Some would call it a classic case of missed opportunity. A company building a distribution center (or perhaps setting up a new manufacturing line) buys simulation software. Week after week, staffers gather around the small screen to watch boxes whizz down a hypothetical conveyor and analyze the patterns traced by tiny Sims-type workers as they go about their virtual tasks. Overtime hours pile up and other projects languish, casualties of an all-consuming quest to design the perfect layout before a drop of concrete is ever poured.
Yet months later, when a newly introduced product line snarls up the shipping process, nobody gives a thought to the simulation software, now gathering dust on a shelf somewhere. That's a bit like using a couples counselor during the blissful prenuptial period but not to de-escalate the inevitable outbursts of marital strife in the months, weeks or years after the ceremony. It's also unfortunate. Solving problems like shipping bottlenecks is arguably one of the things simulation does best.
In its most basic form, simulation software takes data from your warehouse operations—picking, packing, material handling, racking and so on—and allows you to play around with different scenarios. Want to know what would happen to picking operations if you added a new conveyor belt? Curious whether a change in racking configuration would speed up the packing process? With simulation, you can answer these questions by shuffling around electrons, without including the protons and neutrons.
Indeed, simulation can be used for much more than simply rearranging the DC "furniture." Because computers have a boundless capacity for crunching data, you can model an entire warehouse or manufacturing plant, or both together. Proponents even suggest you could use simulation software to play around with designs of an entire international supply chain.
Yet that won't happen anytime soon. With its inherent complexity, simulation has historically proved a hard sell. "Simulation's … a highly technical subject and it remains that. It's difficult to sell at upper-management levels," says Jan Young, product manager at Catalyst International, a software vendor in Milwaukee, Wis. Another hurdle, says Young, is getting people in the habit of thinking of simulation when there's a problem to solve. "You can do a lot and provide a lot of value with the technology, but you have to understand its capabilities and how it fits in with the other technologies that are available," she says. That means managers have to become familiar enough with its capabilities that they'll recognize it when presented with a problem that lends itself to simulation.
Young notes that acceptance of simulation software in DCs is more widespread in Europe than in the United States. But Matthew Hobson-Rohrer says that's starting to change. Rohrer, who's the director of aerospace and defense at Brooks Automation in Salt Lake City, Utah, says his company's customers are branching out in the ways they use simulation. "There's a lot more activity in controls testing simulation," he says. Automated material handling systems in warehouses and manufacturing plants are usually controlled by one or several software systems—ranging from a programmable logic controller to the more sophisticated warehouse management system. Known collectively as control systems, these help keep track of where product is and help make decisions about how conveyors are used to merge or separate items.
"What we see customers doing is linking a simulation model to their control system, using it to test a control system and check [to make sure] it's robust enough before they actually install it," Rohrer says. "It becomes an emulator of the actual warehouse. They can use it to run all sorts of scenarios and can test control systems before they buy."
Emulation means running a test of a system by hooking it up to a decision-making software system and allowing it to run theoretically, but in real time. You present the decision software with apparently real situations, to test how it responds. "What it's doing for our simulation customers is allowing them to extend the model from design to function and allowing them to go toward [using it for] operations," Rohrer says. "They're looking at simulation as more than a planning tool."
Given the complexity, it's probably no surprise that the companies most adept at using simulation software to solve operations problems are often the ones that act as consultants for end users of warehousing and manufacturing systems, or those who design and install them. One example is E2M Inc. (pronounced ee-squared-em), a systems integration firm in Norcross, Ga. E2M and its sister company Polytron Inc. specialize in helping Fortune 100 companies design and operate bottling and packaging operations for their ever-changing products. "When someone comes up with a design, we're the ones who figure out how to run it and make it," says Geoff Mueller, simulation engineer at E2M's emulation modeling division.
The company started using Brooks' simulation software in 1999 and soon began to see opportunities to use it to debug programmable logic controller systems. As the simulation software became more advanced—using color and 3D displays as well as upgraded graphics—the opportunities for emulation became greater and greater.
For Mueller, emulation equates to an opportunity to debug decision-making software before it's set loose in a real warehouse or bottling plant. "You don't have to spend a lot of money to figure out if it's going to work," he says. He admits that emulation isn't perfect—it allows you to get perhaps a 90-percent accurate picture of what would happen if you plugged the software into a real facility's operations. "There's always going to be real world stuff that hits you," Mueller says. "But it's better than before, where you would go in cold, maybe 60-percent ready."
The main benefit of being able to emulate and debug systems is that they can be brought online in the real-world scenario much quicker than before,Mueller says. And emulation can be used for highly complex scenarios. "When a company is adding new software, often there's a distribution center that's hooked straight to the manufacturing plant, and changing something in manufacturing means they need to change the system in the DC and do it fast because it affects the whole plant," Mueller says. Testing new systems in this way has become so popular, Mueller says, E2M now includes it in its standard pricing.
Ongoing testing is one of the main benefits, says Mueller. "Once it's operational, throughout the lifecycle of the system, you can check changes offline."Most companies have a spare programmable logic controller, in case of emergency, and this can be used to run models, which is particularly useful for training and gathering continuous feedback on possible system changes. "Training is a huge benefit here, because you are literally working with the real system. It's like a flight simulator where you have all the controls of a 747, hooked to a computer that shows you what you'd actually see out the window. We've found it to be of great benefit,"Mueller says. In fact, he adds, it's considered one of E2M's leading competitive advantages.
Slow to warm up
All the chest-pounding aside, the acceptance rate of emulation technology remains low among U.S. companies— Mueller estimates it at around 10 to 15 percent. Though both users like E2M and vendors (like Brooks, Catalyst, CACI Products Co., Flexsim Products Co. and Rockwell Software) talk a good game, acceptance has been spotty. "The companies applying it are consulting firms, because the initial investment in time and money for a company to get up to speed is still significantly high,"Mueller says. "So it's used by people with multiple customers. Once you do one facility, the skill set is sufficiently valuable that you have to do another one and another one."
Then there's the matter of simulation's track record. Catalyst's Young says some reluctance to use simulation can be traced to the software's "checkered history in the U.S." "There have been instances where it's been grossly misused," says Young. "It's easy to create a simulation and have the simulation produce a result. But it's partially based on random numbers. You need to run the simulation with 20 different [sets of random numbers], but sometimes that isn't done. They run it once and say: 'Hey, we should do this,' and it turns out not to be the right thing."
It's important to remember that a system's only as good as the data fed into it—or as the saying goes, garbage in, garbage out. In the end, creating virtual scenarios in order to test another software system's responses requires that the fake warehouse be eerily close to the real one. By the same token, it's necessary to separate the factors that make a difference from those that don't. For example, it doesn't matter whether a lift truck is yellow or red, but the speed at which it accelerates does.
"Before you do this," warns Young, "you have to understand what your objectives are and how you're going to measure the simulated world. It's the user's responsibility to figure out what the simulation means. All the data crunching and math just leaves you with a bunch of numbers. The interpretation of that is art."