1142

urcs and breakdowns, nor is it theintent of that standard to do SQ. How-ever, new rules have been added toNFPA 36 as a result of extractorbreakdowns; these will be discussed.

SAFETY & ENVIRONMENT

Extractor failure: safety proceduresIn June 1998, I received a phone callfrom Robert Benedetti, the NationalFire Protection Association's (NFPA)staff liaison for the NFPA 36 Commit-tee on Solvent Extraction Plants. Hehad received a letter from the plantmanager of a peanut solventextraction plant in Johannesburg,South Africa, that had experienced anexplosion and fire in 1996. A man wasburned 10death, and the extractorwas destroyed. This event occurredafter a mechanical failure in anextractor. The extractor was openedand entered, and, at that time, anexplosion resulted. The plant mallagerhad made reference 10 the 1997edition of NFPA 36, but he noted thatsuch incidents and the corrective pro-cedures are 1101covered in NFPA 36.He asked for advice on how to safelyproceed when extractors experiencemechanical breakdowns. Mr. Benedet-ti requested that 1 respond to him. Iwrote a six-page letter covering pro-cedures 10 use when an extractor failsmechanically. A week later, I receiveda very nice letter from the plant man-ager, ill which he thanked me for myreport, but with yet another request.He stated that the information andknow-how that I provided are 1I0tgen·eralty well known in many parts of theworld, and suggested lhat someonepresent a paper at an AOeS meetingcovering this subject. I hope thispaper's contents will satisfy hisdesires.

The subject of handling extractorsthat have failed under load when fullof hexane solvent is not covered intextbooks. It is dealt with on a cese-by-case basis, and from experience

INFORM. Vol. 10,00. 12 (December 1999)

Figure 1. Elftractor damaged In 1966 explosion

obtained during past breakdowns.This article presents four situationsof extractor failures. NFPA 36, So/-vent Extraction Plants, does notspecifically cover such extractor fail-

TIUs ankle by C. w.us KingsbaUr.CL Kingsbder Inc., 7145 Nonhgrult Dr.NE.AIltuItG, GA 30318. is baud on his p"KnltJIion

gMn dMring the 1999 AOCS AIUllMJIMeeting cl Expo in Orlando. Florida.

Case No. IIn September 1966. an extractor hadbeen emptied and air-purged. butwas not purged according to proce-dures to allow welding and burningin the extraction area. This proce-dure requires emptying the plant ofall hexane solvent and securing andisolating the solvent storage tankand its contents from the rest of theplant. All equipment in the plant is

1143

Figure 2. Twisted extracted flake elevator after 1966 explollon

opened and cleaned. and thepipelines opened and drained of allflammable materials. Then, all oftbe equipment piping is purged withlive steam for a minimum of twodays to ensure all solvent has beenevaporated. After steam-purging. theequipment and solvent plant area arechecked with an explosion meier tobe certain there are no flammablemixtures of hexane and air remain-ing in the area. Only then can weld-ing and flame cutting of equipmentproceed.

The extractor entry doors had beenremoved, and maintenance work hadbegun on the ell masse extracted flakeelevator. Some of the chain flightsfrom this unit were being replacedwith an open electric motor wrenchbeing used 10 remove the bolts hold-ing the flights. With no warning, therewas an ignition of solvent that hadremained in the discharge hopperlocated just above the extracted flakeelevator. This ignition Slarted a smol-dering fire of the soy fines that hadbuill up and accumulated on the insidewalls of the extractor. The fire depart-ment was called, and firefightersattempted to extinguish this smold-ering fire. Thirty minutes after thefirefighters began trying to pUIOUIthisfire, a second. and much more violent,ignition occurred, knocking a fire-fighter offhis perch.

Case No.2In March 1982, in Jackson, Missis-sippi, an extractor cell door hadjammed and was bent, causing anextractor to fail under load. Theextractor was allowed to cool, theaccess door to the extractor wasremoved. and the extractor drive wasreversed and run backward to thepoint that the bent cell door was in aposition to be removed. This exrrac-[Or had no purge blower, nor did ithave live steam piped to it. Duringthe time this work was being done,hexane vapors kept leaving the entryopening of the extractor where thedoor had been removed. All plantpersonnel left the extraction buildingand went to the maintenance buildingto locate a new cell door. During thattime, hexane vapors ignited outsidethe extractor, setting off the delugesystem and alarm. The flash from thefire started a smoldering fire insidethe extractor. The flow of delugewater was stopped, the fire depart-ment was called in, and firefighterstried to extinguish the smoldering fireinside the extractor. However, aboutone hour later, there was a secondignition of hexane vapors, withgreater intensity than the first, but notserious enough to cause extensivedamage to the plant, nor to injureanyone.

At that point, the Jackson FireDepartment made a decision to "but-ton up" the extractor, and to fill theextractor with water. Their idea wasto put out the smoldering fire insideby raising the water level inside theextractor to the point where the fire

was located at a safe distance from theextraction area, a smoldering fire,such as had occurred, could be extin-guished without risking lives. Thissafety idea was incorporated intoNFPA 36, and now appears on page36-12:

Par. 5-8.3 "Extractors, Desolven-titers, Toasters, Dryers, Spent FlakeConveyors shall be of a design thatminimizes the possibi/iry of ignitionof product deposits. Such equipmentshall be protected by extinguishingsystems, using inert gas, steam, or acombination of the two, controlledfrom a safe remote local ion."

The plant's general manager madea very wise and critical decision afterthe second ignition. He gave ordersfor everyone to evacuate the extrac-lion plant area until it could be deter-mined what was happening. Theplant officials contacted the companythat I worked for, Btaw-Knox inPlusburgb. Pennsylvania (the firmhas been part of several corporateacquisitions since then), and I wassent to Quincy, Illinois. to help them.But before I arrived, a third. and veryviolent, explosion destroyed thisextractor, blowing building panelsmore than 300 feet from the building.Figures I and 2 illustrate the powerand destructiveness of a hexane sol-vent explosion.

However, since the general manag-er had ordered everyone to leave theextraction plant, nobody was killedand nobody was injured. I inspectedthis plant the following day and foundthat a valve in the 1.inefrom the sol-vent-feed high tank to the extractorhad been left open. Although liquidsolvent was not entering the extractor,hexane vapors were going into theextractor while the open motor powertool was in service, and this hadcaused the explosion.

It occurred to me that (a) if livesteam could be piped to the extractorand the extracted-flake elevator, and(b) if the valve admitting this steam

INFORM.Vol. 10. no. 12 (December 1999)

1144

SAFETY 8< ENVIRONMENT

was burning. However. they were notable 10 fill the extractor with muchmore than about two feet of water.After 24 hours, the fire departmentstated that the fire was out, and theextractor could be entered. The plantwas reentered, and three employeesstood in front of the access door andremoved the bolts holding the door.As the man in the center removed thecover. there was a violent explosion,the quarter-inch-thick cover platewas deformed into a giant "C," andwas blown into the worker, crushinghis skull and killing him instantly.The worker to the left of the fatallyinjured worker was severely burnedand later died from his injuries. Theother worker was blown off the plat-form, but was only slightly injured.The smoldering fire inside the

extractor had not been extinguished.and as soon as the door wasremoved. sufficient oxygen becameavailable to close the fire triangle.and this deadly explosion resulted.The "fire triangle" consist of threecomponents, which in this case are:hexane. which is the fuel for the fire;air, which has oxygen and supportsthe fire; and a source of ignition.such as a spark or flame to stan thefire. These three items are called thethree corners of a fire triangle. Thehexane and air also must be in theproper proportion or quantities forignition to take place. plus ignitionmust occur while that proper propor-tion exists.

The extraction plant was evacuat-ed immediately. No one, includingfire department personnel. would go

SEPARATION ANXIETY?REMEDY: POPE WIPED FILM STILLSFOR HEAT SENSITIVE MATERIALS!

••

351 N. Detor.! WOOdS BOulevardP.O. BOX18saukville, WI 53080 USAPhone: 262·268·9300Fax:262·268·9400E·mall: saiesecccemc.ccm

Sizes range from 2"Laboratory Unit (shown)through 20" ProductionSystems (up to 1000 kg/hr)

... High vacuum, short residencetime for lowest degradation,highest quality product.

... Configurable for Evaporation,Concentration, Molecular Distillation.Can be combined with columns forFractional Distillation.

... StraightfOlward scale-up from labthrough pilot plant and production.

... Proven acceptance withpharmaceuticals, nutraceuticals,cosmetics, foods, oils, polymersand numerous specialties.......

Toll ProcessingServices Available

Fof InformoHon elrde '12 t on Iorm on lOIt page

INFORM. Vol. 10. no. 12 (December 1999)

back in. The plant boiler was shutdown, so steam was not available.Shortly after the fatal explosion. Jreceived a call to come to Jacksonand was at the plant within fourhours. I was able to enter the plantwith another fireman for an inspec-tion, and we secured the plant (whichin itself is another story). I wasplaced in charge of rebuilding theplant. and incorporated all safetydesigns required by NFPA 36 at thattime, including emergency smother-ing steam and a large extractor purgeblower. For the record, the source ofignition for the original explosionwas an open motor that had beenmistakenly installed, about 10 yearsearlier. on an extractor-stage pumplocated near the extractor entry pointwhere the door had been removed.The Jackson Fire Department wasseverely reprimanded for its actionsduring this incident and also for alack of knowledge on safety of hex-ane extraction plants. At that time,there were two solvent plants in oper-ation in Jackson.

There is an important epilogue tothis case. About one year after thisplant had been rebuilt, there wasanother extractor breakdown underload. I was called in and this time,we made the decision not to open theextractor. Instead. all of the miscellaand solvent from the unit werepumped. Live steam was channeledinto the three connections located atthe bottom of the extractor. Theextractor was steamed this way forabout 24 hours. allowed to cool. andopened. All of the solvent in thesolids remaining in the extractor hadbeen evaporated. When the extractorwas manually unloaded. there wereno trace odors of hexane solvent inthe meal as it was removed. Thisshowed live steam can be used tosafely desolvenrize tbe material leftin an extractor after it has failedmechanically.

Case No.3This plant in Hamburg, Germany, waspan of a large complex consisting of asingle building that housed twoextractors with related auxiliaryequipment, plus equipment to processthe miscella and extracted flakes from

1145

Figure 3. HambUrg extraction complex InvotvedIn 1983 explosion

a third large extractor that was locateda distance from the main extractionbuilding. ln July 1983, there was a fireand explosion that totally destroyedthe two extraction systems along withthe auxiliary equipment used to pro-cess the extracted material and miscel-la for the third extractor. However, thethird extractor was not damaged. Thisextraction complex can be seen in Fig-ure 3; the destruction is shown in Fig-ure4.

Although this case technically isnot strictly an extractor breakdown,it does fall into the same categoryas one. The problem here was thata very large extractor-probablythe largest in the world at thattime-was full of hexane-ladensoybean flakes, but there was noway to desolvemize or dispose ofthem. Iwas hired by the plant man-ager to help him resolve his prob-lems with the governmental author-ities, to aid him in securing theplant, and to assist in the rebuildingof the plant

This plant is next to the RiverElbe in Hamburg. The initial planto unload the extractor was todump the contents of a cell of theextractor onto the concrete nearthe extractor. spread the materialout with rakes. and lei it air-descl-vemize. This was tried, but evenafter the solvent-wetted flakes had

been exposed to the air for manyhours, there was still a great deal ofsolvent remaining in them. The plantmanagement also had obtained per-mission from local authorities toload the entire solvent-laden contentsof the extractor onto a river bargeand tow it out to the North Sea fordisposal. I was nOI sure how thismight be accomplished, but thoughtthat this program could be very dan-gerous, starting with getting the

material onto the barge and then offthe barge.

This extractor was about 51 feet indiameter. and had a large and separatedischarge hopper. about 12 feet indiameter, into which the cells woulddump. There were two conveyorsfrom the discharge hopper that fedinto the extracted flake take-awaychain conveyor. Based on what hadhappened in Jackson, Mississippi, theprevious year, I suggested that livesteam be piped into the outlet of thetwo conveyors from the dischargehopper. The plan would be to emptythe extractor, cell by cell, into the dis-charge hopper, and then to steam-des-olventize the contents of each cell.The content of the first cell wasdropped into the discharge hopper,steam-desolventiaed for several hours,and then dumped upon the outdoorconcrete pad located under the dis-charge hopper. This material turnedout to be golden brown, had no sol-vent in it, and even had a few waterballs. The extractor was emptied suc-cessfully in this manner. Although ittook about four days to do so, it wasdone safely with no problems. Thisagain proved that solvent could beremoved from an extractor that hadbroken down under load without haz-ard [Q either plant personnel or to theplant.

Figure 4. Damage et 1983 explosIon in Hamburg, Germany

INFORM, VoL 10. no. 12 (December 1999)

1146

SAFETY Be ENVIRONMENT

H H ~-~- .:::<.~-~H ....

oJ-{~ H

0

~~

00 =00 o O~~H H H

Agure 5. Japanese diagram shows location 01 victim. In 1991explollon.

Case No.4The explosion. fire. and deaths thatresulted from this extractor that failedunder load was. by far. one of theworst that this industry had suffered. Itoccurred Dec. 23,1991, in Japan. Theextractor was a Yoshino unit that didnOI have a purge blower. nor did ithave live steam piped to it from a safelocation. Eight men died in this inci-dent. Figure 5, provided by the JapanOilseed Processors Association(JOPA). shows the location of eachperson killed. One was inside theextractor. two were at the opening onthe lower extractor work platform. oneat the other opening at this level. andfour were at the main plant operatinglevel. After failing under load, theextractor had been allowed to cool forabout 10 hours before all liquid sol-vent and miscella were pumped fromthe unit. The extractor manhole coverswere removed and the first manentered the extractor. Fifleen minuteslater. the explosion occurred and eightmen died.

Figure 6 shows the actual fire, thedamage that occurred. and the loca-tion of the extraction plant ar the sue.Japanese governmental authorities

,

became involved with this accident.and imposed new, and I might add,virtually impossible rules forJapanese oilseed processors. Becauseof this explosion, the Osaka authori-ties, the Japanese Ministry of Labor,Fire Defense Authority. and even theJapanese Diet (parliament), becameinvolved with this disaster. Theyruled that as long as hexane solventremains in an extractor that has failedunder load, the extractor could not beentered or even opened. JOPA mem-bers could not operate with such aruling. JOPA hired me to meet withtheir processors to develop a safe andsensible method to handle this situa-tion. I visited the plants of the 10largest oilseed processors. Iexplained to them the methods ofsteaming out an extractor (as dis-cussed in Cases 2 and 3 above). Iworked directly with MitsuoKurashige. lOPA's Technical Directorat that time. (Mr. Kurashige made anexcellent presentation at the 1995San Antonio AOCS meeting aboutthe Kobe earthquake.) I recommend-ed thai a test demonstration be madefor the authorities to show them thatan extractor full of solvent-laden

Flakes can be desolvemized with livesteam addition without first openingthe extractor.

The JOPA members thought thiswas a great idea. However, when Isuggested that one of them do this atthe end of their processing season.nobody volunteered 10 do this. Andthey still have not done this. TheJapanese authorities are still arguingabout what to do, and as of this date.have not made a decision or a ruling.

Finally, entering an extractor thathas failed under load must be dis-cussed. Unfonunately. I have had todo this several times. and it is not apleasant experience.

Before even considering enteringan extractor that has failed under load,certain actions must be taken.

First, the extractor must be provid-ed with a properly designed, sized,and installed purge fan. Cargill wasthe first processor to install a purgefan in the mid-I 950s. Cargill sized thefan to change the empty air volume ofthe extractor once every three minutes.i.e .. 20 times per hour. With this largevolume of air entering the extractor.the solvent-air mixture inside theextractor will be below its LFL (lower

INFORM, Vol. 10. no. 12 (Oecember 1999)

1147

Figure 6. Japanese magazine displaying cover photo of extraction plant lire

flammable limit) of 1.2% by volume.The latest Edition of NFPA 36

(1997) has a new requirement forpurging extractors (Par. 5-8.1.7),which states: "The extractor shall beprovided with means to remove sol-vent vapors so that the concentrationof vapors inside the unit in the areawhere work is required call be main-tained at or below 25 percent of thelower flammable limit. If a purge fansystem is used to accomplish this. ilshall IIIeel all of the following require-ments." It then explains the capacity,location, type, etc., for the fan.

My question about the wording inthis paragraph is what other practicalmethod or means can be used toaccomplish this except a purge fan. Iwould strongly recommend changingthe wording by removing the word"if." It is my opinion that a purge fanis one of the most important safetydevices you can have when you enteran extractor that has failed underload.

Second, live or open sparge steammust be provided to the extractor toconform to Par. 5-8.3, as was statedearlier. Having live steam available atall limes to send into an extractor

from a safe remote location providesyou with a second method of remov-ing the solvent from an extractor thathas suffered a breakdown under load.

Once you have one of these toolsavailable, you can decide which ofthese will serve you best to correct theproblem. If you plan to use live steamto remove the solvent, prepare yourprogram to accomplish this. If youdecide to use the purge fan to removethe solvent in the extractor, and thenenter the extractor, the following pro-cedures are recommended:

• Allow ample time for the extrac-tor to cool. Do not rush this time,especially if the extractor is insulated.It may take 24 hours or longer to dothis. Take aU the time that is needed todo it properly.

• Pump as much liquid solvent andmiscella from the extractor as possi-ble. Do this many times during thecooling period.

• During the time it takes to coolthe extractor, formulate a detailed planfor the work. Organize a team, andhave the program reviewed andapproved by your plant management.

• Have air packs or air maskschecked and made available for the

person or persons entering the extrac-tor. Check that there is sufficientapproved, breathable air available forthe personnel inside the extractor.

• Follow your company's "Con-fined Space Entry Permit Program."Be sure that a safely harness is worn,and that a rope is attached to the har-ness.

• Use only explosion-proof flash-lights provided with a UL (Underwrit-ers Laboratory) label.

• Do not use ponable electric lightswith extension cords, regardless oftheir rating.

• Begin air-purging the extractorfrom the top down. Remove accessplates and sight glasses from the top:slowly work your way down to thebottom.

• Constantly monitor the air in theproximity of the extractor and the airdischarging from the purge fan withan explosion meter. If the explosionrisk rises, Slap work and leave theextractor.

Remember that the sources of igni-tion for Cases I, 2, and 4, and the onein South Africa, were from outside theextractor, not from work being doneinside the extractor.

• Limit the length of time that anindividual is inside the extractor er anyone time.

All extractors, regardless of thetype or design, can fail when fullyloaded with solids and hexane sol-vent. An extractor can be safelyentered, inspected, and temporarilyrepaired under these breakdown con-ditions. But all necessary precautionsand care must be used. All plantsmust have a written, well thought outconrlngency program prepared,reviewed, approved, and in place forsuch an emergency. Decide whetherthe extractor is to be air-purged orsteam-purged prior to entering theextractor.

Install a purge blower and a steamextinguishing system controlled froma safe remote location if the extractoris not already provided with these.Both systems will save you lime dur-ing normal plant shut-down programs.But better yet, you will have thesecritical systems in place and ready touse in case your extractor fails underload. •

INFORM. VOl. 10. no. 12 (December 1999)