Getting Sensors to Do the Dirty Work in Industry 4.0

Industry 4.0, the fourth industrial revolution, combines the power of automated machinery with the power of the Internet.

High-performance processing and data storage has enabled the emergence of intelligent industrial equipment for roles such as assembly, packaging, and inspection; which is capable of monitoring multiple sensor channels and executing complex algorithms in real-time, such as high-speed multi-axis motor drives or precision positioning. These advances alone have empowered manufacturing businesses to make enormous gains in terms of capability, quality and productivity.  

Although on-board functionality has become extremely sophisticated, it is widely understood that machine sensor data has value extending far beyond basic monitoring and control. The power of The Cloud now provides the means to unlock this extra value. Big Data is coming to manufacturing, and is bringing unprecedented opportunities to improve business performance.

Big opportunities for big data

There are two major fields of opportunity. First, the power of The Cloud can be used in an essentially real-time role, to enable continuous machine optimization and dynamic maintenance scheduling. This can replace traditional maintenance at fixed intervals, thereby helping to reduce ownership costs and down time, and enable equipment vendors to adopt a more flexible “as a service” business model.

In the longer term, historical analysis can provide value-added services based on vast quantities of data, taking advantage of deep knowledge and powerful data analytics in The Cloud. Unlike a machine, or even enterprise-level computing, The Cloud can store data from trillions of samples captured over years, and run applications to identify connections and trends that can give businesses unprecedented insights into their operations to help optimize assets and processes.

Applications and services currently in the market include real-time and historical-analysis from a variety of providers, such as Aeris, GE, IBM, Microsoft and others. Manufacturers can use historical tools, for example, to search and learn from previous experiences with machine settings or processes that may have been acquired anywhere in the organization at any time in the preceding years. They can also compare the performance of individual assets across an entire fleet to aid optimization. In comparison, traditional mechanisms for sharing experiences between professionals, such as quality circles or application workshops, can be unreliable, time-consuming, and expensive to organize. In multinational organizations, language differences can present yet another barrier. Moreover, the Cloud-based service overcomes the challenges associated with centralizing and organizing the information for instantaneous access and use in the future.

Together, the on-board control and monitoring intelligence and the Cloud-based applications present a voracious demand for sensor data indicating machine health and process performance. Optical sensors, including infrared sensors that benefit from high noise immunity, are used extensively for monitoring the status and position of mechanisms such as tape feeders inserted in surface-mount feeders (Figure 1), for counting objects on a conveyor, or verifying contents of boxes or bottles in food-processing or packaging equipment.

Figure 1: Optical sensing to verify correct alignment.

Driving sensing into the toughest environments

In some industries, environmental hazards may impair the performance and accuracy of optical sensors. These can include cutting oils used in machining processes (Figure 2), or ethanol-based detergents dispersed by food-processing machinery. Rugged lens materials used in optical sensors such as the Panasonic CX-400 series provide resistance to such chemicals, thereby preventing degradation leading to inaccurate performance or sensor failure. The latest versions of these sensors also satisfy other environmental imperatives, with revised integrated processing that reduces sensor power consumption by up to 60% compared to previous generations. These sensors also have a strong infrared beam that can be adjusted for long-range use and is also able to penetrate materials for purposes such as detecting contents inside cardboard boxes as part of a packaging application.

Figure 2: CX-400 sensors have been tested with a variety of water-soluble and water-insoluble oils.

A similar type of infrared sensor is the Omron E3F/R series. Both types, CX-400 and E3F/R, have open-collector outputs capable of sinking or sourcing up to 100 mA, which gives equipment builders flexibility to connect the sensor to a device such as a PLC in a variety of ways such as directly to an analog input port or by interfacing multiple sensors to a digital bus.

Other robust and long-lasting sensors include Hall-effect rotary position sensors, which are less vulnerable to moisture, dirt, vibration, and shock than rotary potentiometers. Honeywell HRS series sensors are packaged in rotary-potentiometer form factors, thereby offering a drop-in conversion, and have a voltage output that can be connected directly to the control system.

Data to the cloud

The PLC or other machine-control intelligence responsible for capturing and processing sensor data typically has multiple analog and digital I/Os for connecting a wide variety of sensors and actuators. An Ethernet port is also common, which allows large numbers of devices such as sensors to be connected, over long distances if necessary, using a bus protocol such as Modbus/TCP or Profinet. A sensor-data acquisition module could comprise multiple sensors such as optical or rotary position sensors, with others like an Analog Devices ADIS16223 digital vibration sensor with built-in SPI host interface, to monitor a variety of machine-health parameters and feed data back to the main controller via a local microcontroller and Modbus interface such as the RAPID-NI-V2012 Modbus interface module. An array of such modules can be used for purposes such as multi-axis motor-health monitoring, connected to the main controller via a single bus interface, as Figure 3 shows.

Figure 3: Modbus provides efficient connectivity for multi-axis sense and control.

The PLC’s Ethernet connectivity also provides the connectivity needed for transferring the data sensed at various points on the machine into The Cloud. With the simple integration of a web or FTP server, this Ethernet port is the traditional point for remote access via the local area network or - via a gateway and with proper security provisions - across the Internet. In this way the machine can make its data available to an authenticated Cloud service. From this point, innovative software development taking advantage of the high-performance computing power of The Cloud, and the economies of scale offered by Cloud-service providers, take over the task of unlocking the value in the captured data through real-time and historical analytics.

Conclusion

Industry 4.0 will enable manufacturing businesses to derive the maximum advantages from the value in their own data. Pervasive and highly robust sensing will be key, collecting vast quantities of data from huge numbers of channels to feed Cloud-based applications designed to generate valuable business intelligence. This intelligence will be delivered in the form of real-time machine optimization and reduction of ownership costs, support for better strategic decision making based on historical trends analysis of a breadth and depth that is far beyond the capabilities of traditional human networking.