"...Virtual product development (VPD) tools allow validation of operational effort, load capacity, and abuse testing."

Founded in 1965 as American Sunroof Company and headquartered in Southgate, Mich., ASC helps automakers design, engineer, and manufacture high-impact, low-volume specialty vehicle programs. The company's award-winning programs include the first retractable hardtop on a truck chassis, the 2003 Chevrolet SSR, which won a 2003 Chrysler Group Gold Award and a Gold Award in the 2004 Industrial Design Excellence Awards. ASC has also developed convertible systems for the Toyota Solara and Mitsubishi Eclipse Spyder, and numerous appearance programs and body packages for such products as the Dodge SRT-4, Dodge Viper SRT-10, and Pontiac Grand Am SC/T.

The company's passion for innovation is obvious from its list of 'firsts': the first modern retractable hardtop, the first factory-installed power sunroof in North America, the first inwardly folding convertible top, the first glass panel sunroof, and the first modular sunroof. Its latest advance, the patentpending xpanse™ Convertible-Top System, was unveiled in January 2005 on the ASC Helios, the world's first modern four-door convertible.

In the following case study, Stephen Doncov provides in-depth insight into the evolving role of Virtual Product Development in ASC's design process. A convertible top header latch, which secures the front bow of a convertible top to the windshield frame structure, is a complex mechanism with as many as 10 or 20 parts made of different materials that deform plastically differently. These types of latches are challenging for CAE software because of the many design issues and tradeoffs to be considered, such as contacts, joints, flexible bodies, and multiple material properties.

The ASC design engineering team had been making do with linear analysis, which could show where issues existed but was unable to address the high nonlinear stress ranges ASC's products experience. Additionally, ASC was looking to streamline its product development processes with integrated tools to minimize the time and cost of designing and testing latches.

The ASC design engineering team determined that an integrated CAD-CAE environment provided by embedded CAE tools was best suited to help solve complex linear, nonlinear, and dynamic problems. In turn, these virtual product development (VPD) tools allow validation of operational effort, load capacity, and abuse testing. The VPD tools are part of ASC's strategy for migrating to Design-Analyze-Confirm processes and away from Design-Build-Test- Break processes.

The design engineering team's task was to ensure that the top header latch mechanism functioned within customer-supplied requirements for operational effort, abuse, and load capacity. Although the design was built in CATIA V4, CATIA V5 FEA/DMU Kinematics and MSC.Software's SimDesigner Motion and SimDesigner Nonlinear tools were utilized. During an earlier evaluation, the V5 analysis tools were found to be robust, quick, and easy to use because the V5 generative design approach utilizes feature recognition. When features on a part are changed, it is automatically remeshed. Using V5 models and a generative approach, the productivity gain more than made up for the time lost converting the V4 models.

For this header latch, it was particularly important to identify areas of nonconformance and provide feedback that could be used to make engineering changes. The results were validated in SimDesigner Nonlinear with physical testing from a previous latch design. The analysis closely matched the areas of high stress and failure identified with physical testing. Furthermore, analysis of the redesign indicated no failures, which also correlated with physical tests.

Another factor for consideration was that the production method for many of the parts is different than the one used for making a physical prototype. For example, prototype die-cast zinc parts are made using investment casting. Consequently, a different material could be used in the prototype. Typically, the properties of the material in a physical prototype are lower, which makes the final determination of the production part properties more difficult. The confidence in an analytical model representing a production unit or assembly is much higher because the properties of the die-cast (production) material are used. Without analysis, the possibility existed for overengineering the part.

For this latch, customer requirements included operating range, abuse (vertical and side force), and maximum load. A motion simulation was used to determine if the effort required to open and close the latch met the 40 to 60N requirement. Three nonlinear analyses were run to determine the effects of abuse, including a 600N opening force on latch handle, a side load of 294N, and a maximum load of 4900N applied to the hook. This last requirement was to ensure the mechanism wouldn't break under some extraordinary and unforeseen event.

Motion Analysis Study

Operational Effort
The model of the header latch mechanism was created using CATIA's DMU Kinematics tool to move the handle to the fully opened position. The handle was constrained to ground, which actuates the model. Torsional springs were inserted between the hook clamp body and roll pin 1, and between the lock button and the long roll pin. The leaf springs were approximated using linear springs and dampers for the handle detent spring and the handle-to-idler spring.

The latch mechanism is quite complex and involves a lot of contact and features that cannot be modeled kinematically. Kinematics approximates the operation of the latch but doesn't consider the physics. The motion analysis in SimDesigner Motion considers the physics. For example, it determines if the springs are too strong or too weak and provides force deflection curves. Kinematics or animation software can't do that.

The stroke of the latch was simulated with a prismatic joint, allowing the receiver to travel along the stroke direction of the latch. The approximate load of the top cover was modeled by utilizing a linear spring along the travel path. As the receiver is pulled, the load increases from zero to a maximum value.

Because SimDesigner Motion considers the physics of the model, it allows interaction between bodies. For this analysis, 3D contacts were utilized. All of the contacts were modeled, including the hook-to-roll pin, handle stop to detent spring, lock button to mounting bracket, lock button to handle, idler to handle in two places, handle to lock button, and hook to receiver. The leaf springs were approximated, using linear springdampers with a revolute (hinge) joint where the pivot would be.

The motion study confirmed the proper operation of the latching system, including the interaction of latch handle and idler bracket. The latch capture range (latch stroke) was verified along with the operation of the handle lock. However, SimDesigner Motion identified spikes of 77N and 88N in the effort required to move the handle, exceeding the requirement of 40 to 60N. Looking at the mechanism as it engaged clearly showed the dowel pin bottoming out before the handle locked. This caused the spike in force. By using a finer hook adjustment, the handle locked before the dowel pin bottomed out in its receiver.

The force required to close the latch mechanism was found to exceed the specified range.

Latch handle abuse analysis in SimDesigner Nonlinear on the modified design
indicated the handle satisfied the vertical abuse specification.

Nonlinear Analysis

Abuse – Vertical Force
The abuse analysis was used to determine if the handle would break when an opening (vertical) force of 600N was applied. If the handle bent, it was not considered a failure. A linear analysis was run to determine which parts could be omitted to save nonlinear analysis computing time. Parts not reaching their yield strength and not critical for maintaining the integrity of the mechanism, such as connecting pins, were eliminated, reducing processing time.

Initially, all the parts were meshed with Tetrahedron 4 elements (tet4's), which are linear brick elements. They are very stiff elements that reduce compute time. Since the nonlinear analysis determined all the high-stress area hot spots were located in the handle, all the parts were kept as tet4's. The handle was set to tet10, a higherorder element.

Using the local mesh feature, a finer mesh was used in the critical stress areas. Local sag conditions were implemented around the area of contact and hot spots. The ability to use localized refined meshing provides higher accuracy results in the critical areas identified during the linear analysis and substantially reduces processing time. SimDesigner Nonlinear allows different types of contact between bodies to be defined. For example, when looking for tangency, contact, or an impact, it allows two bodies to be constrained as always glued together, intermittent contact, or never touch. This allows the user to decide how the individual parts will interact, which makes the solver more efficient and further reduces processing time.

The clamp mounting bracket and both handle-to-idler hinge pins were constrained. For bodies that do not collide and bodies contacting themselves, contact was set to inactive, eliminating their consideration. The bodies that did not need to rotate in relation to one another were set to glue. The bodies that could come in and out of contact with one another or would need to rotate relative to one another were set to touch. A 600N load was applied in the vertical direction of the face near the end of the handle.

For the material properties, ZA-8 zinc, the same material used in production parts, was used for all the components, except the connector pins, which were 1010 steel. ASC and MSC.Software provided the nonlinear material properties. Although the curves were not validated with physical testing, this can be an important consideration. Any cold-working or heat treatment of the parts can lower or raise the yield and ultimate strength, dramatically affecting the nonlinear material curve.

The contact bias was set at 90%, allowing the solver to converge faster because it doesn't have to drive down to an exact solution. Contact bias is a numerical method to help convergence in contact analysis using MSC.Marc, the solver used by SimDesigner Nonlinear. For some applications, the contact bias helps improve stability in contact analysis. Coulomb friction was activated because friction affects material deformation or material flow on the contact boundary. This in turn is reflected in force and stress, etc. Additionally, friction generates heat, which affects material properties in thermalcoupled analysis.

As mentioned earlier, the physical prototype was made of ZA-12, which is not as strong as the production material ZA-8. VPD tools enabled the use of ZA-8 production material properties, which with satisfactory correlation would provide data for determining the performance of production parts.

The handle design was simulated using ZA-12 and ZA-8 material properties. Additionally, a physical prototype was made with ZA-12 and tested to validate the results. The vertical abuse test of the initial design indicated the ultimate stress was exceeded in the handle using ZA-8 at approximately 60% of the load. The actual results were:

The handle design was modified and another simulation performed using ZA-12 and ZA-8 material properties. Another physical prototype was made with ZA-12 and tested for correlation. The actual results were:

The redesign using ZA-8 material properties provided an approximately 200% safety margin, which satisfied the vertical abuse specification. Additionally, a simulation was run using the ZA-12 material properties, which indicated a failure at 600N. The physical test using the ZA-12 material determined that failure would not occur until 130% of the load was achieved (approximately 780N). This was within 25% of the simulation results, which was within ASC's correlation target.

Abuse – Side Load
Normally, a failure happens when somebody applies too much force on the handle. Therefore it was important to know what would happen if the handle was left half open and got caught on somebody's shirt, or what would happen if somebody tried to muscle open the latch with too much force. Would the handle bend or break off?

The same boundary conditions, constraints, material properties, meshing, and solver settings were used for the 294N side load analysis as in the vertical load analysis. The original design satisfied the side load abuse design requirement.

Alternate Failure - Maximum Load Analysis
The alternate failure analysis was a customerrequired test to make sure the hook could withstand a high-impact event. In this case, the hook had to withstand a maximum load of 4900N in the latch stroke direction. As in previous simulations, linear analysis was used to determine which parts could be omitted from the nonlinear analysis. In this case, all the parts were eliminated except for the clamp body, hook, roll pin 1, and receiver.

All of the parts were meshed with tet4 elements and, as earlier, a nonlinear analysis was run to identify the hot spots for high stress. The first simulation was used to refine the mesh. The receiver and roll pin remained tet4 elements. However, the clamp body and hook were re-meshed with tet10 elements. The local meshing and sag conditions were implemented around the contact areas and hot spots.

An advanced restraint was applied to roll pin 1, allowing only translational displacement in the latch stroke direction. The two translation degrees of freedom (DOF) in the non-stroke (loading) direction were removed. The same contact constraints were used as in the 600N latch abuse analysis, including inactive, glue, and touch. A distributed force of 4900N was applied across the roll pin in the latch stroke direction. The material properties for the latch analysis were set as in Table 1.

Because SimDesigner Motion considers the physics of the model, it allows interaction between bodies. For this analysis, 3D contacts were utilized. All of the contacts were modeled. The leaf springs were approximated, using linear spring-dampers with a revolute (hinge) joint where the pivot would be.

CATIA's DMU Kinematics tool was used to move the model to the fully opened position. SimDesigner Motion was used to build the model's motion constraints, including handle to ground, and to insert the torsional springs by hooking the clamp body to roll pin 1 and the lock button to the long roll pin.

As in the nonlinear abuse analysis, the solver was set for contact bias at 90% and Coulomb friction was activated. To meet the 4900N load, more expensive materials had to be used for this customer than before. For example, the receiver had been made of plastic, as it is for many other customers. However, the 4900N load requirement is significantly higher and required changing the receiver from plastic to steel. This increased material and process costs.

Since the earlier system had passed all of the requirements, it was possible that the use of a steel receiver was over-engineering. Using simulation, it was relatively easy to compare the customer's earlier design to the current version. By comparing the ultimate strength of the earlier system to the new design, the 4900N requirement was determined too severe. By changing the receiver from steel back to plastic, the new design was as strong, if not stronger, than the earlier system, and a competitive price point was maintained.

New Process, New Advantages

ASC's product development processes are expanding. In the past, normal operating conditions were the focus. However, customers are now asking which parts might fail under extreme conditions. When a mechanism fails a physical test, it can be embarrassing for the engineers and troubling for the customer. By using SimDesigner for CATIA V5, designers and analysts share one common interface and database. Multiple iterations can be virtually tested in the computer without building anything and investing in tooling. Failures and iterations occur virtually, are fixed virtually, and then when the best design is identified, verified with physical testing.

Moving from the old school of 'design, build, test, break' to 'design, analyze, and confirm,' ASC engineers identified and addressed the failure points in the handle without going through several stages of physical prototyping and testing. Physical testing is becoming the last step for verification. Using embedded analysis tools earlier in the design process, ASC's design engineering team was able to determine weak areas in the initial design before physical prototyping and testing, saving a substantial amount of time and money.