ML20217C544
| ML20217C544 | |
| Person / Time | |
|---|---|
| Site: | 05200003 |
| Issue date: | 09/28/1994 |
| From: | Levin A NRC (Affiliation Not Assigned) |
| To: | Mcpherson G NRC (Affiliation Not Assigned) |
| Shared Package | |
| ML20217C479 | List: |
| References | |
| NUDOCS 9803270054 | |
| Download: ML20217C544 (5) | |
Text
.
na Q j
v
,o E
UNITED STATES li NUCLEAR REGULATORY COMMISSION
/
( *i WASHINGTON, D.C. 20555-0001 g
September 28, 1994 MEMORANDUM T0:
G. D. McPherson, Senior Thermal Hydraulics and Testing Expert Division of Systems Safety and Analysis THRU:
Robert C. Jones, Chi f
Reactor Systems Branch j
Division of Systems Safety a d Analysis q
1 FROM:
Alan E. Levin, Senior Reactor Engineer Reactor Systems Branch LU Division of Systems Safety and Analysis >
t
SUBJECT:
OBSERVATION REPORT FOR AP600 CORE MAKEUP TANK TEST 507 On September 16, 1994, I observed the performance of Test 507 of the Core Makeup Tank (CMT) Test Program. This test program supports design certification of the AP600. The test was conducted at Westinghouse's Waltz Mill site, near Madison, Pennsylvania. Although this test is not the last one in the test matrix, Westinghouse changed the order of testing for this part of the program, so that in fact this was the final test in the original test matrix to be performed. However, Westinghouse is planning to rerun some of the earlier tests that were found not to have met acceptance criteria by Westinghouse Test Engineering or Nuclear Safety.
The basic test setup, general testing procedures, and testing personnel were j
covered in my earlier test observation report, and therefore will not be j
repeated here.
I will briefly describe the test procedure for this experiment, review the test performance and results, and report other observations.
The 500 series of tests are nominally the most complicated of the CMT program.
For each test, the CMT is allowed to recirculate until a specified fraction of the inventory is heated to near saturation conditions at the test pressure, and the tank is then drained and depressurized simultaneously.
The way in which the test description is written appears to imply that this procedure is continuous and is similar to the CMT behavior observed in the integral systems tests. This is not exactly the case.
Rather, the CMT is allowed to recirculate, but when the specified fraction of the inventory has been heated, the test is stopped briefly, and then restarted from the test pressure, the tank is drained, and depressurization is initiated when the tank has drained to a specified level.
Depressurization is carried out at a specified rate, which can change during the test. The reason for stopping the test in the middle is the small size of the facility's steam-water reservoir (SWR), which to a degree simulates the reactor vessel. The configuration of the simulated pressure balance line in the facility is not similar to that in the plant, because there is no " cold leg" pipe. Instead, a " dip tube" arrangement is used, where a pipe extends down through the top of the SWR. When the tube is covered with water, the CMT will recirculate; when it is not covered by water, steam is drawn into the CMT and the CMT will drain.
The volume of the SWR is 9803270054 980116 PDR ADOCK 05200003
/;.
A PDR
[
. such ~ that, in the'" draining" mode, it cannot accomodate the. inventory with the SWR water level remaining below the dip tube, unless the SWR is first drained well below the level of the dip tube. Thus, the test starts out with a high SWR level. to cover the dip tube, recirculation is initiated, and the test is then stopped, the SWR is drained to accomodate the CMT inventory, and the test is started again with the dip tube uncovered, which imediately initiates CMT draining.
In addition to the starting, stopping, and restarting of.the t' est, it must be noted that keeping the CMT at the nominal test pressure and beginning-depressurization only when the draining phase was in progress is not the way the CMT operates in the integral tests. There, the facility depressurizes continuously from the inception of a transient due to effects of inventory loss out of the break (if any), cooling of the RCS by the PRHR system, and injection of the cold CMT inventory.
The consequences of the procedure described above are:
1.
There is no possibility of seeing draining occur due to flashing in the pressure balance line, as has been postulated in some integral tests, since the facility remains at high pressure during recirculation.
2.
During the time it took to re-initialize the test after the recirculation phase, the temperature of the heated portion of the CMT dropped significantly, due to a substantial heat loss from the CMT.
3.
Since the draining phase is also started at the nominal test pressure, it is possible that the system does not depressurize fully during the period that the CMT is draining.
For test 507 itself, the test pressure was 1850 psia, the specified fraction
)
of _the inventory heated during recirculation was 20% (2 ft out of the 10-ft-i long tank), and the target depressurization rate was 1.5 psi per second, dropping to about 0.5 psi per second late in the test.
(Since j
depressurization is done by remote manual operation of a control valve, the depressurization rate is difficult to establish with any precision.)
The target temperature in the upper part of the CMT was within 100'F of saturation (T
- 626*F). The target drain rate in the second part of the test was 16
- gpm.
. As far as the performance of the test itself is concerned, it.was almost flawless.
I arrived at the_ test site at about 9:30 a.m.
Heat-up.of the facility was well under way, and was completed by about 10:30 a.m.
The test comenced at approximately 10:54 a.m. with a 100-second _ steady-state data acquisition period,. and recirculation began at about 10:57. The natural
- circulation period ended at 11:03. -Testing recommenced at 11:33 with another 100-second steady-state data set, and draining was initiated at 11:35.. The
- . actual: drain. rate at the. start of this. phase was about 17.25 gpm, slightly over the-target rate, but the flow dropped as the CMT emptied, and the average
' drain rate was-about 16 gpm. ' As-noted above, the combination of high ' test.
. pressure and delayed depressurization meant that the CMT could not be depressurized completely before it drained. The lowest pressure reached in the test was about 700 psia. The test ended by about 11:50 a.m.
After the test was completed, I was able to review some of the data from this test, as well as from previous CMT tests.
Specific observations deriving from this test and review of other data are:
1.
These tests should provide good data to Westinghouse to assist in the modeling of the CMTs in the AP600 and in the integral test facilities.
However, the facility coes have clear limitations:
it is not able to represent the full ranae of CMT operating conditions observed in the integral facilities or--by extension--in the plant.
In particular, it cannot represent transitions between different operating modes. This facility does represent reasonably well the recirculation and heat-up behavior of the CMT.
It also models draining fairly well, and in a manner that can be characterized. Both of these modes can be considered
" quasi-steady," in that they change very slowly.
However, by virtue of the test facility configuration, it cannot replicate the " integral" i
behavior of the CMT, in terms of imposing a realistic depressurization profile or following the transitions from recirculation to draining, induced either by flashing in the pressure balance line or by inspiration of steam from the cold leg to the CMT when the RCS inventory has decreased to the point where the loops are uncovered.
2.
Review of the data showcd behavior consistent with that in the integral tests during the quasi-steady recirculation phase and after draining had become established. Of particular interest is the fact that, according to the test engineer, it was not necessary to supply steam to the SWR from the external source during depressurization; flashing of both the saturated liquid in the CMT and the saturated liquid in the SWR was sufficient to supply _ steam to fill the open volume in the CMT.
3.
The slow depressurization and rapid draining would appear to indicate that relatively little data is being obtained in the low-pressure range of CMT operation, i.e., below accumulator initiation pressure.
Integral tests show that the fourth stage of ADS initiates at pressures between about 100 and 200 psia, triggered by the CMT reaching the 20% inventory mark.
Depending on the transient being simulated in the integral tests, as much as 50% of the CMT inventory drains at low system pressures (less than about 500 psia).
The data from the CMT test program needs to be reviewed in detail to see how much information is available at these pressures, since the low-pressure behavior of the CMT appears to be quite important, and the CMT's role in opening ADS-4 is a key aspect of AP600 safety performance.
4.
I made a point of revieaing the data from the heated RTD level sensors installed-in the test facility. What I saw left me very concerned about i
the performance of these instruments. The raw data includes temperatures from both the reference (unheated) and heated junctions of these sensors (rather than just a differential temperature).
I observed very inconsistent data from these sensors. When both junctions are l
}
. covered by liquid, the temperature difference is normally about 4-6*F.
However, during the recirculation phase, while both junctions of the uppermost sensor were still immersed'in liquid, the temperature difference fluctuated substantially, at times reaching about 20*F. This behavior appeared to be related, at least in part, to the movement of the thermal front past the RTDs during recirculation. Once draining began,'when the RTDs became uncovered, the temperature difference rose fairly slowly to around.I4*F, with the difference increasing with lower -
pressure. The test engineer explained that when these sensors were tested in air, the discrimination between " covered" and " uncovered" was very good, since air's thermai conductivity is much less than water's.
However, steam at high pressure has a thermal conductivity not very much less than water's.
Thus, the sensors do not show a strong signal at uncovery at-relatively high pressures.
The testing crew indicated that evaluation of these data by Westinghouse's I&C design personnel-had just begun, and 'that algorithms were being looked at to try to develop a highly reliable level sensing system. This area should be one that receives extensive attention from the NRC; Westinghouse must be able to show not only that the ADS will actusts wnen it is supposed to, but also that the system will Dg1 actuate, due to a spurious level signal, when it'should not. The implications of this issue with regard to ITAAC for the CMT injection and level sensing systems also need to be considered carefully.
One other point with respect to the level sensors is that I was told that the heating element in one sensor (the lowest one in the test tank) had burned out.
It must be recognized that these sensors have been cycled far more in this test than they would be in the plant, and further, that there are four such sensors at each level in each CMT, with ADS actuation based on 2-out-of-4 logic. Nevertheless, power to these sensors must always be "on" during plant operation, and their robustness and reliability will need to be evaluated carefully, especially since they will be immersed in borated water, which could lead to corrosion or other deleterious effects. The ease of performing maintenance on these. sensors during pir.nt operation, e.g., replacing the heating element, should also be examined closely.
As a closing observation, I was interested to learn that the testing personnel
.at the CMT facility know almost nothing about the results from Westinghouse's
. integral' tests, especially as they relate to CMT performance and integral systems effects.
This may not keep the CMT test from providing valuable information, nor from being a well-run program. Nevertheless, it would appear useful to let the test engineer, at least, see how his program gets integrated into the:" big picture," perhaps by visiting a test in OSU/ APEX, for instance, and'I was surprised and somewhat disappointed to see that Westinghouse has not made this. type'of effort.-
. In summary, the CMT test program appears to have been successful, well-run, and capable of providing very useful information. However, it does have shortcomings in being able to represent a systems-level simulation of the CMT in the AP600, especially during transitions between the two different modes of CMT operation.
Therefore, the data from this program, along with relevant data from the integral test facilities, need to be reviewed carefully by the NRC to ensure that sufficient data have been collected over a broad enough-range of parameters to adequately characterize the performance of the system in all of its aspects.
.