Predictive Maintenance Through Targeted Data Analytics
Many industry reports state that data analytics for predictive maintenance has the potential to provide the highest return on investment for digitalization. I won't debate those reports and I actually agree with part of the premise, but mining value from industrial sensors using advanced data analytics is hard work. However, as an owner's engineer, my job is not to say no to hard challenges: it's to help customers understand the "How's" of designing and leveraging technology to maximize returns from their capital investments. One of the projects we're currently completing is to install an industrial-grade wireless network throughout a large power generating station.
See Enhancing Access and Connectivity to Harness Digital Transformation. Once this network is in place, the potential use cases are bountiful.
Here are a few of my favorite use cases where we think the ROI scales linearly.
Mobile Applications - If you've ever tried to use a smartphone or tablet in a large industrial environment, you know how difficult it is to send “receive/download” information to these devices. Having a Wi-Fi network opens the door to building and integrating custom applications that help employees do their work more productively. Also, mobile applications can help drive towards a paperless ecosystem that reduces costs and helps the environment.
Wireless Data Networks - I must be clear here. There are many wireless protocols used for the industrial internet of things (IIOT). In a future post, I'll explain the advantages and disadvantages of some of the most common protocols used in IIOT networks. I mentioned Wi-Fi earlier, and while that's not specifically what I'm speaking about here, it is an essential element of the data pipeline. Common industrial protocols are WirelessHART, ZigBee, ISA100.11a, Bluetooth, NFC, and RFID, to name a few.
Wireless data networks provide a low-cost option for sensor data such as temperature, pressure, flow, velocity, etc. to be retrieved without the need for costly cabling. The amount saved just in cabling cost, labor and waste is usually in the tens to hundreds of thousands of dollars on a single large project.
I like wireless sensor networks for another reason beyond cost savings, and that's savings of time. There is much less setup time during the initial install. Once the sensor is powered, configured, and establishes communication with the gateway, we can begin tuning or making adjustments. There are no loop resistance issues, termination errors to worry about, or future wiring damage problems about which to be concerned. Also, the money saved in cabling and construction can be used to improve the analytics, which is where the real longer-term value is gained.
Industrial wireless networks are not without downsides and cybersecurity is one of the biggest areas for concern. See an earlier blog post on securing your network perimeter. My advice is to create two networks, a wireless "controls network" and a wireless "analytics network." The wireless controls network is used for critical control data. This network will be enhanced with security, such as network perimeter protection, policy violation detection, logging, and incident recording/reporting. My advice is to keep this network small and simple to help minimize cyber risk and because of the compliance implications that most organizations will require, such as frequent auditing. The analytics networks does need to be cyber secure, but not to the degree of the controls network. Many of the wireless protocols, such as WirelessHART and ISA100.11a, include AES encryption and create a unique encryption key for each message. Even with all the built-in security, there is one important thing to know about wireless networks: they all can be subverted by denial of service (DoS) attacks. I highly recommend not adding any devices that are necessary for protection or emergencies to these networks. These types of devices should be hard-wired from field devices to the node controller or main controls processor.
Use our Digital Transformation Compass as a guide for addressing specific problems within your facility.
In future blog posts, we'll take a deeper dive into the intricacies of industrial wireless networks. This will help uncover how to begin constructing a low-cost architecture for enhanced data analytics in your predictive maintenance environment.