Data collection and analysis can be used in a variety of ways, but for this column I would like to focus on maintenance. Though maintenance may not seem as exciting as all the other production improvement possibilities, I find maintenance to be an often overlooked aspect of manufacturing which can lead to costly, unscheduled downtime and quickly swamp any incremental gains in production improvements.
Maintenance programs tend to fall into one of three categories:
Fix It When It Breaks. This barely qualifies as a program, but there are many folks out there who operate in this fashion. It represents a very expensive way to run a manufacturing facility, as virtually every maintenance event results in that undesirable, expensive unscheduled downtime.
Periodic Maintenance. This is the most common maintenance program and consists of periodically servicing equipment based on an educated guess as to how long it takes things to wear. This period is generally settled on over time, meaning you have to be burned a few times by unscheduled downtime before you are able to settle on a period that takes into account all the failure modes of the piece of equipment. The problem with periodic maintenance is that, to be effective, the period is based on the worst-case scenario and you end up over-maintaining your equipment most of the time.
Preventive Maintenance. This is becoming more common. We are hearing it being considered mosy often in applications where the control system is used to monitor events in the system (e.g., motor runtime, counting strokes on a cylinder, or cycle counts on contactors), which is then cross-checked to manufacturers’ recommended maintenance schedules which triggers maintenance for that piece of equipment. This represents a great maintenance program, but I would suggest it still leaves room for improvement—as manufacturers’ specifications are generally conservative and do not always take into account the application of the equipment.
Looking at these three common practices, there are some very obvious holes ... which means there is significant opportunity for uptime and cost improvements with a better way. The better way is Proactive Maintenance. The idea here is to add sensors and technology to equipment to predict problems and trigger maintenance prior to any real trouble. Advances in sensor, networking and processing technology has made this possible. There is a sensor out there for most every application, networking them is a breeze, and memory is cheap, so storing the data is a straightforward proposition. It then becomes a matter of building the monitoring/notification system, which I would suggest could be handled by any reputable automation solution provider.
Some straightforward examples of this type of maintenance data collection include: vibration monitors on bearing sets, monitoring motor temperature/current, pressure monitoring across filters, and tension monitoring on chains/conveyors. With this data, it becomes a matter of establishing some set-points/tolerances to issue notifications to the proper personnel. In certain applications, more advanced monitoring could be required that would drive the need for a statistical process control (SPC) system or other advanced algorithms to predict failure is looming. Either way, with proper design and consideration, criteria can be established for any piece of equipment to ensure maintenance is happening at the right time, rather than being based on an arbitrary time/cycle count.
Now to take things a little further, with the “Internet of Things” concept, notifying the proper personnel can be more than a pop-up window on your maintenance supervisor’s desk. What about automatically generating a purchase order to your local vendor for the parts that are going to be required for the work? How about an order automatically emailed to your local service provider to schedule the work? This represents a significant deviation from the standard way of doing business, but the technology exists to handle maintenance and service in this fashion, essentially connecting the data on the plant floor to the people that can take action on it.
With the right information, well ahead of a failure, you can make better decisions on how to handle an equipment problem rather than reacting when the pressure is on. Like I said, there is informational gold in all that data, it is time to start digging and collecting the data.