When the automation age began, manufacturers wanted to use robotics at every conceivable step of the process. This improved productivity and output, but production processes still did not yield consistent results. While engineers pondered why automation had not achieved zero defects, production managers had their hands full managing and planning schedules, inventory, maintenance, and energy costs usually with statistical data hand-coded into compute resources.
The Industrial Internet of Things (IIoT) offered solutions as it ushered in Industry 4.0 to optimize operations with networked digital information, networked information processing, and networked control. The “things” in IIoT are typically the sensors built into equipment or added at various process flow points. They are one of four foundational pillars that enable the optimized operation associated with Industry 4.0.
Awareness gains from sensors
Implemented at scale, sensors feed information to computational power such that new actionable insights can be used to improve many areas of operation. They play a key role in:
Increasing flexibility and responsiveness: Sensor information builds the immediate awareness needed to optimize the use of Just-in-Time processes, while ensuring the workforce is well-informed of changes and requirements at all points of operation.
Reducing equipment downtime: Sensors can empower equipment to self-monitor to predict maintenance requirements and self-calibrate to reduce installation, configuration, and maintenance costs.
Increasing yield: Sensors can “examine” products not just at the end of the manufacturing operation to aid in quality assurance but throughout the process flow to enable quality control. They can also ensure only as much material is used as is needed for various processes, thereby minimizing waste.
Demonstrating compliance: Sensor data creates an auditable trace that may be used to demonstrate compliance with various regulations related to good manufacturing practices (GMP) as well as various standards.
Developing new processes and cost structures: Analysis of historical sensor data can help understand how processes can be further optimized. It can also provide insight into costs at process and material levels, helping to develop more efficient processes or adjusting per unit costs elsewhere.
Types of sensors
There are many types of sensors enabling benefits described above. Some of the most common sensors in the manufacturing industry include the following.
- • Flow sensors measure flow rate of a fluid being supplied through a pipe to a process chamber, such as a reflow soldering machine
- • Force sensors or load cells or load cells measure force applied on one or more axes for such applications as ensuring safe gripping by robot arms and measuring when material bins need refilling
- • Humidity sensors are used where controlled moisture levels are necessary, such as in cleanrooms or inventory storage
- • Image sensors as part of industrial camera systems are used in inspection, process control, workflow management, robot guidance systems, and track-and-trace applications
- • Level sensors provide real-time measurements of material quantity in tanks, bins, and containers
- • Position sensors — angular or linear — detect distance traveled and are used in robotics and conveyors to help in component placement, inspection, packaging, and sorting
- • Proximity sensors measure the proximity of objects for handling by robots or even proximity of workers to equipment for safety reasons
- • Temperature sensors, including IR sensors, help detect critical process temperatures and machine condition
Additional sensors, such as accelerometers, particle counters, current or power meters, vibration sensors, and ultrasonic microphones may be used to monitor or predict equipment health. The smart factory is therefore bristling with awareness-enhancing devices that generate data often at millisecond intervals.
Distributed intelligence with smart sensors
The constant data generation in sensor-driven optimized operations requires information be differentiated and prioritized so that process control parameters and safety critical measurements are processed at or near real-time while the rest is analyzed in a longer time window. Moreover, moving all of the data across an industrial Ethernet or 5G network to the datacenter, where a single process station may cough up 10-30 data points every few milliseconds, is often unviable.
Smart sensors address this by embedding compute resources, machine learning (ML) algorithms and artificial intelligence (AI). Information requiring immediate action can thus be processed fast at the sensor node and data that benefits from deeper analysis in combination with upstream or downstream conditions can be handled in the datacenter or a hybrid cloud. Furthermore, embedded AI can give awareness to sensors of their own performance, thereby reducing instances of corrupted data and alerting to damage before a sensor failure occurs.
Implementing feedback loops
The combination of advanced sensors and sensor modules with embedded compute resources means that raw data can be transformed into intelligent real-time decisions and actions. This allows much greater control over processes by weaponizing insights gained on process variation against manufacturing inconsistencies mentioned earlier.
Feeding back downstream process and equipment information upstream to effect early changes in manufacturing can significantly improve yield (Figure 1).
Figure 1: A block diagram representing potential for automated feedback loops between upstream lithography and downstream etch processes to reduce variation in critical dimensions (CD).
The feedback loops shown in Figure 1 can be adapted to almost any manufacturing process. For instance, when populating printed circuit boards (PCBs), manufacturers would typically place industrial cameras in automated optical inspection systems (AOI) before and after the reflow soldering process. Before reflow AOI detects solder paste deposition and device placement errors, while after the reflow oven AOI validates corrections were successful.
Such feedback loops, when automated, remove uncertainties due to slight variations in manufacturing conditions, such as temperature and humidity, or process conditions due to slight changes in material, process gasses, or pressure. The result is not only consistency in the output but the ability to improve process optimization and to make better and faster adjustments to changing production recipes.
Look ahead to Quality 4.0
Quality check and quality control (QC) have always been part of manufacturing, but never before smart manufacturing enabled by sensors and AI have companies expecting zero defects. Although a faded quality management buzzword introduced in the mid-1960’s, the concept is seeing rejuvenation in the Zero-Defect Initiative due partly to the digitalized awareness created by smart sensors.
Going further, Quality 4.0 is taking these scorecard-driven early attempts at eliminating defects by expanding feedback loop systems beyond the immediate manufacturing floor to inventory and site management, and even to partner manufacturing sites, tracking material and products throughout their process lifecycles.
Zero defect is an aspirational target but one that seems increasingly achievable with Quality 4.0. Its framework is coming up on the smart sensor foundations of Industry 4.0.
Ask your nearest Arrow representative today how you can use sensors to optimize your operations.
