Recent Researches in Applied Mechanics

Size: px

Start display at page:

Download "Recent Researches in Applied Mechanics"

Mitchell Thompson
6 years ago
Views:

1 Dynamic Design Verification (DDV) of vertical pump using, combined Modal Analysis (MA), Finite Element Analysis (FEA) & Operation Deflection Shape (ODS) SIAMAK NOROOZI, MIHAI DUPAC School of Design, Engineering and Computing Bournemouth University Talbot Campus, Fern Barrow, Poole, Dorset BH12 5BB UNITED KINGDOM mdupacbournemouth.ac.uk ABDUL GHAFFAR ABDUL RAHMAN Department of Mechanical Engineering University Malaysia Pahang Gambang MALAYSIA Abstract: - When a complex rotating machinery system is commissioned its overall static and dynamic characteristic changes substantially. The overall characteristic of the system is not the superposition of the characteristics of individual sub systems, and failure can occur due structural incompatibility or mismatched stiffness. In this paper a novel ODS procedure is developed and used to monitor relative in-service planar or orbital displacements of machinery for any signs of excess or incompatible displacements. In such cases, the system is taken out of service for full dynamics analysis and subsequent design modification. A combination of ODS, Modal analysis and FEA have been used to verify any design improvements. The undesirable structural dynamics is the root cause of poor reliability. This paper deals with a general problem of dynamic incompatibility exhibited by systems that are manufactured by the 3 rd parties but needs to function reliably to avoid huge losses in production. Key-Words: Modal analysis, ODS, FEM, DDV, design, vertical pump 1 Introduction Conventional condition monitoring such, FFT and orbital analysis, or trending RMS or even modal analysis [4] can be performed to check the condition of machinery and to identify any changes to the dynamics of the system in order to preventing catastrophic failures. However, vibration based condition monitoring on its own cannot predict the root causes of all failures. These causes are due to a variety of reasons such as poor design, bad manufacturing processes, poor assembly, wear and tear, imbalance, missalignment, bad practices, incompetence or inexperience as well as incompatible structures stiffness resulting in undesirable structural dynamic characteristics. This makes machinery un-reliable, reduces efficiency and makes their maintenance and monitoring an important task. Generally these systems are made out of many sub system each manufactured to high standard by different manufacturer and brought to site for final assembly. However, when these, individually sub systems are assembled together, the functional outcome became sub-optimal due to both static and dynamic incompatibility related to the miss-match stiffness or highly non uniform mass distribution. In such cases, the overall dynamic response is no longer the superposition of the individual responses and the system behaves in a completely different or unpredictable manner. Once commissioned and the system is connected to the grid and is in service conventional modal testing is no longer appropriate to test for natural frequencies, mode shapes and damping. Such systems need to be tested under operating conditions [5, 6]. A non-invasive condition based maintenance procedure is developed here that enables monitoring of in-service rotating machinery to detect any early signs of impending failure. Such signs could be due to running speed being close to some of the natural ISBN:

$Forces due to miss-alignment or imbalances can cause large relative displacements that in the long terms can result in fatigue and fracture and well as other forms of damages such as seal failure due$

2 modes of vibration of the system [11] which can also indicates either design problem or damaged structure[2]. When classical condition monitoring is not possible, the operational deflection shape (ODS) technique may be used for detecting or prediction structural weaknesses due to damage or incompatible structures parts [3, 7, 8, 9, 12]. Forces due to miss-alignment or imbalances can cause large relative displacements that in the long terms can result in fatigue and fracture and well as other forms of damages such as seal failure due to forces that are uncounted for during the design stages. These dynamic loads are hard to predict and can be magnified due to the increase in amplitude of the motion as a result of changes in the shape and subsequent loading conditions. The ODS technique is an excellent alternative to the classical monitoring [13]. Moreover ODS provide a relative dynamic 3D displacement due to mode shapes or relative orbits enabling full visualisation of the vibration of the structure [12]. This relative displacement can quickly identify structural weakness that can result in failure. This combined with FEA [12] allows structural modification that shifts close to running speed resonances well away from the running speed. In this paper a dynamic design verification (DDV) technique based on the combination of experimental modal analysis (MA), FEA model and operation deflection shapes (ODS) techniques was applied to calculate the relative movement of a vertical pump that exhibited huge vibration problem. A powerful DaisyLab based DAQ and signal processing and analysis system was developed by the authors and used here to measure the overall dynamic characteristics of the pump, and to visualize its structural dynamics in 3D space for critical frequencies and near running speed resonances identification. It was observed that the DDV process represent a valuable tool for non-invasive condition monitoring of rotating machinery with poor structural integrity resulted from mismatch stiffness and mass distribution of the key parts/subsystems within the final assembly. Moreover, the study allows a better understanding of the role of structural dynamics on systems reliability and how it helps to identify where the mechanical weaknesses results in excessive flexibility and movement causing excess wear and tear and fatigue. 2 Problem Formulation 2.1 Problem description and dynamic design verification (DDV) procedure. The need for a non-invasive testing approach using a dynamic design verification (DDV) procedure was justified by the poor reliability (excessive wear and tear) of a vertical pump shown in Figure 1. (a) (b) (c) Figure 1: (a) the pump in service, (b) solid CAD model of the pump, (c) pump mesh used for the modal analysis procedure. The operation specification of the pump under investigations, a 12-stage vertical pump having a solid shaft electric motor with the running characteristics of 3.3kV/180kW/3000RPM., is shown in Table 1. Handled Fluid C 2 H 4 Input Power (kw) 180 Operating Speed (RPM) 3000 Suction Pressure (barg) Discharge Pressure (barg) Pumping Temp. ( C) Bearing Make & Type SKF 7219 Duplex with QJ219N2MA Motor stand (Kg) Discharge head (Kg) Suction Barrel (kg) Total Weight (kg) Driver - Electric Motor (kg) 1700 Table 1: Vertical Pump Operating Data The modal analysis/frf measurement technique was used to determine the dynamic characteristics of the pump. The dynamic characteristic of a system refers to Natural Frequencies, Shape and Damping [10, 12]. The Operating Deflection Shapes (ODS) analysis was used to determine the deflection pattern of the whole pump while in operation. As mentioned in [8] ODS can be defined as the ISBN:

structure deflection at a particular frequency. To capture it, an accelerometer was used as a reference and at a reference point, while machine is running.

The process will involve the calculation of Frequency Response Function (FRF) needed for curve fitting and vibration feature extraction.

Correlation between the ODS before any design modification and the ODS after the design modifications using FEA allowed for the dynamic design verification (DDV) of the pump.

3 structure deflection at a particular frequency. To capture it, an accelerometer was used as a reference and at a reference point, while machine is running. A second tri-axial accelerometer is used at the other roving points to measure acceleration in 3 orthogonal directions. The process will involve the calculation of Frequency Response Function (FRF) needed for curve fitting and vibration feature extraction. Moreover, as detailed in [8], this applies for any forced motion, either at a moment in time or at a specific frequency. Correlation between the ODS before any design modification and the ODS after the design modifications using FEA allowed for the dynamic design verification (DDV) of the pump. The methodology for the DDV conducted here is summarised in Figure 2. Modal analysis of the system before any modification and after modification was also conducted. FEA model of the system was also created, as shown in Figure 1b. It was used for modal analysis and design modification. Once the FEA is validated and proved to show the same natural frequency and the mode shape as the real system it was used iteratively for all subsequent design modification until a frequency that is away from running speed is identified. 3 Problem Solution The modal analysis of the vertical pump, performed based on the Frequency Response Function (FRF), identified the structural resonance of the pump and the root cause of reliability problem as a near structural resonance problem, due to the low bending stiffness of thrust pot and discharge head. Figure 3a show the mode of vibration at running speed. During the operating condition, the forces generated by the motor, rotating at 3000 rpm (50 Hz) is transmitted to the thrust pot and discharge head section. The modal analysis of the vertical pump also showed the 2 nd bending mode to be at 53.0 Hz and 46.4 Hz as shown in Fig. 3b and Fig. 3c respectively. Comparing Figs. 3a, 3b and 3c, one can observe how close the mode shape at running speed is to the 2 nd bending resonances, which can only be attributed to the poor structural dynamic characteristic. 50 Hz (b 53Hz Figure 2: Schematics of the dynamic design verification (DDV) process. The structure is then dismantled and stiffeners were added by welding them to the external walls of the casing to add more stiffness needed between motor and the flange. For modal analysis the system was divided in to a series of roving point and one impact point. Connecting these roving point, generates the wired mesh representing the pump, in ME scope as shown in Figured 1c. The impact was applied using impact hammer and acceleration from all these points where measure. Every impact was repeated 20 time and results were averaged 20 times. The FRF for each roving point and in 3 orthogonal directions for all the roving points were then collected and transferred to ME Scope for a 3D visualisation of the system dynamics. Hz. Figure 3: Operating shapes close to running speed, (a) operating shape 50 Hz, (b) operating 53 Hz, and (c) operating 46.4 Hz With reference to the design of the pump shown in Fig. 1, a finite element analysis (FEA) of the model was performed and correlated with the FRF modal analysis. This correlations represents an important part of the dynamic design verification (DDV) process, as it allows detailed analysis of any design changes, and their effect on dynamic response, before any design modifications. ISBN:

The optimised FEA model of the pump, shown in Fig. 4 showed the 2 nd bending mode to be of the same order of magnitude to that identified by modal analysis and ODS.

It was also identified to have the anti-nodes located at the thrust pot, which would eventually cause the mechanical seal and other components to wear, and may lead to the failure of the pump.

4 The optimised FEA model of the pump, shown in Fig. 4 showed the 2 nd bending mode to be of the same order of magnitude to that identified by modal analysis and ODS. Figure 4: FEA showing structural weakness The animated model clearly showed that the structural movement during the operation is dominated by the 2 nd bending mode. It was also identified to have the anti-nodes located at the thrust pot, which would eventually cause the mechanical seal and other components to wear, and may lead to the failure of the pump. Since fatigue failure is a function of the stress amplitude, with stresses roughly below 40% of the yield strength, for the presented case the fatigue failure was not considered to be an issue. 4 Results and Discussion Figure 6 shows the modal frequencies around the running speed. It is quite clear from this that static structure has resonances close to the running speed of 50 Hz. 1 st Bending Twisting 1 st Bending 2 nd Bending Figure 5: FRF Modal spectrums for the Vertical Pump The modal analysis results of the two modes of interest, i.e., the 2 nd bending modes which may cause uneven load transmission and subsequently excess wear and tear, are shown in Table 2. Shape Frequency Damping 1st Bending 7.58 Hz 5.35 Twisting 33.9 Hz nd Bending 46.4 Hz 1.3 2nd Bending 53 Hz 2.65 Table 2: Modal frequencies results Operating Deflection Shape Analysis ODS analysis was performed on the vertical pump from the motor non-drive end to the base. In Fig. 5 the ODS spectrum of the pump is presented. It was observed that the motion was dominated by the running speed of the motor at 50Hz or 3000RPM, mainly in-line with the suction-discharge pipe (Figure 4). Figure 5: Vibration velocity spectrums for the Vertical Pump Based on the results of all the above analyses, the overview of the mechanics can be analyzed. During the operating condition, the forces generated by the motor, rotating at 3000 rpm (50 Hz) is transmitted to the thrust pot and discharge head section, which was clearly shown from ODS analysis of the running speed. It is also clear that the pump is operating the 2 nd bending mode which is close to the running speed which dominated the movement. From modal analysis, the natural frequency of the 2nd bending mode occurred at two instances; 46.4Hz and 53Hz, whose anti node is at the region of the thrust pot - discharge head interface causing near-resonance to the system. The thrust pot - discharge head of the pump are dynamically weak burdened by the weight of the massive motor. Therefore, the thrust pot needs to be stiffened while ensuring that the stiffening should avoid other natural frequencies (twisting mode) from entering the running speed region. Dynamic design verification (DDV) criteria indicated the need for corrective measure that would ISBN:

5 shift the 2 nd bending mode and frequencies of the pump from 46.4Hz and 53Hz, away from the running speed of 50Hz whilst ensuring that the 33.9Hz twisting mode will not enter into the pump operating region. A structural dynamic modification, based on the dynamic design criterion, that would shift the 2 nd bending mode of the vertical pump at 50.4Hz, away from the running speed up to above 70Hz frequency while ensuring that the 32Hz twisting mode will not enter into the pump operating region, was considered. The indicated change was initially performed using the verified FE model of the pump, which had similar resonances to the actual pump. The outcome - based on a simple modification of the pump trust pot - helped to shift the resonant frequencies from 46.4 Hz to76.2hz while the twisting mode resonance changed slightly from 33.9Hz to 36.7Hz, as shown in Table 3. Shape Twisting mode 2nd Bending mode Modal FRF (Hz) Modal FEA (Hz) SDM (Hz) & Table 3: Modal Results Summary based on the proposed structural dynamic modification (SDM) 5 Conclusions In this study the reliability of a vertical pump was investigated using dynamic design verification (DDV) technique based on modal analysis (MA) and operation deflection shapes (ODS). Based on the performed modal analysis, it was concluded that the pump has low reliability due to its structural natural frequency being close to the running speed of the pump causing near-resonance situation / excitation during operation. Results from ODS analysis, showed that the structural movements - dominated by the 2 nd bending mode - would eventually caused the mechanical seal and other components to wear and leading to the frequent failure of the vertical pump. Using a dynamic design criterion the 2 nd bending mode was shifted away from the running speed and from other natural frequencies, while ensuring that the twisting mode will not enter into the pump operating region, and thus removing the excessive cause of possible wear and failure. References: [1] I. Bucher, D.J. Ewins, Modal Analysis and testing of rotating structures, Philosophical Transactions: Mathematical, Physical and Engineering Sciences The Royal Society London, Vol. 351, 2001, pp [2] S.W. Doebling, C.R. Farrar, M.B. Prime, D.W. Shevitz, Damage identification and health monitoring of structural and mechanical systems from changes in their vibration characteristics: A literature review, Report No. LA MS, Los Alamos National Laboratory, [3] O. Dossing, C.H. Staker, Operational deflection shapes: background, measurement and applications, 5th International Modal Analysis Conference, London, UK, 1987, [4] D.J. Ewins, Modal Testing: Theory, Practice and Application, Research Studies Press Ltd., 2000 [5] L. Hermans, V.D. Auweraer, Modal testing and analysis of structures under operational Conditions: industrial applications, Journal of Mechanical Systems and Signal Processing, Vol. 13, 1999, pp [6] G.H. James, T.G. Carne, J.P. Laufer, The natural excitation technique (NExT) for modal parameter extraction from operating structures, Journal of Analytical and Experimental Modal Analysis, Vol. 10, 1995, pp [7] W.D. Marscher, C. W. Jen, Use of operating deflection and mode shapes for machinery diagnostics, Proceedings of 17 th International Modal Analysis Conference, 1999, pp [8] P.L. McHargue, M.H. Richardson, Operating deflection shapes from time versus frequency domain measurements, 11 th International Modal Analysis Conference, Kissimmee, USA, 1993, pp [9] R. Pascual, J.C. Golinval, M. Razeto, On-line damage assessment using operating deflection shapes, Proceedings of 17 th International Modal Analysis Conference, 1999, pp [10] S.S. Rao, Mechanical Vibration, SI Edition, Prentice Hall, 2005 [11] E. Swanson, C.D. Powell, S. Weissman, A practical review of rotating machinery Critical Speeds and s, Journal of Sound and Vibration, Vol. 39, No. 5, 2005, pp [12] B.H. Tongue, Principles of Vibration, Oxford University Press, 2nd ed., [13] H. Vold, B. Schwarz, M. Richardson, Measuring operating deflection shapes under non-stationary conditions, Proceedings of the ISBN:

6 International Modal Analysis Conference IMAC, San Antonio, TX, Volume 2, 2000, pp [14] K. Waldron, A. Ghoshal, M.J. Schulz, M.J. Sundaresan, F. Ferguson, P.F. Pai, J.H. Chung, Damage detection using finite element and laser operational deflection shapes, Finite Elements in Analysis and Design, Vol. 38, pp , ISBN:

Calculating Stress & Strain from Experimental ODS Data

Calculating Stress & Strain from Experimental ODS Data Brian Schwarz, Shawn Richardson, Mark Richardson Vibrant Technology, Inc. Scotts Valley, CA Click here for the presentation ABSTRACT In this paper