Bosch Rexroth Celebrates 40 Years of Business in the U.S.

Modeling and simulation can provide big benefits to users, so long as they take care to avoid the pitfalls.

Mechatronics refers to a systems-engineering approach to machine building based on physical and mathematical models and simulations.  Covering the mechanical, electrical, controls, and software aspects of a system, the approach provides a cost-effective way to explore design options and look for the best possible solution to a given set of design constraints. "The core of the mechatronic approach is to brainstorm possible designs," John Pritchard, Global Product Marketing Manager at Rockwell Automation (Milwaukee, Wisconsin). "You conceive the whole potential solution, you spend time modeling, and you build a virtual prototype to see the likely outcome. Obviously, simulation is not perfect but it’s certainly a great indicator.” In part I of this article, we discussed the theory. Now, let's take a closer look at how a mechatronic design approach is implemented and the pitfalls to avoid.

A good approach is to start with a goal and a first approximation, what Pritchard characterizes as a neutral design statement - for example, the machine needs to move a 5-lb load from point A to point B in a given time. Fundamental physical constraints exist that simply can’t be engineered around. The idea is to do a quick sanity check to ensure that the neutral design statement is realistic, then the real work begins.

The idea behind mechatronic is for the mechanical, electrical, controls, and software engineers to collaborate to produce a single integrated model of the whole system.  They model the payload, mechanism, motors, feedback, drives, and controls. The process might start with a mechanical design that gets augmented by the electrical engineers, the controls engineers, the software engineers, all in a collaborative environment.  Different elements may originate in different design tools, but it all gets brought together in a collaborative environment. Once the team has established a model of the entire machine from load through controls, they can begin to simulate performance and evaluate the effects of difference changes (see figure 1).

"You can play around with it. You can see what happens if you change the material of a bracket or the reduction ratio of a gearbox,” says Pritchard.  “You can see what happens if you add compliance, because two of the big things that can give you a surprise tend to be around compliance and backlash.” Even the best analysis of load and torque can become irrelevant in the face of unforseen dynamic effects.  “Say you have compliance in the system. A motor in the system can excite that compliance and you can wind up with the payload vibrating in a way that you didn't really imagine.  Using mechatronics, we can model that now and ideally avoid that scenario entirely.”

The process can sometimes provide a formidable number of candidate designs.  The key is winnowing down those candidates into a handful of choices worth pursuing.  Prichard cites a packaging OEM customer who used a mechatronic approach to explore changes to an existing machine design.  "They were almost overwhelmed with 20 potential solutions,” he says.  “They got some students to come in over the summer and build virtual prototypes of those 20 different machines so they could run them and then understand which were the best candidates.  I think in the end they narrowed it down and built two actual physical prototypes to see which one worked best.  The feedback from the customer was that they were able to fast-track to an optimized design and save over $100,000 in the process.  That has to be a shining example of how to do it.”

The key aspect of it all is that the team can comprehensively test their designs with minimum outlay (see figure 2).  “You can start having your conversation between the various disciplines of mechanical, electrical, control, and software engineers and say, ‘Look, under these operating conditions we see a certain resonance here and we think it's caused by the masses in this part,’ or ‘There's friction in the system causing a lot of hysteresis, we're getting overshoot on the mechanism,’” says Tony Lennon, Industry Manager for Industrial Automation at The MathWorks (Natick, Massachusetts).  “You can have a much more realistic discussion about where the design trade-offs need to be made - can we reduce the mass here, reduce the size of this motor, etc.”

With the right software tools, the testing can extend to a breadboard version of the controller running real-time simulations, or a computer model running on the actual machine with simulated I/O.  The idea is to first test different components with the machine individually, then graduate to integrated testing.  With a well-designed, comprehensive model, the simulation environment can accommodate all of this. Multiple groups of the team can work on their individual parts of the simulation that are then tied together with specialized interfaces.  The individual parts can be connected in a single environment which both simplifies and speeds up the integration process.

Challenging Times
Model-based design is a powerful technique, but like any other, it carries certain challenges.  Often, the biggest problem is not mechanical or electrical or software engineering, it’s the human factor.  The mechatronics approach runs counter to the long-standing, silo-based culture of most engineering organizations.  Whether in the private sector or at a university, separate departments typically exist for the various disciplines.  “It’s a bit of a standing joke,” says Pritchard.  "I was talking with a mechanical engineer about machine design.  I asked him if he talked to the electrical guys often and he said, ‘Not unless I have to.”  Even within a department like electrical engineering, the various sub-disciplines like controls or embedded systems may splinter out into their own groups.  In order to fully realize the benefits of mechatronics, organizations need to work to move past this silo mentality.

The process is already underway at the university level.  Around the globe, engineering schools are opening degree programs and even departments of mechatronics in which students are trained in a holistic, multidisciplinary perspective from day one.  They may not be an expert in any one discipline but they understand how the different areas relate to one another and the interplay between them.

An important element that sometimes gets overlooked is the inclusion of the product marketing staff and the design process.  Most products have cost and performance numbers specified from the very beginning.  During the optimization process, the design team may discover that they can meet cost targets but only at the expense of reducing the performance levels.  If the result is a slower machine that provides no competitive advantage in the marketplace, however, there is no point in going through the exercise. The inclusion of marketing staff at this juncture helps the design team understand which trade-offs are acceptable.

Integrated Tools
For all of the sophisticated software packages available, challenges remain in the implementation of model-based design.  Mechatronics requires a central modeling environment in which the mechanical, electrical, controls, and software engineers can all access the latest results and as a group determine how best to make trade-offs to optimize overall performance.  “Everything should be integrated, so that as soon as the simulation is done, everybody sees it and everybody has access to the results or to the raw data for possible design changes,” says Razvan Panaitescu, Engineering Manager, Mechatronic Support, Siemens Industry Inc. (Norcross, Georgia).  “That is where the true value of mechatronics really starts to show.”

In the silo environment, problems all too often result in fingerpointing.  If a controls engineer working on the controls for a machine finds instabilities, for example, he or she may blame the improper choice of actuators or linkages, or perhaps hidden resonances in the machine.  The mechanical engineer, meanwhile, might question the accuracy of the controls engineer’s model, especially if it leverages controls design tools.

If the controls engineer can use the mechanical data provided through the CAD model and simulate the dynamic behavior of the mechanical design by simply applying their control algorithms to the existing models, it eliminates some of those conflicts (see figure 3). Indeed, the right development tools can foster collaboration.  “The mechanical guys and the controls guys speak two different languages,” says Christian Fritz, Product Manager for Motion and Mechatronics at National Instruments (Austin, Texas). “Having tools that help them collaborate takes a lot of the pain away.  Integrated design tools provide a common language, and support development teams in optimizing the overall design by analyzing mechanical components, electronics and control algorithms at the same time.”

Typically, system and controls engineers use block diagrams to model mechatronic systems.  Ideal software would import mechanical data - component masses, inertia tensors, and physical topology of the linkages and joints - from a 3D CAD model into a block diagram model, Lennon notes.  “The block diagram model is used to simulate the machine dynamics in order to design the control system, unlike the 3D CAD model, which is used to analyze stresses caused by static and dynamic loads,” he says.  “The important point is that both models represent the same dynamic system because they use the same design data.”  The controls engineers can leverage this model to run simulations as part of the control system design process, testing out open-loop and closed-loop controls strategies.  The simulated control system can even be designed using a state model to allow machines to support different tuning parameters during different operating conditions.

In a perfect world, the models would allow engineers to enter the parameters of electrical components like motors and drives so that they can tune the motors fairly easily.  One challenge is that motor parameters may refer to steady-state conditions rather than those experienced by motors used dynamically for speed or position control.  A solution is to characterize a motor using a dynamometer to actually measure some of the values that can allow the dynamic model to be tuned.

Above all, the software tools need to be intuitive and fast. Machine builders operate under perpetual deadline pressure.  Unless software is relatively quick and simple to use, it is unlikely to be adopted by machine builders up against the clock.”

Converting Data To Intelligence
Of course, all of the simulations in the world are useless if you don't know how to interpret and act on the results.  The process of extracting intelligence from the volume of data in model-based design can be the single biggest challenge.  Certainly, software exists to optimize performance by running simulations with variations in motor size, for example, but leveraging intelligence beyond that simplistic level requires engineering skill.

“You need that new type of mechatronics engineer who can open up the simulation and say, ‘We have to do something about this particular spot or that particular component to improve,’” says Panaitescu.  “This is why many companies eventually give up using simulation, because they very seldom have that in-house capacity to immediately spot the root cause of error and give good suggestions. Instead, they rely on trial and error, or on experienced people who don't look at the simulations, they look at already-built machines and go with a screwdriver and fix the problem.”  The problem arises when even that veteran engineer can’t address the problem, or they move on to another company.

“Even if it does work, if you don’t understand why it works, when it breaks - and it will - or when the customer wants more features, better performance - and he/she will - you can’t deliver,” says Kevin Craig, Robert C. Greenheck Chair in Engineering Design at Marquette University (Milwaukee, Wisconsin).

Mechatronic modeling can help companies deliver, but only when the approach is leveraged properly.  “Everybody can simulate nowadays,” says Panaitescu.  “It's harder to find the person who can really interpret the results and correct the original mistake.” Companies need to focus on finding or developing engineers who can combine the modeling savvy of the recent college graduate with the intuitiveness of the veteran engineer with the screwdriver.  Such expertise may require time and money to develop, but the results are worth it.

“If you don't invest in your workforce now, and take them to this new level, you won't be innovative,” says Craig.  “With the economic situation, now is the time to invest in your workforce, now is the time to really change because when the economy turns around, you will come out of it with a competitive advantage.”