ML20237E631

Special Rept 87-014:on 871014,hydrogen Pressurizer Ignited. Caused by Inadequate Pressurizer Draining & Depressurizing Procedure.Procedure Review & Rev & Control Room Personnel Training Planned

ML20237E631
Person / Time
Site:	Wolf Creek
Issue date:	12/18/1987
From:	Withers B WOLF CREEK NUCLEAR OPERATING CORP.
To:	NRC OFFICE OF ADMINISTRATION & RESOURCES MANAGEMENT (ARM)
References
87-014, 87-14, WM-87-0329, WM-87-329, NUDOCS 8712290099
Download: ML20237E631 (6)

Text

-- .

4 Wi$LF CREEK NUCLEAR OPERATING CORPORATION Bart D. Withers President and Chief Executwo Officer I

December 18, 1987 WM 87-0329 U. S. Nuclear Regulatory Commission ATTN: Document Control Deck Washington D. C. 20555 V

Subject:

Docket No. 50-482: Special Report 87-014 Gentlemen:

i The enclosed Special Report is submitted as a voluntary report concerning a hydrogen burn in the pressurizer.

Very truly yours Bart D. Withers President and Chief Executive Officer BDW/llk co: J..E. Cummins (NRC)

R. D. Martin (NRC)

P. W. O'Connor (NRC), 2 8712290099 871218  %

PDR ADOCK 05000482 S DCD /p/,}

P.O. Box 411/ Burlington, KS 66839 / Phone: (316) 364-8831 / g\

An Equal Opporturvty Employer M F/HCNET

, l kttachment to WM 87-0329

. 'Page 1 of 5 December 18, 1987 WXF CREEK GENERATING S'IATION SPECIAL REPORP 87-014 SPECIAL REPORT CONCERNING HYDROGEN BURN IN PRESSURIZER DUE 'IO PROCEDURE INADEQUACY DESCRIPTION OF EVENT At 0915 CDT on October 14, 1987, a utility welder struck an arc which ignited hydrogen inside the pressurizer. The resultant hydrogen burn created a very loud noise that lasted about five seconds. At the time of the event, the plant was in Mode 6, Refueling, with the refueling pool full and the core partially unloaded.

During the plant cooldown and depressurization, the water level in the pressurizer was raised to abaut 95 percent cn October 2,1987. (See Table I for a detailed sequence of events.) The pressurizer level was then lowered to a level correspoMing to one foot below the reactor vessel flange; this emptied the pressurizer. During the draining, nitrogen was introduced to prevent formation of a vacuum in the pressurizer. Upon entry into Mode 6, )

on October 6, pressurizer safety valve testing was started. While a cafety L valve was removed for testing, a piece of sheet metal aM a plastic cover was taped over the flange to preserve internal cleanliness.

Followirg flooding of the reactor cavity to 23 feet above the reactor vessel flange (pressurizer level about 25 percent), work was started on the replawment of the three upper root valves for the pressurizer level instrumentation. All three lines were first cut and the valves and piping assemblies were renoved on October 12, 1987. The first valve aM piping assembly was reinstalled and welding completed at about 0800 CDT on October 14, 1987. The welder then noved to the next valve assembly and ocmnenced tacking the 3/4 inch socket weld. Upon striking an arc for the second tack, the hydrogen ignited.

As the hydrogen burned, its pressure was relieved through the temporary cover where the pressurizer safety valve had been renoved. The noise of the burn wan heard throughout containment and in the Control Room via the refuelirg ccmnunications circuit. A Control Rocxn licensed operator noted a small spike on a pressurizer level recorder. A survey of the vicinity of the pressurizer revealed to change in radiaticn levels. Subsequent sampling for flanmable gasses found none around or within the pressurizer.

CAUSE OF EVDTP The event was caused by the acutining of a flanmable mixture of hydrogen and oxygen and an igniticn source (the arc). Subsequent analysis has shcun that the procedure which first raises the pressurizer level while collapsing the bubble, then empties the pressurizer while supplyire nitrogen, can leave as nuch as nine percent hydrogen in the pressurizer. The renoval of the safety valves ard instrument root valves allowed the introduction of air to produce the flanmable mixture.

1 ,

.Attachntent to hw 87-0329

- Page 2 of 5 December 18, 1987 l

In depressurizing the pressurizer, the level is raised to near the top of the pressurizer level indication range. The upper taps for the level i instruments are located approximately three feet below the top of the l pressurizer. In this volum would he a mixture of water vapor and l hydrogen. The root cause of the event was the procedure which drains and l depressurizes the pressurizer did not adequately remove the hydrogen from j the pressurizer.

The administrative procedure, " Control of Ignition Sources", requires a check-off indicating that closed equipnent such as tanks be purged of l flanmable gasses. A contributing cause to the event was that this procedure does not specifically require a flanmable gas sample to be taken. Since the pressurizer had been filled with nitrogen during the draining, to flammable mixture was believed to be present. Based on this, a Shift Supervisor told the utility welder that filled out the Ignition Source Permit for the pressurizer root valve replacement that the pressurizer had been purged with nitrogen gas.

'Ihe procedure for renoval of the pressurizer safety valves required, in the

" Precautions and Limitations" section, monitoring for the presence of

! hydrogen gas. However, this requirement was not specifically reiterated in the body of the procedure, and therefore a gas sample was not taken. It is recognized that the welding activities and the remwal of the safety valve were not directly related. However, had the gas sample been taken, the presence of hydrogen gas might have been detected prior to the incident.

Therefore, this procedure is considered related, but not causal.

ANALYSIS OF EVENT During normal operation, hydrogen is added to the reactor coolant. Over long pericds of operation, the water in the pressurizer achieves a chemical equilibrium with the Reactor Coolant System (RCS). Above the water level in the pressurizer is a vapor space that is filled primarily with steam, but does contain some ron-condensable gasses (primarily hydrogen) in equilibrium with the high temperature water.

It should be noted that during plant cooldown and depressurization there is little exchange of fluid between the pressurizer ard RCS. Therefore, the pressurizer may have a significantly higher concentration of hydrogen  !

dissolved in the fluid than does the RCS. The hydrogen gas in the vapor space has no ready means for renoval.

3n analysis of the event was conducted by Westinghouse to determine if the system loadings causal by the hydrogen burn exceeded the design loading.

Since it is not possible to specifically determine the composition and the anount of the hydrogen and air that was present at the tine of ignition, the following assumptions were made for the bounding case:

1. The pressurizer vapor space contained 100 percent hydrogen prior to draining.

. Attachment to WM 8'i-0329

. Page 3 of 5 December 18, 1987

2. All of the nitrogen added during draining collected at the botton of the vapor space, creating a pocket of pure hydrogen at the top.

3. During safety valve renoval and root valve replacement, the anount of air necessary to burn all of the hydrogen present entered and completely mixed with the hydrogen. During air entry, only the nitrogen was displaced; all of the hydrogen remained.

Ignition of the resultant mixture was analyzed using the Westinghc>use 0]MPACT code. The results of the analysis slow a peak gas pressure of 61.9 pounds per square inch absolute (psia) was reached 0.505 seconds after ignition. The analysis indicates the temperature of the inner surface of the pressurizer wall started at 100 degrees Fahrenheit and peaked at 121.5 degrees Fahrenheit at 2.02 seconds after ignition.

The results were loser than the design pressure of 2500 psia and temperature of 680 degrees Fahrenheit. The peak thermal stress resulting fran the 21.5 degrees Fahrenheit temperature rise is less than 6500 psi. %is stress when '

m mbined with the pressure stress is well belos the fatigue endurance limit. %us, the contribution to fatigue is negligible. We resulting stresses fran this event were substantially less than all American Society of Mechanical Engineers (ASME) Code stress allcwables for the pressurizer shell an3 nozzles.

CORRECTIVE ACTIONS Welding of the root valves was initially suspended until an investigation and evaluation could be conducted. Radiation surveys and flanmable gas samples taken after the event indicated normal readings.

A walkd>an of the pressurizer, piping and supports was conclueteS on October 24, 1987, by utility engineering personnel. W ey found no specific ananolies that could be linked to danage caused as a result of the hydrogen burn event.

We pressurizer spray nozzle was visually examined on October 29, 1987, by utility Quality control personnel. This exannnation found no loosa parts, no evidence of cracks nor loss of integrity at the welded connections.

The pressurizer emersion heaters were tested for insulation resistance and continuity on October 29, 1987. W e results were satisfactory. he pressurizer level transmitters were checked on October 29, 1987. Two of the three narroa range transmitters were recalibrates. Wese results were consistent with previous calibrations. The pressurizer safety valves were renoved and inspected. hey were not damaged by the heat generated during the hydrogen burn.

. Attachment to WM 87-0329

. Page 4 of 5 December 18, 1987 The procedure for depressurizing and draining the pressurizer will be reviewed to determine the best nethod of renoving the hydrogen from the pressurizer. This procedure will be revised accontingly ard tle Control Pcan personnel will be trained cn the revisions. This will be cmpleted prior to the next time the procedure is used.

'Ihe procedure for control of ignition sources has been revised to require a check for combustible gasses prior to welding on closed or poorly ventilated tanks. Personnel authorized to approve Ignition Source Permits are being trained on the new requirements.

The procedure for renoving the pressurizer safety valves has been revised to include nonitoring for hydrogen within the bcdy of the procedure.

ADDITIONAL INFORMATION There was to damage to plant equipment or release of radioactivity as a result of this event. At no tine did corditions develop that may have posed a threat to the health and safety of the public.

There have been to previous similar events.

l l

l l

1 l

l i

5

' Attachment to WM 87-0329 Page 5 of 5

. December 18, 1987

'IABLE 1 SDQUENCE OF LVINIS Date Time Event 10/02/87 1949 Ccmnencing RCS pressurizer draindcun 10/02/87 2347 Reactor vessel level approximately one foot belos flange 10/05/87 Isolated and tagged out pressurizer PORV, safety and spray valves 10/06/87 1040 Entered mode 6, refueling 10/06/87 Renoved Pressurizer Code Safety 8010A (flange covered and loop seal drained) 10/07/87 2300 All reactor vessel studs detensioned.

10/08/87 1224 Lifted reactor head 10/08/87 Flood reactor cavity 23 feet above reactor vessel flange; pressurizer filled to 25 percent 10/08/87 Reinstalled Pressurizer Code Safety 8010A 10/10/87 Renoved Pressurizer Code Safety 80108 (flange sealed arxi loop seal drained) 10/12/87 day Cut lines for pressurizer level instrument root valves shift V099, V102, and V108 10/13/87 day Weld root pass for V099 coupleted shift 10/14/87 ~0800 Carpleted weld for V099 10/14/87 ~0915 Tack welded V102. This apparently caused hydmgen ignition.

_ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ - -____-