IOMAC'13 5 th International Operational Modal Analysis Conference 2013 May 13-15 Guimarães - Portugal IMPACT-SYNCHRONOUS MODAL ANALYSIS (ISMA) AN ATTEMPT TO FIND AN ALTERNATIVE Abdul Ghaffar Abdul Rahman 1, Zubaidah Ismail 2, Siamak Noroozi 3, Ong Zhi Chao 4 ABSTRACT Modal frequencies, modal shapes and modal damping comprehensively define the dynamic characteristics of a structure. Prior to developing a mathematical model of the dynamic behaviour of any structure, these parameters need to be firstly determined. Currently, the two techniques used to extract these parameters are the classical Experimental Modal Analysis (EMA) and the Operational Modal Analysis (OMA). The fundamental difference in the two techniques lies in the manner by which the system is excited. A novel method called Impact-Synchronous Modal Analysis (ISMA) utilising the modal extraction techniques commonly used in EMA but performed in the presence of the ambient forces is proposed. The differences between EMA and ISMA lie in the averaging technique used. ISMA utilised Impact-Synchronous Time Averaging (ISTA) triggered by the impulse of the artificial force introduced by the impact hammer prior to performing the Fast Fourier Transform (FFT) instead of frequency domain averaging adopted by conventional EMA. Results showed that modal parameters were successfully determined in the presence of periodic responses of cyclic loads and ambient excitation using ISMA in both laboratory rig and industrial machinery. The well correlated results with classical EMA during static condition show the effectiveness of performing ISMA in the presence of the unaccounted forces. Keywords: Experimental Modal Analysis, modal frequency, modal damping, modal shape, Impact- Synchronous Modal Analysis, Impact-Synchronous Time Averaging 1. INTRODUCTION Three parameters namely modal frequencies, modal shapes and modal damping comprehensively define the dynamic characteristics of a structure. Prior to developing a mathematical model of the dynamic behaviour of any structure, these parameters need to be determined. Currently, the two techniques used to extract these parameters are the classical Experimental Modal Analysis (EMA) and 1 Professor, Faculty of Mechanical Engineering, University Malaysia Pahang, dragrahman@yahoo.com 2 Associate Professor, Civil Engineering Department, Faculty of Engineering, University of Malaya, zu_ismail@um.edu.my 3 Professor, School of Design, Engineering and Computing, Bournemouth University, snoroozi@bournemouth.ac.uk 4 Mechanical Engineering Department, Faculty of Engineering, University of Malaya, zhichao83@gmail.com
A.G.A. Rahman,Z. Ismail,S. Noroozi, O.Z. Chao the Operational Modal Analysis (OMA). The fundamental difference between the two techniques lies in the manner by which the system is excited. EMA requires the system to be in complete shutdown mode. In other words, there should be no unaccounted excitation force induced into the system. Measurable impact or random forces are used to excite the system. Cross-spectrum of measured inputs and system s responses produce the transfer functions at a discrete set of geometrical positions. Various curve-fitting techniques are then used to extract the three parameters. In practical situations where the system cannot be shut down completely or the structure is too huge to response to artificial excitation, OMA is sought. The operating forces coming from the machine cyclic loads or ambient forces are used as the exciters. As these quantities cannot be measured, OMA utilized only information from the measurable responses (multi-output data) and some algorithms [1-5] are used to extract the three modal parameters i.e. modal frequencies, modal shapes and modal damping. In OMA, structural modal parameters can be computed without knowing the input excitation to the system. Therefore, it is a valuable tool to analyse structures subjected to excitation generated by their own operation. Presently, OMA procedures are limited to cases where excitation to the system is white stationary noise [6]. The main advantage of OMA is that the measured responses can be used for modal identification of structures under real operation without requiring excitation from a hammer and shaker. Therefore, the measured responses are representative of the real operating conditions of the structure. Although OMA holds advantage over EMA in terms of its practicality and simplicity to carry out the procedure and performing the analysis while the system is in operation, the lack of knowledge of the input forces does affect the parameters extracted. For example, mode shapes obtained from OMA cannot be normalised accurately. This results in approximations of the mathematical models. Over the years, as the part of the effort to improve the estimation accuracy in OMA and EMA, the focus was mainly on the development of modal identification algorithms. Relatively less effort was put into improving the digital signal processing aspects, especially upstream of the collected data. In this research, a novel method, named Impact-Synchronous Modal Analysis (ISMA) that utilises Impact-Synchronous Time Averaging (ISTA) [7, 8] is proposed. ISMA has the advantages of the OMA and EMA combined. It carries out the analysis while the system is in operation and at the same time is able to provide the actual input forces in the transfer functions, hence, allowing for better modal extractions. This novel technique can be applied in both rotor and structural dynamic systems to obtain the dynamic characteristics of structures without disturbing the operations. This is very crucial for the industrial plants especially those high downtime cost rotating machinery. In petrochemical plants, the downtime cost alone can go up to USD 6,000 to 90,000 per hour. 2. IMPACT-SYNCHRONOUS MODAL ANALYSIS (ISMA) In Impact-Synchronous Modal Analysis (ISMA)[8], when analysis is performed while machine is in running condition, all the responses contributed by the unaccounted forces are filtered out in the time domain, leaving only the responses triggered due to the impact hammer. This is achieved by utilising the Impact-Synchronous Time Averaging (ISTA) [7] prior to performing the Fast Fourier Transformation (FFT) operation. The process of modal parameters extraction follows the Experimental Modal Analysis (EMA) procedures. In commonly used time domain synchronous averaging, signal acquisition from rotating machine is triggered at the same rotational position of the shaft using a tachometer for every cycle. The time block of the averaged signal eliminates all the non-synchronous and random components, leaving behind only components that are integer multiples of the running speed. In ISMA, the same and simple averaging concept is used but only to achieve the reverse i.e. to filter out all the speed synchronous and random signatures. In this case, data acquisition is triggered by the impact signature. The periodic signatures and their harmonics are no more in the same phase position 2
5 th International Operational Modal Analysis Conference, Guimarães 13-15 May 2013 for every time block acquired. Equation (1) shows that averaging process will slowly diminish these non-synchronous components hence leaving behind only the structure s response to impulses which are synchronous to the repetitive impact force. N 1 1 y ( t) x( t rt o ) (1) N r 0 where: y(t) is the averaged vibration signal in time domain, N is the total number of impacts, x (t) is the vibration signal in time domain, r is number of impact and T o is the time interval between impacts. Cross spectrum of the averaged time block of impulse responses and the averaged time block of impact signatures is used to generate the transfer function. In discussions on limitations of Impact-Synchronous Modal Analysis (ISMA), it is noticed that ISMA gives more prominent results with higher number of impacts. However, responses from unaccounted forces that contain even the same frequency as that contained in the impulse response could not be diminished if the impact frequency is consistent with respect to the impact signature although high number of impacts is taken. Information of the cyclic force is also important. ISMA has limitation in its application on high speed machines and large structures where the impact response may be too small as compared to the operating cyclic loads while excessive impacts may result in non-linearity. Finally, in signal processing aspect, application of exponential window is necessary in ISMA to attenuate the response signals generated by cyclic loads. Therefore, parameters like number of impacts, impact frequency, impact force level and exponential window are important in ISMA in order to improve the technique. 3. LABORATORY TEST In this research, a motor-driven test rig consists of a motor coupled with rotor shaft system as shown in Figure 1 was used to study the effectiveness of Impact-Synchronous Modal Analysis (ISMA) during operating condition compared to Experimental Modal Analysis (EMA). A total of 20 points were selected on the structural plate of the test rig. Figure 1 Measurement Points and Locations of Motor-driven Test Rig Figure 2 shows the instrumentation set-up. The measured input is force from the impact hammer and the measured output is acceleration from the accelerometer. Data were obtained by using a DASYLab- National Instruments data acquisition system together with an impact hammer and tri-axial accelerometer. Impact hammer was connected to channel 1 of the National Instrument (NI) dynamic analyser and a tri-axial accelerometer was connected to channel 2, 3 and 4 respectively. 3
A.G.A. Rahman,Z. Ismail,S. Noroozi, O.Z. Chao Figure 2 Instrumentation Set-up The experiment was carried out by fixing the impact hammer and roving the tri-axial accelerometer. The sampling rate used was 2048 samples/sec with block size of 4096. This yields frequency resolutions of 0.5 Hz and 2 seconds of time record length to capture every response signal. Five averages or impacts were taken at each measurement point for non-rotating condition using EMA and one hundred averages were collected during rotating condition using ISMA. The signals were processed by a self-developed virtual instruments application programme in DASYLab using different averaging techniques to generate FRFs and Coherence Functions. The modal extraction techniques applied to EMA could also be applied in ISMA. Figure 3 shows a three-dimensional structural model of the test rig which was drawn with coordinate points and connected by straight lines using a modal analysis software called ME Scope. The displayed point numbers show the measurement locations as in the actual base plate. This model was used to display the mode shapes of the structural plate of test rig from the acquired data. In addition to that, the software was also used to perform post-processing of the acquired data and curve-fitting for the extraction of modal frequency, modal damping and modal shape. List of instrumentation are displayed in Table 1. Figure 3 Structural Wire-mesh Model of Motor-driven Test Rig Table 1 List of Instrumentation INSTRUMENTS PCB Impact Hammer, Model 086C03 IMI Tri-axial Accelerometer, Model 604B31 NI USB Dynamic Signal Acquisition Module, Model NI-USB 9234 DASYLab v10.0 ME Scope v4.0 Table 2 shows the summary of the correlation results of first three natural modes obtained between EMA and ISMA. EMA was performed during stationary condition and modal extraction results are used as a benchmark to correlate with Impact-Synchronous Modal Analysis (ISMA) results (Table 3) which was performed during operating condition at 20 Hz. The dynamic characteristics result obtained by ISMA shows good correlation with the benchmarked EMA s result. Differences of natural frequency for the first three modes are less than 10%. The mode shapes correlation using MAC values 4
5 th International Operational Modal Analysis Conference, Guimarães 13-15 May 2013 range from 0.958 for mode 1 to 0.809 for mode 3 also show close correlation of the mode shape vectors. Table 2 Summary of Natural Frequencies and Mode Shapes Comparison between EMA and ISMA Mode Natural Frequency Natural Frequency Percentage of [Hz] by EMA [Hz] by ISMA Difference [%] MAC 1 9.92 9.56 3.63 0.958 2 15.6 15.2 2.56 0.899 3 24.9 23.4 6.02 0.809 Table 3 Dynamic Characteristics of Motor-driven Test Rig during Operating Condition at 20 Hz using ISMA Mode 1 (Knock in Vertical Direction) Natural Frequency [Hz] 9.56 Damping [Hz] 0.886 Mode 2 (Knock in Vertical Direction) Natural Frequency [Hz] 15.2 Damping [Hz] 0.678 Mode 3 (Knock in Vertical Direction) Natural Frequency [Hz] 23.4 Damping [Hz] 1.29 5
A.G.A. Rahman,Z. Ismail,S. Noroozi, O.Z. Chao 4. CASED STUDIED ISMA has been tested in real industry application; performed on a diesel fuel pump package at an offshore platform in Malaysia to investigate the high vibration problem of the package. The instrumentation and procedures used in real industry application was the same as used in laboratory condition. The only difference is the impact hammer was replaced with a larger size model of hammer (PCB Impact Hammer, Model 086D20) was used in the test. ISMA was used to determine the dynamic characteristics, namely natural frequencies, mode shapes, and damping of pump. Two identical units i.e. A and B were sat on the same skid. Pump A was shut down while the adjacent standby unit B was in operation. This will avoid any disturbance on the daily operational processes at platform. Measurements were carried out in April 2012. Vibration measurements were taken on motor, motor support, pump, pump support, skid, pipes, etc. as shown in Figure 4. Initially, measurement points were identified and marked. All the points were then linked to obtain a wire-mesh model representing the machine in ME Scope software as shown in Figures 5. Unit A Unit B Figure 4 Diesel Fuel Pump Package Figure 5 Measurement Points and Locations of Diesel Fuel Pump Package ISMA revealed that several natural frequencies were close to and coincide with operating running speed of 49.5 Hz at motor and pump sides of unit A. (Table 4). This explained the high vibration which was recorded in these components. Generally, unit A was operating in resonance situation. This was the primary cause of high vibrations at motor and pump. As a long term solution, it was recommended to perform Structural Dynamic Modification (SDM) which requires Finite Element Analysis (FEA) for verification prior to fabrication. 6
5 th International Operational Modal Analysis Conference, Guimarães 13-15 May 2013 Mode 1 Table 4 Dynamic Characteristics of Diesel Fuel Pump using ISMA (Knock in Horizontal Direction) Natural Frequency [Hz] 42.0 Damping [Hz] 0.882 Mode 2 (Knock in Horizontal Direction) Natural Frequency [Hz] 46.0 Damping [Hz] 1.340 Mode 3 (Knock in Axial Direction) Natural Frequency [Hz] 49.5 Damping [Hz] 0.445 Mode 4 (Knock in Horizontal Direction) Natural Frequency [Hz] 52.0 Damping [Hz] 0.310 7
A.G.A. Rahman,Z. Ismail,S. Noroozi, O.Z. Chao Currently, the machines are still requested to shutdown before modal analysis is performed. However, in these plants where downtime cost is very high, there are always some in operation and standby units located adjacent to the unit under analysis. The cyclic load and ambient excitation from those running units are transferred to the static unit. Hence, vibration generated by the adjacent running units will contaminate the impulse response when modal analysis is performed on the stationary unit. With ISMA applied, unaccounted forces are diminished and prominent FRF is obtained through manually operated impact hammer. Impact hammer is limited to low impact numbers due to large number of measurement points on the machines and time constraint. However, random impacts with low number of averages have successfully determined dynamic characteristics of structure under analysis. With information of the machinery s characteristics, further analyses are performed to solve the vibration problems. 5. CONCLUSIONS This study has demonstrated the effectiveness of using Impact-Synchronous Modal Analysis (ISMA) in the determination of dynamic characteristics of a system while in operating condition. The well correlated results with classical EMA during static condition show that ISMA could be performed in the presence of the unaccounted forces. It is also noticed that modal parameters were successfully determined using ISMA in industrial machinery. ACKNOWLEDGEMENTS The authors wish to acknowledge the financial support and advice given by Postgraduate Research Fund (PV086-2011A), Advanced Shock and Vibration Research (ASVR) Group of University of Malaya, Advanced Structural Integrity Vibration Research (ASIVR) of Universiti Malaysia Pahang and other project collaborators. REFERENCES [1] Zhang, L.M., R. Brincker, and P. Andersen (2001) Modal Indicators for Operational Modal Identification. In: Proceedings of the 19th International Modal Analysis Conference, Orlando, Florida, USA, 746-752. [2] Brincker, R., L.M. Zhang, and P. Anderson (2000) Modal Identification from Ambient Response using Frequency Domain Decomposition. In: Proceedings of the 18th International Modal Analysis Conference, San Antonio, Texas, USA, 625-630. [3] Schwarz, B. and M.H. Richardson (2001) Modal Parameter Estimation from Ambient Response Data. In: Proceedings of the 19th International Modal Analysis Conference, Orlando, Florida, USA, 1017-1022. [4] Schwarz, B. and M.H. Richardson (2007) Using a De-Convolution Window for Operating Modal Analysis. In: Proceedings of the 2nd International Operational Modal Analysis Conference, Orlando, Florida, USA, 1-7. [5] Whelan, M.J., et al. (2011) Operational modal analysis of a multi-span skew bridge using realtime wireless sensor networks. Journal of Vibration and Control 17(13): 1952-1963. [6] Mohanty, P. and D.J. Rixen (2004) A modified Ibrahim time domain algorithm for operational modal analysis including harmonic excitation. Journal of Sound and Vibration 275(1-2): 375-390. [7] Rahman, A.G.A., Z.C. Ong, and Z. Ismail (2011) Effectiveness of Impact-Synchronous Time Averaging in determination of dynamic characteristics of a rotor dynamic system. Measurement 44(1): 34-45. [8] Rahman, A.G.A., Z.C. Ong, and Z. Ismail (2011) Enhancement of coherence functions using time signals in Modal Analysis. Measurement 44(10): 2112-2123. 8