Maintenance

Health Maintenance of Rotating Machinery Using Wireless Mesh Networks Wireless Condition-Based Maintenance Gerry A. Nadler, Chief Scientist MachineTalker, Inc. 10 September 2007

Background When recorded over a period of time, the pattern of vibrations produced by operating machinery can be analyzed to detect changes that indicate that the equipment is due for maintenance and repair. Detecting and predicting the possibility of equipment failure allows personnel to fix or correct a problem before it occurs. This predictive maintenance strategy which saves resources until needed, is referred to as "Condition-Based Maintenance" or CBM. The goal of CBM is to perform maintenance only upon evidence of need. CBM necessitates the use of vibration sensors that gather the measurement data. The recordings are sometimes made manually with sensors and recording equipment carried from machine to machine on a periodic basis; or the data is taken from sensors affixed to the machinery and wired back to some collection point. It is in this process that the use of wireless techniques can produce significant savings. Introduction This white paper describes the MachineTalker® peer-to-peer wireless mesh network for monitoring the vibration in rotating machinery in an industrial environment. It can be used in both discrete manufacturing (aircraft, auto manufacturing) and process industries such as refineries and chemical plants. The document briefly discusses the economic issues and makes a business case for vibration monitoring. Following a brief overview of the basic concepts the paper will discuss wireless mesh networks as applied to monitoring in the most cost effective manner (as opposed to wirebased implementation). Business Case In general, vibration in rotating machinery is not beneficial. It can cause excessive wear, cracking, loosening of fasteners, excessive noise, fracture of solder joints in electrical machinery and a host of other problems. In the case of aircraft, serious vibration can cause catastrophic failure leading to loss of life. The goal of vibration monitoring is to detect vibration patterns which will lead to failure.

CBM WhitePaper 091007

1

In numerous studies maintenance typically represents 15 to 40 percent of the total cost of an installation. In some cases over a 20 year lifetime the maintenance can exceed the entire cost of the equipment. Let us assume that a company does $50 million in annual sales with a gross margin of 60% (very good). The cost of goods sold is 20 million so the gross profit is $30 million. If one assumes a 28% maintenance budget then the cost of maintenance is $14 million/year. If one can save 10% on maintenance this translates to $1.4 million directly to the bottom line. One would have to increase sales by almost 2.4 million to achieve this level of performance. Is this 10% savings achievable? The answer to this is an emphatic yes. This can be achieved by the installation of a Condition-Based Maintenance capability using wireless vibration monitoring to detect imminent failure in rotating machinery. Figure 1 shows the distribution of costs for rotating machinery maintenance in a petrochemical plant from 1973 to 1982. The actual dollar amounts are: Pumps $22,600,000 Compressors $6,950,000 Blowers $2,230,000 Turbines $240,000 (From Charles Jackson “Practical Vibration Primer”)

PUMPS 70.5%

$22,600,000

ROTATING EQUIPMENT COSTS AT PETROCHEMICAL OPERATION 1972 TO 1982

$240,000 $2,230,000 $6,950,000

TURBINES 1.0%

BLOWERS 6.9% COMPRESSORS 21.6%

Figure 1

CBM WhitePaper 091007

2

Why Do We Monitor Machine Vibration The overall goal for Condition-Based Maintenance is to monitor the health of a rotating machine. There are other parameters besides vibration which are useful in determining machine health. One can measure temperature, flow, and pressure or even do oil analysis to predict eventual failure. The best technique is to use vibration because it is the overall indicator of mechanical conditions and the earliest detector of developing defects. Vibration can detect the following mechanical conditions: Out of balance Misalignment (Bent Shaft) Damaged roller element bearings Damaged or worn gears Mechanical looseness Noise Cracking Basic Concepts of Vibration Measurement The standard technique for measuring vibration is to use a piezoelectric accelerometer which is attached to the rotating device at the appropriate point to measure the vibration. An accelerometer is a sensor that produces an electrical signal that is proportional to the acceleration of the vibrating component to which the accelerometer is attached. The acceleration parameter is a measure of how fast the velocity is changing (see Figure 2). Normally the acceleration input data is converted to a velocity waveform (see Figure 3). (For those mathematically inclined, acceleration is the first derivative of velocity). Spectrum Analysis If you examine the waveforms in Figures 2 and 3 which are based on the direct output of an accelerometer you will see that they seem to be complex waveforms. A French mathematician named Jean Baptiste Fourier (1768–1830) made a unique discovery involving complex waveforms. Fourier discovered that all complex waveforms such as the ones derived from vibration sensors can be broken down into a series of sine waves. The sine waves have different amplitudes and frequencies. In other words, complex waveforms are actually a superposition of much simpler sine waves. By using techniques initially developed by Fourier one can find the constituent frequencies that make up the complex waveform. As computers became more powerful and wide spread it has become feasible to apply Fourier methods using a mathematical algorithm called a Fast Fourier Transform (FFT) to change a timeseries waveform into a frequency spectrum (see Figure 4).

CBM WhitePaper 091007

3

By examining the frequency spectrum of the vibration signal one can examine the health of a vibrating machine. This is because certain frequency spectra may indicate abnormal operation or anomalies in the normal operation of rotating machine. It is often the case that since wire-based sensors are too expensive, due to the cost of installing wires from the sensor point to a control room, a portable vibration analyzer is used. As described earlier, maintenance personal periodically take the portable analyzer to the rotating machine to make vibration measurements. The analyzer has a powerful computer and does the FFT to determine the vibration spectrum as shown in Figure 4. As we will see, by using wireless vibration sensing the need for periodic visits by maintenance personnel is not required.

Figure 2 (Direct output of accelerometer)

Figure 3 (raw accelerometer data converted to velocity)

Figure 4 (Result of FFT (Fast Fourier Transform))

Overview of Peer to Peer Wireless Mesh Networks MachineTalker designs and manufactures peer-to-peer wireless mesh networks. Peer-to-peer mesh networks are very powerful because they are constructed from separate radio nodes that can “talk” and “communicate” among each other just as people can talk to each other in a room. These networks are self forming. They become operational just by turning on the power and the nodes discover each other and maintain contact with their neighbors. Another attribute is that when two radios are not in direct contact an intermediate radio can act as a relay. If a radio fails then an alternate path can be found using another radio. This provides for communication redundancy and the network is “self-healing” (see Figure 5).

CBM WhitePaper 091007

4

Figure 5

Each of the radio nodes can be connected to various sensors including vibration, temperature, pressure, intrusion, gas and other types of sensors. The mesh network can be connected to a backbone network or a single PC or laptop. One of the radio nodes acts as a gateway to the PC or other network attachments. Vibration Monitoring Using the MachineTalker Wireless Mesh Network One of the primary reasons for the low numbers of in situ vibration monitors is the cost of installation. The actual sensors are mounted on the rotating devices that are far away from control rooms or are difficult to install using wired networks. Hundreds of feet of conduit have to be installed and integrated at either remote or central control rooms. The cost of installation far exceeds the cost of the sensors and associated electronics and it can take weeks or months to install. A wireless implementation can be installed in a day or less, even in a hazardous environment. The MachineTalker Class 1 Div 2 wireless monitor can be mounted within a short distance of the sensor that is attached to the rotating machinery. The vibration sensor has to be mounted appropriately on the motor or gearbox, and most sensors require power to be turned on prior to measurement. The MachineTalker unit can be pre-programmed to turn on the power and then

CBM WhitePaper 091007

5

perform its data acquisition phase. There is no configuration required by the operator, as the unit is ready to go upon installation. Intelligent Data Analysis The MachineTalker wireless vibration analysis software package which runs on a PC does complete analysis of the vibration sensor data. It replaces and provides additional features not found on portable vibration analyzers. The software does an FFT as described above to provide a frequency spectrum. In addition it executes an intelligent analysis of the vibration data to look for anomalies in the data which may indicate eventual faults. It compares historical data and attempts to determine eventual failure modes. Predictive analysis at its best.

Contact Gerry Nadler at MachineTalker, Inc. for more information and a demonstration. [email protected] MachineTalker, Inc.

513 De La Vina Street, Santa Barbara, California 93103

Phone: (805) 957-1680

CBM WhitePaper 091007

♦

Fax: (805) 957-1740

♦

www.machinetalker.com

6