Ruthenium doped nickel-alumina-ceria catalyst in glycerol steam reforming
Fuel Process Technol
The role of Pt loading on La2O3–Al2O3 support for methane conversion reactions via partial oxidation and steam reforming
Fuel
Hydrogen production by steam methane reforming in a membrane reactor equipped with a Pd composite membrane deposited on a porous stainless steel
Int J Hydrogen Energy
Hydrogen production from steam reforming ethanol over Ni/attapulgite catalysts-Part I: effect of nickel content
Fuel Processing Technology
Methane dehydrogenation and oxidation process over Ni-based bimetallic catalysts
Fuel
Low temperature glycerol steam reforming on bimetallic PtSn/C catalysts: on the effect of the Sn content
Fuel
Micro-channel reactor for steam reforming of methanol
Fuel
Investigation of autothermal reforming of methane for hydrogen production in a spiral multi-cylinder micro-reactor used for mobile fuel cell
Int J Hydrogen Energy
Methanol steam reforming in a microchannel reactor by Zn-, Ce-and Zr-modified mesoporous Cu/SBA-15 nanocatalyst
Int J Hydrogen Energy
Micro reactor hydrogen production from ethylene glycol reforming using Rh catalysts supported on CeO2 and La2O3 promoted α-Al2O3
Int J Hydrogen Energy
Hydrogen production in a zigzag and straight catalytic wall coated micro channel reactor by CFD modeling
Int J Hydrogen Energy
Microkinetic modeling of aqueous phase biomass conversion: application to ethylene glycol reforming
Chem Eng Sci
Degradation of Fluoxetine using catalytic ozonation in aqueous media in the presence of nano-γ-alumina catalyst: experimental, modeling and optimization study
Sep Purif Technol
Performance analysis of ultrasound-assisted synthesized nano-hierarchical SAPO-34 catalyst in the methanol-to-lights-olefins process via artificial intelligence methods
Ultrason Sonochem
Cited by (14)
Designing catalysts via evolutionary-based optimization techniques
2023, Computational Materials ScienceCitation Excerpt :Small variations in an encoded programme can also greatly alter the time for computing the fitness of a given program and may lead to very long run-times if not mitigated [33]. AD Kiadehi et al. [34] used GP along with multivariate polynomial regression to model hydrogen production by ethylene glycol steam reforming over three different catalysts. J. Zheng et al [35] used GP to accelerate the screening of catalysts for the hydrogen evolution reaction.
High H<inf>2</inf> selectivity with low coke formation for methanol steam reforming over Cu/Y<inf>1.5</inf>Ce<inf>0.84</inf>Ru<inf>0.04</inf>O<inf>4</inf> catalyst in a microchannel plate reactor
2022, International Journal of Hydrogen EnergyCitation Excerpt :From mass transfer point of view, because diffusive mixing under streamline flow is dominate mechanism in narrow channels, the dimensions scaling down (to micron scale) can cause perfect mixing of the reactants in a short time according to Fick's law. Consequently, the overall reaction rate and products selectivity (in some cases) can raise remarkably in micro-channel reactor with enhancement heat and mass transfer efficiency [33]. Therefore, a suitable, economical and promising option is methanolic microformers to supply hydrogen fuel for fuel cells [34].
A complete fuel processor for propylene glycol as hydrogen supply for a 5 kw low temperature pem fuel cell – Interim report on single reactors and system performance
2022, Catalysis TodayCitation Excerpt :Owing to the boundary conditions of the application, highest compactness is required. To date, no polyalcohol reformer reactors and fuel processors have been reported in open literature apart from lab scale reactors for catalyst testing applying fixed bed or even membrane reactor technology [5] or microchannel reactors [6,7]. However, some theoretical analyses have been performed revealing a required reforming temperature exceeding at least 600 °C and a potential for increased efficiency of polyalcohol fuels and methanol compared to higher mono-alcohols, acetic acid and acetone [8].
Towards improved genetic programming based-correlations for predicting the interfacial tension of the systems pure/impure CO<inf>2</inf>-brine
2021, Journal of the Taiwan Institute of Chemical EngineersCitation Excerpt :GP was proposed and developed by [33]. This rigorous ML techniques showed satisfactory results when modeling various parameters and phenomena, such as hydrogen production using ethylene glycol steam reforming [34], oil formation volume factor [35], sandstone permeability [36], the IFT of the system paraffin-carbon dioxide [37], the IFT of the system oil/brine [38], and the compressibility factor of gas condensate [39]. During the iterative process of GP which aims at establishing the most powerful correlation, the so-called tree representation is applied to manage the individuals of the population known as chromosomes.