1999-01-3760
Application of Computer Aided Engineering in the Design of Heavy-Duty Truck Frames
Carlos Cosme, Amir Ghasemi and Jimmy Gandevia
Western Star Trucks, Inc.
Copyright © 1999 Society of Automotive Engineers, Inc.

ABSTRACT
In recent years the heavy-duty Class 8 truck market has become very focused on weight and cost reduction. This represents a major challenge for design engineers since these vehicles are used in a wide variety of vocations from highway line haul to logging in severe off-road environments.
The challenge is to meet the weight and cost
reduction goals without sacrificing durability and performance. This paper discusses the integration of computer aided design and engineering software codes (Pro/Engineer,
ADAMS, and ANSYS) to simulate the effect of design changes to the truck frame .In particular, this paper discuses the development of an ADAMS multi-body dynamics model of a full truck and trailer to simulate vehicle handling, roll stability, ride performance, and durability loading. The model includes a flexible frame model using a component mode synthesis
approach with modes imported from a finite element analysis program. The link between the multi-body simulation and the finite element code is also used to transfer
loads back to the finite element model for stress analysis. Tight links between all the codes ensures that new design iterations can be quickly evaluated for concurrent
design and analysis. A detailed case study showing how this technology has been used is also included.
INTRODUCTION
Recently the heavy truck industry has experienced a large push to develop vehicles with reduced cost and weight. This has been a major challenge for truck manufacturers
as they look for ways to optimize their vehicle designs without sacrificing durability or performance.
Since the truck frame is a major component in the vehicle system, it is often identified for refinement. This paper outlines a computer aided engineering (CAE) procedure for analyzing changes to the truck frame and how these changes affect vehicle performance .The frame of a heavy truck is the backbone of the vehicle and integrates the main truck component systems such as the axles, suspension, power train, cab, and trailer.
The typical frame is a ladder structure consisting of two C channel rails connected by cross-members. The frame
rails vary greatly in length and cross-sectional dimensions depending on the truck application. Likewise, the
cross-members vary in design, weight, complexity, and cost. These variations will depend upon the cross-member purpose and location. Refer to Figure 1 for an illustration
of a truck frame. However, the effects of changes to the frame and cross-members are not well understood.
For example, if the torsional stiffness of a suspension cross-member is lowered, what is the effect on the vehicle’s
roll stability, handling, ride, and durability? Design engineers require answers to these types of questions to guide them in their work. In particular, a concurrent design and analysis procedure is required so that new
designs can be quickly evaluated.

Figure 1. Class 8 Heavy-Duty Truck Frame

COMPUTER AIDED ENGINEERING
In the last twenty years there has been an enormous growth in the development of CAE tools for automotive design. Much of this technology has been adopted by the truck industry as truck manufacturers look to improve their designs in a rapidly growing market. Today structural design is typically performed using two CAE tools: finite
element analysis (FEA), and multi-body system simulation (MSS). These are combined with computer aided design (CAD) software to improve design and analysis
communication.
CAD – In the last fifteen years CAD systems have replaced drawing boards as the method of choice for design. They enable designers and engineers to quickly
create realistic models of truck components, vehicle assemblies, and design drawings for manufacturing.
Advanced CAD systems are rich in features such as parametric solid model and large assembly management. They have evolved to become major databases for engineering information. In particular , CAD systems provide important data for downstream CAE applications.
FEA – Finite element analysis is usually used by engineers to study the strength of structural components.
Typical FEA activity is focused on analyzing structural stresses, deflections, and natural frequencies. The analysis begins with a discretized representation of a structure
known as a mesh. The mesh is composed of nodes and elements and is often created with geometry from a CAD system. The nodes represent points on the structure where displacements are calculated. The elements are bounded by sets of nodes and enclose areas or volumes. They define the local mass, stiffness, and damping properties of the structure. Equations relating these quantities
to the nodal displacements are automatically developed by the software codes. Other inputs, such as boundary conditions, applied loads, and material properties, must  be defined by the user. Each of these quantities requires careful  judgement for meaningful results to be achieved. Results post-processing includes images of deformed structures under load, coloured stress contours, and mode shape animations.
MSS – Multi-body system simulation is used to study the motion of components and assemblies and is often used to study a vehicle suspension or a vehicle’s handling and ride response. A typical MSS model of a full vehicle will be composed of rigid bodies (wheels, axles, frame , engine, cab, and trailer) connected by idealized joints and
force elements. The MSS code automatically develops  the non-linear differential and algebraic equations that define the motion of the bodies in the model. The equations are numerically integrated to produce time histories of rigid body displacements, velocities, accelerations, and forces. Results are viewed as graphs and animations of
the system motion. As with FEA, CAD data is often used to develop a MSS model. Geometry data from a CAD assembly is used to establish the layout of the MSS model such as the location of joints and force elements. CAD solid model data is also used to estimate the location of the center-of-mass and the inertial properties of each rigid body. Forces acting on a rigid body from a MSS can be used as input loads to a finite element analysis to determine the structural stresses in that rigid body.
The CAE tools discussed in this paper include Pro/Engineer for CAD, ANSYS for FEA, and ADAMS for MSS. The following discussion references the specific capabilities of these codes in developing a customized environment for the engineering analysis of truck frames.
CAE CUSTOMIZATION FOR HEAVY TRUCK
MODELLING
As described above, the current offering of CAD and CAE tools provide a great deal of integration. Nonetheless,
these tools are very general in scope and a significant customization effort is required for the analysis of heavy duty trucks and truck frames. To fully understand how changes to the truck frame impact vehicle handling, roll
stability, ride, and durability requires a detailed  MSS model that can simulate all these effects. Using the
ADAMS software code such a model was developed at
Western Star Trucks. Refer to Figure 2 for a view of the model in the ADAMS environment.

Figure 2. ADAMS MSS Model

The model includes the following characteristics:
• 100 rigid bodies
• 180 force elements
• 45 joint elements
• 415 degrees-of-freedom
The rigid bodies include the frame, cab, axles, wheels ,engine, hood, radiator, leaf springs, suspension arms, drive shafts, and the trailer. Mass properties for many of
these bodies were estimated using simplified solid models in Pro/Engineer. The force elements include linear and non-linear bush ielements that model rubber isolators, such as the cab and engine mounts. Non-linear single component forces are used to model air springs and shock absorbers. Property data for these elements are derived from tests performed by component suppliers. Revolute joints and
spherical joints are used to model connection points, such as wheel bearings and torque rod pivots, respectively. Pro/Engineer assemblies are used to determine the geometric location of these elements.
Since the heavy truck industry offers a wide variety of vehicle layouts, the locations of many of the truck’s subsystems were made parametric for easy modification. For example, the front axle subassembly(wheels, axles, leaf springs, and shock absorbers) were linked to a variable defining the longitudinal position of the front axle. Using
this technique, truck models with different front axle positions can be quickly developed by changing the value of this variable. This procedure was duplicated for the following
subassemblies: rear suspension, cab ,engine ,hood, and fifth wheel and trailer .Tire to road contact is handled with the ADAMS built-in tire routines and includes models for tire handling and durability forces. In ADAMS road profiles are represented
as a mesh of triangles similar to a finite element mesh. The geometry and mesh for the road profiles are generated with Pro/Engineer. A custom software program is
then used to translate the mesh into two files for ADAMS :a road file format for the solver to determine the tire/road interaction forces, and a graphics format to view the road
during post-processing animation. These files are stored in a common directory for easy retrieval. Custom control algorithms were developed to control vehicle speed, steering, and drive torque. These functions
can be quickly modified to execute different vehicle maneuvers such as roll stability, a high speed lane change, or durability bumps similar to a proving ground.
After the simulations are run, the forces and torques acting on the frame are written to data files. A custom software program is then used to extract the loads at specific
time steps and write them to an ANSYS load file. The load file is then read into ANSYS and applied to a finite element model of the frame. The frame stresses are then calculated using an inertial relief solution.
In summary, the model uses custom software routines and the existing links between the CAD and CAE codes to create a custom environment for evaluating the performance and durability of a heavy-duty truck. However, the model assumes that the truck frame is a rigid, under formable body. In reality, the truck frame contains a great deal
of flexibility which can impact vehicle performance and stability. As a result, these effects must be captured in the multi-body system simulation.
CAE SOLUTION FOR FRAME FLEXIBILITY
PREVIOUS TECHNIQUES – In the past, several techniques have been employed to capture frame flexibility in a MSS model. Three popular methods are: bushings,
mass beam elements, and FEA super element reduction. In the first method the frame is divided into two or more rigid bodies connected together with force elements having bushing-like properties: stiffness and damping in three translational directions and three rotational directions. The bushing properties are adjusted to give the overall frame bending and torsional stiffness.
As can be expected, this method is cumbersome to use, and if properly tuned, it will be capable of capturing only the fundamental bending and torsional modes of the frame. In the second method the frame is divided into a large number of rigid bodies interconnected by massless beam elements. This is similar to the bushing method but many more rigid bodies are usually used, and they are connected with massless beam elements whose equations (Timoshenko beam theory) are better suited to modelling
truck frame rails and cross-members. Nonetheless, it istime consuming to build a frame with this method and careful tuning of the beam elements is still required to capture the frame’s flexural response. The third method is the most accurate of the three methods and is based on a finite element representation of the frame. In this method the finite element model is reduced
to a super element representation with the overall stiffness and mass properties condensed to a set of master nodes. The reduced model is checked against the original finite element model to ensure that the important frame dynamics are still captured. It is then imported into the MSS environment where the super elements and
master nodes are converted to an equivalent representation of rigid bodies and force elements. Although this method is based on a finite element solution, it can still be difficult to achieve accurate results. For example, care must be taken in selecting the master nodes to ensure that the mass and stiffness condensation process is accurate.
All the methods described above are difficult to use for creating an accurate flexible model of a truck frame. In general, they are only capable of capturing the basic frame response: the first few bending and torsional
modes and the gross frame stiffness. If each method is to work, a significant effort is required to tune its properties to match some reference, such as static deflection testing,
modal testing, or finite element simulation results. Consequently, neither method is suitable for use in a concurrent design and analysis environment - it would simply take too long to make changes to the model, and it would not have adequate spatial resolution to capture subtle design changes to the frame.
COMPONENT MODE SYNTHESIS TECHNIQUE –
Recent advances in the integration of FEA and MSS have overcome the difficulties in the methods described above .It is now possible to use a finite element model directly in a multi-body simulation using a modal superposition technique known as component mode synthesis (CMS).Using modal superposition, the deformation of a structure can be described by the contribution of each of its modes. Normally, a very large number of modes are required to accurately capture the deformations at points where constraints are applied to the structure. CMS was developed to alleviate this problem. It combines normal modes with constraint modes. These constraint modes,
or static shapes, capture the deformation of key areas of the structure without having to maintain an excessive number of normal modes. As a result, they are computationally more efficient. The CMS procedure adopted in the ADAMS code is based on a modified version of the Craig-Bampton approach. In this method the structure is considered to
have interface points where constraints and forces are applied, and each interface point can have up to six degrees-of-freedom: three translations and three rotations. The motion of the structure is then described by a combination of two sets of modes: constraint modes for
the interface points, and fixed interface normal modes. A constraint mode is calculated for each degree-of-freedom
of an interface point, and it describes the static shape of the structure when that degree-of-freedom is given a unit deflection while keeping the degrees-of-freedom of all the other interface points fixed. This procedure is repeated to develop a family of constraint modes for all the interface points. Since the constraint modes are static shapes, their frequency information is unknown. The fixed interface normal modes represent the normal modes of the entire structure when all the degrees-of-freedom of all the interface points are held fixed.
In this form, the Craig-Bampton modes are not ideally suited for integration with the multi-body equations of motion. For example, the constraint modes add rigid body modes which conflict with the ADAMS non-linear rigid body motions. Also, the constraint modes may contain high frequencies that are difficult to solve. In the ADAMS implementation these problems are handled by orthogonalizing the Craig-Bampton modes. This identifies the rigid body modes making them easy to disable. It also
adds frequency information to the constraint modes which is valuable for setting integration parameters during the multi-body simulation. After orthogonalization, a modified set of modes exist: normal modes for the unconstrained structure (free-free like modes similar to those calculated in a typical FEA eigenvalue run), and the interface degrees-of-freedom. See Ottarsson [3] for a complete description of this method.
All the modal calculations described above take place in the ANSYS environment and are performed on a finite element model of the frame. To compute the modes, the
user selects the nodes representing the interface points where forces and constraints enter the frame, and then runs a macro that executes the appropriate ANSYS commands. The number of normal modes to include in the calculations are passed as a parameter to the macro. The final set of modes are written to a modal neutral file
MNF) that can be read by ADAMS.
The advantage of this modal superposition method are many and include:
• The frame is represented by a single modal neutralfile. As a result, it is very easy to reuse the frame in other MSS models. The files can be stored in common
directory for archiving and future use.
• In the MSS model the frame is represented as a single flexible body and not a large number of rigid bodies. bodies. This makes it much easier to manipulate the
frame in the model.
• Every flexible body mode adds only one degree-of freedom to the simulation. Previous methods added many more degrees of freedom since they used a
large number of rigid bodies and each of these added six degrees-of-freedom.
• The linear, flexible characteristics of frame model are more accurate since they are based on a full finite element model and not a collection of rigid bodies and force elements. This makes it much easier to
tune the model to agree with modal test results.
• Damping is added on a modal basis. Thus, damping results from modal testing can be easily added for increased accuracy.
• Modal participation during a simulation can be tracked based on the strain energy contribution. Modes that do not contribute significantly can be deactivated for improved computational efficiency.
• Visualization of simulation results is improved since the FEA mesh exists in the MSS environment and it can be used for viewing frame deformations during
animation.
• The transfer of MSS loads back to the original FEA model for stress analysis is improved since loads are associated with nodes in the finite element mesh. Although, this method has many advantages, it can still be time consuming to implement. For example, not all joint and force elements are supported for direct connection to the frame mesh when it is the MSS environment. These must first be connected to massless rigid bodies that are then locked to nodes of the mesh using fixed joints which remove the degrees-of-freedom added by the massless rigid bodies. Since a truck frame can have
36 or more connection points in a MSS model, it can be very time consuming to connect a flexible frame. Also, if an existing flexible frame needs to be replaced with a
new one to evaluate a design change, more modeling effort is required and the potential to introduce modeling errors is great.
To overcome this difficulty custom programs were developed to integrate a flexible frame in a vehicle model. The process begins with a full vehicle model with a rigid frame. A copy of the model is made and then a series of macro programs are executed that perform the following tasks on the copy:
• Read the flexible body modal neutral file and position the flexible body in the vehicle model.
• Create and connect massless rigid bodies at each node where forces and constraints are applied to the
flexible body.
• Modify all connections to the existing rigid body of the frame so that they connect to the appropriate massless rigid bodies.
• Delete the rigid body that previously represented the frame. In a typical analysis, a simulation will be run with the rigid frame as a baseline, and then after importing the flexible frame, it is repeated. The results of both are studied to understand the influence of the frame’s flexibility. Then another copy of the model with the rigid frame is made
and it is converted to have a different flexible body (a different modal neutral file is read into the modelling database). This new flexible body will have some design change that needs to be evaluated such as a new crossmember .The same simulations are repeated and all the results are compared.
With this process the burden of adding a flexible frame to a model is greatly reduced. For example, it takes less than one minute to convert a model with a rigid frame to have a flexible frame. As a result, this procedure works very well for concurrent design and analysis.
FINITE ELEMENT MESH MODELLING – In order for these methods to work effectively, the frame’s finite element model must be easily created and modified to reflect changes required by designers. The approach adopted here begins with a Pro/Engineer solid model assembly of the frame rails and cross-members similar to
the one shown in Figure 1.The solid models for each component are created specifically
for finite element analysis meshing, and as a result, are simplified versions of the actual designs. The finite element mesh is created with an add-on module for Pro/ENGINEER called Pro/MESH. It contains features to simplifyfinite element meshing such as automatically determining midplane locations for shell elements, applying
global and local mesh controls, and defining element properties. When requested, a mesh of shell, beam, and mass elements is created that reflects the current CAD
Creating the mesh while inside the Pro/ENGINEER environment has many advantages. For example, changes to
the solid models are automatically reflected in the mesh.
Thus changes to the frame rail geometry or the position of cross-members can be quickly meshed and exported to ANSYS. Likewise, beam elements that represent bolt
connections follow components as they are moved to new positions. Subassemblies, such as cross-members, can also be suppressed so that they are not included in
the mesh. This makes it very easy to turn on and off different cross-member designs and develop meshes of each configuration.
After the mesh is created, it is read into ANSYS, and a custom macro is run that renumbers the nodes where forces and constraints will eventually be applied in
ADAMS. This ensures that the same node numbers are always used. Nodes for constraint mode calculation are then selected and the modal neutral file for ADAMS is created. The file is then ready for input to the ADAMS model.
As described earlier, the ADAMS simulation loads are transferred to ANSYS and the stresses acting on a finite element model of the frame are calculated. This can be the same finite element model that was used to develop the modal neutral file for ADAMS or it can be a different model with a finer mesh. For a different finite element model to work, its mesh must have the same node numbers
at the connection points as the original model used for ADAMS. This is easily achieved by executing the node renumbering macro as used earlier. Consequently, a model with a refined mesh can be used for stress calculations, while a model with a coarse mesh can be used to
calculate the modal properties for ADAMS.
A sub modelling approach was also developed so that cross-member stresses could be computed more accurately.
First a finite element analysis is performed on the same coarse finite element model as used for ADAMS.
Custom macros in ANSYS are then used to read the results and extract the forces and torques from beam elements that represent the bolt connections between the frame rails and the cross-members. These are then applied to a detailed finite element model of the cross member alone. The bolt forces are applied using ANSYS RBE3 elements that are configured to distribute the bolt loads to nodes that would be under the bolt head of the two mating components. Again, this is achieved with custom macros and is completed in a matter of minutes. Mesh controls in Pro/MESH are used to ensure that an annular ring of nodes and elements exist to represent the bolt hole and the area under the bolt head. This technique
can be applied to finite element models with shell or solid elements.
VALIDATION OF FRAME FLEXIBILITY
In order to establish the accuracy of the flexible frame model, a modal test was conducted on a bare truck frame. An ANSYS finite element model of the same frame was then built using the procedures described
above and the eigenvalues and eigenvectors were computed for comparison. As can be seen in Table 1, the finite element model agrees very well with the test results. A modal neutral file was then produced for
ADAMS and the modes were recomputed using the ADAMS/Linear option. The ADAMS modal frequencies also agree well with the test data and confirm that the
flexible frame provides an accurate representation of the real structure. Note, only modes up to 56 Hz were extracted from the modal test data. Higher modes exist,
but the test did not contain enough measurement points to clearly define the mode shapes. The frame model was then incorporated into the full vehicle
MSS model using the procedures described above.
The modes for the vehicle were then computed and compared to modal test data for a linehaul truck with a similar configuration. See Table 2. Again the agreement is good and indicates that the dynamics of a typical linehaul truck
are well represented by the model.
CONCLUSION
The aim of the project outlined in this paper was to develop a process by which design changes to a truck frame could be quickly evaluated such that concurrent design and analysis would be possible. As described
above, this goal has been achieved by combining current computer aided design and engineering codes with custom
software routines. The process takes advantage of the strengths of each code to create a high fidelity environment where the impact of subtle design changes to the truck frame can be measured against vehicle performance and durability requirements. Design alternatives can be quickly evaluated and fed back to the designer while it is still possible to make changes. Although this process is being used successfully, there
are many areas where further enhancements can be made and will be the focus of future development. These include:
1. Validation of simulation results using time history data from full vehicle tests. This will be used to understand the simulation weaknesses and tune model parameters. As the simulation accuracy improves, the models will provide better data for componentoptimization.
2. Flexibility using the component mode synthesis technique will be added to other structures such as the cab/sleeper and the trailer. The flexibility of both these structures is believed to play an important role
in ride and durability.
3. New CAE technologies such as fatigue analysis will be added. Computer aided engineering codes for fatigue life estimation have improved dramatically in recent years, and it is now possible to estimate the fatigue damage to a structure using the full time history loads from a multi-body simulation. Fatigue life contours can then be viewed on finite element models just as stress contours are today. This technology greatly enhances durability analysis and the development
of a virtual proving ground.
ACKNOWLEDGMENTS
We would like to thank the management team at Western Star Trucks for having the vision to support the development of an integrated CAE process. We would also like to thank Ken Murray, Don Moore, Bob Perra, and Mark Gobessi for their careful review of the paper.