ML20038A631

Forwards Addl Info Requested in NRC Re Reactor Vessel Level Indication Sys Being Installed at Facility

ML20038A631
Person / Time
Site:	Trojan File:Portland General Electric icon.png
Issue date:	11/06/1981
From:	Withers B PORTLAND GENERAL ELECTRIC CO.
To:	Clark R Office of Nuclear Reactor Regulation
References
TAC-45175, NUDOCS 8111160073
Download: ML20038A631 (18)

Text

.

QW l

A

2,s,\\ s i m

/

w,

~

Portland Genem! Electric Company fy

v, ~ E '7,. '

l lL

ll7,

)

<~

}^

2,, f

-f /

9 G

.s-(-

y~

n,

'I

'N

--(_

y November 6',

1981

.i

.i __

s Trojan Nuclear Plant Docke t 50-344 License NPF-1 Director of Nuclear Reactor Regulation ATTN:

Mr. Robert A.

Clark, Chief Operating Reactors Branch No. 3 Division of Licensing U.

S. Nuclear Regulatory Commission Washington, DC 20555

Dear Mr. Clark:

Yaur letter of September 29, 1981 requested that Portland General Electric Company (PCE) provide additional information concerning the Reactor Vessel Level Indication System being installed in Trojan. Our response to your request is attached. The Westinghouse submittal designated "Microprocesser System" applies to the Trojan plant.

Sincerely, Bart D. Withers Vice President Nuclear Attachments c:

Mr. Lynn Frank, Director State of Oregon Department of Energy OO I I$

8111160073 811106 DR ADOCK 05000344 PDR

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

j t

i Trojan Nuclear Plant R. A. Clark Docket 50-344 November 6, 1981 License NPF-1 i

l.

f 1

i l

i I

i l

)

i 1

l l

t I

r 1

A 1

l i

I l

Attachment A l

i i

4 l

2 l

j y

PGE RESPONSE TO NRC REQUEST FOR ADDITIONAL INFORMATION l,

l ON TROJAN'S REACTOR VESSEL LEVEL INDICATION SYSTEM (RVLIS) i i

{

4 1

i i

\\

t j

i i

I l

i l

l l

i i

n l

2 i

I r

i I

k I

1 l

i h

I l

I k

I i

i

(

-_..,m.n,an-

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

Item 1 Justify that the single upper head penetration meets the single failure requirement of 2'"DEG-0737 and show that it does not negate the redundancy of the two instrument trains.

Response

Redundancy is not compromised by having a shared tap since it is not conceivable that the tap will fail either from plugging or breaking.

Freedom f rom plugging is enhanced by, (1) use of stainless steel connec-tions which preclude corrosion products and, (2) absence of mechanisms (such as flow) for concentrating boric acid.

It is inconceivable that the tap will break because it is in a protected area.

In other cases where sharing of a tap occurs in the RCS, we know of no prior experience reporting deleterious malfunctions of the shared tap.

Also, even if the shared tap does fail, it should be recognized that RVLIS is not a protec-tion system initiating automatic action, but a monitoring system.

t Item 2 i

Describe the location of the level system displays in the control room with respeci

'4 other Plant instrument displays related to ICC monitoring; in particular, the saturation meter display and the core exit the rmocouple lisplay.

Response

The RVLIS displays will be located on Panel C-09 in the Trojan control room. They will be located in close proximity to the Subcooling Margin j

Monitors which are on the same panel. Margin to saturation and selected core exit thermocouple status and readings are displayed on Trojan's Subcooling Margin Monitors. All core exit thermecouple readings are displayed on the Plant computer (Panel C-01) on demand.

1 Item 3 I

I Describc the provisions and procedures for on-line verification, calibra-tion and maintenance.

I

Response

l In general, the system electronics are verified, maintained and calibrated on-line by placing one of the redundant trains into a test and calibrate mode while leaving the other train in operat.lon to monitor for inadequate i

core cooling. The RVLIS requires the normal maintenance given to other l

control and protection systems within the Plant.

A general verification is performed bef ore shipment using non plant-specific data. After installation, Plant personnel can verify operation of the system by disconnecting the sensors at the RVLIS electronics, providing an artificial input, and observing the response of the system on the front panel and remote display.

l l

l t

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

____ - - _- ~

4 l

1 2

i l

On-line calibration of th< system is made possible by the controls available on the main processing unit. The calibration consists of I

entering constants into the non-volatile Random Access Memory (RAM),

along with adjusting the potentiometers on the analog to digital con-

[

]

version cards.

The initial calibration is performed when the system is installed.

Subsequent calibrations can be performed to maintain system accuracy.

Software programs are provided so that the front pcncl controls and display can be used to perform a functional test, j

serial data link tests, calibration tests and deadman timer tests.

Item 4 I

)

Describe the diagnostic techniques and criteria to be used to identify malfunctioning components.

Response

i j

The microprocessor-based RVLIS performs internal diagnostic checks of the non-volatile RAM, non-volatile Programmable Read Only demory (PROM) and i

other microprocessor components.

No operator interface is required for these internal checks which are performed in each cycle.

A " deadman" 3

i circuit is provided to detect microprocessor failure. This circuit will indicate a processor problem on the front panel of the unit and automatic-e i

ally reset the Central Processing Unit (CPU) to restart the microprocessor.

l

]

The remote display unit of the RVLIS indicates the status of the input sensors.

If any sensor is out of range or disabled, a symbol will follow j

the af fected level reading on the summary display page.

In addition, software programs are provided so that the front panel controls and display can be used to perform a functional test, serial data link tests, calibration tests and deadman timer test.

Item 5 Estimate the inservice life under conditions of normal Plant operations i

and describe the methods used to make the estimate, and the data and sources used.

Response

1 The Westinghouse RVLIS is designed for an inservice life of 40 years.

The inservice life of the RVLIS microptocessor-based electronics is dependent upon proper maintenar.ce, including the replacement of individual component parts when necessary-Based on the assumption of normal condi-tions and proper maintenance at the components, the only limitation to 4

the inservice life will be the availability of replacement parts.

It is estimated that in 20 years, some of the components may be technically l

obsolete and no longer 'roduced.

Consequently, the cards may have to be modified in the future so accommodate the current technology.

Some of the equipment is similar to equipment installed in present Westinghouse plants that have been operating for 10-15 years.

I i

1

3 The following valves have been supplied by Westinghouse for the RVLIS for the Trojan Plant:

W Design Code W Valve ID Qty Manufacturer Specification Applicability 3/4 T 78 4

Rockwell G-952855; Rev 0 ASME B&PV Class I 1

1/4 X 281 10 Autoclave Engineers G-955230; Rev 2 N&S 1

l 1/4 N 28I*

6 Autoclave Engineers G-955230; Rev 2 N&S l

• Shut-off valve which is part of the transmitter access assembly.

l l

The 3/4 T78 valve is a stainless-steel, manually-operated globe valve i

l whose basic function is to isolate the flow of fluid. The valve is l

designed for a cycle life of 4000 cycles over the 40-year design life, which satisfies the normal Plant operating requirements established in the above-referenced specification. The valve is a hermetically-scaled valve designed to be maintenance free with no consumable materials making a pressure boundary seal.

The instrumentation vrives (W Valve ID's 1/4 X 28I and 1/4 N281) are stainless-steel, manually-operated valves.

They are designed to meet the requirements of the above-referenced specification which calls for zero leakage (environmentally and across the seats), minimal fluid displacement during stroke and a 1000-cycle life. For normal Plant operating conditions, the metallic parts are designed for a 40-year service life.

Item 6 Explain how the value of the system accuracy (given as +/-6%) was derived.

How were the uncertainties frem the individual components of the system combined? What were the random and systematic errors assumed for each component? What were the sources of these estimates?

Response

The system accuracy of +6% water level was a target value established during the conceptual design. This value was related to the dimensions of the reactor vessel (12% from nozzles to top of core) and core (30%),

and the usefulness of the measurement during an accident.

Subsequent analyses have established a system accuracy based on the uncertainties introduced by each component in the instrument system. The individual uncertainties, resulting from random effects, were combined statistically to obtain the overall instrument system accuracy.

Some of the individual uncertainties vary with conditions such as system pressure. The follow-ing table identifies the individual uncertainties for the narrow range measurement while at a system pressure of 1200 psia.

4 Uncertainty coaponent and Uncertainty Definition

% Level a.

Differential pressure transmitter 1,2.1 calibration and drif t alowance

(+1.3% of span) multiplied by the ratio of ambient to operating water density.

b.

Differential pressure transmitter 1,0. 7 allowance for change in calibration due to ambient temperature change (10.5% of span for +50*F) multiplied by the density ratio.

c.

Differential pressure transmitter 10.34 allowance for change in calibration due to change in system pressure (1,0.2% of span per 1000 psi change) multiplied by the density ratio.

d.

D*fferential pressure transmitter

+0.7 allowance foi change in calibration due to exposure to long-term overrange (10.5% of span) multiplied by the density ratio.

e.

Reference leg temperature instrument

+0.64 (RTD) uncertainty of +5'F and/or

~

~

allowance of +5*F for the difference between the measurement and the true average temperature of the reference leg, applied to each vertical section of the reference leg where a measurement is made.

Stated uncertainty is based on a maximum Containment tempe ra ture of 420*F, and a typical reference leg installation.

t f.

Reactor coolant density based on

+2.3 l

auctioneering for highest water density

~

obtained from hot leg temperature (1,6*F) or system pressure (+60 psi).

Magnitude of uncertainty varies with system pressure and water level, with largest uncertainty occurring when the reactor vessel is full.

I i

I

5 Uncertainty Component and Uncertainty Definition

% Level g.

Sensor and hydraulic isolator bellows 11.46 displacements due to system pressure changes or reference leg temperature changes will introduce minor errors in the level measurement due to the small volumes and small bellows spring constants. The changes, such as pressure or temperature, tend to cancel, ie, the bellows associated with each measurement move in the same direction. Maximum expected error due to differences in capillary line volume and local tempera-tures is equivalent to a level change of about 5 inches, multiplied by the density ratio.

h.

Density function generator output mis-10.50 match with ASHE Steam Tables limited to a maximum or:

1.

Electronics system calibration, overall 11.0 uncertainty limited to less than:

j. Control board indicator resolution; 10.5 microprocessor digital readout to nearest percent of level span.

The statistical combination (square root of the sum af the squares) of the individual uncertainties described above results in an overall system instrumentation uncertainty of 13.9% of the level span.

For the narrow range indication of approximately 40 feet, this corresponds to an uncer-tainty of 11.5 feet at a system pressure of 1200 psis.

Examples of the uncertainty at other system pressures are:

Uncertainty = 13.6% at 400 psia Uncertainty - 14.2% at 2000 psia Uncertainty = 14.25% at 2250 psia.

Item 7 Assume a range of sizes for "small break" LOCAs. What are the relative times available for each size break for the operator to initiate action to recover the Plant from the accident and prevent damage to tha core?

What is the dividing line between a "small break" and a "large break"?

Response

Inadequate core cooling (ICC) was defined in WCAP-9754, "Indequate Core Cooling Studies of Scenario Uith Feedwater Available Using the NOTRUMP Computer Code", as a high temperacare condition in the core, such that the operator is required to take action to cool the core

l 6

before significant damage occurs.

3nring the design basis small LOCA, j

the operator is not required to take any action to recover the Plant

'q other than to verify the operable status of the safeguards equipment, trip the reactor coolant pumps (RCPs) when the primary side pressure has l

decreased to a specific point, and initiate cold and hot Icg recircula-tion procedures as required.

In the design basis small LOCA, a period of cladding heatup may occur prior to atiomatic core recovery by the safe-guards equipment. The heatup period is dependent upon the break size and Emergency Core Cooling System (ECCS) performance.

i An ICC condition may arise if there is a failure of the safeguards equipment beyond the design basis.

In that case, adequate instrumenta-tion exists in the Trojan Plant to diagnose the onset of ICC and to determine the effectiveness of the mitigation actions takea. The instruments which may be used to determine the adequacy of core tooling are a subcooling meter, core exit thermocouples (T/Cs), or the RVLIS.

For a LOCA of an equivalent size equal to approximately 6 inches or less, an ICC condition can only occur if two or more failures occur in the ECCS. As indicated in WCAP-9754, an ICC condition can be calculated by hypothesizing the failure of all high-head safety injection (HPSI) for LOCAs of approximately 1 inch in size.

For a 4-inch equivalent size LOCA, one can hypothesize an ICC condition by assuming the failure of all HPSI as well as the failure of the passive accumulator system.

For LOCAs of sizes 6 inches or less, the approach to ICC is unambiguous to the reactor operators. The first indication of a possible ICC situa-tion is the indication that some of the ECCS pumps have failed to start or are not delivering flow. The second indication of a possible ICC j

situation is the occurrence of a saturation condition in the primary coolant system as indicated on the subcooling monitor. Shortly after the i

d second indication, the RVLIS would start to indicate the presence of steam voids in the vessel. At some point in time the RVLIS will indicate a collapsed liquid level below the top of the core. The core exit T/Cs l

will begin to indicate superheated steam conditions.

If appropriate, the RVLIS and core exit T/C behavior will provide unambiguous indications to l

the operator to follow the ICC mitigation procedure.

j j

WCAP-9754 indicates that the selected core exit T/Cs will read 1200*F at approximately 11,000 seconds after the initiation of a 1-inch LOCA l

with the loss of all HPSI.

In the event of a 4-inch LOCA, the T/Ce will indicate 1200"F at about 1350 seconds. Westinghouce Owners Group is developing Generic Westinghouse Emergency Operating Procedures (EOPs) which will aid in operator recognition of earlier inJication of a possible ICC situation under these conditions. Recovery procedures to depressurize l

the primary system below the low pressure safety injection shutoff head may be followed. These procedures include correction of the HPSI failure, opening steam dump or opening pressurizer PORVs. The RCPs may be restarted to provide additional steam cooling flow.

Large break LOCAs consist of LOCAs in which the fluid behavior is inertfally dominated.

Small break LOCAs, on the other hand, have the fluid behavior dominated by gravitational effects.

For LOCAs which i

i l

__m.____

i 1

i 7

are significantly larger than an equivalent 6-inch break, the ECCS j

has the maximum potential for flow delivery since the primary coolant j

system is at low pressure.

No early manual action is useful in recovering from ICC. Analyses for 1

I LOCAs in this range indicate ambiguous behavior of the core exit T/Cs and RVLIS early in the accident due to dynamic blowdown effects.

I This behavior is temporary, and the core exit T/Cs and the RVLIS will indicate the progress being made by the ECCS in recovering the core.

l When the core exit T/Cs and RVLIS may be temporarily providing i

ambiguous indications, no manual action is needed or useful.

Later I

in the accident when manual action may be useful, the core exit T/Cs and RVLIS will provide an unambiguous indication of ICC if it exists.

This unambiguous indication may be present as early as 30 seconds after the initiation of the LOCA for a double ended guillotine rupture of a main coolant pipe.

It follows from the above discussion that, for ICC considerations, a reasonable definition of large breaks defines them as breaks that are significantly larger than an equivalent 6-inch break. All other breaks q

are small breaks.

4 Item 8 1

l Describe how the system response time was estimated. Explain how the i

response times of the various components (differential pressure trans-1 ducers, connecting lines and isolators) affect the response time.

Response

l The microprocessor reads all the inputs every 5 seconds and updates the q

digital display and analog outputs within 4 seconds after the inputs are i

read. Thus, a worst case time from analog input change to display and analog output change is 9 seconds. Any analog delays due to the front-end electronics, sensor electronics, sensor mechanics, impulse lines, hydraulic j

isolators, etc., have 5 seconds to settle out.

Thus, analog delays only l

add to the 9-yecond worst case response time if they are longer than 5 seconds. The front-end electronics of the microprocessor system have a i

time constant less than 0.5 seconds, and the total analog delays due to the sensor electronics, mechanics, impulse lines and hydraulic isolators 4

are less than 3 seconds. Therefore, the worst case response time is

,l 9 seconds for the system.

Item 9 There are indications that the TMI-2 core may be up to 95% blocked.

Estimate the effect of partial blockage in the core on the differential pressure measurements for a range of values from 0 to 95% blockage.

Response

a Blockage in the core will increase the frictional pressure drop and increase the total differential pressure across the vessel. This will be reflected as a higher RVLIS indication. The increase in reactor i

f A

d

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

1 I

g i

1 vessel level indication will be most significant under forced flow conditions when the reactor coolant pumps are operating.

^

t

]

In order for significant blockage to be present, the core would have to have been uncovered for a prolonged period of time.

A low reactor vessel level would have been indicated during this time.

If the RCPs i

had been operating throughout the transient, there would have been sufficient cooling to prevent significant core damage. Therefore, for significant blockage to exist during pump operation, the operator t

would have restarted the pumps after an ICC condition had existed for a period of time.

Based on the history of the transient, the operator would know that the RVLIS would read higher than expected. Although the RVLIS would read high, it would still follow the trend in vessel inventory. The operator would be able to me..itor the recovery with i

the RVLIS.

Under natural circulation conditions, the impact of core blockage is not expected to be large.

Although the RVLIS indication will read slightly higher than normal, the RVLIS will still tread with the vessel inventory and provide useful information for monitoring the recovery from ICC.

ICC will have been indicated at an earlier time, before a signifi-cant amount of core blockage has occurred. The operator will know that the RVLIS could read slightly high, based on the history of the transient.

Item 10 Describe the effects of reverse flows within the reactor vessel on the indicated level.

i

Response

l Reverse flows in the vessel will tend to decrease the D/P across the vessel, which would cause the RVLIS to indicate a lower collapsed level than actually exists.

The low indication would not cause the operator to take unnecessary actions.

It is important to note that large reverse flows are not expected to occur for breaks smaller than 6 inches in diameter during the time that the core is uncovered. Large reverse flow rates may occur early in the blowdown transient for large a

diameter breaks but, as is discussed in the response to Item 7, it is not necessary to use the RVLIS as a basis for operator action for breaks in this range.

Item 11 What is the experience, if any, of maintaining D/P cells at 300% over-i ranga for long periods of time?

j

Response

i Experience in overranging of D/P instruments has been obtained in i

previous applications of D/P capsules similar to those used in RVLIS.

In Dual Range Flow (D/P) applications, the " Low Flow" transmitter (and/or gages) are overranged to 300% or greater by r rmal flow rates, yet they provide reliable metering when required for startup.

l 1

,--y----,-

w e.

.-e-,9wwe,y._

,w e

-~.

m a__,,

- - - - =,,, -

m _ _ _ _ _ _

l l

9 i

Also, test data exists on the basic transmitter design showing about 0.5% effect on calibration with 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> exposure to 3000 psig overrange.

i f

All units are similarly exposed to this overrange for 5 minutes in both directions as a part of factory testing.

There have been instances involving accidental overrange of these instruments (including RVLIS) as the result of leakage or operator l

errors, where full line pressure overranges hava occurred for up to several weeks with minimal effect on instrument accuracy.

Based upon this experience and test data, we expect to prove statis-tically that reliable measurements can be made by the selected overranged I

instrument designs used for RVLIS.

Online calibration capability is provided if needed to support gathcring of statistical data.

Item 12 4'

Five coaditions were identified which could cause the D/P level system to give ambiguous indications. Discuss the nature of the ambiguities for:

1. accumulator injection into a highly voided downcomer, 2. when the upper head behaves as a pressurizer, 3. upper plenum injection, and

4. periods of void redistribution.

Response

1.

When the downcomer is highly voided and the accumulators inject, the cold accumulator water condenses some of the steam in the downcomer which causes a local depressurization. The local depressurization will lower the pressure at the bottom of the vessel, which will lower the D/P across the vessel, causing an apparent decrease in level indication. The lower pressure in the downcomer also causes the mixture in the core to flow to the lower plenum, causing an actual decrease in level. The period of time when the RVLIS indication is lower than the actual collapsed liquid level will be brief.

f I

An example of when this phenomenon may occur is when the reactor j

coolant pumps are running for a long period of time in a small-l break transient.

Af ter the RCS loops have drained and the pumps are circulating mostly steam, the level in the downcomer will be j

depressed. A large volume of steam will be present in the down-comer, above the low mixture level, which allows a large amount of condensatlon to occur.

For most small-break transients, the reactor coolant pumps will be tripped early in the transient and the downcomer mixture level will remain high, even in cases where ICC j

occurs. When the downcomer level is high, the effect of accumulator injection on the RVLIS indication will be minor.

2.

When the upper head begins to drain, the pressure in the upper head decreases at a slower rate than the pressure in the rest of the RCS.

This is due to the upper head region behaving much like the pressurizer. The higher resistance across the upper support plate relative to the rest of the RCS prevents the upper head from draining quickly. This situation only exists until the mixture level in the upper head falls below the top of the guide i

10 tubes. At this time, steam is allowed to flow from the upper plenum to the upper head, and the pressure equilibrates.

While the upper head is behaving like a pressurizer, the vessel differen-tial pressure is reduced and the RVLIS indicates a lower than actual collapsed liquid level.

This phenomenon is discussed in the summary report on the RVLIS*

relative to the 3-inch cold leg break. Since that time, the upper head modeling has been investigated in more detail.

It was found that the modeling used at that time assumed a flow resistance i

that was too high for the guide tubes. Subsequent analyses have shown that the pressurizer effect has less impact on the vessel l

D/P than was originally shown. There is very little impact on the results after the level drains below the top of the guide tubes.

The pressurizer effect is still believed to exist, and it becomes more significant as break size increases. The interval of time when the upper head behaves like a pressurizer is brief, and the RVLIS will resume trending with the vetael level after the top of the guide tubes uncover. The reducei.

uIS indication will not cause the operator to take any unneces ary action, even if a level below the top of the core is iudicated since the core exit T/Cs may be used as a supplemental indication of the approach to ICC.

3.

Upper plenum injection is not utilized in Trojan.

4.

During the time when the distribution cf voids in the vessel is changing rapidly, thrae can be a large change in the two phase mixture level with very little change in collapsed mixture level.

The use of the RVLIS is still valid for this situation, however.

The only event that has been identified which could cause a large void redistribution is when the reactor coolant pumps are tripped when the vessel mixture is highly voided. Af ter the pump perform-ance has degraded enough that the flow pressure drop contribution to the vessel differential pressure is small, the change in RVLIS indication will be small when the pumps are tripped. As discussed in the summary report, the approach to ICC would be indicated when the wide range indication reads 33%.

If the pumps were tripped at this time, the core would still be covered. The operator would know that the core may uncover if the pumps were tripped with a wide range indication lower than 33%. Prior to pump trip, the core will remain adequately cooled due to forced circulation of the mixture.

When the pumps trip, the two phase level may equilibrate at a level below the top of the core. The narrow range indication will provide an indication of core cooling capability at this time.

Item 13 No recommendations are made as to the uncertainties of the pressure or temperature transducers to be used, but the choice appears to be left to the owner or AE.

What is the upper limit of uncertainties that should be t

• Westinghouse Electric Corporation, " Westinghouse Reactor Vessel Level Instrumentation System for Monitoring Inadequate Core Cooling",

j December 1980.

1

11 allowed? Describe the effect of these uncertainties on the measurement of level. What would be the effect on the level measurement should these uncertainties be exceeded?

Response

The reactor coolant pressure and temperature signals originate from the existing wide range pressure and hot leg RTDs installed in the Plant. The uncertainty goals for these instruments are understood.

As indicated in the response to Item 6, the pressure uncertainty is 160 psi and the temperature uncertainty is 16*F, resulting in a maximum level uncertainty contribution of 12.3% when the vessel is full. This uncer-tainty is smaller when the level is at the elevation of the reactor core.

This contribution to the total uncertainty would increase roughly in proportion to an increase in the pressure or temperature measurement uncertainty.

Item 14 Only single RTD sensors on each vertical run are indicated to determine the temperatures of the impulse lines.

Where are they to be located?

What are the expected temperature gradients along each line under normal operating conditions and under a design basis accident? What is the worst case error that could result from only determining the temperature at a single point on each line?

Response

RTD sensors are installed on every independently run vertical section of impulse line, to provide a measurement for density compensation of the reference leg.

If the vertical section of impulse line runs through two compartments separated by a solid floor, an RTD sensor is installed in each compartment.

The RTD is installed at the midpoint of each vertical section, based on the assumption that the temperature in the compartment is uniform, or that the temperature distribution is linear in the vicinity of the impulse line. As stated !n the response to Item 6, an allowance for the true average impulse line temperature to dif fer from the RTD measure-ment by 5 F is included in the measurement uncertainty analysis. This allowance permits a significant deviation from a linear gradient, eg, 20% of the impulse line could be up to 25'F different from a linear gradient without exceeding the allowance.

During normal operation, forced circulation from cooling fans is expected to maintain compartment temperatures reasonably uniform. During the LOCA, turbulence within a compartment due to release of steam would also produce a reasonably uniform temperature. Note that the impulse lines are protected from direct jet impingement by metal instrument tubing channels.

Item 15 What is the source of the tables or relationships used to calculate density corrections for the level system?

=

4 12

Response

j The relationships used in the microprocessor-based RVLIS to calculate density corrections are based on the the ASME steam tables dated 1967.

These relationships are implemented in the system using two fourth order polynomials, end to end, fit to approximate the steam tables.

I Iter 16 i

The microprocessor nystem is stated to display the status of the sensor input. Describe how this is indicated and what thie actually means with l

respect to the status of the sensor itself and the reliability of the I

indication.

Response

1 The remote display unit of RVLIS (ndicates the status of the input 4

sensors.

If any sensors are out-of-range, regardless of the reason, a symbol follows the affected level reading on the summary display page.

The particular sensor that is out-of-range is identified at the bottom of the summary display page. Due to the redundant sensors and trains, it is possible for the operator to disable some of the sensors without af fectir; the system reliability. The display indicates which level readings are affected. The disabled sensors are also displayed at the bottom of the summary page. A separate sensor status page can be displayed showing all sensors which are disabled or out-of-range and their affected level readings.

Item 17 Describe the provisions for preventing the draining of either the upper head or hot leg impulse lines during an accident. What would be the resultant errors in the level indications should such draining occur?

Response

l The layout of the impulse lines fron t he upper head ind hot leg are i

arranged to prevent or minimize the impact of drainage during an accident.

In general, however, the water in the impulse lines will be cooler than the water in the reactor or hot leg, and there will be sufficient sub-cooling overpressure in the lines so that very little, if any, of the water would flash to steam during a depressurization or Containment heatup. Heat conduction along the small diameter piping and tubing would be insufficient to result in flashing in a significant length of piping.

The connection to the upper head f rom a spare control rod drive mechanism port drops or slopes down from the highest point of the vessel connection to the sensor bellows mounted on the refueling canal wall, so water would be retained in this piping. Draining of the vertical section immediately j

above the reactor vessel has no ef fect on the level measurement, since this section is included

c. the operating range of the instrument.

1 Draining of the horizontal portion of vessel vent piping above the vessel also has no effect on the measurement since no elevation head is involved.

13 The connection from the hot leg to the sensor bellows is a horizontal run of tubing, so draining of this tubing has no effect on the measurement

[

since no elevation head is involved.

The majority of the impulse line 1er.gth is in capillary tubice scsled at both ends with a bellows (sensor bellows at the reactor end, hydraulic isolator at the Containment penetration end), so water would be retained in this system at all times. The water will be pressurized by reactor pressure, and since the reactor temperature will be higher than Contain-ment temperature during an accident, the water in the sealed capillary lines cannot flash.

I'em 18 Discuss the effect on the level measurement of the release of dissolved, noncondensible gases in the impulse lines in the event of a depressurization.

Response

The majority of the impulse lines a;e sealed capillary tubes vacuum-filled with demineralized, deaerated water. The lines contain no noncondensible gases and are not in a radiation environment sufficient for the disassociation of water.

The short runs of impulse line connected directly to the primary system will behave as described in the response to Item 17.

There would be no error due to gases in the hot leg line since the line is horizontal.

f Since there is no mechanism for concentration of gases at the top of che reactor vessel during normal operation, the connection to the top of the vessel would contain, at most, the normal quantity of dissolved gases in the coolant. The subcooling pressure during an accident would maintain this quantity of gas in solution.

I Item 19 In some tests at Semi-scale, voiding was observed in the core while the upper head was still filled with water. Discuss the possiblility of ecoling the core exit T/Cs by water draining down out of the upper head during or after core voiding with a solid upper head.

Response

Realistically, an indication of an ICC condition would not occur until L

the primary coolant system had drained sufficiently for the reactor vessel mixture 'evel to fall below the top of the core.

Westinghouse has performnd analyses which indicate that the upper head will drain below the top of the guide tubes before ICC conditions exist. The guide tubes are the only flow path from the upper head to the upper plenum.

In WCAP-9754, " Inadequate Core Cooling Studies of Scenarios With Feedwater Available, Using the NOTRUMP Computer Code", it was found that inadequate core cooling situations would not result for LOCAs of an equivalent size i

or equal to approximately 6 inches or less without two or more failutc=

in the ECCS.

In both specific scenarios examined in WCAP-9754, a 1-inch "nd 4-inch small LOCA, the upper head and upper plenum had completely drni1ed before the onset of an ICC condition.

14 In the Trojan Plant, the core exit T/Cs protrude slightly from the bottom of the support columns.

In this location, they measure the temperature of the fluid Icaving the core region through the flow passages in the upper core plate.

Flow from the upper head must enter the upper plenum via the guide tube before being able to enter the upper core plate flow passages.

In addition, the LOCA blowdown depressuriza-tion behavior must be such that there is a flow reversal for the core exit T/Cs to detect the upper head fluid temperature.

The upper head fluid is expected to mix with the upper plenum fluid as it drains from the upper head.

The potential for core exit T/C cooling from colder upper head fluid, while the core has an appreciable void fraction, is not viewed as a potential problem for the detection of an inadequate core cooling situation. Although some Semi-scale testa indicated core voiding while the upper head was liquid solid, that does not imply that the core exit T/Cs would give an ambiguous indication of ICC.

Calcula-tions for a Westinghouse PWR and consideration of the core exit T/C design would not result in ambi - as ICC indications.

t Item 20 Describe the behavior of the level measurement system when the upper taad is full, but the lower vessel is not.

Response

During the course of a LOCA transient, the upper plenum will experi-ence vaiding before the upper head. The voids in the upper plenum will be indicated by a lower RVLIS reading. The RVLIS will not indicate where the voiding is occurring, but at this point in the transier.t, it is not necessary to know where the region of voiding is.

In the early part of the transient when the mixture level is above the top of the guide tube in the upper head, it is sufficient for the operator to know that the vessel inventory is decreasing, irrespective of the region where voiding is occurring. As discussed in the response to Item 21, the fluid i

in the upper head does not affect the RVLIS indication after the upper l

head has drained to below the tcp of the guide tubes.

As discussed in the response to Item 19, the upper head will drain before the onset of ICC and there will not be an ambiguous indication during the period of time when RVLIS will be used.

l Item 21 j

One discussion of the microprocessor system states that water in the upper head is not reflected in the plot.

Does this mean that there is no water in the upper head or that the system is indifferent to water in the upper head under these conditions?

Response

The discussion in the system description is contained in the section describing the analysis of the system performance. The statement in question is referring to the WFLASH code calculation of mixture level, 1

_ =... -

t 15 I

rather than how the RVLIS will respond to water in the upper head. The computer code includes calculation of water mass and pressure in the upper head, but this water mass is not included in the calculation of j

mixture level; hence, the mixture level is indicated only below the elevation of the upper support plate.

l j

The RVLIS measurement from top to bottom of the vessel will measure j

the level ta the following regions:

top of vessel to top of guide tube; inside guide tube from top to upper support plate; upper plenum; reactor core; lower plenum. During a LOCA, the RVLIS will measure the water level in the upper head only until the level drops to the top of the guide tubes.

RVLIS would then measure level reduction in the guide tubes and upper plenum. The water remaining in the upper head below the j

top of the guide tubes would not be measured by RVLIS. This water would eventually drain through small holes into the guide tubes and downcomer.

This draining would be accomplished within a few minutes, depending on J

the accident.

In any case, the water temporarily retained in the upper I

head would have no effect on the RVLIS indication.

Item 22 i

Describe the details of the pump flow D/P calculation. Discuss the possible errors.

1

Response

Calculations are performed to obtain an estimate of the differential pressure that the wide range instrument will measure with all pumps operating, from ambient temperature to operating temperature. The calculations employ the same methods used to estimate reactor coolant flow for Plant design and safety analysis. These calculations are used primarily to define the instrument span and to provide an estimate for the function that compensates the differential pressure signal over the full temperature range, ie, that results in the wide range display indicating 100% over ~ 5e full temperature range with all pumps operating, pumping subcooled coolant. During the initial Plant startup following installation of the instrumentation, wide range dif ferential pressure data would be obtained and used to confirm or revise the compensation function so that a 100% output is obtained at all temperatures.

Since the calculated compensation function is verified by Plant operating data, any uncertaicti:s in the flow and differential pressure estimates are eliminated.

Item 23

)

Have tests been run with voids in the vessel? Describe the results of j

these tests.

I

Response

Tests of the Westinghouse RVLIS have been run with voids in the vessel at the Semi-scale Test Facility in Idaho.

Small break LOCA experiments are being conducted at this facility by EG6G for the NRC. The results of i

~. -

t

)

16 l

these tests are used to compare the RVLIS measurements with Semi-scale dif ferential pressure measurements, gamma densitometer data and core cladding surface thermocouple indications.

To date, after correcting for j

differences between PWR reactor vessel internals and Semi-scale modeling, good correlation between Semi-scale level indications and RVLIS measure-ments has been observed.

In cooperation with the NRC, EG6G and ORNL, Westinghouse is nreparing a report summarizing the RVLIS performance during selected semi-scale tests.

t Item 24 Estimate the expected accuracy of the system ai.er an ICC event.

Response

The accuracy of the system as described in the response to Item 6 l

would be the same for any 10CA-type incident, including an ICC event, l

causing a temperature increase within the reactor Containment. Uncer-i taintiee due to reference leg temperature measurements and sensor and hydraulic isolator displacements are included in the accuracy analysis.

i Item 25 Describe how the conversion of RTD resistance to temperture is made in the analog level system.

Response

J 4

l The RTD is connected such that an analog voltage, which is proportional to RTD temperature, is input to the microprscessor system. This analog

[

i voltage is converted to temperature by using a curve stored in memory which relates voltage to RTD temperature.

i 1

J i

LAH/bw 4-66.15A2

..