Simulation and Analysis of Dual Unbalanced Rotor Effects on Natural Frequency in a Digital Twin Shaft Model Fadhel Abbas Abdulla1,* , Ahmed Imad Abbood1 1 Department of Mechanical Engineering, Mustansiriyah University, Baghdad, Iraq Abstract Blades are one of the basic components of a gas turbine and its main function is to rotate the shaft associated with the generator motor. Gas turbine model MS9001E used power plants at south Baghdad station, the blades are subjected to harsh working conditions such as high vibration, temperatures and pressures, thus highlighting the importance of studying the materials used in Manufacture of blades that work under harsh operating conditions. In this research, stress, strain and deformation produced by the centrifugal force that the blade is subjected to be studied, as well as studying the natural frequencies of the blades. Three-dimensional was created through the program solidwork 2018 and then exported to the program ansys 2019 for analyzing. Two alloys of materials (GTD-111) and (IN-738) were analyzed and compared between them, and the results showed that alloy (GTD-11) is the best and is suitable for use in the manufacture of blades. Keywords Rotors, Natural frequency, Deformation, vibrations frequency, Modal analysis 1. Introduction den accidents. Numerous researchers have focused on investigating the impact of cracks on the efficiency of The dynamic rotor plays a crucial role in the behavior rotating shafts. Some have conducted analytical analyses of rotary machines, ranging from large-scale systems to study these issues, while others have approached them like power plant rotors and turbo-generators to smaller approximately. One key factor in minimizing undesired systems such as tooth drills, pumps, and air compressors vibrations is effectively controlling the rotor’s geometric [1, 2, 3]. Understanding the history of rotor dynamics is imbalance [20, 21, 22]. By employing calculations and essential as it highlights the fundamental challenges in understanding the mass of unbalance during rotation, developing and implementing rolling bearings for vari- it is possible to measure the vibration response of any ous applications [4, 5, 6, 7, 8] when stability and quality system. Multiple researchers have conducted studies on must be assured [9]. The study of dynamic behavior the impact of externally applied axial force and torque on in rotating machinery began in the early years of the the lateral vibration of shafts. Alaa et al. [23] derived the 19th century when the industrial revolution increased equation of motion for a flexible rotating shaft subjected the demand for analyzing rotational motion also in the to a constant compressive axial load by incorporating robotic industry [10, 11, 12, 13]. Since the fifties, numer- gyroscopic moments consistently. In the study [24] exam- ous researchers have conducted studies on crack prop- ined the stability of a rotating cantilever shaft carrying agation in shafts, and some of these findings have been a rigid disk at its free end, considering follower axial extended to real-world rotors, providing valuable insights force and torque loads. Chen and Sheu [25] analytically for designers [14, 15, 16]. Typically, rotors operate un- investigated the stability behavior of a rotating Timo- der cyclic pressure, making them susceptible to various shenko shaft with an intermediate attached disk under operational issues, such as fatigue cracks. These cracks longitudinal force, providing frequency equations for var- tend to occur when the rotors’ natural frequencies and ious boundary conditions and numerically determining critical speeds increase as the shaft length decreases and critical axial and follower forces. Chen et al. [26] The the cross-sectional area increases [17]. It is crucial to influence of inertial forces on shafts and beams can result identify the vibration characteristics of cracked shafts in axial stresses. Researchers have also examined the to develop a control system that can detect operational rotation of beams around an axis perpendicular to their errors and early-stage cracks [18, 19] and prevent sud- beam axis, where centrifugal force directly induces axial stress in the beam. In this study [27] utilized the dynamic ICYRIME 2024: 9th International Conference of Yearly Reports on Informatics, Mathematics, and Engineering. Catania, July 29-August stiffness matrix for an Euler-Bernoulli beam subjected to 1, 2024 axial force to analyze the vibration of rotating uniform * Corresponding author. and tapered beams. In the investigation reported [28, 29] $ fadhel975@uomustansiriyah.edu.iq (F. A. Abdulla); derived the governing equation for the linear vibration of ahmed.abbood@uomustansiriyah.edu.iq (A. I. Abbood) a rotating Timoshenko beam, considering the coupling  0000-0002-7840-0659 (F. A. Abdulla); 0009-0006-8763-7160 (A. I. Abbood) between extensional and flexural deformation by lineariz- © 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License ing the fully geometrically nonlinear beam theory. They Attribution 4.0 International (CC BY 4.0). 32 CEUR ceur-ws.org Workshop ISSN 1613-0073 Proceedings Fadhel Abbas Abdulla et al. CEUR Workshop Proceedings 32–37 proposed a power series solution method to determine of 1200mm, while the rotors have diameters of 230mm, the natural frequency of the rotating Timoshenko beam. as shown in Figure (1a). The bearings are positioned at This study introduces a new phenomenon that can influ- three locations, with the midspan position being partic- ence the performance of rotating shafts. The change in ularly interesting for identifying the optimal position. natural frequency is attributed to the position of rotors The system’s geometry was created using SolidWorks and the midspan between double rotors. Unlike previous 2020 and then exported to ANSYS 2019 software. The studies, the disks located equal spaces between fixtures rotor system geometry was discretized in the initial stage and rotors. To accomplish this objective, the lateral de- using tetrahedral elements. The model comprises 160,022 formation and natural frequency need to be calculated, nodes and 40,196 elements, as depicted in Figure (1b). then evaluating the impact of this distance on the lat- eral natural frequency of the shaft. The paper presents a numerical solution to evaluate the systems. 2. Materials and Method A modal analysis was conducted on a structural steel shaft, and commercially available rotors were used. The material properties of the structural steel are provided in Table 1. Finite element analysis (FEA) was chosen as it offers more comprehensive results compared to ex- perimental studies, with the added benefits of speed and cost-effectiveness [30, 31, 32] also in term of missing data reconstruction or imputation [33, 34]. The FEA employed (a) a finite element discretization approach to solve complex structural equations by dividing the structure into spe- cific finite elements[31, 32, 35, 36]. The unbalanced rotors were designed using FEA, utilizing a mesh system com- posed of interconnected nodes. The model, created in SOLIDWORKS 2020, was exported to ANSYS 2019 soft- ware. The ANSYS model then meshed, and boundary conditions were applied. The software solved the system equations to determine the model’s natural frequencies. Table 1 Material Properties of Rotating shaft and Rotors PROPERTIES VALUE Specific Heat (J/Kg K) 485 Young’s Modulus (GPa) 210 Density (kg/m3) 7850 Poison’s ratio 0.30 Thermal conductivity (W/mk) 60 Thermal expansion (oC) 14 × 10−6 Yield strength (Mpa) 450 (b) Shear Modulus (GPa) 80 Figure 1: a. The geometry of rotor system in SOLIDWORKS, b. Meshed model using ANSYS 3. Modelling and Analysis of The geometry of the rotor is affected by the unbalanced mass and the rotating velocity of 200 RPM in ANSYS Rotating shaft and Dual Rotors Modal. The boundary conditions shown in Figure (2), the The model developed aims to simulate rotating machin- two ends of the shaft are fixed support and the position ery with unbalanced rotating components. The model of bearings are BEARING boundary conditions in ANSYS consists of two main components to simplify the sys- Model. The geometry of the rotor and shaft are meshed tem: the rotors and the shaft. The shaft has a length with ELEMENT 186 to adept to different shapes. ANSYS Modal analysis uses the following equations 33 Fadhel Abbas Abdulla et al. CEUR Workshop Proceedings 32–37 (a) 1𝑠𝑡 mode 162.43Hz Figure 2: Boundary conditions in ANSYS. to solve vibration problems: 1. Mass Matrix Equation: [𝑀 ]𝜓 + [𝐾]𝜓 = 0 2. In this equation: [M] is the mass matrix, which represents the distribution of masses in the system. 𝜓 is the vector of mode shapes or modal displacements. [K] (b) 2𝑛𝑑 mode 418.48Hz is the stiffness matrix, which represents the stiffness of the system. 3. Eigenvalue Equation: [𝐾]𝜓 = 𝜆[𝑀 ]𝜓 4. In this equation: 𝜆 represents the eigenvalues, which determine the natural frequencies of the system. [K] is the stiffness matrix. 𝜓 is the vector of mode shapes. By solving the above equations, ANSYS Modal analy- sis calculates the natural frequencies (eigenvalues) and corresponding mode shapes (modal displacements) of the system. (c) 3𝑟𝑑 mode 569.65Hz 4. Results and Discussion The outcomes derived from ANSYS 2019R3 depend on various elements such as the system’s shape, length of the shaft, elastic characteristics, rotor spacing, and boundary conditions. These factors impact the inherent frequency of the system. A higher inherent frequency signifies an improved design by reducing vibration amplitude and (d) 4𝑡ℎ mode 601.04Hz lowering the likelihood of component malfunction [37]. The modal analysis offers a valuable understanding of these outcomes, aiding in assessing and enhancing the system’s performance. Figure (3) shows the mode shapes for the first five modes for the 916mm midspan. Defor- mation (11.69 mm) at a frequency (162.43Hz) 1st mode, and deformation (12.22 mm) at a frequency (418.48Hz) 2nd mode. The following three modes show an increase (e) 5𝑡ℎ mode 648.08Hz in frequency, and deformation will reach the peak at 4th mode for the 3rd ,4th, and 5th modes; the frequency and deformation are 569.65Hz, 35.22mm; 601.04Hz, 35.74mm; Figure 3: The first five mode shapes of rotor midspan 916mm which are extracted by ANSYS19, are labeled (a) to (e) respec- and 648.08Hz, 29.22mm respectively. tively 34 Fadhel Abbas Abdulla et al. CEUR Workshop Proceedings 32–37 (a) (b) Figure 4: Shape modes and the Natural frequency in (Hz). Figure (4a) shows the change in natural frequency of 5. Conclusions the unbalanced rotating shaft with the mode shape num- ber. It is noticed that the 1st and 5th modes are having The following conclusions can be derived from the find- same value, the change in natural frequency occur in ings of this study: the other three modes. Figure (4b) gives a lock at the 1. Using midspan at the highest value increase the three modes 2nd,3rd, and 4th, for midspan 916mm gives value of natural frequency and reduces the effect of highest frequency in the second mode while lowest fre- whirling. quency for the midspan 756mm for the same mode. The 2. The position of the rotors does affect much on the third mode midspan 796mm and 756mm approximately rotating shaft’s natural frequency because of the deforma- have the same frequency about 570 Hz and the 796mm tion in the rotating disk, which has the same dimensions. midspan has same value with 756mm midspan. 836mm 3. The second mode is affected by the position of rotors midspan has a value between them. 4th mode midspan which give high natural frequency at a high midspan 756mm has the lowest frequency 591 Hz and the mid value. span 836mm has the highest value 607Hz. 4. As well as the distance between two rotating discs is higher, the system balance increases too. In future works, new optimization methods based on 35 Fadhel Abbas Abdulla et al. CEUR Workshop Proceedings 32–37 neural approaches will be explored [38, 39]. impaired people with a following robotic assistant by means of end-to-end visual target navigation and reinforcement learning approaches, in: CEUR References Workshop Proceedings, volume 3118, 2021, p. 51 – 63. [1] M. J. Jweeg, S. N. Alnomani, S. K. Mohammad, Dy- [12] G. Capizzi, C. Napoli, S. Russo, M. Woźniak, Lessen- namic analysis of a rotating stepped shaft with and ing stress and anxiety-related behaviors by means without defects, in: IOP Conference Series: Ma- of ai-driven drones for aromatherapy, in: CEUR terials Science and Engineering, volume 671, IOP Workshop Proceedings, volume 2594, 2020, p. 7 – Publishing, 2020, p. 012004. 12. [2] A. A. Shandookh, A. A. F. Ogaili, L. A. Al-Haddad, [13] C. Napoli, C. Napoli, V. Ponzi, A. Puglisi, S. Russo, Failure analysis in predictive maintenance: Belt I. E. Tibermacine, Exploiting robots as healthcare drive diagnostics with expert systems and taguchi resources for epidemics management and support method for unconventional vibration features, He- caregivers, in: CEUR Workshop Proceedings, vol- liyon (2024). ume 3686, 2024, p. 1 – 10. [3] A. A. F. Ogaili, M. N. Hamzah, A. A. Jaber, Enhanced [14] M. L. Cappuccio, J. C. Galliott, F. Eyssel, A. Lanteri, fault detection of wind turbine using extreme gra- Autonomous systems and technology resistance: dient boosting technique based on nonstationary new tools for monitoring acceptance, trust, and vibration analysis, Journal of Failure Analysis and tolerance, International Journal of Social Robotics Prevention 24 (2024) 877–895. (2023) 1–25. [4] L. A. Al-Haddad, A. A. Jaber, Applications of ma- [15] F. Bonanno, G. Capizzi, S. Coco, C. Napoli, A. Lau- chine learning techniques for fault diagnosis of dani, G. L. Sciuto, Optimal thicknesses determi- uavs., SYSTEM (2022) 19–25. nation in a multilayer structure to improve the [5] L. A. Al-Haddad, L. Ibraheem, A. I. EL-Seesy, A. A. spp efficiency for photovoltaic devices by an hy- Jaber, S. A. Al-Haddad, R. Khosrozadeh, Thermal brid fem - cascade neural network based approach, heat flux distribution prediction in an electrical 2014, pp. 355 – 362. doi:10.1109/SPEEDAM.2014. vehicle battery cell using finite element analysis 6872103. and neural network, Green Energy and Intelligent [16] A. A. F. Ogaili, M. N. Hamzah, A. A. Jaber, Inte- Transportation (2024) 100155. gration of machine learning (ml) and finite element [6] A. A. F. Ogaili, K. A. Mohammed, A. A. Jaber, A. E. S. analysis (fea) for predicting the failure modes of Al, Automated wind turbines gearbox condition a small horizontal composite blade, International monitoring: A comparative study of machine learn- Journal of Renewable Energy Research (IJRER) 12 ing techniques based on vibration analysis, FME (2022) 2168–2179. Transactions 52 (2024) 471–485. [17] A. A. F. Ogaili, A. A. Jaber, M. N. Hamzah, Statis- [7] R. Giuliano, F. Mazzenga, A. Vizzarri, Integration tically optimal vibration feature selection for fault of broadcaster and telco access networks for real diagnosis in wind turbine blade, Int. J. Renew. En- time/live events, IEEE Transactions on Broadcast- ergy Res 13 (2023) 1082–1092. ing 66 (2020) 667–675. [18] G. Lo Sciuto, G. Capizzi, R. Shikler, C. Napoli, Or- [8] L. Cotugno, F. Mazzenga, A. Vizzarri, R. Giuliano, ganic solar cells defects classification by using a The major opportunities of blockchain for automo- new feature extraction algorithm and an ebnn with tive industry: a review (2021) 1–6. doi:10.23919/ an innovative pruning algorithm, International AEITAUTOMOTIVE52815.2021.9662907. Journal of Intelligent Systems 36 (2021) 2443–2464. [9] C. Napoli, G. Pappalardo, E. Tramontana, A hy- [19] G. Capizzi, G. L. Sciuto, C. Napoli, R. Shikler, brid neuro-wavelet predictor for qos control and M. Wozniak, Optimizing the organic solar cell stability, in: Lecture Notes in Computer Sci- manufacturing process by means of afm measure- ence (including subseries Lecture Notes in Arti- ments and neural networks, Energies 11 (2018). ficial Intelligence and Lecture Notes in Bioinfor- doi:10.3390/en11051221. matics), volume 8249 LNAI, 2013, p. 527 – 538. [20] G. C. Cardarilli, L. Di Nunzio, R. Fazzolari, M. Re, doi:10.1007/978-3-319-03524-6_45. F. Silvestri, S. Spanò, Energy consumption sav- [10] N. Boutarfaia, S. Russo, A. Tibermacine, I. E. Tiber- ing in embedded microprocessors using hardware macine, Deep learning for eeg-based motor imagery accelerators, TELKOMNIKA (Telecommunication classification: Towards enhanced human-machine Computing Electronics and Control) 16 (2018) 1019– interaction and assistive robotics, in: CEUR Work- 1026. shop Proceedings, volume 3695, 2023, p. 68 – 74. [21] A. A. F. Ogaili, A. A. Jaber, M. N. Hamzah, A method- [11] N. N. Dat, V. Ponzi, S. Russo, F. Vincelli, Supporting ological approach for detecting multiple faults in 36 Fadhel Abbas Abdulla et al. CEUR Workshop Proceedings 32–37 wind turbine blades based on vibration signals and single-disc and single-span rotor-bearing system machine learning, Curved and Layered Structures with rayleigh beam model, Mechanical Systems 10 (2023) 20220214. and Signal Processing 116 (2019) 322–346. [22] S. Bhide, S. Katekar, Dynamic analysis of rotat- [33] G. Capizzi, C. Napoli, L. Paternò, An innovative ing stepped shaft with transverse crack, mechani- hybrid neuro-wavelet method for reconstruction cal dept. skn sinhgad college of engineering korti, of missing data in astronomical photometric sur- pandharpur, Journal of Engineering Research and veys, in: Lecture Notes in Computer Science (in- Applications 5 (2015) 78–82. cluding subseries Lecture Notes in Artificial Intel- [23] A. J. Abdulah, M. Z. Khalifa, A. J. O. Hanfesh, Re- ligence and Lecture Notes in Bioinformatics), vol- ducing vibrations generated in a gas turbine model ume 7267 LNAI, 2012, p. 21 – 29. doi:10.1007/ ms9001e used in south baghdad power plant sta- 978-3-642-29347-4_3. tion by improving the design of bearings with [34] C. Napoli, G. Pappalardo, E. Tramontana, Improv- damper, Engineering and Technology Journal 39 ing files availability for bittorrent using a diffu- (2021) 1454–1462. sion model, in: Proceedings of the Workshop on [24] M. Phani Kumar, P. Samanta, N. Murmu, Rigid Enabling Technologies: Infrastructure for Collab- rotor stability analysis on finite hydrostatic double- orative Enterprises, WETICE, 2014, p. 191 – 196. layer porous oil journal bearing with velocity slip, doi:10.1109/WETICE.2014.65. Tribology Transactions 58 (2015) 930–940. [35] F. A. Abdulla, M. Qasim, A. A. F. Ogaili, Influ- [25] C.-W. Chang-Jian, Gear dynamics analysis with ence eggshells powder additive on thermal stress of turbulent journal bearings mounted hybrid squeeze fiberglass/polyester composite tubes, in: IOP Con- film damper—chaos and active control analysis, ference Series: Earth and Environmental Science, Journal of Computational and Nonlinear Dynamics volume 877, IOP Publishing, 2021, p. 012039. 10 (2015) 011011. [36] A. A. F. Ogaili, A. A. Jaber, M. N. Hamzah, Wind [26] A. Vaziri, M. Patil, Vibration analysis of a cracked turbine blades fault diagnosis based on vibration shaft, International Journal of Advanced Engineer- dataset analysis, Data in Brief 49 (2023) 109414. ing Technology (2013). [37] A. Ogaili, M. Hamzah, A. Jaber, Free vibration anal- [27] L. Canese, G. C. Cardarilli, L. Di Nunzio, R. Fazzo- ysis of a wind turbine blade made of composite lari, M. Re, S. Spanò, A hardware-oriented qam materials, in: International Middle Eastern Simula- demodulation method driven by aw-som machine tion and Modeling Conference, volume 2022, 2022, learning, in: 2023 57th Asilomar Conference on pp. 27–29. Signals, Systems, and Computers, IEEE, 2023, pp. [38] G. Lo Sciuto, G. Capizzi, S. Coco, R. Shikler, Ge- 937–941. ometric shape optimization of organic solar cells [28] G. C. Cardarilli, L. Di Nunzio, R. Fazzolari, D. Gi- for efficiency enhancement by neural networks, in: ardino, M. Re, A. Ricci, S. Spano, An fpga-based Advances on Mechanics, Design Engineering and multi-agent reinforcement learning timing synchro- Manufacturing: Proceedings of the International nizer, Computers and Electrical Engineering 99 Joint Conference on Mechanics, Design Engineer- (2022) 107749. ing & Advanced Manufacturing (JCM 2016), 14-16 [29] S. S. Shijer, A. H. Jassim, L. A. Al-Haddad, T. T. September, 2016, Catania, Italy, Springer, 2017, pp. Abbas, Evaluating electrical power yield of pho- 789–796. tovoltaic solar cells with k-nearest neighbors: A [39] G. Lo Sciuto, G. Susi, G. Cammarata, G. Capizzi, A machine learning statistical analysis approach, e- spiking neural network-based model for anaerobic Prime-Advances in Electrical Engineering, Electron- digestion process, in: 2016 International Sympo- ics and Energy (2024) 100674. sium on Power Electronics, Electrical Drives, Au- [30] W. Qin, G. Chen, G. Meng, Nonlinear responses tomation and Motion (SPEEDAM), IEEE, 2016, pp. of a rub-impact overhung rotor, Chaos, Solitons & 996–1003. Fractals 19 (2004) 1161–1172. [31] O. E. Efe-Ononeme, A. Ikpe, G. O. Ariavie, Modal analysis of conventional gas turbine blade materials (udimet 500 and in738) for industrial applications, Journal of engineering technology and applied sci- ences 3 (2018) 119–133. [32] A. Wang, W. Yao, K. He, G. Meng, X. Cheng, J. Yang, Analytical modelling and numerical exper- iment for simultaneous identification of unbalance and rolling-bearing coefficients of the continuous 37