The optimization of propane-fueled spark-ignition engines is crucial for enhancing efficiency and reducing emissions. While deep neural network offers a powerful tool for this task, their performance is highly sensitive to the training hyperparameters of adaptive optimizers like Adam. This study systematically investigates the impact of the Adam optimizer's learning rate and first-moment coefficient on the performance of a deep neural network for predicting combustion, thermal efficiency, and emission characteristics. A high-fidelity engine simulation model, validated against experimental data, was used to generate a comprehensive dataset. The deep neural network's architecture was optimized, and its training was analyzed under varying combinations of learning rate (0.001, 0.01) and first-moment coefficient (0.8-0.95). Results indicate that a higher learning rate significantly accelerates convergence and improves predictive accuracy. The optimal configuration achieved the highest training accuracy of 93.75%, with coefficient of determination values exceeding 0.98 for all outputs. Transient parameters like effective release energy and NOx emissions were more sensitive to first-moment coefficient variation than integral metrics like brake mean effective pressure and thermal efficiency. The analysis demonstrates that adaptive control of these hyperparameters is critical for the deep neural network to resolve complex combustion phenomena accurately, thereby enabling faster calibration, improved energy efficiency, and more precise emission control in spark-ignition engines.

Ammonia is a promising carbon-free fuel for engines in response to the era of carbon neutrality. This study investigates the combustion and emission characteristics of an ammonia–diesel dual-fuel engine operated under lean and relatively low medium-speed conditions, with diesel injection timing and ammonia energy fractions as the primary parameters. Based on this, a deep neural network model was developed to predict critical parameters of the ammonia–diesel dual-fuel engine. The results indicate that applying delayed diesel injection timing enhanced the efficiency of the ammonia–diesel dual-fuel engine, including brake thermal efficiency. However, for ammonia energy fractions exceeding A50, the highest efficiency was observed at an earlier diesel injection timing of 12°CA BTDC rather than at the most delayed timing of 9°CA BTDC. Delayed diesel injection timing and higher ammonia energy fractions extended ignition delay, yet the shortening of the late combustion phase contributed significantly to reducing total combustion duration, thereby improving efficiency. While delayed diesel injection timing lowered NO, NOx, and CO emissions at the same ammonia energy fractions, it deteriorated N2O and NH3 emissions. NO2 and CO2 emissions displayed either pronounced nonlinearity or minimal sensitivity to diesel injection timing. The optimized deep neural network model achieved R2 values of 0.999, 0.996, 0.9733, 0.9635, 0.9471, and 0.934 for in-cylinder pressure, heat release rate, brake thermal efficiency, indicated mean effective pressure, NOx, and NH3, respectively, demonstrating exceptional predictive accuracy. These outcomes provide valuable guidance for advancing experimental and AI-driven approaches to optimize combustion, efficiency, and emission control in ammonia–diesel dual-fuel engines.

This research investigates the influence of combustion duration on combustion characteristics, emissions, and residual gas in a propane-fueled spark ignition engine under varying load conditions (25%, 50%, and 100%). Utilizing a two-cylinder engine equipped with a control dynamometer and supported by simulations conducted with AVL-Boost software, the study explores combustion durations ranging from 40 to 80 degrees crank angle. The study integrates simulation and experimental methods to address challenges in measuring key parameters such as residual gas and effective release energy under different conditions. Moreover, key performance metrics, including effective release energy, brake mean effective pressure (BMEP), brake-specific energy consumption (BSEC), hydrocarbon (HC), carbon monoxide (CO), and nitrogen oxides (NOx), were systematically analyzed. The results reveal that combustion duration significantly impacts engine performance and emissions, with longer durations improving combustion efficiency at lower loads but increasing BSEC. Residual gas ratio (RGR) varied more prominently at higher loads, indicating its strong interdependence with combustion characteristics. Optimal combustion durations were identified for each load, balancing performance and emissions. At lower loads, longer combustion durations reduced HC and CO emissions but exhibited fluctuations in NOx emissions. Conversely, at higher loads, shorter combustion durations were more effective in minimizing emissions.

This study explores how blending n-butanol with diesel fuel affects macro- and microscopic spray behavior under CI engine conditions. Investigations were conducted in a CVC by varying n-butanol ratio, nB70-nB90 (70-90 % nbutanol content), with different injection pressures and nozzle dimensions. Macro- and microscopic spray properties were analyzed thorough combined experimental and numerical approach. Increasing n-butanol content in diesel reduces both kinematic viscosity and surface tension; consequently, average spray area increases by 7.5 %, 5.1 % and 3.1 % on nB70, nB80 and nB90 respectively, exhibiting significant changes compared to pure diesel. Furthermore, changes in nozzle diameter have a significant impact on cone angle by 25.3 %, wider under 0.33 mm on nB90. Further investigation on numerical study shows linear correlation with the spray penetration, with nB90 reaching the smallest SMD of 8.2 mu m under 120 MPa injection pressure and 0.28 nozzle diameter upon other testing conditions. The smaller droplets are formed due to the poorer surface tension of nB90 fuels that help to break down the fuel droplets with ease, the reducing droplet diameter pattern-thus enhancing atomization of fuel. The results support by the reduction in Ohnesorge number and increase in Reynolds number for nB90 indicate substantially reduced droplet cohesion and enhanced aerodynamic breakup compared to pure diesel. The addition of n-butanol, up to 90 % on diesel fuel, demonstrated considerable potential for spray evolution features, providing direct guidance for injector optimization in alcohol-diesel blended fuel striving to achieve clean combustion applications, a wide range analysis describing the processes behind these phenomena is provided.

A study is conducted with simulation models to optimize the stability of gaseous pressure in a compressed natural gas (CNG) injection system. The simulation models are established for the injection system, including an operating model of a rail pipe, mechanical and electromagnetic models of a CNG injector and a pressure regulator. An ON–OFF controller is applied for the pressure regulator to keep a stable gaseous pressure in rail pipe as the injection processes is activated. The structural parameters of CNG injection system are selected to investigate their effects on the gaseous pressure stability in rail pipe, including volume of rail pipe, spring stiffness and number of coil turns in the pressure regulator. To optimize the stability of gaseous pressure in rail pipe based on above structural parameters, a particle swarm optimization (PSO) algorithm is applied. The study results indicate that the gaseous pressure stability in rail pipe could be improved when increasing rail pipe volume, increasing spring stiffness and reducing number of coil turns in the pressure regulator. PSO algorithm allows to find quickly the best structural parameters to optimize the gaseous pressure stability in rail pipe during injection processes.

Biogas-powered free-piston linear alternators (FPLAs) represent a promising solution for decentralized renewable electricity generation. However, the use of anaerobic-digested biogas in FPLAs and the effects of variable methane (CH4) content and operating conditions on combustion behavior remain poorly understood. This study combines multi-scale computational simulations with a data-driven framework to evaluate and predict the performance of biogas-driven FPLAs for the electrification of livestock waste. The three-dimensional transient model shows that lower CH4 content in biogas enhances delivered gas mass but reduces trapping efficiency by about 11 % (from 70 % CH4 to 40 % CH4 case). Moreover, while the delivered CH4 mass initially rises with concentration, it levels off beyond 70 % due to reduced total delivered gas. On the other hand, higher CH4 fractions accelerate combustion, increasing peak pressure and indicated mean effective pressure (IMEP) up to 3.5 MN/m2 and 0.66 MN/m2, respectively. Spark timing strongly influences combustion phasing, with an optimal advance was found at approximate to 6.7 mm, aligning peak heat release near top dead center. Overall, biogas with 45-70 % CH4 can be effectively utilized in FPLAs, offering a sustainable alternative to conventional engines in livestock farm electricity generation systems, with estimated thermal efficiencies reaching up to 34 %. Complementing these findings, a deep learning approach accurately predicts key performance metrics, demonstrating the potential of AI-driven models to support FPLA design, optimize operation, and enhance efficiency and sustainability in decentralized energy generation.

The article focuses on researching and improving dynamic performance of an electric power-assisted bicycle (EPAB) based on simulation models under human control. The model takes into account structural parameters such as wheel radius, bike mass, and crank length by building a simulation model of EPAB. The article then discusses the development of a mathematical model for the EPAB, which includes a dynamic model of the bicycle under the driver’s control, a dynamic model of the electric motor, and speed control models using a Fuzzy logic controller. These models are simulated and solved using MATLAB/Simulink. Furthermore, the article mentions the use of a Fuzzy algorithm to control the electric bicycle’s speed at 22 and 25 km/h, taking into account factors such as slope and wind speed. The effectiveness of this control is evaluated at each set speed and the external factors as mentioned above. Finally, the article discusses the establishment of an experimental system for electric-assist bicycles and the conducting of experiments in real conditions. The results of these experiments, such as velocity and moving distance of the EPAB, are compared to the simulations.

Biogas from palm oil mill effluent (POME) is a promising fuel that has many advantages as an alternative fuel. The methane content in biogas derived from POME is up to 75% and can be used as an alternative fuel in an internal combustion engine. One of the technologies for utilizing biogas in compression ignition engines is the Diesel Dual-Fuel (DDF) technique due to the different characteristics of fuel and the impact on the environment due to significantly reducing emissions. This study aims to find the effect of biogas POME composition and energy ratio on the DDF engine’s performance and emissions. The simulations using AVL BOOST software were confirmed by experimental engine parameters. The modeling was conducted on the biogas energy ratio (20%, 40%, 60%, and 75% POME) and biogas POME composition (55% and 75% methane). The results showed that the fuel consumption of diesel fuel was reduced by up to 69%, and NOx and soot emissions were reduced by up to 92% and 80%, respectively, with dual-fuel mode operation. Meanwhile, the value of brake mean effective pressure (BMEP) and efficiency was reduced by up to 18%, volumetric efficiency decreased by up to 4%, the increase in brake specific energy consumption (BSEC) was up to 23%, and brake specific fuel consumption (BSFC) was up to 155%. The optimum of the engine’s performance and emission was 40% biogas ratio with 75% methane content.

The laminar flame velocity is a significant metric in premixed combustion modeling of spark ignition (SI) engines. The present investigation was conducted to determine the chance of a dedicated EGR (d-EGR) system being added to a four-cylinder SI engine to increase thermal efficiency, which might be reduced owing to the high EGR ratio for reducing in-cylinder NOx generation by lowering the combustion temperature. Methane and propane were chosen as the test fuels. The numerical findings predicted by the PREMIX algorithm in CHEMKIN-PRO were used to determine flame temperature ( T_f ) and laminar burning velocity ( S_L ). The laminar burning velocities acquired at varied beginning pressures are required for conducting a thorough kinetic investigation of the combustion reaction and testing the actual reaction mechanisms. Thermal efficiency was calculated using the Wocshni’s heat transfer coefficient and Wiebe function. The findings reveal that the d-EGR mechanism boosted thermal efficiency, surpassing that of the typical SI engine’s stoichiometric combustion due to the low flame temperature and fast laminar burning velocity. These findings give essential theoretical references for enhancing the thermal efficiency of SI engines powered by methane and propane fuel.

This study presents a novel biomimetic flow-field concept that integrates a triply periodic minimal surface (TPMS) porous architectures with a hierarchical leaf-vein-inspired distribution zone, fabricated through 3D printing. By mimicking natural transport systems, the proposed design enhances oxygen delivery and water removal in proton exchange membrane fuel cells (PEMFCs). The results showed that I-FF and G-FF significantly improved mass transport and water management compared to conventional CPFF. The integrated design I-FF-LDZ achieves up to 32% improvement in power density at 1.85 A/cm2@0.4 V and delays the onset of mass transport losses. The study also reveals that optimizing the volume fraction Vf significantly affects gas penetration, with lower Vf (30%) improving performance in the mass-limited region. These findings underscore the promise of nature-inspired, 3D-printed flow-field architectures in overcoming key transport limitations and advancing the scalability of next-generation PEMFC systems.

In this study, mixtures of high‐octane gasoline–biodiesel (GB) rating from 0% to 20%, denoted as GB00‐GB10‐GB20, were chosen to be experimented under simulated gasoline compression ignition (GCI) engine inside a constant volume combustion chamber (CVCC) before making comparisons about spray and combustion characteristics. While typical data of the nonvaporization spray characteristics: spray penetration length, spray cone angle, and spray evolution were collected by varying 40–90 MPa of injection pressure at 15 kg/m 3 ambient gas density, data representing combustion characteristics, namely the ignition delay, flame development, and heat release rate (HRR) were illustrated under 90 MPa injection pressure, same ambient gas density, and temperature variation of 900–1000 K. Although the fuel mixture spray development trends followed a similar pattern, GB10 and GB20 had an averagely 6.25%–7.7% faster impingement velocity than that of GB00 except for the lowest considered injection pressure. Additionally, the ignition delay reductions were recognized for both GB10 and GB20, with the duration dropping from 1.09 to 1.06 ms and 0.96 to 0.79 ms, respectively. The results for spray and combustion characteristics under GCI engine condition‐like showed a close relationship, offering valuable insights for simulation, spray, combustion, and ANN modeling, thereby supporting the development and application of GCI engines.

The effects of diesel and the ammonia ratio on the emissions and combustion characteristics of ammonia utilized in AMMONIA direct injection (AMMONIA-Di) engines were investigated through experimental and numerical investigations. A rapid compression expansion machine (RCEM) modified to facilitate the dual direct injection fuel (diesel-ammonia) - compression ignition (CI) method was used to conduct the experiment. A compression ratio (CR) of 19 and an ammonia energy percentage ranging from 10% to 90% were used in the experiment. Changes were made to the start of injection (SOI) from 0o to 40o before top dead center (BTDC) in order to find the best auto-ignition properties of ammonia. In order to facilitate auto-ignition, the diesel’s SOI was maintained at 10o BTDC. Computational fluid dynamics (CFD) modeling was used to establish the detailed emission propagation during the combustion process. During the expansion step, ammonia goes through a second stage of combustion, demonstrating that the fuel cannot burn entirely during the initial auto-ignition process. Emissions of CO2, HC, and NOx rise when direct injection CI engines use up to 50% ammonia. When SOI is applied to ammonia at 0 and 40 BTDC with an ammonia energy percentage higher than 50%, the emissions vary significantly, indicating poor combustion quality that encourages the production of emissions.

Selective catalytic reduction (SCR), a widely used technology to mitigate NOx emissions from diesel engines, faces the challenge of improving NOx reduction efficiency at low temperatures. O3 can improve the low-temperature NOx reduction performance of SCR by oxidizing NO in exhaust gas to NO2, a more reactive molecule. This study investigates the potential of O3 assisted SCR reaction in real exhaust gas condition of diesel engine and compares the differences of two catalysts, Cu-zeolite and vanadium. The results show that NOx reduction performance at low temperature can be improved by assisting O3 in the both catalysts. The improvement of NOx reduction performance was slightly better for the Cu-zeolite catalyst than for the vanadium catalyst. This is because Cu-zeolite catalysts have a larger NO2 or HNO3 adsorption capacity and stronger bonding strength than vanadium catalysts, undergoing better SCR reaction and forming more NH4NO3.

The paper presents research results and design experimental system for electric motorbike. With purpose of this article is to study the input parameters that affect electric motorbikes, to better understand the important factors related to the operation and performance of electric motorbikes. To achieve this goal, an electric motorbike simulation model has been built based on MATLAB SIMULINK software. In addition, the research also focuses on designing and installing an experimental system to measure the speed, power and power consumption of motorbikes. The simulation model is used to study the influence of input factors and working conditions on the performance and power consumption of electric motorbikes. Through simulation research, the article shows that thanks to the integration of important structural parameters, such as wheel diameter and transmission ratio, this model can simulate the vehicle’s operation process effectively, authentic and accurate.

With rising fuel consumption across road transportation, there is growing interest in expanding the market share of renewable fuels, such as ethanol. Ethanol can be produced from raw materials from various starch-rich plants. In CI engines, ethanol cannot be utilized on its own, largely due to its low cetane number. In this study, a constant volume combustion chamber (CVCC) is employed to investigate the effects of adding ethanol in diesel with different proportions (10%, 20%, 30% v/v) on the spray and combustion characteristics. Optical techniques, such as shadowgraph and direct photography using high-speed imaging methods, were employed to reveal the spray and flame development process. This study examines the effects of varying fuel injection pressures (50, 80, and 110 MPa) and ambient pressures (1.5 and 3 MPa) on diesel-ethanol (DE) fuel blends. The study emphasizes the impact of DE blending ratios on the spray’s macroscopic features, while the microscopic characteristics are investigated through computational fluid dynamics (CFD) simulations to provide a comprehensive analysis of spray behavior under these conditions. The spray experiments are further combined with the flame and combustion characteristics of the blended fuel under different temperatures (800, 1000K), oxygen concentrations (15, 21%), and ambient density (15 kg/m3). It is revealed that the increase in ethanol content in diesel alters the fuel's physicochemical properties, resulting in a reduction not only in kinematic viscosity but also surface tension, thereby modifying the spray's behavior. Consequently, the average SPL is reduced, while a broader SCA is observed. Additionally, the Sauter Mean Diameter decreases with an increased ethanol ratio in diesel which indicates an improved atomization process compared to pure diesel fuel. In the case of spray combustion characteristics and the flame development process, the results demonstrate that the ethanol-blended spray flame has an unstable flame boundary, displaying multiple wrinkles at the outermost flame edge. Moreover, as the ethanol proportion increases, the peak combustion pressure experiences a slight drop under most test conditions. However, higher ambient temperature and oxygen concentration could significantly reduce the ignition delay. A thorough discussion of the mechanism underlying these events will be provided.

This research presents a numerical analysis of the environmental impacts associated with using hot steam as a co-product in hydrogen production through Steam Methane Reforming (SMR) of renewable gas sources. As hydrogen production technology advances rapidly, reducing emissions and addressing environmental concerns, particularly greenhouse gas (GHG) emissions, have become essential. This study examines the SMR process with a focus on the environmental effects of utilizing hot steam as a co-product for electricity generation or facility heating. The analysis evaluates renewable feedstocks, including landfill gas, animal waste, food waste, and wastewater sludge, to determine their viability for sustainable hydrogen production. Key pollutants, such as carbon monoxide and nitrogen oxides, along with GHGs, are assessed to identify the most environmentally advantageous feedstock options. This work aims to provide insights to promote sustainable hydrogen production practices.

Ethanol blend has been applied in a gasoline engine for many decades. However, it has a tremendous barrier for a compression ignition engine. Phase stability is one of the crucial factors. This research studies the ethanolblended fuel in the diesel engine using biodiesel as the emulsifier. Because it has been sold commercially, 10 % biodiesel blended in diesel (B10) is selected to mix with ethanol. As a result, 10 % ethanol (B10E10) could add into B10 without phase separation. The effect of the ternary blend on spray structure when varying injection pressure and air density is first investigated. The degree of atomization is analyzed through the Ohnesorge diagram. The influence of the spray on the combustion is last examined. The results found that cavitation is a remarkable feature of the ternary blend. It dominates the spray of ethanol blend over the operating conditions. The internal flow incident alters the tri-blend spray differing from the diesel and B10. B10E10 has the greatest atomization, followed by diesel and B10. The puffing phenomenon occurs and results in combustion. Since fuel properties (heat of vaporization) significantly affect the combustion, the effect of spray could not interfere. Therefore, the correlation between spray and combustion could not conclude.

This study investigates the effects of oxygenated fuels, specifically long-chain alcohols, impact fuel atomization and combustion behavior in CI engines. The objective is to examine how higher n-butanol blending ratios in diesel fuel influence spray dynamics and combustion performance under varying engine conditions using an advanced combustion strategy. Experiments were conducted using a constant volume chamber (CVC) and a rapid compression-expansion machine (RCEM), both designed to replicate CI engine conditions. N-butanol was blended with diesel at ratios ranging from 70% to 90% with 10% increments, and key parameters such as spray formation, cone angle, penetration length, in-cylinder pressure, combustion performance, and efficiency were analyzed. The study also evaluated the effects of varying injection pressures on spray behavior. The results demonstrate that increasing n-butanol content significantly alters spray and combustion characteristics. Higher n-butanol proportions lead to longer spray tip penetration and larger spray areas at higher injection pressures, while the cone angle remains relatively unchanged. The 90% n-butanol blend exhibited the most distinct differences from pure diesel. However, due to n-butanol’s high latent heat of vaporization, in-cylinder temperatures decreased, resulting in longer ignition delays. To mitigate this, a spark-assisted compression ignition (SACI) strategy was employed, with adjustable spark duration to assess its impact. Compared to pure diesel, SACI-applied n-butanol/diesel blends exhibited higher peak in-cylinder pressure and heat release rates, improving indicated thermal efficiency. Additionally, ringing intensity (RI) assessments confirmed that all tested conditions remained below the 5 MW/m2 threshold, ensuring acceptable combustion stability. This study provides a comprehensive analysis of n-butanol/diesel blends under SACI conditions, demonstrating their potential to enhance spray and combustion characteristics. The findings underscore n-butanol’s promise as a sustainable alternative fuel, addressing key challenges in dual-fuel combustion strategies.

Methanol is significantly emerging as a promising alternative fuel in the pursuit of carbon neutrality. This study aims to analyze the combustion characteristics of methanol in a spark-ignition (SI) engine operating under high compression ratios and ultra-lean conditions through both experimental and simulation approaches. The objective is to derive optimized combustion efficiency by employing various ignition strategies based on discharge energy. To this end, experiments were conducted using a Rapid Compression Expansion Machine (RCEM) to replicate realistic engine environments. The effects of discharge energy and spark duration across different spark coil configurations were investigated through both experimental methods and computational fluid dynamics (CFD) simulations. The experimental results showed that the use of multiple spark coils achieved an energy release rate of approximately 239 mJ/s, more than twice that of the single-coil configuration. Simulation results were in good agreement with the experimental findings. In the single-coil setup, combustion indicators such as in-cylinder pressure were significantly influenced by variations in spark duration. In contrast, the multi-coil configuration demonstrated higher peak values in several combustion characteristics, but once a certain threshold of discharge energy was exceeded, the improvements in performance became marginal. This suggests that simply increasing the spark duration is not always an effective strategy for optimizing the spark channel in methanol SI engines. Furthermore, the multi-coil ignition strategy significantly reduced the ignition delay, contributing to improved combustion stability under lean-burn conditions. As a result, the multi-coil strategy achieved an indicated thermal efficiency (ITE) approximately 10% higher than that of the single-coil approach, with the maximum efficiency observed at 30.4% in the 10-coil strategy.

Compact and low-weight structures perform an essential part in enhancing the capability of high-power proton exchange membrane fuel cells (PEMFC). The primary issue for this kind of metallic bipolar plate is that an arrangement of both plates provides three flow fields: an anode side, a cathode side, and a coolant. In this study, a decussate flow field (DFF) was proposed as an improvement over the conventional serpentine flow field (SFF) to enhance the mass transfer capabilities. The effect of the decussate flow channel configurations for both the DFF BPP and the SFF BPP on the performance of PEMFC was investigated using three-dimensional computational fluid dynamics ANSYS modelling. The depth of the channel of the DFF bipolar plate was varied in the range of 0.3 mm-0.75 mm to study the effect of dimension on the overall capability. The simulation outcomes demonstrated that with the rise of the depth of DFF bipolar plate promoted superior oxygen transportation and more efficient water removal compared to the SFF bipolar plate. The bipolar plate DFF with a 0.75 mm depth design shows a uniform dispersion of current density and temperature as compared to the SFF bipolar plate. The simulation also indicated that the PEMFC with the DFF bipolar plate demonstrated better electrochemical performance compared to the PEMFC with the SFF bipolar plate. The BPP with 0.75 mm depth DFF shows a power density with a maximum value of 508.24 mW/cm2, while SFF shows a power density with a maximum value of 395.56 mW/cm2. The aforementioned results offer an important path for designing decussate flow channels that can increase fuel cell power density and progress the creation of useful fuel cell technologies.