Machinery - Repository of the Faculty of Mechanical Engineering, University of Belgrade
Not a member yet
8397 research outputs found
Sort by
Using CFD as a Replacement for Expensive Experiments in Education
No full textIn this paper, the authors analyzed the use of computational fluid dynamics (CFD) in education. The teaching of fluid mechanics today is mostly based on the theoretical approach. Although, throughout history, it has been shown that the earliest knowledge of fluid mechanics was gained through practical experience and experiments. Apart from this advantage, laboratory exercises and experiments also have numerous disadvantages. Experiments require significant financial resources, equipment and device maintenance. Many complex and specific experiments are not easy to perform in laboratory conditions. This leads to repeating the same experiments over generations. Student safety is also an important factor. During certain experiments, an increase in pressure or temperature may occur, leading to the risk of explosion or fire. Here we consider the possibility of replacing laboratory exercises by using CFD software. Computational fluid dynamics is gaining more and more importance as an alternative to classical laboratory exercises. This technology enables reliable virtual simulation of various fluid phenomena. The application of CFD in education would allow students to experiment with different parameters and scenarios without exposure to hazards, with more accurate and deeper data analysis. The paper also compares CFD software. Software is generally classified into two groups: open source and commercial software. Two open source software are presented in detail: OpenFOAM and SimFlow. On the example of airfoil NACA 0012, in both software, the simulation results were analyzed
Development of an Artificial Neural Network-Based Tool for Predicting Failures in Composite Laminate Structures
No full textComposite materials are widely used in aerospace, automotive, biomedical, and renewable energy sectors due to their high strength-to-weight ratio and design flexibility. However, their anisotropic and layered nature makes structural analysis and failure prediction challenging. Traditional methods require solving complex interlaminar stress–strain equations, demanding significant computational resources. This paper presents a bio-inspired machine learning approach, based on human reasoning, to accelerate predictions and reduce dependence on computationally intensive Finite Element Analysis (FEA). An artificial neural network model was developed to rapidly estimate key parameters—laminate thickness, total weight, maximum stress, displacement, deformation, and failure criteria—based on stacking sequence and geometry for a desired load case. Although validated using a specific composite beam, the methodology demonstrates potential for broader use in rapid structural assessment, with prediction deviations under 15% compared to FEA results. The time savings are particularly significant—while conventional FEA can take several hours or even days, the ANN model delivers accurate predictions within seconds. The approach significantly reduces computational time while maintaining precision. Moreover, with further refinement, this logic-driven model could be effectively applied to aircraft maintenance, enabling faster decision-making and improved structural reliability assessment.Grant numbers: 451-03-137/2025-03/200105 and 451-03-136/2025-03/20002
ADATA-DRIVEN OUTLOOK ON THE CHARACTERIZATION OF FERRITIC STEELS IN THE DUCTILE-BRITTLE TRANSITION REGION
No full textSince the 1980s, there has been sustained interest in the size effect, recognized as one of important outcomes of fracture mechanics. In this article, the size effect is investigated through the estimation of the Weibull cumulative distribution function (CDF) for the fracture toughness KJc defined as an elasticplastic equivalent stress intensity factor derived from the J-integral at the point of onset of cleavage fracture, Jc. This fracture toughness of ferritic steels within the ductile-brittle (DBT) transition temperature region is a stochastic extrinsic property characterized by significant experimental variability, making a statistical approach essential for accurate DBT characterization. This inherent variability and irreducible uncertainty in the fracture properties of ferritic steels align with weakest-link theory, which attributes scatter in strength and fracture toughness to flaws, inhomogeneities, or other local imperfections within the material texture. According to this theory, even a small number of defects—especially near the tip of a pre-existing macrocrack or notch can trigger localized microseparation events that may evolve into macro-fracture, particularly when dislocation activity is suppressed due to low temperature. Weibull statistics traditionally provides a framework for estimating the likelihood of encountering such defects and assessing their impact on the overall probability of cleavage fracture. Recently, the two-step-scaling (2SS) method was introduced to model size effects throughout the DBT temperature region. This method is based on weakest-link theory and Weibull distribution, incorporating the size dependence of both the scale and shape Weibull parameters within an appropriate framework. In this article, the 2SS method is examined from a practical standpoint to demonstrate its application in engineering contexts. The 2SS approach is evaluated using the EURO fracture toughness dataset, with the results underscoring its broad applicability and versatility
Prediction of technical systems availability using the simulation models based on the AI techniques and statistical methods. A case study: Bucket wheel excavator
No full textThe availability of technical systems such as the bucket wheel excavator (BWE) is one of the key indicators of their
efficiency. The BWE is a highly productive machine that works 24/7 within the so-called continuous machinery
systems known as the ECS (excavator, conveyor, stacker) and ECC (excavator, conveyor, crushing plant).
Six simulation models for predicting/calculating the BWE availability are developed based on the time picture of
its work over a four-year period. The data from the first three years are used as input data for modelling, while
the data from the fourth year are used for accuracy testing of the developed simulation models.
Simulation models differ in the way they model the times between failures and repair times, i.e., the durations of
the ``UpTime'' and ``DownTime'' states. For modelling the ``UpTime'' state, theoretical distributions (ThD), the NonHomogeneous
Poisson Process (NHPP), and Artificial Neural Networks (ANN) are employed, while for modelling
the ``DownTime'' state, ThD and ANN are utilized. Furthermore, the probability distribution of BWE failures and
stoppage occurrences, as well as the quantification of their mutual relations and influences, is provided.
The predicted change of the BWE availability in a one-year (test) period, for all six simulation models, is presented.
The best prediction results are yielded by the simulation model that uses the NHPP for modelling the ``UpTime''
state duration and the ANN for modelling the ``DownTime'' state duration.grant numbers 451-03-137/2025-03/200105, 451-03-136/2025-03/20012
Outlook for Air Taxis at the 2027 Expo in Belgrade
No full textThe 2027 Expo in Belgrade offers a valuable opportunity to showcase air taxis as an emerging urban mobility solution, a prospect actively recognized and supported by Serbia’s political leadership. This paper addresses the key challenges that must be overcome to take advantage of this opportunity. A comparison of several eVTOL (electric vertical take-off and landing) aircraft currently at an advanced stage of development is made, and their potential performance in the context of use at the upcoming World Expo in Belgrade is discussed
Monitoring of Reinforced Concrete Beam Damages Using HHT-based Acoustic Emission and Digital Imaging Correlation Technologies
No full textA comprehensive experimental method for monitoring and characterizing the damage evolution of Reinforced Concrete (RC) beams was conducted under three-point bending load, using a combination of Hilbert-Huang Transform (HHT)-based Acoustic Emission (AE) technology and Digital Image Correlation (DIC). AE sensors capture the passive elastic waves emitted during damage progression, and their signals are analyzed using Empirical Mode Decomposition (EMD) to extract key features such as frequency and amplitude. Simultaneously, DIC and strain gauges monitor the deformation and stress distribution on the beam surface and reinforcement. The experimental results demonstrate a clear correlation between AE signal characteristics and the various stages of crack initiation, propagation, and structural failure. AE signals exhibit stage-specific spectral changes, evolving from low-frequency, low-amplitude patterns during micro crack formation to high-frequency, high-amplitude signals during major crack propagation. The DIC strain fields visually capture deformation concentration zones that align well with AE signal trends. This integrated monitoring approach enables the quantitative evaluation of structural damage and offers a reliable method for health monitoring of concrete structures. Subsequently, a database of characteristic parameters for different damage stages is established, providing theoretical support for the application of HHT in analyzing non-stationary signals in Structural Health Monitoring (SHM)
Application of Artificial Intelligence in Testing Conformity with Benford’s Law in Chaotic Dynamical Systems
No full textBenford’s Law, a logarithmic distribution of leading digits, has
proven useful in domains such as fraud detection, natural sciences,
and numerical data validation. Despite its widespread empirical
success, its application to data generated by chaotic dynamical
systems remains a relatively unexplored area. Given the sensitive
dependence on initial conditions and long-term unpredictability of
chaotic systems, the question arises whether their numerical outputs
inherently align with Benford’s distribution.No. 451-03-137/2025-03/200105, dated February 4, 202
Novel Design of an Optimized Morphing EDF Duct
No full textElectric Ducted Fans (EDF) provide several advantages over conventional propellers. In hover they can provide additional thrust for the same power
and rotor diameter meaning smaller footprint of the aircraft on the take-off/landing
area. In axial flight the shroud reduces the tip losses of the rotor especially in higher
flight speeds. Unfortunately, due to the difference in velocity streamlines in both
flight regimes, a shroud optimized for hover will have negative effects on the
propulsion efficiency in axial flight. This means that a compromise must be made
in the design of a propulsion system intended for fixed wing Vertical Take-off
Landing (VTOL) aircraft with efficient axial flight capabilities for example, using
separate systems for hover and for axial flight. Here, an optimization methodology is proposed for a novel design of a morphing ducted fan. Firstly, the optimal
geometry for hover was obtained by genetic algorithm (GA) optimization. After
that, GA optimization was used to obtain the optimal geometry for cruise with the
introduction of constraints as to be able to create a geometry capable of morphing.
In the end, the internal structure of the duct which can provide accurate control of
the inlet and outlet was defined by utilizing techniques similar to the ones used in
morphing wing designs
Experimental Analysis of the Deformation and Failure of a Composite Drone Arm
No full textThis paper presents an experimental analysis of the strength of a composite drone arm. The use of composite materials is widespread in the aerospace industry, especially in designing and manufacturing unmanned aerial vehicles (UAVs). The main reason is the need to reduce the UAV’s mass without damaging the structure. Due to their characteristics, composite materials enable the structure to be lighter than it would be if it were made of conventional materials (duralumin, etc.), while at the same time the strength parameters are adequate. Experimental analysis is usually used when it is necessary to check and verify values previously obtained by analytical and numerical methods. In this paper, we first perform an experimental analysis to find the maximum value of force at which cracks will occur. The value of the obtained force will help us in further numerical analyses to optimize the structure
Concept of a reconfigurable CNC machine with distributed control
No full textThe emergence of Industry 4.0 has changed manufacturing by integrating advanced digital technologies, enabling extensive real-time interconnection between information technologies and operational technologies within Industrial Control Systems. This ubiquitous information exchange enables manufacturing systems to operate flexibly and respond quickly to changing market demands and a growing diversity of products. The reconfiguration of manufacturing systems, both physical and functional, is driven by the Industrial Internet of Things, which employ Cyber-Physical Systems as its core enabling technology. This transition initiates a shift from centralized control systems, where a single central unit manages all control tasks, to a distributed control system architecture. Within a distributed control system, the main control task is achieved through the intensive cooperation of smart devices, such as sensors and actuators, equipped with computation and communication modules. In this work, we present a concept of a Computerized Numerical Control (CNC) system with distributed control, based on the cooperation of Low-Level Controllers, which execute local tasks while receiving primary control commands from a High-Level Controller. The proposed approach considers both functional and physical properties to ensure system reconfigurability and scalability