- 1.0 A New Product Development Paradigm
- 1.1 Computational Engineering and Virtual Prototypes
- 1.2 Computational Science and Digital Surrogates
- 1.3 The Computational Engineering and Science Ecosystem
- 1.4 High-Performance Computers: The Enablers
- 1.5 Full-Featured Virtual Prototypes
- 1.6 The Advantages of Virtual Prototyping for Systems of Systems
- 1.7 Virtual Prototyping: A Successful Product Development and Scientific Research Paradigm
- 1.8 Historical Perspective
This chapter is from the book
This chapter is from the book
This chapter is from the book
1.1 Computational Engineering and Virtual Prototypes
Engineering can be defined as follows:
The use of science and mathematics to design, construct, or manufacture physical systems.
Computational engineering is an extension of classical engineering that makes it possible to shift from physical prototypes to virtual prototypes. Computational engineering adds:
Digital models and simulations, often coupled with high-performance computing, to solve complex physical problems arising in engineering analysis and design.
Traditional engineering product design involves the repeated iteration of four steps: 1. requirements and conceptual design, 2. detailed design, 3. construction of physical prototypes, and 4. experimental testing of the physical prototypes. Figure 1.1 depicts this process. The design analysis was initially carried out using hand calculations and heuristics, then calculators, and, more recently, computers (Consortium 2015 and Paquin 2014).
)
Figure 1.1 Traditional product development process based on physical prototypes (Courtesy Software Engineering Institute)
If the experimental physical prototype tests are successful, the product proceeds to manufacture. The manufactured products are tested experimentally to determine whether they meet requirements.
The manufacturing process is guided by engineering drawings, which can be digital, paper documents, and possibly even a physical model of the product.
Computational engineering exploits the use of computers to supplement or even replace the use of physical prototypes with virtual prototypes for the development of complex products (Consortium 2015, Post 2015, and Post 2009). Figure 1.2 illustrates this process. The starting point of a virtual prototype can range from a 2-D CAD drawing to a 3-D NURBS (non-uniform, rational B-spline) description of the product geometry, including all the associated metadata needed to completely describe the product and its attributes. The latter is called the digital product model. It can also include the history of the design process, a complete record of the analysis of the product design, and the data needed to start product manufacture.
)
Figure 1.2 Computational engineering product development process based on virtual prototypes (Courtesy Software Engineering Institute)
Figure 1.3 compares the two approaches.
)
Figure 1.3 Schematic comparison of historical empirical iterated “design, build, test” paradigm and the virtual prototyping paradigm (Courtesy Software Engineering Institute)
What does it mean to pass a “virtual” test? Chapter 10, “Verifying and Validating Science-Based Software,” discusses this in detail. However, remember that Mother Nature always casts the final vote. Her laws of physics enable us to predict her vote before we build a physical prototype of the product and test it. As our understanding and experience grow, the need for physical confirmation decreases.
Although the widespread use of computational engineering is relatively new, the use of computers to design and analyze the performance of products, or at least components of them, dates at least as far back as the early days of the Manhattan Project during and just after World War II (Dyson 2012, Ford 2015, and Atomic 2014). From the end of World War II to the present, the U.S. Department of Energy nuclear weapon laboratories have relied on virtual prototypes to maintain and optimize the designs of nuclear weapons (Energy 2019). When high-performance computers became generally available (1960s onward), industry and government began to use them to develop new, more complex products (Council 2005, Post et al. 2016, and Paquin 2014).
More recently, the concept and capabilities of computational engineering (and science) have been extended to include the whole process of product development and scientific research. This is expressed with terms such as digital engineering and the digital thread (Fei Tao 2019 and Kraft 2016). Starting with the digital product model, the artifacts (surrogates and design metadata, along with maintenance and operational history) of the entire product lifecycle are linked together as a digital thread.
Computational engineering extends the classical engineering approach to make use of software and computers to model the effects that determine the behavior or performance of a product before it is manufactured. Only in the last decade have high-performance computers reached the petascale capability (1015 FLOPS) to handle all the relevant aspects of the design of very complex products such as aircraft and ships. However, even high-performance computer workstations are often adequate for designing and testing smaller, simpler products. The utility of computational engineering is not limited to designing complex machines such as aircraft. It is also being used to design consumer products such as bicycles, golf clubs, bleach and detergent bottles, and even hip replacement joints and other medical appliances.
The strong need for a new approach for the product development of complex systems based on computational engineering can be illustrated by the history of three of the most recent U.S. Department of Defense (DoD) major multibillion-dollar air vehicle procurement programs described in a recent Hudson Institute report (Greenwalt 2021):
F-35 Joint Strike Fighter (JSF)
F-22 Raptor Stealth Fighter
V22 Osprey Tilt-Rotor
Following the present “design, build, test, repeat” (see Figure 1.1) approaches that rely on the development and testing of physical prototypes for these three aircraft, DoD contractors now need more than 20 years from project start (contract award) to deliver fully operational aircraft to the DoD. In 1975, they needed only about 5 years. The Joint Strike Fighter (F-35) program started in 1996. The first fully operational F-35 was delivered 15 years later in 2011 (F-35 2021). The costs almost doubled from project start to acceptance by the Air Force (Wheeler 2012). Without improvement in aircraft procurement methods, it will continue to take the U.S. longer to develop new aircraft. The F-22 program started about 1986. Delivery of the first fully operational aircraft occurred in 2005, nearly 20 years later (F-22 2021). The number of F-22 fighters that the DoD ultimately purchased dropped from 750 to 183, due to the increase in price of each airplane and the delay in the program (Ritsick 2020). The final cost was $340 million for each F-22. The V-22 Osprey Tilt-Rotor program started about 1983. The first full operational capability (FOC) aircraft was delivered in 2005, more than 20 years later (V-22 2021).
Before 1975, DoD defense contractors generally took no longer than 5 to 7 years to develop and deliver new aircraft. The Lockheed F-117 Nighthawk Stealth bomber program, the General Dynamics F-16 Fighting Falcon, and the McDonnell Douglas F/A-18 Hornet were begun in the 1970s and delivered in 5 to 6 years. The McDonnell Douglas F-15 Eagle, started in 1967 was delivered in 1976, 9 years later.
Other types of complex aircraft have not experienced this level of growth in development time. During the this period (1967 to 2011), the time to market for new commercial aircraft such as the Boeing 737 (delivered in 1967), the Boeing 767 (in 1982), and Boeing 787 (in 2011) rose by only two years, from approximately 5 years to 7 years. Similarly, the time to market for new automobile and truck models has remained 4 to 6 years. Commercial aircraft and automobiles and trucks are quite complex, with massive amounts of embedded software, high reliability requirements, and strong cost constraints (Greenwalt 2021).
The recent history of Goodyear Tire and Rubber Co. provides a compelling example of the benefits of computational engineering for reducing the “Time to Market.” Facing fierce competition from Europe (Michelin) and Japan (Bridgestone), Goodyear decided in 1992 to build physics-based tire design software (which we henceforth often refer to as tools). Goodyear entered into an ongoing collaboration with Sandia National Laboratory that enabled the company to combine its knowledge of tires and its materials with Sandia’s knowledge of finite element algorithms for massively parallel computers to develop this capability. In 2003, Goodyear used the tire design tool to reduce its time to market by a factor of 4 (Miller 2010, Miller 2017, and Council 2009). By enabling designers to generate and analyze more design alternatives, the tire design tool also allowed Goodyear to increase the number of new products per year, from 10 to 60. The Goodyear annual reports began to refer to the company’s “New Product/Innovation Engine.” This story is being repeated in many industries, such as Ford automobiles (Kochhar 2010), Whirlpool refrigerators (Gielda 2009), and Procter and Gamble shampoos and hand lotions (Lange 2009).
Today the use of virtual prototypes can supplement and, in some cases, even replace the use of physical prototypes (Post 2015). Design engineers can construct and store thousands or even millions of 3-D virtual prototypes of a potential product in a design option tradespace. The performance and behavior of each virtual prototype can be quickly assessed using simplified physics-based software. Further analysis can identify the most promising design options. More sophisticated high-fidelity computational analysis tools can then be used to accurately predict their performance. Finally, if needed, a physical prototype of the final design can then be constructed and tested for final verification of the design before manufacture.
Another advantage is that virtual prototyping can greatly accelerate product innovation. Final design decisions can be postponed until later in the design process, when more information is available through virtual testing of the design candidates. Usually this can be done orders of magnitude more quickly and cheaply with virtual prototypes than with physical prototypes. Virtual prototypes make it possible for designers to learn by developing and testing a much larger tradespace of designs. Goodyear used virtual prototyping to increase its innovation rate by a factor of 6, from 10 new tires per year to 60 new tires per year. Virtual prototyping enabled the company to rapidly learn through failure, similar to the Silicon Valley mantra (Petroski 2006 and Post 2017).
Numerous studies (Augustine 2007, Oden 2006, and Glotzer 2009) have highlighted the risks to U.S. international competitiveness if U.S. industry falls behind the rest of the world in the use of computational engineering. It is widely acknowledged by senior U.S. industry and DoD leaders (Cordell 2018) that, to remain competitive, U.S. industry and the U.S. government and its contractors must reduce their time to market, costs, and risks while producing high-quality, market-worthy products.