Jay Tedeschi, Senior Solutions Evangelist Manufacturing Solutions Div.,Autodesk

How to integrate simulation within digital prototyping: A sample process

HTC Sweden used digital prototyping to simulate and refine motions of the polishing head on this mobile machine for grinding and polishing stone floors.  “It is time-to-market that is crucial to us,” explains Karl Thysell, HTC founder.  “We have to be able to launch products at a rate our
competitors cannot keep up with.”

For centuries, mankind has struggled to understand complex design problems.  Often, the path to understanding has involved building a physical prototype and testing it to ensure that a design adequately met the demands of the real world.

Leading organizations, such as Mercedes Benz and Boeing, work to speed this process by building digital prototypes. According to a recent Aberdeen Group study, best-in-class manufacturing companies are using digital prototyping (DP) and simulation to test digital models for proper function prior to building a physical model.

Simulation versus building
To start, it’s important to understand the distinction between simulating reality and building reality, and the difference between simulation and modeling tools.

The latter point is important because the structure of a simulation model can be somewhat counterintuitive for a user familiar with standard structure in an assembly-modeling tool. The main difference is that the simulation tool requires components to be grouped according to their use in the simulation model as rigid bodies, whereas the assembly modeler typically handles components based on their participation in an engineering bill of materials (BOM).

To build a better model…  Consider a sub-assembly with inherent motion, such as a damper assembly or shock absorber. The modeling tool sees this as a single component, albeit one capable of compressive motion along the damper axis. This same component would be “seen” by a simulation tool as a single “rigid body” with no motion of its own within the body. Therefore, in the simulation model, this damper must actually be two components, an upper and lower damper, with the motion relative to one another described by a sliding joint.

After configuring the model, the next step is to add joints, which allow and describe the movement between rigid bodies in the simulation. There are many ways to accomplish this, and most simulation tools allow the user to create these joints manually, automatically (from existing assembly constraints), or a mix of both techniques. For the purpose of simulation, proper function of the assembly depends entirely on the types of joints used, and where they are used.

For example, a sleeve that translates along a shaft requires a type of joint that describes this motion. Most engineers intuitively would suggest either a cylindrical joint with one translational degree of freedom (1 DOF) plus 1 rotational DOF.  Fewer would use a prismatic joint (1 translational DOF).  But the conditions these types of joints describe would only exist if the sleeve were manufactured to impossibly tight tolerances in relation to the shaft diameter. More realistically, the sleeve mechanism is something akin to a point on a line (the centroid of the sleeve ID along the axis of the shaft), or even more accurately, 3D contact.

Without this insight, you might construct a model with a highly accurate or realistic representation of the joint required. When you run the simulation, chances are you’ll find no apparent errors with your model, but unfortunately you will see almost no progress during the simulation either. One percent of such a simulation run can take almost 10 minutes to execute…and you stare blankly at your screen wondering what went wrong.
Congratulations… you’ve built a model so accurate that simulation of its motion will require the horsepower of a Cray computer. You’ve built reality.

…Or not to build it.  Let’s assume that the goal of simulation is to find the forces exerted on the sleeve when it comes to a stop at the bottom of the shaft. With the 3D contact model/joint configuration described above, the sleeve traveling down the shaft would tend to rock back and forth, contacting the shaft diameter at the edges of the upper and lower sleeve openings. The parasitic friction from this contact would slow the rate of descent for the sleeve, and yield a fairly good result at the end of the run.

Unfortunately, this configuration also presents the simulation software’s solver function with the task of calculating a solution to a problem with 4 DOF; hence the excruciatingly long run times. There is a better way, and all it takes is keeping in mind what your required results are from the simulation run. In contrast to assembly modeling, simulation requires you to consider in advance what data you require, so that you can find out the information you need to achieve the desired  performance of a designed assembly.

The 3D contact joint would be the most accurate representation of reality; however, it is also the most costly in terms of calculation requirements. Instead, we can choose a much
simpler representation of the assembly motion that is easily handled by the solver.  The prismatic joint, with the addition of frictional coefficient properties, would yield an almost identical result at the end of the simulation run, with much lower requirements for calculation time and resources.

Simulating reality
Based on this insight, let’s consider what is required to design a cylinder head assembly. There are many design problems to tackle. Where do you start? Using a hypothetical valve assembly, the workflow that follows will show how to use simulation as a design tool and as an analysis tool. We will use simulation to evaluate an assembly’s performance even before all of its components have been designed, to understand the component properties required for desired performance—and then generate accurate specifications. This is probably the least-understood use of simulation, but it’s an application for which most simulation tools are uniquely suited.

Start with a 2D sketch, above right, of a cross section of a cylinder head. The many components that make up this assembly are shown in different colors, from the cylinder head itself to valve guides, valve seats, valve stem collets, and collet retainers. To evaluate these components’ motion, think of the assembly as comprising two component groups: the valve assembly itself, which is capable of motion; and the cylinder head, which is grounded. The point of contact between the two is the valve seat, which can be considered part of the cylinder-head component group.

Use a prismatic joint to describe the motion of the intake valve group. Once the joint is created, most simulation tools let you drag the valve dynamically through its range of motion and confirm that its movement is as expected.

Create a cam profile that induces motion of the valve group within the cylinder-head assembly. We sketch just a pair of concentric circles whose radial distance from each other matches our valve lift. Depending on the simulation tool, you may need to place some work or construction geometry on the cylinder head sketch that corresponds to the location of the axis of rotation of the cam. We can use a rotation joint aligned to the valve seat to allow movement of the cam part.  This is a very basic 2D assembly with a grounded cylinder head, a valve component capable of moving along the valve guide axis, and a cam part that is capable of rotating about its axis.

A 2D cylinder-head assembly shows the many components.

Describe motion of the valve and cam component during the course of the simulation. By editing the properties of the joints, we can modify a wide range of parameters to set up the simulation environment and produce the movement desired.We need to define the velocity of the rotating cam and the timed movement of the valve relative to that rotation. We can use a constant rotational velocity for the cam, but a graphing tool is necessary to describe complex valve movement.

Graph complex valve movement for the cam profile.

Most simulation tools include a tool for this type of time/position input, or accept data in ASCII or spreadsheet format. When the motion of these two components is final, we can run the simulation and extract the full profile of the cam.

Trace essential reference data to verify cam profile.  Once we’ve created the movement of the valve mechanism, we can use that movement to create geometry and the parts used to manipulate those mechanisms. Typically some sort of trace function serves to extract information such as acceleration, velocity or position relative to some reference frame. Trace data usually can be viewed in a graphing tool or displayed on the model itself. The display option often allows the trace to be exported to a sketch, which is how we will create our cam profile.

Set the trace point to follow valve motion.

For this simulation, we set the trace point on the valve component to follow its motion. Because we need to “trace” this motion relative to the position of the cam, the cam has to be selected as the reference frame. Running the simulation generates the cam profile, and this geometry can be saved to a sketch in the cam part.

Repurposing simulation
This example shows how simulation can solve a design problem, specifically the generation of cam geometry to drive a valve through a specific range of motion. We essentially reverse-engineered the cam, based on desired motion and timing of the cam and valve. This simulation can be used to solve other design problems as well, such as component sizing and selection.

For our sample cylinder head, we need valve springs that will close the valve with a minimum amount of play or “bounce.” The valve needs to seat very quickly, as the next stroke after
intake is compression, and any loss of seal between the valve and seat will lead to less efficient compression.

Add contact joints to gauge performance of the assembly.  The cam will affect the tappet via a 3D (or 2D) contact joint, and the same condition will be used to define the interface between the valve and valve seat. The same rule of thumb exists here, as before: If it’s possible to describe a joint with fewer degrees of freedom, do so.  The example described above is definitely a 3D contact problem, however if you were to drive a work-plane through the entire assembly, a set of sketches could be created from that cross-section, and in this situation a 2D contact joint could be used instead.  This is a much simpler problem for the solver, which will lead to much better simulation performance and a quicker run.

Valve cam contact joint impacts the tappet.

Once the 3D contact joint is in place, we can add the spring joint to the top end of the valve component below—or the simulation tool you are using may allow you to add a spring rate to the prismatic joint itself.

Start with a very weak spring and evaluate the valve closure performance. Since the cam doesn’t factor into this simulation, it’s helpful to suppress the contact joint and rotation joint that controls the cam. Set an initial position for the valve to its maximum lift to preload the spring. During the simulation run, the spring will force the valve closed, and the graph below helps to evaluate the position and impact of the valve as it seats.

Start with a weak spring set-up to determine valve closure performance.

Simulation shows the weak spring specification affects valve-closure performance.

These data help to understand and achieve the crucial balance between the strength of the spring, which closes the valve, and the cam force required to open the valve. Whether the simulation tool you use allows you to specify spring strength or optimize spring rate by minimizing motion relative to a joint, running the simulation again with the cam and contact joint function shows the amount of motion of the joint versus the force required to open the valve. The graph below shows that with spring rate set at about 10 N, the amount of force required to open the valve—in this case about 400 N—peaks at about the same time the valve reaches its fully open position.

Simulation shows the force and valve position based on initial spring-strength specification.

Force of closure is the next step for evaluation.  In addition to rate and nature of closure, we can look at stresses generated on the valve when it closes. One of the results that we can use from our simulation is the amount of force generated on the contact joint between the valve and the valve seat.
 
From the graphing tool we will find the maximum force generated on the valve contact, a graph that will look very similar to the graph above that detailed the position of the valve itself. We can select that maximum force, pass it to the valve itself and then perform a stress analysis of the valve to extract structural information about the valve design, the maximum forces exerted through the valve from the contact, and its factor of safety.

This safety-factor plot shows that this amount of closure force will result in a valve design insufficient for our needs. At this point, we have made two important discoveries through simulation: the current spring rate will require an excessive amount of power at the cam to fully open the valve, and this same spring rate will place undue stresses on the valve itself.

Initial analysis indicates valve would not meet safety requirements.

Slight modification of spring stiffness resolves stress on the valve.

Revising spring strength improves the balance between force and valve position.

Reducing the spring stiffness to about 75 percent of its current strength results in a much better profile, with maximum force reduced to about 140 N.  What is particularly interesting is that maximum force occurs much earlier in the valve lift sequence, occurring at about 10% of the valve period.  This change also eliminates any factor-of-safety concerns, and maximum stresses on the valve are now at acceptable levels, as the revised safety-factor diagram and graph (below left) show.

Simulation technology has caught up with our big ideas for great product designs. In fact, it can play a much greater role within the digital prototyping workflow than most users realize. With some advance thinking about your workflow, it shouldn’t be too difficult to use simulation software to generate design information as well as input for stress analysis—and make it a best practice.

Autodesk
www.usa.autodesk.com/digitalprototyping

:: Design World ::

 

Source: :: Design World ::