Abstract of Special Research Report (RR-87)

National Institute of Occupational Safety and Health, Japan

Prediction of Fatigue Crack Growth Lives in High Strength Steel Weldments under Low Temperatures

RR-87-1

Yoshio KITSUNAI and Etsuji YOSHIHISA

: Some defects such as undercuts or lack of penetration are sometimes introduced into welds in structures due to a restriction of welding procedure. As a structure containing these defects with high stress concentration was subjected to cyclic loadings under low temperatures, fatigue cracks sometimes initiated from such defects and propagate to such an extent that unstable fracture occurs. In particular, fatigue crack growth is accelerated at low temperature, becoming unstable because of reduced fracture toughness of the welds. To prevent such failures introduced by fatigue cracks growing from welding defects, it is important for us to obtain further knowledge concerning the behavior of fatigue crack growth and fracture toughness of welds under low temperatures.
    Numerous data on the fatigue crack growth of welds at room temperature have been obtained. However, data concerning welds at low temperatures are less commonly reported. As regards the study of fatigue crack growth in welds at low temperatures, it has been shown that the crack growth rate is mainly affected by the fracture mechanism with reducing fracture toughness and the residual stresses. The influence of residual stresses on the fatigue crack growth rate has been evaluated using an effective stress intensity factor range estimated based on the crack closure measurement. Application of this approach to real large welded structures, however, may be often difficult from the practical point of view. As superposition principle is applicable for an elastic body with a crack, hence the crack growth rate in material with residual stresses at room temperature has been evaluated by superposition of the respective stress intensity- for the residual stress field and for the applied stresses. Fatigue cracks at low temperatures mainly behave elastically. Therefore it may be possible to evaluate fatigue crack growth rate, da/dN, by applying the superposition approach. In this study, we assume that the effect of residual stresses on crack growth rate is equivalent to the stress ratio effect, the fatigue crack growth behavior of welded joints and the base metal in a HT 80 steel at temperatures ranging from room temperature down to 93 K is evaluated using the effective stress intensity factor range which is taken into account the residual stresses and temperatures. The effects of low temperatures on the fracture mechanisms during the fatigue crack growth and the fatigue fracture toughness, KfcR, are also examined.
    Two types of specimens, compact type (CT) specimens with 14 mm thickness and 10 mm width, and center-cracked tension (CCT) specimens with 4 or 8 mm thickness and 100 mm width, were machined out from the butt-welded plate of HT 80 steel. Notch in each specimen was oriented so that the crack passes through the weld metal parallel to the welding direction. Heat treatment was not performed for all specimens. The fatigue crack growth tests were carried out at room temperature (around 293 K) and 223, 173, 123, and 93 K using a 196 kN closed-loop servo hydraulic fatigue testing machine attached with a refrigeration chamber. Low temperatures were achieved by controlling the flow of liquid nitrogen into the refrigeration chamber. The stress ratios, R, for the fatigue tests were 0.05, 0.25, 0.5, and 0.8 ; and frequencies were ranged between 10 to 25 Hz.
    The main results obtained in this study are as follows:
  (1) As fracture mechanism during the fatigue crack growth was dominated by striation formation, the crack growth rates of the base metal and the welds were relatively insensitive to temperatures. Hence the crack growth rates of the welds were dominated by residual stresses rather than temperatures. In contrast, fatigue fracture toughness were markedly influenced by temperatures.
  (2) Fatigue crack growth rates in the ranging from 5 × 10^-9 to 7 × 10^-6 m/cycle for the base metal and welds were correlated with the effective stress intensity factor range, ΔKeffR, estimated by superposition of the respective stress intensity factors for the residual stress fields and the applied stresses regardless of residual stress distributions, temperatures and specimen configurations, when the crack growth was dominated by striation formation.
  (3) Fatigue crack growth lives in welds under low temperatures are able to predict using A KeffR, when the initial residual stress distribution in the welds is known.
  (4) Fatigue fracture toughness, Kfc, in the welds in CCT specimen was lower than that in CT specimen regardless of temperatures, because the crack tips in the welds in CCT specimen always existed in the fields of tensile residual stress, whereas the crack tip in the welds in CT specimen was always in the fields of compressive residual stress. The influence of residual stresses on the Kfc is able to evaluate using a superposition approach.
  (5) As the temperatures decreased below 123 K, fatigue crack growth rates of the base metal and welds increased markedly because of the occurrence of cyclic cleavage during striation formation. The cyclic cleavage was formed due to local deterioration of the material caused by cyclic straining ahead of the fatigue crack tip.

Principle and Logic of Safety in Man-Machine System

RR-87-2

Noboru SUGIMOTO

: Despite the fact that reliability of processing machines has been increasing recently occupational accidents show a rising tendency in last few years. At the same time, a newly born machine based on Micro-electronics, like a robot, that triggered Technology Innovation, presented new safety problems which is difficult to be handled by means of a conventional safety technology. These call for a rapid shift from conventional to new safety technologies.
    In this paper, an operation conducted between man and machine is comprehended as a man-machine system, and approached from the logical structure of the system.
    Safety operation is defined as that both the matter of "existence of human" (H) and the matter of "existence of movable parts of machine" (M) cannot happen at the same time and space.
    This definition leads to a principle of safety operation, namely, H∧M=0.
    In addition to that, the requirements of the system aiming at the safety operation are to meet the following principles of safety structure :
  1) Permission : of operation along with confirmation of safety,
  2) Inhibition of error on hazardous side.
    The logical system structure of safety is achieved by means of Interlock and Fail-Safe technology.

Construstion of Man-Machine Interlocking System for Remote Control Robots

RR-87-3

Noboru SUGIMOTO and Hiroyasu IKEDA

: Robots remotely controlled by human operators are powerful tools for developing the automation and safety of complicated jobs. A man-machine system to be built for such a remote control robot is required to be designed after careful consideration of conditions of the robot, operator and task.
Of the functions of the man-machine interface of the remote control robot, those functions that are especially related to safety are described in this report. Based on the assumption that a human operator inevitably makes mistakes when issuing commands to the robot, two types of interlockings are proposed for the remote control robot system. The robot system is equipped with a fundamental, main-interlocking for safety monitoring on the robot side and with a supplementary, quasi-interlocking for exceptional tasks that cannot be covered by the main interlocking.
A manipulator system approach is taken to add to the reliability of the quasi-interlocking.
The quasi-interlocking scheme that does not impair the safety of the human operator when he makes mistakes is developed for the pneumatic manipulator that can control the flexibility of operation over a wide range of conditions.

The Architectionic Principles and Designing of Hazard Control

RR-87-4

Yoshinobu SATO

: The technical terminology "Fail-safe" is now used as different concepts in various engineering fields. The discrepancy sometimes brings about confusion to the discussions on safety among the different engineering systems. It is important for carrying execution into safety plans for mechatronics, especially robotics which consists of overall engineering systems, that a generalized methodology to conceptualize safety measures be established.
    In this paper, first, hazard restraint measures are generalized as the following four principles, based on the "Action-change and Action-chain" hazard production theory :
  (1) Careful selection of system elements,
  (2) Prevention of undesirable changes in each system element,
  (3) Inhibition which prevents the system from transferring to a system phase in which a damage can be brought about, and
  (4) Control which prevents the system from causing a damage after the system has transferred to the system phase.
    Second, hazard control, which materializes the Hazard Restraint Principles (3) or (4), is generalized in terms of a "Dissociation of Action-linkages." Dissociations of action links are categorized by the following dissociation principles :
  P₁ ; Control of an action-source,
  P₂ ; Control of an action-path,
  P₃ ; Control of an action-source and path, and
  P₄ ; Control by substitution for a failed function.
    Third, the relationships between the dissociation principles and the action link dissociations, are given [Theorem 1]. Here, the dissociation actions, which act an element and are necessary conditions for dissociation of an action link from the element, are categorized as follows :
  (a'); Energy-transmission type,
  (b'); Information-propagation type,
  (c'); Agent-transfer type,
  (d'); Supply-obstruction type,
  (e'); Existence-form type,
  (f'); Function-cessation type, and
  (g'&g''); Function-substitution type.
    Next, the fundamental relationships between the dissociation principles and the dissociation actions are given [Theorem 2], and the schematic representation of the dissociations is shown in Fig.2. Control chains, which link a dissociation element and are necessary conditions for generation of a dissociation action from the dissociation element, are introduced. Control chains consist of control action links. Control actions are categorized into types “a''”, “b''”, “c''”, “d''”, “e''”, and “f''” like dissociation actions.
    Then, reversal action, which, are produced by a element's failure to generate a control action or a dissociation action, are categorized into types “a”, “b”, “c”, “d”, “e”, and “f”,. The fundamental properties of control or dissociation action reversals are developed [Theorem 3] : For a reversal chain resulting from a reversal of a control action link, the fundamental properties are given [Theorem 4, 5, and 6].
    Finally, quantitative characteristics of the control actions and dissociation actions are examined. A single-direction fluctuating action link, where the strength or property of the action is controlled in a single direction, and a dual-direction fluctuating action link controlled in dual or more directions, are introduced from the quantitative aspects of action control. Relationships between dual-direction fluctuating action links in a hazard control system are given [Corollary 15]. Relationships between control (or dissociation) action types and single-direction (or dualdirection) fluctuating actions are given [Theorem 7].
    Last, architecture of hazard control systems is demonstrated by examples involving robot failures etc., (Fig.2 - 5).
    Last of all, some discussions are made for carrying execution into effective safety plans.

Simple Shear Testing Method using Torsional Shear Apparatus and Strength-Deformation Characteristics of Sand in Torsional Simple Shear

RR-87-5

Noriyuki HORII

: Many field problems involve the simple shear condition as typically observed in such an element B in Fig.1. The simple shear test is one of the plane strain shear test in which, as different from the plane strain compression test, both strain increments ε_t and ε_r equal to zero.
    In this paper, the method for simple shear simulation using a hollow cylindrical specimen and the torsional simple shear behavior of sand are described. Drained tests at a controlled axial stress were performed auto-matically by means of a servo-system consisting of a personal-computer (PC 9801) and pneumatic-controlling system.
    A hollow cylindrical specimen has dimensions of a height of 20 cm, an outer diameter of 10 cm and an inner diameter of 6 cm. Tested sand is Toyoura sand and its physical properties are G_s = 2.64, D₅₀ = 0.16 mm and U_c = 1.46.
    The main results obtained are summarized as follows :
  (1) Simple shear testing method using torsional shear apparatus can simulate a simple shear deformation with a high degree of accuracy.
  (2) The major part of principal stress rotation takes place only at the early stage of shearing where γ_at is less than 1 %. At relatively large strain, however, the rotation is very small.
  (3) The strength and deformation properties of Toyoura sand at relatively large strain in torsional simple shear (TSS) are very similar to those by a plane strain compression (PSC) test in which the directions of principal stress with respect to the fabric of specimen and value of σ'₃ are similar to those at failure in TSS test.

Analysis of Fatal Accidents by Fall in the Building Construction Work

RR-87-6

Yasuo TOYOSAWA and Hisao NAGATA

: There are a lot of occupational accidents by fall on the building construction sites. Approximately 70 % of the fatal accidents and 30% of all the fatal accidents by fall occur under building construction.
    In order to establish the effective countermeasures against the accidents caused by fall, it is first necessary to analyse the accident statistics.
    In this paper, 1,118 actual fatal accidents by fall (652 accidents in Steel and Reinforced concrete building Construction. 466 accidents in Wooden building Construction) occurred during 1979 - 1984 which were reported to the Labour Standard Inspection Offices are investigated.
    The outline of this paper is as follows ;
  (1) In 1985, accidents by fall at the construction work number 20,349 including 332 fatalities. Among these accidents, the steel and reinforced concrete building construction work occupy 5,258 accidents including 103 fatalities, while the wooden building construction work occupy 7,863 accidents including 66 fatalities.
  (2) The number of accident per construction area at wooden building construction work is increasing. In 1985, 1.03 fatalities by fall/1 million m² at wooden building construction work and 0.93 fatalities by fall/1 million m² at steel and reinforced concrete building construction work are numbered, respectively.
  (3) On the steel and reinforced concrete building construction sites, the deaths mainly came from falls from scaffold (34 %), falls from building beam (15.1 %). As to the type of work, a large number of accidents occur during exterior finish work (23.8 %), temporary work (particularly erection and dismantling of scaffolds) (20.4 %) and steel work (17.2 %).
  (4) On wooden building construction sites, roof (24.4 %), beam (23.1 %), scaffold (16.2 %) and ladder (8.6 %) are the main workplaces where the fatal accidents by fall occur.
    Regarding the type of work, a large number of accidents occur at roofing work (16.1%) and roof truss making work (11.1%).
  (5) Concerning the type of occupation, the scaffold men who suffered from falls account for one fourth of the fatal accidents at steel and reinforced concrete building construction work. On the contrary, the carpenters account for 45 % of the fatal accidents at wooden building construction work.
  (6) The average age of fatal workers on the building work sites is higher than that of the injured workers. This trend is particularly shown at the wooden building construction work. The average age of the fatal wooden building construction workers is 50 year-old.
  (7) The small scale industries tend to experience a greater number of fatalities. 82 % of the fatalities at steel and reinforced concrete building construction and 96 % of the fatalities at wooden building construction occur in the small scale enterprises with less than 30 workers.
  (8) The average height of the falls is 8.5 m (the standard deviation is 6.2 m) at steel and reinforced concrete building construction work, 4.2 m (S.D = 1.6 m) at wooden building construction work.
  (9) Concerning the cause of fatalities, about 70 % of the fatal worker got injuries on their skull.

Experimental Development of the Data Base 'SAFE' (Data Base System for Labour Accident Fact Exploration)

RR-87-7

Yoshimi SUZUKI and Yutaka MAEDA

: It is generally recognized that information on labour accidents is important for establishing countermeasures against other similar accidents ; in other words, the information is necessary for the development of preventive technology against labour accidents. In this regard, it is desirable to establish a data base system concerning the accidents in order to make effective use of information on labour accidents.
    In this context, a fundamental concept about the establishment of this kind of Data Base System was discussed by the experts in a committee founded by the RIIS (Research Institute of Industrial Safety), and the basic guide lines about the establishment of the "Industrial Accident Data Base System" were proposed in a report submitted to the RIIS in 1981.
    However, as the 1st step of development of the data base system, it is necessary to frame a prototype system, not only for examining problems about practical use of the system, but also for realizing a high performance of actual utilization of the system.
    From this point of view, experimental development of a prototype data base system has been made in this study acceding to the guide lines indicated in the report. As main function of this prototype system, information retrieval for data exploration of actual labour accidents is taken into account. In this prototype data base system, 99 information items are picked for postulated information retrieval (38 numeric data items, 58 Japanese character items and 3 Japanese text items).
    This prototype system is named "Data Base System for Labour Accident Fact Exploration (abbreviated as Data Base SAFE in this paper)", and when considering it's characteristics, it will be one of the sub-systems of the "Industrial Accident Data Base System" in the near future.
    This paper presents details of experimental development of the "Data Base SAFE", in particular, concerning actual procedures for development of this system and it's prior design.
    The main substances of this paper are as follows :
  (1) Role and situation of information on labour accidents.
  (2) Contents and characteristics of information sources ; "Mission reports on investigation of labour accidents by the Ministry of Labour" and "Industrial accident reports submitted to the Ministry of Labour by enterprises."
  (3) Proposed form of information in the data base system concerning labour accident.
  (4) Types of assumed methods for information retrieval on actual labour accidents.
  (5) Precondition for experimental development of "Data Base SAFE".
  (6) Selection of DBMS (Data Base Management System) for "Data Base SAFE".
  (7) Establishing of information items of "Data Base SAFE".
  (8) Actual procedure for the development of "Data Base SAFE".
  (9) Basic concept for update procedure of "Data Base SAFE".
  (10) Development of programs for data handling in "Data Base SAFE".
  (11) Management concerning utilization of "Data Base SAFE".
  (12) Practical procedure for information retrieval on actual labour accidents in "Data Base SAFE".

Statistical Analysis of Accidents in Bllet Train Consruction Work

RR-87-8

Shigeo HANAYASU and Yoshimi SUZUKI

: The situation surrounding construction work has been changing rapidly due to the increase of large scale construction, introduction of highly innovated technology, aging of workers in sites, etc.
    Accordingly, the features of accident occurrences in terms of the causative factors to accidents and their combinations have also been changing.
    Hence, in order to establish the effective safety countermeasures, it becomes necessary to understand the accident situation in the construction work thoroughly in accordance with its changes.
    This paper deals with the accident investigation for the purpose of getting the basic nature of accident as well as their changing situation in the recent construction work.
    The construction work under the present investigation is that of the Joetsu-Shinkansen bullet train railway (from Ohmiya to Niigata ; 275 km in length ; 1972-1982). This project covers various types of structure construction work, such as tunnels, bridges, etc.
    The accident data were made available from the Industrial accident reports submitted to the prefectural labour standards offices and the labour standards inspection offices involved. They totaled in number 2150 injuries including 89 fatalities.
    To throw light on the nature of labour accidents in construction sites, various statistical analyses were carried out. The main results obtained in this analyses are described as follows :
  (1) In the Joetsu-Shinkansen bullet train construction work, accidents associated with tunnel construction account for 74 %, and overhead bridge construction takes about 15% of all accidents.
  (2) The number of accidents concentrates in the middle aged groups. However, seeing from the accident frequency rate, accidents are more liable to occur in younger and elderly personnel, which is twice or 3 times higher than the average rate.
  (3) The number of accidents as well as the accident frequency rate are more likely to increase with the decrease in the years of work experience.
  (4) The frequency distribution of accidents classified by time shows that about 27 % of all accidents have taken place in the night time shift. The particular feature of accident occurrence in regard to time also shows the peaks after 2-3 hours after the rest period or the beginning of the work.
  (5) From the comparison of accidents classification by work activities between conventional tunnelling method and NATM method, roof bolting and mortar placing are the activities frequently cited in the NATM tunnelling method.
  (6) CTA analysis was carried out to clarify the frequency of contributive causative factors that lead to an accident occurrence and their combinations for various types of accidents, working time, etc. By making use of the influential transition matrices, the similarity in accident situation between activities, working time, and others in tunnelling work are examined.
  (7) The relationship between workdays lost due to accidents and their frequency can be expressed as a log-log linear straight line. If the slope of a line exceeds -1, the expectation of the lost workdays including fatal accidents can obtained within a limited value. The expectation of workdays lost with fatalities for the Joetsu-Shinkansen tunnel construction was 90 days. No distinguished difference in the expected lost workdays was found between Sanyo Shinkansen accidents and Joetsu-Shinkansen accidents.
(8) The consequence of accidents (magnitude) that indicates the number of injured workers involved in an accident tends to become larger in the Joetsu-Shinkansen tunnel construction compared to the Sanyo-Shinkansen tunnel construction due to the increase of potential risks of fire during tunnelling work.
    The return period of an accident involving more than 40 injury workers in tunnel construction work was estimated about 250 years.

Critical Ignition Temperatures of Chemical Substances

RR-87-9

Takashi KOTOYORI

: Ignition or explosion accidents in which any thermally unstable chemical substances are concerned often occur, due to failure of temperature control, when these substances are processed in manufacturing plants such as reaction or distillation vessels or driers, or when stored in large -scale tanks or warehouses, especially when kept in circumstances a little hotter than normal. Main measures to prevent this kind of accidents are to keep substances at temperature levels sufficiently below their critical ignition temperatures. Critical ignition temperatures (hereinafter referred to as T_c values) of thermally unstable substances become, thus, important data to rely on, when we aim to prevent, by temperature control, the fire or explosion accidents which these substances bring about. Instances, however, have been only a few so far where the T_c values of bulky stacks of chemical substances were actually measured and/or calculated and reported as reliable data.
    In this paper, T_c values of eight kinds of self-heating or thermally unstable chemical substances are calculated and reported, using an adiabatic self-ignition testing apparatus (called SIT), and applying a procedure. Both SIT and the procedure have been developed at RIIS. A detailed description of SIT was given in ref.13), and that of the procedure in ref.14).
    Substances tested are nitrocellulose, lauroyl peroxide, p-toluenesulfonylhydrazide, α,α'-azobisisobutyronitrile, benzoyl peroxide, high test hypochlorite, p,p'-oxybisbenzenesulfonylhydr azide and dinitrosopentamethylenetetramine.
    Measuring conditions are as follows : sample amount, ranging from 450 mg for nitrocellulose to 3 g for high test hypochlorite ; sample container, sealed cell made of Pyrex glass (ca. 2 ml) ; initial sample (starting) temperature, ranging from about 120 °C for nitrocellulose to about 50 °C for lauroyl peroxide.
    Main results are as follows :
  1) By applying SC-SIT (SIT measuring techniques with sealed cell), adiabatic self-heating behaviours of almost all chemical substances become measurable safely, easily and with good sensitivity, without being botherd by vapourization phenomena of sample itself or by endothermic effects accompanying such vapourizations and by evolution phenomena of corrosive or toxic gases.
  2) Self-heating chemical substances are classifiable into two groups when SC-SIT is applied ; one is a group, tentatively named "T-C (thermal combustion) type", and is the one showing such a self-heating behaviour, as shown in Fig.6, that when a sample is exposed to an atmosphere of temperature T, the sample temperature reaches T within several tens of minutes, then it goes on rising relatively slowly with such a rate varying with the T value that Arrhenius equation ordains ; and the other is a group, tentatively named "A-C (autocatalytic) type", and is the one showing such a self-heating behaviour, as shown in Fig.5, that when a sample is exposed to an atmosphere of temperature T, the sample temperature approaches gradually to T over a long interval of time, and in course of time, as soon as T is reached the temperature begins a sudden and rapid rising. These two heating behaviours correspond to combustion modes of the thermal combustion and autocatalytic combustion, so designated by Frank-Kamenetskii, respectively (ref.15). Frank-Kamenetskii's critical condition for thermal explosion is applicable to only chemical substances of T-C type.
  3) The procedure yields T_c values in good agreement with corresponding data calculated or observed by other researchers (Table 3). Therefore the procedure, which was shown to be applicable to calculation of T_c values of oxidatively heating substances such as wood sawdusts in the preceding paper (ref.14), has been now proved to be sufficiently applicable also to calculation of T_c values of self-heating substances. In the case of five substances of T-C type, The T_c value was calculated to range from 56 for α,α 7-azobisisobutyronitrile to 78 °C for dinitrosopentamethylenetetramine, assuming each to be packed in a sealed infinite cylinder, 40.64 cm (16 inch) in diameter. In the case of three substances of A-C type, it was reasoned that there is a fair potentiality for these chemical substances to ignite ultimately, if nitrocellulose is kept at 41 °C, lauroyl perxide at 42 °C and p-toluenesulfonylhydrazide at 78 °C for 60 days, respectively.
  4) By using Equation (11), it is possible for every substance of A-C type to calculate a T value, which can be utilized as a kind of critical temperature to prevent ignition accident due to substances of A-C type, corresponding to an arbitrary Δt value. Examples of such calculations are presented in Table 1.
  5) T_c is a function of only r in Equation (9). Therefore, if (T_c - r ) diagram is provided in advance for every substance of T-C type, it enables us to read readily a T_c value corresponding to an arbitrary r value, though being a rough value. Examples on benzoyl peroxide are exhibited in Fig.15 and 16.

Dust Explosions in a Cyclone System

RR-87-10

Toei MATSUDA and Toshihiro HAYASHI

: Dust explosions in industrial plant can occur in various process units, such as hoppers, dust collectors and ducting. Cyclones are the most widely used industrial dust collectors for separating dust and air from solid-gas streams. Some engineers recognize that dust explosions might be avoided in a cyclone, which is employed usually for the collection of coarse particles with spiral flows. Larger vortex in a cyclone, however, will be expected to increase a burning rate of dust suspensions. To assess the dust explosion hazards, some experiments were carried out in an industrial scale cyclone system.
    The cyclone plant consisted of several units, as shown in Fig.1, and the explosion pressures were released through the vents on the quenching box connected to an outlet ducting from the cyclone. The cyclone vessel, of volume 0.32 m³, was constructed to withstand much higher explosion pressures than those anticipated, and was situated on a dust settling chamber of volume 0.15 m^3. The air velocity was measured in the inlet duct to the cyclone, and dust concentration was given from the amounts of the air flow and the dust supplied. The ignition source was a small amount of Al and Mg powders mixed with barium nitrate and barium peroxide. ABS-resin dust and ethylene-vinylacetate-copolymer (EVA) were used as fuel, which were of industrial grade with average diameters of 183 μm and 42 μm, respectively (Fig.3).
    The first series of tests were to measure explosion pressures in non-vortex dust suspensions in the cyclone without driving an exhaust fan. For the static tests, dust was dispersed through two perforated tubes by the aid of high pressure air (Fig.2). The maximum value of the explosion pressures was 145 kPa, obtained with variations of dust quantities up to 300 g (Fig.8). Comparison was then made between the pressures in dust explosions and in methane-air explosions in the cyclone without turbulent flows. For the static gaseous mixtures, a maximum pressure of 202 kPa was observed in the cyclone when explosions were vented into the duct system with two venting doors on the quenching box (Fig.14).
    Further tests were carried out while the fan was working. The flowing dust-air mixtures were ignited at position X1 in Fig.1 and the explosion flames were propagated into the cyclone vessel. Variations of explosion pressures in the cyclone with dust concentration are shown in Figs.17 and 18, at air velocities of 5.2, 8.3 and 13.4 m/s. An increase in the air velocity made slightly higher the explosion pressures for two kinds of dust tested (Fig.19), yet the maximum pressures attained were lower than those observed under the non-flow conditions. On the other hand, the data obtained from time measurements between ion-gaps showed that the apparent flame speed in the cyclone remained in some 10 m/s (Fig.22). The flame detectors showed that explosion flames were seen both in the inlet and outlet ducts, and fire balls were often formed outside near the mouth of the dust feeding duct. On these observations, any data have not been given for the effect of vortex formation on the explosion to be much serious in the cyclone.
    A working cyclone causes vortex flow, in which dust is concentrated close to the wall. These turbulence and concentration gradient show the conflicting effects on dust explosion hazards by each other. Palmer (Refs.4 and 5) cited only a fraction of the volume of the cyclone being filled with an explosible dust suspension for a probable explanation of the low pressures in their closed cyclone plant. For the cyclone in the present work, it seemed that the reduction of explosion pressures would be explained as the results of effective explosion venting and compulsive suction of the explosion flame towards the fan. Unless appropriate vents were provided with, explosion flame could be markedly intensified by the action of the fan.

Flame Propagation Characteristics of Flowing Dust-Air Mixtures in a Tube

RR-87-11

Toei MATSUDA

: Dust explosion and fire hazards should be taken into consideration in the design and operation of industrial pneumatic conveying systems associated with combustible dusts.
    This paper describes flame propagation characteristics of flowing dust-air mixtures in a 4.
2-cm diameter, 13-m length tube of a suction-type pneumatic conveying system.
    In the small horizontal tube system, flame velocities were measured as a function of dust concentration at several conveying air velocities. The dusts studied most extensively are cork dust and ABS-resin powder. The flame velocities were obtained from the output signals for photodiodes and they merely represent the rate of leading flame front in propagation, for which correction was not made at various air velocities. The ignition source situated in the middle of the test tube length was given by a spark, whose energy was relatively small and at a level from 2 to 3 J, depending on a flow condition.
    Apparently there are two types of flames, namely downward and upward propagation.
Figures 2 and 3 show flame velocity versus distance curves at various dust concentrations and air velocities. Most of the downward expanding flames reach to peak of flame velocity upon ignition.
Then, at low air velocities and higher dust concentrations above - 0.2 kg/m³, the flame velocities are diminished to a velocity such that the flame front almost stopped, meanwhile at the lower concentrations the flame velocities decrease gradually although the tube length appears to be short to observe full deceleration of flame velocity in the downstream. At higher conveying air velocities, the flame velocities certainly decrease with some variation after the peaks, but within a range of relatively thin dust concentrations, they are re-accelerated towards the dust collector.
In the upstream propagation, flame velocities increase relatively slowly at first and then rather rapidly, towards an open end of the tube.
    The maximum values of the peak downstream flame velocities are given at the dust concentrations from approximately 0.12 to 0.16 kg/m³ for the ABS-powder, and from about 0.15 to 0.20 kg/m³ for the cork dust with increasing air velocities from 15 to 30 m/s, respectively. These concentrations are lower than the optimum dust concentration which minimized the minimum ignition energy at various conveying air velocities. The upstream flame propagation is only possible within a narrow range of dust concentrations, around that maximizes the downward peak flame velocity.
    These dust flame propagation behavior would be partly similar to that for gaseous mixtures in a tube. However, the dust flame propagation in this observation is seemingly characterized by a rapid decrease in flame velocity after initial flame acceleration. Although this could be explained by a flow resistance at the tube wall due to high viscosity of the flame, the additional data of dynamic increase in dust concentration before the flame front seem to give another probable reason. Because of the dependence of flame velocity on dust concentration, the increasing concentration in front of the flame will cause a more rapid burning if the dust concentration is much lower than the actual stoichiometric concentration; this in turn will increase the dust concentration, and so on. Thus, the maximum peak values of flame velocities may have been recorded at a concentration on the lower side of the spark ignitable limits. If the dust density in flows is increased even more, the concentration ahead of the flame will be increased more than the stoichiometric composition, possibly near to or over the upper limit, resulting in the collapse of the flame into smouldering particles or in quenching.
    Furthermore, turbulent mixing immediately ahead of the flame seems to be responsible for the variation in flame speeds. The rate of combustion can be considerably enhanced in the flow with a high level of turbulence. Too much intense of turbulence, on the contrary, may cause decrease in burning velocity. Such a dual role of turbulence was tentatively and qualitatively cited to explain the phenomena on the flame propagation in the small tube.
    In conclusion, it is unlikely that the flame will accelerate into detonation with weak spark ignition in comparatively small tubes of pneumatic transportation systems for common organic dusts.

On the Cause of Accidental Explosion of Oxygen Hoses

RR-87-12

Hidenori MATSUI and Kougaku KOMAMIYA

: Many accidental explosions of oxygen conveying rubber hoses for gas welding or steel manufacturing industries have occurred in Japan and other countries. A typical case shows rupture of a hose in many places due to a detonation. The cause of the explosion may be that a fuel, such as acetylene reversely flows into oxygen hose and formes a fuel-oxygen mixture, then it will be detonated. However, as oxygen pressure used in a workshop is usually much higher than that of the fuel gas, the reverse flow of fuel is hardly realized. In order to make clear the real cause of these accidental explosions and to establish the preventive measures of the accidents, three kinds of experiment, reproducing the explosion phenomena, have been carried out.
    In the first experiment, based on the assumption of the possibility of reverse flow of a fuel into oxygen, the propagation of gaseous detonations of acetylene-oxygen mixture in rubber hoses and the effects on rubber hoses were investigated. As a result, it was proved that the oxygen rubber hose was hardly ruptured by oxy-acetylene detonation at usual oxygen pressure unless the hose had been degraded. In an excess concentration of oxygen, the inner surface of the rubber hose was ignited by propagation of gaseous detonation and large amounts of soot, hydrogen and carbon monoxide were produced by the succeeding diffusion combustion of the rubber. These combustibles were re-detonated by another ignition.
    In the second experiment, soot film detonation in oxygen was investigated to confirm the secondary explosion of the soot produced by diffusion combustion of the hose rubber. The inner surface of the test tube was coated with thin layer of soot which was formed by the decomposition of acetylene induced by shock wave. The soot film detonation was directly initiated by an oxy-acetylene gaseous detonation. The wave velocity in a transparent vinyl resin hose was determined from self luminous streak photographs with a rotatory drum camera, and the wave velocity determined from ion current detection with ion gaps in steel pipe. The wave pressure was also determined with piezo electric pressure transducer. Almost steady soot film detonations were observed with the wave velocities of 1,500 - 2,000 m/sec. The peak pressures of the wave were about 56 times the initial pressure of oxygen. Therefore, the soot could be one of the fuel materials for the accidental explosion of oxygen conveying rubber hoses.
    In the third experiment, DTA (Differential Thermal Analysis) was performed with the fats which was initially adhered to the inner surface of a rubber hose in the manufacturing process. This experiment was pre-examining to see the possibility of solid fat film detonation in a hose. Exothermic starting temperature, ignition temperature and wt.% of burned residue of the fats were determined from DTA. Next, fat film detonation in oxygen conveying rubber hose was investigated. Attempts to detonate in smaller hose of 6.5 mm i.d. were not successful, but in larger hoses of 25 mm i.d., the fat film detonation with steady wave velocity was achieved in the range above the oxygen pressure of 1.5 atm. However as a quantity of the fat in the hoses is too small, it may be suggested that the rubber itself could be a fuel material. The wave velocities in these experiments were 1,350 - 1,450 m/sec, and the peak pressures were 31 - 32 times the oxygen pressure.
    From the above experimental results, we may conclude the following on the cause of accidental explosion of oxygen conveying rubber hoses : Rubber is essentially combustible material and is easily ignited in oxygen by various ignition sources, such as static electricity spark, friction spark of particles, heat of adiabatic compression of oxygen and flash back at the torch, etc. Once ignition of inner surface of the rubber hose has occurred, proceeding diffusion combustion of rubber produces other combustibles, such as soot (carbon), hydrogen and carbon monoxide, etc. These combustibles are dispersed into oxygen flow through the hose and form detonable mixture. This mixture will be re-ignited by the first diffusion combustion or flash back at the torch. The flame in soot, combustible gases-oxygen mixture will easily develop into two phase detonation, sometime including fat film detonation, with higher wave pressure and longer reaction zone compared with homogeneous gaseous detonations. For preventive measures, removal of the ignition sources, regular examination and cleaning of hoses are important.

Diagnosis of Degradation of Rubber Insulation Gloves for Use in High Voltage Working --Thermal Degradation of Natural Rubbor Insulationg Material--

RR-87-13

Kenji ICHIKAWA and Ryuji TANAKA

: In order to diagnose the degree of degradation of rubber insulating gloves for use in high voltage working, it is necessary to have much knowledge about the tendency of their degradation in service and relating to various degradation factors. The degradation in service and by voltage and ultraviolet has been described in the previous reports 1),2),3). In this report, the authors describe the experimental results of thermal degradation of rubber gloves.
    These gloves are made of natural rubber insulating material, and the oxidazation reaction is caused by their being kept in atmosphere without actual use. Therefore, it is very important to clarify the relations between the degree of degradation due to oxidazation and the insulation performance, and the mechanical strength.
    The degradation due to oxidazation is a sort of chemical reaction, and this reaction proceeds severely under high temperature atmosphere. Therefore, the heat acceleration tests of natural rubber material being used rubber insulation gloves were carried out and the relations between the degree of degradation due to oxidazation and the insulation performance, and mechanical strength were investigated. In addition, the activation energy and the thermal endurance life formula of rubber insulating material were derived to instigate it.
    Principal conclusions obtained are as follows :
  (1) When rubber material is heated, it causes the degradation due to oxidazation, and it releases disintegration gases from the internal part of it and its weight is decreased. The higher the heat temperature is, the more decrement of weight is obtained with the greater transition of decrement at early stage of heating.
  (2) The deterioration of insulation performance such as the dielectric loss tangent and volume resistivity appears when heat acceleration is given severely to rubber material while the breakdown voltage does not appear. The dielectric loss tangent and volume resistivity show the tendency of deterioration when the decrement of weight becomes about 30 mg/g.
  (3) The deterioration of mechanical strength such as tensile strength and percentage of elongation appears more clearly than that of insulation performance. But that tendency shows that the deterioration occurs suddenly when the decrement of weight becomes about 30 mg/g.
  (4) The activation energy obtained from the decrement of weight due to the thermal degradation is 9.3 - 12.4 kcal/mol.
  (5) If the oxidazation that the decrement of weight becomes about 28 mg/g is the life due to the thermal degradation, the thermal endurance life formula is given as follows:
       1nt = (4.68 - 6.24) × 10³ / T - ( 8.4 - 12.6)

Protection Effects of DC Leakage Operated Circuit Breaker and High Resistance Grounding against Electric Shock Hazard Underwater

RR-87-14

Eiki YAMANO, Tatsuo MOTOYAMA and Ryuji TANAKA

: Working in the sea or other waters and the use of electricity in such areas are increasing in recent years. These situations may lead divers to hazards of underwater electric shock. This paper reports on an investigation of the effect of a protection method for such hazards.
    As a shock-hazard protection method, a circuit breaker which is fitted on an a.c. supply grounded midpoint by high resistance and operated by d.c. leakage signal is suggested, and the effect of the method for protection against electric shock underwater is investigated by inference and experiment. The greater part of the investigation is applicable to other types of earth-leakage circuit breaker, because it is a common defect of circuit breakers that electric shock during operating time is not prevented.
    From the results of investigation, it is concluded that the suggested method protects a person from instant death due to electric shock underwater in the majority of cases and greatly contributes towards the safety of underwater use of electricity.
    The results of investigation are summarized as follows :
  (1) It was inferred that the instant death from electric shock due to single line ground fault underwater can be prevented by the d. c. leakage operated circuit breaker and the high resistance grounding.
  (2) The instant death due to fault of both lines underwater is also prevented in the cases that ;
   i) the time interval between faults is beyond the operating time of breaker,
   ii) the progress of fault (i.e. decrease of insulating resistance) is slow,
   iii) the fault locations are inside of conducting screen (e.g. metal casing, etc.),
   iv) a person is not located in hazardous domain around a fault area.
  (3) When a person is located in hazardous domain produced by a fault of both lines, death from electric shock may occur in a certain range of frequency according to the conditions. In the experiment with rabbits, a serious effect which may lead to instant death occurred twice in 52 shocks by artificial fault underwater.
  (4) Rabbits were instantly killed by shocks in the experiment without protection of circuit breaker.
    When a local protection method such as a double insulation or a conducting screen fails, the supply system may remain in undetected hazardous state. Therefore, it should be considered to protect by a method whose effect covers whole line, such as an earth-leakage circuit breaker, in addition to local protections.

JNIOSH