Simulation tactical tool for system builders

Test drive a new materials handling system digitally before you spend capital to build it.

By Tom Feare, Senior Editor -- Modern Materials Handling, 7/1/2000

Materials handling projects. Roller Coaster Tycoon or Sim-City 3000. And training pilots for the U.S. Navy and Air Force. What do they all share?

You might not think there's much of a connection between materials handling and top-selling computer games. Or any link to becoming a 'top gun.' But there is.

Computer simulation is the common element. Simulations for the military, PC games, and animated films all helped lead to the latest capabilities in industrial systems modeling. And models of systems have become increasingly important in the real world of project design.

Today's 3-D and 4-D digital models of factories and warehouses owe much to stunning advances in computer graphics designed originally for military purposes and for top-selling games and movies. And that's a key advantage.

'Seeing is believing,' says Matt Rohrer, director, simulation products and services, AutoSimulations. 'With graphical simulation, manufacturers and materials handling managers can test control logic, update and verify operating data, and test 'what if' scenarios and new ideas to optimize their current facilities or look out into the future.

'If the graphics don't look real,' Rohrer stresses, 'it will be hard to convince anyone that the model is valid.'

Often, that's a major benefit of simulation: selling the modeled system to management, particularly those who aren't engineers.

'Now, after days, perhaps weeks of model building and testing, we have a simulation that realistically represents system features. And from the model we also can run lifelike, 3-D animations of the system that even the accountants and financial people can understand,' says Deidra Donald, Deneb Robotics product manager.

Today's simulated factory and warehouse models also rely upon the greater processor speeds and number-crunching power packed into newer PCs. Applied to analysis of a materials handling system, simulation can be a powerful tool in decision making, testing the many 'what ifs' behind any detailed process or flow of goods.

'Simulation is process verification,' explains Bob Brown, president, Deneb Robotics. 'Computers now are faster, the graphics better,' he says. 'It's much more affordable to develop and run a 3-D model.'

Model building is a complicated computer technology to learn, as any simulation expert will acknowledge. Yet it's less difficult today than it was 5 or so years ago, says Karen Stanley, AutoSimulations director of sales and marketing. Even so, most software vendors start out new users with a 3 to 4 day training program.

F & H Simulations vice president Roger Hullinger agrees with Stanley about the ease of use of today's systems. But simulation projects in materials handling are complex efforts to carry out, he hastens to point out, in terms of time spent gathering data and understanding the flows of materials.

Significantly, as Deneb's Brown suggests, in many simulation exercises, '80% of the problem is collecting the data necessary to build the computer model.'

Major simulation software programs today are more intuitive, have 'drop-and-drag' and 'point-and-click' features, and use click-on dialog box options to simplify the model building process. There's little, if any, programming expertise required.

Ready for 'what if' games?

But is simulation something for you, your company? Are you ready to become this kind of 'what if' materials handling gamer with a computer? Or do you just need to be smart enough about simulation to look over a consultant's or vendor's shoulder as he or she builds a model?

Should you:

• Start that next handling project by first simulating it on a computer model, then specifying the hardware subsystems, equipment, and software?

• Or, perhaps, construct a model of an existing system to find ways to improve the operation, speed up work-in-process and material flows, avoid bottlenecks, better allocate resources?

• Or take your model and use it to monitor, possibly control and adapt handling processes as changing conditions require?

It all depends, say the experts in computer simulations. Complexity is the key word in the answers to such questions. The more complex the project, the more likely you are to get it right creating a simulation model first, then building the system in the real world. Or to model all the many details of queues, resources, and the like in an existing facility, and then see how to improve upon subsystems within the larger warehouse or factory shop floor you already have up and running.

Meanwhile, for now, only a few cutting edge users are using models to monitor and control their operations.

When paper, pencil work

Paper and pencil, a calculator, a spreadsheet or two, and some other simple planning and design tools still work to design very simple systems. Vendors and consultants can provide good advice without a simulation when your project closely resembles similar and simple systems already installed elsewhere.

'Anyone can perform analysis manually,' suggests Charles Harrell, R & D director and founder, ProModel Corp. But as the complexity of the analysis increases, 'so does the need to employ a computer-based tool.' Spreadsheets can perform many complex calculations, for example. Yet they use averages to represent activities such as work-in-process arrivals at a work station or goods received on a loading dock, and they approximate the availability of labor or machine resources.

Such approximations do not reflect the randomness and interdependence that system resources and other elements share in the real world, Harrell says.

Simulation, however, he adds, accounts for randomness through probability distributions taken from study data on the system or process, for instance.

With static analysis techniques such as queuing theory and spreadsheets, he continues, 'you know the average wait time and number of items in a queue, but there is no way to further examine the data.

'With simulation, you know the wait time, number of items, minimum and maximum values, confidence interval, data distribution, and the time plot of values. It is more valuable to know that the number of items in a queue exceeds 10 only 5% of the time than it is to know that the average number waiting is 2.'

What will it cost?

As much as $50 million a year is spent, worldwide, on purchasing the software for simulation, the vendors say. 'Probably two to three times above that amount,' adds F & H's Hullinger, is spent on fees and paychecks behind all the hours that simulation specialists, in-house engineers, equipment vendors, consultants, and others take in building models for industry.

At a bare minimum, one engineer needs to be assigned to a simulation project, says the F & H exec. That person will need time to learn the software, then to build the model.

Simulations can be completed in a few days or take as long as several months. Costs to the user can run from a bare bones figure of $2,000 to longer projects totaling $135,000 or more.

Even a $150,000 investment in simulation, says Deneb's Brown, can be a worthwhile 'insurance policy' in getting a $2.5 million project right, and not making the mistake of over- or under-building the facility.

One simulation vendor has a pharmaceutical company on retainer spending $100,000 yearly for its simulations. Another vendor suggests its average project runs roughly $30,000 to $50,000. A third vendor says its average is around $30,000 and down from as much as twice that several years ago.

Detroit's automakers spend huge sums on simulation, as do firms in aerospace and other high technology industries such as semiconductor manufacturing.

Even so, as AutoSimulations' Stanley observes, 'not even General Motors does all of its simulations in house.' Nor does GM use the same software from plant to plant, division to division.

Do you want to simulate in house? One rule of thumb is to try to do so when you have 3 or 4 similar materials handling problems to solve each year. Then you probably can justify the expense of developing the necessary internal expertise. What's more, this number of projects will keep the simulating team proficient in the software's usage, suggests Stanley.

Internal simulations, points out Vivek Bapat, simulation product manager, Rockwell Software (formerly Systems Modeling) benefit from the detailed knowledge that users have on their own systems. 'Our goal is to ensure that our product speaks the language of the person closest to the manufacturing line or the warehouse operation, enabling him to contribute as much as possible to the simulation effort.'

Yet, with new projects separated by lengthy gaps of time, consultants and/or vendors with simulating expertise may be required. Expect to find a mix of talents and kinds of software in use.

What the future holds

'Object-oriented computer simulation technology is the wave of the future,' suggests F & H's Hullinger. His firm's software and that of some other vendors build models around an electronic library of objects (conveyors, AGVs, forklifts, and the like). Or the software lets the user create his own objects. With object-oriented software the modeler can make changes in the simulation faster, moreover.

Some simulation programs are evolving toward becoming part of, or integrated with corporate-wide enterprise systems. Bapat of Rockwell Soft-ware, for example, says his firm's 'philosophy is for simulation to truly become an enterprise tool, used across a broad cross-section of applications, from customer-contact management, business processes, manufacturing, and supply chain management.'

At EAI, AutoSimulations, and else- where in the simulation industry, there's progress toward and interest in having models and animations communicating over the Internet. Project team members at separate plant or corporate sites might collaborate on a model, for example.

Or 'multiple models might talk to one another in different parts of the U.S.,' as AutoSimulations Rohrer suggests. In a project for the U.S. Postal System, a high-speed mail sortation model 'talks' to a tilt-tray system via Internet connections.

Rockwell Software's Bapat suggests that Internet applications for simulation appear 'exciting because modeling involves a collaborative effort.' Yet it may be that only reports, data, and animations from simulations will be exchanged. Security issues involved in sharing, and building models on the Web, could be major concerns because a model incorporates an organization's intellectual property and business intelligence, he observes.

EAI, meantime, has launched e-Vis.com, a secure web portal for speeding up business-to-business visual collaboration on simulation and related projects.

Simulation of materials handling systems-it's a serious computer game, it's hard work, it's fascinating when one sees the final models running, and it's more than paying its own way in the real world.

The cat's meow at Friskies

Late last month pet food maker Friskies was set to go live with its new, third party logistics distribution center at Hams Hall, Birmingham, England. Much of the credit for bringing this 3PL DC on line and on time goes to simulation (AutoSimulations, www.autosim.com ).

'In fact, there simply is no other manner in which such large and complex projects (75,000 pallet positions) can be adequately and accurately tried and tested,' says Trevor Chesteron. He is principal project manager for Swisslog Digitron, the systems integrator, which worked in partnership with Wincanton Logistics, the 3PL firm.

SDX simplifies model building