ML20204J191

Forwards Rev 1 to Event Rept 98-17 Re Fire Which Caused Damage to Cascade Equipment Containing Radioactive Matl in X-326 Bldg Side Purge Area at Portsmouth Gaseous Diffusion Plant.Revised Rept Includes Root Cause & Corrective Actions

ML20204J191
Person / Time
Site:	Portsmouth Gaseous Diffusion Plant
Issue date:	03/19/1999
From:	Jonathan Brown UNITED STATES ENRICHMENT CORP. (USEC)
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
GDP-99-2014, NUDOCS 9903300021
Download: ML20204J191 (17)

Text

__ _ _ _ _ _ _ _

j s

USEC A om sn.rsy comp.ny March 19,1999 GDP 99-2014 U. S. Nuclear Regulatory Commission Attention: Document Control Desk Washington, D.C. 20555-0001 Portsmouth Gaseous Diffusion Plant (PORTS)

Docket No. 70-7002 Event Report 98-17, Revision 1 Pursuant to 10CFR76.120 (d)(2), Enclosure 1 provides a revised 30 day Event Report for an event that resulted from a fire which caused damage to Cascade equipment containing radioactive material in the X-326 Building Side Furge Area at the Portsmouth Gaseous Diffusion Plant. The revised event report includes the root cause and corrective actions. Enclosure 2 is a list of commitments contained in this report. Due to the extensive revision of the report, vertical lines have been omitted.

This repor.eplaces the previous report in its entirety.

Should'you require additional information regarding this event, please contact Scott Scholl at (740) 897-2373.

Sincerely, w

JA j

J. Morris Brown Gencal Manager Portsmouth Gaseous Diffusion Plant s

F.nclosures: As Stated cc:

NRC Region III Office NRC Resident Inspector - PORTS

~

9903300021 990319 PDR ADOCK 07007002 C

POR._

I!O. Nx 800. Ponsmouth. OH 45661 Telephone 740-897-2255 Fax 740- M7-2644 http://www.usec.com Offices in Livern. ore, CA Paducah. KY Portsmouth, OH Washington. DC U-

m I

3 GDP 99-2014 Page1of15 i

Event Report 98-17, Revision 1 Description of Event' On December 9,1998,'at 0610 hours0.00706 days <br />0.169 hours <br />0.00101 weeks <br />2.32105e-4 months <br />, Fire Services personnel responded to a fire which occurred in Cell 25-7-2 in the Side Purge Area of the X-326 Process Building.. The fire, which breached process equipment in several places, was contained by the building sprinkler system and was extinguished I

' by plant Fire Services personnel. The fire damage was primarily limited to the Cell 25-7-2.

equipment, housing, and thrn ages of adjacent Cell 25-7-4. Since the fire damaged process l equipment and piping which contained radioactive material, this event is reportable in accordance with 10CFR76.120(c)(4). A description of the event is provided below.

' Operations were normal on the moming of December 9,1998. Cell 25-7-2 was the bottom onstream cell of the Side Purge cascade. The purpose of the Side Purge is to vent most of the light gases, such as oxygen and nitrogen, from the cascade. Heavier intermediate weight gases, such as coolant (R-114), continue up the cascade until they are vented by the Top Purge equipment. Low speed Cell 25-7-2 is used to create a stripping section. The B-Stream exiting Cell 25-7-2 typically contains a low percentage of the low molecular weight " light" gases which enter the Side Purge. The Side Purge configuration was normal with the exception that Cell 25-7-4 was offstream/ bypassed and out of service. A-Stream process gas flow was into Stage 4 A-suction of Cell 25-7-2, from Stage 12 of Cell 27-1-2, through a manual block valve. B-Stream process gas return flow was from Stage l_ of Cell 25-7-2, to the Stage 1 A-Stream of Cell 27-2-1. A coolant " bubble" had moved through the purge cascade at about 0330 hours0.00382 days <br />0.0917 hours <br />5.456349e-4 weeks <br />1.25565e-4 months <br /> and had been purged. The Unit Operator making rounds of the

' cell floor at about 0600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> stated that Cc!125-7-2 equipment was running quietly with no unusual

. indications.

I At about %05 houw, the ACR-6 (Area Control Room' 6) operator was taking hourly readings wher.

the operator noticed the Cell 25-7-2, Stage 2, Amp meter briefly deflect to about 70% full scale, and then rei+n to the normal position of about 30% scale. The operator told the First Line Manager in j

Training (FLMIT), who was leavhig the ACR, that it appeared there was another coolant " bubble" I

entering the Side Purge cascade. The FLMIT said he would tell the Section Manager when he reported to ACR-5. ' Meanwhile, the operator throttled the Sidt Purge vent remote control valve (CV-1261) t %ck the suspected coolant bubble out of the Side Purge cells.

At about 0606 hours0.00701 days <br />0.168 hours <br />0.001 weeks <br />2.30583e-4 months <br />, the Amp meter in ACR-6 again deflected, which indicated that the Cell 25-7-2, Stage 2, compressor surged and then returned to normal load. Cascade Control personnel in the

'X-300 Plant Control Facility (PCF) observed that the Cell 25-7-2 Amp meter, which indicates the total cell Amps from all six motors, was slowly climbing and called ACR-6 to determine the cause.

~

During the phone call, at 0607 hours0.00703 days <br />0.169 hours <br />0.001 weeks <br />2.309635e-4 months <br />, Cell 25-7-2 loaded up and tripped on instantaneous ground overcurrent before the opeiator could trip the cell motors. Adjacent Side Purge Cells 25-7-6,25-7-8,

]

j

c

\\

j GDP 99-2014 1

Page 2 of 15

)

' Event Report 98-17, Revision 1 and 25-7-10 also loaded, with seal exhaust alarms activating. The operator depressed the Cell-Off-Split button (produces full split in Cell 25-7-2) to close the three open block valves. The expected response of a green light, indicating the block valves had completely closed, did not occur and the a

panel had a red and a green light indicating one or more of the block valves was not fully closed.

.I Even if the block valves had all closed, the cell would not have been completely isolated because

the internal'A-Line Stage 4 manual block valve (AB2S4), used to create a stripping section,

remained open.

l The X-326 Building CC-Shift FLM was in route to the ACR-6, when the FLM heard the cell

' compressors and motors wind down. The FLM then proceeded to ACR-6 and found Cell 25-7-2

~

. shut down. - Isolation valve ESP-7 had closed automatically, isolating the Side Purge vent stream j

from atmosphere.

j

- At about 0610 hours0.00706 days <br />0.169 hours <br />0.00101 weeks <br />2.32105e-4 months <br />, sprinkler system %2, which is located in proximity to Cell 25-7-2, alarmed in the Fire Station and the PCF. Operators in the. nearby X-600 Steam Plant heard the external X-326 Building sprinkler system alarm bell, observed white smoke, smelled an electrical type odor, and heard a mt:ffled pop. A Protective Force Officer on the cell floor heard a roar followed by two POPS.

At 0611 hours0.00707 days <br />0.17 hours <br />0.00101 weeks <br />2.324855e-4 months <br />, Cascade Control told the ACR-6 FLM that there were reports of smoke and flowing i

sprinklers in the south end of X-326. The FLM reported a cell coolant alarm, indicating high coolant temperature or low coolant pressure, but did not report smoke. Fire Services arrived on the scene at 0613 hours0.00709 days <br />0.17 hours <br />0.00101 weeks <br />2.332465e-4 months <br />. By 0615 hours0.00712 days <br />0.171 hours <br />0.00102 weeks <br />2.340075e-4 months <br />, smoke was entering ACR-6. The operator and the FLM did not immediately evacuate because their respirators were stored outside the ACR. Firefighters entered 1

X-326, Door 7, and repor"d heavy, light colored smoke.

At 0617 hours0.00714 days <br />0.171 hours <br />0.00102 weeks <br />2.347685e-4 months <br />, the Section Manager and the FLMIT en-route from ACR-5, wu attempting to reach ACR-6 with four respirators, but encountered heavy smoke and called the PCF for information. The Cascade Controller related that there was a sprinkler 462 alarm, smoke coming out of the building

. and that the UF front had been pushed down the cascade into the X-330 Building. The Section 6

Manager and FLMIT continued to attempt to reach ACR-6 while avoiding the heavy smoke.

1 The Unit Operator, wearing respiratory protection, entered the cell floor and proceeded toward Cell 25-7-2. The operator saw there was smoke coming from the cell and valved off the lube oil supply located near the top of the exit stairway before he exited the cell floor. Smoke was light but slowly

. becoming denser above the cell floor.

l

m 4.- a

_g.

=

GDP 99-2014 Page 3 of15

' Event Report 98-17, Revision 1

-- At 0619 hours0.00716 days <br />0.172 hours <br />0.00102 weeks <br />2.355295e-4 months <br />, ACR-6 personnel evacuated the control room without respiratory equipment by entering the ACR-6 basement and going into the tunnel which provided egress from the area. At 0620 hours0.00718 days <br />0.172 hours <br />0.00103 weeks <br />2.3591e-4 months <br />, Fire Services personnel checked ACR-6 and determined that it had been evacuated. - At

-'0621 hours0.00719 days <br />0.173 hours <br />0.00103 weeks <br />2.362905e-4 months <br />, the Section Manager and the FLMIT entered ACR-6 by a rear door and began a search

. of the ACR to be certain the area was empty. The Section Manager pushed the Cell-Off-Split button.

. and received a green light within seconds, indicating the valves were nearly fully closed by the ACR =

operator on the previous isolation attempt. The Section Manager closed motor operated valve 7 ESP-1, the first isolation valve beyor.1 the Cell 25-7.., Stage 4, manual valve (AB2S4), which

-isolated Cell 25-7-2 from the Side Purge supply line and the operating cascade.

At 0622 hours0.0072 days <br />0.173 hours <br />0.00103 weeks <br />2.36671e-4 months <br />. ACR-4 personnel manually activated the building recall horn. At 0623 hours0.00721 days <br />0.173 hours <br />0.00103 weeks <br />2.370515e-4 months <br />, the Section Manager pushed the automatic recall button located in ACR-6. The recall signal is to alert

' building personnel to report to designated assembly locations. At 0625 hours0.00723 days <br />0.174 hours <br />0.00103 weeks <br />2.378125e-4 months <br />, the Section Manager and FLMIT exited the building and proceeded to the Command Post.

At 0626 hours0.00725 days <br />0.174 hours <br />0.00104 weeks <br />2.38193e-4 months <br />, Fire Services personnel entered the cell floor and reported a fire in progress. Flames were reported around the cell housing and from motor openings. Ileavy black smoke was filling the area making visibility poor. An oil / water mixture on the cell floor created difficult footing for the firefighters. The Fire Services personnel returned to the Command Post and developed a plan to extinguish the fire.

it was later determined that seven sprinkler heads had activated. Three at the rooflevel, two under the bypass housing, and one in the aisle way that most likely activated wh en hot gas jets burned through the cell housing at Stages 1,3 and 4. One under the coolant platform most likely activated j

when the cell housing door panel adjacent to Stage 6 became warped and opened. The high pressure fire water sprinkler discharge was successful in containing the fire.

At 0630 hours0.00729 days <br />0.175 hours <br />0.00104 weeks <br />2.39715e-4 months <br />, Fire f'ervices personnel proceeded to lay hoses on the east side of the building and applied water to the outside of the Cell 25-7-2 housing. Other firefighters gained access to the south door of the building'and proceeded to lay hose.

- At about 0730 hours0.00845 days <br />0.203 hours <br />0.00121 weeks <br />2.77765e-4 months <br />, the south door of the cell housing was opened and the firefighters proceeded to spray. water inside the cell housing. The fire was extinguished and upon examination for hot spots, an open process pipe was observed from the West Side of the cell. Fire fighting activities were curtailed and the firefighters returned to the Command Post. After consultation with the Incident Commander on Nuclear Criticality Safety (N("%oncerns, the firefighters went back to the cell housing and performed cool down operations wir,e a' aiding the open pipe. u '749 hours0.00867 days <br />0.208 hours <br />0.00124 weeks <br />2.849945e-4 months <br />, the sprinkler system was valved off. Total. sprinkler dis Mge was approximately 30,0N vallons. At E

g'

• \\

- Enclosure 1 GDP 99-2014

~y

~,

~

Page 4 of15 Event Report 98-17, Revision 1 0815 hours0.00943 days <br />0.226 hours <br />0.00135 weeks <br />3.101075e-4 months <br />, Emergency Squad personnel valved off the Cell 25-7-2 hydraulic oil supply. The fire was declared out by Fire Services at 0817 hours0.00946 days <br />0.227 hours <br />0.00135 weeks <br />3.108685e-4 months <br />.

~ Fire damage was limited primarily to Cell 25-7-2 and consisted of severe damage to the components in the compressors, the converters, and the R-114 coolers. ; At least one stage control valve disk was melted. Holes were burned in process piping elbows in three stages, several process pipe expansi n

. joints were ruptured, and holes were melted in three stage converters. Minor damage also occurred

to three stages of adjacent Cell 25-7-4.

- The material that outgassed was enriched UF at less than 7.0% U-235. The quantity of material 6

released during the event was estimated as approximately 5 Kg of uranium.

Cause of Event On December 9,1998, an investigation team was established to determine the causes for the event.

Outside technical experts were added to the team to assist in the cause determination. The extensive fire damage experienced by Cell 25-7-2 equipment has made it difficult to determine the root cause.

Much of the equipment has been damaged to the extent that evidence needed to determine the root cause was destroyed. The conclusions reached by the team are discussed below and are based on the best available evidence.

The investigation team has concluded that the direct cause for the event was most likely rubbing of internal compressor parts which initiated an exothermic reaction. Friction, resulting from rubbing

.of compressor parts, is believed to have generated enough heat to reach the melting point of aluminum. The molten aluminum then chemically reacted with the UF process gas, generating 6

additional heat. As the cell continued to operate onstream, additional UF was available to feed the 6

reaction and spread it to other stages. At least one gas cooler eventually ruptured releasing coolant into the cell. The release and expansion of the coolant into the cell increased the cell pressure and generated additional heat as aluminum chemically reacted with the coolant. The high temperatures and elevated pr.6sure led to the destruction of the converter tube bundles and breach of the cell

. piping and converters.

A chemical analysis of the predominant Cell 25-7-2 exothermic reaction products concluded that they reculted from aluminum reacting with UF and R-114 coolant at temperatures which likely 6

exceeded 3600 F. The analysis also concluded that most of the heat generated during the event resulted from the aluminum /R-114 coolant reaction.

l E_

4

F 1

w j

GDP 99-2014 Page 5 of 15 Event Report 98-17, Revision I The investigation team also reviewed plant operating conditions that existed prior to the event to j

. determine if there were any abnormal conditions that could have initiated or contributed to an exothermic reaction. Cascade operating parameters for the 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> prior to the time surging was first detected by the operator, such as purge rates, UF front location, feed and withdrawal rates and surge drum bleed back information were reviewed, but did not indicate any activity which could

. have initiated the event. Cascade monitors such as alarms, line recorders, space recorders, oxidant

. monitors and cascade laboratory sample analysis did not reveal any unusual activity. A review of cascade and feed plant valving orders and operational logs also did not reveal any unusual activities that would have initiated the event. Based on these findings, tne investigation team concluded that the exothermic reaction was not caused by any abnormal concentration of gases or unusual operating conditions.

l The investigation team concluded that tne root tause for the event was an equipment failure. The team was not able to identify the exact failure mechanism. However, the most likely failure 1

mechanisms were identified as one of the following: Chemical deposits due to wet air inleakage, compressor impeller failure or foreign material in the process lines.

i Chemical deposits due to wet air inleakage can result in a build-up of chemical reactants in critical 4

clo e tolerance areas of the compressor. The deposits may continue to accumulate until the tolerances close, causing mechanical " rubbing" that generetes sufficient heat to initiate a j

UF/ aluminum reaction.

]

- An analysis of chemical reactants remaining in the cell after the exothermic :caction was performed to determine if deposition of uranium may have occurred. Non-destructive assay (NDA) gamma

scans indicate-that the majority of uranium compounds remaining in the damaged cell are concentrated in the Stage 2 compressor, compressor discharge, and stage cooler, with the next

- highest concentration occurring in the corresponding Stage 1 components. The Stage 1 converter,

. followed by Stage 2 converter, show the highest quantity of uranium of the 6 converters in the cell.

I

. Laboratory analyses of solid grab samples removed from the cell debris show that the Stage 2

-I compressor.:ontains material with the highest uranium concentrations. These findings are consistent

~

with NDA neutron measurements thatlwere also taken on the compressors. The deposition of uranium in the lower stages of the cell wotdd be consistent with air inleakage to the Stage 2

compressor. Based on these results, the investigation team concluded that air inleakage would have most likely_ occurred in the Stage 2 compressor seals or piping expansion joints.

i An inspection of the Stage 2 compressor seals and piping expansion joints was conducted to

. determine where inleakage may have occurred. The compresser seals did not show any signs of

'n GDP 99-2014

{

Page 6 of15 Event Report 98-17, Revision 1 damage which could have allowed wet air to enter the compressor. However, the piping expansion -

joints were found to be damaged.' Metallurgical checks of the expansion joints are being conducted j

. to try to determine if the damage occurred before the event. Due to the heat damage experienced by j

the expansionjoints, the results may not be conclusive and it may not be possible to determine if wet.

I air inleakage was a factor in this event.

1 A second possible cause was the failure of a compressor impeller. Cracks may have propagated through the impeller, changing the dimensions of the impeller such that rubbir.g occurred.

Alternately, the impeller may have separated and rubbing may have occurred between impeller pieces in the compressor and the remaining rotating portion of the impeller. The resulting friction could create enough heat to melt the aluminum impeller vanes and initiate a UF,/ aluminum reaction.

j The investigation team identified compressor failure as a possible cause after it was discovered that

. Cell 25-7-2 had been shutdown in August 1998, as a result of high vibrations in the Stage 4 i

compressor. The compressor was removed from the cell and replaced in November,1998, but was i

not inspected before the cell was restarted. Following the December 9,1998 exothermic reaction, i

a boroscope inspection of the Stage 4 compressor that was removed in November was performed.

The inspection revealed that the Stage 4, first stage impeller vanes, were separated from the impeller.

The compressor was then disassembled and found to have extensive first stage impeller dmnage and chemical evidence indicating that an exothermic reaction had occurred in August 1998.

An examination was also conducted of the Stage 2 compressor that was in service at the time of the Decembar 9,1998, exothermic reaction. Based upon a visual examination, there were indications j

that the vanes separated from the impeller in a manner that may be similar to the August 1998, failure. However, it is not know if the impeller damage occurred before or after the exothermic reaction. Additional metallurgical examinations of the August, Stage 4 impeller and the December 9 i

Stage 2 impeller are planned to determine the cause for the vanes separating fwm the impellers. Any j

similarities of the failure mechanism between these two compressors will be evaluated for e possible common cause. In addition, these examinations may be able to identify if vane separation occurred

~ before the exothermic reaction or if the vane separated due to elevated temperatures generated from I

an exothermic reaction.

The possibility that loose pa ts, metal' debris or other foreign material may have entered the

- compressor was also investigated. Foreign material could migrate into the compressor and cause j

sufficient " rubbing" to initiate an exothermic chemical reaction. Due to the extensive damage experienced by the equipment it was not possible to detemiine if foreign material was present in the system prior to the exothermic reaction. Ilowever, a review of procedure XP3-GP-GP1070,

" Foreign Material Exclusion" indicated that the procedure was adequate to detect debris. Since the

.'1 GDP 99-2014 Page 7 of 15 Event Report 98-17, Revision 1 procedure.was used during the replacement of the compressors that failed in August, the

]

investigation team concluded that it was unlikely that any foreign material would have been left in i

the system following the compressor replacement.

As stated above, an evaluation of damaged expansionjoints and compressor vanes is planned. This event report will be revised if the evaluation results in significant new information or leads to substantial changes to the corrective action plan.

A contributing cause was inadequate procedures and training. The Safety Analysis Report (SAR),

Section 3.1.2.2.2, Isolation of Failures, states that, "If the Amp loading in a single stage begins

. increasing, the problem may have been initiated by compressor parts rubbing, deposits in the compressor, bearing failure, or some other failure associated with the motor itself, which requires immediate shutdown". The Side Purge cascade operating procedure did not contain guidance to help the operator differentiate between load changes due to controllable factors such as compressor surging and load changes due to equipment failure, and it did not contain guidance as to the type of -

Amp increases which, woald require a cell shutdown. Furthermore, no alarm response procedures had been developed and implemented for load alarms due to surging, or high/ low Amps.

-Without procedural guidance in this area, the operator interpreted the Stage 2 Amp meter fluctuations as surging due to a coolant bubble coming up the Cascade. An actual coolant bubble would have resulted in mino; motor Amp oscillations of only a few Amps. These minor oscillations j

would have existed in at least two siages of the affected cell, with the largest oscillations in the stage with the most coolant. The correct evaluation of the Stage 2 motor Amp deflections could have led to a more prompt cell shut down.

i l

Operating experience has shown that the quicker a cell undergoing an exothermic reaction is shut l

. down and isolated, the quicker the reaction will be extinguished. This particular exothermic reaction progressed so rapidly from the time the Amp meter fluctuations were first observed, that it is not i

- known if an earlier cell shut down would have made a significant difference in the extent of the exothermic reaction. However, this event did demonstrate the need to provide improved guidance J

to operators on how to evaluate and respond to unexpected Amp meter fluctuations, and to develop i

and implement alarm response procedures.

The procedures for Top Purge and remaining cascade cells currently contain appropriate guidance 1

regarding operator response to' motor AMP changes. Ilowever, this guidance is fragmented and i

i L

contained in multiple procedures. To facilitate operator response, the existing guidance will also be incorporated into alarm response procedures.

L

~

u

/

Ey.

, _3

,.,g t

GDP 99-2014 Page 8 of 15 Event Report 98-17, Revision 1 During the review of this event, the investigation team also identified issues where further evaluation is needed. Although these issues were not contributing causes to the event, they identify areas where the potential to improve operator and equipment response to exothermic reactions may exist. Evaluation of each of these areas will be performed to determine if enhancements are warranted. - These issues identified by the team are discussed below.-

The olant desien does at orovide temnerature monitorine and alarm instrumentation for each Puree Cell stage:. UF / aluminum exothermic reactions are initiated due to the presence of a high 6

temperature source, such as compressor rubbing. Temperature monitoring capability with an associated high temperature alarm utilized in conjunction with motor Amp readings, could provide o

an indication that conditions are developing which could cause an exothermic reaction or that an exothermic reaction is already occurring. The temperature alarm would allow for more rapid operator evaluation and cell trip, if necessary to prevent or mitigate a potential exothermic reaction.

All compressors depend on throughput flow and either a recycle flow that is cooled or a suction stream that is cooled to control the internal temperature of the compressor. The current temperature monitoring capability for each Side Purge cell consists of temperature indication for the suction and discharge of each stage. These temperatures can be read at the Local Cell Control Panel (LCC).

On Cells 25-7-2 and 25-7-4, Stage I has a stage discharge thermocouple and a recycle valve. Cells 25-7-6,25-7-8, and 25-7-10 have high stage discharge thermocouples on Stages I and 6 and recycle valves at Stages 1 and 6. The high discharge temperature thermocouple causes the cell recycle valves to open on high temperature, and turns on an alarm light at the LCC and an alarm light / horn on the appropriate cell panel in ACR-6. In the case of Cell 25-7-2, the cell recycle valve was already open because the feed to Side Purge was to Stage 4 of the cell instead of Stage 1.

The ACR does have a cell coolant high temperature alarm, which measures the temperature of the

. coolant for the entire cell, not an individual stage. The same alarm light is used for a coolant low pressure alarm.~ The cell coolant alarm does not provide a rapid indication of a stage temperature problem.

The plant design does not provide an internal temperature motor trio function: The purge cascade operation differs significantly from the isotopic cascade. The purge cascade equipment must handle a wide range of non-UF gas mixtures. Changes that occur in the purge cascade are more difficult 6

to diagnose than changes that occur in the isotopic cascade because of the variations in gas raixtures.

- Under certain operating conditions, compressor problems could be misinterpreted as changes in gas

- concentration, resulting in delays in taking action to minimize equipment damage.

-l 1

y s

GDP 99-2014 Page 9 of 15 Event Report 98-17, Revision 1 l

l

.A motor trip function for internal compressor high temperature, coupled with temperature i

l monitoring, would provide a rapid methou of curtailing compressor rubbing that could lead to an l

_ exothermic reaction. This type ofinstallation currently exists er is being installed on Evacuation Booster Station compressors. The purge cascade compressors can be built with an internal l

temperature probe, so no change in compressor design would be required to provide the signal for the trip circuitry. Installation of the automatic trip based on compressor internal temperature would result in an automatic compressor shutdown when the temperature increases to the trip setpoint.

I l

The mechanism by which an isolated hot spot caused by mechanical rubbing became a propagating reaction is not fully understood. In most previous exothermic reactions, the bot spot did not propagate and an isolatd reaction occurred. The mechanism for propagation is likely dependent on the presence or combination of factors such as low partial pressure of UF, abnonnally high Freon 6

concentration, process gas flow rate, and/or relatively new aluminum impellers. By improved monitoring of cascade temperatures and the addition of automatic high temperature shutdown instrumentation, it may be possible to detect hot spots and shut down cells before reactions can propagate.

Ooeration of Side Purge stage control valves in manual mode: The stage control valves had been i

placed in manual mode because they were not operating properly. During norma; operating l

conditions, the Cell 25-7-2 Stage Control Valves (CVs) are operated in the automatic mode to j

maintain the stage pressure at the desired value. The CVs will cycle between approximately 80%

l to 90% open to maintain cell pressure.

l On November 23,1998, the Cell 25-7-2 CVs for Stages 2 thru 6 were placed on manual control and l

100% open. The purpere for this was to properly control Side Purge pressure and UF concentration, 6

in intermediate speed purge Cell 25-7-6, with Cell 25-7-4 o!Tstream for maintenance work. The Cell 25-7-2 pressure taper could not be lowered sufficiently to mainte:u proper pressure and UF6 l

concentration in Cell 25-7-6, with the CVs for Stages 2 through 6 in automatic. This was due to a problem with the stage CV instruments and controllers, which prevented the CVs from gc'ag to the

(

full open position when necessary. The CVs for Stages 2 through 6 were placed in manual and j

100% open.

The Stage 1 CV remained in automatic to assist with pressure control since placing all CVs in manual and 100% open would have lowered pressure too much.

(

A Work Request had been submitted to calibrate tne stage CV instruments and controllers, but was not completed prior to this event. The Work Request was not given a high priority since operation in the manual mode is allowed. Operating ihe cell in this manner may have masked the compressor rubbing problem and precluded the operator shutting down the cell in a timely manner. The ability to detect problems based on changes in Amp meter readings is in part based on the assumption that E

p GDP 99-2014

)

Page.10 of 15 Event Report 98-17, Revision 1.-

the control valve is maintaining the stage pressure at a set value. With the control valves full'open 1

in manual, Stage 2 could have been underloaded. As the rubbing increased, the Amp load may have returned to a near normal value.

lhe Side Purce and Too Purne feed block valves do not have automatic isolation canability: Side Purge Cells 25-7-2,25-7-4 and Top Purge Cell 25-7-20 cre unique in that the "A" line supply can be fed to the center stages of these. cells to set up a stripper section for the Side Purge mode of

- operation.- This is accomplished through the use of manually operated valves located in the cell

. housing. These manual valves for Cell 25-7-2 are AB2S3 to Stage 3 and AB2S4 to Stage 4, for Cell 1 7-4 it is AB4S4 to Stage 4, and for Cell 25-7-20 it is either AB20S3 or AB20S4 to Stages 3 or

- 4 respectively. This configuration, when in use, prevents the automatic isolation of these cells in an emergency condition through the use of the cell-Off-Split push button.

The lack of automatic isolation could enhance the undesirable effects of an exothermic reaction or 1

a piping wet air inleakage problem due to the delay experienced in isolating the cell from its UF 6 supply or from tne remainder of the Cascade. In the case of an exothermic reaction, this could prolong the fueling of the reaction, while, in the case of a piping breach, it could allow more wet air j

to be drawn into the Cascade.

The feed line can be isolated by alternate means that are available to the operators. The first motor operated. valve upstream of the cell can be closed from the ACR. However, as a result of 1

deficiencies'in procedures, the operator in the ACR at the time of this event did not recognize the need to use this alternate isolation method before evacuating the ACR.

The motor load alarm relavs were not oronerly set to activate at the anoropriate levels: The motor load a: arm relays monitor motor current and alarm to changes in current. The alarm set'ings are relative to current operating conditions and are not set to any pre-determined values. 'the alarm j

relays are designed to alert the ACR operator of changing cascade conditions that may require the operator to evaluate overload or underload conditions and take corrective actions. The relays

)

actuats whenever the actual motor load goes outside of the alarm bandwidth. When cascade

. conditions change and new' steady state motor load conditions aie established, the bandwidth must be adjusted so that the new steady state motor load is within the bandwidth.

There has becn no guidance avaRole to Operations on how to maintain the load alarm relays within range to perform their intended function. By not being properly set, it is likely that an alarm could have already been in place as a result of a previous motor load change. With one stage in alarm condition, any additional stage disturbances would cause the affected stage alarm to light, but the audible alarm would not come in. Following this event, maintenance personnel determined that the Q

r GDP 99-2014 Page11of15 Event Report 98-17, Revision 1

' low load alann relays were not functional. The ACR-6 operator stated 'that at the time of the event, the high load alarm did not actuate. However, it is not known if this was due to improper settings

- or to a mechanical malfunction.

. Setting motor load alarms to match changing cascade conditions is difficult because of the design -

and the age of the equipment. The alarms can only be set accurately by maintenance personnel since adjustments to the relay are required. A new motor load alarm has been designed which allows the operator to set the alarm limits without the assistance of maintenance. This new type ofinstrument for the purge cells would make it easier tc maintain the motor load alarm limit set points at the

. proper values. This could enhance the operator's ability to identify upset conditions that could lead to equipment problems.

Corrective Actions

.A.

Immediate Actions Taken

1. On December 23,1998, a briefmg was initiated for X-326 Facility Operations personnel. The briefings included a description of what happened, how the incident was handled and what to l

look for as indicators that such an event is happening. The briefing emphasized the steps necessary for tripping and isolating a cell. Actions to be taken if the cell block valves do not fully close were'also addressed.

2. On December 15, 1998, the development of a training module was initiated to address j

recognition of cell surging, cell loading and cell shutdown requirements.

3. - On January 6,1999, a lessons learned was developed and issued to Cascade personnel to communicate operating conditions that may increase the possibility of a similar exothermic reaction.

4.- On December 9,1998, administrative controls were established to prevent returning the Side Purge cells to service.-

5. ' n December 10,' 1998, a vibration survey was conducted on runrJ ig motors and compressors O

in the Top Purge Cascade. Engineering initiated weekly vibratit. urveys on operating purge cascade equipment.

p l,'

e l

i i

GDP 99-2014 Page 12 of 15

- Event Report 98-17, Revision 1

6. The X-326 Daily Operating Instructions for January '4,1999, emphasized the procedure j

- administrative controh that state "upon cell startup, any cell indicating abnormal vibration shall J

be followed up with a full set of vibration readings unless the vibration is excessive upon which the cell shall be shut down",

)

7. On December 21,1998, an administrative control was placed in the Daily Operating Instructions for cascade process areas to require that, "Any centrifugal compressor that is shut down due to high vibration will not be re-started in the presence of UF ".

6

8. ' On December 11, 1998, as a result of the damage observed on Cell 25-7-2, an ultrasonic inspection was performed on B-line elbows in the other Side Purge cells. No evidence of thinning was observed.

9. On December 11 and 12,1998, a vibration survey was conducted on all operating cells in the cascade. Each cell was physically touched to determine if vibrations were greater than normal.

Vibration readings were taken on cells which exhibited higher than normal vibrations, j

10. On December 23,1998, an administrative control was placed in the X-326 Daily Operating Instructions to require that the motor load alarm relay set points be checked by maintenance any time that cell pressures are changed enough to affect cell load.

1 B. Corrective Actions Planned

1. Prior to restarting the Side Purge cells, applicable Side Purge procedures will be revised to flowdown the requirements of SAR Section 3.1.2.2.2, " Isolation of Failures".

. 2. By June 1,1999, Cascade procedures will be changed to require aligning painters on ACR Amp j

meters with black pointers to previde a norma' baseline for monitoring motor load changes.

3. By November 3,1999. Alarm Response Procedures for Top and Side Purge Cascade will be developed and implemented.

4. By May 19,1999, operators will be provided guidance that identifies causes of motor Amp changes and how each of these causes will be reflected by the cell panel Amp meters in the ACR and other instrumentation available to operations.

t

(T Enclosure l' GDP 99-2014 Page 13 of 15 Event Report 98-17, Revision 1

5. Prior to restarting the Side Purge cells, process gas temperature monitoring, alarm and automatic compressor high process gas temperature shutdown will be installed on Side Purge cells.

I C. Additional Actions Planned l.

1. A metallurgical analysis of the expansionjoints and compressor vanes will be completed. Based on the results of the evaluation, additional corrective actions will be developed if required.

. 2. A formal program will be developed for determining the cause for centrifugal compressor L

failures to ensure appropriate corrective actions are taken to enhance compressor safety and reliability.

L

3. Procedures will be revised to require that centrifugal compressors which are taken ofTstream due L

' to excessive vibration will not be restarted in the presence of quantities of UF,, sufficient to allow l

an exothermic reaction to propagate.

4. The cell statup procedures will be revised to require that upon cell startup, any cell indicating abnormal vibration shall be followed up with a full set of vibration readings or shutdown if vibraiions are excessive.

5. The vibration monitoring procedure will be revised to require that a full set of vibration data be j=

taken upon startup for any cell which had action level vibration when it was shutdown.

6. Alarm Response Procedures for Cascade ACRs will be developed and implemented.

L L

7. A project will be initiated to install process gas temperature monitoring, alarm and automatic compressor high temperature shutdowr: on remaining cells in Unit 25-7.

-8. 'A project will be initiated to install process gas temperature monitoring and alarm on remaining

. cascade cells which do not have this instrumentation.

9. Cascade Operations procedures will be revised to prevent operating Top and Side Purge cells in I

manual for extemp 3 periods without additional monbering to ensure the Amp load is not excessive for thr' < prating pressure.

!!O. An improved design for the Side Purge stage control valve pneumatic controllers will be completed and a schedule to install the new instrumentation will be developed.

t

I l

l GDP 99-2014 l

Page 14 of l5 I

Event Report 98-17, Revision 1

11. Appropriate administrative controls will be established to ensure that the Taylor type "D" pneumatic controllers are repaired and calibrated, so that they can operate in automatic, prior to I

returning the low speed purge cells, which are currently shutdown, to service.

j i

12. Operators will be trained on the difference in the response of the type "D" pneumatic controllers i

versus the type "F" pneumatic controllers used on all other cells, and how to properly adjust the controller.

l

13. Side Purge operating procedures will be revised to require isolation of the feed lines as part of required cell isolation.

14. A propotal for a project to implement replacement of X-326 load alarm relays with new relays

)

which allow setting by the operator will be developed.

Extent of Exposure ofIndividuals to Radiation or Radioactive Materials Personnel in the facility at the time of the fire, and responders to the fire, were monitored for potential ineke of radionuclides. Fifty-two personnel involved with the event submitted urine samples. C f these, nineteen personnel working in the building at the onset of the emergency were placed on precautionary restriction until their samples were evaluated. Each sample was analyzed

)

for uranium and technetium and all results were less than 5 g/l uranium and 6,000 pCi/l for i

technetium. Since all samples were less than the SAR Action levels, no further actions were

{

required. All precautionary restrictions w^re removed for the 19 individuals.

l Area radiation readings during and following the event noted no increase over normal background readings for that area. Area posting requirements remained the same (Contamination Area). With l

the exception of some precautionary expansions of boundaries within already posted " Restricted

. Areas", only minimal boundary changes were necessary. Personnel access to the area was not j

l restricted for radiological protection reasons nor were radiological dose reduction measures required during or following the emergency response. Air sampling results within the fin:ility indicated slight increases in the airborne radioactivity levels during the fire, while monitors outside of the facility indicated no readings above the minimal detectable activity for the equipment. Prior to the 1

termination of the c...t, airborne radioactivity levels within the facility had returned to normal.

Although the fire created several contamination contml cor.cems, worker, environmental and public exposures were not significantly affected. SAR, Section 5.3 requirements were maintained and

p l

1 GDP 99-2014 j

Page 15 of 15

)

l*

Event Report 98-17, Revision 1

)

i 1

nonnal procedural controls were effective in maintaining contamination control and returning the l-area back to pre-fire conditions from a radiological control standpoint.

]

Recovery activities have been successfui in removing the oil and water from the affected area.

Decontamination efforts have restored the operating floor te pre-fire entry requiremer.ts for Personnel Protective Equipment. Approximately eighty percent of area boundaries have been l

returned to their pre-fire positions and decontamination efforts are continuing to restore the remaining boundaries.

1 L

Lessons Learned i

The emergency response to the fire was adequate and plant safety systems performed as designed l

to contain the fire and minimize the extent of damage. Ilowever, the event revealed weaknesses in

. procedures and training of plant personnel. These issues will be assembled and entered into the plant's corrective action management tracking system. Each identified issue will be addressed to i

l ensure actions are taken to improve any future response to similar cvents.

l i

I l

1 j

GDP 99-2014 Page1of1 Event Report 98-17, Revision 1 List of Commitments

• 1.

Prior to restarting the Side Purge cells, applicable Side Purge procedures will be revised to

- flowdown the requirements of SAR Section 3.1.2.2.2, " Isolation of Failures".

j 2.

By June 1,1999, Cascade procedures will be changed to require aligning pointers on ACR

.j Amp meters with black pointers to provide a nonnal baseline for monitoring motor load changes.

3.

By November 3,1999, Alarm Response Procedures for Top and Side Purge Cascade will bc developed and implemented.

4.

By May 19,1999, operators will be provided guidance that identifies causes of motor Amp changes and how each of these causes will be reflected by the cell panel Amp meters in the

~ ACR and other instrumentation available to operations.

5.

Prior to restarting the Side Purge cells, process gas temperature monitoring, alami and automatic compressor high process gas temperature shutdown will be installed on Side Purge cells.

i

• Regulatory commitments contained in this document are listed here. Other corrective actions listed in thh submittd are not ensidered regulatory commitments in that they are either statements of actions completed, or they are considered enhancements to USEC's investigation, procedures, programs or operationa.

9

'