ML20125B519
| ML20125B519 | |
| Person / Time | |
|---|---|
| Site: | Midland |
| Issue date: | 12/07/1983 |
| From: | Little W NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III) |
| To: | Clark J NRC OFFICE OF RESOURCE MANAGEMENT (ORM) |
| Shared Package | |
| ML20125B272 | List: |
| References | |
| FOIA-84-893 NUDOCS 8506110440 | |
| Download: ML20125B519 (1) | |
Text
/p * * *
- kq'% tliilit l' 51 Al t 5
[*,,, j NUCLE AR Ri Gill ATORY COMMISSION
,a nor. ni s ,
t . / 7W koostvt L1 ROAD
% / cit N i t tvu. sLLwois som December 7, 1983 ,,
MEMORANDUM FOR: John W. Clark, Chief ,
Management Analysis Branch, Office of Resource Management FROM: W. S. Little, Chief Engineering Division Branch 2 - Region III
SUBJECT:
10 CFR 50.55(c) ITEM REGARDING MIDLAND CONTAINMENT PROTECTIVE COATING DEFICIENCY t
This is to cxpress our appreciation of the technical assistance which l Messrs. L. Abracson and D. Lurie of your staff provided to Region III.
l Their assistance was instrumental in our assessment of Consumer Power Company's statistieni inethodology iclative to the testing of potentially defective protective coatings.
A E nin, thank you for your assistance in this matter.
'I
.l @
P. S. Littic Chief Engineering Division, Branch 2 l
l JL -993 PM
M ,. m (jyk.)idTk. an+ fh c .
. ;, g'- p Ah -
1 COnSumeIS ' uw,
~
J~
Power J.me. W Ceek Vete hesdent = hojects, Eng4ntenng and Construction
- c. ..i ow isas w..i r.,n.n m..a. s.o . m 4eroi . isin 7ss o4s2 December 17. 1982 79-12 #11 l
Mr J G Keppler. Regional Administrator i
US Nuclear ReEulatory Commission t
Region III 799 Roosevelt Road l Clen Ellyn. IL 60137 I
i MIDLAND NUCLEAR COCENERATION PLANT DOCKET NOS 50-329 AND 50-330 CONTAINMENT INTERNAL STRUCTURES C0ATING DEFICIENCY TILE: 0.4.9.37 SERIAL: 20649
References:
- 1) J W Cook letter to J G Keppler, same subject. dated October 30. 1981 Serial 1459E
- 2) NRC Inspection Report No 50-329/82-10 and 50-330/82-10 dated July 28. 1982 keference 1 was our final 50.55(e) report on the contain:ent internal structures coating deficiency. During the NRC site visit to close out this iten (see Reference 2) it was determined that supplementary information was required to be developed to allow the NRC final closure. The attachnent to this letter is a supplementary final report and provides the inforcation that was nece nary.
JWC/L'RB/1r Attachmenti HCAR 35. Revised Final Report Revision 2. dated Fnvember 10. 1982 CC: Docunent Control Desk. NRC Washit.gton DC RJCook. NRC Resident inspector Midland Nuclear Plant 99 b
y ,x
, pl
( _
m r,1m OC1282-0034A-Hr01 J
f.
4'j 2 l , ,
t Serial 19119
- 81-05 #7 CC
- CBechhoefer, ASLB Panel RSDecker. ASLB Panel FPCowan, ASLB Panel j JHarbour ASLB Panel -
l AS&L Appeal Panel l MMCherry, Esq MSinclair BStamiris CRStephens, USNRC WDPaton, Esq. USNRC FJKelley, Esq, Attorney General SHFreeman, Esq. Asst Attorney General WHMarchall CJNerritt, Esq. TNK&J l
1 .
1 l
l OCl282-0034A-NP01
083779 Attechmint to
'c* 93S5S serial P0a9 Bechtel Associates Professional Corporati5ht2 '22 Management Corrective Action Report (MCAR)
SUBJECT:
NCAR 35 (issued 11/13/79)
Containment Internal Concrete Structures Coating PINAL REPORT REVISION 2 DATE: November 10, 1982 FROJECT: Consumers Power Company Midland Plant Units 1 and 2 Bechtel Job 7220
1.0 INTRODUCTION
During a planned coating inspection in November 1979, a Consumers Power Company engineer noted that there was a loss of adhesion between multiple coats of a decontaminable surfacing system used on concrete walls. The coating system was specified as System 9 in Specification 7220-A-15(Q), Technical Specification for Subcontract for Epoxy or Phenolic Decontaminable Surfacer for the Midland Plant Units 1 and 2 Consumers Power Company, Midland, Michigan. J.L. Manta, Inc, was the coatings subcontractor responsible for installing and inspecting decontaminable coatings included in the specification.
Nonconformance Report M-01-4-9-132 was written on November 8, 1979, requiring identification of the cause and extent of the i nonconforming work. All concrete coated with System 9 was subject to testing. Inspection indicated that the loss of adhesion could be detected in the containments and other plant buildings. No apparent cause could be identified. The adhesion failure occurred between unspecified cultiple coats of NuKlad 117(N), surfacer, resulting in disbondcient of a final surfacing coat and the two finish coats. The first surfacer coat was firmly bonded to the concrete.
Preliminary analysis indicated that a potential safety problem existed. Should coatings be removed during a reactor accident that initiated the containment spray system, the failed coating f rom some areas afght be carried to the sump and cause screen blockage. As a result of this analysis, MCAR 35 was written November 13, 1979. It was determined that the criteria contained in this HCAR were reportable under 10 CTk 50.55(e).
This report incorporates or provides references to data furnished previously in HCAR 35, Interim and Final Reports, which were submitted in accordance with reporting requirements of 10 CTR $0.S$(e).
It includes a description of corrective action that has been leplemented to replace defective coating in coitainment Units 1 and 2 The reporting format follows the guidelines of Bechtel Engineering Department Procedure (EDP) 4.60, Rtv 1, Section 2.0.
0766f
g93779 093959 Bechtel Associates ProfessionalCorporation NCAR 35 Revised Final Report Page 2 of 22
2.0 DESCRIPTION
OF DEFICIENCY Coating System 9 includes four application steps of a th'n film surfacing system recommended by Ameron Protective Coatints Division. NuKlad 117(N) was first applied as required te fill voids 3/4 inch or less. A continuous surfacing coat wr then applied to provide a pinhole-free surface using a rubbre-faced trowel to spread and smooth the coating. Following a specified cure period, Amercoat 90 finish was applied at approximately 5 mils thickness. Pinholes were repaired with NuKlad 117(N) or Amercoat 90 before the final 5 mil coat of Acercoat 90 was applied.
Delamination occurred between two costs of NuKlad 117(N).
Contrary to Ameron's data sheet (R3/78), which was incorporated in J.L. Manta's quality assurance manual, two complete coats had been applied. J.L. Manta confirmed that this practice was followed during most of the work and was permitted by Ameron in a revised data sheet (November 1978). However, this revised data sheet was never submitted for approval to Bechtel.
Containment coating systems at Midland must be prequalified to Midland design basis accident criteria in accordance with Specification 7220-A-45. Syste= 9 was prequalified as a single surfacing coat; requalification of multiple coats would be required before approval for use. This factor resulted in a potential nonconformance status of all System 9 coating that had been applied, and no further work was permitted in containment. From November 11, 1979, to November 30, 1979, random adhesion tests by knife verified that the delamination problem was significant enough to preclude a cause based on a unique set of circumstances, such as local contamination or poor mixing of components. A preliminary evaluation program was developed by J.L. Manta and Bechtel to systematically inspect all surfaces coated with System 9 to determine the extent of the failure. It was known at this time that all coating System 9 did not delsminate and that the problem was most severe on interior shield walls at the lower elevations of the Unit 2 containment.
When the coating deficiency was identified, the investigation was subdivided in the following manner:
- a. Location of specific boundaries of sound and unsound coating System 9 .
0766f
gg3g g
' ' 'V Bechtel Associates ProfessionalCorporation MCAR 35 Revised Final Report 09377g Page 3 of 22
- b. Determination of the root cause of the delamination
- c. Investigation of corrective procedures for removal and replacement of failed coating System 9 with an acceptable coating system These requirements made it necessary to establish a schedule of evaluations in which each activity was assigned and completed independently. Preliminary investigations provided a basis for progracmed work by Bechtel, Consumers Power Cocpany and consultant laboratories. It required approximately 2 years to complete all investigations and perform analysis of tests and data. It was discovered that the cause of delsmination was related to several factors, each of which could independently or collectively lead to coating failure.
3.0 SUyy.ARY OF INVEST 1 CATION AND HISTORICAL BACKCROUND 3.1 PRELIMINARY INVESTICATION 3.1.1 Adhesive Properties of Epoxy Base Products - Generic j Each coating system specified for application to concrete surfaces is cocposed of one or more epoxy coating materials, each with a defined function within the composite coating system. When cultiple coats are applied over each other, each coat must adhere to the previous coat. The first coat must bond to the cementicious substrate.
Adhesive properties to the substrate and between
. coats are measurable by test. However, because the epoxy coating system strength exceeds the concrete tensile strength by an appreciable
-margin, it is unusual'to have adhesive failure within the coating system. Failure analysis requires that the source _of bond strength be examined. Generally, the adhesive strength of coatings is a result of the following factors:
- a. Partial solubility and wetting of the previous coat
- b. Mechanical interlocking to a porous or rough surface, or previous coat 0766f
.~ : . .. - -
VS3STS
"~
Bechtel Associates Professional Corporation MCAR 35 Revised Final Report 000l{Q Page 4 of 22
~
- c. Bonding due to polar or electrical attraction of molecular groups typical of each generic type of resin used in the coating vehicle ,
Epoxy base products can be designed to adhere by one or all of the above factors. Because no '
failure occurred at the substrate interface, it was apparent that the problem was related to wetting and proper bonding of a second coat of NuKlad 117(N) to a partially cured NuKlad 117(N) surface. A reliable bond can be obtained to a previous coat if the surface is uncontaminated.
The degree of cure of the previous coat has a definite influence on maximum bond strength.
'NuKlad 117(N) is not recoatable following I week cure at 73F based on Ameron data. Furthermore, any surface condition resulting from contamination or exudate that may occur during cure of the first coat can partially or completely eliminate adhesion between coats.
The preliminary and progracmed investiEation was designed to methodically evaluate potential causes of reduced adhesion between multiple coats of NuKlad 117(N). It was known in the initial stages of the investigation that at least two coats of NuKlad 117(N) were applied contrary to the specification, but approved by Ameron. This constituted the as-built condition, and cpot checks on adhesion verified that all coating work was not deficient. The principal concern was related to location of failed areas and, if possible, determining why certain areas failed and adjacent locations were acceptable.
3.2 EVALUATION OF INSTALI.ED COATING ' SYSTEM 9 3.2.1 Testing of Coating Failure -Knife Test Method The identification of failed coating required that a method be selected or developed for evaluation of adhesion. Adhesion test methods available are described in ASTM procedures. Few of these methods are designated for field testing of thick films with high tensile strength. The elcometer method, using a bonded aluminum test dolly, 0766f
69395S
" ' Bechtel Associates Professional Corporation "C^^ 35 Revised Final Report 693779 Page 5 of 22 appeared to be the most securate method, but this method had some limitations in areas where access was restricted. By experimenting with knife tests, it was determined that 3-inch parallel cuts 1/4 inch apart (through the coating to the substrate) provided a reproducible method of determining disbondment. If the coating system delaminated, the cut area could be removed by separating layers of the coating using a knife as a probe. The coating could not be separated if adhesion was tight. In a trial area in the auxiliary building, a comparison with e1cometer tests showed that the knife test failed corresponding with e1cometer readings of 200 psi or less [ minimum specified in ANSI N5.12(1974) for acceptable bond to concrete).
J.L. Matta prepared Test Procedure JLM 450-A-5 on the basis of these results, and this procedure was used for determination of pass / fail areas on subsequent testing.
3.2.2 Mapping - Extent of Coating Failure Initially, it was important to determine the amount and location of f ailure -and if there was correlation between the failed area with documented application variables. A single unit of NuKlad 117(N) covers approximately 100 square feet at the recommended coverage rate of 5 to 10 mils dry film thickness. Material losses due to limited pot life af ter mix reduce the practical coverage to under 100 square feet / unit. A 100-square-foot-test area representing a single mix was therefore selected as the initial test area. Sketches were prepared to divide all concrete surfaces into 100 ft2 grids. SKA-205 through SKA-216 were issued for Containment 2 tests. When Unit 2 testing was complete, SKA-217 to SKA-228 were similarly issued for Containment I tests.
Knife tests were performed at a site within the boundaries of each test grid and the test location was marked. This provided an assessment of the approximate percentage of failed area - 18% in 0766f
.93959 ,
"4 Beghtel Associates ProfessionalCorporation MCAR 35 093779 Revised Final Report
- Page 6 of 22 '
l Unit 2 and 6% in Unit 1. Although thls test did not locate the extremities of coating failure, the number of tests and the test rationale provided a '
basis for planned corrective action. It appeared ;
that there was sufficient sound coating to justify
, repair rather than complete removal and recoat.
The total mapping program required several months to complete.
3.2.3 Correlation of Application Variables with Failed Areas i
- During Unit 2 mapping and testing, the daily i
documentation records for the initial application of Systcc 9 were tabulated. This included the following:
- a. Mapping of areas coated on each day for each coat of System 9, including two or more surfacer coats and two or more finish coats.
- b. Batches of each coat used were plotted on the area covered daily by each batch.
- c. Time between coats for each coat was identified (1 week, I to 3 weeks, over 3 weeks).
- d. Teeperature recorded for the surface during application was plotted.
3.2.4 When this data was superimposed over the pass / fail grids on sketches, it was possible to compare application variables with f ailed areas. Although no correlation was evident, certain deviations from specified requirements were noted.
- a. Two full coats of NuKlad 117(N) were applied in most cases. In some instances, more than two coats were applied.
i 0766f
'S3959
' '4 Bechtel Associatas Professional Corporation
~
MCAR 35 Revised Final Report 093779 l Page 7 of 22
- b. Several areas were identified in which the l i
time between coats exceeded manufacturer's recommendations,
- c. In a few isolated steas, Amercoat 90 was
" sandwiched" between coats of NuKlad 117(N).
- d. In some cases documentation was incomplete or inaccurate. Relative humidity, although listed on forms used by J.L. Manta, was not recorded except for a few months after work com=encement.
3.2.5 Frelfeinary Evaluation - Coating Re= oval Methods available for removal of failed ccating were limited because the problem was identified in the advanced staEe of construction. Sandblasting was not practical because of potential damage to valls and adjacent installed equip =ent. AlthouEh water blast by high pressure laser could re=ove the failed coating, it was not co=patible with other construction activities. Minnesota Mining and Manufacturing (3M) produces an impact cleaning tool called a Roto Feen. Trial deconstrations using this equipcent verified that delaminated System 9 was removed from a sound layer of NuKlad 117(N). If dela=ination did not occur, additional cleaning effort was required to recove the entire coating system. Generally, a fresh concrete surface was exposed. NuKlad 117(N) residue remained in pits and depressions. Corners and areas with limited access were cleaned to bare concrete using a needle gun.
The Roto Peen method was adopted for removal because it provided a clean delineation between delaninated areas. These areas generally occurred between steel embeds. This supported the conclusion that an application variable was responsible for delamination. The embed spacing provided a logical boundary to terminate application during vertical drops from upper to lover elevations, and was evidently used to assign work to the painters.
0766f
93dSbS N Bechtal Associates Professional Corporation MCAR 35 093779 Revised Final Report Page 8 of 22
~
All subsequent testing was performed by reapp11 cation of coating systems to surfaces cleaned by Roto Peen. Final cleaning by sanding or Clean-N-Strip Wheels was required.
3.3 REVIEW OF SPECIFICATION 7220-A-15(Q) AND SUBCONTRACTOR QA -
PROGRAM AND PROCEDL7tES 3.3.1 Specification 7220-A-15(Q), Technical Specification for Subcontract for Epoxy or Phenolic Decentaminable Surfacer, was issued for bids on February 17, 1976. The initial work performed by Bagvell Coatings was limited to non-Q-listed work in the Units I and 2 auxiliary buildings. An Azeron coating identified as NuKlad 1871 was used in System 9. J.L. Manta was awarded the contract for all surfacing work not cocpleted by Bagvell Coatings. J.L. Manta's quality assurance program and procedures were prepared on the basis of the products originally specified. The sub-contractor was also required to bid alternative surfacing systems of other manufacturers with products equivalent to A=eron systems. Eased on evaluation of bids, the Carboline alternat!.ve bid by J.L. Manta was not accepted. Therefore, Ameron surfacing systems were used.
Shortly after appifcation was started by J.L. Manta, a safety problem was identified in the A=eron surfacing syste=s. This required a change in the epoxy convertor used, and revised products were tested and approved at Oak Ridge Nat*onal Laboratories (ORNL) using the new systems recoczended by Ameron. The specification was revised to incorporate the product changes NuKlad 110AAB Green in Systems 7 and 8, and NuKlad 117(N) in System 9.
3.3.2 J.L. Manta Concrete Coatings Procedure JLM-450-A-1 was issued and attached to J.L. Manta's Quality Assurance Manual Job 7220, Midland Project, Subcont ract 7220-A-15. Section 6.4.0 contains instructione for application of NuKlad 117(N) and i
e d
0766f 1
-33S59
- 'd B;ght:1 Associatg3gfegsional 7 Corporation MCAR 35 Revised Final Report Page 9 of 22 requires sealing pinholes before topcoating by using either NuKlad 117(N) or A=ercoat 90 depending on the size of the pinhole.
The quality assurance manual and procedures incorporates the specific revision of the -
canuf acturers data sheets that apply to esch individual type of coating; the canufacturer's instructions are candetory.
3.3.3 Nonconfor=ance Identification - Two Cesplete Ccats of NuKlad 117(N)
- a. A review of the doeurents listed in Secticns 3.3.1 and 3.3.2 indf eated that KuKlad ll7(N) applied as two coats was ncnconforming. Neither the specified canufacturers data sheet or J.L. Manta's procedures per=itted two ce=plete coats.. The tested configuration and cost recent revisien of J.L. Manta's quality assurance progra=
pertitted one co=plete coat. In Nove:ter 1978, Ateren issued a new data sheet on NuKlad 117(N) describing cenditions under which two coats were per itted. This data sheet was never submitted to Bechtel for approval and incorporation, but was evidently used by J.L. Manta as a work-controlling docu=ent.
- b. Other ocuconfor=ances are identified as follows:
- 1) Failure to record all infor=ation required or cc=plete all for=s on JLM for=s, which are copies of ANSI N101.4(1972) docu=entation. This includes records of relative hu=idity and dev point calculations on certain f o rms .
- 2) Excessive ti=e between coats was docu=ented on daily application records (see Section 3.2.3).
0766f
093959
' "N Bechtcl Associates Professional Corporation HCAR 35 093779 Revised Final Report Page 10 of 22 3.3.4 Investigations by J.L. Manta and Ameron Ameron Protective Coatings Division was requested by J.L. Manta to assist in evaluation of the problem. Ameron's participation included site visits by technical personnel and laboratory work -
at their Buffalo Laboratory. The correspondence and test reports are summarized below,
- s. Ameron reported that a solvent odor could be detected in areas where no topcoat had been applied. This was apparent immediately following film separation. If topcoat had been applied, it was possible for solvents from the finish to collect at the interface where delamination occurred. However, in accordance with the specification and manufacturers' data sheets, solvent was not permitted in the surfacer.
- b. Ameron reported that the film was soft, indicating incomplete cure. A pink discoloration in the NuKlad 117(N) film was questioned. This was later identified as an iron oxide contaminant in the silica filler used in NuKlad 117(N).
- c. Ameron furnished Laboratory Report 405 on January 29, 1980 The report tended to eliminate two potential causes of delamination, which were subsequently proven to significantly contribute to the problem moisture and Amercoat 6 thinner used in the KuKlad 117(N).
3.4 PROGRA.9 FED INVESTIGATION i
During the preliminary investigation, it was determined that the probable root cause of delanination could be I traced to one or more potential variables during application of System 9. Nonconformances identified did not completely define the extent or location of delamination. Before making repairs, it was essential that the cause of delsmination be established. Laboratory and controlled plant tests were required to provide data 0766f
' 'S Bschtel Associates ProfessionalCorporation MCAR 35 0$3((g gg bbh Revised Final Report Page 11 of 22 for analysis and resolution of this problem. The following studies of application variables were scheduled between May 1980 and August 1981:
- a. Controlled study of the effect of time between coats conducted by KTA-Tator, Inc. ~
- b. In plant testing of solvent additions to first and second coat of NuKlad 117(N). Initial testing included all solvents and thinners used onsite. Final testing included xylol and Amercoat 6 thinner with application scheduled at intervals following mixing of components.
a
- c. Laboratory tests at Consumers Power Laboratories at Jackson, Michigan. These included thinned and unthinned sample preparations to test mixing variables, contamination, and effect of application to wet surfaces. .
- d. Contracted evaluation tests by Inco Laboratories. -
Buffalo, Ncv York. Imco was requested to do sufficient application tests to co= ment on the selection of NuKlad 117(N) as a decontaminable '
surfacer coating.
1 e. Analysis of existing coating at Midland compared with
! prepared samples of known composition (KTA-Tator, Inc.). These tests were run by KTA-Tator, Inc. to verify the type and amount of solvent added to NuKlad 117(N) by J.L. Manta.
- f. Tests conducted by Bechtel in Ann Arbor to determine the influence of relative humidity during the cure of each coat.
3.5 RATIONALE l
The rationale in proceeding with the corrective action investigation was as follows:
3.5.1 All System 9 was nonconforming because of use of unqualified procedures utilizing two coats of f
NuKlad 117(N). Therefore, the two-coat work in the as-built condition was processed through 0766f
. . .. 1
- Bechtel Associates ProfessionalCorporation !
- MCAR 35 093779 "93959 Revised Final Report -
i l- ..
Page 12 of 22 design basis accident (DBA) testing before a decision was made regarding the status of System 9 in areas where delanination did not occur.
!~
3.5.2 A replacement system was required for areas where delaninated coating was removed. Because all -
residues of NuKlad 117(N) could not be removed 1
without damage to concrete surfaces, the replacement system also required requalification
- ! under DBA test conditions.
A
- 3.5.3 Bechtel construction took over the coatings work i and implemented its own quality control program.
j Field coatings engineers were also provided to j monitor and direct the work at the jobsite.
! Consumers Power Company was responsible for the j quality assurance program to provide the required l over-inspections and other necessary functions j vhen the Q-listed work tegan. A specification was j prepared by project engineering incorporating j procedures and controls to prevent recurrence of
- the delsmination problem. The specification
- provided types of controls, such as surface I
preparation, application procedures, points of
- inspection, material storage, and certification of painters. When the specification was issued,
- Bechtel construction was able to perform the i coatings work. Because the root cause was i unknown, all specification requirements were evaluated in non-Q-listed areas where surfacing j was required. Repairs in containment were not t
performed until the cause was determined.
, 3.5.4 Statistical test methods were available to provide i a basis for sampling to established levels of reliability. - Regardless of the root cause.
unacceptable work could be segregated.
i
, 3.6 ACCEPTABILITY OF UNTAII.ED SYSTEM 9 3.6.1 Specification 7220-A-55(Q) was prepared to
- determine the acceptability of those portions of
. System 9 that had not delsminated. This
- specification was designed to provide the following information
i 9
0766f i
,- .m. , ,,, . 4,,e., ,a... ... ___, _..s._m .m ..a - . - .
~
. 6 Bachtel Associatos Profossional Corporation NCAR 35 09377g 093S59 Revised Final Report Page 13 of 22
- a. Reliability of the knife adhecion test
- b. Provide a statistically selected series of core samples suitable for design basis accident (DBA) tests et ORNL to Hidland criteria specified in Specification 7220-A-45.
3.6.2 Specification 7220-A-55(Q) was implemented to provide at least 95% reliability that at least 95%
of the samples collected were representative of the area tested. The following sucmarizes the sampling procedures:
- a. All failed areas as determined Lin preliminary testing and mapping of 100 f t 2grids were excluded f rom tests.
- b. Sample sites were statistically selected from approximately 500 test grids providing 80 sacples.
- c. Each saeple site was tested for delamination by knife on four sides of the immediate site for core sampling. One knife test was run using 1/8-inch spacing between parallel cuts. This accentuated the tendency for delamination.
- d. Special samples were taken in both pass and failed areas and tested by immersion in 180 to 200T spray solution at Consurers Power Laboratory. This test confirmed that blistering, cracking, and peeling can occur during the final immersion stage of simulated DBA testing in accordance with Specification 7220-A-45. Samples that pass the knife test passed the immersion test and samples that fail the knife test failed.
- e. DBA testing was conducted on samples collected under Section 3.6.1.b. Neither frradiated or unirradiated samples failed. A few panels blistered. However, this was -
attributed to the fact that panels were not prepared in a laboratory and represented an
! as-built plant condition.
0766f
> % B;chtel Associatos Professional Corporation MCAR 35 0937 9 93959 Revised Final Report Page 14 of 22 3.7 REPLACEMENT OF SYSTEM 9 IN FAILED AREAS - COATING SYSTEM TESTS 3.7.1 Four principal coating manufacturers that supply prequalified coating system for concrete surfacing were selected as potential suppliers of -
replacement coating in failed areas.
Specification 7220-A-57(Q) was prepared to provide and control sample preparation and collection of potential System 9 replacement specimens. This specification included the following requirements:
- a. Use of delsminated coating removal methods and follow-up cleaning by procedures proposed 2 for plant use
- b. Surface preparation of wall areas in a room in Containment 2 that had exhibited extensive delamination
- c. Selection of saeple sites typical of retained NuKlad 117(N) residue, and ecsked " window" ;
areas to permit recording (photographically) ,
the surface condition under new coatings in the saepled area
- d. Sufficient area of each system applied in one, two, three, or four coats to permit selection of alternative systems that best meet the functional requirements
- e. Application under fully documented conditions with (optional) technical assistance by the coating manufacturer
- f. Test the use of Amercoat 90 as a finish coat over all other coating systems 3.7.2 Sample areas were prepared and sample sites selected after preparation. Core samples were collected for DBA testing in accordance with Specification 7220-A-45. Samples were DBA-tested
, at ORNL and are reported.
0766f
- - Bechtal Associatos ProfcssionalCorporation
- Cn 35 093779 93359 l Revised Tir.a1 Repcrt Page 15 of 22 l
1 .
l 3.7.3 All samples tested were acceptable. Basedet l compatitility with existing Systes 9, a coating l systes was selected that required a sist coat cf
! Amercoat 90 over the cleaned surface, which had a NuKlad 117(N) residae. Folleving this, a surfacer coat cf NcT. lad 117(N) was a;;11ed ad cured -
folleved by a finish ccat of Aserccat 90. This l coating systen was placed in the tectrical l specificatics for reapplicatic= is areas where
! ccating has been removed.
l 3.E FETA 7JLTICE CT TICF51CAI. 5FECIFICATIC5 TCE ATTI.1CATICE CT CICCSTA?.IKA5LE CCATINOS TO CCNCETE 3.8.1 Sreef ficaric: 722C-A-56(0)
This specificatics was issued c: A;ril 14, 1951, l- to replace Specificatice 7220-A-15(Q), previessly used for sutecetracted verk. Iefere implementatic=, draf t prc<edures fer Ceatics Systems 6, 7, E, and 9 in 5;ecificatic:
7720-A-15(Q) were pre;4 red and verified by a;;11eation in ect-Q-listed areas bef ore preparatice cf $;ecificatie: 72TC-A-56(Q). n.i s specificatic: incerpcrated rigid eevire=sental cct.trols to preclofe further prehless.
3.6.2 Fevistees to 5;ecificatic: 722C-A-56(C)
Vtee the roct cause cf Systez 9 fa11ere was determined and a replacetect systes tested, apptc;riate revistens were made to Specificatice 722C-A-56(Q). These acc1cde the follevieg itenst
- a. tecrease relative humidity permitted to 22*.
fer all surfacer systess
- b. Incorporate replacesent Systes 9
- c. Irect; crate testing procedure for lecatics, verlifcation, arJ reseval cf failed ccatieg (this ites is described in detail uMer Corrective Actica) 0766f
~.% Bechtel Associatcs Professional Corporation MCAR M Revised Final Report 093I7S 093959 Page 16 of 22 -
The replacement Systes 9 was applied in accordance
- with the requirements of the revised Specification 7220-A-56(Q).
4.0 ANA1.YSIS OF SAFETY IMP 1.1 CAT 10N 4.1 The total Systes 9 coating that now exists in containment Units I and 2 is comprised of portions from the original subcontractor-applied Systes 9 (old coating)
[ Specification 7220-A-15(Q)], which has been qualified by the testing process, and portions from the new Bechtel-applied System 9 (new coating) [ Specification 7220-A-56(Q)). It is not considered feasible that under accident conditions this existing System 9 coating can degrade performance of the contain=ent sump or sprinkler head system for the following reasons:
4.1.1 The 95/95 testing plan used to qualify the "old Systet 9" (subcontractor-applied) cc,ating results in a 952 confidence that a minimum of 95% of this coating vill be good (i.e., not delsminate).
Using this as the most conservative approach, the '
resulting 5% of the old System 9 coating equals approxicately 2,349 square feet in Unit I and 524 square feet in Unit 2. According to the '
actual test results, 472 grids were tested in Unit I with 5 grids that failed, resulting in a minimum of 97.8% of the old Systes 9 that will be good. In Unit 2 because of the smaller amount of remaining old Systes 9 coating,118 grids were tested with I grid that failed, resulting in a sini=u: of 96% of the old System 9 that will be good. Therefore, the actual resulting percentage of 2.2% in Unit I equals approximately 1,034 square feet, and the 4% in Unit 2 equals 419 square feet of old Systes 9 coating that may possibly delstinate. When the reliability of the current System 9 coating (old plus new) is analyzed, Unit I results in a 95% confidence that 98.4% of the System 9 coating will be good, and in Unit 2, a 952 confidence that 99.4% of the l Systes 9 coating will be good.
4.1.2 As described in the Midland Tinal Safety Analysis l Report, Revision 44, Subsections 6.2.2 and 1.2.2, j and Table 6.2-23, each containment sump is designed in accordance with Regulatory Cuide 1.82 0766f
n- Bechtel Associates Profcssional Corporation 093S59 MCAR 35 093779 Revised Final Report Page 17 of 22 criteria. The sump is designed to maintain a water flow approach velocity to the fine mesh sump screens at less than 0.2 f t/see with the floor level in the immediate vicinity of the sump sloped gradually down, away from the sump. These design ,
t features allow for settlement of debris with a specific gravity greater than 1.05 before reaching the sump screens (refer to Regulatory Guide 1,82, June 1974, Section B).. Each of the System 9 components (Amercoat 90 and NuKlad 117(N)] has a specific gravity greater than 1.4 (based on tests done by Censumere Power Company Laboratory). If an accident condition occurred and lack of i
adhesion was exhibited by the System 9 coatings, the debris would settle during flow to the sump, 1 rather than becoming entrained in the surp screens.
5.0 PROBABLE CAUSE i
5.1 Each of the test programs provided positive indications that the delamination between coats of NuKlad 117(N) could be duplicated by application conditions, which were representative of unusual circumstances that could have occurred during the work.
5.1.1 If the time interval between coats of NuKlad 117(N) was extended beyond 7 days, the frequency of delatination increased. However, the results of this test did not conclusively indicate that this variable in itself accounted for the aeount or location of delatinatica that occurred at the Midland plant.
5.1.2 Reports by KTA-Tator, Inc. (supported by Bechtel tests) indicated that NuKlad 117(N) applied at the end of the useful pot life increased the potential for adhesion loss if warm material was applied to a cooler surface. This reduced the flow property of the applied coat due to a rapid increase in viscosity or solidification at the cold surface.
5.1.3 Solvent addition extended the apparent pot life and in severe cases caused defamination. Some solvents appeared to cause additional exothermic heat when added to the mix, indicating reactivity with the amine convertor.
0766f 49 4 s e
- + B chtel Associates ProfossionalCorporation NCAR 35 093779 693959 Revised Final Report Page 18 of 22 5.1.4 Both KTA-Tator, Inc. and Inco Laboratories concluded that delaufnation increased at high levels of relative humidity. This observation was '
made independently by both laboratories performing application tests. This occurred with and without solvent addition. This work provided important
- evidence needed to explain the delsmination in the plant.
5.1.5 When the results from section 5.1.4 were reported.
Bechtel performed tests to determine the percentage of relative humidity that was detrimental. The results of this testing were as l followst
- a. Without solvent (Amercoat 6), relative humidity above 80% was detrimental to !
adhesion between two coats of NuKlad 117(N).
- b. With the use of Amercoat 6 thinner, delamination does not occur until most I solvent has evaporated. This requires 6 to 8 weeks and did occur at relative humidity t below 80%.
5.1.6 It was concluded that the major root cause of delsmination was due to high levels of relative j humidity during the cure of the first coat of NuKlad 117(N). Cure was retarded and delamination was not readily apparent because of the unauthorized addition of Amercoat 6 thinner to a
NuKlad 117(N). This solvent addition also extends the apparent pot life and, because of hygroscopic properties, makes the uncured coating more susceptible to high relative humidity. Variations i in film thickness accounted for differences in !
solvent retention. This could have accounted for the failure to observe this problem while l application was under way. It is estimated that s !
total loss of adhesion between coats may require up to 6 months on thinned Nuklad 117(N). l 5.2 The sequence of events typical of the installation 'of System 9 is listed below to provide logical explanation for the location of delasimated NuKlad 117(N).
l 0766f
.. c.
. .\ - -'
'. - -- B;chtcl Associates Profossional Corporation
^
MCAR 35 093779 093959 Revloed Final Report Fase 19 of 22 5.2.1 surfacing of all concrete was scheduled during 1973 and 1979 To avoid interference with other
- crafts, work was normally started at 4:00 a.m. and proceeded until noon. Areas cleaned on the l
previous day were coated early on the following
, day. During most of this time interval, the
- l temporary construction openings (doors) were l closed when coating work was in progress. At l 7:00 a.m., when other craf ts started work, the construction opening was in use and the j temperature and relative humidity of the air gradually changed to reflect outside ambient
- conditions. Coating work was then stopped in cold weather due to restrictions spectifed for surface t empera t ure . Air flow during the cure of freshly applied surfacer was free lower elevations where ,
the humidity was greatest to the upper levels through the shield wall area.
i 5.2.2 Units I and 2 were scheduled for coating work based on availability of the uncoated areas. -
However, Unit I had less construction work in progress and environmental conditions were reported to be more stable. This accounts for the fact that delstination was not as extensive in Unit 1. The construction opening on Unit I was also shielded to some extent by other buildings whereas Unit 2 construction opening faced the river.
l 5.2.3 The probsble cause of delatination has been traced
{ to high relative humidity. The separation of r
coats of NuKlad 117(N) did not occur until the I
first coat had cured. Complete cure was delayed for months due to unspeciffei thinning of NuKlad 117(N) with Amercoat 6 thinner. - Righ relative humidity causes the water sensitive amine convertor to migrate to the coating surface. This is further enhanced by water soluble solvent
- residues of the type that can reset with free
, amine groups preventing complete cure within the limits predicted for maximum hardness. At the surface of the coating, any amine exudate can also i
react with carbon dioxide forming complex incompatible products on the surface. Delaying ;
I T^ l l 0766f u ;
.v
- . __ . , r -
l '
t
\
I' -
Becht::l Associates Profcssional Corporation ncAn as 093779 093959 l
Revised Final Report !
Page 20 of 22 i
the time between coats and prolonged pot life may -
have influenced delanination, but could not account for the extensive failures observed.
5.2.4 Although the second coat of NuKlad 117(N) may have i been thinned in the same manner as the first coat, -
f it was not necessary to apply as heavy a film. :
The second coat was used to completely fill (
surfacer pinholes, and there were no heavy i deposits required to fill the concrete depressions i and undercut areas. The thinner film was able to i cure more quickly and release the solvent through l Amercoat 90 finish coats. '
6.0 CORRECTIVE ACTION l L
6.1 Nonconforming System 9 included all system 9 applied by subcontract in containment buildings 1 and 2. Therefore. .
all System 9 was subject to removal or testing to resolve the nonconformance. Previous tests of 10' x 10' grids ;
provided information on the location of failed areas, but !
did not provide the failure boundaries. When delstinated Systet 9 was removed by impact tools, such as the Roto l Peen, it was possible to segregate sound coating by observing the difficulty of removal. l Starting at known
- failed test areas, coating was removed to the point of intersection with sound coating. Field engineering l determined the total area of coating removal based on further knife tests and consolidation of removed coating !
into logical areas for reapp11 cation of System 9, i 6.2 Specification 7220-A-56(Q), Section 12.0, defines the ~
removal and reapplication of systes 9. sections 13.7 and '
13.8 provide inspection and testing criteria for selection f of randos sample sites by a scientific sampling plan. The t laplementation of these sections of Specification 7220-A-56(Q) require removal and replacement of system 9 or requalification of Systes 9 installed under subcontract I and allowed to remain. The results of this program are tabulated in Appendix A. The procedure for implementation i
. was as follows: r l 6.2.1 Drawings A-1000(Q) through A-1015(Q) were issued l to subdivide all coated concrete walls in i j containment buildings I and 2. Each containment * ;
i a
0766f
"" Bechtel Associatcs Professional Corporation 1
HCAR 35 Revised Final Report 093779 093959 Page 21 of 22 area represented by a drawing was a unit for statistical testing.
6.2.2 Each drawing was further marked off in 2' x 2' grids for use as special sample sites.
6.2.3 Bechtel project engineering performed a walkdown of each area before issuing the drawing to construction. All removed coating and areas inaccessible to testing were identified on the drawing. Other coating systems applied in conformance to the subcontract Specification 7220-A-15(Q) were marked. The resulting 2' x 2' grids constituted the population of sample sites for statistical sampling of System 9.
6.2.4 Bechtel reliability engineering generated random grid numbers for performance of knife test adhesion. The criterion used provided 95%
confidence that a minimum of 95% of the unremoved coatings in each t.ontainment could not delaminate. Crids were designated A, B, or C.
- a. Testing of 59 "A" grids permitted no failure
- b. Testing of 59 "A" grids and 34 "B" grids allowed one adhesion failure
- c. Testing of 59 "A" grids, 34 "B" grids, and 31 "C" grids allowed two adhecion failures 6.2.5 Teams consisting of a Bechtel field engineer and a Bechtel quality control engineer performed adhesion tests under a quality control inspection plan. Based on test performance, results were documented and reported as " pass" or " fail." If the area failed, additional coating removal was permitted. A revised drawing was issued and ratested by repeating Steps 3 and 4 above. If in the judguent of field engineering the area represented by the drawing did not justify retest, all coating was removed. .
0766f
Bechtel Associates. Professional Corporation NCAR 35 093779 T93959 Revised Final Report ,
Page 22 of 22 6.2.6 Af ter July 1,1982, testing was limited to "A" grids only. This eliminated the need to locate the boundaries of failure represented by permitted failures of "B" and "C" grids.
6.2.7 All corrective action was completed by September 15, 1982. This includes the reapplication of System 9 to areas where delaminated coating was removed. Final touchup of areas where rework is ,
being performed will be completed after other construction activities are complete.
7.0 REPORTABILITY Eased on the safety implication analysis of this report, the described deficiency was considered reportable in accordance with the Code of Tederal Regulation,10 CFR 50.55(e).
Subritted by: h .0. k.tA %
M.A. Ilughes Architectura Croup Supervisor Approved by -
-L E.M. Hughes Ann Arbor Project Engineer Concurrence by: M
.J. K s( f' Ar hite tural Chief Concurrence by: iN E.H. Smith Engineering Manager Concurrenceby:[M.A.Dietrich
, Project Quality Assurance Engineer 0766f
. . MCAR 35
- Final Report
- Y *4% Revision 2 Appendix A r
CORRECTgVE f 9 /7U ACTION-SYSTEM 9 TEST AND REPLACEMENT i OOorn
,suos a
- E E Total Coating Total Coating g I]g E T Removed Remaining RMs .
8 i{ z OP- 4 su 5 (Sq. F t.) (Sq. Ft. )
685' A 1000(O) 1 A8 1 to 2 718 5,990 659' 685' A 1001(O) 4 A8 1 to 1 1,755 4,953 .
659' 659' A 1002(O) 2 A 2 to 2 1,683 4,479 640' A 1003(O) 2 ABC 2 to 1 283 5,879 640' 685*
A 1004(O) 3 - 3 to 2 12.434 0 All Coating Removed 640' A 1005(O) 3 A.D 3 go 1 1,728 10,706 640' 640' A 1006(O) 6 -
4 to 2 4,130 0 All Coating Removed 626' 640' A 1007(O) 2 AB 4 to 1 1,149 2,981 620' 626' A 1008(O) 10 - 5 to 2 12,050 0 All Coating Removed 593'6" 626' A 1009(O) 1 A 5 to 1 1,534 10,516 593'4 "
640' A 1010(O) 1 6 to 2 10.958 0 All Coating Removed 593'4 "
640' A 1011(O) 0 A 6 to 1 712 10,246 593'4 "
A 1012(O) 0 -
7 Misc 2 6,824 0 Att Coating Removed A 1013(O) 1 A 7 Misc 1 5,126 1,698 A 1014(O) 5 - 78 Misc 2 7.131 0 All Coating Removed A 1015(O) 5 -
78 Misc 1 7.131 0 All Coating Removed Total: For Unit No 1 19,418 46,979 Total: For Unit No 2 55,928 10.469 Totat For Units 1 & 2 75,346 57,448 l Total Area Of Concrete (System 9) per Unit = 86,397
(-2SM
9 Glu _Lmed au3 MW !
Ap(l%-W(p)~ M Tc Nx6co.
so-w;n-,o; so. nope-so
& 4e n. W Pm so.mei A 4, Odis \M +~~~M M @g w&uag m b % -.u dV % a s' y %q r- m -
%.W-tLJn Ao hw ILQ a,nTtt '-
n.9 T
4maiv% d++0.Q' -ftc
<dhb,Mkb Lend .L .
4k Mh eck lu w lv SM
- 4 \ n w ao pwn1 up a >W w ad 4+ Gu g L g :c QQ' avu B d m % s. w E .e , v>e W w h i d ~ L k Mb., m ~ L W Mc.
GMq m % te m P. m .a L L.:-< b + G w ~.a w x .a A l ae a .
6/1