Old Skills, New Skills

After working as a semiconductor process engineer, Hank Hogan hung up his cleanroom suit and now writes about process control and other technologies from Austin.

Start With the Basics
While new technology promises greater capabilities and additional revenue streams, it's important to keep in mind that automation is all about controlling a process, says Jim Campbell, president of system integrator Viewpoint Systems, Rochester, N.Y. That fundamental truth plays a part in what basic skills his company seeks in new employees.

"We typically hire people who are in computer engineering, electrical engineering or mechanical engineering," he says. "We tend to avoid computer scientists because they don't get into the physical world very much. But just basic engineering or engineering technology is important."

Another illustration of the need to satisfy the basics comes from automatic packing system supplier JLS Automation, York, Pa. The company is known for solving problems in the food packaging industry that others can't, claims President Craig Souser.

JLS Automation uses ABB FlexPicker robots for its platform with as many as 20 in a system. In nine out of 10 cases, the solution involves a vision system, as the robots have to deal with an incoming stream of randomly oriented objects of different sizes.

Vision and networking expertise are important to the company's automation engineers, but so too is a grounding in the real world, Souser says. "Historically, we've done best with mechanical engineers, who could then learn about programming and the electrical side of things, as opposed to taking people with really, really strong software background or electrical background, and then trying to turn them into robotics engineers. If you don't understand how physics works and the mechanical world works, you can struggle in robotics."

Fortunately, mechanical engineers coming out of school have a good software background, Souser says. He expects the need for software skills in his company to grow, as the meat industry, for example, begins to embrace increasingly sophisticated automation.

A controls system engineer should know — and understand — the process being worked on, agrees Barrett Davis, partner at system integrator AutoMate of St. Louis. The company specializes in industrial automation and municipal control systems. "Indeed, the best situation is one where the controls engineer understands the process as well as or better than the customer," Davis says. He says this allows the engineer to answer three critical questions: "Are you getting the correct information? Are you getting it at the right time? And are you processing it to give the proper answer?"

When asked to sum up the skill set that's needed in an automation engineer, Davis says "A control system engineer isn't just a programmer. He's got to be a physics guy. He's got to be an electronics guy. He's got to be an electrician. He needs to be a programmer. He needs to be an instrument guy. He needs to be all that balled up into one."

Robert Trask is a TwinCAT 3 specialist with the PC control technology company Beckhoff Automation. He notes that another fundamental skill — that of fully considering a problem — can run up against the understandable desire to begin writing code and attacking a problem immediately. "Increasingly, we're going to have to be thinking about what we're doing before we start pounding." he says.

To illustrate the need for methodical preparation, Trask relates a story about Abraham Lincoln. When asked how to cut down the greatest number of trees in a six hour span, the story goes, the 16th U.S. president said he would spend the first two-thirds of the time sharpening an axe.

Figure 1: Modern automation programs and solutions can be modular, thereby allowing more complicated problems to be attacked. To do this, engineers need expertise in building blocks, such as higher level languages, simulation and mechanics, among others.
Source: Beckhoff Automation

Avoid a Wall
Trask says control software might be growing too complex for old programming skills to work. If programming does succeed, the resulting code could be too complex to maintain or document. The problem is similar to one that hit the banking and airline industries decades ago.

The solution implemented by these other industries was the use of object-oriented methodologies (Figure 1). A control analogy involves treating cars on a highway as individual objects operating autonomously within the same system. This can then lead to flexible and effective traffic control. This is an example of how to tackle control problems.

"Rather than thinking about one huge procedure, you start thinking about these little units of things, and they're responsible for their own behavior," Trask says. "The good news is that most machines can be divided up. We have a folder, and we have a wrapper, and we have a gluer. Those can act as distinct objects."

The object-oriented approach might be helped along by the latest edition of IEC 61131-3, the international standard that covers programmable controllers. Published in February 2013, this third edition of the standard adds extensions that incorporate a clever method for using function blocks in an object-oriented way, Trask says.

A consequence of this increasing emphasis on software is that engineers need to master new skills and acquire familiarity with industrial protocols and communication interfaces, says Christian Fritz, senior product manager for advanced controls and robotics at National Instruments. It also means that software documentation is more critical as programs are now an integral part of the machine. Handled correctly, this reality can bring benefits.

"Proper software design flow and documentation can make it easier to go through the process of certification," Fritz says. "Those are some of the things that automation engineers didn't necessarily prioritize in the past. But today, because the design is so much more software-centric, these are very important topics."

He adds that software is breaking down old barriers. For instance, there is the rise of mechatronics, which combines mechanical, electrical, control and computer engineering. The result is a system that's more flexible and powerful than a simple mechanical system. However, by necessity, this approach requires that traditional information silos between different disciplines be breached. Vendors are trying to ease the learning curve by incorporating system-modeling and other advances into their tools, Fritz says.

Software is indeed becoming more complex, and the underlying control systems are evolving, says Viewpoint's Campbell. The company focuses on measurement and control with a heavy emphasis on test and embedded systems.

Figure 2: A bolt-on controller improved the capability of this gear-lapping machine without a complete redesign. This solution required expertise in distributed communication and the use of programming building blocks instead of a monolithic design.
Source: Viewpoint Systems

Whereas, in the past, everything might be run from a central controller, today — and even more so in the future — systems can consist of intelligent sensors and actuators with distributed decision-making a possibility, adds Campbell (Figure 2). That means control software has to be carefully designed and implemented due to new demands. If an actuator or sensor takes action, then every other component needs to be alerted. This ensures that multiple control loops or state machines do not fight each other, which has the potential to create control chaos.

"Ultimately I think communication buses are going to take that over," Campbell says.

Tie Things Together
"A PLC controls other parts of the system and a touch panel PC coordinates and monitors everything," Campbell continues. "Eventually, they'll need to be synchronized, so the machine operates with tight synchronization instead of pausing and waiting for some message to appear. So, real-time communications are going to be more prevalent as time goes on."

As sensors and actuators get smarter, expertise with distributed systems will become more important. Add to that the need for skill in and knowledge of real-time communications. This combination makes it possible to avoid a priority inversion. "You can end up with a situation that a high-priority task is being held off by a low-priority item if you don't write the software properly," Campbell says. Such an event can tie up a communications bus or any piece of hardware to which the two tasks share access. Discovering these dependencies and debugging them after the code has been written and the system is supposed to be running can be a challenge, Campbell adds.

Issues surrounding real-time communications are just one aspect of needed networking skills. In automation as elsewhere, devices are growing ever more connected. Thus, it's increasingly important to have the skills needed to design, manage and maintain networks.

That can be seen in the products and services being offered by global machine tool provider DMG/Mori Seiki. Tomohiko Hayasi, manager of system design, says that engineers today must know more than just their company's metal-cutting machines. Due to automation and the use of robots for part handling, engineers also have to know about the surrounding peripherals, such as inbound and outbound conveyors, gauging stations, measuring stations, deburring stations and washing stations (Figure 3).

Figure 3: DMG/Mori Seiki U.S. demonstrates highly automated manufacturing, using robotics to produce a machine part. Engineers in charge of the project must be familiar with surrounding peripherals, such as inbound and outbound conveyors, as well as having networking expertise.
Source: DMG/Mori Seiki U.S.

At one time, all of the needed communication to tie everything together into one system would have been handled by discrete, hardwired I/O. That is no longer the case.

Engineers, therefore, have to be up to speed on applicable networking standards. Going forward, wireless is one clear trend. For machine and factory-floor automation, that brings up issues of safety and security. The latter is more than just a concern of the factory being networked. In some automation and control applications, the government gets involved. Integrator Jensen recalls working on the controls for a water-treatment facility in California, and being told that the security of the network would be scrutinized by the U.S. Dept. of Homeland Security. "They come in and inspect them and check out the security," he says. "Down the road, they're really going to start playing a big part in the policing of security."

Several other looming changes promise to demand control engineers pick up new skills. One is cloud computing, another is big data, and a third is the advent of IPv6, the latest Internet Protocol version. The three intersect, courtesy of the 128-bit address used by IPv6. Because of this, the new protocol has about 8 x 1028 times the roughly 4.3 billion addresses permitted by IPv4, the old standard. That means IPv6 will allow every and any type of sensor and actuator to have its own address, and this has control system implications.

"The amount of data is going to be huge," says Ben Orchard, applications engineer at Opto 22. "We get some analysis software built into that, and all of a sudden a building becomes this living, breathing thing. We've got data coming back from low-power, low-cost sensors and every sensor has its unique address."

He doesn't expect that this will be the case in all automation and control settings. Instead, it is most likely to be the case in new installations or complete retrofits. Consequently, automation engineers will need to pick up skills in database management and cloud computing, if only to avoid getting swamped by a sea of sensor data.

Cloud computing requires remote access and server-based applications. The first increasingly is accomplished using smartphones, tablets and other devices that users supply themselves. Knowing how to ensure the security of such access is a skill that automation engineers are likely to need.

As for the second, server-based applications will demand new skills. However, these might pay for themselves in a different way, according to Rod Koning, sales engineer at systems integrator Perceptive Controls in Plainwell, Mich. Using the cloud can lead to fixed costs, much like leasing equipment. This can be attractive to smaller operations that don't have the internal resources to support computer hardware and software. Organizations might go into the cloud because they want a predictable expense.

Those hosting cloud services will realize revenue, and that could lead to the need for a new set of skills for automation integrators, Koning believes. "Some system integrators could actually end up hosting [the service] themselves, putting up their own hardware, putting up their own infrastructure so they can create their own internal cloud, and use it to go after that market."

Back to More Basics
Carl Olson, engineering director at chip-programming solutions provider Data I/O, also notes the need for certain basic skills. For example, he points to root-cause-failure-analysis expertise, an important skill when designing and debugging a product.

The company's automated systems program the memory and other chips that provide the firmware found in cars, cellphones, set top boxes and elsewhere. Because they're sold globally, they have to meet harmonized safety standards, and that requires another fundamental skill engineers should have.

"The idea is that they can evaluate a system and basically do a risk assessment on it," Olson says. "How likely is something to happen? How can I detect it if it happens? If I can't detect it, how severe is the consequence going to be?"

At Rockwell Automation, attention to basics shows up in its external training, which is the responsibility of Georgene Berman, global product manager for training services. The company must train two different audiences, she says.

"The OEM, we think, starts with a blank slate, while a typical end user might work on an existing system, like by doing preventive maintenance," Berman suggests. "System integrators might fall into both camps by doing some code development and some code maintenance."

The best and most effective way to do such training is the old-fashioned, hands-on, instructor-led class, she says. That method is also the one preferred by an overwhelming majority of Rockwell Automation's customers.

But, the march of technology means that training eventually will be delivered on a mobile device with the content broken up into suitably small chunks. An example of how this might work is a troubleshooting procedure that appears on-demand on a tablet. Rockwell Automation has a pilot program in which a tablet is used for training. Some of the participants might not be familiar with tablets or their operation, so instruction includes not only course-specific content, but also how to use the device, Berman reports.

Training is something that can't be avoided, according to Perceptive Controls' Koning. Today, there can be classes offered through distributors. Another training avenue currently available could be at a tech or user group conference held yearly by a vendor.

"There are a couple of methods but the majority of it is you have to take probably about a week out of every year to keep your guys up to speed on the different products if you want to maintain that competitive edge," Koning says.