Aircraft engines are typical complex engineering systems with multidisciplinary cross-cutting and multi-component strong coupling. They work under high temperature, high pressure, high speed, and multi-field load/environment, and must meet extremely high requirements for use such as high thrust, light weight, long life, and high reliability. They are extreme products with great difficulty in development. As the core component of aircraft engines, the structural integrity and reliability of the rotor structure are the weakest links in aircraft engine design and the bottleneck restricting engine development.
On the one hand, the structural system is subjected to harsh and complex aerodynamic, mechanical and thermal loads, while mechanics (gas-heat-solid) interacts and constrains with materials, processes and other disciplines; on the other hand, the structural life shows great dispersion, while safe flight requires a low failure probability. At this time, traditional deterministic design technology faces many challenges. Probabilistic design, as a fine design method, can quantify risks, reduce weight and reduce costs while meeting reliability requirements. It is one of the most promising key technologies to solve the bottleneck of advanced aircraft engine development. Engine structural design development
Aeroengine structural design has gone through the development process of static strength design, safe life design, deterministic damage tolerance design and structural probabilistic design.
The main starting point of static strength design is that the structure does not break under the given design load; after the use load, there is no visible permanent deformation after unloading. For a long time in the past, due to the small engine load, the stress level of the structure was very low, and the requirements for the structural life were not high, and the static strength design could meet the design requirements.
In 1954, the British "Comet" jet airliner exploded and crashed into the sea. The cause of the accident was the fracture damage caused by the fatigue effect of the metal structure of the aircraft fuselage. This shows that designing structural parts according to static strength cannot guarantee their safety in use, and the problem of safe service life must be considered in structural design. After the accident, the safety life method was adopted in the structural design of aircraft engines.
The premise of safe life design is to assume that the structure is a continuous uniform body without defects. Because the material has a certain dispersion, the structural design in engineering often determines the average life of the structure through experiments, and combines experience to give a larger dispersion coefficient, and then obtains the allowable life of the structure.
The British Ministry of Defense standard (DEF STAN 00-971) requires that the curve with a reliability of 99.87% (i.e. -3σ) be taken as the S-N curve corresponding to the safe life. The American Aviation Turbojet and Turbofan Engine General Specification (MIL-E-5007D) stipulates that the typical structure of the engine adopts a safe life design, and the allowable strength and life characteristics of the material are based on -3σ, and the corresponding confidence level is 95%, see Figure 1. In terms of civil aircraft, the airworthiness clause FAR33.14 issued by the Federal Aviation Administration (FAA) of the United States requires that the life of rotor parts such as compressors and turbine disks be determined using the safe life design method; the airworthiness regulations JAR-E of the European Joint Aviation Authority also stipulate that the life of engine rotor parts must be determined by the safe life method.
The safe life design method takes into account the dispersion of the mechanical properties of the engine structural materials, and establishes a high reliability from a probabilistic perspective to ensure the safety of the structure during service. However, the actual structure must contain initial defects such as inclusions, holes (cracks), and cavities. Therefore, the safe life design method does not guarantee the safety and reliability of the structure;
In addition, the safe life design is a very conservative design method. Under the current 95% confidence level and 99.87% reliability, it is possible that only one of the 1,000 structures will have the specified cracks, and the other 999 will also be considered to have reached the fatigue life and withdraw from service, resulting in great waste. It is estimated that more than 80% of the discs designed using the safe life method still have a life of 10 times or more after being scrapped, so it is necessary to develop more economical design and evaluation methods.
Deterministic damage tolerance design can solve the safety and economy problems in engine structure design. This method is different from the safe life design method. It assumes that there are defects or small cracks in the material, and they will gradually expand under the action of fatigue loads, eventually leading to structural failure. Deterministic damage tolerance design determines the maintenance cycle and maximum allowable crack size of the structure based on the use conditions, material properties and maintenance requirements of the structure, controls the structural material defects and damage extension within a reasonable range [7], and effectively improves the economic efficiency of engine structure maintenance.
The U.S. Air Force Engine Structural Integrity Outline (MIL-STD-1783) clearly states that damage tolerance design, testing and control should be carried out on key engine fracture parts to determine the maintenance cycle of the structure. However, the deterministic damage tolerance design method is difficult to accurately consider the impact of various random factors on life dispersion, and cannot accurately predict the failure probability of the structure and the sensitivity of the input parameters.
The criteria, laws and methods studied and established in the above engine structure design process are all deterministic, that is, material properties, manufacturing processes, mission usage and other parameters use fixed values, which cannot accurately predict the fatigue life of the structure. Therefore, it is difficult to optimize and balance the structural weight, service life and design criteria, and it is impossible to give a quantitative reliability index. The life requirement with reliability is a prominent feature of modern engine design, and structural probabilistic design is an important design method to ensure structural reliability.
The structural probabilistic design method believes that the actual external load acting on the structure and the actual bearing capacity of the structure are both probabilistic quantities. It is impossible to give precise values during design. They are called random variables or random processes, which obey a certain distribution. Taking this as the starting point for structural design [10] can better conform to objective reality. In the research of the Integrated High Performance Turbine Engine Technology Program (IHPTET), the United States proposed a probabilistic design method and established a probabilistic design system (PDS) for the rotor system, which reduced the structural design reserve to an acceptable reliability level and ensured a relative balance between safety and working capacity, as shown in Figure 2. Following the IHPTET program, the Multipurpose Economical Advanced Turbine Engine Program (VAATE) launched the probabilistic high cycle fatigue of blades as one of the key technologies. The U.S. Engine Structural Integrity Program (MIL-HDBK-1783B) also pointed out that using probabilistic design margin instead of traditional safety factor or deterministic margin can more accurately represent the changes in component response. The FAA of the United States has promoted the application of structural probabilistic design in the field of civil aviation, replacing FAR33.14 with the airworthiness clause FAR33.70, that is, probabilistic damage tolerance design is carried out for engine life-limiting parts, and the failure risk level is determined by comparing with the design target risk.
In order to meet the performance requirements of advanced aircraft engines, more and more new materials such as powder alloys, intermetallic compounds, ceramic materials, composite materials, etc. are constantly emerging. These materials have obvious improvements over traditional metal alloys in certain properties such as strength, stress, density, etc., but there are many uncertain factors and new failure modes in their material properties, and the use of deterministic design methods will produce large deviations. Therefore, it is more necessary to adopt probabilistic design methods for typical engine structures under new materials, new structures, and new process systems. Key technologies for probabilistic design of typical engine structures
1. Probabilistic model of random factors (1) Load dispersion Aircraft engines, especially military engines, have to complete a large number of flight missions such as cruising, simple stunts, complex stunts, formations, attack and interception, resulting in a large dispersion of the working loads of the engine structure, such as mechanical loads, aerodynamic loads, and temperature loads. The establishment of the engine load probability model needs to be combined with load spectrum processing technology, and the damage conversion principle such as engine load spectrum and structural load spectrum should be studied. By collecting and processing the engine field flight load spectrum and test spectrum, the load variation law of the typical engine structure over time is obtained, and the statistical theory is used to establish the working load probability statistical characteristics of the typical engine structure.
(2) Material dispersion The differences in the preparation process of engine structural materials lead to the dispersion of material mechanical performance parameters (such as elastic model, Poisson's ratio, stress-strain curve, etc.); at the same time, due to the complexity of the processing technology, the material properties of different parts of the typical engine structure are also significantly different. Therefore, it is necessary to combine the test data of standard material specimens and characteristic simulation parts of different parts of the structure to establish a probability model verified by experiments.
For example, to deal with the dispersion of the material cyclic stress-strain relationship, the strain can be decomposed into elastic strain components and plastic strain components, and linear equations of elastic strain components, plastic strain components and stress are established respectively. The above equations are randomized and the probability model of cyclic stress-strain is established by regression method, as shown in Figure 3. (3) Process dispersion Process dispersion refers to the structural dispersion caused by random factors such as material preparation, processing methods, and detection technology.
Initial defect randomness. For material preparation methods such as forging, casting, and hot isostatic pressing, the number, position, and size of initial defects such as inclusions, holes, and cavities in the material are all random, as shown in Figure 4. Therefore, it is necessary to develop equivalent damage characterization methods for surface/subsurface/internal defects and establish a defect probability model for typical engine structures in different regions. Geometric dimension randomness. The geometric dimensions of typical engine structures are random due to the instability of mechanical processing technology. To this end, it is necessary to determine the key geometric dimension parameters of typical engine structures through sensitivity analysis, and to conduct statistics on the key geometric dimensions to establish their probability distribution characteristics.
Residual stress randomness. The instability of processing technology such as engine structure surface treatment technology (such as shot peening, laser strengthening), heat treatment technology (such as solid solution, aging) and other processes (such as linear friction welding, electron beam welding) leads to random residual stress on the surface and inside of the structure. The probability model of residual stress needs to be systematically studied through numerical simulation, process inspection and mathematical statistics.
Randomness of defect detection technology. Different detection technologies lead to randomness in the size and position of the minimum detectable defect. The probability of defect detection (POD) is introduced to describe this randomness. By preparing damaged specimens with sufficient defect types, sizes and positions, conducting defect detection tests and establishing a mathematical model of the detection data, the POD curve of the detection technology is obtained by statistical processing.
2. Typical engine structure Typical engine structures (such as turbine blade structures) are subjected to complex alternating loads/environments of gas-heat-solid multi-fields, and their failure modes are mostly composite fatigue failures, such as fatigue-creep, thermomechanical fatigue, high-low cycle composite fatigue, etc. Therefore, the composite fatigue failure probability model of typical engine structures under complex loads/environments based on experiments has become a key technology for the probabilistic design of engine structures. It is necessary to closely combine theoretical research with experimental verification to reveal the mechanism of the composite failure mode of typical engine structures under complex loads/environments and develop corresponding composite fatigue life prediction methods. Combined with the dispersion of loads, materials, and processes, a composite fatigue failure probability model of typical engine structures that has been verified by experiments is established.
The Structural Strength and Multidisciplinary Optimization Research Group of Beijing University of Aeronautics and Astronautics takes the high-pressure turbine disk of the WP7 engine (made of GH33) as the object to study and develop a set of probability models and design methods for fatigue-creep coupling failure modes of turbine disks that have been verified by experiments, including two damage stages of crack formation and crack propagation, from material properties to design applications (Figure 5). 3. Structural Probabilistic Response Analysis Technology
The probabilistic design of typical engine structures involves many disciplines and complex structures. Directly using numerical methods to solve the structural failure probability/reliability will result in too long calculation time. Therefore, establishing a structural probabilistic response analysis method is a key technology for the probabilistic design of typical engine structures.
At present, the response surface method combined with the Monte Carlo method is often used to perform structural probability response analysis, and a set of high-efficiency and high-precision probability response analysis methods suitable for typical engine structures are developed to solve the problem of massive calculation in structural probability analysis. Sampling methods such as Latin hypercube method, quasi-Monte Carlo method, Latin centroid sampling, and important sampling can be used; approximate modeling methods such as neural network, Kriging model, and support vector machine can be selected (Figure 6).
4. Random factor sensitivity analysis
Through the sensitivity analysis of random factors, on the one hand, the main factors affecting the dispersion of structural life can be determined, providing a design basis for the probabilistic design of typical engine structures; on the other hand, random variables with low sensitivity indicators in structural probability design can be processed into deterministic parameters, thereby saving calculation time and improving solution efficiency; focusing on the dispersion of the main random factors is more direct and significant than other factors in improving the solution accuracy of structural failure probability/reliability.
5. Engine typical structure probability design system Aiming at my country's aviation engine material process system, establishing an aviation engine typical structure probability design platform with independent intellectual property rights is also a key technology for engine structure probability design. The realization of the engine structure probabilistic design platform requires a large number of iterative calculations. Each change in the geometric model during the calculation process will cause the analysis data such as the mesh and boundary conditions to change accordingly. Therefore, the connection technology between the geometric model and the analysis model, the automatic reconstruction of the mesh and the automatic loading of the mesh are technical difficulties that need to be solved.
For example, the structural strength and multidisciplinary optimization research group of Beihang University uses virtual blocks to replace complex geometric structures to establish flexible multi-block meshes, introduces the constraints of follower points to achieve the mesh reconstruction of the geometric structure in the optimization iteration, and solves the high-precision mesh problem in the multidisciplinary optimization (MDO) process of turbine blades, as shown in Figure 7. With the support of the National Defense 973 Program, the structural strength and multidisciplinary optimization research group of Beihang University proposed and established a turbine disc probabilistic design process based on the deterministic design process of turbine discs, and initially established the first turbine disc structural probabilistic design system (TPDS) in China. It includes the main functions of load spectrum processing, random variable processing, geometric modeling, failure mode analysis and design decision-making, and integrates commercial software such as UG, ANSYS, and MATLAB, as shown in Figure 8. Several aspects that need to be further studied
The probabilistic design technology of typical engine structures has become one of the most promising key technologies to solve the bottleneck of advanced engine research and development. It is gradually being used in the development of key parts of advanced aero-engines in China. It has been applied locally and the effect is beginning to show. In the next step of development, the following research work needs to be deepened and strengthened.
1. Material-process-structure probabilistic integrated design Structural design, material, and process integration and iteration are the fundamental paths to improve the level of advanced aero-engine development. The effective application of engine structural probabilistic design technology is inseparable from the integrated consideration of materials and processes. Therefore, the probabilistic design of engine structures needs to start from the source of design, reveal the influence mechanism and regular regulation of typical structural manufacturing sensitive parameters on the reliability of engine structures, and form a probabilistic design method of typical engine structures that integrates manufacturing sensitive parameters, thereby establishing the theory and method of engine material-process-structure probabilistic integrated design.
2. Multidisciplinary optimization design based on reliability The structural design of aero-engines involves many disciplines such as aerodynamics, materials, heat transfer, structure, and strength. The disciplines are coupled with each other. The probabilistic design of a single discipline cannot guarantee that the designed structure meets the reliability requirements of use.





