For far too often maintenance has been treated as expendable. Now, however, the internet of things (IoT) has brought along technology in a useful package to help maintain factories.

New sensor technology can measure pressure, distance, temperature, and much more in a smaller package and from greater distances. No longer are you required to modify a machine or tool. You can now use external laser technology to measure quality and keep track of machine maintenance. If you want to know if the machine is deviating outside of its temperature range, you’ll know. If you need to track power usage, you can even see the readout from a beach in Florida.

Sensors can directly connect to a hub that transmits information to a program, compiles the data, compares it, and then uploads the important results. Sensors can connect directly, but that’s a lot of internet or intranet connections. Hubs compile and report on the devices.

Think about your home and your stove, and a boiling-over pot of macaroni your child started before taking that text message. Imagine sensors determining spillage on your cooktop and automatically shutting off the burner.

This is the first stage of smart technology: self-analysis. You could have a sensor connect to a monitor. That monitor checks for spillage and when it sees it, sends you a text message. You then go online, check into your home network and turn off the stove. This is informational and provides data. You have access to login and make changes; however, don’t you want the device to act on its own in this scenario?

This is what we are talking about today. Open loop means you said something should happen, and it happens, but you’re not sure it happened and you have no feedback. Closed loop gets that feedback to confirm what you expected to happen did happen. Smart technology acts upon this information.

These loops are ever-expanding. Closed loop at sensor level, closed loop at machine level, closed loop at network level. If something is going to be safety- or crash-related, it needs to be as instantaneous as possible. Lower-priority decisions can wait for human intervention.

An important step in all of this technology is closing the loop. Let’s say a machine like a stamping press produces car parts. As the tool gets dull, the tonnage goes up. At a predetermined limit, the press will stop with a warning. It will also alert the tool room to do maintenance on the tool and tell the production team that they need to address the shortage of parts.

Predictive maintenance, though, is the future and the now. Instead of just telling you after it happens (reactive), let’s say it informs you hours, days, weeks or even months in advance.

Unplanned downtime costs far more than planned maintenance. You have the cost of lost work, moving staff, priority shipments and panicked purchasing agents, who are all no longer doing what they were scheduled to do and are now reacting to a problem.

Predictive maintenance looks at the predetermined expectations of failure and starts to learn on the fly. Using smart technology and sophisticated calculations, you are now no longer locked into planned maintenance far before failure or a reactive downtime condition. Now, machines and software link together to design advanced algorithms that can assist you with better information. Smarter machines, smarter calculations, smarter decisions.

Rarely in maintenance do you see a linear wear pattern. Mechanical brakes, hydraulic valves and electronic sensors all have a curve that accelerated over time until failure. This unpredictable nature is what computer systems and smart controls attempt to replicate.

Systems start off with linear parameters, time, distance measurements, temperature readings. Over a short period of time, these systems start to educate themselves as predictable wear patterns start to occur. The systems take things like time to failure rates, temperature changes into account, and monitor the deviation patterns that are occurring.

From there, the machines, controls and data gathering provide information. The computer modeling and software start to determine failure dates based on current information, past history and future planned production.

Imagine sitting at your desk and getting a list with the expected dates of failure for tooling, equipment, materials and even getting these dates far enough ahead of time that you can plan for long lead-time items that can crush a manufacturing deadline.

Information is getting smarter, equipment and controls are getting smarter, and computer systems are getting smarter. Shouldn’t we get smarter? Now, if only I could predict when Bobby the operator is going to take a long weekend.