by Steve Johnson
Computational Systems, Inc
Today’s cost-sensitive maintenance environment dictates an effective, simple-to-use, high payback technology where materials cost and personnel training are concerned. Ultrasonic monitoring is such a technology. Most plant systems and equipment generate operational noise of some sort and during failure modes the noise characteristics can change dramatically. Detecting this change and fixing the problem before failure can result in higher quality production output, reduced downtime, reduced maintenance overtime, and greatly reduced costs.
In its simplest form, noise is a vibration of molecules through a medium, like air, grease, or metal, which moves spherically outward from its source. Vibrations can be broken down to energy levels at discrete frequencies and when the human ear detects these levels and frequencies, it translates them into intensities and tones which are then transmitted to the brain. The human ear can detect frequencies between 20 Hz and 20 kHz. This range is referred to as the audible, or sonic, range. Frequencies above this range are referred to as ultrasonic. Ultrasonic instruments measure the ultrasonic frequencies and, through a process called heterodyning, translate the frequencies down into the audible range. Furthermore, the instruments will typically give a type of visual indication (a digital display or an analog meter) of the noise intensity either as some percentage of the instrument’s output or an actual calibrated sound measurement in dB.
As previously stated, what makes ultrasonic noise monitoring so useful, is the large number of systems and equipment which emit ultrasonic noise when problems develop or failures occur. The most effective industrial applications are detecting leaks in pressurized or vacuum systems, inspecting steam traps, mechanical analysis (especially bearings), inspecting valves, and electrical corona or discharge detection. Due to the nature of high-frequency waves, ultrasonic noise tends to be very localized and highly directional. These two aspects tend to make ultrasonic noise detection and analysis simple enough that, with little training, maintenance personnel can begin making big contributions to the bottom line. One final note: Always follow plant safety procedures when around operating equipment, pressure and steam lines, or electrical equipment.
Leaks in pressurized systems
By far, the most common usage for ultrasonic noise detection is to locate leaks in pressurized gas or vacuum systems. When a gas, such as air, escapes from high pressure to low pressure through an orifice, the flow becomes turbulent, generating ultrasonic noise. The noise is characteristically identified as a ‘rushing’ or ‘roaring’ sound with few discernible frequencies. General guidelines for finding leaks are as follows: first, set the volume level of the gun such that the normal background noise of the plant is near the bottom of the hearing threshold. Then pan the gun along the lines of the air/gas system to be checked. When the typical leak sounds are heard, slowly move the gun up, down, right, and left in the shape of a cross to determine which orientation ‘sounds’ the loudest. With the gun pointing in the loudest direction, begin moving toward the sound until the leaking component is found.
A common question being asked is, “Can the material cost of a leak be determined from the noise level of a leak?” There are many factors involved in this type of analysis, such as pressure difference, hole shape, hole size, whether the flow is choked or not, humidity, temperature, and instrument calibration. While a general relationship between noise and flow rate can be assumed based on empirical data, these numbers are mere guidelines at best. Further research is necessary to determine how the other factors will influence the calculations.
Steam Trap Analysis
All steam systems, no matter how well insulated, will lose some steam to condensation. The condensation must be removed or else it will puddle in the low points of the system. When enough condensate collects, it will begin to ripple into waves due to the steam passing rapidly above it, much like a lake ripples when the wind blows over it. As the rippling continues the size of the ‘waves’ will grow until the steam will actually lift the condensate, forming it into a type of plug, and push it through the pipe. This water plug will proceed to slam into the next bend, or worse, component (i.e. turbine, boiler, etc.) at a high velocity. This is called ‘water hammer’ and is, obviously, something to be avoided. Steam traps are placed periodically through the system to remove the condensate from steam systems. Steam traps are, in essence, ‘smart’ valves, whose purpose is to keep the steam in a system, but let all the condensate, air, and other gasses escape to return lines for reconditioning. Due to the complexities of steam systems, there is no quick formula for determining whether a steam trap is operating correctly or not. While there are some obvious tests which may be performed, no single method will work in every situation. Some of these tests are: listening for the trap discharge cycle, checking the temperature of the inlet and outlet, and comparing the inlet and outlet temperatures.
When a steam trap is sized correctly and operating correctly, it will discharge periodically and close periodically. If the trap is not discharging, then the trap is either failed open or failed closed. The trap discharge cycle can be monitored by listening to the outlet line. Generally speaking, during discharge the ultrasonic intensity level at the trap outlet should be higher than the inlet. However, even if the trap is heard to be discharging periodically, the outlet noise should be characterized to insure that it is indicative of turbulent flow noise and not the rushing sound of steam (this may require experience before reasonable accuracy can be expected).
Another obvious test is to measure the inlet and outlet temperatures and compare the results with some obvious conditions. For instance, if the temperature of the outlet line is substantially above 212 oF (100 oC), then there is a high probability that the trap is allowing steam to blow through and it may have failed open. If the inlet pipe has cooled to ambient temperature, then there is probably little if any flow through the trap. This may be caused by the trap failing in the closed state or by some other problem, such as an upstream valve being closed when it shouldn’t be.
Additionally, there should be a temperature difference between the inlet and the outlet lines of the trap since, when operating correctly, a mixture of steam and condensate enters the trap and only condensate leaves the trap. While each type and size of trap can be unique and should be monitored to determine its own temperature characteristics, a small temperature difference between the two locations is a good indicator that further checking should be performed.
One of the main sources of ultrasonic noise in equipment is friction between two objects moving relative to each other. This is most commonly seen in bearings. Whether due to improper lubrication conditions or mechanical defects, metal-to-metal contact will generate significant amounts of noise. Traditional vibration analysis has shown that most mechanical defects (pits, cracks, spalls, etc.) generate noise in the sonic range. However, ultrasonic guns which have frequency adjustment and are sensitive enough can detect such defects. To effectively monitor bearings, first establish a baseline dB level for the bearing when it is running correctly. Use another technology, such as vibration analysis, or compare this bearing with other bearings in similar applications to make sure that the bearing is indeed in a good state. Next, trend that value through subsequent readings and look for changes. There are some general guidelines for evaluating bearing conditions based on increases in the dB level, i.e. for a 6-9 dB increase check lubrication, a 10-12 dB may indicate the beginnings of a bearing fault.
A common cause of bearing faults is over- or under-lubrication. An under-lubricated bearing will allow metal-to-metal contact and wear, while an over-lubricated bearing will heat up and possibly pop the bearing seals. Under-lubricated bearings will sound like general noise (little or no impacting) and will generate noise around 25 kHz, though heavy loads on bearings may tend to lower this somewhat. To effectively lubricate a bearing, a baseline reading at 25kHz should be established. Then as further checking indicates a dB rise at 25kHz, monitor the bearing while lubricating. As lubrication is added, the noise levels should decrease. When the dB levels return close to the baseline, cease lubricating. Do not over-lubricate a bearing. One word of caution: It may take some time before lubrication spreads itself throughout the bearings inner surfaces, so proceed slowly.
Valves are inspected similarly to steam traps by checking sound intensity levels both upstream and downstream of the valve. If the sound level downstream of the valve is greater than upstream of the valve, then the valve is, depending on the type, probably partially open or maybe completely open. Sometimes if the sound level downstream of the valve is less than upstream (or very small) it means that the valve may be closed. However, there is not always true. Ultrasonic noise is generated when turbulent flow occurs, such as the back currents which form when liquid or gas flows through a partially open valve. Furthermore, the intensity level of the noise is directly proportional to the flow rate. If a valve is completely open and its configuration is such that it does not significantly interfere with the flow through the pipe or the flow rate is slow enough, then the flow will stay laminar and the downstream flow will generate little if any ultrasonic noise. To put it succinctly, hearing nothing below the valve does not necessarily mean that the valve is closed. To be more sure, use a sonic detector downstream of the valve (or maybe on the valve stem) because laminar flow will generate sonic noise. If listening with a sonic sensor and an ultrasonic sensor yields no sound downstream of a valve, then the valve is probably closed.
One final application for ultrasonic monitoring is the detection of arcing (electricity traveling through the air), corona (ionization of air around electrical conductors), or electrical insulation breakdown. Searching for electrical problems is similar to searching for gas leaks, in that, with the airborne sensor in, pan the gun around through electrical equipment and listen for electrical fault sounds like popping, buzzing, or crackling. Then move toward where the sound is the loudest. For situations where the source of the noise is out of reach and/or the exact location cannot be easily determined, most ultrasonic vendors have parabolic reflectors as accessories. A parabolic reflector will usually at least double the effective range of a standard airborne sensor and will greatly enhance the directionality.
Ultrasonic instruments have a wide range of effective applications and can serve very capably as a first line ‘defense’ against breakdowns. While no single technology can provide total solutions for all the possible maintenance issues which could arise, coupling ultrasonic monitoring with other technologies, such as vibration, oil, and thermographic analysis can greatly reduce equipment failures, reduce personnel overtime, reduce energy consumption, and improve system and product quality.
Steve Johnson is the Engineering Product Champion for Ultrasonics and Infrared Thermography at Computational Systems, Incorporated. He oversees research, current and new development, and customer applications for both product lines. Previously, at CSI he worked for four years on the development and support of MasterTrend. He received a BS and MS in Mechanical Engineering from the University of Tennessee, Knoxville.