Conference Opening Remarks & Welcome - Amy Wynn, MFPT ED & Chris Nemarich, MFPT Board Chairman
Reimagining The Model for Reliability and Readiness Data Analytics - Keynote Speaker Larry Burrill President of Readiness and Sustainment Solutions Business Unit and Co-Founder McKean Defense Group
Hydraulic Institute, Turbomachinery & Pump Symposium, & Vibration Institute Introductions
Larry Burrill is President of McKean Technical Services of McKean Defense Group Inc., an engineering and technology solutions firm that provides life cycle engineering, sustainment and modernization support, and information technology/enterprise transformation services to the U.S... Read More →
Peter Gaydon is the Director of Technical Affairs at the Hydraulic Institute. Mr. Gaydon held design, development, and test engineering positions with major pump manufacturers prior to joining the Hydraulic Institute. With the Hydraulic Institute, Mr. Gaydon has technical responsibility... Read More →
Belt-driven mechatronic systems are becoming more popular for a range of applications. A modified robotic manipulator was adapted to allow different faults to be incorporated into the mechanism, with additional sensors to characterize the compromised kinematics. Different data-driven models were used to detect anomalies in motor power consumption and end-effector motion; and a physics-based, lumped-parameter dynamic model was used to identify different faults. Comparative assessment metrics were used to compare the performance of different fault models from sets of laboratory test data.
The present paper describes universal vibration sensor mount devices for attaching sensors to the machinery. Some types of machinery vibration sensors need to be mounted in a particular orientation relative to the machine being monitored. One example is two-axis and three-axis MEMS accelerometers used to detect machine-critical acceleration vector components. Another example is sensors with side connectors or cables whose orientation is restricted by space limitations. In many instances, vibration machinery sensors constructed as could be mounted using a central bolt that going throw the sensor body threads into a bore on the machine body. Such sensors are usually more expensive that similar sensors with solid body. In some cases, proper sensor orientation is achieved by simply gluing the sensor to the machine body. This makes it difficult to remove the sensor if it needs to be replaced. There are also proprietary sensor mounts designed for specific sensors, but these lack versatility. We designed the universal sensor mount that can be used to mount a wide variety of machinery sensors at precise orientations. The idea here is that sensor module mounting member is locked in a selected rotational position using the adjustable locking member. The sensor module is attached to the universal sensor mount by threading the sensor module onto the sensor module mounting member until tight. A determination is made whether the sensor module is substantially aligned with one or more reference axes of the machine. If not, the sensor module is detached from the universal sensor mount. The sensor module mounting member is unlocked and rotated to another selected rotational position that will substantially align the sensor module with the one or more machine reference axes. The sensor module mounting member is then relocked in another selected rotational position and sensor module is reattached to the universal sensor mount.
Variable speed assets such as VFDs and low-cost DC Drives are becoming increasingly common across a wide range of industries and applications, with significantly greater process control and enormous cost savings two driving benefits. The challenges these assets present to the vibration analyst, however, are equally significant. This presentation will cover the data issues before moving on to discussing available tools and solutions, including the benefits and limitations of using advanced techniques such as time synchronous averaging and precise order tracking.
This project, sponsored by the Office of Naval Research, integrates oil and vibration analysis to develop a more complete diagnostic and prognostic model for monitoring bearing degradation and performance while inside a vehicle's differential. The results of this project could be used to optimize repair schedules and minimize the risk of catastrophic failure of machine components for military personnel during a mission through condition-based maintenance. Each test bearing is first introduced with an outer race fault. The fault's degradation is then accelerated by running the bearing under overloaded conditions in the boundary lubrication regime. Accumulated damage to the bearing is characterized in a clean, low-noise environment before being implemented into the differential for in-situ testing. Dynamic characterization of the designed test machines were performed by long runs spanning multiple hours under rated conditions to determine any wear-in effects and ramp-up tests to distinguish order-based bearing frequencies and structural resonances. Modal analyses were performed on the static system to provide additional evidence of structural resonances within the machines. This paper will discuss the design challenges and solutions for creating a test bed to monitor bearings in accelerated-degradation conditions and in noisy environments, such as the differential. Results of run-to-failure data and analysis will also be presented.
This paper describes work related to improving the electrical performance of an accelerometer-based sensor, RotoSense™, used for monitoring rolling stock: the locomotives and cars used in trains. At the 2018 MFPT conference, a paper, “Improved RotoSense™ for Rolling Stock: Locomotives and Cars,†focused on physical improvements to the sensor, although there were improvements in signal performance. This paper describes subsequent improvements to that sensor, with focus on signal quality and battery life. The original version of the sensor described in this paper is the first and, still, only known to survive, intact, three days of testing at the National Test Track Center in Pueblo, Colorado, including a 10-hour, non-stop, 400-mile test run. The rationale, the methods, and the results of those electrical improvements are presented in the paper.
An introduction to the Condition Monitoring professional qualification scheme, organised by the British Institute of NDT._ To recognise personnel demonstrating capability in one or more Condition Monitoring disciplines at an ISO recognised grade, to provide assurance to the empoyer/ owner/ operator of complex and expenive capital assets, that the person is proficient to perform the work
Steve Greenfield has served as regional technical support manager for Eaton’s Aerospace Group since 2007 and has more than 30 years of experience in advanced sensor systems for gas turbine and rotorcraft health monitoring.Greenfield works with airline customers in Europe and Africa... Read More →
Peter Gaydon is the Director of Technical Affairs at the Hydraulic Institute. Mr. Gaydon held design, development, and test engineering positions with major pump manufacturers prior to joining the Hydraulic Institute. With the Hydraulic Institute, Mr. Gaydon has technical responsibility... Read More →
In today's economy, manufacturers strive for the highest level of equipment productivity possible, while achieving the lowest operational costs. Fortunately, open and comprehensive industrial innovation platforms (IIoT) allow us to incorporate decades of reliability and condition monitoring domain expertise along with equipment connectivity, enterprise systems connectivity, people connectivity, expert systems, machine learning, augmented reality and more.
Predictive Maintenance promises the larger returns on investment (ROI) opportunities of a digital transformation. The gap between where manufacturers are today and where new industrial innovation platforms can take us stems from the historic inability to connect systems and people together with all data sources in a single location; thus, giving way to higher visibility, machine learning, and ultimately improved insight into the best next steps and strategy that improve equipment productivity.
However, the laws of physics bounding the ability of our production equipment are not changed by IIoT applications platforms. Equipment and processes continue to fail as physical, chemical, and environmental parameters wear and change. Our equipment and processes fail for known reasons. We can map the failure modes of our equipment and processes to machinery, operating process and environmental parameters. We then detect changes in these parameters that lead to equipment and process failure. We then predict a failure, and work to prevent the failure before it occurs. We are on our way to jump start our digital transformation for predictive maintenance.
This presentation introduces a typical production environment and equipment to illustrate failure modes of equipment and process. It illustrates parameters our IIoT platform can monitor, and the decades of domain expertise can build on for immediate ROI while setting the stage for machine learning and augmented reality to join in as our digital transformation and the tools we use mature over time. Example case studies are offered to illustrate the failure mode approach.
The paper presents a detail comparison between the traditional piezoelectric and MEMs based vibration sensors. The future prognoses of what technology and where will be used in the future is discussed.
Maintenance costs constitute a major part of operating costs in any industry. This has motivated industries to adopt best machine condition monitoring techniques so that costs can be reduced and productivity can be increased. In many rotating machinery system, shaft cracks are frequent and serious malfunctions that may lead to catastrophic failure and financial loss. Fault detection techniques are crucial for safe and reliable machinery and the demand for timely and accurate detection is increasing day by day. Shaft crack detection is an especially critical task due to the complex operating conditions that shafts are subjected to. One of the shaft crack detection approaches is vibration and acoustics monitoring, which can be implemented in an automated fashion. This paper presents the application of recurrence plots (RPs) and recurrence quantification analysis (RQA) in the diagnostics of various rotating shaft cracks. The RP is a two-dimensional visualization technique to investigate high-dimensional dynamical systems. It identifies the times when the state space trajectory of the dynamical system visits roughly the same area in the state space. On the other hand, RQA offers a more objective and quantitative method for the investigation of dynamical systems, which will represent the extracted features that characterize the system response. In addition, RPs and RQA are a modern tool for nonlinear data analysis, which enables us to investigate the various responses of the system (i.e., periodic, quasi periodic and chaotic) and provides valuable information about the dynamics of the system. The computed RQA features are ranked and the optimal set is selected using mutual information. Finally, an artificial neural network is used as a classifier to distinguish between the different shaft conditions.
A laboratory scale rotor test bed was used to investigate shaft crack detection techniques under controlled conditions. The study was implemented on a shaft that was seeded with two damage conditions produced by a crack propagator over 24-hour and 48-hour time periods. The horizontal and vertical displacements were measured for each shaft condition using proximity probes.
The study demonstrates that the RQA provides rich information about the status of the health of the shaft. Furthermore, results show an outstanding performance of the RQA in shaft crack detection with minimal knowledge about the dynamic response of the system.
Analytics Isn't easy let's accept this upfront. While we're at it, let's demystify unhelpful buzzwords, realize It takes more than technology, and requires organizational change: People, Process & Culture. So why do companies continue to mature their analytics capacities knowing it's complex? BECAUSE DATA DRIVEN INSIGHTS ARE TRANSFORMING BUSINESSES! This session will identify ways to explain the complexities of analytics approaches and data science simply, rather than oversimplify the complex. It will help analytics practitioners and business stakeholders speak in a way that inspires collaboration and sustainable best practices. Join Chris MacDonald to learn: - What it means to ask the right questions of data - What everyone should know about the advanced analytics process regardless of automated technologies - How to get value from advanced analytics while building a data driven culture.
The attachment of accelerometers is typically accomplished by magnetic mount, wax,_x000D_ glue, screw-mount, or a hand-held technique. This paper present the results of careful_x000D_ laboratory study of the resulting accuracy of the vibration values across a broad frequency_x000D_ spectrum, based on both impact testing as well as shaker testing in a laboratory environment._x000D_ The magnetic mount and had-held techniques are much more convenient than attachment_x000D_ with screws or glue, but have been shown in various studies to result in tapping on_x000D_ uneven surfaces, and in general to lose frequency response above 500 Hz, at frequencies_x000D_ still very much of interest in machinery diagnostics. This is detailed in ISO 5348. _x000D_ However, the authors demonstrate that the combined use of wax with either the _x000D_ magnetic mounting technique or the hand-held method can provide surprisingly accurate _x000D_ readings for most practical applications.
The diagnosis of rotating and reciprocating machinery has been an area of active research for several decades, for both commercial and military equipment. Most researchers are knowledgeable concerning instrumentation and signal analysis algorithms, but not as experienced with the 'physiology' of various machinery, the details and dynamical behavior of which can be quite complex. Therefore, the tendency has been to use either simple rule-based look-up tables, or in the other extreme ti apply adaptive learning, neural nets, or other data-driven statistics-based approaches that look for anomalies in a manner that attempts to be machine-agnostic. These approaches can be useful in some applications. However, the authors suggest that a superior approach is the use of physics-based set of algorithms that are based on the mechanical and fluid-dynamic (or electrical for motors/generators) details of the machinery in question. Some examples are provided of such algorithms that have application over broad classes of machinery. It is demonstrated how such as approach can be blended with ISO standards based constraints and guidelines to provide software that can mimic much of what an experienced machinery engineer or technician would conclude from a given data set. Examples of machinery successfully diagnosed by the approach will be given.
Industrial sensor data is Big Data and largely of time-series nature for asset intensive industries such as oil & gas, chemicals, paper & pulp, pharma, mining etc., including utility companies in electricity, gas and water production/generation and transmission/distribution. Attend this session as we review the challenges in sensor and IIoT (Industrial Internet of Things) data and meta-data management. This involves working with large volumes of high velocity data from automation and control systems with hundreds of thousands to millions of sensors; each sensor collecting measurements every minute, some every second or even few microseconds. The data may also be from CMS (condition monitoring systems) and several other disparate line-of-business data sources such as planning, work management, quality, weather, web pages etc. IIoT and edge devices now bring in even more data with their own new challenges during data collection, and the need to merge this data with legacy sensor data sets.
This presentation will focus on the lessons learnt during our 35+ years of working with data and information management in manufacturing operations (https://en.wikipedia.org/wiki/Manufacturing_operations_management). And, we will cover how data analytics (with machine learning when appropriate) and visualization are used in customer use cases in energy reduction, predictive maintenance and reliability, process yield improvement, product quality, and others.
A data infrastructure approach (scalable backbone) with a faceted and application-adapted digital model allows flexible and extensible methods for data collection, analysis, and visualization that are fit-for-purpose and finally, actionable insights. Also see https://en.wikipedia.org/wiki/OSIsoft
Attend a maintenance or reliability conference today and you can't fail to miss the many vendors offering wireless sensors for condition monitoring. Wireless sensor networks offer a way to expand the practice of CBM by making more condition data from more machines available for analysis. Triaxial Solid state semiconductor MEMS accelerometers enable low power wireless vibration sensors by easing the design of the signal processing, digital interface and power management with techniques not easily implemented with traditional technology. This presentation will describe a very low power vibration sensor based on a triaxial MEMS accelerometer.
Managing Director, The Machine Instrumentation Group
Ed Spence is the Founder and Managing Director of The Machine Instrumentation Group, a consultant representing a network of contract engineering service providers with expertise in Condition Monitoring, sensor design, signal processing and data engineering. Previously, Ed was the... Read More →
At the MFPT Conference two years ago, the machinery community got its first exposure to the concept of motion amplified video. Since then, at least three organizations have been developing commercial systems that employ one of two approaches, either the so-called Lagrangian Approach, or the so-called Eulerian Approach. With either approach, the concept is to capture high speed video which can then be displayed in slow motion, but more importantly which permits displacements of the order of microns to be amplified so that the physical vibration motion can be easily seen in the video of the operation of the actual components. The authors have been using both approaches in the field and lab, but in their own software developments have concentrated on the Eulerian, because of its ability to better resolve small displacements, which are especially important at the higher frequencies typical of vibrations in certain machinery such as high speed compressors, steam turbines, and gas turbines. Some of the pros and cons versus Operating Deflection Shape testing using accelerometers will be described, and use on specific machinery examples will be provided.
Industry experts suggest that predictive maintenance can reduce breakdowns by nearly 70%. The cost of unplanned downtime in industrial systems and processes can be overpowering. By combining IoT sensor data with advanced ML technologies, enterprises can enable real-time predictive maintenance that can confidently identify and proactively alert operators before mission critical issues arise bringing industrial processes to a halt. Aldo is a global thought leader in delivering Predictive Maintenance and Process Optimization technology solutions for industry's such as manufacturing, transportation, and utilities.
We'll discuss how to take control of the industrial edge and IoT with a low barrier to entry in deploying ML and predictive analytics within the world of maintenance and reliability.
This presentation will also discuss how to shift an industrial systems maintenance and operations paradigm from reactive break/fix to proactive with machine learning and predictive analytics at the physical edge.
In recent decades, many industries, especially in aviation, have been investing in developing efficient maintenance approaches such as condition-based maintenance (CBM). In passenger train cars of the Israel Railways (ISR), some of the bearings of the electrical motors of the air-conditioner suffer repeated failures. Therefore, assessment and prognostics of the remaining useful life of its rolling-element bearing is essential.
One of the most common methods for health monitoring of rotating machinery is based on vibration signals. In rotating machinery, actions of the moving parts take place at specific angular positions rather than at specific times. For this reason, accurate estimation of the instantaneous Angular Speed (IAS) has an important role in facilitating reliable diagnostics of the vibrations providing enriched information about the health status of the machine. In practice, however, direct measurement of the angular speed is sometimes impossible, uneconomical or inaccurate.
This work focuses on the estimation of the IAS of a fan electrical motor for diagnostic of its supporting bearing. In this study, few methods for IAS estimation directly from the vibration signal are compared. Finally, a complete vibration analysis scheme for diagnostics of the axial loaded bearing of the fan is proposed and validated experimentally.
This paper describes the relationship between a bathtub curve, failure distributions, and commonly used, but often misunderstood and misapplied, metrics such as the following: mean-time-before failure (MTBF), mean-time-between failure (MTBF), mean-time-to failure (MTTF), and other related terms. A bathtub curve for an object is related to a failure-distribution curve for that object, and when combined, a continuous curve is created. That curve has commonly named regions: infant-mortality, constant-failure, and wear-out, which comprises two subregions that I call degraded-operation and functional-failure. Associated with those curves are numerous metrics that are sometimes misunderstood and misused such as the following: failure rate, prognostic trigger point, prognostic distance (PD), MTBF (two meanings), MTTF, failures-in-time (FIT), and life time (reliability).
There is a growing trend of using unmanned aerial vehicles (UAVs, or drones) for inspection of industrial facilities and infrastructure. The types of inspections currently done entail remote sensing using cameras: qualitative assessment and photogrammetry using standard cameras, as well as temperature monitoring using thermal imaging cameras. The benefits of using drones are to reduce risk to inspection personnel near operating equipment, reduced cost, and improved auditability of archived data. Machine condition monitoring can benefits from this enabling technology provided that vibration monitoring and lubricant analysis can also be done remotely. Laser vibrometers can be used for remote monitoring of equipment that does not have permanently mounted, dedicated vibration monitoring sensors; but this sensing method is capital intensive, requires a laser head and controller with a typical mass of over 5 kg, and must be pointed from an appropriate direction, which may not be accessible. Lubricant analysis requires access to a lubricant sampling port and a means for collecting and analyzing the sample. A system is described for deploying a drone with a manipulator that engages the equipment of interest for vibration data collection. The same system can use a similar payload for drop-tube vacuum sampling of lubricant by inserting a tube through a fill port or dip stick port and withdrawing an oil sample from the sump cavity. The key technical challenges are collision-free navigation and hovering, followed by control of dynamic deployment of the payload that connects to the machinery. A proof-of-concept system is described with experimental results.
This paper describes the relationship between a bathtub curve, failure distributions, and commonly used, but often misunderstood and misapplied, metrics such as the following: mean-time-before failure (MTBF), mean-time-between failure (MTBF), mean-time-to failure (MTTF), and other related terms. A bathtub curve for an object is related to a failure-distribution curve for that object, and when combined, a continuous curve is created. That curve has commonly named regions: infant-mortality, constant-failure, and wear-out, which comprises two subregions that I call degraded-operation and functional-failure. Associated with those curves are numerous metrics that are sometimes misunderstood and misused such as the following: failure rate, prognostic trigger point, prognostic distance (PD), MTBF (two meanings), MTTF, failures-in-time (FIT), and life time (reliability).
The diagnosis of rotating and reciprocating machinery has been an area of active research_x000D_ for several decades, for both commercial and military equipment. Most researchers are_x000D_ knowledgeable concerning instrumentation and signal analysis algorithms, but not as_x000D_ experienced with the 'physiology' of various machinery, the details and dynamical behavior of _x000D_ which can be quite complex. Therefore, the tendency has been to use either simple rule-based_x000D_ look-up tables, or in the other extreme ti apply adaptive learning, neural nets, or other_x000D_ data-driven statistics-based approaches that look for anomalies in a manner that attempts_x000D_ to be machine-agnostic. These approaches can be useful in some applications. However, the_x000D_ authors suggest that a superior approach is the use of physics-based set of algorithms that_x000D_ are based on the mechanical and fluid-dynamic (or electrical for motors/generators) details_x000D_ of the machinery in question. Some examples are provided of such algorithms that have_x000D_ application over broad classes of machinery. It is demonstrated how such as approach can_x000D_ be blended with ISO standards based constraints and guidelines to provide software_x000D_ that can mimic much of what an experienced machinery engineer or technician would_x000D_ conclude from a given data set. Examples of machinery successfully diagnosed by the_x000D_ approach will be given.
An introduction to Oil Debris analysis as a diagnostic tool in the health monitoring of critical plant and machinery. From simple devices to systems compatible with on board health and usage monitoring systems, the application of traditional early failure detection by magnetic plugs is explained along with the evolution to systems offering prognostic potential.
The presentation is aimed at those new or unfamiliar with the benefits of Oil Debris analysis and those who may not include it as their primary technique in preparation for a new career move or to gain qualifications in Condition Monitoring, as well as those exposed to the technique wanting to bring their knowledge up to date.
Steve Greenfield has served as regional technical support manager for Eaton’s Aerospace Group since 2007 and has more than 30 years of experience in advanced sensor systems for gas turbine and rotorcraft health monitoring.Greenfield works with airline customers in Europe and Africa... Read More →
During testing on a small gas turbine, frequencies related power turbine shaft and a bearing were observed. The frequencies of the fault were not associated with any known bearing on the power turbine shaft. This paper is an investigation of why the observed bearing fault frequency of 8% lower than anticipated. It will be shown that because the faulted bearing was a worn thrust bearing, the contact angle of the roller element had changed. This is an infrequently observed phenomenon which can lead to a missed fault detection on a critical component. A mitigation strategies for this type of failure is discussed.
During operation, it was observed that a specific mechanical system experienced undesirable vibration and it became necessary to understand and mitigate this phenomenon. This document investigates the tools, methodology, and results of the dynamic characterization of the system. The characterization makes use of the experimental modal analysis (EMA) methods of single input multiple output (SIMO) and single input single output (SISO). The validity of the theory of reciprocity is confirmed to minimize measurement error, cost, and time of repeat testing. Finite element analysis (FEA) is utilized in choosing transducer and modal hammer impact locations to adequately characterize the system. Single degree of freedom (SDOF) and multiple degree of freedom (MDOF) curve fitting is utilized to fully characterize the system’s mode shapes and natural frequencies. The EMA characterization results are used to modify and validate the FEA model so that FEA can be used to model potential structural modifications to the system to mitigate the undesirable vibration. Structural modifications are chosen, implemented, and their effectiveness is quantified using EMA. A qualitative evaluation of the methodology of FEA validation by EMA and tuning of the model to match the experimental results is discussed.
Often referred to as the life blood of equipment, Lubrication and Oil Analysis can provide a great deal of information on the condition of equipment, as well as serve as a leading indicator of the future condition of the equipment. Oil analysis can detect lubrication related defects in rotating machine including abrasive wear, oil contamination, viscosity loss, additive content loss, water ingress and more. Historically, oil analysis is performed by periodic sampling with analysis done away from the machine in a laboratory environment. With the advent of new sensing technology, online oil monitoring is now possible and provides real-time condition indicators of oil and machine health. This presentation begins with a introduction to oil analysis and the machinery defects it can detect. It is followed by explanation of route, and online techniques for oil analysis for condition monitoring. An example using predictive analytics is provided, along with a short survey of oil analysis sensors.
A shaft/rotor crack is a common defect in rotating machinery and detection of such crack is a very serious matter. A crack could lead to a catastrophic failure of the rotor if not detected in a timely manner. Vibrations analysis is widely used to monitor the health of the rotating machinery. However, vibration signature of a cracked shaft is not clear, although a considerable literature is available on the vibration analysis of rotor cracks. The problem is further complicated in presence of misalignment and unbalance loadings. This paper examines the effects of misalignment and unbalance on vibration signature of shaft cracks with the aim of developing unique condition indicators (CI) for crack detections. It is also important to gain an understanding of difference between the signature of breathing and open rotor cracks. An experimental investigation was carried out by the authors to determine a unique signature of shaft seeded with different levels of crack. The data is presented for shafts supported on rolling as well as fluid film bearings. Time-Frequency techniques are applied to study rotors operating under non-stationary conditions.
This paper presents a case study in diagnosing an excessive pipe vibration due to a beating phenomenon. The outdoor process pipes in a sewage plant were found to vibrate viciously and result in loud hamming noise affecting the surrounding community. The process pipes were connected to two identical blower units, each driven by a motor via belt and pulley system. Besides the loud noise, the excessive pipes vibration also posed a concern to the plant personnel that the possibility of premature machine failures may occur if the problem persists any longer. A comprehensive vibration investigation was conducted to map-out the vibration of the entire machine train that includes pipes, blowers, motors, skid, plinth and floor slab of the blower house. Vibration investigation found that pipes vibration was most severe when the two units of blowers were operating simultaneously. It was found that the root cause of the excessive pipe vibration was caused by beating phenomenon of which two adjacent machines operating under a slightly different speeds, of which in this case, the two blowers were operating at 41.88Hz and 41.72Hz respectively. Beating is a phenomenon of constructive and destructive interference of two identical waveforms with slightly different frequency. The remedy measure undertaken was thus to fine tune the operating speed of the two blowers. It was found that pipes vibration had subsided considerably when the two blowers speeds were adjusted 7.5 Hz apart, or fine-tuned to operate at the speed of 42.5Hz and 50Hz respectively. As a result, the loud humming noise emitted from the pipes was also noticed to have disappear altogether.
Wear debris sensing technology has in the past been limited to using particle counts to detect faults. Current advancements in particle sizing, particle sensing range, and algorithm improvements have allowed more advanced approaches to detect faults more reliably. Further, these advancements now allow wear debris sensors to be used as a feedback loop for life extension actions. This presentation will use real industrial gearbox data to show how changes in operation and other actions impact gearbox failure rate.
Mark Redding is President and Co-Founder of Poseidon Systems, LLC. He has spent most of his career developing advanced PHM tools and technologies to monitor the reliability of equipment for aircraft, heavy machinery, processing equipment, marine, mining, and power plants. After selling... Read More →
Dr. Suri Ganeriwala is founder and president of Spectra Quest, Inc. He has over thirty-five years of industrial and academic experience in machinery vibration diagnostics and control, signal processing, and viscoelastic materials characterization. Suri has worked for Philip Morris... Read More →
A cycloid drive for gearboxes allows for high reduction ratio and zero or very low backlash. The cycloid gear design is based on compression, whereas most gear interactions are based on shear. Further, the contract of a cycloid gearbox is typically rolling vs. sliding, which is seen in traditional gearboxes. These features of a cycloid gearbox allow for high shock load capacity, high torsional stiffness, and quiet operation.
This paper details the modeling required for correct configuration to perform analysis on the cycloid gearbox and then is demonstrated on a 51:1 ratio, run to failure test. This paper documents the sensitivity of standard condition indicators for gear/bearing during the run to failure test.
Reliability Centered Lubrication (RCL) is an approach to lubrication focused on designing a program for lubrication and oil analysis considering operational conditions, best practices, and modifications required to both machine and facility. Implementation of an RCL program is carried out utilizing an RCL Design Process, Implementation, and Sustainability. A current state analysis is an excellent starting point, comparing one's own facility to best practices. Understanding equipment needs including engineering calculations and operating conditions for lubrications amounts and frequencies. Putting a lubrication program in place includes implementation of sustainable lubrication routes that add value, coupled with lubrication storage best practices. Finally, sustainment of the lubrication program is founded with lubrication metrics and KPIs that guide continuous improvement. This presentation provides an introduction to Reliability Centered Lubrication including a case study and tips for getting started.
The mechanical industry is focusing on modern maintenance techniques to minimize the downtime so that the production can be increased. The fourth industrial revolution also known as Industry 4.0 is aiming to provide fully autonomous systems including condition monitoring fault diagnostics techniques. In order to implement the Industry 4.0, the maintenance department needs automatic condition monitoring and fault diagnostics techniques. In this study, octave analysis has been proposed for early detection of bearing faults. In order to show the effectiveness of the proposed methodology, detailed experimentations have been performed on a Machinery Fault Simulator (MFS). The octave analysis has been implemented on the vibration data acquired through the accelerometers placed on the inboard and outboard bearing housing. The vibration data of a healthy bearing has been compared with a faulty bearing at four different rotational speeds under two operating conditions: no load applied and a constant load applied near outboard bearing housing. The standard deviation has also been calculated for healthy and faulty bearings. The octave analysis has high peaks for all rotational speeds and both operating conditions. The values of standard deviation of faulty bearings were found to be high as compared to a healthy bearing. This research article proposes a simple approach that does not require signal processing. The results proved that octave analysis along with standard deviation can be used for early detection of rolling bearing faults. Keywords: Condition Monitoring, Fault Detection, Industry 4.0, Octave Analysis, Rolling Bearings, Vibrations.
We describe advances in comprehensive forward deployed fluid analysis based on fluid condition assessment with FluidScan® and contaminant and wear debris analysis with the LaserNet Fines® optical particle analyzer. The combined technologies provide comprehensive analysis of fluid condition and machinery health, and are implementable either as a forward-deployed portable instrument or a continuous, autonomous online monitor. The combined technologies provide comprehensive immediate on-the spot actionable information at the operational level about the current condition of platforms and machinery, readiness for deployment or usage, and advanced warning about impending failures or need for specific maintenance. A forward deployed portable instrument is of value for use in monitoring fleets of vehicles or platforms with a single instrument through bottle sampling, while an online monitor is of value in providing continual assessment of the condition of critical, high-value equipment where serious faults can develop between bottle sample intervals, and support reduced manpower or workload environments.
Previously we have described the combination of LaserNet Fines® and FluidScan® into a single portable unit capable of analyzing lubricating and hydraulic oils1. Our current efforts involve extending the capability of the portable instrument to fuel analysis. We have developed fuel analysis algorithms for identifying if unknown fluid is a fuel, distinguishing among multiple types of aviation and diesel fuels and gasoline, and calculating multiple physical properties depending on fuel type. In the current effort, we are incorporating this capability into the portable analyzer utilizing an extended range mid-IR spectrometer that can analyze lube oil, hydraulic oil and fuel in a single instrument. Advances in the online monitor include extending the capabilities of the online LaserNet Fines® particle monitor to include gearbox oils, as well as, hydraulic and engine lube oil and to combine it with an online FluidScan® monitor. We will describe results with both the portable and online units for analysis of engines and transmission.
1 Comprehensive, Field-Deployable Portable Fluid Analyzer , J. E. Tucker*, J. Reintjes*, T. Sebokâ and P. F. Henning , Presented at 61st Meeting of the Society Machinery Failure Prevention Technology April 2007
Planet bearing fault identification is an attractive but challenging task in numerous engineering applications, such as wind turbine and helicopter transmission systems. However, traditional fault characteristic frequency identification and impulsive feature extraction based diagnosis strategies are not sufficient to resolve the problem of planet bearing fault detection, due to complex physical configurations and modulation characteristics in planetary gearboxes. In this paper, a novel discriminative dictionary learning based sparse representation classification (SRC) framework is proposed for intelligent planet bearing fault identification. Within our approach, the optimization objective for discriminative dictionary learning introduces a label consistent constraint called discriminative sparse code error™ and incorporates it with the reconstruction error and classification error to bridge the gap between the classical dictionary learning and classifier training. Therefore, not only the reconstructive and discriminative dictionary for signal sparse representation but also an optimal universal multiclass classifier for classification tasks could be simultaneously learnt in the proposed framework. The optimization formulation could be efficiently solved using the well-known K-SVD dictionary learning algorithm. The effectiveness of the proposed framework has been validated using experimental planet bearing vibration signals. Comparative results demonstrate that our framework outperforms the state-of-the-art K-SVD based SRC method in terms of classification accuracy for intelligent planet bearing fault identification.
One of the limitations of conventional modal testing using a roving impact hammer is that the reference sensor (usually an accelerometer) must remain fixed throughout the test. Since the accelerometer must be connected by a wire to the data acquisition system. a very long wire may be required when testing a large structure. Furthermore, better quality signals are possible if each impact force is applied closer to the response accelerometer. Because it does not require a fixed reference sensor throughout the test, a Rapid Impact Test is faster and easier to use on any size structure.
In this new method, either the impact hammer or the accelerometer can be moved to a different DOF between acquisitions of data. One sensor can be “hopped over” the other in slinky fashion, or both can be moved, provided that a chain of FRFs is calculated from the acquired data. Each FRF has two DOFs associated with it. An FRF chain is formed when each FRF has the same DOF as another FRF.
An FRF chain is a set of multi-reference FRFs. The multi-reference FRFs can be curve fit using single-reference methods, but the resulting modal residues must be further processed to obtain mode shapes. The residue post-processing is based on the relationship between modal residues and mode shapes. Examples using an impact hammer, uni-axial accelerometer, and 2-channel data acquisition, and also using an impact hammer, tri-axial accelerometer, and 4-channel data acquisition are included in this paper.
Research Manager, Kittiwake, Parker Hannifin Manufacturing Ltd
Research manager at Parker Kittiwake since 2007. Responsible for technology scoping in New Product Development for condition monitoring of and quality testing in fuel, lube oil and hydraulic fluids. Main market segments for the products are marine and heavy industry.Graduating with... Read More →
This paper describes a design for an example of a robust Prognostic Health Monitoring (PHM) system for a dual set of electromechanical actuators (EMA) subsystems. The paper is presented as a case study to address some of the major in a design of a PHM system that you face. Given a set of assemblies and subassemblies that are to be monitored for prognostic support, (1) How do you continually and simultaneously monitor them? (2) What considerations are related to sampling rates; prediction accuracy, resolution, and precision; state-of-health (SoH); (3) How do you associate a node or nodes to which one or more sensors are attached to software-program modules to condition data and extract features? (4) Is data to be acquired and processed in real time or near real time (directly from a sensor to program modules to prediction algorithms) or as batched data kept in files? The paper uses historical and experimental data with results from an example PHM system.
This paper describes how to evaluate the accuracy of prognostic estimates for remaining-useful life (RUL), state of health (SoH), and prognostic horizon (PH). The primary goal of prognostics is to be able to accurately predict a future failure in systems, including condition-based-maintenance (CBM) systems that use condition-based data (CBD) to detect degradation and project that degradation to a failing level at a future time. That future time and the time when the data is sampled are used to calculate predictive (prognostic) information such as RUL, SoH, and PH. Evaluating the accuracy of prognostic information is critical and begins with knowing that ideal, zero-error estimates are not practical, primarily of factors such as the following: how long it takes for degradation to progress to a failing level, when the onset of degradation occurs, changes in rate of degradation due to operating and environment conditions, and noise in the sensor-measurement system. Performance metrics to evaluate accuracy are developed and presented: convergence efficiency, prognostic distance (PD), and prognostic-horizon accuracy (PH).
For on-shore applications, the support structure for a rotating machine normally consists of a block foundation. Adequate dynamic analysis of support structures for rotating equipment is necessary to ensure good conditions for the operation of the supported machine and safeguard other machinery near the subject rotating machine, as well. The normally practiced method tries to achieve an under-tuned support structure which considers the machine mass and block foundation mass as one entity. Calculations consider that unbalance of the rotating machinery train generates a centrifugal force, which depends on the total mass of the set distributed in the two points of the axis, on the eccentricity between the centre of gravity of the rotor and the geometric axis of rotation, and on the angular velocity of the train. The design procedure assumes that the unbalance force is transmitted to the foundation. This paper endeavours to highlight that revised design procedures for support structures of rotating machinery should consider the stiffness ratio of the bearing support (casing) and the rotor - bearing system and restrict to assume that all vibrations are transferred to machinery foundation irrespective to the type of machine. Similarly, in the case of equipment with AMB (active magnetic bearings) very less vibration is transferred to the casing and after foundation as the magnetic flux helps rotor to levitate. This paper takes an approach of rotor dynamics to propose a revised thought process of structural support design. The second objective of this paper is to propose the foundation be supported by low cost dampers at all sides of foundation block thereby minimising shear waves transmitted to the ground. This proposal should enable a smaller foundation size with less effort on site construction.
This project, sponsored by the Office of Naval Research, integrates oil and vibration analysis to develop a more complete diagnostic and prognostic model for monitoring bearing degradation and performance while inside a vehicle's differential. The results of this project could be used to optimize repair schedules and minimize the risk of catastrophic failure of machine components for military personnel during a mission through condition-based maintenance. Each test bearing is first introduced with an outer race fault. The fault's degradation is then accelerated by running the bearing under overloaded conditions in the boundary lubrication regime. Accumulated damage to the bearing is characterized in a clean, low-noise environment before being implemented into the differential for in-situ testing. Dynamic characterization of the designed test machines were performed by long runs spanning multiple hours under rated conditions to determine any wear-in effects and ramp-up tests to distinguish order-based bearing frequencies and structural resonances. Modal analyses were performed on the static system to provide additional evidence of structural resonances within the machines. This paper will discuss the design challenges and solutions for creating a test bed to monitor bearings in accelerated-degradation conditions and in noisy environments, such as the differential. Results of run-to-failure data and analysis will also be presented.
Machinery diagnostics is entering a new era, with new urgency, as industry moves toward better asset management and eventually to unmanned operations. Owners and operators are expecting the advanced machines of the future to have the ability to self-diagnose conditions that could lead to catastrophic failure or to unanticipated down time. Next-level, algorithm-driven associations to yield, machine efficiency, and other operating characteristics that can be defined in terms of the energy associated with known machine processes can also be translated into useful parameters for transmission over a digital data bus.
To reach those goals, Dytran has created CAN-MD®, a cutting edge, digital sensor platform that processes raw analog data inside the sensor, enabling a bussed architecture that delivers actionable results, not raw data. Advanced sensors with on-board digital signal processing (DSP) features are the key to this new machine awareness. The advent of smaller, more powerful microprocessors enables a new generation of bus-based digital vibration sensors to process and reduce analog data inside the sensor itself.
The new technology eliminates the long wire runs to each sensor commonly associated with traditional analog test cell arrangements and replaces it with a single-cable, all-digital bus-based schema. In addition, the improved system architecture provides reduced SWaP (size, weight & power) of traditional onboard VHM (vibration health monitoring) systems, easier troubleshooting and more importantly, distributed processing.
CAN-MD® offers a variety of analog sensor adapters that allow users to add existing sensors to the CAN-MD® network. This extends many of the benefits of the CAN-MD® technology to legacy vibration sensors or other measurement node types. By expanding the measurement input possibilities, it allows the system to provide improved sensor fusion, pulling data from a greater number of nodes to allow users to make data driven decisions based on multiple sensor locations or measurements types. Sensor adapters currently include tachometers, optical blade trackers, IEPE sensors (acceleration, pressure, force) and high temperature charge mode sensors.
Our talk will discuss the digital sensor fusion domain of CAN-MD® technology and our strategy of converting all sensors to a common, bus-based environment.
The amplitude of machinery vibrations becomes very high when any of a support structure resonance frequency lies within the regime of machinery operating frequency. During resonant condition, the support structure normally demonstrates vibrations in rigid body modes and aggravating the shaft vibration conditions. Sometimes, the evaluation of resonant frequencies for equipment base frame is not given its due importance by the OEM. For medium sized rotating machines. To identify such condition at site and rectification is cumbersome. Tuning of the base frame can be done either by increasing mass m or increasing k by stiffing the base frame to shift its resonant frequency. A passive and active control of coupled modes of vibration (rocking and pitching) can be controlled by using servo valves and hydraulics. For passive control, some devices can used which can mitigate the structural resonance problem. The proposed methods shall cover any possible discrepancy of structural resonance and on-site grouting issues where a support structure may undergo push pull action by above lacunae thereby aggravating machinery vibration
Steve Greenfield has served as regional technical support manager for Eaton’s Aerospace Group since 2007 and has more than 30 years of experience in advanced sensor systems for gas turbine and rotorcraft health monitoring.Greenfield works with airline customers in Europe and Africa... Read More →
Research Manager, Kittiwake, Parker Hannifin Manufacturing Ltd
Research manager at Parker Kittiwake since 2007. Responsible for technology scoping in New Product Development for condition monitoring of and quality testing in fuel, lube oil and hydraulic fluids. Main market segments for the products are marine and heavy industry.Graduating with... Read More →
Mark Redding is President and Co-Founder of Poseidon Systems, LLC. He has spent most of his career developing advanced PHM tools and technologies to monitor the reliability of equipment for aircraft, heavy machinery, processing equipment, marine, mining, and power plants. After selling... Read More →
This tutorial will provide attendees with the fundamental principles of digital signal processing in a simplified manner without resorting to the underlying complex mathematical structure. Students will learn how to utilize their existing vibration analysis tools to extract the maximum information from the vibration signal. You will learn how to set up your data collector to maximize the benefits. The benefits will include how to determine the sampling rate, resolution, max frequency, and how to distinguish if you are missing a frequency component in your selected resolution, etc. I will also, present the new developments such as Spectra Kurtosis, Cepstrum Analysis, order tracking, and demodulation techniques
Dr. Suri Ganeriwala is founder and president of Spectra Quest, Inc. He has over thirty-five years of industrial and academic experience in machinery vibration diagnostics and control, signal processing, and viscoelastic materials characterization. Suri has worked for Philip Morris... Read More →
Peter Gaydon is the Director of Technical Affairs at the Hydraulic Institute. Mr. Gaydon held design, development, and test engineering positions with major pump manufacturers prior to joining the Hydraulic Institute. With the Hydraulic Institute, Mr. Gaydon has technical responsibility... Read More →
Additive Manufacturing has changed the face of conventional metal forming technologies. Conventional metal casting involves the manufacture of solid forms or patterns that sand is formed around to develop the shape of the cavity that is filled with liquid metal. The patterns or tooling for castings will often cost as much as 100 times that of the needed casting and require weeks to months to produce. The tools must then be stored, maintained and repaired between uses. Because of the amount of space required and the length of time required for producing patterns, the technology is not conducive to a mobile manufacturing platform. Additive Manufacturing (AM) and digital part creation has allowed the production of castings within days rather than within weeks or months. It has eliminated the need for the lengthy production of costly tooling with requirements for storage space. With the proper tools and training additive manufacturing for metal casting will allow replacement of critical cast components and provide for weapon system sustainability on or near the battlefield. The objective of this project is to show a proof-of-concept of in-theater production of replacement parts for long lead time DoD critical components utilizing a sand 3D-printer with actual desert sand to produce sand molds from a CAD drawing of the actual part, followed by pouring molten aluminum into the casting molds. The parts will subsequently be produced by optimizing the casting mold design, with appropriate gates and risers to allow for successful casting.
Although vibration analysis has been used to monitor the health of rolling element bearings for more than half a century, a reliable method to accurately determine the severity of defects from vibration data remains an elusive goal for both academics and industry.
Rolling element bearings are critical to the safe and economic operation of industrial machines in all sectors. Determining whether a defect is present or not only answers part of the problem. Without a knowledge of the severity of a defect, engineers and plant operators have only limited information upon which to base planning and maintenance decisions. Accurate information on the size and location of the defect as well as the current operating conditions are all essential to properly characterize defect severity and to allow effective corrective actions to be determined.
The complexity of the defect impacts in rolling element bearings, the influence of the vibration response and the transmission path between the defect and the sensor, changes in operating conditions and other sources of vibration all conspire to challenge the accurate measurement of defect severity even in the simplest of machines. Historical trending of vibration data from multiple machines and/or multiple failure events remains the only option and even this can have limited success.
In the early 1990s, ground-breaking research by the author into the origins of vibrations generated by discrete faults in rolling element bearings revealed previously hidden features within the impact events caused by defects. These features offered the possibility of a new approach to determine defect severity whereby defect size could be accurately and directly determined from vibration data without the need for historical trending.
In recent years this original research has been further progressed by the author and the original approach has been significantly enhanced to provide more information to characterize defect severity. Additionally, the enhanced method has proven to be more effective in industrial environments where the features of interest can be masked by noise and other source of impact.
This presentation will background the challenge of determining defect severity in rolling element bearings and summarize the findings of original research. The enhancements that have been made to the method will be detailed and the additional condition information provided by the enhanced method will be described. The simplicity of the method and its potential to be integrated as an on-sensor IIOT device will be demonstrated and results from testing of recent prototype devices will be shared.
Iain’s professional career has been focused on innovation and growth by successfully combining new ideas and talented people to bring new technologies to life in a wide range of industries across the globe. On completing his engineering studies, Iain travelled to Europe where he... Read More →
In looking into the components of a human monitoring system, there are three main elements that comprise the system: sensors; data acquisition and communication; and data processing and analytics. Sensors that function with the purpose of sensing body movements or collecting specific physiological or biological parameters of an individual are typically known as wearables. Wearables used for capturing body movements are primarily inertial measurement units (IMUs) which utilize sensor fusion to combine the technology of an accelerometer, gyroscope, and magnetometer. Data obtained from these wearables may provide important insight into subtle differences in body movements that influence performance outcomes. The long-term goal of our work is to develop approaches that enable the prediction of an individual’s performance in an open-skilled environment. The specific aim of this research was to determine how data from a full body IMU-based system could be used in detecting subtle movement differences in the execution of a pre-planned agility test versus a reactive agility test._x000D_ Ten healthy young adult males who were regularly physically active and had played on a sports team within the past four years participated in this study. An Xsens Awinda 17 sensor suit was used to capture body movement data during the two agility tests. In both the pre-planned and reactive agility test, study participants stood facing six programmable illuminating lights, arranged in a 3 meter arc, 1.5 meters apart from each other, with the center being the starting position. For the pre-planned test the participant was informed which three lights in order they would run to, with the requirement that they return to the center before advancing to the next light. For the reactive agility test, the participant was informed that all 6 lights would turn on, but the light they must run to and turn off would not have a color pair (ex. 3 red, 2 blue, 1 green). Participants were also asked to remember the color sequence of the three lights they turned off for an additional stressor._x000D_ Throughout the tests, each participant’s center of mass (CoM) was tracked and compared to the direct path to the light with deviations calculated using standard sums of squares error (SSE). Study participants were broken into two groups for comparison: faster individuals and slower individuals based off of their completion time on the pre-planned agility test. Results from the pre-planned agility assessment indicated that the mean SSE for the CoM deviations to the light averaged 1.66 ± 1.43 meters for the faster performers and then 2.96 ± 2.05 meters for the slower performers. For the reactive agility tasks, the CoM deviations were higher with an average SSE of 5.82 ± 4.16 meters for the faster performers and 6.97 ± 7.60 meters for the slower performers. This suggests that the slower individuals were not just slow, but were likely slow because they were inefficient. Additionally, these findings indicate the notable increase in inefficiency generated by the uncertain scenario of the reactive agility test which corresponds to the 84% time increase from the pre-planned test.
Electrical distribution systems require annual inspection of bolted interconnections to identify loose parts and prevent arcing. This periodic maintenance is becoming increasingly critical on US Navy vessels since many ships are designed with electrical distribution systems of 4,160 VAC (Medium Voltage - MV) and 13,800 VAC (High Voltage - HV). Additionally, new power transmission systems are under development where multiple segments of electricity carrying insulated bus pipes will be connected with bolted hardware. Inspecting electrical system components at these voltage levels involves a time consuming preparation and assessment process to insure safety of the personnel and avoid potentially hazardous conditions. Another issue of these periodic inspections is that an interconnection may be acceptable at time of inspection but loosen up shortly after and cause damage before the next inspection cycle. The optimal inspection system would consist of continuously monitoring the interconnection points to detect the formation of faults at the very early stages. The fiber optic distributed sensing (DTS) technology where the entire length of an optical fiber becomes a continuous temperature detecting device can be utilized to remotely monitor the electrical interconnect point and also monitor electrical cables, machinery, and other electrical devices. This technology is used extensively for applications including oil and gas extraction, transport, and processing, fire detection in buildings and tunnels, monitoring the operation of power cables, and geo-hydrological systems monitoring.
A number of development efforts funded by the National Shipbuilding Research Program applied the DTS technology using Raman scattering for the detection of loose connections in MV and HV electrical panels and to monitor the interconnection points of an electrical distribution system using insulated bus pipes. Both applications presented a number of challenges including insuring that a sufficient length of fiber was used to provide accurate detection, multiple closely spaced measurement points could be accurately identified and measured, and that the fiber was installed in contact with the surface being measured.
The trials confirmed that the Raman DTS can be used to effectively monitor electrical connections. Additionally the efforts also demonstrated that the DTS system can use the same optical fiber cable length to monitor other conditions and devices including temperatures of zones where the cable is routed and machinery, including using the system for fire detection. Challenges to be addressed for the full scale implementation of this versatile technology on naval vessels include: a) facilitating the methods of application of the sensing fiber to the parts monitored; b) connections of multiple fiber segments monitoring individual devices to create a continuous length of fiber connected to one single sensing channel; and c) programming the DTS device to identify the individual devices and zones, with the temperature ranges and alarms specific to each zone and device. The presentation will describe the applications of the DTS technology evaluated through these NSRP programs, the results of the tests performed, the follow on efforts required for wide scale implementation on Navy vessels, and how these can be applied to monitoring commercial electrical systems.
CEO / Chief Technology Officer, RSL Fiber Systems, LLC
Giovanni P. Tomasi is the founding CEO and Chief Technology Officer of RSL Fiber Systems, LLC. He has been involved in the military shipbuilding industry for over 30 years, starting in the 1980’s collaborating with the shipyards and the Navy to define the requirements and develop... Read More →
The tutorial will provide an overview of the fundamentals and principles of Systems Engineering (SE). This includes understanding the processes that are used to assist the engineer in a successful design, build and implementation of solutions. The context of this tutorial will be to describe the involvement of SE throughout the life-cycle of a project from cradle to grave. Due to the ever growing number of complex technical problems facing our world, a Systems Engineering approach is desirable for many reasons. The interdisciplinary technical structure of current systems, technical processes representing System Design, Technical Management, and Product Realization are instrumental in the development and integration of new technologies into mainstream applications. This tutorial will demonstrate the application of SE methodology to these types of problems. Learning objectives include: Understand how to develop and engineer a system; understand the importance of teams, communication, requirements definition, interface control, and focus on deliverables; useful tools and techniques for each aspect of the SE process. The first 30 minutes will be used to present the overview of SE. The remaining 60 minutes will be used to provide special consideration of attendee’s specific concerns.
This project, sponsored by the Office of Naval Research, integrates oil and vibration analysis to develop a more complete diagnostic and prognostic model for monitoring bearing degradation and performance while inside a vehicle's differential. The results of this project could be used to optimize repair schedules and minimize the risk of catastrophic failure of machine components for military personnel during a mission through condition-based maintenance. Each test bearing is first introduced with an outer race fault. The fault's degradation is then accelerated by running the bearing under overloaded conditions in the boundary lubrication regime. Accumulated damage to the bearing is characterized in a clean, low-noise environment before being implemented into the differential for in-situ testing. Dynamic characterization of the designed test machines were performed by long runs spanning multiple hours under rated conditions to determine any wear-in effects and ramp-up tests to distinguish order-based bearing frequencies and structural resonances. Modal analyses were performed on the static system to provide additional evidence of structural resonances within the machines. This paper will discuss the design challenges and solutions for creating a test bed to monitor bearings in accelerated-degradation conditions and in noisy environments, such as the differential. Results of run-to-failure data and analysis will also be presented.
Failure prognosis of the rolling element bearings (REBs) is crucial in the rotating machinery. The damage evolution in the REBs consists of two main phases: damage initiation and propagation. The conventional REB life models address the lifetime of the bearing to the damage initiation, i.e. first defect formation. However, after the first defect formation, the bearing might be fully operational for millions of cycles. There has been a growing awareness of the need to understand the damage mechanism during the propagation phase. Over the past two decades, studies attempting to understand the damage mechanism and to develop damage propagation models have been published. Nevertheless, the damage mechanism is only partially understood, the existing models are inefficient and the physical phenomena are not well represented. Once the damage mechanism is understood a physics-based prognostic tool can be developed. In order to understand the spall propagation phase, a physics-based model has been developed. The model aims to study the material behavior at the trailing edge of the spall during the rolling element (RE) impact. Based on the model results a qualitative damage analysis for crack evolution within the spall edge was conducted. Moreover, a metallurgical analysis of the bearing from endurance tests was carried out. The metallurgical analysis added insights regarding the damage mechanism and was used for model validation. The results achieved from the damage analysis are in good agreement with the experimental observations. To our best knowledge, this is the first study attempting to simulate damage evolution within the spall edge based on physical insight.
Dr. Suri Ganeriwala is founder and president of Spectra Quest, Inc. He has over thirty-five years of industrial and academic experience in machinery vibration diagnostics and control, signal processing, and viscoelastic materials characterization. Suri has worked for Philip Morris... Read More →
Bearing faults in machinery are among the most critical faults that require attention by maintenance personnel at early stages of fault initiation. In many cases it is difficult to directly and accurately identify the fault type and its extent under varying operating conditions. This work demonstrates a novel procedure for bearing fault detection and identification in an experimental set-up. Three seeded faults in the rotating machinery supported by the test roller bearing include inner race fault, outer race fault and one roller fault. The rotor is run at different speeds and with different levels of rotating mass unbalance. Accelerometer based vibration signals are analyzed for the different bearing faults’ signatures using statistical moments, frequency spectra and wavelet coefficients. These are then used as features to create a data-partition using relational data clustering. The composite differential evolution technique is proposed for optimizing the clustering parameters. The objective is to correlate bearing faults to the extracted vibration features. The results of this analysis will be extended for applications in real time bearing condition monitoring system.
This paper describes an multivariate-analysis (MVA) methodology to detect and prognose three types of faults associated with an electromechanical actuator (EMA). The faults are the following: (1) loading faults, such as friction, on the shaft of an EMA motor, (2) shorting faults in the stator windings of the EMA motor, and (3) on-resistance faults in one or more power-switching transistors used to convert direct voltage/current into alternating current. The presented methodology overcome difficulties associated with typical MVA methods such as the following examples: solving simultaneous equations, performing a statistical-based analysis such principal component analyses (PCA), and a K-nearest neighbor (KNN) regression or other Euclidean-based distance methods. Examples of those difficulties are the following: (1) 'noise' in the data containing the signal(s) of interest, (2) method produces information suitable for classification' rather than diagnosis or prognosis; and (3) the data does not include known independent variables, rather all variables in the data are dependent - which is the case for the phase currents of an EMA. A unique root-mean-square (RMS) of quantifying phase current values and a methodology for using those values are presented with examples that demonstrate that phase-current data for each of the three faults can be processed to unequivocally identify and isolate the fault, and to prognose a future time at which functional failure is likely to occur.
Rolling element bearings are essential components in rotating machinery, it is of great importance to detect the bearing fault as earlier as possible during the operation of machinery. In recent years, variational mode decomposition (VMD) has been widely used in signal decomposition and feature extraction from non-stationary signals. However, the determination of the decomposition layers and quadratic penalty parameter of VMD is still puzzling. In this paper, an improved VMD and optimized frequency band entropy (OFBE) method is proposed. Firstly, the energy entropy maximum principle is utilized to select the decomposition layers and quadratic penalty parameter. After that, a detection method of OFBE based on the principle of maximum kurtosis is presented to determine the bandwidth parameters. Finally, envelope analysis is employed for the detection of fault-related frequency components based on the obtained sub-band signal. The performance of the proposed approach is validated by faults in different parts of the bearing. Results show that the new methodology yields a good accuracy in bearing fault diagnosis. Keywords: Fault diagnosis; bearing; VMD; entropy; envelope analysis
Dr. Suri Ganeriwala is founder and president of Spectra Quest, Inc. He has over thirty-five years of industrial and academic experience in machinery vibration diagnostics and control, signal processing, and viscoelastic materials characterization. Suri has worked for Philip Morris... Read More →