Measuring hot-surface minimum ignition temperatures of dust clouds – History, present, future

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries journal homepage: www.elsevier.com/locate/jlp

Measuring hot-surface minimum ignition temperatures of dust clouds – History, present, future

T

Rolf K. Eckhoff University of Bergen, Dept. Physics & Technology, Bergen, Norway

ARTICLE INFO

ABSTRACT

Keywords: Dust clouds Dust explosions Ignition sources Hot surfaces Minimum ignition temperature MIT Laboratory-scale test methods Historical development

Hot surfaces capable of igniting dust clouds can exist in various situations in industry. Typical locations are inside furnaces, burners and dryers of various kinds. Hot surfaces can also arise accidentally e.g. by frictional overheating of bearings and other mechanical parts, and electrically. To avoid accidental ignition of dust clouds by hot surfaces, it is clearly essential to know the minimum temperature of any such surface, at which an explosible dust cloud making contact with the surface, will ignite. The apparatus that is currently used in most countries for this purpose is the Godbert-Greenwald (G-G) furnace. Two pioneering forerunners of the 1935 G-G furnace were developed by M. J. Taffanel and A. Durr (1911) in France, and subsequently by R.V. Wheeler (1913) in UK. However, none of the two methods were intended for assessing maximum permissible temperatures of hot surfaces in industrial plant to prevent hotsurface ignition of dust clouds. The present paper reviews some standardized laboratory-scale apparatuses currently used for this purpose. However, when applying data from such tests, one has to use a safety factor. Current regulations in several countries require that maximum temperatures of hot surfaces in industry that may come in contact with explosible dust clouds, shall not exceed 2/3 of the MIT in oC of the actual dust in air, as measured by one of the current standard methods. It seems unlikely that a more sensible approach will become available in the near future. MIT of a cloud of a given combustible dust in air is not a physical constant. It varies both with dust concentration, turbulence and systematic flow of the dust cloud, and size and shape of the hot surface. Experimental and theoretical evidence shows that MIT decreases with increasing hot-surface size or increasing furnace diameter. Experiments have also shown that MIT increases with increasing dust cloud velocity past relatively small hot surfaces. However, some data indicate that increased dust cloud turbulence can lower MIT. In the literature the two terms minimum ignition temperature (MIT) and auto-ignition temperature (AIT) are sometimes used interchangeably. A possible difference between the two is discussed in the paper with reference to experimental data from hot-surface ignition of explosive gas mixtures. Development of numerical computer models for simulating ignition of dust clouds by hot surfaces is assumed to make further progress. It nevertheless seems unlikely that such models will be widely used in the near future for specifying maximum permissible temperatures of surfaces in industry. One obstacle to general practical use of numerical models seems to be identification of the credible worst-case scenario that can represent a given practical industrial ignition situation. Also, extensive use of sophisticated simulation models will probably be limited by unfavourable cost/benefit considerations.

1. Industrial situation Hot surfaces capable of igniting dust clouds can exist in a number of situations in industry. Inside furnaces, burners, and dryers of various kinds they constitute part of the process equipment itself. Hot surfaces can also be generated by electrical heating and hot work, and by frictional overheating of bearings and other mechanical parts.

E-mail address: [email protected]. https://doi.org/10.1016/j.jlp.2019.02.003 Received 6 February 2019; Accepted 6 February 2019 Available online 15 February 2019 0950-4230/ © 2019 Published by Elsevier Ltd.

In areas in industry where explosible dust clouds may occur, it is clearly important to know the minimum temperatures of any hot surfaces there, at which explosible dust clouds making contact with these surfaces, will ignite. As soon as adequate estimates of these minimum temperatures are available, adequate precautions can be taken to ensure that temperatures of hot surfaces in such areas do not rise to these values.

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

facilitated by photographing the flames (flame size indicated by the dotted pocket below the furnace exit). The results recorded permit the classification of the dusts with reference to their inflammability. The dusts studied by Taffanel and Durr were classified into eight groups, designated A to H in the order of increasing inflammability.”

However, the minimum ignition temperature (MIT) of a cloud of a given dust in air is not a physical constant of that particular dust, but depends both on the concentration of dust in the cloud and the state of cloud movement (turbulence and directional flow), as well as on the size and shape of the hot surface.

The explanation of most of the other letters in Fig. 1 may be as follows: The two glass bottles in the left hand part of the figure comprise the system used for producing the desired, controlled and reproducible flow of air for dispersion the powder/dust sample (o) in the rubber tube (g) into the hot vertical-furnace core (a). The first step in obtaining this was presumably to feed water (w) through the opened valve (s) into the bottle to the far left in Fig. 1. With the valves (r) and (r’) closed, this would raise the air pressure and temperature in that bottle to p’ and t’. By opening and closing valve (r'), with valve (r") left open, and reading the manometer (h), the desired air pressure p in the second glass bottle used for dust dispersion, was obtained. Finally, by releasing the clip (m) on the rubber hose (g) the desired flow of air for dispersing the dust into the hot furnace core was obtained. Brown and James described the furnace used by Taffanel and Durr (1911) in this way: “The furnace consisted of a porcelain tube 10 cm long and 2 cm in diameter, heated electrically, through which dust could be blown downwards by air. The criterion of ignition was the appearance of flame at the bottom of the tube, and the volume of the flame (obtained photographically) was taken as a measure of inflammability. ……… This apparatus is important as the forerunner of many subsequent forms of laboratory furnaces or hot-tube apparatus.” The furnace apparatus developed by the English researcher Wheeler (1913) has sometimes been regarded as the “root” of the modern G-G furnace. However, because of the similar lengths and diameters of the French and English furnaces, it seems likely that Wheeler somehow must have known of the work of Taffanel and Durr. The following introductory note in Wheeler's report gives the background of his work:

2. Laboratory-scale tests for MIT of dust clouds 2.1. Overview by Babrauskas Babrauskas (2003) presented a brief overview of the most commonly used methods for determining MITs of explosible dust clouds. However, the brevity of the review did not permit inclusion of any indepth discussion of the features and performances of the various methods. 2.2. The Godbert-Greenwald (G-G) furnace 2.2.1. The “roots” The G-G furnace is probably the most widely used apparatus internationally for determining MITs of explosible dust clouds. The development history of this furnace started more than 100 years ago. Probably the initiating event was a huge coal dust explosion catastrophe in a coal mine in Courrieres Colliery near the Pas-de-Calais Mountains in France on March 10, 1906. This led to a comprehensive research programme in France, in which the French researcher M. J. Taffanel played a central role. According to Brown and James (1962). Taffanel and his co-worker Durr (1911) were probably the first researchers to use a vertical tubular furnace for obtaining relative measures of minimum ignition temperatures of dust clouds. The present writer was unable to obtain a copy of the original Taffanel/Durr report, but a drawing of their apparatus, reproduced in Fig. 1, is included in a report by Frazer et al. (1913). Explanations of most of the letters in Fig. 1, as given by Frazer at al. (1913), are as follows:

“In consequence of the two serious dust explosions, which occurred in November 1911, at a provender mill in Glasgow and an oil cake factory in Liverpool, a series of experiments has been carried out at the Home Office Experimental Station at Eskmeals by Dr. Wheeler, the chemist attached to the Explosions in Coal Mines Committee, with samples of all kinds of carbonaceous dust, so far as known to the Inspectors, which are liable to be generated on premises under the Factory and Workshop Acts, with a view to determine their degree of inflammability and capacity to transmit explosions … …. Experiments were made with 66 samples in all. The samples were not specially selected, but were in all cases ordinary samples taken by Factory Inspectors from beams, ledges or other projections in the course of their inspections. The Report deserves the careful consideration of all occupiers of factories and workshops in which carbonaceous dusts are generated.”

“A vertical porcelain tube (a) is heated from the outside to the desired temperature, as indicated by the thermocouple (d). The dust under investigation is placed at (o) and is ejected by the compressed air contained in the 1-litre bottle (p). Just previous to the injection of the dust by the current of air, (the tube) (e) is pressed down tightly on (a), (f) making a tight joint between the two. During the passage of the dust laden air through (a), ignition occurs if (a) is sufficiently hot. In addition to determining the temperature of ignition, the investigator should study the size of the flame that issues from the lower end of (a). This study is

Wheeler defined 3 different “inflammability” classes of dusts (= dust clouds in air) as follows: Class I: Dusts which ignite and propagate flame readily, the source of heat required for ignition being comparatively small; such, for example, as a lighted match. Class II: Dusts which are readily ignited, but for which the propagation of flame require a source of heat of large size and high temperature (such as an electric arc), or of long duration (such as the flame of a Bunsen burner). Class III: Dusts which do not appear to be capable of propagating flame under any conditions likely to obtain in a factory; either (a) because they do not readily form a cloud in air, or (b) because they are contaminated with a large quantity of incombustible matter, or (c) because the material of which they are composed does not burn rapidly enough. Wheeler (1913) described two different test methods that he used for determining the class to which a gived dust should be assigned. It is

Fig. 1. Drawing of the Taffanel and Durr (1911) apparatus for measuring the minimum ignition temperatures of coal dust clouds in air. From Frazer at al. (1913). 64

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

Fig. 3. Portraits of the French researcher M. J. Taffanel. and the English researcher R.V. Wheeler from Cybulski (1975).

carbonaceous dusts in both apparatuses, he arrived at the following conclusion: “With a few exceptions the dusts could be grouped in the defined inflammability classes on the principle that Class II contained those dusts which were incapable of propagating a flame when exposed to Test No. 1 (small ignition source of short duration), but which inflamed readily in Test No. 2. The Class I dusts included all the dusts that propagated flame readily even in Test No. 1.”

Fig. 2. Wheeler's “second method”, used for assessing “the lowest temperatures at which ignition can be effected” with dust clouds igniting in Wheeler's first method. From Wheeler (1913).

It is worth mentioning that both Price and Brown (1922), Beyersdorfer (1925), and Brown (1951) also described Wheeler's second test metods in detail, with reference to Fig. 2. Portraits of the two pioneers M. J. Taffanel and R.V. Wheeler are shown in Fig. 3. As already pointed out, the overall purpose of Wheeler's development of his furnace apparatus (Test No. 2) was not to develop a test method for determining MITs of dust clouds in the present meaning of this term. Wheeler did measure the minimum furnace temperatures at which the various dusts ignited, but he did not relate these measurements to the issue of predicting the minimum hot-surface temperatures at which the dusts would ignite in contact with various kinds of hot surfaces in industry. As already said, his objective was to use his two methods for assigning dusts to one of the three flammability classes that he had defined. 16 years after the Wheeler (1913) report Godbert and Wheeler (1929) adapted Wheeler's “dust flammability” concept from 1913 to the more restricted, but indeed very important, area of preventing coal dust explosions in coal mines. More specifically, the idea was to develop a laboratory-scale method by which one could determine “the amount of inert dust required in a mixture with coal dust to suppress the ignition of the dust”. For that purpose they developed and used modified versions of both of the two Wheeler (1913) tests. In both of the modified test methods the dust was blown into the furnace by a blast of oxygen, rather than air. Furthermore, in the case of test No. 2 the temperature of the interior of the furnace was kept constant at 700 °C in all tests. In the Godbert and Wheeler (1929) report no mention was made of the igniting surface being a “loosely rolled spiral of copper gauze” inserted into the heated porcelain tube, as pointed out in the Wheeler (1913) report. It may appear therefore, that in the 1929 version of the furnace, there was no copper gauge spiral. A photograph of the furnace used by Godbert and Wheeler (1929) is shown in Fig. 4. From this report it appears that the length of the heated tube was 11 cm and its internal diameter 2 cm, as opposed to the 10 cm and 2,5 cm of the 1913 furnace. So perhaps a new, slightly different furnace was built for the purpose of the 1929 investigation.

“Test no. 2. Determination of the lowest temperatures at which ignition can be effected”, that is of concern in the present context. The apparatus is illustrated in Fig. 2, and the similarity with Taffanel and Durr's furnace (Fig. 1) is striking. It is clear from the Wheeler (1913) report that his intention with Test No. 2 was not to estimate minimum hot-surface temperatures at which dust clouds would be expected to ignite in contact with hot surfaces in industry, i.e estimating MITs in the current meaning of the term. He described his Test No. 2 as follows: “The igniting surface in this test is a loosely rolled spiral of copper gauze contained in a porcelain tube (A), of 25 mm internal diameter and 10 cm long. This tube is placed vertically as shown in Fig. 2. The sieved and dried dust is introduced into the horizontal dust-tube (B), its weight being 0.2 grams. The wide portion of this dust-tube is then placed vertically over the porcelain tube which is heated by a small electric furnace. A tap which connects with the apparatus for giving a constant puff of air, previously described is now quickly opened. All the dust in the dust-tube is thus projected downwards through the furnace. If the temperature is high enough it ignites there and a flame appears underneath. Before making the series of determinations with the different dusts, a series of experiments was made using dust of the same composition and varying the time of passage of the cloud through the tube. This was done by varying the time of fall of the piston in the apparatus for giving a constant puff of air. It was found, as anticipated, that, up to a certain point, the slower the rate of passage of the dust-cloud through the porcelain tube the lower need the temperature of the latter be to cause ignition.” It is interesting to note that Wheeler considered the “loosely rolled spiral of copper gauze”, inserted into the heated porcelain tube as the real ignition source, rather than the internal surface of the porcelain tube. After having conducted careful experiments with all the 66 65

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

and Wheeler (1929), in which the laboratory-scale data had correlated fairly well with data from some intermediate-scale-gallery coal dust explosion experiments conducted at Buxton in England. In his foreword to the Godbert and Greenwald (1935) report, George S. Rise emphasized that the unique large-scale mine gallery experiments conducted by USBM were of particular significance in the cooperation across the Atlantic Ocean. These experiments had produced invaluable data for the fractions of inert dust that had to be mixed with coal dusts of various categories, to suppress flame propagation on the scale of a real coal mine gallery. The results from these investigations therefore constituted a unique basis for seeking a correlation between large-scale and laboratory-scale data. It was eventually agreed that Godbert, who worked at the Safety in Mines Research Board (SMRB) in Great Britain under the supervision of professor Wheeler, was to visit US Bureau of Mines (USBM) for the two years 1931–32 to work together with the American researcher Greenwald to investigate whether the “second” Wheeler apparatus (Fig. 2), with some modifications, could be made to meet even the American needs. Clearly, Godbert, when joining Greenwald in USA, brought with him the extensive experience he had gained from working with Wheeler on the problem at hand. In their report Godbert and Greenwald (1935) described the laboratory-scale apparatus and procedure that they finally arrived at, as follows:

Fig. 4. Orginal photograph of the Wheeler “Test No. 2 apparatus”. From Godbert and Wheeler (1929).

2.2.2. Development of the “modern” Godbert-Greenwald (G-G) furnace The very first description of the “modern” G-G furnace, having a prolonged heating section of about 20 cm, appears to be that by Godbert and Greenwald (1935). Their report was the end result of a quite long story starting more than one decade earlier. The ultimate substance of the Godbert/Greenwald cooperation seems to be a direct consequence of the work done by Godbert and Wheeler (1929). Again the ultimate objective was to develop a laboratory-scale test that, in a better way than Godbert and Wheeler's 1929 test, could predict whether or not a given coal dust, or a mixture of a coal dust and inert dust, would give dust explosions in large scale mine galleries. In his historical review the American pioneer in the field, George S. Rice (1925) described how this special cooperation between Great Britain and USA actually came about: “Following informal conferences in the summer of 1923 at the Eskmeals Station (on the North-Western coast of England), with Dr. R.V. Wheeler, in charge of the research work under the Safety in Mines Research Board (in Great Britain), ————, a proposal was made by the Secretary for Mines (in Great Britain) to the Director of the Bureau of Mines (in USA), through the Secretaries of State and Interior, for a cooperation on research relating to safety in mining.” It is worth noting that this original formulation of the objective of the cooperation was rather wide and general. The agreement was formally put into force in 1924. Quite a few years should pass before this agreement materialized in the specific cooperation between Godbert and Greenwald that resulted in the G-G- furnace. US Bureau of Mines (USBM) had themselves already been working for some time to develop a laboratory-scale method that could predict whether or not a given dust cloud could propagate a dust flame in large-scale coal mine galleries. As “benchmarks” they were able to use unique data from some tests conducted in their own large-scale experimental mine gallery. However, up to 1931 the efforts in USA to develop a suitable laboratory-scale prediction method had not given satisfactory results. Apparently, therefore, the Americans were very interested in learning from the work conducted by Godbert

“The furnace is mounted on a tripod. The heated central (vertical) tube is cylindrical, of refractory material, and 11/4 inches (32 mm) in inside diameter and 8 inches (203 mm) long. Heat is supplied through a nichrome winding placed in a groove on the outside of the tube. The temperature gradient along the tube is reduced by concentrating the winding towards the ends.” …. “The furnace temperature is measured by a base-metal thermocouple, inserted in the bottom of the tube. The thermojunction is against the tube wall at its midpoint vertically and is kept in that position by a clamp attached to one of the tripod legs.” It is interesting to note that the length of the vertical heated ceramic tube was now about 20 cm, not 10–11 cm as in the previous two Wheeler/Godbert furnaces. Furthermore, no mention was made of the “loosely rolled spiral of copper gauze” inserted into the heated porcelain tube in Wheeler's (1913) original test no. 2. As already mentioned, it appears that this insert was removed even by Godbert and Wheeler (1929). In their report Godbert and Greenwald (1935) concluded that the best agreement between the large-scale mine gallery data and the data from their modified laboratory-scale test method was obtained when all the laboratory-scale tests were conducted with a constant hot-surface temperature in the furnace of 720 °C. As would be expected, the correlation finally obtained between large and small scale tests was not perfect. Therefore, the ultimate question asked, was if it was good enough. At the end of their report Godbert and Greenwald (1935) arrived at the following conclusion: “The results of the tests made in the laboratory apparatus and in the experimental coal mine agree as well as can be expected until experimental mine tests agree better among themselves.” There is a lot to be learnt from this conclusion even today. Large scale experiments do not always provide an ultimate precise answer! 2.2.3. Adoption of the G-G furnace for MIT determination of dust clouds Today the G-G furnace is the most widely used apparatus for standardized experimental estimation of the minimum temperature of any

66

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

hot surface in industry that will ignite a cloud in air of a given combustible dust that makes contact with that surface. A question then is: When and where was the G-G furnace adopted for the first time for this purpose? According to Brown (1951) the exact year of adoption of an about 22 cm long G-G furnace for this purpose in UK, by both SMRB (Safety in Mines Research Board) and FRS (Fire Research Station), was 1949. In this new application of the furnace the automatically controlled temperature of the internal wall of the ceramic furnace core tube was varied systematically in steps of 10 °C and the experiment repeated until a minimum furnace temperature for ignition of the dust tested had been identified. Tonkin and Raftery (1961), in their FRS report on ignitability and explosibility of dusts, refer to their version of the G-G furnace for MIT measurements as “Wheeler's Test No.2 modified”. The most significant modification was probably the doubling of the length of the heated vertical section of the furnace from about 10 cm to about 20 cm, performed by Godbert and Greenwald (1935). The method was also described by Raftery (1968). Palmer (1973) also confirmed that the G-G furnace was in use in UK before 1970 for measuring MITs of dust clouds. He provided a short description of the construction of the furnace and the MIT test procedure used in UK at that time. The earliest mention traced of the use of the G-G furnace in USA (USBM) for measuring MITs of dust clouds, is in a report on inflammability and explosibility of metal powders by Hartmann et al. (1943). In a later summary report on laboratory methods for evaluating ignitability and explosibility of dusts at large, Dorsett et al. (1960) also incorporated the same MIT method as described by Hartmann et al. (1943).

Fig. 5. IEC version of Godbert-Greenwald (G-G) furnace. for determination of MITs of dust clouds. From Eckhoff (2016).

2.2.4. International and national standardization of the G-G furnace in IEC and Europe In 1970 the IEC (International Electrotechnical Commission) decided to establish a working group for development of a series of experimental laboratory-scale test methods for determining ignitability and explosibility data of dusts layers and dust clouds. The exact name of the group was IEC Technical Committee 31, Sub Committee 31H, Working Goup 2. Its first meeting was held at Bundesanstalt für Materialprüfung (BAM) in Berlin, Germany, February 28 - March 2, 1973, and the 2nd meeting at Fire Research Station (FRS), Borehamwood (London), England, September 12–14, 1973. The outstanding secretary/chairman of the group was Ken N. Palmer from FRS. Round about 1970 the present writer had just started the buildingup of a dust ignitability/explosibility test laboratory at Chr. Michelsen Instutute (CMI) in Bergen, Norway. It was therefore a very timely and great privilege and opportunity for him to be allowed to join the IEC working group as the Norwegian delegate at its 3rd meeting in 1974, at US Bureau of Mines (USBM), Pittsburgh, PA, USA. This made it possible also for Norway to take part in all the subsequent “round robin” tests through the following years, of both the G-G furnace, and a series of other test apparatuses/methods developed and standardized by this working group. During the Pittsburgh meeting standardization of the G-G furnace for MIT measurements was a main topic. USBM offered to send each of the participating laboratories 3 tailor-made Alundum core tubes (one for use and two spares) for the G-G furnaces to be built by all the participating laboratories. The special spiral groove on the outer tube surface for guiding the electric-heating wire on the outer tube surface

Fig. 6. Ignition of an aluminium dust/air cloud in the Godbert-Greenwald (G–G) furnace. From Eckhoff (2003).

was an important feature of the tubes. The distance between windings at the tube ends were smaller than at the central parts of the tube, in accordance with the design of Godbert and Greenwald (1935). At the same Pittsburgh meeting FRS, England, offered to send three of their silica core tubes to those of the participating laboratories wanting to consider this option, which included the laboratory in Norway. Even these tubes had the Godbert-Greenwald-type winding grooves on the outside. Eventually a new IEC standard for experimental determination of MITs of dust clouds could be submitted to IEC's Central Office in Geneva. The standard described both the test apparatus and the experimental procedure. Specifications of the way of generating the air

67

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

blast for dispersing the dust were also included. Fig. 5 illustrates a version of the G-G furnace that is essentially in agreement with the first IEC standard. Fig. 6 shows a photograph of a test with fine aluminium dust in the G-G furnace used in Norway. The most recent international standard for determination of MITs of dust clouds constitutes only a minor part of the multi-method standard ISO/IEC (2016). Very few details on apparatus and procedures are specified in the text of the standard. 2.2.5. More detailed studies of the performance of the standard G-Gfurnace 2.2.5.1. Temperature distribution in furnace. It may seem to be tacitly assumed that the temperature along the vertical heated internal surface of the G-G furnace is uniform. However, as Fig. 7 shows, the temperature distribution along the vertical ceramic G-G tube used in Norway, as measured by Scheid and Sand (2000), varied significantly with the location inside the furnace. Fig. 8. BAM furnace for determination of the minimum ignition temperature of dust clouds. From Eckhoff (2003).

The test is repeated until the lowest metal disc temperature that causes ignition of the dust cloud, has been identified. 2.3.2. 1.2 liter US Bureau of Mines furnaces During the 1980-ties and 1990-ties two new alternative furnace

Fig. 7. Temperature distribution vertically along the internal ceramic tube surface of the Godbert-Greenwald furnace used in Norway. From Scheid and Sand (2000).

2.2.5.2. Influence of air pressure used for dust dispersion on measured MIT. Scheid and Sand (2000) also investigated the influence on MIT of the initial pressure of the air in the compressed-air reservoir used for blowing the dust into the standard G-G furnace. The test dust was lycopodium (clavatum) for which they found only a minor influence. However, lycopodium is very easy to disperse in air, and it seems reasonable to expect that for less easy-to-disperse dusts the initial air blast pressure may play a more significant role. 2.3. Other furnaces for MIT measurements than the G-G furnace 2.3.1. BAM furnace As described by Leuschke (1966a, 1966b), an alternative furnace for determination of MITs of dust clouds was developed by Bundesanstalt für Materialprüfung (BAM) in Berlin, Germany. A cross section of this furnace is shown in Fig. 8. The experimental procedure for the BAM furnace is similar to that of the G-G furnace. Originally the generation of the dust cloud was manual, as illustrated in Fig. 8, by pressing a rubber bulb. More recently the dust dispersion system was automated. In the BAM furnace the dust cloud is blown against a vertically oriented circular, concave metal disc (“deflecting surface”) of 20 cm2 area and known surface temperature.

Fig. 9. Cross section of 1.2 liter US Bureau of Mines furnace. for determination of minimum ignition temperatures of dust clouds.

68

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

tests were developed by US Bureau of Mines. The first of these was described in detail by Conti et al. (1983) and a cross section is shown in Fig. 9. The volume of the ceramic furnace chamber is 1.2 liter.

furnace, and the more recent USBM 6.8 liter furnace. ISO/IEC (2016) describes the current ISO/IEC international standard for determination of MITs of dust clouds in a similar way, by providing two specific options for the apparatus, viz. the IEC G-G furnace and the German BAM furnace. It is interesting to note that the BAM, Berlin, Germany, the inventor of the BAM furnace, no longer uses this furnace as their standard method for measuring MITs of dust clouds.

2.3.3. 6.8 liter US Bureau of Mines furnace Conti et al. (1993) described a further development of the 1.2 liter US Bureau of Mines furnace. The modification made was basically a significant increase of the furnace diameter, resulting in a significantly larger furnace volume of 6.8 liters. Fig. 10 shows a cross section of this furnace.

3. Comparison of MIT results from various furnaces 3.1. Standard and modified G-G and BAM furnaces Hensel (1984) compared MITs for a range of dusts, using four different apparatus, viz. the standard and a modified BAM furnace, and the standard and a modified IEC G-G furnace. The BAM furnace was modified by replacing the rubber bulb for manual dust dispersion by an automatic solenoid-valve-operated system, as used with the IEC G-G furnace. Hensel also doubled the length of the G-G furnace from about 20 cm to about 40 cm, which increased the residence time of the dust in the furnace. Hensel's results are summarised in Table 1.

Table 1 Comparison of minimum ignition temperatures of dust clouds in air determined in four different test furnaces. From Hensel (1984). Name of substance

1 Sugar 2 Wheat flour 3 Dextran 4 Lycopodium 5 Tobacco additive 6 Painting powder (a) 7 Painting powder (b) 8 Painting powder (c) 9 Aluminium 10 Alloy (10% Zr) 11 Zircaloy fines 12 Brown coal 13 Pitch coal 14 Lignin 15 Pittsburgh coal 16 Plastic (a) 17 Plastic (b) 18 Plastic (c) 19 Plastic (d) 20 Plastic (e) 21 Chip board dust

Fig. 10. Cross section of 6.8 liter US Bureau of Mines furnace for determination of the minimum ignition temperature of dust clouds. From ASTM (2015a).

In their report on the performance of the 6.8 liter furnace Conti et al. (1993) concluded that in general this furnace gave lower minimum ignition temperatures than the 1.2 liter furnace, but the differences were largely comparatively small (see Fig. 12 in section 3.3 below). From Conti et al. (1983), as reproduced in Eckhoff (2003).

Minimum Ignition temperature of a dust cloud (°C) G.G.-furnace

BAM-furnace

IEC

long

manual

autom.

420 490 400 460 560 570 500 470 560 430 400 410 590 460 580 370 370 490 520 400 510

420 470 380 455 540 570 520 480 515 380 390 400 560 450 570 390 370 480 500 370 480

360 410 370 430 540 530 490 460 510 380 350 440 560 500 590 400 380 480 480 390 450

340 375 340 425 510 490 460 450 510 370 340 410 560 470 580 390 375 450 450 370 450

Maximum temperature difference (K)

80 115 60 35 50 80 60 30 50 60 60 40 30 50 20 30 10 40 70 30 60

With the exception of powders 7, 8 and 16, the prolonged version of the G-G furnace gave the same or somewhat lower values of MIT than the standard version. However, the differences were mostly moderate and not more than 50 °C. In view of this it is interesting to consider Table 2 giving some earlier results obtained by Brown (1951). Brown compared MITs obtained in the Godbert-Wheeler (1929) furnace (heating tube length 11 cm), with MITs obtained in the G-G furnace (heating tube length about 20 cm).

2.4. Current standards for MIT determination of dust clouds ASTM (2015a) describes the standard test method currently used in USA for determining MITs of dust clouds. It is interesting to note that the standard itself only outlines some main required features of the test apparatus, without specifying any specific preferred apparatus in detail. Instead the standard describes four suggested apparatus options, viz. the IEC G-G furnace, the German BAM furnace, the USBM 1.2 liter

69

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

3.2. Standard G-G furnace versus more recent 1.2 liter USBM furnace

Table 2 Comparison of lowest furnace temperatures in oC causing ignition of clouds of some dusts in air, in the original Wheeler furnace (11 cm heating tube length), and in the G-G furnace (about 20 cm heating tube length). From Brown (1951). Dust type

Wheeler furnace

G-G furnace

Sulphur Aluminium laurate Calcium stearate Melamine resin Wood flour Maize starch

270 770 830 770 710 630

290 410 630 540 510 470

Conti and Hertzberg (1987) compared the MITs obtained using their new 1.2 liter USBM furnace with corresponding data from the standard G-G furnace. Table 3 gives some results extracted from a more extensive table in Conti and Hertzberg (1987), which also contains data from other furnace types. As Table 3 shows, the 1.2 liter furnace gave systematically lower MITs than the standard G-G furnace, but the differences vary considerably, from a max. of 90 °C to a min.of 15 °C.

Table 2 shows that 5 of the 6 dusts tested gave a considerably lower minimum furnace temperature for ignition in the G-G furnace than in the original Wheeler (1929) furnace. When combining the data in Tables 1 and 2, it is seen that for most of the dusts the lowest minimum furnace temperature for ignition decreases systematically with increasing furnace length, from 11 cm via about 20 cm to about 40 cm. However, the largest part of the reduction of MIT occurred when increasing the heating tube length from 11 to about 20 cm. Scheid and Sand (2000) found that the blowing of the dust cloud into a G-G furnace in the standard way did not cool the internal surface of the furnace significantly. Wheeler (1913) arrived at the same conclusion for his shorter furnace. This gives support to the suggestion by Hensel (1984) that the reduced MITs found by him by doubling the length of the G-G furnace, was due to increased residence times of the dust clouds in the furnace, rather than reduced cooling of the furnace. In Hensel's experiments with the modified BAM furnace, the modification was just that the manually operated rubber bulb used for blowing the dust into the furnace was replaced by an “automated” air blast generator. As Table 1 also shows, this modification led to just a moderate reduction of the minimum ignition temperatures. In MIT tests with the BAM furnace, ignition is considered to have taken place if it occurs within 10 s after the dust has been blown into the furnace. However, because of the horizontal geometry of the BAM furnace, dusts that do not ignite directly when blown at the vertical heated surface, will settle onto the hot internal bottom of the furnace. Then, with some dusts, smouldering gases may develop, get mixed with air, and get ignited by the vertical hot surface within the 10 s test period. In some cases this kind of gas ignition occurs at a lower hot-surface temperature than that required for direct ignition of the dust cloud. Fig. 11 gives the results of a series of 20 comparative tests with just one dust, showing that the BAM furnace gave a somewhat lower MIT than the G-G furnace.

Table 3 Comparison of MITs obtained in a standard G-G furnace and in a more recent 1.2 liter US Bureau of Mines furnace. From Conti and Hertzberg (1987). Type of dust

G-G- furnace

1.2 liter furnace (oC)

Temp. difference

Pocahontas coal Pittsburgh coal Gilsonite Lycopodium

640 610 580 480

625 540 490 435

15 70 90 45

3.3. 1.2-liter USBM furnace versus 6.8-liter USBM furnace Fig. 12 shows a comparison of MITs obtained for various dust concentrations of lycopodium in the 1.2 litre US Bureau of Mines furnace and the more recent 6.8 litre USBM furnace described by Conti et al. (1993).

Fig. 12. Comparison of minimum ignition temperatures of lycopodium obtained in two US Bureau of Mines furnaces of different diameters and volumes. From Conti et al. (1993).

In Fig. 12 all the experimental points are 6.8 litre-furnace data. The solid line is the borderline between ignitions and non-ignitions for that furnace. For the 1.2 litre furnace no experimental points are given in Fig. 12, only the dotted borderline between ignitions and non-ignitions obtained in an earlier investigation. As Fig. 12 shows, the 6.8 liter furnace gave systematically lower MITs, on average about 50 °C lower, than the 1.2 liter furnace, across the entire dust concentration range investigated. 4. Experimental studies of influences on MIT of dust clouds by various parameters of the hot surface 4.1. Influence of size of hot surface

Fig. 11. Comparison of results from determination of the minimum ignition temperature of American lycopodium by two different furnaces. Volume of dust sample blown into the furnace in each test was 1.6 cm3 for both furnaces. From Eckhoff (2003). Data communicated privately to R.K. Eckhoff in 1975 by G. Leuschke, BAM, Berlin.

As shown by Eckhoff (2016) several independent investigators demonstrated that MITs of explosive gas mixtures decreased systematically with increasing hot surface area. Fig. 13 illustrates that Pinkwasser (1989) found a similar 70

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

Fig. 13. Influence of area of hot surface up to 1000 mm2 on the minimum ignition temperature of clouds in air of three natural organic dusts. Comparison with results from the BAM furnace having a hot surface area of approx. 2000 mm2. From Pinkwasser (1989) as reproduced in Eckhoff (2003).

dependence for dust clouds. The three smallest surfaces were 10 mm long pieces of heated wire of thickness 0.7, 1.2 and 6 mm respectively, bent as a U. The largest surface of 1000 mm2 was obtained by coiling 50 mm of the 6 mm diameter wire. Fig. 13 also gives MITs obtained in the BAM furnace (hot surface area about 2000 mm2, corresponding to a circle of 50 mm diameter) of the three dusts. In view of the results shown in Fig. 13 one may wonder if the lower MITs for the elongated G-G furnace than for the standard one (Table 1) could also be due to a greater hot surface area in the long furnace. However, as discussed above, the difference between the long and standard G-G furnace most probably stems primarily from an increased residence time of the dust cloud in the longer furnace.

Fig. 14. Influence of the residence time of coal dust clouds in the GodbertGreenwald furnace on the minimum ignition temperature. Three different concentrations of a specific brown coal dust. From Griesche and Brandt (1976) as reproduced in Eckhoff (2003).

a minimum ignition temperature of 310 °C. This is lower than all the data in Fig. 14, but about 100 °C higher than the very low value found for a residence time of >1 s and 500 g/m3 dust concentration. This indicates that MITs from standard G-G furnace tests may not be the lowest min. ign. temperatures that may operate in practice in industry.

4.2. Influence of air blast intensity on MIT in standard G-G furnace The first mention of the influence of the air blast intensity on MIT in a furnace apparatus traced, is in the book by Price and Brown (1922). They report as follows: “Before making the series of determinations with the different dusts, a series of experiments was made using dust of the same composition and varying the time of passage of the cloud through the tube. This was done by varying the time of fall of the piston in the apparatus for giving a constant puff of air. It was found, as anticipated, that, up to a certain point, the slower the rate of passage of the dust-cloud through the porcelain tube, the lower need the temperature of the latter be to cause ignition.” Also in their measurements of “relative inflammabilities” of dusts in oxygen, using Wheeler's “second” test method (Fig. 4), Godbert and Wheeler (1929) emphasized “the effect of the pressure of the oxygen blast” used for blowing the dust into the furnace. This was illustrated by different result being obtained using different pressures. Griesche and Brandt (1976) modified the G-G furnace in such a way that dust clouds of known concentrations could be passed through the furnace at a desired known constant velocity. They then investigated the influence of the dust cloud velocity, or the mean residence time of the dust cloud in the furnace, on the minimum ignition temperature of the cloud. The results are given in Fig. 14 and show that MIT decreased quite significantly with increasing residence time of the dust in the hot tube. A conventional Godbert-Greenwald test of the same coal dust gave

4.3. Influence of dust concentration on MIT As shown in Fig. 12, MIT was practically independent of dust concentration across a wide concentration range for both the 1.2-liter US Bureau of Mines furnace and the 6.8-liter US Bureau of Mines furnace. No data have been traced which can confirm whether this independence is generally valid for MITs of dust clouds. In their quite sophisticated studies, Scheid and Sand (2000) found that the dependence of MIT on dust concentration seemed to decrease with increasing relative turbulence intensity of the dust cloud. 4.4. Influence of dust cloud dynamics on MIT In addition to their specific studies of the G-G furnace, Scheid and Sand (2000) also conducted a more fundamental experimental investigation of the ignition of combustible dust clouds by hot surfaces at large. This included studies of the influence on MIT of both the dust cloud velocity parallel with a vertical hot surface, and the dust cloud turbulence intensity. They then used a rather complex experimental facility by which they could determine MITs of dust clouds having various average flow velocities as well as various turbulence levels. A carefully constructed wind tunnel was part of the facility. More than 200 single experiments were performed with lycopodium (clavatum).

71

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

Some experiments were also performed with maize starch. The hotsurface ignition source was a vertically oriented Kanthal plate, of length 50 mm and width 20 mm. The plate temperature was measured by means of a thermocouple inside the hot plate, and also, as a reference, by means of an optical pyrometer located at the opposite wall of the flow channel. Turbulence intensities were reported in terms of relative intensities, i.e. the measured turbulence intensity divided by the mean flow velocity. The turbulence intensities were obtained by means of LDA (Laser Doppler Anemometry) in the wind tunnel, just upstream of the point where the dust was loaded into the air flow. The range of relative turbulence intensities covered in the experiments was 0.015–0.105%. The mean flow velocity range covered was 0.2–7.2 m/s. The main conclusions drawn from the experiments were:

5. Discussion 5.1. “Hot-surface-ignition” versus “auto-ignition” It appears that in the literature on ignition of explosible dust clouds and explosive gas mixtures the terms hot-surface-ignition and auto-ignition are sometimes used interchangeably. The same applies to the corresponding terms minimum ignition temperature (MIT) and auto-ignition temperature (AIT). However, a differentiation between the two concepts would be useful. MIT may then be defined as the minimum temperature that a plane heated surface must have to ignite an explosible dust cloud or explosive gas mixture that is initially at ambient temperature, which makes contact with the hot surface. AIT, on the other hand, may be defined as the minimum temperature to which the entire dust cloud or gas mixture must be heated to ignite spontaneously. It would be anticipated that generally the AITs for given dust clouds would be lower than the MITs. In view of this reasoning, the work of Alfert and Fuhre (1988) on ignition of explosive gas mixtures is interesting. They used mixtures of propane and air (4.5–5.0 vol% propane in most experiments) by hot surfaces in a cubical 50 litre steel box illustrated in Fig. 15. They first covered the top of the box by a sheet of aluminium foil, making provision for two gas outlet tubes. The box was then flushed gently with the propane/air mixture entering the box through an inlet in the box bottom. The rise of the overall fuel concentration in the box was monitored continuously. The hot surface was generated when the overall fuel concentration approached the inlet concentration. When the hot-surface temperature reached 600 °C, the heating rate was reduced to generate a surface temperature increase of a few degrees per minute. Corresponding values of hot-surface temperature and propane concentration in the box were recorded up to the moment of ignition. The results are summarised in Fig. 16 and compared with data obtained using the two standard auto-ignition test methods ASTM (2015b)

a. MIT increased quite strongly and exponentially with increasing dust cloud velocity across hot plate. b. MIT decreased significantly with increasing relative turbulence intensity. c. MITs found in the wind tunnel experiments were considerably higher than those found in the G-G furnace. d. The reciprocal of MITs increased exponentially with increasing dust cloud residence time at the hot surface. e. Ignition mostly appeared to occur at the downstream end of the hot plate. f. Convective heat transfer seems to be the dominant heat transfer mechanism. g. Ignitions seemed to occur within the thermal boundary layer of the hot plate. h. Dependence of MIT on dust concentration seemed to decrease with increasing relative turbulence intensity of the dust cloud. i. Differences between MITs of lycopodium and maize starch in the GG furnace, and in the more advanced set-up used, seemed quite equal.

Fig. 15. 50 litre cubical explosion chamber in which explosive mixtures of propane and air were ignited by various hot surfaces. From Alfert and Fuhre (1988). 72

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

whereas with a hot surface in an open cloud at ambient temperature, a self-propagating flame has to be established in a that cloud. 5.2. Do the standard tests for MITs of dust clouds in fact measure AITs? The 37 mm internal diameter of the heated ceramic tube in the modern G-G furnace may appear sufficiently small for the entire cold dust cloud blown into the tube to be heated to the furnace temperature before the ignition process inside the dust cloud in the tube gets under way. On the other hand, the experiments by Hensel (1984) with a prolonged G-G furnace gave somewhat lower minimum ignition temperatures than the standard furnace, which could imply that in the standard furnace the entire dust cloud may not be heated fully to the furnace temperature before it exits the furnace bottom. As opposed to the G-G- furnace, both the 1.2 litre and the 6.8 litre USBM furnaces are fully closed. It is unclear, however, if the entire dust clouds in these furnaces get heated to the furnace-wall temperature when ignition occurs. A lower temperature in the core of the dust cloud seems more likely in the 6.8 litre furnace than in the slimmer 1.2 litre furnace. In his theoretical analysis Tyler (see 6.4 below) assumed that the entire dust cloud was heated to the vessel wall temperature before ignition got under way. Under these conditions his model predicted that for a given dust cloud AIT would decrease systematically with increasing furnace diameter. A calculated example is given in section 6.4 below.

Fig. 16. Minimum hot-surface temperatures for ignition of stoichiometric propane/air mixture in the cubical chamber illustrated in Fig. 15. Comparison with AITs measured in two standard test and another similar test. From Eckhoff (2016).

5.3. Use of a safety margin on MITs from current laboratory-scale tests In his book Palmer (1973) included a whole, most useful, chapter on “application of results of tests”. The chapter also contains a section on MIT of dust clouds. The following statement by Palmer is indeed relevant for the present discussion: “Under industrial conditions ignition (of a dust cloud by hot surfaces) may in fact occur at a lower temperature than the minimum (ignition temperature) obtained in the (standard G-G furnace) test” Regulations and guidelines in several countries have traditionally required that maximum temperatures of hot surfaces in industry that may make contact with explosible dust clouds, shall not exceed 2/3 of the MIT in oC, as measured by one of the current standard methods. The European standard EN 1127-1 (2007) contains the same requirement. Also in a draft from 2009 of a revision of the same standard, this requirement is maintained. It has, however, not been possible to trace a more recent final standard resulting from this draft. 5.4. Will the 2/3 of MIT requirement remain adequate even in the future? Will a safety margin of 2/3 of MIT in oC remain adequate even in the future? Why could not 3/4, or even 4/5, of MIT in oC be satisfactory safety margins? As long as a clear answer to this question is not available, it may seem advisable to stick to the traditional 2/3 as the general rule. However, there may be situations in industry where the 2/3 requirement may be difficult or even impossible to comply with. In such cases one may be able to justify some reduction of the safety margin, by either conducting tailored experiments, or by numerical simulation.

Fig. 17. Standard IEC apparatus for determination of minimum autoignition temperatures of explosive gas mixtures. From Eckhoff (2016), based on IEC (1975).

and IEC (1975), and a closed-bomb method described by Kong et al. (1995). The IEC (1975) apparatus is illustrated in Fig. 17. As seen in Fig. 16 the lowest temperatures that gave ignition in the experiments in the cubical box (Fig. 15) were 300 °C - 500 °C higher than those obtained in the three auto-ignition tests, including the current American standard ASTM (2015b) and the international standard IEC (1975). A reason for the very low auto-ignition temperatures obtained in the ASTM and IEC standard tests (Fig. 17), could be that in these tests a selfpropagating flame is established in a preheated explosive cloud,

6. Some early theories for predicting MITs/AITs of dust clouds 6.1. Mitsui and Tanaka (1973) Mitsui and Tanaka focused on the influence of particle size on the minimum ignition temperature. They considered a spherical dust cloud inside which heat was generated by combustion, and from which heat was lost due to convection and radiation. They then assumed a combustion rate with Arrhenius type exponential temperature dependence, 73

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

and being proportional to the total particle surface area in the spherical dust cloud. The critical ignition condition was specified as the rate of heat generation due to chemical reaction being equal to the rate of heat loss. The resulting equation seemed to predict an influence of particle size in good agreement with experimental results when using a tuning constant depending on the dust chemistry.

Tyler's theoretical prediction agrees qualitatively with the decrease of experimental AITs when moving from the G-G furnace to the USBM 1.2 liter furnace (diameter ca. 100 mm), as shown in Table 2 above. However, the increase of the furnace size to 6.8 liter (diameter ca. 200 mm) gave a less pronounced reduction of the AITs/MITs of the dust clouds tested, as shown in Fig. 10 above. In practice the situation with very large furnace diameters may approach that of a dust cloud at ambient temperature making contact with a vertical hot surface. Tyler concluded that he had not found any theoretical model by which data from G-G furnace tests could be transformed to minimum ignition temperatures in more complex practical situations in industry. He suggested that stirred reactor ignition experiments, as performed successfully for combustible gas mixtures, may provide a more fundamental understanding of dust cloud ignition processes. Such experiments could yield appropriate activation energies for the ignition processes, which may be used to scale minimum ignition temperatures more reliably. The study of ignition of acetaldehyde/air mixtures, by Harrison and Cairnie (1988) and Harrison et al. (1988), taking this approach, is an example of its potential. However, the present writer is not aware of any such experiments that have been performed at a later time.

6.2. Takigawa and Yoshizaki (1982) A similar study, focusing particularly on the geometry of the Godbert-Greenwald furnace, was undertaken by Takigawa and Yoshizaki (1982). They investigated natural organic dusts and found a reasonably good agreement between measured and numerically predicted dependence of minimum ignition temperature on particle size. The numerical model calculations also revealed that the residence time of the dust particles in the furnace affects the ignition process. It was concluded that a steady-state solution of the minimum ignition temperature is not applicable to the ignition process in the GodbertGreenwald furnace apparatus. Comparison of model predictions with experimental data from other workers confirmed the validity of the predicted effect of the residence time of the dust particles in the furnace on the minimum ignition temperature.

6.5. Higuera et al. (1989)

6.3. Nomura and Callott (1986)

These workers analyzed theoretically the heterogeneous ignition of a cloud of spherical monodisperse coal particles injected instantaneously in the space between two parallel isothermal walls. The effects of the ratio of gas temperature to wall temperature, the conduction/radiation parameter, and the size of the reacting dust cloud relative to the optical length, were explained.

Nomura and Callott modified the Cassel/Liebman ignition theory to account for the influence of the residence time of the dust particles in the hot furnace on MIT. The theory suggested that it is possible for MIT of mono-sized coal particles of about 50 μm diameter to be minimal even for a limited residence time. The theory was extended to dust clouds with a distribution of particle sizes. It was shown that there is a range of size distributions for which the possibility of ignition is at a maximum. The calculated results were presented as Rosin-Rammler charts indicating size distributions being most sensitive to ignition.

6.6. Possible role of theories/numerical models for predicting real MITs in industry Development of numerical computer models of simulating ignition of dust clouds by hot surfaces is assumed to make further progress. Nevertheless, it seems unlikely that such models will be widely used in the near future for specifying maximum permissible surface temperatures of surfaces in industry. One obstacle to extensive use of simulation models will probably be identification of the credible worst-case ignition scenario that represents a given practical industrial ignition situations (worst credible dust concentration, particle size distribution, cloud turbulence level, systematic cloud movement in relation to hot surface etc.). On the other hand, there may be special cases in industry, where compliance with the 2/3 rule-of-thumb is very expensive. In such cases it may be cost effective to undertake tailored numerical model analyses to see if a reduced safety margin can be justified.

6.4. Tyler (1987) Tyler was concerned with the problem of scaling ignition temperatures of dust clouds from laboratory test apparatus to industrial scale. In particular he focused on the Godbert-Greenwald furnace as the laboratory-scale apparatus. Tyler pointed out that there does not seem to be any single physical/chemical pattern for ignition of a dust cloud. In the case of substances such as sulphur and polyethylene, the minimum ignition temperatures are high enough to allow complete evaporation or pyrolysis to form gaseous fuels. At the other extreme there are metals having minimum ignition temperatures at which the metal will notvaporize fully. In the case of complete evaporation the exothermic oxidation process most probably takes place in the gas phase, whereas with most metals ignition most probably occurs at the surface of, or within, the particle. However, as Tyler pointed out, these differences may not be important in the establishment of the unstable state of ignition which precedes a fully developed flame. Tyler produced a Semenov-type mathematical model of the ignition of a dust cloud in a furnace. This implies that the temperature across the dust cloud in the furnace is homogeneous, with the main temperature drop from the dust cloud to the furnace wall occurring at the boundary between the cloud and the wall (Biot number equal to zero). However, validation of the model was difficult. No reliable activation energies were found in the literature. Tyler pointed out that the activation energy could be quite different from that associated with a fully developed flame. Nevertheless useful parametric studies could be performed. For example, the model predicted a significant decrease of the minimum ignition temperature with increasing furnace diameter. The modern GG furnace has a diameter of 37 mm. For a much larger furnace diameter of 300 mm, and a dust with a G-G furnace MIT of 1000 K the MIT predicted by the model was at least 150 °C lower than the G-G value.

7. Conclusions 1. The need for a laboratory-scale method for experimental assessment of minimum ignition temperatures (MIT) of clouds of combustible dusts in air was realized long ago. The basic principle of measurement of the Godbert-Greenwald (G-G) furnace being currently used as a standard method in many countries, was first introduced by Taffanel and Durr (1911) and Wheeler (1913). However, the purpose of their apparatuses was not to estimate MITs of dust clouds in contact with various hot surfaces in industry. 2. The specific G-G furnace design in current use was first developed by Godbert and Greenwald (1935) as a result of a joint British/American research effort. However, even the original use of that apparatus was another than to estimate MITs of dust clouds in contact with various hot surfaces in industry. 3. The G-G furnace was first adopted for MIT measurements of explosible dust clouds in air at large by England in 1949 and some 74

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff

time before 1960 by USA. 4. Later 3 other apparatuses were developed in Germany and USA for the same purpose. 5. Data from MIT tests of the same dusts in different apparatuses may vary significantly. For the time being, therefore, no single existing test method can be identified as the “correct” one. 6. Experimental and theoretical research has confirmed that MIT in air is not a physical constant of a given combustible dust, but depends both on the size, shape and orientation of the hot surface, as well as on both the dust concentration and the dynamic state of the dust cloud (overall velocity of the dust cloud past the hot surface, and inherent dust cloud turbulence). 7. In the literature the two terms minimum ignition temperature (MIT) and auto-ignition temperature (AIT) are sometimes used interchangeably, most often without explaining possible differences. The present paper discusses a possible main difference. 8. It now seems possible to develop numerical models capable of predicting whether a specific hot surface can ignite a specific explosible dust cloud. However, it may be difficult to define the worstcase scenarios to which such models should be applied in various specific industrial situations (hot-surface geometry, dust cloud concentration, particle size distribution, cloud turbulence and bulk velocity). Also, cost/benefit analyses may rule out mathematical modelling as realistic short-term options in most cases. It seems unlikely therefore, that numerical modelling will be widely used in the near future for specifying maximum permissible surface temperatures in the dust industry. 9. For the foreseeable future it seems sensible, as a main rule, to stick to the traditional requirement that maximum temperatures of hot surfaces in industrial plant that may make contact with explosible dust clouds shall not exceed 2/3 of the MIT in oC, as measured by an accepted standard method.

USA, 0-9728111- 3-3pp. 170–173. Beyersdorfer, P., 1925. Staub-Explosionen, Dresden und Leipzig Verlag von Theodor Steinkopff. Brown, K.C., 1951. A Review of Present Methods of Testing Industrial Dusts for Inflammability, SEMRE Research Report No. 21. Safety in Mines Research Establishment, UK April 1951. Brown, K.C., James, G.J., 1962. Dust Explosions in Factories: a Review of the Literature, SMRE Res. Rep. No. 201. Safety in Mines Research Establishment, Sheffield, UK. Conti, R.S., Cashdollar, K.L., Hertzberg, M., Liebman, I., 1983. Thermal and Electrical Ignitability of Dust Clouds. Rep. Inv., vol. 8798 US Bureau of Mines, US Dept. Interior, Washington DC. Conti, R.S., Hertzberg, M., 1987. Thermal autoignition temperatures from the 1.2 liter furnace and their use in evaluating the explosion potential of dusts. In: Cashdollar, K.L., Hertzberg, M. (Eds.), Industrial Dust Explosions, ASTM STP 958. ASTM, Philadelphia, USA, pp. 45–49 Chapter in. Conti, R.S., Cashdollar, K.L., Hertzberg, M., Liebman, L., 1993. Improved 6.8-liter Furnace for Measuring the Auto-Ignition Temperature of Dust Clouds. Rep. Inv. vol. 9467 US Bureau of Mines, US Dept. of the Interior, Washington DC. Cybulski, W., 1975. Coal dust Explosions and Their Suppression, English Translation of the Original Book in Polish. published for the US Bureau of Mines, USA, by the Foreign Scientific Publications Department of the National Center for Scientific, Technical and Economic Information, USA 1973. Dorsett, H.G., Jacobson, M., Nagy, J., Williams, R.P., 1960. Laboratory Equipment and Test Procedures for Evaluating Explosibility of Dusts. Rep. Inv. 5424. US Bureau of Mines. Eckhoff, R.K., 2003. Dust Explosions in the Process Industries, third ed. Gulf Professional Publishing (an imprint of Elsevier), Boston, USA, 0-7506-7602-7pp. 720. Eckhoff, R.K., 2016. Explosion Hazards in the Process Industries, second ed. Gulf Publishing Company (an imprint of Elsevier978-0-12-803273-2 (Chapter 2). EN 1127-1, 2007. European Standard EN 1127-1: Explosive Atmospheres – Explosion Prevention and Protection. Part 1: Basic Concepts and Methodology. vol. 36. European Committee for Standardization, Management Centre: Rue de Stassart, Brussels, pp. B-1050. Frazer, J.C.W., Hoffman, E.J., Scholl, L.A., 1913. Laboratory Study of the Inflammability of Coal Dust, Bulletin, vol. 50 Dept. of the Interior, Bureau of Mines, Governmental Printing Office, Washington DC. Godbert, A.L., Wheeler, R.V., 1929. The Relative Inflammability of Coal Dusts. A Laboratory Study, Safety in Mines Research Board, Paper No. 26. Godbert, A.L., Greenwald, H.P., 1935. Laboratory Studies of the Inflammability of Coal Dusts. Effect of Fineness of Coal and Inert Dust on the Inflammability of Coal Dusts. Bulletin, vol. 389 United States Department of the Interior, Bureau of Mines. Published by US Government Printing Office, Washington DC. Griesche, G., Brandt, D., 1976. Einflussfaktoren auf die Zündtemperatur von Staub-LuftGemischen. Die Technik 31, 504–507. Harrison, A.J., Cairnie, L.R., 1988. The development and experimental validation of a mathematical model for predicting hot-surface autoignition hazards using complex chemistry. Combust. Flame 71 (1), 1–21. Harrison, A.J., Furzeland, R.M., Summers, R., Cairnie, L.R., 1988. An experimental and theoretical study of auto-ignition on a horizontal hot pipe. Combust. Flame 72 (2), 119–129. Hartmann, I., Nagy, J., Brown, H.R., 1943. Inflammability and Explosibility of Metal Powders, US Bureau of Mines Report of Investigations No. 3722. US Department of the Interior. Hensel, W., 1984. Methoden zur Bestimmung der Zündtemperatur von Staub/LuftGemischen an heissen Oberflächen. In: Eine Vergleichende Untersuchung. BAM Jahresbericht Pp. 86–88. Bundesanstalt für Materialprüfung, Berlin, Germany. Higuera, F.J., Linan, A., Trevino, C., 1989. Heterogeneous ignition of coal dust clouds. Combust. Flame 75 (3–4), 325–342. IEC, 1975. Electrical Apparatus for Explosive Gas Atmospheres. Part 4: Method of Test for Ignition Temperature, second ed. Central Office of the International Electrotechnical Commission, Geneva, Switzerland IEC publication 74-4. ISO/IEC, 2016. International Standard ISO/IEC 80079-20-2/Ed1, Explosive Atmospheres, Part 20-2: Material Characteristics, Combustible dusts Test Methods, Pt. 8.1. International Electrotechnical Commission, Geneva, Switzerland. Kong, D., Eckhoff, R.K., Alfert, F., 1995. Auto-ignition of CH4air, C3H8air, CH4/C3H8/air and CH4/CO2/air using a 11 ignition bomb. J. Hazardous Materials 40 (1), 69–84. Leuschke, G., 1966a. Über die Untersuchung brennbarer Stäube auf Brand- und Explosionsgefahren. Staub Reinhalt. Luft 26, 49–57. Leuschke, G., 1966b. Verfahren zur Ermittlung der Zündtemperaturen aufgwirbelter brennbarer Stäube. BAM Mitteilung, Bundesanstalt für Materialprüfung, Berlin, Germany. Mitsui, R., Tanaka, T., 1973. Simple models of dust explosion. Predicting ignition temperature and minimum explosive limit in terms of particle size. Ind. Eng. Chem. Process Des. Dev. 12 (3), 384–389. Nomura, S., Callcott, T.G., 1986. Calculation of the ignition sensitivity of dust clouds of varying size distributions. Powder Technol. 45 (2), 145–154. Palmer, K.N., 1973. Dust Explosions And Fires, Book in Powder Technology Series. published by Chapman & Hall Ltd., London, UK SBN 412 09430-4. Pinkwasser, Th, 1989. Influence of Size of Hot Surface on the Minimum Ignition Temperature of Dust Clouds. Private Communication to R. K. Eckhoff from. Th. Pinkwasser, Bühler, Switzerland. Price, D.J., Brown, H.H., 1922. Dust Explosions: Theory, Nature, Phenomena, Causes, Methods of Prevention. National Fire Protection Association, Boston, USA. Raftery, M.M., 1968. Explosibility Tests for Industrial Dusts. Fire Research Technical Paper No. 21. Her Majesty's Stationery Office, London. Rice, G.S., 1925. Review of Coal-Dust Investigations, Trans. American Institute of Mining

8. Epilogue It seems likely that the development histories of even other methods for measuring ignitability and explosibility parameters of dust clouds than MIT are unknown to many scientists and engineers working in the dust explosion field. Commonly measured parameters include minimum electric spark energy for ignition (MIE), minimum explosible dust concentration (MEC), maximum oxygen concentration for inerting (LOC), and maximum constant volume explosion pressure and rate of explosion pressure rise (Pmax and (dP/dt)max). Therefore, any attempt at producing historical reviews of the development of even other test methods than the one discussed in the present paper, should be welcomed. There is much to be learnt from all the good work done through many years by the pioneers in the field. Acknowledgements As a professor emeritus at the Department of physics and technology at the University of Bergen, I am indebted to the Department for so generously allowing me to keep an office there and for providing me with the necessary infrastructure for carrying on with my writing. References Alfert, F., Fuhre, K., 1988. Ignition of Gas Clouds by Hot surfaces. Which maximum Temperature Is Relevantand Acceptable for Equipment in Classified Areas? CMI Report No. 873301-1. Christian Michelsen Research. (Since 2001: GexCon AS), Bergen, Norway. ASTM, 2015a. American society for testing and materials, standard E 1491 for testing of minimum auto-ignition temperature of dust clouds. In: Annual Book of ASTM Standards 14.02 ASTM, Philadelphia, USA. ASTM, 2015b. American society for testing and materials, standard E659. Method for testing of autoignition temperatures of liquid chemicals. In: Annual Book of ASTM Standards 14.02 ASTM, Philadelphia, USA. Babrauskas, V., 2003. Ignition Handbook. Fire Science Publishers, Issaquah, WA 98075,

75

Journal of Loss Prevention in the Process Industries 59 (2019) 63–76

R.K. Eckhoff and Metallurgical Engineers, No. 1435-F. Issued with Mining and Metallurgy. March 1925. Scheid, M., Sand, I.Ø., 2000. Influence of turbulence and flow-rate on ignition properties of dust clouds. In: Report CMR-00-F30075 from Chr. Michelsen Research (Presently GexCon), Bergen, Norway. Taffanel, J., Durr, A., 1911. Fifth Series of Experiments on the Inflammation of Dusts: Tests of Inflammability. Comité Central des Houilleres de France, Station d'Essais de Lievin, Paris August 1911. Takigawa, T., Yoshizaki, S., 1982. Research of several problems related to the minimum ignition temperature of agricultural dusts. J. Soc. Powder Technol. Japan 19 (10),

582–591. Tonkin, P.S., Raftery, M.M., 1961. Testing of Industrial Dusts for Explosibility, Fire Research Note No. 464, Fire Research Station, Boreham Wood, Herts., UK. (unpublished draft report dated June 1961). Tyler, B.J., 1987. Scaling the Ignition Temperatures of Dust Clouds. Report No. F3/2/ 347, (June) Fire Research Station, UK. Wheeler, R.V., 1913. Report on the Inflammability and Capacity for Transmitting Explosions of Carbonaceous Dusts Liable to Be Generated on the Premises under the Factory and Workshops Act. Factory Department, UK Command Paper 6662.

76