Faster by Design
An SwRI multidisplinary effort produces software, design tools for next-generation combat vehicles
James D. Walker, Ph.D., Sidney Chocron, Ph.D., Michael Moore, Ph.D., and Gregory C. Willden
An analysis of motion from a Tier 1 blast (left) takes seconds to compute, while a Tier 3 blast near the front of a conceptual hull (right) requires about an hour of computation time.
Using the automated meshing feature, a designer can produce a vehicle, including interior structural members, in about four to five minutes using a conventional computer.
These images show a blast computation on a conceptual hull without (top) and with (below) a heat-affected zone (HAZ). Including the HAZ shows how the materials will be affected by a blast.
This computation shows that strong bolts can tear a test panel during loading (left), while weak bolts will break (right).
SwRI validated the blast survivability tools experimentally, such as this test designed to measure the loads on an armor plate produced by a buried charge.
This side-by-side comparison shows a structural test specimen undergoing blast loads using the computational pipeline (left) and from an experiment conducted at the SwRI test range.
From left, Dr. James Walker is director of the Engineering Dynamics Department in SwRI’s Mechanical Engineering Division. His research efforts focus on the mechanical response of a variety of systems and materials to impact loads. Dr. Sidney Chocron is manager of the Computational Mechanics Section in the Engineering Dynamics Department. Chocron specializes in the response of materials at high strain rates. Dr. Michael Moore, a staff engineer in the Tactical Networks and Communications Section in the Communication and Embedded Systems Department of SwRI’s Automation and Data Systems Division, specializes in systems engineering and design. Gregory Willden was formerly a staff member in the Automation and Data Systems Division.
Complex military systems are increasingly expensive to develop. Complexity increases cost when it leads to "schedule slip," which leads to changes to the systems, which, in turn, leads to increased complexity. This complexity versus cost spiral oftens results in systems that meet the higher performance objectives, but are no longer affordable. The industry terms this phenomenon "requirements creep."
Over the past six years, three major heavy vehicle development programs for the U.S. Department of Defense have been cancelled. The vehicles for all three programs were complex, electro-mechanical systems intended to replace currently fielded systems with vehicles having new capabilities and enhanced survivability. The need for these advanced vehicles stemmed from threats encountered in the past decade of conflicts. The fact that the new capabilities are just now getting integrated with vehicles, and in some cases have yet to be integrated, is a testament to the difficulties faced by the DoD with the current vehicle procurement model. The Defense Advanced Research Projects Agency (DARPA) launched the Adaptive Vehicle Make (AVM) portfolio of programs with the ambitious goal of reducing the time from concept to rolling vehicle by a factor of five. By changing the design paradigm, the first vehicle rolling off an assembly line would be fully functional, and thus the long years of fixing, redesigning, and requirements creep would not occur. These vehicles would be "correct by construction" rather than "correct by design," meaning that for complex systems, what rolls off the production line would not be exactly the original design, but it would work and meet the requirements.
SwRI and the AVM efforts
The AVM effort aimed to foster a new design methodology centered around computer-aided design (CAD) automated analysis software tools. These tools would allow designs to be evaluated in the digital space, reducing the number of physical prototypes. The tools were to be tailored and evaluated against designs for ground combat vehicles, specifically an amphibious armored troop carrier. Southwest Research Institute (SwRI) was awarded the contract to deliver analysis models for the AVM effort in the area of survivability. This was a $5.7 million program, with the majority of the work to be completed in one year. The SwRI team provided models to evaluate the survivability of the vehicle designs against ballistic, blast, and corrosion threats. The team drew from several SwRI technical areas for experts in mechanical engineering, ballistics and explosives, materials engineering, and structural engineering as well as in systems engineering, software architectures, model-driven design, electronics, and training. Four subcontractors also were involved. Before SwRI was involved in the AVM effort, the tools developed were used in an open design competition, with DARPA awarding a $1 million prize to the winning design team. The powertrain and suspension were integrated into a chassis, and the mobility performance of the as-built chassis was compared to the predictions made by the design software. In order to further exercise the tools developed, selected vehicle manufacturers used the software developed at SwRI to aid in designing their vehicles. To demonstrate the SwRI-developed survivability software’s capability, the government also performed a design exercise, with blast article fabrication and blast testing performed at SwRI under DARPA’s Fast Adaptable Next-Generation (FANG) Ground Vehicle program. DARPA transitioned the AVM effort to the newly formed Digital Manufacturing and Design Innovation Institute, where the tools continue to be developed.
Advanced tools for advanced vehicles
The SwRI team’s work resulted in significant advances in analytical software tools such as the survivability tools, which analyze blast and ballistic survivability. The DARPA-funded teams used the survivability tools extensively for design exercises and earned favorable reviews from the commercial engineering design and manufacturing companies that produce defense systems.
The SwRI team was able to combine its considerable experience and expertise in impact and blast research with SwRI vehicle and software expertise, plus that of our subcontractors, to produce software tools that are best demonstrated by five major innovations developed during the project: multi-fidelity analysis/varying levels of refinement; automated meshing and connecting of parts for complex vehicle structure; uncertainty quantification and development of 95-percent bounding models; and a sophisticated large deformation/material failure material model library and better blast loads analysis tools. The fifth innovation was to connect the whole design "pipeline" together so it executes automatically.
A major goal was to make the survivability analysis tools easy to use. For example, the team developed a shotline viewer that allows a designer to explore the effects of ballistics impacts on a vehicle and shows results of the terminal ballistics models. For blast analysis, the software pipeline produced movies of the explosive event, showing the resulting deformation and damage to the vehicle. These capabilities allow designers with limited background in survivability to quickly understand how these threat environments affect their design.
Another major goal of the tool development was to remain "CAD agnostic" as much as possible. CAD agnostic means that the tools are not tied to one specific CAD system, but rather use a generic CAD format – referred to as STEP (Standard for The Exchange of Product model data) files – so any CAD system would be able to use the SwRI-developed ballistics and blast tools.
A major innovation that SwRI brought to the DARPA program was a multi-fidelity modeling approach employing different tier levels. Each tier had a different level of accuracy and uncertainty, in exchange for different amounts of computational time. For example, lower tier models are simpler, typically based on more assumptions about the physics. Because they include physics assumptions that then require less detailed computations, they have shorter computation run times, and thus are less expensive, but they also are less accurate and have a higher degree of uncertainty. The multi-tier approach allowed rapid exploration of the design space using fast-running lower tier models that sped up the conceptual design phase. In addition, by developing the different tiers of models, simpler, lower-tier models were quickly completed and thus working survivability models were always available in the software development process. This was beneficial for the concurrent development of other pieces of software and the definition of interfaces. Further, the various tiers of survivability solvers could also be used at a conceptual level, where only an outer "concept hull" needs to be defined. This allows a rapid initial design space exploration to determine the amount of armor and structure needed to survive a specified threat. As a design proceeds through more detail, the survivability analysis can be applied many times to allow the designer to make adjustments, such as at manufacturing seams, to ensure designer-level protections. By performing survivability analysis at the conceptual level, it is feasible to automate the design-space exploration.
Automated meshing and connecting
One of the more tedious and time-consuming steps in survivability analysis is transforming CAD geometry to a format that is amenable to analysis tools. To make the tools more useful and to head toward the goal of "press one button in CAD to get your analysis," the team developed tools to automatically mesh and then connect parts for the blast analysis. Our tools provided automatic meshing of various structural parts with a focus on producing quad-shell meshes, the types of meshes that have been shown to give the most accurate results in structural blast computations. (Structural parts include plates, panels, skins, and cross-section beams such as I-beams, C sections, and brackets.) The various parts are then automatically positioned in 3-D space, or "assembled" to produce the concept vehicle. Many different types of parts, with various meshing schemes, can be assembled this way. The emphasis is on robustness in automatically producing a good mesh for blast loading.
An important part of the assembly of a mesh is its connections. The team developed an electronic "bolter/welder tool," which includes the ability to operate the tool both when connections are specified or in an automatic mode. The combination of the powerful automatic welding feature and the automatic generation of meshes greatly reduces analysis time. Welds are automatically inserted when free edges are seen to be near other objects. Bolts are automatically inserted when holes are found aligned and a bolt for the region has been specified. In particular, the tool can bolt multiple plates together (three or more) if the bolt holes line up.
Another important element of the weld-connection and mesh-production capability is the inclusion of a heataffected zone (HAZ). Not including a HAZ leads to unrealistic strength predictions near joints. The automeshing tool, after connections are complete, passes through the mesh and produces a HAZ near all the welds. The material strength and damage properties are adjusted in the HAZ. It is important to include the HAZ in a mesh when performing blast computations, and the SwRI-developed tools did it automatically. During the connections step, a corrosion analysis also is performed.
Given the complexity of the computations and the multi-fidelity nature of the tools, the SwRI team quantified uncertainties and developed bounding models. The team developed ballistic and blast models to return 95-percent bounding results in addition to the nominal results. This means that, based on historical knowledge of variations in inputs into the models, one can find the performance bounds of the armors and blast mitigation systems. Using these statistical variations, many executions of the models were run a priori for certain situations to allow development of inputs to the physics-based models that would return the 95-percent bounding result. These computations used fast, off-line computations to develop an understanding of the influence of material, geometry, and other variability on the survivability results. With this, one can compute not only the nominal performance of the vehicle under a ballistic or blast threat event, but also the 95-percent bounding or worse-case response. This allows designers to know how close they are to meeting performance objectives, and how much resilience is built into their design. This information can be used in an optimization study where robustness is one of the optimization parameters.
A sophisticated material model library
Survivability systems are used once, and the materials are used all the way through failure. Hence, it is necessary to know the large deformation and failure properties of the materials. As part of this program, the SwRI team compiled a database of survivability materials with constitutive properties that described the largedeformation, plastic-deformation, and flow as well as the damage properties. Large deformation means that solid materials reach a limiting strength. Sometimes the deformation is so significant that it causes the material to "flow" similarly to a viscous fluid. A permanent final deformation is referred to as plastic deformation.
Also, because blast loads for buried charges are still a research topic, the SwRI team performed experiments to further characterize the blast loads on simple structures. In addition, experiments were performed using v-shaped hull body designs because in some instances involving buried charges, v-shaped hulls have been shown to reduce loads imparted to the vehicle by underbelly blast.
The team also tested structural members to validate loads and the computational pipeline. Structural members were held in a test frame and blast loaded with buried charges. These same structures were blast-loaded using the computational ipeline. Through this work, as well as extensive use of historical data, the blast survivability tools were validated and showed good results.
Connecting the whole pipeline
Historically, manually performing all the steps in survivability analysis was quite tedious and time-consuming. A concerted effort to automate the whole process paid off by greatly reducing the amount of time required. During one of the DARPA exercises, teams said the survivability software allowed them to perform detailed conceptual design iterations in 30 minutes per iteration, something that previously may have taken two weeks per iteration. Other design teams praised the tools for their ease of use, speed, and the unique ability to convey survivability results to the vehicle designer. Extensive verification and validation exercises were performed on all the models to confirm their implementation and the implementation of the pipelines.
The survivability analysis modeling tools are an important and successful part of the DARPA AVM effort. They perform as designed, fully automating complex steps that typically take manweeks to perform, and also providing estimates and bounds on the soundness of the answers. The SwRI team successfully demonstrated that automating steps ranging from material properties to meshing, connecting, and uncertainty quantification (UQ) analysis is a useful capability for designing blastand impact-resistant vehicles. In addition to their use by others, SwRI researchers have used these software tools in important follow-on design activities to analyze vehicle survivability.
Questions about this article? Contact Walker at (210) 522-2051 or firstname.lastname@example.org.
Acknowledgments: The authors acknowledge the contributions of SwRI staff members Dr. Charles E. Anderson Jr., Dan Pomerening, Dr. Elizabeth Trillo, Thomas Moore, Donald Grosch, David Riha, Alan Steiner, Scott Mullin, Dr. Alexander Carpenter, Joe Bradley, Andrew Barnes, Carl Weiss, P.A. Cox, John McFarland, Jeremy Zoss, Warren Couvillion, Katie McLoud, James Mathis, Ray Burgamy, Juan Magallan, Joe Elizondo, Danny Mendoza, Andrew Kammer, Dr. Kathryn Dannemann, and Joe Melig.
SwRI acknowledges DARPA funding through contract number D12PC00466. The views, opinions, and/or findings contained in this article are those of the authors and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government. Approved for Public Release, Distribution Unlimited.