ML20084A627

Review of Diesel Generator Bldg at Midland Plant

ML20084A627
Person / Time
Site:	Midland
Issue date:	10/31/1983
From:	Costantino C, Chris Miller BROOKHAVEN NATIONAL LABORATORY
To:	NRC
Shared Package
ML20084A617	List: ML20084A627 ML20084A639
References
NUDOCS 8404250139
Download: ML20084A627 (25)

Text

. .. _ _ _ _ _ . _. . . . _ . ._. _ _ . . ._ . _ _ _

APPENDIX 111 Review of Diesel Generator Building at Midland Plant by C. A. Miller and C.J. Costantino Structural Analysis Division Department of Nuclear Energy Brookhaven National Laboratory Upton, NY 11973 October, 1983

(

8404250139 840412 PDR ADOCK 05000329 '! -

A PDR )

Table of Contents 1.0 Introduction .................................................. 1 2.0 Evaluation of Petinent Work ................................... 2 2.1 History of Structure ..................................... 2 2.2 Settlenent History ....................................... 3 2.3 Crack Patterns ........................................... 6 2.4 Structural Analysis ...................................... 7 2.4.1 Bechtel's Computation of Settlenent Stress (Ref. 2) .......................................... 8 2.4.2 Bechtel's Analysis Using Measured Settlements (Ref. 3) .......................................... 9 2.4.3 Matra's Analysis Using Measured Settlenents (Ref. 4) ......................................... 10 2.4.4 Estimation of Stresses from Crack Data (Ref. 5) ......................................... 10 2.5 Stress Totals ........................................... 11 2.6 Survey Data ............................................. 12.

3.0 Assessnent of the Diesel Generator Building .................. 12 4.0 Response to Concerns of R.B. Landsman ........................ 13 5.0 Conclusions .................................................. 16 References ........................................................ 18 Appendix ..........................................................A ,

1.0 INTHODUCT ION This report describes a study undertaken by Brookhaven National Laboratory (BNL) to evaluate the extent to which settlement cracks observed in the Diesel Generator Building (DGB) at the Midland Nuclear Power Plant impact on the ability of the building to satisfy design requirenents. Dr. R.B Landsman, of Region III, has raised questions regarding this safety issue (Re f. 1). The specific objective of this study is to assess the significance of his canments and to prepare a written response.

This objective was achieved by reviewing the existing pertinent work (published reports, testinohy and analytical studies), and by interviewing key personnel so that a correct inteq)retation of the work performed could be made. Additional calculations were specifically omitted from the scope of this study. All of the conclusions drawn in this report are based on an assessment of calculations and studies performed by others.

The study described herein was carried out during the period of August through September 1983. On August 4, a meeting was held at NRC to discuss the problem and to obtain some of the pertinent literature. Some of this litera-ture was carried back to BNL while other documents were mailed to NRC during the following week. Appendix A contains a listing of all reports used during the program. On August 24, a meeting was held at Bechtel Corporation of fices in Ann Arbor, Michigan. Presentations were made by Bechtel and Consumers Power staff summarizing the work performed by project personnel to demonstrate the adequacy of the DGB. Their consultant's (Dr. M. Sozen of the University of Illinois and Dr. G. Corley of Construction Technology Laboratories) also discussed their work. An inspection of the DG8 was held on the evening of August 24 and during the morning of August 25. At this inspection, the cracks were observed although no new detailed crack maps were made. Discussions were held with construction personnel to determine the sequence of concrete place-ment.

Further interviews were held at NRC on September 8. Individual inter-views were held with Dr. Harry Singh (soils consultant for NRC from 'the Anny Corps of Engineers), Joseph Kane (NRC staff), and Lynan Heller (NRC staff).

4 A conbined interview was also conducted with Frank Rinaldi (NRC staff), John Matra (structural consultant for NRC frum Naval Special Weapons Center), and Dr. Gunnar Haarstead (structural consultant for NRC). The purpose of these interviews was to explore the role each played in the design and analysis of the DGB and to learn of their concerns regarding the adequacy of the DGB.

An audit of the DGB calculations by the task group was held at Bechtel's Ann Arbor offices on September 12 and 13. Dr. Sozen was present on September

13. The following items were reviewed in detail during this audit: nu me ri-cal models used by Bechtel to calculate stresses in the OGB due to settle-ment; the magnitude of stresses due to the various load cases; the method of determining stresses from c' rack data; the accuracy of the survey nethods used to monitor settlments; and the concrete pour data. A meeting was held with Dr. Landsman of Region III on September 13, at which time his specific con-cerns raised in Ref. I were discussed.

This report is organized as follows. An evaluation of the literature is presented in Section 2 of the report. Section 3 contains BNL's assessment of -

the adequacy of the DGB, while specific responses to Dr. Landsman's concerns are given in Section 4. Conclusions are listed in Section 5.

2.0 EVALUATION OF PERTINENT WORK The naterial on the DGB which was reviewed during the course of this study is divided into six categories; namely, historical description of the structure and its settelnent behavior; developed crack patterns; structural analyses to evaluate settelment stresses; treatment of other loads and stresses; and survey data. The material in each category is described and evaluated in this section of the report.

2.1 History of Structure The OG8 is a reinforced concrete shear wall building consisting of five cross walls connecting a north and south wall. The interior walls are 18" thick while the exterior walls are 30" thick. The structure is 155' by 70' in e

m

plan and is 51' high with an intennediate floor slab located 35' above the f oun dat i on. Wall footings are located under each of the walls, the footings being 10' wide and 30" deep. The building is foundeo on about 30' of various fills overlying the natural glacial till.

The fill was placed from 1975 through 1977 with construction of the DGB begun in October 1977. Concrete was placed in 6 lif ts as follows:

October 1977 - to Elev. 630.5 (foundation)

December 1977 - to Eley. 635.0 Ma rch 1978 - to Elev. 654.0 Au gust 1978 - to Elev. 662.0 December 1978 - to Eley. 664.0 Februa ry 1979 -

to Elev. 678.3 Within each lif t the pours were 9enerally made from east to west. Construc-tion joints occur in the middle of the cross walls and at the west end of each bay for the north and south walls.

Large settlements and cracks in the concrete were noticed while the lif t going to Elev. 662 was being poured. Construction was halted while the pro-blem was being studied. It was concluded that the large settlement was due to poor compaction of the fill material. This settlement caused the structure to

" hang up" on the duct banks which penetrate the footings on the cross walls.

The duct banks were cut loose from the DGB foundation in November 1978 and construction of the building restarted. In January 1979, 20' of sand sur-charge was placed on the site to consolidate the fill. This remained in place until August 1979. In September 1980, a permanent dewatering system was in-stalled to maintain the water table below Eley. 610.

2.2 Settlement History The DGB is founded on approximately 30' of fill material, underlain by a very stiff glacial till about 190 feet thick. A dense sand layer about 140' thick lies below the till, which is in turn underlain by bedrock. The tildj0rlty of the !IlI was placed dt t he site between 19/b and ]9//, wj th att ual foundation cunstruction cumpleted by Jariuary 19/8. Duri tig July 19/8, set t le-ments of the order of 3.5 inches (Ref. 7) were noted which were greater than the original 40 year predicted settlements. Apparently consolidation of the fill was taking place as structural dead loads were applied. In addition, the four electrical duct banks under the structural crosswalls were acting as hard points to the foundation since they were in turn being supported by the stiff natural soils below the fill. This caused rotation of the building about the duct bank s.

Construction was halted during August 1978, a soil boring program under-taken to determine the problem with the fill and Drs. R.B. Peck and A.J.

Hendron retained to advise on the remedial action. The exploratory program consisted of 32 borings (with no undisturbed sampling) and 14 Dutch cone penetroneters . These confirmed that the fill had been improperly placed (in an extremely variable density state) and consisted of varying amounts of co-hesive as well as granular backfill. Lean concrete was also encountered in the backfill. The thickness of silty clay backfill was found to be greater under the south-east side of the building leading to the gener. ally larger settlements on this side .

A surcharge program was implemented to attempt to consolidate the fill more unifonnly. In addition, the duct . bank s were cut loose from the founda-tion in November 1978 to eliminate the foundation hard points. Surcharging began in January 1979 and remained in place until August 1979, when it was determined that primary consolidation had been completed. Instrumentation (primarily settlement plates and Borros anchors) placed in the fill'was used to arrive at this conclusion. It should be-noted that the consolidation test results, obtained from undisturbed samples taken after completion of the sur-charge program, did not confirm this conclusion. Data was sufficiently scattered to indicate that the fill may not be uniformly consolidated. Unfor-tunately, the boring ' program conducted after the surcharge program was com-pleted, did not include cone penetrameter soundings for canparison with the readings taken before the surcharge was- applied.

9 At the conpletion of the surcharge program, it was decided that since loose sands still existed in the till, a pernunent dewatering system would be installed to preclude the potential for soil liquefaction during a seismic e ven t . This dewatering caused additional settlenents to be developed at the site, but apparently these were related to deep seated consolidation of the natural soils under the fill, and would be more uniform than the settlements caused by the fill consolidation, it is questionable whether the piezometer data was of any significance in analyzing the excess pore pressure condition developed in the fill during the consolidation process. The readings indicate generally very low pore pres-sures, about 1/20 the magnitude of the applied surcharge pressures. It is not clear in fact whether the fill was ever fully saturated at the time of the surcharge program.

Peak settlenunts anticipated at the end of 2025 (actual settlenents to date plus secondary settlements from now till then) are specified in Ref. 7 to vary from 4.79 inches (under the NW corner) to 9.33 inches (under the SE corner). However, it should be mentioned- that the exact settlement history at the various settlenent markers at the DGB is open to question. For example, it is mentioned in Ref. 7 that the maximum settlements in August 1978 were about 3.5 inches. Yet the data used in the stress analyses for the presurcharge period (Figures ES-14 of Ref. 7) indicates peak settlements of only 1.99 inches. It was stated at one of the Bechtel presentations that prior to cutting the duct banks loose from the footing, footings along the North wall actually lifted off from the soil, with the DGB rotating about the duct bank s. There is no indication of this behavior in any of the settlement data used in the computations. Ref. 8 lists the settlenent increnent from 8/79 to 12/2025 to be 2.36 inches under the SE corner of the building. For the same period Ref. 7 lists this data as 1.89 inches. Thus some inconsistencies appear to exist in the various documents.

__--_m m _______. __.--_

2.3 Crack Patterns Af ter it was detennined that settlement was a problem, Bechtel initiated a program to nonitor cracks in the structure. In general cracks were visually observed and an optical ccinparator used to detennine crack width. Crack widths greater than 10 mils wre of specific interest as this coaresponds to reinforcing stresses of about 10 ksi. Crack maps were prepared based on surveys conducted during December 1978, September 1979, February 1980 and July 1981. Dr. Corely observed the cracking in January 1982 (Ref. 6) and confirTned that the general pattern of cracks agreed with the July 1981 Bechtel crack maps. He prepared a detailed crack map for the center interior wall. A comparison of this center w'all map (Fig. 4.21 of Ref. 6) with that prepared by Bechtel in July 1981 (Fig. 4.17) indicates that more cracking had occurred although the widths of the cracks appear to be about the same.

Cracks were observed during the BNL inspection of the plant on August 25, 1983 and some photographs taken. In general the pattern of cracks appears to be similar to the previously mapped cracks. However cracks, which had not been shown on ar1y of the Bechtel cracks maps, were noted in both the north and south walls. These additional cracks are in the lower level (up to Elev. 664) and run at 45 degree angles to the horizontal up to the cross walls.

The first crack maps prepared from the December 1978 survey indicate vertical cracks in the cross walls etch begin near the bottom of the wall and run up to Elev. 664 (this was the top of the concrete pour at the time the settlement problem was first noticed). The pattern of cracking is more severe in the east side of the building. This crack pattern is compatible with the model that assumes the cracks result from flexural stresses caused by the building " hanging up on the duct banks". No crack maps were prepared for the north or south walls.

Th second set of crack maps were prepared from the September 1979 survey.

In general, many of the cracks which occurred in the east wall prior to placing the surcharge do not appear on these maps. The east center and center walls show the same type of crack patterns as shown on the first crack maps except for the appearance of additional cracks. These maps also show cracks in the upper level of the building These cracks occur near the south side of the building in the cross walls. The cracks tend to be vertical with some inclination of the cracks near the south wall. Some cracks are indicated in these maps for the south wall. Primary cracking occurs in the east side of the wall and are concentrated in the upper portion of the wall. The north wall is shown to be more severely cracked than the south wall and contains mostly vertical cracks in the upper part of the wall. The cracks appear to be centered about the three interior walls.

The third set of crack maps were prepared frcm the July 1981 survey.

These maps indicate the same type of cracking as before although the cross wall now contain more cracking near the north side of the building than was evident before. The west wall contains many more cracks than were shown previ ously. These cracks run from the Elev. 664 level down to the base of the s tructure.

It appears that maqy of the cracks which have occurred nay be attributed to the building resting on the duct banks. Other cracks have occurred, how-ever, which were most likely caused by dif ferential settlement of the wall footi ngs. Comparison of successive crack observations generally indicates that more cracks are occurring, but that the maximum size of the cracks is still about 20 mils.

2.4 Structural Analyses The various analyses which have been used to evaluate stresses in the DGB are discussed in this section. The first analysis described is the method used by Bechtel to estimate stresses due to settlement for use in its load conbination study. This analysis makes use of the straight line approxima-tions to the profiles of the settlenents of the north and south walls. The second 'and third analyses described are the Bechtel and Matra studies, which attempt to use the actual neasured settlenents to estimate settlenent stresses. These analyses, though different in detail, lead to the similar conclusion that the settlenent neasurenents were (and continue to be) in significant error. The fourth analysis describes a cruder model which attempts to approxinate an upper bound to settlement stresses by looking at

~ __ ._ _

.. e the crack neasurements. The first three analyses are based on detailed finite elenent models, while the fourth is based on crack patterns and crack widths.

2.4.1 Bechtel's Computation of Settlement Stresses (Ref. 2)

Since the building settlenents occurred when the structure was in various stages of construction, the settlement stresses were evaluated for four dif-ferent tine periods. The first period spans from the beginning of construc-tion through August 1978 at which time construction was halted. The second tire period extends from August 1978 to January 1979 during which the duct banks were cut loose from the structure and construction resumed. The third time period extends from January 1979 to August 1979 during which time the surcharge was placed. The last time period extends to the year 2025 and includes measured settlements from August 1979 to December 1981 as well as the predicted settlements over the forty year life of the structure.

The actual measured settlements were used to calculate stresses for the fi rst period. Stresses were calculated in each of the walls by determining the arc of a circle which fit aqy three adjacent measured displacements. The radius of the arc was then used to find the resulting bending monent in the wall, and the moment used to calculate stress. The maximum stress in each of the walls was assumed to exist over the entire wall. The stress in the south wall was 11.3 ksi; the east wall 6.6 ksi; and all other walls 2 ksi.

The increments in stress which occurred during each of the other three time periods were evaluated using a finite element model of the DGB. This model was constructed and run on the Bechtel version of SAP (BSAP). The building was defined with 853 nodal points. Plate elenents were used to model the walls, and beam elements used for the footings. Eighty-four (84) boundary elenents were used to model the vertical soil stif fness (equivalent to the coef ficient of subgrade reaction). An iterative process was then used to determine the stif fness of these boundary elements. A best fit straight line was first fit through the measured settlements for the north wall and another straight line fit to the data for the south wall. It was shown that the measured displacenents departure from the best fit straight lines is within the tolerance of the survey data. Dead load reactions were next estinated at

each of the 84 boundary elements. The stif fness of any soil element was then determined as the ratio of the dead load reaction to the displacenent of the best fit straight line. The BSAP program was run and the reaction found at each of these bounda ry elenents. A new stif fness was then calculated as the ratio of the reaction to the displacement of the best fit straight line. This process was continued for several iterations.

It is our opinion that this model will yield unconservative estimates of stresses. If the iteration process were successfully completed, the deforma-tion of the north and south walls will be straight lines. The only stresses that would be computed would then occur due to racking of the structure caused by the difference in the north and south wall straight ines. It should be clear that if a best fit plane could be passed through all the settlement points under both the north and south walls, no stresses would be computed anywhere in the building. The stresses computed by this approach are a function of which iterative cycle is used to define to soil spring paraneters, and bears no resemblance to the actual soil conditions at the site. There is no reason to expect that the soil stif fness should vary from point to point as shown by the analyses. We therefore conclude that this approach to compute settlenent stresses is inappropriate.

I 2.4.2 Bechtel's Analysis Using Measured Settlenents (Ref. 3)

This analysis was performed using the same finite element model described a bove. This time however, the known survey displacement data was input to the program at the ten (10) wall intersection points. The settlenents used were the displacement increments measured for the fourth time period described above. At the remaining 74 boundary element points, the structure was allowed to deform as required to maintain equilibrium (forces equal zero). It was found that computed stresses were very high in those elements adjacent to the wall intersection, but fall of f rapidly away from these points. This indi-cates that the analysis overly penalizes the structure by imposing large con-centrated forces at the wall intersections. In fact, at some points, the soil is required to pull the structure downward to match these known displacements.

. o A modi fied analysis was perf onned by Bechtel at the suggestion of the task group. Rather than input only the ten known displacements, a snoothed curve was generated which matched the known settlement data, but eliminated the sharp profile changes developed in the analysis described above. A best fit polynanial was passed through both the north and south wall settlements, and displacenents computed at all boundary element points of the finite element model. Comparative plots of wall profiles indicate that this approach would still yield high stresses.

2.4.3 Matra's Analysis Using Measured Settlements (Ref. 4)

The analysis performed by Matra is similar in intent to that described a bove . Differences between the two are as follows. First, this finite element analysis was performed for all four time periods described in Section 2.4.1. Three separate finite element models were used to define the DGB at various stages of construction. For each problem analyzed, the known settle-ment data at the wall intersection ooints was input to the models. The report does not specifically state what input was used at the remaining boundary element points between the wall intersection. However, at the interview, Matra stated that a linear displacement profile was assumed between these points. The stress results of the analyses are similar to those described above for the Bechtel study, with similar conclusions reached. In fact, it can be anticipated that the Matra stress calculations would be even higher than the corresponding Bechtel results due to the linear assumption between data points. If in fact this was done, the conclusions reached in that report would be of little value since such high bending stresses would be generated at these discontinuities.

2.4.4 Estimation of Stresses from Crack Data (Ref. 5)

Sozen considered the problem of predicting reinforcement-stresses fraa a knowledge of the crack. patterns. He observed that the usual problem is to predict crack width based upon a given reinforcement stress. When these methods are applied to the DGB center wall, a 20 ksi steel stress is consistent with a crack width of 20 mils. He also adds +he crack widths for a series of cracks in the center wall and equates this to the total elongation 4

• o in the reinforcenent. Using an estimatea gage length over whicn this elongation occurred he obtains an estinated stress of 24 ksi, and indicate, a probable range of 20-30 ksi considering the uncertainties of the method.

(This was presented by Sozen at the August 24 meeting). It is likely that these stress values would be reduced with time. A major cause of cracking was the hard points provided by the duct banks. When these were cut free, one would expect the stresses induced by the uneven support to be relieved. Creep in the concrete would also tend to relieve the settlement-induced stresses.

Rinal di (pg.11086 of the testimor!y) reported at the interview of Septenber 8, that he calculated stresses using Sozen's method in each of the 5 cross walls, as well as the~ north and south walls. He then added these stresses to the maximum stress reported in each of the walls by Bechtel. The resultant maximum reinforcement stress was found to be less than 54 ksi (the allowable limit). It was noted that the Bechtel stresses already included settlenent stresses (to an unknown degree however) from the analyses described in 2.4.1. The crack-based estimates of settlement stresses were added to the maximum of the Bechtel stresses without regard to where they occurred. Whil e this is a conservative approach, there is no documentation of the computa-tions. It should be noted that there would be some question in the applica-tion of this method on those walls where relatively few cracks occurred.

2.5 Stress Totals

~

The finite element model described in 2.4.1 was used to calculate wall forces from all loadings except for the seismic loading. A lumped nass model was used to determine forces resulting from the seismic loading. These forces-were then combined according to the load combirations_ required in ACI 318 and ACI 349. Critical elements were then identified in each of the walls and Bechtel's program OPTCON used to evaluate reinforcement stresses. OPTCON detennines the reinforcement stress resulting -from out-of-plane bending noment plus in-plane shear loading. The shear capacity of the concrete is deducted from the total shear load with the difference assumed to bec'arried by the rei n forcement. The following are peak 'r einforce6ent stresses reported by Bechtel for the critical load cases: north wall - 22 ksi; south wall - 34 ksi; west wall - 29 ksi; east wall - 23 ksi; and interior walls - 20 ksi.

The allowable steel streess is 54 ksi.

o a

2. t, Survey Data Bechtel reports that the accuracy of the survey data describing the DGB settlcnents is 1/8" until the surcharge was removed and 1/16" since that time.

Standard survey techniques and equipment were used.

3.0 ASSESSMENT OF THE DIESEL GENRATOR BUILDING The DGB has undergone very large settlements which have undoubtedly caused serious st ructural distress. This distress is nanifested in the cracks which have occurred in the building. The purpose of this section of the report is to give' an opinion as to (1) whether the building is structurally '

sound and (2) whether the building still meets the criteria as stated in the FSAR.

An important issue is whether the major part of the settlement has occu r red . The settlement data indicate that settlements are well into the secondary consolidation phase so that large additional settlements would not be anticipated. This leads to confidence that predictions of the adequacy of the structure tlased on settlements which have taken place to date should hold for the 11fe of the structure. Certainly, . settlements should be monitored and the problem reconsidered should more than the anticpated additional settle-ments occur. Relative settlements of points on the structure of .005" are significant; The accuracy of the settlenent measurements should be refined to reflect this requirement.

Whila significant cracking has occurred in the structure, it would appear that there/is little evidence to indicate that the structure is unsound. The structure is very massive and is not subjected to large loadings. Even the tornado and seismic loadings do not_ introduce large stresses and usually these stresses occur at locations that are not critical ' locations ,for the settlement stresses.

It is difficult to show that the stresses in the DGB meet the criteria of

'the FSAR. Bechtel's straight line analysis (see 2.4.1) is based on the claim that the settlement ~ survey data is not sufficiently accurate to calculate

(. .

~

I 3 structural stresses. The cdjustnmnt they make to account- for this inaccurocy gives results that are likely unconservative. If c'onservative 4:skumptions are made then the calculated stresses are too large to satisfy the criteria and i

not consistent with the ' crack pt.tterns cbserved in the strdcture (see 2.4.2).

j lt is doubtful whether any analysis could now be developed which would pro-vide.more realistic estimates of settlement stresses witb the required degree of confidence. ,

The most likely source for obtaining reasonable estimates of settlement t

stresses are the crack studies (see 2.4.4). However, these studies must be f documented much more cocipl'etely than has been .donc to date. It is imperative that significantly better methods -be used to monitor crack growth than is currently being considered. - Whitemare strain gages should be used exten-

{

sively. Plugs are attached' to the concrete on a 2" gage. An instrunent is then used to measure the distanca between the plugs. Accuracies of .0001" is r outi ne. Such gages would 'give 'a good picture of the overall behavior of the c rack s. It shoul'd be noted that t'he repair of cracks would not interfere with the use of these instruments. No special " windows" need to be maintained

\

during the crack repair psrogram.. This program of crack monitoring is also important because there is some indication that cracks in the DGB have not stabilized and that the number of cracks mayz in fact be increasing.

4.0 RESPONSE TU CONCERNS OF R.B. LANDSMAN The Regicn 111 inspector has_ raised four concerns (Ref.1) regarding the adequacy of the DGB.

Each of these is, addressed in the following. /

-Concern 1: . FINITE ELEMENT ANALYSIS ^

. 4

~

1 The first concern deals with' the Bechtel finite' element models -(see 2.4.1 1

and 2.4.2) of the UGB used to evaluate stresses due 'to .ettlement.- There are four objections made to the models.

s i, . #

, Concern is raised with regard to the ua of e / dcked section properties while} the concrete iskknowhito be cracked. - All concrete Structures are 7

a

{  ?

i s s a[ ,

. . .f, a 1; il 3 , ,

g

\. [

. e l

cracked and it is standard proctice (specifically pennitted in the ACI code) to determine forces in concrete structures based on gross section properties (i.e., neglect the cracks in the concrete and the reinforcement). If cracked section properties were used then the stresses calculated by Bechtel (2.4.1) would have been smaller. Therefore neglecting cracks in this analysis is a conservative approximation. On the other hand, the analysis reported in 2.4.2 was used to show that the measured settlements result in stresses which are so high that much more severe cracking would be expected than was observed. It was then argued that the measured values must be in error. If cracked sections were assumed for this analysis the calculated stresses would have been snaller, but probably still not consistent with the observed crack l patterns.

t l

I The straight line representation of the settlements along the north and south wall for the analysis reported in 2.4.1 is said to be in error. As in-dicated in that section of this report, it is our opinion that this analysis will result in unconservative predictions of stresses' due to settlements. As l such, it is considered to be an inappropriate analysis.

I-The third part of this concern raises questions regarding the time ef fects of the settlements. Bechtel does calculate stresses for dif ferent phases of the settleaent. The structure was changing during the significant settlement period. Construction was still in progress during the largest set tlenent s. Therefore the structural geometry changed as did the cuncrete properties (while maturing). - The Bechtel models did not account for these cha n ges. This would have been conservative for the calculation of stresses, I

but would result in lower stresses in the analy'ses performed using the ~

measured settlenents as input.

The fourth objection deals with the claim that the NRC' staff did not approve of the Bechtel analysis. It appears that this is the case and the intention of the staff was to use settlenent stress data based on an analysis -

of the cracks rather than the finite element analyses.

b.

9

Concern 2: RELIABILITY OF ME ASURLD StifLEMLHT VALUES The analyses reported in 2.4.2 and 2.4.3 were used to show that stresses computed from structural models subjected to the measured settlerents are very high and would indicate cracking in the structure where no cracks are ob-served. The objection is raised that a linear model was used and that a non-linear model accounting for plastic ef fects would result in a redistribution of stresses and the same conclusion may not apply. This observation is true, but by itself would not change the conclusions drawn from these analyses.

As stated above, however, there are other factors which when coupled with L

this objection may result in a dif ferent conclusion. The other important f actors are: the assumed shape of the settlement between the measured points; and the differing geonetry of the OGB when the various phases of settlenent -

O C Cu r red .

Concern 3: STRESSES DETERMINED FROM CRACK SIZES l

If the finite element analyses are not reliable then one alternative approach is to find settlenent stresses from a study of the crack sizes. The objection raised is that this approach is not consistent with normal engi-neering practice and that there are no equations available to evaluate stresses from crack data when the stress fields 'are as complex as occur in the i DGB. It is true that this would not be standard practice, but "non-standard"

analyses may be used provided they are sufficiently documented and shown to give results that are conservative.

An approach that could predict approximate settlement stresses in the DGB could probably be used to demonstrate its adequacy. This is true for two reasons. First, stresses in the structure due to other. loadings are!rather low and there is a'large reserve for settlement stresses. Second, if large settlenent stresses and local yielding of the reinforcenent occurs, the resulting deformations of the structure will reduce the settlement-induced l oa di ngs.

The docunw.mtation of the crack analyses used to determine stresses is not su f ficient. There is no calculation on record which calculates stresses in d il of the walls using tnis method. There is also no written justification showing that the octnod may be used for structures like the DGB.

Concern 4: CRACK MUNITORING This concern deals with the lack of a good crack monitoring system and specification of action to be taken if the cracks exceed certain limits. As stated in Section 3.0, it is our opinion that the planned crack monitoring systen is not adequate. More reliable gages (e.g., Whitenore Strain Gages )

should be placed in areas where cracking is now evident. These gages can be used even af ter crack repairs are made.

Two limits are now defined in the current crack monitoring program. If the crack width reaches .05" (Action Limit) a meeting will be held to evaluate what stcps to take when the cracks reach the next limit. The next upset limit is set at .0e" (Alert Limit). It is our opinion that the form of this plan is adequate, but that the specific threshold numbers must be based on a resolu-tion of the current settlerent stresses. A safety margin must be lef t for the other potential loading events, such as tornado or seismic loads, with the re-naining allowable stress allocated to future potential settelments.

Once this limit was reached the only solution would Le to make a struc-tural repair. The exact form of this repair would depend on the location and extent of the crack which exceeded the limit. The p'> r ne cesponse could not specify the nature of the repair, but could indicate that an exceedance of the Alert Limit would result in a structural repair rather than performing addi-tional analyses.

5.0 C0tCLUSIONS Based on the review of the studies performed to demonstrate the adequacy of the UGB, the following conclusions are drawn:

. o

1. The settlement data indicates that primary consolidation of the fill is completed. However, it is recommended that the anamolies in the documentation of the settlement history be resol ved. (See last paragraph of Section 2.2).

2. It is unlikely that a satisfactory stress analysis can be perfonned based on the measured settlement data. It is recommended that settlement stresses be estimated from the crack width data. The existing work that has been done in this area must be completely documented.

3. It appears that the number of cracks in the DGB are con-tinuing to increase. It is essential that a better crack monitoring program be established as outlined in Section 3.0.

4. The upset crack width levels specified in the crack monitoring program should be chosen so that a sufficient stress margin is available to resist the critical load combinations.

5. If the Alert Limit (in crack width) were exceeded, specific structural repairs should be mandated.

6. While significant' cracking has occurred in the DGB, it is our opinion that the structure will continue to fulfill its functional requirement. This conclusion is based on the f act that stresses induced in the structure by all other extreme loadings are small.

REFERENCES 1.

Memorandum for R.F. Warnick through J.J. Harrison from R.B. Landsman, Subject Diesel Generator Building Concerns at Midland, dated July 19, 1983.

2. Bechtel Calculation No. DQ-52.0 (Q), Rev. 2.

3.

Bechtel Calculation No. DQ-52.7 (Q) - Finite Element Calculation of Settlement Stresses Using Actual Displacements.

4. Structural Reanalysis of Diesel Generator Building Utilizing Actual Measured Deflections as Load Input, by John Matra, Naval Surf ace Weapons Center.

5. Evaluation of the Effect on Structural Strength of Cracks in the Walls of the Diesel Generator Building Midland Plant Units 1 and 2, hy Mete Sosen, Februa ry 11, 1982.

6. Effects of Cracks on Serviceability of Structures at Midland Plant, by W.G. Corely, A.E. Fiorato, and D.C. Stark , April 19, 1982.

7. Executive Summary, Diesel Generator Building, Midland Plants Units 1 and J

2, August 1983.

8. Letter from CPCo to NRR dated October 21, 1981; Enclosure 1 Tech.

Report, Structural Stresses Induced by Differential Settlement of the DGB.

4

____m. _ _ _ . _ _ . _

APPENDIX A: SOURCE MATERI AL FOR STUDY Site Specific Response Spectra Midland Plant Units 1 & 2 Addendum to Part 1 Re'.punse Spectra--Uruinal Ground '.urfot o Jan 81 Weston Geq4sical Corp Site Specific Response Spectra Midland Plant Units 1 & 2 Part II Response Spectra Applicable for the top of fill material at the plant site April 81 Weston Geophysical Corp Site Specific Response Spectra Midland Plant Units 1 & 2 Part III Seismic Hazard Analysis Feb 81 Weston Geophysical Corp Soil Boring and Testing Program Midland Plant Units 1 & 2 Test Results Foundation Soils Auxiliary Building Woodward-Clyde Consultants Aug 81 Docket Nos. 50-329,50-330 Test Results Perineter and Baf fle Dike Areas Soil Boring and Testing Program Volume II Supporting Data July 81 Docket Nos. 50-329,50-330 Test Results Perineter and Baf fle Dike Areas Soil Boring and Testing Program Volume 1 Woodward-Clyde Consultants July 81 Docket Nos. 50-329,50,330 Estimates of Maximum Past Consolidation Pressure of Cohesive Fill Materials Diesel Generator Building July 81 Woodward-Clyde Consultants Docket Nos. 50-329,50-330 USA /NRC Before The Atomic Safety and Licensing Board 12/7/62 testimony of; Frank Rinaldi John Matra Gunnar Harstead with respect to the Structural Adequacy of The Diesel Generator Building at Midland Official Transcript Proceedings Before NRC Atomic Safety and Licensing Board OKT/ CASE No. 50-329,50-330 OL & OM 12/10/82 pages 11008 through 11228 A-1

)

I

Evaluation Report for Concrete Cracks in the Diesel Generator Building Consusiers Power Company 2/16/82 Evaluation of the Effect on Structural Strength of Cracks in the Walls of the Diesel Generator Building Mete A. Sozen 2/11/82 Relationship of Observed Concrete Crack Widths and Spacing to Reinforcement Residual Stresses Consuners Power Company 6/14/82 Observed Cracks in Walls of Midland Plant Structures 6/14/82 Corley and Fiorato Portland Cenent Association Safety Evaluation Report related to the operation of Midland Plant Docket Nos. 50-329 and 50-330 Consumers Power Company USNRC 5/82 Effects cf Cracks on Serviceability of Concrete Structures and Repair of Cracks Consuners Power Company 4/30/82 Effects of Cracks on Serviceability of Structures at Midland Plant Corley, Fiorato, Starx Portland Cement Association Summary of Sept. 8, 1981 Meeting on Seismic Input Parameters Midland Plant USNRC 12/3/81 USA /NRC Before the Atomic Safety and Licensing Board 50-329,50-330 testinony of Jef frey K. Kimball 9/29/81 NRC Atomic Safety and Licensing Board 50-329 OM,0L 50-330 OM,0L witnesses; Johnson Burke Corley Sozen Gould NRC Before the Atomic Safety and Licensing Board (no date)

NRC staff testinery of Joseph Kane on Stant ris Contention 4.8 Docket Nos. 50-329 OM,0L 50-330 DM,0L Safety Evaluation Report related to the operation of Midland Plant October 82 Docket Nos. 50-329 50-330 USNRC NUREG-0793 Supplenent No. 2 Safety Evaluation Report related to the operation of Midland Plant June 82 Docket Nos. 50-329 50-330 USNRC NUR EG-0793 Supplenent No.1 A-2

l NRC Atomic Safety and Licensing Board 9/29/81 Applicant 's Brief on Compatibility of Site Specific Response Spectra Approach with 10 CRF part 100 Appendix A Safety Evaluation Report related to the operation of Midland Plant May 82 Docket Nos. 50-329 50-330 NUREG-0793

~

Response to the NRC Staff request for Settlement Related Analyses for the Diesel Generator Building b/1/82 Consumers Technical Report Structural Stresses Induced by Differential Settlement of the Diesel Generator Building Consumers Power Company Test Results of Soil Boring- and Testing Program for Diesel Generator Building Docket Nos. 50-329 50-330 7/31/ 81 Consumers Power Company Final Results of Soil Boring and Testing Program for Perimeter and Baffle Dike Areas 7/27/81 Docket Nos. 50-329 50-330 Consumers Power Company NRC Atomic Safety and Licensing Board Docket Nos. 50-329 OM,0M 50-330 OM,0L Witnesses; Hood 12/3/81 Kane Singh Rinaldi NRC Atomic Safety and Licensing Board Docket Nos. 50-329 OM,0L 50-330 OM,0L Witnesses; Kennedy 2/17/82 Campbell Rinaldi Kane Matra Hood Singh CSE Input to the Midland SER Supplement Aug. 82 Geotechnical, structural, mechanical and hydrologic inputs for the Midland Ser Supplement Transcript of Proceedings USA /NRC 1/6/81 Deposition of Frank Rinaldi Transcript of Proceedings USA /NRC 1/9/81 Deposition of Pao C. Huang Transcript of Proceedings USA /NRC Docket Nos.- 50-329 OM, OL 50-330 OM,0L Deposition of John P. Matra 1/7/81 A-3

USA /NRC Bef ore the Atomic Safety and Licensing Board Docket Nos. S0-329 OM-OL 50-330 UM-OL NRC Staf f Brief in Support of the use of a Site Specific Response Spectra to comply with the Requirements if 10 CFR Part 100, Appendix A 9/29/81 USA /NRC Befor-e the Atomic Safety and Licensing Board Docket Nos. 50-329 OM-OL 50-330 OM-OL Testinony of Dr. Paul F. Hadala with Respect to the Study of Amplication of Earthquake Induced Ground Motions and the Stability of the Cooling Pond Dike Slopes Under Earthquake Loading 9/29/81 USA /NRC Before the Atomic Safety and Licensing Board Docket Nos. 50-329 OM,0L 50-330 OM,0L Witnesses; Boos Hendron Hanson Testinony of Ralph B. Peck before the Atomic Safety and Licensing Board, in the the matter of Consumers Power Company (Midland Plant, Units 1 and 2), Docket Nos.

50-329 UM, 50-330 ON, 50-329 OL, 50-330 OL, notarized Nov. 3,1982.

Letter from CPCo to H.R. Denton dated June 14, 1982 with Enclosure " Response to the NHC Staff Request for Additional Information Required for Completion of Staff Review of Soils Remedial Workd dated June 14, 1982.

Summary of August 17, 1982 Meeting on Soils-Related Construction Release, dated September 7,1982, by Darl Hood.

" Structural Reanalysis of Diesel Generator Building Utilizing Actual Measured Deflections as Input", by John Matra.

Letter from CPCo to H.R. Denton dated October 21, 1981 with

Enclosures:

" Structural Stresses Induced by Differential Settlement of DGB",

"Subgrade Modulus & Spring Constant Values for DGB Structural Analysis",

" Bearing Capacity Evaluation of DGB Foundation" "Logterm Monitoring of Settlenent for DGB",

" Relative Density and Shakedown Settelment of Sand under DGB",

"Estinates fo Relative Density of granular Fill Materials, DGB",

" Review and Control of Facility Chagnes to DG8",

"DGB Bearing Pressaure due to Equipnent and Commodities",

Report form Woodward-Clyde to CPCo dated June 10,1981, " Preliminary Test Results, Soil Boring & Testing Program, Perimeter and Baffle Dike Areas",

" Seismic Margin Review, Midland Energy Center Project": Volumne 1, Methodology and Criteria, dated February 1983, Volume V, Diesel Generator Building, dated July 1983, prepared for CPCo by Structural Mechanics Associates.

A-4

i 1

Applicant's Propsed Findings of Facts and Conclusions of Law on Remedial Soils Issue Docket Nos. SU-329-0M 50-330-0M 50-329-OL 1

50-330-OL Testimony of Karl Weidner for the Midland Plant Diesel Generator Building September

, 8, 1982 Docket Nos. 50-329-OL 50-330-OL 50-329-0M 50-330-0M '

Find Report on the ADINA Concrete Cracking Analysis for the Diesel Generator Building by Gygna Energy Services, September 16, 1981 a

d 1

a >

t i

A-5

l

.,  !* I . ~

• e APPENDIX IV -

TNCLOSURE u

[ge**se ugitgD sTAf tg g NUCLEAR REGULATORY COMMISSION

. e l

• ntcloN lil *

' q '

.I too noosavtLv nono

• ..... otsN s tLYN. ILLiwoes ,s,og7, ,

i

UUL T 9' %f HEMORANDUM FOR

R. F. Warnick. Director. Office of Special Cases '

THRU: J.Nk J. Harrison. Chief. Section 2. Hidland .

FROM:

R. B. Landsman. Reactor Inspector

SUBJECT:

DIESEL GENERATOR BUILDING CONCERNS AT MIDLAND .

At the recent hearing before C'ongressman Udall's subco=ndetee . I ' expressed because building over the of years.

numerous structural cracks that havengoccurre u ASLB hearings. I also expressed the same concern during the recent Mr. Eisenhut has requested me to document the basis of my concerns about the building so an iodependent review group can analyze them .

My first concern deals with the finite element analysis that Consu=ers Power Company (CPCo) used to show that the building is structurally und. so Their cracks.model of the building assume,d a very rigid structuren without a y i

structure.The building has numerous cracks, reducing the rigidity of the in the analysis. The effects of these cracks have notdata been CPCo's interpretation of the settlement as taken a into account straight line approximation always stems from their position that the building is too rigid to deform as indicated by actual settlement readings.

t The phases settlement of construction. of the building occurred over a period of time during erent diff used in their model. It is this time dependent effect that was also not Even CPCo expert Dr. Coroly testified at the ASLB hearings that the analysis should have "taken into account cracking dependent effects" in order to give correct results. me and ti of ficial position, as stated by Dr. Schauer., on CPCo's analysis vas. "TheFinall staff takes no position with regard to that analysis."

My second concern deals with the ceceptance of the diesel generator building in the SSER #2 which was subject to the results of an analysis to be performed by the NRC consultants using the actual settlement valuec.

The consultants testified at the ASLB hearing that this analysis gave unacceptable.results 'and this portion of the SSER should be stricken. They are basing their unacceptable results and comments on their finding of 4

go -

1 yp#y ,

i I

)

. 1

~

R. F. Warnick 'JUL i S 983 1 I

1 i .

1 .

very high stresses obtained in areas where no cracks exist. Therefore, i

the actual used in an settlement analysis. values are not accurate enough (are in ' error) to be

• i The consultants, as well as CPCo, ran.a linear analysis (structure always in the elastic range) instead of a plastic analysis which would allow a redistribution of loads in the structure. Th erefore. .

supposed areas of high stress, where cracks are not located, may not ex'ist due to redistribution of loads. Finally, the staff's of ficial position, as stated by Mr. Rinaldi, on this analysis as performed by the consultants,

, was that the actual settlement values could not be relied upon to detercine

if the diesel generator building meets regulatory requirements.

My third concern deals with the fact that we are not follo ing noba1 engineering practice in accepting the building by using a crack analysis approach because there is no practical method available today to analyze a complex structure with cracks in it.

The basis of this' concern is that there are no formulas available that can estimate stresses in a complex stress field like those which. exist in this building. Thus, the evaluation of the structure based on the staff's crack analysis using empirical-unproven formulas to determine the rebar stresses is unacceptable.

4 My fourth concern deals with the staff accepting the building by relying i on a crack monitoring program to evaluate the stresses during the service' life of the building. If cracks exceed certain levels, reco=mendations 4 vill be made for maintaining the structural integrity of the building.

The basis for my concern deals with the lack of crack size criteria and the lack of formulated corrective action to be taken when the allowed crack sizes are exceeded.

These concerns which I have just enumerated are also shared by members of Mr. Vollmer's engineering staff, as well as their consultant. These concerns were documented in the ASLB hearing transcripts of December 10, 1982, prior to my ever expressing my concerns before the ASLB hearing or

Congressman Udall's subcommittee.

• In summary, since it is impossible to analyze this severely cracked j

structure to the total staff's approval. I reco:znend some remedial structural fixes be undertaken to ensure the structural integrity of the building.to provide an adequate margin of safety'.

• f4 1.4 n 1m Ross B. Landsman Reactor Inspector cc: DMB/ Document Control Desk (KIDS) i 4

i i

I

= :--- = . ,::r.~r- . . r. v . . .wwe'-w. * '"--"** ~~

^

_ _ . . _ _ . . _ . . . _ _ . _ . _ . , _ , _ , _ _ _ _ _ _ . _ . _ , _ _ ,_., _ _ , ~ _ . _ - _ . _ __ ___