Violet Leavers

V4L Particles L td.

v.leavers@V4L-group.co.uk

UK

Moreover the replacement by the equipment manufacturer under warranty does nothing to mitigate the cost of unscheduled downtime and lost revenues. A solution to this problem lies with the various fluid and particulate condition monitoring tests that convey information about the current mechanical state of a system. In the front line of these is the collection and analysis of wear debris particles taken from a component’s lubricating or power transmission fluid. Wear debris particle analysis is so important because sampling is relatively simple to execute, the test is non-destructive and it can give a vital early warning of incipient component failure. The current work reviews recent advances in wear debris particle analysis with relevance to the needs of effective, on-site condition monitoring.

Particle Sizing and Counting Hardware

Particle counts can be determined using optical instruments, the first of these methods is to use a microscope. Particles are precipitated from fluid samples which are taken from the component’s lubrication system by draining through a filter patch. Particles are then interactively sized and counted using a microscope [1]. However, because of its labour- intensive nature, this method was superseded in the 1960’s by Automatic Particle Counters (APCs) [2].

First generation APCs contain a laser light source and a detector, which are separated by an optical cell. The oil sample flows through the cell and when a particle passes through it, an area of light is obscured. The detector senses the loss of light and outputs a voltage. The voltage pulse generated increments the particle count and the height of the pulse is used to determine the size of the particle. First generation APCs have the disadvantage that they cannot distinguish between multiple particles and because they are ‘blind’ to the shape of the particle they are only able to report size in terms of a projected area equivalent diameter. Size is defined as the diameter of the disc with an area equivalent to the area of the particle’s shadow. This leads to a significant source of error because the estimated projected area equivalent diameter is a function of the shape of the particle [3] and the size of the particle is increasingly underestimated as the shape becomes more elongated. In particular, long and thin particles will be systematically undersized to the point where they may slip into a size range smaller than their actual size indicates or, even more critically, disappear from the count all together [3].

Typical harsh on-site installation environment for industrial equipment.

A second generation of APCs [4] has emerged, which operate using microsecond duration-pulsed lasers which freezes the image of the particles present in the optical cell. The light sensor associated with first generation APCs is replaced by a CCD sensor. The system is able to collect the silhouette images of multiple particles and image processing is then used to count and size the particles.

Various contaminants such as varnish or fibres have optical properties that make them invisible to APCs. Such contaminants can build up to critical levels without being detected by the APC. Reference [4] section 6.1 gives a list of eleven possible sources of error when using a second generation APC. It would require a relatively high level of skill and experience, which is not generally available on-site, to detect these errors. In addition the limited size of the optical cell requires a filter to limit the size of detectable particle to <100 μm. For much of the large equipment used on-site, particles <100 μm are considered benign and hence the apparatus would filter out the very particles that indicate significant wear.

It is clear that even the addition of technical upgrades cannot solve the problems associated with the use of APCs. This means that on-site personnel must use them with caution and often in tandem with a microscope if they are to have any confidence in the results obtained. To do this a patch is made from each sample passed through the APC. In this way the operator can monitor for contaminants that are invisible to the APC, and also qualitatively verify the magnitude of the particle count by comparison of the patch with a chart showing images of patches over a range of cleanliness codes. If the cleanliness code indicated by the chart is very different from that indicated by the APC, then the particles are interactively sized and counted using a microscope. New technology that removes the need to perform this costly and time-consuming double-checking of results is long overdue.

Particle Sizing and Counting Sofware

There are various particle sizing and counting software packages on the market. However these are generally available only for purchase with top-of-the-range branded microscopes and this takes such equipment beyond the budgetary limits for most on-site applications. Moreover, standardization of microscope design can never fully exploit advances in digital imaging technology. This is because the microscope still has many disadvantages associated with its use in wear debris particle analysis. In particular, a microscope is not well suited to viewing, analysing or imaging particles that vary from macroscopic to microscopic size. This is because implicit the assumption in the design of the microscope is that the dimensions of objects of interest will fall within the same size range.

When viewing microscopic particles, the shallow depth of focus at high magnifications means that microscope stands must be particularly stable in order to view and capture blur-free images. A good quality microscope stand will normally weigh more than 20 kg. Recently, the use of motorised stages to obtain images with artificially extended fields of view or depth of focus, makes such microscopes not only more expensive but also even more weighty and cumbersome. The fact that such microscopes are not easily transported prevents their use in the field or on-site and in addition, they require extra training and know-how in both computing skills and knowledge of image processing techniques to be able to navigate the software in order to acquire the desired quality of image.

Current particle sizing and counting algorithms require a high contrast, grey level image, for which the user needs to input a threshold in order to separate the particles from the background. This adds subjectivity to the analysis and risks the sofware failing to detect low contrast details that might be associated with translucent mineral particles or fibres. Both of these types of contamination can cause significant damage if left undetected. However sophisticated are the microscope optics, hardware, accuracy and repeatability of the system, it will always be dependent on the subjectivity of user- defined thresholds. In addition, the value of the threshold needs to be routinely monitored by the user. This is very tedious and each operator will have a different idea of where the threshold should be. In this sense such software is both subjective and only semi-automatic. The microscope is simply the wrong starting point for dedicated particle sizing and counting using digital imaging technology [5].

Wear Debris Particle Classification Software

In contrast to other quantitative fluid analysis techniques, wear debris particle analysis requires a high level of training and expertise and for this reason is usually carried out in specialized laboratories. However, on-site maintenance professionals are often forced to work in remote locations and are prone to long periods of isolation due to adverse weather conditions. In such circumstances staff may need to make maintenance checks and judgements without the luxury of being able to send samples to a specialist lab for analysis. Moreover, high rotation of personnel and variable levels of skill and training pose an additional problem.

The earliest form of particle analysis software is the wear debris particle atlas, a composition of specially selected images of the idealised form of particles. Such images may bear no relation to the torn and degraded particles that are under analysis in the field and for this reason it may be difficult for the end user to find an instance of a particle that is sufficiently close to the particle under analysis. Thus, wear debris atlases have very limited use in on-site condition monitoring situations.

The next level of wear debris particle analysis software uses segmented images of particles to extract morphological parameters such as the longest length, width, aspect ratio, area and perimeter. This type of software merely transforms the image data into sets of parametric data, which still requires expert interpretation. Moreover, without the information concerning the particles’ surface texture and colour, it would remain difficult for even an experienced analyst to provide a reliable diagnosis. This type of software package is thus not suitable for on-site use where training and skill levels are highly variable.

Another type of wear debris particle image analysis software is detailed in ASTM D7596 -11, Standard Test method for Automatic Particle Counting and Particle Shape Classification of Oils Using a Direct Imaging Integrated Tester [4]. This software uses a neural network to identify the wear debris mode from particle silhouettes captured on a low resolution, 0.3 megapixel CCD array. However, pre-selection of the input to the neural network will bias the identification results and moreover particles rarely present with idealised shapes, instead they are more often shredded or torn. The ASTM standard [4] (section 6.10) recognises that identification using this method may be unreliable and suggests that further, interactive investigation by microscopy, such as examination of particles on a membrane filter patch or analytical ferrography, may be necessary to better determine the exact particle types present. Once again, this requires a level of skills and training that may not be present on-site. Moreover, this type of technology still fails to remove the need to perform costly and time-consuming double-checking of results.

Figure 1. Macro-2-Micro one shot image, at x500 magnification, of an oxidized particle on a filter patch showing surface detail such as would not be visible using a microscope without extended focus capability.

Innovative Particle Imaging Hardware

New technology has recently become available [5], which solves many of the practical limitations imposed by the traditional design of the microscope when viewing and capturing the images of both macroscopic and microscopic particles. The new technology is dedicated to optimising the lateral and axial resolution available at the magnifications and resolutions required to reproduce images in an electronic form, whether that is for data storage, printing in reports or for on-screen viewing. In this way images can be generated in the way that the depth of focus and field of view are optimised for viewing macroscopic and microscopic particles at the same magnification.

Using the new Macro-2-Micro technology [5], it is possible to acquire sharply focused images over a much wider range of magnifications and resolutions than when using a traditional microscope and without resorting to motorised stages or specialised software to create a wider field of view or extended depth of focus. Figure 1 shows a Macro-2-Micro one shot image of an oxidized particle on a filter patch. Without such a sharp image allowing the surface detail to be seen, this particle might be mistaken for a brass/copper fatigue particle, whereas it is a hybrid particle with the striations associated with severe sliding and colours indicating heating.

The new technology can be implemented in a way to be sufficiently stable and compact to be used on-site and it generates images of a size that can be transmitted electronically, should it be necessary to seek more expert advice from a remote specialist laboratory [6].

The V4L FilerPatchScan automatic particle sizing and counting software has also been developed for use with the new imaging technology [6]. This software is uniquely ‘plug-and-play’ and does not require the user to input subjective thresholds in order to distinguish particles from the background image. This makes it ideal for on-site use where the end user may not have the necessary skills or training to set image processing thresholds. The new particle sizing and counting hardware and software technology is compliant with the ISO 4406 and 4407, NAS 1638 standards and also the SAE ARP598 standard.

Figure 2. Magnetic Plug Debris at about x40 magnification, imaged by Macro-2-Micro unit through tape on a card.

Figure 3. Image-2-Information software indicates early stage of lubrication starvation.

From Images to Information

This new concept in wear debris particle analysis has been developed to specifically meet the needs of on-site technicians. The software is compliant with, and uses the particle classifications and nomenclature given in the ASTM D7684-11 Standard Guide for the microscopic characterization of particles from in-service lubricants [7].

The software provides the on-site maintenance professional with access to an expert- knowledge-base of the fundamentals of wear debris analysis in order to assist in the identification of transitions between benign, active and critical wear patterns. By interacting with the software the end user can access the following information:

• the wear debris mode to which a selected particle belongs
• the processes and conditions contributing to a particular wear mechanism
• information about equipmentspecific wear modes
• wear debris analysis using equipment specific baselines
• when and how to correlate the data from other cleanliness tests together with wear debris mode classification in order to identify transitions between normal, active and critical levels of wear
• an alert when equipment health is critical and the on-site professional needs to call for remote, expert support and back up.

Thus the software is ideally suited for onsite situations where the level of training and skill of the attendant technician may require substantial support. Figure 2 shows an image where an inexperienced technician might mistake the particle indicated for brass/copper. The software points out that because the surface is not a uniform colour, see Figure 3, this is in fact a heated particle indicating the early stages of lubrication starvation.

Conclusion

The fluid and particulate condition monitoring needs of the on-site maintenance professional differ significantly from the resources required by the lab based expert. The current work has identified new technology that caters to the need for portable equipment that is both easy to set up and use. Also identified are software resources that address the variable level of skill and training of on-site personnel.

»»References ›› [1] I SO 4407 – Determination of particulate contamination by the counting method using an optical microscope, 2002. ›› [2] I SO 11171 – Calibration of automatic particle counters for liquids. ›› [3] L eavers V. Technological advances in automatic particle counting, Practicing O il Analysis 2007. ›› [4] ASTM D7596 -11, Standard Test method for Automatic Particle Counting and Particle Shape Classification of O ils U sing a Direct I maging I ntegrated Tester. ›› [5] L eavers V. Macro to Micro: L ooking beyond the microscope for wear debris analysis. 22nd I nternational Congress of Condition Monitoring and Diagnostic Engineering management (COMADEM), San Sebastian, 2009. ›› [6] www.V4L-group.co.uk ›› [7] ASTM D7684 -11, Standard Guide for Microscopic Characterization of Particles from I n-Service L ubricants.