ML20138E639

Investigation of Possible Deterioration of Porous Concrete Millstone 3 Nuclear Reactor

ML20138E639
Person / Time
Site:	Millstone
Issue date:	04/11/1997
From:	Burg R, Klemm W, Ost B CONSTRUCTION TECHNOLOGY LABORATORIES, INC.
To:
Shared Package
ML20138E631	List: B16403, Forwards Root Cause Analysis & Residual Strength Assessment, Porous Concrete Matls Properties Evaluation & Containment Structure Seismic Assessment in Response to Rai, ML20138E639 ML20138E647
References
CON-050943, CON-50943 NUDOCS 9705050074
Download: ML20138E639 (67)

Text

!

REPORTIO:

Northeast Utilities System P. O. Box 270 Hartford, CT 06141 Investigation of Possible Deterioration of Porous Concrete -

Millstone No. 3 Nuclear Reactor by Borje W. Ost, Ronald G. Burg, Waldemar A. Klemm, and F. MacGregor Miller 11 April 1997 CTL Project No. 050943 5420 Old Orchard Road, Skokie, Illinois 60077-1030 847/ 965-7500 800/ 522 2CTL Fax: 847/ 965-6541 p

(( $O0Ck Oh000k23 PDR t

,. - . _ . . - - --_ _.- - - .-- - .- .- -.~.- - -

t ,

i EXECUTIVE

SUMMARY

i i ,

Introduction ,

i Nonheast Utilities operates a nuclear power generation station at Waterford, Connecticut.

Millstone Unit #3 of this power station was placed into service in the 1980's, with several concrete 4

l elements playing a pan in the structural and water drainage functions of the plant. Among the l I materials used in the construction are a porous portland cement-based concrete and a porous i i Lumnite@ brand high-alumina cement concrete. Construction Technology Laboratories has been l l asked to investigate reasons for the formation of deposits in the water drainage system of the ESF

! sumps at Nonheast Utilities' Millstone #3 Nuclear Power Station. In addition, CTL has been l asked to determine a " root-cause mechanism" for deterioration of the sub-base concrete, as i l

' reflected in the degradation of " mock-up" specimens created to simulate the in-place concrete, and  !

to provide an estimate of residual strength in the concrete, particularly in the porous concrete l elements.

i

, CTL's concem is primarily with the integrity of the porous high-alumina cement cone:ete. The evidence available is idrential, oased on the petmgraphic and chemical examination of mock-up  :

sainples from Phar,es I through Ill, residues from the filters on mock-up samples, site water, and l five samples of actual residues from the Millstone #3 ESF sumps and weirs.

Based on the results of CTL tests and observations, the root cause mechanism for degradation is multifold. The combination of the two cements (ponland cement and high-alumina cement)in the substructure " sandwich" proved to be unfonunate indeed. The portland cement provides a source of leachable ions which, in conjunction with other sources (ground water, soil and bedrock) of the same and other ions, attack the high-alumina cement matrix. At least four steps in the attack likely are occur 6ng:

1. Leaching of ions by water percolating through surface fill, sandy lenses in the glacial till or fissures /foliations in the bedrock. Significant amounts of calcium, sulfate, sodium, potassium, and chloride are dissolved in the highly aggressive rain waters. Analysis of the water in well borings located outside the plant indicates the waters are mostly very aggressive to ponland cement matrices, as indicated by three separate corrosion indices calculated for the l waters. No such indices are available for the Lumnite@ cement. The assumption that the water emanates mainly from outside the structure seems reasonable, since the amount of flow in general is proponional to the amount of annual rainfall, and plant experience has been that the flow generally increases 3 to 4 days after a heavy rain. The actual flow path has not been determined, but it is likely that the flow is in a southerly-to-southwesterly direction. CTL has also suggested 1

i g

}

4

! that at least some of the water could be highly aggressive condensation water forming at the dew 4

j point line. 'Ihe possibility of penetration of ocean water to the structure cannot be completely eliminated, since one of the downstream wells drilled into the bedrock near the stmeture had an

{

_ elevated chloride level.

~ 2. Leaching of lime and alkali from the portland cement. The acidic waters or water j low in dissolved solids are capable of dissolving additional amounts oflime (calcium hydroxide)

! and alkali (sodium and potassium) from the portland cement (probably chiefly from the porous i l portland cement matrix). The first compound to dissolve when flowing water has access to

! concrete elements is usually calcium hydroxide, owing to its higher water solubility. The resulting i

waters after contact with the portland cement will have a high pH and be aggressive to Lumnite@

j cement.

3. Conversion and subsequent carbonation of the hydrated Lumnite@ cement.

This reaction takes place only in the Lumnite@ cement materials. Conversion is defined as the changing oflow density initial hydration products (hexagonal crystal habit) to a very dense more thermodynamically stable cubic crystr.1 form (hydrogarnet). This reaction is inevitable in calcium aluminate cement concrete. It is quite slow at below room temperatures and with low water-cement ratio pastes, but is accelerated by moisture and by higher temperatures. It results in a densification of the cementitious matrix, and significant ecncomitant increases in the porosity of the cement paste; 50% porosity (solids basis) can be generated. This results in a degradation of the strength.

However, by itself, conversion probably would not have created the degree of deterioration found in the porous calcium aluminate cement as evidenced by the appearance of the residues. Carbon dioxide, either from the atmosphere, or dissolved in water, has reacted with the calcium aluminate

.{

hydrates (both the low density hexagonal hydrates and the dense hydrogarnet) to generate )

aluminum hydroxide and calcium carbonate. These reaction products are not cementitious, but I they probably can coat the surface of the cement paste and give it some protection against further attack, since they are relatively insoluble in water. The cr* m carbonate may act as a filler and produce a minor strength recovery.

4. High pH dissolution of calcium aluminate and aluminum hydroxide in the Lumnite@ concrete The protective aluminum hydroxide is highly soluble in the high pH I

waters. CTL did not definitely establish that the so called " alkali regeneration cycle" reported in the literature is operating in the system. The conditions are, however, favorable for such an attack mode.

Another attack mode in the Lumnite@ concrete (ettringite formation), not well-documented for high-alumina cement in the literature, is likely since Lumnite@, although designed for sulfate 2

l 1

msistance, is prone to sulfate attack panicularly in the pmsence oflarge amounts oflime. Ettdngite fomtation in a high lime and high pH environment can be expansive and may aggravate the amick on the Lumnite@ porous concrete.

1 There is no indication that cement residue is collecting in the weirs or sumps. Thus the reaction l

seems to occur mainly in the water or on the surface of the hydrated cements. This is consistent  ;

with a congruent dissolution of the calcium aluminate or aluminum hydroxide. Neither of the two

{

hexagonal hydrates nor hydrogamet were present in residue samples. Conventional conversion calculations wem not possible except by infemnce.

1 Further Discussion )

CTL identified by X-ray diffraction and thermal analysis the following compounds, in each of the weirs and the sumps: ettringite, Fdedel's salt, thaumasite, calcium carbonate, and a carbonated I

, monosulfate hydrate type compound. The compounds are consistent with the ionic species present in the system. The solubility products of some of these compounds (especially ettdngite) are quite  !

low and the tractions go to near completion without a significant thmshold level for the reactants. It seems that the amount of sulfate available to the system is the controlling factor. Excess lime and I l

. alumina remain in the exit waters. l The Millstone plant does not at the pmsent time have a definitive direct method to determine flow 1

rates through the porous concrete. Based on the estimated ranges for the flow rates CTL has l estimated that approximately 12,000 to 60,0T)Ib of residue is forming each decade. Of this only j about 1,000 lb. have been collected in the weirs and sumps. There is also an additional relatively

large amount of soluble matedal escaping with the exit waters. CTL has provided a method to calculate these amounts when average flow rates and associated concentrations are better defined.

it seems reasonable that due to the low flow rate of water with respect to the volume of available space in the porous concrete the conditions inside the concrete are quite homogeneous. The residence time is long and a near equilibrium is achieved. Due to the relatively greater flow in and near the porous drainage pipe, the conditions there may not be as uniform.

It is CTL's opinion that the loss of calcium from the hydrated portland cement has not seriously impaired the strength of the ponland cement porous concrete. A large proponion of the calcium could be supplied by the ground water and a significant amount could come from the degraded Lumnite@ cement.

It is CTL's opinion that the ten-foot thick massive concrete base mat is not at risk of chemical attack to any significant extent.

3

NUSCO's mock-up tests developed usable qualitative data and was instrumentalin understanding  !

the compLr . .: actions in the "as-built" structure, but encountered common difficulties in accelerated simulation of a complex cementitious system. Aggregate-cement ratios wem not close to those in the "as-built" structure. This was particularly true for Phase I and II. Also between Phases I and II there was a change in the calcic plagioclase rock aggregate (similar to what was i used in the "as-built" structure) to a potassic feldspar. Analyses of entering and exiting waters were incomplete. The internal temperature history of the concretes was not documented. Cudng was i not similar to the "as-built" condition, except perhaps in Phase II. The mock-up constmetion 1 practice of placing each concrete in 3 to 5 in. layers tended to produce horizontal cold joints, a condition not typical of the "as built" monolithic condition. The thickness of the Lumnite@ seal grout on top of the Lumnite@ porous concrete was 2-in. vs.1/8 to 1/2-in. in the "as-built" structure.

The absence of any residue formation in Phase III and the indication of no dissolved ions in the i exit waters in Phase III (due to better curing) makes it particularly hard to translate the best available strength data to the "as-built" constmetion where residue formation has been prevalent. l l

Taken in total, these difficulties make a quantitative translation of mock-up strength data to the i Millstone situation tenuous indeed. However, semi-quantitative data were developed. l Statistical analysis of the compressive strength data from the mock-up study, Phase III, was performed and a linear model developed. In the mock-up study a trend of decreasing compressive strength with exposure to additional water flow and time was obtained. Although applying these test data and the resulting model to the structure in the field is problematic for several reasons, a rough first-order estimate of potential loss in compressive strength in the structure of 380 psi over the next five years is suggested. It models only the effects of conversion and partial carbonation, and does not take the dissolution mechanism into account. Due to the uncertainty of these data, extrapolation to longer time periods is not warranted. Furthermore, because there are no reliable l datu for the current in-place strength of the calcium aluminate cement porous concrete, it is difficult to predict future absolute in-place strength of the porous concrete in the containment mat structure.

Again, as a rough first-order estimate, it may appropriate to estimate the in-place strength in the structure using the lowest measured compassive strength in the mock-up study.

Recommendations Given the uncertainty in these predictions, it is critical that cores be extracted and tested from the actual structure. This can establish a reliable value for current in-place compressive strength and should provide samples that can be examined by various destructive and non-destructive methods to better understand any deterioration mechanism and its basis.

4

4 CIL recommends that NUSCO proceed with the planned coring of the actual concrete at Millstone i i

#3,ifit can be done safely. The corirsm operation should be carefully planned and executed to

ensure that statistically valid samples am cbtained. Documentation on video and immediate j protection of the 5,amples befom exposure to carbon dioxide in air are of prime imponance. j i

Revised April 17,1997 a

5 k I i 4 i i l

• 1 l

l 1 i i

, i i

a d

e i

i i )

i 1

l i

I

-i l

J l

l l

l I

5 l

i TABLE OF CONTENTS Paga INTRO D UCTI ON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 SCOPEOFWORK.......................................................................... 1 POROUS CONCRETES AT MILLSTONE UNIT 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Mill Aone Summary of Porous Concretes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 ,

CTL S u mmary of Additional Concrete Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 i CONDITIONS IN PIPES , WEIRS AND S UMPS .. . . ... .. . . .. ...... . . . . .. . . . . . . .. . . . . 4 PREVIOUSLY REPORTED POROUS CONCRETE PROBLEM........................ 5 PROPERTIES OF THE CEMENTS USED AT MILLSTONE . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

. M ill a n d L ab o r a t o ry R e p o rt s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . 6 I Po rtl an d Ce me n t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1 G e ne ral In form ati on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Portlan d Cement Hydn ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 High-Alumina Cement (HAC).......................................................... 1I Ge n e ral I n form ati on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 High-Alumina Cement Hydration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 A GG RES S I VE WATERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 i PortIand Cement in Aggressive Waters.. ............................................... 14 High Alumina Cement in Aggressive Waters................ ........ ................ 17 ANAL.YSIS OF SUMP AND WEIR RESIDUES, AND LIQUIDS ...................... I8 199 6 an d Ol d er S a m pl e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I8 1997 Samples............................................................................ 20 AN ALYS I S OF WE LL WATE RS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 i ROCK CORE S AMPLES FROM WELL B ORINGS .. . . . . . . . . . . . .. .. . .. .. .. . .. . . . . . .. 25 AMOUNT OF RESIDUE IN POROUS CONCRETE AND M ATERIAL LOST IN EXIT WATERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 M O C K-U P TESTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Mock.Up Test Difficulties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 An alysis of Mock-Up S amples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Ex perimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Residues from Fihers and UnmLMolds............. ......................... .32 Conclusion from XRD and DSC Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 S u mm ary of Petrographic Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Mock-Up Concrete Strength and Predic: ion of Long-Term Strength ......... 36 FINDINGS.................................................................................... 42 R EFE REN C ES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 TABLES.........................................................................................

FIOURES.....................................................................................

i

APPENDlX A. XRD, MOCK-UP CORES B. XRD, MOCK-UP FILTERS AND RESIDUE C. XRD, ROCK CORES AND GLACIAL TILL D. XRD, POROUS PIPE E. XRD, REVIEW OF ABB-CE XRD PATTERNS F. XRD, SOLIDS FROM GROUND WATERS

! G. PETROGRAPHY: MOCK-UP CORES, FILTERS AND RESIDUE; ROCK CORES H. MICROANALYSIS, POROUS PIPE

! I. MICROSCOPIC EXAM OF WEIR AND SUMP RESIDUE J. XRF, MOCK-UP FILTERS AND RESIDUE K. ANALYSIS OF MILLSTONE GROUND WATERS L. TGA AND DSC, MOCK-UP CORES M. TGA AND DSC, MOCK-UP FILTERS AND RESIDUE N. TGA AND DSC, MILLSTONE ROCK CORE O. TGA AND DSC, WEIR AND SUMP SAMPLES P. CTL SPECIFIED METHOD FOR ANALYSIS OF GROUND AND SUMP WATERS, AND SUMP SOLIDS Q. XRF, PARTICLE SIZE, TGA OF LUMNITE FOR PHASE II AND III R. MILL AND LABORATORY REPORTS FOR CEMENTS FOR MILLSTONE AND MOCK-UP TESTS S. ABB-CE REPORTS FOR 1996 4.ND 1997 l

T. GEI REPORTS MILLSTONE WEL;LS: LOCATION, DEPTH, AND INITIAL l WATER TESTS l U. BIBLIOGRAPHY, HAC, AND POROUS CONCRETE V. CALCULATION OF AGGRESSIVITY INDEX W. ESTIMATED SUMP INFLOWRATES X. CALCULATION OF RESIDUE FORMATION ii

i j

j INTRODUCTION j

In early July 1996, Mr. Waldemar A. Klemm, Senior Principal Scientist and Group 3_

Manager, Materials Research & Consulting (MRC) Group, and Mr. Ronald G. Burg, ,

! - Principal Engineer & Group Manager, Materials Technology Group, of Constmetion Technology Laboratories, Inc (CTL) were contacted by Mr. K. Lakshmipathiah, Principal

! Engineer at Northeast Utilities Service Co. (NUSCO) regarding a residue deposition i problem in the porous concrete drainage system undemeath the Millstone Unit 3 nuclear i power plant. Borje Ost, Senior Scientist in CTL's MRC group, visited the plant on July 11 and 12,1996 and was briefed on the problem by Mr. Lakshmipathiah. CTL started a

work on the project in July with Borje Ost as principal investigator. Tne desired work

! was further expanded in meetings between NUSCO and CTL personnel on October 15 and 16,1996 and January 20 and 21,1997.

l SCOPE OF WORK l Purchase orders described the scope of the work as follows:

1. Conduct chemical analysis of water infiltration into the containment mat.

!- 2. Identify potential source of water.

l

! 3. Conduct an independent chemical analysis verification of white residue i

j collected from the ESF sumps and also the samples collected from the mock-up test. Evaluate and determine possible types of cements used in the

]

l construction of porous concrete undemeath the mat.

i

! 4. Review all documents related to the geological site conditions and correlate the

! surrounding substrate material's chemical characteristics with the residue i

collected at sumps.

5. Determine susceptibility of water to react with cements used in the porous concrete.

l 6. Review historical data relating to the strength of calcium aluminate cement with i respect to time / wet attacks.

l

7. Assess existing strength and stmetural integrity to carry design basis loads.

i

8. Participate with Nuclear Regulatory Ccmmission (NRC), media and other I

panel of experts in exchanging technical ideas to resolve the long-term strength effects.

I 4

i

^

\

l l

The actual work was delineated in CTL's proposal dated November 4,1997 and approved by NUSCO in a January 31 purchase order.

POROUS CONCRETES AT MILLSTONE UNIT 3 Millstone Summary of Porous Concretes In an October 14,1994 letter to the Nuclear Regulatory Commission (NRC), J. F. Opeka, l Executive Vice President, NUEC summarized the foundation substmeture construction as l

follows:

"The Millstone Unit No. 3 Containment Foundation Substructure consists of two layers of porous concrete placed between the bedrock and the containment foundation. During construction, a two-layer rubber membrane was installed between the porous concrete layers to act as a barrier and prevent water from and around the construction area from reaching the foundation structure. This allowed surface water due to rain, etc. above the membrane to flow through 6 (six) inch diameter perforated pipes embedded in the top l layer into Engineered Safety Features (ESF) sumps 7A & 7B, while the bottom layer I facilitated collection of any ground-water for de-watering purposes. It was expected that after construction and during plant operation, the ESF pumps would be dry and would not collect any ground water from the top layer of the porous concrete. Additionally,it was also expected that the bottom layer would retain water from the surrounding area, and this l water would be prevented from entering the top layer of the porous concrete by the i continuous waterproof membrane.

To preclude either surface or subsurface water from infiltrating the substructure, all construction joints were sealed with water stops below grade level. In addition all other structures surrounding the containment were designed with drainage systems locate <1 under their foundations to collect surface water, primarily due to absorption of local precipitation. These drainage systems discharge into sumps at designated locations.

Therefore, the combined effect of waterstops, presence of waterproof membrane and the drainage system surrounding the containment should preclude water from infiltrating into the top porous concrete layer directly below the contamment foundation.

, Evidence of water-in-leakage into the ESF sumps from the Containment foundation substructure indicates that either the water stops and/or the water-proof membrane of the substructure have been damaged either during or subsequent to construction. The presence of white cement residue emanating from the drains was initiallyjudged to be from the Calcium Aluminate cement concrete and most likely due to washing of cement laitance from the concrete layer. ..."

UTL' Summary ol AdditionalConcre'te'Information Figure 1 details the porous concretes. The membrane was placed over the 10-inch porous portland cement concrete after a smoothing portland cement (PC) grout had been installed.

A two-inch thickness of PC grout was then installed followed by 9-in. of Lumnite@

(calcium aluminate cement) porous concrete, which was then topped with 1/8 to 1/2 in, of Lumnite@ seal grout. This grout then formed the base for the pouring of the dense portland cement stmetural concrete for the 10-ft thick containment mat.

2

The use of calcium aluminate cement in the top porous concrete at Millstone Unit 3 was approved by U.S. Atomic Energy Commission (Docket 50-423,1974).

Portland cement porous drainage pipe (see Appendices D and H), nominally 4-in. I.D.

and 6-in. O.D., embedded in the porous Lumnite@ concrete was used for draining the water from the containment structure into sumps 7A and 7B. Figure 2 shows the layout of the pipe. Nine hundred lineal feet of pipe was purchased for the installation. X-ray diffraction and microanalysis reports on the pipe are included in Appendices D and H.

The majority of the porous concrete placement started at the end of Febmary and l

continued to the end of March 1975. Minor work continued until mid May. A review of the available construction records by Stone and Webster indicates about 2,198 cubic yards of porous PC concrete were used and 556 cubic yards of porous Lumnite@ concrete were used. The cement content was about 655 lb/cu yd for the portland cement and about 645 for the Lumnite@ cement (reference S & W faxes dated April 3,1997). The large volume 1 of PC concrete was used for preliminary levelling of the blasted rock bottom of the construction area.

i j Table I shows approximate typical Millstone aggregate / cement ratios and water / cement ratios used. The original design called for a 6 to 1 aggregate-cement ratio (by weight), but the actual concretes used had a ratio of 4 to I based on CTL's review of field Q.C.

records. The water /ccment ratio for the PC concrete was raised, but it was kept low for

, the Lumnite@ concrete. Thus, the mix used in the structure was richer and the cement coating of the aggregate was thicker than the original design mix. Additionally, there were some unit weights measured in the field significantly higher (paste run off?) or lower (incomplete mixing or consolidation?) than the average values. The unit weights are highly variable.

Indications are concrete was loosely screeded into place with some tamping of the surface.

The available records state only that curing was by " water spray or ponding. No records were kept of the internal temperature of the placed concretes, either initially or when the massive 10 ft overlay concrete was placed. Air temperatures were presumably typical for the time of year (March) at Waterford, Connecticut. Fairly detailed records of air temperatures and also initial and curing concrete temperatures, are given in a summary table prepared by Stone & Webster and dated February 28,1997.

The target strengths for the porous concretes were 1000 psi. The records indicate a few of the mixes (in general these were cast at low ambient and concrete temperature) did not 3

b i

i l

l achieve the target strength, but still were probably well abose the necessary strength for l the intended load.

1 k'

The waterproof membrane,in addition to being part of the porous concrete substructure, i b

was wrapped around the whole structure, starting some 60 ft below ground and carried j up to near ground ievel. The records do indicate that there were four holes in the installed l membrane which required patching. There were also several problems with splicing of j the membrane. One of these splices was 15 ft long. Another problem splice was caused i by water buildup in the South sump area of the ESF building. The water buildup caused i bulging and rupture of the splice. CTL is unaware of any tests to ensure that repairs to the above prcblems were efficacious.

CONDITIONS IN PIPES, WElRS, AND SUMPS i During the first visit to the Millstone plant, a video was shown to Borje Ost illustrating the 1

l i operation of the drains and the condition of the pipe. Weirs had by this time been installed '

I

. to prevent back-flow from the sump area into the pipes (see Fig. 3 and 4). The available i light for the camera was not strong enough to illuminate long distances into the drainage l pipe, Furthermore, the camera used a fixed focus, further hampering efforts to see the

{ condition of the pipes. It appeared that there was not a gross amount of residue collecting

{ in the pipes; rather the deposit seemed to have formed in nodular manner presumably l around some debris or aggregate particles on the bottom of the pipe. Solids had been

! collected both as deposits in the weirs and the sumps. Approximately 1000 lb have been collected from 1987 through 1996 (reference Frank Matovic memo of March 26,1997).

CTL concluded in July 1996 that there existed a good possibility that, if reliable analysis of the weir and sump sediments could be combined with analysis of the exit waters (with estimated flow rates) and entrance waters, a reasonably good estimate of how much of the two cements was dissolving could be obtained. Using best available estimates for the liquid flows, a mass balance could be set up with the available information assuming no significant sedimentation occurred in the drainage pipes. Even if the exact ratio of the attacks on the cements could not be established, boundary values could be calculated.

When routine cleaning of the sumps and weirs was done in January 1997, a video camera was passed through the pipes. At this time the video showed almost complete blockage of the pipes. A " roto-rootering" type clearing of the pipes was attempted, but another subsequent video showed complete blockage of at least one of the pipes.

4

The evidence for heavy deposition and even blockage of the pipes significantly impacted the possibility of establishing a materials balance for the materials. CTL's assumption that the deposition of residue occurred only in the weirs and sumps was no longer valid. j The analysis of the residues would, therefore, provide only partial insight as to what had occurred in the substructure. The analysis of exit waters could still be done reliably. A )

direct access to the water entering the structure some 60 ft below ground level, or at some l point higher in the structure was not possible. To date the exact source of the water l entering the stmeture has not been defined. There is good reason to believe that the major source of the water is from the outside, but CTL has suggested that at least some of the I water may be condensation inside the mbber membrane. The per minute flow of water is I quite low and the surface area of the structure is large (160-ft circumference and some

60-ft tall). Such condensing water, with a low ionic strength, would be quite corrosive to l

any portland cement matrix it encounters. Condensation could be higher on rainy days I l

since the relative humidity of the air would be higher.

PREVIOUSLY REPORTED POROUS CONCRETE PROBLEM Unfortunately, CTL was unable to locate any literature relevant to a porous Lumnite concrete stmeture. However, Alexander (1993) and Alexander et al(1994) reported on a j residue problem involving a subgrade porous concrete slab somewhat similar to the one at j Millstone. Only one layer of porous concrete was used. The cement was portland l cement. l The slab under a shopping center building was 23 x 35 meters in area and 6-in. thick.

Drainage was provid:d by a 4-in.-diameter agricultural-type perforated drainage pipe emptying into a sump. The annual rainfall in the region is about 40 in. The location is coastal and the soil is sandy and tree-draining. The range for the flow of water was 1 to 10 liters or more per minute. A white residue formed in the pipes sometimes blocking the drainage system and also coated impellers in pumps to the point of seizure.

Alexander obtained samples of the inflow water near the edge of the slab and also a sample of the er.it water. The groundwater sample had a pH of 7.3 ar.d the exit water 12.5. He calculated the Leaching Corrosion Index (LCI) developed by Basson (1989),

Basscn and Addis (1992), and Alexander et al (1992). The LCI was 650 indicating the entrance water was mildly to fairly aggressive. Also an older index, the Langelier Index, indicated a leaching water. The exit water was non-corrosive and " scaling." Alexander attributed the residue formation to leaching of lime from the portland cement followed by carbonation and deposition of calcium carbonate.

5

By maldng some simplifying assumptions, Alexander calculated from the increase in lime as the water passed through the structure that the readily available lime would dissolve in 2.6 years. It would take 11 years for dissolution of all the lime from the portland cement.

Others have indicated that the disappearance of even 50% of the lime produces a very

, weak structure.

One of Alexander's conclusions was that "no-fines concrete is probably unsuitable for use in subfloor drainage elements where leaching groundwaters exist."

PROPERTIES OF CEMENT USED AT MILLSTONE

. Mill and Laboratory Reports Mill or laboratory reports for the cements used at Millstone and also in later mock-up tests are in Appendices Q and R. Table 2 shows for reference CaO/Al 23O percent and molar ratios. Some of the reports give a combined alurnina plus titania value. This is a carryover from older analytical techniques that did not distinguish between aluminum and titanium. In any case, titania is generally considered hydraulically inactive since it is tied up in calcium titanate, a non reactive compound. The actual alumina contents must, therefore, be estimated from typical values for titania from other reports.

Portland Cement 1

General information

Portland cement, as manufactured and sold either in bulk or in 94-lb sacks, is a grey powder ground to a high fineness (usually more than 95% passing through a 325-mesh or 45pm sieve) and capable of hardening into an extremely strong mass when mixed with water When cement is mixed with water it is called paste, with water and sand it is called mortar, and with water, sand, and coarser aggregates it is called concrete. Cement is the glue which holds together the sand and aggregates particles which constitute approximately 85% of the mass of the concrete. The hardening process is called hydration because portland cement gains strength through the chemical process of coniliming with water. Although, et later ages, concrete and monar lose excess water by drying, the material's strength development is based entirely upon the constituent minerals of cement reacting with water.

Cement is a finely ground assemblage of et least four major minerals. It is made by intergrinding a properly proportioned mixture of quarried argillaceous (containi1g alumina, iron, and silica such as clay or shale) and calcareous (containing mainly lime 6

_ _ . ___ . _ ~.

~

l such as limestone) raw materials.. This mixture is burned in a rotary kiln at temperatures of about 2650' F until it partially fuses, and the clay, silica, and limestone combine to form new minerals or compounds in the form of nodules ranging in size from perhaps 1/4 j to 3 in, in diameter. These nodules are called clinker and subsequen Jy axe interground l with approximately 5% of gypsum (calcium sulfate dihydrate) to produce the final i prcduct, portland cement.

For practical purposes, portland cement clinker may be considered as being composed of four principal compounds. The approximate percentage of each of these can be calculated from a chemical analyais of the cement. The major compounds with their chemical formulas and the accepted abbreviations in cement chemist's nomenclature

• are as follows:

tricalcium silicate 3CaO SiO2 = C3S dicalcium silicate 2CaO SiO2 = CS 2

tricalcium aluminate 3CaO Al O23 = CA f 3 tetracalcium aluminoferrite 4CaO Al.4,03 Fe2O3 = C4AF

] It is known that C3 S (primary) and C 2S (secondary) control most of the strength-developing characteristics of cement. These are the mineral phases which, with water.

hydrate to form the gel responsible for the :,trength of concrete End concrete products.

The C 3A (aluminate) phase also rapidly reacts with both water and gypsum in a way to

control the rate of setting and prevent early stiffening and loss of fluidity of concrete. This permits transit mixed trucks to transport concrete for an hour or more without significant

. loss of concrete workability. Cements with high C A 3 contents are more susceptib!e to l chemical deterioration by excessive sulfates in ground water or sea water. Ur. der these conditions, cements with a low C 3A content are required.

It is possible to produce portland cements particularly adapted for use under specific conditions. Differences in cemem type are brought about by changing the relative proportions of the four major mineral phases. This is accomplished by modifying the ratios of lime, silica, alumina, and iron in the original raw material mix design. The major portland cement types are specified in ASTM C 150, Standard Specifi.:ation for Portland Cement. Within this specification, the following five principal pw6ad cement types are described, each with specific requirements:

Type I general use portland cement Type II moderate sulfate resistance or moderate heat of hydration

• Conventional Cement Notation:

C = CaO; S = SiO 2 ; A = A1203; F = Fe2 O 3; 5 = SO 3 ; C = CO 2 ; and H = H 2 0.

7 4

Type III high early strength Type IV low heat of hydration Type V high sulfate resistance  !

Type I cement is the product manufactured for use in general concrete construction when I the special properties of the other types are not required. Type II cement is used for constmetion where moderate sulfate attack may occur. Its composition is characterized by I a relatively low C A3 content (less than 8%), which is responsible for its good resistance to 2

sulfate solutions. Type III cement is used where rapid strength development of concrete is essential. It usually has a higher C S 3 content and is ground to a much greater fineness in

order to increase its reactivity. Type IV cement, which ger

rates less heat than the other l types and at a lesser rate, contains more C2S and less C33 and was developed to reduce the potential for cracking in mass concrete which may result from high intemal temperatures caused by hydration. This cement is rarely produced today, because the same effect can be achieved through the addition of reactive siliceous materials such as fly l

ash or granulated blast furnace slag to Type Il cement. Such additives are called pozzolans and the resulting cement is called a blended cement. ASTM specifications C i 1

595 and C 1157 cover blended hydraulic cements. Type V cement is especially beneficial  !

when high sulfate resinance is needed. It has the lowest C3A content (less than 5%) of all ,

cement types.

]

An important optional requirement in C 150 is a limit on the permitted alkali (sodium and potassium) content of the cement. Low alkali cement (less than 0.6% ulblies expressed l as an equivalent sodium oxide) is required by some state highwny departments to prevent  !

l concrete distress resulting from an expansive reaction between the cement alkalies and susceptible or reactive aggregates. This condition is called alkali-silica reactivity (ASR).

f

~

Portland Cement Hydration The hydration of portland cement is complex. For the present purposes it is sufficient to just cover the basics.

Portland cement, containing the primary clinker minerals tricalcium silicate (C3S) or alite, dicalcium silicate (C2S) or belite, tricalcium aluminate (C 3A), and tetracalcium aluminoferrite (C4AF) or ferrite solid solution (Fss), and interground with gypsum and/or anhydrite, yields upon hydration a hardened mass consisting of a calcium silicate hydrate

, gel (C-S-H), calcium hydroxide (CH), and several calcium sulfoaluminate hydrate phases, principal of which is ettringite. The hydration of the silicate phases usually proceed as follows:

8

2C3S + 6H 4 C S32H3 + 3 CH (1) 2 C2S + 4 H -> C 32 S H3 + CH (2)

Both alite and belite form the same C-S H hydration product, but differing amounts of calcium hydroxide. The actual ratio of C/S is sornewhat variable and may range as high as

^

1.7. In any case relatively large amounts of calcium hydroxide can be produced. For a Type II cement the cdcium hydro ,ide content would be approximately 31% on a dry cement basis, or 22% on a hydrated cement basis Calcium hydroxide can be attacked by aggressive acidic waters or disrolved by pure waters. In more severe exposures the deium silicate hydrate can also be attacked.

Aluminate-sulfate reactions. The hydrr. tion of the aluminate and aluminoferrite phases are somewhat raore complex and go through several possible stages depending upon the amount and solubility c.f the calcium aluminate and calcium sulfate phases present. Pommersheim and Chang (1988). in the development of a mathematical model for the hydration of C 3A in the presence of gypsum, outline the following hydration steps:

C3A + 3 C5H2 + 26 H -> C A 3 3C5 H32 (3)

In equation (3) the reaction of aluminate with gypsum yields ettringite as the first product.

Schwiete and Ni6l (1965) have shown that with a nonnal commercial portland cement, ettringite is formed immediately after gaging with water and this salt is the main initial hydration pnxluct. 1 1 Monosulfate formation. When the molar ratio of gypsum to C A 3 is less than three, and when all of the gypsum is consumed, as would be the case in all ASTM Type I portland cements, the second stage reaction occurs as follows:

C3A 3C5 H32 + 2 C3 A + 4 H -> 3 C 3A-C5 H 12 (4)

The hydration product in equation (4) is the calcium monosulfoaluminate, which is usually considered tne end product of aluminate hydration, in the absence of carbonate. (Taylor speaks of the monosulfate later being incorporated into C-S-H.)

Calci im aluminate hydrates. At this point, most of the ettringite may have reacted, and if C 3A still remains, the third stage of the reaction will follow:

2 C3A + 21 H -+ C4AH 13 + C 2AH8 (5)

This reaction involves the simple hydration of residual tricalcium aluminate, and if it occurs in a cement having little or no calcium sulfate present in the very early stages of hydration, the resulting reaction is quite rapid, exothermic, and results in the early setting 9

s

j phenomenon known as " flash set." The mixture of these two hexagonal hydrates, and sometimes C4AH19, which are initially formed are thermodynamically unstable and will slowly transform to the cubic hydrate, hydrogamet, as follows:

C4 AH 3 + C2 AH8 9 2 C3 AH6+9H (6)

Sulfate attack. If a secondary and later source of sulfate ion is permitted to enter the system, any of the reaction products from equations (4), (5), or (6) will react in a similar fashion as the original C3A to reform the more stable calcium trisulfoaluminate salt, ettringite. In concrete, exposed to high-sulfate ground waters, this disruptive expansive reaction is known as " sulfate attack." In the hydration oflower C 3A-content cements, such as ASTM Type II or Type V cements, the hydration reaction usually terminates with j equation (3), or equation (4) progresses only to a minor amount until complete hydration of the remaining C3A occurs. ,

1 Ferrite-sulfate reactions. It should be noted that the ferrite phase, which is a continuous solid solution series ranging in composition from 2C F to C6AF 2 to C62A F, with an average composition of C4AF, will hydrate to form the same sequence of isostructural reaction products as C O 3 for Al23 3 A, but with some substitution of Fe2 O , as shown in the following examples of ettringite famiation:

1 3 C4AF + 12 C5H2 + 110 H -+ 4 C3(A,F) 3C3 H32 + 2 (A,F)H3 (7) l l

C4AF + 2 CH + 6 C3H2 + 50 H -+ 2 C (A,F) 3 3C5 H32 (8)

The stoichiometry for equation (7) requires some additional lime, which is why hydrous oxides of iron or aluminum also are formed. In the actual case of portland cement hydration, there is an adequate supply of excess lime from the hydration of the calcium silicate phases, as shown in equations (1) and (2), and equation (8) would appear to be more appropriate. Brown (1987), however, in a study of C4AF hydration in gypsum and lime-gypsum solutions, found that early aft formation occurs through solution and the resulting ettringite contains little or no iron Much of the iron appears to precipitate as a FH3. gel. Taylor (1990) opines that the tendency to form iron oxide or hydroxide may be associated with the fact that A13+ species can migrate relatively easily in pastes, whereas those containing Fe3+ cannot, and are mainly confined to the space originally occupied by the parent anhydrous aluminoferrites.

10

High-Alumina Cement (HAC)

General Information Calcium aluminate cement, or HAC is manufactured as a grey-to greyish brown powder ground to a high fineness (usually more than 95% passing through a 325-mesh sieve) and capable of hardening into an extremely strong mass when mixed with water. Paste, mortar, and concrete have the same meanings when applied te a calcium aluminate cement as to a portland cement. The hardening process is also called hydration with these materials, which also gain v.rength through the chemical process of combining with water.

Ahhough, at later ages, concrete and mortar lose excess water by drying, the material's strength development is based entirely upon the constituent minerals of cement reacting with water.

HAC is a finely ground assemblage of compounds of calcium and alumina, together with ,

lesser amounts of silica and iron, and minor amounts of alkalies, sulfur, titanium and  !

phosphorus. It is made by intergrinding a properly proportioned mixture of an alumina source (usually bauxite) and a calcareous material (containing mainly lime such as limestone), typically to a fineness of about 15% residue on a 200 mesh (74 m) sieve.

This mixture is bumed in a rotary kiln at temperatures of up to about 1550*C. There are j two important process variations in this burning step. In one variation, known as the

" fondu" method, the material is all fused in the kiln, while in the other, it is sintered.

I The third material usually present is a source ofiron; often iron is found in the bauxite )

used for the alumina source. Higher iron contents tend to favor using the fusion method l for preparing the product, while if silica is instead the chief impurity, a sintering method may be more suitable.

Typical analyses of high-alumina cement using the fusion method and the sintering method are shown in the following table. It is clear that the higher silica clinkers are often sintered, and the higher iron clinkers are frequently fused:

l l

l 11

Type of HAC Fused (Robson) Sintered (Lehigh)

% Al 2O3 39 46.5

% Ca0 37.5 35.5

% SiO2 4.0 8.5

% Fe2 O 3+ FeO 15.0 6.0

% MgO 1.0 0.7

% SO3 0.I 2.0

% Alkalies trace trace The compound composition of these cements usually feature monocalcium aluminate (CA) as the most important crystalline constituent. The compounds are listed in the following table:

Compound Formula Relatlye Remarks Abundance Monocalcium q CA High Most important component.

Aluminate Needed for strength.

Calcium Dialuminate CA2 Mod. Iow Slowly hydraulic 12-calcium 7-aluminate C12A7 Generally low Highly reactive-flash set risk Ferrite Nominally Mod High Slowly hydraulic C2(A.F)

Belite Nominally Moderate to Slowly hydraulic

-C23 absent Gehlenite C2AS Mod.-to-low Unreactive Wustite FeO Low Unreactive Melilite C2AS w/ Low Unreactive MgO and Fe CA is the most important phase for strength development, although 12 C A7 , CA2, and C2S may also contribute. It is noted from phase diagrams for HAC that the join to C2S is short and the join to C2AS is long. Thus C 2S is not present in all HAC cements. The remainder of this discussion will focus on the behavior of the CA.

High Alumina Cement Hydration The hydration ofli AC is complex. For the present pmposes we will consider only the hydration of the CA. When CA hydrates, the first compound generated is CaO Al 2Oy10H 0, 2 a hexagonal hydrate:

CA + 10 H2O --+ CAH10 This hydrate is the principal strength-forming material, which may also be converted at slightly higher temperatures into another hexagonal hydrate, C 2 AH8 . The other product 12

of this reaction is gibbsite (aluminum hydroxide, AH3 , of variable crystallinity), which has no cementitious properties:

2 CAHi o -+ C 2AHg + AH3+ 9 H 2O These reactions are normal for HAC and should not cause any problems. However, both of these hexagonal hydrates are unstable with respect to a cubic material, C3 AH6 (hydrogarnet). The conversion of the hexagonal hydrates to hydrogarnet is an exothennic process, and may be depicted per Neville (1975a) as follows:

3 CAHi o -+ C3 AH6+ 2 AH 3+ 18 H2O 3 C2 AHs -+ 2 C 3AH6 + AH3 + 9 H2O l This reaction has been termed conversion. The hydrogamet is a denser phase than either of the hexagonal hydrates, and therefore porosity of the system must necessarily increase when conversion occurs. The reduction in volume is greater than 50% (Mehta,1964).

I If a hydrating HAC system is maintained at a low temperature, or dry, or preferably both, the metastable hexagonal hydrates persist, and conversion is slow (though inevitable). On the other hand, higher temperatures and higher humidities accelerate conversica (Neville).

Also, conversion appears to be accelerated by the presence of alkalies. Ordinarily, again, the alkali catalysis would not be a problem in normal HAC systems because tne alkali level is so low. However, the portland cement used at Mill 3 tone #3, even though a low alkali product with less than 0.6% alkali eqt:ivalent, is still high in alkali content compared to that of HAC cements.

Conversion does not completely destroy the matrix; in fact, acceptable durability is often reported in fully converted, or nearly fully converted, systems. Hydrogarnet is susceptible to carbonation (Neville) however, and if a source of carbon dioxide is available, the hydrogamet can be further converted to calcium carbonate and gibbsite:

C3AH6 + 3 CO2 + 3 H2O -+ 3 CACO3+ 3 AH3 Acconfing to Lach (1987), this calcium carbonate may first manifest itself as vaterite ( -

csicite), fmally undergoing further change to aragonite or calcite. This ultimate reaction, termed " carbonization" by Lach, but perhaps better termed carbonation, completes the i degradation of the system, which is now without stmetural strength. At the same time, according to Lach, the hydrated alumina, that may have been amorphous, becomes crystallized into gibbsite crystals.

13 i l

Aluminate sulfate reactions. The HAC system is not very susceptible to sulfate attack, since the hydrating system contains no free calcium hydroxide. Therefore, ettringite is not a normal product of hydration. However, if the water flowing through the hydrated system contains time and sulfate, the sulfate durability of the material will be compromised:

CAH10 + 2 Ca(OH)2 + 3 CaSOr2H 2 O +14 H2 O -4 C 3A 3C5 H32 The reader is referred to Neville's book (1975a) and especially to his summary (1975b) of the book in a Canadian paper (enclosed in Appendix U) for a more thorough description of the stability of HAC concretes.

HAC conversion determinations are described by Brown (1977), Haines (1995), Jambor and Skalny (1996), and also Neville (1975a). Jambor and Skalny indicate mercury intrusion porosimetry is a better estimator for strength than conversion determinations.

Conversion protection measures are described by Ding et al (1996) and Fu et al (1996).

1 AGGRESSIVE WATERS l Portland Cement in Aggressive Waters In discussing the action of aggressive waters en the two cement types, it is important to recognize several generalizations:

• Corrosive action will occur at a significant rate only if the system has high porosity.

Acidic waters are corrosive to both portland and high alumina cement concrete, but less so in the latter case.

• Alkaline waters are very corrosive to high alumina cement concrete, but not nearly as corrosive to portland cement concrete (although alkali-silica reactivity may at times be a factor).

Waters containing carbon dioxide or bicarbonate are aggressive to both types of concrete.

• Very pure water (e.g., rainwater, distilled water, etc.) is very corrosive to portland cement concrete.

The reasons for these differences can be understood in the light of the chemistry of the two cements. In the case of portland cement, the first phase of this deterioration probably occurs via the leaching of calcium hydroxide from the paste. Calcium hydroxide has a 14

I l

i solubility of 0.185 grams per 100 mL water; this is by far the highest solubility of any i normally available calcium compound in the cement. To the extent that the pH of the water is lower than 7.0, the solubility will be higher. Once the calcium hydroxide is dissolved, it can react with carbon dioxide in the water or the air to form the much less soluble calcium carbonate, or it can percolate to the calcium aluminate cement formations, with effects to be discussed later. One of the most common components of hard water is  !

calcium. If the water already contains a significant concentration of calcium ion, the solubility of calcium ion from the portland cement will be correspondingly reduced, while if the water passing through the concrete is very low in calcium ion, the corrosivity will be enhanced.

F.M. Lea (1970) indicates, "...uure water decomposes the set (portland) cement compounds, dissolving the lime from them, and to some extent the alumina; continued leaching eventually leaves on:j a residue of incoherent hydrated silica, iron oxide, and alumina. This action on a mortar or concrete is so slow as to be negligible unless water is able to pass continuously through the mass. Waters that are acidic owing to the presence of uncombined carbon dioxide, of organic or inorganic acids, are more aggressive in their action, the degree and rate of attack increasing as the acidity increases and pH of the solution falls...In general, acid solutions which attack cement mortars or concretes by dissolving part of the set cement do not cause any expansion, but progressively weaken the material by removal of the cementing constituents. A soft and mushy mass is all that ultimately remains.

"The main cause of acidity in natural waters is the yresence of carbon dioxide and humic acid. The solubility of carben dioxide in water unc er various partial pressures, and the pH values of the solutions, are as follows:

Solutions saturatt.t. with CACO 3 CO2content of atmosphere Gms CO2 /1 at pH CACO3 dissolved, pH

(% by vol. ) 18'C grams / lite 0.00 - -

0.0131 10.23 0.03 (normal air) 0.00054 5.72 0.0627 8.48 0.30 0.0054 5.22 0.1380 7.81 1.0 0.0179 4.95 0.2106 7.47 10 0.1787 4.45 0.4689 6.80 As the table shows, in the presence of calcium carbonate, the carbon dioxide becomes combined as calcium bicarbonate and the acidity of the solution is much reduced."

Lea goes onto describethe effects of other calcium salts and of other salts onthe amount of aggressive carbon dioxide in the water. He wams against the effects of humic acid on the aggressivity of the water, and on the fact that percolation through limestone may help ameliorate this effect. He warns especially against pure waters. He states,"...as a broad guide, waters with a temporary hardness above 4-5 parts CACO3 per 100,000 are not likely to be very aggressive unless the free carbon dioxide content is very high, about 5 parts per 100,000 or above. As the temporary hardness decreases, so does the 15

concentration of carbon dioxide required for the water to be aggressive. For a temporary hardness of I-2 parts /100,000 a water will not have a marked solvent action unless the free CO2 is above about 1 part/100,000..., while waters less than 0.25 part/100,000 can be aggreaive even without free CO2." He also makes the point that flowing water is much more aggressive than stagnant water.

Popovies (1987), in the Durability Symposium sponsored by the American Concrete Institute, gives a useful summary of the deterioration mechanisms that may occur with concrete. He subdivides the deteriorations into six classes based on the underlying mechanisms. Under this classification, he characterizes Class I deterioration as that of leaching, caused by water containing carbonic acid or having low carbonate hardne51It leaches out the calcium hydroxide developed by the hydration of nortland cement.

Deterioration of Class I does not reach large proportions unless the dissolvinnewer of the water is high and the concrete is porous. but it can produce unsightly efflorescence.

Class II deterioration consists of non-acidic reactions between the calcium compounds of -

the hardened cement paste and compounds of the aggressive liquid. Here, the reaction products are weaker than the original hydration products, and/or have the tendency for leaching. Thus the process weakens the concrete and makes it more porous. A subclass, that is Subclass II-A, is where the mechanism is bar,e exchange as in the case of magnesium salts. Subclass II-B is the saponification reaction between animal fats and calcium compounds, and Subclass II-C is reactions of sugars forming calcium saccharate.  ;

Based on the discussion offered by Lea and Popovics, it appears that at least two factors were unfavorable for the leaching from the porous portland cement system -- the high porosity of the porous concrete, and the access of carbon dioxide to the system, either via I the water or via atmospheric exposure. This is of course combined with the high corrosivity of the water.

There are fairly standardized methods for determining the aggressivity of waters to portland cement matrices. Basson (1989), Basson and Addis (1992), Alexander (1993),

Alexander et al (1994) describe the use of the Leaching Corrosion Index and also the Langelier Index. The latter is actually designed for water pipe work, but it serves as a secondary index also for portland cement. Another index of some use is the Ryznar Stability Index (Drew Chemical Corporation 1986).

16

(

l l High-Alumina Cement in Aggressive Waters l

l With respect to high alumina cement, the picture is quite different. It will be mmembered I that the portland cement releases calcium hydroxide to the percolating water. Calcium hydroxide can interact with alkeli salts in the ground water to produce alkali hydroxides, or alkali carbonates after interaction with CO2- l Ca(OH)2 (aq) + K2 SO4 (aq) -+ CaSO4 + 2 KOH(aq) i 2 KOH(aq) + CO2 (aq or gas) -+ K2 CO3 (aq)

The situation is similar with sodium salts. Of course, these materials can cause a very high pH in the water. Robson (1962) indicates, "Unlike portland or supersulphated ]

cement concretes, those made with aluminous cement are not resistant to solutions of I alkali hydroxides since the protective alumina gel is readily dissolved. The presence of excess alkali in the ccm nt or in the aggregates must also be avoided in any aluminous concrete designed for structural pmposes because, apart from the effect on setting time l and strength development the alkali may accelerate the rate of conversion of hydrates...In special circumstances alkali hydroxides can play the part of a catalyst in a disruptive cycle involving atmospheric carbon dioxide...Aluminous cement concrete poles had been fixed l in a portland cement concrete block situated below ground level and deterioration was noted in the poles just above the base. It was found that alkali carbonate solution (derived I I

from alkali hydroxide in the portland cement base) had been drawn into an unduly porous aluminous cement concrete by capillary action and that evaporation of the solution had concentrated the salt below the surface of the concrete at the affected area. The following l cycle then occurred:

K2 CO3 + CAHy -+ CACO3+ K 20 Al O 23+ 10 H2O 2 K2CO3 + C2AHg --+ 2 CACO 3 + K 2 0 Al23 O + 7 H 2O K20 A10 23+ CO 2(gas) -+ K CO 2 3+ Al 0 23(or Al(OH)3)

The alkali carbonate thus regenerated by atmospheric carbon dioxide could contum. the above cycle. This alkaline hydrolysis of the cement required the presence of alkali in the concrete, and it should be noted that the direct action of CO2upon the calcium aluminates (as in normal atmospheric carbonatien) tends to improve concrete strength. Furthermore, the resistance of aluminous cement concrete to waters containing aggressive CO2 is very

! high under normal circumstances."

i Neville (1975a) describes also the alkali regeneration cycle and offers the following:

17

. - - . - . - - . . - - - - ~ - - - - . _ . - _ - _.. - -

i. i 4

"The alkali carbonate may also have its source outside the concrete, e.g. when it is placed

in granitic ground or in Portland cement.... the concrete in the interior becomes blackish in j colour and soft. ....the presence of alkalis encourages conversion even in the absence of evaporation or carbonation."

CTL concludes that a relatively low alkali ponland cement in contact with Lumnite@ may

[ be perfectly acceptable in dense monolithic structum, but may not be acceptable in porous concrete. An alkali limit even lower than that specified by ASTM for " low alkali" cement l

~

would be prudent for use in porous structums.

[ Neville continued regarding'the structural failure of the Stepney school in the U.K.:  ;

j " Feldspars and micas were detected in both the coarse and fine aggregate...The amount of

{ 'freeable sodium and potassium' in the aggregate was 0.021%....." He concludes: "It

! seems thus that alkali hydrolysis is a rare cause of deterioration of HAC concrete fairly f easily avoided: we must not allow percolation into HAC concrete through Ponland cement j' and, above all, we must not use alkali-active aggregates."

! Another mechanism for the breakdown of Lumnite@ cement is panicularly expected in l lime waters. Calcium aluminate cement is very stable in sulfate waters, but not in the

! simultaneous presence oflime [Ost,1997]. The reaction to form ettringite

[ (C3A 3C5 H 32)is favored by the presence oflime and sulfate. If the lime content is

{ high, an "in-situ" reaction may be expansive,i.e., it may cause spalling of the calcium  !

i aluminate matrix. Ettringite is very insoluble in high pH environment; it has a solubility i

. product of 1 x 10-45. In the Millstone situation the ground water could supply the sulfate

{ and the portland cement the time, i .

i-For additional general information on HAC perfonnance reference can be made to 4 Sorrentino et al (1995), and to Mangabhai (1990) which contains relevant papers by j . Midgeley, George, and Crammond.

l 4 ANALYSIS OF SUMP AND WElR RESIDUES, AND LIQUIDS i

1996 and Older Samples

{

in July of 1987 an internal Millstone Unit 3 laboratory reported that the residue contained

! 14.2% calcium,6.0% boron,5.3% aluminum,0.25% magnesium, and 0.1% " silica."

i The lab concluded that the material was consistent with a cement type material and funhermore that the alumina level was consistent with an attached specification sheet.

]

[ This specification sheet was not attached to the CTL copy. We assume this sheet may j have covered a high alumina cement. The laboratory recommended that the sample should I8

be analyzed by a laboratory familiar with the ASTM C-114 method for analysis of hydraulic cement. The high boron content is not characteristic for cementitious materials.

An analysis by the ABB-CE laboratory in August of 1991 concluded that the major crystalline phase present was calcium carbonate. Semi-quantitative analysis found the equivalent of 54% calcium oxide and 44% carbon dioxide on a " normalized" dry basis. A small amount of MgO equivalent was also detected, but all the other potential elements were below the detection limits of the techniques used. X-ray diffraction (XRD) analysis indicated that only calcite and an iron oxide sulfate hydrate compound were present.

The presence of a significant amount of alumina in the 1987 report was not confirmed in the 1991 report. This may have been due to a nonhomogenized sample.

No further analytical reports of this type were av " - until ABB-CE reports (see Appendix S) dated July 9 and August 14,1996 ..e issued. These reports did not contain an XRD analysis of the crystalline phases in the residues. The June and July 1996 samplings were done by NUSCO.

A very simplistic calculation was made (see Table 3) to estimate how much Lumnite@ or Type II cement would have to dissolve congruently to support the alumina le vel found in the July 9,1996 report residue. It appeared that if one assumed 1000 lbs of residue had l been produced over recent years it would have required only a couple of bags of Lumnite@ or maybe 20 bags of Type II. It should be noted that this estimate does not i

)

include the much larger net amount lost with the exit waters, nor does it account for any i deposition in the pipes. CTL pointed out that a large amount of soluble materials could I escape with the exit waters when the flow was integrated over the years.

CTL perfomied an analysis of the quality and significance of the two 1996 reports. Table 4 shows the calcia (CaO) to alumina (Al 23O ) ratio for the solid samples as determined by Inductively Coupled Plasma Spectroscopy (ICP). The same ratios are shown for the liquid samples in Table 5. For reference Table 2 shows the ratios for the two cements used at Millstone. The analyses of the cements were taken from the table of mill or hboratory reportsin AppenaixR.When the cakialalumina reslaue ratios were calculatea for X-ray fluorescence (XRF) elemental data (Table 6 included in the August 14,1996 report), it became obvious that there was rather wide disagreement between the oxide percentages determined by ICP and XRF (Fig. 5 and 6) and furthermore, even the ratios of calcia to alumina did not agree (Fig. 7). A calculation on one of the samples gave a low analytical total of 94% even when the larger of the numbers (ICP or XRF) was used.

19  ;

Discussions with ABB-CE indicated that the samples may not have been well homogenized before splitting. Drying techniques at the laboratory had varied, at least in the past. Sometimes drying to 81* C was used and sometimes moisture determinations had been done to a flat point between 105* and 155' C. At other times microwave drying had been used. There was also a question regarding SO 3 analysis, as XRF indicated 1.4% and ICP 0.30% SO3 .

Since the materials balance preposed by CTL would require more reliable data, and the nmples could not be sent to CTL because they were classified as radioactive by NUSCO, CTL proposed that cement analysis experts from CTL assist ABB-CE in standardizing on a fixed cement analysis test method. Such a test method was prepared by CTL and submitted to NUSCO (Appendix P). Overall CTL was convinced that ABB-CE had the required expertise and equipment to follow such a method. CTL also felt that it was important to pro:ect the XRD samples from carbonation by exposure to a limited carbon l dioxide air, or pure nitrogen atmosphere. l l

l 1997 Samples j i

Mr. Howard Kanare, Manager of the Chemical Services Group, CFL assisted ABB-CE ,

through telephone consultations and a personal visit to the laboratory before the next set of samples was analyzed. i ABB-CE subsequently issued preliminary reports dated February 24 and 28,1997 on samples taken at Millstone on January 20 and February 5,1997 (see Appendix S) The data indicated that in general a good quality analysis had been accomplished.

4 The XRF analysis confinned that there were significant amounts of both chloride and sulfate in the five residue samples. Actual sulfate levels may be even somewhat higher than given in the report. Per Mr. Kanare, the temperature for the fusion of the discs for the XRF may be high enough to volatilize chloride and some sulfur. Furthermore the flux matrix for the fusion may not be fully optimized. ABB-CE subsequently released the XRD samples to CTL since the small amounts were not considered to constitute a radiation hazard.

Thermogravimetric analysis (TGA) weight losses were run to 950' C at CTL.

Incorporating these, ABB-CE analytical totals then became 100.79,98.11,100.74, 100.55, and 99.32 percent, respectively, for the residue samples numbered 1 through 5.

CTL concludes that ABB-CE produced good quality and internally consistent analysis on the 1997 samples.

20

It should be noted that very little, if any, cement residue has appeared in the weirs or sumps. The materials now collected seem to be the result of reactions in the liquid phase.

In any case, conversion determinations were not done at ABB-CE since the samples were declared radioactive and ABB-CE lacked thermal analysis equipment for such tests.

ABB-CE recently released the five XRD residue samples (see Appendices I and O) to CTL since they were considered safe to handle because of their small size. The majority of the precipitated materials were derivatives of calcium aluminates in the cements (either tricalcium aluminate in the portland cement, or more likely calcium monoaluminate or hydrogarnet from the Lumnite@ cement). In addition, calcium carbonate was found in several of the samples. ABB-CE computer-generated XRD patterns (see Appendix S) were obtained on residue samples which had bet dried at 35' C, in a nitrogen  !

i atmosphere. Only samples #4 and #5 from the 7B sump, and that from the 7B N.W. weir showed calcite as the major peak. This probably reflects a lower flow rate, longer residence time, and greater chance for carbonation. This may indicate that the calcium l carbonate deposition now occurs mainly in the exit areas of the pipe where carbonated air i has access.

This was the first time samples had been protected from carbonating after sampling due to i carbon dioxide in the atrnosphere. If the previous samples had been severely carbonated it l would explain why the pre-1996 reports stated that calcium carbonate was the major i

crystalline phase. It is notable that the main ultimate decomposition product of most of these type of materials is calcium carbonate. The possibility that the process has changed since the earlier samples cannot, however, be excluded. One major change was the i I

installation of weirs completed in June,1996.

CTL's own review (Appendix E) of the ABB-CE XRD pattems found ettringite (C3 A 3C3 H32), Friedel's Salt (C A 3 CaCl2 H12), thaumasite (CaSi(OH)6 CACO 3 CaSO412H2O), calcite (CACO ), 3 and a mixed carbonate / sulfate analogue of the monosulfoaluminate(3CaO Al2 O3 0.67 CACO 3 0.33CaSO411H 2 O)-here abbreviated as "CCASH". The same five compounds were found in all samples, but in variable abundance. X-ray diffraction and TGA results agreed on the qualitative identification of the components of these residues, but there were some discrepancies on the quantitation, partially due to peak overlap. The following table gives the X-ray diffraction results, with the compounds listed in the order of X-ray peak intensity.

21

Sump / Weir Residue Samples - Compounds Identified 1 Sump 7A 7A 7A 7B 7B

,i Weir. South weir NW weir Sump NW weir Sump

, Compounds CCASH CCASH CCASH CCASH Calcite (in order of Ettringite ' Ettringite Ettringite Calcite. CCASH peak heights) Cricite Friedel's salt Calcite Ettringite . Ettringite j i Friedel's salt Thaumasite Friedel's salt Tharpasite ' Thaumasite l

l Thaumasite Calcite Thaumasite Friedel's salt Friedel's salt

]

1 Approximate Composition:

. % CCASH 33.4

{ %Ettringite 26.2 i  % Calcite 19.5 i-  % Friedel's S. 8.0

% Thaumasite 12.8 The chemical analytical results obtained for these samples by ABB - CE Laboratories are consistent with these findings. The table shows a rough quantitation of the sample from 7A sump, south weir, indicating that, in concert with the XRD results, the solid contains much higher amounts of the "CCASH" and ettringite than of the other components; the calculation was carried out using chemical and thermal analytical data.

The presence of ettringite indicates that the pH of the system was still high, but the presence of calcite shows that hydrated lime has carbonated in the water. The thaumasite may be a degradation product of ettringite in the presence of silica, alkali, and a low temperature. The chloride in the water may have converted some of the aluminate hydrates to Friedel's salt. Thaumasite is generally regarded as a later stage of degradation than ettringite or carboaluminates. In any case, the components of the residue can be understood as products of a dissolution mechanism, principally oflime in the portland cement, and of lime and alumina in the calcium aluminate cement matrix.

ABB-CE analyzed the water samples by ICP and standard wet chemistry. The " grab" samples taken after only a short period of equilibration on January 20,1997 agreed well with other types of samples taken on February 5,1997. Samples tak.en of waters just overflowing the weirs agreed well with the inside weir samples and with those taken from the end of the pipe when the weirs were empty. Replicate samples agreed and so did the QC samples. Only one spurious high magnesium value was reported, but it was possible to disregard it because the sample in all other respects agreed well with a companion sample, 22

ANALYSIS OF WELL WATERS Water samples were obtained on January 15 and February 14,1997 from test wells drilled fairly near the structure and mostly in a direction coincident with the estimated path of water (from North and Northeast of the structure). Figure 8 shows location of the holes.

Appendix T gives information on the drilling operation. Two rain samples were also submitted to CTL for analysis. The CTL analysis of the samples is given in Appendices F and K.

Corrosivity indices for the well bore waters and two samples of rain water are also shown in Appendix K. A map of the wells and information on the drilling opcration is given in l Appendix T. '"hree corrosivity indices previously referenced were used: Aggressiveness Index, Langeliet Index, and Ryznar Stability Index. The first index is designed for portland cement marices. The Aggressiveness Index calculation equation is shown in Appendix V. The follt. wing parameters are used for the index: pH, calcium carbonate saturated pH, calcium hart' ness, total ammonia, magnesium, sulfate and chloride. No index for Lumnite@ concrete was located.

Aggressiveness Aggressiveness ,

Sample. Date Index Sample, Date Index 1 A,1/15 High 4A,2/14 High 1 A,2/14 No 4B,1/15 Non 2A,1/15 High 4B, 2/14 Non 2A,2/14 Very High SA,1/15 High 3 A,1/15 Very High 5A,2/14 Non 3 A,2/14 Very High 5B,1/15 Very High 3B,1/15 Mild 5B,2/14 High 3B,2/14 Mild 6A,1/15 Mild 3C,1/15 High 6A,2n-, Mild 3C,2/14

• Rain,1/25 Very High 4A,1/15 Non Rain,2/25 Very High

• No water in this location on this date Samples I A,2/14 and 4B,2/14 had high pH readings which were confirmed on CFL retained samples, but not fully confirmed on retest samples on 3/3/97 at NUSCO.

Based upon the average Aggressivity Index, locations 3B,4A, 5A, and 6A probably contain mildly corrosive waters. Locations l A,2A,3C, and 5 3 contain highly corrosive waters. Location 4B waters are probably non-corrosive. The 3A and rain waters are very higiily corrosive.

23

1 There are some indications that the January 15,1997 samples were still affected by the drilling operation. Thus, obtaining a third set of samples would be prudent.

The high pH of 11.9-12.6 (see Appendix K) for all the exit waters clearly shows the waters can here be considered as non-corrosive to portland cement matrices. Conversely, at this point the waters would be aggressive to Lumnite@ matrices.

Note that hole SC was a dry hole indicating a well draining area. On Febmary 14 the 3C hole was also dry reflecting a dense bedrock with low permeability. Important parameters analyzed for are given in the simulated three-dimensional views in Figures 11 through 16.

Unfortunately no hole was located in the path of the flowing waters close to the stmeture.

Such a hole was not possible due to subterranean services in this area. It is expected that holes 4 and 5 would not provide upstream waters to Unit #3. Note, however, that hole SA in bedrcck has a relatively high chloride content indicating possible intrusion from saltwater into a foliation layer in the rock.

Based on the geologic description provided to CTL the most permeable layers are the fill and the sandy lenses in the glacial till. The basal till and the bedrock are fairly impermeable. It, therefore, appears that the best estimate for the waters reaching Unit #3 would be the waters passing through holes 6A. l A,2A,3B and 3A.

An analysis of the change in the concentration of the soluble ions as they pass thru the structure shows: ,

l l

Location: Ave. for wells

• Ave. Weirs ** l 1

Date:

Jan.15, Jan.20, Feb.14 Feb.5 Sulfate, ppm 71.3 3.82 Sodium, ppm 35.9 89.9 Potassium, ppm 6.46 40.4 Chloride, ppm 26.6 50.7 i

• Holes 6A, I A,2A,3A,3B

• • Except Weir 7B South, stagnant water From the wells to the sumps, there was a large loss in sulfate, a 2.5 times increase in the sodium, a six fold increase in potassium, and an increase in the chloride. The loss of sulfate is consistent with the formation of sulfated residues. The potassium content in the cement was 0.47% and the sodium was 0.25% The cement could be a source for the alkalies. The increase in the chloride content remains unexplained unless there is another 24

_. . .~ - -- -- - ._ . - .- . .-

E

)

i source of chloride such as salt or sea water in the system to supply the required amount for the formation of the calcium chlorealuminate in the residues. Note that the mineral i

chlorite does not contain chloride.

ROCK CORE SAMPLES FROM WELL BORINGS -

5 Physical inspection by the petrographer of five rock cores and one glacial till are included j in Appendix G. The petrographer concluded that the alkali (potassium) feldspars and

micas (biotite and muscovite) although normally considered stable may be susceptible to 4

incipient attack by waters and be a source for potassium and sodium. This finding is consistent with the elevated potassium and sodium found in the water samples taken from the wells. XRD (Appendix C) confirmed the mica and plagioclase, but that analyst did not think the materials were leachable. Appendix N shows TGA and DSC analysis. See Appendix T for location and depth of the cores.

AMOUNT OF RESIDUE IN POROUS CONCRETE AND MATERIAL LOST IN EXIT WATERS There are three main destinations for the mobile reaction products from the attack on the cements.

Some of the products end up inside the porous structure as precipitaes. The actual amounts are not known, but an estimate can be made. A reasonable estimate for the loss of sulfate as the waters pass through the structure is 67 mg/L (see separate discussion of the loss of sulfate in the waters, also Figure 12, and the weir water analysis in Appendix  ;

K). If this sulfate precipitates a residue of a composition .similar to that previously j calculated for South weir 7A residue which was analyzed to contain 2.87% SO3 the total residue for an average flow rate of 1000 gpd for ten years would be about 59,300 lb (see Appendix X). If the flow rate is only 200 gpd the residue would be 11,900 lb. This is a  ;

first order estimate, but it clearly indicates that a relatively low concentration of sulfate in l the water can produce a large amount of residue in the structure when dealing with compounds such as ettringite whose solubility products are so low in the high pH waters i that the reaction goes virtually to 100% completion even at low concentrations. 4 The amount of residue collected in the weirs and sumps is rather well defined at only about 1,000 lb. over the years 1987 to 1996. This tends to indicate that a large proportion  :

of the precipitation occurs inside the structure and the reaction is almost complete when the water reaches the weirs and sumps.

25

CTL developed a method to calculate the amount of lime and alumina escaping with the waters to be used when better defined annual flow and annual concentrations are known.

This method is illustrated by the example shown in Tables 8,9, and 10. The flow rates used in the example are those as measured using weir and sump fill rate data by NUSCO (Appendix W from NUSCO draft report dated 3-6-97). The large calculated losses may grossly overstate the net losses because they are based on instantaneous flows and concentrations, which may not be representative of average annual values. There are also some uncertainties in the method used for determining the flow rates. Until actual annual flow rates and average concentrations are known, one may want, for instance, to scale down these numbers linearly for a flow rate of 1000 gpd as a first order approximation.

The numbers do, however, illustrate that relatively large amounts of calcia and alumina are escaping with the exit waters. The numbers also indicate that a relatively large amount of the calcia is coming from the ground waters, but the preponderance of alumina comes from the structure.

For comparative purposes it should be noted that the porous portland cement concrete contains about 910,000 lb. of calcia (2198 cubic yards of concrete contaiing 655 lb. of cement per cubic yard with the cement containing 63.37% calcia). the porous Lumnite@

concrete contains about 1.69,000 lb. of alumina (556 x 644 x 47.20%).

NUSCO does not have a definitive direct way of determining the average ant ual water flow rates. They are estimated by integrating run times of pumps in sumps 7A and 7B (see Figure 4). Unfortunately, the pumps are not positive displacement types and the check valves also stick due to residue. Thus, some waters from sump 10 leak back through the check valve and are pumped again. The flow calenlation is complex involving differential flows between pumps in sumps 7 and 6. Flow rate estimates range from 200 to 1600 gpd. Instantaneous flow rates through the weirs are also estimates. They are based on times to fill the weirs, but uncertainties are introduced when the weir water levels are lowered for the fill-time measurements. This permits an in-rush of water from the porous concrete. Instantaneous flow rate estimates have been about 1600 to 2300 gpd.

MOCK-UP TESTS Since actual samples of the Millstone Unit 3 substructure porous concretes were not available, NUSCO pmceeded with three phases of a mock-up study to try to simulate what was happening 60 feet below ground. These test are described in three volumes entitled " Porous Concrete Mock-up Testing for Millstone Unit 3, Waterford Connecticut".

26

These reports were issued by Alden Research Laboratories, Inc (AP1) in final revised form in December,1996.

Phases I (1992) and II (1993) were short term 40-day exposure studies (30 days of flowing water) and Phase III (1995) was a 15-cycle exposure study. Each cycle consisted of 21 days of water flow and 7 days of drainage before coring. The tests were designed to predict ultimate performance at Millstone.

Mock-Up Test Difficulties All accelerated studies of concrete performance are difficult. Portland cement continues to hydrate over years and is subject to external exposure conditions. Lumnite@ cement is also subject to many inter-related changes due- to the effects of time, humidity,

) temperature, carbonation, and other exposure conditions. Even for portland cement, there is no single accepted way to predict ultimate performance under difficult-to-define exposures from short-term tests.

The following additional difficulties or significant differences from the "as-built" constmetion were encountered with the mock-up tests:

j 1. Aggregate / cement ratio 2 With reference to Table 1 it can be seen that phases I and II used an aggregate cement ratio by weight of about 6, very much different than the ratio of 4 in the Millstone concretes. In Phase III, the ratio was 4.75, still significantly higher than in the Millstone concretes. With a porous concrete a change in the aggregate-cement ratio changes mainly the thickness of the paste layer as long as the water-cement ratio is held constant. There was also a change in the aggregate type from Phase I to Phase II and III work (see Table 1 and Appendix G).

2. Concrete temperatures The concretes were cast in the summertime. The Millstone concretes were

, cast in the winter. No record of the intemal temperatures of the concretes were kept.

3. Water compesition Only a very simple analysis of the exposure waters was performed. In fact, on occasion the well resupplying the water "may have been pumped almost dry" possibly allowing some solids or variable composition waters to enter the 27

. . . . - .. -- -. . .- - - - . -- . .~ . - --

I i

. system and confound the composition of residues collected. Some of the waters were recirculated. No measurements of the corrosivity of the waters were taken. Exit waters were not analyzed. j 1

4. Curing procedures i Phase I Lumnite@ concrete curing was not fully described in the ARL report

, except that it was cured for 2 days. No PC grout was used.

. 2 Phase II Lumnite@ concrete curing is not described in detail in the pertinent l t' ARL report. It has been related to CTL as being similar to "as-built" wet curing.

t The curing procedure of the Lumnite@ concrete was optimized in Phase III by l

using a gentle fogging spray as soon as the paste no longer rubbed off on a moistened finger. After the concrete reached an age of 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> no further  ;

! water was supplied for the remainder of the 7-day cure. This procedure is expected to produce a durable product and is totally different from that  ;

j reportedly used at Millstone. I I CTL notes that fog curing was recommended for the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> by the i- cement company, Atlas Lumnite@ & Refcon@...(1987). According to a j company spokesman, this has been known since the 1950's. This avoids heat buildup and evaporation of water in the surface layer. The company j recommends against flooding or wet burlap curing. CTL also notes that Lone

.. Star Lafarge in Netter (1975) recommended " membrane curing water spray, i

or water sprinkling should be used to prevent loss of water of hydration, especially during hot weather."

5. Residues Residues were encountered with both Phase I and 11 concretes, but only when the two cements were in contact (double layer). In the construction of these mock-ups, the concretes were placedin 2 or 3 layers each producing " cold joints" which would have low resistance to horizontal water flow and not provide sufficient residence time for the waters to saturate with the ions.

No residue was encountered in Phase III. According to ARL staff, even water drying on the floor did not leave a residue. The logical conclusion is that the attack mechanism in Phase III was completely different than in the Millstone concretes. In the Phase III mock-up, it is likely that only conversion is 28

l ocurring without appreciable leaching or other reactions. This would make

[

' translation of strength predictions to the Millstone concrete tenuous indeed.

!. In view of the long list of difficulties, one might on a first look conclude that none of the mock-up tests is translatable to the "as-built" structure at Millstone. Considerable effort

had gone into the mock-up studies and they deserved a closer look. The results of the' l - rather intensive study requested by NUSCO clearly shows that although direct j quantitative links are hard to find, usable qualitative data was developed. In fact this latter data was instmmental in understanding the complex reactions in the "as-built"

! structure.

l l Analysis of Mock up Samples t-l Mr. K. Lakshmipathiah of NUSCO selected typical cores and cages from the thr, mock-up phases to be sent to CTL. It should be noted that the cores and cages were not taken at l

j- the end of the flow tests. They were taken in December 1996. Two of the slabs, Phase I j.

and II, had been stored outside for years. Phase III slab had been stored for months in a l~ building which was not always heated. The samples probably are more converted than at j the end of the flow tests. Included with the samples were some residues and filters from Phases I and II. The samples are described in detail in the petrographic report (Appendix l - G) and the X-ray diffraction (XRD) report (Appendices A and B), which shows locations

{ where samples were taken at CTL. Differential scanning calorimetry (DSC) and j thermogravimetric analysis (TGA) traces are included in Appendices L and M. Appendix j J shows X-ray fluorescence (XRF) elemental analysis.

i l Experimental Results

Tables l'1 and 12 give the combined results of X-ray diffraction and thermal analysis of the samples from four cores

Core IA (mock-up phase I), Core IIA (mock-up phase II),

and Cores III 9M and III SL(mock-up phase III).' In general, the results show that the porous calcium aluminate cement concrete in these specimens has undergone complete conversion. Small amounts of calciummonoaluminate decahydrate (CaO A1 O) 2 Oy10H2 were found in a few samples, but no 2CaO Al2 Oy8H2O was found. In more severely weathered samples, the conversion product hydrogarnet (3CaO Al2 Oy6H 2O) had also decomposed to calcium carbonate and aluminum hydroxide (gibbsite). The co-existence of small amounts of CAH oi even in cases of advanced decomposition was also documented by Lach (1987). This carbonation may have occuned from CO2 in the air, or it may have occurred from carbonate in the water, in the less severely weathered calcium 29

i aluminate cement samples, such as the Lumnite@ grout, substantial amounts of hydrogamet and gibbsite together with CAH io, and much lower quantities of calcium

! carbonate, attest to some residual durability in these an:as.

The foregoing has emphasized the problems associated with allowing potentially aggressive waters to contact both the portland cement and the alumina cement porous concretes. The evidence is strong that leaching and degradation have occurred. Apropos l to the discussion is the work of V. Lach, which was published in the ACI Durability f Symposium. In the case he describes, high-alumina cement was used for structural concrete. This structure collapsed reportedly due to a gradual decrease in concrete l strength below the criticallimit. The remaining pan of the structure, which was made from portland cement concrete, was not damaged. Testing indicated that the high-alumina cement concrete had a lower-than desired bulk density and a high porosity. The concrete i

had all changed to a brown color by the time the collapse had occmred. The degrees of

! carbonation and conversion were both very high, as determined by the low (but still l measurable!) amotmts of CAHi o and uncarbonated lime. The striking similarity of his i findings and those in the present situation are appreciated by the findings given in his i " Table 2", which is reproduced as follows:

Stage Binding Phase Composition Remark l

Starting CAH io+C2AH 8+AH3 gel Principal Phases only I CAH io+C2AHg + AH3 gel + Weak initial carbonation (ionic

CACO3(vaterite)+C2AS reaction through solution)

II CAH io+C3AH 6+ AH gel 3 + CACO 3 Conversion (main rxn), visible (vaterite) + FHn + C2AS color change to brown, decrease of strength III relicts of CAHi o+AH3 gel + intensive carbonation added to I vaterite+FHn+C2AS conversion (topochemical) strong decrease in strength IV AH 3gel + AH cryst3 + FHn + CACO 3 Recrystallization, complete (Aragonite, calcite)+C2AS(C2ASHg) degradation,concretecan be crushed by hand Many of these minerals are found in the X-ray diffraction and DSC tests representing stage II and III in the above table, and some calcite and aragonite have also been found.

The author (Lach) concludes that alumina concrete cannot be applied in principle for supponing structures.

30

i i

i-t ,

)

1 l The mock-up portland cement concrete and grout have undergone less weathering than J l

l have the calcium aluminate cement concrete and grout. Amounts of residual calcium

! hydroxide are less than would be anticipated from unweathered structures; some of this i material has apparently been leached out of the specimens. Calcium silicate hydrate (the

! principal component for durability and strength) is apparent in most of the portland cement )

samples. However, especially in the porous portland cement concrete, carbonation has l

3 proceeded significantly. In most cases, it seems to have affected principally the amount of  ;

calcium hydroxide, rather than calcium silicate hydrate. In many cases, ettringite I

(3CaO Al O3 2 3CaSO4 32H2O) is still present. Since ettringite becomes unstable as the i f pH decreases below about 10, it can be surmised that the pH in these areas has been maintained greater than 10.

In some cases, there is evidence of movement between layers in the mock-ups. The presence of gibbsite in the portland cement areas reflec's probable contamination from nearby calcium aluminate cement-based forrr stions. The amount of calcium carbonate in some of the calcium alumin'tu. ment-based formation implies the possibility that some calcium hydroxide leMhed from the n:arby portland cement-based structures may have  !

leached into this area and subsegaently carbonated. 1 i

The calculation of conversions for the HAC mock-up sections is complicated by the fact that the aggmgate contains chlorite, a micaccous mineral giving an X-ray diffraction pattem very similar to that of pay residual calcium aluminate decahydrate (CAHi o). There may also be overbps in tha thermal analytical peaks in the 80-160*C temperature range. ,

Both of these facts introduce uncertainty in the conversion calculations. Nonetheless,it is possible to use the thermal analysis data to set an upper bound on the amount of CAHi o in j the samples. This permits a determination of the minimum conversion rate that must have  ;

occurred. This will be a minimum value, because some clay minerals that may be present  !

also may decompose in this temperature region.

1 The mock-up samples from Phases I and II contain no detectable unreacted CAHi o. The residue samples were generated from aqueous solution, and thus would not be expected to contain CAH10 in the pacipitate. However, there is evidence that some of the CAH10 may still be present in the mockup samples from Phase III. In particular, we have j investigated the relative amounts of CAHio and gibbsite in the samples from Core 9MIII.

The following table shows these calculations:

i '31 1

9 i

Conversion Calculations for Core 9MIII

Compound - Mol. Wt. LOI TGA Fraction Theoretical AH of d

(theoretical) dehydration CAH io 338.18 53.27 % 29.81 % 705.87 i-

AH, J/g H2O Gibbsite 3449 1195.06 Hydrogarnet 2797 599.46 Ca(OH)2 4568 1110.67 CAHoi 1325 CAH to, AH, CAH 10 .%' Gibbsite, Gibbsite, % - Consersion,%

J/g ma AH, J/g ( n)

Reference 12.44 1.76 67.11 5.62 76 %

Core 9MIII-7 26.94 3.82 83.7 7.00 65 %

Core 9MIII-6 11.2 1.59 55.56 4.65 75 %

Core 9MIII-5 9.66 1.37 80.77 6.76 - 83 %

Core 9MIII-4 10.25 1.45 56.57 4.73 77%

Core 9MIII-3 6.76 0.96 24.32 2.04 68 %

Note that the sample numbers are the same as shown in the XRD report. It can be seen that conversion ranges from a minimum of abcut 65% for the Lunmite@ grout to over -

83% in the sampic tht had been known to be exposed to the atmosphere because of a mid formation break. Thus the degree of conversion is greater upon greater exposure, which is consistent with expectations. The actual conversion, of course, may be greater than the numbers cited, to the extent that clay and micaceous minerals may also be undergoing decomposition at temperatures in the range of 80-150*C.

Residues from Filters and Unused Molds Samples from filters and an unused mold were also investigaied using the XRD and thermoanalytical methods. In " Mold C", there was evidence of the presence of the "CCASH" material.or.a very similar material, asjudged by XRD. The " Unused Mold" sample was principally calcite (73%) and gibbsite (7.3%), and showed a high degree of decomposition. Quartz and mica from the water or the aggregate were also present, together with a small amount of what appeared to be aragonite. The sample labeled " Mold A" contained significant amounts of calcite (33%), and lesser amounts of gibbsite and aragonite. The chemical analysis of this material showed significant levels of magnesia, but no MgO-containing compounds were discovered with XRD. The sample labeled

" Meld C" showed the material "CCASH" (>60%) together with gibbsite (3.3%), calcite, 32 j

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

I i I

and vaterite, (28% combined carbonates) consistent with the material found in the

! weathered cores, and also in t'. e sump residues.

1

. Conclusions from XRD and OSC Work

i

I

I l

l In general, the correlation between the two techniques were very good. The residue '  !

. samples (panicularly the sample labeled " Mold C") had compositions generally in )

agreement with those of the sump and weir samples. There is evidence that leaching continues to occur from both portland cement and high-alumina cement stmetures. The mineralogical nature of the precipitates implicates carbonation as a potential weathering

]

!. mechanism. The degree of degradation suggested by the sump and weir samples is less >

, severe than that evidenced in the mock-up samples from Phases I and II, but more severe

than suggested by the mock-up samples from Phase III. The results suggest that leaching, j conversion, chemical interaction between the hydration products of the two cements, and

! carbonation are all occurring and may be contributing to potential durability questions, j panicularly with respect to the porous calcium aluminate cement concrete in the "as-built"

{.

j struture. Probably the strongest mechanistic evidence is that the high pH water from the ,

portland cement concrete (possibly near saturation in calcium hydroxide content), and the-l product of metathesis of dissolved alkali to hydroxide, resulted in a high pH leachant that

dissolved significant amounts of alumina and lime from the calcium aluminate concrete.

l The interaction of the two cement chemistries must be considered potentially deleterious.

! l 8

Summary of Petrographic Work (IAcck-Up Tests) .)

i 1

The following conclusions are extracted from Appendix G: ,

l '!

l Reaction. The apparent extent of reaction between the materials containing Lumnite j 1 cement and the materials containing portland cement is small. Where portland cement concrete or grout is in contact with Lumnite@ concrete or grout, only a thin layer of altered paste with closely spaced scaling cracks is observed. This altered and cracked layer is observed only in the material containing portland cement. None of the samples studied exhibited intimate bo'nding between materials containing the dissimilar cements.

However, where a thin layer of Lumnite@ paste locally coats portions of the porous portland cement concrete, the Lumnite@ paste is soft.  :

Definitive evidence of alkali attack was not observed in the samples. Alkali attack may manifest itself as a black discoloration of the paste accompanied by paste softening and 33

a ,

l 1

!. cracking. According to Neville (1975), deterioration by alkaline hydrolysis (aikali attack)

I < 'of the calcium aluminate compounds is a potential concern in the following scenarios: )

• Water percolates through portland cement concrete into Lumnite@ concrete.

[ = When Lumnite@ (high-alumina cement) concrete or grout is in contact with  !

portland cement concrete or grout (portland cement supplies soluble alkalies I

4 for the reaction). .;

j. '

*- . Lumnite@ concrete contains or is embedded in rock types such as granite,

! schist, or micaccous rocks which may be a source of soluble alkalies, i- '

! .. In general, porous concrete is more susceptible to attack. However, because conversion

[ (transformation of the hexagonal calcium aluminate hydrate phase, CaO 23 Al O 10H2 0, to

[ cubic hydrogarnet,3CaO Al O 236H2O) increases the porosity of the paste, the dense Lumnite@ grout is also susceptible to this reaction. According to Neville (1975), alkali l

j attack proceeds in the following steps:

K2CO3+ CaO Al23 O aq -9 CACO3 + K 20 Al2O3- .j l

.CO2in the atmosphere or dissolved in water regenerates the K2CO3:

j CO2 + K20 Al 23 O + ag -9 K2 CO3 + Al2O3 3H 2O j Alkali serves as a carrier. The final reaction is:

i 1

!: CO2 + CaO Al23 O ag -4 CACO 3+ Al2O3 3H2O 4

l The end products of alkaline hydrolysis are CACO3 (calcium carbonate) and Al(OH)3 a .

, (gibbsite). These am major constituents of the soft Lumn!!e@ paste but may also be l 3 derived by processes other than alkali attack. The portland cement components in the

{ mock-up construction reportedly contained a low-alkali cement. Components of the l

l mock-up constmetien, Phase II and III, containing Lumnite@ cement and the siliceous

[ metamorphic aggregates (granitic gneiss and mica schist) have an internal source of

[ a'kalies. The local rock types at the Millstone site, predominantly granitic gneiss, pegnatitic granite, and mica schist, are also a source of alkalies. Trap rock (gabbro and

[

i diabace), present in some of the mock-up samples (Phase I), is not a significant source of solubk alkalies, i.e. the feldspars in these rocks are calcie plagioclase.

Conversion, Inversion. The calcium clumir. ate cement paste in the Lumnite@ grout and concrete is largely converted based on macroscopic and microscopic properties.

Based on paste hardness and tenacity of the paste-aggregate bond, the strength of the Lumnite@ grout appears acceptable. However, portions of the porous Lumnite@ concrete 34

i i

1

'l i

are crumbling where the paste is soft and weak. The color of the paste fraction in the j mock-up porous Lumnite@ concrete is mottled shades of brown. The uniform dark l brown color of the paste fraction of the reference sample appears to be an indication of a lesser degree of conversion. In the porous Lumnhe@ concrete, Phase I and Phase II, the i relative softness of the paste and the abundance of random microcracks reveals volume l

change and an increase in secondary porosity. The majority of the paste constituents are isotropic. Conversion products, such n 3CaO Al2Or6H20, are isotropic; the primary I hydrated calcium ah

minate phases are not. Identification cf the paste constituents is best

determined by differential thermal analysis (DTA) and X-ray diffraction analysis (XRD).

Neville (1975) presents a formula for calculating the degree of conversion

% conversion = mass of gibbsite / (mass of gibbsite + mass of i CaO Al2 0- 10H2 O) x 100 Conversion of the calcium aluminate hydrates is accelerated by elevated temperatures and i requires the presence of water to dissolve and reprecipitate components. The conversion process is generally regarded as the cause ofloss of concrete strength, but see Jambor and Skalny (1996). Various explanations have been offered including:

i -

Densification. Conversion ofless-dense, metastable pseudo-hexagonal phases to denser, stable isometric phases (crystalline inversion). This results in increased porosity.

Aging of tricrocrystalline alumina gel. Alumina gel transforms to 3

macrocrystalline gibbsite.

4 WM.r. A significant amount of water is released into the pores during  !

.ersion.

~

1 Paste color change from dark brown-gray to lighter shades of brown appears to i 1

accompany conversion of the calcium aluminates and, perhaps, reflect the degree of j conversion The lighter sht. des of brown are attributed to the oxidation of ferrous iron to ferric iron (" inversion") wh!ch according to Neville does not reduce : trergth.

Historically, American high-alumina cement contains approximately equal amour ts of l ferric and ferrous iron [Robson,1962]. The increase in porosity that results from 1

densification (above) facilitates the oxidation of iron.

Leaching, Dissolution. The porous portland cement concrete exhibits evidence of paste leaching around the voids. Calcium hydroxide, and probably some calcium in calcium silicate hydrate (C-S-H), has been removed (dissolved) from the paste. The 35

optical properties of the paste have been altered; the crystallinity of the leached paste is reduced such that the paste appears isotropic. However, the physical properties of the paste are not substantially affected und the strength of the porous concrete has not been reduced. Calcium removed from the paste has combined with CO2 in water to form calcium carbonate deposits precipitated as a thin lining in the voids, but this may have happened during dry storage of specimens before coring. The portland cement grout does not exhibit evidence of paste leaching or dissolution.

In the porous Lumnite@ concrete, the principal evidence of paste leach ng (i.e. removal of Ca and Al ions) is the presence of gibbsite and small amounts of calcium carbonate partially lining the voids. The optical cnaracteristics of the paste do not reveal a well-defined zone ofleaching. An uneven layer of greatly increased porosity is observed adjacent to many voids and is considered evidence of leaching and dissolution (i.e. i l

removalof amineralphase).

Erosion, Cavitation. Erosion includes the processes of solution (chemical erosion),

corrasion (mechanical erosion by moving water), and transportation. Cavitation, sensu i strictu, is the corrasive effect of bubble collapse (Bemoulli effect) at points where pressure is increased due to decrease in velocity. Evidence of solution and corrasion occurs in the porous concrete (Lumnite@ and portland cement). As judged under the microscope, solution (see leaching and dissolution) appears to occur to a greater extent in the portland cement concrete. Rough paste lining the surface of voids in the porous concrete in the Phase 11 samples is considered evidence of erosion, presumably by corrasion. Corrasion occurs to s greater extent in the Lumnite@ concrete; soft paste is more susceptible to mechanical erosion. Fine-grained calcium aluminate paste debris is present in voids, coats the membrane in Phase III samples, anci appears a a loose coating on the portland cement grout in Phase I and Phase III samples. Poro m o Ge porous Lumnite@ concrete in Phases I, II, and HI have been reduced to rubble. 6iis material loss, or loss of integrity, is attributed to a combination of corrasion and soluen enabled by conversion of the calcium aluminate hydrate ccmpounds.

Mock Up Concrete Strength and Predletion of Long Term Strength

. Mock-up tests were performed by Alden Research Laboratories,Inc. (ARL). ARL vras responsible for the construction of the mock-up structures, subjecting the mock-up structures to various flow conditions, and coordinating, obtaining and testing of core s unples removed from mock-up strucmres. While not specifically stated in the ARL reports, t seems apparent the purpose of the mock-up test was to simulate the conditions 36

l 1

i h

. ' in the foundation of the plant in which a high-alumina porous concrete was in near ,

proximity to'a porous portland cement concrete and to determine if water movement  !

through this system could lead to degradation of either porous concretes. l l The tests performed by ARL were designated Phases I, II, and III. Summarized in Table 1

are salient properties of the porous concrete used in construction of the plant and used in i j each of the three phases cf ARL's work. Perhaps the most important parameters for i

. porous concrete are aggregate gradation and aggregate /ccment ratio. Recognizing coarse

' I aggregate gradation was controlled to the extent pcssible in all three phases of work, and no specific data were available with respect to gradation used in construction of the plant

]

other than its conformance with the ASTM C 33 specification, aggregate / cement ratio j becomes the most important parameter to evamate when comparing using mock-up test  !

j- data to predict field performance. Phases I and II of the mock-up test program used an l aggregace/ cement ratios, by volume, of 6.5 to 7.29. This is considerably in excess of that reportedly used in plant construction,4.41 io 4.43. Given the difference in l aggregate / cement ratio between the mock-up tests of Phase I and II and plant construction, i l it is unlikely that data from these phases of the work can be used to predict Seld performance with any degree of certainty. Phase III ARL mock-up work did, however,  !

use an aggregate / cement ratio mom in keeping with reported field values. Unlike Phase I )

. and II work, Phase III mock-up work may provide laboratory data that can be extrapolated to field conditions and potentially used to evaluate long-term performance of the porous concretes. In interpreting Phase III data it is imperative to note the high-alumina cement porous cencrete was cured using a light water fogging. This represents optional curing i for calcium aluminate concrete and likely is not representative of curing that was

perfonned in the field. The significance c,f curing is apparent when considering that nenq of the waters exiting from the mock-up structure in the Phase III work were noted to have j residue or dissolved solids. It was even reported that exit water spilled on the laboratory

i. floor and allowed to dry left nQ residue. This is in contrast to Phase I and II work where

significant residue was produced during test. Thus, it is concluded that the Phase III mock-up cannot be used to predict degradation due to chemical attemative or attack on the i high-alumina cement porous concrete, but can only be used to estimate potential strength I degradation due to conversion of high-alumina cement.

Concrete strength was measured using two methodologies in Phase III. Unconfined compressive strengdi was measured on cores extracted from the mock-up structure in

general conformance with ASTM C 42, Standard Test Methodfor Obtaining and Testing Cores and Sawed Beams of Concrete. This is a well established and recognized method of evaluating in-place concrete strength. While testing cores extracted from a stmeture j 37

I does not provide a direct measure of in-place concrete strength, it does provide usable data for evaluating the suitability of the in-place concrete strength, and is routinely used in the industry to assess the adequacy ofin-place concrete. The Building Codefor Requirementsfor Reinforced Concrete Structures (ACI318) and the Code Requirements for Nuclear Safety Related Concrete Structures (A CI 349) both tefen:nce testing unconfined cores to evaluate suitability of in-place concrete. Both of these codes have the identical provision for evaluating core strength, i.e, core strength shall be considered acceptable if the average of three tests is at least 85% of specified design strength and no single core strength is less than 75% of specified design strength.

The other methodology used in Phase III to assess concrete strength was to test concrete cores extracted from the stmeture in compression while under some degree of lateral confinement. This is an atypical test method that requires a more sophisticated test set-up and a more critical an dysis of the resulting data. While it can be argued that a compression test with some degree of lateral confinement (biaxial compression) is more closely related to the condition in the stmeture, there are comparatively little data on these types of tests to establish appropriate evaluation and acceptance criteria. The manner in which the confined tests were run adds to the difficulty in interpreting the data. In a well controlled biaxial compression test the amount oflateral restraint is known, controlled, and uniform. Lateral restraint for the tests performed in Phase III was provided by enclosing a paper-wrapped porous-concrete 6-in diameter test specimens in a 7-in. ID diameter Schedule 40 pipe and filing the annular area between the pipe and test specimen with a molten sulfur. based compound normally used to cap compression test specimens.

While this system does provide lateral restraint, the magnitude of the restraint and its uniformity are not known. At best, it can only be assumed that the lateral restraint remained constant for all the test specimens. However, in any case, the magnitude of the lateral restraint is unknown.

Given the conditions and limitations enumerated above, it appears that the most meaningful analysis of the data is a determination of potential trend in compressive strength due lo exposum io the flow conditions and aging established in 1he Phase 111 work. This trend analysis does not form the basis for establishing causation in change in compressive strength (this is discussed in other sections of the report), but only addresses the correlation betwcen exposure to flow and change in compressive strength. It is important to note that correlation does not necessary imply causation.

Reviewing the ARL data for the portland cement porous concrete it is apparent that there is no significant change in compressive strength or density of the unconfined compressive 38

1 strength specimens over the duration of the test. Compressive strength tests of the j confined portland cement porous concrete yielded somewhat erratic results with no clear trend with respect to number of periods of flow exposure. Density of the unconfined test l specimens was totally unrelated to test duration. This suggests that the erratic confm' ed compressive strength measurements are more likely the result of scatter in the test data, due to the test method, rather than any correlation with test duration or number of periods of flow exposure. Clearly, the compressive strength or density of the portland cement l porous concrete was not affected by exposure to flowing water as maintained in the Phase l

III work. I Linear regression analysis was performed on the unconfined and confined compressive strength data for the calcium aluminate porous concrete. The data used in these analyses was taken directly from the ARL work and is presented in Table 13. Figure 9 gives the I results of the linear regression analysis of the unconf'med compressive strength data while Figure 10 gives the results of the confined compressive strength data. Several salient observations can be made from these analyses. For both test conditions, i.e. unconfined and confined, there is a statistically significant trend of lower measured compressive strength with test duration. However, the correlation between compressive strength and  ;

periods of flow is somewhat weak. As measured by the R2 statistic, the correlation between compressive strength and periods of flow was 0.27 for the unconfined test and 0.40 for the confined test. In somewhat simplistic terms, it can be said test duration on 1 the number of periods of flow can account of 27% and 40% of the trend for decreasing compressive strength for the unconfined and confined tests respectively. If the correlation were perfect, the R2 statistic would be 1.0. The strength of the correlation can also be  ;

assessed by noting the scatter in the measured test data from the predicted line shown for each regression analysis. The regression analysis also yields a value of decrease in compressive strcngth per period of exposure. For the unconfined tests compressive strength decreases on average 32 psi per period, while for the confined tests compressive strength decreases on average 35 psi per period. These predicted decreases in compressive strength per period represent a best fit to the available test data. Obviously, some test data suggest a greater decrease, while other data suggest a smaller decrease. To account for the scatter in the data it is possible to calculate values for reduction in compressive strength per period that will encompass a given percertage of anticipated values. This was done for both test conditions using a 95% confidence interval. A maximum reduction of 64 psi with a minimum reduction of 0 psi per period will encompass 95% of all anticipated test data for the unconfined test conditicns, while a maximum reduction of 61 psi with a minimum reduction of 9 psi per period will 39

(- . _ . - . - - . - - - . - - - . - - _ _ - . - - - - - -

i i,

4

. encompass 95% of all anticipated test data for the confined test condition. The fact that both the unconfined and confined compressive strength data have very similar correlation i to test periods is additional confirmation that correlation between these parameters exists and modeling by regression in appropriate and meaningful.

~  !

Applying these test data and the resulting model to the stmeture in the field is problematic for several reasons. In order to evaluate the impact of decreasing compressive strength on 4

the field structure a good value for either initial in-place strength or current in-place 1 strength is needed. Neither value is available. During construction compmssive strength was only measured on test cylinders that were iabricated at time of construction. These

test cylinders viere compacted in a manner greatly different than used when placing p concrete in the stmeture, thus their measured strength cannot be assumed to be the same as in-place strength. To date, no cores have been extracted and tested from the as built porous concrete. Testing cores extracted from the porous concrete would provided the i most reliable measure of current in-place compressive strength.

l Because it is probable the only degradation mechanism active in the Phase III mock-up is

, conversion, additional insight into the magnitude of this effect can be obtained by l considering the conversion analysis conducted on mock-up Core Sample 9M111-7. This analysis revealed that conversion was between 65 and 83% complete in the Phase III j mock-up at test completion, ie. after 15 periods of flow exposure. Neville (1966)'

! indicates that at completion (100%) conversion a high-alumina cement concrete with a  ;

! water-to-cement ratio of 0.24 to 0.35 will lose approximately 40% of its unconverted l 1 i i strength. This agrees well with the test data from the mock-up when considenng an initial strength of 1520 psi, a loss of 32 psi per period, and approximately 75% complete

conversion (at test completion) results in a calculated loss of 42% of initial strength at i

complete (100%) conversion).

l 1 i Initial compressive strength 1520 psi (intercept) l l

Reductions per period 32 psi f

Number of periods 15 A

l Reduction of end of test 480 psi l  % Conversion 75 %

j Conversion at 100% (480/0.75) 640 psi Reduction in strength (640/1520 42 %

40

f Given these data, it is reasonable to assume that when, and if, complete conversion occurs in the as-built stmeture, the porous calcium aluminate cement will retain approximately ,

60% ofits original strength, i Without extracting samples from the field it is very difficult to predict to what degree ,

i

' i conversion has taken place in the field. It may, however, be possible to estimate an upper j

bound limit for potential near-term strength loss in the as-built stmeture from the mock-up l test data. Neville (1996) reports that for a water-to-cement ratio of 0.27 to 0.40 after 5 years the average degree of conversion will be 30%. Recognizing in the mock-up test 4 approximately 75% conversion occurred after 15 test periods we may assume 30%

conversion would occur after 6 test periods. Applying the upper bound (95% confidence l interval) limit of 64 psi per test period, a maximum near-term strength loss due to conversion over the next five years of 380 psi is obtained.

Based on the test data from the Phase III work it can be concluded that the anticipated total 1

j maximum loss in compressive strength due to conversion of the porous high-alumina

concrete will be on tiie order of 640 psi, while an upper-bound estimate for potential l strength loss over the next five years is 380 psi. At this time, and with the currently available data, projections beyond five years are not warranted or prudent. Furthermore, I

it is important to note that conversion is a phenomenon with an end-point, it does not i continue indefinitely, therefore extrepolating these data to predict losses greater than 640 psiis not appropriate.

Several issues should be examined in much greater detail in order to develop a more j realistic model to predict in-place compressive strength of the porous high-alumina cement concrete. First, and most important, is the extraction and testing of cores removed from l the actual structure. This can establish a reliable value for current in-place compressive strength and should provide samples that can be examined by various destructive and non-destructive methods to better understand if any deterioration mechanism is active and its basis. Once a conclusive field deterioration mechanism is established, a more realistic assessment can be made of the applicability of the mock-up tests and their interpretation.

Until these tasks are completed, it is neither pessible, nor prudent, to estimate long-term strength potential (in excess of five years) of the porous calcium aluminate cement concrete.

s 41

A FINDINGS I

! Based on quantitative laboratory analysis of water and residue samples obtained from the field and valuable, transferable qualitative observations from the mock-up tests, CFL 5

, believes that the root-cause mechanism for degradation taking place is multifold. The i i combination of the two cements in the substructure " sandwich" proved to be unfortunate

indeed. The portland cement provides a source ofleachable ions which in conjunction

, with other sources (ground water and aggregate) of the same (and other) ions attack the '  !

Lumnite@. At least four modes of attack likely are occurring: leaching oflime and alkalies t from the portland cement; leaching of alkali and other ions by rain water percolating

[ through the ground outside the plant; conversion of the hydrated Lumnite@ cement; and high pH dissolution of calcium aluminate hydration products in the Lumnite@ concrete.

l l Another attack mode, ettringite formation, a mode not documented well for HAC in the i literature, is suggested by CFL as likely since Lumnite@, although designed for good f sulfate resistance in low lime environment, is prone to sulfate attack in the presence of

j. large amounts oflime. Ettringite formation in a high lime and high pH environment can l be expansive and may aggravate the attack on the Lumnite@ porous concrete.

1 j There is no indication that cement residue per se is collecting in the weirs. Thus whatever

^

)

reaction is taking place in the field structures seems to occur mainly in the water or on the i surface of the hydrated cements. This is consistent with a congruent dissolution of the l

calcium aluminate or aluminum hydroxide. Neither of the two hexagonal hydrates nor .

j hydrogamet were present. Conventional conversion calculations were not possible except.

by inference.

Analysis of the water in well borings located outside the plant indicates that the waters are mostly very aggressive towards portland cement matrices. Three corrosion indices  ;

calculated for the waters lead to the same inference. No such indices are available for the Lumnite@ cement. The assumption that the water emanates mainly from outside the structure seems reasonable since the amount of flow appears to be proportional to the amount of rain in a year and plant expenence has been that the the flow increases 3 te4 days after a heavy rain. The actual flow path has not been determined.

Based on geologic descriptions and three dimensional representations of the concentrations of the most important chemical elements it appears that the water can travel through the artificial fill overburden and through sand lenses in the glacial till to reach the vicinity of the reactor building. The basal till and the bedrock are fairly impermeable, but there are some fissures and foliation faults which could also provide a path for the water.

42

i Significant levels of sulfate, sodium, potassium, chloride and calcium can therefore enter the Lumnite@ porous concrete structure through postulated leaks in the rubber membrane.

. It should be noted that some of the concentrations would not normally be considered high when dealing with monolithic concretes, but may be highly reactive in the porous concretes. They become even more significant when they are integrated over the years and they take part in precipitating compounds which have an extremely low solubility product in the high pH environment inside the porous concrete. In effect there may not exist a threshold level for the reaction. Also the alkalies may take part in a self-regenerating cycle as described by Neville.

)

Millstone estimates for the average flow of water have ranged from 200 gpd to as much as .

1600 gpd. The plant is still refining methods for calculating the flow. CTL calculated, j using some simplifying assumptions (based on the almost complete depletion of the -

sulfate in the water), that 12,000 to 60,000 lb. of residue can be formed in the structure every ten years. Of this it is known that only about 1,000 lb deposits in the weirs and sumps. CTL has also pointed out that relatively large amounts oflime and alumina not used in the formation of the residue can escape as soluble ions in the exit waters. CTL has i provided a method to calculate these amounts when more is known about average flows i and concentrations.

Mock-up tests encountered common difficulties in accelerated simulation of a complex cementitious system. Additionally other controllable parameters were not controlled.

Aggregate-cement ratios were not close to those in the "as-built" structure. This was particularly true for Phase I and II. Aggregate type was changed during the tests.

Analyses of entering and exiting waters were either incomplete or absent. The temperature history of the concretes was not followed. Curing was not similar to the "as-built" condition except perhaps in Phase II. The mock-up construction practice of placing each l concrete in 3 to 5 inch layers tended to produce horizontal coldjoints, a condition not typical of the "as-built" monolithic condition. The thickness of the Lumnite" seal grout on '

top of the Lumnite" porous concrete was 2-in, vs.1 /8 to 1/2-in. in the "as-built" structure.

The absence of any residue formation and the absence of dissolved ions in the exit waters ,

I in Phase III (due to atypical optimal curing) makes it particularly hard to translate the best available strength data to the "as-built" com.tmetion where residue formation has been prevalent. Taken in total, these difficulties make a quantitative translation of strength data to the Millstone situation tenuous indeed. However, semi-quantitative data were developed.

43

1 l

Statistical analysis of the compressive strength data from the Phase IU mock-up study was performed and a linear model developed that addresses only strength degradation due to conversion. In the mock-up study, a trend of decreasing compressive strength with time was obtained. Althout,h applying these test data and the resulting model to the stmeture in the fie'd is problematic for several reasons, a rough first-order estimate of potential loss in compressive strength in the structure of 380 psi over the next five years to 2002 is suggested. Due to the uncertab y of these data, extrapolation to longer time periods is not warranted. Furthermore, because there are no reliable data for cu rent in-place strength of the high-alumina cement porous concrete, it is difficult to predict future absolute in-place strength of the porous concrete in the structure. Again, as a rough first-order estimate,it may appropriate to estimate the in-place strength in the structure using the lowest I measured compressive strength in the mock-up study. None of the mock-up studies provided sufficient data to predict potential strength loss due to chemical attack or alteration. Strength degradation for these faecors would be additive to strength loss from conversion.

Given the uncertainty in these predictions, it is critical that cores be extracted and tested from the actual structure. This can establish a reliable value for current in-place I compressive strength and should provide samples that can be examined by various destructive and non-destnictive methods to better understand any deterioration mechanism and its basis.

CTL recommends that NUSCO proceed with the planned coring of the actual concrete at Millstone ifit can be done safely. The coring operation should be carefully planned and executed to ensure that statistically valid samples are obtained. Documentation and immediate protection of the samples before exposure to air are of prime importance.

l It seems reasonable that due :o the low f!ow rate of water with respect to the volume of available space in the porous concrete the conditions inside the concrete are quite l

homogeneous. The residence time is long und a near equilibrium is achieved. Due to the relatively gmater flow in and near the porous dramage pipe, the conditions there may not be as uniform.

It is CTL's opinion that the loss of calcium from the hydrated portland cement has not l seriously impaired the strength of the portland cement porous concrete. A large proportion l

l of the calcium could be supplied by the ground water and a significant amcunt could come from the degraded Lumnite@ cement.

44

1

! It is CFL's opinion that the ten-foot thick massive concrete base mat is not at risk of

chemical attack to any significant extent.

It should be noted that NUSCO was cooperative in providing samples and required information to CTL in a timely manner whenever possible. CTL greatly appreciates this ,

i assistance.  !

The preponderance of the main data for this report was developed over a short period of l time from the middle of January to the middle of March,1997. CTL provides this report )

as a best effort considering time constraints and available data.

! Acknowledgement: i

i The contribution of the following CFL staffis acknowledged

Don Broton (XRF),

JoAnne Delles (chemical analysis, test method specification for waters and residues),

doward Kanare (liaison with ABB-CE, Millstone sampling), Laura Powers-Couche, (petrography), Ella Shkolnik (thermal analysis), and Fulvio Tang (XRD).

4 l l i

4 d

i 4

d i

45 i

4

?

LIST OF REFERENCES i

i Alexander, Mark G., (1993) " Dissolution of No-Fines Concrete Slab Due to Soft-Water Attack," Journal ofMaterials in Civil Engineering, Vol. 5, No. 4, Nov., pp. 427-435.

Alexander, Mark, Addis, Brian, and Basson, Jack, (1994) " Case Studies Using a Novel Method to Assess Aggressiveness of Waters to Concrete," ACIMaterials lournal, March-April, pp. 188-196.

i " Atlas Lumnite & Refcon, Calcium Aluminate Cements," (1987) Bulletin 1-01, July, i

Lehigh Cement Company, Allentown, PA.

1 i Basson, J. J. and Addis, B. J., (1992) "An Holistic Approach to the Corrosion of 4 Concrete in Aqueous Environments Using Indices of Aggressiveness," G. M. Idorn IntemationalSympo.sium, Holm, Jens and Geiker, Mette, Eds., American Concrete Institute, SP-131, pp. 33-65.

Basson, J. J., (1989) " Deterioration of concrete in aggressive waters - measuring

, aggressiveness and taking countermeasures," Portland Cement Institute, Concrett-Durability Bureau, Midrand, South Africa,22 pages.

Brown, Paul Wencil, (1987) "Early Hydration of Tetracalcium Aluminoferrite in Gypsum

and Lime-Gypsum Solutions," Journal of the American Ceramic Society, Vol. 70, No. 7, 1 pp. 493-496.

Brown, Ronald A. and Cassel, Bruce, (1977) "High alumina cements Background and Application of Thermal Analysis Methods," American Laboratory, January, pp. 45-56.

Ding, Jian, Fu, Yan, and Beaudoin, J. J., (1996) "Effect of Different Inorganic Salts / Alkali on Conversion-Prevention in High Alumina Cement Products," Advanced

Cement Based Materials, April, pp. 43-47. j

)' Drew Chemical Corporation, (1988), Drew Principles of Industrial Water Treatment, pp. i

'243-244.

Fu, Yan, Ding, Jian, and Beaudoin, James J., (1996) " Corrosion Protection of Reinforcement in Modified High-Alumina Cement Concrete," ACIMaterialslournal, November-December, Vol. 93, pp. 609-612.

. i Haines, Peter J., (1995) " Thermal Methods of Analysis, Principles, Applications and Problems, " Blackie Academic & Professional, Glasgow, Scotland, U.K., pp. 101-104.

Jambor, Jaromir and Skalny. Jan, (1996) "Another look at the deterioration of calcium aluminate cement concmte," Materiales de Construccion, Vol. 46, No. 241, Febmary/ March, pp. 5-20.

Lach, V., (1987) "The Deterioration of Alumina Cement Concrete," Concrete Durability Katharine and Bryant MatherInternational Conference, John M. Scanlon, Ed., American Concrete Institute, SP 100-97, Vol. 2, pp.1903-1914.

Lea, F. M., (1970) The Chemistry of Cement and Concrete, Third Edition, Che.mical Publishing Co., Inc., New York, NY.

46

i l

Mangabhai, R. J., Ed., (1990) " Calcium Aluminate Cements," Proceeedings of the l e International Symposium, held at Queen Mary and Westfield College, University of ]

London, July 9-11, E. F.N. Spon, An imprint of Chapman and Hall, London.  ;

Mehta, P. K., (1964) " Retrogression in the Hydraulic Strength of Calcium Alurninate Cement Structures," Minerals Processing, Nov., pp.16-19.

I Netter, W. S., (1975) " Working with Calcium Aluminate Cement Concrete," Plant Engineering, October 30, pp. 51-52.

i-i Neville, A. M.,(1996) " Physical properties of high-alumina cement," Properties of Concrete, Fourth Edition, John Wiley & Sons, Inc., New York, NY, pp.93-102.

t- Neville, A. M., (1975a) High Alumina Cement Concrete, The Construction Press Ltd.,

Hornby, UK, John Wiley & Sons, Inc., New York, NY.

l

, Neville, A. M., (1975b) "Is High-alumina Cement a Satisfactory Structural Material?"  !

. Canadian Journal of Engineering, Vol. 2, No. 4, December, pp. 373-380.

I l

Pommersheim, Janes and Chang, Jemei, (1988) " Kinetics of Hydration of Tricalcium  ;

i Aluminate in the Presence of Gypsum," Cement and Concrete Research, Vol.18, No. 6,

, pp. 911-922. '

l Ost, B. W., (1997) " Sulfate Resistance of Ettringite Cements," submitted for publication to World Cement.

i

~

Popovics, S.,(1987) "A Classification of the Deterioration of Concrete Based on Mechanism," Katharine and Bryant MatherIntemational Conference , John M. Scanlon, Ed., American Concrete Institute, SP-100, Vol.1, SP 100-10, pp.131-42.

Robson, T. D., (1962) High-Alumina Cements and Concretes, Contractors Record Limited, London, U. K., and John Wiley & Sons, Inc., New York., NY Schwiete, Hans Ernst and Ni 1, Egid M. G., (1965) " Formation of Ettringite Immediately After Gaging of a Portland Cement," Journal of The American Ceramic Society, Vol. 48, No.1, pp.12-14.

Sorrentino, Danielle, Sorrentino, Francois, and George, Mike, (1995) " Mechanisms of Hydration of Calcium Aluminate Cements," Materials Science of Concrete IV, Skalny, Jan and Mindess, Sidney, Eds.,' Die American Ceramic Society, pp. 41-90.

Taylor, H.W.F., (1990) " Calcium aluminate, expansive and other cements," Cement Chemistry, Academic Press, London and San Diego, CA, pp. 316-335.

47

__m ._m _ . _.m_ _ _ . _ _ . . _ _ _ m .. . _ _ . . . . _ . _ _

t TABLE 1. AGGREGATE / CEMENT AND W/C RATIOS IN POROUS CONCRETES cenere so cement, ai. Aggressee, tse Arc vet none we wt.neue weeer, ese weiericement nodded Ratio Denmar per Orighel Desigm Pertiend Coment 447 2638 6.38 5.90 145.9 0.326 Original Design Celclunf Aluminase 447 2835 6.85 6.34 145.9 0.326 Milletone PC, one soledled mit (trom 231-101) 3936 16119 4.43 4.10 1569.0 0.399 123.0 Milletone CA,5 mhree 9120 pel 3948 16119 4.41 4.08 1271.0 0.322 124.8 Mmetone CA,3 rnitee 4120 pcf 3948 18119 4.41 4.08 1319.0 0.334 114.5 Phase i Mache PO 8-10-92 2012 12107 6.50 6.02 624.8 0.311 110.0 Phase 1 Mockup CA 6192 447 2835 6.85 6.34 133.5 0.299 117.7 '

Phase i Mockup CA 6 $2-92 1410 9135 7.00 6.48 412.9 0.293 116.2 Phase R Mockup PC 8-28-93 2920 17400 6.83 5.96 958.8 0.328 123.8 Phase 5 Mockup PC 8-28-93 3014 17400 6.61 5.77 989.8 0.328 125.8

• Phase Il Mockup CA 7-1-93 1410 8960 7.28 6.35 463 0 0.328 125.0 Phase 11 Meekup CA 7-143 1410 8975 7.29 6.37 463.0 0.328 126.0 Phaeo H Mockup CA 7143 1410 8975 7.29 6.37 463.0 0.320 123.0 Phase NI Mockup PC 560 2670 5.46 4.77 228.6 0.408 127.6 Phase NI Mackup CA 564 2670 5.42 4.73 198.1 0.351 129.2 Rev. 4-7-97

• Estimated denetties of 3.02 for Lumnte,3.15 for Portland Cement, and for aggregates per Dean Whao, AHL: 2.915 for Phase I and Milletone, 2.75 for Phases II seul Nl. (The hoevier rock wee frorn Wellmgford. The lose dense rock vres from Wauregon).

NOTESt Regarding a gg egetWeement ratios- Large change in MWone mix from origbe! desie. Phase I and H close to that design. Phase et intermediate Original design mht taken from page 2-2 of document number u-12179-867d (deted 12/17/927)

Coment, aggregate, and water weights for Millstone Lumnte concretes are estimer es, date for e8 mhies nol available A8ittstone concrete done#ies are from fresh unk weights. PC concrete densities as high as 140 pcf indicate OC problems?

Concrete densities for enock-up teste are from cast cylinder teste (eges 7 to 28 days). .

r

. - _ _ _ _ _ . _ _ . _ _ _ _ _ . ________.____.._...____.________________.__..__________-.m_ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ . _ _ . _ _ _ _ _ . _ _ _ _ _ . _ _ _ - _ _ . _ _ _ _

TABLE 2. C/A RATIOS FROM MILL REPORTS Report kampleID Ca0 Al2O3 CaO/Al2O3 C/A Date  %  %  % Wt. Ratio Mole Ratio 3/24/75 Lumnite Mill Report (Universal Atlas) 34.20 47.20 0.72 1.32 6/8/93 Lumnite Mill Report (Lehigh) 34.16 50.93 0.67 1.22 3/13/95 Lumnite Mill Report (Lehigh) 35.83 50.22 0.71 1.30 11/26/96 Lumnite CTL Report (CTL ID 922971) 35.78 46.80 0.76 1.39 4/4/75 type 11 Pittsburgh Testing Lab (Atlantic) 63.37 4.89 12.96 23.56 6/10/75 Type 11 Pittsburgh Testing Lab (Atlantic) 62.53 4.49 13.93 25.32 7/27/77 Type 11 Pittsburgh Testing Lab.(Atlantic) 61.88 4.66 13.28 ~ 24.14 3/15/93 Type Il Mill Report (Lehigh) 63.80 4.50 14.18 25.77 3/14/95 type II LA. (Blue Circle Cement) 63.40 4.70 13.49 24.52 CaO not shown on PTL reports, Ca0 calculated by CTL Lehigh Lumnite Reports may include TiO2 in Al2O3 valus

l I

TABLE 3. ESTIMATE OF AMOUNT OF CEMENT DISSOLVED SAMPLE # ALUMINA IN ALUMINA IN WOULD NEED WOULD NEED RESIDUE, % RESOUE, LBS LUMNITE, LBS. TYPE II, LBS 3 12.47 124.7 244.8 2723 ,

4 5.71 57.1 112.1 1247 5 9.77 97.7 191.8 2133 AVE. 9.32 93.2 182.9 2034 ASSUMPTIONS:

CEMENTS DISSOLVECONGRUOUSLY ALL ALUMINA CAME FROMTHE CEMENT NO ALUMINA ENTERED WITH ENTRANCE WATER NO ALUMINA WAS LOSTIN THE EXIT WATER 1000 PO8)NDS OF DRY SEDIMENT COLLECTED SINCE 19XX 4

. ___ . . . ~ ... - . _ . . . . . . . _ . - - - _ . - . . .. .. . . . . . -. ~ ._ . . . . . _ . . ..

s TABLE 4. CALCIA TO ALUMINA RATIOS Report SampleID Ca0 Al2O3 CaO/Al2O3 CIA Date  %  %  % Ratio Mole Ratio 7/29/87 *tSF SUMP

• 19.87 10.01 1.98 3.51 9/19/91

• SUMP FLOOR

• 54.00 <1 N/A N/A 7/9/96 3. Pump Side Sump 7A O/S Wier 36.95 12.47 2.96 5.39 7/9/96 4. Ground Water Side Sump 7A I/S Wier 43.18 5.71 7.57 13.76  !

7/9/96 5. Sump 7B taken from drum 42.80 9.77 4.38 7.97 7/0/96 6. Basement Mockup Samplo from U3 35.83 13.38 2.68 4.87 r

Rev. 8-9-96 e

L i

i

_ . . ~ . . _ _ _ _ . _ _ _ _ _ . _ _ _ _ _ _ _ . _ _ _ m . _ _ _ _ _ . _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ . _ _ _ _ _ _ _ _ . _ _ _ _ _

TABLE 5. CALCIA TO ALUMINA RATIOS LIQUID SAMPLES Report Sam le ID* CaO l Al2O3 CaO/Al2O3 CIA Date  %  %

l  % Ratio Mole Ratio 7/9/96 1. AdC Surnp 7A-34 (liquid) 0.0579 0.0003 203 369 7/9/96 2. B&D Sump 78-34 (liquid) 0.0004 0.0022 0.20 0.37 7/9/96 Water leach of 3. 1.3134 0.6747 1.95 3.54 7/9/96 Watet teach of 4. 2.1757 1.5943 1.36 2.48 7/9/96 Water teach of S. 1.0097 0.5905 1.71 3.11 7/9/96 Water leach of 6. 0.8350 0.6922 1.21 2.19 8/14/96 SUMP 7A single weir (south) 7/8/96 0.0507 0.0004 122 222 8/14/96 SUMP 7A double pipe weir (west) 7/8/96 0.0185 0.0097 1.91 3.47 8/14/96 SUMP 7B S weir (single pipe) 7/8/96 0.0457 0.0001 484 880 8/14/96 SUMP 78 NW Weir (2 pipe) 7/8/96 0.0202 0.0137 1.47 2.68 8/14/96 SRW SUMP 1 (by diesel) 7/10/96 0.0094 0.0000 1238 2251 0/14/96 Under drain Sump NE cnr ESF Bldg 7/11/96 0.0077 0.0000 340 618 8/14/96 Elec. Pipe Chase NE cnr Wste Bldg 7/11/96 0.0018 0.0000 118 214

• ** . e *g

i l

9 i

TABLE 6A. CALCIA TO ALUMINA RATIOS . ?D SAMPLES XRF DATA ,

1 Report SampkrID hO Al2O3 CaO/Al2O3 C/A  ;

j Date  %  %  % Ratio Mole Ratio f

a 8/14/96 15. Sump 7B (Drum) 44.50 9.40 4.73 8.61 8/14/96 16. Sump 7A SW Pipo 33.50 9.50 3.53 6.41 8/14/96 17. Sump 78 West Pipe 37.20 13.00 2.86 5.20  !

8/14/96 18. Sump 7A Outside Weir 42.30 13.90 3.04 5.53 8/14/96 19. Sump 78 South Pipe 30.70 7.30 4.21 7.64 8/14/96 20. Sump 7B North Pipe 37.90 9.70 3.91 7.10 8/14/96 21. Sump 7A Northwest Pipe 41.10 7.80 5.27 9.58 l 8/14/06 22. Sump 7A South Pipe 33.00 14.10 2.34 4.25 l

1 TABLE 68. COMPARISON ABB XRF AND ICP DATA

$ l Report Sampio ID Cao,% Cao,% AL203, % AL203, % l Date W ltP w ICP e

8/14/96 15. Sump 78 (Drum) 44.50 36.78 9.40 5.35 8/14/96 16. Sump 7A SW Pipe 33.50 32.60 9.50 9.03 8/14196 17. Sump 78 West Pipe 37.20 26.63 13.00 3.63 8/14/96 18. Sump 7A Outside Weir 42.30 33.16 13.90 12.64 8/14196 19. Sump 78 South Pipe 30.70 28.03 7.30 4.53 8/14/96 20. Sump 78 North Pipe 37.90 40.33 9.70 14.94 8/14/96 21. Sump 7A Norttraest Pi;;e 41.10 38.15 7.80 7.52 8/14196 22. Sump 7A South Pipe 33.00 31.10 14.10 13.53 SUNt 300.20 274.77 84.70 71.17 AVERAGE: 37.53 34.35 10.59 8.90

)

Rato XRF/ICP: 1.0926 1.1899 l l

i I

i

I I 4

TABLE 7. CALCULATION OF POUNDS OF OXIDE LOSS OVER THE YEARS IN EFFLUENT WATERS

^

POUNDS OXIDE LOST = A x B x C x D x E x F x G

)

WHERE:

A = PPM OF ELEMENT FOUND BY ANALYSIS OF WATER l i

i

, B = 1/1000000 = CONVERTS FROM PPM TO FRACTION OF ELEMENT IN  !

LIQUID 1

C = (MOLECULAR WEIGifI' OF OXIDE)/((NUMBER OF ATOMS OF ELEMENT CONSIDERED IN OXIDE) x (ATOMIC WT. OF ELEMENT)) j FOR Ca0 THIS IS 56.08/(1 x 40.08) = 1.3992 FOR Al2 O3 THIS IS 101.96/(2 x 26.98154) = 1.88944 i D = 8.337 POUNDS = AVE. WT. OF A GALLON OF WATER WITH SMALL l

4 AMOUNT OF DISSOLVED SOLIDS l l

E = GALLONS OF FLOW PER DAY F = 365 DAYS PER YEAR G = NUMBER OF YEARS OF WATER FLOW l l

l FOR EXAMPLE, LET US CONSIDER TIIE 7/11/96 SAMPLE j UNDER DRAIN SUMP NE CORNER ESF BLDG (ASSUMING 1000 i 1

GALLONS PERDAYFORTEN YEARS):

Ca0 = (55.05/1000000) x 1.3992 x 8.337 x 1000 x 365 x 10 = 2344 LBS.

Al2O3 = (0.12/1000000) x 1.88944 x 8.337 x 1000 x 365 x 10 = 6.90 i LDS. I

TABLE 8. CIO AND Al2O3 LOST WITH WATERS Report SampleID Ca PPM AI,FPM FLOW, GPD YEARS Ca0 Al2O3 Date OF FLOW POUNOS POUPDS 7/9/96 1. A&C Sump 7A-34 (liquid) 6/29/96 413.62 1.51 1341.6 10 23,627 116 7/9/96 2. 940 Sump 78-34 (liquid) &29S6 3.20 11.62 273.6 10 37 183 ,

2/24/97 7A Aump 2/5/97 307.45 11.98 1684.8 to 22.055 1.160 2/24!97 78 Sump 2/5/97 134.65 68.27 585.6 to 3.357 2.299 S/14/96 7A 8. 7/8/96 362.68 2.20 866.8 10 13.416 110 8/14/96 7A b.W. 7/8/96 132.21 51.24 472.8 10 2,661 1.393 8/14/96 78 8. 7/8/96 326.76 0.50 0 10 0 0 ,

8/14/95 B N.W. weir 7/8/96 144.58 72.83 273.6 10 1,6E i 1.146 2/24/97 7A $. welr 1/20/97 403.80 1.03 842.4 10 14,883 50 2/24/97 7A N.W. weir 1/20/97 144.85 54.35 842.4 10 5,195 2,632 2/24/97 78 8. weir 1/20/97 3.65 24.89 0 to 0 . 0 2/24/97 7B H.W. we!r 1/20/97 120.19 66.13 585.6 10 2,997 2,227 2/24/97 7A d. weir 2/5S7 416.06 1.35 842.4 10 14,923 65 2/24/97 7A N.W. weir 2/5/97 130.16 49.19 842.4 10 4,669 2,382 2/24/97 78 5. weir 2/5/97 0 10 0 0 2/24/97 7B N.W. weir 2/5/97 160.35 70.82 585.6 10 3,998 2.384 Totals for sumps 6/29/96 23684 299 ,

Totals for sumps 2/5/97 25412 3459 lotals for weirs 7/8/96 17762 2648 Totals for weirs 1/20/97 22855 4909 Totals for weirs 2/5/97 23590 4832 r

No semp liquid sample taken on 7/8/96,1/20t97 78 somp liquid taken on 2/5/97 may have been compromised by pump turNng on 7-8-96 flows used for 1996 sampres.119-97 !!ows trsed for 1997 samples 2/5/97 Ca art:1 Al values are from "inside weir" samples '

, r

_ _ _ _ .-. _.___.._.______._________.__n _ _ _ _ _ _ _ - - _ _ _ _ _ . _ _ _ _ _ _ _ . - _ . . _ _ _ _ _ _ _ __ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ -- ____ - __m_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ . -_

.____m

[ TABLE 9. Ca0 AND Al2O3 FROM GROUND WATERS Se @ ID ' Dete Ca, PPM A! PPid  !

i H.OW.GPD YEARS Ca0 - Al2O3 -

i(CTL, GEI, BORE HOLE NO.) l Sampled l 1/16197 OF FLOW POUPOS POLNt3B 923377 gel 96 b1A 1/15/97 87.5 0.5 2318 10 8.638 SS 923379 gel 96 02A 1/15/97 31.2 1.8 2318 10 3.074 236 9233B1 gel 96 03A 1/15/97 46.8 0.2 2318 10 4.618 30 S23383 GEt96 03B 1/15/97 101.8 0.4 '2318 10 10.046 57 923385 gel 96 03C 1/15/97 26.9 0.1 2318 10 2.650 18 923397 GE!96 04A 1/15/97 41.6 5.3 2318 10 4.105 700 923%9 gel 95 040 1/15/97 29.7 1.8 2318 to 2.926 236 l92339: GEI96 OBA 1/15/97 43.1 1.9 2318 to 4,743 255 923393 gel 96 OSB 1/15/97 2S.0 1.0 2318 to 2.564 128 '

??3'WS gel 96 08A 1/15/97 74.6 0.03 2318 10 7.364 4 923397 gel 96 Trbck 32 7 0.4 0.01 2318 1D 39 1  ;

j923399 GEi96 Trtsck 33 7 0.4 0.01 2318 10 '

39 1 l AVEHAGEfTHUCRS NOTINCLUDFD) 51.40 1.30 '2318 to 5,073 173  !

i 1

i L

i d

.t i.

e I

___. . _ __ . _ _ . _ __..._..m . _ _ _ _ . _ . . ._. . _ _ . _. . . . ._ _ _ _ . _ . . . . ..

h TABLE 10. Ca0 AND Al2O3 FROM GROUND WATERS Nample 10 Dete Ca, PfW AI, PPM FLOW. GPD YEARS Cs0 Al203

((CR, G8Et, BCCE HOLE NO.) Sempted __

1/19/97 OF FLOW POUNDS FOL%DS 923377 GEt96 01A 2/14/97 87.4 0.9 2270.4 10 8.449 119 92337S GE'96 OtA 2/14/97 83.7 2.5 2270.4 10 8,091 326 923381 GE!98 03A 2/14/97 38.2 0.1 2270.4 10 3,693 9 923383 GEi96 Ob8 2/14/97 149.9 0.3 2270.4 10 14.404 40 923385 GE!96 03C -2/14/97 0.0 0.0 2270.4 to 0 0 923387 GEI'M 04A 2/14/97 51.9 4.1 2270.4 10 5.017 554 923389 gel 96 C4B 2/14/97 121.0 0.4 2270.4 10 11,697 56 923391 GEI93 OBA 2/14/97 143.0 0.2 2270.4 10 13,824 25 923393 gel 96 0%B 2/ t 4197 34.3 0.8 2270.4 10 3.316 10t r 923395 gel 96 OBA 2/14/97 126.0 0.21 2270.4 10 12.180 27 Rain 1/25/97 1/25/97 0.0 0.45 2270.4 10 0 59  !

Rain 2/1U97 2/14/97 5.0 0.01 2270.4 1 0 _ __ __ 483 1 l AVERAGE fRAIN NOTINCLUDED) 83.45 0.95 2270.4 -10 8,r 67 124 l

)

t

. - . _ _ - - _ . . . - - . _._ m _ , _ _ . - _ . .._____________m_ _ _ _ _ . _ _ - . _ . _ . _ . _ . _ . _ _ _ _ _ . _ - _ w

, Table 11 XRD and DSC Results - Cores IA and llA - Phases I and 11

. Core Location Sample Component XRD Abundance DSC Quantity, No. No. yo IA top-pc mortar 1 Calcite (CACO 3) Predominant 36.5

, Gibbsite (Al(OH)3)! Trace 1.2

, Portlandite (Ca(OH)2) _ Minor trace Possibic trace

{

mbble-lumnite pc 2 Calcite Predominant 27.l*

Vaterite (.p-CACO 3) Small Amount

• Gibbsite Moderate 11.3 bottom lumnite pc 3 Calcite Predominant 33.6*

Vaterite Small Amount Gibbsite Moderate 10.3 lumnite pc- 6" below 4 Calcite Highest 25.7*

lumnite grout Vaterite *

~

Small Amount Gibbsite Moderate 11.2 lumnite pc- 3" below 5 Calcite Highest 14.8*

. lumnite grout Vaterite Small Amount *

Gibbsite Moderate 13.2 i Hydrogamet (C3AH6) Trace 0.9

! , lumnite grout 6 Hydrogarnet Highest 18.4

. Gibbsite Very high 19.7 t

Calcite Trace Amounts 2.6

Strutlingite (C2 ASH 8) Trace Amounts Possible Trace IIA porous pc paste - 3" 1 Cdeite Highest 16.8 from bottom Cement Clinker Phases High No peak Ettringite. Trace Small peak,

(C3A 3CaSO4 32H2O) unquantified Gibbsite! Trace 0.1 i Ponlandite Trace 2.1 por. pc paste near top 2 Calcite Highest 20.8 j Portlandite Minor 0.5 Cement clinker phases High No Peak Gibbsite Trace 1.0

, Hydrogarnet MinorTmce No Peak por, lumnite near bot. 3 Vaterite Highest 16.0*

• 4 Gibbsite High I 8.4 i Calcite Small **

por.lumnite midJJe 4 Vatedte Highest 16.7 "

Gibbsite High 19.3 Calcite Small **

lumnite grout 5 Vaterite Highest 7.3 "

Gibbsite . High 19.9 Calcite Small **

Hydtcpmet Small 7.3

• or ** - Calcite and vaterite are indistinguishable by themial analysis. The value given represents both compounds.

48

j ,

Table 12 .

XRD and DSC Results - Cores til 9M and til 8L - Phase ill -

Core I.xx:ation Sample Component XRD Abundance DSC Quantity, l'  %

i No. No.

III9M Mid - porous pc conc. 1 Calcitee Predominant 25.9 Portlandite small amount 0.1 Ettringite trace amount small amount 3 Jibbsite(?) trace amount 0.1%  :

lower 1/2, pc grout 2 'ortlandite Highest 10.0.

j C-S-H Present Present Ettringite small amount small amount i

Calcite small amount - 6.0 il interface, pc grout & 3 Calcite Highest 15.9*

porous lumnite conc.

• Vatedte - Small amount Strutlingite Smallamount 2.0 l 1.0

Gibbsite Small amount

Hydrogarnet Trace Trace  !

[ Ettdngite Trace Present  !

i C-S-H Present Present I bot. 2" por. lum. conc. 4 Calcite Highest 12.7*

I Vaterite Small amount

' Gibbsite Small amount 3.9

. Hydrogamet Trace amount . 0.6 mid. por. lum. conc. 5 Calcae Highest? - 4. l

• 2 Vaterite Small

- Gibbsite Small amount 6.8 Hydrogarnet Small amount - 1.2 top 2" pot. lum. conc. 6 Calcite Small amount 5.5*

Vaterite small amount Gibbsite small amount 4.6 Monocarboaluminate small amount not. quant.

Hydrogarnet small amount 5.6 ,

lumnite grout 7 Vaterite Small amount 5.6* j Gibbsite Small amount 7.0 l Calcite Small amount I Hydrogarnet Small amount - 1.2 i Strutlingite very small amt. minor peak?  :

III 8L mid. porous pc conc. 1 Cakite Predominant ' 38.0 Porthadite Small amount 0.I 3 Gibbsite! small amount 0.2 top porous pc conc. 2 Calcite Predominant 41.8* '

Vaterite Small amount Portlandite - Tace amotmt 0.8 Gibbsite! Trace amount 0.3 .

bot. por. Icm. conc. 3 Calcite Highest 13.1* l Vatedte Small amount l Gibbsite Small amount 3.2 Hydrogamet' Small amount 4.5 .

l W,

4 4

Table 12 (cont'd) -

XRD and DSC Results - Cores til 9M and til 8L - Phase ill Core Location Sample Component XRD Abundance DSC Quantity, i No. No.  %

mid. por. lum. cone. 4 Calcite Highest 12.5*

Vatedte Small amount

• l Gibbsite Small amount 7.0 i Hydrogarnet Trace amount 1.4 I

] lumnite grout 5 Vatedte ' Small amount 7.1

• l Gibbsite Small amount 10.1 Hydrogarnet Trace amount 3.3 Calcite Trace amount *  ;

4 l

l 8 !I I

• 1 1

i l i

1 4

5 i

e

i l I Table 13 Phase ill- Calcium Aluminate Test Data' l

Unconfined Samples Confined Samples Compressive Compressive Period' Flow. gpm8 Density, pcf Strength, pal Dansity, pcf Strength, psi _

1 32 120.4 1333 117.8 1901 2 27 119.7 1283 120.9 1607 1

3 19 117.6 1450 122.0 1646 4 30 119.1 1597 122.4 1470 22 120.7 1667 122.4 2085 l 6 22 121.7 1303 120.6 1979 1

7 17 119.6 1643 120.0 1842

-8 18 118.8 1230 119.7 1587 i 9 22 115.4 1003 119.7 1391 i

10 22 116.6 877 120.0 1293

, l 11 19 115.2 913 119.1 1489

. 12 18 116.8 1005 118.3 1470 e

13 17 115.1 1517 118.0 1352 14 17 117.4 897 119.8 1392  !

15 18 116.4 1207 118.7 1411 Average 21.3 118.0 126? 120.0 1594 '

i Std Dev 4.6 2.1 276 1.5 247 2 1. Test perfomed by, and all data from ARL

2. Each period comprised 21 days of water flow at noted rate followed by 7 days of no flow.

3. Flow adjusted to maintain water level at interface of Mix E and containment mat concrete.

~

g"

.. y, .

j

/ ,' - . ' . . ,.  ;  !

u . . . . . . .. / uT,,P,00,

• y.......

• uEuarvaE

. . . *... .: '.. ..*~ . . . /

• i X . . . * *  :

' )

POLY SHEET - - . .. '

7, -

.2* COMPRES:BLE MATERIAL q T (MORTAR) m . . ,

~

j/j RIBBED FIBERGLASS PEET I

i *

, s .'. .

s .,' . .

l

, TAR PAPER SEAL *Y' .\ '

, p .' g -

S ./

~

l c' ? '

1 ** CONCRETE BLOCK WITH PROTECTIVE

AEMBRANE SIDE & 36~ +18' WIDE GAP 5, .II ~ -

g, TO MAT EL. f4 2T-3*

q ,,

i

' *3N STONE SIDE FOR DRAINAGE. USE - - l I 74'-6' R 4

• ARPAPER ON TOP TO EXCLUDE  %' y' ,. ,

'~

v i

' IEAL COAT -

,,,2. -

4* LAYER OF COMPACTED NN CRUSHED STONE 1* DIA DRAlfJ HOLES THRu BLOcx in unE wTH J WATERPROOF CONTAINMENT STRUCTURE HOLES IN VERTICAL BLCCK /

'~/ MEMBRANE

e

,s0 Thick FOUNDATION MAT l

/ -

Cort RETE SEALER ' POROUS CONCRETE PROTECTIVE BOARD

! rf, 7 (MORTAR)

~

~

POROUS CONCRETE LEVELING pct; PROvCE DIA DRAlf. HC' Sc ~

THRL COf ORE *E B.00 '. IN h' ~! -

6 DIA ORAIN PIP ~r

' EACh COFE CF THE PASSAGE j OF WATER INTC DOROUS cot 4 CRETE /,

- 1 \

_. I ,, i \ EL. (-) 37-6*

p,, x. ..... . . . .. . .

6. , :... .- . ...

. .. m. ...\..\..

. . . . . 9.

. . . u ,.

~ , ' ..

i w*g=;ff,o? _

l' /; p p l *.'~,'. n

, . , 6.1) y WyQQQfyij, HH

/, nNfSH SMOON TO 7, _g n >\, SEAL SURFACE

\

GAPS FOR 4

DRAINAGE Fig.1 DETAL OF MEMBRANE & FOUNDATION MAT

( NOT TO SOALE )

( l l.

t Q ,'

,_ . l

. i

/e-ESF BU!LD!NG \

m r n:c e:.

p .is. .,. . .

aws^M

~~

y

. , SCP d

i 8'

g,'

l i

. i t

/-

, , o ,

N ,  ! l

. \ l l

\

l { 4 i j l

. f

% (.CONTMT s'fRUCTutt l  : .s', a:woaxuo.

oamin piet

/

N '

\' f cc c.u ai.cn

.  %, h/

w PL A N EL. (-137 '-6 ?

snowiu uw a ca4 =4es

?

~

' TITLE FIGURE 2 ,

CONTAINMENT MAT

• LAYOUT OF DRAINAGE

% i

. 4

~ '

~

~ ~ ~

r

~

~ ~ B ted ae 6 6 s 9 9

/

656D 999/ Wrve p m

t e/0 1 808l 0 //1 D H ds nb u a1 / //1 // T uO S D7 77777 R o r F c ] O Gew

~

i ~ S v ~ ~ N oF l

E o N M

t a

)

o

)

e

~

~

~

~

] m 3

k n

a r

p p ip ip -

~

]. Q i F

g F e e ~

~

~

d d w n

a n s i

n n i

s ~ _

!* ,r

" A h lo F

ioi(

r

}

i i( ~ e _

i c t pt t

~

\ J ta n in r n W8 h

c e

c e e seim im

~

~

~

~

h 4 d n _

T D de de ~ B NBp 8 u

o r

S Sr e56r e Fr 7 m G

t t t t 1 l P u a .

(/"

e Veai5 a P e r

WVWWWW M h A m

U a

D n ~

S -

- i r

e io W t

a@@@@g@ -

S c o

r P L ted ae Wrv w e o M

l ds lF w nb S uO r _

U E T

~

io rr r h; o

r Gow t

e a*

W9 S

l _

O ala oF N d N wg 2 I n u

o F 6560 6 6 9h99

/ /// / /

9 9 d

n6 u3 l G

r S t e9d00 0 0 2 21 1 1 i a////// G o

r g>

E D6 677 7 7 8

mae h

,ar

) ,DA m

u r

o m

o h 6.

9

/

)

ie r f r

r wippeiiii

) ) ) )

ee eep p A

7 A

B p

m A a

e r

)

L 8

! e~w p p p p P P u

m 7 d e ee ee a n i tsiiiii d ddd d M D

o t s s s s ir d nu oosiiii u n n n n U W e

ent:

r st pt i

ui oin~tnnn n n r

a a cseee6b emeiiiiii (i ( ( ( (

mmmmmm t i e

t t S

4 d ddddd mroD s n f eeee e e S5SSBS ei t t t t t t t

ae e eeee e rl p wm WWWWWW lo a FS 1 ..n 2 o i

t L

c@@@@@@

a o ~~

ESF SUMP SYSTEM i RHR CUBICLES i l Pump Pump l Pump j 3DAS-3A 3DAS-3B i 3DAS-11 i #E i1 -

i. -

., ... ._..o .to Waste

l _ __....

Tanks

8 i  !

i 4' Elevation u ,,  :

i

y i , .

I . .i

! Sudp~3DASTA 5 ump 36AS-6N l: h ump 3DAS-10 L

....... .... . . . - - ...= ......... .._ .....

RSS CUBICLES  :.

l Pump Pump  !

j 3DAS-8A 3DAS-8B i

, t

!. N i N i.

i &D!scharge Filter eDischarge Filter i

i

1

?

-34' Elevation t t .

l l 1

, i Fig. 4. ESF Pumps

i i......Sum ......p 3DAS .7A. . .. . .. . . ... . . . . . . .. . . . . . .. . .Sum . . . . . . . . . . . . . . . . . . . . . . . . . . . . .p 3 D A. .S .7 B..................................................!

H,e r . {

.. .- . . - - . . . . . . . . . . ~ . . . . . . . - . . - . - . . . . . . . . . . - . . . . . - . - - . . . . . .

. . i I

- l

j. i l  !

i 50.00 d l

s. a 5

I 45.00 -

2 i

i i f'

f, e

4-

, g 40.00 -

!~- h E r .  !

n 9 O

i 35.00 -

I 4

, a a i d

e i

30.00-

{

. m 1

(.

e  ;

25.00 - - ' '

1 25.00 30.00 35.00 40.00 45.00 50.00 $

Ca0 by XRF, %

Fig. 5..Ca0 by ICP vs. Ca0 by XRF (ABB-CE 1996 Data)'  ;

1 9

l l,'

t 4

1 4

t d

1 l

i-  !

i

1. -

1

! -e 14.00 -

E a

l' 12.00 -

. 10.00 -

i l *. a

, a. .

j 2 8.00 -

j N

n R

q 6.00 -

e a

i i

• 4.00 -

i n 1

2.00 -

t 4

t' O.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00

Al2O3 by XRF, %

e Fig. 6. 41 0 23by ICP vs. Al 0 23 by XRF (ABB-CE 1996 Data)

[a

!'1-5~

i.

?

e 4

r d.

s >

,n-r

. ._ . . . . . . _ . _ _ ~. _ _ _ _ _ . _ . . . . _ . . . . . . . _ .

I i l 1

l l

i

)

' )

4 14 -

l l ,9 i a '

li

• i e 1 12 -

19 1

a 4

. 10 -

N a

i. <

Q a,'

g 8 -

i 2 4 l

g e g6 -

a i

a 4 - '

i 2 -

4 j l J I s

0 ' ' ' '

i 4

0 2 4 6 8 10 12 14 C/A FROM XRF DATA i Fig. 7. Molar Ratio of Ca0/A1023 by ICP and XRF (ABB-CE 1996 Data) 4 1

4 4

4

I g I I /

Ik  ? .

/

pU N #

i @

yf ~~

T ,-

~ ^

gh .

\' .

o 6

gel 96-06A f ,/ i

~

,' t, ,

, c

[ / ~-

1 ,- 0 -

-} 11 5 s

, f/,

M A)

~ '

, \ ,

b

.t) $ r\

g, O gel 96-04A a gel 96-04B L J

'N

/ /f. /

//

~

h

" l

!/ lg me r- 1 i,

~~

2 7,

1 6-05A '

gel 96-058

"" i '

- '/

///5 "

gel 96-03A-f e \ GE196-05C-l 5

'K >

' 5) ,,

g .

gel-9601A gel 06-018 gel 96-03B

-gel 96-03C

,0% il h =

1. -- =1 L- ~

s-y crM acxt ,

y

-n__L ,.,

t-,,,

N 1 ..

. r- g eog4 ( ,.

/ p aj 13 -, f C

/; ' $~ ' "

/ __ > g j L ,,,

{

' O C']U

==

} Q ~

}

c ,,, -,.,i -

~ -

y ,

_l],-f

' ~ i

- j u;;

- _~~a j

3 , ,,,

_ gg n NO S . l ll / ~*

x

1. BORINGS SHOWN WERE PEftFORMED ~ ~

/ 0 180 360 N s'

/

DECEMBER 1996 TO JANUAkY 1997.

2. BASE PLAN PROVIDED BY NORTHEAST UTluTIES Z / l .*

SCALE. FEET SYSTEM WATERFORD CONNfCTICUT.

! LEGEND. Northeast Utilities System Ground Water investigation j Waterford, Connecticut Millstone Unit 3 BORING LOCATION PLAN h gel 96-06A Waterford, Connecticut y $ gel BORING j GEI Consultants, Inc. Project 96199 February 1997 Fig.8

Fig.9 Linear Regression Analysis of Unconfined Compressive Strength and Flow Periods

SUMMARY

OUTPUT Rearession Statistics Multiple R 0.519 R Square 0.270 Adjusted R Square 0.212 Standard Error 245 Observations 15 ANOVA df SS MS F Signdicance F Regression 1 288,322 288,322 4.799 0.047 Residual 13 781,049 60,081 Total 14 1.069.371 Coefficients Standard Error tStat P-value Lower 9S% Upper 95%

intercept 1518 133 11.40 0.00 1231- 1806 Period -32.1 14.6 -2.19 0.05 44 0 Unconfined Compressive Strength i verse Periods of Flow l

2000 -

0 0 1600 o f I

x w

o 200 o o Measured Compressive Strength, psi o n j , o o -- Predicted Compressive Strength, ps!

g 800 - ,

E l

$ I 400 -

0  !

0 2 4 6 8 10 12 14 16  !

l Period  !

l 1

i I

Fig.10 Linear Regression Analysis of Confined Compressive Strength and Flow Periods l

SUMMARY

OUTPUT Regression Statistics Multiple R 0.634 l

R Square 0.402 )

Adjusted R Square 0.355 l

Standard Error 198

Observations 15 i ANOVA I l df SS MS F Significance F

- Regression 343,070 343,070 8.722 1 0.011 )

Residual 13 511,353 39,335 i Total 14 854.423 .

i

, Coefficients Standard Error t Stat P-value Lower 95% Upper 95%

Intercept 1874 108 17.39 0.00 1642 2107 Period 35.0 11.9 -2.95 0.01 -61 -9 I

Confined Compressive Strength verse Periods of Flow

{ ]

2500 -

2000 - o 0 o i 7 i a

$ o o

$500 - o 3 o o o i m o j

$ o Measured Compressive 3 Strength, psi 11000 -

h Predicted Compressive Strength, y psi 500 -

O I O 5 10 15 Period 1

I l

i

+ 40'-O'

'^

S *

, + 30'-O' fii1 48 44 A se se sA

~- w '. - + M '- O' 1gi .

11.3 207 34 0

\ I

\ 72$ ll$

w_ 11.e _ _ . _ . - _

. 3 0._o.

i 10.2 13.2 SM NL (O'-O')

l gj_ _ _ _ .

_ _ 3 0._o.

\ \ \_ ____ ._ __ __ ._ .___._"21_ _ _ .- _

_.i.__._ . . . _ . - - - - -

---.--. - - M'-O' Se \

- 30'-O' A'

y____.___ _ _ - - - 40'-O'

$2 1E7 h4 Yts 142 GRID A s3 UNIT 3 KEY TNL I

e _ _ e -

Fig.11 Cloride, ppm O 150 FEET

__i._____. _...__._...___._.._______._._._________.__.m___ . _ _ _ _ _ _ _ _ _ _ _ _ . _ _ . _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ . _ _ _ _ _ _ . _ _ - . _ _ _ _ _ _ . _ _ . _ _ _ _ _ _ . _ _ _ _ . _ _ _ _ _ _ _ _ _ . . _ . _ _ _ . _ _ _ . _ _ _ _ _ . _ _ _ _ _ _ _ _

y es" _m- p_. ^ _ + + * ,.E* s*

=

~~ -- ~~

+ 40'4 3A  %

+ 30'4 sA

, g.g fy 1h- '~ -

- + 10'4

\ =

2ss

~-

EA "

g 222 g_ 37.s

~~~

- -10'4 ,

g

. - ~ ~ ~ . . . . -\1 ~ ___.__ -- - ~ ~ ~ ~ ~ ~ . - _ __

-. . - - __ _ s'4 s '

_ , . . - - - ~ ~ ~ . . -

_30'4 3'

V r

- 4;84 h2 g R2 W.o i

un1T3 KU 998 T

leur s - _m 8 s

nonm ag.n suitate vem .e. -

8A

+ 40'-0*

^

• 3B -- --

+ M '-O'

• ^ * ** 5^

L 74.6 sta 582

[ '

i Y

126.0 87.5 57.5 8#'4 101.8 29.7 87.5 149.0 121.0 iHE4- ir- _- + 10'-0*

i -

26.0 SEA LEVEL i (O'-O')  :

2*-- - - 10'-O'

• 51 2

\i 46.8

- M'-O' 6 i l

i

- M'-O' t' .

26.9 V - 40'-O' h2 9 h4 1$.0

~I55 5

GRID (TRUE)

S3 UNIT 3 ggy NORM --

t N g . __ __ 95 1

marae 7-  :

1 I PLANT i Fig.13 Calcium, ppm 0 00 150 FEET

sA

+ @'-O'

'^

8

• y,

+ 30'-O' 4A Sc se sA

}

, , , jjgt --- -

+ 20'-O' 8_59 4 g4 3.88 - .

II 624

~

0.00 i 5.43 6.53 13.3 e.ss- s.n- _-- - + 10'-O' i -

5.57

1. SEA LEVEL i (O'-O')

Isi$_- '

- 10'-O'

• i1 721 i_.- . . _ . _ . . . _ _ _ _ _ _ .

. _ _ _ _ _ . - - - - 20'-O' Se i 30'-0*

^'

y _

_ q'_g=

$2 255 h4 IE s.2s j

@ UNIT 3 GRID (TRUE) 3 KEY h6 ^"

1

_a__ _ .. ,_

PLRJT i

" Fig.14 Potassium, ppm 0 150 FEET C- . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

+ 40'-O'

'^

as 3^ . sc

. , O'

• ^ A ,sc se sA

/8.. _ _. ._ 'y_ _ _ _ _ _ "

[ .

___ 4 g*-O' 34.0 m 56.7 ff kh 8 9E5- sa.s- -

+ 10'-O' i -

17.5 SEA LEVEL '

i (O'-O')

El- '

- 10'-o-

\

23.4 ,

\ \ 58.9

.- - i - E'-O' S6

- M'-O' E

V - 40'-O'

$2 1s.2 @

4 ES 2M

\

GRID M S3 UNIT 3 ggy Nonm i h N 9 __ __ s6 mame 1

PL M r i

" Fig.15 Sodium,ppu o so 100 1 sofar w . _ _____ _ _ _ _ _ - _ _ - _ _ _ _ _ _ _ _ _ - - _ . _ _ _ _ _ _ . _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . - _ _ _ - - _ _ _ _ _ _ _ _ - _

4

+ 40'-O'

'^

38

+ 30'-O' l

j_f 48 4A A sc ,

sa sA

,7, 8% . -

- + 20'-O' 6.86 6.45 Ell N 8g 76*es

~

8 46 - sie- ._-- + 10'-0*

\-

7.17 7.2s SM L i (O'-O')

I *E -

- - 10'-o-

\

Di og 7.53

____i_.____ _ _ _ . _ _ - _ _ - - - -

- - 20'-O'

- 30'-0*

__ _ g _ _ _. _ _. - - -- -

V--- - 40'-0"

@ 7E b4 }$

2 7.08 GRID (TRUE) 3 UNIT 3 KEY 15JAN97 I 14FEB97 N & y_. _ __;_

1 5

PLANT "UR*

Fig.16 pH o tsomer

_.__.m _ _ __ _ _ _ _ _ . . _ _ _ _ _ _ _ _ - . _ _ _ _ _ _ . . . - _ _ . _ _ _ _ _ _ _ _ _ _ . . . . _ _ _ . _ _ _ _ _ _ _ . _ _ _ _ _ _ _ . _ _ . . . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ . - . _ _ _ _ _ . - . _ _ _ _ _ _ _ _ _ _ _ _ _ .

-.___.U