eSource 115 IoT and Predictive Maintenance

eSource 115 IoT and Predictive Maintenance

Posted 02 May 2019
Over the past few years, the hot topic at many maintenance and reliability conferences has been the use of IoT also known as ‘the Internet of Things’. When thinking of equipment, we often draw comparisons of equipment to the human body. In the broadest sense, IoT is the nervous system of equipment. Our bodies can sense of touch, movement, balance, smell, taste, sight, pressure, temperature, and auditory vibrations. The information is received by various nerves and transmitted to our brains for processing. How we respond to this data is based on the judgement of how we perceive the information.

The IoT mimics this type of sensory management. IoT is being used on equipment, vehicles, and components having outputs from sensors of temperature, pressure, conductivity, pH, vibration, sound, electrical imbalances, color, or light. The sensors provide electrical signals passed through a network. The network can be a closed, hardwired system, run on electrical transmissions such as Bluetooth, or via the internet. A continuous or scheduled series of data points are sent to the computer system which acts like the brain. This determines how the data is managed using pre-determined judgement parameters. The parameters are established through failure analysis. The application is merely an extension of what has been occurring ever since predictive maintenance was first used. 

Many years ago, a mechanic could listen to a machine and through the experience of witnessing failed machines of the past, could accurately predict the potential for failure. This worked on a very primitive scale, yet machines became more complicated. It wasn’t until computers were invented that data could be stored and calculations made. Predictive maintenance could now be done by taking large volumes of data and making decisions based on comparisons of what was perceived and what had occurred. The first example began in the 1940s during World War II.  Early computer scientists (Turing and Good) developed a decoding machine to understand encrypted messages. An encrypted message can have hundreds if not thousands of misleading data points. But by signaling out the common elements and discarding the others, a prediction could be made yet it required thousands of complicated equations.

Another example were the inventors of the first atomic bomb. Predictive techniques were used in computer simulations to predict an outcome based on previous experiments. The work carried out was limited to a series of very controlled experiments due to the cost and danger of what was trying to be carried out.

Today, lubricated machines can be assigned alarm limits based on levels on allowable or condemning levels of contamination, wear, or oil condition and remaining performance factors. Oil and grease samples can be taken and sent for analysis then compared to the process parameter history and maintenance records. Sensors can be installed inline to provide basic data points that may indicate further analysis is needed. Together, the data from the laboratory analysis, in-line sensing, and the actual maintenance work that is carried out can be compared and accurate assignments made. See how ALS can work with you to establish the data needed for your IoT.

Written By:

Michael D. Holloway, MLA I, MLA II, OMA1, MLT I, MLT II, CLS, LLA I
Principle Consultant, Certified Reliability Leader
ALS Tribology

 

Related Articles

X
alsglobal.com uses cookies to help improve your experience on our website. By continuing you consent for cookies to be used. More info
Confirm