15 October 2021, Volume 61 Issue 10

    

• Select all
  |

FUEL CELLS AND LITHIUMION BATTERIES
• Select
  Advances in metal bipolar plates and coatings for PEM fuel cells

  PEI Pucheng, LI Zizhao, REN Peng, CHEN Dongfang, WANG Xizhong

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1025-1038. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.026

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Metal bipolar plates for proton exchange membrane (PEM) fuel cells have price and processing advantages, but easily corrode which reduces the fuel cell electrical conductivity and durability. This paper reviews recent advances related to the electrical conductivity and durability of metal bipolar plates and their coatings for fuel cells. The review starts from the market demand for PEM fuel cell stacks and the advantages of using metal bipolar plates in PEM fuel cells. Then, the typical metal bipolar plate and coating testing methods are described. Most coatings besides alloy coatings meet the US Department of Energy 2020 target. Then, this paper describes the environmental factors and working conditions that influence the plate electrical conductivity and durability. Finally, this study identifies future research prospects related to test methods, coatings research, and key factors. This paper reviews key advances in metal bipolar plates and their coatings for improving their efficiency and durability for fuel cell stacks.
  
• Select
  Reinforced high-performance membrane electrode assembly for proton exchange membrane fuel cell prepared via direct membrane deposition

  LI Xue, ZHANG Hong, LIN Cheng, WANG Shubo, XIE Xiaofeng

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1039-1045. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.028

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The performance of proton exchange membrane fuel cells is directly related to the properties of the membrane electrode assembly which can be significantly improved by ultra-thin proton exchange membranes. In this study, electrospinning and direct membrane deposition were used to prepare poly(arylene ether sulfone) nanofiber reinforced proton exchange membranes. The resulting composite membrane was very thin, about 13 μm. The fuel cell peak power density with H₂/air with this membrane of 1.18 W/cm² was 23% higher than with a commercial membrane of similar thickness (Nafion^® NC700-CCM), 0.96 W/cm². By optimizing the catalyst ink composition and using gas diffusion layer with better hydrophobicity, the fuel cell peak power density increased to 3.55 W/cm² for H₂/O₂. The relative humidity at the cathode had little influence on the fuel cell performance. Hence, the membrane electrode assembly reinforced with a poly(arylene ether sulfone) nanofiber layer provides excellent fuel cell performance, even with low humidities for fuel cell applications.
• Select
  Wavy channels to enhance the performance of proton exchange membrane fuel cells

  LI Zijun, WANG Shubo, LI Weiwei, ZHU Tong, XIE Xiaofeng

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1046-1054. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.029

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The performance and durability of proton exchange membrane fuel cells (PEMFCs) are limited by factors related to fuel delivery and water management with the flow channel significantly affecting these factors. This study investigated flow in a three-dimensional wavy channel using a single straight channel model consistent with the experimental conditions. The model was used to analyze the performance of PEMFCs with straight channels and wavy channels and the oxygen, liquid water, velocity and current density distributions in the two channels. The results show that for high current densities, the three-dimensional wavy channel enhances the oxygen transfer from the narrow channel to the catalytic layer, improves the oxygen supply, effectively removes the liquid water in the channel, and increases the peak power density by 10.16%.
• Select
  Comparison study of three porous electrode models for the forward design of lithium-ion batteries

  FANG Ruqing, ZHANG Na, LI Zhe

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1055-1065. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.032

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The electrodes of lithium ion batteries are composed of active material particles, binders, conductive additives and pores filled with electrolyte. Optimizing the porous structure of the electrode effectively improves the ionic and electronic transport inside the cell which improves the battery power output. Many previous designs have used the conventional trial-and-error method while the forward design method based on an electrode model is gradually being more widely used in the industry. However, previous electrode models required excessive computations to get the desired prediction accuracy. This paper presents an extended homogeneous porous electrode model for lithium-ion batteries which balances the computational cost and the prediction accuracy. This study compares the computational costs and the electrode modeling accuracy of three electrode models, a traditional volume-averaged pseudo 2D model, a 2D heterogeneous particle-packing model from the previous work of the authors' group, and the extended homogeneous model of this study. Then, this paper discusses how these three electrode models can be used for the forward design of lithium-ion battery electrodes. Finally, the extended homogeneous model is used to analyze the influence of the particle size distribution in the negative electrodes on the cell rate performance with improvements of up to 25% for low C-rates to 100% for high C-rates produced by narrowing the particle size distribution and reducing the particle size.
  
POWER SYSTEMS
• Select
  Dynamic analysis and multi-objective parameter optimization in multi-receiver wireless power transfer systems

  TAN Tian, CHEN Kainan, LIN Qiuqiong, JIANG Ye, ZHAO Zhengming

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1066-1078. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.022

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Wireless power transfer (WPT) can improve power supply safety, reliability and flexibility. Multi-receiver WPT systems have many advantages as high power, contactless power supplies for electric vehicle charging, rail transit and other systems. However, dynamic analysis and optimization methods are not well developed for multi-receiver WPT systems. This paper presents a dynamic model for a typical multi-receiver WPT system to analyze the effects of the system parameters on the steady-state and dynamic responses. Then, a multi-objective parameter optimization method is developed for the system based on the dynamic model. Simulations and experiments show that the dynamic model and optimization method are accurate and effective.
• Select
  Improvement of SiC MOSFET based converters with self-regulating drive

  LING Yatao, ZHAO Zhengming

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1079-1087. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.023

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The active gate drive (AGD) method has been developed to optimize the SiC metal-oxide-semiconductor field-effect transistor (MOSFET) switching behavior by controlling the gate side. However, verification of current AGD methods has been limited to double-pulse tests, which fail to show practical values for AGD methods in converters and fail to show the AGD capabilities for continuous operations. The authors have presented an AGD method targeting insulted gate bipolar transistors (IGBTs). This paper ajusts this AGD method for SiC MOSFETs with parameter compensation. The accuracy of this modified AGD method is then compared with conventional gate drives (CGD) and existing AGD methods via double-pulse tests. In the double-pulse tests, this AGD method reduces the turn-on delay and its loss by 53% and 30%, and reduces the turn-off delay and its loss by 65% and 67%. This AGD method is then applied to a Buck converter to verify how it improves the switching spikes and converter efficiency. For a given spike, this AGD method improves the converter efficiency by 0.3% to 1% compared with the CGD method.
• Select
  Calculation method for parasitic capacitance of high-frequency transformers

  XU Wei, ZHAO Zhengming, JIANG Qirong

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1088-1096. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.021

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  High-power multi-port converters can have high-frequency oscillations that seriously threaten system reliability, with the parasitic capacitance of the high-frequency transformers (HFTs) being one of the important factors. This paper summarizes the existing methods for calculating the parasitic capacitance of solid wires and litz wires and presents a piecewise solution method based on the electric field lines between solid wires to calculate the HFT parasitic capacitance. This study also corrects the shortcomings of the existing analysis method for litz wires. The analytical method for calculating the parasitic capacitance for typical C-type and Z-type winding arrangements and arbitrary multi-turn, multi-layer winding arrangements is re-derived taking into account the influence of the isolation layer in high-voltage, high-capacity transformer windings. Comparisons with finite element predictions and experimental data verify the accuracy of this method as a foundation for calculating the parasitic capacitance of HFTs for system design and analyses.
• Select
  Low voltage ride through function for the high voltage AC ports of four-port power electronic transformers

  NIE Haozhe, SHEN Yu, ZHAO Zhengming, WEN Wusong, YUAN Liqiang

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1097-1105. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.035

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The widespread use of power electronic transformers in power systems has led to more research on their operational safety and stability. The voltage sag in the alternating current (AC) port of the power electronic transformer caused by a short circuit fault of the grid connection point can lead to serious instabilities which can slow the power grid recovery. The instability mechanism of the high voltage AC port of a modular four-port power electronic transformer is analyzed in terms of the energy characteristics when a voltage sag fault occurs. Then, a low voltage ride through control scheme is given to maintain safe, stable operation of the power electronic transformer during the fault. This scheme provides a low voltage ride through function by switching the power balance port and the power electronic transformer control strategy. Finally, this low voltage ride through control scheme is verified using DSIM simulations which shows that the scheme effectively improves the safety and stability of the power electronic transformer and provides low voltage ride through in multi-port power electronic equipment.
MECHANICAL ENGINEERING
• Select
  Motor servo control parameter tuning for parallel and hybrid machine tools based on a genetic algorithm

  WANG Liping, KONG Xiangyu, YU Guang

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1106-1114. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.009

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The motor servo control of machine tools influences the processing efficiency and processing quality of the machine tool. Most servo motor controllers use the proportional integral derivative (PID) control which depends on accurate determination of the control parameters. However, accurate determination of the control parameters requires much time, effort and experience with manual tuning usually required to provide the desired accuracy. This paper presents a method for tuning servo control parameters based on a genetic algorithm. A theoretical model is developed and refined based on the machine tool conditions. The fitness function is developed based on the parallel and hybrid machine tool characteristics with alarm indexes and optimal indexes. The servo control parameters given by the genetic algorithm are then tested on a 5-axis hybrid machine tool. The results show that the optimized parameters give better servo motor following accuracy than the engineer's parameters.Thus, this parameter tuning method based on a genetic algorithm saves time and effort while giving more accurate servo control parameters.
• Select
  Optimization of servo matching for a five-axis machine tool based on the RTCP function

  CHEN Yanyu, GUAN Liwen, CHANG Jiahao, HU Lan, WANG Linquan

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1115-1123. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.007

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The performance of a five-axis machine tool not only depends on the dynamic characteristics of each single-axis servo feed system, but also how well the dynamic characteristics of each axis are matched. An optimization method is developed for A/C axes included precision five-axis machine tools with the rotation tool center point (RTCP) function using multi-axis servo matching based on roundness testing to improve the overall machine accuracy. This method uses matched single-axis servo feed systems to reduce the cost of building each single-axis electromechanical coupling system while ensuring the dynamic accuracy. A kinematics analysis of the machine tool structure is used to build an electromechanical coupling model for each axis servo feed system that is implemented in SIMULINK. Then, a specific trajectory is designed for the RTCP function to analyze the motion relationships for each axis for the trajectory. The method is evaluated by selecting matching single-axis servo feed systems for servo matching between three linear axes of X, Y, and Z based on the roundness test method. The SIMULINK model is then used to compare the dynamic performance of the machine tool before and after optimization for the specific trajectory. The results show that this optimization method improves the machine dynamics with little cost for each single-axis servo feed system.
• Select
  Servo stiffness optimization of torque motor direct-drive CNC machine tool feed systems

  CHEN Yanyu, FU Meng, GUAN Liwen, CHANG Jiahao, DIAO Lei, HU Lan

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1124-1131. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.019

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The stiffness characteristics of feed systems directly driven by torque motors mainly depend on the servo stiffness. Therefore, the servo stiffness is a key factor affecting the feed system disturbance rejection, dynamics, speed and accuracy. This study analyzes a torque motor direct-drive feed system using a servo stiffness optimization method based on the Routh criterion. The optimization simultaneously sets multiple servo control parameters based on the mutual constraint relationships between the servo control parameters for a stable control system. The analysis uses a complete coupled electromechanical model of the servo feed system based on the direct-drive torque motor model. Then, the influence of the servo system control parameters on the servo stiffness is used to develop a simplified solution method for the system stability Routh criterion inequality and to define the constraint relationships between the system control parameters. The method is verified in Simulink. This research gives a servo control parameter optimization method using the servo stiffness as the optimization objective. The method is then applied to an A-axis direct-drive torque motor feed system in a five-axis computer numerical control (CNC) machine tool to verify the optimization method. The results show that the optimization significantly improves the servo stiffness while maintaining system stability and has little effect on the dynamics of the servo system such as the dynamic tracking accuracy.
• Select
  Wall climbing welding robot for automatic welding of spherical tanks

  FENG Xiaobing, PAN Jiluan, GAO Lisheng, TIAN Wei, WEI Ran, PAN Baiwa, CHEN Yong, CHEN Suyun

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1132-1143. https://doi.org/10.16511/j.cnki.qhdxxb.2020.26.037

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Automatic welding of large steel structures, especially three-dimensional spherical surfaces, requires a mechanism that can walk on the spherical surface and a stable welding tracking control system. This paper describes a wall-climbing robot that combines a crawler with a wheel to balance the required flexibility of the wall-climbing mechanism and the load capacity. The robot can automatically identify the weld position through a visual tracking system and intelligently determine the weld path to enable autonomous movement and tracking of the weld. The flexible, trackless crawler body adapts to large curvature changes in the spherical tank and can walk freely on the tank surface to provide efficient automatic welding of spherical tanks.
• Select
  Analyses of preload variations in aeroengine rotor bolted connections

  ZHAO Bing, ZHANG Shouyang, WANG Hui

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1144-1151. https://doi.org/10.16511/j.cnki.qhdxxb.2020.22.035

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The preload consistency of bolted connections plays an important role in ensuring the safety operation of aeroengine rotors. A model was developed here for the torsional tension relationship with a self-locking torque of closing to improve bolt tightening control of flanged joints of aeroengine rotor assemblies. The model was used to investigate the influence of changes introduced by the self-locking torque attenuation for various processing conditions. Tests were then used to verify the model. The results show that the remained self-locking torque is reduced by repeated use of the tightening nut resulting in some loss of the preload on the flange. In addition, the lubrication conditions greatly impact the preload. Also, repeated use of the tightening nuts changes the preload along a known trend and different batches of bolts have different preloads. This study shows the key factors affecting the consistency of preloads of aeroengine rotor bolted connections and how to accurately evaluate the preload variations caused by each factor as a reference for optimizing the bolt processing.
• Select
  Walking speed estimation based on empirical mode decomposition method and swing foot parameters

  WANG Wei, YANG Kaiming, ZHU Yu, QIAN Yuyang

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1152-1158. https://doi.org/10.16511/j.cnki.qhdxxb.2021.21.029

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Predicting walking speed is an effective way to improve the functionality of a feedback-controlled treadmill. A walking speed estimation method is proposed as a solution to this problem based on swing foot parameters using the empirical mode decomposition (EMD) method. The relationship between the user's average swing foot velocity and real walking speed is declared by theoretical derivation. To solve the problem of integral drift and accumulated error during acceleration double integration, the EMD method is used to decompose the acceleration signal into signals of different frequencies. After removing the drifting component, the velocity signal is reconstructed from signals whose frequencies are close to the intrinsic gait frequency. For the modal aliasing phenomenon caused by integral accumulation error, the ensemble EMD method is used to reconstruct the swing foot displacement signal. Based on the heel-strike and toe-off events, the period and foot displacements during the swing phase can be obtained. Thus, the average swing foot velocity is then calculated for walking speed estimation. The proposed method is tested under five different walking speed conditions and compared with the traditional double-integral method. Experimental results prove the effectiveness and benefit of the proposed method.
AUTOMATION
• Select
  A syntax analysis method of PLC instruction list program and its application in static testing

  LI Xueliang, ZHAO Qianchuan, YANG Wen, Syed Naeem HAIDER

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1159-1165. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.012

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Programmable logic controllers (PLC) are automatic controllers widely used for industrial control. Because PLC program testing is lack of general testing tools, developers can only manually check code syntax errors, which is inefficient. Thus, industrial users need a standardized PLC code testing method to automatically complete the PLC program syntax detection and analysis. This paper presents an instruction list syntax model based on the Backus-Naur form (BNF) and the IEC61131-3 standard for industrial users. A PLC code syntax fault detection algorithm with linear complexity is then built with an abstract syntax tree to automatically check the PLC code syntax. Compared wich the CODESYS Static Analysis, analysis of two industrial PLC programs demonstrates the usability of this PLC code static analysis method.
• Select
  Rolling scheduling of reheating furnaces in steel production when some furnaces may not be available at times

  QIAN Peng, WU Cheng, SONG Shiji, DING Jianya

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1166-1176. https://doi.org/10.16511/j.cnki.qhdxxb.2020.26.036

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  A model is presented here for scheduling of reheating furnaces that are used after continuous casting and before hot-rolling when some of the furnaces may not be available at certain times which leads to constraints on the scheduling. The constraints that occur in actual production lines are modeled in a mixed-integer programming model to minimize the combined costs of the adjusted schedule, select the transport path for each job and determine the furnace allocations. A heuristic algorithm for rolling scheduling is used to update the model and decompose the problem with partial reuse of existing results to improve the dynamic scheduling efficiency. Simulations show that the algorithm can provide real-time and effective scheduling.
COMPUTER SCIENCE AND TECHNOLOGY
• Select
  IPv6 active address discovery algorithm based on multi-level classification and space modeling

  LI Guo, HE Lin, SONG Guanglei, WANG Zhiliang, YANG Jiahai, LIN Jinlei, GAO Hao

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1177-1185. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.017

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The enormous IPv6 address space makes it impossible to apply traditional IPv4 brute-force scanning for IPv6 active address discovery. This paper presents an IPv6 address discovery algorithm based on multi-level classification and space modeling. The multi-level classification algorithm uses multi-dimensional information for fine-grained division of the seed addresses. The space modeling uses four representation strategies to model any address set with pattern representation used to balance the low detection efficiency caused by the large modeling space and the sample error caused by the small modeling space. New active IPv6 addresses can be discovered by heuristic traversal of the pattern representation. Tests show that this address discovery algorithm has a higher hit rate than previous methods and verifies that the fine-grained division of the seed address improves the hit rate of the address discovery algorithm.
CONSTRUCTION MANAGEMENT
• Select
  Improved experimental measurement method of construction worker fatigue and unsafe behavior

  GUAN Zhongyao, XIANG Tian, FANG Dongping, GUO Hongling

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1186-1194. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.030

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Construction accidents can be reduced by measuring worker fatigue and the relationship to unsafe behavior. However, previous research has mainly used subjective instruments such as subjective rating scale or questionnaires to collect worker fatigue and behavior data, which are highly subjective. This study improves an existing safety experiment method by analyzing the limitations of the existing safety experiment method and identifying worker physical and mental fatigue. A trial experiment to verify the effectivity and feasibility of the improved experimental method shows that this improved method better reflects the worker conditions and causes fatigue. This study also identifies key physiological indicators related to worker fatigue to support future research in this field.
• Select
  EPC project owner and contractor design decision analysis based on game theory

  ZHAO Yue, QIANG Maoshan, WANG Hao

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1195-1201. https://doi.org/10.16511/j.cnki.qhdxxb.2021.22.027

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  The engineering-procurement-construction (EPC) method is frequently used in construction projects because this method integrates all aspects of design and construction into one process. This method differs from the design-bid-build(DBB)method in which the design and construction steps are sequential. An important factor for smooth and effective implementation of an EPC project is how involved is the owner early in the design process. This study used game theory to analyze the effects of the owner's level of involvement and the contractor's design optimization decisions in an EPC project with a numerical analysis of the more applicable two-sided optimization strategy for engineering practice. The results show that increasing the profit and cost ratio in the optimization process reduces the importance of the owner's level of involvement in the design process. For profit and cost ratios less than 1.2, the owner being heavily involved in the design has an optimal compatibility with a specific proportion of the risk assumed by the owner while for ratios greater than 1.2, the importance of the owner's level of involvement in the design process increases with increasing proportion of the risk assumed by the owner. Overall, the owner having limited involvement in the design with an design optimization by the contractor is the most advantageous strategy. Practical suggestions are given to help the industry better understand and more effectively use EPC.
CIVIL ENGINEERING
• Select
  Three-dimensional laser profilometer survey system of pavement slip characteristics

  XU Li, GUO Runhua, PENG Huiting, SHI Pengcheng

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1202-1211. https://doi.org/10.16511/j.cnki.qhdxxb.2021.25.001

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  A 3D laser profilometer was used to measure the average height (R_a), average section depth (MPD), average structure depth (MTD), root mean square height (R_q), peak volume (V_mp), wavelength (λ), and two-dimensional power spectral density (PSD) of eight asphalt pavements. The results show that the anti-slip characteristics correlate with R_a, MPD, and MTD with over 80% correlation. PSD decreases with increasing wave vector q, which can also be used to independently evaluate the anti-skid characteristics of the pavement. The micro- and macro-texture wavelengths of the AC13 and AC16 series samples mainly range from 0-3 mm, with the texture band below 1 mm correlating positively with the anti-slip characteristics with up to 80% correlation, with less correlation in the 1-1.25 mm band, and even less in the 1.25-2.5 mm band. Rutting reduces the roughness as seen in the lower PSD than for the original pavement.
AEROSPACE ENGINEERING
• Select
  Optimization of a micro turbojet engine combustion chamber

  LI Dongjie, ZHOU Bohao, LIANG Qian, LAN Xudong

  Journal of Tsinghua University(Science and Technology). 2021, 61(10): 1212-1220. https://doi.org/10.16511/j.cnki.qhdxxb.2020.22.034

  Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save

  Micro turbojet engine combustor designs cannot be directly scaled from large aero-engine combustion chamber designs due to the different fuel supply modes and inflow conditions that affect the micro turbojet combustor efficiency. Bad designs lead to high wall temperatures, poor combustion stability and nonuniform outlet temperature distributions. Both ANSYS and CFX were used in this study to analyze the air flow and combustion characteristics in the combustion chamber of a micro turbojet engine. The numerical simulations were validated by comparisons with cold state tests. A "gas swirling liquid" design was then developed to improve the combustion. The numerical simulation results show that this design reduces the combustion chamber outlet temperature distribution coefficient from the traditional 0.54 to 0.15, which greatly improves the combustion chamber and turbine efficiencies and the thermal efficiency, and the thrust of the micro turbojet.