1. TART: Monte Carlo Radiation Transport in Industrial Applicationsby
Dermott E. Cullen
University of California
Lawrence Livermore National Laboratory
L-128/P.O. Box 808
Livermore, CA 94550
Tele:
E.Mail:
Website:
This paper will soon be available at my website

2. What is TART? What’s it’s Pedigree?
TART: Monte Carlo Radiation Transport in Industrial Applications
What is TART? What’s it’s Pedigree?
Livermore production code for 40 years
Livermore’s equivalent of MCNP
Only recently released outside Livermore
Now available Internationally
The currently released version, TART2000,
Coupled Neutron-Photon
3-D Combinatorial Geometry
Time Dependent
Energy Range: Very Low up to 1 GeV

3. Why should you use TART? It is VERY Fast
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART?
It is VERY Fast
an order of magnitude or more faster than other codes
Uses Latest ENDF/B data
neutrons & photons
It runs on ANY computer
Mainframes
UNIX Workstations
IBM PC – Windows/Linux
Power MAC
My wristwatch (next year)
The TART System is VERY User friendly
Only about % is the Monte Carlo code
The remainder is tools to make your job easier

4. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
The TART System is a complete system
It helps you prepare and check input
Runs your Monte Carlo calculations
Helps you analyze your results
Interactive graphics is extensively used
In input preparation and checking
Overlaying results on your geometry
Viewing neutron and photon data
Interactive graphics is used
Not just to produce pretty pictures
But rather to improve understanding

5. Example 3-D plot of experimental setup
Cylindrical
Detector
Cylindrical Filter
Spherical Source

6. Example 2-D plot of experimental results
Note, the reflection
on the Filter
from the detector
Source
Filter
Detector

7. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
A well known advantage of Monte Carlo
is its ability to handle complicated geometry
TART try to optimize this advantage
Other codes allow first & second degree surfaces
Planes, spheres, cylinders,…
TART allows third & fourth degree surfaces
Cubic & quartic splines
Very fine, detailed surfaces
Torus
There is no limit to the detail of geometry
Everything is dynamically dimensioned
Here’s an example of a complete seven story building

8. Example 3-D plot of seven story building (the entire NIF facility)

9. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
TART does source and criticality problems
What makes TART so fast is experience
40 years of use at Livermore
Hundreds of man-years of use
Very Responsive to day-to-day User Needs
If you have a need, just ask
Very versatile Results
Allows 22 different types of tallies
If this doesn’t meet your need tells me; I’ll make it 23

10. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
Many Utility codes to analyze output
If you have a specific need, just ask
Extensive Verification
Against Measured Results
And Other Codes – MCNP & Sn Codes
Other codes stress variance reduction
That you must be an expert to use
TART includes variance reduction, but
I recommend you rely on built-in expert system
to allow non-experts to use it

11. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
Other codes start from basic principles
In my opinion this approach, ignores,
Everything we were taught in school
Everything we’ve learned since then
For TART I use every trick of the trade

12. Why should you use TART? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use TART? (continued)
For example, TART uses
continuous energy neutrons
and continuous energy kinematics
but multigroup cross sections
700 groups: 50 per energy decade
Multiband parameters in all group
Because sampling continuous energy cross
sections converges too slowly
and isn’t necessary
we can use self-shielding theory
and the multiband method

13. U-238 total cross section, 1 to 2.15 keV.
Over 3 orders of magnitude variation,
with low probability of resonance

14. Result: difficult to reproduce average
cross sections and distance to collision
10 samples
1 million samples

15. TART: Monte Carlo Radiation Transport in Industrial Applications
Garbage In, Garbage Out!!!
Any code is only as good as the data it uses
for Neutrons & Photons
I am amazed at how much effort goes into
designing radiation transport codes
And how little thought seems to be given to
the data the codes use
Your results cannot be better than the data you use
Before deciding to use any code you should
know the pedigree of the data it is using

16. Garbage In, Garbage Out!!! (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Garbage In, Garbage Out!!! (continued)
In the case of TART, it always uses the most
recent ENDF/B data
Neutrons: ENDF/B-VI, Release 8
Photons: EPDL97 (now ENDF/B-VI)
Incorporating new data into TART is now
completely automated and simple
I can apply data the same day I receive it

17. What Monte Carlo codes do I use?
TART: Monte Carlo Radiation Transport in Industrial Applications
What Monte Carlo codes do I use?
Experience has shown that no code is perfect
So it is important to verify results
I use a combination of codes to do this
During the day I use TART
Because of its speed
At home, at night, I verify results using
MCNP
I strongly recommend you also verify results

18. Why should you use Monte Carlo?
TART: Monte Carlo Radiation Transport in Industrial Applications
Why should you use Monte Carlo?
Only Monte Carlo can accurately Model
Complex Geometry
Exact Reaction Kinematics
For Example of Kinematics,
Neutron Elastic Scattering
Photon Compton Scattering
Exact with Monte Carlo – approximate Sn
If this is important in your applications
you should be using Monte Carlo

19. Can you Afford to Use Monte Carlo?
TART: Monte Carlo Radiation Transport in Industrial Applications
Can you Afford to Use Monte Carlo?
Historically it has been too expensive or
time consuming to use Monte Carlo
Needed many hours/days
on expensive super computers
Those days are gone forever
Recent increases in computer power and
decrease in computer costs make Monte Carlo
very competitive for use in today’s applications

20. Can you Afford to Use Monte Carlo? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Can you Afford to Use Monte Carlo? (continued)
In 1995 the first computer independent
version of TART was released
In 1995 it took the fastest available PC only about
three times as long to run a set of test problems,
as it took on a CRAY YMP super computer
These results showed great potential for small
computers
But nobody could foresee what was about to happen

21. Can you Afford to Use Monte Carlo? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Can you Afford to Use Monte Carlo? (continued)
I gave a similar talk in 1999 – then available PCs
ran these same problems 50 times faster than could
be done in 1995
Today’s PCs run these problems 207 times faster
than in 1995 [28,487 seconds vs. 89 seconds]
What took at entire 9 to 5 day (8 hours) in 1995
today takes less than 2.3 minutes
THAT BOGGLES MY MIND!!!

22. Can you Afford to Use Monte Carlo? (continued)
TART: Monte Carlo Radiation Transport in Industrial Applications
Can you Afford to Use Monte Carlo? (continued)
It wasn’t too long ago that with Monte Carlo
we could afford to runs thousands of histories
& more recently this became millions (106)
Using TART a PC today can runs billions (109)
histories per day
Using multiprocessor computers, TART
can run trillions (1012) Histories per day
Statistical convergence is no longer a problem

23. Conclusions Today Monte Carlo is very practical
TART: Monte Carlo Radiation Transport in Industrial Applications
Conclusions
Today Monte Carlo is very practical
Today the major cost of applications is
your salary
NOT computer costs
I recommend that you use several Monte
Carlo codes to verify your results
Naturally I recommend that one of these
codes be TART – see my website for details