ML20245B340

Technical Assistance for Fort St Vrain Task Ii:Pcrv & Pcrv Tendon Evaluation, Final Technical Evaluation Rept

ML20245B340
Person / Time
Site:	Fort Saint Vrain
Issue date:	12/31/1986
From:	Anderson C, Fugelso L LOS ALAMOS NATIONAL LABORATORY
To:	Office of Nuclear Reactor Regulation
Shared Package
ML20235N115	List: ML20235N112 ML20245B340
References
CON-FIN-A-7290 NUDOCS 8706090425
Download: ML20245B340 (17)

Text

'

Technical Assistance for Fort St. Vrain Task II. PCRV and PCRV Tendon Evaluation by L. Erik Fugelso, MEE-13 Charles A. Anderson, MEE-13 Los Alamos National Laboratory, MS J576 Los Alamos National Laboratory, University of California Los Alamos, New Mexico 87545 ,.

Fin No. A-7290 Cognizant.NRC Personnel - Mr. Ken Heitner ORB-3 Phillips Building, MS 428 Nuclear Regulatory Regulations, Nuclear Regulatory Commission Hashington, DC 20555 Final Report December 1986 i

l l

l i

i

/

go c 0 74tw l

5&

V

/ ff.

l l -

NOTICE This report was prepared as an account of work sponsored by an agency of the' United States Government. Neither the United States Government nor any agency thereof, or any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third party's use, or the results of such use, of any infor-mation, apparatus, product or process disclosed in this report or

~

represents that its use by such third party would not infringe privately owned rights.

I. INTRODUCTION During a routine lift off test of the end caps of selected prestressing tendons in the PCRV of the FSV reactor, several individual steel wire strands were observed to be broken and others had been corroded to some extent.I This degraded tendon condition was observed during a scheduled periodic tendon surveillance procedure and program established before initial operation of the Fort St. Vrain reactor. After this first appearance of broken strands, an augmented surveillance program was proposed by Public Service Company of Colorado (PSC) and implemented.2,3 This report is a review of the current tendon surveillance program being implemented by PSC at the Fort St. Vrain reactor.

In addit.on to an assessment of the enhanced surveillance program, review of several finite element and other analyses to assess the current and future state of stress in the concrete in light of the occurrence of tendon degreda-tion was done. The adequacy of the analyses to predict the course of the stress-history and its dependence on the state of degradation of the tendons is examined. Additional finite element calculations were made to clarify some further points and are presented here. The predictions yield very conservative estimates for limiting advances of the state of degradation to maintain con-crete integrity.

In Section II our task objectives are spelled out.Section III summarizes the available PCRV surveillance data and its applicability to assessing the current and long term stress state in the PCRV.Section IV icviews previous analytical evaluation of PCRV integrity--two of which 1.iclude an examination of the effect of ruptured tendons.Section V presents some further analytical and numerical results on PCRV integrity after tendon rupture. Finally,Section VI summarizes our conclusions.

II. PROGRAM OBJECTIVES The original objectives of this work as defined in the FIN plan were (1) the review of all available PCRV surveillance data, (2) the evaluation of the individual surveillance data relative to the performance of the PCRV and the prestressing tendons--in particular, to relationship with the degraded tendon I

configurations, and (3) the overall evaluation of the effect of tendon degra-datie on PCRV integrity during the futu.'e operation of the Fort St. Vrain reactor. These objectives were later modified by NRC to concentrate on PCRV and tendon evaluation and analysis.

III. SURVEILLANCE DATA Specific data that have been measured since the onset of PCRV prestressing are load cell data at a selected subset of tendons, strain gage data at se-1ected positions within the concrete mass at selected rebar locations and at liner and at thermocouple penetrations, periodic deflection measurements at variable internal pressurization, periodic surface crack surveys, and helium permeation and concrete moisture measurements.

The main usable data of the above list are the limited number of long term load cell histories and the current lift off force and visual inspection data.

These data are discussed further later in this section. Even though specified in the Fort St. Vrain reactor FSAR,4 the strain gage data and deflection test data made available to us were inadquate to be quantitatively useful in assessment of longer term PCRV deformation behavior. No concrete moisture data were available to us. The limited helium permeation data indicate that neither major structure-wide concrete failure nor liner failure have occurred.

The crack survey data show a number of surface cracks of limited extent that exhibit very slow growth and that should be expected on the exterior of this structure. The cracks may have mainly developed at vessel prestressing.

In early 1984, a routine tendon examination showed that several of the prestressing tendons in the Fort St. Vrain PCRV had broken or were ineffective because of extensive corrosion.I While the fraction of tendons exhibiting any broken strands was small, and the fraction of broken strands in any tendon was also small (typically one to five of 169, although there were two cases with 12 of 169 and one with 21 of 169) there is legitimate concern as to the effect of the degradation of the tendons on the integrity of the PCRV and to the stability of the tendons with respect to further rupture. Because of this, a more extensive tendon surveillance program with visual observation for cor-roded and/or failed tendon strands and more frequent lift off force measure-ments has been recently implemented by the Public Service Company of Colorado.

Table I illustrates the number of tendons and their type covered by the origi-nal and enhanced surveillance programs. In addition, criteria for initiating a more intensive engineering study of potentially failing tendon and more pre-cise criteria for replacement of a failing tendon were also established..

TABLE I NUMBER OF TENDONS INSTRUMENTED FOR LIFT OFF TEST Top Bottom Circumferential Crosshead Crosshead yfrtical Total Tendons 310 24 24 90 Original surveillance program tendons 16 2 2 6 Control group tendons 3 1 1 3 Additional surveillance tendons 13 1 3 12 The revised tendon surveillance program and current data are reported in Ref. 2. The complete program and selected earlier data are given in the sur-veillance plan.2,3 The data detailed in these reports are the visual in-spections, the load cell measurements and selected lift off tests.

The visual data reported include observations of rust and corrosion and the number of ineffective strands in each tendon. In these reports, no reference was made to the other data (moisture, strain gage, etc.) originally requested '

to be reviewed. Prior to 1984, a small number of tendons had been visually inspected and had lift off tests performed on them at fairly infrequent in-tervals, while almost continuous load cell monitoring on a few other selected tendons was done. Figure 1 illustrates raw data from one of each type of tendon load cells. .

1 IV. PREVIOUS ANALYTICAL EVALUATIONS OF PCRV INTEGRITY l 1

Previous structural calculations of the PCRV response to internal pres-sure, thermal stress and prestressing loads with complete and degraded tendon )

l I

i

4 4

l p)vaancAL p)omcuwtMrWn4L taan  ! teso yb y--"-

~

Ndb s >

m -L ."" .

me sm TV 4 i I

. o - o I

soo, ison see seoo . !

w rRowinmpys) N F80"'"* A) kjenoucaosweAo I

snoo J

. - i

( .-,.

i g  %. .

tien ese see toes TM FROW 1MM (DAYS)

Fig. 1. Typical load cell data for tendon hold-down force versus time.

L.--_____- - - -

.: , .. .. .:... . . . . . - . . .o . . . . . . . ... .

D configurations have determined the state of stress in the concrete, liner and tendons and in this way, have examined the integrity of the PCRV. The FSAR (Ref. 4) reports a three-dimensional finite element solution for a 300 sector of the PCRV. This solution gives elastic stresses in response to prestress, to combined prestress and temperature, and to combined prestress, temperature and pressure loads. Approximation for liners and reinforcement are in-cluded. These solutions indicate the regions of maximum and minimum stress and the regions of stress concentrations. .These calculations do not treat concrete creep and tendon relaxation, nor do they examine the change in stress state caused by the rupture of a tendon.

In the FSAR is also presented a two-dimensional axisymmetric finite ele-ment solution (ignoring the buttress) to estimate the time-dependent creep response of the PCRV over a period of approximately 35 years. Measured creep properties of concrete and prestressing steel appropriate to this structure are utilized. Response to the reference pressure and thermal loads as well as thermal response only are calculated. The prestressing of circumferential, vertical and radial tendons is included. Calculated output is the circum-ferential concrete stress at the inner wall at mid-height as well as circum-ferential and vertical tendon forces. The analysis showed that concrete creep and tendon relaxation were predicted to be well within acceptable values over the life of the PCRV.

As mentioned previously the main data on the tendon capability and current capacity is the load cell data. Examples of the raw load cell force-versus-time curves are shown in Fig.1. Load cell measurements have been made on one set of tendons since the startup of the reactor (over approximately 4000 days) and on an additional set since mid-1984. The slow decrease in the hold-down force with time for both sets of tendons is consistent with the analytic solu-5 tion presented by Fugelso and Anderson that is shown in Fig. 2 wherein they analyzed the hold-down force for a tendon in a concrete whose creep law (at constant stress) is linear with the logarithm of time. The calculated tendon hold-down force exhibits a logarithmic decay for the first 4000 days, which closely matches the data for sets of monitored tendons. The analytic solution is based upon a fit to the concrete creep data over a period of 1000 days.

Based on the agreement of the data and the solution, the extrapolation of the analytic solution to 4000 days seems excellent. Extending the calculation of y

(

the hold-down force to 10,000 or 20.000 days (27 to 55 years) using the same creep law indicates only a very small decrease in the hold-down force over the remainder of the reactor's life. Figure 2 shows the calculated hold-down force extrapolated to 20,000 days. Comparing this limiting value with logarithmic extrapolation of the measured average hold-down force over the same period shows that the difference of these values is less than the standard deviation ,

of the measured estimate. Both are significantly higher than the minimum value of 1160 kips required at the end of service.

The viscoelastic properties of the steel (at the temperature at the tendon) are such that the stress relaxation that will occur due to steel and wire prop-erties would have occurred within the first year after prestressing. No further significant stress relaxation in the steel wires is anticipated.

y 5 B I 3 WE ASUMED AVO ltXTRAPOLATtDI Sapo -

g WE ABURED

• 14

\

y

• q YW17 g g N vatM CALC (*Stc6,7  % N v=r1 #

2

- , ._. '"m

.." , , f' N *s[s[x v=>

i<= <

• • *"" "'"

• Mt ASURED WEATVRt0-16  %

=

,tfee -

9 9 e i 1 to too 1000 10,000 TIME + 1 == DAYS Fig. 2. Theoretical hold-down force versus measured hold-down force.

1 5

Fugelso and Anderson ,6 also calculated the redistribution of concrete i stresses in the barrel part of the PCRV wall (including the buttress) when a fraction of one tendon was suddenly removed. Measured creep properties )

were taken into account and a three-dimensional finite-element calculation j over a 100-year period were carried out. Global and local changes in the nor- ]

mal and transverse stress components were made. Additional calculations on a I more realistic three-dimensional model were carried out mainly to investigate the effect of a change in tendon configurations on stress components in the regions of stress concentrations. For the concentrations in axial stress near the inside cavity wall at the upper corner of the cavity, removal of one com-plete vertical tendon changed that stress by about 30 psi. Changes in the axial stress component due to alteration of the circumferential tendon were smaller by a factor of at least three and at most five, similar to the changes noted in the one dimensional model. Thus, removal of a single tendon changes the concrete stress by only a small amount, and maintains safe levels of stress in the PCRV.

The analysis of Ref. 7 calculates the effect of the removal of either circumferential or a combination of crosshead and circumferential. tendons on the ultimate load capacity of the PCRV. For the first case, only circumferen-tial tendons are considered. The failure modes of a horizontal. slice through the core are either tensile failure of the liner or tensile failure of the prestressed tendon in response to the internal pressure, pg . The former mechanism is dominant. The failure loads were calculated by an ultimate load analysis. This model indicates that when the internal pressure is at the reference pressure, pr - 845 psi 58% of the circumferential tendon strands can be removed without PCRV failure. When the internal pressure is 1.5 times the reference pressure, 25% of the tendons can be removed. The failure mechanism in the top and bottom heads was failure in the concrete in flexure and was calculated by yield line theory. In this region of the PCRV, the effects of removal of circumferential and crosshead tendons are coupled.

Simple inequalities for the percentages of required intact tendons are given by the equations:

Nc + 0.45 N, 2 64 for pgpg i (1)

Ng + 0.45 N, 2 136 for pg = 1.5 pR i 1

l

i where N g and N, are the percentages of intact circumferential and cross-head tendons, respectively.

These numbers are very similar to those that are estimated by calculating the percentage of circumferential tendons required such that the tangential stress in the concrete in a horizontal slice through the core is just at tensile yield. A simple expression for this percentage f is given in Eq. (2) below.

/ AP bt i i ~ 't) f - 100 N OA 's k0 /

where A = a +b 2 2 b2-h2 and B = 2b2 / b2-a2 and where a, b are the inner and outer radii, e, is the prestress, 2 is the PCRV height, A0 is the cross-sectional area of each tendon, N is the number of effective tendons and et is the tensile strength of the concrete.

This formula is derived with the approximation that all the prestressing tendons are replaced by a thin exterior steel shell with the same total cross-section area as all the effective tendons (not all circumferential tendons are present in a given sector) and that the exterior shell has the same prestress.

Using the value of 550 psi for the tensile strength of the concrete 294 in.

and 186 in, for the outer and inner radii, 168000 psi for the prestress, 1116 in. for 1, and 207 for N, we find that 58% of the circumferential tendons are required to prevent the concrete from failure at the interior wall.

These tendon percentages required are estimated as average values over a volume containing many tendons. As shown in Ref. 4, the crucial number for structural failure is the average fraction of intact tendons. The tendon sur-veillance program in Ref. 2 and 3 suggests that any individual tendon be re-placed when 20% (15% in critical regions) of its strands have been degraded, i This choice of number is certainly conservative when applied to the whole set of tendons and, since it is applied to each one individually, will be even safer.

V. ADDITIONAL ANALYSES OF PCRV INTEGRITY UNDER TENDON RUPTURE During the current contract period additional numerical evaluations of stresses in the PCRV were undertaken and accomplished to further understand the re'iations between the concrete stress and the status of the prestressing tendons.

Previous studies of the redistribution of the triaxial stress state at various locations in the PCRV structure due to degradation of the tendon struc-ture were calculated after several types of tendon degradation in a fairly simplified version of the structure (5,6) Tendon degradation considered was by removal of a fraction of an individual tendon or by complete removal of an entire tendon from a system with many tendons. Degradation of the tendons was considered in an environment with only one set of tendons, e.g., vertical OR axial, and with intersecting tendons. Concrete creep over a 30-year period was included, but steel relaxation, which normally is complete within a year or so, was ignored, and the computations started with the end of the steel relaxation. The additional calculations were performed with a goal of examining the effect of the tendon degradation on the stress concentrations in the structure.

4 From the 3-dimensional elastic solution in the FSAR discussed before, the largest stresses in the system occur at the inner radius at midplane at the thinnest wall section. Without prestressing, the maximum stress will be the hoop stress whose value under the condition of no prestress will be tensile and in excess of the tensile strength of the concrete. Hith the prestress, this stress is compressive under all postulated load conditions.

The time dependent solution examining the creep response reported in the FSAR was restricted to an axisymmetric wedge (4) . The previous LANL study considered the three-dimensional nature of the PCRV midplane region (5,6) ,

The finite element code, NONSAP-C, was used to evaluate the long time creep response of wall section with the buttress and with circumferential and axial prestressing. In addition to the normal operating stress configurations throughout the structure's life, selected or circumferential tendor.s were allowed to fail and the corresponding stress changes near the midplane were examined in detail. The change in the critical concrete stress, the hoop stress at the inner radius at mid-height, changed to more tensile values with

-g-

a decreasing fraction of intact circumferential tendons (initially pre-stressed). The increase in this hoop stress was of the order of 40 psi per circumferential tendon. Degradation of a vertical tendon do not affect the critical hoop stress significantly.

The finite element code, ABAQUS, was used for a further evaluation of the intersection of a corner generated stress concentration and the degradation of the prestressing system. Figure 3 shows a finite element mesh for the whole PCRV structure and Figs. 4 and 5 show the details of typical layers of the 0

finite element mesh for a 30 sector of the lower half of the structure.

Figure 4 showing the level through the core, and Fig. 5 showing a typical section through the head. Each element is a 20-node, linear strain brick with the wedge-shaped elements being degenerate bricks. Coarse and fine versions of each type of mesh were prepared, the latter having approximately eight times s

N

.- j g gs

~ ~

..' ~

N r

4 ,'

\

d v .. ,

N Y) *z -

, ~

\- -

ry. , ,

~

~

- y s '-

N ' '

4

, s .

N. r -

g, 1

\

Fig. 3. Finite element mesh for the complete Fort St.,Vrain PCRV.

N > )>

z[h y# ,

7*% >

' l

)

L z[.y%w-> < <y>;,

L L ' y> > .

• h[

Fig. 4. Finite element mesh for the 300 sector of the lower half of the Fort St. Vrain PCRV.

as many nodes and elements as the former. The geometric structure was selected so that the nodes and element edges coincided with tendon locations. In ABAQUS, the tendons were modeled by truss elements (no bending stiffness of the tendons was allowed) which coincided with the edges of three-dimensional solid ele-ments. The assembled stiffness matrix was the sum of the superposed stiffness matrices for the bricks and the trusses. The material models for either type element could be chosen from the extensive models available in the code.

One of the several possibilities for calculation to examine the inter-action of the concrete stresses and the tendon degradation is detailed here.

The stress concentration near the inner radius-intersection-with-the inner cap  !

(opposite the thinnest wall section) is calculated in response to the internal

. reference pressure and vertical tendons. Two calculations were performed, one with the full tendon area and prestressing and one with one tendon removed.

These two problems represent only a portion of the loading in the struc-ture, and are chosen such that slight differences in the stress state due to

- the change in a tendon structure (which is small in itself) will be illustrated 1

pp N

l i g 2<

lf ;bkib

[i,f'&>>i5b 3j$

• \[

Fig. 5. Finite element mesh for a slic.e of the 300 sector of a half of the Fort St. Vrain PCRV through tie head.

with as few interference effects es possibla. He consider only a subset of 0

the vertical tendons, namely (the four tendor,t in the inner row of the 30 sector, and the internal pressure. Omission of Ba circumfarantial tendons in this calculation will result in unrealistic stre.ss distribution relative to the actual structure and its operation. Comparison of the differences between the stress states with the intact tendons and the degraded tendons, in partic-ular in regions 01 li ?ss concentration, is the goal of this particular cal-culation. Analysis of the reincipal stresses in the vertical cross-section containing the buttress for the intact prestressed tendon were compared with the same stress for the PCRV with the tendon in the inner row opposite the buttress decreased in number of strands by 25%. The contour plots are almost identical, with position shifts of the order of the plotting line width at most. At any interior point the vertical. stress changes by about 5-6% uni-formly throughout the section, the remaining stress components are unaltered.

The Table presents the percentage change of several selected stress components at various locations throughout the sector. Absolute' stress magnitudes are not reported as the calculations were made for a reduced and idealized set of tendons. For our purposes here, only the difference in the stress components has any significance. Further, the numbers presented in Table II are given to only one significant figure to reflect the approximate nature of the model-ing of those included tendons. This result is similar to those reported by fugelso and Anderson (5,6) , where changes in stress mid-height due to removal of a complete tendon were observed to be slight. In particular, the values of the stress concentrations in the concrete near sharp interior corners or edges changed less than 10 percent with changes in the intact tendon population.

TABLE II Differences in Stress Components at Various Locations Due to Removal of One Vertical Tendon Location Stress Component Difference (%)

Inner wall at mid-height Vertical +6 Inner wall at mid-height Radial 0.

Inner wall at mid-height Hoop + 0.1 Outer wall at mid-height Vertical 5. ',

Outer wall at mid-height Radia) 0.0 Outer wall at mid-height Hoop 0.1 Near head and wall Vertical 6.

Intersection Near head and wall Radial 0.0 Intersection near head and wall Hoop 0.1 Intersection Top center of head Vertical 0.0 Top center of head Radial -1.0 l c . _ _ _ _ _ _ - _ _ - _ _ _ _ _ _ _ _ _ - _ _ _

L A

, VI. CONCLUSIONS

1. A program of. tendon and tendon load cell surveillance to monitor the prestressing system.in the Fort St. Vrain reactor was established in the orig-l inal safety report and was further enhanced by PSC when'the original monitoring l program found degraded prestressing tendons, including corroded and broken tendon strands. Our models and the load cell data show that stress relaxation in the monitored tendons is decreasing with a log time relationship. Normal concrete creep and tendon relaxation will be within the original specifica-tions.

2. Analyses of the concrete stresses have shown that under the current i

state of tendon degradation (involving rupture of tendon strands), and includ-l ing'the effects of concrete creep and tendon relaxation, all stresses in the l PCRV will remain within safe bounds throughout the lifetime of the reactor.

3. Based on the analysis of Ref. 7, very conservative limits for the number of broken tendons that can be structurally tolerated for both the reference pressure and 150 percent of the reference pressure have been established. From these, an even more conservative tendon replacement policy has been invoked by the Public Service Company of Colorado. He concur with this policy.

4. Stress concentrations in the concrete near sharp interior corners or edges change less than 10 percent under the condition of a rupture of a single tendon.

VII. REFERENCES

1. Letter, Harembourg, D. W., (Manager Nuclear Production), Fort St. Vrain Nuclear Generating Station to John Collins (Regional Administrator), U.S.

Nuclear Regulatory Commission, Region IV, dated April 12, 1984.

2. " Tendon Surveillance," report No. 85084, and references therein, Public Service Company of Colorado (1985).

3. "PCRV Tendon Interim Surveillance and Status Report," Public Service Company of Colorado, report No. 86042 (1986).

4. Fort St. Vrain Nuclear Generating Station, Finite Safety Analysis ReDort, Vols. 3 and 6 (1980).

_14 1

-- .-. -.-.....-.. ,....L...~ *..

,:..Y* :._...u ..e... . .. .. .

5. Fugelso, E. and Anderson, C.A., " Evaluation of Concrete Creep and Stress-Redistribution in the Fort St. Vrain PCRV following Rupture of the Pre-stressing Tendons," Los Alamos National Laboratory technical evaluation report to NRC (1985).

6. Fugelso. E. and Anderson, C.A., " Evaluation of Concrete Creep and Redis-tribution in a Prototypic PCRV following Rupture of the Prestressing Tendons " Paper D-8/6, Proceedings of the 8th International Conference on Structural Mechanics in Reactor Technoloov, Brussels (1985).

7. " Fort St. Vrain - Tendon Requirements based on Safety Consideration,"

Public Service Ccmpany of Colorado, report No. 85053 (1985).

l l

l

. _ _ _ -