ML19296B098
| ML19296B098 | |
| Person / Time | |
|---|---|
| Issue date: | 10/24/1979 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Bates A Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1181, NUDOCS 8002200058 | |
| Download: ML19296B098 (3) | |
Text
,,
A3'A$CF.T i m.cic:: sm a.
g.gTo ACM N@! 2,
24 October 1979 7.
Dr. Andrew Bates g gg,g,.
.m s.-
ROM:
Ivan Catton g y_
r COPY TO: Dr. M. S. Plesset SUBJ:
Advanced Code Review Group Meeting 6 September 1979 Silver Springs, Maryland The TRAC code is still evolving. The latest addition is a droplett field. The droplett field will accomodate an arbitrary initial drop size distribution. The various heat transfer relationships used in the code are still being modified and many are "make-do" empirical expressions. The flow regimes allowed and how they are used is somewhat elementary. LASL argues that the "make-do" approach allcws them to move forward and that changes can be made later.
It is my concern that some of the correlations or physical models needing modification will be forgotten and will beccme integral carts of a verified and accepted code. At present, criteria for changes in TRAC are not well defined and many times are made without proper justification. This somewnat chaotic approach was noted by a review group member to be a potential problem. The need for a systematic approach is apparent.
It was suggested that a table be made up showing models and regimes of application to better demonstrate where experiments or models are needed. At present there is no effort by the code developers devoted specifically to spelling out what areas need more experimental backup or analysis. An effort of this type at INEL clearly pointed out the lack of heat transfer data in some key parameter ranges.
Several areas that TRAC is weak were spelled out. They are as follows:
a) flow regimes - present flow regimes are reasonable but not necessarily compatible with multi-dimensional analysis nor do they accomodate transients.
b) bubble sizing - need to improve criteria, c) droplett entrainment - present model does not lead to proper precursor cooling during reflood, d) minimum contact temperature - (minimum film boiling temoerature) presently varies in a non-chysical way.
Its relationship to precursor cooling not well defined.
It was noted by Dr. Lahey that TRAC is now at a stage where tne physics look the same as the SCORE code develooed several years ago at INEL. The difference accarently is just the numerics. The SCORE code had numerical problems and efforts to develoo it were terminated.
It is not clear that TRAC is without some difficulties of its own.
8002200 OQ
The TRAC code does not always converge on a steady s.Ite result.
This infomation was surprising. Both SANDIA and BNL personnel using TRAC have had difficulties.
It is not clear what the difficulty is. The LASL code developers claim that they have no problems.
If one cannot obtain steady state solutions to steady state phenomena, then one cannot have a great deal of confidence in predictions made by the code.
It is hoped that serious attention will be given to tne TRAC code numerics.
Some time was spent discussing'small break models. Several aspects seem to be relevant and deserve some discussion here. Dr. Shimway's study of semi-scale demonstrated the importance of the steam generator, particularly during the early phases of the small break. This is in contrast to many discussions that we have had with the staff. The arguments about homogenization and increased flow at the breaks and its relationship to whether or not one should turn off the pumps was brought out.
It's not clear that we understand break flow well enough to make any statements about the effect of homogenization on the break flow. Present use suggests knowledge about the break flow I'm not sure that we have. The turning off of the pump forces reliance on the reflux mode of natural circulation. We know very little about the reflux mode and in particular in U-tube steam generators, a great deal of work needs to be done, or at least some simple scoping type experiments. RSR is having Professor Peter Griffith run experiments which should answer some of the questions about the reflux mode.
In TRAC, friction is always applied to the liquid. At present the data that is used comes mostly from pipe flow studies.
It's not clear where one obtains data for other components.
It's also not clear where or what the experimental program is that backs this up.
It seems to me that there needs to be some further separate effects studies to close the gaps. For example, shear associated with cross flow in rod bundles is a necessary element in studying UHI.
At present, shear and cross are not being handled in more than an approximate way.
Hence, the near tem upper head injection results must be viewed with caution.
There is some data cited in Wallace's book that a rod bundle can act as a steam separa tor. There is further evidence of this in measurements of phase separation in Y's and T's.
Some results given by Lahey in NUREG CR-0557 show that vapor follows ne pressure drop and liquid coasts by. The fraction being diverted is proportional to the square root of the density ratio. One should be able to make an estimate of flow out of a break using Lahey's work. Accounting for the steam secarator effect in a small break analysis may lend some insight as to whether or not pumps should be turned off homogenized ficw out the break may be an error.
Most of uw data that's available on pressure drop or friction factor is for circumstances where a solid wall is nearty.
It's not clear how one obtains data for the shear acting on a node that is away from boundaries.
When one uses flow regime maps to detemine what friction factor or pressure drop relationship should be used one can get into a great deal of difficulty. An example of tnis is a filling tank.
If you have a node in the middle of the filling tank, the void fraction in that node will run from zero to one as the tank fills up.
Depending on the flow rate and mass flux rate, this can lead you through every possible flow regime and lead to chaotic results. TRAC does not have a scheme for avoiding this problem.
2
~.
A study was made to assess how well TRAC could analyze the TMI incident. At oresent the TRAC code cannot handle the pressurizer with a leak.
It uses a drif t flux model to handle the surge line and does not have the proper flow regimes. There is no diffuser plate in the pressurizer. TRAC is some distance from being aole to handle the surge line and the pressurizer properly.
Very little of the advance technology in TRAC can be brought to bear on the small break problem. Further, it is not clear that pump energy is properly put into the flow.
If small break calculations are to be made that follow the accident scenario for any length of time, the reflux mode either in once through or in IJ-tube steam generators must be an integral part. At present the TRAC code cannot handle such a situation.
D e
3