Flexible assembly - manufacturing's newest frontier.
About nine years ago, the Ford Motor Co and Unimation Inc, robot-builder in Danbury, CT, had a hush-hush project underway. Together the companies developed a two-armed assembly robot with hand-to-hand coordination governed by a single controller. When all the parts and software were put together, the robot was able to assemble 14 parts of an automobile governor. Moreover, the robot could perform the task one and one-half times faster than a skilled human.
From the outset, though, the project was deviled by technical problems. Accurate, consistent parts feeding and parts removal were difficult to achieve, and the software presented formidable challenges. After giving it a good try, the companies shelved the project.
The Ford-Unimation attempt typified flexible assembly projects in the US a decade ago. Problems presented themselves at every turn, and success seemed out of reach. Interested organizations kept at it, though--not only manufacturers and robot builders but also vision developers and private, university, and government laboratories--and one by one the obstacles were either overcome or circumvented.
A number of progressive manufacturers, particularly in the automotive and electronics industries, succeeded in developing efficient, flexible assembly systems (FAS) in their laboratories, and then moved the systems onto the production floor. One of the more notable examples is IBM Corp, which developed FAS for assembly tasks in its own plants. Today IBM not only uses FAS but also markets them and gives courses on their use.
For the most part, though, the subject of flexible assembly was more discussed than demonstrated. True, the forecasters were making optimistic predictions, and at trade shows some of the robot-builders had their offspring performing cute tricks, but by and large an interested, prospective user couldn't "go out and buy an FAS.'
The turning point
In January, 1982, something occurred that seemed to push the germinating FAS technology out of the ground. IBM Corp introduced their System 7535, the first packaged, recognizable FAS to hit the US market. Since then--and particularly during the past 12 to 18 months-- flexible assembly has become the hottest thing going in manufacturing.
"Flexible assembly is a very significant activity, and it will be exploding,' says Stanley Polcyn, vice president of Unimation Westinghouse. "That's not really surprising, though, when you consider the large portion of our work force involved in putting things together.'
According to the University of Michigan's 1982 Delphi Forecast on industrial robots, the percentage of assembly systems sold in the US that will be programmable (flexible) will rise from 1980's modest 15 percent to 45 percent by 1990. The automotive, electronics, and aerospace industries are expected to be leaders in adopting the FAS (see bar chart).
Robot builders that have targeted assembly as a primary market are optimistic about their sales prospects. "Of all the robots we're selling, about one third are going into the FAS,' says Guy Potok, vice president of engineering and manufacturing at GMF Robotics Corp. "Some of those systems incorporate 40 to 50 robots each. Once a user has installed a successful system, its design can be cloned in other, similar plants. And much of the technology is generic; that is, it can be transferred not only within an industry but between industries.'
Three major forms
Generally speaking, each FAS is unique, but FAS seems to fall into three major categories according to configuration. The first of these--the one that gets the most attention--is the cell in which one or more robots perform a number of assembly operations in sequence.
Increasingly, the robots are being aided by programmable sensors such as machine vision and ultrasonic arrays, and by simpler devices for sensing parts presence, torque, and other phenomena. A description of a robotic assembly cell appears in the large box headed, "How Fanuc uses robots to build motors.'
A second, less-publicized form is the flexible, multistation, synchronous system. In some respects, this type of system resembles a fixed or dedicated assembly line with transfer mechanisms. The system achieves flexibility through the use of computer control and individually programmable, selectable stations.
An outstanding example of this type of FAS is the pump-assembly system recently delivered and installed by Alliance Tool Corp, Rochester, NY. Containing 98 stations, the system can assemble any of 64 pump configurations, each consisting of as many as 37 components. Top throughput rate is 1000 pumps/hr.
The third major form of FAS being offered today is the module. In this configuration, the equipment, tooling, and local controller for a given task or limited set of tasks--e.g., drilling holes, inserting screws, and driving the screws--are bundled into one unit. This may be contained in an enclosure such as a NEMA 12, or may be left open but protected by guardrails, light curtains, or other means. Modules can even be mounted on wheels.
A simple screwdriving module, for instance, may contain a robot, bowl feeder, track, and controller, all within a box. If you are bringing a new product into production, and it calls for stacking and mating parts, drilling holes, and driving screws, you could conceivably build your line by attaching one multipurpose module to the conveyor carrying palletized parts, or by attaching three simple, single-purpose modules.
Currently, both IBM and Pansonic tend toward different approaches. IBM units tend to be multipurpose while Panasonic leans toward the simpler designs.
"Ours is what could be called the Japanese approach,' says Michael McCraley, robotics project manager for Panasonic Industrial Co. "The Japanese dedicate a single station or module to one or two tasks only. This contrasts with the typical western or American/European approach, in which the FAS designer tries to have one or more robots in a cell replace a human in all aspects, performing 10 to 15 different tasks in one station.
"Furthermore, if you use assembly modules, you don't have to redesign the cell every time the model or product changes,' he continues. "You simply add, subtract, or exchange modules. The flexibility comes from this interchangeability, and also from fast-change tooling and local programming.'
McCraley also maintains that the use of modules will benefit users in lower equipment costs. "As the modules become standardized and production of them increases, unit costs will drop. Both producer and user will gain.'
GMF Robotics plans to design flexible assembly modules, too. "Like IBM, Panasonic, Toyoda Machine Works, and a few others, we see a promising future for modules,' says Guy Potok. "Alternators, pumps, motors: these and other products, in a number of industries, lend themselves to the modular approach.'
"If you look at how the Japanese are actually using modules,' argues Stanley Polcyn, "you see them in lines producing videocassette recorders, household fans, assemblies for major household appliances, and other consumer products. These are usually long production runs, not batches, and certainly not units of one.
"We think that the use of simple pick-and-place arms in a production run is not true flexible assembly,' he adds. "True FAS is for short or medium runs, where you want to minimize work in process and build the products to order. For this type of FAS, you need sophisticated controls, software, vision, and other advanced technology.'
What's in a name?
Semantics undoubtedly enters the picture; nobody has nailed down a firm definition of flexible assembly systems that everyone accepts. Part of the problem stems from the fact that flexibility can be gained in many ways: From people, programmability, modularity, fast-change tooling, nonsynchronous material-handling systems, selectable dedicated stations, and others.
"I think that to be called an FAS, a system must be programmable, expandable, modifiable, and capable of man/ machine interface,' says Bart Huthwaite, vice president and general manager, VSI Automation. "Total automation is not essential; you can have a flexible system in which robots, hard automation, and humans work side by side and interact.'
Huthwaite's company, VSI Automation, is a partner with Nitto Seiko, Japanese builder of scara-type robots, and he travels to Japan two or three times a year. "The Japanese are not as eager to fully automate the assembly functions as we are,' he reports. "They have the edge in labor, as their workers are both low-cost and diligent.
"We, on the other hand, have assembly workers who are generally speaking high cost and not as diligent. The only way for American companies to beat the Japanese in cost and quality, therefore, is to remove labor from assembly. That will come through the gradual, selective, increasing use of advanced technology.'
Much R&D still needed
The FAS being designed and installed today performs a wide variety of assembly operations. Somewhere between 70 and 80 percent of these are placement, stacking, and mating of mechanical parts, as well as insertion of screws and bolts.
Also critical to the overall assembly function, though, are the tasks of inspection and verification. This is where programmable sensors--machine vision and ultrasonics--play critical roles. These sensors, sometimes wielded by robots, are being used to verify that parts are present, measure deviations of holes and parts from their programmed locations, identify misoriented or bad parts, and verify that work has been performed correctly. Feedback to the cell or system controller is essential.
"It's important to consider how well the work is being done,' stresses Guy Potok. "Inspection and in-process quality control are vital to the FAS concept. That's one reason we need continuing development in gray-scale vision, high-speed data processing, and local-area networks.'
Potok and other insiders also think that much R&D remains to be done in feeding, sealant application, and fastener insertion.
For a long while, it was said that one technical hurdle that needed to be jumped was integration of data from both machine vision and tactile sensing. At least one company has made the hurdle successfully.
"We're starting to see more marriages of vision and tactile sensing,' says Keith Kerstetter, automation consultant with IBM's Industrial Systems Organization. "For instance, at our facility in Austin, TX, in a lab environment, we are using both vision and tactile sensing to identify and pick parts from a matrix.
"The coordination of data from cameras and tactile sensors enables us to egg crate the parts loosely and present them to the FAS in a random orientation.'
Costs still high, but so are returns
Thus far, the industries that have been adopting flexible assembly most avidly are automotive and electronics, along with aerospace. Not coincidentally, automotive and electronics are two industries that have been feeling much pressure from foreign producers.
In automotive, applications extend from component-supplier plants to subassembly plants, and on into final assembly facilities. FAS is now being used to assembly door panels, hydraulic pumps, air-conditioner controls, radios, speakers, valve trains, camshafts, cylinder heads, instrument panels, and other components and subassemblies.
In electronics, the FAS is employed in assembly of disk drives, printed-circuit boards, and other components. Among US suppliers, companies such as IBM, Automatix, Intelledex, Machine Intelligence, Microbot, Panasonic Industrial, and Unimation have aimed much of their effort toward this industry.
"One of the big challenges here is meeting the standards for clean-room work,' says Unimation's Stanley Polcyn. "We've become involved heavily in wafer handling and chipmaking with our Puma robots, but it has been possible only because the Pumas give off so little dirt during operation.
"For the most part, electronics manufacturers are not as concerned with removing humans from the clean room as with reducing the reject rate. It takes very little dust or dirt to spoil a chip.'
Regardless of which industry you're in, buying and installing an FAS is still a costly venture. Quotes ranging from $1/4 million to $1 million, depending on the FAS size and complexity, are not uncommon. "That's another good argument for the module approach,' says Michael McCraley. "You can get into it inexpensively, relatively speaking, and merge it with existing human labor and hard automation.'
According to Bart Huthwaite, the payback period for FAS has been averaging about three years, versus the two years or less for some other forms of flexible manufacturing. "The need to install new palletized conveyors is the main factor behind the longer period,' he explains. "Those three years shorten quickly, though, if a plant runs two shifts. Payback is also quicker in automotive, where labor costs are so high.'
Despite the high initial costs and relatively long payback periods, however, many companies are realizing significant gains from the use of FAS. One such company is a Japanese manufacturer of auto seat-belt retractors. By installing an FAS to assemble the retractors, the company was able to reduce the number of operators from 21 to four. When you consider the former high labor-content in retractors, that's quite a savings.
Many other benefits
According to users and suppliers, the benefits of adopting FAS include not only productivity gains and dollar savings, but also:
1) Higher, more consistent product quality.
2) Ability to build to order, to handle small batches economically.
3) Ability to change over quickly for new models and products.
4) Reduction of scrap and rejects.
5) Reduction of work-in-process inventories and related costs.
6) Ability to predict manufacturing costs more accurately because assembly costs are fixed rather than variable. "The cost of labor tends to inflate,' points out Guy Potok. "With FAS, though, once you've made the investment, the cost remains constant.'
7) In cyclical industries, the ability to recover more quickly from down phases.
"Some of the benefits are hard to quantify, but they're real nonetheless,' says Bart Huthwaite. "For instance, FAS demands high-quality, accurately made parts. The manufacturer puts greater demands on his suppliers of parts and materials, and that in turn reduces the indirect costs of inspection, returns, warranties, and so on.'
Much of what has been said and written about FAS sounds good, but many barriers--some of them significant--remain in the way of its more widespread adoption. High-investment cost is number one; lack of knowledgeable engineers is also near the top of the list.
Another barrier is the tendency in some companies to attempt retrofitting FAS in old, troublesome systems. "Quite often, and interested company will take on the most difficult tasks first,' says Michael McCraley. "That's a sure way to failure, and failure jeopardizes the future use not only of FAS but also of advanced technology in general.'
Even more fundamentally, many prospective users of FAS don't understand the basics of robotics, particularly the different robot anatomies, their capabilities and limitations. "Some FAS applications call for scara-type arms, others for Cartesian coordinate, still others for vertically articulated,' points out Bart Huthwaite. "No one anatomy solves all problems.'
Another barrier--and this may prove most critical--is a lack of understanding of and commitment to design for assembly (DFA). Experts point out that to get the most from an FAS's potential, the parts and indeed the entire end product must be modified or designed for automated assembly.
Several major US manufacturers, including General Motors and General Electric, have adopted computer software that gives accurate evaluations of parts for their suitability to flexible assembly. Three major software packages for DFA are now available and in use. One, called the UMass method, was developed by Professors Boothroyd and Dewhurst of the University of Massachusetts; the second is known as the Hitachi method; the third is the GE method.
Steps you can take
If you're interested in digging further into flexible assembly and FAS, you may want to consider obtaining literature from suppliers in the field; the box at the end of this article will give you a start. Another step might be to dig up published articles, books, and technical papers on the subject. The Society of Manufacturing Engineers would be a good source.
Experts in the field also advise that you do the following:
Take a long look at your entire procurement, warehousing, and manufacturing process, not just at isolated assembly stations. "The user should be aware of everything that happens to each part from the time it leaves the fabricating shop or the receiving department,' says Keith Kerstetter. "Also, plan ahead for what's going to happen to assemblies after they leave an FAS.'
Learn all you can about the various robot anatomies, and about what they can and can't do.
Send yourself and your key people to seminars on FAS. "Immerse yourselves in the technology,' urges Michael McCraley. "Learn its concepts and terms.'
When you think about FAS, think of lab tryout and familiarization first. Then, when you're ready to implement an FAS on the shop floor, keep it simple. "Pick an application where you're fairly certain you can succeed,' McCraley says, "before you tackle a tough one. You want at least one success behind you before you risk the more difficult applications.'
Start with simple, user-friendly technology only.
Bring your shop-floor people into the picture early, and get their input. If they feel they're part of the project, they'll sell themselves on it. Spring a surprise on them, though, and you're bound to meet opposition.
Flexible assembly is certain to become important in US manufacturing during the years ahead. As a potentially valuable tool in your company, the technology is worth at least a look.
Photo: Percentages of assembly systems in the US that will be programmable, will rise from 1980's modest 15 percent to 45 percent by 1990, according to the 1982 Delphi Forecast on industrial robots. The survey and forecast were conducted by the University of Michigan and the Society of Manufacturing Engineers.
Photo: At General Electric Co's power-meter plant in Somersworth, NH, a GE Model A12 two-armed robot performs assembly tasks. The relatively simple flexible assembly system has increased productivity in this operation by 130 percent, and has yielded an average annual savings of $80,000.
|Printer friendly Cite/link Email Feedback|
|Author:||Quinlan, Joseph C.|
|Publication:||Tooling & Production|
|Date:||Sep 1, 1984|
|Previous Article:||Nix on clicks, bet on chips.|
|Next Article:||Lasers advance the precision welding art.|