Achieving optimum performance of unit processes and complex manufacturing systems is only possible if accurate and timely information is available concerning the unit process, particularly with respect to the five process components—equipment, tooling, interface, workzone, and workpiece material. Three strategies (not mutually exclusive) can be envisioned for ensuring that this information is available: data bases and knowledge bases that capture empirical data and heuristic rules; models of the process and its components, from which performance can be predicted; and direct measurements from sensors for process monitoring and feedback to controllers. All focus areas are subject to monitoring needs, often with competing requirements for time response or location of sensors. Sensors play a critical role in the development and implementation of advanced and new unit processes. Today many sensors are selected for use in unit processes on the basis of their reliability and low cost, not because they measure parameters that are fundamental to the material being processed. As the processes that make up the manufacturing systems become more complex, the challenge and the benefits of developing new, enhanced, and reliable sensing systems has increased (NRC, in press). Sensors and sensor systems that are inexpensive, robust in the processing environment, and easy to use are required.
The chapters in Part II of this report identified a large number of unit processes, and components within those processes, in which sensors are required to advance the technology or enable the existing technology to perform to its limits. Sensor-based inspection and quality-control technologies are widely recognized as crucial for major advancements in manufacturing capability and productivity. Thus, sensors are key elements in the other enabling technologies of process control (Chapter 12) and in process precision and metrology (Chapter 13).
There are a wide range of applications for sensors within the domain of unit processes. Some examples, in roughly the order of increasing sensor-system sophistication, are
- interfaces or nodes within the system to pass information (e.g., tolerances, orientation, material characteristics, sequence, etc.; Ayres, 1988);
- quality-control schemes involving only pre- and post-process inspection; that is, finding the defect before the part enters the unit process (when nothing is happening) or after it exits the process;
- in-process inspection, in which deviations can be detected in real-time and perhaps corrected before additional processing occurs; and
- self-directed processing, in which the control settings are determined by the response of the material, as measured by sensors, to the processing conditions, as opposed to following a pre-established schedule of process parameters.
Sensors are vital for intelligent processing. The requirement for improved sensor systems for process monitoring and control will become increasingly important as more-complex processes are developed for small-quantity production of precise and expensive components made of advanced materials. Because research on process modeling and material characterization cannot remove all uncertainty in the manufacturability of advanced work pieces, sensors will always be necessary for successful production. Part II contains examples of unit processes that would be extremely difficult to control without in-process feedback or of which it would be difficult to accurately predict the outcome in advance due to a number of material, tooling, and environmental variables, such as tool-workpiece interface conditions that govern the rate of energy transfer from the equipment and tooling into the workzone. For example, precision machining requires adaptive control based on sensor readings.
The committee envisions manufacturing systems that are so finely controlled that they achieve very stringent requirements of new materials, processing techniques, tolerances or shapes, production speeds, and yields that would be extremely difficult to achieve without the real-time aid of computers in which process models are embedded and to which an array of sensors constantly communicate the process state and provide system diagnostics of potential problem areas. The speed of response and level of attention required for these new unit processes exceed the capabilities of the most adept machine operator.
The adoption of sensor technology can make viable those unit processes that are inherently so complex or unstable that they demand continuous monitoring and control to ensure acceptable yields. All unit processes for advanced materials with intricate near-net shapes, high-quality surface treatment, and compositions
controlled exactly to precise specifications share these characteristics of complexity and potential instability as they are pushed to their full potential. Research on sensors should address the challenges posed by these applications.
Some of the best information regarding performance of sensor systems in industry comes from the automotive manufacturer Mercedes-Benz (1990). A survey was conducted of approximately 120 sensor systems applied to drilling (35 percent), multiple spindle drilling (7 percent), turning (57 percent), and milling (1 percent). The results are summarized in Table 11-1. The benefits derived from the successful applications far outweighed the costs of the failures. For example, Mercedes-Benz reports significant realized savings for the machining of a transmission shaft using a sensor-based process monitoring system (Mercedes-Benz, 1990).
Machining of advanced materials such as ceramics provides another example of the need for sensing as part of the process. For the most part, ceramic materials are not intrinsically expensive. The principal production costs are in the shaping of the ceramic part, including initial shaping (such as injection, casting, or pressing), green and white machining, and the finish machining after sintering (König and Wagerman, 1993). Up to 90 percent of the cost can be attributed to the final finish machining of the sintered parts by grinding; lapping; or other less conventional techniques, such as electrical discharge or laser-assisted machining.
The machining process itself may induce surface defects in the component, which severely limit its usefulness. Because it is not possible to predict the circumstances under which these defects occur (due to machine performance, tool performance, and process variability, sensors must be utilized in-process or post process (König and Wagerman, 1993). Sensor-based monitoring could address
Table 11-1 Results of Mercedes-Benz Manufacturing Sensor Implementation
Sensor Implementation Result
Fully functional (performed as expected)
Conditionally functional (had one or more deficiencies)
Technical failures (poor resolution, false alarms, etc.)
Replaced by alternative systems
the machine (e.g., diagnostics and performance monitoring); the tools or tooling (e.g., state of wear, lubrication, and alignment); the workpiece (e.g., geometry and dimensions, surface features and roughness, tolerances, and material damage); or the process itself, including the interface between the tooling and the workpiece (e.g., chip formation, temperatures, and energy consumption).
The objectives of machine condition monitoring also include safety, prevention of damage to the machine, prevention of rejected workpieces, prevention of idle time on the machine, and optimal use of resources (Tonshoff et al., 1988). Tonshoff et al. point out the importance of the sensor as part of a system that includes signal conditioning, models relating measured values to monitoring and control variables based on fundamental process physics, and strategies for information utilization. This technology is often referred to as an ''intelligent sensor'' and implies much more capability than just the transducer and preamplifier might in a simple monitoring setup. Many of these intelligent sensor systems employ multiple sensors for process monitoring. A variety of sensors are used to provide a range of coverage of process characteristics with the goal of ensuring higher reliability. This multisensor approach requires more attention to real-time feature extraction, information integration, and decision making to be effective.
The increasing demands of sensor systems have encouraged the development of systems using a variety of sensors. This is especially true for monitoring processes that are categorized as precision manufacturing, that is, for which the tolerances on form, dimension, or surface features are very stringent. In the case of machining, these processes are characterized by very small material removal rates, very low process power consumption (often in the presence of high tare power consumption), and the requirement that the sensor not intrude in the process or require substantial physical modification to the machine. These requirements tend to reduce the effectiveness of, or eliminate consideration of, a large number of traditional sensing methodologies—for example, in the case of tool condition sensing in machining, force, torque, and motor current or power measurement.
Unit process equipment can be viewed as platforms for advanced sensors and controls. High-speed machining is an excellent example of the need to integrate the development of machines, controllers, and sensors and the tooling to advance the process technology. This type of integrated development is more likely than not to be essential for all advanced unit processes. Sensor development has been most advanced recently for sensing methodologies that can take advantage of the microlevel silicon machines and devices now available. For example, researchers have developed and fabricated miniature vibration sensors and accelerometers that, if appropriately applied to a machine structure (like the
resin concrete materials now in use as machine tool structures), could provide in situ sensing capability for control of machine stability and deflection.
Sensing techniques exist that, individually, are not sufficiently reliable for process monitoring over the normal range of process operation. However, if several different sensors are used, each of which is effective over a particular portion of the operating range, effective real-time process monitoring can be realized over the entire operating range.
There is a tremendous need to incorporate sensors in existing unit processes. For many applications, the research for relating sensor output to process characteristics has been completed but not yet exploited in a real production environment. For example, solid-state sensors are commonly used to ensure consistency for many chemical processes. These same sensors have the potential to monitor and report on interface variations due to lubricant contamination or loss in a variety of unit processes for which the interface is a key to product acceptability.
A persistent barrier in applying sensor technology has been matching application needs to sensor capabilities. A recent National Research Council study suggests a methodology that employs a set of descriptors that could be used to match needs with capabilities and thus provide a rational basis for evaluating candidate sensor technologies (NRC, in press). There are many advanced sensor technologies available which range from optical, infrared techniques to high-frequency ultrasonic and acoustic emission technologies (Shiraishi, 1989; NRC, in press). The motivation to increase the use of sensors is high. For example, estimates of the impact of sensing systems on process performance of existing systems indicate that a sixfold increase in effective operation time is possible (Eversheim et al., 1984). If sensors also enable increased product yields due to prevention of defects, the total benefit will be even greater.
Key aspects of sensor technology to be emphasized for R&D include:
- sensor systems with digital architectures readily integratable with machine controllers;
- intelligent sensor systems that employ advanced sensor fusion and feature extraction techniques for reliable process-state determination and diagnostic decision making;
- process-specific sensor developments that address the need to monitor aspects of all five process components;
- new sensor materials and techniques capable of monitoring nontraditional processes and the processing of nontraditional materials;
- exploitation of sensor technology that was developed for applications other than unit manufacturing processes;
- digital signal processing techniques that can accommodate uncertainty in sensor-produced data;
- methodologies for readily assessing the economic viability of applying sensor systems in unit processes; and
- vehicles for speeding industrial evaluation and commercialization of new sensing technologies.
Ayres, R.U. 1988. Complexity, reliability, and design: Manufacturing implications. Manufacturing Review 1 (1):26-35.
Eversheim, W., W. König, M. Weck, and T. Pfeifer. 1984. Tagungsband des AWK'84. Aachener Werkzeugmaschinen-Kolloquium.
König, W., and A. Wagerman. 1993. Machining of ceramic components: Process-technological potentials. Proceedings of the International Conference on the Machining of Advanced Materials held July 3-16 in Gaithersburg, Maryland. Gaithersburg, Maryland: National Institute of Standards and Technology.
Mercedes-Benz. 1990. Fertigungssicherheit und-qualität durch Intelligente Technologien, Aachen Machine Tools Colloquim, May 1990, Wettbewerbsfaktor Produktionstechnik . Düsseldorf: VDI-Verlag GmbH.
NRC (National Research Council). In press. Expanding the Vision of Sensor Materials. National Materials Advisory Board, NRC. Washington, D.C.: National Academy Press.
Shiraishi, M. 1989. Scope of in-process measurement, monitoring and control techniques in machining processes—Part 2. Precision Engineering 11(1):27-47.
Shiraishi, M. 1988. Scope of in-process measurement, monitoring and control techniques in machining processes—Part 1. Precision Engineering 10(4): 179-189.
Tonshoff, H.K., J.P. Wulfsberg, H.J.J. Kals, W. König, and C.A. van Luttervelt. 1988. Developments and trends in monitoring and control of machining processes. Annals of the CIRP (International Institute of Production Research) 37(2):611-622.