Video Vibration Limitations To Be Aware Of
By Bill Marscher
A great new approach has arrived on the machinery diagnostics scene, based on using the pixels of a high speed video camera to act as thousands of simultaneous non-contact vibration sensors. Our company’s technique called VibVue™ gets very high resolution in the video picture, and can reliably detect and measure vibration displacements down to about 100 millionths of an inch (2.5 microns) p-p, while looking at an entire 10 foot (3 m) length of a feed pump and DE of its driver. VibVue™ uses motion magnification (also called “video vibration amplification”) to magnify this same level of infinitesimal motion to observable levels to facilitate comprehensive troubleshooting.
But we are the first to admit video vibration and motion magnification are not without limitations, just like any other sensor technology. Based on our experience, there are a few applications where the technology doesn’t shine. And, we use VibVue™ as an effective tool with our accelerometers, not always in place of them.
Higher Frequency, Lower Velocity
100 millionths of an inch sounds pretty good, and it is. But allowable vibration specs for machinery are generally in terms of velocity, not displacement. This can be a problem for measuring techniques (like video vibration) that depend on light modulation, which directly detects displacement, not velocity.
Let’s say the vibration is occurring at 1000 Hz, which is at the upper end of the API or ISO vibration measurement frequency measurement, and in the range of where blade passing or piping acoustic-based vibrations are common. For 100 millionths of an inch p-p, what kind of velocity are we talking about? Plug this into an on-line vibration calculator and the result is 0.22 in/sec rms (5.5 mm/s rms)! Whatever spec you are using,- ISO, API, HI- this “barely detectable” result is likely above what that spec requires. For example, API-610 for pumps requires under 0.12 in/sec rms (3.05 mm/sec rms), and ISO 10816, depending upon type and class of machine, requires about 4.5 mm/s rms (0.18 in/sec rms). The good news is that at lower frequencies, the velocity able to be detected gets progressively better, so that at 100 Hz, 100 millionths of an inch p-p permits detection of nearly 1/10th of the ISO limit.
Shrinking the Field-of-View Only to Do What Accelerometers Do
Unfortunately, some video vibration detection systems are way less sensitive than 100 millionths of an inch, or in some cases even 250 millionth’s of an inch, for a ten foot field-of-view. So, does this make the system worthless? No, but it does limit where and how it should be applied. For example, using a lens or adjusting the distance to the target to focus the camera’s pixels on a smaller field-of-view (like the bearing housing alone, as an interesting item) allows detection of vibration levels of proportionally smaller values. For example, for a video vibration system marginally capable of 250 millionths of an inch, if you set the field-of-view to 1 foot (300 mm), you can marginally detect 0.055 in/sec rms (1.41 mm/s rms). This is about half the API acceptance level. But at what price? If I am just going to focus on a bearing housing, why do I need a video system? Why not use a quick and simple accelerometer paired with a walk-around analyzer?
The justification for using video vibration magnification/ amplification technology is that, sure, it can detect the vibration (Job One!), but it can put the component’s vibration in the context of the movement of any attached or nearby components. This includes near-field (like the bearing housing) as well as far-field components (like the casing and pedestal, and maybe piping strung up and out of reach). Being able to see what is moving and what isn’t is extremely valuable to condition monitoring personnel as well as equipment troubleshooters. Vibration Motion Magnification is VERY useful in this circumstance, BUT … only if it can detect ISO-interesting vibration levels across a large field-of-view. Looking at 10 inches (25 mm) of an elephant’s flank through a telescope doesn’t give you much useful information about the elephant!
Other Factors to Be Aware Of
Another issue: a video system’s sensitivity to picking up small displacements depends on
- the camera resolution and sensor quality (no, your iPhone is not the same as a high quality machine vision camera, no matter how many pixels it has),
- the specific mathematical/statistical technique used to translate light intensity variation into vibration displacement versus time or frequency, and
- the lighting intensity, and contrast within the surface and against the background.
By purchasing a system that has a superior camera, and superior magnification algorithms (e.g. check out the excellent algorithms of Wu or Rubenstein of MIT), you can achieve sufficient sensitivity, and sub-ISO vibration levels can be detected, even for a large field-of-view. However, if you do not have good lighting on the equipment you wish to observe (and if you cannot add effective temporary lighting) then video vibration amplification will probably produce disappointing results. Also note that the higher in frequency you go, the faster the frame rate, and consequently the sensor has less time exposed to the light, requiring more of it.
Besides detection, resolution (pixel density) is important if video display of vibration is going to be effective. Interpolation of one sort or another is used to create a magnified vibration scene across the camera’s full field-of-view, and to provide a displacement frequency spectrum at (in some systems) a large number of locations, and in others, very limited locations. We prefer to work at the sub-pixel level, for minimal interpolation bias. If the vibration magnification or amplification system you have purchased only includes limited locations/ pixels that are used to create the video, the result could easily be misleading, and likely not very useful. In such a case, performing an Operating Deflection Shape (ODS) is a better approach, even though it is time-consuming and tedious if enough points are employed to catch most issues of interest (like soft foot, for example).
Right Tool for the Job
All that said, video vibration and motion magnification are still powerful tools that we recommend for your troubleshooting tool set. And, we think they work well when combined with other sensors (example) to help provide a comprehensive understanding of what’s going on. Let us know if you’d like to learn more about our VibVue™ solution.