The World is paving its path towards Industrial Internet of Things (IIoT) with full throttle and the initiative like Smart factories and Industry 4.0 is increasing the level of dependency on electronic machines for work, with a greater pace than yesterday. Consequently, this dependency has been one of the major catalysts in pushing technology forward with advancements. Now, labor-intensive methods are being switched with machines having the latest technology and with a promise of a far better performance than humans, in almost every industry. However, these advanced machines are complex, and once suffered a breakdown, it can take a lot of time as well as manpower to recover from that breakdown. Such breakdowns in a manufacturing unit can cost a huge fortune to companies. Predictive maintenance has probably been around as long as someone first said: “That’s going to break soon.” From a simple refrigerator in a single-family home to complicated industrial equipment supporting thousands of dollars of infrastructure, predictive maintenance is the silent hero behind the scenes
Previously, when machines were not too complex, there was more maintenance through manual labor on contrary to what it is now. The manufacturing units and factories want to stay competitive by the rapid assembly lines and tremendous automation through complex machines. These machines deliver the best results when they are being regularly checked for maintenance. Besides, the routine maintenance check-ups, the breakdown of a machine can also be anticipated, before it even completely shuts down.
Predictive maintenance has become a very efficacious and competent tool in this fast-moving technology-led world by minimizing disruptions. It is an important component of Industry 4.0and Smart Factory vision and mainly consists of increasing efficiency by planning maintenance activity based on detection or prediction of failures in the process or equipment based on data analytics methods.
Predictive maintenance facilitates the ability to properly allocate and distribute resources and alternate paths for processing so that the production doesn’t stop, extends the uptime of the equipment, reduces overall maintenance costs, and allows optimization of spare part inventory by enabling preventive maintenance based on the equipment’s actual needs.
Predictive maintenance involves a combination of techniques such as Condition-based monitoring (CbM), machine learning, and analytics to predict upcoming machine or asset failures. When monitoring the health of a machine, it is critically important to select the most suitable sensors to ensure faults can be detected, diagnosed, and even predicted. There are many sensors currently used to sense and detect faults, in rotating machinery and their loads, with the end goal of avoiding unplanned downtime. Ranking each sensor is difficult as Predictive maintenance techniques are applied to a multitude of rotating machines and nonrotating machines.
The application of sensors for predictive maintenance detects anomalies in equipment before those turn into system-critical failures, allowing maintenance to be scheduled before the equipment actually breaks down. In contrast to preventive maintenance which is characterized by regular inspection intervals for maintenance replacements and repairs, predictive maintenance can detect system failures outside of set maintenance intervals.
The application of sensors as predictive maintenance
Some sensors can detect certain faults, such as bearing damage, much earlier than others. The sensors most commonly used to detect faults at the earliest possible moment are discussed, namely accelerometers and microphones. Most PdM systems will only employ some of these sensors, so it is imperative to ensure potential critical faults are well understood along with the sensors that are best suited to detecting them. The key to the successful implementation of a predictive maintenance program is utilizing the proper sensors to determine machine conditions. Sensor selection starts with an understanding of a machine’s potential failure modes and the warning signs associated with these modes. Typical warning signs on equipment with rotating parts include unbalance, bearing damage, cavitation (pumps), increased machine vibration levels, increased temperature of machine components, loss or reduction of lubrication flow, and loss or reduction of cooling water flow.
Meandering around the sensors for predictive maintenance, I was fortunate enough to get the opportunity to have an insightful conversation with Deepak Bachu, Director & Country Head for Sales – Power & Sensor Systems, Infineon India about the application of Sensors as predictive maintenance.” In predictive monitoring, constant monitoring of system performance and state anticipates potential failures outside the planned inspection intervals of preventive maintenance. However, leveraging the potential of predictive monitoring can require what is referred to as condition monitoring. Condition monitoring is a prerequisite for predictive maintenance and as such is a component part of predictive maintenance, not a maintenance strategy in itself. Condition monitoring can be compared to an employee who constantly monitors a system’s condition and records any irregularities during operation but does nothing more-” said Deepak.
“Sensors are the backbone in the data collection process of condition monitoring and predictive maintenance. Placed inside systems, they collect data on the most crucial system parameters such as current, vibrations, sound anomalies, airflow, and many others. Combined with microcontrollers, connectivity, and embedded security, collected data can be processed to detect anomalies before they turn into costly failures- Deepak further added.
Deepak also enlisted a variety of sensors that can be implemented such as current sensors, 3D magnetic sensors, pressure sensors, MEMS microphones, and hall sensors, and switches:
- Airflow measurement at the compressor based on the XENSIV™ DPS368 barometric pressure sensor
- Current measurement at the fan and compressor based on the XENSIV™ TLI4971 current sensor
- Position sensing of the motor with XENSIV™ TLI493D-A2B6 3D magnetic sensor
- Sound anomaly detection in the unit with the XENSIV™ IM69D130 MEMS microphone
- Linear vibration measurement with XENSIV™ TLE4997E Linear Hall sensor
- Opened and closed lid detection with XENSIV™ TLE4964-3M Hall sensor
- Speed and direction measurement with XENSIV™ TLI4966G Double Hall sensor
- Data processing with XMC4700
- Secured connectivity on and authentication with OPTIGA™ Trust M
PdM for a semiconductor manufacturing unit is altogether a different ball game
In the context of Semiconductor Manufacturing, the work item is a thin disc of Silicon referred to as a wafer. The wafer undergoes a series of highly controlled and complex processes such as cleaning, deposition, patterning, etching, metallization, measurement, and electrical testing. The wafer contains many “chips” patterned on it and the percentage of good chips at the end of all processes indicates the wafer yield. Nearly every electronic device that was ever created started off as a chip on the wafer. These processes are highly interdependent and the equipment used is often embodiments of state-of-the-art technology. Predictive maintenance also helps reduce wastage and avoid abrupt breakdowns of the process flow. It is especially critical in Semiconductor Manufacturing as the processes are highly dependent and cost-intensive. In addition, it also significantly impacts productivity and yield.
Shashank Shrikant Aghaese, who currently leads SESG (Smart Equipment Solutions Group) as Director and Manish Goel, Senior Director of an image sensor, multimedia, and LSI R&D teams at Samsung Semiconductor India R&D Center, had earlier spoke with us and highlighted that for successful PdM set up, the quality of data is one of the paramount requirements to anticipate the health of the equipment for a successful PdM setup. Because sensors detect a comprehensive caricature of the physical factors of the equipment, it becomes imperative to have a continuous sensing process as well. Both the experts reflected and enlisted some key requirements for setting up a successful PdM process in our conversation.
- Data quality: Sensor data carries important signatures of the health of the equipment, process as well as wafer. It is imperative that the data be accurate and free of noise or influence by extraneous factors as it forms the main input for the rest of the PdM setup.
- Pervasive sensing: Sensors convert the physical characteristics of equipment and the process to the digital domain. In order to get a comprehensive picture of the physical factors, it is important to cover all the important interactions between wafer, equipment, and process. This requires increasing the sensor coverage to monitor pressure, temperature, vibrations, humidity, the flow of gas, power, etc. at multiple locations.
In the case of using images as data for anomaly detection or root cause analysis in semiconductor manufacturing, as we move into smaller geometry process nodes, imaging-based fault detection techniques require higher resolution cameras to be able to identify the smaller features and any anomalies. Also, a higher frame rate is required to be able to keep up the normal movement of the objects under test. In addition, the camera or sensor should be able to adapt to the environment of the manufacturing and be able to capture high-quality images in any lighting conditions. It is challenging to upload and store this amount of data in real-time. Therefore, an edge AI platform is highly desirable which can analyze and filter the required for further analysis.
- AI modeling: The factors that affect the wafer or the proper functioning of the equipment are quite complex. However, the data available is huge. This is a good case for the utilization of AI modeling that can learn the complex and non-linear relations between the physical state of the process and sensor data. The AI models play an important role in detecting and predicting anomalies, bad process conditions, and the remaining useful life of the equipment or its module.
The accuracy of the models becomes an important pillar of PdM. In addition to creating the models, a strategy for time-based model updation is required since the relation between different process parameters changes over time and is highly dependent on the placement of sensors and equipment layout. In addition, physics and domain knowledge infusion in AI models is also desirable to improve accuracy and gain from the experience of experts.
Most of the recent research effort in PdM has been in the area of creating advanced and complex AI models. Another field of interest in recent years has been explainable AI whereby the predicted failure modes and the reasons for the same can be described in terms of physical factors that can be easily understood by the stakeholders.
As human beings, we have this perpetual and stringent desire to peep into the future to get an idea about the fatal anomalies that can happen to us. Because if we know them before handed, then we might be able to minimize the damage. Such is the case with modern machinery’s life as well. Sensors are a crucial guide and the question of their applications in modern machinery has become a matter of its life or death. Setting up an effective predictive maintenance system via sensors can save an industry, a business, or a manufacturing unit from huge losses.
Mayank Vashisht | Sub Editor | ELE Times