ML20207F044
| ML20207F044 | |
| Person / Time | |
|---|---|
| Issue date: | 06/02/1999 |
| From: | Marsh L NRC (Affiliation Not Assigned) |
| To: | Gupta V HOLTEC INTERNATIONAL |
| References | |
| NUDOCS 9906080016 | |
| Download: ML20207F044 (5) | |
Text
t-1 June 2, 1999 Mr. Vikren Gupta, Qu;lity Assurcnce M nig:r Holtel intemational 555 Lincoln Drive West Marlton, New Jersey 08053
SUBJECT:
REQUEST FOR A TECHNICAL REVIEW OF A DRAFT INFORMATION NOTICE REGARDING ERRONEOUS RESULTS FROM AN ANSYS COMPUTER CODE Mr. Gupta:
The U.S. Nuclear Regulatory Commission (NRC) is plarining to issue an information notice discussing a problem reported by Holtec Intemational regarding erroneous results from an Ansys computer code. We ask that ycu review the enclosed draft of that information notice to ensure the technical information is accurate. Your cooperation in this matter is appreciated.
Please return any comments you may have as soon as possible. A copy of this' request and your response will be placed in the Public Document Room for review by the public. Your response should be mailed to:
U.S. Nuclear Regulatory Commission ATTN: Thomas Greene, NRR/REXB MAIL STOP: 0-12D1 Washington, DC 20555-0001-Please address any questions you may have on this matter to Tom Greene of my staff.
Mr. Greene may be reached by phone 301-415-1175. If no comments are received by close of
' business on June 10,1999, we will assume the technical information in the notice is correct.
yb. Ntarsh, Chief.
Events Assessment, Generic Communications and Non-Power Reactors Branch Division of Regulatory Improvement Programs Office of Nuclear Reactor Regulation
Enclosure:
Draft Information Notice ip p
REXB R/F MM PDR fy-VV DOCUMENT NAME: G:\\REXB\\ tag \\letholte.wpd To receive a co >y of this document, indicate in the box C= Copy w/o attachment / enclosure E= Copy with attachment / enclosure N = No copy OFFICE PECB: DRIP C:PECB:DftlP 1
NAME TGreene M LMarsh h h OFFICIAL RECORD COPY
,9 f
~
9906090016 790602 PDR ORG NRRA 1
(.' O " " Q
- y. O j -) s PDR g76 c
V\\,~
UNITED STATES
)
NUCLEAR REGULATORY COMMISSION OFFICE OF NUCLEAR REACTOR REGULATION WASHINGTON, D.C. 20555 June XX,1999 NRC INFORMATION NOTICE 99-XX:
ERRONEOUS RESULTS FROM A COMMERCIALLY AVAILABLE COMPUTER CODE Addressees j
t u.
All holders of operating licenses or construction permits for nuclear power reactors,.,..
Purpose pl
[gy g
7m The U.S. Nuc! ear Regulatory Commission (NRC) is issuing this in p
n sensitive nonlinear phenomena, could yield erroneous results. It is
$ N used in modeling addressees of the use of a commercially available computer code t that recipients will
^
review the information for applicability to their facilities and consi og as appropriate, to avoid similar problems. However, suggestions contained in this in rmation'
'ce ats not NRC requirements; therefore, no specific action or written respo ti~ required.
?g, gy Qpscriotion of Circumstances
.i
{,
g l,
9 4 r s.
4v, HOLTEC International (HI), in eccordance with 10 CFR Part 21,'submped a notification to the NRC in November 1995 idennfying computatioru:! problemswith the computer code, ANSYS.
ANSYS is a commercially available three-dimensional finite element computer program for
-engineering analysis and design (e.g., linear and nonlinear static and dynarnic analysis for civil structures and mechanical components under different loading conditions such as dead weight, pressure, thermal, displacement, wind, impact, and e9thquake loads). The notification involved the following two distinct computational problems that arose during the process of performing a nonlines(transient anal' using ANSYS 5.0A and ANSYS 5.1: (1) disparate results for calculated disiqlacements are ined depending on the type and version of personal computer employagi and (2) dispa '
results for forces and displacements are obtained depending upopn. 'tochniques emp
- to launch the analysis runs. Examples of the computer platformpend the
_ing techn ~ ques used in the analysis are shown in.
{w..,,y With regard to the first problem, the M3ili$ ear analysis has been tried on five different computers usN ANSYS 5.0A undef a disk operating system (DOS) and resulted in two distinct sets of answers on the five compplers. In addition, the same analysis has been run using ANSYS 5.1 pone of the fiveyputers and this gave a third set of results.
Qx-m 6
)
I e
IN 99-XX June XX,1999 i
Page 2 of 3 With regard to the second problem, the nonlinear analysis has been run on one of the five computers using ANSYS 5.1 with different techniques to launch the analysis runs, and different answers were obtained depending on the launching techniques used.
HI reviewed the procedures and results, <,f the analysis, end concluded that the magnitude of the discrepancies in both cases is considered significe and should not occur. Hi advised ANSYS, Inc., the developer of the ANSYS code, of these disparities.
)
7 The NRC staff reviewed the displacement results of the analysis, and found that thi
/
magnitudes of the results are small; however, one result is 2.5 times as much artdhe smaller f
result. Although the analysis involves small magnitudes, in a nonlinepi analysis this much difference in result from the same computer code is a concern. It rpy well be posifq that
. some users with an inadequate computer platform may have obtairped results that mayipry by as much as 250 percent or more without realizing the significance the enf bse of the y*
erroneous results.
p:.
Discussion ay g y, <
Part 21 to Title 10 of the Code of Federal Regulations (10 CFR Part 21) requires'that suppliers of basic components or cervices associated with basic components, evaluate' deviations.
Deviations must be evaluated for potential safety significance to determine whether the deviation could create a substantial safety hazard, and thus be classified as a defect.
i f: y In this case, HI notified the NRC in accordance with 10 CFR Part 21 that the deviation was considered significant, but its evaluation concluded that the deviation does not affect any safety-related equipment or services previousl/ supplied,to licenseesby Hl.
The ANSYS codo has been used for engine $ering analyses by many architect-enginee f
/
consultants, and licens.ees of nuclearpower plants forlmany years. Consequently, there is a potential for the use of erroneous results in safety-related applications. The ANSYS User's Manualdoes provid& cautions regardlng the use of the nonlinear option of the computer code, and an experienced an'dqualified analysishould be able to determine the sensitivity of the end results to initial conditions %qd the choice ol' cific values of input parameters. However, the NRC staff has no way to determine the actual fety significance of the problems because there is insufficient informatiori reGarding which nuclear licensees actually performed a nonlinear analysis using the ANSYS code, the types of computer platforms used, or to what extent licensees' actual use of the cbde would nave contributed to inaccurate results that could potentially cause an unsafe condition. Therefore, it is expected that recipients win review the informatien for applicability to their facilities and corrective actions, as appropriate to address potential problems.
e w"
sO
s-i.
IN 99-XX June XX,1999 Page 3 of 3 Related Generic Communications Information Notice 96-29
" Requirements in 10 CFR Part 21 for Reporting and Evaluating Software Errors," May 20,1996 (Accession nt:mber 9605150209).
This information notice requires no specific action or written response. If you have any questions about the information in this notice, please contact the technical contact listed below or the appropriate Office of Nuclear Reactor Regulation project manager.
p
/
j e,
5 p.
y e.-
,f
- ?p,
.(.
3 Ledyard B. Marsh, Ch,.
jy D.
Events Assessment, Generic Communications T~
Division of Regulatory % Branch and Non-Power Reac lpiprpvement Programs I-Office of Nuclear Reactor R ulation v
Attachment:
Examples of The Computer Platforms and LaunchiiipTechniques L.i '
[5$
}h p
- -^
A[
~
,,s
- w..
l j'
(,;
,:t b'
f A
l f
(
c I 7,3 s.
1
.~ +
E ' ',
vM J
~
a j,..
f r
-5 k-';_
u
',.s f,.:
J l
c o i
\\
l.
m Att:chmint IN 99-XX 1
June XX,1999 Page 1 of 1 i
EXAMPLES OF THE COMPUTER PLATFORMS AND LAUNCHING TECHNIQUES i
A.
Computer Platfor. git: IBM-compatible PCs j
1, Pentium-1 90 MHZ,64MB ram,2GB SCSI Hard Drive, no disk compression, e
DOS 6.22 or Windows NT 3.5 (workstation) 2.
Pentium 7' g
~ 90 MHZ,32MB ram,1GB IDE Hard Drive, no disk co, ~
- sion, DOS 6.22 4P'
/
[
3.
Pentium-3 90 MHZ,16MB ram,540 MB IDE Hard Drive, Stacker 4.0 disk compression, \\;
DOS 6.22
~. v.
g e.,.,
4.
Pentium-100 4
100 MHZ,64MB ram c
[:~[
/(
5.
486 Machine qq
/'
a B.
Launchina Techniaues:
s, 7
1.
Run RAxxUSE.INP individua via the b,
launcher indle input filANS.INhc 2.
Run RAMUSE.INP via using the batch launcher 7;
n % w.
~
d s
a g
g}'9 0
c r
- N+3 jf?
m, p',}'e -
K;*
ty
,D; s.,