Determination of Minimum Inerting Concentrations for Combustible Dusts in a Laboratory-Scale Chamber

0957–5820/02/$23.50+0.00# Institution of Chemical Engineers

www.ingentaselect.com=titles=09575820.htm Trans IChemE, Vol 80, Part B, November 2002

DETERMINATION OF MINIMUM INERTINGCONCENTRATIONS FOR COMBUSTIBLE DUSTS

IN A LABORATORY-SCALE CHAMBERA. DASTIDAR* and P. AMYOTTE

Department of Chemical Engineering, Dalhousie University, Halifax, Canada

E xperiments were performed to compare intermediate-scale (1 m3) and laboratory-scale(20 L) inerting results. In general, laboratory-scale inerting levels were higher thanintermediate-scale values. This can be attributed to the use of a strong ignition source to

initiate the test, which may have overdriven the explosions in the smaller test vessel. Previouslyreported agreement between the smaller test vessel and full-scale experiments may be due tooverdriving in the 20 L chamber, leading to high inerting levels similar to those encountered infull-scale tests due to � ame acceleration. Use of weaker ignition sources in the laboratory-scalechamber did produce inerting levels similar to those observed in the intermediate-scale vessel.A new � ammability limit parameter has been de� ned as the minimum inerting concentration(MIC; in units of mass concentration, i.e., g m¡3). This is the concentration of inertant requiredto prevent a dust explosion regardless of fuel concentration. Previous experimental work in a1-m3 spherical chamber has shown this � ammability limit to exist for some fuels. In the currentwork, inerting experiments were performed in a 20-L Siwek chamber using identical materialsto those used in the 1-m3 chamber. The results show that an MIC can be determined in thesmaller test chamber; however, there is a strong dependence on ignition energy strength used toinitiate the explosion.In the 20-L tests, as in the 1-m3 tests, not all combustible dust and inertant mixtures showed ade� nitive MIC, although they did show a strong dependence between inerting level andsuspended fuel concentration. As the fuel concentration increased, the amount of inertantrequired to prevent an explosion decreased. Even though a de� nitive MIC was not found for allof the dusts, an effective MIC can be estimated from the data. The use of MIC data can aid inthe design of explosion suppression schemes.

Keywords: dust explosions; inerting; explosion mitigation.

INTRODUCTION

Dust explosions are a serious hazard in the process indus-tries and suppressants are widely used to provide protectionagainst such explosions. For suppression to be effective, it isimportant to determine the amount of inertant required toprevent an explosion of the dust at any concentration.Inerting concentrations can be determined by performingsuppression tests with laboratory-scale equipment or withlarge- or full-scale equipment. Laboratory-scale tests areoften carried out because of the cost and effort required forfull-scale testing.

An additional � ammability limit parameter can be de� nedfor inerting dust clouds with particulate material. This newparameter is called the minimum inerting concentration(MIC; in units of mass concentration, i.e., g m¡3). Theminimum inerting concentration is the lowest inert dust

concentration at which a dust explosion is not possible forany fuel concentration. MIC values can be determined bymeasuring the full � ammability curve for the fuel–air–inertdust mixture. This approach differs from suppression testingin that the fuel and inertant are premixed and collectivelyinjected into the test vessel.

The use of the MIC is an important component in thedesign of explosion suppression systems for industrialequipment and chemical isolation barriers in pipes. Suppres-sion and isolation designs having inert agent concentrationsbelow the MIC will have the possibility of allowing dustexplosions to occur. Therefore, the MIC provides guidancefor the safe design of suppression and chemical barriersystems.

Previous experiments1,2 conducted with the Fike 1-m3

spherical chamber (Blue Springs, MO) used � ve differentcombustible dusts: aluminum, anthraquinone, polyethylene,Pittsburgh pulverized coal, and cornstarch. The inertantsused were limestone, monoammonium phosphate, andsodium bicarbonate, all of which have been shown to beeffective explosion suppressants. Because it is economically

287

*Currently at Safety Consulting Engineers, Inc., Schaumburg, Illinois, USA60173.

advantageous to perform MIC tests in smaller explosionvessels, these same dusts (fuel and inertant) were tested inthe current work in a 20-L Siwek chamber (Kuhner AG,Birsfelden, Switzerland). The primary consideration was tostudy the in� uence of test vessel volume on the MIC. Toosmall a vessel will not allow for proper � ame front devel-opment and may underestimate the inerting requirement. Arelated parameter that was investigated is the applied igni-tion energy. An ignition source that is too large may lead tooverdriving the explosion, thus overestimating the inertingrequirement.

BACKGROUND

Figure 1, from Moore and Siwek3, is a pictorial repre-sentation of suppressant requirements to mitigate an explo-sion coupled with suppressant delivery from high-ratedischarge canisters. The quantity of suppressant requiredto mitigate an explosion begins at some initial value at zerotime, and increases rapidly as the explosion � reballincreases in size. This initial amount of suppressant isdependent upon material properties and the nature of theignition source. The upward curvature of the line is depen-dent upon the nature of � ame propagation of the material ina given vessel geometry.

The delivery of the suppressant from the discharge canis-ters begins after an initial delay, which is dependent upon theexplosion detection system. The curve then increases rapidlyand begins to level off as the suppressant propellant is spent.For adequate suppression of an explosion, the suppressantdelivery curve must cross and exceed the suppressantrequirement curve. The nature and curvature of the sup-pressant delivery curve is dependent upon the designcharacteristics of the suppressant device (e.g., number and

type of detectors, number and size of discharge canisters,and � ow rate from the discharge canisters).

A key factor in determining if the two curves cross is theintersection of the curve showing the quantity of suppressantrequired to mitigate an explosion with the ordinate (i.e., thesuppressant requirement at zero time). This value isthe inerting level, or the suppressant requirement when thesuppressant and the fuel are intimately mixed. In a dustexplosion situation the quantity of fuel present in theoperating unit may not be known; therefore it wouldbe bene� cial to know the suppressant requirements at zerotime regardless of fuel concentration. This is provided by theMIC. Being able to experimentally obtain MICs similar tothose found in the 1-m3 inerting work1,2 in a smallerchamber (e.g., 20-L Siwek chamber), would greatly assistin reducing the cost accrued in explosion suppressiondesign.

Previous work by Dastidar et al.4 has shown that there is astrong interdependence of the relative amount of premixedinertant required to prevent an explosion and the fuelconcentration. The authors report from tests of Pittsburghand Pocahontas pulverized coals that fuel concentrationsnear stoichiometric volatile-air mixtures required more lime-stone to prevent an explosion than higher fuel concentra-tions. These stoichiometric volatile-air mixtures representthe fuel concentration that produces the maximum explosionoverpressure. At volatile-air ratios in excess of stoichio-metric there is insuf� cient oxygen for complete combustion,and the unreacted material acts as a thermal sink, helping toinert the explosion. As a result, a decrease in overpressure isobserved. For inerting work this implies that at high fuelconcentrations the excess fuel dust acts to increase theeffective thermal sink in the dust mixture. Since the fuelaids the inertant, there is less inertant needed to prevent anexplosion.

The same point was argued by Mintz5; by increasingthe concentration of fuel in a � xed volume chamber, a densedust cloud is formed. In such a cloud, when a fuel particleburns, all the oxygen in its immediate vicinity is consumedin the reaction zone. This, however, will also consume theoxygen that would be required by the neighbouring fuelparticle for combustion. The transient nature of the combus-tion propagation wave also means that there is insuf� cienttime for oxygen to be transported to the particle in a densedust cloud. This implies that even though there may besuf� cient oxygen for combustion globally, there is notenough oxygen in the local vicinity. As a result, theneighbouring particle does not burn. This unburned particlethen acts as a heat sink reducing the � ame temperature andthe explosion overpressure. By adding an inertant to the dustmixture, a dense dust cloud is formed at low fuel concen-trations thereby hindering the transport of oxygen to the fuelparticle in time for combustion to occur, as well as acting asa thermal sink.

Inerting experiments (1 m3) at Fike1,2 have shown asimilar concentration dependence of explosion inerting.By plotting the data as fuel and inertant concentrations,the results form an inerting envelope similar to a traditional� ammability envelope. Figure 2 shows the inerting enve-lope for Pittsburgh pulverized coal1,2. The data pointsrepresent the explosibility limit of the fuel–inertant mixture.They are interpolated by averaging the highest inertantconcentration tested that produced an explosion with theFigure 1. Suppressant requirement and delivery curve3.

Trans IChemE, Vol 80, Part B, November 2002

288 DASTIDAR AND AMYOTTE

lowest inertant concentration tested that did not produce anexplosion for a given fuel concentration. The area to the leftof the curve represents the explosible area. In this area thereis insuf� cient inertant to prevent an explosion. The area tothe right of the curve represents the nonexplosible area;here there is suf� cient inertant to prevent an explosion. The‘nose’ of the envelope represents the least amount ofinertant that would prevent an explosion regardless of fuelconcentration. This point is the MIC, the least amountof inertant required to mitigate an explosion, when mixedintimately with the fuel. It is also the concentrationof inertant (independent of fuel quantity) at which, fromFigure 1, the suppressant-required curve intersects theordinate of the graph in the ideal situation where � amepropagation is independent of ignition strength. In practice,however, it should be noted that factors such as initiationpoint location(s) and ignition energy density may contributeto this intersection point. Thus, the intersection point onFigure 1 may be greater than the MIC.

The data trend shown in Figure 2 provides a usefulplatform from which adequate protection measures can beimplemented. Since it is dif� cult to determine at any giventime the actual suspended dust concentration in processequipment, or even the potential suspended dust concentra-tion from a process upset, estimates are made in designingexplosion protection equipment. One of these estimates is totake the worst case scenario and perform inerting experi-ments at the fuel concentration that produces the highestoverpressure. This results in a relative amount of inertantrequired to prevent an explosion. The inerting level sodetermined cannot be used as an absolute concentration ofinertant because higher fuel concentrations would requiregreater amounts of inert material to prevent an explosion(even though the relative concentration may be less). If theinerting level was to be maintained as a constant ratio, largequantities of inertant would be required at high fuel concen-trations. While this may guarantee safety, it poses certainequipment design problems. Suppression devices have tobe very large and complicated to be able to introduce suchlarge quantities into the process unit, and the process unitsthemselves have to be designed to accommodate these largersuppression devices.

If, however, an MIC were to exist for a dust, suppressionstrategies could be designed to introduce a � xed concentra-tion of inertant. A � xed number of high-rate dischargecontainers could be attached to a process unit to deliverthis � xed concentration of inertant. At both low and highfuel concentrations there would be enough inertant toprevent an explosion. Thus, if the concentration of explo-sible dust were to increase or decrease in the unit, therewould always be suf� cient inertant to suppress an explosion.

A key point that must be made is that the MIC is anignition limit, not a suppression limit. This is due to the factthat the inert matter is premixed with the fuel when injectedinto the test vessel. This intimate contact between the fueland inertant represents an ideal case. For suppression theremay be losses in inertant effectiveness due to the type ofdispersion from the discharge canister and incompletemixing between the inertant and the fuel.

EXPERIMENTAL

Experiments were performed in a 20-L Siwek chamber6,7.A schematic diagram of the chamber and auxiliary equip-ment is shown in Figure 3. The spherical test chamber has avolume of 20 L and is made of stainless steel with amaximum allowable working pressure of 20 bar(g). Thevessel is surrounded by a water jacket, which is used tocontrol the initial temperature of the test chamber. The topaccess cover is � tted with ignition leads. Chemical ignitors,similar to those used in the 1-m3 chamber, are connected tothe ignition leads. The ignitors are manufactured by Sobbe(Germany) and are composed of 40% zirconium, 30%barium nitrate, and 30% barium peroxide. An ignitor witha stored energy of 5 kJ has 1.2 g of this mixture and releasesits energy in about 10 ms.

The desired amounts of fuel dust and inert dust for agiven experiment are mixed and placed in the dust storagechamber (volume of 0.6 L). The ignitors are attached to theignition leads and the test chamber is sealed and partiallyevacuated to 0.4 bar(a). The computer control program isthen initiated and the dust storage chamber pressurized to20 bar(g) with extra dry compressed air. The computer then

Figure 2. Inerting envelope for Pittsburgh pulverized coal dust with sodiumbicarbonate, monoammonium phosphate and limestone as inertants. Experi-ments conducted in a spherical 1-m3 chamber using two 5-kJ ignitors asignition source.

Figure 3. 20-L Siwek chamber.


MINIMUM INERTING CONCENTRATIONS FOR COMBUSTIBLE DUSTS 289

opens the solenoid valve between the dust storage chamberand the test chamber, dispersing the air and dust mixtureinto the test chamber through a Kuhner rebound nozzle andraising the chamber pressure from 0.4 bar(a) to 1 bar(a).

The computer then energizes the ignition source after atime delay of 60 ms. The computer records the pressure–time history of the explosion using two piezoelectric trans-ducers installed � ush with the interior of the chamber. Thepressure–time trace provides the maximum overpressure(Pex), the maximum overpressure due to the explosion ofthe dust mixture (Pm) using a correction algorithm tocompensate for the pressure increase from the ignitor aswell as vessel cooling, and the maximum rate of pressurerise ((dP=dt)m) for a given test. A Pm value of 1 bar(g)(2 bar (a)) was used as the explosion criterion to delineate anonexplosion from an explosion. This is the same criterionas used in our previous work (1 m3 and 20 L)1,2,4.

As previously mentioned, the dusts used were aluminum,anthraquinone, cornstarch, Pittsburgh pulverized coal, andpolyethylene as fuels, and monoammonium phosphate,sodium bicarbonate, and limestone (for cornstarch andPittsburgh pulverized coal dust) as inertants. The choice offuels was made to represent materials commonly used in theprocess industries: a metal dust, a chemical intermediate foran organic dye, a food product, fuel used in energy produc-tion, and a typical polymer. The inertants, monoammoniumphosphate and sodium bicarbonate, are commonly usedsuppressants in the explosion protection � eld. Limestone

is commonly used in mines to prevent coal dust explosions.The experimental test matrix is given in Table 1.

Particle size measurements of all dusts were made using aMalvern Instruments 2600 Series analyzer based on theprinciple of laser diffraction. The results of the particle sizeanalyses are given in Table 2, in which the arithmeticvolume or mass mean diameter, DW, is shown.

The explosion parameters (overpressure and rate of pres-sure rise) of the pure fuel dusts were determined as afunction of concentration. This was achieved by varyingthe amount of dust placed in the discharge chamber. Theeffect of inertant was then determined for each fuel concen-tration by measuring the explosion parameters as inertantwas added to the mixture. Data treatment was carried out asdescribed by the following example.

Figure 4 is a graph of the explosion data previouslydetermined for Pittsburgh pulverized coal dust in the Fike1-m3 chamber using sodium bicarbonate as an inertant. Theopen circles represent experiments where the combustionoverpressure of a given fuel and inertant mixture did notmeet, or exceed, the 1 bar(g) explosion criterion. The closedcircles represent fuel and inertant mixtures where thecriterion was met. The inerting level was interpolated byaveraging the highest inertant concentration tested thatproduced an explosion with the lowest inertant concentra-tion tested that did not produce an explosion for a given fuelconcentration. In this manner the sodium bicarbonate inert-ing envelope given in Figure 2 was constructed. The 20 L

Table 1. Test matrix for experiments conducted in the 20-L Siwek chamber.

Dust Concentration range (g m¡3) Inertant Ignition energy (kJ)

Aluminum 250–1000 Monoammonium phosphate 5Sodium bicarbonate 5

Anthraquinone 250–1000 Monoammonium phosphate 5Sodium bicarbonate 5, 1

Cornstarch 250–1250 Monoammonium phosphate 5Sodium bicarbonate 5, 1Limestone 5

Pittsburgh pulverized coal 250–1500 Monoammonium phosphate 5, 2.5, 1, 0.5Sodium bicarbonate 5, 1Limestone 5

Polyethylene 250–1250 Monoammonium phosphate 5Sodium bicarbonate 5

Table 2. Particle sizes of fuel and inertant dusts used for MIC testing.

Particle size

Dust<100mm(wt%)

<25 mm(wt%)

Volume meandiameter, DW (mm)

Aluminum 90 50 17Anthraquinone 92 32 46Cornstarch 100 87 17Pittsburgh

pulverized coal93 14 60

Polyethylene 83 17 64Monoammoniumphosphate

83 18 62

Sodiumbicarbonate

97 32 37

Limestone 94 49 41 Figure 4. Development of inerting envelope using Pittsburgh pulverizedcoal as fuel and sodium bicarbonate as inertant (1-m3 chamber).



data plots constructed in the current work (Figures 5 to 20)were prepared in a similar manner.

RESULTS AND DISCUSSION

In� uence of Vessel Size

In this section we report on the comparison of inertinglevels measured in the current work (20 L) and in ourprevious 1-m3 experiments1,2. The initial inerting experi-ments in the 20-L chamber were performed with a 5-kJignition source. The selection of this ignition strength wasbased on our previous experimental work4. Results from thisprevious study4 showed good agreement between the mine-scale inerting of coal dust with rock dust, and 20-L experi-ments, when a 5-kJ ignition source was used in the smallerchamber. The standard ignition strength used in the 1-m3

chamber is 10 kJ8,9. Thus the data legends shown in Figures5 to 16 indicate an ignition energy of 5 kJ for the 20-L plotsand 10 kJ for the 1-m3 plots.

A dif� culty often encountered in the � eld of explosionmitigation is the dearth of available experimental data in theopen literature. These data are typically proprietary and theirscarcity not only inhibits the design of explosion mitigationsystems, but also the development of the science in this� eld. In an attempt to widely disseminate such data, we have

provided comprehensive � gures detailing all the experi-mental results from the current work.

Figures 5 and 6 show the results of the inerting tests foraluminum using monoammonium phosphate and sodiumbicarbonate, respectively, as inertants in the 20-L and 1-m3

chambers. For low aluminum dust concentrations (below1000 g m¡3), the inerting levels are in good agreement. Forfuel concentrations of 1000g m¡3 and greater, the concen-tration of inertant required to prevent an explosion in the20-L chamber, using a 5-kJ ignition source, is greater thanthat required in the 1-m3 chamber.

In the 1-m3 curves shown in Figures 5 and 6 there is aslight upward trend of the plotted lines; however, there is noobserved minimum inerting concentration. The curvature isan indication that the higher aluminum concentrationsrequire relatively less inertant to prevent an explosion thanlower concentrations of aluminum. The in� uence of fuelconcentration on inertant requirements for the mitigation ofan explosion is different in the 20-L than in the 1-m3

chamber. In the 20-L vessel, low concentrations of alumi-num required relatively less inertant to prevent an explosionthan higher concentrations of aluminum. From Figure 5(20-L), for 250g m¡3 of aluminum, 50% inert material isrequired to prevent an explosion, and 60% inert material isrequired for 500 g m¡3 of aluminum. The relative inertinglevel then remains constant at 65% for the remaining

Figure 5. Comparison of 1-m3 and 20-L chamber inerting curves foraluminum using monoammonium phosphate as inertant.

Figure 6. Comparison of 1-m3 and 20-L chamber inerting curves foraluminum using sodium bicarbonate as inertant.

Figure 7. Comparison of 1-m3 and 20-L chamber inerting curves foranthraquinone using monoammonium phosphate as inertant.

Figure 8. Comparison of 1-m3 and 20-L chamber inerting curves foranthraquinone using sodium bicarbonate as inertant.



concentrations tested in the 20-L chamber. This is theopposite of the trend observed in the 1-m3 chamber,where the relative concentration of inert matter graduallydecreases.

This difference observed between the relative inertinglevels in the 20-L chamber and the 1-m3 chamber may beattributed to lower explosion overpressures in the 20-Lchamber for pure aluminum. The low explosion overpres-sures observed in the 20-L chamber at low fuel concentra-tions such as 250 g m¡3 indicate an approaching minimumexplosible concentration. Since test conditions are near thelimit � ame concentration, less inert material is required tomitigate the explosion in the 20-L volume than in the 1-m3

volume. Lower explosion overpressures in the 20-L chamber(than in the 1-m3 chamber) at higher fuel concentrationsmay be an indication of wall quenching. The reasoningbehind the need for relatively less inertant to mitigate anexplosion at higher fuel concentrations, above stoichio-metric fuel=air concentrations, is that the excess unburnedfuel in the oxygen-lean propagating � ame front acts as aninerting agent. However, if the � ame front is prematurelyquenched (or partially quenched) by the wall, suf� cientoxygen is present to facilitate combustion. Since theunburned fuel does not assist in the abstraction of heatfrom the reaction zone, the required level of additionalinertant to mitigate combustion remains the same.

A phenomenon observed in the inerting of aluminum inthe 1-m3 chamber was that if insuf� cient inertant is usedin the fuel=inertant mixture, the explosion overpressure isenhanced to levels greater than that for the pure dust1,2.This phenomenon has been termed SEEP (SuppressantEnhanced Explosion Parameter). It can be attributed todecomposition of the inertant by heat from the combustionof the aluminum. In the case of sodium bicarbonate, carbon-dioxide is produced and for monoammonium phosphate,ammonia and hydrogen, both � ammable gases, areproduced. This raises a concern for the inerting of hotburning metal dusts. If insuf� cient suppressant is present,this could lead to greater dangers than if no suppressantwere used. This phenomenon has also been observed in thecurrent work involving 20-L chamber inerting tests,although the extent of enhancement of the explosion over-pressure was not as great as in the 1-m3 chamber. Inboth cases (20-L and 1-m3) the phenomenon was moreapparent in the aluminum=sodium bicarbonate system thanwith aluminum and monoammonium phosphate.

Figures 7 and 8 show the results of the inerting tests foranthraquinone using monoammonium phosphate andsodium bicarbonate, respectively, as inertants in the 20-Land 1-m3 chambers. In all cases, the inerting level in the20-L chamber, using a 5-kJ ignition source, is much greaterthan that observed in the 1-m3 chamber. Although there is

Figure 9. Comparison of 1-m3 and 20-L chamber inerting curves forcornstarch using monoammonium phosphate as inertant.

Figure 10. Comparison of 1-m3 and 20-L chamber inerting curves forcornstarch using sodium bicarbonate as inertant.

Figure 11. Comparison of 1-m3 and 20-L chamber inerting curves forcornstarch using limestone as inertant.

Figure 12. Comparison of 1-m3 and 20-L chamber inerting curves forPittsburgh pulverized coal using monoammonium phosphate as inertant.



some curvature in the plotted inerting levels in the 1-m3

chamber, and even a minimum inerting concentration in thecase of inerting with sodium bicarbonate, this curvature isnot observed in the 20-L results.

In Figure 7 a concentration of 500 g m¡3 of anthraqui-none required 770 g m¡3 of monoammonium phosphate toinert in the 1-m3 chamber, and 1333 g m¡3 to inert in the20-L chamber. These data correspond to a concentration of60% inertant in the 1-m3 chamber and 73% inertant in the20-L chamber. At a fuel concentration of 1000 g m¡3,anthraquinone required 1111 g m¡3 of monoammoniumphosphate to prevent an explosion in the 1-m3 chamber(a relative inertant concentration of 53%). In the 20-Lchamber a higher concentration of inertant, 2667 g m¡3,was required. This is a relative concentration of 73%, thesame relative inerting level as that observed at the lower fuelconcentration.

As seen in Figure 8 at a concentration of anthraquinoneof250 g m¡3, the inerting level in the 1-m3 chamber is875 g m¡3 (relative concentration of 78% inertant). Theinerting level in the 20-L chamber for the same fuelconcentration is 1833 g m¡3 (relative concentration of 88%inertant). Comparison with similar data from Figure 7indicates that in the 20-L chamber, as in the 1-m3 chamber,monoammonium phosphate is a more effective inerting

agent than sodium bicarbonate. Although an MIC of1700 g m¡3 is observed (Figure 8) in the inerting envelopeof anthraquinone with sodium bicarbonate in the 1-m3

chamber, a minimum inerting concentration is not seen inthe 20-L chamber inerting curve using a 5-kJ ignitionsource.

Figures 9, 10 and 11 show the results of the inerting testsfor cornstarch with monoammonium phosphate, sodiumbicarbonate, and limestone, respectively, as inertants in the20-L and 1-m3 chambers. As also seen in the anthraquinoneresults, the inerting levels for cornstarch in the 20-L cham-ber, using a 5-kJ ignition source, are much greater thanthose observed in the 1-m3 chamber.

In Figure 9, which shows some curvature at high fuelconcentrations in the 1-m3 chamber, there is a similarcurvature in the 20-L chamber inerting envelope. The1-m3 results, while not forming a complete envelope,approach an asymptote near 700 g m¡3 of monoammoniumphosphate. The 20-L inerting results have a similar curva-ture and appear to approach an asymptote near 1200 g m¡3

inertant.In Figure 10 a distinct envelope is observed for the

inerting levels in both chambers. The MIC from the 1-m3

chamber inerting envelope is 450 g m¡3. The MIC deter-mined in the 20-L chamber, using a 5-kJ ignition source, is

Figure 13. Comparison of 1-m3 and 20-L chamber inerting curves forPittsburgh pulverized coal using sodium bicarbonate as inertant.

Figure 14. Comparison of 1-m3 and 20-L chamber inerting curves forPittsburgh pulverized coal using limestone as inertant.

Figure 15. Comparison of 1-m3 and 20-L chamber inerting curves forpolyethylene using monoammonium phosphate as inertant.

Figure 16. Comparison of 1-m3 and 20-L chamber inerting curves forpolyethylene using sodium bicarbonate as inertant.



much greater at 1500 g m¡3. As seen in Figure 11 for boththe 1-m3 and 20-L volumes, the plots of the inerting level asa function of fuel concentration do not form an envelopeover the concentration ranges tested.

Figure 12 shows the results of the inerting tests forPittsburgh pulverized coal with monoammonium phosphateas inertant in the 20-L and 1-m3 chambers. As with the othercarbonaceous dusts tested, the inerting level and the MIC aremuch greater in the 20-L chamber, using a 5-kJ ignitionsource, than those observed in the 1-m3 chamber. The MICfor the 1-m3 volume is approximately 100 g m¡3; the 20-LMIC is signi� cantly higher at about 700 g m¡3. In bothchambers there is a distinct curvature of the line depictingthe general trend of the interpolated inerting levels. Thiscurvature is an indication that higher concentrations of fuelpresent in the combustion chamber require less inertantto prevent an explosion than intermediate concentrationsof fuel.

On a relative basis, this concentration level dependencecan be seen in the quantity of inertant required to preventan explosion at 250 g m¡3, at the low end, and 1500 g m¡3

at the upper level of fuel loading. At low fuel concentra-tions the relative concentration of material required to

prevent an explosion is 60% in the 20-L chamber usinga 5-kJ ignitor to initiate combustion in the vessel. At thehigher fuel concentration the relative amount of inertmaterial in the total mixture required to prevent an explo-sion is only 25%.

Figures 13 and 14 show the results of the inerting tests forPittsburgh pulverized coal with sodium bicarbonate andlimestone, respectively, as inertants in the 20-L and 1-m3

chambers. As with the inerting of Pittsburgh pulverized coalwith monoammonium phosphate, the inerting levels deter-mined in the 20-L chamber, using a 5-kJ ignition source, aremuch greater than those measured in the 1-m3 chamber.However, unlike Pittsburgh pulverized coal inerting withmonoammonium phosphate, the lines used to depict thetrends observed in the interpolated inerting levels do notcurve suf� ciently in the concentration ranges tested to forma distinct envelope in the 20-L chamber. Inerting envelopesare, however, formed in the 1-m3 chamber.

Figures 15 and 16 show the results of the inerting tests forpolyethylene with monoammonium phosphate and sodiumbicarbonate, respectively, as inertants in the 20-L and 1-m3

chambers. Again, the inerting level is much greater in the20-L than the 1-m3 volume.

Figure 17. Comparison of 1-m3 and 20-L chamber inerting curves atdifferent ignition energies for anthraquinone using sodium bicarbonate asinertant.

Figure 18. Comparison of 1-m3 and 20-L chamber inerting curves atdifferent ignition energies for Pittsburgh pulverized coal using sodiumbicarbonate as inertant.

Figure 19. Comparison of 1-m3 and 20-L chamber inerting curves atdifferent ignition energies for cornstarch using sodium bicarbonate asinertant.

Figure 20. Comparison of 1-m3 and 20-L chamber inerting curves atdifferent ignition energies for Pittsburgh pulverized coal using mono-ammonium phosphate as inertant.



In� uence of Ignition Energy

As previously mentioned, the initial choice of using a 5-kJignition source in the smaller chamber to determine theinerting level was due to the good agreement this energyproduced with mine-scale results4. Our more recent work2,11

has shown that for coal dust, the use of a 5-kJ ignitor tendsto result in higher inerting levels in the 20-L chamber thanthose observed in the 1-m3 chamber. Additionally, otherresearchers have found that lower ignition energies arerequired to produce minimum explosible concentration(MEC) values in the 20-L chamber comparable toresults obtained from larger scale experiments10. (TheMEC is another limit � ame condition similar to the MIC.)In this section, we therefore report on the comparison ofinerting levels between the 20-L and 1-m3 volumes whenthe applied ignition energy in the 20-L chamber is varied.

Figure 17 depicts the inerting envelope for anthraquinoneusing sodium bicarbonate as an explosion mitigating agent.The 1-m3 data and the 20-L data using a 5-kJ ignition sourceare from Figure 8. However, in the current � gure data frominerting experiments conducted in the 20-L chamber using a1-kJ ignition source are also given. The use of the lowerignition energy to initiate combustion signi� cantly reducesthe inerting level in the 20-L chamber. For example, at a fuelconcentration of 500g m¡3 the 20-L inerting level wasreduced from 2416 g m¡3 to 1750 g m¡3, which is closerto the 1-m3 value of 1333g m¡3. The inerting level decreaseis much greater at higher fuel concentrations. At 1000 g m¡3

the inerting level in the 20-L chamber using a 5-kJ ignitionsource is 3625 g m¡3. This level is decreased to 2000 g m¡3

when a 1-kJ igniter is used in the chamber.This phenomenon may be caused by two factors. First, the

squib ignitors are not point ignition sources but releaseenergy volumetrically. The 5-kJ ignitor behaves as a multi-point volumetric ignition source occupying about one-thirdof the 20-L volume of the smaller chamber12. This can leadto overdriving of the smaller chamber at limit � ame condi-tions where combustion of the dust particles within the large� ame volume generates enough heat to change the testconditions in the chamber, namely temperature and pres-sure. This elevated temperature and pressure condition canrender combustible a fuel=inertant mixture that is noncom-bustible under ambient conditions. Additionally, the burningof the fuel in the � ame volume without true propagation canincrease the pressure of the chamber so as to exceed thelimit explosibility criterion of 1 bar(g). The lower strength1-kJ ignitor would not occupy as much volumetric spaceand therefore not be able to change the dust cloud conditionsas greatly, thereby reducing or eliminating overdriving of thesmaller chamber.

The second factor that may contribute to this phenom-enon involves oxygen diffusion to the reaction zone in densedust cloud conditions. As previously noted, under densecloud conditions a burning particle of fuel may rob itsneighbour of oxygen, thus inhibiting propagation of � ameto it. The larger ignitor ejects more hot particles ofpyrotechnic material to a greater number of multiple loca-tions in the chamber. This induced combustion at multiplepoints does not rob neighbouring particles of oxygen to asgreat an extent as would a lower energy ignitor. The hotejected material from the lower energy ignitor would bemore concentrated in the center of the vessel and would be

‘less multipoint’ in nature. This localized combustion, underdense cloud conditions, may lead to � ame quenching byoxygen starvation of the � ame zone.

Even with the lower ignition energy, the formation of adistinct inerting envelope and subsequent MIC as seen inthe 1-m3 chamber is not observed in the 20-L chamber(Figure 17). The good agreement in the inerting levelobserved between the 20-L chamber using a 5-kJ ignitionsource and mine scale results may be due to overdriving inthe small chamber, thus simulating the in� uence of pressurepiling that is observed in the mine-scale tests. Pressurepiling occurs during � ame propagation in structures thathave a high L=D ratio13.

Figure 18 shows the inerting curve for Pittsburgh pulve-rized coal using sodium bicarbonate as an inertant. The dataof note are the inerting results for the 20-L chamber using a1-kJ ignitor. The trend observed in Figure 17 is also seenhere; the inerting levels at low concentrations more closelyapproximate 1-m3 inerting results. For example, at250 g m¡3 of fuel the 20-L chamber result using a 1-kJignitor is similar to the 1-m3 level: 278g m¡3 and340 g m¡3, respectively. The difference in the inertinglevels from the two chamber volumes increases as the fuelloading increases. Unlike the 1-m3 chamber, no distinctenvelope is formed in the 20-L chamber, even with the 1-kJignitor. The trend does make it appear as if an envelope isbeing formed; however, the high inerting level at1250 g m¡3 of fuel prevents the enclosure from developingcompletely.

Figure 19 shows the inerting of cornstarch using sodiumbicarbonate. Unlike Figures 17 and 18, an envelope andMIC are formed in the 20-L chamber when testing with thelower ignition energy (as well as with 5 kJ). Tests conductedin the 20-L chamber using a 1-kJ ignition source producedinerting results similar to those measured in the 1-m3

chamber. Decreasing the ignition energy in the 20-L cham-ber from 5 kJ to 1 kJ lowered the apparent MIC from1500 g m¡3 to 500g m¡3; the latter value is comparable tothe 1-m3 MIC value of 450g m¡3.

Figure 20 shows the inerting envelope for Pittsburghpulverized coal using monoammonium phosphate as aninertant. Several additional ignition energies were tested inthe 20-L chamber: 2.5 kJ, 1 kJ, and 0.5 kJ. Clearly, decreas-ing the ignition energy from 5 kJ results in inerting levelssimilar to those observed in the 1-m3 chamber. In severalinstances, the inerting level plotted on the graph for the 20-Lchamber, using a 0.5-kJ igniter, overlaps data points fromthe 1-m3 chamber.

It is readily apparent that the ignition energy required toproduce inerting results in the 20-L chamber similar to a1-m3 volume is dependant on the material being tested.Aluminum, a hot burning metal, did not need a reduction inthe ignition energy to produce similar results in bothchambers. On the other hand, the carbon-based dusts clearlydid. This requirement for low ignition energies (0.5–1 kJ)in the 20-L chamber for carbonaceous dusts is an indicationthat ignition energy density is an important considerationin testing for inerting levels. The standard ignition strengthused in the 1-m3 chamber produces an ignition energydensity in the chamber of 10 kJ m¡3. When using a 5-kJignitor for inerting level testing in the 20-L chamber, anignition strength that gives good agreement with mine-scaleexperimental results, the energy density is 250 kJ m¡3.



Decreasing the ignition energy in the smaller chamberbrings the energy density closer to that existing in thelarger chamber. A 1-kJ ignitor produces a nominal energydensity of 50 kJ m¡3 in the 20-L chamber, a 0.5-kJ ignitorproduces half that value. Both these lower energies produceenergy density values of the same order of magnitude as thatin the 1-m3 chamber.

When reviewing this dissimilitude between 20-L, 1-m3

and mine-scale inerting levels (for coal dust) it appears thatenergy density, in conjunction with vessel geometry, mustbe considered. For vessels with high L=D ratios where � ameacceleration due to pressure piling can occur, as in a minegallery, a high energy density is required in the 20-Lchamber to produce similar inerting results. In chamberswith lower L=D ratios where the propagating � ame front inthe burning dust cloud will reach the vessel walls atapproximately the same time (making allowances for buoy-ancy and turbulence factors), a lower ignition energy densityis required.

As demonstrated by the data in Figures 19 and 20, theMIC observed in the 1-m3 chamber can be reproduced in the20-L chamber when low ignition energies are used forinerting level testing. This fact permits the use of the smallervessel, with attention to experimental conditions, in deter-mining the amount of inert matter that will prevent anexplosion regardless of fuel concentration. This quantityof inertant, when mixed intimately together with the fuel,can be used as the zero-time suppressant requirement(Figure 2). This suppressant quantity can then be of assis-tance in constructing a complete suppression scheme. Theuse of the smaller 20-L chamber in determining this valuewill result in time, monetary, and labour savings. Also, animportant point to note when examining the inerting resultsis that the relative effectiveness of the inerting material inmitigating the fuel dust explosion in the 1-m3 chamberis maintained in the 20-L chamber. Sodium bicarbonateis more effective than monoammonium phosphate in miti-gating aluminum dust explosions and monoammonium

phosphate is more effective for carbonaceous dusts inboth chambers.

Table 3 summarizes the minimum inerting concentrationresults from both the 1-m3 chamber and the 20-L Siwekchamber. The cases of cornstarch inerting with monoammo-nium phosphate and sodium bicarbonate, and Pittsburghpulverized coal dust with monoammonium phosphate, showthe presence of an MIC in both chambers. For the caseswhere an MIC was observed in the 1-m3 volume but not inthe 20-L volume (anthraquinone and Pittsburgh pulverizedcoal inerted with sodium bicarbonate), a large reduction ininerting level was observed when the ignition energy wasdecreased from 5 kJ to 1 kJ. Further reduction of the ignitionsource to 0.5 kJ (not tested) may result in the formation of anMIC. Decreasing the ignition source from 5 kJ for testing ofPittsburgh pulverized coal dust with limestone may also leadto the formation of an MIC for this material since an inertingconcentration level dependence on ignition strength hasalready been observed4.

CONCLUSION

The smaller 20-L chamber can be used to produceinerting levels similar to those observed in the 1-m3 cham-ber. For carbonaceous dusts this requires performing explo-sibility tests using ignition energies of 0.5 kJ to 1 kJ. Theonly metal dust tested in the current study produced similarinerting levels in the 20-L chamber as in the 1-m3 chamberwhen using a stronger 5-kJ ignition source.

The dependence of the inerting level on fuel concentra-tion observed in the 1-m3 vessel was observed to a lesserextent in the 20-L chamber. However, for some dusts whichshowed a distinct MIC in the 1-m3 chamber, the presence ofan MIC of similar magnitude was demonstrated when usinglow strength ignitors. This permits the use of the 20-Lchamber in MIC determination for some materials withcareful consideration of test conditions such as ignitionenergy.

Table 3. Summary of MIC results for various fuel and inertant combinations.

Existence of minimum inerting concentration (MIC) (Value g=m3)

Fuel Inertant1-m3 chamber*,10-kJ ignitor

20-L chamber,5-kJ ignitor

20-L chamber,2.5-kJ ignitor

20-L chamber,1-kJ ignitor

20-L chamber,0.5-kJ ignitor

Aluminum Monoammoniumphosphate

No No

Sodium bicarbonate No No

Anthraquinone Monoammoniumphosphate

No No

Sodium bicarbonate Yes (1700g m¡3) No No

Cornstarch Monoammoniumphosphate

Effective (650g m¡3) Effective(1150g m¡3)

Sodium bicarbonate Yes (450gm¡3) Yes (1500g m¡3) Yes (500g m¡3)Limestone No No

Pittsburghpulverized coal

Monoammoniumphosphate

Yes (100gm¡3) Yes (700g m¡3) Effective(300g m¡3)

Yes (150g m¡3) Yes (100g m¡3)

Sodium bicarbonate Yes (600gm¡3) No NoLimestone Yes (1100g m¡3) No

Polyethylene Monoammoniumphosphate

Effective (900g m¡3) No

Sodium bicarbonate Effective (900g m¡3) No

*Experimental data from the 1-m3 chamber presented in this paper are from previous work1,2.



FUTURE WORK

The value of conducting explosion mitigation experi-ments in a 20-L chamber is to minimize the time andexpense of inerting level and MIC determination over thatin the 1-m3 chamber. Further minimization of designexpenses may be achieved by using a mathematical modelto estimate or predict the inerting level and, speci� cally, theminimum inerting concentration. To that end we arecurrently working on a report of the results of a such amodel based on the concept of calculated adiabatic � ametemperature.

REFERENCES

1. Dastidar, A. G., Amyotte, P. R., Going, J. and Chatrathi, K., 1999,Flammability limits of dusts: minimum inerting concentrations,Process Saf Prog, 18: 56–63.

2. Dastidar, A. G., Amyotte, P. R., Going, J. and Chatrathi, K., 1998,Scaling of dust explosion inerting, Archivum Combustionis, 18: 21–45.

3. Moore, P. E. and Siwek, R., 1992, New developments in explosionsuppression, VDI Berichte, 975: 481–505.

4. Dastidar, A. G., Amyotte, P. R. and Pegg, M. J., 1997, Factorsin� uencing the suppression of coal dust explosions, Fuel, 76:663–670.

5. Mintz, K. J., 1993, Upper explosive limit of dusts: experimentalevidence for its existence under certain circumstances, CombustFlame, 94: 125–130.

6. Siwek, R., 1980, Experimental methods for the determination ofexplosion characteristics of combustible dusts, Proc of the 3rd Inter-national Symposium on Loss Prevention and Safety Promotion in theProcess Industries, 3: 839–850 (Basel, Switzerland).

7. Siwek, R., 1977, 20-L Laborapparatur fur die Bestimmung derExplosionskenngrosser brennbarer Staube. (20-L laboratory apparatusfor the determination of the explosion characteristics of � am-mable dusts (Basel-Winterthur Engineering College, Winterthur,Switzerland).

8. ISO 6184/1–1985 (E), 1985, Explosion protection systems—part 1:determination of explosion indices of combustible dusts in air (Inter-national Standardization Organization, Geneva, Switzerland).

9. ISO 6184/4–1985 (E), 1985, Explosion protection systems—part 4:determination of ef� cacy of explosion suppression systems (Inter-national Standardization Organization, Geneva, Switzerland).

10. Cashdollar, K. L. and Chatrathi, K., 1992, Minimum explosibleconcentrations measured in 20-l and 1-m3 chambers, Combust SciTechnol, 87: 157–171.

11. Dastidar, A. G., Amyotte, P. R., Going, J. and Chatrathi, K., 2001,Inerting of coal dust explosions in laboratory- and intermediate-scalechambers, Fuel, 80: 1593–1602.

12. Zhen, G. and Leuckel, W., 1997, Effects of ignitors and turbulence ondust explosions, Journal of Loss Prevention in the Process Industries,10: 317–324.

13. Kumar, R. K., Bowles, E. M. and Mintz, K. J., 1992, Large-scale dustexplosion experiments to determine the effects of scaling on explosionparameters, Combust Flame, 89: 320–332.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the � nancial support of the NovaScotia Technology Transfer and Industrial Innovation Agreement, and theNatural Sciences and Engineering Research Council of Canada. The authorswould also like to acknowledge the support and assistance of John Goingand Kris Chatrathi of Fike Corporation (Blue Springs, MO) for the use oftheir 1-m3 spherical explosion chamber and for supplying the fuel andinertant test materials.

ADDRESS

Correspondence concerning this paper should be addressed toProfessor P. Amyotte, Department of Chemical Engineering, DalhousieUniversity, P.O. Box 1000, Halifax, NS, B3J 2X4 Canada.E-mail: [email protected]

The manuscript was received 6 September 2002 and accepted forpublication after revision 14 November 2002.



Determination of Minimum Inerting Concentrations for Combustible Dusts in a Laboratory-Scale Chamber

Documents

Transcript of Determination of Minimum Inerting Concentrations for Combustible Dusts in a Laboratory-Scale Chamber