Theses supervised by Prof. Dr. İsmail Lazoğlu
37 theses · Koç University
Development of an open-architecture process control system for the direct metal laser sintering (DMLS)
Do˘grudan metal lazer sinterleme (DMLS) i¸slemi kullanılarak metal par¸caların katmanlı imalatı, geometrik olarak karma¸sik modelleri imal etmek i¸cin toz malzemenin katman bazında eritilmesini kullanır. Burada, y¨uksek g¨u¸cl¨u bir lazer ı¸sını, metal toz katmanındaki tarama vekt¨orleri boyunca hareket ettirilir ve bu, biti¸sik malzeme ile birle¸sen bir eriyik havuzunun olu¸sturulmasıyla sonu¸clanır. Tipik olarak, DMLS makineleri pahalıdır ancak hassas metal par¸calar ¨uretmek i¸cin eklemeli ¨uretim se¸cenekleri. Di˘ger avantajlar daha az malzeme israfı, iyile¸stirilmi¸s ¨ur¨un geli¸stirme d¨ong¨us¨u, hızlı prototipleme, ¨ozelle¸stirmeler ve i¸slevsel olarak derecelendirilmi¸s metalin k¨u¸c¨uk parti boyutlu ¨uretimi par¸calar. Bu ¨ozellikler, havacılık i¸cin uygun bir se¸cim olan DMLS s¨urecini, ¨ozelle¸stirme gerektiren otomotiv, di¸s, alet ve tıp end¨ustrileri d¨u¸s¨uk toplu ¨uretim. Orne˘gin, ortopedik implantların imalatı DMLS, gerekli ¨ozelle¸stirmeyi ¨ sunar ve her iki hastaya da fayda sa˘glayan ¨ozelliklere izin verir ve cerrahlar. Bununla birlikte, bu hassas metal katkılı imalatın avantajları teknolojinin hala k¨u¸c¨uk ve orta ¨ol¸cekli end¨ustrilerde ¸co˘galması gerekmektedir. y¨uksek maliyeti, par¸ca kalitesi sorunları ve bireysel yapılar arasında d¨u¸s¨uk tutarlılık. Metalik ¨ur¨unler ¨uretmeye y¨onelik DMLS se¸cene˘ginin, genel s¨ure¸c zinciri boyunca yayılan teknik zorluklarla birlikte geldi˘gi belirtilmektedir. Aslında, metal tozlarının imalatı sırasında ve s¨uper ala¸sımlar, 50'den fazla i¸slem parametresi vardır son b¨ol¨um¨un kalitesini etkilemek i¸cin etkile¸simde bulunan. Burada birincil odak noktası, par¸calı eritme sırasında metalik tozların karma¸sık termal davranı¸sı, her bir katmandaki sınır ko¸sullarına ba˘glı olarak i¸slem parametrelerinin dikkatli bir ¸sekilde se¸cilmesini gerektirir. Bu nedenle, DMLS tabanlı eklemeli ¨uretim teknolojisi, kapsamlı par¸ca kalitesini etkileyen bu s¨ure¸c parametrelerini tasarlamak i¸cin ara¸stırma yapın ve tekrarlanabilirlik. Aynı zamanda, ara¸stırma odaklı deneyler yapmak i¸cin a¸cık eri¸simli makine platformu ve yazılım deste˘gi mevcut de˘gildir. Bu nedenle, metal tozunun erimesini ke¸sfetmenin zorlu˘gu, b¨uy¨uk ¨ol¸c¨ude esas olarak end¨ustriyel kullanım i¸cin satılan ticari sistemler. Bu sistemler sadece pahalı olmakla kalmaz, aynı zamanda temel operasyonlara ¸cok sınırlı eri¸sim sunar ve donanımları, dolayısıyla kullanıcı i¸cin bir kara kutu gibi g¨or¨unmektedir. Bu zorlu˘gun ¨ustesinden gelmek i¸cin, kullanıcıyla birlikte b¨uy¨uyebilen a¸cık eri¸simli bir ara¸stırma platformu sistemi kabul edilir i¸se yarar. Bu ara¸stırmada, MarcSLM adlı a¸cık kontrol mimarili bir DMLS sistemi geli¸stirilmi¸stir. ˙I¸s, MarcSLM makine mekanizmasının geli¸stirilmesini i¸cerir, sistem kontrol yazılımı ve olu¸sturulmu¸s s¨ure¸c planlama yazılımı. Bu sistem metroloji sens¨orleri i¸cin mod¨uler eri¸sim ve aray¨uz sa˘glamak ¨uzere tasarlanmı¸stır. altta yatan eritme s¨urecini yakalamada faydalıdır. Bu ama¸cla, ilk olarak, metalik katmanlı ¨uretim dijital ipli˘gi i¸cin bir ¸cer¸ceve ¨onerilmi¸stir. Bu i¸slem zincirinin ayrı mod¨uller dizisi olarak g¨or¨unmesini sa˘glayarak derleme i¸slemcisi ve sistem kontrol yazılımının yazılım uygulaması. Mod¨uller, imalat tarafından sıkı testlerin yapıldı˘gı bir ¸calı¸san makineye entegre edilir numune modelleri ba¸sarıyla y¨ur¨ut¨ulm¨u¸st¨ur. Daha da geli¸stirmek i¸cin yerle¸sik i¸slemci yazılımının i¸slevselli˘gi, manip¨ulasyon h¨ukm¨u dikme tabanlı kafes yapıları da dahildir. Bu h¨ucresel manifoldlar ¸cok maksimum avantaj elde etmeyi sa˘glayan katmanlı imalat tasarımında kullanı¸slıdır DMLS teknolojisinin. Geli¸stirilen model kompakt bir veri kullanır kafesleri i¸slemek i¸cin yapı ve azaltılmı¸s hesaplamayla verimli do˘grudan dilimleme sunar makine bilgisayarında uygulanabilmesi i¸cin maliyet. Ayrıca, bu ¸calı¸smanın bir par¸cası olarak kafeslerin tasarlanması i¸cin yapısal topoloji optimizasyonuna dayalı yeni bir y¨ontem ¨onerilmi¸stir. T¨um derleme verileri ¨ozelle¸stirilmi¸s bir makine i¸slemlerini sorunsuz bir ¸sekilde t¨uretmek i¸cin tasarlanmı¸s makine kodu.
Fonksiyonel malzemeleri kullanarak eklemeli üretim
The ability to additively manufacture using functional materials such as electrically conductive, dielectric, refractory, and hard materials leads to the development of smart and active products. There is a scarcity of functional materials that can be used in the additive manufacturing technique. In this work, an additively manufacturable and electrically conductive polymer matrix composite (ePMC) is developed and characterized. The electrical conductivity of the developed ePMC is tuneable with a maximum conductivity of 128 S.m-1. An application of the developed ePMC is presented for additive manufacturing a smart disinfection system to contain the spread of SARS-CoV-2. The thesis also proposes a multi-material additive manufacturing approach for the additive manufacturing of ceramic materials. The proposed approach involves the layer-by-layer additive manufacturing of a negative mold geometry. The ceramic slurry is cast into the mold gradually as it is built up layer-by-layer. The ceramic green body is demolded by dissolving the mold in an organic solvent. The multi-material approach eliminated the shape retention issue of the slurry-based additive manufacturing technique.
Development of a novel hybrid frost detection and defrost system for refrigeration systems and its applications
The repetitive collection of biological samples and their preservation for analyses at a later stage is a key aspect of biomedical research. Millions of samples collected every day around the world may degrade if proper storage conditions are not provided. Therefore, it is imperative to have a storage facility that is efficient enough to preserve the integrity of these samples over time. Refrigeration systems play a crucial role in delaying the degradation process of samples by maintaining suitable thermal conditions in storage facilities. However, the unremitting operation of the refrigeration system and the presence of moisture inside the storage facility may result in frosting on the surface of the evaporator. The frosting is a phenomenon most detrimental to the performance of refrigeration systems, as it directly affects the heat transfer process inside the refrigerator cabin. The workload on the compressor increases many folds under frosting conditions and the refrigerator struggles to maintain the desired temperature. Consequently, the energy consumption and the probability of stored samples degradation over time increases. The commercial refrigeration systems use a blind and periodic defrosting cycle without any quantification of frost, which leads to lower efficiencies. Therefore, there is a need for an intelligent system that not only detects the presence of frost but also takes countermeasures to defrost the evaporator on-demand, without affecting the quality of stored samples. In the first part of this research study, a hybrid frost detection – defrosting system (HFDDS) is developed that is comprised of a novel photo-capacitive sensing technique and a dual-purpose additively manufacturable sensor and defrosting heater. The HFDDS can detect the formation of frost, measures the thickness of frost from 1.3 to 8 mm with a 5% margin of error, and triggers a defrosting response once a critical frost thickness is attained. The HFDDS is targeted to provide a defrosting on-demand instead of the inefficient blind and periodic defrosting for the refrigeration systems. In the second part of this research study, a novel real-time thickness of the frost-based defrost-on demand technique is presented for refrigeration systems. The hybrid frost detection and defrost system developed in the first part, is employed to quantify the thickness of frost in real-time and to defrost the evaporator using a 12 W heater. The effect of the thickness of the frost-based defrost threshold on the energy consumption of the refrigerator is evaluated. The defrost threshold of 6 mm yields the maximum energy conservation of 10% as compared to the default blind and periodic defrost strategy of the test refrigerator. In the third part of this research study, a novel, frost feedback-based distributed defrosting approach to minimize the defrost desynchronization is proposed. The evaporator is divided into three regions based on the frost distribution pattern. Each of these regions was equipped with an additively manufactured low-powered defrost heater controlled by an optical frost feedback sensor. The frost feedback-based distributed approach proved to be effective in eliminating the defrosting desynchronization. The frost feedback enables the system to terminate defrosting as soon as the frost in the discretized region melts, which reduced the defrost energy significantly. The distributed approach minimizes the rise in the cabin temperature during defrosting and therefore, leads to the reduction of energy spent in the subsequent recovery cycle. The results of the frost feedback-based distributed approach were compared to the default temperature-based strategy using a single defrost heater. The frost feedback-based operation of the three distributed defrost heaters at 6 W each demonstrated maximum overall energy conservation of 18.9% as compared to the default temperature-based strategy.
Active and adaptive damping of chatter vibrations in the boring process using magnetorheological damper
Chatter is a limiting factor during boring of deep holes with long slender boring bars. This thesis introduces a new magnetorheological (MR) damper to suppress the chatter and increase the stability of the boring process. A novel design that utilizes a minimal amount of MR fluid filled in a sponge layer surrounding the boring bar is presented. The MR fluid layer and the electromagnetic circuit are externally applied to the boring bar, which allows easy installation and adjustability of the overhang length of the bar. A custom made, bidisperse MR fluid which has an improved sedimentation resistance property is used to ensure a long lifetime of the damper. A new sliding mode control with variable gain super twisting algorithm is designed and implemented for active damping of chatter vibrations in the boring process using the developed MR damper. Simulations of the controller show its fast response and robustness against disturbances and parameters uncertainties. The proposed controller outperforms the commonly used PID controller for disturbance rejection at wide range of frequencies. Experimental validations of the developed system were performed on CNC turning center. The modal analysis of the boring bar with the new MR damper shows improvements in both the damping and the dynamic stiffness of the system. This enhancement significantly increases the chatter-free depth of cut on the stability lobe diagrams. Validation cutting tests were performed under various machining conditions on boring of Al7075 and Inconel 718 workpieces which are materials widely used in many aerospace applications. The improvements achieved by this work are illustrated in terms of measured acceleration values in both time and frequency domains, in addition to the quality of the machined surfaces.
Incremental forming of light weight sheet materials at elevated temperature: An investigation of formability, failure, and output quality of the process
Abstract: Sheet metal forming is an integral part of the manufacturing field that is widely used in the aerospace and automotive industries. Today, the highly competitive field of manufacturing demands a higher level of product customization, a shorter design-to-release time cycle, and higher productivity. A novel sheet forming technology to cope with the requirements of agile manufacturing is deemed to have a flexible and highly capable tooling system. By employing the computer numerical control (CNC) technology and relatively simple tooling, Incremental Sheet Forming (ISF) offers a die-less forming possibility. More specifically, heat-assisted ISF has been appeared promising in terms of sheet formability and dimensional accuracy of the final sheet component. Due to the global warming concerns, aerospace and automotive industries are urged to implement lightweight materials and design principles. This has led to continuous growth in the consumption of thermoplastics, for example, in the automotive sector. During the last decade, a great deal of effort has been devoted by scholars to understand the mechanism and the governing rules of the ISF process. These studies have mostly targeted metallic alloys namely aluminum, steel, titanium, and magnesium. Polyoxymethylene (POM) is the least investigated thermoplastic among the engineering polymers employed for ISF, possibly due to its low formability at ambient temperature. However, POM is known as a semicrystalline thermoplastic of good creep and fatigue durability. POM also offers high resistance against chemicals, oil, and fuel corrosion. In addition, POM provides a good surface finish and friction resistance quality. This thesis aims to investigate Friction Stir Incremental Sheet Forming (FSISF) of Polyoxymethylene. Therefore, a comprehensive experimental, analytical and numerical (FEA) study is conducted. The proposed analytical and FEA frameworks are tuned to provide fairly accurate predictions with an affordable simulation cost in terms of the computation time and the required hardware. A comprehensive experimental campaign is conducted by adopting Design of Experiment (DOE) and response surface methods. Individual and interactive effects of the process parameters, namely tool size, step size, feed rate, and spindle speed on the process outputs are discussed in terms of forming force, contact temperature, heating rate, and deformation rate. An empirical model is provided to predict the total forming force as a function of the process parameters and time. Also, similar models are developed for the maximum force and the tool-tip temperature. Relatively high sheet formability is achieved, whereas heat-induced defects are avoided. The effects of geometrical parameters on sheet formability are briefly discussed and a new formability criterion is proposed for ISF. Friction in FSISF of POM is experimentally evaluated. Dimensional integrity, thickness distribution, surface quality, the evolution of the material crystallinity, surface hardness, and elastic modulus of POM under specific forming conditions are measured and discussed. A discussion is also provided on fracture in FSISF of POM based on the morphology of the cracks acquired by Scanning Electron Microscopy (SEM). A joint numerical-analytical method is proposed to efficiently predict force and contact temperature in the FSISF of POM. For this purpose, the mechanism of deformation and stress-strain state is defined based on a membrane analytical framework including the stretch-bending effect. Also, the pronounced bending effect in the initial stage of the forming process is investigated by extending an available analytical framework for deflection of a circular sheet under a lateral concentrated load. The friction-induced heat and temperature distribution in tool and sheet are derived by adopting the Finite Difference Method (FDM). The area of the tool-sheet contact interface is estimated by proposing two geometrical approaches. Predicted forming force and contact temperature are validated by experimental measurements. Friction-induced heat rate and meridional stress component are presented as a function of the equivalent plastic strain for some FSISF benchmarks. The ratio of the plastic strain energy density to the triaxiality is also presented as a function of the final equivalent plastic strain. Accordingly, a brief discussion is provided on the sheet failure in the FSISF process. It is aimed to establish a framework for a fairly accurate FEA of the FSISF process with an affordable simulation cost. For this purpose, the material constitutive models namely Two-Layer Viscoplastic (TLV), Parallel Rheological Framework (PRF), Three Network (TN), and Johnson-Cook (JC) are discussed and calibrated. A series of comparative FEAs is conducted to identify the best meshing strategies, FSISF tool definition, element type, element size, and mass scaling. Heat partitioning between tool and sheet is discussed in detail by conducting a precise review on the available approximate analytical methods in the literature. Benchmark FSISF tests are considered for simulation. To demonstrate the capability of the proposed FEA methodology, the simulation case studies reflect a broad range of sheet formability, temperature, and stress triaxiality. The simulation results are validated by experimental measurements. A brief discussion and numerical investigation are provided on damage evolution and failure in FSISF of POM.
Design and implementation of process control for deep drawing using flange draw-in
In sheet metal forming processes, using new materials with high strength and low formability and an increase in part design complexity leads to narrowing the process windows where high-quality parts can be formed without any defects. Moreover, these reductions in process limits increased the sensitivity to process variations such as changes in material properties and lubrication conditions, resulting in increased scrap rates and decreased productivity in mass production. In recent years, process control strategies and different blank holder systems have been developed and utilized in deep drawing to increase part formability and productivity. This thesis mainly focused on developing design methodologies for inline proportional plus integral (PI) process control to improve part quality and reproducibility in the existence of process variations. Moreover, a smart-compact cushion system containing a segment-elastic blank holder with multiple hydraulic actuators and process sensors for inline monitoring is designed and adapted in a servo-spindle press to conduct experiments. Finite element analysis (FEA) software tools are utilized to determine optimal process parameters such as blank holder force and material flow which are necessary for producing high-quality parts. Additionally, process variables such as flange draw-in and punch force are evaluated numerically and experimentally to determine their relationship with the failure, such as tearing and wrinkling. This work focuses on flange draw-in as a control variable that is manipulated by the blank holder force adjustments based on tracking a reference flange draw-in with the help of the PI process controller. The critical novelty presented is the use of draw-in as the reference parameter for PI-based process control, a combination that has not been previously investigated in the literature. Systematic design and implementation of a process controller for deep drawing with a semi-complex rectangular sheet metal part are presented. First, according to the FEA simulation tool and experimental work, the deep drawing process model structure is derived in the form of discrete-time transfer functions. Second, the dynamic model parameters, which can vary with the die geometry, are estimated via system identification techniques based on experiments. Later, numerical simulation tools are utilized to develop a process controller based on the dynamic process model. The process controller is fine-tuned by using root-locus and frequency-response analysis. Eventually, the proposed inline process controller constructed in the simulation environment is implemented in the servo-spindle press via PLC. The experiments were performed in a servo-spindle press with the smart-compact cushion system that includes 16 hydraulic actuators to control draw-in outputs measured via displacement sensors placed around the sheet periphery. The results show that the proposed process control design has an excellent performance in tracking the reference flange draw-in and significantly improving the part quality and reproducibility in the presence of process variations.
Autonomous truck-trailer parking - path planning and path tracking control
Maneuvering a truck-trailer system while docking is extremely challenging. This study aims to alleviate this problem by presenting a novel cascade path planning framework and an enhanced path-following control framework for autonomous semi-trailer docking. In the proposed system, the cascade path planning framework generates kinematically feasible and drivable maneuvers for trailer parking and the path-following control framework introduces adaptive controllers that utilize gain scheduling for forward and reverse path-following tasks in docking maneuvers to increase the robustness and path-following performance. In the cascade path planning approach, a realistic and deterministic parking behavior model, iterative analytical method (IAM), is proposed and combined with an enhanced Closed-Loop Rapidly Exploring Random Tree (CL-RRT) approach. Cascade path planning approach combining CL-RRT with IAM mimicking real-world parking practice enables generation of both kinematically feasible and deterministic parking maneuvers with obstacle avoidance. For evaluation, different parking scenarios are generated and selected through a developed case generation tool. The proposed path planning approach is evaluated through MATLAB simulations for performance evaluation. The results achieved a noticeable success with a high rate of generated feasible maneuvers for truck-trailer parking. In the proposed path following control framework, the system includes an improved pure pursuit controller with adaptive look-ahead distance for forward path following; a cascade controller of reverse pure pursuit, and a gain-scheduled linear-quadratic (LQ) control for reverse path-following. In the evaluation of the path-following performance of forward and reverse controllers, the closed-loop system of path-following controllers with the truck-trailer kinematic model is simulated in MATLAB/Simulink for various test cases, and the results are compared with those of other studies. Furthermore, different docking scenarios are generated via the cascade path planning algorithm for autonomous semitrailer docking. These are tested with a high-degree semi-trailer model within the IPG TruckMaker simulation environment, and with a full truck-trailer vehicle in the test field. The results of both simulations and physical testing clearly demonstrate improvements in terms of the control problem formulation, i.e. the stabilized path-following is obtained with acceptable path-following errors.
Quality failure detection technique in simultaneous and sequential multi-valve plastic injection molding
The plastic injection (PI) process has the nonlinear and time-varying complex dynamic characteristics of the batch processes, and the continuity of the process depends on the long-term experienced and skilled operators. The increase in the design complexity requirements, the use of new materials to reduce environmental effects, and energy consumption minimization requirements increase the control studies about PI. Failure detection in the PI process is critical for increasing the automation level of the process and reducing the scrap rate. In recent years, quality failure detection methods and smart injection strategies have been developed and utilized in PI to increase control over the process. This thesis mainly focused on developing failure detection techniques for simultaneous and sequential multi-valve PI molding. Three main criteria are questioned in the proposed method: high quality failure detection rate, sustainability in the production, and feedback substructure. Different trials are conducted, and an iterative approach is used to construct the method. Confusion matrices and hypothesis evaluation metrics such as accuracy (Acc), precision (Pr), and sensitivity (Sn) are used to measure the proposed technique's success. An iterative approach is used to satisfy the criteria, and studies started with time series-based approaches. Time series approaches neither have feedback substructure nor sustainability. Therefore, data windowing approaches are implemented, and criteria are satisfied. Cavity sensors (CS) and valve timings are the primary sources of information, and the only collected input from the injection molding machine (IMM) is the injection start signal. Thus, a failure detection method that is independent of the IMM is proposed in this study. Quality failure detection in multi-valve PI molding is presented in this study. Timings of the valves are collected via a programmable logic controller (PLC), while sensor outputs are collected with a custom-developed data acquisition device (ADCR). Collected sensor profiles and timing information are matched and transferred to the electrical panel, which can store, analyze, and configure the upcoming data. Labeled data is used to construct a learning-based failure detection algorithm and valve time feedback. The study focuses on analyzing the cavity pressure (CP) sensor and derivative profiles with the sliding data windowing technique. Additionally, the labeled time series CP profile is analyzed with Python's Time Series Feature Extraction on Basis of Scalable Hypothesis library (TSFRESH), and the most critical parameters are evaluated and selected for the learning algorithm. To evaluate the quality of the product, features are extracted from the windowed CP sensor profile, windowed pressure derivative profile, cavity temperature (CT) sensor, valve opening times, and TSFRESH. Different learning algorithms such as k-nearest neighbor (KNN), support vector classifier (SVC), and logistic regression (LR) are used for failure detection. Thus, a learning algorithm is created, and a new method that has high quality failure detection rate, sustainability in production, and feedback substructure is proposed for multi-valve PI processes.
Dynamic model development for highly efficient inverter compressor
The heart of a household refrigerator can be assumed as the compressor since it receives the refrigerant, compresses it and allows gas to circulate inside the refrigeration cycle. Even though there are different types of compressors in the market, reciprocating compressor is the most common one at the refrigerators. Due to energy regulations that become more strict day by day because of global warming, the challenge between compressor manufacturers is to producing the most energy efficient compressor. In order to reduce the energy dissipation and increase the efficiency, the fundamentals behind the pyhsics used in the compressor must be understood clearly. Moreover, even though dynamics and mechanics disciplines are investigated in this thesis, the relationship with thermodynamics and flow dynamics must be considered as well. In this thesis, a comprehensive mathematical model of a reciprocating compressor is established by including all design parameters and considering all force sources for both steady state and startup behaviour where the highest vibration and noise are observed. The model validations are conducted by specially developed experimental setups. Moreover, the components that play a significant role in dynamic response of the compressor such as springs or discharge tube are investigated individually, further optimizations are proposed to reduce the design time and both vibration noise levels. A novel design methodology is proposed for the discharge tube for the first time in the world and mathematics is included in the design process. Furthermore, to reduce startup and shutdown vibration and noise, a novel design addition, called transient vibration reducer (TVR), is utilized. TVR is a passive damper that operates as a pseudo-active damper due to its stepwise structure, interacting with the motion of the compressor only above predetermined limits of body displacement without affecting or compromising steady-state performance. Finally, the startup and shutdown position map data of the crankshaft is formed with an all-round experimental setup for the first time in the literature. Suggestions are made to have a smooth startup behavior, implemented on a compressor and validated with experimental studies.
Modeling and analysis of distortion in milling of aerospace parts
The issue of distortion of parts manufactured by machining is a long-standing problem in the aerospace industry. Especially in the case of large, thin-walled machined structural components, post-machining distortions result in a loss worth billions of dollars to the aerospace industry. Control of distortions is, therefore, very critical to ensure the conformity of these parts and the efficiency of the process. Effective control of the process necessitates efficient models and simulation techniques for predicting distortion of the workpiece after machining. Very high cutting force and temperature loads generated during machining affect the workpiece in several ways. These cutting loads are dictated by the parameters of machining and the material characteristics of the cutting tool and the workpiece. Moreover, the initial stress of the blank also dictates the distortion behavior. Accurate modeling of all these factors is crucial to the precise prediction of distortion of the workpiece. Milling is a very important process for the machining of aerospace parts. In this thesis, modeling of the milling process is carried out for efficient prediction of various aspects of the process. A novel analytical algorithm is proposed to predict the cutting temperature of the workpiece. The proposed algorithm allows accurate calculation of the workpiece temperature by duly incorporating the drop in temperature of the workpiece during the non-engagement periods of the cutting tool. Another novel contribution of this thesis is developing a hybrid FEM-analytical model for predicting distortions of machined thin-walled parts. The model considers loads due to cutting as well as the effect of the initial bulk residual stresses of the material. The measurement of the initial stresses of the material is carried out by the crack compliance method, whereas machining-induced loads are calculated analytically. A novel strategy is devised to incorporate these loads in a FEM model. The results of the proposed models are validated with machining experiments. Another contribution of this thesis is the development of a novel method for monitoring of the machining process using a low-cost infrared sensor. Three critical aspects of the machining process, i.e., tool wear, chatter, and workpiece deformations, are detected using a single infrared sensor. Different signal processing techniques are applied in time and frequency domains to analyze the data collected from the sensor. The results of the proposed method are verified by carrying out machining experiments.
Hermetik pistonlu kompresörün tribolojik analizi
A hermetic reciprocating compressor is one of the most critical parts for the energy efficiency of a household refrigerator. The tribological performance of the compressor's piston-cylinder pair could be enhanced through introducing micro-texture on the compressor's piston surface. In this thesis, an experimental study is presented to tribologically assess the effect of the micro-textured piston on the performance of the hermetically sealed reciprocating compressor. Effects of the micro-dimples texture of the piston on the compressor's coefficient of performance was investigated. Refrigerant leakage from the piston-cylinder clearance was decreased by 35% and the compressor's cooling capacity was observed to be increased slightly. In addition, this thesis also presents a novel solution to the lubrication problem of an inverter-type hermetic reciprocating compressor used in household refrigerators during a low-speed (less than 2000 rpm) operation. The lubrication system is primarily based on an additively manufactured unibody pump. Moreover, the thesis also aims to discuss the development of a cermet material for the manufacturing of piston via a multi-material negative additive manufacturing approach. The suggested methodology entails the layer-by-layer 3D printing of a negative mold geometry. The cermet slurry is cast into the mold gradually as it is built up layer-by-layer simultaneously. The cermet green body is then demolded by dispersing the mold in an organic solvent such as DCM. The multi-material 3D printing approach eliminated the shape retention problem of the slurry-based additive manufacturing system. A piston produced from cermet material through this method is found to be 10 times lighter than a conventional piston.
Development of a modular pulmonary resuscitation device for chronic and acute respiratory support
Noninvasive ventilation (NIV) and invasive ventilation are two distinct approaches to managing respiratory failure and supporting individuals with impaired breathing. These techniques are crucial in critical care medicine and respiratory therapy, offering life-saving interventions for various pulmonary disorders. Noninvasive ventilation represents a method of respiratory support that aids patients without requiring invasive procedures such as endotracheal intubation. It typically involves the use of a noninvasive ventilator or positive airway pressure devices to deliver air or oxygen through a mask or interface, aiding patients in maintaining adequate oxygenation and carbon dioxide removal. NIV has emerged as a preferred choice in various clinical scenarios, including chronic obstructive pulmonary disease (COPD) exacerbations, acute respiratory failure, congestive heart failure, and sleep-related breathing disorders. This approach offers several distinct advantages, such as reduced risk of ventilator-associated pneumonia, improved patient comfort, and the possibility of patient self-management, making it an increasingly valuable tool in respiratory medicine. The emergence of the COVID-19 pandemic in 2020 prompted the creation of numerous affordable, open-source, or readily constructed ventilators for urgent deployment. However, most of the proposed setups constituted of either automation of a hand-operated resuscitator called Bag Valve Mask (BVM), which is limited in its ability for pressure and volume control, or the development of pressurized valve-based mechanisms that require a constant high-pressure input source, making them redundant in scenarios where pressurized air/oxygen source is not available. Furthermore, these solutions are generally catered towards invasive and intubated mechanical ventilation techniques. Turbine-based positive pressure ventilators bridge the gap between portability and advanced therapy modes for pulmonary therapy via non-invasive ventilation. The existing body of literature exhibits a notable paucity of comprehensive examinations concerning the conception, evaluation, and viability of turbine-based non-invasive ventilators as an economically efficient and readily deployable platform for respiratory support. This deficiency in research extends to the exploration of their potential applications in contexts extending beyond emergency situations, particularly in the domain of home care ventilation. This thesis delves into examining and evaluating the creation of a modular respiratory system that can provide continuous positive pressure and Bilevel positive pressure therapy. In the first phase of this research study, a high-pressure turbine-based positive-pressure support, non-invasive ventilator design is proposed. The developed hardware can deliver multi-mode respiratory therapy, namely pressure support and volume-assured pressure support ventilation. The novel modular device works as a non-invasive respiratory support device with novel algorithms developed to maximize patient machine synchronization and robust closed-loop control strategy for pressure therapy during varying pulmonary compliance. In the second part of this research, a novel real-time wireless sensory module technique is presented for advanced respiratory monitoring and control during non-invasive ventilation. The proposed techniques focus on improving the current standards of monitoring and therapy in continuous positive pressure systems utilizing single-limb circuits with passive leak ports. Implementing the proposed wireless pressure and flow monitoring, in conjunction with the developed modular respiratory device, yields enhancements in the device's responsiveness in detecting distinct patient respiratory phases. Furthermore, this setup effectively alleviates the necessity for computational modeling of dynamic leakages, consequently facilitating the precise monitoring of inspiratory and expiratory tidal volumes in Non-Invasive Ventilation (NIV). Notably, these capabilities significantly advance compared to conventional homecare ventilatory support utilizing single-limb passive leak circuits. In the third part of this research, real-time assessment for pulmonary mechanics is performed via Machine learning techniques from data acquired in real-time during non-invasive pressure therapy mode. The pulmonary mechanics are key monitoring parameters for ensuring correct therapy mode and therapy pressure or volume levels are provided for patient care. To date, for pulmonary mechanic measurement, an inhalation pause maneuver is adapted. This maneuver is only possible during invasive mechanical ventilation, where patient-derived breathing is absent, and the total breathing is controlled via the ventilator. The absence of an inhalation pause maneuver or any active spirometry during non-invasive ventilation makes monitoring pulmonary mechanics extremely challenging. Thus we present a novel framework for the development of machine learning techniques that can predict pulmonary mechanics like compliance and resistance with an overall accuracy above 95% with and without system leakages, leading to a novel digital framework that allows for lung mechanics monitoring without clinical intervention and can help save ventilator induced lung injuries.
Machine learning based cooling time prediction in plastic injection molding process
This thesis presents a new approach to develop a machine-learning model for the cooling profile prediction of a planar part produced through the plastic injection molding. Design parameters related to the cooling stage of injection molding, such as distance between the cooling channel and plastic part surface, the distance between cooling channels, the cooling channel diameter, the thickness of the part, as well as plastic material properties including density, mass, thermal conductivity, and specific heat are considered. A wide range of scenarios are created considering the design parameters, ensuring the creation of a comprehensive dataset. To manage this extensive collection of cases, scripts are used to automate designs and simulations. The scripts first generate the required cooling channels for each scenario, then simulate the cooling stage of the injection molding process and collect the relevant results. Then, the physics of the cooling is discussed to estimate the time-dependent temperature of the part in the plastic injection process. An LSTM machine-learning model is used to forecast the simulation results and a regression model is used to predict the cooling profile of the parts. The validation of the developed machine-learning model is done by estimating the cooling time of two industrial parts, produced by plastic injection. One of the parts is cooled with the conformal cooling channels and the estimated cooling time is found with a 2.67% error. On the other hand, the second part is cooled with conventional cooling channels and the estimated cooling time is found with an 11.44% error. In addition, the impact of each design parameter on the cooling time was examined. After the examination, it was noticed that the plastic part should be designed as thin as possible during the design to reduce the cooling time. Also, designing cooling channels as close to the surface of the plastic material as possible was very important to reduce the cooling time.
CNCmachine tool digital twin applications with advancements of edge computing
Digital Twin application development for CNC machining processes is challenging since collecting real-time high-frequency process data from data sources such as dynamometer sensors and processing collected data into data infrastructures to develop analytical models without latency and time-synchronization is required. Dynamometers are used to measure cutting torque and cutting forces occur between the workpiece and cutting tool during machining process. Collecting dynamometer data during machining is expensive in terms of sensor mounting, data acquisition and processing. To overcome this challenge, a method for cutting torque estimation that utilizes edge computing application to collect internal machine data is proposed in this thesis, to establish preliminaries for machining Digital Twin applications with edge computing. The cutting torque is estimated during drilling experiments and prediction errors are calculated for experiments with same workpiece material as 0.31% and 0.9%, and with different workpiece material as 1.02% and 1.25%. Edge device allows the collection of high-frequency machining process data coming from spindle and feed-drive motors by providing inbuilt network components that enables data acquisition to Cloud sources from machine. Edge device signals are also presented in servo motor current and position closed-loop control loop block-diagrams to demonstrate data flow stages from CNC machine tool's internal servo system to Edge device. The block-diagrams are validated with Edge signals collected during milling and drilling experiments.
A comprehensive investigation of the influence of vacuum conditions on the drying characteristics of textiles
Pulsed vacuum drying (PVD) is a novel drying technique that has gained attention in recent years. It is one of the techniques that has emerged to address the limitations of conventional drying methods. This study proposes a comprehensive investigation into the influence of vacuum conditions on the textile drying process. The research focuses on the utilization of both centrifugal vacuum pumps and pneumatic vacuum generators to create a controlled vacuum environment within a custom-designed vacuum drum. It also aims to examine the influence of PVD on textile drying characteristics. This will be achieved through a series of steps: first, a thorough analysis of the impact of the constant vacuum and pulsed vacuum process on the drying kinetics of textiles in a vacuum environment; second, identification and evaluation of optimal drying parameters, such as temperature, duration of the vacuum phase, and period of the atmospheric phase; and finally, development of a practical framework for simulating thermal and mass transfer during vacuum assisted drying. The accuracy of the simulations is confirmed through experimental results. The comparative study will extend to conventional dryers, with the aim of demonstrating the advantages and limitations of vacuum-assisted drying in terms of drying rate and total drying time. Through this investigation, the thesis aims to contribute valuable insights into the potential advancements in textile drying technology, with a focus on vacuum-assisted methods.
Temperature-controlled organ cooling jacket system for open and robotic-assisted surgeries
Cold ischemia plays a crucial role in open and robotic-assisted kidney transplants and partial nephrectomy surgeries, not only minimizing the risk of renal injuries but also extending the duration of the operation. Traditionally, the cooling process relies on the application of ice slush to the donor kidney during operation, a method prone to variability as the amount of ice slush is determined arbitrarily by the operating doctor. To address this issue, a novel approach is proposed, introducing a temperature-controlled organ cooling jacket. This innovative design incorporates auxiliary liquid circulation and refrigeration units, providing a controlled and consistent cooling environment. The cooling jacket, crafted from biocompatible materials through negative molding with 3D printed molds, represents a significant departure from the temperature-uncontrolled procedures prevalent in current practices. The development of the device is detailed in this study, covering the entire process from design conception to manufacturing techniques. The materials and methods employed in the production of the organ cooling jacket are discussed, highlighting the adherence to design constraints. The efficiency of the device is examined through ex vivo testing, shedding light on the promising results achieved. This comprehensive exploration encompasses the evolution of the proposed solution, offering valuable insights into its potential application in enhancing and implementing temperature-controlled cold ischemia in open and robotic-assisted kidney surgeries.
Modelling and simulation of the cardiovascular system and the Istanbul heart ventricular assist device
Heart transplantation rates are limited due to high donor heart rejection rates and the rising number of heart failure (HF) patients due to a growing and ageing population. Left Ventricular assist Devices (LVADs) are increasingly being implanted as a buttress against advanced-stage heart failure. However, they are associated with a multitude of complications. LVAD design and development is a thorough pro cess involving many steps before they can acquire approval from regulatory bodies, involving virtual testing in-silico, verification of the results in-vitro, validation of performance in animals, i.e., in-vivo, and ultimately clinical trials. As we progress along its different phases, the development process becomes more complex, less controllable, and more costly. Under such circumstances, reliable pre-clinical evaluation becomes imperative. The current work makes use of two pre-clinical models namely numerical modelling and mock circulatory loops. An intuitive and knowledge-based approach to numerical modelling is adopted whereby model components are con nected in a topological manner without explicitly coding model equations. Mathworks' Simscape (TM) modelling environment is used to develop a comprehensive and real-time executable model of the cardiovascular system (CVS) and the Istanbul Heart (iHeart) VAD. The CVS model is developed by combining existing models and the VAD model is an analytically-derived model calibrated using iHeart VAD's characteristic data. A baseline is established by calibrating the CVS model with healthy heart parameters. Parameters for Dilated Cardiomyopathy (DCM), a sub set of Heart Failure with Reduced Ejection Fraction (HFrEF), are collected from published patient data and disease modelling is conducted using statistically pooled values. To simulate iHeart VAD support, it is anatomized to the left ventricle apex and the ascending aorta in Simscape (TM). CVS-VAD modelling is applied for in-silico adaptation of hemodynamic ramp testing, an invasive clinical procedure which can potentially support clinical decision-making before conducting such an invasive procedure by exploring the possibility of improvements preemptively. It is also used to conduct a full factorial design of experiments to study the functional relationships between CVS parameters and LVAD suction speed which can be used to guide the development of suction detection and avoidance control algorithms. CVS-VAD model is also used to develop a numerical-hydraulic hybrid mock circulatory loop to study the behaviour of the iHeart VAD in-vitro. The adopted modelling approach results in a more realistic replication of the physiological environment an LVAD is exposed to. A comprehensive framework for the pre-clinical evaluation of mechani cal circulatory support devices is provided. Due to its object-oriented and modular nature, the featured model can be readily modified for other cardiovascular diseases. In addition to facilitating LVAD research, the presented work provides a teaching tool for understanding the pathophysiology of heart failure, diagnosis rationale, and degree of assist requirements. Utilization of in-silico and in-vitro approaches provide complementary information, allowing device behaviour and performance prediction and enhancing model accuracy.
Development of Istanbul heart version 2
Heart Failure (HF) continues to be a leading cause of death due to the shortage of donors for transplantation. This shortage has led to the development of several alternative solutions/therapies. Mechanical Circulatory Support Devices (MCSD) are one possible solution that has been proposed. Left Ventricular Assist Device (LVAD) is one of the most common MCSD available. LVADs are generally used as a bridge-to-transplant, bridge-to-recovery, or destination therapy for HF patients. Of the many LVAD design considerations, ensuring blood compatibility and hemolysis prevention is of paramount importance. This study aims to develop a small and compact version of the Istanbul Heart (iHeart). Computational Fluid Dynamics (CFD) has been employed in reducing the size and optimizing the impeller geometry while maintaining the required performance attributes. The LVAD impeller diameter is reduced to 30 mm while wrap angles are varied to minimize the hemolysis index and to deliver a 5 L/min flow rate at 100 mm-Hg. In-silico tests are performed to find the best impeller geometry achieving the set conditions. An experimental study was performed to validate the simulations. The experiments measured the hydraulic performance in terms of pressure and flow rate as well as the hemolytic performance in terms of hematocrit and plasma-free hemoglobin. Finally, an initial in-vivo study is conducted for the selected LVAD design. Two adult sheep weighing around 50 kg have been used to evaluate the anatomical fitting, biocompatibility, and performance of the LVAD in the short term.
Gelişmiş plastik enjeksiyon kalıplama ve endüstriyel uygulama için ML odaklı biliş
For a long time, plastic injection molding has been an important part of mass production in many fields. But in today's competitive market, traditional molding methods don't always work well when you need high efficiency and quality that stays the same. We need better and smarter ways to make things because of the need for more sustainable practices, the use of more recycled materials, shorter production times, and rising labor costs. This thesis describes a cognition-based method that is meant to work without depending on the shape of the parts, the type of material, or the specific production equipment used in the injection molding process. To do this, cavity pressure sensors were carefully put in key parts of the mold to collect data that was unique to each cycle. This information helped us find a stable operational range, or "reliable zone," that guarantees that high-quality parts are always made. One important thing this work does is show that changes in the cavity pressure curve are related to both the quality of the parts and the settings of the machine. Using this connection, a convolutional neural network (CNN) was trained to create a basic knowledge system that helps operators by suggesting ways to fix problems when they notice that something is wrong. The model that was made was able to correctly classify 98% of the time. After the baseline was made, the proposed method was used on two real-world case studies that involved parts with different shapes, materials, and quality standards. For each application, a method for adapting knowledge to specific tasks was used, and when the system was put into real-world production environments, it had an average accuracy rate of 95%.
Smart manufacturing in machining process using industrial edge device
In modern manufacturing, smart machining solutions leverage advanced data analytics, real-time monitoring, and adaptive control strategies to optimize production efficiency, minimize costs, and ensure the quality of the product with remarkable precision and reliability. This thesis, titled "Smart Manufacturing in Machining Process Using Industrial Edge Device," explores innovative approaches for integrating industrial Edge devices into machining operations, enabling real-time monitoring, predictive maintenance, and cost-effective process optimization. The research is structured into four main items, each addressing a critical aspect of smart manufacturing in machining processes, specifically focusing on tool wear prediction, sensorless cutting forces and torque measurement, tool chipping detection, and runout monitoring. First, an AI-assisted digital shadow was developed to predict tool flank wear using data acquired from an industrial Edge device in a drilling operation. The study combines high-frequency data acquisition from an industrial Edge device and a rotary dynamometer, followed by extensive feature engineering. A recurrent neural network (RNN) model utilizing bidirectional long short-term memory (Bi-LSTM) and bidirectional gated recurrent unit (Bi-GRU) architectures is employed to analyze tool wear regions. The developed digital shadow enhances cost efficiency by reducing the dependency on expensive multi-sensor systems and mitigating unnecessary tool replacements, aligning with the principles of smart manufacturing. Second, a novel sensorless method is introduced for real-time measurement of cutting forces and torque in the milling process. Accurate measurement of cutting forces and torque is critical for process monitoring, tool condition assessment, and optimization in machining operations. However, conventional dynamometers, while precise, are costly and introduce complexities in experimental setups. This study leverages an industrial Edge device to extract spindle current and torque data in real time. Experimental validation during the milling of Titanium alloy Ti6Al4V demonstrates strong correlations between dynamometer and Edge device cutting data, achieving mean errors of less than 12%. This sensorless approach significantly reduces setup time and cost while ensuring precise force monitoring, making it a viable alternative to traditional measurement techniques in smart machining environments. Third, tool condition monitoring was investigated with a focus on detecting cutting-edge chipping in the milling of titanium alloys. As a critical factor, tool chipping significantly impacts machining efficiency and tool life. The study analyzes spindle current and torque signals obtained from an industrial Edge device operating at a high-frequency sampling rate of 500 Hz in the time and frequency domains. Fast Fourier Transform (FFT) analysis reveals significant spectral differences in Edge device signals before and after tool chipping occurrences. These findings establish industrial Edge computing as a reliable tool for tool condition monitoring, minimizing the need for expensive multi-sensor systems and enabling early detection of chipping events to prevent catastrophic tool failure. Finally, an innovative sensorless approach for in-process runout detection was proposed. Runout, a key factor influencing machining accuracy, surface finish, and tool wear, is traditionally assessed using contact-based sensors. This study demonstrates that spindle current and torque signals, captured through an industrial Edge device, are highly correlated with dynamometer measurements, enabling real-time frequency-domain analysis for runout detection. The proposed method is experimentally validated during milling of Ti6Al4V, presenting a cost-effective and efficient approach to improve machining precision and stability in smart manufacturing environments. In summary, this thesis contributes to the advancement of smart manufacturing in machining processes by harnessing the potential of industrial Edge devices for real-time monitoring, predictive maintenance, and sensorless process optimization. The proposed methodologies reduce reliance on expensive sensors, improve machining efficiency, and enhance cost-effectiveness, thereby paving the way for the next generation of intelligent and autonomous manufacturing systems.
Integrated framework for hip joint simulator design and validation under realistic loading conditions
Accurately replicating physiological hip joint loading is essential for the reliable evaluation and optimization of total hip replacement implants. This thesis presents a comprehensive methodology integrating finite element modeling, topology optimization, and fatigue sensitivity analysis with experimental validation using a custom-built, ISO 14242-compliant hip joint simulator. The mechanical system features zero-backlash Harmonic Drive® and planetary gear mechanisms, a high-precision six-degree-of-freedom load sensing platform, and adaptive control strategies that precisely reproduce complex multi-axis motion and force profiles. Beyond conventional ISO-based testing, this work incorporates measured in vivo daily-life activity data from Rydell, Paul, and Duff-Barclay alongside the ISO 14242 standard, enabling accurate simulation of gait cycles and other functional tasks. This integration highlights discrepancies in force magnitudes and loading patterns between standardized and physiological conditions. Monte Carlo simulations with stochastic gait load vectors were conducted to evaluate the simulator's structural response under rare high-angle or high-magnitude loading scenarios not captured by existing standards. These analyses informed subtle reinforcements in the force-application mechanism and guided parametric studies assessing the influence of key design variables on stiffness, actuator demand, and service life. Results show that the optimized arm structures achieved a significant weight reduction of 55% without compromising structural integrity or fatigue resistance. The hybrid control architecture, leveraging high-resolution feedback and zero-backlash actuation, ensures high-fidelity reproduction of gait cycles under both deterministic and random load conditions. Validation experiments confirmed the system's ability to maintain accurate force tracking, minimize actuator energy consumption, and sustain long-term wear and fatigue testing reliability. The developed platform satisfies ISO 14242 compliance and enables extended, physiologically realistic testing scenarios, providing a robust foundation for future studies in implant wear mechanisms, lubrication dynamics, and predictive data-driven modeling for next-generation prosthetic designs.
Havacılık endüstrisinde kullanılan alüminyum alaşımlarının karbür parmak frezeler ile işlenmesinde takım ömrü analizi
In today?s competitive aerospace and a ircraft industries, demand in manufacturing products with high speed, high precision and lower cost is increasing. According to the Airbus reports, number of commertial aircrafts is doubled every 15 years. Milling is one of the major manufacturing processes in producing aerospace and aircraft parts such as all structural parts and jet engine components. In aerospace and aircraft industry, removing of 90-95% of the bulk material with milling is very common to lower the weight of parts. Therefore, milling is perhaps the most critical manufacturing process in these critical parts. Selections of optimal machining conditions and the appropriate tools based on scientific analysis are very critical for high quality part manufacturing with lower cost in shorter cycle times. When machining velocities are increased in order to reduce the process time, machining forces, heat generated and process temperatures all increase. The harsh machining environment accelerates the tool wear. As a result, worn tools generates higher forces and more deformations on parts and tools, may cause bad surface quality and scraped parts, and sometimes may results into catastrophic failures. Along with negative effects on the work safety, machining center and final product, the tool costs are one of the main contributors of the overall cost. Therefore, accurate prediction and optimization of the tool life are very important both from the scientific and industrial perspective. In this thesis, tool lives of different helical end mills in milling aerospace grade aluminum alloy Al-7050 which is used in structural parts of aircrafts and aerospace systems are investigated experimentally. This research was sponsored by the Turkish Aerospace Industries (TAI). The workpiece material and milling tools are supplied by TAI. In this research, the forces acting during the milling of Al-7050 are modeled using mechanistic calibration technique taking the bottom edge cutting affects into account. The cutting force model is validated with the cutting experiments conducted on multi-axis CNC milling processes in Mori Seiki NMV5000 DCG. Moreover, the heat generated and temperature distributions in the tool geometry are predicted using the mechanic and thermal modeling methods. The thermal model outputs are compared with the available data in the literature and with the infrared thermal measurements performed in the Manufacturing and Automation Research Center. After the temperature field is established, an analysis on wear is performed. The predicted wear values are also compared with the experimental measurement results. The surface roughnesses of the manufactured parts are measured to observe the effect of wear on the finished part. Besides the scientific analysis on milling mechanics and thermal analysis, this research concludes that by selecting appropriate cutting tools and machining conditions significant saving (up to 85%) on tooling cost in industry is possible.
İnsansız hava aracı gözlemlerinde gerçek-zamanlı görüntü mozaiklemesi ve stabilizasyonu
Using mini Unmanned Aerial Vehicles (UAVs) equipped with camera for aerial surveillance is an gaining popularity worldwide. However severe vibrations and fast movements coupled with size and weight constraints presents a limit for effectiveness of UAV surveillance. Although there are several studies investigating image stabilization and mosaicing to provide a solution to this problem, most of them are at experimental phase requiring movement constraints or additional hardware installed on UAV. In order to provide a hardware independent solution to work at actual operational conditions, a novel real-time aerial image stabilization and mosaicing system is developed. System is developed for Baykar mini IHA which is the main UAV used by Turkish Military on rural operations. In order to achieve required standards, factors affecting the performance of real-time operation in real-world conditions were analyzed. Classifications of scenery encountered during flights and differences between infrared and day light images were investigated. A survey on current state of art registration algorithms is conducted and selected algorithms are tested in both in-door experiments and flight tests. Necessary optimizations and modifications are done in order to achieve a robust, accurate, real-time mosaicing and stabilization algorithm. Comparisons of several different approaches are done by using a novel mosaic quality measurement method employing printed high resolution images for ?Ground Data? and 5 axis CNC for positioning. Resultant methods are able to increase effectiveness of mini UAV surveillance beyond its current limitations and can be applied to any basic UAV configuration having a Ground Control Station computer.
Yeni heart turcica centrifugal yapay kalp pompası sisteminin ın vitro kan testleri
Heart failures are the most common diseases causing deaths globally. According to the report of International Society for Heart and Lung Transplantation, there are approximately 50,000 patients all around the world who are candidates for heart transplant and only about 5,000 heart transplantations can be performed each year in the world. According to the data announced by Turkish Ministry of Health, about 300 patients were registered in the waiting list for heart transplantation and only 63 heart transplantations were performed in 2013. Whereas, the unregistered number of patients waiting for heart transplantation in Turkey are estimated around 3,000 by the cardiovascular surgeons. Therefore, artificial heart pump and support systems become the only solution for these patients. Ventricular Assist Devices (VADs) play a vital role in the survival of these patients, however imported heart pumps are not affordable for every heart patient due to their high prices. Consequently, research on the development of VADs is of vital importance. This research is a part of developing New Heart Turcica Centrifugal (NHTC) as the first implantable left ventricular assist device (LVAD) produced in Turkey. It covers in vitro blood tests performed with the aim of evaluating the hemolytic performance of the developed LVAD and investigates the effects of different bearing designs on hemolysis. Five prototypes with different bearings were designed and manufactured for the blood tests under various conditions. The tests were conducted in a small scale test loop which was built in order to minimize the amount of blood to be used. Normalized index of hemolysis (N.I.H.) was calculated for these tests according to American Society for Testing and Materials (ASTM) F 1841-97 standards. The final prototype of this study showed excellent performance by achieving N.I.H. values of 0.002 g/100L where blood pumps are considered antitraumatic as long as their N.I.H. values remain under 0.01 g/100L.