ML20199H664

Forwards Request for Addl Info Re 970324 Response to NRC RAI Re GL 92-08, Thermo-Lag 330-1 Fire Barriers, for Peach Bottom & Limerick.Requests Response within Sixty Days to Support Review Schedule

ML20199H664
Person / Time
Site:	Peach Bottom, Limerick
Issue date:	11/14/1997
From:	Rinaldi F NRC (Affiliation Not Assigned)
To:	Hunger G PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC
References
GL-92-08, GL-92-8, TAC-M82809, NUDOCS 9711260187
Download: ML20199H664 (7)

Text

Mr. George A. Hunger, Jr. November 14, 1997 Manager-Licensing, MC 62A-1 PECO Energy Company Nuclear Group Headquarters Correspondence Control Desk P.O. Box 195 Wayne, PA 19087-0195

SUBJECT:

REQUEST FOR ADDITIONAL INFORMATION FOR LIHERICK GENERATING STATION, UNITS 1 AND 2, AND fcACH BOTTOM ATOMIC POWER STATION, UNITS 2 AND 3 - THERMO-LAG RELATED AMPACITY DERA11NG ISSUES AND GENERIC LETTER 92-08 (TAC NO. M82809)

By letter dated March 24, 1997, you submitted a response to the NRC Request for Additional Information (RAI) related to Generic Letter (GL) 92-08, "Thermo-Lag 330-1 Fire Barriers," for Peach Bottom Atomic Power Station (PBAPS) and Limerick Generating Station (LGS). The NRC staff, in conjunction with its contractor, Sandia National Laboratories (SNL), has completed the preliminary review of the licensee's submittal, and has identified a number of open issues and concerns requiring clarification (Enclosure 1).

We request that the licensee )rovide its response w; thin 60 days, to support our review schedale. If you 1 ave any questions regarding the enclosed RAI (Enclosure 2), please contact me at <l5-1447.

Sincerely,

/S/

I I

Frank Rinaldi, Project Manager Project Directorate I-2 Division of Reactor Projects - I/II Office of Nuclear Reactor Regulation Docket Nos. 50-352, 50-353 50-277, 50-278

Enclosures:

1. SNL Letter Report )/

2. Request for Additional Information cc w/encis: See next page DISTRIBUTION:

Docket File OGC f " yaep y '7 r,m -

PUBLIC ACRS - e, _-c Fa me PDI-2 r/f WPasciak, RI BBoger JCalvo JStolz RJenkins M0'Brien LTran FRinaldi lllllllllll, l

• MPadovan OFFICE PDI-2/PH , J PDI-2/l@ PDI-2/ b NAME FRinaldh h YO'Bhen JStolzk DATE // / /4/97 ll / N/97 Il /N /97 Y "*d AL RECORD COPY l 7 A. uT NAME: AMP 2. LGS l 9711260187 971114 D l PDR ADOCK 05090277 P PDR

1 I

.- . paa4 9 3 t UNITED STATES i

-lo j NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. soteH001 4,,,,,* November 14, 1997 Mr. George A. Hunger, Jr.-

Manager-Licensing, MC 62A-1 PECO Energy Company Nuclear Group Headquarters i Correspondence Control Desk P.O. Box 195 Wayne, PA 19087-0195

$UBJECT: REQUEST FOR ADDITIONAL INFORMATION FOR LIMERICK GENERATING STATION, UNITS 1 AND 2, AND PEACH BOTTOM ATOMIC POWER STATION, UNITS 2 AND 3 - THERMO-LAG RELATED AMPACITY DERATING ISSUES AND GENERIC LE1TER 92-08 (TAC NO. M82809) .

I By letter dated March 24, 1997, you cubmitted a response to the NRC Request i for Additional Information (RAI) related to Generic Letter (GL) 92-08, i "Thermo-Lag 330-1 Fire Barriers," for Peach Bottom Atomic Power Station I (PBAPS) and Limerick Generating Station (LGS). The NRC staff, in conjunction 1 with its contractor, Sendia National Laboratories (SNL), has completed the '

preliminary review of the licensee's submittal, and has identified a number of open issues and concerns requiring clarification (Enclosure 1). i We request that the licensee provide its response within 60 days, to support our review schedule. If you have any questions regarding the enclosed RAI  :

(Enclosure 2), please contact me at 415-1447.

Sincerely,

!. bf Frank Rinaldi, Project Manager Project Directorate I-2 Division of Reactor Projects - I/II Office of Nuclear Reactor Regulation Docket Nos. 50-352, 50-353 50-277, 50-278

Enclosures:

1. SNL Letter Report

2. Request for Additiona Information -

cc w/encls: See next page w

J

e _.-

]

PECO Energy Company- Peach Botton Atomic _ Power Station,' I Units 2 and 3 Limerick Generating Station, t Units 1 and 2 cc:-

J. W. Durham, Sr., Esquire Chief-Division of Nuclear Safety- -

Sr.-V.P. & General Counsel PA Dept. of PEC0_ Energy Company _

Environmental Resources 2301 Market Street, S26 P.O. Box 8469 Philadelphia, PA 19101' Harrisburg, PA 17105-8469 PEC0 Energy Company ,

ATTN: Mr. T.-N. Mitchell, Vice President Board of Supervisors--

Peach Botton Atomic Power Station Peach Botton Township 1848 Lay Road R. D. il Delta, PA -17314 Delta, PA 17314 PECO Energy Company Public Service Commission of Maryland ATTN: Regulatory Engineer, A4-55 Engineering Division Peach Bottom Atomic Power Station Chief Engineer 1848 Lay Road 6 St. Paul Centre <

Delta, PA 17314- Baltimore, MD 21202-6806 Resident Inspector Mr. Richard McLean U.S. Nuclear Regulatory Commission Power Plant and Environmental Peach Bottom Atomic Power Station- Review Division

. P.O. Box 399 Department of Natural Resources Delta, PA 17314 B-3, Tawes State Office Building Regional Administrator, Region I U.S. Nuclear Regulatory Commission Manager - Limerick Licensing, 62A-1 1475 Allendale Road PECO Energy Company King of Prussia, PA 19406 965 Chesterbrook Boulevard Wayne, PA 19087-5691 Mr. Roland Fletcher Department of Environment Mr. Walter G. MacFarland, Vice President 201 West Preston Street Limerick Generating Station Baltimore, MD -21201 P.O. Box A Sanatoga, PA -19464

, A. F. Kirby III External Operations - Nuclear Plant Manager Delmarva Power &_ Light Company Limerick Generating Station

-P.O. Box 231 - P.O. Box A Wilmington, DE 19899 Sanatoga, PA 19464 Chairman Board of-Supervisors of Limerick Township 646 West Ridge Pike Linfield, PA- 19468

2-r PEC0. Enert, Company Peach Botton Atomic Power Station, l' Units 2 and 3 Limerick Generating Station,

. Units 1 and 2 Director Site Support Services Senior Resident Inspector-Limerick Generating Station U.S. Nuclear Regulatory Commission P.O. Box A -

Limerick Generating Station-Sanatoga, PA -19464 P.O. Box 595

- Pottstown, PA 119464 Manager - Experience Assessment Library _

Limerick Generating Station - U.S. Nuclear Regulatory Commission

-P.O. Box A -

Region ~I Sanatoga, PA 19464 475 Allendale Road King of Prussia, PA 19406

-Senior Manager - Operations Dr. Judith:Johnsrud-Limerick Generating Station National Energy Committee P.O. Box A Sierra Club Sanatoga, PA 19464 433 Orlando Avenue State College, PA 16803 Director - Site Engineering Manager-Peach Bottom Licensing Limerick Generating Station PECO Energy Company P.O. Box A Nuclear. Group Headquarters

'Sanatoga, PA 19464 Correspondence Control Desk P.O. Box No. 195 Manager-Business & Co-owner Affairs - Wayne, PA 19087-0195 Public Service Electric and Gas Company Mr. George A. Hunger, Jr.

-P.O. Box 236_ Director-Licensing, MC 62A-1

'Hancocks Bridge, NJ 08038-0236 PECO Energy Company Nuclear Group Headquarters PECO Energy Company ~

Correspondence Contro1~ Desk Plant Manager P.O. Box No. 195 Peach Bottom Atomic Power Station Wayne, PA 19087-0195 1848 Lay Road Delta, PA 17314 ,

O Manager-Peach Bottom Licensing PECO Energy CompanyE LNuclear Group Headquarters Correspondence Control Desk P.0, Box 195 Wayne, PA 19087-0195 A.'A.--'__. , . _ _ _ _ _ _ _ . . _._..N - .

,.N.,. ., ,, , . . . . .- +v.,,- +, r- ,, , _ , _ . . . + , , . , . . -

c , ,n,. .

-.,14 .

< 'An Initial Review of the Proposed PECO Ampacity

- Assessment Methodology for Limerick and Peach Bottom l

A Letter Report to the USNRC ,

N P

September 23,1997 Revision 0 Prepared by:

. Steven P.Nowlen Risk Assessment and Systems Modeling Dept.

Sandia National Laboratories Albuquerque, New Mexico 87185 0747 ,

Prepared for: .

Ronaldo Jenkins Electrical Engineering Branch Office of Nuclear Reactor Regulation U. 5. Nuclear Regulatory Cc.amission <

Washington,DC 20555

, USNRC JCN J2503 QAjV}i '

. Enclosure 1

a a_w - . ~ . , -

._ ~ . - a_

TABLE OF CONTENTS:

Smaion 'Eaan

FORWARD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv .

1.0 INTRODUCTION

. . . . ................................................I 1.1 Objective . . . . . . . . ..............................................1

- 1.2_ Overview of the Licensee Ampacity Derating Approach . . . . . . . . . . . . . . . . . . . . . I 1.3 Organization of Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2.0 THE LICENSEE THERMAL MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.1 Objective of this section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.2_ Intended Model Applications and Approach to Review '. . . . . . . . . . . . . . . . . . .-. . . 3 2.3 The Licensee Modeling Approach . . . . . . . . . . . . . . . . . . . . . . . . . -. . . . . . . . . . . . 3 ,

' 2.4 _ . Specific Features of the Thermal Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.5 - Specific Model Features of Potential Concern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.5.1 Modeling ofInternal Convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

-2.5.2 Radiation Shape Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

'2.5.3 Concerns Specific to Junction Box and Gutter Models . . . . . . . . . . . . . . . . I1 3.0 VALIDATION OF THE LICENSEE APPROACH . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1 A Potential Pitfall to the Licensee Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.2 The Evaluation of Conduits . . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . 14 3.2.1 Consistency in the Overall Methodology . . . . . . . . . . . . . , . . . . . . . . . . 14 3.2.2 Conduit Case Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.2.3 . Recommended Supplemental Validation Cases for Conduits . . . . . . . . . . 16 3.3 The Evaluation of Cable Trays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.3.1 Consistency in the Cable Tray Clad and Baseline Analyse . . . . . . . . . . . . . 16 3.3.2 The Licensee Cable Tray Case Studies . . . . , . . . . . . . , . . . . . . . . . . . . . . 17 3.3.3 Recommendation for Supplementti Cable Tray Validation Studies . . . . . . 19 ~

3.4 Analysis Applications for Other Commodities . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4.0

SUMMARY

OF FINDINGS AND RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . 22 4.1 Thermal Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.2 Adequacy of Licensee Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 5.0 REFEREN C ES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .' . . . . . . . 2 4 iii

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

. .c '

FORWARD ,

~

l The United States Nuclear Regulatory Commission (USNRC) has solicited the support of Sandia National Laboratories (SNL) in the review of ficensee submittals associated with fire protection *

and electrical engineering. This letter report represents the first in an anticipated series of review reports associated with ampacity derating submittals from PECO Energy for the Peach Bottom i

Atomic Power Station (PBAPS) and the Limerick Generating Station (LGS). The submittal reviewed by SNL' documents an analysis methodology proposed by PECO for use in the assessment of the ampacity derating impact of Thermo Lag 330-18 fire barriers when installed on cable trays, conduits,junsion boxes, and cable gutters. This report document SNL's findings and ,

recommendations based on a review of this licensee submittal. The documents were submitted by the licensee in response to USNRC Generic Letter 92-08 and a subsequent USNRC Request for Additional Information (RAI). This work was performed as Task Order 4, Sub-task 4 of USNRC JCN J 2503, 9

4 2Thermo-Lag 330-1 is a registered trademark of Thermal Sciences Inc.

iv wr e , - - . .

. ~ .

1.0 INTRODUCTION

1.1 Objective In response to USNRC Generic Letter 92-08 and a subsequent USNRC Request for Additional Information ofJune 22,1995, PECO Energy has provided documentation of the methodoiogy that the licensee intends to utilize in assessing the ampacity derating factors associated with its installed fire barrier systems. The methodology is to be applied at both the Peach Bottom Atomic Power Station (PBAPS) and the Limerick Generating Station (LGS). SNL was asked to review the licensee's submittals under the terms of a general technical support task ordering agreement JCN J 2503, Task O. der 4, Sub-task 4. The licensee submittal review by SNL was documented in:

- Letter, G, A. Hunger, Jr., PECO, to the USNRC Document Control Desk, March 24, 1997 with two attachments.

The objective of this report is to document SNL's findings and recommendations regarding this licensee response.

1.2 Overview of the Licensee Ampacity Derating Approach The licensee approach is based on a direct application of ampacity derating factors (ADFs) to tabulated ampacity limits for the installed cables. The ADF values to be used will derive from one of two sources:

For plant configurations which correspond to configurations tested by other members of the industry, PECO will directly apply the experimentally determined ADF value.

For untested configurations, the licensee applies a thermal model to assess the clad case ampacity limits and determines an ADF value for the application based on a comparison of the calculated clad case ampacity to tabulated nominal ampacity limits for the same application. The intent is that these ADF values.would then be applied to each cable in the given application.

The SNL review has focused, naturally, on the licensee th:rmal modeling approach and its technical validity. The direct application of a test-based ADF to like cenfigurations is considered common and acceptable practice, so long as a sufficient basis for " thermal similarity" is established. In this submittal, the licensee has not provided specific case examples to illustrate the in-plant application of either metiiod; hence, no judgements in this regard have been made by SNL.

1.3 Organization of Report i

j Section 2 provides a detailed review of the licensee thermal model, including a discussion ofits l development and technical merit. Section 3 examines the specific validation case examples included in the licensee submittal, and assesses the adequacy of these assessments as a basis for I

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

--  ?. . ,

the acceptance.of the thermal modeling approach. - Section 4 summarizes the SNL fmdings and .

l recommendations. : Section 5 identifics the referenced documents.

f-

'I a

5 a

4 S

• n 2

2-

i

. 2.0 THE LICENSEE THERMAL MODEL i

2.1 Objective of this Section This section of the report will focus on an assessment of the technical merits of the licensee thermal model itself. This will begin with a general discussion of the licensee thermal modeling approach, an overview of the thermal model development and features, and a discussion of individual points of potential technical concern identified in the SNL review. Major findings and recommendations will be summarized in Section 4 below.

2.2 Intended Model Applications and Approach to Review The objective of the licensee thermal model is primarily to analyze untested banier configurations.

Hence, the standard single tray or single conduit cc igurations would not, in practice, be analyzed using this approach although the validation ;tudies do include some examples of this type. Instead, the thermal modelis intended to address unique configurations such as multiple commodities in a single barrier envelope. This could include combinations of trays and conduits, boxed conduits (as compared to conduits clad with pre-formed conduit sections of the barrier materiall nd clad electrical enclosures such as junction boxes and wire-ways. The model is also intended u treat barrier enclosures that include concrete walls as one or more sides of the barrier system.

It is quite clear that no licensee, nor indeed even the industny as a whole, could perform enough tests to encompass all such applications. In the view of SNL it is when one must assess these types -f unique installations that thermal modeling is most appropriate, and the SNL resiew has been based on this premiss. Given this " global" view, the questions which remain are the issues of the modeling details, conservatism, overall technical acceptability, and validation. It is these questions which have been the focus of this SNL review. The first three items, those related to the development of the thermal model are taken up here. The issue of validation will be taken up in Section 3 below.

2.3 The Licer,see Modeling Approach The ultimate objective of the PECO methodology is to predict the ADF impact for a specific commodity (conduit, cable tray, junction box, etc.) in a given fire barrier installation. The role of the thermal model in this process is to estimate the clad case ampacity limit for a " generic" commodity. The corresponding baseline limit is taken from standard industry tables of cable ampacity (IPCEA P-46-426 [1] and ICEA P-54 440 (2]). The ADF is then based on a comparison of the predicted clad case ampacity limit to the nominal tabulated baseline ampacity limit for the same " generic" commodity. It is the intent of the licensee that this same ADF value would then be applied to baseline ampacity limits for specific installed cables for a final assessment of ampacity loads. The method for calculation of the ADF values is critical and bears repeating:

- The licensee approach is to derive a fire barrier ADF by comparing a predicted clad utse ampacity limit from the thermal model to tabulated ampacity limits for the corresponding baseline case.

3

.i As a general observation, SNL has noted in several previous reviews of other licensee submittals ~

that this can be a very tenuous and potentially inappropriate approach to assessment. The  ;

concerns focus primarily on issues of consistency between the clad and baseline cases. This point will be taken up in greater detail in Section 3 below. The balance of this section of the report will focus on the technical merits of the licensee clad case analysis thermal model only.

2.4 Specific Features of the Thermal Model In some senses, the thermal model is tailored to each specific case under consideration. In particular, the analysis of a given case will use specific information on the physical geometry of the installed fire barrier system such as the number and size of the conduits and cable trays housed within the barrier, the oversil dimensions of the fire barrier enclosure, and the physical orientation of those commodities in couparisor, to each other and the fire barrier system.

In other senses the thermal model uses some common simplifying assumptions for all cases. In particular, the analysis utilizes " generic" assumptions regarding the cable size and cable fill characteristics of the commodity under study. That is, the analyses are not based on specific in-plant cable loading conditions, but rather, on a generic cable loading condition for each commodity in the barrier enclosure. It is the intent that the generically derived ADF values would then be applied to individual cables in the actual installation as a final basis for assessmant.

As a general observation, SNL finds that this is an acceptable and conservative practice. In particular, this approach allows no credit for cable load diversity. All cables in the generic simulation are assumed to be fully and uniformly loaded within a given commodity (tray or conduit e.g.). This, in and ofitself, is a conservative approach to analysis and does avoid a number of potential complications that would result if the model were to include diversity effects.

With regard to the thermal model itself, the licensee has incorporated assumptions that are both conservative and non-conservative in nature. Generally, the licensee has utilized widely accepted, though somewhat dated, heat transfer correlations and concepts. SNL did examine certain of the case examples presented in the submittal. While these examples are somewhat limited in the level of presentation detail, it appears that all aspects of the thermal model have been implemented consistent with the text of the submittal. No serious numerical errors were noted, and SNL was able to reproduce the numerical results of those calculations reviewed in detail. (The example calculations are taken up in greater detail in Section 3 below).

General features of the PECO thermal model are as follows: ,

- For all surfaces, both radiation and convection are credited in the analysis. This includes both within the fire banier envelope and at the external surfaces of the envelope.

- In the analysis of heat transfer at the external surfaces of the barrier enclosure all surfaces of the fire barrier system are credited equally. This includes the sides of the barrier system.

4

- In convective heat transfer, no corrections for surface orientation have been applied.

Rather a single " conservative" correlation has been applied to all surfaces regardless of orientation.

-- Heat transfer to' and through concrete walls is credited if such walls form one or more sides of the fire barrier system. The treatment is based on steady-state one-dimensional conduction within the concrete and convection and radiation heat transfer at both the inner and outer surfaces of the wall. This treatment is considered appropriate and conservative.

- The model has at least nominally treated all of the important thermal features including heat transfer effects within a cable mass or bundle, heat transfer between the cables and the barrier, heat transfer through the barrier including potential air gaps, and heat transfer from the barrier to the ambient.

In section 2.4.3 below, specific features of the thermal model that SNL finds to be questionable wil' be discussed. Hewever, there are many specific features of the thermal model that SNL will not take any exception to. For some of these items the licensee treatment is sufficiently unique that some explanation of SNL's review findings is con:idered appropriate to clarify SNL's conclusions as to why the treatment is considered acceptable. These include the following observations:

- The licensee has uted a linear heat transfer coeflicient form of radiative heat :ransfer correlations. This is a common practice, but if not properly handled can introduce some significant error in problems where radiative heat transfer is important. The licensee treatment in this case is appropriate because iterative procedures have been established to recalculate the linear coeflicients until a " temperature match" is obtained. This iterative procedure adequately accounts for the non-linear effects of radiation and eliminates any potential error associated with the linear treatment.

- For the treatment of convective heat transfer, the licensee has applied a simplified analysis that is ,,omewhat at odds with common practice. First, SNL notes that the cited general correlation, while somewhat dated, is appropriate to the treatment of external surfaces. (External here rcfers to surfaces in an open, unrestricted ambient environment such as on the exterior of the fire barrier system. Specific concerns related to internal surfaces will be discussed below.) The correlation chosen is of a form common to simplified correlations for such treatment.

In a typical analysis a different set of coefficients would be applied in the general correlation for each surface of the system depending on the surface orientation (e.g., top, bottom, sides, round versus flat, upward versus downward facing). This is because orientation is critic' .e efliciency of convective heat transfer. In contrast, PECO has utilized a single x . coefficients for all surfaces. While somewhat unusual, SNL agrees with the licensee that the specific coefficients chosen for use in the general correlation are, in balance, conservative for the external surfaces. In particular, the cited heat transfer coefficient is conservative in comparison to either an upward facing hot plate (such as the top of a barrier system), a vertical hot plcte (such as the sides of a barrier system), or heated cylinder (such as an individual clad 5

^

I

' ~

conduit). While the cited correlation is somewhat non-conservative for a downward facing hot plate (such as the bottom of a fire barrier system), the anticipated impact will be compensated for by the conservat:sm inherent in the treatment of vertical surfaces and upward facing surfaces. SNL recommends acceptance of this simplification for this application given that the licensee has cited its basis for the chosen values, and that the values will be conservative in net effect.

- For certain of the licensee barrier systems, one or more sides of the enclosuie may be comprised of concrete walls. The licensee has credited heat transfer to and through these surfaces in a relatively conservative manner. In particular, the licensee has treated the concrete walls in a manner similar to the treatment of the other fire barrier panels crediting only one-dimensional heat transfer through the walls. This will be somewhat conservative because in reality two-dimensional heat transfer effects would enhance the rates of heat transfer into the walls. The licensee has not credited any two-dimensional effects in its analysis, and this is conservative. It should be noted that any attempts to credit two-dimensional heat transfer would quickly become quite complex and is likely beyond the scope of any simple ampacity calculation. SNL finds the licensee approach to this aspect of the analysis to be appropriate to the calculations objectives.

- In the treatment of conduits, the thermal resistance betwcen the cable bundle hot spot and the surface of the conduit itselfis a very important parameter. The evaluation of this parameter is often based on the works of Neher [3,4] but this approach can lead to inconsistencies between the clad and baseline case unles: the baseline case is also fully evaluated. The licensee approach to this problem was somewhat unique, and did resolve this potential concern. This issue is taken up in Section 3.2 below.

- One important parameter in the thermal analysis of a cable tray or cable bundle is the effective thermal conductivity ofv e bundle. This value is critical to characterizing heat transfer effects within the cables. In each of the example calculations which cite a numerical value for this parameter, the licensee has cited a value of 0.09 BTU /hr*ft**F. While the source for this value is not cited,it apparently derives from a USNRC SER cited elsewhere in by the licensee (see PECO'ref. 9). The SER had, in turn, cited the results of certain calculations performed by SNL that were later verified by testing'.

The licensee cited value is consistent with that used by SNL in our own modeling, but is somewhat lower than that commonly used in such assessments, and in particular, the value assumed by Stolpe [5] and used in the development of the ICEA tables [2); namely,0.15 BTU /hr*ft* *F. As a general observation, the use of a lower value for thermal conductivity will increase the temperature rise predicted within the cable bundle and therefore reduce predicted ampacity limits. Because the licensee model only considers the clad case, use of the lower conductivity value will imply some level of conservatism in comparison to Stolpc's model and the ICEA tables. It 2See "An Experimental Assessment of Thermal Conductivity in a Composite Electrical Cable Mass", A letter report to the USNRC, January 20,1995, forwarded under cover S. P.

Nowlen, SNL, to R. Jenkins, USNRC/NRR/EELB, work performed under USNRC JCN J2018, 6

should also be anticipated that the level of relative conservatism introduced would increase with increasing fill depth. Hence, SNL finds that the cited value is appropriate and does, in fact, represent a source of conservatism in the licensee cable tray calculations.

2.5 Specific Model Features of Potential Concern SNL does take excep'. ion to certain of the licensee treatments. These are features that in SNL's judgment may lead to non-conservatism in the calculation results. These points are considered of particular concern to this licensee submittal because the licensee is attempting to directly calculate clad case ampacities, and is then comparing those ampacity lirNts to tabulated ampacity values.

This implies a need for a his,her level of accuracy in the thermal model, and minor non-conservatism might significantly compromise the analysis results. The two areas of technical concern identified by SNL are discussed in the following sub-sections.

2.5.1 Modeling ofInternal Convection SNL has identified two potential problems in the licensees treatment ofinternal convective heat transfer In this discussion internal convection refers to convective heat transfer between the protected commodities and the trapped air pockets within the banier system, and in turn between the trapped air pockets and the inner surface of the fire banier itself.

The first and most serious problem is the fact that in treating convection within the fire barrier system the licensee has applied the same correlation as that applied to the exterior of the barrier system. The cited correlation is intended only to address convection from an open and unobstructed surface to a general ambient. The general application of this same correlation to the interior air spaces will be both inappropriate and non-conservative for many typical fire barrier installations.

For a general interior air space one mu:t consider that large-scale convective air flows typical of an external surface may not develop. Rather, a system ofless efficient, smaller-scale convective cells may develop, or air circulation may be suppressed altogether. For a typical fire barrier enclosure, the barrier panels are relatively tight fitting, and there is little room for the development oflarge scale convective air flows. For these cases it is appropriate to utilize confined space convection correlations (see for example Holman's section on " Free Convection in Enclosed Spaces" (6]).

SNL does acknowledge that the current *reatment may bejustified for larger enclosures that have large gapes between the commodities, and between each commodity and the surfaces of the enclosure system. (Examples would include the "large box" configurations tested by TVA/ Watts Bar (7-10] in which the barrier endosures were significantly larger than the protected commodity.) Basically, ifit is possible for large scale circulation patterns to develop within the fire barrier envelope, then the current treatment would be appropriate. This would in turn require that the air trapped within the barrier is free to circulate throughout the enclosure and that the air has free and unobstructed access to the majority of the internal surface (for example, cable trays in close proximity to a side wall would disrupt heat transfer to that surface, and it would be appropriate not to credit convection to the obstructed portion of the surface).

7

To resolve this concern, SNL recommends that the licensee be asked to justify its current ~

treatment ofinternal convection on a case-by-case basis for each application under study. For any case where the current treatment cannot be justified the thermal model should either not be applied or the intemal space convective heat transfer should be modified to confonn to accepted methods for the analysis of convection in a confined space.

The second concern is that it appears that the licensee analysis ofinternal convection for cases involving multiple commodities in a common barrier enclosure has treated the heating sources as independent rather than simultaneous. This treatment does simplify the analysis of each commodity, but may be inappropriate. It should be noted that this discussion will be primadly relevan' 'o those cases involving large internally unobstructed enclosures for which the internally trapped air can be realistically treated as a single air pocket rather than as separate individual air pockets. In particular, if the air pockets are treated as confined spaces, then each air pocket should be treated individually (for example, treating the air pocket above a clad tray as separate from the air pocket below the tray), and this discussion would not apply.

The problem arises in that there is in reality only one net internal enclosure surface available for the trapped air to communicate with. Two cases are identified by the licensee, and the problem is slightly different for each:

- The licensee treatment of general enclosures assumes that "each raceway interacts convectively with the full area of the enclosure." By treating each source independently, the thermal model will underestimates the air-to-barrier temperature drop needed to support the total rate of air-to-barrier heat transfer for all commodities.

This is because the air space must simultaneously transfer the heat load from all of the sources to the enclosure surfaces. In effect, the same heat transfer correlation applies, but the heat load should be the total heat load for all protected commodities, not that of any one individual commodity, (The problem is analogous to treatment of heat transfer through and away from the barrier itself which must also consider the total heat load from all sources, not that ofindividual sources, to obtain the correct overall temperature drops.)

- For"large enclosures containing more that two trays" one mitigating factor has been incorporated. For these cases "it has been assumed that each raceway . , exchanges heat with the enclosure wall over an area no larger than the surface area of the raceway itself." This may mitigate the impact of this concern, but may also be -

inappropriate, In particular, this practice would only be appropriate if the sum of the area for all the protected commodities is less than or equal to the total internal surface area of the barrier system. This is again an issue of over-crediting the available surface area in this step of the analysis which may occur if the area of the commodities exceeds that of the barrier enclosure, a possibility when there are multiple commodities in a common enclosure.

(Note that the quoted citations above appear on page 11 of the calculation.) Th licensee does appear to have some appreciation of the discrepancies being introduced based on the discussion in the submittal. However, SNL disagrees with the licensee assessment that the impact of this treatment is not significant. While radiation does tend to be somewhat dominant in the heat 8

p + i transfer process, convection can account for 25J 50% of the overall herat transfer. Hence, it is a important that appropriate methods of analysis be employed.-

- To resolve this concern, some modification of the licensee thermal model is recommended. 1 However, there is no one clear-cut method to achieving this goali The "most correct"_ a;.proach j would be to implement an iterative procedure to ensu_re that a single intemal air space temperature j is determined that is consistent with all of the projected commodities simultaneously, This would 2

require that the licensee iterate on the her .oad of each commodity until a balance between the --

assumed air temperature and the commodes is achieved. This would be a rather tedious and j cumbersome process because each commodity would add to the number of parameters to be -  ;

" juggled" to achieve balance.-

i

. There is, however, an altemative approach that is somewhat less accurate but would involve only  ;

a minor modification to the current model. This approach is analogous to, but an extension of,

'i the treatment currently used by the licensee for "large enclosures." That is, the licensee could continue to treat the commodity to barrier heat transfer as independent for each source, but in the _

convection analysis, each commodity should be assumed to communicate with only a limited portion of the intemal fire barrier surface area. The interaction area should be determined so that the ratio of the commodity surface area to the surface area of all commodities combined is equal '

to the ratio of the assumed convective area to the total intemal surface area of the barrier:

surface commoditYp bbarrier effective

y arrier total surface commodityg This approach works for this model only because the exact same heat transfer corrdation is used for all surfaces. Hence, the rate of heat transfer is scaled primarily by surface area. In eff'ect, this

- approach apportions part of the available enclosure surface area to each commodity in proportion to its own surface area. This approach will leave some uncertainty in the results, but this uncertainty will be relatively modest. The reccmmended supplemental validation cases discussed in Section 3 below should ensure that the final results are acceptable. ,

2.5.2 Radiation Share Factors 1

On page 12_of the licenses. submittal, the calculation of radiation shape factors is discussed. In general the discussions appear appropriate. However,in one specific case, the licensee '

calculation appears questionable.

This is the case of radiative exchange between a tray that is obstructed by another trsy (above or ,

below) and the side walls of the fire barrier system. The licensee cites a value of 0.3 for such

. situations.L While the licensee has cited an appropriate basis for calculating these shape factors i (the Handbook of Heat Transfer, perpendicular surface correlations) it is not clear that a value of

( ' O.3 is bounding for all likely cases. In particular, the calculated value would be dependent on both

. the width of the tray and on the separating distance between trays, and the licensee h

s not demonstrated that the cited value of 0.3 is bounding for its applications.-

L 9

1 L. = . _ - -. . - - .

It is unclear if the licensee's intent was to apply the cited value of 0.3 ' generically to all such cases, or wether thh, - 13 simply intended as an example.LThe text implies that this is intended as a .

generic value for all cases, and one of the example calculations appears to support this interpretation. In particular, consider the example case illustrated in Table 8 of the calculation. ,

This case involves al stack of two' power cable trays in a common enclosure. The' net radiation view factor for each tray is cited as 0.8, which is fully consistent with the licensee practice as

~

- described on page 12, including the assumption of a 0.3 tray-to gap view factor, but is not ~

consistent with the cited physical dimensions.

In this case,it is possible to directly calculate the correct view factor. Note in particular that the t

~ cable trays are cited as being 24" wide and 4" high each. The fire barrier is assumed to be 1.5"

.  ; thick, so two panels (top and bottom for example) have a net thickness of 3". The overall height of the exterior of the fire barrier is cited as 22.5" Hence, one can infer that the separation distance between the two trays is 11.5" (22.5"-(2'4")-(2

• l .5")= 11.5"). .

The calculation process for this shape factor is illustrated below, First, the physical configuration ofinterest is illustrated in the following figure:

l gTray-to-tray gap'(surface 2)

Cable Tray (surface 1)

W Each surface is assumed to be infinitely long (extending into and out of the page) consistent with

' the licensee assumptions. For this particular example, h=ll.5" and w=24". Given this

. configuration, we first define the variable (H) such that:

3 N= -

- 0.479 w 24 .

Using this variable, the shape factor for the tray-to-gap is given by [12)

F ., =

3 1+R-/1+H 2 and for this case in particular:

F3 ., = 1 ' + 0. 479 - /1 + 0. 4792 ) = 0.185-Hence, for this case, the net view factor for a given tray to the inside of the fire barrier should

! "have been cited as 0.685 (F=(l'O.5)+( 185*,5)+(.185*.5)) rather than the licensee cited value of l 10 i

L

--4 _ - - - , .- - .. . ~ . _ . . - . . . . . . _ . . _ __g-...-

1 .

0.8. This reduction i.i the view factor would represent a 15% reduction in the rate of radiative l heat exchange for this particular case.

i SNL finds that use of a generic shape factor of 0.3 for all tray-to side gap applications is technically inappropriate. ?t is recommended that tb licensee be asked to either (1) demonstrate and use a bounding value for all cases considered by the licensee or (2) ensure that shape factors i are appropriately calculated to correspond to the actual physical configuration on a case-by-case basis.

2.$.3 Concerns Specific to Junction Box and Gutter Models The licensee has also applied its thermal model to the analysis of both junction boxes and cable

" gutters." Both of these commodities have been treated in a manner that is unique in comparison to either the conduits or cable trays, and aspects of this treatment have not been adequately explained. For these applications, it is recommended that the licensee be asked to answer the following questions:

- The treatment for cables in junction boxes and cable gutters assumes that the cables remain in a bundled configuration. This assumption is acceptable and conservative.

However, while the treatment of heat transfer does consider the thermal conduction within the cable bundle itself, it is unclear how heat transfer between the cables and the enclosure (thejunction box or gutter), and between the enclosure and the barrier have been ha. 31ed. Please provide a more explicit discussion of these aspects of the thermal model for these applications. Please include a more detailed example calculation for each commodity that illustrates how these factors are analyzed.

- Please explain how the assumed baseline ampacity impacts the analysis ofjunction boxes and gutters. In particular, for conduits the baseline ampacity was used to estimate the cable to conduit thermal resistance values. Is a similar procedure used to assess the cable bundle to junction box or gutter thermal resistance values based on the cited baseline ampacity limits?

- What is the rational for treating the cable gutters uniquely in comparison to cable trays? Based on the physical descriptions provided by the licensee the gutters appear to be, in effect, narrow open top solid bottom cable trays. The one case example presented in Table 9 of the submittal appears to involve a 5" square gutter with a total fi'.1 depth of about 3.1" if treated as a cable tray. How would treatment of these commodities as cable trays alter the results of the analysis?

In considering this aspect of the licensee submittal,it should also be noted that the licensee analysis ofjunction boxes is rather unique. Most licensee's will simply assume that the derating associated with raceway installations (clad conduits) leading un to the junction box is the limiting condition. This is generally considered by SNL to be acceptable because (1) the cables spread out in a junction box minimizing mutual heating effects and maximizing the available heat transfer surface area, (2) the increased Sternal enclosure volume will enhance convective heat transfer rates, and (3) the available heat transfer surface area is vastly increased in the junction box as compared to a conduit. Further, the NEC standards specifically allow for the neglecting of 11

installation details over a short length of an overall cable run, and junction boxes typically meet

. this requirement for ampacity assessment exemption. For this reason, SNL considers that the licensee assessments ofjunction box ADF values is not necessary to a demonstration of adequate ampacity margin. An assessment of the ampacity loads for raceways leading up to these enclosures should be sufficient to ensure overall compliance. (This argument does not apply to the cable gutters.)

i t

12 I ._ - - - _ _ _ _

3.0 VALIDATION OF THE LICENSEE APPROACH 3.1 A Potential Pitfall to the Licensee Approach As noted above the licensee thermal model compares a clad case ampacity limit estimated by the thermal model to a baseline ampacity limit taken directly from standard tables of ampacity. In previous reviews SNL has cited such practices as a significant point of potential concern. SNL's concerns in this regard focus on the potential that such an analysis might be comparing " apples to oranges" because the tabulated baseline and the calculated clad ampacity limits may be based on entirely different thermal models.

SNL has repeatedly cited that consistency between the baseline and clad case analyses is critical to a reasonable estimation of fire banier ADF values. Our own experiences have also shown that absolute calculations of actual ampacity limits are difficult to achieve with accuracy due to inherent uncertainty in heat transfer modeling. In contrast, an estimate of ADF impact based on comparison of a clad case model to a self consistent baseline case model is relatively easy to achieve. Hence, in the view of SNL, the PECO approach carries an inherent uncertainty that should be explicitly addressed. As a general observation SNL notes that there are three identifiable paths to addressing this concem:

1. Implement a thermal model for the clad case analysis that is fully consistent with the models used to develop the tabulated ampacity limits (e.g. use a Stolpe-like model for cable trays if the P-54-440 tables are to be applied). This consistency should be demonstrated by analysis of the baseline case and demonstration of consistency with the ampacity tables. The clad case model should then be developed as an extension of the baseline case model in which all of the critical assumptions and correlations are maintained intact. This would then allow for a direct comparison of the clad case analyses and the tabulated baseline ampacity limits.

2. Develop and apply models for both the baseline and clad cases that are fully self-consistent and base the ADF estimates on comparisons between the two model evaluations. In this case consistency with the tables is not an issue, simply self-consistency between the clad and baseline case models. This is, perhaps, the most common approach utilized in practice.

3. Perform sufficient validation studies to demonstrate that the proposed approach is conservative in comparison to either the tabulated ampacity limits or in comparison to tested configurations for which an ADF limit has been determined directly.

This remainder of this section will consider the general approach take to the analysis of both conduits and cable trays, and will examine the relevant case examples for each as documented by the licensee. These discussions will include and assessment of the adequacy of these studies to resolve the concern identified immediately above; that is, self-consistency in the ADF assessments.

SNL will also make specific recommendations for supplemental validation studies.

13

3.2 The Evaluation of Conduits ,

3.2.1 Consistency in the Overall Methodology As was noted briefly in Section 2.3 above, the licensee thermal model for conduits has employed a key feature that effectively addresses the issue of consistency between the clad and baseline assessments. That is, as a part of the clad case analysis, the licensee has in effect performed a self-w sistent analysis of the assumed baseline case as well.

One of the critical parameters in the assessment of conduit heat transfer is the cable to conduit thermal resistance (or intemal hest transfer coefficient). Considerable attention has been focused on this problem as evidenced by t se works of Neher, et.al. [3,4). One common method of analysis is to apply the Neher correlations to estimate the cable to conduit thermal resistance. The same value can then be applied to assess both the baseline and clad case ampacity limits, and this ensures self-consistency.

PECO has followed a somewhat different approach. PECO has assumed the baseline ampacity limit using standard ampacity tables. This value is then used to estimate the internal heat transfer coefficient by analyzing the baseline case and matching the rates of cable heating to the rate of external heat transfer. Self consistency with the clad case is ensured because the same convection and radiation correlations are applied to both cases. The licensee can then "back-calculate" the internal heat transfer coefficient for the baseline case, and simply applies that same value to the clad case analysis.

As a secondary note, the licensee practice in this regard also introduces one source of conservatism for the conduit analyses. That is, in the analysis of the baseline case and the assessment of the internal heat transfer coeflicient, the licensee has assumed a conduit emissivity of 0.8, an upper bound estimate for this parameter. This results in a maximized estimate of the cable-to-conduit thermal resistance. While the thermal resistance value is carried directly to the clad case analysis, the emissivity of the conduit is reduced to 0.4 in that analysis, a lower bound estimate of this parameter. This practice introduces a significant additional level of conservatism into the clad case analysis.

This practice is critical to SNL's assessment of the technical acceptability of the licensee conduit thermal model. In effect, the licensee has fully analyzed the baseline case tising a thermal model that is fully self-consistent with the clad case analysis model. The only difference is that for the baseline analysis the licensee starts with a "known" ampacity and estimates the corresponding thermal properties, rather than starting with the "known" thermal properties and estimating ampacity. This includes use of the same external heat transfer correlations and the same internal cable-to-conduit thermal resistance factors for both parts of the analysis. This practice also ensures that the thermal parameters assumed in the model are self-consistent with the assumed baseline ampacity.

By virtue of this practice, SNL finds that the licensee thermal model as currently applied to conduits has adequately addressed the issue of baseline and clad case analysis consistency. While SNL has identified one unrelated technical concern relevant to the conduit analyses (the treatment 14

ofinternal convection as discussed in Section 2.5 above), SNL recommends that the overall approach as currently applied to conduits is acceptable.

3.2.2 Conduit Case Examples The licensee has presented three case examples invoMng the analysis of conduits, ont for a 2" conduit clad with a 1-hour barrier, one for a 5" conduit clad with a three hour barrier, and one for a group of four conduits in a common 3-hour enclosure. Each of these calculations is documented in a Table in the PECO calculation. These case examples and the insights gained from them are taken up in this sub-section.

- PECO Table 6 2" Conduit with a 1-hour barrier. This case example involves a conduit individually clad with a 1-hour barrier assumed to be comprised of pre shaped conduit sections. The predicted ADF for this case was 8.3%. This value was compared by the licensee to a value from a USNRC SER for Texas Utilities (TUE)

Comanche Peak Station (see licensee reference 9]. The licensee cites a test ADF of 6.6% for a nominally identical case. Unfortunately, PECO apparently failed to fully appreciate that the methods used by TUE left considerable uncertainty in the test results. The same SER cites a worst case estimate of the fire barrier ADF for this case of 21.5%. Hence, there is considerable uncerte:nty in the TUE test result, and that test does not represent a proper basis for model validation.

Unfortunately, there are no alternate tests of a 2" conduit available for direct comparison. As a generalindication of the anticipated ADF, SNL would cite both the TVA/ Watts Bar [7-10] and FPC/ Crystal River [11] tests. Both licensees have tested I" and 4" conduits, and have cited 1-hour ADF results ranging from (-)3.31% to

(+)3.5%. Hence, in comparison to these results the PECO estimates appear conservative for this particular case (a single clad conduit). SNL would anticipate that much of this conservatism derives from the 'ne of a high conduit emissivity for the baseline case and a low value in the clad case analysis as discussed in 3.2.1 above.

- Table 7: 5" Conduit with a 3-hour barrier. This case is quite similar to that given in PECO Table 6, except that the conduit diameter is increased to 5" and a three-hour barrier is assumed. The PECO estimate of the fire barrier ADF was 19.8%. No comparison to other calculations or to experiments have been made by the licensee.

The closest available experiment for a nominally similar barrier system would be the FPC/Crysta' River tests in which a 4" conduit with a 3-hour barrier was found to have an ADF ofjust 2.7%. In comparison, a 4" conduit tested by TVA/ Watts Bar with a base 3-hour barrier plus an upgrade using 2 layers of 3/8" thick Thermo-Lag 770-1 had an ADF of13%. In comparison to these results, the PECO estimate is

, again quite conservative.

l As an additional note, SNL did attempt to reproduce the licensee results cited in this case example. This exercise was largely successful within a reasonable limit of numerical error.

- PECO Table 11 Four conduits in a common 3-hour enclosure. This case example considers a common three hour enclosure assumed to be 1.25" thick, surrounding a l

! row of four,5" diameter conduits. The baseline case analysis should be essentially 15 I

s~ .

. identical to the baseline case analysis presented in the Table 7 example discussed : "

immediately above.- However, the result of this analysis is apparently different.

That is, for this case, the cable to conduit heat transfer coefficient is ' cited as -

2

' O.62 BTU /hr*ftF as_ compared to the Table 7 cited value of 0.59 BTU /hr*ft * *F.

The_only difference that SNL was able to note was that a different value of the cable d

electrical resistance was used (8.75'10 fin Table 11 versus 8.17* 10 (ohm /ft) in Table 1 7). This difference is sufficient to cause the change in values. In this case, self _

consistency is more important than; absolute values, and the self-consistency appears to -

have been maintained within each individual case. Hence, this is not considered a

- point of significant concern. .

The net result of the calculation was an estimated ADF of 19.8%. While there i

. is no experimental data for a four-conduit enclosure, TVA/ Watts Bar did test a 3-conduit enclosure in its own test program. The TVA results found an ADF of 8% for.

a nominally similar common enclosure with three 1" conduits. While the case considered by PECO would likely suffer a somewhat_more harsh ADF due to the presence of four conduits instead of three, it is likely that the PECO modeling result in ,

this case is conservative.

3.2.3_ Recommended Supplemental Validation Cases for Conduits The only direct comparison between the licensee thermal calculations and actual experimental results was that for a TUE test that had considerable inherent uncertainty. The results of this comparison, while interesting, are not considered a sufficient basis for validation of the thermal

- model for the intended applications. Hence, SNL finds the validation cases cited to be inconclusive. In particular, the licensee has not explored the full range of applications which are intended for the model, nor taken advantage of the full range of experimental results now

. available.

This is considered a serious weakness of the PECO submittal that can easily be rectified. That is, a direct comparison is possible given the availability in particular of the TVA/ Watt _s Bar test results. TVA performed three tests involving multiple conduits in a common enclosure (TVA test articles 7.4,7.5 and 7.8). Each test involved either three or six 1" conduits. For the six conduit array, tests were also performed for a small tight fitting enclosure and for an enclosure much larger than nominally required. It is recommended that the USNRC ask the licensee to validate its calculation method for roultiple conduits in a common enclosure by a direct comparison to these three specific TVA/ Watts Bar test sets. Also ofinterest would be the TVA tests involving a

- single conduit it either a large or small boxed enclosure (TVA test articles 7.7a and 7.7b). This case example would illustrate wether or not the thermal model is properly capturing the role'of the barrier surface area in the calculations. Validation against these two test cases is also recommendedi

_3.3? The Evaluation of Cable Trays ,

3.3.1 Consistency in the Cable Tray Clad and Baseline Analyses The consistency case for the PECO cable tray applications is markedly different from that of the PECO conduit applications. -In particular, the PECO cable tray method includes no equivalent 16 N.. ,n,,e.,.,,.w r. y,, .

e,.,.

analysis of the baseline cable tray case such as that incorporated into the conduit analyses. For cable trays, the PECO model has not implemented either of the first two approaches to ensuring ,

consistency as identified in Section 3.1 above; hence, it is by default dependent on the third option, direct comparative validation.

The licensee has, in fact, implemented a set of validation calculations, and has documented the results in the submittal. Unfortunately the validation cases documented are limited, do not directly correspond to any known experiments, and have not adequately explored or demonstrated the capab'.lity of the model to conservatively assess the intended applications involving multiple trays in a common enclosure.

b discussed in Section 2.4 above, one aspect contributing some conservatism to the PECO cable tray calculations in comparison to the ICEA ampacity tables is the fact that all of the PECO tray simulations use a cable mass thermal conductivity value of 0.09 BTU /hr* A* *F as compared to the Stolpe value of 0.15 BTU /hr*fP'F used in the development of the ICEA P-54-440 tables. The use of a lower value will result in higher predicted cable bed temperature rises, and somewhat '

lower clad case ampacity limits. Because the PECO model considers only the clad case in practice, this willintroduce some degree of conservatism. The degree of conservatism should increase with increasing fill depth.

However, other factors may be non-conservative. In particular, the licensee analysis of the fire barrier includes full credit for heat transfer through all sides of the barrier system. This may be optimistic if sections of the barrier are thermally blocked by cable tray side rails or by un-powered control and instrumentation trays. Also, as noted in 2.5 above, the licensee has utilized potentially non conservative methods in assessing internal heat transfer behaviors.

The net effect of these offsetting factors cannot be determined without adequate validation. The following sub-sections will discuss the license validation studies that have been presented, and will make recommendations as to what additional studies may be needed to resolve the remaining Concerns.

3.3.2 The Licensee Cable Tray Case Studies The licensee has documented several case studies for situations involving cable tray installations.

The insights derived from these case studies are summarized below:

- Table 3: Unprotected Trav Case Examele. This example case considers a single cable tray,24" wide, in a baseline configuration. This appears to be the only case considered by the licensee that was specifically intended to address a baseline configuration of any commodity (no baseline conduits, junction boxes, or " gutters" appear to have been explicitly considered). As such, while this case is not representative of the intended model applications, it is an important example that illustrates to at least some extent the degree of consistency one should expect : " ween the licensee thermal model and the baseline ampacity tables. That is, in acra applications the baseline conditions will be assumed to correspond to the k,cA P54-440 tables and will not be directly analyzed further This case example provides a 17

direct assessment of how the licensee modeling assumptions would compare to those i tables for at least one case.

The ultimate result of this calculation is an estimated baseline ampacity of 32,7 A for the case considered. This value can be compared directly to a nominal tabulated ampacity limit 34 A taken from Table 3-6 of P-54-440. The result of this particular calculation nominally indicates some conservatism in the licensee calculation as compared to the standard tables. SNL would attribute this largely to the lower assumed value for the cable mass thermal conductivity as has been discussed above.

This would also translate to a nominal conservatism in the ADF estimates because in practice the higher 34 A table value should be used in the ADF calculation.

- Table 4 24" Trav with 1" fill and 1-hour wrap. This case study considers a single 24" wide cable tray with a 1" depth of fill enclosed in a nominal 1-hour fire barrier wrap.

This case corresponds directly to a nominal single tray enclosure such as that tested in an IEEE P848 type ampacity test with the exception of the assumed cable fill depth. A P-848 tray will have a nominal fill depth of 3". As ::ited by the licensee and by SNL, the ADF impact will tend to decrease with increasing fill depth.

The results of the case study are an estimated ADF impact of 31.9%. In comparison, the licensee has cited certain results presented in an USNRC SER for the Texas Utilities Comanche Peak Station (licensee reference 9). The cited results were actually based on thermal modeling results generated by SNL. In this case, SNL considered an essentially identical tray configuration at various depth of fill values.

The licensee has apparently mis-quoted the SNL results. For the 1" fill case, the licensee cites an ADF from the SER of 31.5%. In reality, the SNL analyses had estimated the ADF impact for a 1" fill depth of 36.2%.

In this case, the licensee results are considerably more optimistic than are the SNL modeling results. It is suspected that the differences are directly attributable to (1) the licensee's non-conservative treatment ofinternal convection as discusse Section 2.5.1 above and (2) to the licensee crediting of heat transfer from the sides of the tray / barrier system. In contrast, the SNL results were based on confined space convection treatment and no heat transfer from the sides of the tray were credited.

The SNL results were also based on the direct comparison of modeling results for both the clad and baseline cases without any reliance on the ampacity tables. Most other features of the two models are nominally similar given this particular application.

Neither case can be compared directly to experiments because no known experiments have been performed using a 1" fill depth of cables. However, fer comparison, TUE obtained an ADF of 31.6% for a standard IEEE P848 tray that included a thin fiberglass blanket placed on top of the cable mass as a protective barrier. Removal of this blanket would very slightly reduce the ADF, but reduction in the fill depth should slightly increase the ADF. In balance, a modest increase in the ADF should be anticipated due to these changes; hence, the licensee estimate would appear modestly optimistic.

- Table 5: 24" Trav with 2" fill deoth and 1-hour barrier. The case presented in Table 5 is identical to that of Table 4 except in that the cable fill depth was increased to 2" instead of 1". The estimated ADF impact for this case was 32.3%. The licensee again cites the TUE/ Comanche Peak SER and compare:; this limit to the corresponding SNL I8

modeling result in which anniysis of the same case yielded a 30.5% ADF (conectly cited by th licensee in this cate). Again, no comparison to data is possible because no known tests have taed a 2' fill depth.

In this partiwlar case the licensee result is somewhat more conservative than the corresponding CE result. However, the trend with depth of fill is incorrect. That

!s, the 1.censee predicts an increase in ADF with an increase in fill depth when the SE results indicated a decrease in ADF with increasing fill depth. Intuitively, one should anticipate that the SE predicted trend is correct. Recall that ADF reflects the relative change in ampacity due to addition of the fire banier. For a cable tray, the limiting condition for the cables is at the center of the cable mass. As the thickness of that mass increases, a slpnificant ampacity penalty is realized. Hence, while the addition of a fire barrier does add an additional penalty, the relative importance is lower giun the penalty already "palo" due to the cable fill depth. The licensee calculations appear to contradict this expectation. It is suspected that again the inappropriate treatment of internal convection is a likely cause for this anomalous behavior. The optimistic treatment of banier heat transfer surface area may also have contributed to this result.

- Table 3: Two 24" trays in a common 3-hour enclosure. This particular case is the first involving multiple commodities, in this case two cable trays, in a corranon enclosure.

The two trays are assumed to be sta:ked one upon the other, both have a l fill depth, and tath are essumed to be powered with the same driving ampacity. The estimated ADT for this configuration was 48.8%. As noted in Section 2.5.2 above, this calculation was compromised to some extent by use of an overly optimistic radiation shape factor. This result is also impacted by the overly optimistic treatment ofinternal convection (there appears to be no justification for application of an open air conelation in this case because the size of the barrier is only nominally larger than the trays themselves implying a highly confined space). Hence, the cited result is likely somewhat optimistic.

The licensee has provided no comparison to experimental data, and indeed, no data for such a two tray configuration is currently available. The closest approximation to this case would be the TVA test of a three tray stack (in which enly the to,7 two trays were powered) clad by a 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> Thermo Lag fire banier. For reference, the TVA derived ADF for this case was 36%. The impact for a lower cable fill combined with a 3 hour3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> barrier should result in an ADF impact signifiantly higher than this value. While nominally consistent with this expectation, because several imponant parameters are differert (the number of trays, the depth of fill for each tray, the tray to tray gap height, the external surface area of the barrier, etc.) a direct comparison is not possible.

3.3.3 Recommendation for Supplemental Cable Tray Validation Studies The licensee calculations for cable trays are heavily dependent on an appropriate validation as has been discussed above. SE finds that the current validation studies are inadequate to demonstrate the acceptability of the licensee cable : ray model. In panicular, the licensee example cases do not allow for a direct comparison between expenmental data and the thennal ruodel.

l 19

'lhete are a number of experiments now available for which a direct comparison should be easily

• accomplished. This includes single cable tray wrap results for one test from TUE/ Comanche Peak, two Thermo Lag tests at FPC/ Crystal River, and two Thermo. Lag tests at TVA/ Watts Bar of which SE recommends that at least one be directly simulated. In addition, there is the one test by TVA of a common barrier, stacked tray configuration. This case woald be of particular interest because it is the only known experimental investigation of ampacity for a multiple tray fire barrier enclosure. It is recommended that the licensee be asked to validate their ther through a direct comparison to these experimental studies.

3.4 Analysis Applications for Other Commodities Only one case example for a cable guter (Table 9) and one example for a junction box have been presented. It is very difficult to draw meaningfulinsights from these two examples because there are no known tests or alternate analyses upon which comparisons can be drawn. In addition, as discussed in 3.4.1 above, there is eome considerable uncertainty as to how certain aspects of the model were implemented. Hence, SE has not spent extensive effort in the revie.v of these cases.

For the sake of completeness, the following observations are made for each of these two cases:

- Dble 9: A Cable Gutter This particular case invol"es the analysis of a cable " gutter".

In this case, the reference ampacity is apparently taken as the cable's open air ampacity limit (287 A). The result of the calculation is an estimated ADF of 61.2% in comparison to the open air ampacity limit. While this sounds rather harsh, the reliance on the open air limit a the ba.:.line ampacity has inflated the ADF limit.

SE has also considered the cited clad case ampacity limit in comparison to other potential sources of the baseline ampacity. For example if the ICEA P54-440 cable tray standard is applied, the calculated depth of fill is 3.1", and the corresponding ampacity limit would be 146 A. The clad case limit was estimated at 111.4 A, so the net ADF would be 23.5%. In comparison to known cable tray ADF values, this would appear nominally reasonable. Recall that this case involves a 5" by 5" gutter, and hence, the available surface area is quite large in comparison to the enclosed volume.

Hence, an ADF more modest that that of a typical IEEE standard tray should be anticipated.

An altemate treatment would consider the conduit baseline ampacity Using the IPCEA P46-426 limits coupled with the NEC correction factors for more than three conductors in a conduit, a baseline ampacity limit of 127.5 A was calculated by SE. Using the same clad case ampacity limit would imply a 14.5% ADF. Again, this would appear nominally reasonable in comparison to knc,w conduit ADF values.

- Table 10: Just, ion Box .121 The treatment of this junction box appears quite similar t s that of the gutter presented in Table 9 of the submittal. However, in this case the b;seline ampacity limit appears to have been taken as the corresponding conduit ampacity limit using the IPCEA P-46-426 value and correcting for more than three conductors in the conduit using the NEC correction factor of 0.35 for 41 or more conductors. Using this as the baseline, an ADF of 28.6% is kriveo.

There is no firm basis for assessing the merits of this value. Nominally, the value appears quite conservative. in particular, the junction box ADF should be markedly lower than would a corresponding conduit ADF because of the vastly 20

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

H, .

i 4

i

?

Increased surface area of the enclosure in comparison io a condnit, No clear l explanation for the apparently very harsh ADF impact was identified by SNL.

i t

i l

t i

j l

1 T

I 1

t l

i i

-i L

I i

c e

+

4 r

I h

21-

__, , . , _ . . .._ .,...__.;...;._,, . _ _ , , .... .,.,__.. ,, ,.._, .._ ,. _,..._ a._,

4.0

SUMMARY

OF FINDINGS AND RECOMMENDATIONS .

4.1 Thermal Model Development In general, SNL found that the PECO thermal model is well documented and has accounted for all of the significant thermal effects ofinterest. With two exceptions, as noted below, the licensee has utilized r>ppropriate assumptions and conclations. Based on a review of the case examples, the model appears to have been implemented consistent with the text discussion, and SNL was able to reproduce the licensee numerical results for certain selected case examples.

There were, however, two specific areas of technical concem identified by SNL that may contribute to non-conservatism in the model results. In addition, SNL was unable to determine how certain aspect of the heat transport problem were handled for applications involving junction boxes and cable gutters. It is recommended that the licensee be asked to address these concems as follows:

- SNL fmds that the licensee treatment ofinternal convection is not appropriate for general applications. Two points of specific concern were identified:

- The licensee has treated internal air spaces using correlations only appropriate to the analysis of open unobstructed surfaces. SNL finds that this treatment is inappropriate as a general model ofintemal convection behavior. It is recommended that the licensee be asked to either (1) justify its current treatment on a case by case basis for each case considered, and/or (2) implement an attemate treatment ofintemal ccnvection based on accepted methods of confined space convection analysis.

- The licensee has treated multiple heat sources in a common enclosure independently rather than simultaneously. This will overestimate the rates of convective heat transfer for many typical appucations. To correct this problem, it is recommended that the licensee be asked to modify its assumptions regarding the effective barrier heat transfer area available for convective exchange to each individual commodity. A specific approach to resolution has been documented in the text above. ,

- In the analysis of radiative heat transfer for stacked cable trays, the licensee apparently intends to utilized a generic value to characterize tray to side panel radiation view factors for intennediate (blocked) trays. SNL finds that the cited value has not been shown to conservatively bound the antic' pated value of this coefficient, and was non-conservative for the one relevant examp!: case cited in the licensee submittal. It is recommended that the licensee be asked so either (1) demonstrate that the generic value used is conservatively bounding for all cases to be analyzed, or (2) calculate the correct value on a case by case basis for each configuration (simplif.ed correlations for this process were cited by SNL).

- SNL finds that the thermal model as applied to junction boxes and cable gutters was not as ti.oroughly documented as were the conduit and cable tray models. SNL recommends that the licensee be asked to clarify its treatment of the following aspects of these applications: (1) How has heat transfer between the cable bundle and the 22

e .

l l

enclosure (the junction box or gutter) and between the enclosure and the fire barrier been handled? (2) How does the assumed baseline ampacity impact the results of the

. analysis? (.,) What is the rational for treating cable gutters in a manner unique from that of a general cable tray? Recommended questions to address each of these points l have been provide in Section 2.5.3 above.

4.2 Adecuacy ofLicensee Validation l i

SNL finds at adequate validation of the licensee model is critical to a demonstration of acceptability, especially with regard to the cable tray and multiple commodity calculations.

However, SNL also found the currer t licensee validation studies to be unconvincing because the calculations have not considered sufficient cases for which a direct ecmparison to available  :

)

experimental data can be made. SNL recommends that a more thorough validation of the model

- be requested. -In panicular, SNL recommends that the licensee be c.sked to a cur.1ent validation results for the following cases: ,

- In the case of the conduit calculations, it is recommended that the licensee be asked to document validation results for the conduit barrier enclosure tests perfo med by TVA for the Watts Bar plant as follows: (1) test anicle 7.4, three 1" conduits in a common, tight fitting enclosure, (2) test anicle 7.5, six 1" conduits in a common tight fitting enclosure, (3) test article 7.8, six 1" conduits in an oversize enclosure, (4) test article 7.7a, single 1" conduit in a small boxed enclosure, and (5) test article 7.7b, single 1" conduit in an oversized boxed enclosure.

- For the cable tray calculations, it is recommended that the licensee be asked to document validation results which are directly comparable to available experiments from TUE, TVA, and/or FPC. These should include at least one representative case for a single tray enclosure and the TVA 3 tray stack test (TVA test article 7.3).

No specific recommendations for funher validation of the gutter and Junction box models have been made because there are no known tests of such systems available for direct comparison.

A 4

23 m....-- - ,,.,x. -. - . . . . . , . , . - . , - - -- ,,...,--.w, , n.....c, -,. , , - , . . 7, , .-,, , , , , .,

.  ?

5.0 REFERENCES

1. Pawer Cable Ampacities, IPCEA P-46-426, AIEE S 1351, a joint publicatior of the Insulated Power Cables Engineers Association (now ICEA) and the Innlated Conductors Committee Power Division of AIEE (now IEEE),1962.

2. Ampacities ofCables in Open Top Cable Trays, ICEA P 54-440, NEMA WC 51,1986.

l

3. F.11. Bulles and J. II. Neher,"The Thermal Resistance Between Cables and a Surrounding Pipe or Duct Wall," AIEE Transactions, Volume 69, pgs 342-349,1950.

4. J. ll. Neher and M. II. McGrath,"The Calculation of the Temperature r'se and Load Capability of Cable Systems," AIEE Transactions, pgs 752-772, Oct.1957.

5. J. Stolpe, "Ampacities for Cables in Randomly Filled Trays," /EEE Transactions on Power Apparatus and Systems, Vol. PAS 90, Pt. I, PP 962 974,1971.

6. J. P.11olman Heat Transfer, McGraw liill, Fourth Edition,1976.

7. TVA Watts Dat: Testing to Determine Ampacity Derating Factorsfor Fire Protected Cablesfor Watts Bar Nuclear Plant, Central Laboratories Services Repon 93-0501, Revision 0, July 6,1993.

8. TVA Watts Dar: Ampacity Derating of Cables EncloseJin One-Hour ElectricalRaceway Fire Barrier Systems (ERFBS), Omega Point Laboratories Report 11960 97332,97334-6,97768 70, March 28,1995.

9. TVA Watts Bar: Ampacity Derating of Cables Enclosed in Cable Tray with Thermo-LagD 330-1/770-1 Upgrade Electrical Raceway Fire Barrier Syster s (ERFBS), Omega Point Laboratories Report i1960-97333, June 30,1995,

10. TVA Watts Dat: Ampacity Derating of Cables Enclosed in Condu'Its with 1hermo-LagD 330-1/7701 Upgrade Elect?lcalRaceway Fire Barrier Systems (ERFBS), Omega Point Laboratories Report 11960 97337 & 97338, August 21,1995.

1l. FPL Crystal River: Ampacity Test Innstigation ofRaceway Fire Barriers For Conduit and Cable TraySystems, Underwriters Laboratory Report Number 95NK17030NC1973, May 7,1996,

12. R. Siegel, and J. R. Howell, 7hermalRadiation Heat 7ransfer, McGraw liill, Second Edition,1981 (tee Appendix C item 13 for cited correlation).

24

t -

[INERICK GENERATING STATION. UNITS I AND 2 m J PEACH BOTTOM ATONIC POWER STATION. UNITS 2 AhD 3 )

DOCKET NOS. 50-352. 50-353. 50-277. AND 50-278 l REQUEST FOR ADDITIONAL INFORMATION REGARDING GENERIC LETTER 92-08 FIRE BARRIER AMPACITY DERATING ISSUES

1.0 BACKGROUND

l By letter dated March 24, 1997 PECO Energy Company (the licensee) submitted a response to the NRC Request for Additional Information (RAI) i related to Generic Letter (GL) 92-08, "Thermo-Lag 330-1 Fire Barriers,"  !

for Peach Bott e Atomic Power Station (PBAPS) and Limerick Generating  !

Station (LGS).

The licensee will apply to plant installed fire barrier configurations

! the applicable ampacity derating factor (ADF) values which vere tested ,

by the industry. The licensee approach for any untested configurations is to apply a fire barrier ADF by comparing a predicted clad or fire barrier enclosed case ampacity limit from its thermal model to the tabulated am)acity limits for the corresponding t,ase line case. The licensee's tiermal model utilizes heat transfer wthodologies to predict or estimate the clad case ampacity limit for pneric electrical ,

raceway commodity. The corresponding base ~ e limit is taken from standard industry cable ampacity tables (i.. , Jr.sulated Power Cable Engineers Association (IPCEA) and Insulated Cable Engineers Association (ICEA)). The ADF value for the fire barrier configuration is then based on a comparison of the predicted clad am>acity limit to the nominal tabulated base line ampacity limit for tie same generic comodity. The Itcensee would then apply the'same ADF value to base line ampacity limits for specifically installed cables as part of a final assessment ,

of the acceptability of ampacity loads.

The staff, in conjunction with its contractor, Sandia National  ;

Laboratories (SNL), has completed the preliminary review of the licensee's submittal, and requires that the following questions be ,

addressed by the licensee.- 1 2.0 OUESTIONS f

After a review of the licensee's cable ampacity assessment methodology, SNL raised the following concerns: ,

p i

Enclosure 2

i -t-2.1 - SNL finds that the licensee thermal model development had the following

- areas of technical concern which may contribute to non-conservatism in the model  ;

t Modeling of Internal Convection l The licensee has treated internal air spaces using correlations only ,

appropriate to the analysis of open unobstructed surfaces. SNL  :

finds that this treatment is inappropriate as a general model of internal convection behavior. It is recommended that the licensee i either 1 justify its current treatment on a case-by-case basis  ;

for each c(a)es constriered, and/or; implement an alternate  ;

treatment of interns 1 convection ba(2)d se on accepted methods ef confined space convection analysis.

The licensee has treated multiple heat sources in a common enclosure independently rather than simultaneously. This treatment will overestimate the rates of convective heat transfer for many typical 4

applications. To correct this problem, it is recommended that the ,

licensee modify its assumptions regarding the effective barrier heat transfer area available for convective exchange to each individual  :

commodity. A specific approach to resolution has been documented in  ;

the SNL Letter Report dated September 23, Igg 7, (Enclosure 1).  ;

Radiation Shape Factors

- In the analysis of radiative heat transfer for stacked cable trays, the licensee apparently intends to utilize a generic value to ,

characterize tray-to-side panel radiation view factors for intermediate (blocked) trays. SNL finds that the cited value has not been shown to conservatively bound the anticipated value-of this coefficient, and was nonconservative for the one relevant example i case cited in the licensee submittal. It is recommended that the l licensee either: (1) demonstrate that the generic value used is conservatively bounding for all cases to be analyzed; or -

(2) calculate the correct value on a case-by-case basis for each configuration (simplified correlations-for this process were cited i in the subject SNL Letter Report.

I i Concerns Specific to Junction Box and Gutter Models

- SNL finds that the thermal model as . applied to junction boxes and ,

. cable gutters was not as thoroughly documented as were the conduit and cable tray models. SNL recommends that the licensee clarify its

-treatment of the following-aspects of these applications: (1) How has heat transfer between the cable bundle.and the enclosure (the junction barrier beenboxhandled?;

or gutter)(and 2) Howbetween the enclosure does the assumed baselineand the fire ampacity

+

u____.. ,. .,-...,_-....,.,_....m_-,, ..,,._.,....._.,.__.-,,,.--c.,.m._...,.y-._y, , . - . , ,_..m,.,

L ..

i

. i l

r f

impact the results of the analysis 7; and (3) What is the rationale for  !

4 treating cable gutters in a manner unique from that of a general cable tray?  !

In conclusion, the licensee is requested to address the concerns f identified or proposed an alternate approach which addresses the thermal ,

modeling issues. (Ses Sections 2.5 of the SNL Letter Report  !

(Enclosure 1) for'further details).  !

l 2.2- SNL found that the current licensee validation studies to be t insufficient because the licensee calculations do not make direct  !

comparison to. all of _the available experimental data. SNL recosmiends  !

that the licensee document validation results for the following cases: j

- In the case of the conduit calculations, it is recommended that the licensee be asked to document validation results for the conduit l

barrier enclosure tests performed by Tennessee Valley Authority 1 (TVA for the Watts Bar plant as follows: (1) Test Article 7.4, '

three) one-inch conduits in a common, tight fitting enclosure (2  :

Test Article 7.5, six one-inch conduits in a common tight fitting)  !

enclosure,-(3) Test Article 7.8, six one-inch conduits in an oversize enclosure, (4) Test Article 7.7a, single one-inch conduit in a small boxed-enclosure, and (5) Test Article 7.7b, single one- ,

inch conduit in an oversized boxed enclosure.  ;

- For the cable tray calculations, it is recommended that the licensee be asked to document validation results which are directly  ;

comparable to available experimients from Texas Utilities Electric,  ;

TVA, and/or Florida Power Corpcration. These evaluations should #

include at least one representative case for a single tray enclosure And the TVA 3-tray stack test (TVA Test Article 7.3).  :

1 The licensee is requested to reconsider its validation of its thermal a model and to pNvide example case calculations in light of the specific SNL findings and the thermal modeling concerns identified in Item 2.1 -

above.. (See Sections 3.1.through 3.4 of the SNL Letter Report '

(Enclosure 1) for further details).

2.3 The licensee is requested to identify the industry ampacity derating values being a> plied-to P8APS and LGS plant installed configurations.

The licensee siould also explain the technical basis used to ensure that '

-any installed plant configurations which utilize industry test data are ,

representative in-terms of design and construction of the applicable '

tested fire barrier configurations.  ;

t I

k I .