Analysis of Ground-Transducer Coupling in Monitoring Vibration from Railways: A Case Study

Ground vibrations induced by railway traffic have received particular attention in recent years since they may generate adverse effects both on people in the surroundings and on adjacent buildings or structures (Alfaro Degan et al. [1]); nevertheless, the energetic efficiency of railway transportation, together with the related improvement of traffic conditions and the reduction of airborne pollutant emissions, has determined a huge implementation of new railway lines, a development that has caused a growing interest among the scientific community for the environmental effects due to the passage of trains. In particular, many research studies about noise and vibration have been carried out in terms of both measurement techniques (Kurzweil [2]; Kima [3]) and analytical methods (Alfaro Degan et al. [4]); moreover, efforts have been made in order to improve passengers’ comfort by reducing vibrations perceived in the vehicle, but the issue of railway-induced ground vibration is far from being solved (Ryue [5]; Adam [6]; Zoccali [7]). Because many parameters affect this phenomenon, different approaches of growing complexity have been proposed to model vibrations determined by train passage (Costa [8]), and among the main causes of uncertainty in modelling train-induced waves, the different transmission paths along the three directions in a fundamentally inhomogeneous medium (Nejati [9]) constitute the most relevant one. Many researchers have focused their efforts on performing prediction models (Ju [10]; Triepaischajonsak [11]; Lui [12]), whereby the propagation of ground vibration waves is analysed considering different generation mechanisms such as rail deflection and track irregularity rather than vehicle dynamics (Volberg [13]); moreover, the validation of propagation models performed by field tests (Connolly [14]) has highlighted some further features connected with sampling methods, and in particular, two relevant aspects for achieving good measurements are represented by positioning and ground-coupling transducer. An efficient coupling between the transducer and the ground is often difficult to achieve, especially when the transducer mounting options are restricted to those allowed by local ground conditions; a bad coupling may be represented by the slipping of the transducer, thus leading to signal distortion in amplitude and phase, which consequently implies that vibration levels are erroneously measured, whereas an efficient ground transducer coupling is guaranteed by a good adherence to sampling tests. Scientific literature reveals that the ground coupling is a resonant phenomenon in which the transducer and ground coupling create a resonant system (Washburn [15]), a feature that has been investigated by several studies focused on the parameters influencing the resonance frequency, and the base area and the weight of the transducer together with the ground conditions have been found to be among the most important parameters that affect the amplitude and the phase of the signal (Omata [16]); another study concluded that the length of the section, the positioning of the transducer and the characteristics of the soil affect the measurements in a deeper way (Krohn [17]). Research based on theoretical models and results from laboratory tests can describe transducer ground coupling in a limited way because the boundary conditions of field tests are very complex to be replicated; in particular, since soil properties, density gradients, quality of mounting and so forth (Faber [18]) vary from one case to another, field validation represents the only way to test, at the local scale, the efficiency of any model. In order to suggest good practices in ground transducer coupling, many national institutions have developed specific documents or technical norms (UNI 9916 [19]; BSI [20]), in which the recommended solution was that of fixing the instrument, an approach found to be effective especially when the maximum sampled acceleration falls in the interval between 0.2g and 1.0g (Segarra [21]). Another coupling method is performed with the use of steel spikes, though their effectiveness is largely debated: some researchers claim that this method modifies the recorded ground motion (Dowding [22]), thus contributing to overestimation of the real value by 46.5% (Blair [23]), while other studies highlight that the above method shows good results for accelerations below 1.0g (ISEE [24]); moreover, long steel spikes are probably the cause for increasing the coupling resonance frequency, a point of view not shared by some authors who instead have shown by means of a specific model that the resonant frequency decreases by increasing length and radius of the spike section (Drijkoningen [25]). Another simple way to perform the coupling of the transducer to the soil is the use of a weight placed on the top of the instrument, a method that appears to be effective (Stagg [26]) in some cases (acceleration values less than 1g) although its application may not be considered affordable under other conditions (Robertson [27]). Such an overview of the scientific literature reveals a lack of a shared approach to the ground transducer coupling, and in this article, a case study is presented in which two different sampling methods are tested, the obtained results are compared, and finally some considerations are carried out on the most effective solution for the selected case study.

This work aims to compare different methods in order to measure ground vibrations induced by train passages. A metal cube and steel spikes were employed. These spikes are characterized by different lengths and an ‘L’ shape, according to field-guidelines UNI 5783-66. In order to investigate some of the issues mentioned above, the adopted methodology consisted in considering the actual signal input as unknown (vibration induced by train impinging on the mounting system). The effectiveness of each coupling method was then assessed by feeding the same impulse into the mounted transducers. This was an effective way of investigating if there were internal flaws or relevant differences in the structure (e.g., resonances) where the transducers were placed (Fig.1). A square metal plate of 5-mm thickness was welded on each steel spike. The transducer was fixed on it through a permanent magnet. Table 1 shows the technical characteristics.

The sampling campaign was carried out by mean of a tri axial accelerometer whose characteristics are summarized in Table 2. The signal has been acquired through SAMURAI software and analysed with Noise Vibration Work software. The instrument was calibrated with 1,250 Hz sampling frequency and 3.15 V gain. Strict adherence to the correct alignment of the transducer in the ground is very important in order to obtain correct measurement; an incorrect alignment may affect the measurement itself. The accelerometer should be positioned on an orthogonal plane with respect to the ground surface. Krohn [17] shows that at frequencies lower that 60 Hz, the vertical alignment of the transducer contributes to creation of a distortion greater than that generated by the coupling mechanism. Before each field test, the vertical and horizontal alignments with respect to the ground were verified through a bubble level placed on the transducer.

The site consisted of a railway section in the southern part of Rome (see Fig. 2). The sampling points were located at a distance of 20 m at a lower level with respect to the railway (approximately 3 m). Three distinct tests were carried out to set the instruments; the first one was aimed to sample the component of the signal along each axis; the second and the third one were used to record vibration levels through steel spikes.