Abstract of Special Research Report (RR-88)

National Institute of Occupational Safety and Health, Japan

Application of Fracture Mechanics to Estimating Strength of Lifting Link Chains

RR-88-1

Masazumi TANAKA

: In order to prevent sudden fracture of lifting chains containing flaws introduced during service, fracture toughness and fatigue crack growth behavior of Mn-B steel lifting chain with 25.4 mm dia. were evaluated using C-shaped specimens which were machined from the chain link. Elastic-plastic fracture toughness, J_IC, test for the C-shaped specimens was carried out in accordance with stretched zone method proposed by JSME 001. Fracture toughness of C-shaped and compact type specimens which were made from the same materials that used to make chains were compared. The effect of temperature on the fracture toughness of link chains was examined by 1T compact type specimens at the temperatures ranging from room temperature down to -80°C. The fracture toughness test and the fatigue crack growth test were carried out using a 196 kN closed-loop servo hydraulic fatigue testing machine. The stress ratio for the fatigue test was 0.05 and frequencies were ranged between 10 to 25 Hz. Fracture mechanics calculation of critical flaw size for the Mn-B steel chain was performed. It was found that the value of elastic-plastic fracture toughness, J_IC, determined by the C-shaped specimens was almost the same with that obtained from the compact type specimens. Moreover, corresponding K_IC(J) value converted from J_IC was almost agree with K_IC value determined from linear fracture toughness test. Hence J_IC test for C-shaped specimens machined from chain link was suitable method to evaluate the fracture toughness of lifting chains. Fracture toughness, K_IC, of the fracture-mechanics calculation, the critical flaw size of Mn-B steel chain was found to be 3.15 mm at room temperature. Fatigue crack growth rate of C-shaped specimens machined from chain link was slightly higher than that of compact type specimens machined from the chain material.

The Basic Safety Design for Control of Intelligent Mobile Robots

RR-88-2

Yoshinobu SATO

: Robots are composed of wide-ranged engineering systems covering not only hardwarebut also software creating artificial intelligence. An overall safety plan must be carried intoexecution for the safety improvement of robotics. The safety plan consists of the following stages:
  (1) Hazard identification,
  (2) Conceptualization of hazard-restraint measures,
  (3) Qualitative and quantitative systems analyses, and
  (4) Specification of system gesign.
    A systematic methodology for stage (1) and (2) analyses, which enable us to identify hazardsand to conceptualize hazatd-resteaint measures involving hazard-control systems, have beendeveloped. In this paper, hazards typically produced by autonomous mobile robots are identifiedand the structure of hazard-control systems for the hazards is discussed based on the methodology.
    First, hazards 'a robot collides with a static object' and 'a robot collides with a mobileobject' are identified using A-C models [Fig.1 and 2].
    Second, hazard-restraint principle 3 is applied to the both hazards, and hazard-control systemsare conceptualized, given that the robot is initially in an area where there is no possibility of collision [Control Chains (1) and (6), and Fig.3 and 4].
    Next, the fundamental difference between a safety-confirmation information-processing systemand a danger-detection information-processing system is defined [Control Chains (2) and (3), and Reversal Chain (4) and (5)].
    Then, hazard-control systems, which materialize hazard-rsetraint principle 4 under thecondition that the system phase is in the modes 'a robot is moving' or 'a robot exists on a collision course', are structured based on the dissociation theory of action linkage. These hazard-control systems involve the following control forms:
  (1) Stop by simply braking [Control Chain (7), and Fig.5],
  (2) Stop by braking control against a skid [Control Chain (8), and Fig.6 and 7],
  (3) Stop by multiple control against a turnover [Control Chain (9), and Fig.8 and 9],
  (4) Course control against a single static object [Control Chain (10), and Fig.10],
  (5) Course control against multiple static objects [Control Chain (10), and Fig.11], and
  (6) Multiple control against multi-mobile objects [Control Chain (ll), and Fig.12],
    Last, the applicability of safety-confirmation information-processing systems and danger-detection information-processing systems is examined. And the relationships among the identified hazards, the whole hazard-space, and the domains in which the conceptualized hazard-control systems are applicable, are schematically represented [Fig.13].

A Study on the Time Intervals between Accidents (4)

RR-88-3

Shigeo HANAYASU

: The accident frequency rate has been widely used as a measurement of safety performance in many workplaces over a long period of time.
    In order to explore the significant changes in the accident situation in succeeding intervals of time, testing hypotheses for the accident frequency rate were employed. In the analysis, the time intervals between occupational accidents were utilized as a useful indicator to give expression in safety performance in workplaces having a certain accident risk. This paper deals with the evaluation function of the testing hypothese such as the expected number of accidents and the expected time to reach a decision. In particular, relative efficiency of sequential probability ratio test was examined through the analysis of these evaluation functions.
    The main features of this paper are as follows:
  (1) The parameter of probability distribution function of the time intervals between labour accidents can be connected to the accident frequency rate. Hence, statistical evaluation of the time period to the occurrences of accidents with reference to the probability for a specific accident frequency rate can be performed.
  (2) The procedure of testing hypotheses for the accident freqency rate to evaluate the changing tendency in the accident situation was described. Necessary sample number of accidents satisfying the testing requirements as well as critical region times to conduct testing hypotheses were analyzed.
  (3) The procedure of sequential probability ratio test for the accident frequency rate was also presented. Numerical examples of statistical evaluation by sequential tests were demonstrated by making use of the serious accidents in building construction.
  (4) Evaluation functions characterizing the testing hypotheses were derived in terms of the expected sample number of accidents to reach a decision and the expected time to a decision both for conventional testing hypohteses and sequential probability ratio test.
  (5) The expected number of accidents and expected time to complete a decision fluctuate in accordance with the changes of the null hypothesis Ao.
    In order to avoid tedious calculation due to the change in the values of null hypothesis Ao, normalization of the expected number of accidents as well as the expected time to complete a decision was achieved. Concerning to the expected number of accidents, the normalization was accomplished simply by dividing the expected accidents number by minimum necessary number of accidents Ko. Similarly, the expected consumption time to reach a decision being divided by the critical region time Tc yields the normalized expected reaching time.
    In addition to the normalization, by making use of the relative accident frequency rate (ratio of arbitrary accident frequency rate to the accident rate of null hypotheses Ao), unified representation of these normalized expected functions free from the change of the null hypothesis Ao), can be attained.
  (6) Relative efficiency of the sequential test compared to the conventional testing in terms of the expected accidents number and the expected time were analyzed. The relative accident frequency rate also gives unified representation of the relative efficiency.
    It was found that the efficiency of a SPRT test was depend on both descriminant ratio and the relative accident frequency rate. Analytical results also showed that the efficiency of a SPRT test was at least more than 10% compared to the conventional test.

An Experimental Development of the Data Base "SAFE" (Data Base System for Labour Accident Fact Exploration)(2nd Report) --Development of the Information Retrieval Supptot System--

RR-88-4

Yoshimi SUZUKI, Yutaka MAEDA, Shigeo HANAYASU and Takayuki ANDO

: Development of the Information Retrieval Support System for the Data Base System for Labour Accident Fact Exploration (abbreviated as Data Base SAFE in this series of studies) has been carried out in this study.
    It is recognized that making effective use of available information on labour accident is essential for establishing countermeasures against similar accidents. In this connection, the development of the Data Base SAFE had been made as the prototype system, and outline of the development was discussed in the first report¹⁾ in 1988.
    This data base system (Data Base SAFE) has already been in service for assisting research activities in RIIS (The Research Institute of Industrial Safety).
    During the course of the use, however, it was pointed out that this prototype system has some disadvantageous points : the first one is the complexity of accessing to this system (particularly for end users who do not have sufficient knowledge of usage procedure about this database system), and the second point is the insufficient functions for treatment of the Japanese keywords. These keywords are automatically produced by means of Japanese keywords production rule supplied in this data base system.
    Therefore, it is desirable to improve the Data Base SAFE which includes the development of the support functions for facility of actual information retrieval operation, and support functions for management of Japanese keywords as well. So, the goal of this study was set up to furnish these functions as a conversational interactive system for the Data Base SAFE.
    The properties of new functions which have been developped in this study, are systematized as the Information Retrieval Support System. This system can also be applied widely not only for the Data Base SAFE but also for other data base.
    The main substances of this paper are as follows:
  (1) The purpose of development of the Information Retrieval Support System was clearly identified.
  (2) Description and reconsideration about various functions in the Data Base SAFE were summarized.
  (3) Actual frequency about use of various commands supplied in the Data Base SAFE was reviewed (Table 1)
  (4) Selection of retrieval commands, relational operators and logical operators most frequently used in practical retrieval operation were investigated (Table 2).
  (5) The function structure in the Information Retrieval Support System was determined (Fig.1).
  (6) Details of functions and sub-functions in the Information Retrieval Support System were described.
  (7) Practical utilization of the Information Retrieval Support System for information retrieval operation concerning labour accident were demonstrated (Fig.2).
  (8) Practical use of support information which was furnished on management of the Data Base SAFE by means of utilities (sub-functions) suplied in the Information Retrieval Support System were examined (Table 3, Table 4).

Thermal Stability of Sodium Azide

RR-88-5

Yasuhiro FUJIMOTO, Takayuki ANDO and Shigeru MORISAKI

: Though chemical accidents have occured over and over, some part of these accidents are based on unstable chemical substances. These chemicals easily decomose or igniteby heats or mechanical shocks under an atmosphere of not so much high temperature.
    In this paper, sodium azide was chosen as an example of unstable chemical substances. Sodium azide is used in the preparation of hydrazoic acid, lead azide, pure sodium or as propellant for inflating automotive safety bags, etc. It is no doubt that sodium azide is more stable than other azides, however, the danger not been assessed satisfactorily.
    Thermal decomposition of sodium azide is shown as follows.
          2NaN₃ ⇒ 2Na + 3N₂                           (1)
    It is known that the decomposition takes place at 300 °C. In addition, the heat of decomposition has been reported. However, these results were obtained about half a century ago and the other thermal analytical results, in particular under an adiabatic condition, have seldom been reported ever since.
    And sodium azide reacts chemically active organic halides as follows.
          RCl + NaN₃ ⇒ RN₃ + NaCl                     (2)
    But reactions between sodium azide and organic halide polymers have not been reported yet. Hence, this paper describes the results of the following two subjects about thermal stability of sodium azide.
  1. Thermal decomposition of sodium azide
  (1) Decomposition temperature and heat of decomposition
    The experiments were carried out by DSC in air and argon atmosphere. The decomposition temperatures were about 430 °C in air and about 400 °C in argon atmosphere. The heats of decomposition were about 90 kcal/mol in air and about 10 kcal/mol in argon. Obviously, the mechanism of thermal decomposition in air is different from that in argon. This is because metal sodium, which is produced through thermal decompositon of sodium azide, is oxidized by oxygen in air.
  (2) Weight decrease by thermal decomposition
    The experiments were conducted with TG-DTA in the above two different atmospheres. In air, the weight goes back considerably after a decrease in the weight by the decomposition of sodium azide to pure sodium and nitrogen. But in argon, such a phenomenon was not observed.
    These results show that the sodium is certainly oxidized in air after decomposing of sodium azide. The following reaction scheme is estimated according to the above results.
          4NaN₃ + O₂ ⇒ 2Na₂O + 6N₂                    (3)
  (3) Self-heat rate and pressure under an adiabatic condition
    The experiment under an adiabatic condition was carried by ARC only in an inert gas. Beginning of self-heat of sodium azide was observed at around 335 °C, and the following decomposition was too fast to follow the increase in the self-heat rate with ARC.
  2. Reactions between sodium azide and organic halide polymers
  (1) Measurements under programmed temperature
    DSC was also used for the experiments in an atmosphere of air. The decomposition temperatures of any organic polymers, such as polyethylene, polypropylene, polyvinylchloride and polytetrafluoroethylene, were not changed in the presence of sodium azide, and no DSC-peaks due to exothermic decomposition were observed.
  (2) Measurements under fixed temperatures
    The experiments were carried out by DSC for providing another evidence in confirmation of no reactions between organic halide polymers and sodium azide. No exotherms were observed at the constant temperature of 300 °C for 2 hrs which might arize by the reactions between the polymers and the azide. The above results show surely that sodium azide may be thermally stable with organic polymers at relatively high temperature.

Experimental Study on the Methods of Explosion Venting (4th Report) --On the Behaviours of Flying-off Type Explosion Relief--

RR-88-6

Hidenori MATSUI, Takayuki ANDO, Yasuhiro FUJIMOTO and Shigeru MORISAKI

: Explosion venting is one of the useful method of explosion protection, which is applied to various industrial equipment of light structure such as dryers and dust collectors processing flammable gases, vapours or dusts. There are two prototypes of relief vent; one is the rupture diaphragm type and the other is called here as flying-off type. In the latter type, a vent cover flies off when it is subjected to an explosion pressure generated in a vessel to be protected, so that the combustion products are expelled through a vent opening. Characteristic features of the rupture diaphragm type vent were reported in the previous paper¹⁾. For the relief vent of flying-off type, an empirical equation has been proposed by Simmonds and Cubbage²⁾ through tests with cubical drying ovens of satisfactorily large internal volume (0.2--14m³ ). It seems, however, that the weights of vent cover they tested were rather small (0.15--3.4 g/cm² ) and that the vent openings were relatively large (K = 1--3), both from a viewpoint of practical use.
    The present report describes the effect on vented explosion pressure of vessel size, vent ratioand the weight of vent cover through gaseous explosion tests in smaller vessels with a widerange of vent opening covered by heavier vent covers. The pressure range of vented explosion, up to 1 kgf/cm², is far higher in this paper than that in the cited paper.
    Three cylindrical vessels with internal volumes of 0.92, 6.8 and 94 litres are used for tests,the ratio of the height to the internal diameter of each vessel being unity. The circular ventopening on the top of vessel is covered with a disc-shaped weight which simulates a vent cover (Fig.1). The diameter of bottom surface of the weight is just slightly larger than that of thevent opening, so that the weight of vent cover per unit area is assumed equal to W /S , where W is a weight of vent cover and S is a vent area, respectively. A 4.5% propane-air mixture isignited at the center bottom of the vessel, and a peak pressure (P ) of a vented explosion isdetermined.
    Simmonds and Cubbage derived their equation on the assumption that P is proportional to W /S for a constant volume (V ) and constant K ; vent ratio K is defined as (D / d )², where D and d are diameters of a test vessel and a vent opening, respectively. Their assumption isapparently not valid, as shown in Fig.2, except for very small W /S , but P and W /S are mostsuitably related, for a wide range of W /S, by a linear relation on a logarithmic plotting. Thus,an eq. (1) stands for each vessel and for a given K ; the constants A and B ₂ are functions of vessel size as shown in Fig.7, while A is independent on K when a vent opening is not so small (see Fig.3).
      P = B ₂ ( K·W ) ^A                            (1)
    Introducing a factor K ^1/2 , P can be closely related to K·W by eq.(2) for a given vessel (Fig.5). Further introducing V ^2/3 as a factor of vessel size, P is expected to be predicted by eq.(3).
      P / K ^1/2 = B ₃ (K·W )^A                     (2)
      P ·V ^2/3 / K ^1/2 = B ₄ (K·W )^A               (3)
    The result shows a fairly good agreement, as seen in Fig.6, but is not so satisfactory as to predict vented explosion pressures in larger vessels than tested. The reason is attributed not to the process of derivation of equations, but to the dependency of power constant A on vessel size; that is, the venting behaviours of flying-off type relief vent is somewhat dependent on the vessel size. In other words, the same W /S and K will give a relatively higher vented pressure as the vessel size increases, especially for a larger range ofW /S , probably because of an increase of inertia of vent cover itself with an increase of vessel size. The pessimistic conclusion is that the prediction of vented pressure in a large vessel based on explosion tests in a smaller vessel is not reasonable. However, it is found, as shown in Figs. 8 and 9, that there exists a limiting value ofW /S or K·W, less than which a vented explosion pressure in a smaller vessel is higher than that in a larger vessel. Such a limiting value of K·W will be of use in designing relief vents, and then is related to the reference volume V ₁ and R ; V ₁ is a volume of the vessel in which explosion venting tests are made, and R is a ratio V /V₁, where V is a volume of a larger vessel to which the result of test in a reference vessel is to be applied. On the assumption that the equation derived in the present work could be extrapolated to a vessel volume of 1 m³, a relation is obtained which enables to estimate the effect of vessel size on limiting K ·W (Figs.10 and 11). As is seen from these figures, limiting K ·W is large enough for practical use, but decreases rapidly with an increase of R .
    Discussions are also made on the comparison between empirical equations by Simmonds and Cubbage, eq. (9), and by the present work, eq. (10), where a , b , A and B are experimental constants.
      P·V ^1/3 = a ·K (W /S ) + b                        (9)
      P·V ^2/3 / K ^1/2 = B (S ·K )^A ·(W /S )^A          (10)
   Assuming A =1, then eq. (10) leads to eq. (12), and eq. (11) is derived from eq. (9) for a given vessel volume; a ₁, b ₁ and B ₁ are constant values.
      P = a ₁ ·K (W /S )+ b ₁                          (11)
      P = B ₁ ·K ^1/2 (W /S )                          (12)
    Taking into consideration the low value of b ₁, eqs. (11) and (12) have substantially the same meaning especially when the value of K is nearly unity. This fact suggests that the power constant A increases with an increase of vessel size but that it would not be so larger than unity. If this could be proved true, the design of relief vents for industrial equipments would become more easier by the method described in this paper.
    The effects of the weight of vent cover and vent ratio on venting behaviours have been explained through the present work, even though the effect of vessel size is not yet certain.
Experiments in larger vessels than tested here with vent covers heavier than tested by Simmonds and Cubbage will clarify whether V ^1/3 or V ^2/3 is a more suitable factor of vessel size.

Influence of Turbulence on Dust Explosion in a Closed Vessel

RR-88-7

Toei MATSUDA

: In recent years there has been a tendency to use new kinds of combustible dusts as materials in higher advancement of technology, without full investigation of their explosion hazard properties. Testing of dust explosibility is needed for safety in processing plants of combustible dusts. Turbulent flow motion is often inherent in a dust-air mixture in closed explosion testing bombs. The explosion data are imfluenced with turbulence intensity.
    Experiments have been carried out to assess the influence of turbulence on the explosionpressures and maximum rate of pressure rises in a 419-L closed near-spherical vessel. Dust was dispersed through two perforated semi-circular tubes with air blast. The tubes were connectedto dust chamber then via a solenoid valve to an 888-L compressed air reservoir charged to 10 bar. Cornstarch and two other chemical dusts were used as the fuels. Ignition source was achemical ignitor whose energy was easimated to be about 5 kJ. After placing the dust in the chamber, the outlet tube of the dust chamber was sealed with a 0.05 mm Al-foil and then the vessel was evacuated to the pressure corresponding to the air discharge time. The air discharge brings the pressure back to the 0-bar(gauge) in the explosion vessel.
    The variables in forming dust clouds and ignition are the air discharge time for dust dispersion and the delay time to ignition from closing the discharge valve. The time delay between discharge and ignition controls the turbulence level at the time of ignition. Figs. 3 and 4 show the optical transmission traces with pressure-time curves, illustrating a relatively good uniformity after a short dispersion pulse. Fig.5 shows the variation of the maximum rate of pressure rise with the time delay between discharge and ignition for comstarch-air mixtures. The influence of turbulence intensity is found in that the maximum rate of pressure rise decreases sharply with rapid decay of dispersion turbulence. However, the decay of the explosion pressure with increasing time delay is much slower for the dust concentrations between 500-600 g/m³ (Fig.6), indicating good distribution of the dust for the shorter time delay.
    To assess the relative hazards of a dust-air mixture, K_st value has been accepted in various countries. The K_st value is the maximum rate of pressure rise scaled with test volume. The standard method to determine K_st values in a 1 m³ has been established in ISO-6184²⁾. To have the same K_st values as those with the ISO, a time delay of 70 ms is needed in our vessel with the ignitor currently used. Further experiments with some other dusts were continued to match theresults for the standard, on search of turbulence level of the mixtures. The present data are compared in Table 2 with those reported by others, showing a fairly good agreement of the K_st between them irrespective of their different detailed structures of turbulence.
    These data for the cornstarch dust from the present experiments are also compared with those of others (Figs. 9 and 10), although the particle properties of the each cornstarch would be greately different. The present data for the explosion pressures show quite good agreement with those of Bond et al.³⁾ in 333-L sphere vessel, but the K_st value is the highest for the present data among the values from different investigators. The comparison suggests that the K_st value does reflect turbulence level of the mixture even in transient flow when the dust is in full dispersion. Although it will be possible to produce reproducible K_st values for a given dust in different methods, it is not clear whether the turbulence level in determining the K_st value of the mixture is pertinent to evaluate the explosion hazard of combustible dusts for industry.

Consideration of Prevention Method against Electric Shock Underwater --Shield Effect of Fault Current by GroundedMesh Set Underwater--

RR-88-8

Tatsuo MOTOYAMA and Eiki YAMANO

: With a widespread of an ocean development, various electric apppliances have widely been used in sea and river, but workers such as divers are exposed to electric shock hazards caused by fault current from the electric appliances at work on the spot underwater. A few fundermental studies have been carried out to prevent the electric shock hazards underwater.
    One of them is to shield the fault current using a grounded mesh set underwater.
    It has already become clear that the shield with grounded mesh is effective to reduce the fault current and that its effect depends on properties of mesh, electrical conductivity of water and objects existing in the vicinity of the mesh.
    Furthermore, the shield effect of fault current underwater is generally known to depend on an inter facial impedance between the grounded mesh and water, especially in a case that conductivity of water is high. However, an influence of the inter facial impedance on the shield effect has hardly been investigated since it is imposible to be deduced theoretically and must be investigated experimentally according to the combination of the mesh and water.
    From the background described above, the purpose of this study is to investigate the shield effect of the grounded mesh, including the inter facial impedance, for reducing the fault current underwater. The shield effect of square meshes made of a few kinds of material with various pitch and diameter was examined experimentally using a typical experimental setup, a water tank simulating the fault current underwater, which consists of twe electrodes and grounded mesh set in water of various condutivities. On the other hand, the inter facial impedances between a few kinds of meshes and water also measured, and thier influence on the shield effect were considered from an electrical circuit theory.
    The results obtained from the above experiments and considerations are summarized as follows:
  (1) An inter facial impedance between the mesh and water depends primarily on a kind of materials, contact area with water and conductivity of water. For example, in the contact with water of conductivity 5.9 S/m and temperature 20 °C, a value of the inter facial impedance of metal plate per 1 cm² was 9Ω in steel, 18 Ω in copper and 56 Ω in stainless steel, respectively.
  (2) An inter facial impedance affects directly the shield effect of the fault current underwater since it increases in inverse proportion to the contact area of the mesh with water and the contact area of mesh with water is small.
  (3) The shield effect of the grounded mesh depends on the shape and material of mesh, conductivity of water and inter facial impedance. For example, evaluating the shield effect as ratio decrease fraction of the fault current underwater suppressed with mesh to that unsuppressed without mesh, the shield effect of the grounded mesh, pitch 20 cm and diameter 2mm, hardly depends on the material of mesh and was approximately 80 % in water of conductivity 0.1 S/m, but it depends on the materil of mesh due to the inter facial impedance in water of conductivity 5.9 S/m. According to experiments the shield effect was 74 % in steel, 67 % in copper and 46 % in stainless steel mesh, respectively.
  (4) The grounded mesh is useful to relaxate the electrical shock underwater, but not perfect to prevent it. Consequently, in addition to the grounded mesh, other methods must be applied in order to prevent the electric shock hazards underwater.

JNIOSH