Final Report Summary - AGITATORCBM (Condition based maintenance of agitators using advanced condition monitoring and Internet/wireless technologies)
This project aims to develop an online system for condition-based maintenance (CBM) of agitators. Agitators are used to mix contents inside vessels which have been widely applied in various manufacturing sectors, such as chemical and petrochemical, pharmaceutical, polymer, biotechnology and food and grain processing industries. A typical agitator system consists of power input, a gearbox and an impeller. The failure of the agitator causes the failure of the processing system, which may cost millions of pounds. In current practice, the inspection / maintenance is carried out in fixed schedules which is costly. With the CBM system developed, the maintenance is carried out based on the health condition of the agitation system, rather than fixed schedules. This will not only considerably reduce the maintenance cost, but also greatly enhance the safety of the processing systems.
The CBM system developed in this research includes three modules: data collection, diagnosis and prognosis, and CBM. Within the data collection module, the embedded sensors collect the signals from the agitator and transmit the data, in a digitised form, over the Internet to a centralised diagnostic and prognostic module, for data analysis. Based on the experience developed, prognostic and diagnostic decisions are made and sent directly to the CBM control. The required maintenance is then scheduled for a suitable time, and replacement of parts or maintenance can then be completed accordingly.
The work conducted in this project include the following:
(1) Refinement of the data collection module
Prior to this Marie Curie International Incoming Fellowships (IIF) project, a pilot project for development of an agitator CBM system was carried out at Nottingham Trent University which is the host of this Marie Curie IIF project. The pilot project developed a prototype data collection module. Refinement from the prototype was carried out in this Marie Curie project, in order to overcome the shortcoming of the prototype and to enhance its functions. Major work conducted include: adjustment of the measurement system and test bench, identification and elimination of the vibration noise sources existing in the test bench, determination and relocation of the sensors to the best positions in the gearbox in order to properly extract vibration features via the test bench, and conducting a number of tests to determine the parameters of the data acquisition system and the connection mode of the A / D board.
(2) Prognosis and diagnosis of the agitator faults
(i) Novel technologies developed
In the literature, there have not been a successful methods for prognosis and diagnosis of agitator faults. It is a challenge task to accurately detect the agitators faults, and it is particularly difficult to detect the faults in their early stage and at low rotation speeds. To overcome this challenge, the following novel technologies have been developed:
- The method based on a new concept for working condition data collection, using RMS values in narrow angular rotation ranges per revolution. With this concept, two novel algorithms were established, which form the major parts of the method: a fault diagnosis algorithm based on encoder time impulses for the low rotation speed helical gear, and an algorithm for early fault feature extraction using a narrow band filter and wavelet-envelope analysis. Simulation and experimental results have shown that these new algorithms are more effective than traditional diagnosis methods.
- The approach integrating multiple intelligent and efficient filtering techniques to extract the most sensitive and representative features of the fault and agitator working condition, including evolutionary digital filtering, envelope analysis based on band-pass filtering (central frequency in low frequency peak areas of vibration signal), band-pass filtering, wavelet analysis coupling methods, and angle domain analysis. This approach greatly enhances fault diagnosis and prognosis at the early stages of fault development.
(ii) Analyses
A huge amount of data analysis work was completed using the time-domain waveform analysis, angular-domain waveform analysis, statistical parameter analysis, wavelet analysis, short-time fast Fourier transform (FFT) analysis, envelope analysis, time-domain synchronous averaging, angle domain synchronous averaging, and order tracking analysis. This work formed the basis for completing this module.
(iii) Experimental work
A large number of experimental tests have been carried out to test and verify the new technologies. The experimental tests include the early stage, growth and late stage faults and component failure. These faults / failure and deterioration tests included pitting faults of four types (25 %, 50 %, 75 % and 90 % pitting areas), broken tooth faults of four types (25 %, 50 %, 75 % and 100 % tooth length) and the coupling misalignment, amongst others. The running conditions for agitator gearbox included five different helical gear rotation speeds (28 rpm, 56 rpm, 98 rpm, 140 rpm and 210 rpm) and four different torque loads on the input shaft of the gear box (2 Nm, 4 Nm, 6 Nm and 8 Nm). The measurement conditions included samples with and without encoder and using different recorded data lengths.
(iv) Benchmarking databases
With the technologies developed and the experimental results, the databases for various fault patterns and features were developed which can be used as benchmarking for comparison with agitators working condition data for prognosis and diagnosis of agitator faults. The databases not only overcame the problem of lacking benchmarking data in the subject area, but also created a solid base for validation of the agitator CBM system developed by this research.
(3) Internet communication
The work conducted includes: implementation of mobile communication for transmitting data from the data collection module to the server, and for the user to access the data analysis results from the server, which particularly benefits the users who are away from the office or work in the field; and the platform for the three modules to communicate with each other.
(4) System integration and validation
All the modules, methods and subsystems developed were integrated to form the agitator CBM system. Due to the limited time, this one year project developed a prototype of the system. The CBM system was validated in the laboratory at Nottingham Trent University to ensure that it worked properly, and it was then demonstrated at Chmineer Ltd. With the feedback given from the demonstration, the system was further refined, and final demonstration was given to the key staff members of Chemineer Ltd, proving that the system has achieved designed functions.
In addition to the above, the following are also resulted from this project:
- Four research publications, including one paper published in the 'International of Design Engineering' (Inderscience), two papers submitted to 'IMechE proceedings Part I - Journal of Systems and Control Engineering' (Sage) and 'Renewable Energy' (Elevier) respectively, and one paper presented at the Fifth International Conference on Engineering Failure Analysis, 1 - 4 July 2012, Hague, the Netherlands.
- A patent application has been submitted.
- The project contributed to the training of two Doctor of Philosophy (PhD) students and on visiting scholar under the supervision of the Marie Curie fellow.
- The project fault prognosis and diagnosis technology and Internet communication methods have been utilised in the teaching module of CAE and internet technologies for the Master course of Advanced Product Design course at Nottingham Trent University.
The CBM system developed in this research includes three modules: data collection, diagnosis and prognosis, and CBM. Within the data collection module, the embedded sensors collect the signals from the agitator and transmit the data, in a digitised form, over the Internet to a centralised diagnostic and prognostic module, for data analysis. Based on the experience developed, prognostic and diagnostic decisions are made and sent directly to the CBM control. The required maintenance is then scheduled for a suitable time, and replacement of parts or maintenance can then be completed accordingly.
The work conducted in this project include the following:
(1) Refinement of the data collection module
Prior to this Marie Curie International Incoming Fellowships (IIF) project, a pilot project for development of an agitator CBM system was carried out at Nottingham Trent University which is the host of this Marie Curie IIF project. The pilot project developed a prototype data collection module. Refinement from the prototype was carried out in this Marie Curie project, in order to overcome the shortcoming of the prototype and to enhance its functions. Major work conducted include: adjustment of the measurement system and test bench, identification and elimination of the vibration noise sources existing in the test bench, determination and relocation of the sensors to the best positions in the gearbox in order to properly extract vibration features via the test bench, and conducting a number of tests to determine the parameters of the data acquisition system and the connection mode of the A / D board.
(2) Prognosis and diagnosis of the agitator faults
(i) Novel technologies developed
In the literature, there have not been a successful methods for prognosis and diagnosis of agitator faults. It is a challenge task to accurately detect the agitators faults, and it is particularly difficult to detect the faults in their early stage and at low rotation speeds. To overcome this challenge, the following novel technologies have been developed:
- The method based on a new concept for working condition data collection, using RMS values in narrow angular rotation ranges per revolution. With this concept, two novel algorithms were established, which form the major parts of the method: a fault diagnosis algorithm based on encoder time impulses for the low rotation speed helical gear, and an algorithm for early fault feature extraction using a narrow band filter and wavelet-envelope analysis. Simulation and experimental results have shown that these new algorithms are more effective than traditional diagnosis methods.
- The approach integrating multiple intelligent and efficient filtering techniques to extract the most sensitive and representative features of the fault and agitator working condition, including evolutionary digital filtering, envelope analysis based on band-pass filtering (central frequency in low frequency peak areas of vibration signal), band-pass filtering, wavelet analysis coupling methods, and angle domain analysis. This approach greatly enhances fault diagnosis and prognosis at the early stages of fault development.
(ii) Analyses
A huge amount of data analysis work was completed using the time-domain waveform analysis, angular-domain waveform analysis, statistical parameter analysis, wavelet analysis, short-time fast Fourier transform (FFT) analysis, envelope analysis, time-domain synchronous averaging, angle domain synchronous averaging, and order tracking analysis. This work formed the basis for completing this module.
(iii) Experimental work
A large number of experimental tests have been carried out to test and verify the new technologies. The experimental tests include the early stage, growth and late stage faults and component failure. These faults / failure and deterioration tests included pitting faults of four types (25 %, 50 %, 75 % and 90 % pitting areas), broken tooth faults of four types (25 %, 50 %, 75 % and 100 % tooth length) and the coupling misalignment, amongst others. The running conditions for agitator gearbox included five different helical gear rotation speeds (28 rpm, 56 rpm, 98 rpm, 140 rpm and 210 rpm) and four different torque loads on the input shaft of the gear box (2 Nm, 4 Nm, 6 Nm and 8 Nm). The measurement conditions included samples with and without encoder and using different recorded data lengths.
(iv) Benchmarking databases
With the technologies developed and the experimental results, the databases for various fault patterns and features were developed which can be used as benchmarking for comparison with agitators working condition data for prognosis and diagnosis of agitator faults. The databases not only overcame the problem of lacking benchmarking data in the subject area, but also created a solid base for validation of the agitator CBM system developed by this research.
(3) Internet communication
The work conducted includes: implementation of mobile communication for transmitting data from the data collection module to the server, and for the user to access the data analysis results from the server, which particularly benefits the users who are away from the office or work in the field; and the platform for the three modules to communicate with each other.
(4) System integration and validation
All the modules, methods and subsystems developed were integrated to form the agitator CBM system. Due to the limited time, this one year project developed a prototype of the system. The CBM system was validated in the laboratory at Nottingham Trent University to ensure that it worked properly, and it was then demonstrated at Chmineer Ltd. With the feedback given from the demonstration, the system was further refined, and final demonstration was given to the key staff members of Chemineer Ltd, proving that the system has achieved designed functions.
In addition to the above, the following are also resulted from this project:
- Four research publications, including one paper published in the 'International of Design Engineering' (Inderscience), two papers submitted to 'IMechE proceedings Part I - Journal of Systems and Control Engineering' (Sage) and 'Renewable Energy' (Elevier) respectively, and one paper presented at the Fifth International Conference on Engineering Failure Analysis, 1 - 4 July 2012, Hague, the Netherlands.
- A patent application has been submitted.
- The project contributed to the training of two Doctor of Philosophy (PhD) students and on visiting scholar under the supervision of the Marie Curie fellow.
- The project fault prognosis and diagnosis technology and Internet communication methods have been utilised in the teaching module of CAE and internet technologies for the Master course of Advanced Product Design course at Nottingham Trent University.