A low-cost vibration analyser, and several other new tools , are proving particularly good at picking up bearing damage and helping identify other faults. Tom Shelley reports
A small, hand-held device can now determine from machine vibrations whether a bearing is damaged and in need of replacement, or whether the vibration comes from some other source – such as an out-of-balance load.
Similar smart instruments, incorporating expert knowledge, could be of equal benefit in many other kinds of predictive maintenance and environmental monitoring tasks.
“The problem with predictive maintenance is that the tools are particularly expensive and difficult to use,” says Russell Sion, managing director of C-Cubed, “so we saw the opportunity for a low-cost vibration analysis meter.”
Building on the development reported in Eureka’s March 2000 edition, concerning instrumented washers that pick up car engine detonations – and the development of a general sensor interface to work with pocket PCs and PDAs, revealed in October 2001 – the company has now brought out a ruggedised Pocket PC customised and dedicated to vibration analysis.
The Pocket VibrA Pro is designed, in Sion’s words, “to do a lot of what the consultant does – in his head – in the meter”. Noisy machines with problems produce vibrations at a range of frequencies.
“Bearing noise is very often swamped by other vibrations – out of balance and misalignment are the main ones,” he says.
This machine demodulates the crude noise signal, extracts the envelope and then separates the strengths of the main frequencies. It identifies which is which, on the basis of information input into an ‘Asset Wizard’, which includes industry-standard settings for motors, pumps and fans.
Regular diagnoses of machines are expedited with the help of an optional bar code scanner. A vinyl barcode is attached to each measurement point. The user then has only to scan the barcode, upon which the instrument automatically takes a reading, and logs this as relating to a particular point on a certain machine.
Output on the instrument is in the form of four bands of data, which might typically be identified as: ‘unbalanced’, ‘misaligned’, ‘bearing defect’ and ‘bearing damage’ - or the same information can be displayed as a frequency spectrum.
In a demonstration, Sion measured input from accelerometers placed on two machines, one of which had been set up to be slightly out of balance, but with good bearings, and another, which was balanced, but with a bearing known to be damaged.
Despite the noises made by the two machines sounding fairly similar to the ear, the instrument had no problem in correctly identifying the faults in each case and giving an indication of how serious each was.
The power of predictive maintenance, however, comes from recording trends, since worn bearings can run happily for years under the right circumstances. But, if they are deteriorating, then they will need to be replaced fairly quickly. Such trends can easily be ascertained by downloading data from the instrument on to a PC equipped with software bundled with the device.
The instrument is resistant to being dropped, water, humidity, and sand and dust, to the various requirements of MIL-STD-810F and IP67. Battery life is typically 8 to 20 hours. Input and dynamic range is ±50g to ±0.0004g. Alarms are user-settable on any vibration parameter, while the cost is £3,950, which seems very reasonable in view of the money it can save in the event of a machine-bearing failure.
Similar approaches could also be applied to a range of other predictive maintenance and monitoring tasks by attaching sensors for measuring, say, temperature or chemical pollutant concentrations and linking this to expert information built into the instrument.
In addition to C-Tech’s handheld device, a number of other techniques exist to check the condition of machines. Corus Northern Engineering Services (CNES) has used acoustic emission techniques to monitor the condition of hoist gearboxes in a steelworks.
The technique is ideal for monitoring slow-moving, high-value plant – especially where there are fluctuating loads, and where breakdown would have a significant effect on production. CNES first monitored the condition of vessel tilt bearings, then two 500-tonne overhead cranes that are driven through two epicyclic gearboxes.
Ian Taylor, business development engineer for plant condition monitoring at CNES said: “We positioned five acoustic sensors on the outside of each gearbox and four on the inside – as close to the internal bearings as possible.”
Once a month, an engineer performs a complete download of the acoustics emissions.
But condition monitoring stretches beyond that of just the bearings. One device – dubbed MCM (Motor Condition Monitor) by its developer, Artesis – keeps an eye on systems just by monitoring current and voltage.
“It allows us to monitor the condition of virtually any equipment driven by a three-phase electric motor,” says Artesis director Andy Bates.
The device is permanently connected to the electrical supply to the motor – and is typically installed the switch room rather than on the plant itself. It then builds up a ‘mathematical model’ of the relationship between motor current and voltage. When problems occur, there will be small, but detectable changes in the relation between the two signals. The device can analyse the changes and identify the likely cause – which could be in either the motor or in the driven equipment – and suggest remedial action.
“It effectively uses the motor as a sophisticated transducer to monitor the performance of the driven equipment,” says Bates.
He adds that MCM’s simplicity and cost-effectiveness opens up condition monitoring to wider fields of applications, and sees potential applications the building services industry.
Further up the chain – certainly in terms of scale – are industrial gas turbines. These are traditionally monitored by setting an upper limit for the exhaust gas temperature (EGT). The engine shuts off if the temperature is exceeded. But subtle changes in EGT – such as through a combustor running hotter or cooler than usual – may go undetected until a “gross error” occurs, says Alstom.
But Alstom’s Novus system uses a different type of analytical method to prevent this – by employing pattern recognition to detect ‘normal’ behaviour. The system first ‘learns’ what a normal range of EGT profiles looks like It then analyses EGT in a multivariate way, so can impose tighter threshold limits – allowing it to detect abnormalities “before they become significant”.
“Novus is able to identify the onset of faults even when all variables are within the normal working range,” says the company.
This material is protected by MA Business copyright
see Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the