CSB Case Study Hoeganaes Dec9 Final

8/3/2019 CSB Case Study Hoeganaes Dec9 Final

1/31

CSB Hoeanaes Corporation Case Stdy

Hoeganaes Corporation Case Study December 2011

Case StudyU.S. Chemical Saety and Hazard Investigation Board

Hgn Ctin: Gtin, TNMetal Dust Flash Fires and Hydrogen ExplosionJanuary 31, 2011; March 29, 2011; May 27, 2011

5 Killed, 3 Injured

N. 2011-4-I-TN

KEY ISSUESHazard reconition and traininEnineerin controlsFire codes/enorcementRelatory oversiht


2/31



1.0 INTRODUCTION

This case study examines multiple iron dust ash fres and a hydrogen explosion at the

Hoeganaes acility in Gallatin, TN. The frst iron dust ash fre incident killed two workers

and the second injured an employee. The third incident, a hydrogen explosion and resulting

iron dust ash fres, claimed three lives and injured two other workers.

1.1 HoeGaNaes CorporaTIoN

Hoeganaes Corp. is a worldwide producer o atomized steel and iron powders. Headquartered

in Cinnaminson, NJ, Hoeganaes has acilities in the U.S., Germany, China, and Romania.

The Hoeganaes Corp. is a subsidiary o GKN, a multinational engineering company headquar-

tered in the United Kingdom. GKN has businesses in addition to powder metallurgy, including

aerospace and automotive driveline industries. GKN acquired the Hoeganaes Corp. in 1999.

The largest consumer or the powdered metal (PM1) product is the automotive industry,which presses and sinters2 the powder into small metal parts.

1.2 FaCIlITy DesCrIpTIoN

The Hoeganaes Gallatin acility (Figure 1), located 30 miles northeast o Nashville, Tennessee,

employs just under 200 employees. Since becoming operational in the 1980s, they have increased

their manuacturing capability over 550 percent rom 45,000 to over 300,000 tons.

TABLE OF CONTENTS

1.0 Introdction 2

2.0 Process Discssion 3

3.0 The Incidents 3

4.0 Analysis 6

5.0 Key Findins 24

6.0 Recommendations 24

Appendix A: Determination oIron Powder Explosibility rom

Pressre Ratio Calclations 28

Appendix B: Determination o

Iron Powder Classication rom

Explosion Severity Calclation 30

FIguRE 1

Satellite view o the

Hoeanaes gallatin acility.

1PM is te accepted acronym by te powdered metals industry.2Sintering is te process o solidiying PM via eat and/or pressure to orm a component.


3/31



3Ductility is te pysical property o a material were it is capable o sustaining large permanent canges in sape witout breaking.

2.0 PROCESS DISCUSSION

Hoeganaes receives and melts scrap steel. Various elements are added to the molten metal

to meet customer specifcations, but the workhorse product, Ancorsteel 1000, is over 99

percent iron. The molten iron is cooled and milled into a coarse powder that is processed

in long annealing urnaces to make the iron more ductile.3 The urnaces are called band

urnaces, or the 100 oot conveyor belt, or band, that runs through them. A hydrogenatmosphere is provided in the band urnace to reduce the iron by removing oxides and

preventing oxidation. The hydrogen is supplied to the acility by a contract provider, onsite.

Hydrogen is conveyed to the urnaces via pipes located in a trench under the oor and

covered by metal plates.

In the process o going through the urnace, the coarse powder becomes a thick sheet called

cake. The cake is sent to a cake breaker and ultimately crushed into the fne PM product.

The majority o the fnished PM product has a particle diameter between 45-150 microns,

or roughly the width o a human hair (Figure 2).

3.0 ThE INCIDENTS

3.1 JaNuary 31, 2011 (Two FaTalITIes)

PM product is transerred through the plant by various mechanisms including screw convey-

ors and bucket elevators. Bucket elevators have a tendency to go o-track when the belt

pulling the buckets becomes misaligned. Once sufciently o-track the strain on the motorincreases until the torque is too great and the motor shuts down. On January 31, 2011, at

about 5:00 am Hoeganaes plant operators suspected bucket elevator #12 o being o-track

and a maintenance mechanic and an electrician were called to inspect the equipment.

FIguRE 2

Fine PM collected rom

the Hoeanaes plant

(penny shown or scale).


4/31



Based on their observations, they did not believe that the belt was o-track and re-

quested, via radio, that the operator in the control room restart the motor (Figure 4).

When the elevator was restarted, vibrations rom the equipment dispersed fne iron

dust into the air. During a CSB interview, one o the workers recalled being enguled in

ames, almost immediately ater the motor was restarted.

City o Gallatin emer-

gency responders arrived

with ambulances and

transported the mechanic

and electrician to the

Vanderbilt Burn Center

in Nashville, TN. Both

employees were severely

burned over a large

percentage o their bodies.

The frst employee died

rom his injuries two dayslater. The second employee

survived or nearly our

months beore succumb-

ing to his injuries in late

May 2011.

3.2 MarCH 29, 2011 (oNe INJureD)

As part o an ongoing urnace improvement project, a Hoeganaes engineer and an outside

contractor were replacing igniters on a band urnace. The pair experienced difculty in re-

connecting a particular natural gas line ater replacing an igniter. While using a hammer to

orce the gas port to reconnect, the Hoeganaes engineer inadvertently loted large amountso combustible iron dust rom at suraces on the side o the band urnace, spanning 20 eet

above him. As soon as the dust dispersed, the engineer recalled being enguled in ames. He

jumped and ell rom a rolling stepladder in his attempt to escape the freball. He received

frst- and second-degree burns to both thighs, superfcial burns to his ace, and scrapes rom

his all. Ater seeing the initial ash o the dust igniting, the contractor took evasive action

and escaped without injury.

FIguRE 4 (RIgHT)

Scene o Janary 31,

2011, incident area.

FIguRE 5

Compter raphic o

Janary 31 iron dst

fash re.

FIguRE 3 (LEFT)

Compter raphic o

maintenance workers

inspectin bcket

elevator #12 jst prior

to Janary 31 fash re.

Elevator Enclosre

Motor


5/31



The engineer was wearing the Hoeganaes-designated personal protective equipment, which

included pants and a shirt that were rated as ame resistant clothing (FRC). He was also

wearing an FRC rated jacket that provided extra shielding to his upper torso rom the ash

fre.

3.3 May 27, 2011 (THree FaTalITIes, Two INJurIes)Around 6 am on May 27, 2011, operators near band urnace #1 heard a hissing noise that

they identifed as a gas leak. The operators determined that the leak was in a trench, an area

below the band urnaces that contains hydrogen, nitrogen, and cooling water runo pipes,

in addition to a vent pipe or the urnaces. The operators inormed the maintenance depart-

ment about the hissing, and six mechanics were dispatched to fnd and repair the leak. One

annealing area operator stood by as the mechanics sought out the source o the leak.

Although maintenance personnel knew that hydrogen piping was in the same trench, they

presumed that the leak was nonammable nitrogen because o a recent leak in a nitrogen

pipe elsewhere in the plant and began to try to remove trench covers. However, the trench

FIguRE 6 (LEFT)

Compter raphic o

the as line connection

involved in the March 29

fash re.

FIguRE 7 (RIgHT)

The as line connection

involved in March 29, 2011,

incident.

FIguRE 8

Compter raphic o

March 29 iron dst

fash re.


6/31



covers were too difcult to lit without machinery. Using an overhead crane, they were able

to remove some o the trench covers. They determined that the leak was near the southern-most trench covers, which the crane could not reach. Shortly ater 6:30 am, maintenance

personnel acquired a orklit equipped with a chain on its orks, and were able to reach and

begin removing the southernmost trench covers.

Interviews with eyewitnesses indicate that just as the frst trench cover was wrenched rom

its position by the orklit, riction created sparks, ollowed by a powerul explosion.

Several days ater the explosion, CSB investigators observed a large hole (approximately 3 x

7 inches) in a corroded section o piping that carried hydrogen and ran through the trench

(Figure 11).

As the leaking hydrogen gas exploded, the resulting overpressure dispersed large quantities o

iron dust rom raters and other suraces in the upper reaches o the building. Portions o this

dust subsequently ignited. Multiple eyewitnesses reported embers raining down and igniting

FIguRE 9 (TOP)

Scene o the May 27

incident beore (let, taken

drin the CSBs Janary

31 incident investiation)

and ater (riht). Note

visible accmlations o

iron powder on sraces.

Circled areas show trench

cover location.

FIguRE 10 (RIgHT)

Compter raphic o

maintenance crews

startin to remove the

trench covers sin a

orklit jst prior to May 27

explosion.

Feb 07 2011 May 28 2011


7/31



multiple dust ash fres in the area. They also reported visibility so

limited in some instances that ashlights were required; one eyewit-

ness said that even with a ashlight, he could see only 3 to 4 eetahead due to extensive dust and smoke.

The hydrogen explosion and ensuing iron dust ash fres injured our

o the responding mechanics and the annealing operator.4 The two

mechanics near the orklit were transported to a local hospital where

they were treated or smoke inhalation and released shortly thereater.

Two other mechanics and the operator who stood by during the

operation were rushed to Vanderbilt Burn Center. Less than a week

ater the incident, two employees succumbed to their injuries. The

third seriously injured employee died rom his injuries almost seven

weeks ater the incident.

Due to the extensive nature o the injuries, and the abundance o both hydrogen and com-

bustible dust present at the time o the incident, it is difcult to specifcally determine which

uel, i not both, caused the atal injuries to the victims.

3.4 eMerGeNCy respoNse

The Gallatin Fire Department (GFD) has responded to 30 incidents o various types over

the past 12 years at the Hoeganaes Corp., including the January 31, March 29, and May 27

incidents. In June 1999, the GFD responded to a fre caused by iron dust that ignited in a

baghouse. One person suered smoke inhalation injuries as a result o the incident.

Beore the GFD arrived at each o the 2011 incidents, Hoeganaes volunteer frst respond-

ers cared or the injured. Hoeganaes volunteers participate in annual training that coversfrst response, CPR, and frst aid. They are instructed to provide care until GFD and EMS

responders arrive.

Immediately ollowing each incident, the volunteers provided frst aid and comort to the

injured by applying water to cool the burns and covering the victims with a burn blanket

to keep them comortable. EMS arrived within minutes o the initial 9-1-1 call and trans-

ported the injured personnel to hospitals.

4At te time o incident, two o te mecanics and te operator were standing near te trenc wile te oter two mecanics were positioned and

possibly sielded by te orklit wen te eplosion occurred.

FIguRE 11 (TOP LEFT)

Hole in 4-inch pipin ater

the May 27, 2011, incident.

FIguRE 12 (TOP RIgHT)

Compter raphic o the

May 27, 2011, hydroen

explosion.

FIguRE 13 (BOTTOM)

upward distrbance o

trench covers cased by

the hydroen explosion inthe May 27, 2011, incident.


8/31



4.0 ANALYSIS

4.1 CoMbusTIble DusT TesTING

According to National Fire Protection Association (NFPA) 484, Standard or Combustible

Metals, a acility that handles metal dust should commission one o two screening tests to

determine i a metal dust is combustible and the provisions o the standard apply (Section4.4.1). I results rom either o the two tests show that the dust is combustible or explosible,

NFPA 484 would apply to the acility either as a matter o voluntary good practice or as a

requirement by a relevant regulatory body.

The frst screening test or the determination o combustibility, also known as the train test,

measures the burning rate o a dust layer over the length o a sample.5 I there is propagation

beyond the ignition point or heated zone, then the sample is considered combustible.6

The second test, or explosibility determination, serves as a basis to determine i a metal

powder or dust is capable o initiating an explosion when suspended in a dust cloud. This

test, perormed in a Hartmann apparatus, determines the minimum ignition energy o a

dust cloud in air by a high voltage spark.7

I either o the screening tests produces a positive result or combustibility or explosibility, NFPA

484 requires urther explosibility testing be conducted in a 20-L sphere. Several values (below)

rom the explosibility test results can be used to characterize the severity o a dust explosion.

4.1.1 CSB COMBUSTIBLE DUST TESTING

4.1.1.1 Combustibility Demonsttion

In order to visually demonstrate the combustibility o the Hoeganaes iron samples, a modifed

Go/No-Go test was perormed by the CSB. Generally, this test is perormed in a closedvessel, but the CSB was interested in directly observing any ames the dust may produce.8

EXPLOSIVITY VALUES

KSt

: calclated vale that compares the relative explosion severity and conseqence to

other dsts (bar m/s). The hiher the KSt

nmber, the more eneretic the explosion.

Pmax

: maximm explosion overpressre enerated in the test vessel (bar)

P/t: maximm rate o pressre rise, predicts violence o the explosion (bar/s)

Explosion Severity (ES): Index to determine i Class II electrical eqipment is reqired

as an OSHA reqirement.

ES > 0.5, Class II Combstible

ES < 0.4, Combstible bt not Class II

5UN Recommendations on te Transport o Dangerous Goods: Model Regulations Manual o Tests and Criteria, Part III, Subsection 33.2.1

6NFPA 484 denes a combustible metal dust as a particulate metal tat presents a re or eplosion azard wen suspended in air or te process specicoidizing medium over a range o concentrations, regardless o particle size or sape.

7ASTM E2019, Standard Test Metod or Minimum Ignition Energy o a Dust Cloud in Air

8Te Go/No-Go test is typically perormed in a modied one-liter hartmann tube; also known as te eplosibility screening test as described in NFPA 484.


9/31



This test dispersed about 30 grams o iron dustsampled rom the baghouse9 associated with

the bucket elevator rom the January 31, 2011, incidentabove an 8 inch burner. Upon being

released, the dust auto-ignites in air due to the heat given o rom the burner below (Figure

14). An intense white ame was produced that reached a peak diameter o 18 inches.

4.1.1.2 20-Lite (20-L) Test Method

CSB investigators collected iron powder samples rom various locations in the Hoeganaes

acility and commissioned testing to characterize its combustibility using two dierent

test methods, the 20-liter (20-L) and one-meter cubed (1-m3) test chambers. The 20-L test

laboratory used the standard test method, ASTM E1226, Pressure and Rate o Pressure

Rise or Combustible Dusts, or the selected iron powder samples. Each dust sample wasinjected and ignited in a 20-L spherical test vessel equipped with transducers to record a

pressure-versus-time profle o the dust deagration in the sphere.

Table 1 shows data rom the CSBs combustibility tests o the Hoeganaes dust and a compari-

son to dust testing the CSB commissioned or previous dust incidents at other companies.

The CSB test data indicate that the iron powder is combustible and is covered by the

requirements o NFPA 484 (Section 4.4.1). Although values indicate that the dust produces

a weak explosion relative to other dusts, the dust is considered combustible by the OSHA

defnition10 and can result in a ash fre capable o causing injuries and atalities.

FIguRE 14

CSB iron dst

combstibility

demonstration, see

Section 4.1.1.1.

9Ventilation equipment tat removes airborne particulate by orcing air troug a specially designed ltration bag.

10OShA 3371-08 2009: a solid material composed o distinct particles or pieces, regardless o size, sape, or cemical composition, wic presents are or defagration azard wen suspended in air or some oter oidizing medium over a range o concentrations.


10/31



20-L COMBUSTIBLE DUST TEST DATA FROM CSB INVESTIGATIONS11

Hoeganaes

Iron Dust

20L test12

Granulated

Sugar

Aluminum

Dust

Polyethylene

Dust

phenolic

Resin

Pmax

(bar) 3.5 5.2 9.4 8.34 7.58

P/t

(bar/s) 68 129 357 515 586

KSt

(bar m/s) 19 35 103 140 165

Explosion

Severity (ES)

0.077 0.22 1.08 1.38 1.43

Classifcation Combstible Combstible Combstible,

Class II

Combstible,

Class II

Combstible,

Class II

Frequently, the hazards o dierent combustible dusts are evaluated by their potential

explosive capabilities. However, the hazards o combustible dusts are not limited to explo-

sions. The Hoeganaes iron powder propagates an explosion less rapidly compared to other

dusts, so there is less overpressure damage, consistent with observations by CSB investiga-

tors. Dust testing results rom Hoeganaes and prior CSB investigations illustrate that dustswith low K

Stvalues can cause ash fres that result in deaths and serious injuries. Although

combustible dusts can lead to explosions, combustible dust ash fres also pose a risk that

must be addressed in industry.

According to the 20-L standard test method, E1226, a dust sample can be defned as

combustible or explosible based on a calculated pressure ratio (PR) using the pressure data

recorded in the 20-L test chamber. For sample concentrations o 1,000 and 2,000 g/m3, a

pressure ratio value greater than or equal to 2 is considered explosible. I the pressure ratio

is less than 2, the sample is considered non-explosible. However, the test method cautions

that the dust can still burn and a dust cloud may experience a deagration depending upon

conditions such as the temperature and particle size.13 The iron dust sample rom baghouse

#4 had a PR o 4.0 and 4.7 at concentrations o 1,000 and 2,000 g/m3

respectively, indicat-ing that the dust sample is explosible.

4.1.1.3 One-Mete Cubed (1-m3) Test Method

The CSB collected a subsequent sample14 rom baghouse #4 and subjected it to combustibil-

ity testing using the one-meter cubed (1-m3) method, ISO 6184-1 Explosion Protection

Systems: Determination o Explosion Indices o Combustible Dusts in Air. The 1-m3

test vessel is larger than the 20-L vessel, and the dust, along with air and a uel source, is

injected into the system dierently.

The iron powder rom baghouse #4 underwent an explosibility screening test in the 1-m3

vessel in an attempt to ignite the sample. At several dust concentrations, none o the tests

produced signifcant pressure which exceeded the test qualifcations or ignition and there-ore the dust sample was considered non-explosible according to this method.

TABLE 1

Combstibility data or

selected materials.

11For a more detailed discussion on caracteristic combustible dust values, see CSB 2006-h-1 Combustible Dust Hazard Study.

12CSB investigators collected tis iron dust sample rom bagouse #4 at te hoeganaes Gallatin acility ater te January 31, 2011, incident.

13American Society or Testing and Materials (ASTM), E1226-10, Pressure and Rate of Pressure Rise for Combustible Dust, ASTM International, 2010.

14At te time tis sample was taken in August 2011, te Gallatin acility ad not been ully operational or about tree monts. As suc, te samplecollected did not contain nes representative o te environment at te time o te 2011 fas re incidents.


11/31



4.1.1.4 Coming Dust Testing Methods

Both the 20-L and 1-m3 tests are accepted methods that can characterize dust explosibility;

however results rom the two tests may dier. There are several actors that can contribute

to varying results among the dust test methods. Dust characteristics, such as particle size,

moisture content, and degree o oxidation (or metals) can aect the ignitability o the

sample in the test chamber.

The 20-L and 1-m3 test chambers were designed to simulate dust explosions in acility

settings, but each test has limitations. The main dierence between the two tests is the

chamber size and the dust dispersion mechanism. Since the 1-m3 test is larger, theoretically

it can better simulate an open-space dust cloud explosion. However, that larger volume also

makes it harder to create a uniorm distribution o dust within the testing chamber. In the

smaller 20-L test chamber it is easier to create a uniorm distribution; however, it is possible

that the smaller chamber also creates an overdriving eect. Since the 20-L chamber is

smaller, the energy exerted by the igniters15 may combust enough dust creating the appear-

ance o ignition16 a situation that would not occur in a acility setting.

NFPA revised the 2012 edition o NFPA 484 to state that explosibility screening tests shall

be perormed in accordance to the 20-L test standard, E1226. However, NFPA added to

the standard annex that the results o the 20-L test can be conservative and an owner oroperator o a acility may elect to use a 1-m3 test or dust explosibility testing as the 20-L

test may result in alse positives or dusts with lower KSt

values.

Despite the discrepancies between the two test methods, the empirical evidence rom the

ash fre incidents at Hoeganaes shows that dusts with lower KSt

values are capable o

ueling ash fres with severe consequences. This urther suggests that acilities should not

rely on the 1-m3 test as a sole determination o dust combustibility hazards. Dusts with

lower KSt

values and characteristics similar to the iron powder at Hoeganaes may not ignite

in the 1-m3 chamber but still have the ability to result in atal ash fres.

It should be noted that both tests are or explosibility screening, and alone may not convey

the ull combustibility hazard.

4.1.2 HOEGaNaES COMBUSTIBLE DUST TESTING

4.1.2.1 Minimum Ignition Enegy (MIE)

In 2010, Hoeganaes contracted to test iron dust samples rom the plant or combustibility

as a result o an insurance audit recommendation. The test had one sample that was similar

in particle size, moisture content, and location to the dust involved in the 2011 incidents.

That sample gave the results seen in Table 2.

The minimum ignition energy (MIE) testing determined that a continuous arc did ignite the

representative samples rom 2010, but a 500 mJ source did not. The conclusion rom the

testing was that the minimum ignition energy was greater than 500 mJ. This inormation is

valuable in determining potential ignition sources or each o the incidents.

15Te igniter may increase te temperature in te smaller 20-L camber, raising te overall temperature o te system and allowing a non-eplosivesystem to appear eplosive.

16Going, J et al., Flammability limit measurements or dust in 20-L and 1-m3 vessels.Journal of Loss Prevention in the Process Industries. May 2003


12/31



4.2 IGNITIoN sourCes

Witnesses indicated that the May 27, 2011, hydrogen explosion was ignited by sparks generated

during the liting o the trench cover. This is reasonable considering that the MIE o hydrogen is

0.02 mJ, and the energy o mechanical sparks rom metal to metal contact can be several mJ.20

The testing contracted by Hoeganaes in 2010 determined that the minimum ignition energy

or representative iron dust samples was greater than 500 mJ, and that a continuous arc

would ignite the samples. One witness at ground level reported hearing an electric sound

at the time o the incident. The motor operating bucket elevator #12 was a likely source o

ignition since it had exposed wiring, was not properly grounded, and was within a ew eet

o the dust cloud source. The wiring was exposed because the electrical conduit supplying

power to this motor was not securely connected to the motors junction box.

HOEGANAES MINIMUM IGNITION ENERGY (MIE) TEST RESULTS

Sample Iron Dst

Particle size (%500

MIT18 (C,Cloud) 560-580

MEC19 (g/m3) 200-250

TABLE 2

Combstible dst test

reslts commissioned by

Hoeanaes in 2010.

FIguRE 15

Exposed electrical

wirin on elevator motor

near Janary 2011

incident site (motor

panel cover was rotated

post incident by re

department).

Motor

Exposed wires

17mJ is an abbreviation or millijoules, wic is a unit o energy. One Joule is equal to 1000 millijoules or approimately 0.24 calorie.

18Minimum Ignition Temperature.

19Minimum Eplosible Concentration.

20V. Babrauskas, Ignition handbook, Fire Science Publisers, Issaqua, WA, 2003.


13/31



Prior to the CSB notiying Hoeganaes that evidence rom the incident area needed to be pre-

served, the company removed and modifed evidence rom the scene, including the elevator

motor, wiring, and conduit. However, on examination, there were spots that appeared to be arc

marks both inside the junction box, and on the outside o the motor housing.

4.3 HOEGANAES

4.3.1 HazarD rECOGNITION

In general industry the combustibility o metal dust is a well-established hazard, but metal

dust fres and explosions continue to claim lives and destroy property. The CSB reviewed three

publications dating back to the 1940s and 1950s that addressed metal dust (including iron dust)

hazards and explosion protection methods. The National Fire Protection Association (NFPA)

code or the Prevention o Dust Explosions, published in 1946,21 lists general precautions or all

types o dusts, including metal powder, and specifc provisions or certain types o dusts.

The Building Construction section o the code states, Avoid beams, ledges or other places

where dust may settle, particularly overhead. The Gallatin acility, built in the 1980s, was not

designed to avoid signifcant overhead accumulations o dust. The code calls or designing and

maintaining dust-tight equipment to avoid leaks and, where this is not possible, to enorce good

housekeeping procedures.

The code also cautions against sources o ignition in areas containing dust and recommends

locating dust collectors outdoors or in separate rooms equipped with explosion venting.

In 1957, the NFPA published the Report o Important Dust Explosions which included a

summary o over 1,000 dust explosions between 1860 and 1956 in the U.S. and Canada. 22 The

report listed 80 metal dust fres and explosions, including one iron dust incident that resulted ina atality in 1951.

A 1958 article in an American Chemical Society publication states, Powdered metals dispersed

in oxygen or air orm explosive mixtures their ammability and explosibility have been

reported in considerable detail23

FIguRE 16

Monds o iron dst

alon elevated sraces

at the gallatin plant,

Febrary 3, 2011.

21National Fire Codes, Vol. II. Te Prevention o Dust Eplosions 1946. National Fire Protection Association, Boston, MA., 1946.

22National Fire Protection Association, Report of Important Dust Explosions, NFPA, Boston, MA, 1950.

23Grosse, A.V., and J.B. Conway. Combustion o Metals in Oygen. Industrial & Engineering Chemistry50.4 (1958): 663-72.


14/31



In the 1990s and 2000s, the Pittsburgh Research Laboratory o the National Institute or

Occupational Saety and Health (NIOSH) conducted a study o the explosibility o various

metals, including iron. The results o these experiments, published in scientifc journals,24

showed the explosibility characteristics o iron powder to aid hazard evaluation in metal pro-

cessing industries. However, management within the Hoeganaes Corp. and GKN Corp. did

not commission an analysis o its own potentially combustible PM products and constituents

until January 2009, as a result o an insurance audit conducted in 2008. This combustibledust testing concluded that all three iron powder samples collected rom various locations in

the plant were explosible (Section 4.3.6.1). However, these results did not trigger an eective

overhaul o the dust containment and housekeeping procedures at the Gallatin acility.

Representatives rom Hoeganaes told the CSB that the dust analysis results did trigger

an operator training program or the recognition o combustible dust hazards. However,

Hoeganaes did not mitigate the dust hazard. Since Hoeganaes did not control the combustible

dust hazard, operators were orced to tolerate the conditions at the acility. Over time, these

ash fres became normalized, since they did not result in any serious injuries prior to the atal

incident on January 31, 2011.

Operators and mechanics reported being involved in multiple ash fres during their employ-

ment at the Gallatin acility. At the time o the incidents, many were aware that the iron dustcould burn or smolder. However, they were not trained to understand the potentially severe

hazard when accumulated dust is dispersed in air. Rarely would operators report the minor

ash fres and near-misses that periodically occurred.

4.3.2 ENGINEErING CONTrOLS

4.3.2.1 Combustible Dust

Thorough hazard recognition is key to eectively managing the risk rom combustible dust.

Once the hazard is recognized, applying the hierarchy o controls or fre and explosion

prevention helps address the ugitive dust issue at the source: the material itsel, the process-

ing equipment, and the work procedures. The hierarchy o controls is a saety concept inwhich a hierarchal ordering o control mechanisms is applied to reduce risk. It covers the

spectrum rom elimination at the source, at the top o the hierarchy, through engineering

and administrative (procedural) controls to personal protective equipment (PPE), at the

bottom o the hierarchy.25

Installing and maintaining engineering controls to eliminate ugitive dust accumulation

is the most eective method to prevent dust fres and explosions. Conveyance systems

and appropriately sized dust collection equipment are examples o engineering controls

that eliminate or mitigate ugitive dust generation at the source. Engineering controls are

preerred over housekeeping, but a robust housekeeping program is important to man-

age ugitive dust accumulations in areas where engineering controls need maintenance or

improvement. Additionally, administrative controls, such as worker training and operating

procedures, complement robust engineering controls.

Signifcant quantities o iron dust escaped rom equipment throughout the Hoeganaes acility.

Enclosures on the conveyance equipment leaked ugitive emissions o iron dust.. In addition,

the dust collection systems were historically unreliable and did not prevent large amounts

o combustible iron dust rom becoming airborne and accumulating on elevated suraces

throughout the processing areas.

24Casdollar, K., Flammability o Metals and Oter Elemental Dust Clouds. Loss Prevention Symposium. AIChE 1994.Casdollar, K., Zlocower, I., Eplosion Temperatures and Pressures o Metals and oter Elemental Dust Clouds. Journal o LossPrevention in te Process Industries 20 (337-348) 2007.

25Kletz, T., Amyotte, P., Process Plants: A Handbook for Inherently Safer Design. CRC Press, 2010.


15/31



4.3.2.2 Hydogen

The trench involved on the May 27, 2011, incident contains many pipes including nitrogen

and hydrogen supply and vent pipes or the band urnaces. In addition to housing pipes, the

trench also acts as a drain or the cooling water used in the band urnaces. At the time o

the incident this water came out o the urnaces hot and drained directly onto the pipes and

into the trench (Figure 17).

According to ASME B31.3, design concerns about ambient conditions around process pipes

ocus on environments and changes that can create physical stresses in the piping. Section

10.5.3 o NFPA 55 requires annual maintenance including inspection or physical damage

and leak tightness. CGA G-5.4 similarly requires regular inspection or physical damage

and leak tightness. However, Hoeganaes did not regularly inspect the pipes in the trench.

The design and maintenance o this trench, should have addressed the issue o slow corro-

sion over time caused by the hot water runo and solids accumulation.

Both NFPA 55 and NFPA 2 state, Provisions shall be made or controlling and mitigat-

ing unauthorized discharges. NFPA 2 urther requires that the storage, use, or handlingo [hydrogen] in a building or acility shall be accomplished in a manner that provides a

reasonable level o saety rom illness, injury or death However, the CSB ound no

evidence o a Hoeganaes procedure to inspect piping within the trench to ensure that cor-

rosion had not compromised the piping systems which would allow an uncontrolled release

o hydrogen. Moreover, Hoeganaes had no written procedure or protocol to mitigate gas

leaks, and maintenance crews were allowed to begin investigating a suspected leak without

testing the atmosphere or concentrations o explosive gas.

4.3.3 aDMINISTraTIVE CONTrOLS

4.3.3.1 Housekeeing

Observations by CSB investigators at the Gallatin acility shortly ater the frst incidentindicated that combustible dust was leaking rom equipment and that housekeeping was

ineective (see Figures 16, 18, 19, 20 and 21). Combustible iron dust coated almost every

STANDARDS THAT ADDRESS THE

HYDROGEN SUPPLY AND VENT PIPES IN

THE TRENCH INCLUDE:

ASME26 B31.3, Process Pipin

CGA27 g-5.4-2010, Standard or Hydroen

Pipin Systems at user Locations

NFPA 2, Hydrogen Technologies Code

NFPA 55, Compressed Gases and Cryogenic

Flids Code.

FIguRE 17

Water fow (pper let

o photo) and externally

corroded pipes in the

trench involved in the

May 27, 2011, incident.

26Te American Society o Mecanical Engineers (ASME) is a proessional body ocused on mecanical engineering. Te organiza-tion is known or setting codes and standards or mecanical devices. Te ASME conducts one o te worlds largest tecnicalpublising operations troug te ASME Press. Te organization olds numerous tecnical conerences and undreds o proes-sional development courses eac year, and sponsors numerous outreac and educational programs.

27Te Compressed Gas Association (CGA) develops and publises tecnical inormation, standards, and recommendations or temanuacture, storage, transportation, distribution, and use o industrial gases.


16/31



surace up to 4 inches deep and was visible in the air. Mitigation o the combustible dust haz-

ard by Hoeganaes was limited to a less-than-adequate vacuuming service, sparsely enclosed

conveyance equipment, and an inadequate baghouse fltration system.

Although bucket elevators and some conveyance equipment were enclosed, ugitive dust

emissions were evident throughout the acility. Moreover, the CSB investigators observedleaks o ugitive dust to the atmosphere when the bags used in the baghouse fltration sys-

tem were pulsed, which allowed dust to escape into the work areas many times each hour.

The baghouse flters are designed to collect the smallest, and consequently most dangerous,

dust particles. Yet, the CSB ound that the baghouses were oten out o service. Employees

reported that the baghouse associated with bucket elevator #12 was out o service sporadi-

cally or the 7 days leading up to the atal incident on January 31, 2011, allowing fne

combustible iron dust to remain in the area, rom which it was dispersed when the elevator

was restarted during maintenance.

4.3.4 pErSONaL prOTECTIVE EqUIp-

MENT

4.3.4.1 Flme resistnt Clothing (FrC)

Workers in production-related op-

erations wear ame resistant clothing

(FRC) to reduce risk o thermal injury

rom ash fre incidents. As part o the

Personal Protective Equipment (PPE)

Standard (29 CFR 1910.132), OSHA

requires employers to provide workers

with FRC in workplaces when ash fre

or explosion hazards are present.

FRC can reduce the severity o burn injuries sustained during a ash fre when engineeringand administrative controls ail. FRC, usually worn as coveralls, is made o treated natural or

synthetic fbers that resist burning and withstand heat.

There are two NFPA standards that provide guidance on the design and use o FRC. NFPA

2112, Standard on Flame-Resistant Garments or Protection o Industrial Personnel Against

Flash Fire, provides the minimum requirements or the design, testing, and certifcation o

FRC. NFPA 2113, Standard on Selection, Care, Use, and Maintenance o Flame-Resistant

Garments or the Protection o Industrial Personnel Against Flash Fire, provides guidance

or the selection, use, and maintenance o FRC. The 2009 edition o NFPA 484 included a

requirement or workers to wear FRC i working in metal dust-handling operations, but it did

FIguRE 20

Iron dst on strctral

spports, Febrary 3, 2011.

FIguRE 18 (LEFT)

Iron dst on raters,

Febrary 3, 2011.

FIguRE 19 (RIgHT)

Iron dst on overhead

sraces, Febrary 3, 2011.


17/31



not specifcally reerence NFPA 2112 or 2113 in the standard. The 2012 edition o NFPA 484

requires that new and existing acilities covered by the standard adhere to the requirements o

NFPA 2113 or FRC.

Hoeganaes employees were required to wear FRC, and the injured and atally injured

employees were wearing the Hoeganaes-designated FRC at the time o the 2011 ash fre

incidents. Though FRC is intended to reduce the severity o thermal injuries, fve severelyburned employees died ollowing the January and May incidents. The specifc FRC worn did

not provide any signifcant protection against the combustible iron dust ash fres and the

hydrogen explosion at Hoeganaes.

4.3.5 1992 INCIDENT

On May 13, 1992, a hydrogen explosion and iron dust ash fre similar to the May 2011

incident in Gallatin severely burned an employee working at the Hoeganaes acility in

Cinnaminson, NJ. CSB investigators interviewed the injured employee rom the 1992 incident

and learned that a hydrogen explosion event in a urnace dispersed and ignited signifcant

accumulations o iron dust which resulted in thermal burns over 90% o his body. The injured

worker spent a year in a burn unit and is still recovering rom his burn injuries.

4.3.6 INSUraNCE INSpECTIONS

4.3.6.1 allin

In November 2008, Allianz, a German-based risk insurer, conducted a routine audit o the

Hoeganaes acility. The audit report noted that improved housekeeping was needed in several

areas o the acility. In the list o risk improvement proposals, the Allianz report stated, The

potential or explosions caused when clouds o powdered metal are aroused in equipment

should be analyzed by an independent consultant. The proposal recommended an independent

dust hazard analysis and a subsequent hazard study to identiy suitable mitigation techniques,

should the iron dust in the acility be ound to be explosible.

In January 2009, Hoeganaes collected samples o base iron dust, urnace-eed dust, and

baghouse dust and commissioned explosibility testing as Allianz recommended (Table 3). In

September 2010, Hoeganaes requested another test o various powdered metals. Test results

showed that 5 o the 9 iron samples had KSt

values greater than 1.

The Allianz audit fndings initiated several action items as part o the Hoeganaes Combustible

Dust Program at the Gallatin acility. The scope o the program was to understand and align

company practices with the OSHA Combustible Dust National Emphasis Program (NEP)

(Section 4.5.1). Action items included combustible dust training or employees and understand-

ing relevant NFPA codes at the acility. Although the majority o the Hoeganaes Combustible

Dust Program action items had planned completion dates prior to the 2011 ash fre incidents,

the program did not eectively mitigate the combustible dust hazards at the acility.

HOEGANAES IRON DUST EXPLOSIBILITY TESTING

(20-L)

Sample Base Iron Frnace Feed Bahose Dst

Pmax

(bar) 1.9 2.8 3.8

P/t

(bar/s) 2.8 63 17

KSt (bar m/s) 3.8 80 22

TABLE 3

Dst explosibility test

reslts commissioned by

Hoeanaes in 2009.


18/31



4.4 FIre CoDes

4.4.1 NFpa 484

NFPA 484, Standard or Combustible Metals, an industry consensus standard, applies to

acilities that produce, process, fnish, handle, recycle, and store metals and alloys in a orm

capable o combustion or explosion. NFPA codes involving combustible metal dusts have

evolved several times since the 1946 The Prevention o Dust Explosions (Section 4.3.1). In the1950s, the NFPA divided the 1946 document into several codes or specifc materials, such

as magnesium and titanium. In 2002, all o the NFPA combustible metal dust standards were

combined into NFPA 484. NFPA 484 describes the tests and methods or determining metal

dust combustibility and provides guidelines or preventing dust explosions and ash fres or

all types o metal dusts.

The CSB commissioned testing similar to the 2009 edition o NFPA 484 explosibility deter-

mination test requirements (Section 4.1). The testing concluded that the Hoeganaes metal

dust sample is explosible and thereore NFPA 484 applies.

Had Hoeganaes voluntarily ollowed, or been required to ollow, NFPA 484 by the GFD

(authority having jurisdiction28) the January and March incidents may have been prevented,

and the eects o the May accident could have been reduced. As with many NFPA standards,NFPA 484 has a retroactivity clause or certain

chapters, stating that requirements or all

existing equipment, installed prior to the cur-

rent edition o the code, are not enorceable by

the authority having jurisdiction, unless it is

determined that the existing situation presents an

unacceptable saety and health hazard.

Neither the City o Gallatin nor the GFD identi-

fed combustible metal dust as a concern or

hazard during previous inspections conducted

prior to the 2011 incidents (Section 4.4.3).Chapter 12 o NFPA 484, Requirements or

Combustible Metals includes provisions to control or eliminate dust fres and explosions.

It requires engineering controls or dust-producing processes such as enclosures and capture

devices connected to dust collection systems. The standard describes recommended house-

keeping practices and requencies, and how to control ignition sources.

Practices at Hoeganaes did not conorm to the saety recommendations set orth in NFPA

484. Under Building Construction, NFPA 484 requires that oors, elevated platorms, and

gratings where dust can accumulate be designed to minimize dust accumulations and acilitate

cleaning.29 The Hoeganaes acility has numerous at suraces overhead upon which the CSB

investigators observed signifcant accumulations o combustible iron dust. Since Hoeganaes has

an iron powder-producing operation, specifc engineering controls outlined in NFPA 484 applyto the machines that manuacture and convey the PM. All machines that produce fne particles

o iron should be connected to a dust collection system that has the appropriate velocity to

capture all dust. The CSB investigators observed that some o the PM conveyance equipment at

FIguRE 21

Photo o eqipment

obscred by airborne dst,

taken by CSB investiators,

Febrary 7, 2011.

28Te organization, oce, or individual responsible or approving equipment, materials, an installation, or a procedure.

29Capter 12 is not retroactive to eisting acilities.


19/31



Hoeganaess was not enclosed; as such, it was not designed to control signifcant dust emissions,

and employees urther reported that baghouse dust collectors were oten down or maintenance.

This section o the standard also requires that dust collection systems be located outdoors; at

Hoeganaes, the baghouses are located inside, posing a serious fre and explosion hazard.

Chapter 13 o NFPA 484 includes provisions or housekeeping and applies to all new and ex-

isting acilities. It requires that accumulations o excessive dust on any portions o buildings ormachinery not regularly cleaned in daily operations be minimized and that ugitive dust not be

allowed to accumulate. Hoeganaes used a vacuuming service to reduce quantities o dust that

had accumulated. However, inadequate dust collection systems and dust leakages rom equip-

ment produced accumulations beyond what could be controlled by the limited housekeeping

service that was being provided.

4.4.2 ELECTrICaL CLaSSIFICaTION

The classifcation o combustible dust hazardous locations is based on the criteria established

by article 500 o NFPA 70, National Electric Code (NEC). The NEC defnes hazardous loca-

tions as areas where fre or explosion hazards may exist due to ammable gases or vapors,

ammable liquids, combustible dust, or ignitable fbers or yings. The classifcations are

broken down into three hazardous material classes:

Class I ammable gas or vapor

Class II combustible dust

Class III bers and yings

Each class is urther categorized into one o two divisions, based on operating conditions:

Division 1 Normal: areas where the classied hazardous material is likely to be

present under normal operating conditions

Division 2 Abnormal: areas where the classied hazardous material is likely to be

contained and present only through accidental release

With the proper evaluation o electrically classifed areas in an operating acility, appropriatelyrated equipment can be installed.

4.4.2.1 Combustible Dust

NFPA 499, Classifcation o Combustible Dusts and o Hazardous (Classifed) Locations or

Electrical Installation in Chemical Process Areas, specifes the type o electrical equipment ac-

ceptable in atmospheres containing combustible dust. It applies to locations where combustible

dusts are produced, processed, and handled, or where surace accumulations o dust could be

ignited by electrical equipment. Based on the requirements o NFPA 499 and the NEC, clas-

sifed electric services and equipment would be required in a acility where combustible metal

dust was present. Specifcally, NFPA 499 states that Division 1 electrical equipment should be

used in areas where combustible dust can accumulate to 1/8 o an inch (3 mm).OSHA 1910 Subpart S includes defnitions and requirements or hazardous or electrically clas-

sifed locations. To determine whether classifed electrical equipment is needed or a combus-

tible dust, a Class II combustibility test is conducted with dust samples rom the acility and

the explosion severity (ES) ratio calculated. An ES o greater than 0.5 signifes an appreciable

explosion hazard, which means either that Class II electrical equipment must be installed or

dust accumulations near electrical equipment must be prevented. ES values less than 0.5 are

generally considered to be lower explosion hazards, and non-rated electrical equipment in

those atmospheres is acceptable.


20/31



In January 2009, Hoeganaes submitted samples or explosibility testing. Although several

samples were determined to be explosible, the ES values were low, precluding the need or

classifed electrical installations in the Hoeganaes acility. The CSB tested iron dust samples

and ound an explosion severity ratio o 0.01 to 0.1, signifcantly less than the ratio that

would require classifed equipment under existing codes.

4.4.2.2 Flmmble Gses

Test results on the iron samples in the vicinity o the Hoeganaes incidents did not require

the installation o classifed electrical services. However, the ammable hydrogen in the

band urnaces did. NFPA 497, Recommended Practice or the Classifcation o Flammable

Liquids, Gases, or Vapors and o Hazardous (Classifed) Locations or Electrical Installations

in Chemical Process Areas, lists hydrogen as a Class I ammable gas. Areas with Class I

materials are urther classifed as Division 2 i the material is normally contained inside the

equipment, or Division 1 i the material is normally present in ammable concentrations

outside the equipment, such as during maintenance. Because hydrogen normally vents into the

work area at one end o the annealing urnace, the area around the annealing urnace should

have been designated as Class I Division 1, and the electrical equipment in that area designed,

installed, and maintained to meet those recommendations.

Moreover, the standard states that i no physical boundary surrounds a Division 1 area, the

transitional area between Division 1 and an unclassifed area is designated as Division 2. In

addition, because the hydrogen piping system includes potential leak points, such as valves

and anges, these areas should have been designated as Class I Division 2. As such, large

areas in the annealing building would be required to have Class I Division 1 or Class I

Division 2 electrical service installed.

Despite these classifcations and their recommendations, the CSB observed inappropriate electri-

cal installations, including large electrical cabinets that were open to the atmosphere and that

had signifcant iron dust accumulations, incomplete conduit, and regular 110-volt cord-plug

outlets instead o ignition-proo electrical devices approved by the NEC or Class I atmospheres.

4.4.3 INTErNaTIONaL FIrE CODE

The International Fire Code (IFC) establishes minimum requirements or fre protection

and prevention systems. The International Code Council (ICC), a membership association

responsible or developing saety codes in residential and commercial buildings, publishes

the IFC. Chapter 13 o the IFC, Combustible Dust-Producing Operations, (2006 edition)

briey addresses the prevention o ignition sources and housekeeping or areas where com-

bustible dust is generated, stored, manuactured, or handled.30 The IFC also reerences several

NFPA standards, such as NFPA 484, but language in section 1304 is vague as to whether the

compliance with the listed NFPA standards is mandatory or voluntary. The IFC authorizes the

authority having jurisdiction, such as the fre department or municipality, to enorce ap-

plicable provisions o these NFPA standards; the word authorizes does not carry the same

weight as shall enorce and might be interpreted as a discretionary rather than mandatorycode requirement.

Companies are required to comply with the IFC only i promulgated through local, state, or

ederal regulations.31 The State o Tennessee Division o Fire Prevention and the City o Gallatin

both adopted the 2006 version o the IFC into their codes. Though the general precautions or

housekeeping and ignition sources are required through code adoption o the IFC, the noted

30Combustible Dust-Producing Operations is located in Capter 22 o te IFC 2012 edition.

31In a previous investigation o a serious dust eplosion at West Parmaceutical Services in Nort Carolina in 2003, te CSB recom-mended tat te state require mandatory compliance wit te detailed provisions o te relevant NFPA dust standard (NFPA 654)rater tan te muc brieer and less prescriptive requirements o IFC Capter 13.


21/31



NFPA standards could be interpreted as voluntary. The Tennessee Fire Code specifcally declares

that the state does not enorce optional or recommended standards or practices.

Practices at the Hoeganaes acility did not conorm with the requirements set orth in Chapter

13 o the IFC. The code prohibits devices using an open ame and the use o spark-producing

equipment in areas with combustible dust. It also states that accumulated combustible dust

will be kept to a minimum inside buildings. However, adhering to the much more detailed

design and engineering requirements o NFPA 484 would have urther reduced the likelihood

that the three serious incidents would have occurred.

At the time o the 2011 ash fre incidents, the Hoeganaes acility was operating under the

provisions o the 2006 IFC. The GFD has the authority to inspect acilities against the IFC, issue

violations, and stop-work orders i buildings or operations are declared unsae based on the

codes provisions. All construction and design provisions apply to new or existing structures i,

in the opinion o the code ofcial, a distinct hazard to lie or property exists. All administrative,

operational, and maintenance provisions apply to new and existing conditions and operations.

For the City o Gallatin, the fre chie is responsible or enorcement o the IFC. CSB investi-

gators reviewed the GFDs inspection history at the Hoeganaes acility. The fre department

conducted three inspections in the previous 12 years, in 1999, 2002, and 2011. The 2011inspection was perormed just two weeks prior to the May 27, 2011, incident. The report or

this inspection documented observations at the acility related to fre suppression and emer-

gency egress, but did not mention combustible dust hazards even ater the January and March

2011 incidents. The CSB ound no evidence that the GFD inspected Hoeganaes against the

provisions o the 2006 IFC or the hazards associated with combustible metal dust, electrical

installation, and operations that use ammable gases.

4.5 reGulaTory oversIGHT

4.5.1 OSHa

4.5.1.1 Combustible DustIn 2006, ollowing three catastrophic dust explosions that claimed 14 lives in 2003, the CSB

issued its Combustible Dust Hazard Study. The study identifed 281 dust fres and explosions in

the U.S. between 1980 and 2005 that resulted in 119 atalities and 718 injures.

The absence o an OSHA comprehensive combustible dust standard was a key fnding in the

2006 study and resulted in a CSB recommendation to OSHA to initiate rulemaking or a

general industry combustible dust standard. The recommendation remains open-acceptable32

as o the publication o this report.

A signifcant reduction in grain dust incidents resulted rom a prior dust regulation enacted by

OSHA or grain handling acilities. In 1987, OSHA promulgated a grain acilities standard

in response to a series o major grain dust explosions. A 2003 OSHA analysis o grain dust

incidents showed that atalities dropped 60 percent ater the regulation was enacted.

Another recommendation rom the 2006 CSB Combustible Dust Study was or OSHA to

develop a national special emphasis program33 (SEP) to address dust in industry while the

comprehensive standard was being developed. In October 2007, OSHA initiated a Combustible

Dust National Emphasis Program (NEP) to target industries that generate, store, or handle

combustible dusts. The program provides guidance to OSHA inspectors about how to apply

32In 2009, te CSB voted to cange te status o te OShA recommendation to open-acceptable ater OShA initiated an advancenotice o proposed rulemaking (ANPR) or te combustible dust regulation.

33SEPs and NEPs are not regulations but rater are enorcement tools tat allow OShA to ocus resources on inspections.


22/31



existing saety statutes or standards, such as the General Duty Clause (OSH Act 5(a)(1)) and the

Walking-Working Suraces standard (29 CFR 1910.22), to acilities with combustible dust.

Although the scope o the NEP applies to various types o combustible dusts including metal

dusts at Hoeganaes, only those acilities assigned to particular North American Industry

Classifcation System (NAICS) codes are specifcally targeted or inspections by OSHA under

the NEP. These NAICS codes classiy acilities by primary business activity. I a acility thathandles combustible dust is not included in the NAICS code list, OSHA can initiate an NEP

inspection only as a result o a complaint, reerral, or occupational injury. The NAICS code

or the Hoeganaes Gallatin acility (331111, Iron and Steel Mills) is not included in the list o

industries targeted by the NEP, although similar industries that handle metals are. According

to the U.S. Census Bureau 2007 Economic Census, there were 352 acilities in the Iron and

Steel Mills industry code (331111) employing over 106,000 workers.

The CSB 2006 dust study ound that 20 percent o documented dust incidents rom 1980 to

2005 occurred in the metals industry. During that period, the CSB documented three iron dust

incidents that resulted in three atalities and our injuries. One o those documented incidents

occurred at the Hoeganaes Gallatin acility in 1996 when iron powder caught fre in a dust

collector and one worker received a smoke inhalation injury.

In February 2008, a catastrophic sugar dust explosion at Imperial Sugar killed 14 and injured

36. A month later, OSHA revised and reissued the Combustible Dust NEP to include acilities

that handle sugar. As a result, OSHA notifed all acilities covered by the sugar industry codes

that they were subject to inspection.

The Combustible Dust NEP is the only national OSHA program to specifcally promote eec-

tive combustible dust hazard management. OSHA did not initiate rulemaking on combustible

dust, as the CSB had recommended in November 2006, until April 2009. In its fnal report on

the Imperial Sugar disaster in September 2009, the CSB recommended that OSHA proceed

expeditiously with the new dust standard. Although OSHA issued an advance notice o

proposed rulemaking (ANPR) or combustible dust in October 2009 and has since convened

various stakeholder meetings, no proposed or fnal rule has been published.

Combustible dust fres and explosions have continued to occur at industrial acilities across the

country; since issuing the 2006 CSB Combustible Dust Study, the CSB has recorded a number

o signifcant combustible dust incidents. Until a combustible dust standard is enacted, the NEP

remains OSHAs primary tool or addressing combustible dust in the workplace.

4.5.1.2 pocess Sfety Mngement

The Process Saety Management Standard, 29 CFR 1910.119 (PSM), is an OSHA regulation

or processes that contain highly hazardous materials or signifcant quantities o ammables.

The intent o PSM, as stated in the standard, is preventing or minimizing the consequences

o catastrophic releases o toxic, reactive, ammable, or explosive chemicals. PSM applies

to processes using or producing any o 137 listed highly hazardous chemicals at or above

threshold quantities and processes with ammable liquids or gases onsite in quantities o

10,000 pounds or more in one location. I applied to Hoeganaes, elements o PSM would

have required practices and procedures that could have prevented or lessened the severity o

the May 27, 2011 incident.

In the May 2011 incident, there was a hydrogen leak and subsequent hydrogen explosion and

dust ash fres that ultimately killed three workers. The hydrogen provided to the Hoeganaes

acility originates rom an onsite generation and storage unit, owned and operated by a


23/31



contractor on land leased rom Hoeganaes. Since the hydrogen generation acilitys total

intended operational capacity exceeds 10,000 pounds o hydrogen, it is covered by PSM.

4.5.2 TOSHa

Tennessee is one o 24 states with a state-specifc occupational saety and health plan. It

develops and operates its own saety and health programs with approval rom ederal OSHA.

In November 2011, TOSHA issued citations to the Hoeganaes Gallatin Facility or the May 27,

2011, incident. Hoeganaes received 15 OSHA PSM Standard violations related to the hydrogen

system. OSHA concluded that the company lacked appropriate procedures to ensure mechani-

cal integrity o the hydrogen piping, ailed to develop an emergency response plan or leak

detection and response, and did not perorm a hazard assessment on the hydrogen process.

Although the Combustible Dust NEP encourages but does not require state plans to adopt the

NEP, Tennessee OSHA adopted the dust NEP in March 2008.

The NEP allows OSHA Area Ofces to add NAICS codes to the list o acilities targeted by

the NEP. However, TOSHA has not added the NAICS code that includes Hoeganaes as o the

issuance o this study.

4.5.3 METaL DUST awarENESS

Since Hoeganaes has been in operation, several opportunities have arisen to increase aware-

ness and address metal dust issues at the acility through technical literature, audits, inspec-

tions, and regulatory oversight. These resources have not been eectively used by Hoeganaes.

Gaps in codes and regulations, inadequate inspections, and poor hazard recognition all

contributed to the three incidents at the Gallatin acility.

Table 4 lists a timeline o events rom 1956 to the present related to combustible metal dust

and the lack o eective controls at Hoeganaes until the third incident in May 2011.


24/31



4.6 MeTal powDer proDuCers assoCIaTIoN

The Metal Powder Producers Association (MPPA) is one o six trade associations that make up

the Metal Powder Industries Federation (MPIF). MPPA membership is open to manuacturers ometal powders, metal akes, metal fbers, or non-metallic powder additives used with these ma-

terials. The stated objective o the MPPA is to arrange or the collection and dissemination o

inormation pertaining to the metal powder producing industries; provide technical acts, data,

and standards, undamental to metal powders and to the applications o metal powders...36

TABLE 4

Timeline o metal dst

pblications, oversiht,

and opportnities to

address metal dst

hazards in indstry and at

Hoeanaes.

34OShA Oice o Program Evaluation, Regulatory Review o OShAs Grain handling Facilities Standard (29 CFR 1910.272).February 2003.

35OShA regulated program to prevent noise induced earing loss (29 CFR 1910.35).

36ttp://www.mpi.org/aboutmpi/mppa.asp.

KEY MILESTONES FOR COMBUSTIBLE METAL DUST CONTROL IN INDUSTRY AND AT HOEGANAES

Year Action

1946 NFPA pblishes Code or the Prevention o Dst Explosions that incldes eneral reqire-

ments or metal dsts

1958 American Chemical Society pblication discsses powdered metals and iron dst

explosibility

1980 Hoeanaes gallatin acility established

1987 OSHA promlates grain Dst Standard, which decreases the nmber o explosions 44%

and atalities 60%34

1992 Hydroen explosion and iron dst fash re severely brns employee at the Hoeanaes

acility in Cinnaminson, NJ

1996 Employee o Hoeanes gallatin acility sers smoke inhalation in dst collector re.

Metal dst inites inside dst collector (inited drin a cttin operation)

1999 gallatin FD inspects Hoeanaes; no mention o dst accmlations

2002 gallatin FD inspects Hoeanaes; no mention o dst accmlations

2002 NFPA 484 is issed which addresses additional combstible metals that wold inclde

iron dst

2006 The CSB isses recommendation to OSHA to promlate a comprehensive combstible

dst standard

2006 City o gallatin adopts International Bildin and Fire Codes

2007 OSHA isses Combstible Dst National Emphasis Proram (NEP)

2008 Tennessee OSHA adopts ederal Combstible Dst NEP (March)

2008 Tennessee OSHA inspects Hoeanaes acility. Condcts respirable metal dst samplin.

Cites Hoeanaes or hearin conservation35 (Oct.) No observation o a combstible dst

hazard

2008 Allianz condcts insrance adit at Hoeanaes, recommends combstible dst testin

and independent consltant (Nov.)

2009 Hoeanaes condcts combstible dst testin o three iron powder samples as recom-

mended by 2008 Allianz adit; all three samples ond to be combstible.

2010 Hoeanaes condcts combstible dst testin o 23 powdered metals; 5 o 9 iron samples

ond to be combstible. No sbstantial actions to mitiate combstible dst hazard

2011 Janary 31 incident; atal combstible dst fash re at Hoeanaes gallatin acility

2011 March 29 incident; combstible dst fash re at Hoeanaes gallatin acility

2011 gallatin FD inspects Hoeanaes, no mention o dst accmlations (May 11)

2011 May 27 incident; atal hydroen explosion/combstible dst fash re at Hoeanaes

gallatin acility


25/31



To achieve this, the MPPA oers a monthly newsletter and bi-annual meetings (all and spring)

to promote shared learning within the PM industry. In recent years, the spring meeting, which

spotlights many saety topics, has begun ocusing on the issue o combustible dust hazards in

the PM industry. The MPPA has sought out external combustible dust expertise and OSHA has

presented and participated in its saety meetings.

5.0 KEY FINDINGS

Over the course o investigating the events at the Hoeganaes acility, the CSB made the ollow-

ing key fndings:

1. Signifcant accumulations o combustible iron powder at the Hoeganaes acility ueled

atal ash fres when loted near an ignition source.

2. Hoeganaes acility management were aware o the iron powder combustibility hazard

two years prior to the atal ash fre incidents but did not take necessary action to

mitigate the hazard through engineering controls and housekeeping.

3. Hoeganaes did not institute procedures such as combustible gas monitoring or train -ing or employees to avoid ammable gas fres and explosions.

4. OSHA did not include iron and steel mills (NAICS code 331111), the industry classifca-

tion code or Hoeganaes, in its Combustible Dust National Emphasis Program when it

was frst issued in 2007 or when it was re-issued in 2008.

5. The 2006 International Fire Code Chapter 13, Combustible Dusts, which was adopted

by the City o Gallatin at the time o the incidents, does not clearly require jurisdictions to

enorce the more comprehensive and rigorous NFPA standards or the prevention o dust

fres and explosions.

6. The Tennessee Fire Code and the City o Gallatin do not enorce optional or recom-

mended standards or practices o the IFC.

7. The Gallatin Fire Department inspected the Hoeganaes acility ater the frst two iron

powder ash fres but did not cite or otherwise address combustible dust hazards present

at the acility just weeks beore the third atal hydrogen explosion and dust ash fre.

8. The ame-resistant clothing (FRC) supplied by Hoeganaes to its employees did not

provide any signifcant protection against the combustible iron dust ash fres and the

hydrogen explosion that caused the atalities.

9. GKN and Hoeganaes did not provide corporate oversight to ensure the Hoeganaes

Gallatin acility was adequately managing combustible dusts prior to and throughout the

succession o serious incidents at the Gallatin acility.

6.0 RECOMMENDATIONS

6.1 osHa

2011-4-I-TN-r1

Ensure that the orthcoming OSHA Combustible Dust Standard includes coverage or com-

bustible metal dusts including iron and steel powders.


26/31



2011-4-I-TN-r2

Develop and publish a proposed combustible dust standard or gen-

eral industry within one year o the approval o this case study.

2011-4-I-TN-r3

Revise the Combustible Dust National Emphasis Program (NEP) to add industry codes or

acilities that generate metal dusts (e.g., North American Industrial Classifcation System,NAICS, code 331111 Iron and Steel Mills, and other applicable codes not currently listed).

Send notifcation letters to all acilities nationwide under these codes to inorm them o the

hazards o combustible metal dusts and NEP coverage.

6.2 INTerNaTIoNal CoDe CouNCIl

2011-4-I-TN-r4

Revise IFC Chapter 2237 Combustible Dust Producing Operations; Section 2204.1 Standards,

to require mandatory compliance and enorcement with the detailed requirements o the

NFPA standards cited in the chapter, including NFPA 484.

6.3 TosHa

2011-4-I-TN-r5

Revise the state-adopted Dust National Emphasis Program (NEP) to add industry codes or

acilities that generate metal dusts (e.g., North American Industrial Classifcation System,

NAICS, code 331111 Iron and Steel Mills, and other applicable codes not currently listed).

Send notifcation letters to all acilities statewide under these codes to inorm them o the

hazards o combustible metal dusts and NEP coverage.

6.4 HoeGaNaes

2011-4-I-TN-r6

Conduct periodic independent audits o the Hoeganaes Gallatin acility or compliance with

the ollowing NFPA standards, using knowledgeable experts, and implement all recom-

mended corrective actions:

NFPA 484, Standard or Combustible Metals, Metal Powders, and Metal Dusts

NFPA 499, Recommended Practice or the Classifcation o Combustible Dusts and o

Hazardous Locations or Electrical Installations in Chemical Process Areas

NFPA 497, Recommended Practice or the Classifcation o Flammable Liquids,

Gases, or Vapors and o Hazardous (Classifed) Locations or Electrical Installations in

Chemical Process Areas

NFPA 2, Hydrogen Technologies Code

NFPA 2113, Standard on Selection, Care, Use, and Maintenance o Flame-ResistantGarments or Protection o Industrial Personnel Against Flash Fire

37Combustible Dust Producing Operations, Capter 13 o te IFC, was moved to Capter 22 in later editions.


27/31



2011-4-I-TN-r7

Develop training materials that address combustible dust and plant-specifc metal dust hazards

and train all employees and contractors. Require periodic (e.g., annual) reresher training or all

employees and contractors.

2011-4-I-TN-r8

Implement a preventive maintenance program and leak detection and leak mitigationprocedures or all ammable gas piping and gas processing equipment.

2011-4-I-TN-r9

Develop and implement a near-miss reporting and investigation policy that includes the

ollowing at a minimum:

Ensure acility-wide worker participation in reporting all near-miss events and opera-

tional disruptions (such as signifcant iron powder accumulations, smoldering fres, or

unsae conditions or practices) that could result in worker injury.

Ensure that the near-miss reporting program requires prompt investigations, as appropri-

ate, and that results are promptly circulated throughout the Hoeganaes Corporation.

Establish roles and responsibilities for the management, execution, and resolution of all

recommendations rom near-miss investigations

Ensure the near-miss program is operational at all times (e.g. nights, weekends,

holiday shits).

6.5 MeTal powDer proDuCers assoCIaTIoN (Mppa)

2011-4-I-TN-r10

Communicate the fndings o this report to all your members, e.g. through a saety article in

an upcoming monthly newsletter.

6.6 CITy oF GallaTIN, TN

2011-4-I-TN-r11

Require all acilities covered by IFC Chapter 13 (2006 edition) to conorm to NFPA standards

or combustible dusts including NFPA 484.

6.7 GallaTIN FIre DeparTMeNT

2011-4-I-TN-r12

Ensure that all industrial acilities in the City o Gallatin are inspected periodically against the

International Fire Code. All acility inspections shall be documented.

2011-4-I-TN-r13

Implement a program to ensure that fre inspectors and response personnel are trained to

recognize and address combustible dust hazards.


28/31



APPENDIx A: DETERMINATION OF IRON POWDERExPLOSIBILITY FROM PRESSURE RATIO CALCULATIONS

baCKGrouND

The CSB commissioned testing o our Hoeganaes iron dust samples in a 20-L Test Chamber

(ASTM Standard E1226-10, Standard Test Method or Explosibility o Dust Clouds). Thetest method states that dust explosibility can also be characterized through the calculation o

a pressure ratio (PR). The pressure ratio calculation is also known as the explosibility screen-

ing test (Section 13). The 20-L test chamber records the maximum explosion pressure reached

during a single deagration test at a dust concentration o 1,000 and 2,000 g/m 3 .

I PR > 2, the dust is considered explosible

I PR < 2, it is classifed as not explosible under those test conditions and the standard

goes on to caution that it is not necessarily not combustible and still may be capable o

deagrative combustion.

CalCulaTIoNs

The CSB calculated the pressure ratio based on the 20-L test results rom the baghouse dust.

According to E1226-10:

Pressure ratio = PR = (Pex,a

Pignitor

) Pignition

Where:

Pex,a

= maximum explosion pressure (bar absolute)P

ignitor= maximum pressure rise in chamber due to igniter = 2.5 (bar absolute)

Pignition = absolute pressure at the time o ignition = 0 bar gauge = 1.0123 (bar absolute)

The test value Pmax,a

already corrects or the igniter pressure:

Pmax,a

= (Pex,a

Pignitor

)

Pmax,a

values obtained rom 20L testing o Hoeganaes dust:

Convert bar gauge to bar absolute:

0 bar gauge= 1.0125 bar absolute3.1 bar gauge + 1.01325 = 4.1 bar absolute

3.8 bar gauge + 1.0125 = 4.8 bar absolute

Pmax, a

DUST CONCENTRATION

3.1 bar ae 1,000 /m3

3.8 bar ae 2,000 /m3


29/31



The equation reduces to:

PR = Pmax,a

/ Pignition

Thus:

PR1,000 = 4.1/1.01325PR

1,000= 4.0

PR2,000

= 4.8/1.01325

PR2,000

= 4.7

The pressure ratios at 1,000 and 2,000 g/m3 are both greater than 2 and the iron dust sample

is considered explosible based on ASTM E1226.


30/31



APPENDIx B: DETERMINATION OF IRON POWDERCLASSIFICATION FROM ExPLOSION SEVERITYCALCULATION

baCKGrouND

The CSB commissioned testing o our Hoeganaes iron dust samples in a 20-L Test

Chamber (ASTM Standard E1226-10, Standard Test Method or Explosibility o

Dust Clouds). OSHA cites the National Materials Advisory Board (NMAB) 353-3-80,

Classifcation o Combustible Dusts in Accordance with the National Electric Code, or

the determination o a Class II combustible dust location. The NMAB 353-3-80 states that

Class II dusts can be characterized through the calculation o an explosion severity (ES).

According to the NMAB 353-3-80:

I ES > 0.5, the dust is considered an appreciable explosion hazard that requires suitable

electrical equipment or Class II locations.

CalCulaTIoNs

The CSB calculated the pressure ratio based on the 20-L test results rom the baghouse dust.

According to NMAB 353-3-80:

Explosion Severity = ES =

Where:

Pmax

= maximum explosion pressure, (bar gauge)

= maximum rate o pressure rise, (bar gauge per second)Reerence dust = Pittsburgh coal dust

Pmax

values obtained rom 20L testing o Hoeganaes dust:

DUST SAMPLE Pmax

P/tmax

Hoeanaes Bahose 3.5 bar ae 68 bar ae per second

Pittsbrh Coal 7.3 bar ae 426 bar ae per second

Pmax(sample)

x

Pmax(reerence dust)

x

tmax(sample)

tmax(reerence dust)

P

P

tmax(reerence dust)

P


31/31


Thus:

The explosion severity o the baghouse dust is less than 0.5 and the iron dust sample is not

considered a Class II combustible dust.

238

3,110ES =

ES = 0.077

(3.5 bar gauge) x (68 )s

bar gauge

ES =

(7.3 bar gauge) x (426 )s

bar gauge

CSB Case Study Hoeganaes Dec9 Final

Documents

Transcript of CSB Case Study Hoeganaes Dec9 Final