IR 05000237/1987029

Insp Repts 50-237/87-29 & 50-249/87-28 on 870807-25.No Violations or Deviations Noted.Major Areas Inspected:Similar Occurrences to 870807 Event,Root Cause Determination & Operations & Health Physics Personnel Reactions

ML17199T388
Person / Time
Site:	Dresden
Issue date:	10/15/1987
From:	Forney W NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III)
To:
Shared Package
ML17199T387	List: IR 05000237/1987029 ML17199T386
References
50-237-87-29, 50-249-87-28, CAL-RIII-87-14, NUDOCS 8710230279
Download: ML17199T388 (29)

Text

'

U. S. NUCLEAR REGULATORY COMMISSION.

REGION III

Report Nos. 50-237/87029(DRP); 50-249/87028(DRP)

Docket Nos. 50-237; _50-249 License Nos. DPR-19; DPR-25 Licensee:

Commonwealth Edison Company P. 0. Box 767 Chicago, IL 60690 Facility Name:

Dresden Nuclear Power Station, Units 2 and 3 Inspection At:

Dresden Site, Morris, IL Inspection Conducted:

August 7 through 25, 1987 Inspectors:

NRC Augmented Inspection Team Team Leader:

M. A. Ring Approved By:

Team Members:

S. DuPont T. M. Tongue I. Villalva I. Yin W. Grant Assistance Provided By:

P. Y. Chen A. J. H; Lee M. Russell P.. D. Kaufman W. L. Forney, Chie~f?*

Reactor Projects Branch 1 Inspection Summary 10/ls--jg7 Date Inspection on August 7 through 25, 1987 (Report Nos. 50-237/87029(DRP);

50-249/87028(DRP))

Areas Inspected:

~pecial Augmented Inspection Team (AIT) inspection conducted in response to the reactor feedwater event of August 7, 1987, on Dresden Unit 3 and related activitie The review included similar previous occurrences, root cause determination, operations and health physics personnel reactions, damage evaluation, and a portion of assessment and preparation related to the restart of the reacto PDR ADOCK 05000237 G

PDR i

l

Results:

No violations or deviations were identified; however, the licensee has committed to conducting a comprehensive test to determine the adequacy of the feedwater system modifications and to collect data for determination of the root cause.

ii

Augmented Inspection Team Report 50-237/87029; 50-249/87028 Introduction Synopsis of Event AIT Formation AIT Charter Persons Contacted I Description - Feedwater Event of August 7, 1987, and B. Previous Events Narrative Description Sequence of Events Recent Feedwater Events Comparison II Investigative Efforts I V VI VII IX. Health Physics Activities Performance of Operators Damage Assessment Feedwater Regulating Valve (FRV) and Pneumatic Controller Operation Feedwater Control System System Description Post Event Monitoring Root Cause Assessment Modifications and Testing AIT Conclusions Inspector Finding Associated with FRV Maintenance Open Items Exit iii Page N *

\\

Attachment N Figure 1 Figure 2 Figure 3 Figure 1 Figure 2 Figure 3 Table 1 Table 2 Table 3 Attachments Description Confirmatory Action Letter (CAL)

Letter lifting CAL Condensate and Feedwater Diagram Feedwater Transient (Level/Flow)

Materials Analysis Report on Breaks Feedwater Regulating Valve and Pneumatic Diaphragm Operator Feedwater Regulating Valve Control Feedwater Regulating Valve with Damping Feedwater Control Block Diagram Feedwater Regulating Valve Control Feedwater Control System Feedwater System Instrumentation Li st of Network 90 Parameters Test Point List Pump Curve (System)

LER 87-013-0 iv

,J

• Introduction Synopsis of Event On Friday, August 7, 1987, around 8:00 a.m. CDT, the Dresden Unit 3 reactor was in the process of being shut down to repair a contain-ment purge isolation valve when (at approximately 25% power)

significant feedwater flow and pressure oscillations and feedwater piping vibrations were experience The reactor was manually scrammed when leaks were reported in the turbine and reactor building area Because of the reported leaks, the main steam isolation valves were closed and the isolation condenser was used for reactor coolin Two pipe breaks were subsequently determined to be the leakage source One break was in the sensing line to a feedwater flow transmitter in the turbine buildin The other break was in a reactor water cleanup (RWCU) drain line in the reactor buildin Surface contamination was present in the areas of the two leaks, pipe insulation had been knocked off in several areas and asbestos dust was present in several area The licensee had previously notified the NRC at 2:51 a.m. CDT that an Unusual Event had been entered for the shutdow At 9:11 a.m. CDT, the licensee notified the NRC of the feedwater oscillations, the manual scram and the piping damag AIT Formation On August 7, 1987, the Acting Senior Resident Inspector (SRI),

S. G. DuPont, who was onsite at the time of the event, proceeded to the control room and observed the licensee's recovery action W. B. Grant, Radiation Specialist, was dispatched from Region III offices to monitor activities related to the contamination and any potential radioactive release T. M. Tongue, Senior Resident Inspector, formerly assigned to Dresden, now assigned to the Braidwood site, was dispatched to Dresden to assist in the monitoring of recovery activitie Subsequently, on August 10, 1987, an Augmented Inspection Team was formed which included the above three Region III individuals and M. A. Ring, Chief, Reactor Projects Section lC and Team Leader, I. T. Yin, Assistant to the Director of the Division of Reactor Safety (DRS) and I. Villalva, Technical Support Section, who arrived onsite on August 10, 198 As the team progressed through the investigation of the event, additional expertise in the areas of pipe stress and vibration was requeste P. Y. Chen, and A. J. H. Lee from the Office of Nuclear.

Reactor Regulation (NRR), M. Russell from Idaho National Engineering Laboratory (INEL), and P. D. Kaufman, Dresden Resident Inspector, were assigned to the AIT in these areas and to assist in monitoring plant damage assessment and restart preparation activitie Concurrent with the AIT activities, Region III issued a confirmatory action letter (CAL-RIII-87-014) which was received by the licensee on August 12, 1987, and is included as Attachment 1 to this report.

The CAL confirmed certain licensee actions related to the event and in support of the AIT, and established conditions to be met prior to restart of Unit *

• ' AIT Charter On August 10, 1987, a tentative charter for the AIT was formulated as follows: Examine the circumstances surrounding the August 7, 1987, feedwater transient event which resulted in two small pipe breaks in the feedwater and RWCU system.

• Concentrate efforts in the following areas: Pipe breaks -

Fracture mechanism System stresses Support adequacy *

Vibration Relation to previous events Feedwater Control System - Ability to control Stability at low powe Relation to previous events Operator Performance - Prior to and during event Recovery action Feedwater Regulating Valves - Contribution to root cause and Check Valves Relation to previous event Performance of RWCU System - As it relates to feedwater Licensee Actions/Analysis in Evaluating Event system Contributions to fracture Expand areas of concentration as evaluation of data direct The AIT charter was communicated by telephone to Regional management on August 10, 198 The charter was not further formalized as developing information directed the course of the inspectio Persons Contacted Commonwealth Edison Company (CECo) Corporate Personnel D. Galle, Vice President, BWR Operations N. Kalivianakis, General Manager, BWR Operations L. DelGeorge, Assistant Vice President, Licensing and Plant Support Services

• M. Turbak, Nuclear Licensing Assistant Manager L. Gerner, Administrative Engineer, PWR Operations

• R. Mirochna, BWR Engineering

• S. Javidon, BWR Engineering R. Grams, Production Services

• R. Janecek, Nuclear Safety J. Abel, BWR Engineering Manager

• M. Schreim, BWR Engineering

• G. Frizzell, BWR Engineerin L. Butterfield, Nuc1ear Licensing Manager Dresden Station Personnel

• E. Eenigenburg, Station Manager J. Wujciga, Production Superintendent

• R. Flessner, Technical and Administration Superintendent

• J. Brunner, Assistant Superintendent, Technical Services J. Kotowski, Assistant Superintendent, Operations

• D. Van Pelt, Assistant Superintendent, Maintenance

• J. Achterberg, Technical Staff Supervisor

• R. Jeisy, Station Quality Superintendent

• E. Armstrong, Regulatory Assurance Supervisor

• M. Moy, Technical Staff

• R. Geier, Quality Control Supervisor

• S. Lawson, Shift Foreman K. Brennan, Regulatory Assurance D. Booth, Master Instrument Mechanic A. Zapatocky, Instrument Maintenance General Electric J. Nash, Field Engineer Sargent and Lundy (S&L)

• J. DeMarco, Control and Instrumentation D. Landers, Consultant

• E. Branch, Consultant A. Dermenjian, Consultant J. Nosko, Consultant

• Denotes those attending the exit meeting on August 25, 1987.

In addition, several Dresden operations, maintenance, health physics and other technical personnel were interviewed by the AI I Description - Feedwater Event of August 7, 1987, and Previous Events Narrative Description On August 7, 1987, Dresden Unit 3 commenced an orderly reactor shutdown at 2:30 a.m. CDT in accordance with Technical Specifica-tion requirements because containment purge isolation valve A0-3-1601-63 would not close during its surveillance tes The valve was manually closed and an Unusual Event was declared in accordance with Dresden's emergency pla This situation played no part in the subsequent feedwater transient events except to serve as the reason for the shutdow Unit 2 was operating routinely at power at the time of the event and continued to do so throughout the event and recover *

Unit 3 was being shut down at 195 MWe (approximatley 25% power)

with the feedwater system in single element control using the 3A fe~dwater regulating valve (FRV) in automatic and the 38 FRV about 2% open in manual with the 3A reactor feedwater pump (RFP) runnin The event started at about 8:08 a.m. CDT as evidenced by a feedwater heater annunciator alarm followed by the Nuclear Station Operators (NSOs) hearing relay chatter sounds in the control panel These indications were followed by a partial containment Group I isolation signa The next observations were a lockup (open) alarm on the 3A FRV and high vibration alarms on the FRV and on the 3A and 3C feedwater pump These alarms were accompanied by full scale swings on the pressure instruments on the feedwater pumps and condensate booster pump Simultaneously, alarms were received on the Reactor Water Clean Up (RWCU) system which resulted in a manual isolation of RWC Later, a broken drain line was identified on the RWCU return to the feedwater syste Since the 3A FRV locked up (open), the reactor vessel water level started rising rapidl This rise was countered by prompt. action by the unit NSO, who manually shut the 3A and 38 feedwater (FW)

motor driven isolation valves, then reopened the 38 FW motor driven isolation valve and stabilized the reactor vessel water level at 30 inches (normal level) using the 38 FRV after a rise to 54 inches (1 inch short of the automatic turbine trip on high level at 55 inches).

At this point, the excursion appeared to be over; however, reports from the reactor and turbine buildings indicated steam and water leaks in the area of the RFP room and the shutdown cooling (SOC) heat exchanger roo Therefore, the Assistant Superintendent for Operations and the Shift Engineer ordered the Unit 3 reactor to be manually scrammed (see Attachments 3 and 4).

In order to minimize the loss of inventory and to limit damage from the steam leaks, the main steam isolation valves were shu The isolation condenser system was placed in operation to provide core cooling and it functioned appropriately except for a Group V isolation which occurred at 10:29 a.m. CD This isolation, which automatically closes the isolation condenser valves, is believed to have resulted in moderator shrinkage and a subsequent reactor scram on low water leve Since control rods had already been inserted on the previous scram, this scram had no effect on the reacto The isolation condenser operation was restored, but the root cause for the isolation had not been determined as of the AIT exit dat The isolation condenser operations, however, had little effect on the feedwater transient except to introduce an additional unexpected action which hampered the efforts to investigate the even Accordingly, except for health physics effects, the isolation condenser operation will not be discussed further in this report but may be further explored in subsequent inspection In addition, during the initial manual scram, 22 control rods were reported to have stopped at the 02 position versus full in (00 position).

This phenomena has been a common occurrence at Dresden and the licensee has performed a safety analysis which shows

• TIME adequate shutdown margin even if all rods had stopped at position 0 This condition also had no effect on the feedwater event; therefore, it will not be discussed further in this repor During the event, personnel in the plant noted various loud noises similar to vibrating piping, waterhammer, and/or W?ter jet sound In addition, water was found leaking in the reactor building in the vicinity of the RWCU systems and shutdown cooling systems and in the turbine building in the vicinity of the reactor feedwater pump Following the decision to scram, the reactor and turbine buildings were evacuated because of both the leaks and a considerable amount of suspected asbestos dust from dislodged insulation which was in the air in these building Later, walkdowns revealed a broken instrument line on the feedwater system (3/4 inch instrument line for feedwater flow transmitter 3-614C) and a broken 1 inch drain line on the RWCU system that interfaces with the feed system via the High Pressure Coolant Injection (HPCI) syste In addition, a number of base plates for feedwater piping hangers were found with concrete anchor bolts loosene Further damage assessment and clean up activities will be discussed later in the repor Sequence of Events A determination of the sequence of events for the feedwater transient of August 7, 1987, was hampered by the fact that the sequence of events recorder was out-of-service at the time of the even Further, the alarm printer, which was available, prints out the time it 11 sees 11 an alarm in its scan cycle (not necessarily the actual time of the alarm).

This means that even through several alarms could come in simultaneously, the printer would assign them a time and a sequence as it scanned through its cycl Consequently, alarms could be out of sequence merely by relative position in the scan cycl Times and sequence of events in the previous narrative description were derived by the AIT from a combination of interviews and hard dat The following sequence of events was prepared by the license Both should be viewed as approximat DISCUSSION OF 8/7/87 EVENT EVENT 0230 Unit 3 at approximately 650 MWe, commenced an orderly, expediti-ous reactor shutdown in accordance with Technical Specification requirements:

Containment purge isolation valve A0-3-1601-63 would not close during surveillance tes (Valve was manually closed.) Declared GSEP unusual even Unit 3 at approximately 285 MWe, 3C reactor feed pump was taken of A reactor feed pump o A FW Reg. valve controlling level (single-element) and 38 FW Reg. valve in manual (2-3%

open).

• 0808 0808+

0809 -

0811 0810 -

0812 0816 0817 -

0833 Unit at 195 MW Received alarms for 3A FW Re A FW Reg. valve vibration, A & C RFP vibratio noticed indication swings for condensate booster pressure valve lock-up, Operator pump and RFP Received alarms for high 3A RFP flow and high 3A RFP discharge header pressur Reactor water level risin Attempted to reset 3A FRV lock-up -- would not rese Closed MO isolation valve for 3A FR Reactor water level increased to +54 11, then began to decreas Operator opened M.O. isolation valve for 38 FW Reg. valv Reactor water level stabilized at normal level (+30 11 ) with 38 FW Reg. valve in auto contro Received reactor water cleanup system alarms for leak detection and cleanup pump tri Received cleanup suction auto-isolatio Shift Engineer received reports of steam leaks in* the Unit 3 RFP room area* and in the SOC heat exchanger roo Decision made to scram the reacto Done manually.

Announced to evacuate turbine and reactor building Secured and isolated condensate/feedwater syste Recent Feedwater Transients at Dresden Units 2 and 3 In less than four weeks, three significant feedwater transients have occurred at the Dresden nuclear facility:

the first and third at Dresden Unit 3 on July 11 and August 7, 1987, and the second at Dresden Unit 2 on July 17, 198.

Event of July 11, 1987 The event on Unit 3 on July 11, 1987, began as Unit 3 was going through a controlled shutdown for generator maintenance and power level was equivalent to about 190-200 MWe with two turbine bypass valves ope The 3A reactor feedwater pump (RFP) was inservice with feedwater flow being controlled by the 3A FRV in three element control (similar conditions to the August 7, 1987, event except in three elements).

At about 0218, the NSOs noted the open indication light blinking for the 3A RFP motor operated discharge valv This was followed by an increase in reactor water level, an associated reactor feedwater flow increase and a full open indication on the 3A FR In addition, half scale oscillations were noted on the pressure instruments for the 3A RFP and the 3A con.densate booster pump, and high vibration alarms were annunicated on the

l -

• A RFP and the feedwater regulating statio The NSO promptly shifted to single element control, placed the FRV controller in manual and inserted a zero demand signal causing the FRV to clos The NSO also reset the FRV to level control from runout flow contro When this seemed ineffective, the NSO commenced throttling the 3A FRV 1s motor operated isolation valv The NSO was unable to stop the level increase before receiving a turbine trip on high reactor vessel water level followed by a closure of all of the main steam isolation valves which caused an automatic reactor tri The running reactor feedwater pump failed to trip at +55 inches level as design intended; however, the remainder of the event, (e.g. scram recovery, etc.) proceeded in a normal manne The licensee 1s root cause evaluation attributed this event to operating the FW control system in 3 element control (feedflow, steamflow and reactor level) at too low a power level where the feedflow and steamflow signals are relatively small and small changes in feed flow and steam flow can cause control instabilitie As a corrective action, the licensee restricted feedwater control to single element (reactor level) operation onl Following a walkdown of the system, the only damage identified was to piping insulatio* Event of July 17, 1987 The event on Unit 2 on July 17, 1987, began as reactor power had been decreased and was steady at around 80%.

The 2A and 28 reactor feed pumps were in service with the 28 FRV operating in single element control and controlling reactor level. *At 11:28 p.m. CDT, the unit re~eived a scram on low reactor water level due to closure of the 28 FR Rea~tor level dropped to -3 inches and the operator noticed that FRV demand was zero, no feedwater was entering the reactor, feedwater control was in automatic and the 28 FRV indicated a 11 loss of air 11 locku The operator took manual control of the minimum flow feed valve and demanded a full open signal which was successful in providing

~bout 1 million pounds per hour feed flo At 2331, the operator cleared the 28 FRV 11 lockup 11 and immediately the valve began openin Reactor level recovered rapidly until both feed pumps tripped on high level at +55 inche The remainder of the scram recovery operations were norma The licensee 1s analysis.determined that the unexpected closure of the 28 FRV (which caused this event) was apparently due to a fault in the feedwater control system which commanded the valve to clos This conclusion was based on the observed zero demand to all FRVs and high level alarms on all FRV control stations when low reactor level conditions actually existe However, even after a series of post ev~nt tests, the licensee was unable to determine the root cause of the control system fault or to duplicate conditions observed in the even Consequently, the eight parameters associated with Unit 2 FRV

• control (A & B reactor level, A & B pressure, A & B FRV demand and A & B FRV position) were connected to a continuous monitoring recorder and a series of special precautions and actions were developed for restar These actions included additional personnel stationed for feedwater control system monitoring, a preferred valve usage sequence and a course of action if abnormalities occurred in the leve1 control syste Comparison The common elements between the Unit 2 and Unit 3 transients are insignificant compared to those of the two transients involving Unit 3, the more notable being that each transient occurred while reactor power was being reduced, and each included a malfunctioning feedwater regulating valv These similarities, however, are minor compared to the following differences:

0

0 The Unit 2 transient resulted in low reactor water level but the Unit 3 transients resulted in high reactor water leve The Unit 2 transient occurred while a drag valve (2BFRV) was being used to regulate flow whereas the Unit 3 transients occurred while a gate (plug-type) valve (3AFRV) was being used to regulate flo The Unit 2 transient occurred at a much higher power level than the Unit 3 transient Finally, and most significantly, it appears that the Unit 2 transient did not cause significant pipe vibrations nor structural damag In contrast, both Unit 3 transients resulted in severe pipe vibrations and damage to pipe insulation, and the Unit 3 transient of August 7th also caused piping and structural damag Because of the significance of the Unit 3 transients compared to the Unit 2 transient, and the differences between the Unit 2 and Unit 3 transients, the remaining discussion is limited to the Unit 3 feedwater transient The common attributes of these.two transients are of particular interest, the more noteworthy of which are:

0 Both occurred on Unit 3 with the 3A Reactor Feed Pump (RFP)

and the 3A condensate/condensate booster pump in service, the 3A FRV in automatic and the 38 FRV in manua The unit was being shutdown for both events and steam flow was in the 190-200 MWe range when considering use of turbine bypass *

valve Each caused severe pipe vibrations and damage to insulatio (Apparently no structural damage occurred during the first transien However, the second transient caused piping and structural damage, e.g., a one-inch drain line in the reactor water cleanup system and a 3/4 inch instrument line in the

0

0

0 fe~dwater system were levered at t~eir connection to a larger pipe, a loose yoke was found on the operating feedwater regulating valve, loose snubber supports and base plate anchors were found, and _anchors with deformed grouting were discovered.)

Each caused the annunciation of high vibration alarms from the feedwater regulating station and the feedwater pump Both resulted in severe pressure oscillations in the feedwater and condensate system Each included rapid changes in feedwater flow and reactor water leve Each had indications of erratic operation of the feedwater regulation valve Both resulted in high reactor water leve (The high water level of the*first transient caused a turbine trip *which lead to the spurious closure of the MSIVs that apparently caused an automatic scra In contrast, the high water level of the second transient was apparently brought under control by prompt operator intervention, thereby precluding a turbine trip and possibly an automatic scram; however, because of the afore-mentioned pipe breaks the reactor was manually scrammed.)

Each experienced malfunctions of the controls for the feedwater regulating valv (During the first event the range spring assembly on the valve stem feedback arm to the positioner malfunctioned such that the regulator valve could not be fully close During the second event, the range spring assembly had disassembled.)

Some differences between the two Unit 3 events were:

the resultant vibration or system shocks were more severe during the August 7 event; on July 11, the feedwater regulating system was in three element control versus single element control on August III. Investigative Efforts Health Physics Activities Through review of records and interviews with Health Physics (HP)

and Operations personnel, the inspectors determined that HP personnel responded to the event in a prompt and efficient manne When the turbine and reactor buildings were evacuated, HP personnel set up appropriate sampling stations and monitored those personnel involve They also took prompt action to work with station men to contain the water that was being released from the two broken line Historically, the isolation condensers have caused the release of measurable quantities of low level radioactive contamination outside the reactor building during their operatio Prior to this event, the Unit 3 isolation condenser had undergone an internal cleaning process to eliminate most of the residual contaminatio With this known to station personnel, the HP personnel responded promptly and thoroughly during and after the event to isolate and monitor those areas that had previously been contaminated during operation of the isolation condense Following the event, they expeditiously surveyed and released those areas found uncontaminate The actions of the HP personnel are believed to have been effective in preventing a large number of plant personnel from becoming contamin-ate The Radiation Specialist inspector arrived onsite at about 1:30 on August 7, 198 By this time the isolation condenser reactor*

cooling mode had been shifted to the shutdown cooling syste Licensee rad/chem personnel were monitoring areas outside adjacent to the isolation condenser for radioactive contamination and areas inside in the vicinity of the pipe breaks for radioactive contamin-ation and asbesto Areas were being released as they were deemed free of radioactive contamination/asbesto No detectable radioactive contamination was found on the outside areas using portable survey instruments or smear However, samples of water which had collected in puddles on the concrete roadway beneath the isolation condenser vent showed 1.0 E-6 microcuries per milliliter of Co-60 when counted on a GeLi syste The contaminated water ~as remove The maximum contamination within Unit 3 was about 22,000 dpm/100 cm2,

mainly in the areas immedi~tely adjacent to the pipe break Lower levels of contamination were detected on the Unit 2/3 turbine dec An Assistant Superintendent for Production had his pants and shoes slightly contaminated (<100 cpm) while checking for the leak The superintendent was whole body counted; no internal uptake was identifie Possible asbestos contamination from pipe insulation caused the licensee to require protective clothing and full-face respirators in areas of the reactor and turbine building No significant problems were identifie No violations or deviations were identifie Performance of Operators During the inspection, the inspectors interviewed a number of licensed Nuclear Station Operators (NSOs) including those involved in the events, and senior licensed personnel such as the shift foremen, shift engineers, and the shift control room engineers (SCRE).

In addition, the General Electric Field Engineer, various instrument maintenance (IM) personnel, auxiliary operators, et were interviewe The inspectors also reviewed various logs and written accounts related to the event and affected equipmen *

This was done to obtain an in depth perspective of the sequence of the events, actions of various personnel involved, and a knowledge of the history of activities leading up to the event Since the sequence of events computer was out-of-service during the event, interviews were one of the few means available to provide a number of options to pursue as potential root cause The inspectors found that all personnel interviewed provided compatible descriptions of the events which were as described previously in this repor Based on the prompt recognition of the event, the actions taken to regain level control, the manual scram and isolation, and the evacuation and recovery actions, the AIT believes that the operators acted in a professional, responsible manner with the safety of the plant in min The operator actions in recognizing the event and regaining level control also demonstrated a good understanding of the symptoms of the July 11, 1987 even However, several operators also expressed an extreme distrust in the operation of the feedwater and feedwater control systems* and questioned the success of the maintenance activities which had been performed on these system No violations or deviations were identifie Damage Assessment The AIT members performed a followup inspection of the CECo feed-water transient damage assessment and repair effor In the week of August 9, 1987, the inspectors reviewed the CECo

"Walkdown Summary, Feedwater" dated August 7, 1987, and observed photographs of piping damage, including broken lines at the 3/4" feedwater flow instrument line tubing and at the 1 11 RWCU drain pipe, snubber support base plate grout damage, loosened bolts and nuts, and feedwater (FW) piping insulation damage at various locations extending from the condensate booster pump discharge to the reactor drywe 11 penetration (X-area) where RWCU and High Pressure Coo 1 ant Injection (HPCI) piping was connecte As a result of the inspectors'

walkdown of the damaged piping areas and review of the FW records, the following comments were made:

0

The CECo walkdown summary of August 7, 1987, did not record all the damage conditions at all the specific piping system location The past design/evaluation of the FW system was considered questionable due to incorrect piping configuration shown in the S&L record *

The past FW transient history was unavailable for review to place the present event in its proper perspectiv *

The CECo management stated that additional actions would be carried out in response to the NRC comment In the following week of August 16, 1987, after additional walkdowns has been performed by CECo site personnel, support damages at a 24

FW pipe riser located between the reactor feed pump discharge and the FW regulating valve station, were identifie During the past weekend, a scaled model of the FW system, including RWCU and HPCI, had been built showing the correct piping configuration and identi-fying all the damaged support, piping, and insulation location The FW transient history was also traced, documented, and discussed during a subsequent CECo presentatio The NRC staff completed inspection of all the damaged areas including extraction steam to the FW heaters, repair of the 3/4

tubing, and repair and modification of the 1 11 RWCU drain pip The NRC staff also reviewed the CECo report, 11System Materials Analysis Department Report on a Failed Reactor Water Clean Up Line and a Feedwater Instrument Line at Dresden Station, Unit 3, 11 dated August 13, 1987, which concluded fatigue cracking and.ductile overload to be the failure cause upon examination of the fracture surface: (see Attachment 5).

The NRC staff also discussed CEC0 1s planned corrective actions for the piping and pipe support The licensee committed to conduct a number of tasks, including:

0

0 S&L will evaluate all the small bore lines (1 11 and smaller)

connecting to large bore pipes including the reactor feed pump suction, discharge, regulating valve station, high pressure heater bay, and the X-area piping system The evaluation will be based on flexibility and oscillation support criteri Any small bore lines that fail either criterion will be non-destructively examined for possible fatigue cracking and would be modified to meet the criteri~.

Non Destructive Examination (NOE) will be performed on regulating valve station piping, RWCU piping in the X-area, and other potential high stress piping locations determined by S& Damage will be repaired on support structures, grouts, and pipe insulation, and loosened bolts/nuts will be tightene S&L will calculate forces that were required to bend the pipe clamp and crack the support welds below the clamp at the 24

• FW rise The forces and the related pipe stresses will be evaluated to determine any safety effects on continued system ope rat ions.

The licensee discussed results and progress of the above activities during a presentation held in NRC Region III offices on August 21, 198 The licensee further stated:

• *

0

The planned FW testing will include engineering evaluation and test verification of FW regulating valve station pipe displacements at reduced power level Determination on whether or not snubbers will be in locking up positions under normal operating conditions will also be mad NRC followup on this item is planned. * This is an open item (237/87029-01; 249/87028-01).

The S&L safety system evaluation of the 24 11 FW riser damages was forwarded to NRC for revie No violations or deviations were identified as a result of the NRC revie Additional dynamic piping analyses were being considered to assist in determining the underlying causes of the FW transien No violations or deviations were identifie Feedwater Regulating Valve (FRV) & Pneumatic Controller Operation The 3A FRV is a Copes-Vulcan double seated 14 inch regulating valve with a 12 inch reversible diaphragm operator (Attachment 6. Figure 1).

The FRV is positioned by either reactor vessel *water level demand (single element control) or a combined demand of steam flow, feedwater flow and vessel water level (three element control) by changing the pneumatic signal across the E/P (electronic-pneumatic)

transduce This pneumatic signal is applied to the positioner 1s bellows where a counterbalance of the pneumatic signal and the feedback force from the valve position linkage attached to the bellows through the range spring arrangement (Attachment 6. Figure 2) is mad This counterbalance will result in either the above diaphragm or below diaphragm port of the positioner opening and directing a pneumatic signal to the diaphragm chamber of the valve operator through the booster relays and the air lock valves. *The pneumatic system is designed to result in a continued loading change on the valve's diaphragm by the positioner until the valve has actually repositioned itself and has rebalanced the initial pneumatic sigrial chang Following the event of August 7, 1987, the licensee disassembled the 3A FRV in an attempt to find information which would assist in the root cause determinatio Photographs were taken of portions of the disassembly to aid both the licensee and the AI The following abnormalities were noted during the valve examination:

0 The position indication spring had broke This item was not considered significant in the root cause assessment in that it only affected indication.

The range spring which is involved in the feedback circuit had disassemble This spring was also found loosened following the July 11, 1987 event, consequently, was considered potentially significant and was factored into the licensee's root cause assessmen \\.

The yoke assembly was found to be loos This finding was also factored into the.licensee's root cause assessmen For both the disassembled range spring and the loo~ened yoke, there was no way to determine-whether or not they had occurred before the transient and caused or contributed to its severity or rather had occurred as a result of the vibration produced during the transien Feedwater Control System System Description The Dresden feedwater control system is a power generation system designed to maintain reactor vessel water level within

• predetermined high-and low-level limits by controlling feedwater flo To accomplish this, the original Dresden feedwater control system included GEMAC analog controllers designed by General Electric (G.E.).

Because of operational and maintenance problems, Dresden's GEMAC feedwater controllers were recently replaced with computer-based digital controllers d~signed by Bailey Controls (hereinafter referred to as Network 90).

The operational problems included system instabilities when transferring to different control modes such as from manual to automatic and single element to three element control, and balancing problem The principal maintenance problems were based on obsolescence, (i.e., since GEMAC is no longer being manu-factured by G.E., spare parts and/or major components are not readily available).

The Network 90 control system is one of two recent modifica-tions made on the Dresden Unit 3 feedwater system (the other being the rework of the impeller for feedwater pump 3A).

Because the Network 90 control system is a new and major modification to the Dresden feedwater system, both the licensee and the AIT consider it to be a possible contributor to, but not necessarily the root cause of, the recent feedwater transient Consequently, a great portion of the post event investigative efforts by the licensee and the AIT were devoted to the feed-water control syste The Network 90 control system for Unit 3 was placed in service on August 23, 1986, subsequent to the recirculation pipe replacement outage, and for Unit 2 on April 9, 1987, following its 1987 refueling outage. Although the Network 90 control system was purchased directly from Bailey Controls, its design specifications and other engineering services, including establishing the interface requirements with existing subsystems, were prepared by The other engineering services performed

by G.E. included reviewing the Network 90 specifications to assure that the like-for-like requirements were fulfilled; verifying the compatibility of the Network 90 system with existing interf~cing components; and assuring that the start-up and acceptance tests of the Network 90 system were identical to the original feedwater control system test The Network 90 control system was designed to operate with the existing input sensors (e.g., level and flow transmitters) and existing output devices (e.g., regulating valve operators).

A brief description of the maj"or components of the Network 90 control system and its function in the overall feedwater control system follow The basic functional requirements and features of Dresden's new feedwater control system are identical to the old system, the major difference being that the original analog control system (GEMAC) was replaced with a computer-based digital control system (Network 90).

Attachment 7, Figure 1, Feedwater Control System Block Diagram, depicts the major components of the feedwater control system (the Bailey Network 90" Distributed Control System is shown within the dashed lines in Attachment 7, Figure 1)~ All process control functions and required signal conditioning are performed by the Network 90 system, including the conditioning and processing of incoming signals from the reactor water level transmitters, steam and feedwater flow transmitters and the associated pressure transmitters for pressure compensation for steam flow and reactor water leve The Network 90 system generates output control signals for the feedwater regulating valves and the regulating bypass valve based on the input signal The feedwater regulating valves (Attachment 7, Figure 2) are controlled by air signals as follow The Network 90 Master Controller transmits an electrical signal to an electrical to pneumatic (E/P) conve~ter based on level changes or on differences between feedwater and steam flo The E/P converter sends a 3-15 psig control air signal to a valve positioner which, in turn, directs operating air through one of two normally open air lock valve One air lock valve supplies air to the upper side of the piston diaphragm of the valve operator for valve closing and the other supplies air to the lower side of the piston diaphragm of the valve operator for valve openin Upon an air system failure or the loss of the controller's signal, the air lock valves close and lock the feedwater regulating valve in its existing position (lock-up).

Dresden's feedwater control system includes two redundant, density compensated reactor water level instrument loop (Density compensation is computed within the Network 90 software configuration.) The initial selection of the control loop is by a manually operated selector switch; however, upon a failure within the selected loop, control is automatically transferred to the other loo Each of the four main steam

. 15

lines has a steam flow transmitter and a pressure transmitte The signals from the instruments in the four steam lines are processed, compensated and summed by the Network 90 system to provide mass steam flow informatio Each of the three main feedwater lines has a feed flow transmitter whose signals are processed and summed by the Network 90 system in the same manner as steam flow to provide mass feedwater flow informatio The Network 90 system controls feedwater flow in one of four control modes:

manual, single-element, three-element and runout flow contro Manual control allows the reactor operator to control the actual flow of feedwater into the reactor vessel irrespective of actual reactor water level or changes therei Manual control is implemented by selecting 11Manual 11 at either the Master Control station or at the Manual/Auto Control stations for the individual feedwater regulating valve Singl~-element cont~ol uses reactor water level as the con-trolling paramete Single-element control is implemented by selecting 11l-Element 11 on the 11 Level Control Mode Switch 11 and is designed to operate at low power levels onl After the desired control level is set, the control system only responds to changes in actual level with respect to the*desired level.*

In contrast, three-element control uses reactor water level, mass feedwater flow and mass steam flow as the controlling parameter Three-element control is implemented by selecting 113-Element 11 on the 11 Level Control Switch 11 and is designed to operate at high power level Runout flow control is designed to allow for maximum feedwater flow without overloading and tripping the feedwater pump motor A brief description of the basic operation of the feedwater control system follow (Refer to Attachment 7, Figure 3.)

Under normal operation, one of two redundant reactor level signals is selected by use of the reactor level selector switc The selected level signal is transmitted to the level vs flow error network and also to the Master Controller via a control mode selector switc This selector switch determines whether the signal to the Master Controller is a single-element (level) signal or a three-element (level, steam flow/feedwater flow error network) signa The individual Manual Auto Transfer Stations for each feedwater regulating valve receive control signals from either the Master Controller or the Runout Flow Controlle The regulating bypass valve normally receives its control signal from the Manual Auto Controlle During normal operation, the Master Controller provides the control inpu Should a runout condition occur, the Runout Transfer Switch automatically shifts the input from the Master Controller to the Runout Flow Controlle During abnormal conditions, such as an excessive mismatch between total steam flow and first stage pressure, an alarm is actuated following a 30 second time delay, or upon

\\

receipt of a full scram signal, the Master Controller dual setpoint source output is reduced by 50% to limit reactor water level overshoot following a scra The major instruments, alarms, interlocks, trips and controls associated with the feedwater control system, are listed in Attachment 7, Table 1, Feedwater System Instrumentatio.

• Post Event Monitoring The discussions that follow address some of the post-event monitoring acti~ns undertaken or planned at Dresde On August 12, 1987, the licensee monitored the performance of the Unit 2 feedwater control system while the plant was at power in order to take advantage of the instrumentation installed after Unit 2 1s event and to obtain dynamic performance informatio The parameters monitored were reactor water level from the 2A and 28 reactor level instrument loops; reactor pressure from the 2A and 28 pressure transmitters used for density compensa-tion of reactor water level; demand signals to the 2A and 28 feedwater regulating valves; and the position of the 2A and 28 feedwater regulating valve Data was recorded while both feedwater regulating valves were in single-element control and while in manual control with no changes in set-poin The recorded data indicated close tracking of the actual position of both valves to the demand signal while the Network 90 system was in single-element control.* When the Network 90 system was in manual control, with no change in set-point, feedwater regulating valve 2A exhibited a close-open oscillatory motion of approximately two percent; however, the 28 va.lve did not exhibit such oscillation Members of the licensee 1s staff attribute the aforementioned oscillations to air leakage between the upper (closing) and lower (opening) air chamber of the regµlating valve 1s operato They further stated that these oscillations are not uncommon and that they can be readily eliminated by providing a tighter package (ring seal) between the two air chamber Although such maintenance action appears to be straightforward, it raised a concern regarding too tight a packing or sea Namely, that a tight seal could require excessive force to start the valve to open or clos Such excessive force could, in turn, cause a significant overshoot in the open or close direction once the valve starts movin Such a scenario is not unreal-istic when one considers that the Network 90 system includes a proportional gain circui A characteristic of a proportional gain circuit is that its output demand signal keeps increasing in the absence of a positive response by the element being controlle Thus, if a valve is temporarily stuck in a given position, successive increases in the moving force would ultimately cause the valve to breakaway from its stuck position.

The excessive breakaway force could in turn cause the valve to overshoot its requested position, after which the valve

• would tend to oscillate about its requested controlled positio If such oscillations are of sufficient magnitude and if they are in resonance with the piping system's natural frequency, they could conceivably induce or initiate the vibrations encountered during the Dresden Unit 3 feedwater events of July 11 and August 7, 198 On August 10th and 12th, 1987, the licensee monitored the performance of the Unit 3 feedwater control system while the plant was.shut dow Consequently, the only parameters monitored were manually initiated demand signals from the Network 90 system and the responses (positions) of feedwater regulating valves 3A a.nd 3 On August 10th, the 11as-found

performance of both 3A and 38 valves was monitore The monitoring of the 38 valve included its response to the following demand signals:

a slow ramp (i.e., a 2%/second open demand to fully open, followed by a 2%/second close demand to fully close); a fast ramp (i.e., a 20%/second open demand to fully open, followed by a 20%/second close demand to fully closed); and step changes (i.e., open step change demands of 20%, followed by a dwell time of from five to ten seconds until the valve demand reached 100% open, followed by similar close demand signals until the demand reached 100% close).

The 11as-found 11 response traces of feedwater regulating valve 38 closely tracked its demand traces (i.e., within its acceptance requirements).. Nevertheless, adjustments were made for the control of feedwater regulating valve 38, followed by nearly identical 11as-left 11 tests on August 12, 198 Although the traces of the 11as-left 11 responses appear to more closely resemble the demand curves than the 11as-found 11 responses, the 11as-left 11 traces appear to*have more lag in responding to the demand than the 11as-found 11 trace The August 10th, 11as-found 11 response tests performed on feedwater regulating valve 3A were not exactly in the 11as-found 11 condition (i.e., the range spring assembly that had fallen off during the August 7th event was replaced in order to obtain valve position traces).

Thes.e tests were nearly identical to those conducted for feedwater regulating valve 38 except that the 20% step demand signals were replaced by a series of 2%/second ramp demand signals of approximately ten seconds duration (i.e., a ramp change of 20%) followed by dwell times of ten seconds until a 100% open demand signal was generated, after which the cycle was reversed until a 100% close demand signal was generate The initial review of the response traces for regulating valve 3A created a concern because they indicated that the valve was only tracking the demand signal to the 80% open position rather than to the demanded 100% open positio It was, therefore, assumed that the valve's operator was not functioning properly and that the valve could not be fully opene However, it was later determined that such was not the case and that the valve was actually closely tracking the demand signal~ and that the

• I *

erroneous traces were caused by improper calibration of the positioner subsequent to installing the aforementioned range spring assembl Root Cause*Assessment Shortly after the event, the licensee established two task forces to review the event, the Damage Assessment Task Force discussed previously and the Root Cause Assessment Task Forc The inspectors reviewed the progress of the Root Cause Assessment Task Force and, in addition, developed an independent list of possible causes to pursue based on findings from personnel interviews and record revie Efforts of both the licensee and the AIT were hampered by the fact that the sequence of events recorder was out-of-service during the even As each of the potential causes was evaluated the 3A feedwater regulating valve (FRV),

its interactions with the 3A feedpump, and the Bailey control system appeared to be highly likely causes. The opinion expressed by a number of Operations personnel is that the Bailey control system, of recent design and*installation, has a much quicker response time than the FRV and its operator. This differs from I&C personnel opinion where they expressed confidence that the components were compatibl The assessment determined several possible causes as follows, but not necessarily limited to:

0

0

0

0

System hydraulic and hydro-dynamic desig Feedwater pump design and operatio Station operating procedure Feedwater regulating valve characteristic Actuator design and operatio Piping support syste Feedwater level control syste Feedwater/condensate recirculation flow syste Potential interaction effects between various combinations of the abov In performing the investigation of the above, the licensee made the following observation '.

Within the 1 1/2 year period preceding the 1987 feedwater transient events, the feedwater level control system had been changed -from an analog electronic control system to a digital electronic control syste Within the one-year period preceding the 1987 feedwater transient event, the 11A 11 reactor feed pump rotating element (shaft/first and second stage impeller assembly) had been replaced with a spare rotating element which had been delivered to the site in.197 The point of operation (in terms of percent power, percent throttle flow, and percent pump capacity) during which the feedwater transient events

-

were experienced is a relatively unstable operating poin The feedwater regulating valve design implemented on Dresden Unit 3 has been outdated by current technolog The operation of the similar feedwater regulating valves at Dresden, Quad Cities and several other plants has not been optima Instabilities in flow through the valve body have been noted, and instances of actuator failures over the years have been noted'.

In addition, similar valves in operation have *

experienced stem fatigue, and ultimate failur The valve/actuator combination has been subject to failure due to vibration caused by hydraulic instabilities and mechanical coupling to vibrating equipmen Within the last year, policy changes regarding strict adherence to operating procedures were put into effec A review by the licensee's root cause determination team of the station operating procedures did not reveal procedural steps that describe/define power level actions to be taken regarding condensate/feedwater system operation to support the feedwater hydraulic system as it is currently designe This review also determined that neither operating procedures nor operator actions were likely root causes to the even The AIT concurred with the licensee's observatio The significance of the above observations are as follows:

Feedwater Control System Replacement The fact that the control system was changed from an analog (GEMAC) to a digital (Bailey) system is.not, in itself, considered by the licensee to be a significant contributor to the transien However, the potential problem areas concerning the control system replacement may be in the interface with the portions of the old (GEMAC) system, such as the E/P transducer, the Moore valve positioner and other components in the pneumatic system, which were not change The licensee has added instrumentation to monitor these components during the feedwater system performance testin Feedwater Pump Rotating Element Replacement The replacement of the 113A 11 RFP rotating element in mid-1986 was performed to minimize seal leakag Since the rotor was quite large, the Dresden maintenance shop did not have equipment to accommodate this

rotor for a dynamic balancing and only a static balance and alignment was performe The licensee 1s analysis of the pump 1 s vibration curves revealed high amplitudes at the five vane-pass frequency, nominal frequency and the frequencies associated with the pump-motor gear box couplin System Hydraulic Operation The licensee 1s analysis of the latest available pump operating curves also revealed that at the 20-25% power condition, where the two feedwater transients occurred, both the reactor feed pumps and the condensate/

condensate booster pumps are operating in conditions where low flow through the pump could result in the development of cavitatio Initially, Dresden had been designed in the late 1960 1 s for a minimum flow condition on the RFPs of approximately 10% capacit Consequently, the minimum flow recirculation lines had been designed for only 10-15%

of the maximum flow through the pump based upon only short periods of operating in the low flow condition However, in the early 1970 1s, the pump manufacturer and Dresden were made aware that in nuclear plants, unlike in fossil plants, the feedwater system is operated at low flows for extended periods during startups and shutdown Tests and studies were conducted to determine the continuous safe minimum flow through the RFPs, rather than the temporary minimum flow It was determined that the minimum flow conditions corresponded to 20-25% flow, rather than 10-15% flo At Dresden Unit 3, the feedwater pumps are required to operate for greater than four hours at flows less than 20%.

The setpoints and line sizing for the minimum flow recirculation path for these pumps are designed for less than 10% flow requirement Continuous operation during these conditions for sustained periods can lead to cavitatio Such cavitation would be observed on the suction and discharge pressure gauges for the pumps, in the form of large swings of the indication deflection (conditions that were observed during both transients).

The licensee 1s composite curve of feedwater pump capacity (total developed head for the RFP discharge header) shows that cavitation could occur at less than 28% flow when one RFP is running in conjunction with two condensate/condensate booster pumps (see Attachment 8).

Regulating Valve Design The licensee 1 s investigative team determined that the designs of both the 11 3A 11 and 11 38 11 FRVs and actuators are not ideal for low flow condition These designs are partially based upon the original plant design for the specified duty conditions given to the valve manufacturer and are not in agreement with the current actual duty conditions of the syste The maximum differential pressure and flow conditions across the valve at low flow are higher than originally specifie At low flow with these conditions, the forces on the valve plug are quite large, acting in the upward stem directio This would have the effect of fighting the actuator pi~ton pressure.

In addition, the licensee has determined that the operation of the pneumatic controls on the valve actuator is susceptible to vibration

._...

condition The pneumatic controls are comprised of spring and diaphragm combinations (the Moore valve positioner), that are sensitive to vibration and could cause valve lock-out on low air pressure (which was observed during both transients).

As previously described in paragraph III.D., the licensee found two abnormalities on disassembly of the 3A FRV which were considered for potential root cause The loose yoke on the upper internals was determined not to have been loose following the event of 7/11/8 Further, evaluation of the possible effect of this condition concluded that while p6sition indi~ation may be interfered with it was unlikely that the loose yoke would alter valve operation. The disassembled range spring was determined to have the ability to alter valve operation in that the expected response to the disconnected spring would be for the FRV to want to open furthe The licensee's root cause team, however, believes the spring may have exaggerated the event but doesn't believe the spring by itself would have been sufficient to cause the even By the time of the AIT exit, the licensee's root cause team had concluded that no specific root cause could conclusively be assigned to this event, but, rather, several possible causes were identified which. separately or acting in combination with each other may have produced the transients observed. Therefore, the licensee proposed to perform certain modifications and.then startup the unit to conduct extensive feedwater operations testing (described in paragraph V of this report).

The AIT agreed with the licensee's determination of no conclusive root cause and the proposed testing. The existing CAL was subsequently terminated via letter, A. B. Davis to Cordell Reed, dated September 4, 1987, (Attachment 2) in order to allow the licensee.to precede with the test progra The licensee has since issued LER 87-013-0 (Attachment 9) which provides further information on this even Subsequent to the AIT exit, the NRR contractor, Idaho National Engineering Laboratory, postulated that a specific type of failure of the check valve just upstream of where HPCI and RWCU connect to the 118 11 feed line could have caused the type of phenomena observed as a result of the August 7, 1987 transient. The specific failure would be a separation of the valve disc such that at most flows the disc would lie undisturbed in the bottom of the valve, but at the 25-30% flow regime the turbulence could be enough to pick up the disc and cause it to block the outlet port. This blockage would then cause an almost instantaneous water hammer (pressure spike) which could have resulted-in the damag The observed oscillations and system parameter traces would then be indicative of a system trying to stabilize itself after the even As flow was secured, the disc would then slide back into the valve body and disappear as a cause until the particular flow pattern was again established. While this theory does explain same of the observations, such as the bent 24" riser clamp, it also relies on several postulations for which there is no corroborative data. Since this theory was not advanced until the license was ready to startup, it was not explored thoroughly by the AI However, should a check valve separation be the actual root cause, the feedwater system should be sufficiently instrumented during the testing to detect and pinpoint the

• *

location of such a failure. Consequently, the plan of starting up the unit and performing the test program was not altered by the check valve theor Modifications and Testing Since the licensee's hypothesis of the cause of the transient is the possible combination of the observed conditions noted in the previous section, the FRVs actuators have been modified to dampen the internal hydraulic forces {Attachment 6, Figure 3) by installing a large mechanical spring to counter the upward direction force and minimize the effect of the hydraulic lift on the actuator piston pressur In addition, a test program has been developed to facilitate the root cause determination of the transients and to provide data that could be used to evaluate the design of the completed modifications and any additional modifications to be made on a long term basis. These long term modifications also include plans to replace the 3B FRV with a low flow stable drag valve, to improve the existing Unit 2 2B drag val~e and modified "3B" FRV with electro-hydraulic control (EHC) actuators to provide improved high flow stability and to modify the existing 2A FRV with a hydraulically dampened actuato *

To facilitate the testing, the licensee has purchased a computer based monitoring system from Bailey Controls that will interface with the existing Network 90 syste The hardware (computer and engineering display station) for the computer-based monitoring system is on sit Software programs are being developed by Bailey Controls for using the combined systems as an events recorder to aid in diagnosing the cause of feedwater transients. The tentative parameters to be monitored are listed in Attachment 7, Table 2, "Tentative List of Network 90 Parameters to be Monitored."

As previously stated, the licensee has implemented an extensive monitoring system to assure that proper and timely corrective actions are taken in the event of any unusual behavior of the feedwater system during the restart program for Dresden Unit The variables to be continuously monitored not only include those of the feedwater system but also those

. which interface or can interact with the performance of the feedwater syste The variables to be monitored and the type of instruments being used are list in Attachment 7, Table 3, "Test Point List, Dresden Unit 3 *II The objectives of the feedwater operability performance test are to collect and evaluate data to determine the root cause{s) of the feedwater transients, evaluate the ability of the feedwater and condensate booster recirculation flows to maintain the pump flow in the stable region (no cavitation) of the pump operability curves, develop operating limits for RFP flows, recirculation valve operation, pump or valve configuration and to monitor the operation of the digital (Bailey) feedwater level control syste *

To achieve these objectives, data will be collected from the feedwater system in the power range between 15 and 100% during the tuning of the feedwater level control system and the characterization of the new hydraulically dampened FRV Power level will then be reduced to 15% and a series of power changes between 15 and 35% will be made with different combinations of pumps and valve An additional test section will verify the new FRY response during valve lock-out conditions. The final phase of the testing will be to reproduce the initial conditions of the July 11 and August 7, 1987, events to evaluate the design of the modifications and to collect data for root cause determination.*

The results and performance of the scheduled testing.will be monitored and evaluated by the NRC during the Unit 3 startu V AIT Conclusions The AIT finds that in the absence of a definitive determination rif a root cause or causes for the feedwater transients of July 11th and August 7th at Dresden Unit 3, the licensee's evaluation of the potential contribu-tions by the feedwater and feedwater control systems has been virtually exhaustiv As noted in paragraph III.c the walkdowns and damage assess-ment were initially not thorough. Subsequent walkdown efforts were judged satisfactor As noted in paragraphs IIIA and IIIB the AIT *

believes the performance of Operations and Health Physics personnel was excellent. The AIT also finds that the post event actions in this regard such as the initial investigations and testing of the feedwater control system responses were reasonably performe The AIT believes the likely causes of this event involve the interactions between the feedwater control system, the feedwater regulating valves and the feed system hydraulic instabilities. As discussed in paragraph IV, in exploring the root cause possibilities, the licensee uncovered conditions where cavitation is possible in the feed system and inadequacies in the feedwater regulating valve design with respect to current usage. These findings need to be explored further by the licensee for long term corrective actio Finally the AIT concludes that Dresden's restart program provides reasonable assurance that the public's safety will not be exposed to an undue ris VII. Inspector Findings Associated With FRV Maintenance During the course of the AIT, the inspector reviewed the Copes-Vulcan technical manual for the FRVs and determined several inadequacies as follows:

Many of the drawings referenced by the manual as pertaining to Dresden, are incorrec The "Valve Assembly" drawing, B-130920 shows the positioner as manufactured by Bailey when, in fact, a Moore manufactured positioner is currently used in the pneumatic valve actuator. Also, the drawing does not indicate the correct configuration of the pneumatic system with the Moore booster relay Drawing B-142063 also shows the incorrect positioner and the incorrect physical mounting of the pneumatic actuator component \\

0 The installation and maintenance instructions for the ASCo

_(manufacturer) three-way direct acting solenoid valve indicated the wrong model (G) for the 2A, 3A, and 3B FRV The correct model is a Form "F".

The manual recommended preventive maintenance has not been performe Review of maintenance history revealed that the valve actuator cylinder head had not been checked for air leakage, the piston sealing rings and valve fittings had not been inspected, the positioner had not been periodically inspected and the three-way solenoid valve had not been cleaned on a periodic basi These observations are considered to be an open inspection item (249/87028-02) in that the inadequacies of the manual do not accommodate good maintenance practices and that a lack of a preventive maintenance program could degrade the performance of the FRV However, the inspector was not able to confirm that any of these conditions contri-buted to the transien VIII. Open Items Open items are matters which have been discussed with the licensee, which will be reviewed further by the inspector, and which involve some action on the part of the NRC or licensee or bot Open items disclosed during th inspection are discussed in Paragraphs III & VI I Exit Interview (30703)

The inspectors met with licensee representatives (denoted in Paragraph I.D)

informally throughout the inspection period and at the conclusion of the inspection on August 25, 1987, and summarized the scope and findings of the inspection activitie The inspectors also discussed the likely informational content of the inspection report with regard to documents or processes reviewed by the inspectors during the inspection. While the inspectors did review some proprietary material, none of the areas expected to be contained in the report were identified by the licensee as proprietar The licensee acknowledged the findings of the inspectio