Machine-vision systems - what can they do for you?
Machine-vision (MV) systems can be applied to many manufacturing operations; where human vision now is required. These systems are best for applications in which their speed and accuracy over long periods of time enable them to outperform humans.
But what about your plant? Can you cash in on the promised benefits? To answer these questions, let's examine MV systems to determine where they can be applied and the types of problems they can eliminate.
The most important use of machine vision is precluding defects during a manufacturing operation. Consequently, cost justification for an MV system stems from savings attributed to the cost of poor quality.
A recent Booz Allen & Hamilton study emphasized there are two elements to the cost of quality--the cost of control and the cost of failure. The essence of the study is that we must include the savings from both in any justification equation for systems that improve quality.
The cost of control is fairly easy to identify and quantify. It includes the cost of preventing and finding defects before products are shipped. The most obvious costs are the labor used in QC activities, and the investment in inspection equipment. These might be large or small, depending on the size and complexity of your operation.
The costs of failure are much more difficult to quantify. They include the internal cost resulting from material scrap and rework, and the external cost stemming from warranty claims, liability actions, and recall orders, as well as hidden costs, such as loss of customers.
Properly applied, machine-vision systems can be a primary means of avoiding both internal and external failures. For example, human-based inspection techniques normally are structured to detect out-of-tolerance parts after they are produced. MV can spot trends--precursors of producing scrap--while parts are still within tolerance.
Moreover, laser gages and linear-array sensors can measure part dimensions during, or immediately after, a machining operation. This data then serves as a guide for adjusting a machine tool or replacing cutting tools before generating scrap.
The automotive industry is moving quickly to adopt these statistical processcontrol techniques. One example is Chrysler Corp's Windsor, Ont, Minivan assembly plant, where a Perceptron system is inspecting every door/hinge assembly (see box "Assembly-line QC by MV'). Trend analysis and frequency distributions developed from data collected by sensors in the system report changes in production quality. This ability to track production data and take corrective action often requires applying MV to inspect every part.
A side benefit--without regard to the amount of in-process data available from other monitoring methods--is reduced paperwork because recordkeeping is automated. Also, MV systems have the ability to transfer data between themselves and process controllers, or even a master computer.
In some manufacturing operations it's impossible to completely eliminate defects, even with a machine-vision system. In such cases, MV can be of value in separating good parts from scrap. Or, it can separate defective assemblies into two groups--those that can be reworked and those that can't. One example of the latter is found in the electronics industry, Figure 1.
Industry experts estimate that a faulty printed-circuit board detected immediately after fabrication can be repaired for as little as 25^. Once the same board is fully loaded with components, rework cost jumps to about $40. The cost to locate and repair the defect becomes even higher after the board is installed in a control system.
Similarly, with most parts or assemblies, the cost to scrap or rework a defective unit spirals up with each value-added step in the manufacturing process. Therefore, the sooner you catch and correct an out-of-tolerance condition, the better off you will be.
In the case of machined surfaces, parts with dimensions exceeding the maximum size tolerance usually can be reworked. On the other hand, undersize parts are very difficult, sometimes even impossible, to salvage. Vision systems can be used to make the distinction between reworkable parts and scrap.
Save on tooling
MV systems can be valuable in applications where expensive hard tooling is required to hold a part during a machining, forming, welding, or similar operation. Many times such tooling can be eliminated, or replaced by less expensive, flexible tooling in conjunction with a machine-vision system.
For example, an Automatix Partracker is used at a Norfolk and Western repair depot in welding new wear plates to railroad wheels, Figure 2. After the wheel is mounted on its welding fixture, MV provides location analysis to compensate for positioning variations and differences from wheel to wheel that show up after old wear plate removal.
Another good application for machine vision is where you are experiencing a high incidence of machine breakdown caused by oversized, undersized, warped, misshaped, or misoriented parts. An MV system upstream of a feeder mechanism can reduce, maybe eliminate, downtime by rejecting unacceptable parts before they impact machine operation.
A situation definitely warranting machine vision is one that requires maintaining a parts inventory because inspection may result in rejecting a complete lot run based on statistical sampling. The 100 percent inspection level--practical with MV--assures you that virtually every part passed by the system is good. This permits applying "just-in-time' inventory-control techniques with a corresponding reduction in material handling time and damage that might be experienced during handling. Likewise, MV can provide savings wherever inspection is a bottleneck.
Similar to one of the basic justifications for robotics, machine vision can be applied to operations in hazardous or unhealthy environments. It generally has a better tolerance for loud noise, elevated temperatures, heavy parts, and airborne contaminants, such as metal dust or toxic vapors, than do humans. Conversely, vision systems aren't as prone to introduce contaminants into an operation, whereas humans often are the source of dust, oil, and other debris carried on hands and clothing.
Lastly, if you are faced with an operation that's subject to a number of errors caused by operator judgment, fatigue, inattentiveness, or oversight brought about because of a dull job, MV can provide a quality boost and cost savings. Whenever undertaking a major capital expansion program, take a hard look at machine vision in lieu of alternative, less effective, often more costly methods.
But is it for you?
Most metalworking plants perform operations where MV can improve quality and cut costs. The checklist of factors to consider when selecting and implementing a machine-vision system (included in this article) will help you decide where MV is a practical solution to your vision-sensing problems. There also is a lost of factors to consider in selecting and installing MV systems.
Being systematic when evaluating needs, developing specifications, planning projects, and arranging for procurement and installation virtually guarantees you will receive a properly designed system that meets your machine-vision requirements.
Photo: 1. This Orbot system is designed to inspect bare printed-circuit boards. It checks the inner layers, artwork, and resist patterns to spot a reject condition at the point of 10 west value added.
Photo: 2. At a Norfolk and Western repair depot, this Automatix Partracker system uses stereo techniques to precisely determine location of a railroad wheel in a welding fixture. Data is fed back to correct the path of the welding gun during placement of a new wear plate.
|Printer friendly Cite/link Email Feedback|
|Publication:||Tooling & Production|
|Date:||Oct 1, 1984|
|Previous Article:||Squelching stamping noise.|
|Next Article:||Rolling to shape: low cost, high potential.|