3D Optical Metrology Helps Prevent Leaks in Rotary Dynamic Seals
Advances in the design and materials used in rotating shafts and their corresponding seals have extended the life and reliability of vehicles and power equipment. To minimize fluid leakage and friction in rotary dynamic seals, however, these designs have a critical dependence on shaft surface texture and machine lead angle. Without proper control of these factors seals can prematurely fail.
Surface roughness, for instance, affects both the lifetime and effectiveness of a seal. The lip of a new seal makes contact with the shaft and abrades as the shaft rotates. If the shaft is too rough, the seal abrades quickly and will begin to leak. If the shaft is too smooth the seal will not bed correctly and will also leak. The shaft must be rough enough that the initial lip wear will allow a small film of fluid (typically between 1µm and 3µm thick) to enter the shaft/seal interface. When this occurs, the seal begins riding on a thin layer of liquid and further wear ceases. The fluid’s meniscus at the seal’s outside edge prevents subsequent fluid leakage.
Machine lead also affects the seal’s effectiveness. All machining and polishing processes for shafts leave some degree of residual grooving on the shaft’s surface. If these grooves are at an angle (called the machine lead, lead angle, or simply lead) to the shaft’s axis they can move the fluid within the shaft/seal interface. Angled one way they pull the fluid out past the seal, creating a slow leak. Angled the other way they push the fluid in the interface zone back into the housing, depleting the film layer and thus leading to excessive wear and early failure. Current standards for shafts and seals call for shaft leads less than 0.05° (see Table 1).
Table 1- Standards covering shaft and seal design and test call for lead angles less than 0.05° and roughness ranges that allow fluid films to develop.
The traditional measurement method for surface roughness uses a contact or stylus profilometer. This device rests a precision tip on the surface of the shaft and measures the tip’s vertical movement as it translates along the shaft parallel to the axis. This action yields a two-dimensional measurement of surface height along the line of translation. By combining the results of multiple measurements from different points along the shaft’s circumference the profilometer produces figures of merit describing the surface’s roughness. The parameter Ra, for instance, is the average magnitude of deviations from the mean profile height. Rz is an average value of the five greatest peak-to-valley distances in five samples, and Rpm is the average peak-to-mean height for the five largest peaks in five consecutive samples.
Traditionally, measuring lead angle requires looping a cotton quilting thread around the shaft, hanging a weight from one end, and measuring with calipers or a micrometer any translation of the thread along the shaft as the shaft rotates. To obtain an accurate measurement the thread must be allowed to move a significant distance, which may mean that the measurement includes regions outside of the critical lip position on the shaft. Many observers also report that this method exhibits a "dead zone" with leads between ±0.05°, resulting in this range often being categorized as zero lead.
3D optical metrology devices are able to measure both surface roughness and machine lead simultaneously. The measurement unit is essentially a modified microscope with an illumination beam that splits along two pathways (see Figure 1). Along one path the beam reflects off of a reference mirror. Along the other path, the beam reflects off the surface of the shaft. When the two beams recombine they interfere with one another, producing a pattern of light and dark bands that mimic the surface texture and can be captured by a CCD camera. These bands have their maximum contrast when the surface feature that created them is in focus. Thus, by vertically sweeping the microscope’s focal depth and recording where maximum contrast occurs for each point on the surface within the camera’s field of view, the instrument can build a full 3Dmap of the target surface.
Figure 1– A 3D optical metrology instrument uses interferometry to measure altitude at points all across a surface area to determine surface roughness and lead angle.
The added dimension that optical metrology produces helps provide additional insight into the shaft’s surface characteristics. The ISO has recently released a standard(ISO 25178) that defines 3D surface parameters and measurement methods. The nomenclature for these parameters – Sa, Sz, and so on – corresponds to the traditional 2D measurements but also helps describe any structure or pattern to the surface.
One way to detect such structure is to perform an autocorrelation function between each point on a surface map and the entire map. If the surface is isotropic, or essentially the same in all directions, the autocorrelation map will have a circular peak in its center. If there is any structure to the surface, such as a groove, autocorrelation will yield a peak that is elongated in the direction of the structure.
3D optical metrology instruments can determine the angle of lead grooves on a surface by performing a fast Fourier transform on the surface map to determine the surface’s angular power spectral density. Grooves will show as strong peaks (see Figure 2) at an angle corresponding to their pitch relative to the camera’s image grid. By fitting a surface curve model to the texture map, the instrument can determine the degree and orientation of any misalignment between the image grid and the shaft axis, allowing the instrument to then calculate the angle of those grooves relative to the shaft axis and thus provide an accurate measure of lead pitch.
Figure 2 – An FFT of the surface map, expressed in polar coordinates, can provide a measurement of lead angle.
Tests using Bruker’s (Tucson, AZ) NPFLEX-LA 3D Optical Microscope have demonstrated the accuracy and repeatability achievable using optical metrology. The measure of repeatability, for instance, involved performing measurement sequences 30 times without removing or replacing the shaft being tested. The 1σ standard deviation of lead angle results was less than 0.005°. The 1σ standard deviation for Sa was less than 1.4nm while for Sz and Spm it was 25nm and 18nm respectively.
To study system reproducibility, a single operator loaded and unloaded the test shaft after each group of 10 measurement sequences. When the starting location of these measurements was not controlled, representing an operator randomly approaching the system with a part to be tested, the 1σ standard deviation of lead angles across the runs was 0.02°. When the starting location was roughly controlled, such as when trying to measure only the critical seal lip contact region, standard deviations dropped to between 0.005° and 0.008°. The 1σ standard deviation of Sa, Sz, and Spm were 1.2nm, 76nm, and 38nm respectively.
Cross System Repeatability
To obtain the most accurate comparison of results across three different systems, the test first calibrated all systems against reference shaft samples with known lead angle samples. Then three different operators measured the leads and roughness on test shafts three times on each of the three different systems. For test shafts with a known 0.4° right-hand lead, the maximum deviation in results was 0.08°. For test shafts with known 0.4° left-hand lead, the maximum deviation was 0.04°. Sa agreed to within 30nm across all systems with a standard deviation of 7nm.
These results show that optical metrology is capable of achieving precision in lead measurement well within the tolerances required by the ISO and RMA standards. Optically obtained surface roughness measurements are an order of magnitude more accurate than the requirements of traditional methods. In addition, optical metrology avoids several of the drawbacks associated with traditional measurement techniques. For example, string-based lead measurement requires clamping the test shaft in a fixture that precludes the existence of non-cylindrical components, such as yolks and geared ends on the shafts. These components must be cut off prior to testing. Thus, the string-based lead measurement is a destructive test and does not allow the testing of parts that are to be put into actual operation.
Another drawback of string-based lead testing is that increasing lead means fewer shaft rotations for a given amount of translation and thus less resolution in the measurement. The accuracy of optical lead measurements, however, is independent of the lead angle. Further, optical metrology can concentrate its measurement in the critical area where the seal lip is to ride rather than taking measurements across a relatively long distance along the shaft, as is necessary with string-based lead and contact profilometer techniques.
The high precision, repeatability, and nondestructive nature of optical metrology ensure that the technique can handle the ever-increasing precision of rotating dynamic seal design in automotive and power equipment. This ability will help to reduce the incidence of premature seal failure, reducing maintenance and warranty costs. Further, 3D optical metrology’s ability to perform multiple tasks simultaneously helps speed the throughput of manufacturing test, reducing production costs.
About the author: Javier Vera is a Product Marketing Manager at Bruker, Nano Division, and has held various positions as product manager, metrology manager, and application engineer. Currently he has responsibilities as marketing manager for the Stylus and Optical Metrology Unit. Javier has more than two decades of experience in the field of industrial and optical metrology, during which he has written several articles and white papers on the subject.