ML20117M526
| ML20117M526 | |
| Person / Time | |
|---|---|
| Site: | 05000000, Dresden |
| Issue date: | 05/19/1983 |
| From: | SARGENT & LUNDY, INC. |
| To: | |
| Shared Package | |
| ML20114G054 | List:
|
| References | |
| FOIA-84-616 EMD-043781, EMD-43781, NUDOCS 8505170118 | |
| Download: ML20117M526 (13) | |
Text
,.
I CARGENT 0 LUNDY
- ~
ENGINEERO cmcaso ATTACHMENT B 4
DRESDEN - 2 MAIN STEAM MONITORING TEST TEST-ANALYSIS CORRELATION t
i
?
hDR F@h8840914 8ARFIEL84-616 PDR May 19, 1983 i
EMD-043781 Page 1 of 13
~1
CAR 0 ENTO LUNDY l
ENCINEERO I
CMICASD l*
Page 2 of 13
- 1.0 Introduction 4
2.0 Piping Thermal Expansion Movements 2.' 1 Walkdown Measurements 2.1.1 RPV Nozzle Thermal Expansion Movements 2.1.2 Support Spring Resistance (Variability) Effect 2.1.3 Thermal Movements Measurement Accuracy
'2.2 LVDT Measurements i
3.0 SRV Transient Loads 3.1 Analytical Assumptions 3.1.1 Forcing Function 3.1.2 Structural Andlysis 3.2 Snubber Load Measurements
+
g
\\
O en
~-
e --
+.,n-,,---,~m,.,
n.-,-p.,.m,
- -v--.--e.v.r,----
-,,-.,-,,,,,-.--n...E-
,,, ~, - -,,,,
e
7-,
5 *Y CAR 3 ENTO LUNDY GNOINEERO CHICAGO Page 3 of 13 1.0 Introduction The results of the Dresden-2 main steam monitoring pro-cedure (SP 83-4-54, Rev. 0) revealed no measured response
-that.could be considered a source of'the failure of the five main steam snubbers.
However, some discrepa'ncies between measured and analytically predicted piping responses were observed for the SRV discharge loadings and for the. piping thermal expansion movements.
Analyses were performed to assess the effects of these discrepancies.
These analyses demonstrated that the piping stresses
' remained within code allowables for all the monitored events.
The "Dresden-2 Main Steam Monitoring Procedure Seven-Day Evaluation" (Sargent & Lundy Calc. No. EMD-043449, Rev. 0) documents the results of these evaluations.
The purpose of this writeup is to offer explanations for the discrepancies between the analytically predicted and measured piping responses.
9
._.-.___m.
Paga 4 of 13-r d
2.0 Thormal' Movements' I
- 2.1 Walkdown Readings
'The recorded. thermal movements for the constant and spring supports on Main Steam lines are within the design values with some minor deviations.
The maximum deviation was found to be within 1/2"'for the spring supports and 5/8" for snubbers.
These sdeviations likely result from the factor described below.
2.1.1 RPV Nozzle Thermal Movements In general, the spring supports movements are either smaller or bounded by the design values.
The measured upward vertical move-ments were less than the analytically predicted movements.
This smaller than predicted upward movement was most prevalent on the hanger nearest to the RPV nozzle.
This discrepancy j
between predicted and measured movements likely results from the use of conservative (larger than actual) RPV movements in the piping thermal expansion analysis.
A comparison was made between Dresden and LaSalle RPV thermal movements to identify the source of f
e 4
y.
. m
-w..
.,-w.,,,,-.__--,.,,,-%-.-,
.,.,-.+,,e, w
,_-.%,.--,-.,----,...-...,ww..-%.,-
ee.-
._,..-.--.---,--m--, e~--
- -,-=.--ew-t-wv
4 Pags 5'of 13 the discrepancies.
The Dresden RPV nozzle movements were calculated by assuming linear expansion.and a uniform temperature over the vessel lengthi Using GE's formulae (Spec. 22A3828, Rev. 1, MPL No. A42-3670), which were provided for the LaSalle Station RPV, the nozzle thermal move-ments were found to be very close to the originally.
calculated values.
However, GE stated that "the application of these formulae should result in conservatively high values for the vessel growth."
Therefore, with less conservative RPV nozzle movements, the actual support movements would be very close to the design values.
Note that during the plants previous operation, piping shakedown to elastic behavior and material creep could also result in smaller than predicted movements.
2.1.2 Support Spring Resistance (Variability) Effect' A variable (spring or constant) support would require ~ applied load in order for the spring to move up or down.
This applied load with respect to the dead weight is called variability, which would differ from one support to another and would depend on the support design and its condition.
A sensitivity analysis was performed
Pagn 6 of 13 e
for Main Steam C line, based on 10% variability, to assess the effect of the piping resistance to thermal movements.
As a result, the piping thermal movements at some locations changed by 3/16".
With a higher variability ratio (could reach 20%) the deviation in piping thermal movements are expected to be even greater than 3/16".
2.1.3 Thermal Movements Measurement Accuracy The thermal movements were visually recorded during cold as well as hot (460*F) conditions.
The smallest division on a typical snubber is 1/2" and on a constant ~ spring support is either 3/32" or 1/4" depending on the total travel distance of the support.
A typical support or-snubber reading would require a combination of judgement and approximation.
A 1/4" reading tolerance is considered reasonable for both cold and hot conditions.
This could result in a combined tolerance of 1/2" on the thermal movement.
This measurement tolerance is con-sistent with the hanger manufacture (Bergen-Paterson) recommendation.
Paga 7 of 13 Taking'into account'the above factors, thermal-
' deviations'within~1/2'" are viewed as acceptable, as Iong.as the spring supports and snubbers are positioned within their working ranges.
Therefore, the walkdown thermal readings are considered within the design values and no significant deviation has been identified.
2.2 -LVDT Measurements The piping vertical 1 movements as recorded by the LVDT's were consistent with the movements indicated by the hanger readings.
The LVDT readings also indicated that
-the piping. upward mov6 ment was somewhat less than, analytically predicted.
The probable reasons for these discrepancies were discussed in:Section 2.1.
LVDT's also were used to measure lateral movement at snubber locations.
As reported in the Seven-Day Evaluation l
there was a discrepancy between the LVDT measured movements and the walkdown measured movements.
However, in all but one case the' discrepancy was less.than 1/2 inch, which is within the measurement tolerance of the walkdown.
At snubber 51 the discrepancy is close.to one inch.
In the opinion of Wyle Labs, this discrepancy could be caused by a bad connection in the butt splicer'or a bad connection 6
' :. z.-- -.. '
~~
Pago 8 of 13 internal to the LVDT.
The change in containment ambient temperature, which reached E l$0'F, could then result
' in an errant LVDT reading.
The MS header lateral movements as measured by the LVDT's are greater than the predicted values by approximately one inch.
An analysis for these discrepant movements was performed and the pipe stresses were shown to remain within Code allowables (RE: Seven-Day Data Evaluation).
.The explanation of the-discrepant movements is currently being investigated.
J e
4
e e*
+
Pagn 9 of 13 y
e 3.0L SRV Transient Loads The. design basis analysis indicated that the SRV transient
' loads on the MS header piping'were not significant.
The test results confirm that these loads are not significant.
p
-For example, snubber number 52 had the largest mea'sured SRV' load.
For snubber 52 the addition (via SRSS) of the design basis SRV loads to the SSE loads increased the snubber design load by less than one percent.
Adding _a conservative interpretation of the measured SRV load (4300 lbs compression) to the SSE loads increases the snubber design load by less than eight percent (less than 650 lbs.).
The' measured loadings did, however, differ from the analytically predicted loads.
For six out of nine of the measured cases the measured loads were less than predicted, and for three cases the measured loads slightly exceeded their predicted values.
This section provides probable sources of the_ discrepancy between measured and' predicted
. loads.
3.1 Analytical Assumptions In_the design basis analysis various assumptions are made to account for unknown parameters and to arrive L '
Paga 10 of 13
}
at a' feasible analytical approach.
These assumptions '
are made to approximate.the actual system behavior.
Deviations between the actual and assumed behavior
' N, will res. ult in discrepancies between the predicte'd and actual response.
?
~
The analysis for SRV discharge loadings is performed in two, steps; first the loadings are generated'and
- g then-these loadings are input into a structural
. analysis.
The primary assumptions,made in both steps
- y x
are described below.
s 3.1.1 Forcing Function The primary assumptions'usedsin the generation of the forcing function involve the behavior of the SRV valve.
Assumptions are made con-a cerning the valve opening; e.g. the rate of area opening, the time of opening, and the flow l
rate.
Variance between these assumptions and E
actual valve behavior result in variance in the
'N
+
4' predicted and actual forcing functions.
In addition, small loadings-in the header piping
'are generated by the depressurization wave that
+-
.is created in the header when the SRV valve opens.
4 1
'1
-Paga 11 of 13
+
~
These depressurization loadings were not considered in the design basis analysis.
These loadings were calculated to be small..
3.1.2 Structural-Analysis The primary assumption made in the structural
' analysis involve the piping and support behavior.
.A-linear piping ~ structural analysis is performed with an assumed damping value.
The snubbers are assumed to be infinity rigid struts without gaps.
The-effect of spring and constant hanger supports is not considered.
The actual system behavior is somewhat different than these assumptions, and some~ discrepancy between pre-dicted and measured responses is therefore expected.
For example, snubbers do not behave as rigid struts.
A snubber, per design, limits the piping accelera. tion to a value less than 0.02g.
If the acceleration is less than this, the snubber will see no load.
If a particular snubber allows the piping to move, then some of the loading will be absorbed in the piping and some of the loading will be redistributed to other snubbers or supports.
Ii
'D ;
L,
~
'- 'M Page 12 of 13 1
Actual snubber behavior alone, especially at low loading values, will account for a good deal of variance between predicted and measured responses.
Loads will be distributed differently between snubbers and some loading will be absorbed in the ii P P ng.
3.2 Snubber Load Measurements The snubber calibrations and the setup of the test equipment was designed for the measurement of large loads, of a magnitude that could fail a snubber.
Specifically, there were no calibration data points for loads less than 5,000 lbs. and the oscillograph recorders were set on a scale of 10,000 lbs. per inch.
The measured responses were typically well below these values.
However, to get-a rough idea of the system response, attempts were made to read the low level signals.
Note that the variance in accuracy,as a percent of reading, increases as the signal level decreases.
i Tabulated below are the estimated accuracies of the measured values for the three snubbers that experienced the largest loadings.
The combined accuracy of the V
- signal conditioning and strain gauge is 3.78% of reading (per Wyle Lab. document " Field Dat'a Acquisition Accuracies").
s w
w
.-e..r-.
--s-m ww n.
- v,
.,.,,,-,.,,,-,,,,.,c.,,n..
,.v-,,.
.r.
-,. ~.,.,._
-w.e.n..-
..,,,,.,,,-.--..-an..---,.s.,,--r..~.
r:
6
. ". ~.
CARGENT O LUNDY Paga 13 of 13 ENGINEERO CHICAbD This does not include the accuracy of the calibration
, curves, which is not defined,but would increase this
~
percentage.
The table includes, in terms of the measured parameter, the value of 1/2 the width of the noise band.
The recorded measured loadings are felt to be conservative (especially in the 5ase of compression) estimates of the actual response.
Measured SRV
'lensio N ession Snubber
. Transient Icads(lbs) 3.78%
1/2 3.78%
1/2 Notes No.
. Tension Ocznpression Reading Noise Band Reading Noise Band 46
+ 1500
- 3700 57 244 i 140 615 According to subsequent 50
+ 1700 800 1 64 260 30 320 calibrations the compress 52
+ 2000
- 4300 1 76 452 i 163 1778 readings cou be conservat by an additional 14 *w 36%
s t
o
[
,