Abstract of Special Research Report (RR-90)

National Institute of Occupational Safety and Health, Japan

The Influence of Hydrogen Gas Atmosphere on Fatigue Crack Growth Behaviours in Structural Steel SB42 --In the Case of 1 atm Hydrogen--

RR-90-1

Masazumi TANAKA

: In the previous study, the environmental effects of pure and salt water on fatigue crack growth (FCG) behaviours were investigated under low cycle loading condition for low carbon structural steels. The result showed that such aqueous environments remarkably accelerated their FCG rates in high ΔK range, which was supposed to be caused by cyclically induced hydrogen embrittlement.
    In this study, therefore, to certify such supposition, FCG experiments were carried out on one of the same steels SB42 in 1 atm hydrogen gas atmosphere under similar low cycle and high cycle loading conditions, and the effects of both aqueous and hydrogen environments were comparatively investigated on the FCG rates and the feature of fracture surfaces.
  (1) Under the low cycle condition, 1 atm hydrogen atmosphere remarkably accelerated fatigue crack growth rate dl/dN of mild steel SB42 by a factor of about 25 in high ΔK range (30-80 MPa·m^1/2), compared to that in air. This acceleration effect was fairly larger than that in the aqueous environments (a factor of about 5). The similar brittle features of fractured surfaces in hydrogen and aqueous environments supports the aforementioned supposition on the mechanism of FCG acceleration in aqueous environments. The larger FCG acceleration effect in hydrogen than in water environments corresponds to the higher areal percentage of brittle fractures in hydrogen than in the latter.
  (2) On the other hand, high cycle FCG rate dl/dN in the lower ΔK range (10-20 MPa·m^1/2) and the ΔK_th value in hydrogen coincided with those in dry air, that is, hydrogen atmosphere does not affect the very low FCG rate of this steel. Although not decisive, this result also corresponded to little differences in fracture morphology in dry air and hydrogen.
  (3) With regards to the above results, it seems that the effect of hydrogen gas on dl/dN appears at rather low ΔK region and gradually increases with increasing ΔK value, showing a different type from any of three types of corrosion fatigue crack growth behaviours proposed by McEvily et al.

On the Structuring of an Inherently Fail-Safe System and Multiphasic Safety Design

RR-90-2

Yoshinobu SATO

: System safety engineers have developed and utilized the methods for identification, analysis and assessment of hazards and risks produced in a designed and assumed system, such as hazard-operability studies and fault-tree analysis. While, we have considered that the designing of systems involving hazard-control systems is a job as an expert in each technological field. In general, the systematization of designing is difficult because we can not express mathematically the knowledge of systems synthesis.
    Recently since systems are getting more complicated and flexible, the cause of large percentage of major hardware-failures that led to accidents has been hardly established. If the cause of accidents is unpredictable, this suggests that the most productive way of increasing safety and ameliorating risk is to mitigate the effect of an accident, i.e., to reduce its consequences. An automobile seat-belt, for example, does not prevent accidents, it, however, reduces risk of injury from an accident.
    A hazard-control system (:the system which prevents damage, given that failures have occurred) materializes the seat-belt approach. The author has made efforts toward categorizing and systematizing the design of hazard-control systems in terms of an action-change and action-chain (A-C) model for hazard-control.
    This paper systematizes the structuring of an inherently fail-safe system and describes the systematic procedure for multiphasic safety design based on the A-C model. The paper obtains the following results:
  1. An entropy model for system state division is proposed. In the entropy model, the entropy of system elements with regard to energy, information, disorganization of shape, etc. divides system state into two states, ordered and disordered state. The ordered state evokes ordered-state actions between system elements. Disordered state produces a function-failure action. The entropy model clarifies the physical background of the A-C model. Next, the mode of systemstate transition is divided into two modes, ordered and disordered transition. Only the ordered state system-elements materialize the ordered transition. Then, an inherently fail-safe system is established by the following definition; if a failure in a system brings about a disordered system transition from ordered state and if this transition prevents a damage which could be caused by a hazard, then the system is an inherently fail-safe system with regard to the failure and the hazard.
  2. The paper generalizes hazard-restraint strategy as the following four principles; 1) exclusion or elimination of undesirable action source, 2) prevention of undesirable changes, 3) controls which prevent a system from transferring to a damage-producing systems phase, and 4) controls which transfer a system from a damage-producing system phase to a damage-avoidable one. Principles 3) and 4) are materialized by hazard-control systems. Designing of fail-safe and fault-tolerant structures into hazard-control systems is hazard-restraint tactics. Multiphasic safety design is the implementation of the hazard-restraint strategy and tactics.
  3. The paper defines the rules for structuring a fail-safe, fault-tolerant, and inherently fail-safe systems, and establishes the method of elaborate and systematic multiphasic safety design.
  4. Examples involving a chemical batch-processing plant demonstrates the new technology and confirms its effectiveness.

Analysis of Labour Accidents in Tunnel Construction Work --According to its Work-Stages--

RR-90-3

Yoshimi SUZUKI and Shigeo HANAYASU

: In order to establish the effective safety countermeasures in the construction sites, it becomes necessary to understand actual situation of labour accidents thoroughly in accordance with its work-stage.
    This paper deals with the accident investigation for the purpose of providing a better understanding of the basic nature of recent labour accidents in tunnel construction sites. The accident data were made available from the Industrial accident reports submitted in 1987 to the prefectural labour standard offices or the Labour Standards Inspection Offices involved, which totaled 161 injuries in number including 9 fatalities.
    In this study, actual recent situation of labour accidents in accordance with the difference of the work-stage (i.e.; excavation work-stage, shaft work-stage, lining work-stage and so on) was firstly investigated. The tendency of the lost workdays due to accidents in accordance with difference of work-stage was also investigated, and expected lost workdays of each work-stage were obtained. According to the results of statistical analysis, the accident-type was one of main factors which strongly influenced on the difference in expected lost workdays. The results obtained in this study are described as follows;
  (1) The actual situation of labour accidents in accordance with the work-stage was obtained (Fig.2).
  (2) The relationship between lost workdays due to accidents and their cumulative frequency can be expressed as a log-log liner straight line (Fig.3).
  (3) If the slope of a line mentioned above is exceed -1, the expectation of the lost workdays including fatal accidents can be obtained within a limited value. The expectation of workdays lost with fatalities for accidents investigated in this study was about 90 days.
  (4) About expected lost workdays, some distinguish difference in expected lost workdays was found between each tunnelling method (Fig.4.)
  (5) Similar tendency was also observed between each work-stage (Fig.5).
  (6) Difference in expected lost workdays according to age of workers was also distinguished, and that of middle aged group was larger than that of the others (Table 3, Fig.6).
  (7) According to the result of statistical analysis (Table 5-9, Fig.8), one of the main factors which strongly influenced on the difference of expected lost workdays was attributable to the difference of accident-type.

Study on Slip-Resistance Measurements for Footwear (2nd report) --Development of Measuring System for Slip-Resistance--

RR-90-4

Hisao NAGATA

: The aim of this report is to introduce a new measuring system for the slip resistance of footwear which was developed by the Research Institute of Industrial Safety (RIIS). In our country, no appropriate slip resistance meter for footwear had existed until recently. RIIS has developed two such meters one a prototype and one its improved version. In a pilot test conducted using the prototype on a lubricated floor, it was confirmed that the coefficient of dynamic friction and the slipperiness felt by test participants who walked on the floor were convincingly correlated. Based on the test results of the prototype, I developed a second one with much improvements made on the mechanism, including the measuring system and an artificial foot. The device introduced here is considerably different from the type of slip meter which is often used by architects when they choose floor materials. In the case of the meter, the maximum static friction plays an important role as explained in my previous report (Nagata, 1989) This primary purpose is to find out the most desirable degree of slipperiness between the floor surface and the shoe sole. The RIIS meters were meant to be used for measuring at a place where actual slipping occurs. With an understanding that slower slipping is less risky, I tried to evaluate with our new measuring instruments less dangerous slips of footwear as its user walks on a slippery surface. A lot more attention had to be paid to dynamic friction than to static friction, as many researchers tried to evaluate the frictional properties of footwear.
A pilot study of slip-resistance using a mechanical model to simulate the human leg with its knee bending and the lower section fixed with a slender rod was carried out. The artificial foot with footwear was placed on the sliding floor on which a kistler force plate was installed. The prototype slip meter was improved to achieve further accuracy with a wider floor area and a servomechanical operation. The artificial foot can be placed in various directions. Further explanations are given on: principle of measurements, artificial foot, measuring system and controlling system of servomechanism.

Explosion Characteristics of Flammable Gas-Dust Mixtures --Explosion of Hydrogen-Cornstarch-Air Mixture--

RR-90-5

Toshihiro HAYASHI, Toei MATSUDA and Hidenori MATSUI

: There have been many reports on explosion characteristics of either flammable gas, vapour or dust. However, explosion hazard of flammable gas mixed with flammable dust, or hybrid mixture, has not been well understood yet. With the advancement of techniques in pulverizing solid materials and with the increase of powder handling processes, the hazard assessment and the establishment of preventive method for explosions of these hybrid mixture will be more important from the viewpoint of industrial loss prevention.
This report describes, as the first step to the study of hybrid mixture explosion, the effect of cornstarch powder content on such explosion characteristics of hydrogen-air mixture as maximum attainable pressure and pressure rise rate. Explosion tests were carried out in a 23-liter cylindrical vessel, in which a required mixture of hydrogen-air was filled; cornstarch was blown up from the bottom of the vessel so as to produce a dust cloud, by means of a pressurized gaseous mixture with the same composition as in the vessel. Cornstarch-air and hydrogen-air mixtures were also ignited at the center of the vessel by a red hot wire, and the results were compared to those of hybrid mixtures.
The maximum explosion pressure of cornstarch-air mixture was found comparable to that of stoichiometric hydrogen-air mixture. Cornstarch dust dispersed into rich hydrogen-air mixture gave decreased explosion pressures as the increase of dispersed mass of dust. For the stoichiometric hydrogen-air mixture, dispersion of dust gave slightly higher explosion pressures probably because of a partial contribution of thermal decomposition of dust. On the other hand, explosion pressure always increased when any amount of dust was dispersed into lean hydrogen-air mixture. Effect of dust dispersion on the maximum rate of pressure rise showed a similar trend to that on explosion pressure. In this case, however, no hydrogen-cornstarch-air mixture gave a higher pressure rise rate than that of stoichometric hydrogen-air mixture. From a viewpoint of explosion safety in industries, one of the most hazardous situation was found in the mixture of smaller amount of hydrogen and cornstarch, which gave relatively higher increasing both in explosion pressure and pressure rise rate. Calculation trial for explosion pressure of hybrid mixture showed the importance of reasonable model of combustion and efficiency of dust combustion.

Measurement of Lower Explosibility Limits of Dusts Dispersed in Air in a Large Scale Vertical Tube

RR-90-6

Toei MATSUDA and Toshihiro HAYASHI

: A better knowledge of the flammability of combustible dusts is essential for prevention of the fire and explosion in their use. Many published recommendations or guidelines on explosion protection methods carry the lean flammability data of dusts for the sake of industry. However, the data are often quite different from one apparatus to another, and seem not to be reproducible even in a certain international round-robin tests with the accepted testing method in ISO or ASTM. Thus, accurate lean limit data are important for industrial safety considerations, as well as for intrinsic scientific interest.
    In this report, lower flammability limits were obtained with dusts dispersed in air in a vertical tube of 225 mm in diameter and 189cm in overall length. The top of the tube was closed and the bottom was open for observation of upward flame propagation. The method of forming dust clouds in the tube was followed with Lee et al. (1987). who used a circular perforated tube and a rod to suppress turbulence level of the mixtures. The dust divided in two small chambers was dispersed through two perforated tubes with linear array of holes by high pressure air. The directed momentum of the dispersion jets was broken with a circular rod to provide a more uniform distribution of dust inside the explosion tube. The igniting source was the chemical powder of aluminium, barium peroxide and barium nitrate. The uniformity of the dust-air mixtures was examined with an optical monitor. The optical transmission data showed for the mixtures to keep even dust distribution after about 0.5s elapsed from the dispersion pulse. Ignition was then activated with the time delay.
    Upward flame propagation was detected with phototransistors on the tube wall, and the radiation signals from dust flames indicated low-frequency vibrations. These observations were found to be similar to the results for corn-starch dust flames of Veyssiere et al. (1988). Then, the top flange of the tube was replaced by a 50 mm orifice to damp the vibration.
    23 kinds of antioxidant or light-absorbing powders additive to plastics and lycopodium powder were used as organic fuels. The lower limits for these dusts were between 70 and 200 g/m³, and the lycopodium and the other two dusts showed the lowest value of 70 g/m³ . Any relationship between average particle sizes and the lower limits was not suggested, but it seemed the dusts having good dispersability may show the value as high as 70 g/m³ for the "true" value of the lower explosibility limit for the dusts used in these experiments. Data from a Hartmann apparatus and a standard 20 L spherical bomb vary from 26 to 45 g/m³ for the lean limit of lycopodium dust. The validity of industrial application of these data is discussed within the framework of safety consideration.

Prediction of Thermal Hazard Data of Benzene Derivatives by Multivariate Statistical Analysis on Their Chemical Structure

RR-90-7

Yasuhiro FUJIMOTO, Takayuki ANDO, Haruhiko ITAGAKI and Shigeru MORISAKI

: Evaluation of chemical hazards is important in the chemical industries because explosions or fires occasionally occur in the batch processes where fine-chemicals including pharmaceuticals or functional resins are produced. If the evaluation of these chemical hazards is possible by easy calculation, this may save chemical industries' money and time.
    Thermal hazards of chemical substances must be evaluated from the two aspect; sensitivity, which shows how easily chemicals may decompose or ignite, and severity, which shows how much or how fast energy is liberated when chemicals decompose. One of typical methods for evaluation of thermal hazards is DSC, which gives information on sensitivity (exo-thermic onset-temperature) and severity (decomposition heat) at the same time with easy operation. This report consists of results of multivariate regression analysis and discrimination analysis between DSC-data and chemical structures of benzene derivatives (307 samples). If these results could predict chemical hazards accurately enough, they would be used for direct evaluation of thermal hazards from chemical structures which are easily known without any experiments.
(1) Regression Analysis
    The following three sets were selected as suitable variables for pre-regression analysis in mono-substituted nitrobenzenes (51 samples) for the prediction of decomposition heat (kcal/mol).
  SET 1: Types of functional groups
  SET 2: Atoms in functional groups
  SET 3: Chemical bonds in functional groups
    From the results of calculating of the coefficient of determination (R ²) and the correlation coefficients, SET 3 was found out to be the most suitable model. The regression analysis using the variables in SET 3 in benzene derivatives (307 samples) gave relatively good result, showing R ² of 73.5%. However, another set of variables including influence of interaction between functional groups may be needed to improve the accuracy of this prediction. For the prediction of exothermic onset-temperature. SET 1 and SET 3 were also selected as suitable sets of variables in the same way. But the both models didn't show good prediction. This reason may due to the fact that exothermic onset-temperature is primarily influenced by minimum bonding energy in chemical substances.
(2) Discrimination Analysis
    Identification of thermal hazard rank of each di-substituted benzenes (128 samples) was carried out using discrimination technique on decomposition heat (cal/g). The kind of chemical bonds in functional groups was adopted as variables and the correct discriminate ratio was over 90%.
    However, our chemical knowledge is not enough to explain the meaning of the coefficients in the discrimination analysis. And more, it was difficult to decided the rank of chemicals which had the values of decomposition heat near the border (ex. 200 cal/g). It may be effective to apply fuzzy theory on this discrimination analysis to solve the above problem.

Resistance against High-Voltage, Low-Current Dry Arc for Electrical Insulating Materials of Organic Solid in Pressurized Air

RR-90-8

Tatsuo MOTOYAMA

: Some electrical insulating materials of organic solid have been required a characteristic which resists against higi-voltage, low-current dry arc. This characteristic is ranked according to an arc-resistance time, which is the time maintained insulation of surface of those materials against arc when they are exposed to the arc. The arc-resistance is tested in atmospheric pressure in accordance with the method described in the ASTM-D-495 or the JIS-K-6911.
    Recently, electric devices have come to be used in higher pressure atmospheres such as inside of caisson, marine vessels, etc. Moreover, the devices often include circuits of higher voltage than the maximum voltage, 12.5 kV, stipulated in those codes, for example, 30 kV such as the case of circuit in the TV-set. However, it is little known about the characteristics of dry arc resistance in high pressure atmospheres, and there is, therefore, a possibility of electrical hazards caused by arc discharge in such atmosphere.
    This study aims at investigating experimentally high-voltage, low-current dry arc resistances of organic solid insulating materials in high pressure air.
    The measurement of arc-resistance time was carried out in dry air compressed up to 1.9 MPa, in accordance with the methods described in the ASTM-D-495 or the JIS-K-6911 except for some test condition, such as open-circuit operating voltage at 50 kV. The materials tested were melamine-glass laminated sheet, phenolic-resin sheet and etc.
    The test results showed that the arc-resistance time of all materials decreased sharply with the increase of atmospheric pressure of up to about 0.4 MPa. The reason can be explained mainly from the fact that the electric arcs approach more closely the surface of test specimen with the increase of atmospheric pressure and accelerate, thereby, the carbonization of specimen.
    Thus, the author summarizes that the electric devices installed in high pressure atmosphere such as inside of caisson, in which pressure range is approximately 0.1 MPa - 0.3 MPa, have to considered about arc-resistance time of materials, which compose of electric apparatus, in high pressure atmospheres before those are installed.

Electrostatic Charging of Dielectric Liquids Mixed with Powders Caused by Agitation

RR-90-9

Tsutomu KODAMA and Yasuyuki TABATA

: Mixing of powders and flammable liquids is often conducted in chemical industry for various purposes such as suspension, emulsion, resolution, blending and so on. It is well known that a presence of dispersed phase due to solid particles or water droplets in poorly conductive liquids produces a large amount of electrostatic charges during an agitation and thereby may lead to fires or explosions due to electrostatic discharges.
    In order to investigate experimentally electrostatic charging due to the agitation of dielectric liquids loaded with various kinds of powders and charge suppression by an increase of liquid conductivities, experiments have been made using a cylindrical stainless vessel, 310 mm in diameter and 315 mm in depth, with an impeller of paddle type. The electrostatic charging of liquids was measured with an electrostatic field meter set on the top cover of the vessel, and the field strength measured was transformed into the charge density of liquids by multiplying a certain factor.
    Results obtained from the experiments are as follows:
    Mixing of any kinds of tested powders such as glass powders (frit), glass beads, adipic acid, titanium dioxide and epoxy resin, into the poorly conductive liquids, kerosene and xylene, caused not only a high level of charge density above 10μC/m³ during the agitation but also a large increase of electric potential after a stop of agitation when the particle size was below 200μm in medium diameter. These increases of electrostatic charging are ascribed to the separation of charges due to relative movements of solid particles and liquid at the interface between them during and after the agitation.
    Electrostatic charges increased with the increase of the quantity of the powders in liquid, but turned to decrease and reversed the polarity of charge with the further increase of the powders.
    Both additions of a conductivity additive "ASA-3" into kerosene and methanol or acetone into xylene indicated great reduction of the charge density down to a level of 0.1μC/m³ during and after the agitation when the conductivity of liquid was raised up to about 1 nS/m.

Historical Review of Occupational Accident Data Acquisition System in Japan and International Comparison of Accident Indicators

RR-90-10

Shigeo HANAYASU

: In Japan, the collection and compilation of statistical data on occupational accidents were officially initiated after the promulgation of the Japanese Factory Law in 1911. Since then the data acquisition system for the occupational accident statistics has been developed into a more elaborate and sophisticated system. This report presents a brief historical outline of the data acquisition system for the occupational accidents in Japan. In addition, an international review and comparison of the occupational accident indicators is presented. This analysis is based on the investigation of resolutions of the International Conferences of Labour Statisticians, which was promoted by the International Labour Organisation. Furthermore, necessary cares to be considered in the use of the accident statistics and another development needs in relation to the occupational accident data acquisition system are recommended.

JNIOSH