Adams Basic Package > Adams View > Adams View > Testing Models > Simulation Basics > Simulation Basics

Simulation Basics

After creating your model, or at any point in the modeling process, you can run a Simulation of the model to test its:

■Performance characteristics

■Response to a set of operating conditions

The entries in this section of the table of contents explain how to define the output desired from simulations and perform simulations

During a Simulation, Adams View performs the following operations:

■Sets the initial conditions for all the objects in your model.

■Formulates appropriate equations of motion based on the laws of Newtonian mechanics that predict how objects in your model will move given the set of forces and constraints acting on them.

■Solves the equations to within your specified accuracy tolerance for such information as part displacements, velocities, and acceleration, as well as applied and constraint forces.

■Temporarily saves the data calculated so that you can investigate your results using animations, plots, and numerical signal processing. You can also permanently save your results in your Modeling database.

As Adams View simulates your model and solves equations, it displays the calculated results as frames of an animation. The animation helps you graphically view the overall behavior of your model and pinpoint specific problems, such as improper connectivity or misapplied motions or forces. After the simulation is complete, you can replay the animation. For more information, see Animation Controls Basics.

Adams View can display this information in Strip charts through measures or you can view the information in Adams PostProcessor for more in-depth investigation and manipulation. See the Adams PostProcessor online help.

Types of Simulations

You can run five types of Simulations in Adams View:

■Dynamic simulation

■Kinematic simulation

■Static equilibrium

■Initial conditions simulation

■Linear simulation

About Dynamic Simulations

Unlike kinematic and static simulations, which involve the solution of only algebraic equations, dynamic simulations are more complex because they involve the solution of differential and algebraic equations (DAEs). Two basic types of algorithms are available in Adams Solver to perform the numerical integration required for dynamic analyses:

■Stiff solution methods that use implicit, backward difference formulations (BDF) to solve the DAEs.

■Non-stiff solution methods that use explicit formulations to solve ordinary differential equations (ODEs) that are obtained from the DAEs by way of coordinate partitioning methods.

In both cases, implicit methods are applied to the formulations to find a solution.

There are four stiff integrators and one non-stiff integrator currently available in Adams Solver. The four stiff integrators are:

■Gear (GSTIFF)

■Modified Gear (WSTIFF)

■Constant BDF (Adams Solver (FORTRAN) only)

■RKF45 (Adams Solver (FORTRAN) only)

The non-stiff integrator is Adams-Bashforth-Adams-Moulton (ABAM) (Adams Solver (FORTRAN) only).

There are two new integrators, HHT (Hilber-Hughes-Taylor) and Newmark, that are only for Adams Solver (C++).

About Controlling the Dynamic Solution

When you first build a model and decide to test it by performing a Dynamic simulation, you should always run two tests: the first using the default integration accuracy and the second using an accuracy 10 times tighter. Comparing results of these different simulations indicate if the numerical results are good approximations of the true solution.

If you see noticeable differences in your corresponding plots between the two simulations you ran with different tolerances, you should tighten the integration tolerance by a factor of 10 again, perform another test, and compare the results of the last two simulations. Repeat this process until you see no noticeable differences in results. When this happens, use the looser integration accuracy because it typically results in a faster simulation.

You can change several dynamic simulation parameters to help you overcome convergence failures:

■Set accuracy - You should always begin by loosening or increasing the convergence tolerance that must be met by all changes in part displacements and forces during the corrector phase.

■Change number of iterations - You can increase the number of iterations attempted by Adams Solver during each corrector phase.

■Recalculate the Jacobian matrix - You can increase the frequency with which Adams Solver recalculates the Jacobian matrix (the matrix of partial differentials) during the corrector phase. In an attempt to gain greater efficiency, Adams Solver uses a modified Newton-Raphson approach that does not update the Jacobian at every iteration.

■Control the maximum step size

Note that you may not always help the solution when you change the default parameters for convergence tolerance, maximum number of iterations, and pattern for updating the Jacobian. For example, if you loosen the convergence tolerance, you can allow too much error to build up in your solution over time and your overall solution accuracy could suffer.

If you increase the number of iterations that Adams Solver attempts during each corrector phase, you might be making the solution less efficient. Often, when Adams Solver cannot get the corrector to converge using the default number of iterations, it is better to let the solution step back in time and predict forward using a smaller time step rather than attempt more corrector iterations.

For more information on the effects of making these changes and tips for controlling the dynamic solution, see the INTEGRATOR statement in the Adams Solver online help.

Comparison of Integrators

The integrator:	Has the following characteristics:
GSTIFF	■Uses backward differentiation formulas. ■Uses fixed coefficients for prediction and correction. ■Is the default method.
WSTIFF	■Uses backward differentiation formulas. ■Uses variable coefficients for prediction and correction.
Constant BDF	■Uses backward differentiation formulas. ■Uses fixed coefficients for prediction and correction. ■The maximum allowed step size controls the integrator accuracy. ■Does not calculate local integration error at each step.
ABAM	■Uses coordinate partitioning to reduce the full set of differential and algebraic equations (DAEs) to a smaller set of ordinary differential equations (ODEs). ■Uses Adams-Bashforth for prediction; uses Adams-Moulton for correction.
RKF45	■Single-step method. ■Primarily designed to solve non-stiff and mildly stiff differential equations when derivative evaluations are not expensive. ■Not highly accurate.
HHT	■Adams Solver (C++) only ■Expected to result in a smaller number of Jacobian evaluations. ■Unlike BDF-type formulas, it behaves like a low pass filter; it cuts high frequency spurious oscillations, while accurately preserving low frequency oscillations. ■The cutoff frequency can be controlled by adjusting the option Alpha; the smaller value, the lower the cutoff threshold. ■Stable at small value of the integration step size.
Newmark	■Adams Solver (C++) only ■Behavior similar to HHT. ■Lower order. ■Has two control parameters.

Equation Formulation Comparison

The following briefly compares the different equation formulations. For more information, see the INTEGRATOR statement in the Adams Solver online help. See Solver Settings - Dynamics dialog box help.

The equation formulation:	Has the following characteristics:
I3	■Ensures that the solution satisfies all constraints. ■Does not ensure that the the velocities and accelerations calculated satisfy all first- and second-time derivatives. ■Monitors integration error only in system displacements, not in velocities. ■Is fast. ■The Jacobian matrix can become ill-conditioned at small step sizes.
SI2	■Takes into account constraint derivatives when solving for equations of motion. This process enables the GSTIFF integrator to monitor the integration error of velocity variables, and, therefore, renders highly accurate simulations. ■Jacobian matrix remains stable at small step sizes, which in turn increases the stability and robustness of the corrector at small step sizes.
SI1	■Takes into account constraint derivatives when solving for equations of motion. ■Monitors the integration error on the impulse of the Lagrange Multipliers in the system. These additional safeguards enable the integrators to monitor the integrator error in velocity variables and the impulse of the Lagrange Multipliers. ■Very accurate. ■Jacobian matrix remains stable at small step sizes, which in turn increases the stability and robustness of the corrector at small step sizes.