ML20148H046
| ML20148H046 | |
| Person / Time | |
|---|---|
| Issue date: | 11/08/1987 |
| From: | Catton I NRC |
| To: | Boehnert P NRC |
| References | |
| ACRS-CT-1905, NUDOCS 8801270107 | |
| Download: ML20148H046 (5) | |
Text
s 27~ Mo5
@s ipHW
. DO :
Paul'Boehnert pNov1987
~
FROM:
Ivan Catton
- r3 te R U.RC SU BJECT: Semi-Annual 2D3D Meeting Mannheim, Germany lb,,. i Y t ~
' ac' 18-21 May 1987 p*
The 1987 Semi-Annual 2D3D meeting was held in Mannheim. The meeting was attended by representatives of the three participating countries (US, FRG, and Japan. The Upper Plenum Test Facility (UPTF) is completed and ready for serious use. Several interesting tests have been run as part of the facility. readiness testing. In particular, the ECC bypass tests have shown conclusively that our present codes are extremely conservative in their treatment of ECC Bypass during the refill stage of a LBLOCA. I think we can look forward to a very productive period for the 2D3D program.
The availability of the UPTF data is presenting some challenges to codes like TRAC. A new period of tuning is upon us. The following paragraphs contain some highlights of the four day meeting.
FRG PROGRAM STATUS Final acceptance testing was completed by early December 1986. The tests show that all systems work. Some entrainment, de-entrainment tests have been completed (CS-1 and CS-2). Quick look reports have been issued.
Bypass test 3 shows that cold leg ECC water gets into the core very quickly. The delivery is intermittent.because of loop to loop interactions. The cold water appears to just fall into the lower plenum. Lots of oscillatory behavior was observed. Hot leg flow intermittently breaks through the upper core support plate near one hot leg then the other. One has to wonder if the core simulator is producing some of the observed phenomena. The codes will have to predict the observed behavior or their benefits may not be helpful.
The data clearly show the resolution that will be needed to predict the downcomer behavior, this translates into code noding requirements. Clearly on the order of 8.or more azimuthal nodes will be needed. The type of oscillations seen in the downcomer are not unlike those observed by the English (presented at the International Heat Transfer Conference, Aug 1986, San Francisco).
It seems to me that TRAC calculations should be done straight away.
The entrainment de-entrainment tests are not considered official tests. This was stated several tines. I do not understand the reason. The process is of sufficient importance that the results should be used anyway.
Condensation in the cold leg causes flow reversal in the hot leg which results in drawing the hot leg injection fluid into the hot leg. This stops upper core support plate penetration at the hot leg 8801270107 871108 DESIGnTID ORIGIIAL 19 PDR Certified By
-. ~.
where the flow reversal has occurred. This process oscillates between ho legs. It would be interesting to correlate plena behavior at a given cold leg with hot leg behavior. Condensation appears to be very important.
U.S.
PROGRAM STATUS The US supplies support to.the experimental part of the 2D3D program as follows:
Two-Phase Flow Instrumentation ORNL/INEL TRAC Analysis LANL Data Analysis and Test Conditions MPR For the most part the instrur.entation supplied by the US is working well. There'have been some difficulties with signal to noise ratio.
A few of the turbine flow meters had to be replaced (9 out of 112).
All but one are now working well.
TRAC analysis' carried out was described by Boyack. Some code improvement has been done as a result of the work. A large number of documents were noted but none have been published. LANL has done an awful lot of calculations, probably at great expense.
If their work is to be meaningful, it has to be published. LANL is promising calculations with TRAC Mod 2 while trying to get NRC to fund its development. Either the NRC plan is unclear or LANL is obtaining funds from other sources.
The complex hot leg - upper plenum physics observed in UPTF have not been addressed by LANL. This leaves the US part of the program lacking.
JAPANESE PROGRAM The Japanese believe that they have satisfied there part of the 2D3D agreement with SCTF and CCfF. They have carried out experiments based on conservative conditions. It is now necessary to understand the phenomena. CCTF test results do not show oscillations like those observed during the UPTF tests. One has to ask whether there was something that precluded oscillations or were oscillations in UPTF non-typical.
The Japanese noted distinct dif ferences in PWR/SCTF/CCTF and UPTF.
They implied that the PWR, SCTF, and CCTF are scaled properly whereas UPTF is not. This may or may not be true. Their observations do, however serve warni.ng that UPTF results must be used with caution when taken as system results.
DOWNCOMFR BYPASS TESTING The results of two tests were described. The first was with combined cold leg and hot leg injection. For the cold leg injection 2
case, the water literally fell into the lower plenum. It fell as a plume from the cold leg iarthest from the simulated break. The cold legs nearest the cold leg with the' break fed it intermittently.
High steam flows lead to high void fraction as expected. The usual finite differencing used in the codes will not handle this very well.
I think treatment of a falling plume will be beyond their modeling capability. This means one will have to be careful to see that the process is treated conservatively if advantage is to be gained from the observations. It is clear, however, that the present code calculations, TRAC or RELAPS, treat downtomer bypass very conservatively.
The hot leg injection (HLI) led to some strange behavior. Slugs of water were swept into the simulated steam generator making it a much more important component during a LBLOCA than it will be for US PWRs. The slugs were intermittent and not simultaneously in both hot legs which led loop to loop oscillations that may have impacted on the downconer bypass phenomena as well.
Murao,JAERI, indicated that the Japanese would like to carry out some tests that are counterpart to the UPTF bypass tetts. This would enable one to confirm scaling ideas and methods as well as more fully test the codes. The discussions following Murao's proposition dealt with several aspects of the UPTF simulation concept.
The UPTF core simulator is made up of 17 steam / water injection zones (a 17 tube steau caliope). The steam and water are controlled separately by a computer feedback system. The desired flow rates come from SCTF III tests and TRAC analysis. Corrections are based on measured core level and data from SCTF III. This is a rather sophisticated process for something whose physics are not all that well understood. A complicating factor is the water plugs in the hot leg getting blown back into the vessel. It is not clear how the steam generator simulators and the core simulator will interact while being controlled by a computer. All four loops will have to be modeled by the codes if this complex process is to be simulated. Each steam generator will act like a phase change bubble yielding all sorts of oscillatory behavior. If you recall, an experiment at Semiscal.e had similar characteristics and has stubbornly resisted explanations via code calculations. I would have been more comfortable with a simpler more controlled experiment even if it did not simulate everything as well.
CODES The German PWR (GPWR) seems to be the vehicle most are using in their 2D3D efforts. It is interesting to note that two nodalizations have been legislated for TRAC in its use on the 2D3D program. The fine nodalization is 8 azimuthal nodes. The data from the downcomer Lypass experiments seem to suggest that there should be more. Of course one should not loose sight of the SNL work by Buxton that shows more nodes is worse not better. Here we have a classic example of tuning. Something malignant in the code has been overcome by 3
adjusting parameters to yield LOFT like behavior and it is now forgotten. Calculations of GPWR behavior during the first 120 seconds took 68 hours7.87037e-4 days <br />0.0189 hours <br />1.124339e-4 weeks <br />2.5874e-5 months <br /> of CPU time on a computer that is the equivalent of a CREY-1.
The US (Lou Shotkin) presented the NRC guidelines for future TRAC-PWR improvements. Release of TRAC-PF1/ Mod 2 will be delayed until.all documentation is completed. Here I can only infer that documentation refers to the long awaited QA document. It was estimated that the documentation would be completed by December 1987-(we now know that this is a very optimistic estimate of the time of completion). The Mod 2 version will be made available when it is used on the program.
A re-evaluation of the need for a Mod 3 is underway. NRC will continue to provide the project with the best state-of-the-art codes possible. Code deficiencies uncovered by 2D3D will be incorporated as soon as possible into the code. Each proposed improvement will be tested against improved reactor safety, cost effectiveness, likelihood of a successful improvement and timeliness of an improved code.
A number of problems still causing computational difficulties were described. Some of them are:
- 1) sensitivity to time step,
- 2) plugging and flow oscillations in the hot leg that are real or otherwise, 3)penetratien of upper plenum liquid into the core is underpredicted, and
- 4) sensitivity to noding detail is such that significant differences in results can occur.
Mod 2 of TRAC-PF1 will be a slightly improved Modl. The improvements do not address some of the major deficiencies noted as part of the CSAU effort.
Use of TRAC to study the behavior of stratified flow in the hot leg yielded some interesting results. It was concluded that TRAC could calculate CCrL in the hot leg conservatively. This is strange because TRAC's two fluid model does not have a stratified flow model in it. It is conceivable that suitable adjustment of the interfacial shear will allow one to get the correct answer (tuning;.
One has to ask if the interfacial shear models are problem dependent.
The calculated course of the refill transient was found to be strongly dependent on the noding. Both too course and too fine led to trouble. The base case noding of the downcomer lead to little water getting into the lower plenum. Reducing the number of nodes yielded much better results. This brings out the need for all such i
studies to be done with the same noding so that proper tuning of the codes will result. The problem is lack of physics and too much tuning to the wrong scale. Further flow regimes based on node thermal hydraulics will always lead to problems as the overall geometry determines the flow regime. These troubles should be no surprise as the codes are lumped parameter based and as such 1
tuning is expected. Tuning, however must be physics based and component specific if it is to be reliable for scaling to full 4
4
)
e-scale LWRs.
-1 GENERAL UPTF has been underway for five years. Mayinger summarized the program by noting that the three countries had spent as follows:
FRG DM 380M JAPAN Y 15977M USA 90M It is not clear how to put the above amounts of money on a common footing. Before the dollar crashed the amounts were about equal.
Mayinger noted that the plans for the UPTF facility have been changed dramatically to shift away from the large break LOCA. The plan is to study the small break and accident management. The Germans plan to keep the facility active well beyond the 2D3D program obligations.
Nine B&W type vent valves have been installed in UPTF. Here is another opportunity for coordination of NRC research plans. I would, however wager that those responsible for the MIST program have no plans for integration of UPTF into their program. They should be asked about this as one of the big questions about the MIST program was the vent valves and.their effects on the transients of interest. There were some (myself for example) who felt that without the vent valve question being addressed, the MIST results were only of marginal
- value, t
e I
a 5
4 4
5 1
- -