View Full Version : Summary: Multibody Dynamics Software

Morris Alan R
12-19-1995, 01:45 AM
Summary of responses --- MULTIBODY DYNAMICS SOFTWARE

Thanks to all that responded!


From: Victor Ng-Thow-Hing
To: morrisa@ecf.toronto.edu
Subject: Re: Multibody Dynamics Software

It's Victor from dgp at UofT. We use SD/FAST and have been satisfied
with its code. I've personally have used it for modelling muscles using
point to point actuators. The big problem is that actual muscle dynamics
is highly nonlinear. Many numerical integration schemes (such as the
RungeKutta methods in SD/FAST) do not perform quickly for these stiff
systems. Furthermore, Joint limits create stiff systems also. As for
muscle models. If you intend to approximate a muscle with a spring-damper
system, you would get reasonable behaviour. I tried using models based on
Hill and Zajac's work and got very nonlinear behaviour. I'm currently
working on muscle geometry, but intend to go back to physical modelling
in the near future.


With regard to your message on BIOMCH-L concerned with multi-body
modelling, we have used SD/FAST for many years to model a motor-
cycle and dummy rider in a computer simulation of them crashing
into a wall or a moving car. SD/FAST produces the fortran
code which describes the dummy and the bike and sets up the
equations of motion. We then add contact equations to describe
the impacts between the various parts and use the ACSL simulat-
ion package to solve the system.

We have found that when you get used to it the process works
well. If you'd like to know more you could contact our
researcher, Ms L-S.M.Mo who would be happy to tell you of
her experiences with the software.
Her e-mail address is: Lai-Sheung.Mo@Brunel.ac.uk

Yours sincerely,

Professor Alan. L. Yettram,
Dept. of Mech Eng.,
Brunel University,
Middlesex, UB8 3PH,



From: Scott Delp

We have used SD/FAST to develop a number of muscle-actuated dynamic
simulations. A graduate student who works with me, Steve Piazza, has
written a paper that gives the details of the musculoskeletal model and the
simulation results for the swing phase of normal and stiff-knee gait. I
would be happy to send you a pre-print of this article if you are
interested. In addition to SD/FAST, we used the "Dynamics Pipeline,"
software that was written by Peter Loan while he was working at the VA
Medical Center in Palo Alto and at MusculoGraphics Inc. You may want to
conatact Pete at jloan@merle.acns.nwu.edu if you want additional

Scott Delp


From: Rene Steiner

although not quit up-to-date the following book proved to be useful
when we had a similar decision a couple of yeras ago:

Schiehlen, W., ed.: Multibody Systems Handbook, Springer, Berlin, 1990

This is essentially a comparision of a couple of commercial multibody
dynamics software packages. As far as I remember ADAMS, DADS and AUTOLEV
are covered and a few others are analyzed.

At the time (circa 1992) we found that none of the commercial systems
that would provide C-code for the equations of motion was able to
handle a system as lage as ours (17 segments, 42 DOF). Therefore we
decided to write our own software (based on algorithms from
Wittenburg, J: Dynamic Systems of Rigid Bodies, Teubner, Stuttgart, 1977).

However, in this business three years are a long time and I would not
be surprised if the capacity of these packages had been substantially
improved. Especially SD/Fast seemed to be very promising and has been
used by Scott Delp et. al. for their commercial biomechanical simulation
package (as far as I know).

Sorry if this is not of much help to you but I am still interested
in the subject an curious about the other replies you get. Please
keep me posted.

Cheers, Rene


From: "Kim B. Blair"

Moldyn, Inc. is interested in continuing the application of
multibody dynamics to problems in the area of biomechanics.
Biomechanics researchers are becoming increasingly interested in
more than just motion information. By having access to multibody
dynamics analysis, researchers are able to calculate quantities
such as force, torque, momentum, energy and work.

Our multibody dynamics code is a computationally efficient
analysis software package designed for use by researchers for the
analysis of the dynamics and control of multibody systems. The
code is compartmentalized with user defined subroutines. These
subroutines can be modified for specific applications such as
muscle force transmission or surface contact models.

Typically, our company has utilized sources of government
funding, in particular the Small Business Innovative Research
(SBIR) program, for our research and development efforts. Often
these programs have involved cooperative research efforts with
academic institutions.

I have attached more detailed information about our company and
our dynamics analysis capabilities. As described in the
enclosure, Moldyn has been developing dynamics analysis software
modules for a variety of different technologies. In the area of
biomechanics, Moldyn is developing a version of its dynamics
analysis software for a product called Human-Machine-Task
Computer Aided Design (HMT-CAD). HMT-CAD focuses heavily on the
task analysis aspect of the human performance issue.

Today's researchers are moving toward laboratories with
state-of-the art biomechanics analysis systems including
components for data collection and systems analysis. It is
envisioned that researchers will find it advantageous to have
access to Peak's superior Motion Measurement Systems and
HMT-CAD's unique performance analysis capabilities with Moldyn's
dynamics analysis package providing a data and analysis bridge
between the systems.

It appears that our technologies in biomechanics may be complementary.
Moldyn is always interested in cultivating new research relationships
and aggressively pursuing research initiatives.


Kim B. Blair, Ph.D. Senior Staff Scientist
Moldyn, Inc. blairk@moldyn.com
955 Massachusetts Ave., 5th Floor (617) 354-3124 ext. 18
Cambridge, MA 02139-3180 (617) 491-4522 (fax)
++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++++++++++++++

I saw your posting on BIOMCH-L and I thought I should fill you in on the latest
information about our software products. MusculoGraphics is about to release
the Dynamics Pipeline (we are shooting for a January 1 release). I believe
you requested information about its availability from us earlier this year.

The Dynamics Pipeline connects our musculoskeletal modeling software, SIMM, to
SD/FAST. The software enables you to perform forward and inverse dynamics
simulations on models created within SIMM. For forward dynamics, you specify
on/off times of the muscles, and the Dynamics Pipeline will compute the
resulting motions of the body segments. The Pipeline has built into it several
models for computing the dynamic forces in muscles. You can choose among them
when you start a simulation, based on your accuracy and computation demands.
For inverse dynamics, you specify the time histories of the joint angles during
some activity, and the Pipeline will calculate the torques required to generate
that motion.

The basic steps to use the Dynamics Pipeline:

- Develop a musculoskeletal model in SIMM
- The Dynamics Pipeline translates it into a model that SD/FAST can understand
and then executes SD/FAST in order to generate the equations of motion for
the system.
- The Dynamics Pipeline then writes out additional code necessary to implement
the dynamic model, and compiles all of the code together to create a complete
simulation program.
- You can then run the simulation and load the results into SIMM for animation
and post-processing.

I hope this helps. Please let me know if you have any questions.

-- Art Wong
Director of Marketing and Operations
MusculoGraphics, Inc.
1840 Oak Ave.
Evanston, IL 60201
(708)866-1808 FAX


From: Geerling EJ Langenbach

In our lab we are using ADAMS and I-DEAS for respectively dynamic and FE modelling purposes.
Although ADAMS doesn't give you the equations of motion code, I want to give you my experience with this software package.
Because ADAMS is divided into two main packages (VIEW and SOLVER) it is very workable. The VIEW section is used to build the model and visualize its shape. SOLVER is the actual calculator of the dynamics. VIEW can then again be used to animate the movemen
t and to get all the graphical output. The graphical output is just great and very helpful during the building and testing process. I think the package is really powerfull and easy to use. I am using a SGI workstation for the analysis, but the package is
also available for PC's and the files are interchangeable.
I use ADAMS to model the mechanics of the human masticatory system: cranium, lower jaw, hyoid, and 18 muscles working in between these bodies. Modelling muscle actuators (fibre and tendon), including velocity-tension and length-tension relationships, is p
ossible using user defined functions. These are activated by B-spline functions, generated in another package and imported into ADAMS.

Hope to help you with this information.

Geerling EJ Langenbach, PhD
Research Associate
Department of Oral Biology,
Faculty of Dentistry,
The University of British Columbia,
2199 Wesbrook Mall,
Vancouver, BC,
Canada V6T 1Z3