Model-Based Automation & Control Creation

The next giant step in mechatronics-based design of automationsystems, at least for machine designers who haven't yet made the leap, could bemoving beyond 3D CAD as a machine design tool and into advanced simulation andanalysis. Most OEMs have invested in 3-D models of machinery components, butthere remains a major opportunity to use 3D for multi-domain simulation andmodel predictive control.

"Digital modeling and simulation is incredibly relevant toOEM machine builders," says John Pritchard, global products marketing manager,Kinetix motion control, for RockwellAutomation. "We have been looking at how customers are,or are not, using digital modeling and simulation to consider the applicationof automation to their machines."

"Companies may say they are doingmodeling and design in 3D, but mechatronics-based design has got to be morethan that. What we have been concentrating on is notjust modeling one aspect of a machine, but actually modeling the whole machine."

A key hurdle to doing this is howcompanies are using 3D CAD packages. Years ago, engineers began moving fromdesigning machines in 2D to using 3-D design software. Many machine builderswho have adopted 3-D design practices are content to simply use 3D CAD todesign what the machine looks like and how the components fit together.

"Many machine designers are not yet realizing theextraordinary benefits of advancedsimulation and analysis," says RachelJacobs, mechatronics design specialist for Rockwell Automation. "That's the portion of mechanical design weconsider part of mechatronics, using 3D CAD for digital modeling andsimulation."

Over the last three years, RockwellAutomation has been working with OEM customers who are using virtual prototypingto develop their machines. Pritchard says that OEMs who have made thetransition to digital modeling and simulation are finding value with the toolsin four basic areas: collaboration, innovation, reduced risk and reduced timeand cost to market.

Connecting the Disciplines

To compare and contrast with customers who are not usingmodeling, many OEMs are often designing in silos. The mechanical, electricaland controls designers are not really thinking about what the machine will dountil it all comes together, while a model forces collaboration earlier andleads to insight.

Pritchard says that, with respect toreduced time-to-market and cost issues, machine builders did say modeling tooklonger when designing their first machine using this new method. Everyone hadto figure out how to use the new tools and, because projects are front loaded,more time is spent upfront on the design. But on the next machine,
it was significantly quicker and they started to use modular design and pullpieces into the new projects. By the second or third machine, they had dramaticreductions in time-to-market.

"We see model-based design as a subset of graphical systemdesign," says Javier Gutierrez, senior product manager for control andsimulation at NationalInstruments. "Typically model-based design is all about software andsimulations, but at some point the rubber has to meet the road, and there's aneed to implement a hardware prototype or hardware loop testing."     

Gutierrez says that when customers do modeling design,typically there are two main components. One is the control algorithm and thesecond is the system that you are controlling, or the plant model. Oneapplication trend is customers using model-based design to design a controlalgorithm and deploy it into a PLC. Butthere are some challenges, mainly the processing speed of PLCs.

To test the generated code, users can reuse the plant modeland deploy into a real-time system that simulates that plant. An example isNational Oil Wells, where engineers were designing a new type of drillcontroller that runs on a PLC. Because the system needs to run offshore, thecustomer actually simulated the oil rig into one of NI's real-time platformsand connected that to the PLC to simulate a surge in the pressure or if the biton the drill starts slipping.

Another example of model-based control for manufacturing isTyco Electronics' use of model predictive control for manufacturing coaxialcable. There is more model predictive control in automated manufacturingbecause the control algorithm is able to control complex systems. With theprocessing performance that is available today, it's being applied to faster andfaster systems where it used to be used only in really slow processapplications.

Developing OEM Design Trends

"One trend we are starting to see is use of multi-domainsimulation tools," says Brian McCleery, product manager for clean energytechnology at National Instruments. "Basically the machine has a mechanicalstructure, and there is a desire to do electrical simulations and possiblyhydraulics control. So we are starting to see machine builders embracing theseco-simulation technologies to better design that control algorithm."

"The idea is that the better yousimulate your machine, the better you can define your control algorithm,"says McCleery. "Another trend is hardware-in-the-loop testing. Now that I havea machine developed, why don't I reuse the model to actually test my controlalgorithm?"

Rockwell Automation's Pritchard says thetransition from 2-D to 3-D design has already occurred at major global accountsand machine builders. Among larger machine builders, the transition to 3D iscomplete because they made the investment and some have gone on to domulti-physics modeling and simulation.

"Many small and medium-size machine builders have not madethat transition," says Pritchard. "My opinion is that the major issue is easeof use, and might even be about perceived ease of use."

Rockwell Automation has partnered with SolidWorks tointegrate Motion Analyzer, its motion system sizing and selecting software,with analysis tools found in SolidWorks. SolidWorks communicates with MotionAnalyzer through API (application programming interface) calls, which provide alive link between the two software packages. Users can create motion profilesin Motion Analyzer, and pass the profiles to SolidWorks to visualize how themachine moves. SolidWorks will then calculate the torque or force required tomove the load through its profile, which Motion Analyzer uses to size andselect motors and drives.

The Motion Analyzer software creates motion profile outputsthat users can "copy/paste" into rungs of ladder logic used to commission aprogrammable automation controller. Motion Analyzer's next step is to add an XML Exchange feature, which will simplify makingsmall design changes by automatically updating the code.

"In the past, the computer hardware needed to do this type ofsimulation was too expensive, but that's not true anymore," says Pritchard. "Itcomes down to how easy it is to use, and engineers learning aboutmultidisciplinary design."

He says that a key to greater use of digital modeling andsimulation for automation system design is getting OEMs to learn about what isavailable.

"There have been huge strides made in this field and, once ananalysis has been completed in a 3D CAD package, the users wants to know thatit is right," Pritchard says. "In the past, it has been hard to interpretnumerical data and difficult to interpret vector plots. But today, the data ismuch more graphically and visually easy to validate and users can correlate onthe screen with what your instincts are telling you. Building confidence in thesimulations makes it easier to interpret the results."