Prediction of Subgrade Modulus Using Artificial Neural Network of New Feed Forward

Siegfried Syafier

doi:10.32816/tiarsie.v18i3.103

Siegfried Syafier Puslitbang Jalan dan Jembatan, Universitas Langlangbuana

DOI: https://doi.org/10.32816/tiarsie.v18i3.103

Abstract

The value of subgrade modulus is an important parameter used for the evaluation or design of a pavement system. The standard method in collecting this value is through the test using the Dynamic Cone Penetrometer (DCP) when doing test pit up to the top level of subgrade. Another method is using an approach dealing with the surface deflection that is collected by the Falling Weight Deflectometer (FWD) and applying the formula of AASHTO 93 or Boussinesq. This paper purposes to propose a method for predicting the value of subgrade modulus by applying an Artificial Neural Network of the type of New Feed Forward. The program is written in Matlab. The training data is about 3125 sets of surface deflection. A set of data consists of 7 surface deflections and spacing of 0, 200, 300, 450, 600, 900, and 1500 mm. This set of spacing conforms to that of FWD geophone spacing. The proposed method is applied to 3 road links i.e. Cawang-Cibinong, Soreang-Cipatik, and Tanjung Sari. It is found the plotting between the results of the proposed method and that of AASHTO 93 mostly falls near the equality line. It is then concluded that the proposed method in predicting subgrade modulus using the FWD deflection data is acceptable.

References

[1] Transit New Zealand Authority, Pavement Deflection Measurement & Interpretation for the Design of Rehabilitation, no. 117. 1992.
[2] AASHTO, AASHTO guide for design of pavement structures. American Association of Street Highway and Transportation Officials, 1993.
[3] B. Marga, “Manual Desain Perkerasan Jalan no. 04/SE/Db/2017,” Jakarta. Dep. Pekerj. Umum Direktorat Jenderal Bina Marga. Jakarta, 2017.
[4] G. Jameson and K. G. Sharp, “Technical basis of Austroads pavement design guide,” 2004.
[5] M. Gribble and J. E. Patrick, Adaptation of the AUSTROADS pavement design guide for New Zealand conditions. Land Transport New Zealand, 2008.
[6] T. F. Fwa, The handbook of highway engineering. CRC Press, 2005.
[7] M. Abambres and A. Ferreira, “Application of ANN in Pavement Engineering: State-of-Art,” SSRN Electron. J., pp. 1–61, 2019, doi: 10.2139/ssrn.3351973.
[8] M. Abo-Hashema, “Artificial Neural Network Approach for Overlay Design of Flexible Pavements,” 2009.
[9] AASHTO, AASHTO Guide for Design of Pavement Structures. AASHTO, 1986.
[10] A. M. Ioannides, “Theoretical Implications of the AASHTO 1986 nondestructive testing method 2 for pavement evaluation,” Transp. Res. Rec., no. 1307, 1991.
[11] A. Ranganathan, “The levenberg-marquardt algorithm,” Tutoral LM algorithm, vol. 11, no. 1, pp. 101–110, 2004.
[12] Syafier, S. Penggunaan Light Weight Deflectometer Pusjatan untuk Quality Control Pekerjaan Pemadatan Tanah Dasar. Jurnal Tiarsie, 15(2), 45-48. 2018