ML20212H506

Ro:On 860708 & 09,start Failures of Emergency Diesel Generator 14 Occurred.Caused by Air in Fuel Lines & Personnel Error by Not Resetting Exciter.Fuel Oil Sys Purged & Personnel Educated Re Correct Procedures

ML20212H506
Person / Time
Site:	Fermi
Issue date:	08/07/1986
From:	Lenart R DETROIT EDISON CO.
To:	James Keppler NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III)
References
NP860387, NUDOCS 8608140088
Download: ML20212H506 (7)

Text

. .

Om3 Robert S. Lenart Plant Mzmager Fermi-2 Edison EEE** 1 r a:1.

August 7, 1986 NP860387 PRIORITY ROUTING bb Ekd*"

kc EIC h

6 Mr. Jamcc G. Koppler Regional Administrator Q-na Region III FILM-U. S. Nuclear Regulatory Commission 799 Roosevelt Road Glen Ellyn, Illinois 60137

Reference:

Fermi 2 NRC Docket No. 50-341 NRC License No. NPF-43

Subject:

Emergency Diesel Generator Start Failures Please find enclosed our report on two start failures of Emergency Diesel Generator (EDG) 14 on July 8 and 9, 1986.

These start failures were determined to be invalid tests as defined in NRC Regulatory Guide 1.108, Section C.2.e. This report is submitted to you in accordance with Fermi 2 Technical Specification 4.8.1.1.3 and 6.9.2.

If you have any questions regarding this report, please contact Mr. Lewis Bregni, Compliance Engineer at (313) 586-5313.

l Sincerely,

/

u~V 1

cc: M. D. Lynch l

W. G. Rogers yQ l

G. C. Wright USNRC Document Control Desk

' l l Washington, D. C. 20555 8608140088 860807

{DR ADOCK 05000341 PDR AUG 1119E t

e-Mr. James G. Keppler August 7, 1986 NP860387

.Page 2

-ENCLOSURE 1 Emergency Diesel Generator Start Failures

1. Emergency Diesel Generator Involved:

EDG 14

2. Identification of Failures:

Failure Dates: July 8, 1986, 2138 hrs July 9, 1986, 1826 hrs

Description:

These were the first and second start failures of EDG 14 since issuance of the Fermi-2 operating license (OL). Before these two start failures, there had been twenty (20) previous valid test starts since the OL was issued. The July 8 and 9 start failures are considered failures in invalid tests under Section C.2.e of NRC Regulatory Guide 1.108, primarily because operating error was involved in both start failures. In addition, the July 9 start failure involved failure to reset the exciter, a feature automatically reset during emergency starts.

3. Cause of Failures:

3.a. July 8 Start Failure EDG 14 had been out of service for approximately two weeks prior to July 8, 1986 for maintenance surveillances required every eighteen months by Fermi-2 Technical Specifications. Post-maintenance testing required a fast-start of the EDG.

The EDG air start system commenced its startup cycle, but the EDG tripped on a " start failure" approximately seven seconds after the start signal was initiated. The trip was initiated by the timing relay which terminates a diesel start if the EDG does not achieve 250 RPM within seven seconds after a start signal.

-w p- -,-~r--- - -

w -.m,-e w --w.,. -p ,- ---------m ,.w-- - , - - --

9 Mr. James G. Keppler August 7, 1986 NP860387 Page 3 During the 18-month surveillance, Maintenance personnel had performed work on EDG 14's fuel oil system. This work introduced air into the fuel oil system at the two fuel oil filters and at the tubing between the injector pumps and the injectors. During the annual' preventive maintenance (performed at the same time), air was also introduced at the strainers to the engine-driven fuel oil pump, the motor driven fuel oil pump, and both fuel oil transfer pumps.

The fuel lines must be refilled with fuel before the EDG is restarted. Plant Technical personnel informed the Nuclear Shift Supervisor (NSS) that the EDG fuel oil system had to be purged and primed before starting the EDG. The NSS was informed that it was necessary to run the motor-driven standby fuel oil pump (SFOP) until the fuel oil pressure stabilized and to run the pump for one minute after starting the EDG.

At shift turnover, the offgoing NSS did not communicate this to the oncoming NSS. The oncoming shift ran the motor-driven SFOP for 5-10 minutes, but turned it off before starting the EDG. Running the motor-driven SFOP circulates fuel through the fuel oil system and back to the fuel oil day tank. Air upstream of the engine driven pump is not purged by this method. The fuel filter also was not vented.

When the EDG was given a start signal, the starting air system rolled the engine, but the engine failed to achieve sufficient speed in the first seven seconds because air in i the fuel lines prevented normal flow of fuel to the cylinders. The motor-driven SFOP did not purge the air l and provide pressure on the fuel header before the l

cranking time control relay timed-out. This sequence of events was confirmed by contact with the EDG vendor representative.

The operator at the EDG local control panel was then I informed that the motor-driven SFOP needed to be operated j prior to starting the engine and for one minute after I startup. Subsequently, the local operator ran the SFOP as prescribed and the EDG started correctly.

l l

L _.

Mr. James G. Keppler August 7, 1986 NP860387 Page 4 During starts of the other EDGs, the SFOP was run prior to and after the EDG start signal was initiated. The other EDGs started succescfully after their l maintance/ surveillance outages, verifying that there was  ;

no problem with air in their fuel systems.

The root cause of the July 8 start failure of EDG 14 was an insufficient purge and prime of the fuel oil system after maintenance. This failure would not have occurred if the operator at the EDG had received proper information from the previous shift regarding operation of the I motor-driven SFOP during the engine start. Because of this human error, this start failure is considered an invalid test.

3.b. July 9 Start Failure EDG 14 tripped on " field failure" during a fast start surveillance test on July 9 at 1826 hrs. The EDG was fast started from the local control panel and was operating at nominal speed (900 RPM) for appproximately 30 seconds before it tripped on " field failure". The EDG tripped on

" field failure" because there was no excitation or field voltage. After communicating with personnel in the main Control Room, the operator at the EDG immediately realized that the exciter had not been reset after the previous engine run.

The exciter should have been manually reset during the EDG shutdown immediately preceding the start attempt. In addition, the operator should have verified exciter reset during the prerequisites for the surveillance run. In this event, the exciter was left tripped. As a result, the time delay relay (40A) tripped EDG 14 approximately 30 seconds after the engine started because no excitation voltage developed across the generator field. After the trip, the local operator manually reset the exciter. The EDG was then started successfully.

p

Mr. James G. Keppler August 7, 1986 NP860387 Page 5 At present'there is no indication at the EDG local control

-panels to inform the operator that the exciter is not reset.- However,. there are;two-alarms in the Control Room which indicate the exciter is not reset. During planned

" slow" starts, the two Control Room alarms ("Div II EDG 14 Exciter Trip", Annunciator 10D12; and "Div II EDG 14 Not Ready For Auto Start", Annunciator 10D8) are present because the exciter remains shutdown until engine speed reaches 900 RPM. The EDG procedure (24.307.17, "EDG 14 Start and Load Test") requires the operator to verify that there are no alarms present at the local EDG control panel or in the Control Room before starting the EDG.

Investigation of this event revealed that the Control Room Nuclear Supervising Operator (NSO) observed the " Exciter Trip" alarm and the "EDG Not Ready For Auto Start" alarm were annunciating in the Control Room before the attempt to start EDG 14. Per procedure, the local operator at the EDG control panel verified that no alarms were present at his panel.

Per procedure, the Control Room NSO informed the local operator that no " abnormal" alarms were present in the Control Room. The Control Room NSO thought that test i personnel were preparing for a slow start as originally l planned. Unbeknownst to him, plant Technical personnel had decided to repeat the " fast" start portion of j procedure to verify the results of the previous l surveillance run. -The Control Room NSO thought a " slow" start was underway. The slow start in performed using a different Attachment to the same procedure, 24.307.17.

The Control Room NSO expected to see the two alarms which l were present in the main Control Room during a " slow" l

start. Thus, the exciter remained deenergized when the EDG l

start signal was given. This led to the EDG trip on field i failure.

l The July 9, 1986 start failure on EDG 14 is considered an

, " invalid test" start failure under Regulatory Guide 1.108 because:

(1) The circuit logic for the EDG would have automatically reset the exciter if an emergency start signal had been received. Also, the field failure trip is bypassed during emergency starts.

I

l\

Mr. James G. Keppler August 7, 1986 NP860387 Page 0 I

i I

(2) The miscommunication between the Control Room NSO and '

test personnel was the result of human error. Personnel involved in the test erred in not ensuring that all participants acknowledged that a " fast" start was being undertaken.

4.a. Corrective Measures Taken For the July 8, 1986 Start Failure l The maintenance procedure for work on the EDG fuel system is being revised to require the fuel system to be vented -

and purged of air after maintenance on the fuel system.

In addition, plant procedures will be revised to require a "no-credit" run of the engine (considered troubleshooting

, and, therefore, not a valid test, under Regulatory Guide 1.108, Section C.2.e.7) after maintenance on the fuel system. This will include operating the motor-driven SFOP until the fuel oil pressure stabilizes (usually about five minutes) before starting the engine, and will include running the SFOP for one minute after starting the engine. After completion of the "no-credit" run, the engine will be shutdown and the surveillance " credit" run will be performed. The "no-credit" run will allow plant personnel to inspect the engine for fuel leaks and other abnormal conditions before performing a survellance credit run. The "no-credit" run will not significantly increase the number of starts experienced by the EDGs because of the low frequency of the maintenance that requires draining, venting, and purging the fuel oil system.

4.b. Corrective Measures Taken for the July 9, 1986 Start

! Failure:

This report will be placed in required reading for l licensed and non-licensed operators to ensure their awareness of the causes of this event. In addition, a plant modification (Engineering Design Package 1633) has been developed which will provide " Exciter Tripped" indication at the local EDG control panel. This modification will be installed during a future outage.

i l

Mr. James G. Keppler i August 5, 1986 NP860387 Page 7

5. Lenath of Time Unit Unavailable:

EDG 14 was out of service for the maintenance / surveillance activities and remained out of service until after the July 9 start failure. Therefore, unit unavailability does not apply.

6. Current Surveillance Test Interval:

The current surveilance test interval is once per thirty

, one days.

7. Verification of Conformance of Test Interval:

Because these two start failures are considered " invalid tests", as defined by Regulatory Guide 1.108, the EDG surveillance test interval is not affected. The current surveillance test interval is in conformance with Regulatory Position C.2.d of Regulatory Guide 1.108 and with Fermi-2 Technical Specification Table 4.8.1.1.2-1.

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