Application of the LS-DYNA Solver to Engineering Challenges Presentation to the ANSYS User's Group of Western PA 
on October 11, 2001 
by Phillip H. Burnside, Ph.D.
PPG Industries, Glass Technology Center

Why do you need to understand LS-DYNA ?

The ANSYS/LS-DYNA Prep/Post only support some of the advanced features with LS-DYNA.
LS-DYNA used a different technology that lends itself to simulations of high speed events very efficiently. As a result what effects convergence is different.
ANSYS solvers work very well static or nearly static, but has problems with very short duration events.
LS-DYNA solvers work very well for short duration problems, but do not work well due to round off error with long duration or static events.

Some of the Features that You have to "Manual" Turn On

Material Failure (element death)
Most of the different Contact algorithms.
Material models for special materials like Foams, Honeycomb and advanced material models. (over a 100 material models)
Special functions like airbags and seatbelts.
For more see the LS-DYNA Manuals

Structure of ANSYS Models

Structure of LS-DYNA Models

What Controls Solution Time ?

Because LS-DYNA works totally in memory the speed of hard drives is not a factor.
The time step is based on the speed of sound (density) to cross the shortest side of any element in the model.
The number of elements can be in the tenís of thousands without significantly effecting the time of one iteration.

Concerns

Hour glassing of the elements is possible which will can cause divergence of the solver or, simply put, error into the solution.
The same assumptions about the various types of elements are still true. (i.e. a shell element needs to be at least 5 times longer on the two sides as it thick.)

Example of Modifying the Input deck to Change Material Model

FROM ANSYS:
*MAT_ELASTIC
  5  0.471E-04  0.300E+08  0.300000 0.0 0.0 0.0
*MAT_ELASTIC
  6  0.725E-03  0.300E+08  0.300000 0.0 0.0 0.0

CHANGE TO:
*MAT_ELASTIC_FLUID
  5  0.471E-04  0.0 0.0 0.0 0.0  320000.0
      0.3  1.0E+20 
*MAT_ELASTIC_FLUID
  6  0.725E-03  0.0 0.0 0.0 0.0  320000.0
      0.3  1.0E+20 

Example of Modifying the Input deck to Add Failure Criteria

FROM ANSYS:
*MAT_PLASTIC_KINEMATIC
     1 0.732E-03 0.300E+08 0.300000 0.100E+06 0.226E+06 0.00 
0.000E+00 0.000E+00 0.000E+00 

CHANGE TO:
*MAT_PLASTIC_KINEMATIC
     1 0.732E-03 0.300E+08 0.300000 0.100E+06 0.226E+06 0.00 
0.00 0.00 0.1473 

Example of Modifying the Input deck to Change how Stress Relaxation is Used

FROM ANSYS:
*INITIAL_VELOCITY_GENERATION
   7  3  0.00 651. 0.00 0.00 
   0.0000 0.0000 0.0000 1.0000 1.0000 1.0000 0

CHANGE TO:
$HAVE VELOCITIES ONLY APPLIED AFTER RELAXATION.
*INITIAL_VELOCITY_GENERATION
   7  3  0.00 651. 0.00 0.00 
  0.0000 0.0000 0.0000 1.0000 1.0000 1.0000 1

What to do If want Use LS-DYNA

Take the ANSYS/LS-DYNA prep/post class to learn how build a model in ANSYS so that it can be exported to the LS-DYNA solver.
Get a LS-DYNA set of manuals so that you can determine what you need to add/modify to the input deck if need a feature that is not currently supported by the LS-DYNA interface.

Repair of USS COLE

Dropping of a 70,000 lb Section of the Propeller Shaft

The area of interest is circled in red

What Happened ?

While the shaft was being rigged inside the hull of the ship a chain hoist failed. 
This dropped one end of the shaft about twelve inches onto a temporary support that was welded to a bulk head.
If this section of shaft had to be replaced it would take six months or more to get.  This would effectively delay the repairs to the USS COLE by those six months since the shafts being used were the battle spares.

Field Observations

The shaft did not appear to be damaged from the fall.
The temporary support on the bulkhead (a 12 C channel) was deformed about three inches.
A smaller support on the bulkhead was bent completely out of the way.

The model was built in ANSYS/LS-DYNA Prep/Post

FEA Results of Shaft Drop

Results of Shaft Drop

The model illustrated that the shaft experienced no plastic deformation at all. This agreed with the field observation.
The Model predicted that C channel would be permanently deformed about 3.2 inches. This also agreed with the field observations.
The model indicated that bulkhead would be deformed about 0.060 thousands which with acceptance criteria for flatness for a DDG. This was confirmed by second onsite inspection.

Potential Savings

Since the shaft was not damaged the potential savings (in time) was six months of yard time that the government and the ship yard would have had to agree on how to share the cost.
The shaft itself is a precision, 70 ft long piece of metal that would have cost over a $100,000 to replace.

Improving the Crash Worthiness of a Propane Semi-Trailer

The Dangers of the Current Design of a Propane Tanker
About every three years a 65,000 lb propane semi loses control and crashes sufficiently violently to rupture the tank usually killing 2-3 people.
These tankers are designed to meet DOT requirements while minimizing their weight so that the amount being hauled is maximized.

Initial Model Results

The current tanker only survived up to a 20 mph impact on either flat wall, cylindrical column, or a 45 degree angled impact on a flat wall.

Results of the Current Design

(Buckles inward due to impact)

Results of the Current Design

(Buckles inward due to impact)

 

Current Design

(into a Column)

Current Design

(into a Column)

Energy Absorbing Material: HEXCEL

This is hexagonal celled material made out of thin aluminum that crushes in bucking mode. Material properties were obtained from Hexcel Inc. 
For this application the underlying shell buckled before the maximum potential energy absorption could take place

Illustration of the HEXCEL

Energy Absorbing Material: LAST-A-FOAM 

This is rigid polyurethane that is cast in a simple mold and then glued onto the head of the tanker. To protect the foam from environmental deterioration, a thin sheet metal cover is applied over the foam.
This foam spreads the impact energy over a larger area, therefore, the foam can be used to it maximum potential.

Impact into Flat Wall

Impact into a Column

A 45o angled impact on a flat wall

Validation of the Small Model by Building Full Model

Results of LAST-A-FOAM 

The crash worthiness was increased from 20 mph for the original design to +55 mph for all three cases.
The impact simulation using the small model was validated by building a full length model that survived to the same crash worthiness.

Economics of Improving Crash Worthiness

The cost of the foam is $10,000 per tanker and 3,000 lb lost load capacity.
There are about 7,000 of these tanker on the road for about twenty years of service
The US DOT states that one death cost 2.5 million.
Based on these conditions it is cheaper to leave the tanker designs alone and just pay off when someone dies.

Replacement of Experimental Testing with FEA Models

Example:  A High Energy Door Slam to see if the Glass Breaks

Where is FEA going in the Automotive Industry ?

FEA is being used to determine if a new design will meet the customer requirements before PPG agrees to be DESIGN RESPONSIBLE.
In addition, this system can be used in the initial design phase with the automakers to determine if changes are needed in the systems to eliminate possible sources of failures.
Finally, this system can be used to determine probable causes of field failures.

Initial work on the Lincoln Explorer Laminated Sidelite

Ford wanted to use a laminated sidelite as a drop-in replacement for tempered glass to reduce the noise that is getting to the driver in the new Lincoln Explorer.
The model indicated that we would need highly heat strengthened glass in order to survive the door slam test that our competitorís glass failed.
Experimental testing validated that the high stresses found in the model are actually present during the slam test.

Example of a Door Slam

Potential Cost Savings

By doing testing virtually, the $300,000 estimated cost of testing and validation of the part can be reduced to about $20,000.
By getting the whole picture of what a design may see, redesigns can be done in the design when the cost of re-doing the design is relatively small.