Application of robust machine learning methods to modeling hydrogen solubility in hydrocarbon fuels

Having accurate information about the hydrogen solubility in hydrocarbon fuels and feedstocks is very important in petroleum refineries and coal processing plants. In the present work, extreme gradient boosting (XGBoost), multi-layer perceptron (MLP) trained with Levenberg–Marquardt (LM) algorithm, adaptive boosting support vector regression (AdaBoost?SVR), and a memory-efficient gradient boosting tree system on adaptive compact distributions (LiteMORT) as four novel machine learning methods were used for estimating the hydrogen solubility in hydrocarbon fuels. To achieve this goal, a database containing 445 experimental data of hydrogen solubilities in 17 various hydrocarbon fuels/feedstocks was collected in wide-spread ranges of operating pressures and temperatures. These hydrocarbon fuels include petroleum fractions, refinery products, coal liquids, bitumen, and shale oil. Input parameters of the models are temperature and pressure along with density at 20 °C, molecular weight, and weight percentage of carbon (C) and hydrogen (H) of hydrocarbon fuels. XGBoost showed the highest accuracy compared to the other models with an overall mean absolute percent relative error of 1.41% and coefficient of determination (R²) of 0.9998. Also, seven equations of state (EOSs) were used to predict hydrogen solubilities in hydrocarbon fuels. The 2- and 3-parameter Soave-Redlich-Kwong EOS rendered the best estimates for hydrogen solubilities among the EOSs. Moreover, sensitivity analysis indicated that pressure owns the highest influence on hydrogen solubilities in hydrocarbon fuels and then temperature and hydrogen weight percent of the hydrocarbon fuels are ranked, respectively. Finally, Leverage approach results exhibited that the XGBoost model could be well trusted to estimate the hydrogen solubility in hydrocarbon fuels.     

Modeling solubility of sulfur in pure hydrogen sulfide and sour gas mixtures using rigorous machine learning methods

Accurate determination of sulfur solubility in pure hydrogen sulfide (H₂S) and sour gas mixtures has a leading role and a fundamental importance in handling and addressing sulfur deposition issues. In this study, rigorous paradigms based on two artificial neural network (ANN) types, namely multilayer perceptron (MLP) and cascaded forward neural network (CFNN), optimized by Levenberg–Marquardt (LM) algorithm were proposed as machine learning (ML) modeling tools to predict the solubility of sulfur in sour gas mixtures and pure H₂S. Besides, explicit and simple-to-use correlations were established using gene expression programming (GEP). The paradigms derived from the methods aforementioned were developed using widespread experimental database. The obtained results indicated that the outcomes gained from the proposed MLP, CFNN and GEP-based correlations are in a high coherence and agreement with the experimental data. In addition, it was found that among the all suggested schemes, CFNN models are the most accurate paradigms for estimating the solubility of sulfur in sour gas mixtures and pure H₂S with root mean square error (RMSE) of 0.0232 and 3.8101, respectively. Furthermore, a comparison between the performance of CFNN and the prior alternatives demonstrated that the CFNN models predict the solubility of sulfur in sour gas mixtures and pure H₂S more accurately. Moreover, based on the trend analysis, it was concluded that the predictions of CFNN follow the real tendency of sulfur solubility in pure H₂S and sour gas mixtures with respect to the input parameters. Besides, the sensitivity analysis dictated that pressure and temperature have the most significant impact on sulfur solubility calculation in pure H₂S and sour gas mixtures. The results reported in this investigation revealed that implication of the considered soft computing approaches in the estimation of sulfur solubility in sour gas mixtures and pure H₂S can lead to the generation of more reliable predictive paradigms which can be integrated in other related applications. Lastly, the findings of this study can help for effective prediction of the solubility of sulfur in sour gas mixtures and pure H₂S while simulating different natural gas processes.     

Effect of rapid solidification on hydrogen solubility in Mg-rich Mg-Ni alloys

The hydrogenation properties of Mg_100−xNi_x alloys (x = 0.5, 1, 2, 5) produced by melt spinning and subsequent high-energy ball milling were studied. The alloys were crystalline and, in addition to Mg matrix, contained finely dispersed particles of Mg₂Ni and metastable Mg₆Ni intermetallic phases. The alloys exhibited excellent hydrogenation kinetics at 300 °C and reversibly absorbed about 6.5 mass fraction (%) of hydrogen. At the same temperature, the as prepared Mg_99.5Ni_0.5 and Mg₉₅Ni₅ powders dissolved about 0.6 mass fraction (%) of hydrogen at the pressures lower than the hydrogen pressure corresponding to the bulk Mg-MgH₂ two-phase equilibrium, exhibiting an extended apparent solubility of hydrogen in Mg-based matrix. The hydrogen solubility returned to its equilibrium value after prolonged hydrogenation testing at 300 °C. We discuss this unusually high solubility of hydrogen in Mg-based matrix in terms of ultrafine dispersion of nanometric MgH₂ precipitates of different size and morphology formed on vacancy clusters and dislocation loops quenched-in during rapid solidification.     

Raman spectroscopy study of molecular hydrogen solubility in water at high pressure

We have measured the Raman spectra of gaseous molecular hydrogen dissolved in liquid water at room temperature and as a function of pressure. Vibrational spectra of molecular hydrogen have been clearly detected. Band intensities and profiles have been carefully measured using, for calibration purposes, the water OH stretching band. From the measured intensities of the Raman band, we have obtained the behavior of hydrogen concentration in the liquid water, as a function of the gas partial pressure. The observed behavior is presented and compared to Henry’s law predictions. Additionally, we present a detailed analysis of the spectral band features from which important information on the interaction of hydrogen with water molecules could be derived.     

Effect of thermal history on the terminal solid solubility of hydrogen in Zircaloy-4

The terminal solid solubility (TSS) of hydrogen in zirconium alloys has a hysteresis. The TSS of hydrogen in Zircaloy-4 during cooling and heating were studied using differential scanning calorimetry (DSC) with a hydrogen content of 40–731 wppm. A significant hysteresis gap was observed between the TSS for dissolution (TSSD) and precipitation (TSSP). It was confirmed that the hydrogen dissolution temperature was unaffected by the previous thermal history in comparison with the hydride precipitation temperature. The TSSP temperature increased with a decrease in the maximum temperature, but a significant temperature gap remained even when the maximum temperature was equal to the TSSD temperature. The terminal solid solubility of hydrogen in Zircaloy-4 can be represented by the following equations.     

Application of advanced correlative approaches to modeling hydrogen solubility in hydrocarbon fuels

In petroleum and petrochemical refineries, having precise knowledge regarding H₂ solubility in hydrocarbon fuels and feedstocks is critical. In this study, the hydrogen solubility in hydrocarbon fuels was estimated using genetic programming (GP) and group method of data handling (GMDH), two exemplary robust advanced models for generating correlation. To do this, 445 observations derived from labratory findings on hydrogen solubility in 17 different hydrocarbon fuels such as bitumen, atmospheric residue, heavy coking gas oil, heavy virgin gas oil, light virgin gas oil, straight run gas oil, shale fuel oil, dephenolated shale fuel oil, diesel, hydrogenated coal liquid, coal liquid, and coal oil, over a large interval of P- operating pressures and T-temperatures were collected. Temperature, pressure, as well as density at 20 °C, molecular weight, and weight percentage of carbon (C) and hydrogen (H) in hydrocarbon fuels, were used as input parameters in developing robust correlations. The outcomes showed the GMDH approach is more precise compared to the GP, with a root mean square error (RMSE) of 0.053302 and a determination coefficient (R²) of 0.9641. Additionally, sensitivity analysis showed that pressure, followed by temperature and H (wt%) of hydrocarbon fuels, has the greatest impact on hydrogen solubility in hydrocarbon fuels. Ultimately, the Leverage method's results suggested that the GMDH model could be relied on to predict hydrogen solubility in hydrocarbon fuels.     

Effect of nickel addition on the solubility of hydrogen in tantalum

A study of isothermal as well as isobaric P–C–T equilibrium measurements has been investigated for the solubility of hydrogen in tantalum and its alloys with nickel (1.7 and 4.9 atom % Ni) in the temperature range of 673–873 K and hydrogen pressure range of 0.60–1.20 atmospheres. The alloys were prepared by arc melting in an inert atmosphere. The dissolved hydrogen was within the solid solubility range corresponding to the temperature and followed the Sievert's law. The hydrogen solubility in tantalum decreased on the addition of nickel as an alloying element. The change in enthalpy and the change in entropy of solution for hydrogen in the tantalum metal and its alloys were calculated. The heat of reaction for hydrogen solution in all the samples was exothermic. The enthalpy of solution for hydrogen in the tantalum matrix increases on the addition of Ni as an alloying element.     

Data-driven diagnosis of the high-pressure hydrogen leakage in fuel cell vehicles based on relevance vector machine

The hydrogen pressure inside tanks and its adjacent pipes can reach up to 70 MPa in fuel cell vehicles. This is the weak links of hydrogen leakage. The diagnosis time of mainstream hydrogen leakage diagnosis method based on hydrogen concentration sensors (HCSs) is easily affected by the number and location of installed sensors. In this study, a data-driven diagnosis method is proposed for the high-pressure hydrogen leakage. Fisher discrimination analysis and linear least squares fitting are used for data preprocessing, relevance vector machine is used for pattern recognition. When the total volume of tanks is 82 L and the hydrogen leakage flow rate is larger than 2 g/s, the diagnosis accuracy of the proposed method is higher than 95% and the diagnosis time is constant. When the leakage location is far away from HCSs, the proposed method can the diagnose hydrogen leakage in a shorter time than mainstream method.     

Modelling of hydrogen induced pressurization of internal cavities

Internal cavities can constitute a crack initiation site especially if filled with hydrogen at high pressure. A new refined equation of state based on recent NIST database has been introduced in order to model the equilibrium pressure. It is based on a thermodynamic definition of fugacity and uses the NIST data relating hydrogen density and pressure to define a new fugacity pressure quadratic dependence. The resulting Equation Of State (EOS) is compared to the standard Abel–Noble EOS and it is shown that for a given fugacity, imposed by a Sievert's law, the corresponding pressure is significantly higher. This new refined EOS was introduced into a previously developed numerical model of hydrogen diffusion and desorption and applied to evaluate the kinetics of pressure build-up within a cavity and its equilibrium pressure. It has been shown that the kinetics of pressure build-up at room temperature, which reaches values close to equilibrium in some hundreds of hours, is compatible with the industrial quality control procedures. The calculated pressures are in the range 4500–8650 bars depending on hydrogen solubility, which differs between the matrix and the segregation bands, and tend to equilibrium values obtained from mass balance approach.     

Models,methods and approaches for the planning and design of the future hydrogen supply chain

The infrastructure of hydrogen presents many challenges and defies that need to be overcome for a successful transition to a future hydrogen economy. These challenges are mainly due to the existence of many technological options for the production, storage, transportation and end users. Given this main reason, it is essential to understand and analyze the hydrogen supply chain (HSC) in advance, in order to detect the important factors that may play increasing role in obtaining the optimal configuration. The objective of this paper is to review the current state of the available approaches for the planning and modeling of the hydrogen infrastructure. The decision support systems for the HSC may vary from paper to paper. In this paper, a classification of models and approaches has been done, and which includes mathematical optimization methods, decision support system based on geographic information system (GIS) and assessment plans to a better transition to HSC. The paper also highlights future challenges for the introduction of hydrogen. Overcoming these challenges may solve problems related to the transition to the future hydrogen economy.     

Hydrogen solubility in furfural and furfuryl bio-alcohol: Comparison between the reliability of intelligent and thermodynamic models

This study compares the reliability of intelligent and thermodynamic modeling of hydrogen (H₂) solubility in two bio-derived compounds (furfuryl alcohol and furfural). The intelligent modeling phase conducts using seven different scenarios. The most accurate approach selects employing the ranking analysis over various statistical indices. The general regression neural network appears as the best intelligent model for the given purpose. This model presents the relative absolute deviation (RAD) of 0.6%, mean square error of 3.87 × 10⁻⁷, and regression coefficient of 0.99912 for predicting experimental measurements in the literature. The general regression neural network accuracy is far better than the perturbed-chain statistically associating fluid theory (PC-SAFT), Peng-Robinson, and Soave-Redlich-Kwong correlations. The most accurate thermodynamic approach, i.e., PC-SAFT, predicts H₂ solubility in furfuryl alcohol and furfural with the RAD = 4.58% and 4.62%, while the general regression model has RAD = 0.79% and 0.5%. Indeed, the proposed model improves the prediction accuracy by more than three hundred percent.     

Predicting the hydrogen uptake ability of a wide range of zeolites utilizing supervised machine learning methods

This study predicts the hydrogen uptake ability of 28 zeolites using artificial neural networks. The topology-related features of four artificial neural networks and their hyper-parameters determine utilizing 349 experimental data reported for zeolites with different surface areas at wide ranges of pressure and ?196 °C. Ranking analysis over various statistical criteria confirms that the most accurate model is the cascade feed-forward neural networks with twelve hidden nodes and logarithm and tangent sigmoid activation functions. This model predicts the experimental data with the absolute average relative deviation of 7.24%, mean absolute error of 0.041, root mean squared error of 0.058, and regression factor of 0.99429. The leverage method approves that 94% of the databank is reliable. Furthermore, the relevancy analysis shows a strong direct relationship between the hydrogen uptake ability of zeolite (HUAZ) and pressure and surface area. This study is the first to develop a model for predicting the HUAZ.     

Studies on the solubility of hydrogen in molten Pb83Li17 eutectic alloy

Lead–lithium eutectic (Pb₈₇Li₁₇) alloy is a candidate material to be used as a secondary tritium breeder, neutron multiplier and heat transfer agent in the fusion reactor. The tritium thus produced in the alloy may be soluble or appear as a new phase of lithium-tritides and/or lead-tritides, which eventually affect the performance of Pb₈₃Li₁₇ eutectic. Therefore, solubility of tritium in the alloy at the operating conditions of the fusion reactor is a subject matter of investigation. Tritium being the isotope of hydrogen behaves more or less similar to the hydrogen. In the present investigation the solubility of hydrogen in the Pb₈₃Li₁₇ has been investigated as a function of temperature and pressure. It was found that, hydrogen solubility in the Pb₈₇Li₁₇ alloy is almost constant above 350 °C. Hydrogen solubility increases with increase of temperature up to 400 °C. Hydrogen solubility is 120 ppm at 400 °C and 800 Torr hydrogen pressure. The solubility decreases on further rise in temperature from 400 °C. However, at all the temperatures hydrogen solubility increased with increase of partial pressure of hydrogen.     

Modeling interfacial tension of the hydrogen-brine system using robust machine learning techniques: Implication for underground hydrogen storage

During the last years, there has been a surge of interest in cleaner ways for producing energy in order to successfully handle the climate issues caused by the consumption of fossil fuels. The production of hydrogen (H₂) is among the techniques which have grown up as attractive strategies towards energy transition. In this context, underground hydrogen storage (UHS) in saline aquifers has turned into one of the greatest challenges in the context of conserving energy for later use. The interfacial tension (IFT) of the H₂-brine system is a paramount parameter which affects greatly the successful design and implementation of UHS. In this study, robust machine learning (ML) techniques, viz., genetic programming (GP), gradient boosting regressor (GBR), and multilayer perceptron (MLP) optimized with Levenberg-Marquardt (LMA) and Adaptive Moment Estimation (Adam) algorithms were implemented for establishing accurate paradigms to predict the IFT of the H₂-brine system. The obtained results exhibited that the proposed models and correlation provide excellent estimations of the IFT. In addition, it was deduced that MLP-LMA outperforms the other models and the existing correlation in the literature. MLP-LMA yielded R² and AAPRE values of 0.9997 and 0.1907%, respectively. Lastly, the trend analysis demonstrated the physical coherence and tendency of the predictions of MLP-LMA.     

Insights into the solubility of H2 in various polyethylene matrices at high pressure: A coarse-grained MC/MD study

After isothermal crystallisation with coarse-grained (CG) molecular dynamics simulations, the solubility of H₂ molecules in different amorphous/semi-crystalline polyethylene(PE) systems was investigated by hybrid GCMC/MD simulations with explicit crystalline phase in high-pressure environment. The crystalline phase is successfully obtained by isothermal crystallisation for branch-free PE, while the existence of the branch hindered the formation of the crystalline phase in PE. In the amorphous phase, the density of end groups turns out to be a critical factor of H₂ solubility. As the branch density increase the density of the end groups, the branch density, but not the branch length has a strong positive correlation with H₂ solubility. While H₂ is excluded from the crystalline phase and the majority of them dissolve within the amorphous region, H₂ molecules were observed to dissolve in the crystalline-amorphous interface and directly neighbouring with the crystalline phase, with a preference for dissolving in the amorphous phase without being directly linked to crystalline phase via chemical bonds. Such behaviour could be explained by the rigidity of bonded crystalline-amorphous interface.     

State estimation of a batch hydrogen production process using the photosynthetic bacteria Rhodobacter capsulatus

This paper addresses the problem of estimating the states of an anaerobic photosynthetic process used for biohydrogen production by the photosynthetic bacterium Rhodobacter capsulatus. The process is described by a non-linear, time-discrete model and the state estimation is solved using an observer based on the Moving-Horizon State Estimation Method (MHSE). This approach is based on the minimization of a criterion (a non-linear function), in this case, the difference between the estimated output and the measured output of the system over a considered time horizon, where the solution is computed by using a numerical interval method. The observer was successfully applied to hydrogen production by R. capsulatus strain B10 in a batch process.     

Optimal operation of a wind-electrolytic hydrogen storage system in the electricity/hydrogen markets

Wind power, the most promising renewable energy source in the world, plays an important role in the electricity markets. Wind power curtailment cannot be avoided in some countries due to its output has a special feature of randomness and volatility. Since the excess wind power being converted into hydrogen and sold to the hydrogen market will be the future trend. This study proposes a wind-electrolytic hydrogen storage system to participate in the electricity/hydrogen markets for selling electricity and hydrogen, which can help to improve the benefits of wind power in the electricity markets and addree the wind power curtailment effectively. With considering the uncertainties of wind power outputs and electricity prices, the optimal operation strategy is proposed with the objective of maximizing profits. The scenario-based stochastic method is adopted to describe the uncertainties, and the financial risk is evaluated using conditional value-at-risk. The operational problem of the proposed system is formulated into a mixed-integer linear programming model. Finally, the feasibility of the proposed operational strategy is validated by a case study. The results show that the expected revenue increases with the increase of the hydrogen selling price, indicating that investors can obtain profits by converting electricity into hydrogen. The optimal expected revenue increases by 33.42% when hydrogen price increases from 1.2 DKK/kWh to 1.8 DKK/kWh and the risk factor is equal to 0. Based on the analysis of the results, the importance of hydrogen can be proven.     

Comparative study of eleven equations of state in predicting the thermodynamic properties of hydrogen

Eleven equations of state are employed to predict the vapor pressures, liquid and vapor densities, liquid and vapor heat capacities, and vaporization enthalpies and entropies of normal hydrogen along the coexistence curve. The volumetric and thermal properties of gaseous hydrogen together with the speeds of sound, Joule–Thomson coefficients and inversion curves for wide ranges of temperature and pressure are predicted as well. The results are compared with experimental data and the recommended values of standard thermodynamic tables. The best equations of state in predicting the properties of hydrogen (saturated and supercritical) are introduced and reported.     

Genetically engineered deletion in the N-terminal region of nifA1 in R. capsulatus to enhance hydrogen production

NifA is the primary activator of nitrogenase, and the N-terminal domain of nifA is sensitive to ammonium concentration. In this work, a mutant Rhodobacter capsulatus ZX01 with a genetically engineered deletion in the N-terminal region of nifA1 was constructed by employing overlap extension PCR to mitigate the inhibition of ammonium on nitrogenase expression in photosynthetic bacteria. The effects of different ammonium ion concentrations on the growth and photo-fermentative hydrogen production performance of wild-type strain R. capsulatus SB1003 and mutant ZX01 with glucose and volatile fatty acids as the carbon sources were studied, respectively. When the ratio of NH₄⁺-N was 20% and 30%, the hydrogen yield of the mutant ZX01 was enhanced by 14.8% and 20.9% compared with that of R. capsulatus SB1003 using 25 mM acetic acid and 34 mM butyric acid as the carbon source, respectively. In comparison, using 30 mM glucose as the carbon source, the hydrogen yield of ZX01 was increased by 17.7% and 22.2% compared with that of R. capsulatus SB1003 when the ratio of NH₄⁺-N was 20% and 30%, and the nitrogenase activity of ZX01 was also enhanced by 38.0% and 47.6%, respectively. When using 10 mM NH₄⁺ as a single nitrogen source, ZX01 showed a 2.6-fold increase in H₂ production. These results indicated that ZX01 demonstrated higher ammonium tolerance and better hydrogen production performance than the wild-type. The deletion in the N-terminal region of nifA1 could partially de-repress the nitrogenase activity inhibited by ammonium.