LEE- Uçak ve Uzay Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 5 / 21
-
ÖgeDevelopment of a nonlinear sonic boom propagation code(Graduate School, 2023-01-24)Civil supersonic flight is still one of the most challenging research topics in the aerospace industry. Since Concorde's last flight in 2003, researchers tried to find efficient solutions to make supersonic flights more affordable and reliable. Meanwhile, with the advance of computational power, computational fluid dynamics (CFD) has been implemented in advanced optimization studies involved in elevating supersonic aircraft design processes with given operational criteria and requirements. However, reducing the cost of a supersonic flight by increasing aerodynamic efficiency is not the only concern in civil supersonic transport. The second most important factor for a supersonic aircraft is the noise produced on land due to the shock waves that propagate through the atmosphere to the ground. This phenomenon is called sonic boom which is addressed in this thesis study. A sonic boom generated by a supersonic aircraft can cause very loud noise on the ground that may exceed 100 decibels. This loudness value is not acceptable due to its effects on people's daily life. Therefore, to enable civil supersonic flight over land, sonic boom loudness must be eliminated or reduced below a certain level. This effort is called sonic boom minimization and there are several methodologies that are provided in this study. Lots of studies for sonic boom minimization utilize optimization algorithms that call sonic boom prediction tools along with the CFD solvers. Therefore, to reduce sonic boom loudness, a sonic boom propagation code that accurately predicts sonic boom loudness is essential for the multidisciplinary design optimization of civil supersonic aircraft. In this regard, a new nonlinear sonic boom prediction code, named ITUBOOM, is developed in-house to be incorporated into our design optimization studies to achieve a low-boom aircraft geometry. ITUBOOM is developed in Python programming language for ease of implementation for design studies. A sonic boom calculation process can be broken down into three main steps; a near-field solution with CFD to generate an initial acoustic signal, atmospheric propagation with acoustics methods, and loudness calculation. Unlike other sonic boom codes, ITUBOOM can also be used to generate a near-field pressure directly from CFD outputs by surface slicing or in-flow signature extraction. Then, it can be used to perform atmospheric propagation by taking into account nonlinear effects such as molecular relaxation and thermoviscous attenuation. Results of ITUBOOM are validated against NASA Langley Research Center's well-known sBOOM code for different conditions on benchmark problems and presented in this thesis in detail.
-
ÖgeScalable planning and learning framework development for swarm-to-swarm engagement problems with reinforcement learning(Graduate School, 2022-12-20)Development of guidance, navigation and control frameworks/algorithms for swarms attracted significant attention in recent years. Since existing conventional aerial defense systems are optimized for a small number of heavy-hitting adversaries such as cruise missiles or fighter aircraft, these systems are often in a critical disadvantage against large-scale aerial swarm attacks that cover a wide area. Thus, defending against the aerial swarm attacks is one of That being said, algorithms for planning swarm allocations/trajectories for engaging with enemy swarms is largely an understudied problem. Although small-scale scenarios can be addressed with tools from differential game theory, existing approaches fail to scale for large-scale multi-agent pursuit evasion (PE) scenarios. To solve this problem, two main approaches are presented in this study. First, a reinforcement learning (RL) framework that controls the density of a large-scale swarm for engaging with adversarial swarm attacks is proposed. Although there is a significant amount of existing work in applying artificial intelligence methods to swarm control, analysis of interactions between two adversarial swarms is a rather understudied area. Most of the existing work in this subject develop strategies by making hard assumptions regarding the strategy and dynamics of the adversarial swarm. The main contribution is the formulation of the swarm to swarm engagement problem as a Markov Decision Process and development of RL algorithms that can compute engagement strategies without the knowledge of strategy/dynamics of the adversarial swarm. Simulation results show that the developed framework can handle a wide array of large-scale engagement scenarios in an efficient manner. Secondly, a reinforcement learning (RL) based framework to decompose to large-scale swarm engagement problems into a number of independent multi-agent pursuit-evasion games is proposed. Variety of multi-agent PE scenarios are simulated, where finite time capture is guaranteed under certain conditions. The calculated PE statistics are provided as a reward signal to the high level allocation layer, which uses an RL algorithm to allocate controlled swarm units to eliminate enemy swarm units with maximum efficiency. This approach is verified in large-scale swarm-to-swarm engagement simulations.
-
ÖgeAeroacoustic investigations for a refrigerator air duct and flow systems(Graduate School, 2022-02-16)Noise has become an important public health problem with industrialization, and has become a crucial design problem for engineering. For this reason, noise reduction studies have became the focus, especially in the white goods, automotive and aviation sectors, which requires interaction with human. Among the vehicles and products in the aforementioned sectors, the refrigerators, unlike the others, are located in the center of the living area and work throughout the day. Therefore, possible sound problems are observed more quickly by the users and are found to be disturbing. At this point, the investigation and reduction of the acoustic propagation of existing products by various numerical and experimental methods is a valuable contribution to both industry and literature. Within the scope of this thesis, the freezer compartment of a refrigerator with a No frost cooling system was investigated from an aeroacoustic perspective. The freezer compartment consists of three drawers where food will be placed, an axial fan that provides air flow, an evaporator cover that separates the evaporator pipes and the interior volume, and plastic walls surrounding them. The main source of air flow noise in the system is the axial fan. For this reason, in the first step of the study, solo aeroacoustic examination of the axial fan was made. Afterwards, the entire freezer volume was examined and the study was completed with three different model proposals in which acoustic emission was reduced. The flow field analysis of the axial fan with an operational speed of 1200 rpm was carried out with commercial software ANSYS Fluent. In this numerical model, Shear Stress Transport 𝑘 – 𝜔 turbulence model was used. Governing equations was solved under three-dimensional, transient, viscous, incompressible flow assumptions. The rotation of the fan was defined by the sliding mesh method. The numerical flow solution was validated with experimental volumetric flow rate data. According to the numerical and experimental results, the flow rate of the axial fan under the specified conditions was determined as 19 L/s. A hybrid aeroacoustic model is created by giving the pressure outputs of the flow solution as input to the acoustic model. For the acoustic solution, Ffowcs Williams & Hawkings (FW-H) model defined in ANSYS Fluent was used and the result of the solution was compared with the sound pressure data collected in the full anechoic acoustic room. Although there is some difference between the numerical and experimental sound pressure curves, it was observed that the hybrid model established to understand the general trend and to catch the blade passing frequency was successful. It was predicted that the difference between experimental and numerical measurements occurred for two reasons. The first is absence of the fan motor in the numerical analysis. Another reason is that the acoustic propagation resulting from the excitation of the air flow to the system structures cannot be predicted with this model. In the second step of the study, the model validated with axial fan solutions was applied to the freezer compartment. The aim here is to reveal the air flow distribution in the freezer volume and to identify the regions where turbulence effects increase. In the numerical model, the axial fan was rotated at an operational speed of 1200 rpm and this rotation was achieved by the sliding mesh method. As a result of the analysis, it was seen that the turbulence formation started at the wing tips as observed in the solo fan analyses, and the vortices coming out of the trailing edge tips were especially concentrated in the region between the upper wall of the freezer volume and the upper two drawers. In addition, a turbulent area was detected at the bottom of the evaporator cover (which is the fan suction area). As a result of the hybrid aeroacoustic model solution, the sound pressure data collected from 1 meter away from the front, rear and side surfaces of the freezer and the sound pressure data collected from the same locations in the full anechoic acoustic room were compared. When the total sound pressure in the range of 10-10000 Hz is compared, it is seen that there is a difference of 3-7 dBA between the numerical model and the experimental results. As a result of the investigations of the axial fan in the solo and freezer volume, three different freezer models have been proposed to improve air flow, reduce turbulence and reduce the resulting noise caused by air flow. In the fist suggested model, the bottom part of the evaporator cover has changed and the acostic propagation has decreased 0.24 dBA at 1200 rpm rotational speed. The position of the axial fan and its distance from the structures in the suction and discharge directions are the parameters affecting the acoustic propagation. In the second model, it is aimed to provide acoustic gain by changing the fan position. In this context, the fan was moved on the shaft by 5 mm and brought closer to the blowing region. With this modification, total sound power level was decreased 2.18 dBA. The final model is the superposition of the first two models. Here, it was aimed to see the combined effect of two mentioned model. At 1200 rpm rotational speed, 3.27 dBA gain was achived by the third model.
-
ÖgeFailure analysis of adhesively bonded cfrp joints(Graduate School, 2021-01-04)Joints are critical areas where load transfer occurs and should be designed to provide maximum strength to the structure. The adhesive bonding process is widely used as a structural joining method in aerospace applications. There are many advantages of using adhesively bonding joints instead of classical mechanical fastening. Some of these can be listed as joining of similar and dissimilar materials (metal-to-composite, metal-to-metal, metal-to-glass), providing a more uniform stress distribution with a significant decrease in the stress concentration in the structure since there will be no fastening holes, a considerable weight gain compared to mechanical fasteners, strong in terms of fatigue strength due to the absence of fastener holes in the structure. In addition to the above-mentioned positive aspects of using adhesives as a structural joining method, strength prediction is vital for an optimum design process in the initial sizing and critical design phases. The fact that adhesively bonded joints have various failure modes makes failure predictions complex. According to ASTM D5573, adhesively bonded composite joints have seven typical failure modes, but they can be listed under three main headings: adhesive failure, cohesive failure, and adherend failure. Adhesive failure occurs at the adherend and adhesive interface, and usually, the adhesive remains on an adherend. These failures are generally attributed to the poor-quality bonding process, environmental factors, and insufficient surface preparation. The other kind of failure, adherend failure, occurs when the structural integrity of the adherend breaks down before the joint structure and means that the strength of the joint area exceeds the strength of the adherend. On the other hand, cohesive failure is the type of failure expected after an ideal design and bonding process, where failure occurs within the adhesive structure. After cohesive failure, the adhesive material is seen on both adherends. Structural joining with adhesive has been used in the aerospace industry since the early 1970s and 1980s. Since these dates, many analytical and numeric methods have been used to study the failures of adhesively bonding joints. Analytical method studies to analyze the failures of adhesively bonded single lap joints, known in the literature, started with Volkersen in 1938. Volkersen did not include the eccentricity factor in the calculations due to the geometric nonlinearity of the single lap joint. This factor was first taken into account by Goland and Reissner in their calculations in 1944. Goland and Reissner made a remarkable study in analysing the adhesively bonded single lap joint, calculating the loads in the joint area and subsequently the stress on the adhesive. Afterwards, analytical method studies were continued by Hart Smith, Allman, Bigwood & Crocombe and more. In addition to analytical method studies, the continuum mechanic approach, fracture mechanic approach, and damage mechanic approach can be given examples to the numerical method studies. The fracture mechanics approach used in this thesis examines the initial crack propagation in the adhesive under three different loading modes. Crack propagation occurs when the adhesive's critical strain energy release rate equals the strain energy release rate under that load. After the three different modes' strain energy release rate values are calculated separately, an evaluation is made according to the power-law failure criterion. There are many types of joint configurations in the literature, and the common ones can be summarized as single lap joints, double lap joints, stepped joints etc. The single-lap joint type is the most widely used joint type in terms of ease of design and effectiveness. Within the scope of this thesis, it is aimed to obtain a general solution that can be applied to all joints after first making a study for the single lap joint geometry and validating the results of this study experimentally. Studies have been carried out to predict the failure load of adhesively bonding CFRP joints. They include two main steps, which are to find the loads at the edges of the joint area and to evaluate the failure criteria by calculating the strain energy release rate with these loads. As the first step, loads at the joint edges are found analytically and with the finite element method, respectively. While calculating the loads analytically, the Modified Goland and Reissner theory is used, which differs from the classical Goland and Reissner theorem by taking the adhesive thickness into account. While calculating the loads with the finite element method, the modelling technique first studied by Loss and Kedward and then described by Farhad Tahmasebi in his work published with NASA is used. The primary purpose of using this modelling technique is to simulate load transfers in overlap regions accurately for complex and analytically challenging to calculate geometries. Especially in aerospace, since modelling the large components with solid elements is not effective in terms of time and resources, a practical modelling technique that can produce results with high accuracy is needed. In the modelling technique used in the thesis, adherends are modelled with shell elements while the adhesive region is modelled between coincident nodes with three spring elements to provide stiffnesses in the shear and peel directions, and the nodes of the adhesive elements are connected to the adherends with rigid elements. The modulus values of the adhesive material are used in the stiffness calculation of the spring elements. After obtaining the loads with the analytical and finite element method, the second step, the calculation of the strain energy release rate values on the adhesive material, is carried out with reference to two different studies. Firstly, linear fracture mechanics formulations were studied by Williams, assuming that the energy required to advance an existing crack unit amount is equal to the difference of performed external work with internal strain energy, and the laminate containing crack performs linear elastic behaviour is used. Conventional beam theory is used for the 1D case, as the deformation will occur like beam deformation. Using beam theory, he formulated the external work and internal strain energy at the beginning and end of the crack. And using these two equations, he found energy release rate formulations in relation to bending moment and axial load. Then, mode separation is made to calculate the energy release rates in the mode I and II directions separately because the critical strain energy release rate value in these two directions is different and needs to be evaluated independently. This study's disadvantage is that the transverse shear load is ignored, and calculations are made only with bending moment and longitudinal force. Within the scope of the thesis, the strain energy release rate is calculated both with the loads found analytically and with the loads found by the finite element method. Shahin and Taheri did the other reference work, and with overlap edge loads, the stress on the adhesive first and then the strain energy release rate is calculated. In this study, two assumptions are made, and the first is that the shear and peel stress change is zero along with the thickness of the adhesive, and the other is that the stress on the adhesive is as much as the displacement difference of the adherends. As a result of the derivations, the stress distribution on the adhesive is found in the joint structure consisting of CFRP adherend and adhesive. Then, according to Irwin's VCCI approach, as if there is a virtual crack, the integration crack length is rewritten so that it converges to zero and the displacements are in stress. Thus, the stress and energy relationship equation is obtained, and strain energy release rates in mode I and II directions of the adhesive are calculated. As a result of all these studies, mode I and mode II strain energy release rate calculations are made according to two different methods with the loads found analytically and with the finite element method. The strain energy release rate values found and the critical strain energy release rate values, which are allowable, are evaluated according to the power-law failure criteria, and failure load predictions are made. For specimens with different overlap lengths, experimental failure load values and predicted failure load values are compared, and inferences are made about the accuracy of the FEM modelling technique and the methods used in SERR calculation. All these results are interpreted in detail, and it is obtained that the FEM modelling technique gives high accuracy results with Method 2 used in SERR calculation. Finally, a bonding analysis tool has been developed with the python programming language. This tool first detects the finite elements corresponding to upper and lower adherends in the model from NASTRAN .bdf file. Then reads the element loads from the .pch file, which is a NASTRAN output and contains the element loads, then calculates the SERR using Method 2 and calculates the reserve factor and failure load, respectively. This tool has been prepared so that these calculations can be made in a short time and accurately for tens of elements in the overlap zone in complex and large models.
-
ÖgeAnalysis of bird strike on metallic panels(Graduate School, 2023-06-15)This thesis investigates the phenomenon of bird strikes, using a combination of literature analysis, statistical analysis, and theoretical models. The study focuses on the potential damage that bird strikes can cause to various parts of an aircraft, which are wind-facing components such as wings, stabilizers, engines, and windshields. The variety of possible outcomes from a bird strike poses a significant threat to aviation safety, as bird strikes account for 90% of Foreign Object Damage (FOD) incidents. As a result, aviation regulations require aircraft to meet specific levels of bird strike tolerance for critical components, and there are a number of certification requirements that airplanes must meet to be regarded safe to fly. To investigate the bird strikes on aircraft, the study uses numerical models, including the Smooth Particle Hydrodynamics (SPH) model, which was used to simulate sandwich plate bird impact experiments. The study concludes that the SPH model may be useful for finite element bird strike case analyses, which can help to improve aviation safety by identifying potential vulnerabilities and developing effective prevention measures. When using a new numerical approach, it is important to compare the results to experimental data to ensure that the simulation accurately reflects reality. Many research studies have included both numerical simulations and experimental data to understand how well the simulation corresponds to real-world scenarios. Experimental studies have traditionally guided aircraft designers in creating structures that are tough enough to withstand bird strikes. However, as aircraft components have become more complex, it has become necessary to develop bird strike simulation programs to design aircraft parts that are both airworthy and can be produced quickly and economically. Furthermore, the optimization process typically involves many iterative steps, which makes computer-based analyses more efficient and cheaper than experiments. However, conducting experiments with real birds, which are often dead or drugged chickens, presents a number of issues. The reproducibility of experiments, the health of researchers, and the availability of suitable bird models are all concerns. Real bird torsos vary greatly, making it difficult to obtain consistent results. While certification regulations only define the mass properties of the bird, different bird species have different densities, leading to variations in pressure loads between tests. As a result of these difficulties, researchers have begun using substitute bird materials instead of real birds. Advancements in computer technology have led to the development of cheaper and more advanced finite element software since the 1980s. This has allowed scientists to analyze bird strikes numerically due to the low cost, speed, and repeatability of the analyses. Various substitute bird models have been investigated in studies, and results have been compared with experimental data. The simple cylinder geometry is still a valuable approach to compare simulation results with experimental data. Different geometries such as spheres, cylinders with flat or hemispherical ends, and ellipsoids may also be used in simulations. When birds are struck at high speeds, their behavior is different from that of a simple elastic solid, and it is the responsibility of scientists and engineers to study the behavior of bird materials both theoretically and experimentally. Statistical data related to bird strikes is provided in the thesis, and it is emphasized that front-facing components of aircraft are the most critical as they are most likely to encounter a direct bird strike. The most frequently struck parts of an aircraft are the fuselage, nose, radome, windshield, wing, rotor, and jet engine. Approximately 70% of bird strikes occur at altitudes between zero and 152 meters, which is primarily during takeoff and landing. This information is useful in avoiding bird strike accidents. As the altitude of an aircraft increases, the natural habitats of birds become further from the plane. The velocity of the projectile has a significant impact on how it responds upon impact. The behaviour of the projectile can be divided into five categories based on the internal stresses it experiences: elastic impact, plastic impact, hydrodynamic impact, sonic impact, and explosive impact. Elastic impact occurs when the projectile material strength is well above the internal stresses caused by the low speeds and accelerations, resulting in the projectile bouncing back from the surface. As the impactor velocity increases, the projectile enters the plastic behavior region, yet the velocity is still low enough to maintain fluid-like flow behavior, causing the bird to spread in every direction parallel to the plate, and the load to expand to a larger area. The theory behind bird strike at velocities that cause the bird to act in the hydrodynamic region is investigated. When the impactor with the initial velocity hits a surface, materials in contact with the rigid plate would immediately come to rest, generating a shock wave with velocity normal to the plate and towards the impactor body. There would be a significant pressure gradient at the outer surface because there is shock load pressure on the inner side and free surface pressure on the outer side. Soft objects impacted at high velocities behave differently than at low velocities, such that even elastic solids behave like liquids. However, testing with real birds can yield scattered data and it is not ethical to kill animals for scientific purposes. Gelatine has been found to be a suitable artificial substitute material with uniform characteristics and can be shaped into simple geometries such as cylinders and spheres for easy handling. Finite element programs offer various solution methods for bird strike simulations. Lagrangian method involves nodes attached to the material while Eulerian method uses fixed nodes in a defined space where material flows through it. Arbitrary Lagrangian Eulerian method is another option that allows for the defined space to change with the material flow, leading to faster computation time. Additionally, the meshless method called smooth particle hydrodynamics allows for particles to move freely without mass distortion. Various basic shapes of birds can be examined for bird strike impacts, including a cylinder, a cylinder with hemispherical ends, an ellipsoid, or a sphere. For a bird with a mass of 1.8 kg and specific geometric parameters, the density of the bird can be determined to be 900 kilograms per cubic meter. Conversely, by using a standard density of 950 kilograms per cubic meter and entering the mass of the bird, a specific volume value can be determined and used to specify the bird's geometry. Honeycomb materials provide stiffness to the structure while not adding too much mass. Hence, honeycombs are a kind of deformable shock absorbers that is widely used in the aircraft industry. In the reference tests, they used single and double core honeycomb sandwich metal plates as specimens under bird strike. They made a correlation between test results and simulation results which can be beneficial. Modelling the material of honeycomb in LS-DYNA has a number of challenges. Firstly, honeycomb has a complex geometry which is expensive to model and simulate with shell elements. Therefore, its effective behavior can be modelled under homogenized solid elements. Out of plane stress strain curve up to crushing was given at reference. Which can be inserted as a stress strain curve to the solid elements. Particle node quantity for the bird impactor and element number for the aluminum sheets and honeycomb is limited with the computer power. Therefore, node numbers are generally about 20519 for the bird material. The simulations provide spatial displacement values and nominal strain curve values that are generally similar to the experimental results. However, there are slight differences, which may be due to errors in both the simulations and the tests. Overall, the strain values align well with the experimental data for both simulations. Therefore, the SPH method can be effectively used to simulate bird strikes on honeycomb sandwich plates, which is advantageous since experimental studies can be time-consuming and costly, especially in the initial design phase of aerospace vehicles.
- 1 (current)
- 2
- 3
- 4
- 5