System architecture, remote monitoring and data-driven maintenance

Factory assets require some type of maintenance to keep machinery in operation and avoid downtime. Manual walkthroughs to gather data on equipment health is a time-tested method for predictive analytics. However, on-machine sensing technology with remote connectivity can augment a maintenance technician’s ability to determine machine health.

Deciding which sensors to install and how to communicate or interface with them are determinations that could require collaboration with the IT department.

If instruments do not need to interface with the programmable logic controller (PLC), then a web server may be used. Then IT can feed those alarms to the maintenance people via a predictive maintenance application. Say there is a need for both: then feed the instruments into the PLC system via a remote I/O situation and read the data via a Node Red interface or using an OPC server and provide input to a reliability screen that can notify mechanics via telephone or mobile application.

The capabilities of such a system allow real-time alerts, trend analysis, asset health scoring and remote access to machine parameters. The idea would be to make the mechanic’s job easier, so that if he is on a troubleshooting call, and there is a parameter that keeps dropping out of specification, then the machine is not dependent on manpower to walk a route. The data is sent to the operator, the maintenance tech for the area and the maintenance manager.

This way, there is definitive data that there are recurring high temps or high current or blocked filters. The manager can use the data to plan a down or to send a maintenance person on a set route to investigate, after the maintenance person is done on the troubleshooting call. At least that is one way to use this type of system locally. What about remote monitoring via cloud technologies?

Where can remote monitoring systems be used? Water and wastewater use vibration and flow anomalies with supervisory control and data acquisition (SCADA) access to trend water problems at unmanned plants. Oil and gas operators have pipeline pressure trending, as well as leak detection and compressor health, to monitor remote assets. Manufacturing uses vibration analysis to know when to rebuild spindles and motors for CNC or large presses. Power-generation units monitor turbine conditions with remote sensing. This is also prevalent in the fuel cell industry for monitoring fuel cells that are stacked for use in power applications at buildings and or microgrid-type functions.

On-machine monitoring may include individual components, group components and infrastructure components. In most cases, a software package may be interfaced, or the current PLC, SCADA and HMI interfaces will be utilized. The local engineer can purchase components and install feedback locally. This is dependent on the end goal of using remote monitoring and whether there are one-off situations, or if there are bulk failures due to lack of manning to monitor equipment.

The positive response to cloud-based remote monitoring is that a system can tell the maintenance manager that there is a problem without a person having to walk it down. The negative response to cloud-based remote monitoring is that the maintenance manager still needs qualified personnel to investigate in person and alleviate the cause of the readings trending high on the remote indications.

Get your subscription to Control Design’s daily newsletter.

Industrial automation companies are creating value streams from selling packaged remote monitoring solutions, and others are paying for engineers at remote operations centers to assist companies in understanding electromechanical failures while acknowledging that there is a reduction in personnel.

This means an end user could pay for a system to be put in that allows real-time asset monitoring health dashboards, remote troubleshooting, performance optimization, cross-site benchmarking and the integration of Pi historians with analytics tools. Next step? Artificial-intelligence (AI) analysis and down planning based on machine input.

There is hardware available for remote monitoring without PLCs that can interface with AI applications. Transformers can be monitored without a PLC using remote inputs, Node Red and an MQTT broker. If it’s necessary to monitor the raw data, then log on with the web server. Requests for data can be managed with local user permissions or via a lightweight directory access protocol (LDAP) server.

Some automation suppliers also partnered for remote monitoring an unmanned plant, which ran 35 consecutive days without personnel. The main objectives of the initial study were to show that they could use IIoT technology and edge computers to improve productivity, maintain yield and quality, save energy and respond to sudden disturbances. Cloud-based monitoring can reduce operational costs based on reduction of down time. The response time to a failure is quicker. From observations, oil and gas, chemical and pharmaceutical industries have the funds to install remote monitoring and to alleviate labor costs. Other industries that have not spent the time to modernize may be forced to, not because of labor reduction, but because of the decline in the number of people becoming maintenance technicians. Regardless, the use of remote monitoring for early detection is only one part of the solution.

Machine-mounted sensors can be used to send data to alert the maintenance manager that there’s not enough manpower to fix all the problems, but that is not going to win over the maintenance manager. However, new technology might make people interested in becoming mechanics, reduce unplanned down time and improve yields, quality and energy. As a minimum, the new control systems engineer can learn how to utilize I/O, update the PLC, add data screens to the HMI and create a website to show trend data that will show why the plant is experiencing an uptick in motor failures.