Abstract of Special Research Report (RR-27)

National Institute of Occupational Safety and Health, Japan

Analysis of Accidents in Tunnel Construction Works (3rd report)

RR-27-1

Ikuo MAE, Shigeo HANAYASU and Noriyuki HORII

: It is well known that accidents associated with tunnel construction have a very high accident frequency rate and also are of extreme severity. The authors have been conducting accident analysis in this field in these three years in order to provide a better understanding of the nature of recent accidents at tunnel construction works.
    In the present work, study was planned mainly in three parts : first, to determine the basic relationships between work accidents and various tunnel characteristics ; second, to throw light on the variations in workdays lost through accidents and on the relation between workdays lost and tunnel characteristics ; and thirdly, statistical and stochastic analyses of 2 kinds of accident rates (number of injuries per 1,000 m and number of injuries per month), assuming that these rates are random variables.
    The tunnels under investigation at this time and the available data are about the same as in the previous report, (i.e. New Sanyo Line railway tunnels constructed in Hiroshima and Yamaguchi prefectures : 86 in number, total length 158 km, total injuries 1931)
    The main points obtained in this study are as follows :
  (1) Relationships between work accidents and tunnel characteristics picked up in this study show positive correlation in all respects and also have a relatively high intercorrelation, except for mean rock grade. The number of accidents could not be related adequately to one characteristic. (Table 1, Fig.1, 2)
  (2) An analysis of variance on the 2 kinds of accident rates show that there are significant differences between excavation methods and excavated length. Hence, the accident rate increases in the following order : Upper half section advanced method, Bottom drift and upper half section cutting method, and Side drift and upper half section cutting method. Furthermore, the longer the tunnel excavated, the higher the accident rate indicated. (Table 2, 3)
  (3) Each of the cumulative accident curves show the convex curves. This reveals that accidents do not occur uniformly throughout tunnel length constructed and also that accidents are concentrated at the workplaces in which the accident rate is highly indicated. (Fig.3, 4, 5)
  (4) Workdays lost through accidents (not including fatal accidents) are influenced neither by excavation method nor tunnel length. (Fig.6, 7) On the other hand, the proportion of fatal accidents (fatal accidents/total accidents) changes greatly in relation to the length excavated. As reported in reference 7, the inverse relationship between accident rate and the number of days of absence is also recognized in this study. (Table 4, Fig.8)
  (5) If accidents take place at random, the frequency of occurrence in fixed intervals of length excavated or construction period have Poisson distribution. For the convenience of calculation, Normal distribution can be employed to test significant changes in the accident rate. (Fig.9, 10, 11) Using regression line, approximate statistical assessment can be conducted without calculating confidence intervals. (Fig.12) The difference of influence of tunnel characteristics on accidents between significant and non-significant workplaces was studied briefly. (Fig.13)

Construction and Performance of a Spontaneous Heating Apparatus

RR-27-2

Takashi KOTOYORI and Michio NAITO

: To study more precisely than before the spontaneous heating properties of thermally unstable substances, it was tried to make such a new adiabatic apparatus that the operation is easy and the accurate data can be obtained with good reproducibility and a wide applicability can be expected, applying a more completer adiabatic control technique than those of most apparatus in the past. Thus, the purpose of this note is to place on record the constructional and procedural details and the experimental performances of this apparatus.
    Main features of this apparatus are as follows.
  1) Adoption of the PID-SCR temperature control technique. Some instruments were also used together with this technique to realize the control as complete as possible, such as a pre-amplifier to amplify the ΔT value, a zero suppression circuit to compensate the stray thermal e.m.f. of the differential thermocouple (Fig.2), a cold junction based on the thermoelectric effect, a voltage shock absorber, some noise filters, etc. Ultimately, the adiabatic control of ±0.2 μV(±0.005 °C) could be achieved at the initial slow heating stages of the sample.
  2) Attempt to control adiabatically between the sample and the atmosphere. In other words, the establishment of the true physical adiabatic condition around the sample was intended, rather than the apparent stability of the ΔT base line as observed in such cases that the control is made between solid and solid or between solid and liquid.
3) Adoption of air bath. This is for the control to be able to follow easily even at the rapid Heating stages of the sample.
    To check the sensitivity of this apparatus, the effect of the rate of heat evolution on the rise rate of sample temperature was measured by evolving Joule heat of 150 to 800 μW in the cell at 60 °C. As the result, 55 μW was found as the lowest limit of the sensitivity of this apparatus (Fig.5).
    H.J. Pasman also made an adiabatic vessel under the similar conception as ours, using a Dewar as the cell¹⁰⁾. He reported he had succeeded in detecting a heat production of 10 mW / kg by using a large sample amount of 1 kg or so. However, the mentioned experiment was actually carried out under such a condition as brings about a heating rate more than 100 mW/kg. In the present case it is probably owing to the strict PID-SCR control that the sensitivity comparable to that of the large scale test could be obtained, in spite of the sample amount less than 1 g and the cell of only a tentative design. Therefore, it will be easy to get a more improved sensitivity on this apparatus, should a new cell structure be devised and should a larger amount of sample be used.
    Pure linolic acid/absorbent cotton mixture, in the form of uniformly scattered fiber assembly in the cell, was used throughout this study as a model of spontaneously heating stacks. The cotton acts as the substrate and was practically inactive to oxygen under the conditions of this study. Oxygen of 1 atm was used as the atmosphere. The standard experimental conditions are as follows : the sample amount, linolic acid, 50 μl, absorbent cotton, 50 mg ; the flow rate of oxygen, 2.3 ml/min ; the range of the pre-amplifier, ±50 μV ; the initial ambient temperature, 60 °C.
    Main experimental results are as follows.
  1) The reproducibility of the heating curve is very well. For example, the scatter was within ±1 min for a temperature rise of 8 °C during 70 min at 70 °C (Fig.7).
  2) The spontaneous heating process is, in the same way as experienced in the usual reaction; affected sensitively by factors such as the temperature ; the surface area, the loading density or the extent of pre-oxidation of the sample ; the coexistent substances ; as well as the range of adiabatic control.
  3) An excellent linear relationship was found to hold between In ΔT and 1/T, assuming the reaction to be of the zero-order at the initial stages, where T is the initial ambient temperature and Δt is the time taken to produce a given temperature rise of 100 μV (2.5 °C) from the initial temperature (Fig.9, Eq.4).
  4) Under the condition that the linolic acid/cotton ratio is kept constant, the rise rate of the sample temperature decreases gradually at successively lower loading densities of the sample (Fig.11). This fact means that, as the sample happens to be the fiber assembly, the thermocouple for the sample does not indicate the temperature of the assembly itself but shows that of the space in the cell which contains the distributed linolic acid/cotton mixture. In other words, the observed heating curve does not reflect the progressive state of oxidation reaction of the sample itself, but only tells of the accumulation process of heat of reaction evolved within the time, during which the sample has been exposed to oxygen. Thus, the initial minute temperature rise after a given elapsed time is proportional to the sample amount (Table 1), and the time taken to produce a given temperature rise of 50 μV (1.25 °C) is inversely proportional to the sample amount (Fig.12), assuming the zero-order reaction to take place at the initial stages, other conditions being equal.
    This apparatus showed to be very useful for studying the spontaneous heating behavior of a wide range of substances and the effects of temperature, coexistent substances and atmosphere on it, with relatively small sample amounts and with easy procedures. However, the modification of the design may naturally be made as desired, to suit special requirements and to get a more improved sensitivity.

Accident Prevension on Walkways and Stairways (2nd Report) --Experimental Study of Human Motion on Stairs--

RR-27-3

Hisao NAGATA and Kinichi KINOSHITA

: According to the first report about the statistical research on stair accidents, many stair accidents are occurring at every stairway, and rapid descent on stairs or wearing high heels is reported to cause many stair accidents.
    Human motion is itself the two leg motion which is dynamically unstable, therefore stair accidents are liable to occur at any stairs. Stair accidents are supposed to occur by one of these slipping, stumbling and miss-footing. It would be necessary to investigate these fundamental factors by the way of human motion study in order to decrease the inevitable stair accidents. In this report the following values accordingly were derived from human motion studies. Force values applied to the tread by the foot, and the foot angle just in contact with the tread were measured as to slipping. The short distance between the edge of the step and the tip of a shoe as to stumbling (defined as the G value). The foot contact length to shoe length as to miss-footing (defined as the Lf value).
    The experimental stairs which may be varied by changing the combination of tread and rise, and the force plate by which three dimensional force values applied to the floor can be measured, were planned and turned out. In this experiments four settings of combinations of tread and rise were used, and two females and four males were chosen as experimental variable conditions. Especially for females the experiments of different foot wears were carried out so as to find out the sharp differences between the high heel and the low heel. Stride time, contact time, force values, the Lf values, foot angles were obtained through the experiments.
    Judging from these results, safe walking is closely related to the dimensions of stairs and walking velocity. It would be best to step within the limited safe speed on the allotted treads. Subjects wearing high heels, in descent, can not rotate their ankles enough to soften the footing on the tread. Many stairs are chiefly planned to climb up with ease. But dimensions of safe stairs should be decided with due consideration to descent motions.

Test-Manufucture of Artificial Fingers (3rd Report) --Angle and grasping force control in the Artificial Fingers--

RR-27-4

Noboru SUGIMOTO, Taiji KONDO and Kiyoshi FUKAYA

: As an effective safe measure of any dangerous or detrimental works, remote controlled manipulators should be expected to be introduced in place of men. But any manipulators could not be said to have met our expectation. We can point out that any manipulators, which have been developed and are now used at any spot, have so small degrees of freedom that they cannot be used in effectiveness.
In our study, an artificial multi-finger system, which might be said unique, were developed as an extremity of manipulators. The system has three fingers which correspond to the thumb, the forefinger and the middle finger, so that the system has nine degrees of freedom.
In order to control functionally so many degree of freedom, Electro-Viscous-Fluids Clutches are developed and used in our study which are different from any existing control methods.
The control of angle and grasping of the fingers can be accomplished by regulating the electric voltage applied to the clutch. In the control system, further, to improve response and stability, feedback compensation (angle and force pickups), differential compensation and integration compensation are systematized into the closed system. In the result, it was confirmed that angle was controllable within 1% error and as for grasping force within 4 % error.

Effect of Microstructure on Fatigue Crack Growth Rate in a High Carbon Steel

RR-27-5

Yoshio KITSUNAI

: It is known that fatigue crack growth rate is related to the range of stress intensity, ��K, by an equation of the form
      dl / dn = C(ΔK)^m
    where C and m are constants.
    Values of the exponent 'm' have been found to lie mainly in the range 2 _˜ 3, but values as high as 10 are obtained, particularly in high strength steels with low fracture toughness. Variation of values of exponent 'm' may be associated with micromechanisms depending on microstructures of the material during fatigue crack growth. The fatigue fracture surface appearances of metals and alloys are important in determining fatigue crack growth mechanisms and post failure analysis. In this study, the influence of micromechanisms on fatigue crack propagation in a high carbon steel is investigated in conjunction with changes of metallurgical structures.
    The material used during this study is an equtectoid steel with prior austenite grain size of 32.6 μm. The composition is given in Table 1. The roughly shaped specimens were cut from the as-received plate with 7 mm thick and these specimens were heat treated to give different microstructures. The heat treatments and static properties are summarized in Table 2.
    A centrally slotted 4 mm thick sheet specimens were machined with the direction of loading parallel to the rolling direction. One side of specimen's surfaces is polished and lines scribed 0.5 mm apart to aid observation of the growing cracks.
    Fatigue tests were performed using a Vibrophore test machine at a frequency about 148 Hz. The crack length was measured using a traveling microscope with 50 magnification and stroboscopic illumination. The crack growth rates were determined graphically by taking the slopes of the crack growth curves at various crack length. The results were expressed in terms of stress intensity factor range, ΔK. Two stage chromium shadowed carbon replicas were obtained from the fatigue fracture surfaces, and examined using a transmission electron microscope to characterize the fracture morphology. A scanning electron microscope was partially used too.
Considerably higher growth rates are obtained for lamellae pearlitic structure annealed at 850 °C, as compared with mixed lamellae and spheroidized structure annealed at 700 °C. Presence of micro-cleavage in addition to striations is observed on the fatigue fracture surfaces of the both structures. In this case, area percentage of micro-cleavage facets in the lamellae pearlitic is correlated with K_max, and the percentage in the mixed structure is dependent on ΔK. These relationships may be useful for the failure analysis, when striations can not be observed. From above the result, the higher rates in the lamellae pearlitic are achieved by occurrence of micro-cleavage dependent on K_max during striation growth. As revealed by precision matching, no lamellae or spheroidized cementite is found on the facets of micro-cleavage. Therefore, micro-cleavage is not produced by fracturing between cementite and ferrite but occurred by separation of ferrite in the micro range.
    Changing of lamellae cementite spacings obtained by varying the cooling rates from 850 °C is little influenced on the fatigue crack growth rates with the ranges from ΔK = 40 to 150 kg mm^-3/2 . This result may be explained by the fact that fatigue crack growth in the lamellae cementite tends to propagate along the pearlite colonies, as shown in Fig.7. No systematic change in the fatigue crack growth rates above the value of ΔK = 400 kg mm^-3/2 is also observed with changes of spheroidized cementite spacings obtained by isothermal heating for 5 to 110 hours at 700 °C. The growth rates in the lamellae cementite, however, are considerably higher than those of spheroidized cementite on account of the different fatigue crack growth mechanisms. In the spheroidized structure, subgrain crackings connected with fine cementite particles appear at values of ΔK below 46 kg mm^-3/2 . The area percentage of the subgrain facets are increased with decreasing ΔK. The transition from structure sensitive to structure insensitive fatigue crack propagation in the spheroidized structure occurs when the ratio of cyclic plastic zone size to the subgrain size is approximately 10.
    The fatigue crack propagation at ΔK ranging from 50 to 150 kg mm^-3/2 in the materials tempered at 250 °C, 400 °C and 600 °C, respectively, is predominant in striation formation mechanism. In this case, the area percentage of non-striation (micro-cleavage, intergranular, and void coalescence) is less than 40 percent. Therefore, the growth rates of the materials tempered are a little influenced by the microstructures or strength of the materials.

JNIOSH