I hate it when I get home after a long day of work and turn on my TV only to have the bulb go out. I know the TV has a sense of bulb life since it tells me hours remaining when I install a new one, but it does not let me know when I’m nearing end of life. The TV is older, but it is still a bit of an annoyance.
However, my TV bulb going out while disappointing is a minor annoyance compared to having a jet engine fail midflight or other critical operational assets fail without warning. Operational equipment requires maintenance to maximize lifespan and efficiency which have obvious impacts to the bottom line. These impacts are multiplied exponentially when injury or loss of life is a potential outcome from an asset’s failure.
The field of predictive analytics is not new but is undergoing rapid change due to the advent of more advanced internet enabled monitoring devices and big data solutions. Manufacturers of equipment also publish evolving guidelines for maintenance schedules under various operating conditions for equipment.
To use a car analogy, no longer do you need to just blindly add oil every several thousand miles. Now, you can wait for the car to tell you it is thirsty for oil. Oh, by the way, how about the air filter too. Take that analogy forward to a turbine driving a natural gas power plant. Given correlations between failure rates at various operational temperatures, excessive vibrations, and amount of continuous uptime an organization could predict an upcoming failure and take a unit offline for preventative maintenance saving big bucks in asset repair, future lost revenue, and even potentially danger to operational employees.
Many organizations want or have this data but still struggle to perform the predictive analytics necessary to do advanced predictive maintenance. In my mind, there is a ‘simple’ process to follow:
- Capture data leveraging your connected monitoring devices
- Tie asset management system and operational monitoring systems together for analytics
- Ask questions to identify root cause failure correlations
- Add root causes to your monitoring logic to alarm operators before failures occur
In the meantime, I need to wash my car right after I switch out that TV light bulb.