The ability to accurately simulate on a computer the operationrobots, transfer machines, conveyors, machine tools, welding machines,and controls under actual operating conditions can be a valuable toolfor analyzing manufacturing operations. To be able to then integratethese elements into a simulation of an entire manufacturing systemoffers significant economic benefits. Manufacturing simulation can be used to identify and solvemechanical problems before machines are built, optimize plant efficiencyby investigating alternative approaches, select the “leastcapable” (least expensive) machine for a given task, anddemonstrate the integration of an entire facility.
Such a simulation capability has been developed by MechanicalDynamics Inc. Two programs, Automatic Dynamic Analysis of MechanicalSystems (ADAMS) and Dynamic Response of Articulated Machinery (DRAM),have shown their ability to simulate the large displacement ofmechanical systems. These programs determine displacements and reaction forces ofmechanisms under actual operating conditions. This includes theapplication of large forces that drive the analysis into the nonlinear domain where finite-element methods either do not work or havediscontinuous effects. Both programs can simulate mechanical stops,friction, nonlinear component characteristics, impact, and largemotions.
To use this system, the engineer creates a model of the mechanismbeing evaluated using standard joints and geometric elements, and thenexamines its behavior. Both tabular and graphic outputs are availablefor study and analysis. Of particular concern in the simulation of whole plants are robotsand transfer-type machine tools. The ADAMS program can match individualrobot capabilities to job requirements by evaluating the motionsnecessary to accomplish a given task. Examples of analysis In fusion welding, for example, differences in joint gap or fitupcaused by manufacturing variations are a major problem for robotwelders. Common solutions package a sensory feedback device–likevision, magnetic sensing, or electromagnetic tracking–with a verymobile robot in hopes that this will cover all the possible weldingsituations. By examining the range of seam configurations and the motionsrequired to weld each of them, an ADAMS-based analysis can select themost cost-effective robot for the job. In many cases, this has provento be a less capable and less costly unit than was initially considered.
A relatively simple vision device can scan the actual joints prior towelding, and the robot control can select the appropriate weld-tip pathfrom a preprogrammed library to reliably weld the parts. Painting is another common robot application where simulation canmake a productivity contribution. Requirements like reach, number ofaxes of freedom, spray pattern for the desired coverage, and correlationof motions to avoid collisions can all be defined and evaluated on thecomputer using an ADAMS-based simulation. Even the displacement caused by the reaction force of thepaint-spray jet can be accurately predicted.
Actual paint coverage iscalculated by a simple algorithm, and the simulation pinpoints potentialtrouble spots in the automated painting operation. The result of the analysis may be that a different spray pattern isrequired, or that a product-design change should be made. Whatever thecase, simulation of the painting operation can often catch qualityproblems before they end up in the actual product. Deflection is also a problem with certain types of lasers used forwelding and inspection. Suspended at the end of a robot arm, theselasers can generate forces that cause significant displacement.
AnADAMS-based simulation predicts the pattern of this deflection andallows the robot control to compensate for it. This is particularlycritical in laser-inspection applications where any variation in laserposition produces incorrect readings. ADAMS simulations are particularly useful in inspectionapplication. By examining a number of different approaches to theinspection task, it is possible to find a solution using a less capablerobot than was originally selected. Interferences between the robot andworkpiece can be predicted. Preventing just one collision between a$50,000+ laser and the part it is measuring would pay for the computersimulation many times over.
In assembly situations, simulations can predict the deflectionpattern of the robot over time and affect the selection of theappropriate robot for the task. Again, this may result in choosing aless sophisticated unit than might have been specified. Machine loading often requires complex motions. By examiningalternative approaches, computer simulation can often eliminate problemsand production bottlenecks before they happen.
When more than one partis handled, the program can help optimize effector design by determiningthe least expensive device that will effectively grasp all partconfigurations. Computer simulation offers another unique capability. By running asimulation faster than real time, they can be used to predict wear andfailure modes.
Fed back into the design cycle, this information can beused to yield performance, reliability, and end-user productivityimprovements. Fewer changes The design of complex metalworking machines has always been aprocess that is more of an art than a science. A significant portion ofthe final cost of these machines is attributed to all the engineeringchanges that are required to make the machine perform as expected. Simulating the operation of a transfer machine on the computergives the machine designer an opportunity to identify and eliminate manypotential problems. Factors like machine/workpiece interference,distortion caused by excess clamping force, and vibration can beexamined and evaluated before any commitment is made to hardware.
Metalcutting machine tools can benefit greatly from accelerated wearsimulation to predict both accuracy drift and failure modes before themachine is even built. Obviously, the ADAMS and DRAM programs are not the onlycomputer-based tools for addressing such problems. Finite-elementanalysis is widely used to optimize part designs. The problem with FEA is that these methods only work for displacements within a very narrowrange, whereas ADAMS and DRAM evaluate a system over its full operatingrange. Examining whole plants In many ways, a transfer machine is a micro example of whole-plantsimulation.
The interaction of its various stations and transfermechanisms as it processes parts is very similar in principle to whathappens in a real manufacturing operation. All that’s missing arethe detail operations like welding, assembly, and inspection. Infactories of the future, most of these functions will be performed byrobots. The ADAMS and DRAM programs have been applied successfully tosimulate most of the equipment found in a typical manufacturing plant.The next step will be to tie this experience together into a singlesimulation covering the entire facility. Before the first concrete is poured or the first machine ordered,alternative approaches to each requirement will be tested and compared.Each piece of equipment can be located for maximum productivity, withits capabilities exactly matched to its job requirements.
Withsimulation eliminating bottlenecks and optimizing work flow beforeanything is built, the result can be dramatic productivity improvementsand reduced costs. Computer-based simulations have held out much promise from the verybeginning, but the reality of simulation technology has never quitelived up to the need. Now, that appears to be changing, and this couldchange the way we approach manufacturing from this point on. For more information on the ADAMS and DRAM programs, circle E1.