ML063040693

E-MAIL: (PD) Firebar-D Material

ML063040693
Person / Time
Site:	Oyster Creek
Issue date:	10/04/2006
From:	Ouaou A Exelon Corp
To:	Ashley D Office of Nuclear Reactor Regulation
References
%dam200611, %dam200612, TAC 8261
Download: ML063040693 (57)

Text

D. Ashley - Firebar-D material C

I Paqe 1 1 From:

<ahmed.ouaou @ exeloncorp.com>

To:

<DJA1 @nrc.gov>

Date:

10/04/2006 6:20:45 PM

Subject:

Firebar-D material Donnie.

The Firebar-D material and its impact on the drywell shell were addressed in a GE report "Corrosion Evaluation of the Oyster Creek Drywell" dated March 6, 1987. The report was transmitted to the NRC in a letter dated May 29, 1987 (5200-87-006). I have extracted the following information from the report, which I'm send in 2 parts because of its size.

Section 2.1.3.2 states that gap material used above elevation 23' 6" was Firebar-D, a proprietary asbestos fiber-megnesite cement product applied as a spray coat. The manufacturer of the product was All Purpose Fireproofing Corporation. The material was modified to achieve a reduced density. The solid material, asbestos fibers, magnesite and magnesium sulfate (approx. 75% asbestos), were premixed and combined in a mortar mixing machine with water, to control density, with foam (aerosol PK, a protein, as a foaming agent) to form a slurry suitable for spray application.

The impact of Firebar-D on drywell shell corrosion is discussed in section 6.2.1 of the report. The report concludes that the lack of ?-Fe2 03 in the oxide on the core plug surface/crust, the relative low amount of Mg in the sand samples and the absence of corrosion at the 51' elevation level suggests that the role of Firebar-D in the degradation of the OC drywell corrosion phenomena is not significant.

Regards. Ahmed

<<GE Report on corrosion analysis Part 1.pdf>>

This e-mail and any of its attachments may contain Exelon Corporation proprietary information, which is privileged, confidential, or subject to copyright belonging to the Exelon Corporation family of Companies.

This e-mail is intended solely for the use of the individual or entity to which it is addressed. If you are not the intended recipient of this e-mail, you are hereby notified that any dissemination, distribution, copying, or action taken in relation to the contents of and attachments to this e-mail is strictly prohibited and may be unlawful. If you have received this e-mail in error, please notify the sender immediately and permanently delete the original and any copy of this e-mail and any printout.

Thank You.

CC:

<fred.polaski @ exeloncorp.com>, <donald.warfel @ exeloncorp.com>,

<john.hufnagel @ exeloncorp.com>

I c:\\temp\\3WJ}00001.TMP Page 111 Mail Envelope Properties (452433AB.989 : 21 : 18825)

Subject:

Creation Date From:

Created By:

Firebar-D material 10/04/2006 6:18:55 PM

<ahmed.ouaou @exeloncorp.com>

ahmed.ouaou @exeloncorp.com Recipients nrc.gov OWGWPO01.HQGWDO01 DJA1 (D. Ashley) exeloncorp.com john.hufnagel CC donald.warfel CC fred.polaski CC Post Office OWGWPO01.HQGWDO01 Route nrc.gov exeloncorp.com Files Size MESSAGE 2302 TEXT.htm 3320 GE Report on corrosion analysis Part 1.pdf Mime.822 8609330 Date & Time 04 October, 2006 6:18:55 PM 6285055 Options Expiration Date:

Priority:

ReplyRequested:

Return Notification:

Concealed

Subject:

Security:

None Standard No None No Standard Junk Mail Handling Evaluation Results Message is eligible for Junk Mail handling This message was not classified as Junk Mail Junk Mail settings when this message was delivered Junk Mail handling disabled by User Junk Mail handling disabled by Administrator

NPU Nuclear E N uclear 100 Interpace Parkway Parsippany, New Jersey 07054 201 263-6500 TELEX 136-482 Writer's Direct Dial Number:

May 29, 1987 5200-87-0061 U.S. Nuclear Regulatory Commission Attention:

Document Control Desk Washington, D.C.

20555 Gentlemen:

Subject:

Oyster Creek Nuclear Generating Station Docket No.

50-219 NRC Meeting - June 11, 1987

References:

1) NRC LTR 12/12/86, Meeting Minutes 12/1/86

2) NRC LTR 1/5/87, Meeting Minutes 12/10/86

3) NRC LTR 1/4/87, Meeting Minutes 12/19/86

4) NRC LTR 3/13/87, Meeting Minutes 2/26/87

5) GPU LTR 5000-86-1116, 12/18/86

6) GPU LTR, 5/12/87 In anticipation of our June 11, 1987, meeting to discuss the Oyster Creek Containment Program, we are transmitting two reports*'which were prepared by outside consultants on the Drywell Corrosion mech'nism.

These Reports are:

1. Microbiological Analysis of Oyster Creek Corrosion Samples, Carolyn F. Mathers, Ph.D.,

York College of Pennsylvania, Febraury 2, 1987.

2.

Corrosion Evaluation of the Oyster Creek Drywell, B. M. Gordon, General Electric, March 6, 1987.

If there are any questions onthe contents of these Reports, we will be glad to discuss them further at our meeting.

Very trI yours, T.*horpe Director, Licensing &

Regulatory Affai rs JRT/1fpa (4825g)

Attf cc's on next page GPU Nuclear is a part of the General Public Utilities System

cc:

Mr. William T. Russell, Administrator Region I U.S. Nuclear Regulatory Commission 631 Park Avenue, King of Prussia, PA.

19406 NRC Resident Inspector Oyster Creek Nuclear Generating Station Forked River, N.J.

08731 Mr. Alexander W. Dromerick, Jr., Project Manager U.S. Nuclear Regulatory Commission Division of Reactor Projects I/Il 7920 Norfolk Avenue, Phillips Building Bethesda, Maryland 20014 Mail Stop No.

316

.4 MICROBIOLOGICALANALYSIS OF OYSTER CREEK CORROSION SAMPLES Prepared by Carolyn F.

Mathur, Ph.D.

York College of Pennsylvania For GPU Nuclear Corporation 2 February 1987

REPORT

SUMMARY

The purpose of this study was to determine if microorganisms are involved in the corrosion process occurring in the Oyster Creek drywell.

The criteria used include:

1) presence of viable cells in or around corroded areas;

2) presence of conditions at the site that are believed to promote microbiologically influenced corrosion (MIC); and 3) presence of specific types of microbes that have been associated with corrosion in other systems.

Four sand or corrosion samples adjacent to metal plugs removed from the Oyster Creek plant drywell were analyzed for microbes that could be influencing the

• corrosion occurring in that area.

Viable bacteria were found in all samples.

Cells were observed closely adhering to the corrosion materials, and a variety of microbial types were isolated.

The temperature, chemistry, and environment were all conducive to microbial growth.

However, we did not detect significant numbers of the types of bacteria believed to play a

dominant role in microbiologically influenced corrosion.

We did not detect sulfate reducing

bacteria, Gallionella organisms, or Pseudomonas (slime formers).

Some filamentous types and spore-formers were present.

We concluded that microbes probably played a

role in the early stage of the corrosion,,

but are not likely the most important factor at this time influencing the corrosion rate in the most,heavily wasted metal areas.

We recommend future monitorinq for NIC (microbioloqically influenced corrosion) and the eventual implementation of corrosion control procedures that

also inhibit microbial growth.

METHODS Four of seven core plugs taken from the Oyster Creek drywell on December 6 and 7,

1986, were analyzed for the presence of microorganisms that could be involved in microbiologically influenced corrosion.

Corrosion or sand material taken from directly behind each of the metal plugs was transferred to a sterile plastic bag in preparation for analysis.

SAMPLE PREPARATION Part of each sample was suspended at a

10% concentration in sterile water and incubated with the vital stain INT (p-iodonitrotetrazoluim phosphate) and substrate for one hour.

The samples were fixed in formalin and transported to York for microscopic analysis..

Another part of the sample was suspended in sterile water and used for innoculated into SRB (Sulfate Reducing Bacteria) media and for standard plate counts.

Representative colonies were isolated and characterized from each sample.

MICROSCOPIC ANALYSIS Samples were stained with fluorescein isothiocyanate (FITC) and analyzed using a Zeis epifluorescence microscope.

The total cell counts were done on the FITC stained

cells, while the percent viable cells (aerobic) were quantitated by countinq cells positive for TNT.

Samples were microscopically examined for bacteria that can be.- ssociated with corrosion.

RESULTS Table 1

includes a

physical description of the sample material and a brief description of its microscopic appearance, cultured results and other relevant information.

Samples 1, 4, and 5 were associated with

wastage, while sample 2

had only pitting.

The sand from sample 5,

however, was not directly adjacent to the metal plug.

Table 2 gives the total cell counts (FITC cells/gram),

the percent viability (as measured by INT staining),

the standard plate count (as colony forming units/gram),

and the SRB growth results.

The cell counts and viability were at levels reasonable for soil samples.

The lower counts and viabilities in the corrosion material samples indicate less microbial activity

here, compared to the straight sand materials.

It could also mean more anaerobic organisms were present in the corrosion samples.

The SRB culture results indicate that

little, if

any, sulfate reducing bacteria were present.

These are believed to play a

role in some carbon-steel corrosion processes.

No Gallionella, a

microbe associated with steel corrosion, was detected by visual inspection.

Many small vibria shaped cells were observed closely adhering to the corrosion material from the Bay 19 samples.

Table 3

contains information on some of the dominant colonies of bacteria isolated from the standard plate counts.

A variety of Gram positive and Gram negative rods were detected.

Pseudomonas, an organism sometimes associated with corrosion, was not detected.

Some filamentous

forms, sometimes seen with corrosion, were observed.

Gram positive smoore formers were

present in sample 5.

Table 1 Samp 1 e Physical and Microbial Descriptions I-19C Physical Description corrosion product; wastage dry sand; pitting corrosi on product; wastage Microbial Description 2-15A corrosion material dominant in sample; slight growth in SRB media; long Gram positive and Gram negative rods; 50% of population aerobically viable long and short rods, both Gram positive and Gram negative; some filamentous.

types; no growth in SRB media.

71%

aerobically viable variety of rod-shaped-Gram positive and Gram negative cells; short vibroid cells closely attached to corrosion strands; may be Gallionella but could not confirm; 40% aerobically vi+/-able Griam positive spore forming rods; weak growth in SRB media; 74%

aerobically viable 4-19A 5-11A wastage; moist taken

from, distant from corrosion material sand area

Tabie Bacteriological Results From Core Plug Samples Collected From Oyster Creek Drywell on December 6 and 7, 1986 Sample 1-19C 2-15A 4-19A 5-IIA FITC cel 1 s/.q

(% viable) 5 x 104 (50)

I 1 IO (71) 6 x 10" (40) 4 x 10l' (74)

SPC (cf u/g) 3x10,*

2 x

10,0 4 x 104-1 x10ý SRB (culture results) little growth negative negative little growth

Table 3 Descriptions of Colonies Isolated From Oyster Creek Samples Samp i e (5) 1 1A-1 (5) 1 IA-2 (2) 15AD-1 (2) 15AD-2 (1) 19C-E (1) 19C-F (1) 19C-G (4) 19A-1 (4) 19A-2 (4) 19A-;

(4) 19A-4 Colony Morpb.

white, irregular

white, irregular

clear, irregular, spreading med.

white dry, irregular med.

clear white

colony, concentric same as 19C-E med.

clear white oblong, mucoid small clear white bright yellow smooth small clear white Gram Stain gram + rods gram + rods gram

+

rods large with long thin filaments long Gram -

.rods long Gram -

rods long Gram -

rods long Gram +

rods Gram fat rods with pointed ends Gram

+

med rods singles,

pairs, clumps same as 19A-1 med.

to long Gram rods sinqles & end to end pairs Misc.

4 spore

+ spore

oxid, cat motility

+ oxid,

+ cat

+

motility oxid. 4 cat

+ motility

oxid,

+cat

+ motility

oxid,

+ cat

+

motility

oxid,

+ cat

oxid,

+ cat motility

oxid,

+ cat

+ oxid, cat

+ motility

tan, dry

wrinkled,

DISCUSSION AND CONCLUSIONS Microorganisms are one of several factors that can influence corrosion rates in

metals, such as carbon-steel.

Specific organisms such as the sulfate reducers (SRB's) and Gallionella can ble active in the corrosion

process, but Microbiologically Influenced Corrosion (MIC) probably involves a complex system of different microbial types.

The presence of cells not obviously involved in corrosion can stimulate the growth of forms that can contribute to the electrochemical changes occurring.

To evaluate MIC in a

given

system, a

variety of factors need to be considered.

In the samples reported on'" here from the Oyster Creek drywell, MIC probably has played a

role, especially in the earlier stages.

However, MIC does not appear to be the primary cause of the corrosion that is currently occurring in the heavily wasted samples (i.e.

from Bay 19).

It should be noted that the relatively non-corroded samples (i.e.

Bay 15) show signs of early MIC

activity, based on the type of small pitting that is occurring and the general conditions of the sandbed environment.

RATIONALE FOR ABOVE CONCLUSIONS I.

Noisture, dirt, and the alternating wet and dry conditions that have been and are occurring in the Sandbed adjacent to the metal are generally conducive to MIC.

Water serves as a source of the microbes and of the nutrients needed by them.

The materials above the sand (i.e.

firebar D) contain protein and other nutrients that could be carried into the Fand 3nd Allow ior the qrrwth of various microbes.

Aater serves 3s a

reactant in microbial metabolism and in corrosion.

It can spread the microbes and corrosive products throughout the system.

2.

The presence of sulfates and chlorides in the water that passed through the sand bed and the differential concentrations of chlorides in the crevices of the corroded metals indicate the possibility of MIC.

3.

Black corrosion deposits are frequently associated with MIC.

4.

Previous studies (i.e.

Vermont Yankee Nuclear Power Plant) indicate that MIC involving Gallionella was most significant early in the corrosion

process, and that the influence of microbes on the corrosion rates reduced considerably over time.

5.

The presence of microbes in the sand adjacent to the metal and in close contact with the corrosion materials itself indicates some microbial influence on the corrosion process.

The presence of microbes in sand is expected, and because of moisture and other conditions, such as temperatures of l1O-F in which many microbes can grow, their numbers could increase.

We detected

little, if

any, SRB's in the samples, and no Gallionella were observed.

Although these two organisms can play a

role in corrosion, their absence does not necessarily rule out MIC as a factor to consider.

Unidentified microbial cells were observed closely attached to the corrosive materials from the Bay 19 samples.

However, the microbial cell counts were not that

high, and this, along with our inability to detect large numbers of organisms believed to be directly involved with MIC, indicates that MIC is rnot a major force affecting the corrosion rate at this time.

.IIC

should, however, be considered when control

procedures are initiated and in future corrosion monitoring studies.

RECOMMENDATIONS

1.

Recheck areas again for MIC during future analyses for corrosion rates.

Note especially the areas which had relatively litle

wastage, as some of these showed pitting and signs of the early stages of MIC.

2.

Consider corrosion control mechanisms that would also be antibacterial, such as elevated temperatures,

drying, flushing with a biocide gas or

liquid, improved

drainage, or cathodic protection.

NUCLEAR ENERGY BUSINESS OPERATIONS GENERAL 0 ELECTRIC F&PMT Transmittal No.- 87-178-003 Rv._

I 1.,

FUEL & PLANT MATERIALS TECHNOLOGY CORROSION EVALUATION OF THE OYSTER CREEK-DRYWELL Date:

March 6, 1987 Prepared By:

Approved By:

B.M. Gordon, Principal Engineer Plant Materials Performance G.M. Gordon, Manager Fuel & Plant Materials Technology GENERAL ELECTRIC-COMPANY LEGAL INFORMATION NOTICE The only undertakings of General Electric Company (GE) respecting information in this document are contained in the agreement between GPU Nuclear and GE and nothing contained in. this document shall be construed as changing the agreement.

The use of this information for any purpose other than that for which it is intended, is not authorized and with respect to any unauthorized use GE makes no representation or warranty, (express or implied) with respect to the completeness, accuracy, or usefulness of the information contained in this document, or that the use of such information may not infringe privately owned rights, nor does GE assume any responsibility for liability or damage of any kind which may-result from the use of any of the information contained in this document.

BMG8706.11 NEO 807 (REV 10/81)

F&PMT Transmittal No. 87-178-003 Rev. 1

1.0 INTRODUCTION

GPU Nuclear's Oyster Creek BWR is characterized by a Mark I containment as shown in Figure 1. During the 1980 refueling outage water was noted around penetrations at elevation 86'0" and running down the wall to floor elevation 75'3".

Water was also observed at a penetration at elevation 47'0" and running down the wall to floor elevation 23'6".

The presence of water at these locations indicated that an intrusion of water Jnto the annular space between the drywell shell and concrete shield wall had occurred.

Water collection was also observed during this outage on the torus room floor which originated from the leak drains in bays 3, 11 and 15, as shown in Figure 2.

Water on the torus room floor-was also noted following construction.

When water samples were withdrawn from the drains in 1980 and were subsequently radiologically analyzed, the results indicated an activity level similar to primary water.I It was concluded at this time that the probable sources of water were the (1) equipment storage pool, (2) reactor cavity, or (3) fuel pool.

It was further concluded that the leakage occurred only during refueling when the reactor cavity, the equipment storage pool, and the fuel pool are flooded.

When water was again'roted leaking from the sand bed drains during the refueling outage in 1983, it was decided that corrosion of the drywell shell could be a concern and an inspection would be performed during the next outage (1986).

However, prior to discussing the details of the program, it is critical to examine the geometrical configuration and construction of the Oyster Creek drywell.

2.0 OYSTER CREEK PRIMARY CONTAINMENT GEOMETRY The Oyster Creek Mark I containment consists of a pressure suppression system with two large chambers as illustrated in Figure 3.

The main chamber is 70' diameter spherical shell with a 33' diameter by 23' high cylindrical

F&PMT Transmittal No. 87-178-003 Rev.

1 shell extending from the top.

The pressure absorption chamber is a shell in the shape of a 30' diameter torus located below and around the base of the drywell.

The two chambers are interconnected by 10 vent pipes 6'6" in diameter equally spaced around the circumference of the pressure absorption chamber, Figure 4.

The drywell interior is filled with the concrete to an elevation 10'3" to provide a floor.

Concrete curbs follow the contour of the vessel up to elevation 12'3" with cutouts around the vent lines, Figure 5.

The drywell exterior is encapsulated in concrete of varying thickness from the base elevation up to the elevation of the top head, Figure 3.

From this juncture, the concrete continues vertically to the level of the top of the spent fuel pool.

The proximity of the concrete surface to the shell varies with elevation.

The concrete is in full contact with the shell over the bottom of the sphere at its invert elevation 2'3" up to elevation 8'11.25".

At that transition, the concrete is radially stepped back 15" to create a pocket which continues up to elevation 1213", Figure 4.

This pocket is filled with sand which creates a cushion to smooth the transition of the shell plate from a fully clamped condition between two concrete masses to a free standing condition.

The sand pocket is connected to drains to allow drainage of any water which might enter the sand.

It is within this sand cushion contact area with the drywell shell where corrosion was identified.

Above elevation 12'3" the concrete is radially stepped back 3" from the shell.

This gap is created during construction by applying a compressible, inelastic material to the outside of the shell prior to concrete installation.

This material was later permanently compressed by controlled vessel expansion to create a gap between the vessel and the concrete F&PMT Transmittal No. 87-178-003 Rev. 1 2.1 Drywell Materials 2.1.1 Drywell Steel The drywell shell is fabricated of ASTM A212 B made to ASTM A-300 requirements.

This material is basically a high tensile (70 Ksi) carbon-silicon steel with a basic composition listed in Table 1 and is equivalent to SA-516 Grade 70 steel.

The shell was coated on the inside surface with an inorganic zinc (Carboline Carbozinc 11) and with "Red Lead" (Pb 3O4 ) primer identified as TT-P-86C Type I on the outside surface.

The red lead coating covered the entire exterior of the vessel from elevation 8'11.25" to 94".

2.1.2 Sand Cushion The sand cushion was filled with sand specified as ASTH 633 from elevation 8'11.25" to elevation 12'3".

The sand was a natural sand composed primarily of silica (Si0 2 ) with some alumina (A1203).

Since the sand was stored at a local dealer uncovered and exposed to the environment during storage and installation, it is a safe assumption that the sand was placed into the sand cushion region in a moist or wet condition.

There Is no information concerning the method or condition of the backfilling of the sand into the sand cushion gap.

Both GPUN and GE performed leachate studies on the sand.

Table 2 presents the GPUN leachate results on various sand samples plus an insulation sample which will be discussed in Section 2.1.3.

The results of the GPUN analysis of the wet Bay 11 sand indicates measurable quantities of Na, K, Ca, Pb, Mg and Cl which are naturally occurring in sand.

The Mg and Cl are also present in the insulation as will be discussed in Section 2.1.3.

The Fe present is probably from any corrosion product incorporated into the sand; the Pb is from the red lead primer.

F&PMT Transmittal No. 87-178-003 Rev. 1 The sand samples from core samples 19C and 15A (to be discussed in detail in Section 5.0) also indicate nominal values of contaminants with the exception of plug 19C which has a significantly higher Fe content.

Since this plug is characterized by considerable corrosion, this result is not unexpected.

Plug 15A, which initially was considered "pitted" but actually had inclusions with essentially no corrosion, has a significantly lower Fe content in the sand behind it.

GPUN also performed an energy-dispersive x-ray analysis (EDX or EDAX) with a scanning electron microscope (SEM) on some sand and small pebbles as shown in Figure 6.

The EDAX spectrum confirms the presence of silica and alumina plus chloride.

The presence of C1 is consistent with the leachate analysis.

The GE leachate ehemical analysis of the sand ýcushlon specimen (plus 2

other samples to be discussed later) is presented in Table 3..

Table 4 lists the ionic constituents of leachate samples in units of-milliequivalents per liter.

These values are calculated from the chemical analysis results using some assumptions of metal species.

Note that this charge balance calculation sum of the anions differs from the sum of the cations by no more than 14% for any of the test solutions.

This degree of agreement tends to verify the quality of the chemical analysis.

Finally, Table 5 provides the chemical analysis results of Table 3 expressed relative to the original samples by multiplying the concentrations reported for the leachates (in mg/L) by. the leachate volume (L) then dividing by the weight of the leached material (g).

A comparison of the GE sand leachate analysis "torus sand", with the Table 5, GPUN analysis Bay 11 sand, Table 2, reveals similar results, (note ppm vs. ppb units).

The major difference, albeit of little technical consequence, is the level of Fe.

The Bay 11 sand contains 1.0-5.0 ppb of Fe while the GE analysis of the torus sand contained <0.04 ppb Fe. Since the sand sample may be from different locations the results are not significant.

The lead content in the Bay 11 sand sample also appears to be higher.

F&PMT Transmittal No. 87-178-003 Rev. 1 2.1.3 Insulation At all elevations above the sand layer, the external concrete mass is set back from the surface at the steel shell an amount calculated to allow unimpeded expansion of the shell during any design condition.

As noted in Section 2.0, this gap was created by applying a compressible, inelastic material to the exterior surface of the vessel prior to pouring concrete.

The material properties were selected to provide resistance to crushing by the pressure induced by the head of concrete but of low compressive strength to allow collapsing by induced vessel expansion.

Design considerations 1 necessitated that a gap of 2" was required from elevation 12'3" to elevation 23'6" and a gap of 3" above 23'6".

The criteria used to select the gap material was as follows:

1.

Tight adherence to curved, painted steel plate surfaces in horizontal and vertical positions.

2.

Insignificant deformation under fluid pressure of wet concrete estimated at 3 psi.

3.

Would be reduced in thickness inelastically by approximately one inch from an initial thickness of 2 to 3 inches under a pressure of not more than 10 psi.

4.

Dimensionally stable at the reduced thickness without significant flaking or powdering.

5.

Unaffected by long term exposure to radiation and heat.

6. Unaffected when exposed on the vessel prior to concrete installation.

F&PMT Transmittal No. 87-178-003 Rev. 1 2.1.3.1 Duct Insulation The.2" gap discussed above was formed by using Owens-Corning Fiberglass SF Vapor Seal Duct Insulation and was applied to the vessel shell from elevation 12'3" to 23'6".

The material was applied as Individual boards 2" thick with a factory applied laminated asphalt kraft paper (high in sulfur and chlorine) water proof exterior face and was attached to the vessel with mastic and insulation pins.

The fiberglass strands were embeddjed in phenol formaldehyde.

Joints between the boards, edges and penetrations were sealed with glass fabric reinforced mastic.

2.1.3.2 Firebar-D Insulation 2.1.3.2.1

Background

The gap material used above elevation 23'6"'was Firebar-D, a proprietary asbestos fiber-magnesite cement product applied as a spray coat.

The manufacturer of this material was All Purpose Fireproofing Corporation.

The material was subsequently modified by Certified.Industrial Products, Inc.

to achieve a reduced density.

The solid materials, asbestos fibers, magnesite and magnesium sulfate (0'75Z asbestos), were premixed and combined in a mortar mixing machine with water and., to control density, with foam (aerosol PK, a protein, as a foaming agent) to form a slurry suitable for spray application.

The first coat (Q inch thick) was standard densitywhile the second and third coat (one inch thick each) was at a reduced density.

After application and drying, the material surface was faced with Griffolyn (chloride content not known) 4 mil. clear polyethylene sheet with all edges sealed by tape and held in by insulation pins.

The polyethylene sheets formed the bond-breaker for the concrete pour.

It is important to note that the Firebar-D is said to be 75% asbestos in magnesite.

To a geologist, magnesite is the mineral form of magnesium carbonate, MgCO 3.3 Complete calcination or dead burning of magnesite produces magnesium oxide, MgO.

Commercially however, magnesite refers to "dead burnt" t

F&PMT Transmittal No. 87-178-003 Rev. 1 magnesium oxychloride, also known as Sorel's cement.3 This material is produced by the exothermic action of a 20Z solution of magnesium chloride, MgC1 2, and a blend of magnesia, MgO, by calcining magnesite and magnesia obtained from brines 3 MgO+MgCl 2+11 H120 ---

3 MgO*MgCl2 01l H 20.

The resulting crystalline oxychloride contributes the cementing action to the commercial cements.

The product is hard and strong but is dimensionally unstable, lacks resistance to weathering, and most importantly is readily attacked by water which leaches out the MgCl 2 and thus is highly corrosive.

2.1.3.2.2 Firebar-D Analysis Leachate analysis of the Firebar-D was performed by both GPUN and GE.

As shown in Table 2 for the GPUN analysis of this Insulation, Firebar-D consisted of a mixture of fiber, foam and concrete had high levels Na, K, Ca, Mg, Cl, NO3, SO4 and total organic carbon (TOC).

The Na, K and Ca are contained in asbestos.

The Mg is also present in asbestos and of course the Firebar-D.

The C1 and SO4 are major compositional factors of the Firebar-D.

The TOC of 1056 ppm is most likely due to the foaming agent Aerosol PK which is a protein.

[This material along with the sulfate could serve as a food for any microbiologically influenced corrosion (MIC).

However, this subject is beyond the scope of this report.]

Although the source of the nitrate is not obvious, it is present in small quantities in seawater.

The GE analysis of a 14 gram insulation leachate is shown in Tables 3-5.

A comparison between the GPUN and GE results, Table 2 and 5, respectively, revealed similar results.

For example, 573 vs 610 ppb Cl, 1936 vs. 1400 ppb Mg, 2850 vs. 2500 S04, 132 vs. 130 NO3, 1056 vs. 900 ppb total organic carbon, etc. were observed for the GPUN and GE analysis, respectively.

F&PMT Transmittal No. 87-178-003 Rev. 1 3.0 POTENTIAL SOURCES OF WATER INTRUSION 3.1 Leakage Paths and History As noted in Section 1.0, leakage of water from the sand bed drains were observed during the 1980 and 1983 refueling outages.

A series of investigations were performed by GPUN to identify the source of the water and its leak path.

Since the same range of radioactivity was found In this leakage water as is found within the reactor, the leak path wai believed to have been from the reactor cavity located immediately above the drywell.

This cavity is filled with water during refueling operations.

It was believed that a leak from this cavity through the bellows seal, Figures 7 and 8, at the bottom drained into the space between the drywell and the space filled with Firebar-D.

Extensive leak tests finally revealed that the most probable source of the water was the drain line gasket, Figure 8.

This gasket was replaced and subsequent leak tests performed on the bellows revealed no additional leaks.

Inspections of the areas previously characterized by leakage indicated that the leakage had been arrested.

However, this history of leakage, which may have initiated at the first refueling outage plus any condensation in the gap between the Firebar-D and the drywell shell, means that the Firebar-D could have been intermittently wetted and leached of corrosive MgCl2 which collected in the sand cushion.

As will be discussed later, the establishment of this electrolyte in the sand is considered the key factor in the drywell corrosion phenomenon.

3.2 Leakage Water Analysis During the 1986 Oyster Creek refueling outage, water samples were obtained from a drain line and analyzed by GPUN and GE.

In addition to tritium, these samples were analyzed for contaminants.

The results of the GPUN analysis is shown in Table 6.

Significant amounts of Na, K, Mg, Cl and SO4 are present.

The sources of these substances F&PMT Transmittal No. 87-178-003 Rev. 1 include the natural substances found in sand, a marine environment and the Firebar-D.

The conductivity is high (680-1100 uS/cm as compared to 0.2, 1, 70 and 1000 VS/cm for good reactor water, good quality distilled water, excellent quality raw water, and 0.05% NaCI solution, respectively) and thus clearly indicates that the drain line water would serve as a suitable electrolyte for corrosion.

The results of the GE analysis, Table 3, of the leakage water again reveals similar results for elements K, Na, Ca, Mg, Al, Cl, NO3, So4i TOC and conductivity.

The only measurable differences between the two analyses was in the Fe and Sr.

3.3 Deposit Analysis Various scrapings from horizontal and verticaýlsurfaces were obtained between the torus and drywell at Oyster Creek.

Theseedeposits were analyzed by inductive couple plasma by GPUN, Table 7, and by EDAX and leachate by GE, Tables 8 and 3, respectively.

Table 7 reveals the pre ence of various metal oxides with Fe 203, hematite ("rust") being the dominant material for material removed by Bay 7 and 11.

The only unusual oxide is the B203 which suggests the presence of reactor water.

The GE analysis of a scraping from Bay 7 only as investigated by EDAX, Table 8, revealed high percentages of Fe, Cl, K, S and Pb.

The results of the GE analysis is fairly consistent with GPUN's investigation although different analytical methods were utilized.

The quantity of Fe is consistent and anticipated.

The presence of lead, Pb, is consistent with the red lead primer coat.

Manganese may be due to the.

presence of manganese sulfides in the steel.

Although the existence of Na, Cl, and K are consistent with the presence of Firebar-D, the presence of these three elements plus bromine is suggestive of a marine environment since Br is also found in seawater.

The results of the leachate analysis of the Bay 7 deposit is presented in Table 5.

The results are consistent with the other analytical results from F&PMT Transmittal No. 87-178-003 Rev. 1 this sample in that significant amounts of SO4, Cl, Ca, K and Na leached out of the specimen.

Pb, B, Sr, Ba and Al were also identified.

With the exception of Pb, all of these elements are present in seawater.

4.0 DRYWELL THICKNESS MEASUREMENTS Motivated-by the presence of water in the drain lines and other penetrations, GPUN performed extensive ultrasonic thickness measurements of the drywell.to determine if degradation was occurring.

Approximately 1000 ultrasonic testing (UT) measurements were obtained through the use of an ultrasonic thickness gauge device (D-meter).

The D-meter measures the time for a longitudinal ultrasound wave to travel to a reflection (backwall or midwall reflector) and back. ýExpanded UT examinations were accomplished through the use of a "A-Scan"JUT technique where the character, location and amplitude of various ultrasound reflectors are displayed on a cathoderay tube.

The initial UT measurements (D-meter) were made from the inside of the drywell at elevations of 51' and the 11'3" sand cushion, Figure 9.

The sand cushion measurements obtained in the bays corresponding to known water leaks indicated that wall thinning had occurred.

Measurements just above these ares in the same plate and at the 51' elevation indicate nominal plate thicknesses.

Measurements were obtained with both the inside surface coating of Carbo zinc 11 in place and removed.

As a result of the initial low thickness readings, additional 1

thickness measurements were obtained as described in detail elsewhere.

To determine the vertical profile of the thinning, a trench was excavated into the concrete floor in Bay 17 and Bay 5.

Bay 17 was chosen since the extent of thinning at the floor level was the most severe.

The additional thickness measurements indicated that thinning below the initial measurements were no more severe and become less severe at the lower portions of the sand cushion.

Bay 5 was selected to determine if the thinning line was lower than the floor

~1 A...

F&PMT Transmittal No. 87-178-003 Rev. 1 level in areas where no thinning was identified.

No significant indication of

-thinning was found in the sub floor region of Bay 5.

Aside from UT thickness measurements performed by the GPUN staff, independent analysis was performed by the EPRI NDE Center and the GE Ultra Image III "C" scan topographical mapping system.

The EPRI investigation verified GPUN's thickness and mid-wall reflector 7 results and the GE mapping confirmed a corrosion transition at seven to eight inches up from the concrete curb in Bay 19.

The Ultra Image results will be used as a baseline profile to track continued wastage.

GPUN also used a UT integration method (30-70-70 technique) to detect minor changes in back wall surface conditions.

This technique was able to verify the roughness condition of wastage and the light corrosion areas of the containment wall as compared to reference standards.

Finally, UT investigations of various plate to plate welds and heat affected zones revealed no indications of wastage or cracking.

5.0 CORE SAMPLING To evaluate if the UT measurements were valid, characterize the form of damage, and determine the cause (i.e., due to the presence of contaminants, microbiological species, or both), it was considered prudent to obtain core samples from various bay locations.

Areas that were characterized by sharp deviations in thickness of less than half the 1.154" nominal wall were designated "pitted/inclusion" areas.

Regions that had UT indications of thinning were designated as "wastage" areas.

Regions above the wastage area and within the sand cushion region that appeared to have no thinning or "pitting" were also selected as candidate core sample sites.

Table 9 summarizes the UT characterizations by bay number.

Core samples of the drywell wall-were obtained at seven locations.

To produce an adequate sample size, an opening large enough to allow removal of sand samples and insertion of a miniature video camera and allow a simple plug design, the sample diameter was optimized at 2" in diameter.

Table 10 I'

F&PMT Transmittal No. 87-178-003 Rev. I summarizes the seven core sample locations, the type of samples obtained and the organization who performed the subsequent analysis.

The core samples were cut in such a manner to eliminate any possible contamination from the cutting operation.

Distilled water was used during the initial cutting operation as a coolant.

The final cut through the wall was performed without coolant and the shell temperature was maintained below U15*0F to prevent the premature death of any viable microrganisns.

Biological samples were taken from four plugs and analyzed by another party for GPUN.

The next five sections present the results of the core sample analysis.

5.1 Core Sample 15A -

Minimum Thickness Specimen-GPUN Core sample15A which was removed from Bay 15 was the key specimen for detailed analysis.

This particular area was characterized by the lowest through-wall thickness (0.490") as observed randomly by UT examination, surrounded by adjacent areas with nominal thickness of 1.17".

Therefore, the question was whether this area was suffering from some sort of localized pitting" attack or did the plate in this location contain inclusions.

The removal of plug 15A immediately revealed that there was no pitting or in fact any serious corrosion attack.

The sample measured 1.17" average thickness and was covered with a uniform dark brown (magnetic) scale.

Elemental analysis of this oxide by EDAX indicated that Fe was the major

(>10 W/o) constituent, followed by Pb (>I w/o) from the red lead primer and traces (10 W/o) with minor amounts of Al and Fe (>1 W/o) and trace amounts of Cl, K, Pb and Ti

(<1 W/o), Table 12.

~1 ~

F&PMT Transmittal No. 87-178-003 Rev.

1 GPUN also prepared metallographic specimens from this core plug in both the rolling direction and perpendicular to the rolling direction, Figure

13.

As shown in Figure 10 and 11, minor pitting (<5 mils) was observed on the surface.

The mid-plane of the specimen was characterized by a band of aluminide stringers, Figures 14 and 15.

These inclusions are sufficiently dense to produce a reflection for ultrasound.

In fact, the measured depth of these inclusions correlated with the depth determined by the initial UT examination.

The validity of the overall UT thickness measurements was also confirmed by actual thickness measurements.

The ability of the'A-scan to identify areas of inclusions, as opposed to pits, was also confirmed.

5.2 Core Sample 19C.- Wastage -

GPUN When core sample 19C was cut, GPUN noticed that a hard black crust remained in the hole at the sand interface.

The crust was approximately 0.5" thick and was subsequently removed for analysis.

Other wastage samples were also characterized by this corrosion product crust.

Figure 16 presents an overall view of plug 19C.

The surface has the classic appearance of general corrosion.

The measured thickness was approximately 0.825" which corresponds with the UT determination of 0.815".

The surface was covered with a thick black powder deposit which varied in thickness up to 030 mils.

EDAX analysis, of the surface, Table 11, revealed that again Fe was the major constituent (>10 W/o) as was the case of plug 15A.

However, for this wastage sample the minor element (>1

/o) was Cl and not Pb.

Trace amount of Al, Si and Mn were also identified, Figure 17.

A cross-section of plug 19C, Figure 18, prepared through one of the valleys on the corroded surface reveals the corrosion product.

An EDAX analysis along the indicated profile location, Figure 19, reveals a chloride peak in a 2 mil thick region adjacent to the steel surface.

EDAX analysis of the magnetic crust/flake deposit removed with plug 19C revealed that the primary con-stituent was Fe (>10 W/o) with only trace amounts of Si and Cl (<1 W/o), Table

13.

The pH, as determined by litmus paper, of the scale was measured at 4.

L F&PMT Transmittal No. 87-178-003 Rev. I Metallographic examination of the plug sample 19C also revealed that scale contained manganese-sulfide inclusions, Figure 20, Manganese inclusions were also found beneath the surface of the plug, Figure 21.

These two figures clearly indicate that the wastage is proceeding through the wall and is capable of retaining inert materials in the original position and orientation.

This result also explains the presence of Mn in the EDAX analysis presented in Figure 17.

5.3 Core Sample 17D -

Wastage -

GE This plug sample was also characterized by general corrosion/wastage.'

The pre-removal UT thickness was determined to be approximately 0.840".

Upon removal actual thickness measurement revealed an average thickness of "%0.860ý.

SEM examination of the surface of plug 17D, Figure 22, revealed a fairly uniform distribution of oxide particles.

An EDAX analysis of this surface revealed a high concentration of Cl (3.71 -

4. 9 2 W/o) and Fe (92.73 -

94.60 W/o), Table 14.

Similar results were obtained for an analysis, Figure 23, of the cross-section of the oxide, Table 15, where 3.45 W/o Cl and 94.40 W/o Fe was identified.

This analysis confirms the GPUN studies on wastage sample 19C where a high chloride peak was associated with the significant general corrosion attack.

The corrosion product crust removed from plug 17D was analyzed by both EDAX and x-ray diffraction (XRD).

The EDAX analysis of the crust reveals that Fe is present in the highest concentrations (88.32 - 98.26 W/o), Table 16, followed by Mn (1.54 -

10.42 W/o), Si (0.00-0.63) and Cl (200 - 3800 ppm).

XRD analysis of this dark brown to black crust was performed on magnetically separated material as discussed below.

The results of the XRD analysis revealed that the non-magnetic aliquot was composed of major amounts of alpha quartz (a - SiO2 ) with small amounts of face-centered-cubic (FCC) M30 4 spinel type phase plus trace amounts (<2 W/o) of an unidentified phase.

The magnetic aliquot composition was essentially is F&PMT Transmittal No. 87-178-003 Rev. 1 just the opposite of the non-mnagnetic sample, that is, the magnetic aliquot consisted of a major phase (>90 I/o) of FCC M3 04 spinel with small to trace amounts of a SiO 2.

The lattice parameter value for the spinel phases was determined to be a

- 8.387 +/- 0.004 A.

This value is close to the lattice 0

parameter of stoichlometric Fe 3 04 at a 8.3963 A.

The slightly smaller measured lattice parameter of this magnetic phase can be most likely attributed to small amounts of other transition elements in substitutional solid solution with the major element Fe.

It should also be noted that the error on the lattice parameter could indicate no change in composition has occurred and the spinel phase could be pure Fe 3 04.

Other compounds such as FeCl 2, FeCl 3, a Fe20 3 and y Fe 203 were specifically analyzed for in the sample, but none were identified with the possible exception of a weak trace of a Fe203' The detection limit for these" types of phase in this type of material is estimated to be one to two weight percent.

Metallogrpahic examination of plug 17D revealed similar corrosion product buildup as seen on wastage plug 19C, as seen by comparing Figure 18 with Figures 24 and 25.

The microstructure of the steel, Figure 24, and the hardness values (RB 81-84) were typical for this type of steel.

The leachate analysis of the sand behind plug 17D revealed significantly less contaminants than observed for plugs 19C and 15A.

The only contaminant present in significant quantities is 19 ppm K, 9 ppa Na and 4 ppm Ca.

The chloride content in this sand, 1.8 ppm, is significantly less than observed in the sand behind plugs 19C and 15A at 45 and 93 ppm, respectively.

It should be noted, however, that the plug core sand samples were received in plastic jars and not as a core sample per se.

Therefore, any higher concentration of contaminate adjacent at the plug/sand interface could have been diluted by mixing.

F&PMTTransmittal No. 87-178-003 Rev. 1 5.4 Core Sample 19A - Wastage -

GE This sample was the second wastage sample received by GE for analysis.

The pre-removal UT thickness measured by GPUN averaged 0.830".

The post-removal average thickness was 0.847".

SEM examination of the surface of plug 19A, Figure 26., revealed a surface which is quite different than that observed on the previous wastage sampleplug 17D.

Only a very fine powder deposit is observed op this surface.

An EDAX analysis of this surface revealed primarily Fe (96.07 - 97.45 W/0 )

with lower amounts of Cl (0.34 -

1.25 w/o) than plug 17D, Table 14.

The lower chloride content could explain the difference in surface morphology.

The cross-section analysis, Figure 27, of plug 19A, Table 15, revealed the absence of many of the elements observed in plug 17D.

Again Fe dominates the analysis (98.37 W /o) followed by Mn (1.24 W/o).

The source of jMn is most likely the manganese-sulfide inclusions in the steel.

The corrosion product crust removed from plug 19A was also analyzed by EDAX and XRD, Table 16.

In this case, plug 19A crust was characterized by Fe (64.69 - 93.36 W/o),

Si (

3.81 - 30.34 U/b),

Mn (up to 1.50 w/o), Ti (up to 2.98 W/o) and Cl (3300 to 19,300 ppm).

The XRD analysis revealed a non-magnetic and magnetic phase compositions that are nearly identical to that obtained on plug 17D.

The only difference found was that the lattice parameter for theM 304 spinel was a - 8.396 +/- 0.003 for plug 19A which is 3 4 0

exactly the value for stoichiometric Fe 3 04.

As was also the case of the crust from plug 17D, no FeCl 2, FeCl 3, a-Fe2 03 or y-Fe2 03 were identified in any measurable amounts.

Metallographic examination of wastage plug 19A, Figures 28 and 29, revealed similar results as observed on plug 17D, that is, thick corrosion product on the surface, normal microstructure and hardness values (RB 80-81).

F&PMT Transmittal No. 87-178-003 Rev. I The leachate analysis of the sand behind plug 19A reveals similar results to that obtained behind plug 17D, Table 5, but again different results as compared to the two GPUN analyzed plugs.

5.5 Core Sample 11A -H

-Above Wastae - GE This core sample was removed from above the wastage region of the drywell but still in a region in contact with the sand cushion.

The thickness measured by UT was 1.170".

After removal of the plug, the thiekness measurement measured at the center of the plug was 1.19".

Thus there was essentially no corrosion on this specimen.

SEM examination of the surface of plug 1IA-H (H - high, i.e., above sample 11A), Figure 30, revealed a surface with isolated Islands of deposits.

Higher magnification examination (1500X) revealed the presence of a non-uniform powdery scale.

The EDAX analysis indicated the Pb (52.61 -

59.77 w/o) and not Fe':(21.72 - 28.87) dominate the surface, Table 14.

This indicates that the red lead paint (Pb 304 ) was still present on the surface.

This result is anticipated since this plug suffered no corrosion.

It is important to note that the Pb is present because no corrosion occurred at this area and not that the red lead inhibited the corrosion.

The high presence of S in both the surface analysis and cross-section analysis (Figure 31 and Table

15) may be due to the affinity of sulfate to combine with the red lead paint to produce PbSO4.

The sulfate may be a leachate from the Firebar-D or from the marine environment.

Significant amounts of chloride are also present.

Since plug 11A-H had essentially no corrosion, there was no crust to analyze by EDAX or XRD.

Metallographic examination of plug 11A-H, Figures 32, 33 and 34, revealed the absence of severe corrosion.

There was only mild attack observed at higher magnifications (125X) on the cross-section of the plug, Figure 33.

Hardness measurements again revealed nominal values CRB 80-81).

F&PMT Transmittal No. 87-178-003 Rev. I The leachate analysis for the sand behind low-corrosion plug 11A-H reveals some interesting differences as compared to plugs 17D and 19A.

For example, despite the fact that this sand was characterized by an order of magnitude higher chloride (26 vs. 2 ppm), sulfate (40 vs. 4 ppm) and magnesium (16 vs. A,3 ppm) content, this plug had essentially no corrosion.

This result is consistent with the GPUN results for plugs 19C and 15A sand, where no-corrosion plug 15A was characterized by higher chloride, sulfate and magnesium in the sand.

The key difference in corrosion response lies not with the relative contamination levels in the sand, but rather the moisture content.

As is shown in Table 5, the sand sample behind plug 11A-H was dry as opposed to 1.1 - 2.6% moisture for plugs 17D and 19A, respectively.

The absolute difference in the contamination levels of the sand are significant on a percentage basis, but not on a corrosion basis.

The key here is the absence of an electrolyte.

6.0 DISCUSSION The results presented in the previous sections on the analysis of various sand, plug, deposit and water samples indicate that a suitable environment for the corrosion of carbon steel is present in the sand cushion area.

In other words, the corrosion of the drywell as exposed to this particular environment could not be considered unexpected.

The question is whether the amount of corrosion is particularly high and what role did other factors such as the Firebar-D insulation, contaminants, differential aeration, red lead primer, or concrete play in the corrosion phenomenon.

6.1 General Factors Affecting the Corrosivity of the Sand Cushion There are numerous factors which would affect the corrosivity of the sand cushion relative to the carbon steel drywell.

These factors include the sand porosity, electrolyte conductivity, contaminant level, moisture level, acidity/alkalinity and the presence of bacteria.

-1 A-

F&PMT Transmittal No. 87-178-003 Rev. I The relative porosity of the sand cushion would be affected by the method of back-filling the sand into the sand cushion region during construction, the settling-of the sand, the Initial moisture content of the sand, whether it was subsequently wetted after installation, etc.

The porosity of the sand would affect the moisture that could be retained in the sand cushion and the establishment of local areas of high aeration.

The more porous the sand the more moisture would be present over an extended period of time and the more optimum the degree of aeration.

Both of these factors would tend to increase the initial corrosion rate.

The degree of aeration of the sand would also affect the type of corrosion products formed on the steel surface.

6 For example, studies by Romanoff have indicated that In well-aerated soils the rate of pitting/corroslon, although initiall.y high, falls off rapidly with time because in the presence of an abundant supply of oxygen, oxidation and precipitation of iron as ferric hydroxide [Fe(OH) 3 ] occurs close to the metal surface, and the protective membrane formed in this manner decreases the subsequent corrosion rate.

As noted on the plug specimens from the Oyster Creek drywell only shallow pitting was observed on some of the specimens.

In poorly aerated regions, Romanoff noted that the initial rate of corrosion decreases slowly, if at all, with time.

Under such conditions the corrosion products, remaining in the deoxidized state, tend to diffuse outward into the soil, offering little or no protection to the corroding metal.

(The actual corrosion products observed on the drywell will be discussed in more detail below.)

The role of conductivity of the sand cushion is more straight forward.

The higher the conductivity, the greater amount of corrosion would be anticipated.

The conductivity of water samples removed from various drain lines at Oyster Creek ranged from 680 to 1100 WS/cm.

The conductivity of pure water at similar temperatures is three orders of magnitude lower than these values.

Hence, the sand/water environment was sufficiently conductive to establish a viable electrolyte for corrosion.

F&PMT Transmittal No. 87-178-003 Rev. 1 As noted in Tables 2 through 6, the sand, scrapings and drain water had high levels of contaminants which would be expected to increase the corrosion rate of carbon steel.

In particular, high levels of detrimental chloride and sulfate were noted in virtually all the samples analyzed.

The mere fact that corrosion occurred at Oyster Creek indicates that moisture was present in the sand cushion.

As discussed in Section 3.0, the sources of moisture include a known leakage of water from the fuel pool which most likely occurred through a-drain line gasket, installation of moist sand during construction, water "squeezed" out of the Firebar-D slurry during pressure testing of the drywell and 'condensation.

The moisture content of the sand samples as measured by GE ranged from 1.1 to 12.6%.

The only dry sand sample was from plug llA-H, which did not suffer any significant corrosion.

High pH is beneficial for the corrosion resistance of iron base alloys.

The pH observed from sand and drain water samples ranged from 5.99 to 8.90, Table 2, 3 and 6.

Host of the pH values were somewhat greater than neutral pH 7.

However, average pH values alone can belmisleading.

As will be discussed later, the establishment of local anode and cathodic sites due to differential aeration will affect the local pH values.

Deaerated anodic regions will have a lower pH while the cathodic regions will have a higher pH.

Also sections of the drywell adjacent to the concrete would benefit from the high alkalinity of concrete.

Corrosion induced by microbes is a widely recognized phenomenon in a number of systems such as oil wells, pipe lines and municipal sewage.

Microbiologically influenced corrosion (MIC) has also been identified in nuclear power plants.

However, the role of MIC in this particular system is being independently investigated and is therefore beyond the scope of this report.

It should be noted that preliminary evidence presented during discussions of the drywell corrosion at Oyster Creek have indicated that the role of MIC, if any, is not considered to be significant.

F&PMT Transmittal No. 87-178-003 Rev. 1 6.2 Specific Influences on Oyster Creek Drywell Corrosion 6.2.1 Firebar-D 3)4 Due to the known high corrosivity of Firebar-D on steel, one of the primary motivations in the investigation of the corrosion of the Oyster Creek drywell was the determination of the role of Firebar-D on the corrosion mechanism.

As noted in Section 2.1.3.2, Firebar-D is composed of MgO, MgCl2 and water.

Studies by Billnski, et al 7 have revealed that 5MS(OH2 )*MgCI 2 -

8H20 is the predominant reaction product in mechanically-sound hardened magnesium oxychloride cement.

This material is extremely sensitive to exposure to'water since there is an extremely narrow concentration range of magnesium and chloride ions in solution in which 5Mg (OH) 2 MgCl 2

• 8H2 0 is stable.

It is the presence of leachable MgCl 2 which can produce severe corrosion problems.

The specific corrosivity of magnesium oxychloride cements has been 8

investigated by Kawaller.

Observations of steel exposed to direct contact with damp magnesium oxychloride reveals a distinctive dark black rust (y-Fe 2 03 ), typical of corrosion which occurs in either a low oxygen or a caustic environment.

XRD investigations by GE specifically designed to determine the presence of Y-Fe 2 03 were negative.

An analysis of the chemical structure by Kawaller revealed that when magnesium oxychloride is exposed to 100% humidity, leaching of surplus magnesium chloride results in the formation of magnesium hydroxide.

Carbon dioxide extracted from the atmosphere combines with this material to form a surface layer of magnesium chlorocarbonate [Mg(OH)j MgCl 2. 2MgCO3.

6H2 01.

This surface layer slows the leaching process.

As additional MgCl 2 is leached, a surface crust of hydromagnesite (5MgCO2.

4C02

• 5H 20) is formed.

These insoluble carbonates and hydromagnesites help to improve the weathering stability of magnesium oxychloride materials.

I F&PMT Transmittal No.

87-178-003 Rev. 1 The lack of Y-Fe 2 03 in the oxide on the core plug surface/crust, the relatively low amount of Mg in the sand samples and the absence of corrosion at the 51' elevation level suggests that the role of Firebar-D in the degradation of the Oyster Creek drywell corrosion phenomena is not significant.

The levels of chloride and magnesium identified in the various laboratory samples may be as much the result of the marine environment as the leaching of the Firebar-D.

The formation of the insoluble carbonates and hydromagnesites discussed above may have reduced any potential pontrlbutlon of Firebar-D to the corrosion reaction.

6.2.2 Contaminants Table 17 presents the typical constituents of seawater..

A comparison of Table 17 and the results of leachate analyses, Tables 2 and 3, the drain water analysis, Table 6,.and the deposit analyses, Tables 7 and 8, reveal that many of the contaminants observed in these analyses could be from the Oyster Creek marine environment.

In partilcular, the presence of Ba, Al, Br, B, Ca and Sr can be explained.

However, the boron and strontium may be from the fuel pool as discussed in Section 3.0.

The primary role of any of the ions in the corrosion of the Oyster Creek drywell would be the enhancement of the electrolyte, that is, an increase in the conductivity.

A secondary role for these ions, and in particular, chloride and sulfate, would be the breakdown of any passive film established on the carbon steel surface.

As seen in Figure 19, higher concentrations of chloride are observed at the drywell wall-oxide layer interface.

The presence of the higher chloride at this interface may be the result of the alternate wetting and drying of the sand cushion.

Regardless of the source of the contamination, that is, the marine environment and/or the Firebar-D, the presence of such known "bad actors" as chloride and sulfate will increase the corrosion rate of the drywell.

.2~ ~...

F&PMT Transmittal No. 87-178-003 Rev. 1 6.2.3 Differential Aeration In most systems which are in contact with atmospheric oxygen, geometrical situations arise where transport of oxygen through the solution by convection (natural or forced) and diffusion to one part of the metal occurs rapidly, whereas It is slow or even negligible at another.

The areas characterized by high oxygen will serve as cathodes where the reduction of oxygen to hydroxyl occurs:

02 + 2H20 +4e-40H Areas depleted in oxygen will become anodic with the corrosion of the carbon steel:

2+

Fe -

Fe+

2e Therefore, areas of the sand cushion adjacent to ready oxygen access such as lower regions near the drain line and upper regions near the insulation gap would become cathodic while areas in the middle of the sand cushion would become anodic.

UT measurements appear to-verify this topographical evaluation.

Also, differences in local concentrations of NaC1 may result in differences in oxygen concentration as suggested by Schaschl and 10 Marsh.

The higher the salt concentration the lower the solubility of oxygen so that these depleted zones become the anodic zones of the differential aeration cell.

6.2.4 Role of the Red Lead Primer The outside of the drywell was painted with red lead which is lead oxide, Pb 3 04, or more accurately Pb 2PbO4, in linseed oil.

Water reaching the surface dissolves a certain amount of pigment and makes the water less "corrosive."

In general, corrosion inhibiting pigments must be soluble enough to supply the minimum concentration of inhibiting ions necessary to reduce the corrosion rate, yet not so soluble that the are soon leached out of the paint.

F&PMT Transmittal No.

87-178-003 Rev. i The inhibiting ion for red lead is probably PbO4" 4 which can passivate steel.

However, in the presence of S04 or CO2 the passivating effects of red lead can rapidly disappear.

Sulfate was identified in many of the analyses and carbon dioxide is readily available in the atmosphere.

It was noted throughout the analysis of the removed core plugs that lead was found on the surfaces of the plugs that suffered minimal or essentially no corrosion.

It is strongly believed that lead was found on such samples because no corrosion occurred due to the lack of moisture (dry sand) and not due to corrosion inhibition of the red lead paint.

Red lead paint alone'simply does not provide long term corrosion protection.

6.2.5 Role of Adjacent Concrete Concrete provides an alkaline environment and, under moist conditions, passivates iron and steel.

Regions of the sand cushion/drywell adjacent to the concrete could be benefited by this local alkaline'environment.

This factor can explain why the lower regions of the drywell below the 8'11.25" elevation which are in direct contact with the concrete did not suffer any measurable corrosion.

Since part of the drywell is in contact with the passivating concrete and part of the drywell in contact with the moist-high conductivity sand;,a macro-galvanic cell is established.

This will result in the acceleration of the corrosion of the drywell in contact with the wet sand cushion.

As will be discussed in Section 6.3, the presence of chloride in the sand will only amplify this effect.

6.3 Relevant Corrosion Reactions It is considered prudent to briefly examine the possible corrosion reactions which are occurring on the surface of the drywell embedded in moist sand.

Iron (steel) ions will go into solution at anodic areas in an amount electrochemically equivalent to the reaction at the cathodic areas.

As noted F&PMT Transmittal No. 87-178-003 Rev. I earlier, the anodic areas of the drywell are characterized by the following basic oxidation reaction:

Fe -

Fe++

2e The relevant cathodic reaction in aerated solutions is the reduction of oxygen to hydroxyl ions:

02 + H2 0 + 4e 40H However, the corrosion of iron or steel Is not as straight forward as illustrated above.

As shown in Figure 35, numerous corrosion reactions can occur depending on the local oxygen concentration, inter alia.

As will be discussed below, the presence of chloride and sulfate as observed in the Oyster Creek sand cushion, also affects the corrosion reactions.

In the absence of chloride and sulfate, Figure 35, hydrous ferrous oxide (FeO.

nH2 0) or ferrous hydroxide [Fe(OH) 2 ] composes the diffusion barrier adjacent to the iron surface through which oxygen must then diffuse."

The pH of saturated Fe(OH) 2 is approximately 9.5.

Pure Fe(OH) is typically 2~2.

white in color but rapidly oxidizes in air to green to greenish black.

At the outer surface of the oxide film, access to dissolved oxygen allows the ferrous oxide toreact to form hydrous ferric oxide or ferric hydroxide:

Fe (OH) 2 + 1/2 H20 + 1/402 --

0 Fe (OH) 3 Hydrous ferric oxide is orange to red brown In color and is the main constituent of "rust." It primarily exists as non-magnetic aFe2 03 (hematite) or magnetic YFe 2 03 where hematite is more stable.

Saturated Fe(OH) 3 has a nearly neutral pH.

A magnetic hydrous ferrous ferrite, Fe3 04

• nH20, often forms a black intermediate layer between the hydrous Fe 203 and FeO, Figure 35.

Therefore, as observed on some of the core samples from the drywell, rust films of various colors (states of oxidation) can exist simultaneously.

F&PMT Transmittal No. 87-178-003 Rev.

1 Motivated by the denting of carbon steel support plates in PWR steam generators, more sophisticated studies had been performed on the "rusting" of 9

carbon steel.

It is believed that this work performed by Pourbaix, et al is particularly relevant to Oyster Creek.

In particular, Pourbaix, et al were looking for conditions which would produce acid chloride attack of the carbon steel.

The mechanism proposed to explain this formation of acid is the hydrolysis of soluble corrosion products with formation, inter alia, of non-protective magnetite which is found in large quantities where denting has occurred.

If no contaminants are present (contaminants are ions other than

+~ OH ndF 2+

H OH-and Fe+), no acid hydrolysis would occur.

When contaminants (such as Cl, Br, SO 4 =) are present, no acid hydrolysis will occur provided there are no oxidants (such as dissolved oxygen) and no concentration by evaporation.

Problems may result from the presence of contaminants when concentration by evaporation occurs even without oxidants and from the presence of contaminants when oxidants are present, even without evaporation.

Since the Oyster Creek sand cushion is most likely characterized by all three factors (contaminants, oxidants and alternate wetting and drying concentration mechanisms), acid formation is expected.

The hydrolysis of ferrous ions in the presence of chloride or sulfate leads to acid and concentrated ferrous chloride or ferrous sulfate solutions:

2+

Fe -- ) Fe

+ 2e 2+

++H Fe

+ H20 --+ FeOH

+

FeOH+ + 2H+ + 3CI- --4 FeCl2 + HC1 + H20 FeOH++

4H ++

2S0-4 FeSO4 + H2S04 + H20 The corrosion rates of iron in these solutions are higher than in neutral or alkaline solutions.

For example, instantaneous corrosion rates were measured by Pourbaix, et al, at 212OF (higher than the drywell) in 4 molar FeC12 solution in closed system without an oxidant was 1.6 mils per year (mpy).

When in contact with magnetite, the instantaneous corrosion rate of iron increased to 8 mpy, and F&PMT Transmittal No. 87-178-003 Rev. 1 was well over 120 mpy when ferric contaminants were present.

Magnetite, as was identified in the plug crusts at Oyster Creek, is considered by Pourbaix as such an oxidizer and not a stable form of iron in mildly oxidizing environments.

The oxidizing power of magnetite is illustrated in Figure 36.

The stable form of iron is a ferric oxide or a ferric hydroxide.

Magnetite was considered as the normal and stable corrosion product of iron in boiler conditions because most boilers operate satisfactorily.

However, the opinion that protection is due to ferric oxide, and not to 9

magnetite, now receives more and more support.

At room temperature it has also been more and more generally accepted that magnetite Is not protective in the presence of aqueous solutions.

The passive films on iron in aqueous solutions at room temperatures appears to consist of Fe3 04 at the metal-oxide interface of of YFe 2O3 maghemite at the oxide-solution interface.

Although YFe 2 03 is difficult to distinguish from magnetite since maghemite is also black and magnetic, and has-the same XRD lines as magnetite, it was not identified in the GE analysis.

However, this similarity between magnetite and maghemite could be responsible for the long accepted opinion that magnetite is the protective oxide in boiler waters.

When magnetite particles are removed from the steel surface, they can be oxidized to hematite (WeFe 2 03 ), maghemite (WFe 2 03) or goethite (aFeOOH),

in 9

the presence of water containing as little as 1 ppb dissolved oxygen.

6.4 Corrosion Rate of Oyster Creek Drywell It is mandatory that the corrosion rate of the Oyster Creek drywell be estimated so that the present design life can be calculated.

A comparison of this value with corrosion rate data available in the open literature will also be useful in determining the relative corrosion performance of the drywell.

An estimation of the Oyster Creek drywell corrosion rate is straight forward since the reduced shell thickness as measured on the removed core plugs, Table 18, was approximately 0.85" and the initial thickness was F&PMT Transmittal No. 87-178-003 Rev.

1 approximately 1.15", the typical loss in thickness is "0.3".

If it Is assumed that the corrosion initiated with the installation of the sand 17 years ago, the average corrosion rate is approximately 20 mpy.

The assumed initiation date is considered realistic since the sand was installed in at least a "moist" condition, was wetted/rewetted during the expansion of the drywell which squeezed out the water from the Firebar-D slurry, and exposed to numerous condensation cycles.

If it is assumed that corrosion only initiated 6 years ago when the first fuel pool leak was noted, then the dstimated corrosion rate increases to approximately'50 mpy.

12-15 A review of the open literature on investigations concerning the corrosion of carbon steel in air saturated environments is semmarized in Table 19 and Figure 37.

Data was selected for only tests with reasonable exposure periods, that is, corrosion test data based on 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> exposure were not used.

In some:cases, ho*ever, the exposure period was not provided.

A more recent literature review performed for GPUN/EPRI on this subject by Pednekar 16 of Battelle Columbus Division also reveals similar corrosion rates.

It is interesting to note that the 20 mpy corrosion rate estimate for the Oyster Creek drywell falls among the data for carbon steel exposed to water ranging in quality from distilled to ambient seawater to a mixture of soils.

If the 6 year-based average of 50 mpy is used, the results are compa~able to Warsaw tap water or warm seawater.

The results of this comparison may, at first, appear somewhat surprising.

The discussion and results from Section 6.3 suggests that the sand cushion environment with high chloride and sulfate, oxygen, high conductivity, etc. would create an environment which would produce higher corrosion rates than specimens immersed in air saturated high quality water.

However, there are a few factors which may be reducing the overall corrosion rate of the drywell.

First, when a metal corrodes in a substance like sand, the sand tends to retain the corrosion products in place which physically stifles further corrosion.

For a specimen immersed inwater, the corrosion products can be transported away from the surface allowing corrosion to

F&PMT Transmittal No.

87-178-003 Rev. 1 continue physically uninhibited.

Second, during operation the sand cushion adjacent to the drywell may dry out and thus temporarily terminate any corrosion reaction.

When the sand is rewetted due to condensation and/or leaks, corrosion reinitiates.

This last scenario would evolve an overall lower average corrosion rate, that is, a combination of high corrosion followed by long periods of dormancy, 16 Pednekar notes that the corrosion rates, corrosion ptoducts, and pH changes observed in the Oyster Creek drywell corrosion are those that are observed for corrosion of carbon steels in aerated, chloride solutions.

6.5 Possible Corrosion Scenario of Oyster-Creek Drwel=-Degradation As illustrated in Figure 38, a series of factors/events most likely affected the corrosion of the Oyster Creek drywell.

Such a corrosion scenario is listed below:

1. Backfilling of moist sand into the transition zone creates an initial electrolyte.

Sand is contaminated by open exposure to marine environment during storage and installation.

Backfilling also affects porosity of sand which affects moisture retention quality and creates random air pockets.

2.

Expansion of drywell during pressure testing "squeezes" water out of the Firebar-D slurry which flows down into the sand bed.

This water contains initial high quantities of chloride and sulfate.

3.

Corrosion of the steel drywell initiates.

Red lead primer

-4 provides some initial protection due to the formation of PbO4 However, carbon dioxide from the air and sulfate from the sand/or Firebar-D accelerate the breakdown of the limiting inhibitive qualities of the red lead primer.

F&PMT Transmittal No. 87-178-003 Rev. i

4.

Areas with more ready access to oxygen such as the insulation gap and drain become local cathodes.

5.

Areas adjacent to concrete are provided some corrosion protection due to local alkalinity.

A macro-galvanic cell is established between the steel adjacent to the concrete and the steel'adjacent to the sand cushion.

6.

Condensation cycles and leaks from fuel pool bellows gasket contribute air saturated water to maintain moist sand cushion.

Additional chloride and sulfate may leach out of Firebar-D and be carried into the sand cushion.

7.

Some regions of the sand cushion see alternate wetting and drying during startup/shutdown cycles.

This results in a concentration of chloride at the metal/sand interface.

8.

Sand maintains corrosion products close to metal surface and thus physically stifles corrosion rate.

9.

Corrosion proceeds intermittently.during "wetting" periods (condensation, leaks) or on a continuous basis.

7.0 RECOMMENDATIONS The loss of containment integrity at Oyster Creek is an obvious concern.

The corrosion mechanism is fairly well defined and an estimated overall corrosion rate of %,20 mpy has been established.

It is now time to address this problem and identify potential mitigation steps for this phenomenon.

At the specific request of GPUN, three areas of mitigation have been analyzed by GE;

1) polymer replacement/addition to the sand cushion;

2) corrosion inhibitors, and, albeit superficially, 3) cathodic protection.

F&PMT Transmittal No. 87-178-003 Rev. 1 7.1 Polymer Replacement/Addition to Sand Cushion' 7 It has been suggested that the removal of the sand cushion could be accomplished by sluicing, If the sand cushion was removed and if the subsequent void was dessicated, corrosion of the drywell would essentially stop.

However, due to structural requirements on the containment, which are beyond the scope of this paper, it may be necessary to refill the sand cushion void with an alternate material which would have suitable mechabical properties.

Therefore, GPUN has requested that a brief review be performed on candidate cushion materials with particular emphasis on polymers.

The first concern for a polymer replacement would be identifying a suitable means of injecting the material into the void.

It would be possible to spray pellets of plastic through numerous core holes cut through the containment.

Although there would be some procedural -difficulties, it should be possible to obtain a relatively uniform distribution of plastic pellets.

Scrap material such as polycarbonate resin (e.g. Lezan) and thermoplastic resin (e.g. Noryl) are available.

Lexan and Noryl can withstand doses of approximately 8X10 6 and 1X10 8 rads, respectively, before any structural damage occurs.

Above these total dose levels, the materials would experience degradation by cracking.

However, this cracking and eventual fracturing would probably have little effect an its structural qualities to serve as a transition cushion.

Since, in this particular application, the sand and plastic would obtain their respective spring constants more from the voids in the cushion rather than any intrinsic material property, both materials should have similar spring constraints.-

However, it is recommended that this assumption be confirmed by a soil geologist.

If it is desired to have a cushion with more support strength, then any candidate polymer should be able to be applied in a sufficiently fluid state so that it could be poured into place.

This material could then completely cover the steel surface and fill the sand cushion void.

Since

F&PMT Transmittal No. 87-178-003 Rev. 1 there will probably be no opportunity for heat curing, then the candidate material should be characterized by an ambient temperature cure.

If a particular polymer is not completely wettable, it may form a crevice against the steel surface which can promote localized corrosion if any electrolyte is allowed access.

Therefore, assurance that any poured-in-place polymer adheres well to the steel must be obtained.

Good adhesion will also depend on the skill with which the monomer or partially polymerized resin is installed.

Epoxies would probably be the best candidates for an intrawall resin injection since any remaining sand would behave as a filler.

The epoxies would also be likely to adhere to steel surfaces.

The short "pot life" and the viscosity of the epoxy resins would make application troublesome; in addition epoxies are relatively costly.

Coal tar epoxies would probably be the best candidates.

Presumably a "Nuclear Grade" material (i.e., one especially low in halogens, sulfur, and embrittling metals) would not-be needed.

An epoxy spray paint could be used if the main concern is to protqct the steel surface.

If the sand can be removed, possibly a coal tar epoxy paint could be sprayed or poured into the intrawall region (if.e, Napko 538 Amine Coal Tar Epoxy).

Then dry sand might be re-installed into the intrawall area for mechanical support, if necessary.

Napka 539 Aluminum Mastic Epoxy is an aluminum powder-filled polyamide epoxy paint that is good for application to "minimally cleaned" surfaces.

It satisfactorily penetrates residual rust on steel surfaces and generally wets steel surfaces thereby assuring more thorough coverage.

Napko 682 Splash Zone Barrier Coating is an epoxy amide capable of application under water, if required.

If the sand is not removed, a paint may still be used since spray paint versions of epoxies or other resins would be more fluid than the corresponding resin and would be more likely to penetrate the sand and reach the steel surface.

There is no obvious way to assure that complete steel

-.17-

FSPMT Transmittal No. 87-178-003 Rev. 1 surface coverage in the sand area is obtained.

The most that might be accomplished would be assuring that excess paint is introduced to the intrawall region i.e., there Is enough paint present for the sand to be saturated and coat all the entire drywell wall.

For fillers that may provide support as a substitute for or an addition to the sand, materials generally used as temporary sealants for valves, flanges, and pipes might be suitable.

These materials *ould be the elastomers (fiber-reinforced, the fiber usually being glass) used for leak sealing by Leak Repairs Inc. (Division of Team Inc., Houston, TX) or by Furmanite Inc. (Virginia Beach, VA).

It would not be possible to use these materials with fiberglass if the sand was not removed.

However, it may still-be desirable to paint the steel surfaces first.

If a polymer with good mechanical strength is desired, then materials might be used that are applied like "potting" polymers 'used for electrical insulation of motors (i.e., pouring of the prepolymerf ked material into place in a large holding container).

However, the highest strength material, (20 ksi UTS) would be 20% glass-reinforced polyaryletheretherketone (i.e., PEEK).

The polymer is castable at 700*F but is currently rather costly.

It is available from ICI Americas Inc., (Wilmington, DE) or from a licensee (Greene, Tweed Engineered Plastics, Harleysville, PA) under the trademark "Arlon".

Arlon is injection-moldable.

Arlon 1260 (carbon fiber-reinforced PEEK) has a 30 ksi UTS.

Addition of polymer resin to existing sand precludes the use of fiber-containing resins.

There is no assurance that sand as a polymer filler would add to the strength of a polymer; such a filler, in fact, usually results in a weaker product.

A high hardness polyetherurethane polymer would provide up to 7 ksi UTS and a 15% carbon fiber-reinforced aromatic polyetherurethane would provide approximately 18 ksi UTS.

A styrene-maleic anhydride copolymer with 20% glass fiber reinforcement and proper processing may have a 10 to 14 ksi UTS.

The only other high strength materials approaching that of PEEK, are the fiber-reinforced epoxies.

Injection-grade, 20% glass fiber-reinforced ABS

F&PMT Transmittal No. 87-178-003 Rev. 1 have a 10 to 13 ksi UTS.

Silicone/nylon 6, 6 pseudo interpenetrating networks (i.e., Petrach Systems, Bristol, PA), made by mixing the two components into a powder or granular form, may achieve 10 to 12 ksi UTS.

7.2 Corrosion Inhlibitors17 The primary problem with corrosion inhibitors involves obtaining a uniform distribution over the entire surface of the steel or, ap with the paint discussed above, corrosion may become focused at unprotected areas.

At the same time, some prohibited inhibitors (i.e., chromates) may have to be avoided.

Limited life or short-term inhibitors are not useful unless the sand cushion area is made virtually airtight.

Therefore, inhibitors that operate by scavenging oxygen may not be usable.

However, those that promote protective oxide formation on steel surfaces appear to be the most promising.

The difficulty is in identifying all of the required properties for this inhibitor in one inhibitor.

Volatile inhibitors are usually of the type that scavange oxygen thereby making them limited-life inhibitors.

Yet water-soluble but volatile corrosion inhibitors would be most likely to provide complete coverage of the steel surface of the sand cushion area.

Molybdate could be used as a replacement for chromate to provide an inhibitor that promotes protective oxide-film formation on steel; but molybdate is not available in a volatile form.

Sodium molybdate is available from Noah Chemical, Farmingdale, NY.

Molybdate corrosion inhibitors, but only for cooling water, have been studied by Houseman (Burnham) Ltd. of the Portals Water Treatment Group in Great Britain.

Molybdates are also available from Climax Molybdenum Co., Calgon Corp.,

Exxon Chemical Co.,

Magna Corp., Nalco Chemical Co., Newage Industries Inc., and R.T. Vanderbuilt Co.

A water-soluble (in case of the presence of any liquid-phase moisture), volatile corrosion inhibitor such as one that might be used for packaging or in long-term storage is the only type feasible for sand cushion use to inhibit further steel wall corrosion.

Co rtac Corporation (St. Paul, MN; contact Boris Miksic) is outstanding in this area.

They have produced a I

F&PMT Transmittal No. 87-178-003 Rev. I volatile, water-soluble inhibitor dicyclohexylammonium chromate (U.S.

Patent 4,275,835; June 30, 1981).

They may also have the molybdate version of this inhibitor or the chromate may possibly be acceptable for Oyster Creek since it is not likely to escape the sand cushion region).

Just as was the case for coatings, incomplete coverage by an inhibitor can concentrate corrosion in unprotected areas.

However, some corrosion inhibitors pose another problem.' Nitrites, for example, shoul4 be avoided since there-are certain moderate concentration ranges (depending on other environmental parameters) which would promote corrosion Instead of inhibiting it.

If the presently existing sand is not removed, volatile corrosion inhibitors may not work.

The sand will readily absorb this type of corrosion inhibitor.

In fact, this would also be true of any inhibitor applied as a solution, aqueous or otherwise.

The same problem exiits here as it did for considering the use of resinous fillers or paints in the presence of the existing sand: a sufficient excess of inhibitor, as a solution or as a vapor, must be used so as to assure that the inhibitor reaches the steel wall and coats it completely.

Otherwise localized corrosion may occur.

Since liquids will be absorbed throughout the sand more readily than vapor, an oil-soluble version of an inhibitor may be suitable for application in this case.

Cortec Corporation has oil-soluble versions of its inhibitors.

Using one that is oil-soluble and volatile may be suitable since it would help ensure coating of the steel wall with the inhibitor in one manner or another.

Cortec VCI-320 would be one such product.

Preservativepetroleum lubricating oils could also be suitable.

Examples are Oilcoat VT and Oilcoat A (formerly Gulf products but now Chevron products), Mobil VaporTech Light Oil, SACI-100 (Witco Chemical Corp.) and Tower 640RP (Tower Chemical Corp., Palmer, PA).

Similar materials may be available from other sources, but it is best to use a product containing a volatile corrosion inhibitor.

(Note that these materials are bound to very flammable.)

F&PMT Transmittal No. 87-178-003 Rev.

1 Pednekar also provides a list of organic and inorganic inhibitors for carbon steel in aerated chloride solutions16 7.3 Cathodic Protection Basically, cathodic protection is a means of reducing the corrosion of a component by making the metal a cathode by means of an impressed current or attachment to a sacrificial anode (such as magnesium, aluminum br zinc).

Since the cathodic protection (CP) system forces electrons into the metal creating this cathode, the basic principle of applying CP is quite simple.

In

.general, the practical application of this corrosion Control method is much more difficult.

For the specific case of the Oyster Creek drywell, it may be-extremely difficult.

For example, CP systems are designed to protect coated structures, that is, provide protection against any defects (holidays) in the coating.

This minimizes the required applied current for protection.

For Oyster Creek, the drywell is presently uncoated and therefore significant and perhaps prohibitive currents may be required.

Other concerns include what source of direct current should be use; can a suitable anode be designed and, in fact, installed around the entire sand cushion; and how can it be ascertained, on completely buried structure, whether or not the entire surface has, in fact, been made a cathode and all corrosion mitigated.

Information which can answer such questions are beyond the scope of this report.

7.4 Mitigation Recommendation It appears that a multiple approach should be used for the mitigation of corrosion of the Oyster Creek drywell.

Since it appears that the main source of the corrosion problem is the wet chemically contaminated sand, the most suitable mitigation step would be the removal of the sand and drying of I

F&PMT Transmittal No. 87-178-003 Rev. 1 the cavity.

This, by itself, would reduce the corrosion rate of the drywell to a vanishingly small level.

If no structural support is required, a further corrosive mitigation improvement would be a back spray painting of the drywell to provide coating protection with an aluminum powder-filled polyamide epoxy paint followed by application of a volatile corrosion inhibitor to mitigate any holidays in the coating.

If the sand cannot be removed, then the application of an excess quantity of an oil-soluble vapor phase inhibitor may be the best approach.

If excess water is a problem, then an application of an excess of a sufficientl.

diluted epoxy paint such as Napko 682 Splash Zone Barrier epoxy amide may be the best choice.

This paint application could then be followed by the excess application of an oil-soluble volatile corrosion inhibitor.

As noted above, the use of cathodic protection as a suitable corrosion mitigation step is considered beyond the scope of this review and therefore will not be discussed.

8.0 CONCLUSION

S The results of metallurgical analysis by both GPUN and GE, data from the open literature and the above discussions have indicated the following conclusions concerning the corrosion of the Oyster Creek drywell:

1. The corrosion/wastage of the drywell is due to the presence of oxygenated moist/wet sand and exacerbated by the presence of chloride and sulfate in the sand cushion.

A contaminate concentrating mechanism due to alternate wetting and drying of the sand cushion may have also contributed to the corrosion phenomenon.

1, *7

F&PMT Transmittal No.. 87-178-003 Rev. 1

2.

Although Firebar-D is a known corrosive agent to steel, its role in this phenomenon is probably secondary.

The source of contaminants in the sand cushion may have been primarily from the local marine environment.

3.

Since the wall thickness measured by UT are extremely close to those measured on actual removed specimens, UT appears to be an accurate non-destructive method of monitoring wall thickness.

4.

The estimated corrosion rate of the Oyster Creek drywell is '20 mpy.

This rate reflects the average corrosion over 17 years of service regardless of the relative continuity of the corrosion reaction, i.e. there may be periods of high corrosion rate activity during wetting cycles followed by dormancy during "dry" periods.

5.

Excluding cathodic protection which is beyond the scope of this report, the optimum method of mitigation of the corrosion of the Oyster Creek drywell appears to be the combination'of sand removal, back spraying of a protective paint and application of a volatile corrosion inhibitor.

F, ~

V F&PMT Transmittal No. 87-178-003 REFERENCES

1.

GPUN Safety Evaluation No. 000243-002, December 12, 1986.

2.

C.R. Judd, "Leaching Tests and Chemical Analysis Results for Oyster Creek Drywell Samples," FMT Transmittal 87-212-0004, January 23, 1987.

3.

G.T. Austin, Shreve's Chemical Process Industries, Fifth ed.,

McGraw-Hill, New York, 1984.

4.

C.A. Sorrell and C.R. Armstrong, "Reactors and Equilibria in Magnesium Oxychloride Cement," Jour. of ACS, Vol. 59, No. 1-2, Jan-Feb.,

1976.

5.

J.E. Lewis Letter to B.M. Gordon, "X-Ray Diffraction Analyses of Corrosion Crust. from Inner Surface of Oyster Creek Drywell," December 19, 1986.

6.

M. Romanoff, Underground Corrosion, National Bureau of Standards, 1957.

7.

Bilinski, et al, "The Formation of Magnesium Oxychloride Phases in the Systems MgO-MgClI-H 0 and NaOH-MgCl 2 -H_0,"

Journal of the American Cermaic Society, VoL. 64, No.

4, April 1984.

8.

S.I. Kawaller, "Update on Magnesium Oxychloride Fireproofing," Fire Technology, Vol.

13, May 1977.

9.

M. Pourbaix, et al, "Chemical Aspects of Denting in Steam Generators,"

NP-2177, EPRI, Palo Alto, CA, December 1981.

10.

E. Schaschl and G.A. Marsh, Corrosion, Vol. 16, 1960.

11.

H.H. Uhlig, Corrosion and Corrosion Control, John Wiley and Sons, New

York, 1971.

12.

H.H. Uhlig, editor, Corrosion Handbook, John Wiley and Sons, New York, 1948.

13.

W.E. Berry, et al, "Survey Report on Corrosion Behavior of Carbon Steel in Pure Water at Ambient Conditions,"

June 30, 1975.

14.

L.L. Shreir, editor, Corrosion, Newness-Butterworths, London 1976.

15. N.E. Hamner, Corrosion Data Survey, NACE, Houston, Texas, March 1974.

16.

S.P. Pedneckar, "Corrosion of Carbon Steel in Aqueous Environments at Temperatures Below Boiling - A Literature Review," Battelle Columbus Division, February 10, 1987.

17.

R.S. Tunder letter to B.M. Gordon, "Information Relating to Drywell Containment Wall Corrosion at Oyster Creek," January 8, 1987.