JNIOSH

Abstract of Special Research Report (RR-85)

National Institute of Occupational Safety and Health, Japan

Explosion Characteristics of Mixtures with Air of Propane, Ethylene or Hydrogen --With Special Reference to Testing Method for Flameproof Enclosure of Electrical Apparatus--

RR-85-1
Toshihiro HAYASHI, Ryuji TANAKA and Katsuhiro SAKANUSHI

: Explosion characteristics of mixtures with air of propane, ethylene or hydrogen are studied in three cylindrical vessels of the same length/diameter ratio. The flammable mixtures are ignited at one end of each vessel, and the maximum explosion pressure attained (P m), the time to attain the maximum pressure (t m) and the average rate of pressure rise (P m/t m) are determined for a wide range of flammable content under atmospheric pressure ; further tests are carried out under elevated pressure conditions for hydrogen-air mixtures only.
    The general tendencies of P m, t m and P m/t m in relation to flammable gas content are those as predicted, although their numerical values have been scarcely reported except for maximum pressures for nearly stoichiometric mixtures. Pressure rise time (t m) can be related to the size of vessel, and the present result shows that the equation proposed by Zabetakis for the prediction of t m may not be applied to larger vessels, even when stoichiometric mixtures of saturated hydrocarbon are ignited at the center of spherical vessel. The prediction of t m for such gases of higher burning velocity as ethylene or hydrogen may be often desirable in practical situations, and thus the relation between t m and the distance from ignition source to vessel wall is formulated for several mixture compositions.
    Effect of initial pressure (P i) on P m is determined for the hydrogen mixtures, and P m is always proportional to P i irrespective of hydrogen content. Some statistical treatments lead to a conclusion that the prediction of P m is possible for a wide range of hydrogen content and for the variation of initial pressure, based on a few tests made with some selected compositions under atmospheric and elevated pressures. P m/t m proves to be proportional to P m, thus to P i, which implies that the dependency of t m on P i can be considered negligible.
    The results of these discussions are applied to compare the characteristics of flammable mixtures specified in the IEC standard to be used in the tests of a flameproof enclosure of electrical apparatus.

Experimental Study on the Methods of Explosion Venting (3rd Report) --Practical Method of Design for Ruputure-Diaphraghm Type Explosion Reliefs--

RR-85-2
Toshihiro HAYASHI

: Explosion venting, or explosion relief, is one of the useful methods of explosion protection, which is applied to various industrial equipments of light structure such as dryers and dust collectors in which flammable gases, vapours or dusts are handled. Even though the effectiveness of its use is generally understood, the present status is far from the practical use in this country by lack of any comprehensive design guide.
    Although there have been proposed several experimental formulae for the prediction of the vented explosion pressure, or the maximum explosion pressure generated in a vented vessel, most workers have used their own test vessels or equipments with arbitrary dimensions and shapes, so that none of them could have discussed, with theoretical background, the effect of the size of a vessel on the vented explosion pressure.
    This report describes a reasonable method of calculating a vented explosion pressure in a larger vessel based on explosion tests in a rather small vessel. Experiments were carried, with 4.5% propane-air mixture, in four cylindrical vessels with different internal volume, each of them having a circular vent of various diameter covered by a polyvinylchloride diaphragm (PVC sheet). From the results of those experiments with smaller three vessels, a simple relation was derived between a vented explosion pressure (P), a vent ratio (d /D ) and a internal volume of a test vessel (V ). Such a relation was applied to predict vented explosion pressures in the largest vessel used to find a good agreement between observed pressures and calculated ones. Theoretical explanations were also satisfactory.
    Four vessels used in explosion tests are shown in Fig.1 and Table 1. Diameters of flange opening (d ) were arbitrarily selected in order to obtain almost the same diametral vent ratio (d /D ) for each vessel ; additional vent ratios were tested with No.1 vessel. In every test, a PVC sheet punched with a central hole of 1.5 mm diameter, and an open flange under test were fastened previously with bolts to the flange of the vessel, and then the blank flange was bolted to complete the test vessel. After evacuating the whole volume, the premixed test gas was fed to a pressure slightly higher than that of atmosphere. The blank flange was removed and the hole in PVC sheet was closed with a piece of cellophane tape. Then the mixture was ignited at the far end of the vessel from the vent, and the pressure development was recorded. In the present work, the ignition site was so selected that only one pressure peak was observed as shown in Fig.2. The fact that the pressure-time record showed no apparent pressure deflection just before attaining the maximum implied the equality of vented explosion pressure to bursting pressure of PVC sheet.
    The relation between d /D and P is shown in Fig.3, from which is seen the linear proportionality, on logarithmic graph, between them for any one of three vessels. Then the following equations can be written :
      P = B/(d /D )A = B (D /d )A                 (1)
        = BK A/2                            (2)
where A and B are positive constants, the latter being the pressure for d /D = 1, and K is the ratio often used in the study of explosion venting and is defined by K = (D /d )2 .
    If the gradients of lines drawn in Fig.3 are assumed to be equal, the constant A in above equations is a fixed value irrespective of vessel size. Thus, providing that the constant B is related to vessel size, the vented explosion pressure can be formulated as a function of the vent ratio and the vessel size. Fig.4 shows a logarithmic plot of the dependency of P on vessel volume for some vent ratios. The following relation is derived, where E and F are positive constants :
      P = E /V F                      (3)
    The numerical value of F calculated by least squares method for four lines lies between 0.331 and 0.335. This result means undoubtedly the dependency of P on the reciprocal of the cubic root of V. In other words, with vessels of L /D = 1, the ratio of the vented pressure for one vessel to that for another vessel is equal to a reciprocal ratio of L (or D ) of these two vessels for any d /D tested.
    These discussions lead to Fig.5, in which the relation between d /D and converted pressure P' is plotted on a logarithmic graph. P' is calculated by the following equation :
      P' = P (Vn /V1)1/3               (4)
where P is a vented explosion pressure observed in a vessel of internal volume Vn, n being the number of vessel shown in Table 1, and V1 is the internal volume of No.1 vessel, which is used arbitrarily as a reference volume in this study.
    As shown in Fig.5, equation (4) explains well the effect of vessel size on vented explosion pressure. Reminding that P and d /D have been related by equation (1) and that Vn /V1 is a constant value for a given vessel, the following relation is derived :
      P' = P (Vn /V1)1/3 = b (d /D) a           (5)
where a and b are positive constants inherent to both explosive gas mixture and vent cover material. Numerical values of a and b for this study are 1.22 and 0.98, respectively. From the viewpoint of practical use, Vn is a internal volume of an equipment to be vented, and V1 is a volume of test vessel in which explosion tests are carried to obtain above two constants. Equation (5) is then applied to No.4 vessel, internal volume of which being 100 times larger than that of No.1 vessel, and the result is shown in Fig.6. The difference between calculated pressures and observed ones can be explained by the expansion of vent cover itself before bursting when subjected to an explosion pressure.
    The validity of experimental formula (5) is also proved theoretically on generally accepted assumptions that an explosion pressure develops as a function of t 3 (t being the time after ignition) and that the time required to attain a maximum pressure in a closed vessel is directly proportional to a cubic root of the vessel volume.
    In the latter half of this report, discussions are made on the prediction of vented explosion pressure by means of pressurization test other than explosion test. As experimentally shown, the vented explosion pressure can be assumed equal to the bursting pressure of a vent cover with the same vent ratio. Therefore, if a pressure other than by an explosion can be applied with a same rate of pressure rise as that of an explosion, the vent cover is expected to show the same behaviour as if it was subjected to the explosion. Apparatus to determine bursting pressures of PVC sheet is schematically shown in Fig.7. No.1 test vessel with decreased length was used to obtain the relation between bursting pressure (P") and average rate of pressure rise (Δ) by nitrogen gas injection, the result being shown in Fig.8. This relation is then extrapolated to the average rate of pressure rise observed for an explosion generated in No.1 vessel with full length and closed entirely.
    Fig.9 shows a comparison and good agreement between bursting pressures subjected to dynamic pressure of pressure rise rate of 200 kgf/cm2 /s and vented explosion pressures observed in explosion test using No.1 vessel. Such a result means that the constants in equation (5) can be obtained through pressurization test mentioned above. This method of predicting vented explosion pressure can be also applied to full-scale equipments directly, even though are necessary some experimental techniques and informations concerning the characteristics of explosions to be occurred in those equipments.

Safety Assessment of Automated Production Systems Using Microelectoronics (2nd Report) --The Comprehensive Logic Models for the Analysis of Accidents Caused by Robots (Part 1)--

RR-85-3
Yoshinobu SATO

: In spite of the fact that some workers were inadvertently killed or injured by industrial robots, we are obliged to use them, because we have yet to overcome such risks and increase productivity. Nevertheless, the robot will be developed and utilized not only for the industrial sectors, but also for our daily life, and in this context the safety of robots will become more important with greater social interest in the near future.
    The purpose of this work is to improve the safety of robot systems, ranging from their design to use.
    In the previous paper, a general technique to identify the hazards to be created in human-robot systems was discussed. The next step of the safety assessment is to analyze the logic of hazard causation mechanisms.
In general, such analysis is made to given practical systems. On the other hand, the function to check the oversight in any implemented analysis is very important. Such function will be given by investigating, so to speak, comprehensive logic models made by assuming all cases of the hazard causation mechanisms.
    Targets in this paper are two typical hazards, i.e., "Struck by robot body" and "Struck by robot arms".
The comprehensive logic models which are used as such general-purpose fault trees are established with their hazards as top events (Fig.1 - 3). The comprehensive logic models resemble the MORT (Management Oversight and Risk Tree), and an individual fault tree can be obtained as a subset of the models, given an actual specification of the human-robot system.
    Further, the statistical independence among the basic events of the models is estimated, and the author concludes that the statistical independence is not expected under the general condition, but that three factors which dominate the interdependence among the basic events under that condition are extracted, and in addition, the assumption of statistical independence will be approximately satisfied in each fragmentary system phase defined by combining the factor-modes that are gained by dividing each factor into several modes (Fig.4).
    Finally, minimal cuts of the comprehensive logic models in some important system phases are then extracted to obtain the main system failure modes (Table 2).

Critical lgnition Temperatures of Wood Sawdust Layers

RR-85-4
Takashi KOTOYORI

: There exists many research reports on various aspects of pyrolysis and combustion phenomena of wood or woody materials. Instances, however, have been only a few so far where the critical ignition temperatures of bulky stacks of wood sawdusts were measured actually. As for that point, the works by Akita1) or by Gross et al.2), and lately by Anthony et al.3), by John4) and by Schliemann5) are quite precious.
    In the present work, critical ignition temperatures of fifteen species of wood sawdust layer, assuming each to be stacked in hot surroundings in air at normal pressure, were estimated with an adiabatic self-ignition testing apparatus (called SIT) following a procedure. Both apparatus and procedure have been developed at RIIS.
    A detailed description of the apparatus has been presented elsewhere9). So the procedure to estimate the critical ignition temperature of any self-heating substance and the results obtained when the procedure was applied to wood sawdust layers are mainly described in this report.
    An equation [eqn. (2)] holding for the self-heating process under an adiabatic zeroth-order assumption is first shown. Then a relation [eqn. (9)] between Frank-Kamenetskii's critical condition for thermal ignition and equation (2) is derived. If we use eqn. (9), we may dispense with direct knowledge of the individual values of molar heat of reaction, apparent activation energy, frequency factor, thermal conductivity or specific heat of the substance, so far as the calculation of critical ignition temperature is in question. The only data required to estimate the temperature is the thermal diffusivity associated with the material, besides data obtained with SIT.
    Wood species tested are Telaling (a sort of the Southeast Asian arbor), Meranti (ditto), Sawara cedar, Japanese cedar, Japanese red pine, Japanese cypress, Zelkova, Paulownia, Western red cedar, Douglas fir, Port Orford cedar, Alaska yellow cedar, Western hemlock, Sitka spruce and Eli ayanskya (a Siberian arbor).
    Wood chips are pulverized on an ultra centrifugal mill, and size-graded to 35 - 60 mesh by sieving.
    Measuring conditions are as follows: sample amount, 300 mg; sample container, silica cell (ca. 2 ml) ; starting temperature, 150 - 178 °C ; the SIT apparatus is left for 270 minutes at the starting temperature in nitrogen atmosphere, to remove all of the moisture from wood sawdust and to establish thermal equilibrium. The other measuring procedures have been described elsewhere9).
    Main conclusions are as follows :
  1) The oxidative heating process of wood sawdusts proceeds, showing an almost linear temperature rise rate which varies with the starting temperature, at temperature ranges up to about 180 °C under adiabatic conditions.
  2) The relative liability of wood species to heat oxidatively is scarcely recognized by thermal analysis such as TG-DTA, meanwhile it can be clearly distinguished from one species to another under adiabatic conditions.
  3) Assuming wood sawdust to be stacked in a form of infinite layer at a thickness of 60.96 cm (2 feet). the critical ignition temperature is estimated to range from 118 for Zelkova to 142 °C for Sitka spruce. Zelkova and Western red cedar are relatively easy to ignite, meanwhile Sitka spruce and Western hemlock are relatively hard to ignite. Paulownia may also belong under a category that is hard to ignite. Douglas fir and Port Orford cedar are medium in ignitability.
  4) The critical ignition temperatures of wood sawdust layers estimated by calculation, following the procedure developed at the RIIS, agreed in the result reasonably with a few real data observed on wood sawdust stacks of similar sizes. So we can say that the procedure is applicable to estimate relatively and quantitatively the ignitability of any stacks of wood sawdust, at least.

Safety Assessment of Automated production Systems Using Microelectronics(3rd Report) --Quantification of the Priority-AND-Failure Logic Consisting of Repairable Input Events--

RR-85-5
Yoshinobu SATO

: Any Fault Tree can be expressed as the equivalent of the OR-conjunction of minimal cut-AND-structures. Therefore, if every element of the minimal cuts is quantified, the Fault Tree can also be quantified.
    One of the most fundamental methodologies for the quantification of minimal cuts is the Kinetic Tree Theory (KITT) which was proposed by Vesely, W. E.
    The KITT yields the existence probability and the expected number of occurrences of the output event of a minimal cut, when the input events are assumed as follows :
  a) The input events are statistically independent,
  b) The input events have exponential failure and repair distributions, each with known failure and repair rates,
  c) The input events are not generated at the time t=0,
  d) The output events are generated only when all the input events are in occurrence, and
  e) The occurrence of output events are independent of the sequence of input events.
    There are, however, cases for which the sequence of input events essentially dominates the occurrence of an output event in actual system, especially in man-machine systems. For example, let's examine the cut which is composed of the following input events :
  Event A 1 ; fault of sensors to detect a human who is entering dangerous zones,
  Event A 2 ; signals for operation of machine,
  Event A 3 ; entrance of a human into dangerous zones.
    It is natural to suppose that an interlock device which has detected the entrance of a human into the dangerous zones and, as the result, has stopped the machine will be kept working, even when the sensor may cause a fault afterwards, until the human goes out of the zone and pushes the reset button to start the machine again. In this case, the event requires to occur prior to the event A3 in order to produce the output event.
    It has been maintained to use fundamentally a Markov model for the quantitative analysis of such consecutive failure logic. But, using the Markov model, no solution has been obtained for the arbitrary number of input events, and it is a little troublesome to get the expected number of occurrences of output events. However, for safety or risk assessment, the expected number of occurrences of output events has often much more importance than the existence probability.
    In this paper, the exact method of calculating the existence probability and the expected number of occurrences of the output event for such consecutive failure logic are given as Theorems 1 and 2, under the condition that the input events have the same characteristics as the KITT except the assumption e) above. Moreover, both the approximate solutions at the unsteady state under the condition that the repair rates of inputs are much greater than the failure rates and the exact solutions at the steady state are given for the arbitrary number of input events in formulae (24),(25),(32),(33), and (20),(21),(30),(31), respectively.
    It is concluded that this methodology gives much more simple and easier algorithms than the Markov model, especially for the calculation of the expected number of occurrences of the output event.

On the Stress of Bolts of the Slewing Circle of a Truck Crane

RR-85-6
Yutaka MAEDA

: The slewing circle is a rotating mechanism to make a crane slew. Most of slewing circles are made of large-sized ball bearings, and they are installed by several pieces of bolts onto revolving super structures and also onto base carriers.
    The purpose of this study is to clarify the stress distribution of each of the bolts in a slewing circle experimentally. The date will be of use to establish the method to calculate the stress about them.
    For this, measurements of the stresses of the bolts of a truck crane of 11 tonf (108 kN).capacity were carried out, and the stress values obtained from the experiment were applied to a model that each of them was divided into such six stress elements as i) initial tightening stress, ii) stress by thrust (vertical) load, iii) stress by a component of moment that would act to overturn the crane to right- or left-side, iv) stress by a component of moment that would act to overturn the crane to forward or backward, v) stress by both of the magnitude of the thrust load and relative position of the bolt to revolving super structure, and vi) stress by both of the magnitude of the moment and relative position.
    Measurements were made for the amount of 228 cases, with a combination of varieties of weight, slewing angle, and slewing radius. Measured data were calculated with a digital computer by the least square method, to get factors to explain the stress distribution.
    The conclusions in this report are: i) stress concentration was recognized to some specific bolts which were near the web member in the base carrier, and ii) revolving super structure's position has some influence to the stresses of bolts on the base carrier. From these, it was assumed that the surface of the slewing circles is not kept up a flat plane while forces act on a system by bolts and slewing circle.

Pre-conditional Function of Sensors for Securing Safety

RR-85-7
Souichi KUMEKAWA, Noboru SUGIMOTO and Kiyoshi FUKAYA

: The remarkable progress in reliability of Micro-electronics (ME) in recent years has resulted in sharp increase of the number of microcomputer-controlled machines, but their safety does not follow appropriately.
    Generally speaking, ME controllers consist of very unstable systems in the sense that their large power is controlled by much smaller electric signals. Therefore, there exists a large possibility that a slight signal disturbance might cause a serious effect, creating a hazardous condition. This means that ME has a large possibility of creating not only safe side errors, but hazardous ones.
    In ME controllers, Interlock Mechanism, in particular, should be built into the system so as to protect people from danger created by hazardous errors.
    On the other hand, a high-performance interlocking mechanism and sensor used in ME does not necessarily become safe. It should not be forgotten that if it is to be applied to a safety device to protect people, the Mechanism-itself including sensors must be made fail-safe.
    In this study, the authors debate pre-conditional function and mechanism of Fail-safe Interlock, and evaluation of safety of the system by the concept of "asymmetrical error rate".

Study on the Possibility of Safety Belt's Slipping out of Buckle

RR-85-8
Kiyoshi FUKAYA

: There was an accident that a worker fell down from a ladder with leaving his wearing safety belt on it because the belt slipped out of the buckle. Similar accidents might occur if safety belts do not work well in the case like this. In this context an experimental study was conducted to investigate the cause of the accident.
    Although the safety belt was later found to have some faults, they were not attributable to the cause of this accident. Because it could experimentally suspend a human dummy, it was suggested that the cause was related to the other facts such as a relative position of the buckle to the body, an angle between buckle and belt (draw angle), etc.
    First, draw angles were measured in the condition that a man actually wears the belt. It was found that the draw angle was greater in condition that the buckle is at the side of the body than in front of the body, and the value varied 27 degree to 35 degree.
    Following this, the slip resistances which are the force needed to slip off belt out of buckle in a fixed draw angle were measured for various draw speeds. When the draw angle was decreased stepwise, the resistance increased suddenly at a certain draw angle. These angles (limit of draw angle) were between 30 and 60 degrees according to the belt.
    In case of the same buckle as the type used in the accident case, the limit was not greater than that under actual wearing condition, but the slip resistance for the draw angle greater than the limit was 1 kgf at most.
This result seems to be applied to the folding type buckle.
    In conclusion the cause of the accident is presumed as follows : Something happened to touch the buckle, leading to the buckle in the put-up condition, i.e. its draw angle became greater than the limit. As the result, the victim fell down from the ladder while the buckle remained loose.
    There is a possibility of belt slip-off accident regarding folding type buckles. Therefore some measures, such as the use of a loose stopper, are recommended.

Analysis of Accidents in Tunnel Construction Work (4)

RR-85-9
Shigeo HANAYASU

: 1. Introduction
    Every year many casualties take place in the construction industry. In particular, labour accidents associated with construction work account for about one third of all occupational accidents and represent nearly a half of the number of death in all the industries. Among them, the tunnel construction sector has a higher accident frequency rate and a higher severity rate than those of other construction sectors.
    This paper presents statistical analysis of labour accidents occurred in a large construction project, with particular emphasis on tunneling work, in order to clarify the basic labour accident situation in construction site.
2. Data collection and monitoring
    The tunnels under the present investigation are those in a part of the New Sanyo bullet train construction project ranging from Hiroshima to Yamaguchi prefectures, which numbers 75 construction sites. The accident data were made available from the reports on industrial accident, which were submitted to the labour standard inspection offices concerned. In addition, information concerning construction site features was collected from the project records compiled by Japan National Railway.
    So far as the variables on construction site features are concerned, 13 kinds numerical variables (e.g. length of tunnel excavated, construction periods, index of rock grading of site, total hours consumed in excavation, etc.), including the number of accidents and 3 kinds of categorical data such as method of excavation, transportation system, and on type of entrances in each construction site, were obtained from the construction record.
    After compiling these data on all construction sites, numerical variables in each site were divided by the length of tunnel excavated in order to make these variables comparative between each site. In order to find the outliers in these normalized data 3rd moment (skewness) and 4th moment (kurtosis) of these variables were used. This detecting resulted in 6 construction sites had some discordant values with respect to the construction variables. Hence, for the purpose of assuring the homogeneity of data structure, variables within this 6 construction sites were excluded from the data set, so that the final number of construction sites for the analysis was to be 69.
3. Analysis and results
    Various multivariate analysis methods were applied for the study of the basic relationship between labour accidents and tunnel construction features on site. The main conclusions reached in this work are as follows :
  1) From the correlation analysis, there is a strong correlation between the total number of accidents and the total volume of construction materials consumed. However, the accident frequency rate (The number of accidents per 1 km) has no positive relation to the unit volume of the construction variables concerned. In contrast to the overall accident frequency rate, the fatal accident frequency rate depends on the geological characteristics of the strata on site and decreases in accordance with the increase of unit volume of construction variables.
  2) According to the principal component analysis, many selected numerical variables can be explained by such a few factors as of the size of construction work, the geological factors of the tunnel, and labour accident factors. Since accidents constitute a major component in PCA analysis, statistical evaluation, with reference to accidents, between construction sites can be conducted by making use of the accident frequency rate. The poisson distribution also could be employed for the statistical analysis of the number of occurrences of accidents in a site.
  3) A discriminant function was developed to distinguish with high accuracy whether or not work place may have fatal accidents during its construction periods. From the analysis, fatal accidents are more likely to occur in a work place where the geological conditions are relatively better, and the execution of the construction work was accomplished within a short or period of time. Interpretation of the variables employed in the proposed discriminant function were explained similar to the ones given in the correlation and the principal component analyses.
  4) Finally, multiple regression equation was proposed to predict the number of occurrences of accident in a construction site with a relatively high multiple correlation coefficient. Total volume of steel arches erected, method of excavation, and index of rock grade have played an important role for establishing the regression equations. These explanatory variables could also give the same implications of the accident situation as revealed in the preceding paragraphs.

Analysis on Processes of Labour Accidents --Case Study of the Joetsu-Shinkansen Construction Work--

RR-85-10
Yoshimi SUZUKI

: It is commonly recognized that there are many factors which contribute to the occurrence of labour accidents.
These factors constitute various combinations and concatenations in the processes of the occurrence of labour accidents. Therefore, in order to establish the effective safety measures, it is necessary to examine in detail the processes leading to labour accidents in individual case.
    For the purpose of examinating the process of the occurrence of labour accidents, 316 samples are selected ; viz 234 labour accidents occurred in the construction work of the Joetsu Shin-kansen (Joetsu New Rapid Trunk Line) and 82 labour accidents occurred in the cutting work sites of road-repairing and road-constructing (see Tables 1 - 5).
    The selected samples of labour accidents are examined by the technique of the C.T.A. (Causal Tree Analysis) method which is an a posteriori method of analysis of accidents and is regarded as useful approach to predicate clearly and facilely the process of occurrence of an accident, particularly concerning the relationships between various contributive factors to an accident.
    In this study, firstly, in order to describe the process graphically, leading to an accident, C.T.diagrams for each accident were drawn out (see Fig.1). Then the linkages of events in the C.T.diagrames were investigated with reference to the factors which might contribute to the occurrence of an accident. These contributive factors are identified and classified into five categories given below: I (human, individual), T (task), M (material), Ep (physical environment), and Es (social environment).
    In accordance with the definition of classification of these five contributive factors, frequencies of these factors in the process leading to labour accidents, and relationships between these five factors in that process leading to labour accident, are examined.
    As the results of the investigation, "the distribution of the frequency ratio of contributive factors to accident" are obtained (see Table 6, Fig.2), and "the influential transition matrices" which mean the situations of reciprocal actions between five contributive factors to accident are acquired (see Tables 7 - 11).
    By making use of these results, the differences of the processes leading to accidents are discussed for different scale of work, different types of accident, different types of structure constructed, and different supervisory constructors.
    The main conclusions obtained are as follows :
  (1) Comparing two construction works different in scale ; the Joetsu Shin-kansen construction and the cutting work sites of road-repairing and road-constructing; it was recognized that they do not have the same distribution of the frequency ratio of contributive factors to accident. Particularly the material factor (M) and the physical environment factor (Ep) are different with two types of work.
  (2) So far as the types of accident concerned, the distribution of the frequency ratio of contributive factors are also different each other. Especially the slope-failure accident in the cutting work sites of road-repairing and of road-constructing is considered to be one particular type of accident.
  (3) Concerning the influential transition matrix, there is no similarity between two construction works ; the Joetsu Shin-kansen Construction and the cutting work sites of road-repairing and road-constructing.
  (4) The similarity could not be found between influential transition matrices which were completed according to the types of accident.
  (5) Also, the similarity could not be verified between nine matrices which were completed according to supervisory contractors.
  (6) On the other hand, the difference is not seen between five matrices according to the types of structure constructed.
  (7) These conclusions mentioned above are supported numerically by the chi-square (χ2) test and the cluster analysis (see Table 12, Fig.3).

Durability of Plywood Scaffold Planks and Nethod of Strength Test

RR-85-11
Yoshimasa KAWAJIRI

: Plywood scaffold planks must have sufficient stiffness and strength to sufficiently endure load of workers and materials on them.
    Occasionally, accidents caused by the fracture of planks (most of them are long-used planks) occur, and workers are wounded or killed. A safety standard for used plywood scaffold planks was established by the Plywood Scaffold Plank Industry Association in 1979, and then constructional requirements of the Ministry of Labour for plywood scaffold planks was promulgated in 1981. However, data to support the standard were not necessarily enough.
    This paper deals with the following studies:
(1) Durability of plywood scaffold planks.
  Nondestructive tests and bending fracture tests on commercial and order-made plywood scaffold planks ( 9 - 16 plies, Apiton or Kapur) were carried out after 2 years' outdoor exposure.
(2) Relation between bending properties and moisture content.
  Bending fracture tests on commercial plywood scaffold planks in various soaking conditions were carried out.
(3) Relation between bending properties and span.
  Bending fracture tests on commercial plywood scaffold planks in various spans were carried out.
    Results of these studies are summarized as follows:
(1) Modules of rupture and elasticity in bending degrade gradually in proportion to the increase of exposure period or moisture content.
  The life of commercial plywood scaffold planks are about 3 - 4 years.
(2) Special order-made plywood planks are stronger than commercial plywood planks ; therefore, there is much room for further improvement in the durability of commercial planks.
(3) Correlation coefficient between modules of rupture and that of elasticity is 0.76, meaning that since their relation is fairly close, it is possible to evaluate approximately the strength of planks by the nondestructive test using flexure devices.
    In conclusion, the method of strength test in the above-mentioned safety standard has been proved to be appropriate.

Threshold for Electric Shocks under Water Causing Suffocation of Rabbits --Change of Threshold due tu Conductivity of Water and Body Arrangemsnts--

RR-85-12
Eiki YAMANO and Tatsuo MOTOYAMA

: Working in water such as the sea or a river and the use of electricity in such areas are increasing in recent years. These situations may lead to hazards of underwater electric shock for divers and other workers. This paper reports on an investigation of such hazards.
    An experiment was conducted with rabbits, using plexiglass water tank to find the threshold of suffocation for underwater electric shock. The rabbit was submerged in water and placed between two parallel plate electrodes arranged in the tank. Then, electric shocks were applied to the subject by passing the current through the water between the electrodes. Shock currents were 50 Hz a.c. and 10 s in duration. The temperature of water was at 30 °C throughout the tests.
    Salt water of 3.5% or tap water was used in the tests, and the body of the rabbit was so arranged that it was parallel with or perpendicular to the direction of current. The threshold, in this study, was expressed in 6 parameters and the changes of the threshold in different parameters due to the test conditions were compared with each other.
    The results of the investigation are summarized as follows ;
  (1) When the rabbit body was arranged in parallel with the direction of current in salt water, the mean threshold of suffocation was observed to be 22.1 V/m in underwater field intensity, or 127 A/m2 in current density, or 6.0 V in potential difference on the subject.
  (2) The thresholds for perpendicular body arrangement were greater than those for parallel by about twice in the field intensity and in the current density.
  (3) When the conductivity of water was decreased from 6.0 S/m to 27 mS/m, the threshold field intensity increased by 3 - 4 times, and the threshold current density decreased to a 70th - 60th.
  (4) Change of threshold in potential difference on the subject, depending upon conductivity of water and body arrangement, was observed to be less than the changes in other parameters.
  (5) The lower limit of threshold of suffocation for rabbits was estimated to be 10 V/m in the field intensity of 10 s shock duration.
    The upper limit of field intensity which allows a certain degree of respiration is less than 10 V/m and assumed to be about 5 V/m. (Free respiration may be sustained at 2.5 V/m which will allow the subject for voluntary movements.)

Degradation of Rubber Insulating Gloves used in High Voltage Electrical Working

RR-85-13
Kenji ICHIKAWA

: The previous reports1)2) described the results of degradation of electrical and mechanical properties of rubber insulating gloves for use in high voltage electrical working, when they were degraded artificially by the application of voltage stress or ultraviolet radiation. This report presents the insulation performance and mechanical strength of them, after their actual use in high voltage electrical working on site for a certain length of time. Actual tests were conducted on high voltage distribution lines in cooperation with an electrical contractor, because the frequency of use is large and the environment is most severe.
    Principal conclusions obtained are as follows :
  (1) Allowable life of rubber gloves is considerably shorter than expected because of their mechanical damages, that is, the average length of life is about 265 days, i.e. 9 months.
  (2) It seems that degradation of rubber gloves with the elapse of time is not so important as considered in the above result. There are however cases where a few gloves within one and a half years of use do not pass the voltage proof test of 10 kV. Therefore, it can be said that rubber insulating materials of gloves may contain degradation contents to the extent of some degrees.
  (3)It was found that there is a close relation between the period of use within one year and the service day, and also between the service day and the total hours of wearing.
  (4)The period of wearing per day is from 30 minutes to one hour at most, and about 75% of those periods are within about 2 hours.
  (5) Electrical and mechanical properties of degraded gloves scattered on a wide range, but they showed the degradation tendencies fairly well.
  (6) Is was clearly demonstrated that the longer the period of use was, the lower the breakdown voltage became.
  (7)Points which caused electrical breakdown are located on not only the trunk of glove, but also finger parts that are additionally covered by a protective glove for mechanical protection.

Considerations of Electrostatic Potential of Charged Insulator Measured by Use of Potentiometer of Induction Type

RR-85-14
Yasuyuki TABATA

: The electrostatic potentiometer of the induction type is often used to assess electrostatic risks caused by the charged material in industry. Although the measurements are carefully carried out, the measured potential, the indicated value of the potentiometer, fluctuates generally whenever the measurements are made. However, the reasons of fluctuations of the potential and the physical meanings obtained from the measurement have hardly been discussed.
    Investigations have therefore been made mathematically and experimentally to scrutinize electrostatic potentials of the charged insulator measured by the use of a potentiometer of the induction type. As the result, it has become clear that the measured electrostatic potential of a charged insulator is proportional to the total charges as one in the case of a conductor and that the dispersion of measurements occurs chiefly due to the potential distribution and fluctuation of the real charged area on the insulator even though the total charge is equal. The apparent capacitance of the insulator is also deduced mathematically and experimentally in this paper.

Other Publications

Kiyose District Map

Umezono 1-4-6, Kiyose,
Tokyo 204-0024 Japan
TEL: +81-42-491-4512
FAX: +81-42-491-7846

Noborito District Map

Nagao 6-21-1, Tama-Ku,
Kawasaki 214-8585 Japan
TEL: +81-44-865-6111
FAX: +81-44-865-6124