AMAZON multi-meters discounts AMAZON oscilloscope discounts
FIG. 7 shows another type of application for robotics and machine tending, which is its use as the master control for managing machining processes on multiple machines that previously have been serviced by more-dedicated forms of automation specifically designed to bring product to the machines and take away finished product, with a unmanned philosophy. Parts are held in tombstone fixtures and these fixtures are stored in a storage area where they wait to be retrieved in accordance with the robot's production schedule. As compared with more traditional dedicated automation such as this example and certainly many others (for instance, horizontal or vertical packaging equipment) robot systems benefit the firm in the following ways:
- Reduced floor space
- Six axes of robot motion that provide superior
- Future applications of the robot for other tasks
- Easy-to-adapt, custom tools for later use
- Significantly less moving parts, meaning higher mean time maneuverability due to the robot being more dexterous between failure (MTBF)
====
FIG. 7 Using the Robot as the Master Control Servicing Machine Tools and
Storage Retrieval System
Machining centers; Robot and Track; Part Transfer System into and out of the work-cell; Storage retrieval system
====
Automating the machining process whether on a single machine or on multiple machines, exploits the weaker human efficiency of simply performing the tasks. Human operators will rarely achieve zero defects and the side benefit for robotic automation is that zero defects can be achieved. It is an advantage to find one cell or more to automate and let an operator continue to service another cell where automation did not make sense because of the part mix, and the cost of the operator being spread across multiple systems and commodities. Profit per hour significantly increases with this strategy. With the addition of the robot operator replacing the conventional shop floor role in servicing the machines in one system, the shop floor operator becomes more versatile as well. Productivity per head and cost per piece often are enhanced.
When machine cutting times are long, for instance 15 minutes or more, exploitation of the robot efficiency versus manual efficiency loses some steam. Unless there are a lot of machines, as in a high- volume example to be covered in the next Section, it is difficult to take advantage of the robot's uptime because the machine requires attention much less frequently. A process like wire EDM is a classic example of a long cutting time. A single- or even dual-EDM machine process is not often automated with industrial robots because cutting time is usually in terms of hours. One advantage that can be exploited is re-configuring a system, with parts held in a storage retrieval system. With the long cutting times, a production schedule over a weekend can be set up. The robot has the ability to change the tooling (work-holding) on the machine, the part type being run, and all the programs, both at the robot and at the machine. There is some value in setting up a robotic machining system with this unmanned, long production duration, with or without changeovers.
If a work-cell is not automated with a robot, there are still many improvements available, including quick- change tooling, common fixturing attributes for handling, and storing programs on a database that can be remotely accessed. These examples of productivity tools that complement the machining to allow the work-cell to adapt should be used for the manual process but are also useful if the cell is automated.
How much capacity does a firm need to be able to justify the use of a robot operator for machine tending? The minimum conditions that need to exist include:
- Single shift of production if currently running manually (approximately 1400 annual hours)
- Product mix following the Pareto rule, meaning 20 percent of the product makes up a significant amount of the production hours
- The product is compatible for robot vision (i.e. has unique identifying features)
- The way product is handled is common across the part mix (i.e. slugs, shafts, cylinders)
- Existence of other areas to exploit that are motivations to automate (i.e. heavy parts, scrap problems due to mis-loading, and other tasks (i.e. deburring, marking, inspection, etc.)
- Miscellaneous criteria; safety, ergonomics, quality
- A sound machining process in terms of chip management and tool life
- Automating the welding process for small batch runs
A manually-intensive process is a process in which the operator has to perform the task of converting the raw state of the work-piece to a completed work-piece, such as with arc welding and material removal. Comparing these two procedures with manual machining, the machine performs the task of converting the raw material stage to the finished component, and the robot system exploits the inability of the manual operator to utilize the machine consistently to its maximum potential. The two new examples allow the robot to exploit an even larger portion of the time taken to change a part from the raw to the finished state. Production efficiency in arc welding and material removal is lower than in any other process including palletizing, where documented efficiencies in the 25 - 40 percent range. In other words, in a ten-hour single shift with one hour for breaks and lunch, the operator can contribute 0.25 to 0.040 * 9 hours = 2.25 to 0.36 hours of productivity. The rule of thumb is that a single robot will contribute productivity of the equivalent of three welding operators. With this large differential between robotic and manual efficiency, the aspects of changeover and setup can become irrelevant.
Systems with large assemblies might use a pair of tandem robots welding on an assembly at the same time, or a very quick welding process with a high deposition rate called tandem metal inert gas (MIG) welding. Tandem mig welding deposits weld metal at very high rates by depositing filler from two welding arcs at the same time. One arc is the lead and the other is the follower. Welding travel speeds is what drives the welding TAKT time. The tandem mig process can attain weld speeds significantly faster than a robotic single-welding-wire process, and weld travel speeds up to 200 inches per minute can be achieved. The potential gains in productivity with these tools can be imagined.
FIG. 8 illustrates an example of tandem mig welding.
Unfortunately there is a not an exact tool like tandem mig for material removal, but robotic material removal exploits other human weaknesses when it comes to this process. For instance, the gains in time and especially quality of surface finish in buffing, trimming, polishing, and grinding are great. Polishing times for impeller blades have been reduced from 20 hours to 4 hours time, utilizing a robotic solution, and the surface finish was consistent and in compliance with requirements. Additionally, the volume of polishing media is reduced, due to the even wear of the media that only a robot can achieve.
The demand for robotic welding systems is much higher than for machine tending for several reasons. First, the changeover setup is simpler for arc welding because the robot always uses the same tool, regardless of the part geometry, whereas, in machine tending, the robot requires a specific tool for each unique part geometry. The second reason is the perception that exists about how arc welding robots are more compatible with changeovers than machine tending. For example, in arc welding, a welding fixture is always required for conventional welding whether the fixture is designed for robotic tacking and welding or is a holding fixture designed for welding manually-tacked assemblies.
FIG. 8 Tandem Gas Metal Arc Welding- Lincoln Electric Model
The term "conventional welding" describes the classic procedure of an operator standing in front of a robotic work-cell, loading and unloading parts at either side A or side B while the robot is welding on the other side. An alternative way to develop a cellular approach for welding is to utilize a six-axis material-handling robot to pick, position, and present the work-piece to the welding robot(s).
Coordinated robot systems using multiple robots are rising in popularity because of the flexibility, and certainly the ability to adapt to future product changes. Additionally, robots have few moving parts so the material-handling device in terms of the industrial robot versus another type of positioning device will be more reliable.
If a firm has fifty part numbers to weld robotically, then there are in principle 50 fixtures. There is a cost for the design, hardware, and maintenance of each of these fixtures. In machine tending, work- holding changes may be needed for producing different parts, especially for a machining center, but a lathe would typically use a common chucking system with perhaps change out of chuck jaws.
In machine tending, the robot does not require a gripper for every part style but generally a few gripper styles cover a range of parts.
The point is that the equipment (device) used to secure a work-piece for robotic machine tending is less in content than the equipment to secure a set of parts for arc welding. In arc welding the perception is that it is acceptable in manual welding to have a welding fixture for a new part style and robotic arc welding simply follows the same precept. In machine tending, the perception of multiple part styles creates the concern that small batch runs are a challenge for the robot.
Product life cycles are also different between the two applications. Welding is a more mature robotics application and as a result there is a broader range of "standard" platforms, and acceptance among manufacturers. Because of the maturity of robotic welding systems there are many tools available to support changeover and to maximize utilization of the system.
In robotic arc welding, the weld fixture contributes to the most time-consuming task in changeover of a system. FIG. 9 illustrates examples of how weld fixtures can be quickly disconnected and connected back into the system. Once the weld fixture is under control, the rest of the components are simple to changeover. For families of parts that exhibit similar geometry such as box-shaped assemblies for electrical panels, computer enclosures, and electrical breaker boxes, a universal fixture can be employed to secure a common geometrical shape that either expands or contracts in length, width, and height.
FIG. 6 also shows an example of a universal fixture. Taking the fixturing changeover to the maximum flexibility would produce a programmable fixture that is configurable to whatever part comes next. Essentially the part surfaces that require to be secured in order to present the work-piece repeatability to the robot, are included within the recipe of every cycle. The fixture positions are programmable either through the robot or another device. The fixture locations also provide feedback to the system to validate that the fixture locations are in fact where they are supposed to be positioned.
FIG. 9 Examples of Managing Fixture Changeover for Robotic Arc Welding ---
Picture A Picture B Picture C
FIG. 9 shows three examples of ways to manage quick changeovers in robotic arc welding. Picture A is an example of a "picture frame" that is attached mechanically to the positioner as shown. The positioner is a headstock/tailstock, single-axis positioner, controlled by the robot. The picture frame serves as a mechanical structure that permits attachment and reattachment of another fixture shown in picture B. The attachment to the picture frame can be with bolts, or even quicker pneumatic couplings that allow more of a lock/unlock function as the fixture is located to the frame.
Quick-change fixturing is critical to maximizing robot utilization and certainly as discussed for other applications the machine itself, as with a press brake or machine tool. Picture C illustrates a series of fixtures that could be arranged as a progressive set of clamps that secure the work-piece in various orientations as welding proceeds sequentially from fixture 1 at the left, to fixture 2 in the middle, and to fixture 3 at the right. Each fixture has a base, and each base can be attached mechanically to a base plate that is welded or bolted to the table. As in the example of the picture frame, each fixture can be quickly located and locked in place; a new weld program is then selected and the job begins.
Another item to note in FIG. 9 is that the fixtures are located on one side of a turntable. On the opposite side of the turntable (not shown) could be a completely different set of fixtures with an entirely different set of products. The robot is programmed to know which side of the turntable is in position to be welded for a given cycle, and the appropriate programs are then implemented. The turntable has a capability similar to that of the machine tool pallet changer in terms of increasing efficiency as the load/unload time of the fixture proceeds concurrently with the robot process.
Examples of other aspects of the welding system to be set up are as follows:
- Robot programs. Consist of the file containing the robot motion sequence and instructions to execute a new process
- The welding procedure may need changing if the base material, joint type, or material thickness changes. The procedure recipe would be part of the robot programming, and automatically managed by the robot programming structure
- Changes to the welding consumables (welding wire, welding torch, and accessories (feeder drive roll, liners, tips). These items would need to be exchanged, and the welding process also must be changed because of the aforementioned parameter changes
The idea of loading/unloading the weld fixturing while the robot is welding is to balance the time between the two so that the robot time and operator loading time are designed to meet the throughput requirements. If the daily production rate is high, and the welding time is long enough, then the operator can be called upon to tack assemblies together, prior to loading the parts in the holding fixture, or to do visual inspection of finished welded parts, and service multiple weld cells. It is essential to avoid severe mismatch in load-unload time where the robot can weld the parts faster than the operator can feed the system. In that situation it may be possible to find ways to lengthen the weld cycle again, to balance the operator's ability to load-unload and perhaps perform other tasks as just mentioned. Often a single operator will be on the shop floor taking care of feeding several robot systems, and that is a good thing in terms of ROI.
=========
FIG. 10 Considerations of the Components Requiring Changeover for Robotic
Arc Welding; Side A; Robot program - robot positions; Welding Procedures;
Welding wire feeding system
=========
FIG. 10 illustrates a typical standard welding system and incorporates a shopping list of items to be considered for changeover when there is a product change. The system in FIG. 10 was designed to weld various "shaft" type parts. The work holding in this situation is similar to that for a machine tool work- holding fixture because the part family is all shaft based. There are two positioners to allow loading-unloading while the robot is welding, and the positioners again are single-axis headstock/tailstock, positioners, controlled by the robot.
Additional tasks that operators should complete while the robot is welding include preparing or completing hard-to-reach welds and/or small welds that may have benefited from fixturing but were not worth the expenditure because the welds were too small. An example might be a simple bracket that has only a couple of inches of weld run to the main body of the work-piece. By the time the operator has tacked the bracket in place, the bracket could have been welded to the main body without the cost of a fixture. The law of diminishing returns is often seen in welding, meaning at some point it's just worth keeping some welds manual because the company will never see any real economic benefit from robotizing every weld.
The shortcoming of a standard welding system is that an operator is always required to be present to feed the system. In other words, unless the welding system is fed by another form of automation, traditional standard welding systems will not support unmanned production. It is not a negative because again, the material handling and welding are occurring at the same time, and the weld productivity is 3 to 4 times greater than the manual process ever could be.
Custom (non-standard) welding systems (meaning that the part types do not fit a standard platform) are probably large assemblies.
Robot-welding of large assemblies or parts with long manual welding cycle times, makes a lot of sense because the welding cycle times are exploited through the robot efficiency and arc-on time, and no operator is standing in front of the system to load and unload material. These systems are more expensive, and can be more complex than standard systems, but their return on investment can be substantially quicker because of the increased labor that would otherwise be needed to meet the customer's demands. Imagine loading a tacked assembly into a weld fixture that normally takes two operators four hours, or longer to weld. The robotic system completes the cycle in about a third of that time, with consistent quality. That facility provides a huge competitive advantage to the firm.
The changeover involved with long cycle-time work-pieces also becomes less of a factor because welding is going on most of the time and there is no changing from part to part. What is interesting is the feedback: "We are not concerned about the setup or changeover time with my robotic welding system because we have so much increased throughput". Again, imagine the economies that can be gained with a single operator feeding two or more welding cells. Additionally, productivity gains make more time for torch maintenance (i.e. contact tip, torch reaming, and liner changes), which becomes minor in terms of the overall throughput.
Automating the material removal process for small batch runs
Moving over to the material-removal side of the process intensive applications, the equipment to perform these tasks always seems to require a higher investment than arc welding, and what is interesting is that operators in the cleaning area of the shop floor are generally paid less than a skilled welder. Material removal, in the author's opinion, is an application that should be automated simply because it 1) causes a high level of turnover on the shop floor, 2) ergonomics, and insurance concerns, and 3) has the worst operator efficiency. Additionally, in the home fixtures, and medical industries, material removal in the form of grinding, buffing, and polishing is what ultimately gives the product the quality surface finish that is the first thing the consumer sees.
The companies that manufacture medical components (knee and hip replacements), and home fixtures, automate the material removal process with robots. Their motivation is consistency, quality, safety, and throughput. The types of products seen in bathroom and kitchen fixtures are all over the board in size and configuration.
Even knee and hip components vary among patients. The robotic material removal systems used to buff and polish these parts are designed to use parametric programming to enable quick changeovers from one batch to the next.
The challenge in material removal is not only that the parts vary in size and shape, but also the individual parts from one to the next in the same batch are different because of the inherent surface tolerances assumed as the part was produced. In plastics there is shrinkage, in foundries and die casting there are varying parting lines, flashing, casting size, and risers. The tools for overcoming tolerances in welding are "through the arc seam tracking", touch sensing, vision, and laser guidance. The tools for coping with tolerances in material removal are active and passive force compliance, sensors and probes, and vision. The tools for the latter are very expensive to purchase. Additionally, higher levels of training, maintenance, and setup are required. Material removal is one of those applications with which the return on investment is either really good or a struggle. Material removal is most compatible with robotics in the following ways:
As a post-process event to the main process. For example, the work-piece is extracted from an injection-molding machine, a blow-molding machine, or a machine tool, etc. The investment in the robotic system was covered in the material handling or the loading of the machine. The incremental investment for additional functionality in deburring, trimming, or whatever is minimal
When there are large batch runs of the same or "like" family of part styles. This way, the process that is developed using the process tools is dialed in for all the parts after setting up one part. Then the programming becomes much easier to manage than it is when programs are developed for many varying part styles. Additionally, parametric programming can be implemented.
The work-piece is large and/or manually time consuming, allowing exploitation of the robot efficiency to become a significant financial benefit
Where the discussion of quality and visual appearance previously mentioned are the consumers priority and they are willing to pay for the quality of finish, or the firm cannot afford the costs of scrap as a result of inconsistent quality
FIG. 11 Dedicated versus Flexible Material Removal Methods --- A dedicated
approach to material removal; A flexible approach to material removal
There are alternatives to robotic material removal that do excellent jobs of removing specific features of a work-piece, and there are dedicated machines that remove a specific amount of material for specific part geometry. Dedicated devices are ideal for higher, more-consistent volumes, or like or similar parts. Both forms of dedicated and flexible automation have their places. If the removal process is the same every time for a family of work-pieces, the value of the changeover capability and flexibility of the robot offers less value to the application. FIG. 11 shows examples of robotic and dedicated automation designed to perform material removal operations on engine blocks.
The same "recipe" methodology and parametric programming can be applied to material removal systems, just like any other application. The scope of work covered in Section 5 used the robotic cleaning of engine cylinder heads as an example for outlining a scope of work. For the engine-head cleaning system, the following tools were used to support changeover for the varying casting styles and management of tolerances:
- Vision to identify part type that automatically configures the system process recipe
- Probes to identify casting surface locations
- Robot grippers designed with variable strokes to handle varying part sizes
- Robot grippers that incorporate multiple functional mechanisms such as pneumatically-powered fingers to grip a casting, and vacuum cups to handle corrugated boards without making a programming or gripper change
- A system control that serves as the coordinator to ensure that the process recipe for the incoming casting is ready for each device. When the cycle is over, the coordinator purges the recipe and prepares to launch another
Tool changers on the robot to pick up various cutting tools throughout the cleaning program. The robots exchange various milling tools and cup grinders automatically as they process each casting
Vision to identify location and orientation of each casting as it is presented to the system, providing a reliable method for handling and loading process equipment
Figures 12a and 12b show some aspects of work-cell component arrangements that facilitate changeovers used for small batch runs. The same plan is used, regardless of the application, when designing for changeover. There is always a system recipe that controls how the process will be run. Each part style requires its own process recipe, and variations in the part characteristics as a raw part or as a complete part, will proportionally affect any changes to the recipe:
|
|
FIG. 12a Considerations of Changeover for Robotic Material Removal --- Robot
tool changers Types of tools Robot paths Process recipe (parameters, procedures)
==========
FIG. 12b Considerations of Changeover for Robotic Material Removal
=========
Automating small batch runs for press tending and palletizing
Before examining the tools that enable industrial robots to behave like human operator in handling small batches, and the associated system changeover, two other applications that have been only touched on will be reviewed. The first is robotic press tending.
Press brakes and stamping presses are designed to form or change metal parts that start as flat blanks. From the robotic point of view, the flat blank is nice because it allows a vacuum or magnetic gripper to handle many style parts easily. Changes in the base material, for example, from steel to aluminum, will require a different gripper because a magnet gripper will not work on non-ferrous metal.
In press tending the next biggest contributor to cost in having to change out a gripper is the pattern in the blank because, prior to bending, the blank is often processed in a turret punch machine that uses a laser to cut out the hole pattern. If the hole pattern is substantial, there simply may not be enough surfaces to grip with vacuum cups. Another driver that would affect gripper selection is the size difference in blanks. For instance picking up a 24 x 24-inch blank weighing 8 lb. is different from picking up a 40 x 60-inch blank weighing 134 lb. Palletizing is similar to press tending in that the robot is being asked to grip a common surface side, usually flat, in order to pick and place the work-piece throughout the process. An engineer who is designing grippers will usually overlay the cases, in the palletizing example, or overlay the blanks in press tending, to see how much overlap and shared surface area exists among the product styles. Through a process of elimination, gripper selection becomes an exercise of matching products that can use the same grippers.
Differences in base material, or between buckets , bags, and cases, immediately separate each product from the group into an individual entity. Glass handling is another separate item in this discussion.
Press tending can certainly become more complex than pick/place palletizing because the process causes the work-piece to be formed into a shape different from when it started. Additionally, the work-piece may need to be handled several times at different locations to ensure that the robot grips the part without interfering with the brake or press tooling. The brake tooling frankly is the biggest opponent to changeover for a manual or robotic process.
Thus, changeovers associated with press tending are a function of the machine setup.
In an earlier machine tending exercise, the changeover was a wash between the robot and the manual operator, but the robot would exploit running production through lunch and breaks, and exhibit an efficiency that exceeded the manual process from between 40-50 to over 90 percent. Additionally, the press-tending system can be set up with a new stack of blanks at the end of a shift for the robot to work through, so that, at the end of every day, the firm could see 20 to 50 additional parts produced, essentially for free, with the exception of the utilities cost.
Press brakes can be configured with a long enough tool bed to allow for multiple punch and die setups to be installed in the press without the need for changeovers of one tool setup for another. The tool setup often is the limiting factor in press-tending efficiency, not the robot system. Additionally, manufacturing machines are similar, and it is in a firm's best interests to pursue a practice of quick- change tooling for any type of process equipment.
Small batches are ideally suited for applications such as palletizing and press tending. However, it's the production machine's changeover and not the robot that makes possible an even higher efficiency through quicker tool changes. There are readily available solutions for quick change tooling in press tending, and if the press bed is wide enough the bed will handle multiple setup's so that no changeover is needed. FIG. 13 illustrates components involved in changeovers in a press tending cell.
FIG. 13 Example of Components Considered for Changeover Press Tending
FIG. 14 illustrates various types of press tending work-cells, all designed to suit the way in which material is transferred into and out of the work-cell. The differences are affected by how much unmanned production is desired, as well as how the system manages the running of different part designs. For example, there are multiple pallets on the floor for incoming and outgoing material.
Each pallet represents a unique part style, enabling the robot to form blanks for a kit or a setup. A setup would include specific quantities of varying parts that make up the kit. The pictures in FIG. 14 show systems with and without conveyance. The conveyance permits a much longer raw material queue, for longer unmanned production. The robot track enables the robot to become even more flexible, because its reach is considerably greater when the robot is not lagged to the floor.
=====
FIG. 14 Multiple Examples of Robotics Press Brake Tending Cells
Robot on track multiple pallets on floor for material in and out of the cell Floor mounted robot with product conveyed in and out Robot on track, product conveyed in and out Robot on track with material in and out on pallets Material fed by gantry and finished material on floor pallets
=====
The following example compares conditions for changeover between a manual and a robotic press-tending application. The example includes the following conditions:
- Seventy part styles
- Parts are made in kits. For this example, five (5) part styles are selected:
Part 1 60 pieces using tooling A
Part 2 60 pieces using tooling A
Part 3 90 pieces using tooling B
Part 4 120 pieces using tooling C
Part 5 60 pieces using tooling D
Part 6 120 pieces using tooling D
- Parts 1-3 are aluminum, and part 4 is steel
- Parts 5 and 6 are stainless steel with a plastics coating on
- 230T 10-foot bed press brake, 6-axis back gage
- Daily production is 16 hours per day - 2 shifts
- One operator per shift
- Production halts at break time and lunch - one hour per one side of the blank shift total time. So available manual production is 7 hours per day
- It takes all day, including tooling changes to make the kit
- There are four (4) changes of brake tooling per day
- Customer does not use quick change tooling on the press brake so a typical tooling setup is 40 minutes to retrieve the tooling, locate tooling in the press bed, and test a part
- Profit per piece or kit is unknown and not factored into this example
- A single operator services the brake and the heaviest blank is 35 lb.
What can be forecast, using this example:
- In a 16-hour day the amount of setup time is approximately 160 minutes, so that there are 1280 min. (total daily time) - 160 min. (setups) - 120 min. (lunch breaks) = 1000 minutes per day available for production
- 452 pieces are formed per day
- On a given day the throughput in 16 hours of production is 1000 minutes/452 pieces = 2.21 minutes average forming time per piece at an estimated 55-percent efficiency. The 55 percent is the estimated manual efficiency when the operator is picking blanks, forming, and then placing finished parts on a pallet. The operator goes to the bathroom, gets something to drink, talks to others, and generally will inevitably waste 45 percent of the available production time
- The robot will operate at 98 percent efficiency within the available production time because the robot is built and designed at 98 percent efficiency
What could this user do to increase productivity and lower costs:
- Quick change brake tooling would decrease setups from 40 minutes to 10 minutes. With four changeover's daily, this savings would gain daily extra production of 120 minutes
- Purchasing a press brake with a longer bed would enable multiple tooling setups to be accessed at any time without changeovers to tooling in the first place
- Compared with the 98 percent efficiency of the robot tending the press brake,
the following results can be achieved:
- 1000 production minutes yields 452 pieces manually at 55 percent
- 1000 production minutes yields 804 pieces robotically at 98 percent
- The additional production that can be gained by running through breaks on a 2-hour shift, and running the system with additional products on the off or third shift can be as much as an additional 100 plus pieces
- The cost savings to the firm is the labor of two shifts, and the ability to reduce one shift, while achieving the same throughput using the same press brake
- Quick-change tooling, or a more flexible press brake, in addition to the robot tending, would provide even further capacity on a single-shift basis
What does it mean to the firm to remove an entire shift? The manual process efficiency could certainly be increased if it was felt that the operator would really produce at a higher level than 55 percent, day after day for a total of 250 days per year. That scenario would certainly offer a valid counterpoint to justifying robotic automation, but the reality is that the press brake asset can not be utilized on an equal basis by a robot process versus a manual process.
The return on investment in this situation is completely hypothetical. It does not matter here because the point was to show that even with an extreme amount of changeover, the robotic system along with some efficient tool-changing methods and/or right-sized equipment can double output with less labor content, improve quality, and provide a dependable operator every day.
FIG. 15 Robotic Press Tending System Designed for Single Piece Flow
One of the most interesting projects relative to press tending is illustrated in FIG. 1 5. This robotic system provided the following functionality:
- The robot is integrated directly with a punching machine so that blanks were cut from a master sheet, their hole pattern punched, and then conveyed to the robot forming system
- Over 200 part (sku) numbers were automated and caught in the system, using online programming
- A bar code label for part identification was placed on each blank as it was transferred into the robot work-cell.
- The part identification established the system recipe including the robot programs, robot tools, and press brake tooling that the robot exchanged automatically, and the press brake forming program. Every cycle used a unique recipe, providing a batch-of-one manufacturing process, for single-part flow with zero operator intervention.
- There were ten (10) robot end-of-arm tools designed to handle various shapes of parts and hole patterns in stainless, phosphate coated, and plain steel blanks, and to handle the press brake tool sections (there were four sections of the die and punch that needed exchanging in the event that a tool change was required)
- The system worked on a three-shift basis, six days per week
- The robots were chosen to handle blanks weighing up to 150 lb. (previously two operators were required to handle the heavy blanks)
As mentioned, many material-handling applications can be designed to handle single-piece flow, and the part characteristics drive the level of flexibility around which the robot system must be designed.
Automating the palletizing process for small batch runs
Palletizing for small batch runs is also a fairly straightforward process. FIG. 16 shows examples of end-of-arm tools designed to accommodate various vacuum and mechanical methods for picking up things for the purpose of palletizing. Palletizing robots use end-of-arm tools that handle pallets, cases, buckets, bags, divider sheets, etc. The robot control system tracks how every layer for each unit load is palletized, including the product patterns and when divider sheets are required between product layers.
Palletizing lends itself nicely to parametric programming. After the robot is taught one position for the first product placed on a pallet, that information can be used as the reference to offset every additional product placed throughout the unit load.
There are situations where multiple pallet types are required to be used, and the system needs to be designed to transfer varying pallet sizes into and out of the system. Where the robot is palletizing products having significant physical differences that affect how they can be picked and placed, the robot may be required to swap gripping tools, or pick up an adapter that is attached to the primary gripper but allows for picking something new.
Palletizing robots have no problem building a partial pallet where there isn't enough product to build a complete unit load at the end of a batch run. In the reverse, robots can also begin palletizing on a partial pallet that was left over from a previous production run. In this situation, sensors, vision, or a manual input may be needed to establish the starting point for the robot to begin its task.
By application, robots will exhibit varying numbers of cycles per minute which ultimately affect throughput. For case palletizing, the robot will typically exhibit 10-14 cycles per minute, depending on the application, and where the pick/place locations are located in the work-cell. The pick/place locations are related to where product is transferred into the system, and where the pallet is located.
Additionally, the robots take longer for palletizing at the beginning and end of the unit load because of the distance for the robot to travel, relative to the elevation of where the product is entering the system. Some users will use a robot servo-controlled, scissor-lift to maintain a constant height location, to maximize robot speeds throughout the palletizing process. Bag palletizing can attain 20 plus cycles per minute, given the fact that the location of bags coming into the system is generally very close to where they are palletized, and the tool acts much faster when gripping the product versus vacuum and picking cases. On the other hand, if the product can't be picked on the top surface so that the robot must use a fork tool to lift up the box from the bottom surface, the tool will take longer to operate than the vacuum system, and the robot's cycles per minute will be less.
====
FIG. 16 Examples of Functional Robot Palletizing End of Arm Tools
Multi-Case Gripper Bag Gripper; Bucket Gripper
====
Optimizing pick/place locations in the work-cell, that is, reducing the amount of motion the robot needs to accomplish for a cycle is very important. The slowest axis of the robot is the waist axis, and in palletizing every second counts considering that the robot may palletize thousands of cycles per day. The robot is capable of picking and placing something at every cycle t (products or a pallet, sheet, etc.). Using the robot to pick something other than the product obviously reduces the rate, but also provides greater flexibility.
The question may be when to use an automated pallet dispenser and when to have the robot pick pallets from a storage stand or use an automated divider (tier) sheet dispenser.
Throughput rates in robotic palletizing are based on how much product the robot can effectively pick and place to build the appropriate pattern. The robot is capable of picking a single product, multiple products, and if needed an entire layer or layers of product.
When examining many part numbers with varying physical sizes the tough process is to provide a universal robot tool that will handle all product without changeover of the tool. The quantity of product to be picked for a given cycle then is a function of the pattern.
A columnar pattern of 2 rows of 4 cases for example, is much simpler and quicker to palletize than a spiral pattern of 8 cases. The more varied the orientation of the products as they are presented to the system, as compared with how they need to be placed on the pallet, the more the palletizing rate is reduced, because additional robot motion is needed to compensate for the orientation differences. Patterns vary because users are required to optimize packing density and unit load height and weight for transportation to their customers.
The following list describes the characteristics that affect robot palletizing throughput:
- The tasks the robot has to perform (pick pallets, divider sheets, number of unit loads being built at one time )
- The time required to take away a full unit load (pallet) and bring an empty pallet
- Orientation of the product coming into the system compared with the orientation pattern required on the pallet
- The quantity of product being picked at a given time, which could be a single unit or an entire layer
- The location of the incoming product (elevation from floor and location to robot) relative to where the finished product is to be placed
If the robot was de-palletizing, meaning removing product from a pallet, and then placing the product somewhere such as on a conveyor, throughput is also affected by the time for the robot to go and search where the product is located before picking it off the pallet.
Switching gears to non-conventional palletizing is a market that is increasingly popular in warehouses and distribution centers because the volume of product handling is extremely high, as well as the labor content. Non-conventional palletizing as it is called, deals with palletizing random product sizes that are fed to the robot in single-flow fashion so that one product after another can vary in physical size and weight.
FIG. 17 Random Order Palletizing Examples
There is another challenge in robotic random order palletizing, and that is building a stable load that will not tip over as the pallet is transported. FIG. 17 illustrates an example of a randomly-palletized unit load. Random-order unit loads are used in the real world in the following ways:
Unit loads consist of a set of individual orders of products.
The unit load then must be built to support the last order to be delivered at the bottom of the pallet and the first order to be delivered on the top. The unit load then is designed to handle small partial pallet orders
In the warehouse or distribution center, the random order mixed unit loads are built to satisfy immediate order needs so that the product coming to the robot palletizing operation is completely varied, being based on customer demand
Random-order palletizing is a great example of utilizing the flexibility of robotic automation. The robot has to grip any box size without changeover, as well as make decisions ahead of time to where the box will be stacked in an ordered sense, to plan for a stable unit load. The robot can't simply pick and place boxes at random. The robot receives information regarding the product that is coming, and has to make decisions on how the product will be placed. The robot's decision making software will have to decide:
1) how and where the product will be arranged on the unit load, and 2) does the robot hold back the product for a later palletizing time slot that makes the load more stable, while trying to optimize packing efficiency and stability of the unit load.
Tools for small batch and high changeover production
The robot itself is a programmable, adaptable device that mimics human motion. The limitation of the robot is its payload capacity.
Even its reach is unlimited, in the practical sense that the robot can be adapted with additional axes of motion if the standard six axes are not sufficient.
It was established earlier in the argument comparing manual versus robotic processing, that the changeover to the process excluding the robot is a wash between the two ways of doing something.
However, there is always the flexibility versus complexity and diminishing returns argument for automation. In other words, beyond the investment for a "basic" functional robotic system, for X amount of added investment, what else is gained in terms of reducing cost and increasing productivity. The robot itself is a completely flexible device, which is why the same robot model is seen doing hundreds of different tasks across a myriad of applications and industries. The robot is simply equipped with and surrounded by the tools necessary to perform the tasks, and these aspects, combined with the level of flexibility in the peripheral tools, are what enable the robot to manage anything from a single part flow to batches of thousands. The peripheral tools are the key to success for managing whatever flexibility is required within the firm's manufacturing philosophy. With the advent of lean manufacturing, producers of robots and peripherals continually develop tools aimed at lowering the time and cost for changeover.
Robots are inherently blind devices that repeat movement to their taught positions within plus or minus thousandth's of an inch, and that is a problem for the real world because normally nothing ever repeats. The risk, cost, maintenance, training, and implementation to enable this blind device to place a welding torch or a hot knife in the location it needs to occupy in space is the science. Loading the chuck in a machine tool with a part that has to be loaded radially within a few thousandths of an inch is certainly a nice challenge for someone who is blind. It can be thought of as having a human operator perform a series of tasks in terms of providing value to a workpiece while blind-folded and expecting a zero-defect product at the conclusion of the task. That is a problem.
In the words of "Amazing Grace," "I once was blind and now I see." Robotic vision among other sensory tools, has forever changed the world of the application of robotics, specifically in overcoming the aforementioned challenges. Vision is used across every type of robotic application. Examples are as follows:
- Validation of the presence or lack of presence of a feature (for instance, the hole was threaded or not)
- Orientation and location of a static or moving work-piece
- Character reading of a part identification as a 2D or 3D dot matrix
- Inspection of a feature characteristic (i.e. does the weld visual appearance comply with quality standards in terms of appearance, undercut, profile)
- 2D and 3D location of an object in space, whether it be located in a bid container or stacked in a random orientation in space
- Finding the starting location to begin a process, given that the tolerance for locating the work-piece is not repeatable Part identification by comparing a unique part feature to a table of features among a series of parts
FIG. 18 Example of Robotic 3D and 2D Vision
There are other tools that accomplish tasks similar to those that robotic vision can accomplish, such as probes, sensors, WID tagging, bar coding, and mechanical means of capturing a work-piece in a repeatable location. Every application is different, and in many situations, vision is used in conjunction with other peripheral tools that give the robot "sight" and "intelligence". FIG. 18 illustrates many applications of robotic vision.
The potential to implement robotics has dramatically increased as a result of vision because, not only can the robot itself become an intelligent device that was already flexible, but the level of investment to implement vision, versus more mechanical, rigid/fixed devices is significantly less. Changeover in a world of fixed mechanical equipment that is designed to enable the robot to overcome blindness is a challenge, and often not very practical for small-batch runs. Changeover now becomes a factor beyond the process "stuff' that the operator has to deal with anyway, like a tooling change on the press brake or other machine.
The cumbersome task of changing out hardware every time a new batch of product is to be introduced to the robot, so that the robot could pick the new part style versus the previous style is no longer needed. The investment in a robotic system is highly weighted in terms of how product enters the work-cell. Simplifying the part presentation with a universal versus dedicated means of presentation for a broad range of product styles is a major advantage for the user because there is no physical changeover.
Furthermore, what if the product could be presented to the robot exactly as it is shipped from the supplier, without any operator intervention? An example is random order bin picking of product that is randomly located in a container newly arrived from the supplier.
FIG. 18 illustrates an example of random bin picking.
What if there was no welding fixture and positioner but instead a material-handling robot serving as the fixture and positioner, presenting the work-piece to a separate welding robot(s). FIG. 19 shows an example illustration of this idea. The material-handling robot and the welding robots in the examples shown in FIG. 19 are all controlled with one robot CPU, which enables movements on the six axes of each of the robots to be coordinated and controlled together. The flexibility of this system through the use of tool changers, and vision for the material-handling robot, would enable a work- cell to be configurable in positioning and securing the work-piece, which are the two of the areas that make the largest contribution of hardware in a robot system. Reducing the hardware not only makes the system more flexible, but also more reliable, with less moving parts and maintenance. Robots servicing robots to accomplish a manufacturing process is the ultimate, flexible work-cell.
FIG. 19 Robots Serving as the Positioner and the Fixture
In examining a robotic system in terms of what the robot can accomplish, the following criteria are used to outline the system component and tools to achieve single-piece work flow.
Component Category -- Robotic Tools and Technology
[ Inbound raw material to system
End of arm tool
Programming Fixturing ) (welding, material removal)
Positioner (work-piece manipulation)
Robot hand-shaking with peripheral devices Safety
]
[
Robotic Vision Automated part identification (i.e. bar coding, WID) Sensors or probes, internal or external to the robot
Programmable gripper
Universal gripper
Tool changers with multiple grippers
Multi-functional gripper that accomplishes multiple tasks (i.e. cutting and welding, or picking pallets and cases)
Parametric-based programming for a process that can be applied to one or more part styles/models Robot paths automatically generated from another source Programming is typically sub-routine based for each process step. Routines can be automatically turned off and on, depending on the recipe Programmable fixturing controlled by the robot or external control Using a robot to hold the work piece versus an external device Using a robot to position the work piece- Programmable positioners that adjust for tooling length, or swing diameter controlled either by the robot or by external controls Changing the process program or a parameter at a peripheral machine/device is standard and changes can be made on the fly by the robot system The robot can be dynamically zoned In on a program change and peripheral safety devices such as light curtains and banners can be dynamically controlled to limit access by an operator
]
There are other components within the robotic system that are classified as peripherals. A generic list of peripherals is as follows:
Bowl feeders Stretch Wrappers Stamping presses
Conveyor Machine tools Gages
Press brakes Inspection equipment Shuttles
Dial tables Case erectors Lasers
Feeding equipment Pallet dispensers Punch presses
Washers Shears Grinding backstands
Part fixtures (tack and hold) Die casting machine
For the sake of argument these peripherals are out of control of the robot, relative to managing changeovers of these devices. When it comes to the process equipment used in feeding, or performing the processes on the work-piece, these components of the robot sys- tem are limiting factors in achieving single-piece flow. For instance, an assembly operation for an engine cylinder head requires the assembly line to assemble up to 14 part models in a single-piece flow design. For instance, one of the processes is inserting guide tubes into the machined cylinder head casting. Guide tubes are bowl-fed to the robot for the assembly. In achieving single-piece flow, the work-cell that inserts the guide tubes requires a unique bowl feeder for every style of guide tube, of which there is a total of eight (8), requiring 8 different bowl feeders in the work-cell to cover any possible demand based on any cylinder head coming into the system.
=====
FIG. 20 Examples of Peripherals in Robotic Welding Systems Requiring Changeover
for Small Batch Runs
=====
FIG. 20 illustrates some other examples of peripherals that may need to be configured in the system for small-batch runs.
As noted, process equipment has its own work-holding or tooling configured for the part style being produced, and will require changeover at some point whether the equipment is serviced manually or robotically. Depending on the application, the robot is capable of performing an exchange of tooling or work-holding in the process equipment, as in exchanging the punch and die tooling for a press brake. The robot exchanging a 12-inch chuck in a lathe is another example that is probably not as practical, but with tombstone fixtures used to secure work-pieces in a machining center there is increased demand for using the robot capability to change the work-holding at the machines in addition to loading and unloading work-pieces. An example is shown in FIG. 21, where the work-holding fixture is common for all part styles, enabling use of a single gripper style for the robot to handle any part. In both FIG. 21 examples, the simplicity of the work-holding and the fact that it is common for the robot to handle, the systems can store hours of production containing a myriad of part styles.
FIG. 21 Examples of Flexible Work-Holding
Manufacturing changeovers to meet varying customer demand should be viewed as an opportunity more than producing exactly the same part every time. Changeover is often "a wash" between manual and robotic operators, so time is even more precious in terms of efficiency and maximizing utilization of the firm's assets.
Someone has to perform the changeover. The implementation of robotics eliminates the possibility that the changeover will be configured incorrectly. The future of the application of robotics is being driven toward single-piece flow with zero or minimal changeovers.
There are countless examples, and this section showed a few of the robotic systems that have been designed for small-batch and low- volume production.
Keeping the robot busy across at least a shift's worth of production is important in terms of achieving a reasonable return on investment. Not every program is the right program to automate, especially when the complexity of the parts doesn't fall into the Pareto 80/20 rule; care needs to be taken to avoid automating these types of programs because the flexibility requirements are not justified.
Finding programs that can be automated is a good exercise for a firm looking to implement its first robot system, to get rid of the dedicated hardware and the traditional model and move to programmable and agile equipment designed to adapt to real world tolerances and a robot system that can run unmanned small-batch production.