Abstract of Special Research Report (RR-24)

National Institute of Occupational Safety and Health, Japan

Study on Maximum Safe Gap with Propane/Hydrogen with Air Mixtures

RR-24-1

Heizaburo TSURUMI

: According to the study on the measurement of experimental safe gap with Propane/Hydrogen with air, the behavior of safe gap of this mixtures that is very few reports which give scientific support to at present is concluded clearly as to be discussed in this paper together with each chapter of this report.
    The outline of this report will be described below.
  1 Experimental safe gaps with 25 mm flanges were determined in a standard spherical vessel and the parameters of the experiment were the concentration of the mixtures and the gap widths.
    The vessel was made of stainless steel and the internal volume of vessel was 20 cc. As ignition source, electric sparks were discharged between the electrodes of an ignition plug- in the center of the vessel. Using the ion probe, it was determined whether gas ignition occurred outside the chamber.
  2 The experimental safe gaps largely depend on the concentration of Propane in the mixtures.
    It is said that the effect of a small quantity concentration of Propane in the mixture appeared sensitively to the increasing of experimental safe gap width and from this results of experiments it should be said that the explosion grade of the mixtures depended sensitively on the concentration of Propane in the mixtures.
  3 According to my observation of this study the curve of the experimental safe gap with this two components mixtures are not linear line connected with safe gap width of Propane and the safe gap width of Hydrogen. But the clarification of this growth mechanism is not clear yet.
  4 Fundamentally, in the behavior of the safe gap, there are many factors by which the probability of ignitions are influenced severely, therefore the percentage external ignitions curve on various concentrations mixtures was not the same figures in the range which became the purpose of this study. By this reason, it is impossible that the statistical maximum safe gap can be obtained theoretically by the same formula without experiments.
  5 The relation between the safe gaps which measured on this study and the stoichiometric concentration of the fuels are closely related under the range of the fuels belonged to the first grade of explosion. And the relation among those quantities is below.
      Safe gap ∝ C st^-1
    There are no doubt many other problems to be considered under this head which I can not take up within the limited scope of this study.

Some Studies on Combustion of Irradiated Polyethylene

RR-24-2

S. MORISAKI

: Oxidative thermal degradation products of polyethylenes at various temperatures crosslinked with electron beams have been analyzed with gas chromatography and mass spectrometry techniques. Carbon monoxide and carbon dioxide are determined at a temperature range of 200-340 °C, and the activation energies of the unirradiated and the irradiated polyethylene (at 100 Mrad) are 13.5 and 11.4 Kcal/mole, respectively. C₁ to C₈ hydrocarbons produced in air and in nitrogen are determined at temperatures from 400 to 540 °C for the polyethylenes. The irradiated polyethylene produces less hydrocarbons in air than the unirradiated polyethylene, contrary to the fact that the crosslinked polymer evolves more hydrocarbons than the unirradiated polymer in a nitrogen atmosphere. Aldehydes and ketones are observed in the volatile oxidative degradation products, and these carbonyl compounds increase quantitatively with increase of temperature up to about 460 °C. It is concluded that irradiated polyethylene is thermally more unstable in the absence of oxygen and more easily oxidable at low degradation temperatures in air than unirradiated polyethylene. Irradiated polyethylene, however, is more heat-stable than unirradiated polyethylene from the standpoint of the ignition process.

A Quantitative Model of Industrial Accident --An Assessment of Industrial Risk Based on a Method of Nominal Multivariate Analysis using Conception of Entropy and it's Actual Example of Analysis--

RR-24-3

Yoshinobu SATO, Taiji KONDO, Soichi KUMEKAWA, Noboru SUGIMOTO, Ikuo Mae and Eiki YAMANO

: In this paper we deal with a method relating to quantitative assessment of a certain risk and show an example making use of it.
    When we try to assess the risk of system by data based on any statistics given by investigation of accidents the data which are given us as explanatory variables cover various kinds and are often given in the form of discrete variates, especially nominal scales.
    There are some methods of what are called multivariate analysis which treat these nominal scales yet apart from them we divesed a method of multivariate analysis adopting the conception of amount of information and studied actual accidents which came about by destructions of grinding wheels.
    The fundamental conception of the analysis is as follows. We express the situation of the system when the accident occurred in a certain explanatory vector e i, and transform it to vector S _i which has R components. As each data has been observed in one of R events which indicate the degrees of damage, we can assess the rate of risk by knowing the density of each data in S _i space.
    Now one of the simplest transformation is a linear transformation, and it is expressed as S _i = [ M ]e _i. We apply a kind of entropy to the definition of the matrix [ M ].
    The matrix [ M ] is expressed as follows.

       J is the number of the factors.
       l _j is the number of the categories in the j -th factor.
       φ _rjk is defined as follows.

    n _rjk is the number of the data which replied to the k -th category in the j -th factor and to the r -th event.
    H _j(y/x) is the conditional entropy of the events (which indicate the degrees of damage) under the condition that we are informed of a category in the j -th factor.
    H _j(y/x)_M is the maximum value which H _j(y/x) can takes when the distribution of the data in each category are changed. The vector e _i is expressed as follows.

e _ij is defined as follows. If i -th datum replied to the k -th category in the j -th factor, the i -th component of e _ij is equal to 1 and the other components are O.
As a result of study in the analysis of actual destructions of grinding wheels, it became possible to evaluate the risk of grinding operation system in some degree.

Smouldering Temperature of Combustible Dust Layer

RR-24-4

T. MATSUDA and M. NAITO

: A layer of combustible dust on a hot surface may ignite if the temperature of the surface is sufficiently high. Although a considerable amount of experimental investigation by many workers have been carried out, the method has not been defined of measuring the minimum temperature, at which smouldering combustion, rather than ignition itself, occurs. The temperature leading to smouldering were taken to be the smouldering temperatures of the dust layers.
Temperature-time records, which gave arrival time to the smouldering were obtained on the centre of the upper surface of dust layers for wood, corn-fibre, soybean protein, cocoa-husk and other powders. The dust layer was either put on the pre-heated plate of constant temperature, or formed with cylindrical metal sleeve of various diameters and depths at room temperature and then heated up to a pre-determined smouldering temperature. Plots of arrival times versus reciprocal of the smouldering temperatures on semilogarithmic co-ordinates were made to compare the smouldering characteristics with kinds of dusts and heating methods. These representations gave straight lines with slopes, which also depended on the size of dust layers in most cases. For relatively small sizes of dust layers, the frequent occurrence was observed of smouldering from the peripheral edge of the layers with free convection of air, corresponding to the measurement of temperature distribution in the interior of the layer. This makes it difficult to correlate to the other factors.
The other results are as follows. According to the infrared spectrum of heated pulp powders the absorption of carbonyl vibration bands was increased only just prior to the pyrolysis of the layer. The application of thermal theory to the results of wood dust layers suggests the reaction with apparent activation energy 30 kcal/mol, which is compared to that of pyrolysis of wood of large amounts. The charred residues of some kinds of dusts were not brought to surface combustion at their smouldering temperature. It was assumed that the combustion reaction of the gaseous pyrolysis products did not occur in the dust layer, containing air at first.

Interruption of Explosions by Flame Arresters (4th Report) -Factors Affecting on Flame Quenching by Wire Gauzes (1)

RR-24-5

Toshihiro HAYASHI

: Wire gauges have been considered as one of the useful quenching elements for flame arresters because of their availability, economical cost, low flow resistance and ease of processing, but the experimental evidences are not yet satisfactory.
    This report describes about several factors affecting on quenching abilities of wire gauges against flames propagating in stationary hydrogen-air mixtures whose initial pressures are up to 2 kg/cm² (G) and hydrogen content ranges between 10 and 60 vol.%. Experiments are carried out in a horizontal 1-inch short enclosed tube (Fig.1, l1 and l2 being about 12 cm). Four kinds of stainless wire gauze are used as quenching elements (Table 1), and maximum number of gauges in pack is twenty.
    In each explosion test, the whole assembly is evacuated before the premixed mixture is fed in up to a desired initial pressure, then the mixture is ignited. An observation is made by the help of a thermo-couple whether an explosion occurred in the protected chamber.
    As it has been known that the higher the initial pressure is, the more difficult the quenching is, the maximum initial pressures, at which no flame transmission occurs, are determined under various conditions. These initial pressures (denoted as Limiting Safe Pressure or L.S.P.) are defined arbitrarily as pressures which indicate flame quenching abilities of gauges tested. Pressure drops (mmAq) across gauges in pack are also measured for air flow rate of 100 l/min.
    Main results as follows ;
  1) It is found most difficult to quench flames of the mixtures containing a little higher than stoichiometric hydrogen (Fig.2). Such mixtures may be used for quenching tests of flame arresters.
  2) L.S.P. increase with number of gauges ( N ), but in some cases there exists certain limit for N beyond which an increase of N has no effect on L.S.P. (Fig.5). For a substantial range of N, L.S.P. can be related to N by the equation :
      L.S.P. = aN - b                    (1)
where a and b are constants peculiar to mesh number of gauze and hydrogen content of gas mixture.
3) The relation between pressure drop (P.D.) and N is approximately shown by the equation :
      P.V. = kN                      (2)
    The constant k is calculated for each kind of gauze from mesh width and thickness of the gauze.
    The fact that k is found to be proportional to a / b in equation (1) shows a direct relation between flame quenching ability and pressure drop, independent of mesh number, of wire gauges in pack.

Interruption of Explosions by Flame Arresters (5th Report) -Factors Affecting on Flame Quenching by Wire Gauzes (2)

RR-24-6

Toshihiro HAYASHI

: As flame arresters are safety barriers to minimize the development of flames after explosions started in closed systems, their failures are deadly, and their performances should have been verified through field tests, if possible, or at least equi-scale explosion tests.
    This report describes about several factors affecting on quenchings of acetylene-air flames by wire gauges. Those results obtained may be of use for the development of methods of design and test of flame arresters, including those equipped in systems unsuitable for on-site tests.
    Experiments are carried out in horizontal 1-inch enclosed tube of various length. Each explosion tube is composed of an ignition chamber, in which the premixed mixture is ignited through an electric discharge, and a protected chamber in which an explosion occurs when gauges under tests fail to quench flames (Fig.1). Nominal mesh numbers of gauges range between 60 and 120. Two kinds of initial gas pressure are used as quenching ability indices of gauges. One is Limiting Safe Pressure (L.S.P.), at pressures lower than which no flame transmit through gauges. The other is Minimum Pressure of Flame Transmission (M.P.F.T.), the lowest pressure at which a flame begins to transmit when initial pressure is gradually elevated. Thus, the higher those pressures are, the more easily flames are quenched. Main results are as follows ;
  1) It is found most difficult to quench flames of the mixtures containing acetylene of about 1.1 times of the stoichiometric concentration.
  2) L.S.P. increase with decreasing a spacing distance between neighbouring gauges in pack, but beyond a certain spacing distance, it has no effect on flame quenching (Figs.3, 4 and 5). The result shows that one can obtain some safety margin by decreasing the spacing distance.
  3) As the length of the protected chamber, which is assumed as the space to be protected from explosion flames in actual equipment, increases, flames become more easily quenched (Fig.6). Then the result of tests with a shorter protected chamber could be applied, on safe side, to the same flame arrester with a longer protected chamber.
  4) Flames transmit more easily through gauges with the increase of either the relative distance between an ignition source, and gauges, or the length of ignition chamber in case of the far end ignition (Figs.7, 8 and 9). The result shows that, first of all, flame arresters should be equipped as nearer as possible to potential sources of ignition.

Safety for Electrical Equipment under Artificial Environments(2nd Report) Ignition of Flammable Solid Materials by Low Voltage Inductive Sparks

RR-24-7

Ryuji TANAKA and Kenji ICHIKAWA

: In a hazardous area where an explosive gas be or may be present, the technique of "intrinsic safety" has been applied all over the world. In this paper, the authors aim at applying the technique to designing of safe low voltage circuits which are incapable of igniting a flammable solid in "Oxygen-hazardous Areas".
    Flammable solid specimens are ignited by low voltage inductive sparks in pure oxygen, oxygen-nitrogen or oxygen-helium atmospheres, at various absolute pressures of up to 1.3 MPa and at various oxygen concentrations, and minimum ignition currents are measured.
    The specimen is attached on the fixed electrode of the 'Break-flash No.3 Apparatus' and ignition of the specimen is observed visually through a glass window of pressure vessel. The circuit voltage is 50 volts d.c. and circuit inductance ranges from 3 to 1,000 mH.
    Of the specimens tested, "Japanese paper" showed the lowest ignition current.
    The minimum ignition currents for the Japanese paper decrease with the oxygen concentration and the total pressure, and the minimum ignition current corresponding to pure oxygen at an absolute pressure of 0.1 MPa and to circuit inductance of 1,000 mH are 0.19 amperes. Various combinations of oxygen concentration and total pressure, that give the minimum ignition current of 0.19 amperes are obtained with 1,000 mH inductance.
    The relations between the minimum ignition current and circuit inductance are also given for pure oxygen at 0.1 MPa and 0.7 MPa, for 60% O₂ - 40% N₂ and 60 % O₂ - 40 % He at 0.7 MPa. These curves show similarity in shape and slope, to those known of ignition of explosive gases in inductive circuits.
    It was concluded that an intrinsically safe inductive circuit suitable for use in "oxygen-enriched atmospheres" could be designed within a certain limit of oxygen concentration, atmospheric pressure, circuit current and inductance.

Test-Manufucture of Artificial Fingers (2nd Report) --Winslow's Effect in Ionic Exchange Resin Disperse System--

RR-24-8

Noboru SUGIMOTO and Taiji KONDO

: Winslow's effect is defined as an essentially instantaneous reversible change in apparent viscosity when a fluid is subjected to an externally applied electric field.
    In our previous papers, we reported that the mechanism of Winslow's effect was able to be explained fully by electro-double-layer hypothesis.
    Namely, the particles (disperse phase) in an electroviscous fluid should have many ionic dissociable groups. And after absorbing moisture, then discoiating ionic groups, electro-double-layer is constructed at the surface of each particle. When voltage is externally applied in this condition, the dissociated ions of electro-double-layer move to the direction (or the opposite) of the electric field along the layer boundary, and this becomes that electric polarization is induced. This polarization gives rise to static-electric-attraction among particles, thus, as a result of inducing resistance to flowing, the apparent viscosity increases.
    Under this hypothesis, the disperse phase in electroviscous fluid essentially should have many ionic dissociable groups inside. And ionic exchange resins are able to satisfy this indispensable condition.
    So, as was expected, we could recognize the greatly large increase of apparent viscosity in ionic exchange resin disperse system.
    Giving an instance, in strongly acidic cation exchange resin the following reaction exists.
      R·SO₃H ←→ R·SO₃^- + H⁺
where, R is polystyrene basic, SO₃H is ionic functional group, ·SO₃^- is fixed ion and H⁺ is counter ion.
    The more moisture becomes involved, the more this reaction moves right ward. So, corresponding ionic exchange resin disperse system to the hypothesis mentioned above, electro-double-layer is of fixed ion and counter ion.
    In ionic exchange resin, counter ion which is one parameter of electro-double-layer can be selected relatively easily on behalf of its essential usage. So, in the result of measurements about many kinds of electro-double-layer by changing counter ion, we could recognize that Winslow's effect was relatively clearly connected with the electro-double-layer. And when the counter ion was Cl^-, SCN^-, SO₄^-2 , the large induced shear stress was measured.
    As of the effects of disperse medium, though we could not concretely find which property of disperse medium have an effect on Winslow's effect, we recognized that tri-2-ethylhexyltrimellitate was superior to the other materials tried here.
    Like this, using ionic exchange resin as it's disperse phase makes consideration about not only viscosity increment itself but also stability and response of electroviscous fluids possible.
    As a result of this study, strongly basic anion exchange resin and tri-2-ethylhexyltrimellitate dispersion we were developed recorded the most typical shear stress induced, moreover had the most enough stability in ever developed.

The Critical Energy for Direct Initiation of Spherical Gaseous Detonations

RR-24-9

Hidenori MATSUI

: The paper reports experimental results on the critical energies for direct initiation of spherical detonations using electrical sparks under various electrode geometries and spacings. The results indicate that for large spacings the detailed electrode configurations have no influence on initiation energy. The critical energy per unit length reaches a minimum asymptotic value corresponding to the value found previously for cylindrical detonations. For electrode spacing less than the characteristic explosion length, the electrode geometry has an effect on the initiation energy. The flanged and the pointed needle electrodes from the lower and upper bounds respectively for the critical energy for various electrode configrations. The results indicate that the effects of the electrode geometry are essentially those corresponding to the severity of gasdynamic expansion generated. The gasdynamic effects fall between the cylindrical and spherical symmetries. This paper also describes some new results for the critical energy for direct initiation of detonations in acetylene-oxygen mixtures using flanged electrodes. The present spark energies represent the true effective initiation energies, independent on discharge characteristics and electrode configurations and gap spacings.
Comparisons with existing data from exploding wire, laser and electrical sparks, as well as initiation by firing a planar detonation in a larger volume, indicate no qualitative differences in the dependence of the critical energy on mixture composition. As far as pressure dependence is concerned, the present results agree with the previous data on laser sparks and electrical sparks. However, for exploding wire and initiation by a planar detonation, much stronger pressure dependence is obtained. A simple model is proposed for direct initiation of spherical detonation by firing a planar detonation into a larger volume. The model enables critical tube diameter data to be reduce to equivalent energies, and good agreement between the critical diameter data of Matsui and the present critical spark energies is obtained on the basis of this model.

JNIOSH