Connected Plant

Don’t Fix That … Yet: How IoT-Enabled Sensors Predict Maintenance and Streamline Operations

The traditional approach to maintenance has two motors that keep it running. The first motor is proactive service: regularly maintaining equipment, following a schedule to ensure each element is running smoothly before repairs are needed in order to prevent a breakdown. The second motor is to repair once a breakdown happens.

While the first part of this approach is proactive, steady, and oftentimes common practice, it is probably not the best use of time. Repairing on a time-based schedule often means that equipment will be replaced or retired before it needs to be, effectively shortening its use and spending more than necessary. It is also a poor allocation of human resources, as the time spent repairing or replacing functioning equipment could be used for more urgent or necessary tasks.

The second part of the approach—waiting for a breakdown—means stopping production, waiting on repairs or parts to come in, and therefore reducing the runtime of the equipment and bringing work to a halt. So, what is the alternative to these two approaches? Predictive maintenance.

What Is Predictive Maintenance?

Predictive maintenance means making repairs based on the actual condition of the equipment, with the goal of preventing breakdowns and emergencies and, in turn, extending a machine’s lifetime. Often built on a network of Internet of Things (IoT) sensors, data is gathered, analyzed, and reported to support informed decision-making on when to repair or service machinery.

Because it’s built on IoT sensors, data is collected remotely and in real time. With this frequent flow of real-time data, indications of malfunctioning equipment can be identified from across the room, or without even being physically in the plant. In fact, these indications are tracked continuously and can automatically alert plant personnel of unwanted changes.

Collected data varies depending on the type of equipment, but typically includes information like temperature, power, and sounds that indicate vibration or wear. This data can be connected through a number of connectivity options. One of the most common connectivity options for IoT-connected equipment is LoRa, Semtech’s long-range wireless technology. LoRa is a widely-adopted long-range, low-power, and low-cost solution. LoRa devices are also very low bandwidth, meaning they have the ability to penetrate dense materials and transmit signals more than 30 miles away in rural areas. They’re also incredibly low power, lasting up to 20 years in some use cases.

Alternative wireless connectivity technologies like Bluetooth or Wi-Fi have their benefits as well. Where LoRa is low bandwidth, Wi-Fi and cellular are high bandwidth, providing the ability to transmit larger payloads like video or audio files (Figure 1). Wi-Fi tends to be best suited for consumer IoT applications, where there are fewer points of connection and a smaller range needed to communicate across. The low bandwidth of LoRa is more effective in industrial applications, since small payloads mean servers using LoRaWAN connectivity can handle a high capacity of messages at once—millions per single gateway using LoRaWAN.

LoRa-bandwidth-vs-Wi-Fi-cellular
1. LoRa technology is a low-bandwidth, long-range solution; whereas, cellular and Wi-Fi technology are high-bandwidth technologies. Courtesy: Semtech

To provide continuous real-time monitoring, LoRa-enabled sensors gather equipment status data and either periodically send this data by secured uplinks to the gateway using LoRaWAN or interpret values and send uplinks only if a value is exceeding a specific threshold or immediately if an alarm has occurred. From there, the gateway forwards the payload of uplinks to the associated network server utilizing LoRaWAN to be decrypted and analyzed by an application server. Finally, this server sends updates to the facility manager via their mobile device or computer, alerting the manager to upcoming or necessary maintenance (Figure 2).

predictive-maintenance-with-LoRaWAN
2. Data flows from sensors to gateways to network servers, where the data is analyzed and alerts are sent, when necessary, to the facility manager’s computer or mobile device. Courtesy: Semtech

Predictive Maintenance in Action

A great example of this process in action is how wastewater treatment facilities responded to a change in consumer behavior at the start of the pandemic. From disinfectant wipes and paper towels to napkins, sewage systems saw an influx in clogs, an unintended consequence of an increased focus on hygiene. According to the California State Water Board, wipes are among the leading causes of sewer system backups.

Using a predictive maintenance solution by myDevices, facilities were able to gather equipment performance data on sewage pumps. By placing the sensor on the equipment, it learned and analyzed vibrational patterns—and then sent alerts when it detected anomalies.

This approach of using IoT sensors to detect failures and secure a better uptime of utility infrastructure also works outside of the plant. Smart energy grid manufacturer and solution provider CAHORS integrated the LoRa devices into the line default detector “Sentinel” for monitoring voltage of power grid networks (Figure 3). The Sentinel sensors transmit accurate, up-to-date data on grid functionality, enabling the detection and location of line faults to improve efficiency and prevent failures. Grid managers receive an alert in almost real time that a repair is needed, and where they must deploy a crew to repair the failure.

LoRaWAN-fault-detection
3. LoRa sensors can be used to monitor for line faults in power grid networks. Courtesy: Semtech

Future-Proofing the Connected Plant

Setting up an IoT network for predictive maintenance doesn’t just start and end there. Building on an open protocol means that the system has interoperability among other existing and future applications. Once it’s established, additional sensors can be added, opening up opportunities for nearly endless applications, like asset tracking and geolocation without the need for GPS or additional power consumption.

Moving from preventative to predictive maintenance requires a shift in mindset. It’s moving from proactive, schedule-based service to repairs based on historical information and data-driven decisions. But once it’s in motion, an IoT-connected predictive maintenance approach reduces waste, improves workplace safety, and prevents expensive repairs, streamlining operations for years to come.

Rémi Demerlé is marketing director for LoRa Developer Ecosystem, WSP LoRa Alliance and Community at Semtech, plus chairman of the Smart Utilities workgroup in LoRa Alliance and member of DLMS User Association.

SHARE this article