How the IIoT and edge computing are solving the “Short Board” problem

April 22, 2022 | Provided By Avnet

SEE ALL AVNET ARTICLES

Why Developers Are Constantly Striving for Integration and Differentiation

Industrial manufacturing processes generate vast amounts of data. This data can be used in meaningful ways – to predict failure, optimize the lifespan of equipment, and improve process efficiencies to keep pace with market demand. In any industrial network, the first step is data collection. Next comes an attempt to strike a balance between local real-time data processing and long-term offline data storage. Subsequently, various measures are undertaken to optimize the industrial manufacturing process.

Information is collected and sent to a central location. If sent relatively early, brief IT outages are usually fall within the acceptable range. But as companies around the world become increasingly reliant on IT, they are becoming less tolerant of time spent on the maintenance of equipment, and are debilitated by the response time achievable with current technology. However, in advanced technology teams, modern IT systems use a powerful suite that combines AI and machine learning (ML) to make their IT infrastructure respond faster to changes in sensor data packets.

Intel is already utilizing IIoT edge computing in a predictive approach to maintenance that has reduced factory downtime by a staggering 300 percent. By monitoring the health of its fan filter units (FFUs) in its semiconductor production facilities through the deployment of industrial IoT (IIoT) sensors and edge computing, technicians are alerted to potential problems. They are also able to develop proactive maintenance schedules and reduce unscheduled downtime. FFUs, which filter and clean the air inside industrial machines, are typically monitored and maintained manually, which makes it challenging to predict failures.

In particular, Intel placed an accelerometer at the top of each FFU to measure variations in the fan’s function, creating a baseline for comparing behavior across both the tool and fleet. It also integrated accelerometers with gateways and edge applications, and developed machine learning algorithms around it. This created a baseline performance for each FFU, facilitating the measurement of changes and the generation of alerts for anomalies and potential problems. Summary data is then sent to the cloud to give tool owners a view into baselines and trends, and a chance to respond to alerts of anomalies in the system. This has enabled the ordering of replacement parts and scheduling maintenance ahead of time, thus increasing FFU uptime by over 97 percent. It has also effectively eliminated process excursions in the manufacturing process, which can result in damage to materials.

Though FFUs represent a single component in a complex process, they are small enough in scope yet large enough in impact to demonstrate a return on investment in the factory. They clearly demonstrate the ROI potential of an edge-computing and cloud-based IIoT predictive-maintenance solution.

What exactly is IIoT?

IIoT (Industrial Internet of Things) is the continuous integration of various sensors, controllers, mobile communication networks, and intelligent analysis technologies with sensing and monitoring capabilities into all levels of the industrial production process. IIoT greatly enhances manufacturing efficiency and improves product quality, reduces product costs and resource consumption, and takes traditional industries to the next level through smartification.

Cloud computing solutions form the lion’s share of the industry's comprehensive network solutions.

And, what is edge computing?

Edge computing refers to an open platform that unifies networks, computing, storage and core application capabilities in proximity to the source of the object or data. It allows local processing and the storage of crucial data, which can then be transmitted to the central data center or cloud storage. Edge computing helps to optimize cloud computing systems and prevent data transmission interruptions. Cloud servers become control nodes for smart edge devices and perform summary analysis.

In a traditional scenario, when a key part of a machine is dislodged or damaged, the entire machine will malfunction. If that part eludes immediate detection, the entire machine will then require checking, thus delaying production and monopolizing human resources. Effectively, one single “short board” has halted the operation of the entire machine. In many ways, IIoT is just like that machine. However, if there are professional monitoring tools that can solve those problems in a timely manner, or send out reports or warnings of likely malfunctions, personnel can overhaul the entire machine before it becomes totally inoperable. Hence, the development of edge computing today is to solve this “short board” problem. It also explains why many people believe that edge computing can redefine IIoT.

However, real edge computing does not stop there. With ever-increasing demand, Industry 4.0 is no longer at the center of the debate. The big question right now is what form the industry will take to accelerate the digital transformation of the manufacturing industry. The main driver of capital investment in IIoT infrastructure is the prospect of facilitating production optimization and reducing maintenance costs, both of which deliver immediate and measurable benefits.

In the traditional IIoT model, the job of the sensor or hardware is to collect data and send it to the upper-tier IoT server or platform through the built-in network connection, and then perform data analysis, data visualization and application development accordingly. Finally, management uses the results of the analysis and visualization to formulate plans for machine maintenance and production process optimization.

Most of the data has instantaneous value and there is no need to wait for it to be transmitted to the upper tier servers. The need for moving data processing to the edge of the network and the ever-increasing processing power requirements have created a different type of IIoT network – one that may not have a strict hierarchy but will incorporate a wide variety of methods for connectivity and processing in many forms of edge devices.

A handy analogy for cloud computing

When you scald your hand, what happens first – do you feel the pain or pull your hand away?

In fact, when someone touches a hot object, the hand automatically pulls away from the object before anything else. It is an instinctive response that is transmitted to the central nervous system through the spinal cord. Only after that can the brain feel the pain. Imagine if the brain felt the pain before pulling the hand away... Maybe you would need to see whether the hand is “well done” before reacting?

Cloud computing is like the brain, and edge computing is like control of the body by the central nervous system. When the hand encounters an obstacle, like something very hot or sharp that needs to be avoided right away, the brain's reaction time is just too slow. This requires that body parts have their own "edge devices". The computing power shifts from the cloud to the edge, birthing the concept of edge computing.

So, what exactly is an edge device? The devices mentioned earlier for monitoring Intel FFUs are edge devices. For example, sensors and AI cameras that are able to accurately measure and record local temperatures are all edge devices. When these edge devices are used for computing, it is known as “edge computing.” In the future, the edge layer will become increasingly blurry, and edge devices will become increasingly intelligent, and increasingly diverse.

The "brain" is ready, but the "central nervous system" is essential

Cloud computing is being increasingly used in industry, making it easier to address and solve problems as they arise and sometimes even before they arise. In summary, edge computing is essential in the following scenarios:

Poor connectivity of IoT devices
Applications rely on machine learning (ML) and require a lot of data to respond
Data needs to be kept in the factory for purposes of security and privacy
Raw data at the edge requires preprocessing to reduce computation

Although the smartification and blurring of the edge tier has brought about great improvements in efficiency, it has also brought great risks to security. Some of these risks arise from a lack of standards and specifications, or uncertain security quality. While the industrial network is closed, the edge tier equipment is exposed to the Internet. Since edge devices break through the constraints of centralized security management, there must be easily exploitable security loopholes. However, it makes sense for computing to be built in edge devices, which can solve the time and capital costs caused by delay issues. Despite the fact that it was originally conceived as an industrial network that integrates centralized management, future development will undoubtedly diverge and diversify. Edge computing provides an opportunity for equipment suppliers to sell large quantities of new software, hardware and solutions. Not surprisingly, many software vendors and chip vendors are already moving in that direction.

It may be that only products that share a common developmental trend are the final winners. The differentiation of software and hardware and the reduction of coupling, the shift from cloud computing to edge computing and from integration to differentiation, are inevitable. So far, the best converged product is probably none other than the smart phone, simply because of its convenience. No doubt other products will gradually diverge after the initial convergence, driven by convenience, and sometimes there may be further cost reduction.

As developers race to embrace the virtually infinite possibilities of edge computing, it may not be long before the “short board” problem becomes a thing of the past.