1     Project Summary

Machinery for manufacturing and plant/asset operation needs to be maintained to ensure continued functionality. Preventative main­tenance is not sufficient to avoid unscheduled and costly downtime of equipment, particularly those that involve heavy loads and intermittent service.

Traditionally maintenance has followed the philosophy of either run-to-failure or planned maintenance at regular intervals. Each of these approaches has been found to be more expen­sive and time-consuming when compared to condition-based methods, under which the condition of a machine is monitored and maintenance is only undertaken if conditions warrant it. This method equally applies to manufacturing processes where the settings of some machines or components may need to be altered based on the monitored condition of the process.

There exists a great need within the DoD and industry for reliable, cost effective condition-based maintenance (CBM) systems. CBM when correctly implemented reduces costs, improves safety, extends life and yields products of better quality in manufacturing. Such systems can only emerge from adequate research and development in the areas of monitoring, diagnostics and efficient modeling of the CBM system.

Improved prognostic data would enable more accurate development of depot and intermediate level work packages and save man-hours by enhancing existing CBM methodologies to improve machinery availability and reduce costly emergent repairs. It is widely recognized that unplanned emergency maintenance is up to ten times more costly than planned maintenance. Even in the case of “emergent” work during planned availability (Naval vessels) the cost is 175% of work bid in an initial work package.

In the case of the CVN/CV, fleet studies show that some 20% of funding for each ship availability is spent on various repairs to rotating machinery ($2 million). An additional $10K is spent on express shipping of critical parts. NAVSEA estimates regarding earlier detection of needed work, suggest a saving of in excess of $250,000 per ship per availability for emergent work. In addition, substantial savings would be generated from reducing failure of equipment at sea. It has been forecast that total savings for the CVN/CV fleet would be in the order of $32 million.

In the modern military and industrial setting, there exist many methods for assessing the con­dition of machines, with vibration analysis being most commonly used and understood. To a lesser extent, other techniques such as thermography and wear monitoring are also adopted.

The significant advancements in development of high-frequency acoustic emission (AE) have provided another option for monitoring the condition of rotating equipment. During controlled laboratory based tests, AE has shown the ability to detect faults earlier than more traditional and established methods of condition monitoring.

The aim of the Damage and Wear project was to investigate the comparative effectiveness and ease of use of AE, vibration analysis and some supplementary signals in the areas of machine condition monitoring and diagnostics. The goal was to develop a simple and cost-effective method of early detection of failure in rotating machinery to:

• Qualitatively assess a machine’s condition

• Rank identical machines against one another

• Monitor the rate of degradation

• Provide information as to when intervention was required.

The breadth of applications ranged from traditional rotating equipment (motors, pumps, and fans that operate as low as ½ rpm) to leaking valves, seals and even cracks in machinery structures.

1.1     Project Partners

• IMES Inc. (formerly Water Weights)

• NCMS

• Cincinnati Machine

• Ford Motor Company

• Ford Windsor Aluminum Plant

• Ford Sharonville Transmission Plant

• Ford Van Dyke Transmission Plant

• Ford Essex Engine Plant

• Ford Windsor Casting Plant

• Wayne State University

• U.S. Marine Corps – Albany

• U.S. Marine Corps – Barstow

• U.S. Army – Red River

• U.S. Navy – Portsmouth

• U.S. Navy – Cherry Point

1.2     Main Findings

Use of AE technology by project partners was new compared to years of experience using vibration analysis. In particular, there was concern over the reliability of some of the collected AE data due to the technology’s dependency on a good coupling between the sensor and the machine under test.

From the results, it appears that the use of an AE signal is indeed viable, and serves as a strong complement to vibration analysis. While the science of detailed diagnostics by AE is still being developed, a combination of AE and vibration analysis seems to be a viable solution that is ready for application now.

The challenges of this application have been explored with hands-on usage results, with respect to a prototype AE monitor, the PUMA^™ Bearing Health Monitor. In summary the main findings of the project were:

•    The PUMA^™ Bearing Health Monitor AE sensor and data acquisition system appears to be accurate and useful, although a number of improvements to the prototype unit are suggested. These mainly relate to sensor size, cabling, hook-up and some software issues (see Section 6.1 for detail).

•    AE signals appear to be generally faithful vectors of machine condition.

•    RMS AE signals also appear to be generally useful as vibration signals when used to trend machine health. In specific cases, such as for diesel equipment, where background vibrations usually make vibration monitoring problematic, AE signals may be more useful.

•    AE monitoring appears to have the advantage of ease of use, if the major concern is to ascertain whether a machine fault exists and/or is progressing. As at now, vibration-based diagnostics may then be used for more detailed investigation of a particular, identified fault situation.

In 53% of cases AE detected a change in machine state before vibration analysis. Vibration analysis only detected a change in machine state before AE in 43% of cases. Overall the two technologies performed equally in 4% of cases.

From examining the results, specifically the first technology to detect changes in the trend of readings, it can be seen that AE does provide advanced warning over vibration analysis. The consistency of the results obtained from AE is generally lower than those obtained with vibration analysis technology; however, the variation in the performance of AE technology across different sites may be attributed to the processes applied to taking the readings when using the AE instrument.

The most appropriate technology for each machine type monitored during the project was collated from the results of the analyses performed on the data collected during the project. The results are shown in Table 1-1.

It can be seen that AE outperformed vibration for the majority of machine types. Vibration analysis appears to perform marginally better than AE for the assessment of the health gearboxes and low-speed axles.

1.3     Benefits

Depot users agreed that the PUMA^™ Bearing Health Monitor was a useful addition to their suite of machine-monitoring tools.

1.3.1     Albany MC³

Albany has reported the AE equipment provided by the Damage and Wear project is still in use and considered a success. The technology works even better with a well-planned preventative maintenance program. The equipment proved to be a value in the fact that it identified pending failures in production equipment before an actual failure occurred. A great potential is seen in a networking system if a permanent mount sensor could be manufactured at a reasonable price. Future plans are to expand usage and possibly automate the process throughout the depot. In addition, there is an interest in development of a project to see if this technology will work on diesel engines.

Table 1-1.   Summary of Most Appropriate Technology for Each Machine Type

1.3.2     Portsmouth Naval Shipyard

Portsmouth used the Damage and Wear equipment on cranes and indicated that it is far better than Fast Fourier Transform (FFT). However, at this time, without the support of their business office and the BRAC outcome, Portsmouth cannot move the technology forward towards real-time monitoring using wireless technology.

1.4     Conclusions

From the recorded data, AE correlates to a level considered “Good” or ”Fair” in 80.5% of cases. However, considering that the operators are inexperienced in the use of AE equipment, this statistic becomes more significant. Using a simple AE instrument, relatively novice operators were able to obtain a similar level of trending information to that obtained from vibration analysis.

During the project period AE performed better than vibration analysis or at a similar level on the majority of machine types with a large enough population from which to draw conclusions.

The consistency of results obtained from the AE instrument does vary across the sites that partici­pated in the project. It is believed that this is due to the processes used to acquire the AE readings, because as project partners became more familiar with the instrument the consistency of results improved.

It appears, from this project, that AE is an effective tool for the assessment of machine health. AE has performed more effectively than vibration analysis in terms of being the first technology to detect a change in machine state and in the relative magnitude of response for a change in machine health. However, it should be recognized that the AE instrument supplied does not have the same diagnostic capabilities as the more widely available vibration analysis technology, although the AE instrument is simpler and more suitable for a novice operator. From this, it would be logical to conclude that, for optimum machinery fault detection, a combination of the earlier detection of AE technology and vibration analysis technology’s diagnosis capabilities should be used.

The prototype AE PUMA^™ Bearing Health Monitor has performed well overall. Unfor­tunately some project partners ceased using equipment driven more by logistic, economic and other issues rather than “lack of performance.”

1.5     Further Work

As expected, there is work to be done on the existing prototype AE hardware before it is ready for full-scale commercial use. Most importantly, the sensor size has to be reduced, but there is also work to be done on the process used to take readings. It is hoped that a reduction in AE sensor size will permit it to be positioned closer to the component under test, and by optimizing the process used to take readings, that the consistency of the readings taken with this particular piece of AE technology can be improved so that it becomes a valuable tool for plant managers.

This project has determined that AE provided warning of a change in machine state before vibration in a majority of cases, and equally that the relative magnitude of response from AE technology is higher for a change in machine state than it is from vibration technology. In a majority of cases these assessments have been obtained qualitatively.

It is strongly suggested that a project be initiated to fully quantify improvements AE offers over vibration technology and at the same time calculate detailed projected savings for specific military and industrial use.