A Method for Detecting Surface Defects in Railhead by Magnetic Flux Leakage

comments

responses

I find the presentation of leakage magnetic field in fig. 1 unfaithful, maybe you could add a figure of the fields from the FEM calculation. On this figure, you could also add sensors.

Thanks for your comments. A sensor is added to the fig.1. A new figure is added to show the magnetic lines near a defect. We use the local graph, because the leakage magnetic field is small, it might be less obvious in a whole system graph.

you could explain how the sensors are built and what are their key geometric properties

The type and dimensions of the sensor are given in Section 4.2.1.

you should give more data about the experimental setup (what are the dimensions, properties of the setup, are they equal to FEM analysis, etc.)

The introduction of the experimental setup is added in Section 4.2.1.

comments

responses

In the introduction, the authors present a review of similar topics based on studies mainly from Asian and Russian literature. I suggest adding an extension of the introduction and description of the method from more international sources and applications.

Thanks for your comments. We replaced a Chinese manuscript with an Australian manuscript and two European manuscripts were added.

In this place, I have also a formal comment - because the form of references used in the paper does not meet the standards.

Sorry. We have revised the form.

I have few questions for the authors about the designed system. Description of created system lacks a more detailed description about the type of used pickup sensors. What are the type and dimensions of used pickups?

The introduction of the experimental setup is added in Section 4.2.1 include the type and dimensions of the sensor.

What is the smallest defect significant for the railway usage and is the measurement system able to detect and evaluate as a possible threat?

We have cooperation with CHINA ACADEMY OF RAILWAY SCIENCES. The defect dimensions are chosen by them and most of the defect samples they provide are these dimensions. China has just started to detect rail surface defects, and the smallest defect has not been definitely decided. This manuscript tries to find defects and has not considered how to evaluate a threat. This needs further study.

Why was the distance of auxiliary sensors based on the chosen defect dimensions from simulation and not practice demands?

The defect dimensions are chosen by CHINA ACADEMY OF RAILWAY SCIENCES based on practice. Simulation and experiment are based on the demands. We are trying to get some samples with smaller defects for further research.

Are the transversal defects the main issue? Were you considering the occurrence of longitudinal defects?  Would the proposed system be able to detect such defects?

Yes, the transversal defects are the most common defects on the surface of the rail head. We believe that the longitudinal defects can be detected by rotating the yoke and sensors 90 degrees. At present, we have no longitudinal defects samples and have not made further research.

I have also formal comments. Presented results are not using the same horizontal axis distribution. Also, the usage of a grid would help to better orientation in presented figures.  Figures in the present state do not support the claim of for one or three sensors configurations.

Sorry, the horizontal axes of some graphs are inconsistent. We have modified the axis of Fig. 11 and 12. In the new figures, the positions of the zero and maximum values in the z and x-axis are about the same.

Fig. 6 a) does not support the claim "The increase in width makes the maximum and minimum of the signal gradually increase." because the widest defect signal is lower than the 2.5mm and very similar to 2mm width.

Yes, the widest defect signal is lower than the 2.5mm width. This is our negligence, and the manuscript has been modified. As can be seen from Fig.6a), when the sensors are arranged according to the minimum defect width, the main and auxiliary sensor signal difference of a wider defect is greater than that of the minimum width defect, so our proposed method is correct.