The Sensors Revolution
Published on : Wednesday 01-09-2021
Connect, Control and Monitor. How modern sensors are contributing to next-generation industrial automation.
Sensors provide critical information about various parameters such as pressure, temperature, flow, gas, position and humidity, etc., and help monitor processes, enabling automation. With the evolution of the internet and its application in industry, sensors have become even more important in taking automation to the next level. With the evolution of IIoT, sensors are integrated throughout manufacturing and industrial processes and used extensively for monitoring and detection of anomalies, leading to not just preventive but predictive maintenance. All this results in all round savings in terms of labour, materials and overall production costs, adding to the bottom line. According to a recent market report, the global industrial sensors market is expected to reach USD 35.47 billion by 2027 from USD 18.81 billion in 2019, growing at a CAGR of 9.77% from 2020 to 2027.
The role of sensors
So what are the roles sensors are performing in the field of industrial automation today? “Sensors are one of the essential elements in the smart manufacturing chain rendering intelligence in true sense. They are the devices enabling detection, analysation, measurement of varied parameters and changes on the production sites like length, height, position, appearance, temperature thus helping a lot in creation of a sturdier predictive maintenance solution,” says Sameer Gandhi, Managing Director, OMRON Automation India. Omron has a wide range of sensors in its portfolio: vision, photoelectric, fibre, displacement/measurement, contact, pressure, ultrasonic, condition monitoring, OCRs, and rotary encoders, to name just a few.
“Sensors play a very important role to make the products intelligent and highly automatic. Sensors allow one to detect, analyse, measure and process changes with physical effect, providing this information to the control system. Sensors are input devices which provide an output signal with respect to a specific physical quantity input,” explains Dr Makarand Sawant, IT Professional, Technology and Business Visionary, who has led Innovation and alignment of IT with business by driving Digital Transformation through Analytics, Process Automation, Mobility, and Cloud including leading Smart Factory, IoT and Industry 4.0 initiatives.
“Connect, Control and Monitor. This is how modern sensors are contributing to next-generation industrial automation,” states Anand Gijare, Founder and Principal Analyst, Sensors Research and Publications, who is the Founder and Principal Analyst at Sensors Research and Publications; a dedicated market research service provider for wireless sensors and sensor-based applications in healthcare, industry, and defence. According to him, modern sensors are truly providing a real meaning to automation. These sensors provide real time operational control and with minimised human interference. Smart sensors make machinery, tools and other devices learn, function and operate by themselves. Talking of smart sensors, Anand maintains these are not just providing automation of the machines or tools but provide control and management of information systems for the entire factory infrastructure. “Modern sensors contribute by continuously providing machine generated data and patterns. This data is processed and analysed using modern machine learning technologies, and data analytics for insights, intelligence and actions,” he elaborates.
According to Harish G Kashyap, Principal Oil & Gas Consultant & Digital Transformation Enthusiast, sensors are the eyes and ears of any industrial automation setup be it process plant automation or manufacturing automation. They are primarily used to detect the input and transmit an output electrical signal. This input could be further processed using the emerging technologies Cloud, Fog or Edge computing as per business needs and then an appropriate output signal be transmitted. They basically serve any one of the 3 outcomes – Insight, Foresight or Aid in better and informed decision making. “In today’s unprecedented VUCA landscape, for sustainable, efficient and competitive industrial operations it’s imperative to understand, believe and realise that sensor technology plays an active role in industrial automation. From simple detection to real time process monitoring to asset condition monitoring or overall asset performance management, sensors today serve an exhaustive list of industrial automation needs. They also aid in continuous measurement/monitoring and precision control in safety and mission critical industries such as process plants, especially, oil & gas, refineries, specialty chemicals, petrochemicals, etc.,” says Harish, who has more than 15 years of experience in the Oil & Gas domain, delivering multiple large scale, high budget capital intensive projects.
“Sensors are the key elements used in almost every intelligent device in industrial automation. The ability to gather critical field device information has allowed sensors to be deployed to simplify and automate industrial processes in multiple ways. Diagnosing the assets’ health status by means of signal noise to prevent failures, triggering alarms for functional safety, etc., are some of the important operational aspects taken by the sensors. Beginning with condition-based monitoring and power management, to image sensing and environment monitoring, the list is just endless,” Jegajith P T, Embedded Practice Head, Utthunga, who is also the company’s R&D manager of product development for various electronics and embedded systems. Throughout his career spanning 14+ years, Jegajith has led and coordinated teams of embedded system professionals in the development of software, firmware and integrated embedded systems and IoT devices.
Trade-off in low cost sensor
Low cost sensors are becoming popular in the IIoT era for many applications in areas like logistics, transportation, pollution control, etc., as many enterprises find affordable options in these solutions. But are there any risks involved in the use of such low-cost sensors? Is there a trade-off here and what are the implications?
“Innovative IoT solutions based on these low-cost sensors are making it possible for various companies to get affordable options. Low-cost sensors have acted as the catalysers for transformation. One of the major obstacles in adapting to cutting edge solutions empowered by IoT technology is the deployment of the needed sensors. Using the most sophisticated sensors for every type is not possible as it goes far beyond the revenue generated,” asserts Dr Makarand Sawant.
Sameer Gandhi shares the view that sensors are used for many critical functions on a production line and they are like the 'eyes' and 'ears' of the control systems. “Any single sensor failure can bring the production line to an abrupt halt. Also, proper functioning of the sensors may also be critical to the safety of workers on the line. Hence, while adopting 'low cost' sensors, one needs to bear in mind that quality and reliability should not be compromised,” he opines.
“Increasing use of sensors has brought down their costs from about $1.5 per unit to about $035 per unit and some come even cheaper, which are use-and-throw sensors that are used in RFID tags and beacons. Smart sensors are easy to install, calibrate and need no or nominal wiring and hardware becomes very cost effective,” says Anand Gijare, who also feels that sensors are becoming cheaper in price but if they are of cheap quality then it is a matter of concern. “Microchips used in the sensors are the heart and brain of that sensor and it is very important to see what quality of microchip is used in a sensor. Again, there are multi-functional sensors that can bring down the workloads,” he adds.
But there are other perspectives as expressed by Harish G Kashyap, who says nowadays we could get a sensor for just $5-$6. Is this cheap? Or rather is this cost effective? “Well it depends on the business and scale/volume of application. If you plan to produce 20 million products with $5 sensors each then the sensor budget would mount to or even exceed a whopping $100 mn,” he says. “The pressure on cost cutting for the hardware is more in recent times and it’s believed that value addition is more from the technology/software embedded in it than the hardware involved. In general as is the case, a low cost sensor is found to have a compromise in its performance be it range, precision, noise, life, etc., which are posing as challenges in industrial operations,” he adds cautiously.
For Jegajith P T, sensors are becoming more cost-efficient because these are produced in mass. This also implies that the usage of sensors is going to increase exponentially. “As far as the low cost sensors are concerned, the quality issues will always exist posing challenges in maintaining the sensor lifetime. Limited durability, difficulty in data management/analysis, etc., are some of the primary challenges faced when low-cost sensors are deployed. A factor that would help reach cost and productivity trade-off is ‘downtime’. As downtime reduces, the maintenance cost also reduces, causing increased sensor productivity,” he elaborates.
Sensors are getting smarter. What exactly is smart about smart sensors? Is it really smart to opt for a smart sensor? “Of late it is ‘Smart’ everything and sensors are not an exception. We are at the dawn of a new era of smart sensors. While normal sensors detect and send data for remote processing, smart sensors on the other hand consist of a sensor, a microprocessor and some kind of communication technology mostly wireless and have some computing embedded in them,” says Harish G Kashyap. “These smart sensors provide quality data in terms of less noise, more accuracy, and more energy saving but come with a cost impact than their non-smart peers.”
“Smart sensors are built as IoT components that convert the real-world variable that they're measuring into a digital data stream for transmission to a gateway. Smart sensors use built-in compute resources to perform predefined functions upon detection of specific input and then process data before passing it on. Smart sensors provide self-identification, smart calibration and compensation, digital sensor data, multi sensing ability, sensor communication for remote monitoring and remote configuration,” says Dr Makarand Sawant.
“Using smart sensors needs a different environment than using conventional mechanical sensors. Smart sensors are effective only if a comprehensive policy of where the entire manufacturing process is automated. It is very necessary that the automation process flows through every part of the manufacturing cycle from vendors to dealers and after sales markets,” says Anand Gijare.
“Smart sensors are trending in the industrial world. The feature that makes a sensor smart is the use of a microcontroller and FPGA to handle the logic operations and take certain actions or store various configurations that can be modified to change specific parameters. TinyML is gradually catching pace to make small embedded devices even smarter,” says Jegajith P T. “Coming to the next question, yes, deploying a smart sensor is a smart decision. Let’s try to understand this with an example, say a Proximity Detection Sensor. A proximity detection sensor can be configured to trigger distance-based alarms for conveyor belt applications.
“They are called smart because they not only sense the ambient information/data but also perform pre-defined functions (read data processing) before passing it on to the next device in the smart manufacturing/IoT chain over network or internet,” says Sameer Gandhi. “This leads to generation of more accurate and automated results rendering them very useful for varied monitoring and control applications. The data processing ability, provided by a microprocessor, is what differentiates a smart sensor from a normal sensor. Such smart sensors can also help in predictive diagnostics and alert for maintenance,” he elaborates.
Effective data capture
The topics of digital transformation like data analytics and AI/ML all depend on availability of 'good' data. What are the advancements in sensor technology to capture data which is difficult to obtain? “In industrial automation, edge devices such as machines and equipment embedded with smart sensors will be the key data suppliers. We should note that machine generated data comes in terabytes and it is unstructured format. Such data is hard to process by human efforts, and therefore needs platforms such as big data analytics and artificial intelligence for data storage, processing, analytics and insight generation,” explains Anand Gijare. He also makes three important points in this connection. First, data quality also depends on the placement of sensors, therefore, preparing sensors to adapt the environment in which they sit is very important for generating good quality data. Second, the parameter that hampers data quality is congested and unstable wireless networks – poor networks can jeopardise the sequence and cause issues at the edge. Finally, data analytics platforms where data is processed and stored in different compartments – design and architecture of data filtration and compartmenting data and releasing unwanted data from the system is crucial for retaining good data and removing unwanted ones.
According to Harish G Kashyap, it all depends on the data quality (reliability, accuracy, precision and noise-free) from the sensor which is then pushed to a cloud infrastructure for generating recommendations or making correct business decisions. The confluence of smart sensing with AI/ML will aid in performance improvement and dynamic decision making. Generating ‘good data’ will support easier feature extraction while building ML models and AI products. “The sky is the limit even for development of this sensor technology but it could be targeted towards sensing difficult to obtain data such as atomic/cell level data, human thoughts and emotions, infant and cognitive behavioural studies, etc.,” he explains.
“Good quality sensor data is crucial for IIoT applications as missing data and poor quality data can cause errors in detection and prediction processes. To deal with scenarios that are unusual, the datasets for abnormal instances can be created in a training environment and a reinforcement algorithm can be developed from the existing dataset to make prediction and enable accuracy,” says Jegajith P T.
What are the emerging trends in sensors that will further increase efficiency? “Use of sensors will increase in vehicles, sports, smart cities, precision agriculture, to improve the environment, improve energy management, and build green office buildings. In the future, maintenance will rely on sensors instead of being carried out according to a needs-based timetable, unsafe situations will be easily predicted,” opines Dr Makarand Sawant, and points out how autonomous sensor technology will become possible with wireless connections over long distances having an integrated power supply. “Sensors will be self-learning over the entire lifespan without maintenance, modifications, or calibration. The possibilities and areas of application for robot technology will increase significantly,” he adds.
Anand Gijare feels batteryless or self-powered sensors, which harvest energy from environmental resources, are the next big thing to watch. These sensors are reducing efforts on monitoring and changing batteries. In industrial automation batteryless sensors are available to monitor steam traps and machine health. Then there is Sensor Fusion technology where it builds the ability of sensors to monitor multiple data inputs. “Sensor fusion helps to put together data from various sensors to create a larger picture, a holistic view which certainly provides more clarity,” says Anand.
“Leveraging edge computing for certain business needs such as anomaly detection and cloud computing for more in depth needs such as pattern formulation, building ML models, etc., are already in the pipeline but the OT & IT industries are yet to unleash themselves to their maximum potential in the convergence,” says Harish G Kashyap. “As the transition to digital transformation accelerates, enhanced process automation would be the need of the hour and hopefully sensor technology would address this need by evolving at a faster pace than now in the not so very distant future of the 4th Industrial Revolution/Industry 4.0,” he predicts.
Summing up, Jegajith P T says sensors are adopting various industrial automation protocols and getting smarter by being able to incorporate AI/DL technologies. There is a range of smart wireless SoCs available for single chip solutions with wireless and data processing functionalities. Auto calibration is another feature available in most of the signal conditioning chips. “A leading semiconductor manufacturer recently announced a SoC which has a built-in signal conditioning unit specifically to handle multiple sensors and a CPU to process the process data. The CPU platform can be used to incorporate various industrial communication protocols. Sensor systems are trending to single chip low-power systems with Industrial protocol support, this will yield the Low cost high reliable sensor,” concludes Jegajith.
(Note: The responses of various experts featured in this story are their personal views and not necessarily of the companies or organisations they represent. The full interviews are hosted online at https://www.industrialautomationindia.in/interviews)