Pictures of the Future   Fall 2007

Prototype for Perfection

Mail sorting machines have an insatiable appetite. In just one hour, they can process up to 40,000 items, which fly through their sorting gates at lightning speed, whereby soft pressure is applied at each gate to send envelopes along on their proper track. A rigid envelope, for instance, one with a CD inside, can do great damage in such a high-speed system, as it can get stuck in one of the gates, causing a huge backup of hundreds of letters in just a few seconds. The machine then has to be shut off, resulting in costly downtime.

Giant sorting units are therefore equipped with precision mechanical instruments for measuring letter stiffness. Like a small finger, such instruments briefly tap each letter to measure its resistance. Envelopes deemed to be too rigid are removed before they can cause damage. The stiffness measuring instruments have to be both sensitive and fast in order to be able to touch each envelope as it flies past without damaging it.

Around a year ago, engineers responsible for the production of sorting machines at Siemens Industrial Solutions and Services’ (I&S) Postal Automation division in Konstanz, Germany, found that they needed a particularly fast module that could measure stiffness in just five milliseconds. What was required here was a high-tech device whose electric motor would move the sensor probe back and forth very rapidly and with extreme precision.

Siemens specialists in Konstanz knew that developing such a a sophisticated piece of equipment would be a tricky assignment, which is why they called in experts from Siemens Corporate Technology’s (CT) Virtual Design Center in Munich to work with them on the planning process. The Center designs complex products on computers and brings them to life in the virtual world, where they are then tested before even one prototype is built. Among other things, you can hear the sound of washing machines running at the center, even before any such machines have been built.

Concurrent Engineering. The ability to bring products to life as realistic 3D computer models is nothing new. Computer-aided design (CAD), for example, has long been a workhorse in industrial design departments, and the simulation of flows and acoustic oscillations is standard technology today. "What we’ve done here at CT is to link all these virtual modeling and simulation tools to create an integrated approach," says Bernd Friedrich, head of the Virtual Design Center.

Friedrich’s work focuses on mechatronics systems development—i.e. designing and linking mechanical components and electronic control systems in parallel. Engineers in Munich refer to this design model as FINE (functional and integrated engineering of mechatronic systems). FINE is used to simultaneously develop mechatronic components, whereby specialists from many fields, such as mechanical, electrical and software engineers, work together on virtual models. Such activities make it clear at a very early stage whether, for example, motors and control units will operate together harmoniously.

"In the past, mechanical components were built first, then the electronics were added, and at the very end the control system was tested with the finished hardware," Friedrich explains. "But that approach simply takes too long." That’s because errors such as a motor with insufficient power or a slow control unit generally aren’t discovered until all the components are operating together in a finished machine—by which time it’s too late. "It was often the case that several prototypes were built and tested before a production-ready design was ready," says Friedrich. The new parallel—or "concurrent engineering"—approach has engineers from all disciplines working together from the beginning, which means a fully functional product model is stored on a computer before anything is built. The computer can thus be used to simulate and run through several product variations. Moreover, customer requests can be taken into account right up until shortly before the conclusion of the development process. "We’ve found that this approach cuts development time by about one-third," says Friedrich. "It doesn’t matter whether it’s automotive components or power plants—new products can be brought to market more rapidly, and this shortening of time-to-market is crucial for sales success."

The stiffness measuring model for postal automation systems was developed less as an effort to cut development time than to achieve the necessary dynamic performance of the product in question. Engineers in Munich were able to demonstrate this performance with their virtual model, in which all components worked perfectly with one another at the required speed and precision. "What’s really remarkable is the linkage between the various mechatronics aspects," says Dr. Thomas Baudisch, who is responsible for Mechatronics Product Development at CT. "Ultimately, it was our interdisciplinary approach that enabled us to design the unit in an optimal manner."

Multi-Physical Construction. All of this integration is backed up by mathematical knowledge, because an approach as complex as the one pursued here is only possible if you are capable of developing the necessary algorithms yourself. That’s why the Virtual Design Center team includes several mathematicians who developed the so-called "multi-physical approach" together with engineers. The concept takes into account many different physical properties, such as temperature distributions within materials, oscillation characteristics, and strengths. Parameters that determine a real product’s future functionality and quality are therefore incorporated into its virtual model. This is the only way to determine in advance whether, for instance, a washing machine will actually spin quietly after it’s built.

The multi-physical approach even goes beyond the product itself, as it takes into account the entire process chain from the first design drafts all the way to future production. It therefore starts with CAD and CAE (computer-aided engineering), moves through simulations and modeling of the products, and ends with CAM (computer-aided manufacturing). For example, when planning the production of turbine blades, CT experts calculated the amount of force the milling and cutting tools apply when machining the blade surfaces. This cutting force calculation enabled engineers to accurately design the dimensions of the clamps that hold the blade in the milling machine while it’s being processed.

CT mathematicians are also looking at natural fluctuations—conditions in a gas turbine combustion chamber, for example, that are not always the same. Changing gas compositions, temperatures, and component tolerances have a major impact on the optimal geometry of the blades. The mathematical optimization approach takes into account precisely these fluctuations, including all uncertainties in the calculation, thereby enabling an optimal design. Mathematicians refer to this as Robust Design Optimization—or RoDeO (see Pictures of the Future, Spring 2006, Paths to Perfection). "This probability approach borders on pure mathematics," says Friedrich. "That’s something that’s never been done before in product development."

Global Turbine Development. Work carried out by experts at Siemens Power Generation (PG) in Berlin involves turbine blades for complete gas turbines. These machines, which are as heavy as several locomotives, consist of thousands of components, including several hundred precision blades that must be joined together exactly. Recently, engineers in Berlin used virtual planning tools for the first time on a major scale while developing the brand-new 340-MW turbine for a new gas and steam facility in Irsching, Bavaria (see World’s Largest Gas Turbine). The most important goal here was to reduce development time through better coordination of staff working in departments housed at several locations, such as designers in Orlando, design engineers in Mülheim, Germany, and production specialists in Berlin.

Up until recently, design drawings were sent back and forth by courier, with engineers writing down comments on the documents. In other cases, sketches were scanned and sent electronically. Experts also frequently had to travel to meet with colleagues in other locations. Today, development project participants conduct videoconferences. In the case of the Berlin-based turbine project, each of the three locations was equipped with a Powerwall VR system, which was used for presenting the virtual turbine model, and which was linked via data connections. This enables participants to view and discuss the same model simultaneously. "Development discussions have improved tremendously as a result, and the entire process has been accelerated," says Michael Schwarzlose, who introduced virtual turbine development at PG. Unlike abstract design sketches, virtual models enable joint communications that enhance understanding of the situation at hand. Component installers, for example, recognize very quickly whether some components might collide during the assembly process. Virtual models also make the entire development process more vivid and dynamic, says Schwarzlose.

The product development process for a new turbine is generally a difficult undertaking that consists of many different steps. It basically begins with a draft design in 3D-CAD programs. These 3D models are created before detailed 2D drawings are made and mainly serve as a means of assessing the availability of required components, and the feasibility of production and assembly processes. Production sketches are not drawn up until later on in the product development process. "We don’t need this redundancy in producing drawings or sketches anymore," says Schwarzlose, "because we can now go directly from CAD to a virtual model." The virtual reality (VR) software used here is from ICIDO, a spin-off of the IPA Fraunhofer Institute in Stuttgart that specializes in planning large machines and even entire factories. Schwarzlose introduced this tool at PG in 2003, at a time when the 340-MW high-performance turbine for the Irsching power plant was still at an early development stage.

In 2005, a further element was added: a VR system for gas turbine final assembly. Since it takes weeks to assemble a giant turbine, and the process is almost as complex as building an aircraft, VR technology has accelerated the assembly process. The technology allows specialized mechanics to practice manual assembly maneuvers in advance using virtual final assembly programs—something that would have been inconceivable just a few years ago. Schwarzlose recalls how things used to work during the early stages of work on the Irsching turbine. Back then, in order to test assembly operations, a full-scale model of a turbine combustion chamber had to be built in Berlin. What’s more, it took months from the moment an order was place until a model could be fully assembled. And, of course, it wasn’t possible to test the assembly process during that time.

Tremendous Savings. The amount of time gained through the use of new virtual tools is tremendous. Depending on the complexity of individual turbine components, it used to sometimes take weeks or even months before researchers could determine whether it would even be possible to install or manufacture certain components. "While it’s true that virtual reality can’t replace a real operation in every single case, the fact remains that an actual model cannot depict or make noticeable the smallest tolerances," Schwarzlose explains. All in all, the virtual planning process can reduce development times by several months, according to Schwarzlose. The Irsching turbine will be operational next year after only seven years of planning and construction. Projects in the past took much longer to complete.

VR is is set to become a key part of product lifecycle management at PG. A roadmap for establishing a PLM process is currently being worked out. The goal here is to permanently incorporate all development processes, combine various development platforms, and simplify the exchange of data. New simulation tools, such as those made by UGS (a major PLM player recently acquired by Siemens—see ( UGS and Siemens), will further develop virtual reality into a key development component whose depiction of reality will become increasingly exact. Such precision has long since moved beyond individual products to include entire factories that are developed in computers (see (Factory Planning), allowing industrial companies to save oceans of time and money.
                                                                                                                   Tim Schröder