ML20132A569

Evaluation of Long-Term Post-Accident Core Cooling of TMI-2

ML20132A569
Person / Time
Site:	Three Mile Island
Issue date:	04/30/1979
From:	Harold Denton Office of Nuclear Reactor Regulation
To:
Shared Package
ML20114E436	List: ML20132A569 ML20132C938
References
NUREG-0557, NUREG-557, NUDOCS 7905230274
Download: ML20132A569 (212)

Text

{{#Wiki_filter:- 'NUREG-0557 a. NUCLEAR REGULATORY COMISSION STAFF REPORT EVALUATION OF LONG TERM POST ACCIDENT CORE COOLING OF THREE MILE ISLAND UNIT 2 April 1979 Y o ( License No. DPR-73 v Docket No. 50-320 oY4 2 . _ _ _ k-, t i v 4 7fc)fs?3@7lKF ] ~

.s Available from National Technical Inforn,ation Service Springfield, Virginia 22161 Price: Printed Copy $9.25 ; Microfiche $3.00 The price of this document for requesters outside of the North American Continent can be obtained from the National Technical Information Service. l 1 1 e { i s

_ Table of Contents Pag 1. Introduction and Summary..................................... 1-1 o 1.1 Introduction.................................... 1-1 1-3 1.2 Summary.................................................. 2 2. C o re Co o l i n g...................................... 2.1 Assessment of Core Conditions............................ 2-1 2.2 Staff Analysis of Core Cooling........................... 2-2 2.3 Discussion of Staff's Thermal-Hydraulic Model............ 2-3 2.4 B&W The rmal-Hydraulic Analysi s........................... 2-9 2.5 Staff Evaluation of Core Cooling......................... 2-13 3. Long Term Natural Circulation................................. 3-1 3.1 Alternative Steam Generator Configurations............... 3-1 3.2 Staff Evaluation......................................... 3-2 4. Core Cooling by Natural Circulation........................... 4-1 4-1 4.1 B&W Analysis............................................. 4.2 Staff Analysis.......................................... 4-2 4.3 Criteria and Conclusions................................. 4-3

1 Table of Contents (Continued) P

Page, i

5. Effects of Non-condensible Gas Evolution on Natural 1 5-1 Recirculation Cooling Capability............................. i 4 6-1 6. Long Term Control of Reactor................................. i i 6-1 6.1 Reactor Coolant Level and Pressure Control.............. 6-2 6.2 Core Cooling without Primary System Convection.......... 1 6-4 6.3 Overpressure Protection / Thermal Shock................... l 6-5 6.4 Solid Conditions in Secondary System Piping............. i 4 6-6 i Steam Generator Tube Integri ty.......................... 6.5 6-7 6.6 Condenser Flooding...................................... 6-8 6.7 Containment............................................. 6-5 6.8 Criticality............................................. i 7-1 7.0 Contingency for Loss of Natural Convection................... 7-1 7.1 High Reactor Pressure................................... 7-2 7.2 Low Reactor Pressure.................................... 1 7-3 7.3 Reflux Boiling.......................................... 7-4 7.4 RHR System.............................................. 7-5 7.5 Contingencies........................................... } J e i i 1 f 1 6 e n. .r.- - + -, - - -, , -,,.-,e

Table of Contents (Cotinued) I fag 8.0 Radiological Considerations................................... 8-1 8.1 Purification Demineralizer............................... 8-1 8.2 RHR System............................................... 8-2 8.3 Steam Generator Leak..................................... 8-3 e S 9 O 4

NRC STAFF REPORT - TMI -2 Achieving and Maintaining Long Term Core Cooling 1. INTRODUCTION AND

SUMMARY

1.1 Introduction This NRC staff report addresses acceptability of the proposed method for long term core cooling of Three Mile Island, Unit 2. The licensee i has proposed" to adopt the B&W recommendation to utilize natural circulation core cooling in accordance with the " Base Case Summary," see Figure 1.1. Other alternative methods of cooling are available, including high and low pressure injection or recirculation. They da not offer the same assurance of reliability and fission product contain-ment as the proposed natural recirculation mode of long term cooling. The present method of cooling with one reactor coolant pump running is also preferred over the other alternatives, but it has the uncertainty of eventual pump and instrument degradation by the environment inside of containment. The proposed mode of long term cooling involves a sequence of events initiated in early April following initial actions to stabilize the reactor after the accident eq March 28. This sequence is designed to place both steam generator secondary cooling systems and the reactor coolant system in a water solid condition for a closed cycle cooling mode, thus keeping the highly radioactive primary coolant inside containment while preparations for plant decontamination are completed. For this preferred mode of operation, B&W has recommended that a feedwater flow of about 5000 gpm per steam generator be provided. Heat removal to the ultimate heat sink will be via intermediate heat exchangers, " Safety Analysis Report for Transition to Natural Circulation," transmitted by letter from J. G. Herbein of Metropolitan Edison Company to Denwood Ross of NRC, April 12, 1979. ,v..

• 4%

1-2 including the condenser for steam generator A in the first few months. These conditions on the secondary side of the steam generators are proposed to optimize primary side core flow, core temperature rise, and average reactor coolant temperature and to result in adequate natural circulation cooling of the core. The staff has evaluated the proposed " Base Case Summary" plan and has considered contingency plans proposed by the licensee. Our evaluation has included consideration of various subjects, summarized below in this Section and discussed at greater length in later Sections, which might affect natural circulation cooling capability. The staff approves the licensee's plans with certain conditions as described in this report and finds that there is no undue risk to public health and safety in the preferred mode of long term cooling. On the basis of current understanding of the accident scenario and available data, the staff reports here on its evaluation of the condition of the core and the core flow resistance as it might affect ability to cool the core by natural circulation. The natural circulation cooling capability of THI-2 for the estimated core flow resistance and a variety of other conditions is evaluated and a comparison of the Base Case and off-nominal plant configurations is presented. The potential for and effects of natural convection core cooling are addressed, and the staff recommendations for reactor performance acceptance criteria upon initiation of natural convection are presented. ~ Also, a discussion of the short and long term potential for evolution of noncondensible gas is includad in this report. Based on current information, it is not expected that quantities of gas large enough to affect natural circulation cooling capability will form at the expected operating conditions, as explained in this report. The staff has also addressed other potential questions regarding long term cooling in any mode such as boron precipitation, and boron dilution

1-3 and recriticality, and has reviewed the proposed contingency alterna-tives in the event of a loss of natural circulation cooling. The planned hardware modifications in the TMI-2 balance of plant for implementing the long term natural circulation cooling mode have been evaluated day be day by a team of NRC engineers working closely at the site with the Licensee and its contractors. The staff has also considered the process and diagnostic instrumentation requirements associated with operation in the natural circulation mode. summarizes the results of this review effort. 1.2 Summary The staff has performed a safety evaluation of the transition to natural circulation shutdown cooling of the TMI-2 reactor core. Our evaluation consisted of a review of the Met Ed/GPU SAR submittal of April 12, 1979, several telephone conversations with B&W technical personnel to clarify the technical content of that SAR, and extensive independent calculations by the staff, national laboratories (PNL, ORNL, INEL, Sandia) and others. We have concluded that there is a high probability that natural circulation cooling of the TMI-2 core can be accomplished using either one or both steam generators in either the steaming or water solid modes. Criteria to accomplish the transition and to evaluate the acceptability of natural circulation cooling performance using only instrumentation which is expected to remain functional for the long term have been defined. Alternative cooling modes have also been considered for the unlikely event that natural circulation cooling fails. The staff evaluation has considered core conditions ranging from a normal unblocked core with an average resistance factor (K) of 9.4 to a core highly blocked (>99%) in the central region by approximately five feet of debris consisting of fuel and zirconium oxide fragments

1-4 and with 90% blockage in the peripheral regions, which gives an equivalent average core K of 3760 (96% blockage) or 400 times normal. Our best estimate of the core resistance model after considering both flow and thermocouple data can be represented by an average core K of 1810 (93% blockage) or 200 times normal. Predicted core flows for this range of core resistances for the current one pump operating condition with one steam generator solid and for the minimum natural circulation condition with one steam generator solid follows. Average Core One Pump (lbs/sec)

• Natural Circulation Resistance Core Flow Flow (lbs/sec)

Normal 11,060 380 Best Estimate 3,610 231 (200 X Normal) Maximum 3,145t 200** (400 X Normal) The calculated natural circulation flow rates as a function of core resistance (K) are linear on a log-log plot. The minimum natural circulation flow rate of 200 lbs/sec corresponds to an average core temperature rise of 13.5F at 3 Mwt decay heat level. It should be noted that the ratio of calculated one pump flow rate to the calculated natural circulation flow rate for a high core resistance is approximately 15.7:1 and independent of the high core resistance considered. The core AT indicated by individual incore thermocouples during late April remained approximately constant after a large drop in the core inlet temperature (234F to 175F). Thus, a closed channel enthalpy ri.te model provides a reasonable basis for predicting the incore thermocouple temperatures. Assuming no change in core flow distribution

• Based on Core Inlet Temperature = 103 F and 3 Mwt Decay Heat Level
  • Extrapolated from Table 3.3 t Extrapolated from Figure 2.9 k

v

• = -

-r e n-. - ~-- 9 ,,a w.g-

1-5 when the operating pump is tripped, the enthalpy rise in each thermo-couple channel can be expected to increase by a factor of 15, cor-responding to the decrease in flow. Predicted equilibrium thermocouple temperatures as a function of decay heat rate on the day the pump trips are given in Figures 1.2 and 1.3. The staff has estimated that an incore thermocouple reading of 1000F is indicative of some core material at 1300F or higher. Due to poor strength properties of core structural materials at this high temperature and due to an increasing propensity for additional oxidation of zirconium, 1000F on the incore thermocouples is considered to be an important limit for effective core cooling. Since up to two thermocouples are expected to approach this limit in the event of early initiation of natural circulation, appropriate precautionary actions are specified in Table 4.3 for the occurrence of two or three incore temperatures above 1000F. A second important limit for acceptable natural circulation cooling is the prevention of bulk boiling in the core. For operation with two steam generators, a criterion to maintain 100F or greater subcooling at the hot leg RTDs is acceptable. For one steam generator operation, low flow or no flow condit1cns would best be indicated by incore thermocouples. The staff reco,mmends that natural circulation cooling be terminated if the average of incore thermocouple readings exceeds saturation temperature. Additional criteria are discussed in Section 4.3. The staff has also concluded that the alternative operating modes of high pressure injection or decay heat removal are less desirable and probably no more efficient than natural circulation cooling. Therefore, natural circulation should not be terminated prematurely. l v

i l-6 l Trip RCP i C (1-2 days) B (1-2 days)A m 2 1000 - Q2 0 a. v, Go solid 8 secondary E (2-3 days) ~ E (1-2 days) W N500 o l 5 A I f hGosolid %100 - - D primary l l l R.C. Temp (*F) e-140-160 220 280 l (1) Degas at A; Lower Pressure (A+A ') while degassing, then return to A. (2) Continue Design / Installation of static and active systems for primary makeup / pressure control and secondary cooling system for "B" S/G. (3) Reduce temperature (A+B) by steaming on "A" S/G. (4) Take "A" S/ so id -(i.e., all liquid, no vapor) drop primary temp. to minimum B+C (5) Trip RC Pump "A" - Establish natural convection - Establish cooling to "B" S/G if available. (6) Drop primary pressure to selected value (C-+D) (7) Take primary system solid - (i.e., all liquid, no vapor) Control pressure 6 makeup with static or new active system END POINT Primary - Natural Cire, all liquid, Long-term P/V Control Secondary - All water, Long-term He'at Dump System Figure 1.1 m

1-7 Core Exit Coolant Temperature vs. Time during Natural Circulation with one Steam Generator Steaming and one Steam Generator Isolated H-5 hottest thermocouple 2400 <- H-8 next hottest thermocouple H-9 representative of the next hottest group 2200 ' ' L-11 representative of peripheral region 2000.. C 1800 - - Pressure = 1000 PSIA v w Flow = 150 lb/see h 1600, Tin = 175'F I 1400.. m Eh fde 1200.. g H-5 5 1000 - * ~ = 8 800

• H-8 600 - -

400 - - - H-9 L-ll 200 < T-inlet 0 ~ 26 28 30 2 4 6 8 10 l i APRIL MAY i Figure 1.2

1-8 CORE EXIT COOLANT TEMPERATURE VS. TIME during Natural Circulation with one Steam Generator in a Water Solid Condition H-5 hottest T/C H-8 next hottest T/C H-9 representative of next hottest group ~ L-ll representative of peripherial region 1000 - - ,E Pressure = 1000 PSIA Flow = 231 lb/sec y Tin = 175'F g 800, A W H-5 s 11 - s 600 - - 3 -Tsar 8 (1000 PSIA) U 400 d n O H-9 200 - L-11 Tin O 26 28 30 2 4 6 8 10 APRIL MAY Figure 1.3

l 2.0 CORE COOLING 2.1 Assessment of Core Conditions i i The staff assessment of the present TMI-2 core geometry is presented in Attachment 1. It is based upon the preliminary data available from the plant as of April 13, 1979. An understanding of core damage is an important factor in assessing the adequacy of core cooling in the i proposed natural circulation mode, as explained in subsequent sections. I The assessment can be summarized as follows: The cladding for many or all fuel elements may have ballooned and ruptured early in the accident. This mode of initial defecting is probably irrelevant in light of later, more extensive damage by oxidation and embrittled fracture of many fuel elements. In the hot upper central region of the core, fuel temperatures probably exceeded 1750 C releasing large quantities of fission products; radio-chemical analyses indicate that about 30% of the total core inventory of noble gases was released into the primary coolant system. The 1750 C temperature is less than the U02 melting point. Temperatures sufficient to cause fuel pellet melting (2800 C) were probably not reached. It is estimated that about 40% of the Zircaloy cladding reacted with water. The region of most severe oxidation probably was localized above the 2 to 6 ft elevation, with more severe oxidation in the central bundles than in the peripheral bundles. Significant melting of partly oxidized cladding may also have occurred, and this material would have solidified when core cooling was re-established. The I severely oxidized (and perhaps fused) cladding probably fragmented upon quenching. The resulting fuel element debris is probably composed of pieces ranging from millimeter size to whole sections of fuel rods. O

. -. - ~ - 2-2 The temperature of unfueled components would lag the temperature of fuel rods by only about 20'F so that they also should have experienced j temperatures above about 1700*C. Consequently, Zircaloy components in the hot region of the core should have oxidized, and components with i Inconel, stainless steel, and Ag-In-Cd should have melted. Because of i many layers of protective sheathing in the instrument tube the incore thermocouple tubes have survived even in the damaged core region, although the outer sheath of the instrument tube may be badly damaged. 7 Nearly all of the broken and oxidized fuel element debris should j, remain trapped in the upper core region because the fuel assembly end fittings at the top of the core have a grillage that would act as a I screen. Furthermore, the compaction of fuel debris is limited because the fuel pellets are fabricated with a packing fraction of about 46% and the theoretical maximum packing fraction (for a bed of spherical particles) is only about 63L It is very likely that fuel debris is also trapped in some mixing cups surrounding the incore thermocouples, contributing to non-uniform thermocouple readings, i

2. 2 Staff Analysis of Core Cooling 1

) In order to evaluate current core cooling and the planned transition ) to natural circulation cooling, calculations were performed by the NRC l staff, and results of calculations performed at Pacific Northwest Laboratory (PNL), Idaho National Engineering Laboratory (INEL), and Oak Ridge National Laboratory (ORNL) were discussed with the staff. The calculations can be summarized as follows: (1) The staff performed analyses with the computer code Cobra IV to develop a thermal-hydraulic model of the core based on the core exit coolant temperature measurements obtained from the post-accident In-core thermocouples with one reactor coolant pump running. Pacific Northwest Laboratory performed similar calculations using Cobra IV and ORNL attempted similar calculations using the I i - - - - ~. ~~ --.---,.-c.-. ,,-e-,,,,,, --~,-,w

2-3 SABRE code. The latter calculations are incomplete; however, preliminary results are consistent with PNL and NRC calculations. (2) Hand calculations have been performed to further develop the core model based on the knowledge obtained from temperature and flow measurements, thermal-hydraulic design data, and COBRA IV calculations. Considerations in the modeling included: (a) an agglomerate of fuel particles packed in the mixing cup surrounding the in-core instrument thimble at the axial core exit location of the thermocouple. (b) suspension of fuel particles in a partially fluidized bed with the present one pump flow conditions, (c) accumulation of fuel debris in the bottom of the reactor vessel. (3) Analyses are in progress to estimate the kinetics of core damage by modeling the events that occurred in the course of the accident. This effort includes hand calculations by the staff to obtain preliminary estimates based on the transient data and more detailed computer calculations using IRT at Brookhaven National Laboratory (BNL) and RELAP4 at INEL. (4) Calculations were performed at INEL using RELAP4 to evaluate natural circulation flow with several possible medes of steam generator operation during long term cooling. Natural circulation test data from Oconee were studied to confirm the calculated flow rates. (5) Calculations of the effect of core flow blockage on water flow in the active and inactive loop have been performed by INEL (using RELAP), Technology for Energy Corporation (TEC) and by the staff.

2-4 These calculations have been compared with post accident and startup test measurements of flow in the TMI-2 reactor coolant system in order to estimate the extent of core flow blockage. (6) The time history of post-accident thermocouple data from TMI-2 has been examined using a simulation of the damaged core to evaluate the capability of the model to predict cooling conditions. The behavior of thermocouple data during the transfer from reactor coolant pump 1A to pump 2A on April 7 has been interpreted by the staff to generally confirm the estimated condition of loose fuel-i particles suspended in the upper regions of the core. As shown in Figure 2.1, high thermocouple readings existed in Region A prior to tripping pump 1A. After start-up of pump 2A, an improved flow condition around the core exit thermocouples of Region A was indicated by decreases of greater than 100F in exit temperature from several of the instrumented assemblies. At the same time, core exit temperatures rose on the order of 50F in several sym-metrically located fuel assemblies in Region B of Figure 2.1. This cannot be totally explained if the fuel in the upper region of the core were stationary since the switch in inlet flow distri-bution would account for a maximum of about 30 percent increase or decrease in core AT. However, it can be partially explained by collapse of the fluidized bed during the flow coast down transient and by a redistribution of fuel particles axially or radially for the new flow distribution obtained with Pump 2A. The trends of the thermocouple data both prior and subsequent to the pump trip also are consistent with this conceptual core model. (7) Calculations were performed to determine if the few high thermo-couple readings that have existed since stable core cooling was reestablished could be caused by debris in the mixing cup sur-rounding the thermocouple. For calculational purposes, a spherical mass was assumed to form around the end of the thermocouple. The )

2-5 calculations show that even for the maximum diameter sphere which fits into the mixing cup, the temperature difference from inside to outside the sphere is only about 60*F. This is not enough to explain the high thermocouple readings that have been observed during the first two weeks following the accident, assuming heat transfer to subcooled fluid surrounding the mixing cup. Assumptions used in the calculation are given in Table 2.2. 2.3 Discussion of Staff's Thermal Hydraulic Model The staff's thermal-hydraulic model consists of a one-eighth core, coarse mesh COBRA IV model with one radial mesh point per assembly and 13 axial nodes. Flow area reductions were included in the assemblies which showed the largest temperature difference across the core. The amount of area reduction was chosen so that the calculated core exit coolant temperature matched the thermocouple readings taken at 5 PM on i 3/30/79. The measured thermocouple readings were corrected for the bias which was observed at the plant (see Section 2.4, below). When the thermocouple readings are corrected for such bias, the core tempera-ture rise in the peripheral assemblies appears to be approximately consistent with a relatively unblocked core. The measurement uncertainties I for the thermocouples and the resistence temperature detectors in the coolant loops (RTDs) are so large that the temperature differences between a normal core and a slightly blocked core would be masked. The staff model takes account of increased resistance to cross-flow by greatly reducing the cross-flow areas. The extent of flow area blockage in the staff model is shown in Figure 2.2. This model has been used to calculate the core exit coolant temperatures for the period 3/30/79 1 to 4/10/79 and has also been used to calculate the coolant conditions under several possible modes of operation for natural circulation. Figures 2.3 and 2.4 present the calculated and measured core exit coolant temperatures as a function of time. ~ t

2-6 The staff's consultants at Pacific Northwest Laboratory also performed thermal-hydraulic calculations for the damaged TMI-2 core. The PNL thermal hydraulic model consists of a coarse mesh, full core model with 21 radial regions each representing eight or nine fuel assemblies. The PNL radial modeling is shown in Figure 2.5. The model includes increased flow resistance and reduced flow area in the hot regions at high elevations, as shown in Figure 2.6. As was the case with the staff calculations, it was necessary for PNL to severely restrict cross flow in order to match the measured core exit coolant tempera-tures. The area changes and flow resistances shown in Figure 2.6 were chosen to match the thermocouple readings from 4/7/79. Both the staff model and the PNL model indicate that extensive blockage to both axial and radial flow is required to explain the measured temperatures. In order to infer a core condition from the area reductions and resistance increases used in the models, we have used the flow and pressure drop characteristics of various possible flow blockage configura-tions. For example, calculations were performed to estimate what type of particles might be suspended in the core due to upflow from one reactor coolant pump and to estimate the depth of particulate matter required to cause the inferred flow blockage (i.e., blockage inferred from readings of thermocouples above the core). ~ The average velocity of fluid in the core is estimated to be about 3.5 ft/sec with one pump running and with a normal core geometry. As can be seen from Figure 2.7, this velocity is high enough to suspend particles of 0.15 inch, or less, in equivalent diameter. Fuel pellet cracking, as is known to occur during normal operation, would produce U0 fragments in a size range that could be be suspended by a 3.5 2 ^ ft/sec upflow. As fabricated, the pellets are about 0.37 inches in diameter and approximately 0.5 inches long.

2-7 The pressure drop required to fluidize a particulate bed of U02 ranges from 2.5 psi /ft for 40% porosity to 2 psi /ft for 50% porosity. There-fore, if the frictional core pressure drop is between one and twelve psid, a fluidized bed between one-half foot and six foot thick could ~ be suspended. The hydraulic resistance of a fluidized bed depends on the porosity of the bed and the shape of the particles. Therefore a range of shapes and porosities were considered in calculating the resistance coefficient for the fluidized bed. The porosity was considered to vary from 40% to 50% and shapes including cylindrical sectors, wedges and parallel-opipeds were considered. If the pressure drop per foot of bed (psi /ft) is expressed as 2 AP _ K W C J2 288 gc6 Where W = mass flow, lb/sec 2 A = area occupied by bed of particles, ft 8 6 = fluid density, 1b/ft 2 g = gravitional constant, ft/sec c K = resistance coefficient, 1/ft i L = bed length, ft j 'then the most likely range for the resistance coefficient, K is ~ 37,000 < K < 138,000 for each 12 inches of debris. In summary, we found that a flow resistance (K-factor) of 37,000 to i 138,000 could be produced by each 12 inches of debris, where the debris' consists of fuel and clad fragments ranging down to 1/8 of an l inch in diameter. The flow resistances used by PNL (600,000 and 100,000) would be equivalent to at least 4.3 feet of debris'for the l hottest region and between 2.7 feet and 0.7 feet of debris in the lower temperature regions. The peripheral region is assumed to be k i I l v -,-,-ww, --,._r

---

,-,~v-----,-y-4-r,-

2-8 free of a significant amount of debris in the PNL model. The staff model was used to study peripheral regions with two possible configura-tions, unblocked and 90% blocked. Using a blockage of 90% in the peripheral region, the total core flow resistance as inferred from the thermocouple readings is 3760 or 406 times normal. In order to calculate the core exit temperature readings following the change to pump 2A on April 7, it was necessary to change the flow blockage model. Some or all of the observed temperature changes may ~ have been associated with axial motion of debris; that is, material that was near the thermocouples (e.g., just under the fuel bundle upper end fitting) might have settled to a lower elevation. If this had been the case, then the temperature of the material might not have been affected when pump 1A tripped; rai.her, the temperature at the thermocouple location might have been lowered as a result of cross flow from cooler regions. For this reason both the early blockage model (with pump 1A rur.ning) and the later blockage model (with pump 2A running) have been used in extrapolating to natural circulation conditions, as described in Section 4, below. Figure 2.8 shows a comparison between the temperature behavior predicted by the later blockaga model and the measured thermocouple readings since the startup of pump 2A. The figure clearly indicates periods of time during which the readings are changing more rapidly than would be expected from the changes in decay heat with time. This could be the result of continuing movement of debris in the region of the hot assembly (H-8). However, there appears to be a smooth and steedy decrease in the readings from 1600 hours on 4/9/79 until 0800 hours on 4/11/79. This redut. tion parallels the calculated results and suggests at least a temporary end to the movement of material near the exit of assembly H-8. This trend has continued up to the present time.

l2-9 2.4 B&W Thermal - Hydraulic Analysis

• A conservative evaluation of core blockage was made by B&W, based on the use of the core exit thermocouples.

Based on the measured temperature data, the core flow in the current mode with one reactor coolant pump ~ cperating was estimated to be less than 1 x 108 lbs/hr. This represents very nearly total blockage. The comparable calculations by the staff indicate about 2 x 106 lbs/hr. In order to better understand the apparently high core exit thermocouple readirgs, B&W performed analyses of the thermocouple data. The chromel-alumel thermocouples are located in instrument thimbles which extend through the core. The thermocouple junction is within a mixing cup in the fuel assembly upper end fitting, approximately 9 inches above the active core. The analysis postulated the presence of fuel accumu-lation in the upper end fitting and the mixing cup. Fuel debris with a radius of 3 to 4 inches is needed co produce the highest observed temperature readings in the early days following the accident when core temperatures had stabilized (AT = 300F from thermocouple to subcooled coolant external to the instrument thimble). This amount of fuel could fit within the upper end fitting, which has an interior width of seven inches. Debris within the smaller diameter mixing cup could result only in a small (10F) increase in temperature. Tests were performed by B&W to determine the effect on thermocouple accuracy of exposing the instrument to 2000F for four hours. Four test thermocouples read witnin 5% of their calibration values over a range of 200F to 1000F after this exposure. Thermocouple data from TMI-2 during normal operation were examined for evidence of a systematic bias. The possibility of de-calibration since the accident was also evaluated. It was concluded from the results of the thermocouple tests and evaluations that the temperature resdings are generally accurate, but have a possible +5F bias. Further upward bias of the thermocouple readings is believed by B&W to be due to fuel debris

• See Licensee's SAR of 4/12/79, Section 3.7

i ,2-10 4 packed around the thermocouple. They cite the shift in thermocouple i readings during the switch from pump 1A to pump 2A to further support ) the debris theory. ~ l From these studies, B&W concluded that the existance of flow blockage to reduce the core flow to 1 x 108 lbs/hr, as can be predicted from I the thermocouple data, was improbable. i j Rather than a blockage model based on the thermocouple data, B&W )_ advances another model of core thermal hydraulics. This model applies the B&W PUMP code to the TMI-2 plant for.different reactor coolant l pump configurations and corrects for the pre-accident flow split j measurements. Core resistance was then increased until the calculation i predicted the hot leg, post-accident flows as measured by the two flow l meters in the present one pump operating condition. This calculation i results in a prediction of core flow of 13 x 108 lb/hr and a core I pressure drop of 18 psi for the one pump operating condition. These j conditions are calculated with an average core flow resistance approxi-mately 200 times the normal resistance, or a form loss coefficient of-1100 for a single-node core representation or 1650 for separate modeling of core and bypass. This model thus yields a more optimistic view of l core flow than do the models based on the thermocouple data. B&W considered.the current estimate of material available for core blockage to determine if their best estimate core resistance was l feasible. They concluded that 167 cubic feet of debris spread evenly i-j, across the core in a three-foot-thick packed bed would produce a form loss coefficient of 1700, which is in good agreement with the total core resistance calculated using the measured flow splits. Therefore, ) this is considered to be their best estimate of current core conditions. i. The Licensee's submittal of the B&W analysis does not address the validity of the post-accideni;, one pump flow data tubes. These tubes i-i 1 i d *

r g.

2-11 are located in the hot legs; typical post-accident readings for the one pump operation are 24% forward flow in the active loop and 17% reverse flow in the inactive loop. The staff has been informed in telephone conversations with B&W engineers that the one pump flow measurements, including reverse flow, are accurate to within one The validity of this information is of primary importance in percent. evaluating the actual core resistance that now exists. B&W has calculated an unrecoverable core pressure drop of 18 psid for the present one pump operation. The pressure drop was determined by adjusting the core resistance in the hybrid computer code PUMP until the calculated flow in the hot leg of the loop A matched the measured flow in the "A" hot leg of TMI-2. The calculation was confirmed by comparing the reverse flow calculated for loop B with the measured reverse flow for loop B. B&W claims that measured and calculated reverse flow agreed very well. Further confirmation of these relatively large loop flows can be obtained by comparing the post-accident power input to the pump with its normal power input. The pump is drawing approximately 10,000 HP or 7 46 Mw with approximately 6.3 Mw of pump heat input to the water. Normal pump heat is 4.5 - 5.0 Mw. Because B&W's calculated core coolant velocity is low (1.26 ft/sec on the average) most of the core pressure drop must be due to form loss. B&W calculated a dimensionless form loss coefficient, K, of 1100 for the combined core flow and bypass flow, as follows: 2 AP = K 288 6 gc where AP = core pressure drop in psi V = core coolant velocity in ft/sec 3 6 = coolant density in ibm /ft 2 g = gravitational constant in ft/sec c

=.. 2-12 The measured core pressure drop for four pump operation under normal l conditions is 14.5 psi. Initially, B&W assumed that 50% of the pressure drop was due to friction and 50% due to form loss. They thus calculated o a normal form loss coefficient of 5.47 and noted that the form loss f coefficient for the damaged core was approximately 200 times the normal form loss coefficient. Other form loss coefficients reported 'I by B&W (1100 and 1650-1700) correspond to the calculated 18 psid pressure drop across the damaged core using different models in the i pressure drop calculation. The K=1100 was for a model which lumped the core and bypass regions. The K = 1650 -1700 was for a model whch considered separate core and bypass flow. B&W provided a parametric study of natural circulation for form loss coefficients ranging from 18 times normal with 5% bypass to 1000 times normal with 5% bypass. The best estimate case of 200 times normal with 30% bypass was also included. They conclude that even with the most pessimistic form loss coefficient of 1000 times normal, the core flow during natural circulation is adequate co cool the core. The staff has evaluated the core and loop flow analysis by B&W. The measured flow rates in the active and inactive hot legs have been studied in an attempt to infer core blockage on the basis of changes in these values from those expected in an unblocked core. Calcula-tions of loop flow rates with various amounts of core blockage have been performed by the staff, by INEL and by TEC. Figure 2.9 shows the calculated changes in loop flow with increasing core blockage. As discussed above, the measured flow rates in the active and inactive loops are 24% forward flow and 17% reverse flow, respectively. Table 2.1 presents the measured inactive loop flows on each of three reactor protection system channels for pump 1A running and for pump 2A running. i. These measured inactive loop reverse flow rates of Obout 17% were used to infer an amount of core blockage by using the infonation on Figure 2.9. The core flow blockage corresponding to 17% reverse flow in the I-1

+r}~nz.

2-13 inactive loop is approximately 96%. This is in reasonably good agree-ment with the results of B&W calculations which indicated that a core resistance of 200 times normal is required to produce this value of o inactive loop flow; i.e., a core resistance of 200 times normal cor-responds to approximately 93% core blockage. Based on measured core pressure drop of 14.5 psid and a calculated frictional pressure drop of 3.8 psid for normal four pump operation, the staff has calculated a normal-core form loss coefficient of 9.4. Assuming that the reported TMI-2 core pressure drop of 18 psid is 6 correct and that the core flow is 13 x 10 lb/hr as estimated by B&W, the resistance coefficient is 1810 or approximately 200 times normal. The most optimistic estimate of core conditions is based on the assumption that the peripheral region of the core is unblocked. This assumption gives a core frictional pressure drop of 1.5 psid and a core average 6 flow of 21 x 10 lb/hr; this corresponds to a form loss coefficient of 57.5 or 6.1 times normal. The value of core resistance which is inferred from the thermocouple readings is 3760 or 400 times normal. The rar.ge of core resistance factors for the cases discussed above is given in Table 2.3. This complete range of core resistances was considered in the natural circulation calculations reviewed in Section 3, below. 2.5 Staff Evaluation of Ccre Cooling On the basis of the available measurements, our own calculations, our consultants calculations, and B&W's calculations, we have developed the following conclusion. It is reasonably certain that the central region of the core is severely damaged and almost entirely blocked to normal coolant flow. This blockage is probably due to a 3 ft to 6 ft thick layer of fuel and structural debris. The lower elevations in this central region of the core might be relatively undamaged and unblocked. The condition of the core peripheral region is more difficult to evaluate. The core exit thermocouples in the peripheral region n

2-14 indicate temperatures approximately 10 F higher than the indicated cold leg temperature. However, these readings are thought to be slightly biased as described in Section 2.4. Therefore the AT measure-ments in the perip.1eral region are not sufficiently accurate to distinguish between unblocked and partially blocked (i.e., up to approximately 90% blockage). The hot leg venturi flow meter readings have been used by B&W and by the staff to infer a total core blockage of approximately 95%. This amount of blockage is consistent with the core exit thermo-couple readings if the center region is more than 99% blocked and the peripheral region is approximately 90% blocked. This appears to be an intuitively reasonable core description and is consistent with or conservative relative to other core descriptions that have been developed (Attachment 1). We conclude that the core resistance is in the range of 200 times normal to 400 times normal. Since the loop flow measurements are expected to provide better information on the average core resistance, we conclude that the best estimate of core resistance is close to the 200 times normal salue as calculated based on the measured loop flows. The value of core resistance inferred from the thermocouple readings - (i.e., 400 times normal) has also been considered in the bounding calculations of 1ccal core temperatures which might occur during natural circulaticn. 9 e c-

2-15 TABLE 2.1 INACTIVE B LOOP FLOW RATE MEASUREMENTS April 1,1979 reactor coolant pump 1A running Reactor Protective System Channel Flow Rate A -6850 lb/sec B C -6889 lb/sec D -6794 lb/sec Average Value -6844 lb/sec (16.9% of full) April 9, 1979 reactor coolant pump 2A running Reactor Protection System Channel Flow Rate A -6870 lb/sec B C -6844 lb/sec D -6541 lb/sec Average Value -6752 lb/sec (16.7% of full) e d I

= 2-16 } TABLE 2.2 Assumptions Used in The Calculation ~ j of The Spherica1> Mass l-(1) One dimensional conduction 1 (2) Decay heat = 0.5% of 2772 Mw 4 l (3) Thermal conductivity of UO2 = 2 Stu/hr-ft *F ) ~ (4) The ratio of voltmetric heat generation to the thermal conductivity does not vary with porosity. 4 ]' i 1 2 I t 1 h 1 { 3 i a e i 4 i 4 I t t 2 1 i i

• U jy,-

$[0 bh.~6, -

t TABLE 2.3 f STAFF CALCULATED CORE RESISTANCE FACTORS V i W AP (AP) (288)a_ W .s K_ CORE, Ib/Hrx10 ft/sec Condition Total Form 6 x _s y lb/Hrx10 (PSID) (core area = 2 49.17 ft ) t 4 Pump 130.I 10.7 9.4 (normal) 120.2 15.47 (Design) 7 U 1 Pump (Normal) 62.02 9.4 (normal) 39.8 7.38 u 1 Pump 39.7 18 1810 (200 X normal) 13.0 1.26 B&W Calculated Core AP 1 Pump 40% 1.5 57.5 (6.1 normal) 21 2.04 Core Blockage 1 Pump, 95% Blockage from 3760 (400 X normal) TCs E 4

i 2-18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 A 292 292 B 287 288 M 294 C 289 303 289 997 '4%1 297 D [ 295j 293 E 300 $8 @320 E0'5] 47 999 287 291; 40/ 214 ? 285' '3I23Ml 291; F 301 265 207 l A 2,93

281, 294 255 oc 6

~ $[3 292 ['jff 1 68 3[ "J ][341 G 293 '372 324

J23_

288 H 285; 289 ,,393i t__. _, m m__ a Ju" b JJo' K M l317; M 297 30M 3T3 i3ddj 295 t 318 B L2_81 288 316 290 M M i 316 295 2 290 ,,,.,j 302 Zo9 N M N 292 300 294 M 305 2 917 g 0 294 296 290 7g7 I$N B p ' 29d r. { R 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 XXX 4/06/79 @ 1000 hrs w/ PUMP 1A (Tin = 285F) XXX 4/07/79 @ 1205 hrs w/ PUMP 2A (Tin = 285F) XXX 4/11/79 @ 0800 hrs w/ PUMP 2A (Tin = 280F) sgssw\\ A indicates blocked region w/ Pump 1A B indicates blocked regions w/ Pump 2A f,fffff Figure 2.1 TMI-2 Incore Thurmocouple Indications of Flow Blockage

M 2-19 TMI-2 Flow Blockage Model Based on T/C Readings from 3/30/79 N 8 9 10 11 12 13 14 15 g 90% 964 994 994 90% 95% 90% 90% .87 1.19 1.30 1.31 1.22 1.22 1.32 .94 N 99% 99% 99% vo4 98% 90% 90% K 1.27 1.30 1.29 1.20 1.22 .97 .79 s 99% 98% 99% 90% 90% '90% ~ L 1.30 1.25 1.03 .96 .62 .49 s 99% 9u4 90% 90% M 1.15 1.08 .93 .72 s 90% 90% 90% N .99 .85 .55 N 90% 0 .59 N P R X.XX RADIAL RPD 76 Assemblies Blocked 101 Assemblies Unblocked CONSUMERS POWER COMPANY MIDLAND PLANT UNITS 1 & 2 4 EFPD 100-30-100 Design Transient Power Distribution at 136 EFPH, 100% FP Figure 2.2

7a V I ar b @8 o C 1 1 9 1 1 E K E 0 C C C 2 /// ) T T T er oc fo 8 0 1 2 ts e t to 6 I %l 1 h fo egar l 4 e 1 va ( no ig 2 e 1 ~ R to ) H sy 3 a a D 2 t 0 ( l a 1 p D e r m m i I s T i V v v A l R e Ae B d 8 O o C M I M T

6

~ g eA eA A 4 G

2 b

0 0 0 0 0 0 0 0 6 5 4 3 E.v 832e&0 yy= l],

h YM V 3I 4 a 0 - r 1 6 b o M G C A I C C S '0 / / P 2 T T 0 F0

• 0 01 82 =

" 6 = s 1 s ne i r TP 6 " 1 4 ^ f n 2 o " 1 ig e R e ) g s a 0 y a r 1 a t D a ( D 4 e 2 m s i e V v T r I u l 8 g A e i R d B o F O M C IM T '6 "'4 OA h A O 0 0 0 0 0 0 0 0 0 6 5 4 3 g*v8B0$3"n 3 l

i 2-22 A B 1 2 3 C t D E 4 5 6 7 8 ~ ~ 3 '. '. " 463 's ao r F G 9 10 11 12 13 H ^ ava aca Jyv 1-Jz4' r I am ummme um mumm mammmmummm mummammmeammme e summer J M 14 15 16 17 18 302*1' 312' F N / \\ P 19 20 21 Q 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Region I Channels 8, 10, 16, 17 ~ Region II Channels 6, 11 Region III Channels 7, 12 Unblocked Area Channels 1, 2, 3, 4, 5, 9, 14, 15, 18, 19, 20, 21 1 1 PNL COBRA IV Model for TMI Figure 2.5 l L

2-23 Flow Resistance and Flow Areas for PNL COBRA IV Model Region I Region II Region III e _ End Fitting K = 500,000 K = 50,000 K = 50,000 Grid 8 Area =.43 x Area =.43 x Area =.43 x Nominal Nominal Nominal K = 50,000 K=25,000 K= 25,000 Grid 7 Area =.43 x Area =.43 x Area = Nomina: Nominal Nominal K = 50,000 K = 25,000 K = 25,000 Crid 6 Area =.43 x Area = Nomina. Area = Nomina: Nominal Grid 5 Grid 4 Grid 3 Grid 2 Grid 1 End Fitting Region I Gap Size = 10~ in. Region II Gap Size =.03 in Region III Gap Size =.03 in. Figure 2.6

2-24 Figure 2.9 0 Fluid Velocity Required to Suspend Spherical UO Particles 2 in 280*F Water at 1000 PSIA 6.. 5- 'd %)

4..

8 0 tu 3-g 1 E un 3 2_ _ D 2 O O a O b.1 0'. 2 0 '. 3 0 '. 4 0 '. 5 Particle Diameter, In. Figure 2.7

o o Core Exit Coolant Temperatures vs Time t @ data from thermocouple H-8 500 - .C E s %u E Y e w 450 g 3 O O O .~ OO MI Q@ OO E Model o e 400 - ~ 4 ~ ~ 3 W W 10 11 12 13

1. days 4/8 4/9 4/10 4/11 date Time Fismre 2.8

2-26 Active and Inactive Loop Flow as a Function of Core Blockage 30 g ~s \\. 's %El N \\ Q 25( \\ Active Loop \\ s ? 8o I 20 ,m $j Measured g Inactive Loop Flow l [ 15 - - C w o i 10. / ! B / l 5 / B C I / l / S 5 2 InactivMoop(ReverseF[ow) o N / A -$ s l [3-I 0 8 s i i i 0 20 40 60 80 100 Blockage % O Technology for Energy Corp. Calculations d Staff COBRA IV Calculations -- Q- - INEL RELAP Calculations Figure 2,9

1 3.0 Long Term Natural Circulation Natural circulation flow rates in the reactor coolant system for post-accident e TMI-2 conditions have been calculated by B&W and by the staff's consultants at INEL for four different plant configurations. The following section addresses these different modes of natural circulation. The INEL calculations were performed with RELAP. The B&W calculations are reported in the Licensee's SAR of April 12, 1979. Each of these models calculates natural circulation flow by modeling the system volumes and elevations and by balancing the frictional and elevation pressure drops throughout the system. The RELAP model used for these calculations is the same model used by INEL for Loss of Coolant Accident analysis. 3.1 Alternative Steam Generator Configurations a) One Steam Generator Steaming and One Isolated Both B&W and INEL calculations indicate that this is the least effective mode of natural circulation. One B&W calculation has been reported for this condition; four INEL calculations were performed. Table 3.1 presents the results of these calculations; they are reasonably consistent. The lower flow in the B&W calculation appears to be the result of the higher value of core resistance that was used. b) Both Steam Generators Steaming Table 3.2 presents the results of the B&W and INEL calculations for this plant configuration. As with the previous case, the results of the B&W m and INEL calculations are reasonably consistent. j c) One Steam Generator Water-Solid and One Isolated 4 This configuration is an expected long term mode of operation while modifications are being made to the secondary system. The results of the 4 i ,,.r--

i 3-2 B&W and INEL calculations for this configuration are given in Table 3.3; they are reasonably consistent. The importance of the core flow resistance can be seen by noting the sensitivity of the results to changes in this parameter. d) Both Steam Generators Water-Sclid. ~ This is the configuration proposed for the long term cooling mode. Table 3.4 presents the results of B&W and INEL calculations for this mode of operation. The B&W results show approximately half as much flow as the INEL results. This difference is primarily due to the differences in core flow resistances used in the calculations. 3.2 Staff Evaluation On the basis of comparisons with the natural circulation test data taken f at Oconee (a plant similar in design to TMI-2) we conclude that either the B&W or the INEL codes provide reasonable estimates of the reactor coolant system flow rates during natural circulation as a function of core flow resistance. In addition, we conclude that the mode of operation proposed in the Licensee's SAR of April 12, 1979, which calls for the use of both steam generators in the water-solid closed-cycle cooling mode, is the most favorable mode of long term operation in that it provides larger core flow than the others studied. Based on the calculated flow rates we conclude that any of the above modes of natural circulation operation will provide sufficient flow to maintain the average temperature at the core exit below saturated conditions. The expected local coolant tempera-tures with natural circulation using one or two steam generators in the water-solid mode is addressed in the next section of this report. a

3-3 Table 3.1 Natural Circulation with One Steam Generator Steaming and One Isolated Source S.G. Level Core Resistance Power Flow Rate in Core B&W 30 ft. 60 x Nominal 5 MWt 145 lb/sec to 306 lb/sec INEL 30 ft. Nominal 5 MWt 430 lb/sec INEL 30 ft. Nominal 3 MWt 350 lb/sec INEL 30 ft. 35 x Nominal 5 MWt 320 lb/sec INEL 3 ft. Nominal 5 MWt 350 lb/sec INEL 12 ft. Nominal 5 MWt 380 lb/sec INEL 30 ft. 400 x Nominal 5 MWt 98 lb/sec INEL 30 ft. 10,000 x Nominal 5 MWt 12 lb/sec e e

, 4 Table 3.2 Natural Circulation with Both Steam Generators Steaming Source S.G. Level Core Resistance Power Core Flow Rate B&W 30 ft. Nominal 5 MWt 417 lb/sec to 638 lb/sec B&W 30 ft. 10 x Nominal 5 MWt 306 lb/sec to 444 lb/sec B&W 30 ft. 60 x Nominal 5 MWt 222 lb/sec to 333 lb/sec INEL 30 ft. Nominal 5 MWt 500 lb/sec f e N

, -QW ~ .~ 3-5 t l Table 3.3 I Natural Circulation Flew with One l Steam Generator Solid and One Isolated Source S.G. Level Core Resistance Power Core Flow Rate B&W Solid 60 x Nominal 3 MWt 194 lb/sec to 306 lb/sec B&W Solid 60 x Nominal 2 MWt 167 lb/sec to 278 lb/sec B&W Solid Nominal 3 MWt 380 lb/sec B&W Solid 200 x Nominal 3 MWt 231 lb/sec B&W Solid 1000 x Nominal 3 MWt 142 lb/sec INEL Solid Nominal 5 MWt 420 lb/sec I e ch i,

3-6 s i 1 f Table 3.4 Natural Circulation Flow With Both Steam Generators Solid i Source S.G. Level Core Resistance Power Core Flow Rate l B&W Solid 60 x Nominal 5 MWt 222 lb/sec to 333 lb/sec INEL Solid Nominal 5 MWt 568 lb/sec INEL Solid 3 x Nominal 5 MWt

• 560 lb/sec l

h-I 4 e i I .i

• I i

4 E b

f 4.0 Core Cooling by Natural Circulation 4.1 B&W Analysis Operation of the THI-2 core with natural circulation cooling has been evaluated by B&W and judged acceptable, as indicated in the Licensee's SAR of April 12, 1979. Minimum core flow requirements have been established based on a margin of three times the flow rate required to prevent bulk coolant temperature from exceeding the saturation temperature. Predictions of natural circulation core flow for assumed flow resistances of 60 to 1000 times the resistance of a normal 177 fuel assembly core show adequate flow to meet the required flow rate for the operating mode in which one steam generator is water-solid on the secondary side (see Table 3.3). The effects of local boiling due to flow blockage and high temperature at fuel debris locations have been considered by B&W. It is expected that any increased local flow resistance due to boiling will be compensated by increased local convection due to the density change of vaporization. If fuel debris is located near the incore thermocouple junctions, thermo-couple readings will indicate the localized heating effects rather than the bulk fluid temperature. Therefore, B&W and the Licensee recommend that it not be required that all incore thermocouple readings be maintained below the saturation temperature in the natural convection mode. Transient natural circulation analyses were performed by B&W at core decay heat levels of 2 and 3 megawatts to evaluate the change in the core flow rate and the change in temperatures following the loss of forced circulation flow. The core flow rate calculated by B&W drops to a minimum at about one minute into the transient, then rises and stabilizes between 10 and 20 minutes. The core outlet temperature reaches a maximum value at 4 to 5 minutes into the transient and a lower stable condition at about ten minutes at a value that will depend upon the initial primary coolant temperature. Equilibrium values for flow and temperature should be ~ reached within the first one-half hour of the transient.

4-2 The B&W analysis has also considered the time available to establish and I evaluate natural circulation flow conditions during the transition from one pump operation. For the preliminary values of reactor temperature and pressure at the time of transition, more than one hour would be required to heat water in the core region to the saturation temperature even if no core flow could be established. An additional hour would be required (at 3 MWt) to boil out enough coolant to drop the level from the outlet nozzle level to the top of the core. If the core were to become instantaneously uncovered, more than one hour would be required (at 3 MWt) for the adiabatic heat-up of the core average temperature from 200 F to 2000 F. These estimates indicate that reasonable time is avail-able for decisions regarding alternative modes of core cooling (see Section 7, below). Criteria for the core exit thermocouples and the hot leg RTD's have been proposed by B&W and the Licensee for monitoring the transition to and maintenance of adequate natural circulation cooling. These criteria are presented in the Licensee's SAR of April 12, 1979 (Figure 6 of Attachment 1 and Reference 2 of Attachment 4). The staff believes these criteria may be too conservative in the sense of causing natural circulation to be discontinued prematurely. The staff's recommended criteria are provided in Section 4.3 below. 4.2 Staff Analysis j t The staff's model of core thermal hydraulics described in Section 2 was used to calculate coolant temperatures during natural circulation. A range of natural circulation flow rates was derived from Section 3 of I this report. I Coolant temperatures were calculated for three assumed plant configurations, corresponding to a flow rate of 231 lb/sec for the case of natural circula-tion with one steam generator-in a water solid condition and a flow rate of 251 lb/sec for the case of natural circulation with both steam generators in a water solid. condition. The coolant temperature calculations were - c'

4-3 performed for various decay heat levels. The results of these calculations are summarized in Table 4.1 and Table 4.2. These results were obtained from COBRA IV calculations and checked with closed-channel hand calculations. These calculations for each plant configuration and each decay heat level were performed with two different sets of assumptions. The first set used the core exit thermocouple data in a conservative manner by assuming that during natural circulation the fraction of natural circulation flow in each region of the core relative to the average natural circulation flow rate was i the same as the fraction of pumped flow calculated for each region of the core relative to the average pumped flow. For this calculation it was assumed that only about 6% of the normal pumped flow rate is going through the i core. In this calculation the effect of local void generation on the elevation head is conservatively ignored. In addition the increased flow resistance due to two phase flow conditions is also ignored. The second calculation used core flow thermocouple data in a more realistic manner by using the calculated natural circulation flow rates directly from the RELAP calculation. In this case the increased core flow resistance causes an increased core exit temperature and increased local void generation and therefore an increase in natural circulation flow which partially offsets the adverse effects of the local core blockage. These analyses indicate that extensive local boiling might occur under natural circulation flow conditions. In addition, several cases analyzed show superheated steam conditions in one or more assemblies. The maximum coolant temperatures were calculated for the case with one steam generator operating. A temperature of 2960F was calculated at the core exit for the core decay power level on April 16, 1979. As indicated in the tables there is a significant sensitivity of this temperature to the reduction in decay heat as a function of time. By April 30, 1979 the highest temperature is calculated to be 1390F. These calculations also show a significant sensitivity to the assumed core flow rate. As seen from the previous section, the value of core flow rcte is dependent on the total I e l

];'v*pjy mp g m 7.g 4-4 ' core flow resistance. It is reasonably certain that the central portion of the core, which has been showing high thermocouple readings, is very nearly completely blocked to flow. However, there is considerable uncertainty as to the condition of the peripheral assemblies. The difference between an unblocked region and a 50% blocked region is only about 1 F in core exit temperature. Since the exit temperatures are not measurable to within one degree, the condition of the peripheral region is somewhat uncertain. The core exit temperatures presented in Tables 4.1 and 4.2 assume 90% blockage in the peripheral region. With this assumption, the total core blockage is approximately 94%. It should be noted that the hot assembly temperatures in Tables 4.1 and 4.2 only apply to a relatively small fraction of the core. For all of the cases calculated, most of the core exit temperatures are predicted to remain below 1000 F. Temperatures below 1000 F would not be expected to have any adverse influence on the ability to keep the rest of the core cooled. In addition, these temperatures would mi result in any loss of strength in the intact structures nor would they be expected to cause any significant movement in the fuel debris. Temperatures in excess of 1000F would, however, continue to drive more fission products out of the U0,2 thus tending to prolong the eventual cleanup of primary coolant. Prolonged local boiling also will lead to boron precipitation tending to furt.her block areas already difficult to cool. 1 In summary, local boiling and local superheated conditions are possible if the transition to natural circulation takes place during the next few weeks. The maximum local superheated temperatures are all expected to be well below the U0 melting point, the Inconel melting point, and the 2 threshold temperature for zirconium water reaction for natural circulation after April 16, 1979 with two steam generators in the water-solid condition. After April 30, 1979 comparable performance would h expected for natural circulation with one steam generator in the vater-solid con'diticn.

4-5 There is some potential that small pieces of uranium oxide and zirconium oxide particulate debris which are now suspended in the core by forced circulation cooling could fall through the core to the lower plenum once the pump is tripped and core flow drops to natural circulation rates. A simple one-dimensional conduction analysis (conduction in the vessel wall is negligible) was performed to determine the depth of material needed in the lower plenum to cause incipient melting of the vessel wall. The results of this conservative approach are shown in Figure 4.1 fer several values of the controlling parameters. Since approximately 10% of the core U0 v lume is needed to produce a 5 inch layer of debris in the 2 lower plenum, and since little suspended matter is expected to fall to the lower plenum upon loss of forced circulation, tripping '.he pump to go to natural circulation is not expected to add more than one inch to the debris layer already in the lower plenum. The amount of debris in the vessel lower head is judged to be small (see Attachment 1). An additional analysis performed at Sandia Laboratory and reviewed by the staff shows that a debris layer in the lower head would be cooled by cellular convection and that the heat generation in the pile would be too low to cause dryout of the debris. Sandia concludes that the temperature in the debris pile, and hence at the liner of the lower head, is only a few hundred degrees higher than the water temperature. Therefore, we conclude that the debris pile can be cooled sufficiently for ary depth of debrit of practical concern at THI-2. 4.3 Criteria and Conclusions The transition from forced convection core cooling using one reactor coolant pump to the natural circulation mode should be possible with minimum risk of further core damage (zirconium oxidation, fuel melting or structural degradation) and minimum risk of unacceptable activity release. o The functional criteria to accomplish these goals should be af, feiluws:

4-6 (a) Avoid bulk boiling of core coolant (b) Avoid prolonged conditions where local temperatures of core materials would be predicted to exceed 1000F (c) Avoid system conuicions that could result in pressure instability with susceptibility to large pressure pulses, e.g., extensive boiling in large regions of the core. (d) Avoid operations that have the potential for relocation of fuel debris, e.g., starting more than one reactor coolant pump. (e) Avoid conditions that could result in noncondensible gas evolution which could affect the natural circulation cooling, e.g., pressures below300psig,Errapiddepressurization. (f) Avoid boron dilution and the potential for return to criticality. (g) Avoid rapid or unne'cessary additions of cold water to the reactor coolant system (e.g., through normal makeup system) because of the possible adverse effect on natural circulation. (h) Confirm that a stable core AT of 5F to 150F is established during the first half hour of natural circulation. (i) Maintain sufficient void in pressurizer to accommodate heat up of fluid between core and steam generator when going into natural circulation. (j) Maintain sufficient level in pressurizer to accommodate primary system shrinkage if reactor coolant pump is restarted. (k) Keep pressurizer spray line open to vent any gases that could collect i in pump or cold leg. (1) Minimize potential for local boron precipitation which could clog flow passages in local boiling areas of the core. The B&W criterion for avoidance of bulk boiling in natural circulation is to assure 100F subcooling in the hot leg as measured by the RTDs. The staff believes this to be adequate for operation with two steam generators. However, the thermocouples at the core exit are a more reliable indication of core boiling while bypass flow exists in an idle steam generator. We conclude that natural circulation should be discontinued with a return to l l

4-7 forced circulation operation with one reactor coolant pump if the average value of incore thermocouple readings should exceed the saturation limit indicated on Figure 4.2. Additionally, the B&W criteria provide for a maximum coolant temperature rise of 150F across the core and require a hot leg temperature rise within the first hour during transition to natural circulation. In addition, the hot leg temperature is not to exceed 250F for pressure greater than 500 psia (180F for pressure near atmospheric). The staff finds these limits acceptable. However the Licensee's recommended limits of ten or more incore thermocouples reading lower than saturation tempera-ture is not needed in view of the staff requirement on the average readings. Likewise, the Licensee's local limit of 800F on two thermocouples appears arbitrary since local high temperatures are acceptable. However, no more than three of the thermocouples should be permitted to remain above 1000 F for more than one hour if restart of a reactor coolant pump is a viable option. The criteria that no more than three thermocouples should be allowed to indicate local temperatures above 1000 F is based on the following considera-tions: 1. It is our best estimate that only one instrumented assembly will approach 1000 F if natural circulation is achieved on April 24 or soon thereafter; up to five others could show approximately 100 F of superheat. Therefore, three thermocouples with temperatures above 1000 F would indicate that natural circulation is not being achieved as expected. 2. It is likely that during natural circulation many thermocouples will read temperatures lower than the maximum local temperatures. This conclusion is based on early thermocouple data; changes in the thermocouple data following the startup of RCP 2A; and on PNL and ORNL calculations. The early thermocouple data indicated temperature

4-8 differences of 300 F or more in adjacent assemblies. After following the trend of the thermocouple data for a long period of time it seems clear that these differences do not represent significant differences in the conditions within the adjacent assemblies, but rather differences in the thermocouple location relative to the maximum temperature in the assembly. While the higher temperatures are probably the result of material filling the entire assembly region up to the thermocouple location, the lower temperatures are probably the result of having some space between the debris bed and the thermocouple location. Both PNL COBRA IV calculations and ORNL SABRE calculations indicate that flow redistribution in a small region (6 to 9 inches) above the debris bed would be expected to produce a difference of 100 F in the temperature at the top of the debris bed. In addition, this mechanism is a likely explanation of the 120 F drop in indicated temperature of the hot assemblies when RCP 2A was started-up. That is, the 120 F drop in indicated temperature may have been the result of the debris bed settling down to an elevation several inches below the thermocouples. We conclude that an indicated temperature of 1000 F during natural circulation probably corresponds to a maximum temperature of approxi-mately 1300*F in the debris bed. Exceeding 1300 F over a significant portion of the core is undesirable since the structural Inconel loses strength rapidly as a function of temperature at this elevated temperature. The degree to which these structures are needed to support the core in its present condition is unknown. Soms additional zirconium oxidation could also occur at this temperature and would be undesirable. 3. It is believed that the 1000 F criterion is not overly conservative since our thermocouple predictions are based on current thermocouple readings with no core displaceinent when the pump is tripped. We

4-9 expect that some fuel debris will settle under the low natural circulation flow rate and resultant thermocouple readings may be lower than predicted. The primary system should not be in a water-solid condition when transition from forced to natural circulation cooling is accomplished. It is anticipated that local boiling will occur during the transition to natural circulation. Local boiling at present decay heat levels will not be capable of sustaining large voids and the collapse of small voids should not be sufficient to cause core structural damage. However, the resultant small pressure pulses could lead to small scale redistribution of the core debris which would affect flow blockage and the settling of debris. Minimum primary system pressure required to prevent noncondensible gas evolution and an evaluation of the effects of a coolant voids on natural circulation are discussed in Section 5 of this report. Calculations of core exit coolant temperatures based on thermocouple data, indicate that local temperatures could exceed 1000*F if transition to natural circulation is attempted prior to about 5/5/79. The high temperatures will not necessarily be indicated on core exit thermocouples since the debris or blockage which is producing the high temperature may drop away from the thermocouple locations under the low flow natural circulation conditions. The planned transition to natural circulation during the first two weeks of May is consistent with the decay time for local debris that should preclude measured temperatures in excess of 1000 F and will minimize the potential need to restart the coolant pumps which could lead to further redistribution of debris. The transition to natural circulation provides less potential for local boiling and coolant superheating the latter it is accomplished. So long as the reactor coolant pump now running remains in a satisfactory status, forced convec-tion cooling should be maintained to take maximum advantage of fission product heat decay, consistent with the need to accomplish an orderly and well planned transition to natural circulation. L

4-10 The staff has concluded that the ECC injection modes (HPI ano RHR) do not offer significant advantages over natural circulation cooling and have several disadvantages (see Section 7, below). Therefore, we do not recommend interruption of the natural circulation mode prematurely in favor of HPI or RHR unless the above criteria are exceeded and improved conditions can be reasonably expected in the alternative cooling modes. In addition to the stated criteria, systematic failure of core exit thermocouples in a given region should be viewed with caution during the transition to natural circulation as it might signal high local core temperatures and a warning that effective cooling is not being accomplished. The limitations on process parameters to assure adequate core cooling in natural circulation are summarized in Table 4.4 and Figure 4.2. Provisions for continuous monitoring and recording of incore thermocouples should be included in the plant modifications for n&tural circulation. The staff has developed criteria for detecting loss of natural circulation during long term core cooling: 1) Increase in the average reading of all operable core outlet thermo-couples by more than 10*F, or 2) Increase in average of the two hot leg temperatures by more than 10*F, or 3) Increase in reactor pressure corresponding to a 10*F rise in satura-tion temperature, or 4) Marked and unexpected decrease in total heat removed by the secondary side of the steam generators (FAT decrease). .... ~ -. -, _.., _ ~ -.. - -.., -

4-11 l Once natural recirculation is lost, it may well re-establish itself through the mechanisms described in Section 5.0. Criteria for detecting re-establishment of natural circulation cooling are: i 1) Constant or decreasing average core outlet temperature (average of all core outlet thermocouples) over a period of hour or more at a temperature below the saturation pressure corresponding to the safety valve set point (assuming an unvented system), and i 2) Constant or decreasing average of the two hot leg temperatures over a hour period, and 3) Decreasing reactor pressure over a hour period, and 4) Increase in total FAT on the steam generator secondary side. The staff concludes that the proposed transition to natural circulation can be accomplished with minimal risk to the public health and safety, and that natural circulation is the preferred long term cooling mode. I The planned transition in early May should minimize the potential for local fuel temperatures in excess of 1000F. 4 i I .-.n -.,---,--,,-,_.,.w..

4-12 E Table 4.1 j i Core Exit Coolant Temperatures During Natural Circulation With One Steam Generator Operating in a Water Solid Condition Flow = 231 lb/sec Tin = 125F Pressure = 1000 PSIA X = Steam Quality Calculated Temperatures Calculated Temperatures Case 1 for Case 2 + Decay Heat Day Power Level (No credit for local buoyancy) (Including local buoyancy effect) 4/23 0.08% Hot assembly = 2060F 545F 1 2nd hottest assembly = 545F 345F J Peripneral Region = 183F 140F 4/30 0.065% Hot assembly = 1390F 460F l 2nd hottest assembly = 545F 300F Peripheral Region = 172F 137F i 5/2 0.060% Hot assembly = 1150F 430F 2nd hottest assembly = 53"F 284F f Peripheral Region = 168F 136F I o k i 4 + 4 ,y-#r- -,4--+, w -,oe m .., - ~ - -... - -r..-

m 4-13 Table 4.2 Core Exit Coolant Temperatures During Natural Circulation Operation With Two Steam Generators Operating in a Water Solid Condition Flow = 251 lb/sec Tin = 125F Pressure = 1000 PSIA X = Steam Quality Calculated Temperatures Calculated Temperatures Case 1 for Case 2. Decay Heat Day Power Level (No credit for local bouyancy )(Including local bouyancy effect) 4/23 0.08% Hot assembly = ll20F 545F 2nd hottest assembly = 545F 330F ^ Peripheral Region = 168F 139F 4/30 0.065% Hot assembly = 600F 415F 2nd hottest assembly = 545F 280F Peripheral Region = 160F 137F 5/2 0.06% Hot assembly = 545F 375F 2nd hottest assembly = 444F 245F Peripheral Region = 157F 134F 9 I wa 4

4-14 Table 4.4 Criteria for Core Cooling Operating Modes ) 1. One Reactor Coolant Pump (RCP) Operating 2. Transition to Natural Circulation During First Hour following RCP Trip 3. Natural Circulation Thru One Steam Generator 4. Natural Circulation Thru Two Steam Generators 5. High Pressure Injection 6 Decay Heat Removal 2perating Limits Applicability to Operating Modes 1. Core Exit Thermocouples < 1000F 1,3,4,5, & 6 1,2,3,5, & 6 2. Average of Core Exit Thermocouples < Tsat 3. Hot Leg Temperature > 100F Subcooled 1,3,4,5, & 6 4-Hot Leg Temperature < 250F All Modes when P>500 PSIA 5. Hot Leg Temperature < 180F All Modes when P<l5 PSIA 6. Core aT < 150F All Modes 7. Hot Leg Temperature Increase Within One Hour After Reactor Coolant Pump Trip 3 Monitor Limits Operating Limit Action (See below) 1. Average Incore Thermo-couples > Tsat (II9"#8 4*2) 2 l 2. Two Incore Thermocouples > 1000F 1 2 i 3. Three Incore Thermocouples > 1000F 1 1 4 Hot Leg RTO Exceeds 100F Subcool Limit of Figure 5.2 3 1 i 5. Hot Leg Temperature > 250F 4 1 6. Hot Leg Temperature Minus old Leg Temperature > 150F 6 1 7. lwo or More Incore Thermocouples Fail Within 30 minute interval 1 3 ~ ,e

d 4-15 4 i l Table 4.4 (Continued) I-i Actions _ l. Return to Mode 1, if possible. Otherwise consider transfer to an alternative cooling mode. j 2. Validate incore thermocouple operability by comparison to other nearby thermocouples and by other available means. 3. Danger of local temperatures high enough to melt incare instrument structures, Evaluate and take Actico 1 or other emergency actions as warranted. i

i l

l i

l l

f 4 4 i i 1 1 a -,r

Thickness of Debris Layer to Heat Liner of Lower Plenum to Melting Point (T = 2550*F) 10.. insulation removed y 9 8-conduction through 4 inch insulation 7. n $.co .5 6 L no conduction through vessel ~ .5 5- .3 13

4..

E Fluid Temperature = 280*F 0 water at 280*F h = 1000 BTU /Hr-ft

• F 2

3' h = 1000 k = 1.5 BUT/Hr-ft ** uo2

• o debris k = 10 BTU /Hr-ft *F IP *

.. * *e C e,e.

• • • e.

ss ,3/16 /S liner k = 20 BTU /Hr-ft *F 2-Iff f f f f /f, 5" vessel wall cs 2 c $ f f f f f f4" insulation h g 2 BTU /Hr-ft

• F 1-h ='2 air at 80*F k$sg05 BTU /Hr-ft*F 0

O 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Days After Reactor Trip Figure 4.1 s-

s s 800 Maximum Allowable Indicated Temperature vs Pressure 3 for Natural Circulation with one or two Steam Generators %u Eo 700 9 Average Thermocouple Temperature Limit for operation with one S.G. E (Tsat) o g 600 ueL C 8 L 3 g 500 - .a 5 E % o ",* tc. oo Hot Leg Temperature Limit a S U 400 - (T - 100*) 1 u sat N o o

=

0m -a 8 8 300 < t c F s eo g 200 i'< 100 - - 0 l l 200 400 600 800 1000 1200 1400 1600 1800 Indicated RCS Pressure (PSIG)

l } 5.0 Effects of Non-Condensible Gas on Natural Recirculation Cooling Capability Measurements during the first week of the accident indicated the presence of a large volume of non-condensible gas in the reactor coolant system. Subsequently the gas was removed through a degassing procedure involving pressurizer spraying and letdown. The procedure included a series of incremental pressure reductions from 1000 psig down to 300 psig to remove any gases trapped at high points in the reactor coolant system. It is currently estimated that no bubble will form by evolution of non-condensible gases now in solution or still residing in RCS pockets, as long as the primary coolant system pressure remains in excess of 300 psig. Consideration has been given to the potential effects of radiolysis of the reactor coolant water, i.e., the decomposition of the water to form hydrogen and oxygen. In gamma and neutron fields typical of nuclear power reactor coolant systems, a hydrogen concentration of 17 cc/kg water is needed to suppress radiolysis of the primary coolant (Ref: US Patent 2937981,5/24/60). In operating plants the usual concentration of hydrogen is maintained between 25 cc/kg and 35 cc/kg. In the TMI plant, the saturation concentrations of hydrogen are listed below for the preliminary points in the Base Plan (described in Figure 1.1): Operating Point "A" (280 F, 1000 psia) 1670 cc/kg Operating Point "B" (220 F, 1000 psia) 1540 cc/kg Operating Point "C" (140 F, 1000 psia) 1430 cc/kg Operating Point "D" (140 F, 100 psia) 140 cc/kg. These values are considerably higher than the concentration of hydrogen required for the suppression of radiolysis of water at the operating conditions of the plant. It can be concluded, therefore, that no signi-ficant radiolysis will take place at the operating copditions defined by n

5-2 points A, B, C and D as long as the primary coolant remains saturated with hydrogen. Some radiolysis may be expected if the concentration of hydrogenisreducedbelowthesaturationlimitandjtreachesavalue lower than 17 cc/kg. The Licensee obtained pressurized samples of primary coolant from TMI-2 on 4/22/79 to determine the amount of hydrogen and other gases in solution. The preliminary analysis of this coolant sample indicates the following: Hydrogen - 23.8 cc/kg Oxygen - <l.1 cc/kg Nitrogen - <9.9 cc/kg Natural convection cooling in the primary system (as described for the Licensee's proposed long term cooling mode and evaluated above) assumes the presence of no free gas volume in the primary coolant system for pressures greater than 300 psi for the reasons described above.

Howevet, the staff has nevertheless evaluated the effects of non-condensible gas f

that might unexpectedly come out of solution, as described in the following paragraphs. If the system pressure and the core exit plenum and hot leg temperatures allow gas to come out of solution, the formation of gas pockets could interfere with natural circulation. The gas could be driven back into solution by increasing pressure to maintain sufficient subcooling. The time required for plant operators to accomplish this reverse process is on the order of hours which is within the available time frame for operator actions following loss of natural recirculation. Nevertheless we arbitrarily assumed that the releaqed gas might remain out of solution for the following evaluation. The released gas would collect at high points. These points are (a) the upper portion of the reactor vessel, (b) the " candy canes" near the top of the steam generators, and (c) the reactor coolant pumps. For natural circulation flow, the gas volume in the steam generator " candy. canes" is 9 a

5-3 J of particular concern since it could inhibit coolant flow to the steam generators. If the volume is small as shown in Figure 5.la, there should be little flow impedance. However, if the volume is large enough to fill the piping, as illustrated in Figure 5.lb, the flow could be at least temporarily stopped. Natural circulation flow tests with an intact core and steam generators in a steaming mode at conditions where the decay heat rate is about 0.2 percent of design power have shown that the ratio of coolant flow to core power is roughly five times that for the reactor at full power with all pumps running. The resulting small temperature increase across the core (10-15'F) produces a small driving head for natural circulation ~ which could be stopped by a gas volume large enough to produce a liquid depression, L, in the " candy cane" of less than 6 inches (see Figure 5.lb). D However, with cessation of natural circulation flow, the core and hot leg temperatures would increase and produce a higher driving head. In addition, the cold portion of the system (including most of the steam generator and RCS piping cold legs) will be kept at about 100 to 150*F as the result of the flooding of the secondary side of the steam generators. The resulting increase in driving head should be sufficient to reestablish natural circulation flow for relatively large volumes of gas. Preliminary calcula-tions indicate that gas volumes that could extend down into the upper plenum of the steam generator may not permanently stop natural circulation flow at hot leg temperatures well below the saturation temperature, providing RCS pressures are brought back to the 1000 to 1500 psig range. As noted previously, the gas volume would shrink and eventually disappear as the gas goes back into solution, provided sufficient subcooling is obtained by the increase in system pressure. O e 1 b, k w

-54 gas bubble Liquid D / ' I //777 u u, v -m x v x Tube Sheet / 7s (a) (b) Gas Bubble in Steam Generator " Candy Cane" Figure 5.1 L --

6.0 Long Term Control of Reactor 6.1 Long Term Control of Reactor Coolant Level and Pressure Pressurizer level indication is most important during the transition from forced to natural convection flow. During this transition, RCS tempera-ture and pressure will change. Pressurizer level indication will be desirable to control the pressure and to indicate the evolution of gases in the system which might inhibit establishment of natural recirculation as discussed in Section 5. If the plant is water solid when the transi-tion is made, the control of system pressure and inventory is more difficult. In the long term, when the RCS is operating in a natural circulation mode, the pressurizer level indication is not as desirable as above because the system can be kept water solid. The Licensee has proposed a method for RCS pressure (and inventory) control during long term operation in the water solid natural circulation mode. This system would serve as a backup to the CVCS and maintain reactor coolant system pressure with the pressurizer filled solid with water. Prim,ary coolant system pressure would thus be maintained even with the loss of pressurizer instrumentation and inoperative pressurizer heaters. Also, the pressure control system will be designed to provide adequate NPSH to the reactor coolant pumps if they are needed, and to suppress potential bubble formation which could inhibit natural recirculation as discussed in Section 5. l The standby reactor coolant pressure control system will consist of passive components (a series of water storage tanks and a surge tank with nitrogen blanket for pressure control) and active components (charging pumps). The system will control reactor coolant pressure over the range of 100 psig to 750 psig. The intent is to use the passive reactor coolant pressure system which would be at first locally controlled. Later, additional instrumentation +

~ l 6-2 and remote control will be incorporated to automate this system. The active pressure control portion would resupply water to the surge tanks with added capability of providing additional makeup water directly to ~ the RCS if needed. The above modification will establish a flowpath of makeup water and pressure co$ trol through the normal makeup lines that interface with the reactor coolant loop cold legs. Chemical control of the degassified borated water used in the pressure control system will be provided by the present chemical addition system. Connections will be provided to accommodate the addition of boric acid, H, demineralized water and hydrazine, LiOH, and NaOH. 2 6.2 Core Cooling Without Primary System Convection After some months of fission product decay, an alternative to natural convection coding will be available. It should be possible to cool the TMI-2 core without circulation (forced or natural convection) through the reactor coolant system. That is, the core could be cooled by simply maintaining an essentially water solid primary system and conducting and convecting heat from the reactor vessel to the containment atmosphere. From the calculations based on the specific operating history of TMI-2, the decay heat (Figure 6.2 shows decay heat from 0 to 130 days following shutdown) is expected to decrease to about 0.3, 0.14, and 0.05 MW at six months, one year and two years after the accident, respectively. If the reactor coolant system is kept at a temperature of about 100 F, the heat loss by conduction or convection to containment would be small relative to the heat that can be removed by the steam generators. However, after six months, it appears that decay heat could be removed by heat loss to the air and earth surrounding the reactor vessel in the containment, o L

6-3 while still maintaining the reactor coolant temperature well below normal operating temperatures (the primary system would require pressurization to suppress boiling). Since data for TMI-2 were not available, the heat loss to containment from the vessel was estimated by scaling values from other PWRs. It is estimated that the heat loss to containment from the reactor vessel, the primary coolant piping, control rod drive mechanisms, steam generators, steam piping and uninsulated parts of the rector coolant system pumps is about 1.5 MW at design power conditions. Roughly one half of this heat loss can be attributed to the control rod drive mechanisms which have cooling water coils and forced convection air cool.ing. The heat loss from the control rod drive mechanisms is herein conservatively assumed to be negligible due to conservatively assumed loss of the cooling water and forced air flow. On this basis, the heat loss from the reactor coolant to containment air is estimated to be about 1.6 x 10 3 MW/*F. Assuming the containment air ventilation fans are lost, a ccnservative estimate of the heat loss by conduction from containment to the surrounding air outside containment is 3 x 10 8 MW/*F. From these values, it is estimated that the 0.3 MW decay heat at six months after shutdown could be trans-ferred directly to 70*F air outside of coatainment with a reactor coolant temperature of about 400 F. Better estimates of the heat losses under these conditions can be inferred later from plant records at TMI-2. It is possible that a significant increase in heat removal could be obtained by using the cooling water to the control rod drive mechanisms which would be available for the long term operation. It is noted, however, that natural circulation flow in the reactor coolant system during this mode of operation (steam generators isolated) would be impeded since the major portion of the decay heat would no longer be removed at j = 4 e

6-4 the top of the steam generators. Further analysis of the natural circula-tion flow and adequacy of core cooling will be needed to demonstrate the feasibility of this backup mode of operation. a 6.3 Overpressure Protection / Thermal Shock With the RCS temperature low, around 150 F as in natural recirculation cooling, the reactor vessel pressure must not be excessive due to the possibility of brittle fracture. System pressure should be carefully monitored and controlled to avoid overpressurization. The rapid addition of cold RCS makeup should be avoided since the makeup-inlet points are located in the RCS cold legs, and the cold fluid would travel into the reactor inlet nozzles, down the downcomer, then into the inlet plenum. The cold makeup could cause thermal shock and brittle fracture potential for the reactor vessel since there is no heating of the fluid until it enters the core. Therefore, continuous rather than intermittent RCS makeup should be considered. Consideration should also be given to provisions for heating of makeup water to the long term primary coolant temperature. Fracture mechanics calculations have been performed for several cases that could be encountered in the planned cooldown of TMI-2. In all cases, the possible atypical weld metal in the lower head is limiting. .Nevertheless, assuming reasonable mixing of the water, our calculations show that there is no need for concern about brittle fracture of the ~ vessel unless extremely unlikely conditons occur. Appendix G calcu W. ions were first performed using all of the conservative Appendix G assumptions. These include a 1/4T flaw, the Appendix G K IR curve, and a factor of 2 margin on pressure stresses. This gave a minimum temperature of 160 F for 1000 psig pressure and a cooldown rate of 50 F/hr. Next, thermal stresses and stress intensity factors were calculated for the proposed cooling parameters. This gave a slightly higher cooldown rate and slightly higher thermal stresses and stress intensity factors. h.

6-5 Again, using the Appendix G factor of 2 margin on pressure stresses, the K curve, and the 1/4T flaw, the minimum temperature to comply with IR Appendix G was 170*F. If the pressure is below about 900 psig, Appendix G requirements and margins would be met at 150 F. a Calculations were also performed assuming a pressure increase to 2500 PSIG. Using K instead of KIR, with a factor of 2 margin on pressure stress, IC and a 1/4T flaw, a temperature of 185*F would be required. With no factor of 2 margin on pressure stresses a temperature of 140*F could be tolerated by the vessel. Therefore, the staff's conclusion is that there is a very low probability of vessel failure under conditions postulated to occur during the planned cooldown. 6.4 Solid Conditions in Secondary System Piping The piping systems affected by operation with water-solid conditions on the steam generator secondary side, out to the first isolation valve, are the main steam line, main feedwater line, and the auxiliary feedwater line. The design of both feedwater lines is predicated upon being filled with wate' during operation and therefore, normal code allowable stresses will not be exceeded. Although the main steam line is not filled with water normally, the entire secondary system was filled with water during pre-operatonal hydro-static testing. The additional dead weight contri- ~ bution to the piping stresses for the water-solid condition is accommodated within normal code limits for that portion inside of containment (based ~ upon oral input from Burns and Roe, the architect engineer for TMI-2.) The spring hangers (one on one main steam line and three on the other) will bottom out and act as rigid restraints. For the main steam piping in the Turbine building the spring hangers will be pinned so as to carry the additional dead weight load of the water in the piping within normal code limits. + e .] 4 ?.k6

l t l y f.[. ( ' s ~ 6-6 It should be noted that all ASME Section III CL.2 components used in the cooling system were designed for seismic Category I service. However all of these components, both those that are part of the original TMI-2 Main . Steam and Feedwater piping system and those obtained from other nuclear sites to be incorporated into the OTSG cooling system, are being utilized in a system with different response characteristics from that for which l they were initially designed or are operating with a fluid media different from that for which they were seismicially qualified, i.e., some components designed for operation on steam during a seismic event as opposed to water filled as in the present-system. Thus because of these differences from the original seismic design requirements, which can affect seismic response, these components should not be considered seismically qualified as installed as a part of the OTSG cooling system, sol'ely on the basis of 3 their original qualifications. Seismic capability of these system modifications is not an acceptance i criterion; therefore, no additional seismic evaluation of this system is planned. l 6.5 Steam Generator Tube Integrity Steam generator tubes are required to maintain their integrity during . postulated design basis accidents including a loss-of-coolant accident 1 (LOCA) or a main steam line break (MSLB) in combination with a safe shutdown earthquake. The design basis LOCA results in a calculated 925 psia secondary-to primary pressure differential and the design basis MSLB results in a calculated 2200 psia primary-to secondary differential pressure at approximately 600*F. The required margins of safety against j-tube failure during these postulated accidents are consistent with the margins of safety determined by the stress limits specified in NM-3225 of Section III of the ASME. Code. Furthermore, a factor of safety of three against tube rupture is required during normal operating conditions which i corresponds to a 1250 psia primary to secondary pressure differential at l approximately 600 F. s 4 ww n-w r v --r-, w, -r , e w v v,,,--- ,,.,,-,--w ---n-w--

6-7 Babcock and Wilcox has provided results of laboratchy tube burst and l collapse tests. The burst tests conducted on specimens with defects up to 70% through-wall resulted in no tube failures at pressures less than 3900 psi and the collapse tests on similarly defected tubes resulted in no tube failures below 3500 psi. Three Mile Island Unit 2 conducted a baseline inspection of 100% of the tubes in both steam generators in November 1977 following the hot func-1 tional tests. Tubes with imperfections of 40% or greater were plugged which is consistent with the basis delineated in Regulatory Guide 1.121, to maintain the factors of safety described above and provide an addi-tional margin for possible operational degradation. There has been no indication of primary coolant leakage through the "A" steam generator tubes since March 28, 1979. Therefore, based on the above design bases and the steam generator inspections and tube plugging which was conducted prior to that date, there are no further special precautions necessary at this time for operating the 'A' steam generator at the proposed conditions of 700 psid primary to secondary pressure differential and at temperatures below 200 F. However, the development of a leak, although considered to be highly unlikely due to the small primary to secondary system pressure differentials, should be considered and sore precautions similar to those described in Section 8.3 for the 'B' steam generator should be implemented. Under the proposed operating condit'ons the secondary coolant pressure on the 'B' steam generator will be maintained at a level greater than the primary system pressure such that any tube leakage in the 'B' steam generator would not permit highly contaminated primary coolant to enter the secondary coolant. 6.6 Condenser Flooding Potential safety concerns associated with flooding of the condenser were ~ considered. Since condenser integrity is not normally included in NRC's

6-8 safety review, little informution was immediately available to determine the safety margins for static or dynamic flooding forces. However, operation of the condenser in the spraying mode, with the hot well at normal operating water levels is anticipated, and potential design limitations of operating the condenser in a flooded mode were not relevant. 6.7 Containment The containment internal pressure has been slightly lower than atmospheric pressure since a few hours after the accident. Current operating procedures indicate that the water flow to the fan coolers should be terminated if ~ the reverse pressure differential reaches 2.0 psi. This action would effectively terminate further cooldown of the containment atmosphere thereby terminating the transient. In any case, this would be a rather slow transient allowing sufficient time for proper action. We believe, however, a more severe transient should also be considered. This transient is the inadvertent operation of the containment sprays. Initiation of the sprays would result in rapid cooling of tihe containment atmosphere causing a corresponding rapid decrease in containment pressure. The magnitude of the pressure decrease will depend upon the inlet spray water temperature (BWST water temperature). To assure that the containment does not exceed the design reverse. differential pressure of 2.5 psi, the containment parameters should be maintained above minimum values as shown in Figure 6.1. The figure indicates that for.a'given inlet spray water temperature, the containment temperatures as well as containment pressure should be maintained above minimum values. The pressure.could be controlled by the addition of a noncondensible gas such as nitrogen or dry air. Control of containment temperature could also be achieved by terminating the water to the fah coolers. Fan operation should continue, to assure ~ proper mixing of the atmosphere while eliminating the heat removal mechanism. Since the' consequences of exceeding the reverse design pressure differential are unknown, we conclude that containment conditions should be maintained in the conditions described above to allow for inadvertent spray operation. T e, 9, e ~ W

6-9 6.8 Criticality From an examination of the physics startup tests (boron concentration and reactivity worth, control rod worths, reactivity coefficients, etc.) for TMI-2 and Rancho Seco (approximately the same physics parameters), reactivity states may be estimated for TMI-2. A brief summary of the results of such estimates, results from some B&W calculations on fuel redistri-bution, and some details of these estimates and calculations follow. From the startup tests, if the fuel pellet configuration is essentially undisturbed from its normal condition, it is estimated that the boron concentration for a just critical cold reactor (k=1.0) ranges from about 500 ppm to 1500 ppm, depending on the presence or absence of the fuel cladding, burnable poisons and control rods. The highest required boron concentration is for no control rod or burnable poison in the core but with fuel cladding in place. The lowest boron need is for a normal, rods in, undamaged geometry. If there were no cladding, substituting borated water for clad raises the low boron requirements and lowers the high requirements. Local rearrangement of the fuel, i.e., from a normal pellet lattice to a more homogeneous water-fuel mixture, reduces reactivity. Based on B&W calculations for gross rearrangements of fuel and borated water into optimized geometries and fuel-water ratios (with no structural material or control rods present), it is estimated that at 1500 ppm boron, criticality could occur (k=1.0) with an optimized compaction of about 40% of the fuel, and at 2200 ppm with about 60% compaction of the fuel. At a boron concentration of about 3000 ppm 100% of the fuel compacted would not be critical. These results for the basically intact fuel geometries are based on the followinginformationandextrapolalion'of'datafortheTMI-2startup reports and the Rancho Seco report. ~ 1. Base. TMI-2 was initially critical;at 530*F, zero power, with all 5 control rods out (ARO), at boron: concentration of 1500 ppm. c i

.a3 6-10 2. Burnup effect. Reactivity data indicate that during the approxi-mately 95 full power days of operation the critical boron concentra-tion had dropped about 100 ppe. 3. Temperature effect. From Rancho Seco data the effect of going to 300*F was to decrease the critical boron concentration by about 60 Thus for TMI-2 at the time of the accident the 300*F critical ppm. boron concentration was about 1340 ppm with ARO. 4. Moderator reactivity coefficient. At 300'F the moderator (only) ~ -5 *F (rods out) at 1400 ppe / temperature coefficient is about +5 x 10 -5 oF (rods in) at 700 ppe boron. The total j boron and about -5 x 10 -5 7, j (including fuel) coefficient at about 1300 ppa is about +2 x 10 5. Temperature effect. From the coefficient results the reactivity is not very sensitive to the temperature in the 100-300*F, 1000-1400 ppm boron range. The critical boron concentration with 100'F, with AR0, is about 130 ppm. 6. Control rods in. The control rods were measured to be worth about 10% ak. The boron worth in the 100-300*F, 500-1300 ppe, rods in and out range is about 1.1 to 1.3% ak/100 ppe. Thus the all rods in (ARI), cold, critical boron level is about 500 ppe. This is approxi-mate because of the large extrapolation in the estimate. 7. Burnable poison. The initial burnable poison reactivity worth at full power was about 4.5% ak. At cold conditions and at the approxi-mately 95 full power day burnup preceding the accident, this reactivity is about 3% ak, and is equivalent to about 250 ppe boron. Thus, if the burnable poison is no longer in the' core the cold, ARI, critical boron level is about 750 ppm with ARI and about 1550 ppm with ARO. ~ 8. Loss of cladding. Loss of cladding and replacement by the borated water is similar to a moderator density increase (small effect from Zr). Replacement of the cladding would be'approximately equivalent ~ Qg.h:&V ' ,va m

t 6-11 to a 20% increase in moderator density. At a 5 x 10 5/*F moderator temperature coefficient state, this density change would be about a 3.7% Ak reactivity change. O The average temperature coefficient over the range of boron levels produced by clad loss for (1) ARI and (2) ARO, no burnable poison states, is about -3 x 10 5/*F and +3 x 10 5/*F respectively. Thus the boron changes are about + 150 ppe and the cold, no burnable poison, no clad critical boron concentrations are thus about 900 ppm for ARI and 1400 ppm for ARO. 9. Local fuel-water mixing. Based on B&W criticality calculations and on open literature criticality measurements and calculations, a change in local pellet configuration from the normal pellet and lattice structure to a more broken up and homogeneous fuel-water mixture would result in reactivity loss and a lowered boron critical level for either the ARO or ARI configurations. Thus it is estimated that if the fuel is not grossly redistributed the required boron concentration is probably under about 1550 ppe, and if the control rod material is still reasonably well distributed in the core, under about 900 ppe. If fuel is grossly redistributed the resulting system could, under conditions of optimum moderation, go critical at 1500 ppe boron or above. (Note: moderation is required since solid U02 spheres require enrichment over 5% to be critical.) The B&W naval criticality group has calculated the effects of gross redistribution of the fuel. They used Monte Carlo. calculations (KENO) and cross sections tested in many critical configuration calculations. ~ The calculations used pellet nuclear parameters (because these maximize reactivity) and fuel-water (boron) ratios which had been optimized by sensitivity studies. The calculations assumed only fuel (core average .a ~ v-+ e-yea -,,,y-_q

6-12 enrichment) and moderator mixtures with no structure or control material (except for the boron in the moderator) or burnup. They did a series of calculations in which it was assumed that the fuel between grids (seven regions of about 21 inch thickness) stayed between the grids but were rearranged into optimum fuel-water ratio " layers", and in which (in successive calculations) top grids disappeared permitting the combination of the several "iayers" of fuel. Calculations were done with baron concentrations of 2100, 3000 and 4000 ppm. Estimating leakages for these calculations and extrapolating to some other boron levels, the following results can be estimated. For no collapsing of " layers" the system is well subcritical at a 1. boron concentration.of 1500 ppm. 2. For a collapse of 3 " layers", giving a combination of about 42% of the reactor fuel, criticality would be approached at 1500 ppm but it would be about 4% subcritical at 2200 ppm. 3. For a collapse of 5 " layers", giving a combination of about 71% of the reactor fuel, the system would be several percent supercritical at 2200 ppm but several percent subcritical at 3000 ppe. 4. For a complete combination of all fuel, either in a cylinder or sphere the system would be slightly subcritical at about 3000 ppm. This last result is the basis for B&W advocating a boron level of 3000 ppm to cover the most extreme configuration. BNL (NRC physics consultants) have also performed some calculations of uppar limit, optimized fuel-borated water configurations (reflected s9heres for gross geometry). A preliminary examination of these calculations shows results similar to the B&W results. e d t l' a

1 i 6-13 In summary, calculations indicate that some postulated gross rearrange-ments of the TMI-2 fuel could conceivably be critical at boron concentrations of less than 3000 ppe. These configurations appear highly unlikely, however. With more likely configurations the required concentration would appear to be about 1500 ppe or less, with the requirement for expected configurations under 1000 ppe. I As of 4/7/79 the boron concentration in TMI-2 coolant was believed to be about 2200 ppm and B&W was advocating an increase to 3000 ppe. As of i 4/19/79, based on an evaluation of the results from the second primary water sample (4/11/79), the boron concentration was above 3000 ppe (probably about 3400 ppm) and boron feed concentrations were adjusted to maintain [ the concentration above 3000 ppm. For the present expected core configuration and boron concentration the core k,ff is well under 0.9. At this multiplication, changes in configura-tion or boron concentration producing reactivity changes of about 1% generally would not be unambiguously indicated by the usual (excore) nuclear instrumentation. Reactivity changes of about 1% would normally 4 be indicated once K,ff is above about 0.95. However, in the case of TMI-2 such changes could be partially or highly masked by possibly destroyed neutran sources, gamma backgrounds, or disturbed or changing geometries or water densities. 4 1 Thus the prediction of the ability to see an approach to critical would be highly speculative. The excore startup range nuclear detectors have been reading higher than normal (compared to post cold start ups) for the expected reactivity state of the core. While there have been no apparent correlations of count rate with boron concentration changes in the reactor, thus indicating the higher counting level is not indicative of high l multiplication in the core (which should be significantly affected by 1 boron changes which have occurred), this does, at least, present some evidence of the uncertainty of monitoring approach to critical. m men o6*-we e --w.,---.._ --m, _v.---n _y-,- mn

6-14 Because of the possible difficulties on monitoring an approach to critical and because there apparently are possible (even though unlikely) critical configurations of the fuel material (and water) with boron concentrations under about 3000 ppm, it is recommended that the boron corcentration not be allowed to fall below 3000 ppm. If, during any operation, there are steaming processes which may concentrate boron to greater than 3000 ppm such that plugging of some part of the system may become a significant possibility, the changes should be monitored as closely as possible and the boron feed level adjusted accordingly. I e e d a 9 4 ow e + e e-e= r a m -- 4

a i e Minimum Allowable Containment Pressure vs BWST II 0 tempterature - TMI-2 2 Design External Pressure = -2.5 psi 14.Q. E:u a Ea T 0 $0 0*F .3 13.0 - 0*F g 4 70*F a 8 w C M 4 60*F Minimum Containment Design Pressure N(-2.5 psig) w w w 30*F 40*F 50*F 60*F 70*F 80*F 90*F BWST Temperature

• F Figure b.1

6-16 Percent of Full Power (2772 MWt) infinite operation Use Nominal Values for ANS-5.1 (1971) - 98% FP

1. 5.,

Corrected for: (1) finite " history" 15d 15d 55d t 1.4<. (2) new U35, U36 Pu 39 decay heat information (ANS1978) (3) actinides 1.3- % 6% (2a) (absciute)< Uncertainty expected N correct) g (1:! " history" is l 1.2-_ 1.1 -- - NOTE: Escat ed gases will reduce these sonx what, especailly in tine frare wher a nobic gas 411 innnvenne. 2ctiuity c, 1.0~ .9. - .8 5 l r: A .7= \\ m ( \\ s o .6; a 5 8 .5-v .4- \\ ~ \\ .3 = ~ \\ \\, ~ .2 w .1-i 0 l l l. t l [ O 2 4 6 8 10 12 14 16 18 20 22 24 hours 0 2 4 6 8 10' 12 14 16 days 0 10 20 30 40 50 60 70 80 90 100 110 120 130 days Figure 6.2

7.0 Contingency for Loss of Natural Convection By the time the natural circulation cooling mode is achieved at TMI-2, decay heat levels will be sufficiently low that time will be available to consider alternative ways of cooling the core if natural circulation cooling is lost (on the order of hours would be required to reach satura-tion temperature in the RCS, and several more hours would be required to boil away water above the core elevation). Some of these alternatives are discussed in the following subsections. 7.1 High Reactor Pressure If reactor pressure is high enough to provide adequatt net positive suction heat (NPSH) for the reator coolant pump (RCP), the quickest and most obvious alternative would be to jog and/or run one RCP. Since the most likely cause of natural circulation stoppage is some type of flow blockage (e.g., gas collection at some vital location such as the upper loop of the " candy cane") this action might sweep the bubble or other blockage to a less critical location permitting natural circulation to resume. 7.2 Low Reactor Pressure If reactor pressure is not high enough for the alternative above, then an attempt should be made to increase pressure. Pressurizer heaters and/or an increase in flow from a pressure controlled charging pump should be attempted. The pressure increase by itself might cause resumption of natural recirculation flow by decreasing the size of any collection of gas that might be blocking natural circulation and allowing it to resume. If not, when sufficiently high primary system pressure is reached the first alternative above can be attempted. O

7-2 i 7.3 Reflux Boiling L= Reflux boiling is the production of saturated or slightly superheated steam in the core (thereby cooling the core), circulation of that steam to the steam generator by a combination of diffusion and pressure differ-f ~ entials in the system, and condensation of the steam in the steam generator tubes by transfer of heat to the secondary water through the tube walls. 1 If a gas bubble large enough to stop natural recirculation is formed, ~ that bubble is also large enough to greatly inhibit condensation effi-3 f-ciency in the steam generator (s). Thus the core temperature / pressure needed to achieve a stable " reflux" cooling cycle might be unacceptably high (an analytical basis should be provided before this cooling mode is attempted). The only obvious ways to improve the condensation efficiency

1) cool the steam generator - if in the steaming mode, possibly the are:

j mode could be changed to the flooded mode; 2) lower primary system water I level to increase the condensation surface available to the primary steam. (This would be of questionable value since the steam would have i to penetrate a considerable distance down very small tubes to utilize the added surface area.) It is not clear how effective this penetration 1 would be, but again analyses should be provided before this mode is tried. Lowering water level would be an unlikely manuever to try under [ the poorly instrumented conditions likely by the time this situation might arise. In summary, core cooling by refluxing might be possible, j although it seems unlikely and we do not recommend wasting time trying to optimize conditions to achieve this mode. 7.4 RHR System .i Core cooling utilizing the existing RHR or utilizing the skid mounted RHR i as a backup could be attempted in the event' natural circulation is not } available. Pressure should be increased as much as possible within the f RHR range to minimize potential RHR pump NPSH problems and gas evolution problems. Pressure could be slowly decreased after initiation of RHR to minimize leakage to the environment. Since RHR suction from and discharge t + 1 I ~... ~ -....., _ -, -. ,,,,. ~

7-3 to the primary system is at a lower elevation than the top of the pressure vessel and upper loop of the " candy cane" where gas bubbles are most likely to collect, those cooling modes should still be possible even after a bubble has stopped natural recirculation flow. In any event, water returned to the vessel should be hot enough to avoid a thermal shock problem at the vessel welds (temperature can be controlled by l controlling cooling in the RHR heat exchanger, i.e., reduce secondary water flow to the exchangers.) 7.5 Contingencies If primary system pressure and level indication is available, and if controlled venting capability is retained, HPI can be used intermittently to provide makeup flow for boiloff cooling for operation over a wide pressure range. (Heated suction flow should be provided to avoid NOT problems on the vessel welds). Flow to containment would be small by the time this situation arises - on 04/12/79 calculated boiloff flow is only i 225 gpm to remove all decay heat and will decrease to half that value by about 05/04/79. This amount of water could be allowed to accumulate for an extended period in containment, or could be removed and possibly recycled through HPI (if leak tight equipment can be made available). If controlled venting is not possible but level indication is available, intermittent makeup flow (as above) can be provided to keep the core covered, but system pressure will go to the pressurizer safety valve ~ setpoint whbre venting will occur. Primary system pressure indication would be desirable, but not necessary, as it could be inferred from HPI injection flow and pressure as discussed immediately below. ~ For the more degraded conditions discussed below, it would be desirable to provide HPI with a throttled output, bypass return line, or a surge tank controlled pressure so that injection flow and pressure can be carefully controlled and monitored. Rate of pressure change versus flow can be used to calculate steam and gas voids in the system. If experience has shown no gas evolution, then this method can be used in a closed

7-4 system to determine volume that is not filled, i.e., a level indication. Injection flow rates would have to be low so that pressure measurement of the injection flow would closely approximate primary system pressure, i.e., little pressure loss in the injection piping inside containment where pressure losses cannot be measured. If level indication is lost and controlled venting is possible, HPI flow can be adjusted along with venting rate to keep the primary system nearly full at any desired pressure. HPI flow rate vs. pressure change can be used to calculate " steam" (non filled) volume, and control can be based ~ on this calculated volume, as discussed above. If controlled venting is not possible, the above cooling mode can still be accomplished but the pressure would have to be at the safety valve set pressure. This could be accomplished by slowly increasing HPI flow rate while monitoring HPI discharge pressure vs flow rate until calculations Es discussed above show the system to be nearly solid (i.e., full). Periodic repetition of the slow HPI flow increase procedure could be used to " benchmark" times when the system was known to be full. At 20 gpm or less boiloff rate, these exercises could be very infrequent - it would take a full day (24 hours) to boil away 1/2 the primary system inventory. 3 O e O 6 e 4

8.0 Radiological Considerations The potential radiological consequences of loss of let-down flow, use of the RHR system, and steam generator leakage are consideration in this section. 8.1 Purification Demineralizer Substantial radioactivity may have built up on the purification demineralizer such that if the flow is stopped, the bed will heat up due to decay heat. ~ Rough calculations indicate that the relief valve will lift and discharge small amounts of water and possibly traces of steam to the Reactor Coolant Holdup Bleed Tanks (RCHBT) if the system is isolated. As long as some flow is maintained, there.should not be any steam. If water and traces of steam are relieved to the RCHBT, the offsite consequences should be nil because these tanks vent to the waste gas vent header which can be placed at a negative pressure by venting back to containment. Procedures should exist for venting the waste gas vent header back to containment should this become a problem. Heat in combination with radiation damage could result in degradation of the demineralizer resin. Radiation degradation that would lead to physical property changes should not occur within the next few weeks. If there has been more fuel degradation than the 0700 3/30 primary coolant sample indicai.cd. it is possible that the resins could physically break down. l This could lead to plugging of the demineralizer lower retention screens, thus blocking flow. It is our understanding that the valve operator for the inlet to the purification demineralizer has failed thus making realign-ment of letdown flow difficult. We recommend that procedures be considered for flow blockage in the purification system. The radiation exposure for the demineralizer resins will also decrease l their ability to ion-exchange. It is expected that decreased ion-exchange is now taking place and that radioactivity could leach off of the resins in the future. This should not be a significant concern because down-stream components are heavily shielded; however, local radiation levels in this area of the plant could increase. t

8-2 i 8.2 RHR System If it is necessary to use the RHR system, iodine releases from leakage could occur. A method to minimize radioiodine releases would be to install a skid mounted charcoal filter system in the RHR room. Such units already exist and could be lowered through the RHR pump room equip-ment hatch. This has been considered and installation prior to reactor systems operation which could lead to a likelihood of RHR system operation is being pursued. The design flow rate of air from the RHR pump rooms is only 350 SCFM. This is a small flow and a small charcoal filter system could be installed in the exhaust ducting if room exists. This would supplement the large Auxiliary Building Filter Units which may become degraded with time. A small fresh charcoal filter would reduce iodine releases by at least a factor of 100 if the RHR had to be used. 8.3 Steam Cenerator Leak Consideration should be given to methods of detecting "A" steam generator leakage with a flooded secondary side condition. Procedures should exist similar to those which follow relative to the operations of the "B" steam generator for minimizing releases should leakage occur--e.g., use of condensate polishers on recirculation to the hotwell and maintaining the condenser at a pressure negative to the condenser circulating cooling water. The secondary coolant presently in the "B" steam generator is contaminated due to the initial primary to secondary leakage which occurred on March 28, 1979. Under the proposed operating conditions the secondary coolant pressure will be maintained at a value greater than the primary system pressure such that if steam generators leakage flow paths are available, the highly contaminated primary coolant will not enter the secondary coolant. However, it is expected that transients of short duration may occur such that a reverse pressure gradient could introduce additional radioactivity into the secondary coolant. To alert the system operators

8-3 to such a condition, indicators and alarms for pressure and radioactivity in the secondary coolant have been provided. These indicators should alert the operator to such an adverse condition so that prompt corrective action could be taken prior to significant additional contamination of the secondary coolant. The steam generator secondary coolant system and secondary services closed cooling water system should be periodically sampled and analyzed to determine if heat exchangers are starting to leak. Since laboratory analysis provides a higher level of accuracy than the continuous radiation monitors in terms of sensing increases in radioactivity concentrations, samples should be taken at a routine frequency of at least weekly or at any time there are indications that possible leakage may be occurring, e.g., increase in the steam generator closed cooling loop surge tank levels. Design features should be provided to monitor and control radioactive effluents from the secondary coolant system to other less contaminated portions of the facility, e.g., secondary services closed cooling water system (SSCCWS) and turbine building floor drains. A radiation monitor should be installed on the SSCCWS to alert the operator to a leak from the secondary cooling system into the SSCCWS. Potential leakage paths from the secondary cooling system to the turbine building environment must be monitored and controlled. The relief over-flow from the surge tank is a relatively high probability source for a spill from the secondary cooling system. To prevent the uncontrolled overflow from the system the licensee has provided a holding tank to contain an overflow. Indication and alarm of high surge tank level would alert the operator to a potential overflow condition to permit corrective action prior to overflowing the holding tank. l Leakage may also occur at various mechan' cal connections in the secondary i cooling system. To the extent practical, locations where leakage is f b

( 8-4 likely, e.g., drain valves, valve stems in valves with a known leakage history, and turbine stop valves, should be isolated from the turbine floor drain system. Isolation in many areas could be accomplished by plugging local floor drains and/or by the addition of curbing to contain spillage. Contingency plans should be provided to transfer collected contaminated liquids to appropriate waste systems. It should be noted that once contaminated water enters the turbine building floor drain system, it is very difficult to prevent its flow to the river. Isolation between the secondary cooling system and services systems, e.g., nicrogen and demineralized water supplies to the surge tank, should be provided to prevent back flow of contaminated water. At least a soft seated isolation valve (or equivalent) and a check valve should be provided. Gaseous effluents from this system should be negligible. The noble gas inventory in the "B" steam generator is negligible because the steam generator was vented. Airborne radioiodine releases should also be negligible because the secondary cooling system is not vented (a nitrogen blanketed surge tank) and the low secondary cooling system temperature (110*F) results in a low air / water partition factor which reduces the volatility of the radioiodine. The pressure regulating valve from the surge tank also vents to atmosphere through a charcoal filter. The staff has estimated the radiation dose rates due to piping in a system filled with diluted secondary water, e.g., such as would occur when the B steam generator is placed in a water solid condition. Table 8-1 shows the results ORNL obtained for calculated dose rates at various distances from various sizes of pipe. As a check, the staff performed hand calculations of the dose rate at one meter from the various sizes of pipe. The hand calculations gave dose rates of 5-11 mr/hr, which is good agreement with the ORNL results. 9 4 ?,

8-5 The source term in 6.5 x 107 cc of water was: I-131 1.4 pCi/cc -3 Cs-134 8.6 x 10 pCi/cc -2 [ Cs-136 1.1 x 10 pC1/cc i -2 Cs-137 3.4 x 10 pCi/cc i cc. The j This source term is diluted to a total water volume of 1.7 x 10s piping is infinite lengths of Schedule 40 with nominal sizes of 12",16", and 20". The dose rates are due almost entirely to I-131 (985). i e e S S 4

8-6 Table 8-1 DOSE RATES VERSUS DISTANCE FOR INFINITE LENGTHS OF SCHEDULE 40 PIPE OF VARICUS NOMINAL SIZES (mr/hr)* Distance from Pipe Center (feet) 12" 16" 20" Contact 53 50 45 2.5 10 12 13-1/2 4.5 6 7 8 10 3 3 4 18 2 2 2 Radionuclide concentrations in water I-131 5.3 x 10 1 pCi/cc Cs-134 3.3 x 10 8 pCi/cc Cs-136 4.2 x 10 3 pCi/cc Cs-137 1.3 x 10 2 pCi/cc h O j

a L_ / umT DSTATES y NUCLEAR REGULATORY COMMISslON ATTACHMENT 1 [ 3 i WASWNGTON, D. C. 20068 kW# APR 131979 p l MEMORANDUM FOR: Roger J. Mattson, Director, Division of Systems Safety, NRR FROM: R. O. Meyer, Reactor Fuel Section Leader, Core Perfomance Branch, Division of -Systems Safety, NRR

SUBJECT:

CORE DAMAGE ASSESSMENT FOR TMI-2 Attached is our assessment of the core damage at TMI-2 for use in the SER for natural circulation. It represents our independent evaluation of the facts available and of the industry / vendor / licensee analysis, which we have heard in several briefings. An earlier estimate of fuel damage was made by Rubenstein et al, and a recent meeting was held at NRC with industry experts. Memoranda describing those evaluations are attached to this document. i 4 vC2 Ralph 0. Meyer,/Section Leader Reactor Fuels Core Perfomance Branch Division of Systems Safety

Attachment:

i ~ i As stated i cc: R. Tedesco K. Kniel P. Check C. Berlinger D. Crutchfield (FOIA File) O. Houston M. Tokar V. Stello B. Grimes G. Knighton J. Voglewede D. Powers PDR ~ -s n AsIo ~ ~K t , VJTA TC/XPJ /.. 3S >f. y p

r CORE DAMAGE A. Introduction For the usual analysis of hypothetical accidents, initial core ~ conditions are assumed and consequences are calculated. This would involve complex thermal-hydraulic calculations and fuel behavior analyses. At Three Mile Island, however, some of the consequences are known (i.e., some information on fission product release, hydrogen generation, and instrument readings is available), so we will use " reverse engineering" as our principal method of backing out an assessment of core damage. We start with the assumption that the core was uncovered and allowed to heat up for significant periods of time. Figure 1 shows the system pressure history for March 28, which includes three periods of significant uncovery. The periods of uncovery correspond approximately with the major periods when system pressure was below the saturation pressure. We will assume that the first core uncovery began shortly after 92 minutes into the accident at which time excore ion chambers show a response spike corresponding to the loss of water shielding. Although the two later periods of uncovery may have produced additional core damage, we will focus on the first period because decay heat was larger then and because that period produced the large radiation instrument reading (at 150 minutes) in the containment indicating major fuel damage. Because the fuel damage to be discussed below is so extensive, we will 9 6

5 g 1 ) g 4 e g 1 L to H. 3 \\ 1 F E \\ R0 U2 S6 2 S g 1 E= RT P ( N O g 1 I 1 TAR l i U E T 0 R A g U S 1 S S D E E R M P U g S 9 S S C A s R ru I 8 o F E M ? 7 I T 6 l 5 I I 4 7\\% P I 3 ] f 2 I I 1 \\ \\ p 0 m 0 0 0 0 0 0 o 0 0 0 0 0 0 0 0 0 0 0 o 0 0 0 0 0 7 5 3 1 9 7 s 3 1 9 7 5 2 2 2 2 1 1 1 1 ,E iEE s 1-7

g a conclude without demonstration that virtually all of the fuel rods in the core failed in the sense of experiencing defects large enough to release Furthermore, the rods probably failed by a LOCA-like ballooning-and-gas. rupture mechanism. Because of the massive oxidation that followed, the mode of failure is probably inmaterial. As a point of reference, Table I lists melting temperatures of the various materials used in the fuel system, i B. Fuel Rods Fission product and hydrogen measurements at TMI-2 give important clues about the condition of the fuel rods. We will deal with fission product releases first. Air and water samples containing fission products have been analyzed. While we have analyzed both for indications of fuel conditions, we have concentrated on the Xe-133 concentration in the air sample. This isotope was selected for analysis for several reasons: (a) it is a noble gas and will not react, plate out or condense, (b) it has a relatively long half life (5.29 days) and a high production rate (6.8 atoms per 100 fissions) and therefore will be abundant thts reducing measurement errors, and (c) fission product release correlations are much better established for noble gases than for other fission products. i Bettis (BAPL) has evaluated the Xe-133 activity and concluded it is equivalent to 31% of the total core inventory. We have independently checked e this calculation (but, of course, not the sample activity) and agree (31.5%).

t e s TABLE 1. KLTING TEMPERATURES ~ ~ FATERIAL TElPEFATURE, C TBPERATURE, F 2805 5080 ZIRC-4 1850 3362 2715 4919 INCOEl. 718 1260-1286 2300-2346 30ClSS 1399-1421 2550-2590 2030 3686 AL23-BC 0 q As-IN-CD 8W 1472 2750 4982 uae>B3 r

• Two fuel assemb. lies contained gadolinia test rods.

1 e 9 9 8

• weM ae

+d-O N n --g-n,n-7- n n-w

r f'~ Fission products including gases are normally retained by the UO2 pellets. A normal pellet release to the fuel rod internal voidage is only 1 or 2% (even for a successfully terminated LOCA) so that a 30% release indicates additional release from fuel pellets not just a ~ release of the gap activity. Fuel pellet releases are strongly dependent on temperature, and Figure 2 shows a correlation of release versus temperature for Xe-133 (from a recent ANS-5.4 draft standard). The correlation, however, is for steady-state releases and we are dealing with a transient. Further errors are possible because of kinetics changes due to oxidation to U 0 or U 0. Nevertheless, it is a reasonable 4g 38 approximation and is consistent with recent short-time annealing ex-periments (private communication 4-10-79, R. A. Lorenz, ORNL) and earlier annealing work (G. W. Parker et al., ORNL-3981 - See attachment A). Parker heated irradiated samples in a furnace for 5.5 hours. The samples had burnups ranging from trace to 4000 mwd /t (about the same as TMI-2). Parker measured releases of about 5% at 1600'C,15% at 1800 C and 40% at 2000 C with an uncertainty of about a factor of 2 in release. These experimental releases for conditions roughly similar arein to TMI-2, but for different isotopes, fair agreement with Figure 2. Using Figure 2 we could conclude that (a) the fuel was uniformly heated 0 (uniform in axial and radial directions) to about 1750 C, or (b) 0 30% of the fuel melted while 70% remained below 1200 C, or (c) 1 any intermediate condition existed. Because of the core uncovery sequence,

i I Xe-133 FISSION GRS RELERSE ~' 2-DRRFT RNS-S.4 STERDY-STRTE FISSION GAS RELERSE NODEL ,/ tu en c ta in.. a / O tJ l x JT Z l

a. +

H d' o x l M l'~ n d-t 26N 2000 22'0 2400 1000 12b0 1400 tii00 18b 2000 0 FUEL TEMPERATURE (DEG Cl FirmE2. l

l 6 [os the fuel rods probably did not heat up uniformly in the axial direction. It is reasonable, however, to treat the fuel rods as isothermal in the radial direction because of the low heat flux. Figure 3 illustrates this point with a comparison of a full-power radial temperatur.e profile and a decay-heat-power temperature profile. 1 There are physical limits on how hot the fuel can get during the ~ periods of core uncovery because the fuel rods have a large heat capacity and a low heat generation rate. If one assumes I zero heat removal (this would produce the most rapid heatup rate possible) during the first period of uncovery, the heatup rate is still fairly slow. Figure 4 shows the adiabatic temperature increase with time for the peak-power axial location, for the low-power ends of the rods, and for the average location. Since there must have been some heat removal thus further slowing down the temperature rise, pellet i temperatures probably did not reach the melting point. Figure 5 shows the temperature changes with time for a surface heat transfer i 2 ~ coefficient of 0.5 BTU /hr-ft _o, which is a very small value. F The results on temperature distribution are, therefore, not conclusive. It is unlikely that fuel temperatures were uniform and no lower than 1750 C, and it is also unlikely that any fuel (UO ) melting took place. 2 The fuel, however, did get very hot compared with its normal operating l i temperatures. Oxidation of Zircaloy.by steam and the attendant decomposition of water 'd provided the major source of hydrogen in the TMI-2 vessel and containment. .y. y .,w w, ..,p, m- .,,c- ,-.w w -. -, ~

i RRDIRL FUEL TEMPERATURES 8 e = $~ G d o 88-u5 t-E 8-w e Eu = = = = = = = = = a $~ LEGEND o - PRE-ACCIDENT a - 1.5 HOURS POST-RCCIDENT (CORE CGVERED) C3 i a I 0.0 0.1 0.2 0.3 0.4 0.5 RADIUS (CM) FicURE3.

O 4 g -8 a 4 II .g o_ CCU D sss s 311 c a N tD(O M e to y I-e ee4 C "N w" I ~ w H E C aa c w b -A c 1 a = s s e a M M M W M M ~ (3'G12OI N I W '~

1 1 i l m 8' I d -B i DH C -9 Id CCC x sss g zzz g e *s : ws e g -gg CD w C W wa i: c _.3 .g xC [d Z u; Oe a a s a 6 M M M W W M C i ~ (0*M) MI N I M 1 i o..

/ / The Containment Systems Branch has estimated the amount of hydrogen present in the plant (Attachment B) after the periods of core uncovery that caused fuel damage. They included amounts (a) consumed by the hydrogen ~ explosion (226 lb mole), (b) remaining in containment after the explosion (80 lb mole), and (c) in the primary system bubble (76 lb mole), which was corrected for radiolysis. Comparing the above amounts with the total amount of hydrogen that could have been produced if all of the Zircaloy in the fueled region reacted with water, we get 41%. As with the temperatures, an ambiguity exists. This coulo mean that (a) about 40% of the cladding wall thickness is unifomly oxidized throughout the core, or (b) 40% of the fueled region of the core has fully oxidized cladding, or (c) any intennediate condition exists, i Figure 6 shows the time required for total wall thickness oxidation as a function of temperature (Cathcart-Pawel correlation). It is clear from Figure 6 that complete oxidation is possible in cladding 0 segments that reached temperatures of around 2000 C during the period of core uncovery. It is also clear from Figure 6 that all of the cladding 0 did not experience sustained temperatures of around 1750 C else it would all have oxidized. This is further evidence that fuel temperatures o were not unifonn throughout the core, and that temperatures locally were very high. Based on early estimates by the Analysis Branch of core uncovery, we will assume simplifieo uncovery histories shown in Figures 7 and 8 for the

e Ma p*%meam e m-ee ..m,,- h 1 TOTAL OXIORTION TIME a-e: 9 \\., TIME FOR TOTAL CLADDING RERCTION AS A FUNCTION OF TEMPERATURE g_ g9: ta H3 - z E n ~ D lh N 'e, Ea ] '=, %g S g e t g %g o-U1 5 3 3 I 3 I I ( I 4 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 TEMPERATURE (DEG. C1 FIGURE 6.

i o CORE UNCOVERY CASE 1 o, -................., ~. .N o - na - cn - u I c O Zo .-. o - J y i >o_ Me _3 i-. z a o- ~, l J* o 's i o o y o. e o ~, a , ~, R- ~. ~. ; i ~ o i 1.5 2 2.5 3 POST-SCRRM TIME (HOURS) Firme 7. )

P I e n .l was. m U

• m,

.m u rn v-(c 5 O ox I tar 5-- CC r U x O o I a p r_. z u, a -n a. w xaO co I t e m. LL. Oh1 Ob1 r$1 de d9 d6 de o (53HCNI) 13A37 1NW1000 3B00 8 .4 e iag

// e folicwing calculation. Fuel that is covered will be considered to be cold (i.e., r.o cladding oxidation). Fuel that is uncovered will be 4 allowed to heat up; fuel that heats up will be given a heat transfer ~ coefficient that is adjusted such that the total integrated oxidation is 40%. These calculations give the oxidation distributions shown in Figures 9 and 10, and these distributions are insensitive to many of the assumptions that were made. Figures 9 and 10 thus are more probable distributions than 100% oxidation over 40% of the core or 40% oxidation over 100% of the core. Figure 11 is a recent best-estimate embrittlement correlation (Kassner et al., ANL) that,shows hig'1-temperature fragmentation of quenched tubes at about 30% oxidation. Using this correlation, Figures 9 and 10 indicate that a fragmented region of about 5 ft. in height exists near the top of the core. It may well be right at the top of the core as a result of simplifications in our analysis. In any event, at least a 4 to 6 ft. of intact (but partially oxidized) fuel rods remain standing at the bottom of the core. Figure 12 shows fragmented Zircaloy cladding after oxidation in a simulated-LOCA test. Kassner (ANL-78-25 and ANL-78-49 reports that at 0 high temperatures (> 1250 C) many fragments are produced whereas at lower temperature the rod may simply break into two pieces. Inasmuch as TMI-2 temperatures were higher than 1250 C and oxidation was severe. O small fragments of the size shown in Figure 12 should be expected along with t larger tube-like pieces. 1 +

- n l 0 _.........- - -..........- ~......-.- .o -m ~ [.il (.D ~* [ H O u U L l -A _ 8 ZO - m r_. ~ pg y O w -w wx \\ W ,a O a U i (E i O -n o .n m u o i i 1 9'O 9'O &'O 2'O O 03ZIO1X0 SNI00973 JD NOIl098J t..-......

e e e CORE OXIORTION - CASE 2 e 9 9 9 0 8 9 0 'I 4 O sn u d- = N O l w a O l I .Z. o l g .e o [ e] e o l ,r.. l' L_ O O I e i Z I O 8 b o / l [m M d~ l" \\. L l . ~ e g m p. %s ,e*

• • • e..

ee## _,,,e CD 0 1 2 3 4 5 6 7 8 9 10 11 12 CORE ELEVATION IFEET) FIGURE 19.

o ) FAILURE MAP FOR ZlRCALOY-4 CLADDING BY THERMAL SHOCK OR NORMAL HANDLING ISOTHERMAL OXIDATION TEMPERATURE (*C) 900 1000 1100 1200 1300 1400 1500 100 g ) g l l l g g i g O CLADDING INTACT 90 - 4 FAILED ON QUENCHING 80 SURVIVED QUENCHING BUT ~ 2 FAILED DURING HANDLING ~ 70 O e m O 8 E O O O O ,e 8 w O O o d F" 40 - 17% LIMIT go ,0 e, ,e e f f e r i O O O k 8 20 L O - -O - 0.O O -O g 10 Ih [ o db O O 1477 K LIMIT !I I i I i O 1 i I i I i i 1200 1300 1400 1500 1600 1700 ISOTHERMAL OXIDATION TEMPERATURE (*K) FIGURE 11.

F*I( -i 1-: a CI 3 Eg j 33 Cg s3 [*II3 d _s y is ~

'!4 c.ga Fi Id ba j Et 1-ir TC #5 L* gj TC #6 gg f3 h l,

. _ i. _ -..- 85 ~ i. ;j. E8 m a p g hglgg hi th W' [ _ l-J y g hIf.,3-3[ $ I{- k i4 Y FI -E;,3 ' t -1 : f-i r3 I4 -.._

1. 3

~~ his- --I s-3 FI ag -~ '1 #2 h: I,-Is-[ [-I s-s J ' 3 'i. g1 y ~ h3 83 C ,i

-I. 8-!

HI, s-: .. _ __ i pg ',3. - g,3, 7. o: - ..4 .,_.m Fig. III.12 Fig. IH.13 Zircaloy-4 Cladding after Thermal-shock Fail-Zirealoy-4 Cladding after Thermal-shock Fall-ute Showing location of Therrrucouples That ute Showing Location of Thermocouples That Produced the Temperature-vs-Time Curves in Produced the Temperature-vs-Time Curves in Fig. III.10. ANL Neg. No. 306-78-223. Fig, m.11. ANL Neg. No. 306-78-224. 6 FIGURE.12. ~ C

Fuel pellets normally crack during operation and crack healing can occur at power. Figure 13 is a typical example of a cracked pellet. Quenching during core flooding may also promote fragmenting of the pellets. Severely fragmented regions are commonly seen in fuel cellets as a result of extreme tencerature conditions in test reactors. Powdered regions in fuel pellets have also been seen in some PBF tests,-but these tests are characterized by very high powers (> 20 kw/f t) and very steep temperature gradients unlike the lov-power uniform (radial) temperature TMI-2 fuel. Therefore we would expect the TMI-2 fuel to be in millimeter-size granules and larger pieces including whole pellets. C. Unfueled Components (Control rods, guide tubes, etc.) Figures 14 through 17 show the control rods, the burnable poison rods, the power shaping rods, and the central instrument tube. All of these rods and the instrument tube are inserted into Zircaloy guide tubes in the fuel assembly. The materials of which these ccmponents are made are indicated on the figures. An important clue about the condition of unfueled components is provided from instrument readings. The fact that all 52 thermocouples worked ,wcat throughout the accident and, continue to give credible information suggests that a central tubular structural member survivied. It is tempting to conclude that all Zircaloy guide tubes also survived, but this may not be the case since the therm 0 couple is well protected by multiple barriers. e e

w

yr r-4

..-i gb IE . : c.c,/ ~ g - .. v.., (, .a . ' '3.g vp .:j K- < s.'.'* . ?, 7

• I

~.!,,('.j!.?

a. 9 4.x:.

W I '...

w:

-.vq,.,3 ' ',. t.r. .u: :. - ;.,?., _'_ g% < 5;:;.; ::~;_,. g pp.,, .y-y.gn y,hh #l ',, h' 1 c...' 5*.C ;.. ' ', f.g.. k{ '%;i ',);f?M; py p'v*;g. ~ r .w.,n- .h-i .. e

L..

x r.. u~ Tj,_ ;A.a.,..- - w .g. e v. = ' W.$. .a. w m, .%..,ma - - \\ l 4.: c'% r 'i",r'. N fl j. i j ,mA x' i, i -{ l' ' ,) ',f s / 1 i ,,4A'. ' 'i. t ~ ..o e umium a mmemm ime u sim umsi m muu mm iiim >J kg{ c. m: r, Q :( e,p s-4 s ane+

R 1 .s COUPLING c-r s . { '5 0 9 SS) h E ~ Eil 2 n: 2 sE: i l:D flElb EElEh spioER (' SS.Ce rad e C F*1M) s 4T c N -~ ~. - + .~..%, -+- I e 'O -+"*- o. o.2 / f",,, + K 3 D + CS ( '[ (0. */ 4 0 is. 0' 0.) e i c a r =u. i g TOP VIEW .s b..~. .,4. c,. /a ~7~ s +r Y +' a +. ~ '75 O* Y 9 I J.D. X 0, f10 D O NEUTRON ASSCRese8 MATERIAL { T O Ky - / 5"]"., 5 C.d) CONTROL ROD - ' ~ / .C b Control Rod Assembly L FIGRE 14, tjh g y gg l f e L.

s.' t COUPLING gq y . (3.04 SS) _ _j_-. e a = n

==i SPIDER .(ss&-.aen47 O [ [ _~. ___ -*-~_ ~*- , p g l g y -+-+. t Jo + ss (0 0 3 I '%-) + dD ?. b TOP VIEW e -- sr h y e,lelY ~) // = 3 HELJTRON AS$ ORBING MATERIAL (soAy,irz,,,re!) ( 3 c. /.. / - s f 4 ) -~ .+. '+' +-+. AXlAL POWER SHAPING ROO w' ?h 'p e Axial Power Shaping Rod Assembly l-a (q 7 g gg FIGURE 15. /g: b'

coupuhe p (Jo+ SS) \\j y / ff $PloER N / (SS G e ds C F3 g)gg W," M S sLRNABLE Poison . W, ;- Z r- + -i neo ~ ( 6, 01S* s%.) ?"D Y y-j o.a 0 0) { .) wc ~ / h gi 'i' T ( .[! d 1 'Q[ o +-. -+- +... ~ "* +- Tor vita y I L x 7 sunNAsLE Poison- - MATERIAL X , ~ -(. B + C - A lg Os) T 54W l TF e 1 Burnable Poison i FIGURE 16. w. -> G -)\\) d \\] U Rod Asser.bly i s.....

Fixed SPND assembly cross section INCONEL OVERSHEATH CAllBRATION TUBE hl BACKGROUN.D N DETECTOR l \\ ~ g _ e.oes - h= f l -SELF POWERED NEUTRON DETECTOR (7) C.;.

- t y

I h! oo (V s THERMOCOUPLE (1) FIWRE 3. 1 'I i R

INEL has made calculations of guide tube temperatures by parametrically varying heat transfer conditions (see Attachment C). Their results show that guide tube temperatures lag the fuel rod temperatures by only about 20 F. Babcock & Wilcox has performed similar calculations and concluded that there i.s a much larger spread in temperatures. We believe the INEL calculations are more nearly correct and that temperatures of unfueled components were close to fuel rod temperatures. 0 Since fuel rod temperatures are believed to have exceeded 1750 C in the hot region of the core, then in that region (a) Ag-In-Cd and its stainless steel cladding would have melted, (b) Inconel spacer grids would have melted, (c) Zircaloy guide tubes would have oxidized, and (d) Zircaloy cladding of the burnable poison rods would have oxidized. In the cooler parts of the core below about the 4 to 6 ft. elevation, we would expect all unfueled components to be intact, although perhaps damaged, just as the fuel rods are expected to be intact. Control rod segments could have only fallen about 3 inches if severed by melting in the hot region, and the Ag-In-Cd absorber should be in place because it is an insoluble metal. Although the burnable poison rods would also be expected to be in place, their poison is probably lost; boron is known to leach out of 8 C-A1 02 3 pellets when exposed 4 ~ to water in a radiation envifonment. l s '

• beesin es...,.

,,g g 7 + 4-r e L

n. D.

SUMMARY

Many or all fuel rods may have ballooned and ruptured, but this mode of initial defecting is probably irrelevant in light of later more extensive damage. In the hot upper central region of the core, fuel temperatures probably exceeded 1750 C releasing large quantities of fission products; about 30% of the total core inventory of noble gases was released. About 40% of the Zircaloy cladding reacted with water. This region of severe oxidation was localized above the 4 to 6ft elevation and may not have included peripheral bundles. The severely oxidized fuel probably fragmented into pieces ranging from millimeter size to whole sections of rods. The temp 2rature of unfueled components lagged the temperature of fuel 0F rods by only about 20 so that they also experienced temperatures above about 17000C. Consequently, in the hot region of the core Zircaloy components should have oxidized, and components with Inconel, stainless steel, and Ag-In-Cd should have melted. Because of many layers of protection, the thermocouple tubes have survived even in the damaged core region, although the outer sheath of the instrument tube may be badly damaged. Nearly all of the broken and oxidized fuel debris should remain trapped in the upper core region because the ppper end fittings have a grillage that would act as a screen. Furthermore, the compaction of fuel debris t

's limited because it is fabricated with a packing fraction of about i 46% and the theoretical maximum packing fraction (for a bed of spherical particles) is only about 63%. It is very likely that fuei debris are also trapped in some mixing cups (See Figure 18) contributing to non-uniform thermocouple readings. An earlier estimate of fuel damage in TMI-2 was made at NRC by Rubenstein, Meyer, Tokar, and Johnston. That estimate is in general agreement with the present estimate although our current evaluation is more refined. A memorandum swanarizing the earlier estimate is attached as Attachment D. E. Recorcendations Reactor fuel is rugged, and it is unlikely that limits for natural circulation conditions will be related to fuel behavior. The general criterion with regard to the fuel should be that additional Zircaloy oxidation and fission gas release should be avoided. 0 Significant oxidation rates do not occur until 900 or 1000 C (See Tigure 6). Significant fission gas releases do not occur until even higher temperatures (See Figure 2). These temperatures should be avoided in the (relatively) undamaged regions of the TMI-2 core, but these temperatures are so high that other limits will probably prevail. By now the adiabatic heatup rate is low (See Figure 19) and ample time will be provided to detect fission gas or hydrogen releases. Therefore, on-line methods of such detection, if feasible, should give adequate A n-

. =. . ~.. p aw .k.bb hb bb VIV3 {*- ..ms on - - ' 5....E /1 DYit. WA 3 -:OCV ? MifC0Y utasanov N / m 1 4 O h b i N Q i o k o W \\ 'l @*^ a k g

k 2

] N ~~ i, Y//////////// N /:>xV

l

RDIRBRTIC HERTUP FOR RPRIL 15, 1979 R 8_ N l do I u 8-a~ ue R @- c-ran. r N8 I O aW .)' a' LEGEND o - 9 W/FT o-6 W/FT i

• - 4 W/FT t

l 0 t d 5 I d E 10 i TIME (HOURS) FIGURE 19. h e

warning of fuel damaging conditions. A discussion of instrument responses relevant to fuel behavior was held with a group of fuel experts from across the industry. A summary of those discussions was prepared by W. V. Johnston and is attached as Attachment E. One consensus of that group was that in-core thermocouple readings should be recorded continuously. A recommendation for such data recording was made and is attached as Attachment F. 4 9 9 e e

• enutses @ S. w e b

1 l ATTACFf1ENT A. s. >e s~- 4 g ( -..s '$ }\\\\ s. ^ c ..~ i ' t7C7CP *C Lc?. ' t: r 1., E.,, f,.1 s....E. f.,;,..:i f .t ~ 2' o. .. P o T. !!7 Y Uq p u,r G CP ClP 1,N.;.0,02.1 J 29 -l'.U..U U-f k'E {i'!RC. l e - .,. ~ a.1'v<,. w, ,p y'jugu. n.,a . o. f19 04*J5' ADP if.ha rY".M4^B 0f V"IME' TU'S 4* /Y. .'N' 3' ' # i~#' , !. / rn e s.i mr - 'cav

h. -

< - r- . TC "

• ErYE'".e' CE "" dVf S #0 '

" * *"I" " *"' #

1. 1. H -..'

..'. '.. g ':. n? ~.:.... ' j.:,' UstD'ecIFirvd.0 [ % th.,&'.Q.~; Q.y.,...*,.; f......N) p* ;..' .p ..se. r-s- e .,g,.t 9T '. . f,. ~ 4 'N ~ '71 }J 04Th en " 5.NPnct M ' W p,:f 9 N. 8

p.-

n C " '~.- .1 ~ c 'n I.nk.P5PhCF wEbMee A@ijppLhkg.9 eat.'f YG;6PIFIP ' M;V c aTIof i N OppD TO MT,?TFSTS,F'dE9E' L\\,'I)d;g yE 'b ' -.I '$5 NE A e 'CEsJ.WMS FLDtSED.FWUm 4M F ~ %ESE .3 .D a CO'etr Tet '4ITF 40? f?RADI-ATgo.Jo nUP NUPS 'pF T9 8CEt,TrF A000 M'f? MT ui aich irTER Inc api AT-104,:I? Fi;04'ING Img@ aTWSF.Esy M: i s'.5 de. P E:JEcTF STt"'#18M7/H"7ED M >P 7 D D".J"EP09T 0"'U \\ 3"I;* 'TWE enLLe 41'% Jkt47E9h.

• TFE PNCD'TeEartpAggs.A yEm.

4.,', % **. i E sy 4 R'MCP$%.:.NhPM7PA. MAS. ' CORP E4P OTI 6%' TO Ay* 'd.4f'.h O 'NM'L'D Nk%'d^@'7.h: U@Y'. 150xFttt^ 4 f.a e b *r:B5.8.m'N # 0.h .2' 't.!l,; Po no $7.'t;E.*.) ELE AS EY. 5' \\ E W MT G RE onrac. %.t: 'a 'aT tro6 cEs c e % E TI mE aa w a e. n.s AT.w.%mu

s. q

, g g, 3,.

• CEy.e n.y.p[d..n.V0;94, nWE PE5m.Ta g 79 ev >P.T t s. '

r y y d.c f fgqM

Q%ckhm,;f...%

G ...,i..-'$ NM,.. ji M

g. d M M.,~. }}i D. N h

%v. g., % w)W,j. m .

• N*

?.E' f' Y' b'l{'h*) Q a . p. & M[hyhyw&j#%?. $ diT .:iN$. Nb p u. Avn .y... . V :,... %+k..w. 3;%. x . p%m u+p,. 6.:,.v,p;4.%,s: Wiih h Kc..

5. ta W%. W 'i !" '.' - :

. vi -@.. -4 m :'*...c f. a... w + n,.,t..W.q.n ?.N+. : J. f f. y,. y,... c z a 1 v'. t %.,;.. u ..,-g,... T e (l-r' 6 '.*,'., ', k, ':~,, g ~ '. * + e ('.. 2 v,$'!, &u'l,.L,"y vy.. f *?.? r,'i s fD. l' (-. ~ ,',--}. - (s ' n,.. es ~ ~. ....,. rs. se '. ~ . ~.,. ~ '.. G;: ~ l .e 8 - _ _ _ _. - _ _ _ _ ~

Attachment B See memorandum from W. Butler, Chief of the Containment Systems Branch, to R. Tedesco, Assistant D* rector for Reactor Safety, and entitled, "Three Mile Island, Unit 2: Analysis and Evaluation of Selected Containment Related Issues," to be issued on or about April 16, 1979. G 6 e o e w l l

1,, Ju iln.p. - +'I i'i 1:.: j i0 '. I T Ix n r b,., ! (.. Arrnassur C. 1 l 2.3 h f5 - ' " I I/..l .l- ' s,.. s. q- .A c i 't.,. i.e t A.< t s. t ." t c. s. gv o,3 .-ll.c,,,,..I (..s ,.., l l., ij 1 (t. (..\\ ) \\'!.1 t. .J.>tt,. l ~ (*In e..e> l e s ,t

. o, f.!., p. )-

[,,,. g,, ; p,s, g O ./ r..n....... w [ :........ ,1 6 j j..._ .........g... t, g... .......g_..... .:..a

c.. -,.. 4

.:....4." 4 A.t.4.. a ".. '. u.: c i, w s M wk,. p t,., p 3 t.,,.:,:..A.,

.,,;y. d ' C,,,,v.cbve. ju.tt l'O i / i bc 4 b *% . (, c,.. (4 = /* i i i i i O e.. ) 'E.&c.nw (_T = Aj;*/ i /o l. 2u l' i Ve... e4 s j, -(t,

l. e, v. g g &,,,/,> l v... h

\\L.t 6, c..., \\ c. s.,, \\ \\, t s. g'e $160

t. / e,/

g -(,g { g,. ggg k (,$ ,., g j l {O a (. ..c( c,,.,,, ,,) . i.. b.' a ( L. ...- r.... {, ( < '. i I / )..

1.. 5, j i,,,,

. ! I s ) (' P. ft ti ) ~ ~ ~ - /. J'....... ,,r$. 6// A"* pg, 4 , '.. - (,,, ). ,, i <,l, ,.,., s ,I 4

• )( $ g I
• *'.f 3 s, i..{

l q'

, g, o{ g {y I

'O I/ I q 'e l i e e p .e g .e-.4 h. g* g

utelTED $TATES [ NUCLEAR REGULATORY COMMISSION y W ASHINGTON, D. C. 20565 3 j 'e,3 April 6, 1979 / / o* '4 y %*e,*/ ,Q ATTACHTNT 0, MEMORANDUM FOR: E. G. Case, Deputy Director Office of Nuclear Reactor Regulation FROM: TMI Fuel Team ESTIMATEOFFUELDAMAGEINTHREEMILEISLAND(TMI)

SUBJECT:

Enclosed is a brief report describing the preliminary conclusions of the team formed to analyze the probable damage to ^ the fuel system at TMI. he ); L. S. Rubenstein, PSS/NRR R. . Meyer, D /NRR %. 7~A-s-M. Tokar, DSS /NRR W. V. Johnston, RSR/RES l WY Enclosure - As stated l hh (2Y)f) ^ S _ i r bffh

• • e e-pe e

C------__--___-__;______-

9 On Tuesday, April 3,1979 a team consisting of L. S. Rubenstein, PSS/NRR R. O. Meyer, DSS /NRR M. Tokar, DSS /NRR W. Y. Johnston, RSR/RES was fomed to survey the fuel groups analyzing the damage to the fuel system of TMI and draw some preliminary conclusions from their deliberations regarding that damage. The fcilowing individuals and organizations were contacted on April 3 and 4,1979: E. L. Zebroski (EPRI) - (representing the Metropolitan Edison Group) J. Taylor /J. Tulenko, B&W R. Denning, BCL D. McCloskey, Sandia J. Scott, LASL In addition to the infomation obtained from conversations with these organizations and the NRC staff, the team obtained a " sequence of events" fra B&W (Enclosure 1) a group of curves describing the pressure, temperature changes at TMI-2 during the first 15 hours from D. Eisenhut, and a BAPL radiochemical analysis of the primary coolant taken at 1600 hours March 29, 1979 and decay corrected to 0700 hours March 30, 1979. The primary infomation used in our analysis of fuel system damage was obtained from the B&W Company, the Metropolitan Edison Industry Group, and from calculations of the NRC staff (Reactor Fuel Section, CPB; Fuel Behavior Branch, RES). System Effects Using the chronology of events obtained from B&W and the control room strip chart tracing of system pressure for the first 15 hours of operation, we were able to detemine that there were three periods in which the primary system pressure was below a saturation pressure corresponding to a temperature of 620*F. The system changes which caused these periods are described in the sequence of events provided by B&W enclosed with this report. The details of what occurred to cause the pressure changes in the primary system are not discussed here as these are considered in other staff reports (see e.g., IE Bulletin 79-05A, Nuclear Incident at Three Mile Island) and will be evaluated by others. f I

s l 2-Examination of figure 1 shows that the first period in which the system pressure was substantially below saturation pressure occurred spproxip.ately 1.75 to 3 hours after start of the transient. The second period, which was relatively short in duration, occurred in the 4.5 to 5.5 hour time frame and resulted in a small decrease in primary system pressure below saturation pressure. The final period of decreased primary system pressure extended from approximately 8-14 hours af ter start of the transient. It was during these 3 periods that the core was exposed to extensive amo~ nts of steam cooling and experienced l u fuel damage. The group was able to infer from examination of these pressure histories, reports of fuel channel temperature changes with time obtained from the incore themocouples, the behavior of the incore rhodium self-powered neutron detectors (SPND's), and 3'-lcng Intemediate Range Ex-Core Detectors. and the containment radiation monitors some details of when the fuel pins lost their integrity, the depth of the core which was exposed to steam cooling, the probable time periods of that exposure, and the amount of damage to the fuel. As previously stated, the evidence for the level of uncovering was obtained from a B&W analysis of the incore SPND's. It can be shown that : Above about 700Y, incore SPND's (Rh) act as themionic elements and generate currents which are correlatable to temperature. Thus, if a discontinuity is observed in current measurement, a transition in temperature may be l inferred. It was assumed that this discontinuity rep-resents an elevation at which voiding of the coolant i has occurred. l Similarly, the excore Intemediate Range Detectors may be used to provide an indication of voiding. i l ihe infomation obtained from these detectors was consistant with the results from the Industry Group calculation that, in approximately one i hour without introduction of rakeup water, the core could boil down to full uncovering. 1 l 1 l 4..e .u...... ..s i S e, b

. Fuel System Conditions During Period of ist Uncovering During the first period of major uncovering of the core (at least 5 feet of the core was uncovered for about an hour, and perhaps all of the core may have been uncovered for about one-half hour), the uncovered portion reached temperatures high enough to fail fuel rod cladding. At this point, fission products were released into the primary coolant as evidenced by the subsequent alaming of the containment activity monitors. Based on the measured coolant activity and the amount of hydrogen release from reaction of the Zircaloy cladding with water, all of the fuel rods probably defected and released fission products. Fuel temperatures were estimated from calculations based on the fission product analysis of the sample of primary coolant, apd also from heat Based on back-calculations that accounted transfer considerations. for temperatures and temperature-dependent release rates that would be required to~ produce the measured level of activity, fuel temperatures Estimates by ORNL based of 1400 to greater than 1600*C were obtained. on their experiments indicated that the Cs and I releases measured would The have required fuel temperatures of at least 1300*C for an hour. heat transfer calculations indicated, on thg other hand, that the fuel temperature may have been only about 1100*C. In either case since is 2840*C, fuel melting was unlikely. These the melting point of UO2 temperature differences can be rationalized by considering that a small There is portion of the core may have been at the higher temperatures. and Zr0 also a possibility of some eutectic formation between UOat temperat attached to the occurrance of such a eutectic. Later analysis by members -ROM) indicates of ANS-5.4 fission gas working group (including one ofg data and the fuel pellet temperatures as high as 2000'C based on Xe While noble gas assumption that half of the core remained cool. activities lend themselves to smaller analytical uncertainties than todine or cesium activities, the uncertainty in the core fraction that is responsible for the release still renders this result inconclusive.. 3 of the Hydrogen balance calculations indicate thaj from 15 to 301 total Zircaloy inventory has been oxidized. Some of the oxidation, The however, undoubtedly occurred during the latter uncoverings. extent of the oxidation probtbly varies as a function of height in the core, with the greatest amount of oxidation having occurred in the uncovered (upper) portions of the fuel rods. Later calculations accounting for hydrogen in the bubble, in the containment, lost in the hydrogen explosion, and gained by radiolysis suggests that almost 40% of the Zircaloy in the fuel region may have been oxidized. TP8 Staf f~ Calculation 2 CPB Staff Calculation Industry Group calculation 3 B&W, Industry Group and NRC Staffs 4 Industry Group & NRC Staffs BAW Calculation f \\

3 ' As the primary coolant level was restored during the latter portion of the time period of the first uncovering, themal and mechanical shock loadings of the oxidized and c:brittled cladding are believed to have occurred and to have resulted in cladding fragmentation. At the end of the period of first uncovering, virtually all of the fuel rods had defected and released fission products. Although temperatures had been high enough for a long enough time to have caused severe cladding oxidation, continued operation of incore instruments strongly indicates that fuel assembly structural members such as guide tubes remained intact. Control rod materials are believed to have remained in place, as indicated by the absence of silver in the primary coolant. Fuel System Conditions and Effects During Period of 2nd Major Uncovering At about 41/2 hours into the event, the core level again decreased to expose the upper 5 feet of the fuel assembifes. The duration of this additional uncovering was shorter than the first, the system pressure was higher, and the overall temperature effects were less severe, as evidenced by the fact that the themocouples in the outer periphery of the core remained on-scale. Because of the reduced severity of the core conditions during the second uncovering, as compared with the first uncovering, less damage is believed to have occurred to the fuel system. Fuel System Conditions During Period of 3rd Uncovering At about nine hours into the event, the core coolant level again decreased, possiblity down to 7 to 71/2 ft. from the top of the active fuel level.* The core remained uncovered at this level for about one to three hours, after which the coolant level was again raised and covered the core. The low system pressure (a.450 psi minimum), the rather lengthly period of ' uncovering, and the additional length of fuel surface uncovered, undoubtedly resulted in additional fuel system damage due to Zircaloy oxidation and embrittlement (followed again by more fragmentation due to themal shock during the recovering of coolant level), although the amount of additional damage is presently unquantifiable. Fuel System Damage Sumary The picture of the core that has emerged is that the core configuration currently consists of a basket-ifke shape of relatively intact assemblies that surround a central region of severely oxidized, and probably fragmented, fuel rods in the upper central part. The fuel

• Based on information received via telecomunication from BW ( Aprf) 3)

5-rods are less damaged in the lower central part of the core. Although the fuel rods in the upper central reg'on may be completely fragmented, the guide tubes, grids, and end plates are believed to be intact thus providing a skeletal structure which supports the remaining portions of the damaged assemblies. Partial flow blockage caused by accumulation of fuel debris is thought to be responsible for continuing elevated The assymetry of the incore thennocouple readings t: ermocouple readings. suggests that a region of the core is more heavily damaged than the average. S e 9 e 8 e em e O O 4 J

) ,4 4/i/IS

• i Ecb.cse,te 4 3

D. \\!. Lcbelle l'& 1 y4. PRELIMINARY SEQLJENCE , g j 0F Evtnis 2 t m j N i (T111-2,3/28/79 INCIDENT) i ~ ) J 1 i The follwing sequence of events for the THI-2 incident of 3/28/79 has been ) { formulated by BW engineers using available plant data. This chronolo y pY j Ms be n constructed from numerous sources and has not been totally i J .ccaffrud. It si y not be pructse in either event occurrtoce or sequence. 'r - ,L Tica, Minutes Event M j E Prior to. The initiating events could have come from nunerous postulated tuttine trip causes. For purpcses of this sequence, they are relatively I l i enferportant. The prime'effect is that it led to a loss of main E feedwater (MFid) booster punps. [ 0 f*ain feedwater pumps are tripped. Almost simultaneously. [ the turbine trip occurs. 9 ( 0.10 Pressurizer pressure increases to the DDV setpoint of h 2270 psig. l 0.15 Secondary side pressure peaks at 1070 psis and is 11mitad I by steam reifer velves. 1~ 0.20 RC pressure trip setpoint rwached (2355 psig at hot leg e i tap) and system pressure peaks at about this value. '. t. g ~ t feedwater pungs (pump discharge pressure are that.aumf1f ary' I I Indicatfons from one turbine driven, two electric) are y ^ i running at this point; however, no level change occurs in i 1 steam generators. ',t O.25 Pressurizer level peaks at 255.'fn'ches (fadicated) and starts e E to decrease with system contraction. h 0.30 quench tank pressure is increasing. ( j g 0.90 Pressurizer level is at a minimum of 158 inches and starts r i i to increase. Hot leg temperature is at a minimum of 5770F W . and starts to increase slowly. i ~ i j l j 1.0 OTSG 1evel indication on the startup range is 10 inches. OT5E pressure holds at about 1025 psig. P. ~ 7 j K by E$FAS on low RC pressure (HPI setpoint = 1600 psig). j; 2.0 ' 013G pressure starts a steady decrease. HP! flow is initiated i I 3.0 The quench tank's increasing pressure levels off at 120 pstg. 1 y. Relief valve setpoint is 150 psig. The hot and cold leg tcrperatures start increasing at a more ' h; '4 75 rapid rate. Analytical simulation indicates that this occurs j when the HP! is turned off. $lte information potes that 3 operator tarwinates HPI fully at 5.1 minutes. J -a. l 1 -w, ,-------..r-n r r ,-----,,-r----,-- --,.+--,-.-,--n,,n-

1 T-c,. .r.. ~ _r. Tire. Minutes Event l 5.0 ' Pressurizer level indicates a slowing and then continues to h increase as the hot leg tes:perature if increasing. $7 l 1 6.0 Pressurizer level indicates s' full pressurizer and the 1 quench tank pressure increases beyond the relief valve Q setpoint of 150 psig. 4 .i?. ?- i RC pr.ssure rt$hes a rainime of 1350 psig with a hot les pi tocareature of SS40F. This indicates hot leg is in i satu. ration condition. j ./ L.i 8.0 Auxiliary feedwater flow is initiated to both OTSG's. This f is indicated by feediate OT5G repressurization to $1025 psia j] ~ and OTSG 1evel change. j 9.0 RC pressure peaks out at 1500 psig and sta'rts to decrease. y Hot leg tagerature peaks out at 597oF. 7 11.0 pressurizer level indictation is restored. It stabilizes B g out at 375 inches at 15 minutes. 16.0 Quench tank pressure drops suddenly, indicating the rupture f disk has blown (setpoint.= 200 1 25 psig).

• s j

18.0 .The decreasing RC pressure stabilizes at 1115. psig. j-22.0; The RC5 tewerature stabilizes at a hot leg of 553'F and a ~ cold leg of 5480F. The tenperature decrease from start of j i auxiliary feedwater to this stabilization represents a b 2000F/hr cooldown. Reactor building pressure is 1.4 psig 3 and increasing. Two feet level is restored in both OT5G's. j The startup level indication shows 0T5E B 1evel increasing 4> 50.0 and 0TSG A level decreasing. pressure in. creases in both 0T5G's. l) 60.0 During the 22-60 minute period, the systee parameters have stabilized in the saturation condition of a prtssure of s1015 k psig. temperature of 4500F. RC flow indication is decreasing E from 60 (initial) to 50 x 106 lb/hr. The reactor building ( pressure is 2.2 psig and increasing. g j 73.0 Two RC pumps are tripped (in Loop B). Reactor coolant flow rete decreases in Loop B. f \\ 78.0 - STSG B pressure drops frors g50 psig to 140 psig in 18 minutes. M.0 T foH ows T AT across the core equals about 5'F. [ hot sat. C 100.0 8oth reaining RC pumps are tripped. 4 l, 114.0-120.0 T and T diverse espidly.' That

• I' '85 D*"

f hat cold l 15 einutes. i i i i 132.0 Site information notes that EE0V relief line was isolated 5 t initially. R8 pressure starts decreasing rmre rapidly. [ t p. ~

l 1 O f m- .N, A.. l Time. Minutes Event 'J 135.0 RCS has depressurized to 670 psig and RCS hot leg 9 ]f temperature is at maximum scale of 620 F. At 620 F, system 0 would have superheating at upper clevations as long cs Q pressurt was below saturation pressure of,1772 psig. 3 RCS shavs rapid re ' pressurization. y J i 650.0 OTSG B level raged up from 5% to 65% in 48 rainutes. / J I J .IC3.0 OTSG B rain steam isolation valves end turbine bypass 1 valves are closed. RCS pressure peaks at 2120 psig. Q 1 103.0-204.0 Regulation by ENOV block valve reduces RCS pressure. 3l.6 HPIcomeson(1600psigsignal). h ~ 204.0 i lL t 216.0a HPI pumo Ic to Loop A turned off. RC pressure decreases stepwise. RB pressure increases steperise. i, e 290.On RB pressure hits 4 psig. Building fan cooler comes on. I l Ci j (4.83 hr) O 6 ) 318.0m RCS pressure increases rapidly from 1250 to 2120 psig in (5.3hr) 35 minutes. The DOV block valve is closed, one HP1 (1A) 1 f is on. a 354.0. OTSG A level is ramped up from 50% to 95% on operating range 3 ~ (5.9hr) in 1 hour and to 1001 in 1.5 hour. 01SG A pressure starts to decmase toward zero. g 3 e i 450.0 The DOV block valve is opened. RCS pressure starts to l-(7.5 hr) decrease (2050 psig to 480 psig in 1 hr. 45 min). [7 519.0 RC system pmssure reaches 600 psig.' core flood tank (8.65 hr) setpoint. 4 j 588.0 RB pressure spike to 28 psig occurs. 3 (9.8hr) .a Loop A twappears on scale, decreases to 525'F in 1/2 hr. 3 630.0 T (10.5 hr) g. hat loop A increases in about 5 minutes from 190'F to 400'F. 4 } 678.0 T aid .(11.3hr) i 750.0 HPI flow increased to 400 gpm. T in Loop A decreases. hat 4 (12.5 hr) ] i 610.0 T W p A dec nases. cold (13.5hr) 3 3 948.0 Pusip 1A is started. -j (15.8 hr)- H i t [ t w. -.-..-.n---,., - - -. ,-,--,-,,-,,-,,-----------,..c-e.---,--- ..--,--n--. + - - - ..,~-n,-..w,- --n- ? l- - ---gn,-

e ' *o. g. l

)

g:. 4 4 l) Event t Tire, Minutes _ Condenser vacuin re-established. [J Thereafter SG A begins steaming to condenser. RCS cooled to approximately 300DF,1000 psi. 4 Letdo>m line ceased to pemit fief and relief valve being L,- esed (estir.ated 14-16 gpm fla). 0 . Sc:ne fuel incore thernacouples reading about 600 F. t R8 Pressure below i psf. High radiation in reactor containment and auxiliary building.? (l I-z 4 A y ) 1 e i E' 5 g,, ? 4 t1 e .e. v. = s B h e* p? e ..e s F, b 's t 4 w l s

t 8 t 1 t ( I d Q d ) s I d h % 4 ' f} g l g 4; ~ ?a $,.- N i i, ( O s / g 4 R-g g g k. ( s 4 s- \\ ,f ,\\

• \\

t o u w' ~~ k \\ k es. 6, a ~- 3 .,.. c s s G ,,.---~-'.. - s tg N W] u s 2 (~ .) a . u s>. 6 g,.> 3 1 t. [p .>{-3 g ~ -p) . gi..- '-<a \\s. f1 3 4 _,,. -. a sn ~. ~ - f,, et 3N .) r ~. ){ -__--w,-,,..v_.f . L. g,__ . -- - - - O I .g g ),_..- e 9 o O 9 G O O 3 fl e

. y~ 7g 6 4 t; p + aarar0 D Core Coolant Conditions At 2 hours after turbine trip the core had become o partly uncovered and remained uncovered for about one hour. During this period activity alarms came on indicating e significant fuel failure. Core was recovered when high pressure injection pump came on. o Two additional periods of extensive core uncovering followed at about 5 and again at 9 through 12 hours af ter turbine trip. e 9 V ^' s e O O 8 1 1

Number of Fuel Rods with Defects o Based on measured coolant activity, all of the fuel rods probably released fission products. Amount of hydrogen released from oxidation of cladding e (metal / water. reactions) also indicates all fuel rods are damaged. Maximum Fuel Temperatures Calculations based on Fission product analysis indicate fuel e terperatures of 1400 to 1600*C. Heat transfer calculatio.ns indicate temperature of about 1100'C. s 9 The melting point of 002 fuel is 2840*C so that core meltdown 'ias not approached, o The absence of Sr and Ba activity.in the coolant confirm the avoidance of fuel meltTH'g. t O e 9 N # $e F GMg a S e$M @ > me e M

I s. d .+ Extent of Fuel Damge W Hydrogen balance calculations indicate from 15 to K% of the Zircaloy s cladding has been oxidized. l i Continued operation of incore instruments indicates that fuel e assembly structural members remain intact. Absence of silver in coolant suggests that control rod materials l e remain in place. Continued low thermocouple readings at periphery suggest that ~ e peripheral fuel assemblies retained much of their original geam. The picture that emerges is that the upper central part of the core is severly oxidized; probably fragmented, and largely) confined to e .i the core region (based on loose parts monitoring data. Partial flow blockage caused by accumulation of fuel debris has r o proba51y occurred and is responsible for elevated thermocouple readings. i t 1 ti I p e f t G e d s "#Oi 88M , W 4 'ly 9 gge -ge-w,w m _..,~ u- ---.,.v-- r,-,

enm, c

..m ,,,.~..,.,.n_..- .... n.

-s m =e S e e A .%e./ 2 g t' g? f / v..a e 8 N s s O h 49- ~ S $? 'y n g V4 / (a}#p ~ u s ( 4 ( ./* /}Y e es A TV,w k' sn v.. f y' ef.. c s 4 g-a o j ;/n i 1. 1

g..

x tj$>s;+ / p'/l e ,1 G rp mv ,d+,s-r b 3-, a f / ~ ,1 %s a c m.. ,' ~ ~,- Q 34 ~~' ' ci

1. 1/

,5 / ~.+', byh f p va

• • rh *'

4;*p s 5 i g, )' .f.- ~. a' o g o k@ 6 k Q a o N ") -

/e,,,

t t v-. ' ~ - _ '~g

~ Y**.g ~ q 1 ?jj, -y. -=.=d<e

== g _.y..... .f. e - e s: tes l Tf T T T T T T

:: ::, {.

t-~ I l} _%'N / N N.' / y-m /j-s D F~~ a T. ynn gw

q-,

F 400 (9 00* a 1ls rr.1 sur ~- r~~ ! O. W, O 7 g%,,-~.m h' m-i e= =. I .canam ' s h 9 n 14 N 'ld' 5 r ( 1 'wP'ca:t];\\ ,t y,,, 3 kl .+ N FUEL DEBRIS--

j 3

. 7s e$ 'T,[,7 E \\ g, s o J OUTER ASSEMBLIES-- dp d \\, \\ l Y \\\\ \\ i;

N "'= "'=

FUEL CLADDING i; 'y PERFORATED BUT j; k- \\ 'k \\A s s.-au, - 2 L' w xi Ax#1 t 'N RODS. INTACT j = i 1 k P ALL GUIDE TUBES ' } i :: AND GRIDS INTACT I 3 p u ,,-r 3: / W f i V-

i

_i / b. y} l'yL. ~ t".a% n,,, s <.,. us.n. i REATOR VE55El fI!:1[Rl!Als asn<w

UNITED 3TATES [ g NUCLEAR REGULATORY COMMISSION g j W ASHING TON, D. C. 2C?55 e \\, / APR 131979 MEMORANDUM FOR: D. Ross, Deputy Director Division of Project Management Office of Nuclear Reactor Regulation FROM: W. V. Johnston, Chief Fuel Behavior Research Branch Division of Reactor Safety Research Office of Nuclear Regulatory Research

SUBJECT:

FUEL EXPERTS MEETING ON CONDITION OF THE TMI CORE A meeting of nuclear fuel experts was held on April 12 to update the estimates of the damage to the TMI core and to consider its effect on the desirability of moving to natural convection cooling of the core. The Experts Group consisted of the following persons: J. S. Tulenko, B&W; R. DeMars, B&W; T. Kassner, ANL; R. A. Proebst'e, GE; K. A. Jordan,}{; R. Duncan, CE; T. Fernandez, EPRI; T. Buhl, NRC; R. Meyer, NRC; W. Johnston, NRC, Chairman. Additional attendees included L. Rubenstein, NRC; C. Berlinger, NRC; M. Tokar, NRC; R. Majors, ACRS Staff and T. Mott, TEC. Sumary The group concluded that although the core is badly damaged, essentially all of the fuel has remained in the core and that the overall packing density of the settled portion is not expected to exceed 70%. Therefore, shutting off the RC pumps should not seriously threaten further damage to the reactor. It was further concluded that the thermocouples (Tc's) located in the upper end fittings are the most important indicator of core condition during transi on to natural convection cooling. If feasible, the addition of a spectrometer to monitor the activity of the loop coolant for new fis on products released the transition to natural convection cooling will provide'an indepen nt plert to possible difficulties. W g Two questions were considered: Is a pump trip likely to lead to an unsafe condition? and What signals will indicate undesirable conditions in the core? An 4 / I n '- -1 m An f l f[.A Th N ).

O. Ross APR 131979 Suninary of Core Status Suninaries of amount of <famage to the core based upon measurements or calculations of fission gas release, hydrogen produced by zircaloy oxidation, coolant analysis, coolant boil-off rates, incore and excore instrumentation were presented by J. Tulenko and R. Meyer. A relatively large pressure drop across the core is inferred by TH calculations. If the pressure drop is real, blockage must also exist in the peripheral assemblies (perhaps by ballooning). The shift in location and magnitude of the high reading core Tc's following the pump trip on April 6 was believed to indicate either a change in the core flow path through more heavily ^ damaged sections of the core to a redistribution of debris surrounding some of the thermocouple beads. An alternate explanation for the change in Tc temperature distribution patterns was presented by T. Mott of TEC. He suggested that the Tc temperature difference nay be due to non uniform flow distributions caused by operation of a single pump rather than non uniformity within the damaged region of the core.. Due to this non uniform flow distribution portions of the core may already be experiencing similar cooling to that expected during natural convection. Mott estimates smaller core pressure drop and suggests the B&W estimates may include substantial external pressure drops. The group visualizes the core as consisting of a heavily damaged region resembling an inverted bell extending across nearly the full width of the top of the core and reaching down about five - six feet into the core at the center and a less damaged remainder of the core. In the heavily damaged region,100% oxidation of the zircaloy and less of a regular geometry is expected. The guide tubes and poison rods are damaged similarly to the cladding. Spacer grids should be located.at or near their original locations. The important coolability conclusions are that although some settling may have taken place, the overall packing density of the settled portion is not expected to be greater than 70% and that 85% to 98% of the fuel and cladding from this region is believed to have remained in the " core" region including the upper end fitting. The remainder of the core is less damaged although considerably oxidized. The original flow geometry is probably retained although the rods may be twisted or warped and broken in a few places and the spacer grids may have collected some loose debris. The above conditions should not preclude satisfactory achievement of natural convection flows. What should be monitored to determine undesirable changes in the core during the transition to natural uenwetico t mling? The temperature distribution of the ccre exit t'mixmies are 'Ac rest important D '

• D. Ross NR 131979 condition monitoring signals. The group believes that all exit Tc's should be continuously tracked and recorded. B&W suggested the following criteria for remaining in natural convection cooling: No more than 2 Tc's above 800*F and at least 10 Tc's below Tsat. There were some reservations among the group about allowing so many Tc's to read above Tsat (as many as 39) and radiolysis was an expressed concern. There was a lot of discussion but no concensus on how many interior Tc's should be permitted to exceed Tsat. Tc's in peripheral assenblies should not exceed Tsat.

The following table surmarizes the available instrumentation and its possible application to monitoring core condition. Detector Event-Core Overheating Criteria Basis

1. Exit Tc's Limit no. in film boiling Not to exceed previous Limit no. above 800*F core damage reverse i

procedure.

2. RTD Maintain positive AT across No flow reversal Hot leg core.

permitted. Cold leg i

3. Ion e.iambers Voio fonnation Ambiguous signal since 6 and N If +, record for future some local superheat interpretation, watch Tc's.

may be permitted.

4. Noise detection If + indicates bubbles in Same as above.

core or loop, check Tc's, SG. /

5. System Pressure If increasing system effects Not direct indication Branch shwld review this, of core condition, but for gas bubble formation detection.

6. Pressurizer Level Same as above.

Same as above. ~ Additional Detection - Feasibility needs to be established. Overheated core alert spectroscopy Increasing activity of Xe, 12 of coolant via for major error in sampling line procedure. Core criticality and H O analysis on Boron 0 ' H2 2 2 ITne mon'ltor chemistry control radiolysis and H content control.2 W. V. Johnstoni)ChiefW4%V LL '

• b- - - - ' ~ ~

vs i

1 U*.lTED ST,*.T;; NUCLEAR REGULATOF.Y CO.'.r.*lSslO:. i j.' j wAsMir GTo.'. D c. 20.t: (.,3]T;. f .... f ~ April 13, 1979 ATTACHMENTF. NRC ADVISORY -- THERMOCOUPLE READINGS 1 We recomend that all incore thermocouples be read in a continuous manner o with provisions for rapid retrieval and permanent storage. It is clear e that a continuous reading of thermocouples will be needed during the transition to natural circulation. It is also clear that temperature trends, which were not recorded during the pump changeover on April 6, would have given additional clues to core behavior in natural circulation. Since future flow transients cannot be ruled out and since the transition to natural circulation could occur involuntarily, we recommend that the continuous recording of all thermocouple readings be initiated as soon as possible. Y Ralph 0. Meyer, Leader Reactor Fuels Section - Core Performance Branch a f B l i i ..c,_. _ -. - - - -..,,, -. - - -,.,.}}