Effects of the range and frequency of vibrations on the momentary riding comfort evaluation of a railway vehicle

Transcription

1 Japanese Psychological Research 1998, Volume 40, No. 3, Effects of the range and frequency of vibrations on the momentary riding comfort evaluation of a railway vehicle HIROAKI SUZUKI Ergonomics Laboratory, Fundamental Research Division, Railway Technical Research Institute, Hikari-cho, Kokubunji, Tokyo , Japan Abstract: When trains pass level-crossings, turnouts, and rail joints, they are momentarily subjected to extreme vibrations. In railway engineering, evaluation of the riding comfort under such occasional vibrations is called the momentary riding comfort evaluation, as distinct from the long-term evaluation, which addresses the riding comfort of passengers for certain lengths of train operation. In order to identify the effective vibrational characteristics of the momentary evaluation, an experiment was performed with a riding comfort simulator. Ten adult subjects for each condition, 80 in total, participated in the experiment. The effects of differences in the range of stimuli, frequency of each stimulus, and scores on a rating scale of discomfort were studied. Differences in the range and frequency affected the evaluation such that subjects tended to make a relative judgment on discomfort. They made almost an absolute judgment when the rating scale was well defined, with a small number of categories. Key words: vibration, riding comfort, perceptual judgment, psychophysics, range-frequency effect. The riding comfort of railway vehicles is determined by vibration, noise, temperature, humidity, seat design and structure, and other factors, of which vibration has most extensively been studied in quantitative form. In railway engineering, riding comfort generally means feeling or evaluation related to vibration and acceleration that occur when trains run (Suzuki, 1998). In the research on riding comfort evaluation, one of the objectives is to clarify the correspondence between vibration (physical quantity) and human response. Therefore, a psychophysical approach is applicable. In contrast to the objects of classic psychophysics the vibration of railway vehicles, a complicated physical quantity, calls for clarification of how to express its characteristics. Different countries and regions have different gauges, ratios of curves and gradients and speed restrictions, and have to deal with different weights of passengers. As it is extremely difficult for this reason to establish a universally applicable international standard for the riding comfort of railway vehicles, different countries have different standards, to reflect their own conditions (Japanese Industrial Standards Committee, 1990). For example, in Japan, where the railway is responsible for a large share of passenger transport, 1 trains in urban zones are so crowded that many passengers are obliged to stand, and accordingly it is common practice in setting the allowable limits of vibration, 1 For example, in 1993, railways shared 35% of the total passenger transport in Japan in terms of distance traveled, while the corresponding figure was 6 8% in west European countries and about 1% in the US (Ministry of Transport, 1996) Japanese Psychological Association. Published by Blackwell Publishers Ltd, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

2 Factors affecting vibrational discomfort 157 Figure 1. An example of a vibration in a railway vehicle. This is a waveform of lateral vibrational acceleration measured at the floor surface of the car body. deceleration under braking, or excessive centrifugal acceleration to refer not to the seating posture but to a more stringent standing posture (Suzuki, 1997a). Vibrations of railway vehicles are a composite of various vibrations of different types, such as permanent vibrations (e.g., those caused by track irregularities), temporary vibrations (e.g., those caused by air-conditioning units), and occasional vibrations (e.g., those caused by level-crossings and turnouts). All these aspects should be considered in formulating a method of evaluating riding comfort. The present study, through basic experiments, examined the factors determining momentary discomfort under occasional vibrations. Riding comfort as a perceptual judgment Expression of vibration Acceleration is generally adopted as a preferred measure of quantifying the severity of human vibration exposures (International Organization for Standardization; ISO, 1997). Figure 1 shows an example of a waveform of vibrational acceleration in the lateral direction measured on a running train, with the horizontal and vertical axes representing the time and the lateral vibrational acceleration, respectively. The vertical arrow indicates a point where an extreme vibration is observed at a levelcrossing or a turnout. ISO/DIS10056 describes the features of railway vehicle vibrations as follows: Railway vibration signals are of a random nature, may include periodic features and cover a wide range of frequencies but energy level inside the vehicle is relatively low, [and railway vibration signals are] not stationary but may be considered as partially stationary (ISO, 1996, p. 5). If the waveform is a simple sinusoidal vibration, it can univocally be expressed in terms of amplitude, frequency and phase. A number of psychophysical studies have been performed on simple sinusoidal vibrations for ease of control in experiments (e.g., see Suzuki, 1998). In the case of vibrations in Figure 1, however, how to describe their characteristics is one of the study items. An index widely used in engineering is the root mean square (r.m.s.), here the root mean square of a particular variable related to acceleration over a specified time. When the mean value of amplitudes is zero, the r.m.s. value is equivalent to the standard deviation. We observe a difference between the maximum amplitude in one direction and one in the opposite direction in each period during a specified time and repeat this observation for the entire section concerned. The maximum value of the differences thus observed is called a peak-to-peak value. The maximum amplitude on one side of the reference axis is called a peak value. Peak-to-peak value (or peak value) is not an ideal index from a statistical viewpoint, because an exceptional amplitude, if any, may be taken as representing the vibrational characteristics of the entire section. In railway engineering, however, the r.m.s. value and peak value are used in parallel.

3 158 H. Suzuki Railways are a fixed guideway system with vehicles running on dedicated tracks. Business entities normally own and maintain both tracks and vehicles. As trains run, the effects of train loads are accumulated on the track. To improve the riding comfort of railway vehicles, therefore, it is important to improve the control of tracks. A key issue in controlling tracks is to improve points where extreme vibrations take place. This is based on the idea that pointby-point elimination of vibration eventually leads to an improvement of riding comfort over the entire section. For this purpose, the peak value is an effective index, as it indicates points where extreme vibrations occur and which therefore require maintenance. The peak value divided by the r.m.s. value is called the crest factor and is used as an index to indicate the degree of vibrational shock. Momentary and long-term riding comfort In view of such specific features, it is necessary, in order to discuss the riding comfort of railway vehicles, to separate the momentary riding comfort at limited points from the long-term riding comfort over an entire section. Although the term riding comfort is normally associated with the long-term comfort, the momentary comfort is equally important in railway engineering. For the purpose of controlling track conditions, therefore, the peak value is used for the momentary evaluation and the r.m.s. value for the long-term evaluation. It is not clear, however, how the momentary riding comfort evaluation is governed by the absolute peak value alone. Suzuki (1996) proved that the anchoring effect on the evaluation of vibration intensity is just the same as in other perceptual phenomena. This implies that the evaluation of momentary extreme vibrations is affected by the frame of reference. Effect of the range and frequency of stimuli Until today, the riding comfort of passengers has generally been tested by having vibration intensity evaluated by them according to rating scales prepared by the experimenter. From a methodological viewpoint, this is a sort of perceptual judgment task. In relation to a number of perceptual phenomena, it is well known that what is perceived is dependent not only on the physical properties of the stimuli, but also on one s frame of reference. For example, Helson explained the problems of the frame of reference, including the anchoring effect, with the concept of adaptation level, and this influenced studies on perceptual judgment thereafter (e.g., Helson, 1964). According to his theory, perceptual judgment is determined by the pooling of all present and past stimuli, and the scale of judgment is set around a central adaptation level, as the average of all effects of stimuli. In constructing a more precise model of perceptual judgment, however, a number of questions arose about Helson s hypothesis, such as whether the reference point could be limited to the average alone, and whether only one adaptation level was enough as a reference point. According to adaptation theory, the judgment of the effects of different stimuli would not change as long as their geometric mean was the same, even though the stimuli followed different patterns of distribution. In response to this question, Parducci and his colleagues proposed a range-frequency model (R-F model), based on a series of experiments performed for more widely ranging stimuli and varied frequency conditions (Parducci, 1963; Parducci & Haugen, 1967; Parducci & Perret, 1971). This model assumes two principles to be followed by subjects in judging the categories. One is a range principle, in which the range of the stimuli is divided into several sub-ranges of a lower order, to which each category corresponds. The other is a frequency principle, in which different categories tend to be used at the same frequency. The model is based on the assumption that actual judgment is a compromise between these principles. The method of formulating the R-F model is briefly introduced below (Kakizaki, 1974; Parducci & Perret, 1971). 1. We denote the ith stimulus in a series of stimuli by S i, and define the range value R i for S i as the ratio of the distance between S i

4 Factors affecting vibrational discomfort 159 and the lowest point in the series to the length of the total range (i.e., the distance between the highest and the lowest points). 2. We define the frequency value F i for S i as the ratio of the sum of stimuli lower than S i to the sum of all stimuli in the series. 3. The judgment J i for the stimulus S i is given by relating R i and F i as J i = ωr i + (1 ω)f i. The coefficient of weight, ω, is set at 0.5 unless another particular value is required for evaluation. The value of C i for actual evaluation is given by a linear transformation, C i = bj i + a, where a is the value of the lowest category for evaluation and b is a value in the range of evaluation. Parducci proved that the R-F model is applicable to various perceptual judgments, such as the size of a square, number of dots, heaviness of a weight, by comparing experimental results obtained for different ranges and distributions of stimuli. He also proved that the contextual effects are conspicuous, even in experiments in which the range and frequency of stimuli are gradually changed. It is known that the contextual factors are influential in the riding comfort evaluation of railway vehicles (Suzuki, 1996, 1997a; Suzuki & Kobayashi, 1994). In this respect, the studies by Parducci and his colleagues are very interesting. Range and frequency of occurrence of railway vehicle vibrations Running trains and passengers on board are exposed to vibrations in various directions, of which lateral vibrations are the most important in the evaluation of riding comfort on railway vehicles (Suzuki, 1997a, 1997b). For this reason, we adopted lateral vibrations as the theme of this study. The white bars in Figure 2 show an example of the distribution of peak values of vibrational acceleration in the lateral direction obtained when an express train runs at the scheduled speed on a section with a number of curves in a mountainous area. Bars represent the frequencies of occurrence of the peak value. The peak values used in this figure were calculated for each block in a 20-min running Figure 2. An example distribution of peak values of lateral vibrational acceleration measured on a running railway vehicle. The white bars represent a case where an express train ran on a section with a number of curves at the scheduled speed, while the black bars represent a case where a test train of the same type ran on the same section at a speed of about 15 km/h faster than the scheduled speed. Approximation curves are also drawn approximating each histogram according to a log-normal distribution. section divided into successive 5-s blocks. Dividing the total section into 5-s blocks has been standard practice in momentary evaluation experiments in Japan. The same approach is also adopted in calculating the r.m.s. values in international standards (European Rail Research Institute, 1993; ISO, 1996). As seen in the figure, the distribution of the peak values is positively skewed. The thin line represents an approximate curve according to a log-normal distribution. The black bars in Figure 2 show an example of the distribution of peak values obtained when a train of the same type is operated on the same section at a speed 15km/h faster than that in the above case. 2 The thick line is an approximate curve according to a log-normal distribution. Although the pattern of the distribution is more or less the same, the frequency 2 On sections with curves of small radii, the maximum running speed is strictly limited to prevent rapid changes in acceleration. Therefore, the test train could not run through the whole section at speeds exactly 15 km/h faster than the scheduled speed.

5 160 H. Suzuki Figure 4. An example of a waveform used in this study. Figure 3. Four-axis vibration apparatus as a riding comfort simulator. of occurrence of extreme vibrations increases due to the higher speed. As it is difficult to control occasional vibrations in trains, we use a vibration simulator by applying the range and frequency of occurrence of stimuli observed in field experiments. Experiment 1 The purpose of this experiment was to clarify the effects of differences in the ranges of vibration intensities and frequencies on the momentary riding comfort evaluation. Method Apparatus. We used a simulator that vibrates independently in four directions, longitudinal, lateral, vertical, and roll (rotation around the longitudinal axis) to simulate various vibration modes of running trains in a laboratory (Suzuki, Omino, & Fukushima, 1997). The vibration plate ( mm) of the apparatus was vibrated with a subject standing on it (Figure 3). As this simulator was being equipped with an anti-noise device, the effects of noise on the evaluation results were negligible (Suzuki, 1996). Subjects. 40 adult subjects (20 men and 20 women, aged years) participated in this experiment. Ten were assigned to each of the four conditions described below, with no one participating in multiple conditions. Stimuli. We made an experimental design simplifying the vibrations observed on actual trains. As shown in Figure 4, we excited the apparatus to a periodic sinusoidal waveform, into which only one period of sinusoidal vibration of a different amplitude was inserted. Hereafter, the inserted vibration that makes the peak value is called the peak vibration, and the periodic sinusoidal vibration is the background vibration. Background vibrations before and after the peak vibration were set for the same periods. We set four distribution patterns (A, B, C, and D) of different ranges and frequencies of occurrence of peak vibration and presented stimuli 48 times in all patterns. Table 1 shows Table 1. Stimulus frequency distribution used in Experiment 1 Acceleration (m/s 2 ) Total Pattern A Pattern B Pattern C Pattern D

6 Factors affecting vibrational discomfort 161 the frequency distributions of stimuli presented in each pattern. Patterns A and B correspond to white bars and black bars in Figure 2, respectively. Pattern C is obtained by shifting pattern A toward the right so that the upper limit of the range takes a value of 2.0 m/s 2, the same as that of pattern B. Pattern D represents a case where all stimuli are presented at the same frequency in the same range as that of pattern B. Patterns C and D would never actually occur on running trains. We set 12 levels of vibration patterns, from 0.35 to 2.00 m/s 2 with a spacing of 0.15 m/s 2 in each range, by taking into account the fact that a difference threshold for lateral vibrations has been proved to be approximately at this level (Suzuki & Kobayashi, 1994). This setting allowed us to clarify the effects of: 1. changes in vibration distribution due to higher train speeds on the evaluation of riding comfort (comparison between patterns A and B); 2. different ranges at the same stimulus presentation (comparison between patterns A and C); 3. different frequencies of stimulus presentation in the same range (comparison between patterns B and D). Changes in the crest factor also affect the evaluation results. To eliminate this effect, therefore, we adjusted the background vibration so that the value of the crest factor was maintained at 2.5, a value typically observed on running trains. Although the effect of different frequencies of vibration should be taken into account, we set the frequency at 2 Hz for all conditions, which is predominant and most frequent on trains. Rating scale. The scale comprised six category ratings: These categories were used as a reference for the purpose of evaluating the vibration of public transport facilities in Annex C to ISO 2631 (ISO, 1997), widely used in various countries. Procedure. The task of subjects was to evaluate on the rating scale the intensity of vibration patterns which were successively presented. Each vibration pattern was presented for 5 s. After each vibration pattern, subjects were required to evaluate the degree of discomfort regarding the presented vibration patterns as vibrations of trains. The interstimulus interval for subjects to write down the rating was set at 10 s. This was adopted as sufficient but not too long for the purpose, based on preliminary experimental results. To avoid the effects of the order in which stimuli were presented, vibration patterns were presented at random for different subjects. Subjects stood with their feet spread to shoulder width and gripping the handrail on the vibration plate with one hand. To avoid the effects of the difference in footwear, they wore slippers of the same design. Results Figure 5 shows changes in the mean scores recorded by all subjects when the peak value intensified in each vibration pattern. As suggested by the figure, evaluation results differed across the vibration patterns. In this experiment, the four steps 0.95, 1.10, 1.25, and 1.40 m/s 2 were presented in all vibration patterns, but their frequencies of 1. not uncomfortable; 2. a little uncomfortable; 3. fairly uncomfortable; 4. uncomfortable; 5. very uncomfortable; 6. extremely uncomfortable. Figure 5. Mean scores as a function of the peak acceleration in Experiment 1.

7 162 H. Suzuki presentation were different. By using the data obtained for these four step stimuli, therefore, we tested the effects of two factors pattern and acceleration by ANOVA. As pattern was a between-subjects variable and acceleration was a within-subject variable in this experiment, a split-plot ANOVA was used. The main effects of the pattern, F(3, 36) = 8.56, p.01, and acceleration, F(3, 108) = 9.00, p.01, were both significant. However, the two-way interaction was not significant, F(9, 108) = The significance of the main effect of acceleration confirms that the discomfort scores rose as acceleration increased. Tukey s HSD tests on pattern, differences between patterns A and C, between B and D, and between A and D were significant, but they were not significant for combinations of other patterns. Discussion Ratings of patterns A and C were significantly different, despite being for the same frequency of stimulus presentation, although the ranges were different. At the same time, the ratings for patterns B and D also differed significantly, despite these patterns having each stimulus in the same range, although the frequencies of stimulus presentation differed. These facts prove that the evaluation of discomfort differs for different ranges and frequencies of stimuli. On the other hand, the difference between patterns A and B was not significant. This means that the rating does not differ much, depending on the running conditions, when the difference in running conditions is limited to the extents assumed in this experiment. Next, we applied the concept of the R-F model to the experimental results. Figure 6 shows a predicted function from R-F model (lines) and empirical evaluated scores (points) as a function of peak acceleration. Although there are some discrepancies between the predicted and empirical values, they are in fairly good agreement as a whole. The correlation coefficients between the predicted and empirical values are as high as for pattern A, for B, for C, and for D. However, the results did not exactly correspond to the R-F model; for instance, the maximum Figure 6. Predicted function from the R-F model (lines) and empirical evaluated scores (points) as a function of the peak acceleration. evaluation point observed in this experiment was 4.0 or less for all vibration patterns against the maximum point of 6.0 among the defined evaluation categories, whereas the model assumes a frequency principle based on which the subjects tend to quote all categories at a fixed ratio and to match the defined maximum point (6.0 in this experiment) to the maximum stimulus in the range of stimuli. The reason for this discrepancy is the difference in the tasks set to test perceptual judgment. Parducci used artificial stimuli designed for the experiment, such as the size of a square and number of dots, while this experiment required judgment on the vibration discomfort experienced on railway vehicles routinely used by subjects. Therefore, evaluation results in this experiment seem to reflect subjects comparisons with their long experience, in addition to the effects of the ranges of stimuli and the frequencies of stimulus presentation. On the other hand, it is known that the endurance of postural stability against rapid changes in acceleration differs by 30% depending on whether the stimulus is signaled in advance or not (Ohno & Nagata, 1994). It is understandable, therefore, that scores for extreme vibrations are lower in experiments than on actual trains. For these reasons, the scores in this experiment seem to be lower by about 40% on average than those predicted from the R-F

8 Factors affecting vibrational discomfort 163 model. In drawing the function in Figure 6, therefore, we used points 40% smaller than those predicted from the R-F model. In this conversion the original values are multiplied by a fixed ratio, and so the correlation coefficient between the empirical and predicted values is not affected. In discussing the fit to the R-F model, the stimuli assumed in the present experiment may not be adequate. Nevertheless, at least as far as the daily problem of the momentary riding comfort of railway vehicles is concerned, it may be worth suggesting here that such a model may be suitable. Figure 7. Mean scores as a function of the peak acceleration in Experiment 2. Experiment 2 As mentioned above, we clarified the effects of the ranges and frequencies of stimuli on the momentary discomfort evaluation. However, what we have found emphasizes the difficulty of making absolute evaluations of perceptual judgments, and it is impossible for us to make a definitive scale for this purpose. Is there not a method by which to make the evaluation more exact? One means may be to change the rating scale and terms. It is known that the results of evaluating the vibration intensity depend on the scale and terms used (Dempsey, Coates, & Leatherwood, 1977; Suzuki, 1998). In Experiment 1, the evaluation scale had different adverbs employed to modify uncomfortable. The terms used were relative and similar to those used in Parducci s experiments. When a subject gave an evaluation category of 3 to a vibration pattern, for example, he or she would give a rating higher than 3 when a stronger vibration pattern was presented thereafter (Parducci & Perret, 1971). The purpose of the study on riding comfort has mainly been to clarify the vibration intensity that subjects feel to be uncomfortable or unacceptable (Suzuki, 1997a). By combining two questions whether the vibration is uncomfortable and whether it is in the unacceptable range as a vibration on railway vehicles we devised the following ratings: 1. not wholly uncomfortable or slightly irritating; 2. uncomfortable, but in the acceptable range of riding comfort for railway vehicles; 3. extremely uncomfortable and not acceptable as riding comfort for railway vehicles. By asking the subjects whether the vibration was uncomfortable or unacceptable, we studied the effects of the question on the evaluation result. Method The experimental method was the same as that for Experiment 1, except the rating scale was composed of the above three evaluation categories. 40 subjects, 20 men and 20 women, again participated in the experiment. Results Figure 7 shows the changes in the evaluation results for the four patterns A D. Compared with Figure 5, there are no large differences between the different patterns. In the same way as in Experiment 1, we tested by ANOVA the four steps from 0.95 to 1.40 m/s 2 which were presented in all patterns. Only the main effect of acceleration was significant, F(3, 108) = 7.08, p.01. The main effect of pattern, F(3, 36) = 0.135, and the two-way interaction, F(9, 108) = 0.09, were not significant. Therefore, it was proved that the different ranges or frequencies did not affect the ratings in this experiment.

9 164 H. Suzuki General discussion Effect of the number of response categories Although the judgment of human beings tends to be relative, the above result proved that we can make an approximate absolute evaluation when we use an appropriate rating scale and terms. Experiment 2 also shows that subjects ratings on scales with a smaller number of categories are less affected by the context of the judgment category. In contrast, Parducci stated that judgment tends to become incorrect and is affected more by the context of the judgment category when smaller numbers of categories are used (Parducci, 1964; Parducci & Sandusky, 1965). We discuss this discrepancy below. Parducci stated that subjects tend to switch categories when the later stimulus can be distinguished from one presented just before, and that the effect of the order of presentation can be eliminated by decreasing the number of stimuli present for each category or by increasing the number of categories available for selection. He verified his statement in several experiments. However, the categories used in these experiments were relative, being modified by adverbs such as very and fairly or attributes such as large and small. The number of categories was increased or decreased by adjusting the types of descriptor. In addition, the objects to be judged were the size of a square, length of a line segment and other neutral stimuli little related to everyday affairs. In contrast, the experiments in this study were meant to evaluate the momentary riding comfort on railway vehicles in relation to experiences in the daily life of the subjects, with emotional words such as uncomfortable, or unacceptable. In this context, our experiments were completely different from those of Parducci. Therefore, the discrepancy between the two types of experiment cannot simply be attributed to the difference in the number of categories used, but requires an investigation into the contents of evaluation. We should further discuss the difference between the evaluation of physical attributes such as large and small and that of emotional elements such as uncomfortable and unacceptable. Acceleration felt as uncomfortable or unacceptable We cannot conclude that ranges and frequencies of stimuli do not affect judgments based on the results of Experiment 2 alone. Moreover, we have to determine the degrees of acceleration perceived as uncomfortable and unacceptable in order to meet engineering requirements. Therefore, we scale all data obtained from Experiment 2 according to Thurstone s method to find the boundary value for judging uncomfortable or unacceptable. Taking the value of acceleration at which 50% of responses are ratings of 1 and 50% are of 2 or 3 as the boundary deciding not uncomfortable or uncomfortable, we arrived at a value of 1.16 m/s 2. Taking the value of acceleration at which the 50% of responses are ratings of 1 or 2 and 50% are of 3 as the boundary deciding acceptable or unacceptable, we arrived at a value of 1.83 m/s 2. The reliability of these values will have to be verified through accumulation of data collected under a wide diversity of test conditions. Conclusions Based on various experiments on the evaluation of riding comfort on railway vehicles, a number of criteria and guidelines have been offered. However, the results obtained have often been inconsistent; the effects of different methods, procedures, and rating scales have not been studied sufficiently. There are a number of railway engineers, therefore, who believe that evaluating riding comfort by requiring subjects to answer in words does not produce accurate results. Experiments introduced in this paper were performed to examine the effects of different stimulus ranges, frequencies, and rating scales on perceptual judgments. Although perceptual judgment tends to become a relative one, affected by the context, these experiments prove that we can obtain almost absolute evaluation results if we adopt an appropriately designed experimental method. For this purpose, it is important to perform an

10 Factors affecting vibrational discomfort 165 evaluation in well-defined words that reflect the daily experiences of the subject, as suggested by the experimental results. Not only ISO2631 but also projects to apply it to the evaluation of riding comfort on railway vehicles aim to establish a generally applicable evaluation scale (ISO, 1996, 1997). However, it may be appropriate to choose evaluation terms more easily understandable in the evaluation of routine affairs, including the riding comfort on railway vehicles. The experiments introduced in this paper adopted simplified stimuli, as they were performed according to the experimental design method. It is necessary to repeat the experiments by applying stimuli more closely reflecting actual vibrations on railway vehicles. Though we used a three-item scale in Experiment 2, further debate will be needed before finally deciding on the measurement scale to be adopted. References Dempsey, T. K., Coates, G. D., & Leatherwood, J. D. (1977). An investigation of ride quality rating scales. NASA Technical Paper Hampton, VA: National Aeronautics and Space Administration. European Rail Research Institute (1993). Application of ISO Standard 2631 to railway vehicles (B 153), Report No. 19, Final report. Utrecht: European Rail Research Institute. Helson, H. (1964). Adaptation-level theory. New York: Harper and Row. ISO (1996). Mechanical vibration Measurement and analysis of vibration to which passengers and crew are exposed in railway vehicles, ISO/ DIS Geneva: International Organization for Standardization. ISO (1997). Mechanical vibration and shock evaluation of human exposure to whole-body vibration, Part 1: General requirements, ISO Geneva: International Organization for Standardization. Japanese Industrial Standards Committee (1990). Vibration characteristics of railway rolling stock Measuring methods (JIS-E4023). Tokyo: Japanese Standards Association. Kakizaki, S. (1974). Perceptual judgment. Tokyo: Baifu-kan. (In Japanese.) Ministry of Transport (1996). Facts and figures of railways. Tokyo: Japan Transport Economics Research Center. (In Japanese.) Ohno, H., & Nagata, H. (1994). Safety assessment of motion environments from a view point of postural stability. Proceedings of the Third Transportation and Logistics Conference, Japan Society of Mechanical Engineers, (In Japanese.) Parducci, A. (1963). Range-frequency compromise in judgment. Psychological Monograph, 77, 2 (Whole No. 565). Parducci, A. (1964). Sequential effects in judgment. Psychological Bulletin, 61, Parducci, A., & Haugen, R. (1967). The frequency principle for comparative judgments. Perception and Psychophysics, 2, Parducci, A., & Perret, L. F. (1971). Category rating scales: effect of relative spacing and frequency of stimulus values. Journal of Experimental Psychology Monograph, 89, Parducci, A., & Sandusky, A. (1965). Distribution and sequence effects in judgment. Journal of Experimental Psychology, 69, Suzuki, H. (1996). Effects of frame of reference on the judgments of whole-body vibration intensity. Japanese Journal of Psychology, 67, (In Japanese with English summary.) Suzuki, H. (1997a). An applied psychological approach to riding comfort evaluation of railway vehicles. Japanese Psychological Review, 40, (In Japanese with English summary.) Suzuki, H. (1997b). Effect of vibrational factors on the evaluation of whole-body vibrational intensity. Japanese Journal of Psychology, 68, (In Japanese with English summary.) Suzuki, H. (1998). Research trends on riding comfort evaluation in Japan. Journal of Rail and Rapid Transit (Proceedings of the Institution of Mechanical Engineers Part F), 212, Suzuki, H., & Kobayashi, T. (1994). Perceptual psychological study on evaluation of riding comfort of the train passing a curve. RTRI Report, 8, (In Japanese.) Suzuki, H., Omino, K., & Fukushima, N. (1997). Improvement in performance of riding comfort simulator. Japanese Journal of Ergonomics, 33, (In Japanese.) (Received April 28, 1997; accepted Nov. 8, 1997)