A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs

Abstract Punching shear failure constitutes a critical structural concern in conventional concrete slabs, prompting the development of various equations to determine their maximum shear strength. Voided concrete slabs have gained prominence for their ecological advantages and sustainability. However...

Full description

Saved in:
Bibliographic Details
Main Authors: Alireza Mahmoudian, Mussa Mahmoudi, Mohammad Yekrangnia, Nima Tajik, Mostafa Mohammadzadeh Taleshi
Format: Article
Language:English
Published: Springer 2025-02-01
Series:Discover Civil Engineering
Subjects:
Online Access:https://doi.org/10.1007/s44290-025-00181-4
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850237837554745344
author Alireza Mahmoudian
Mussa Mahmoudi
Mohammad Yekrangnia
Nima Tajik
Mostafa Mohammadzadeh Taleshi
author_facet Alireza Mahmoudian
Mussa Mahmoudi
Mohammad Yekrangnia
Nima Tajik
Mostafa Mohammadzadeh Taleshi
author_sort Alireza Mahmoudian
collection DOAJ
description Abstract Punching shear failure constitutes a critical structural concern in conventional concrete slabs, prompting the development of various equations to determine their maximum shear strength. Voided concrete slabs have gained prominence for their ecological advantages and sustainability. However, the calculation of punching shear strength for voided slabs lacks a standardized approach due to limited experimental data. In this study, an innovative application of machine learning is proposed to address this challenge. Initially, a predictive model for punching shear strength in conventional slabs is established using the Random Forest Regressor (RFR). This model treats voided slabs as solid ones to generate a reduction factor, representing the reduction of punching shear strength in voided concrete slabs compared to the corresponding conventional ones, for estimating the punching shear strength of voided slabs. The coefficient is multiplied by the shear strength of the slabs based on the distance from the exterior side of the column to the first existing hole in the slab and the effective depth of the slab (d). The efficacy of the approach is showcased using the Random Forest Regressor model, finely tuned through a Grid Search technique, with performance evaluated using the R-squared coefficient and Root Mean Squared Error metrics. If the aforementioned distance is less than 0.3d, a factor of 0.55 should be applied. For distances equal to or greater than 0.3d but less than d, a factor of 0.61 should be used. Lastly, for voided concrete slabs where the distance between the first void and the column face is equal to or greater than d, a factor of 0.8 should be used. Additionally, an alternative approach utilizing a direct Random Forest model with fivefold cross-validation has been developed to enhance the robustness of the predictions. The results indicate that this model can effectively estimate the punching shear strength of voided concrete slabs, contributing to the advancement of structural engineering practices.
format Article
id doaj-art-d0bccc313bd34439bf53cc3a16f2de9f
institution OA Journals
issn 2948-1546
language English
publishDate 2025-02-01
publisher Springer
record_format Article
series Discover Civil Engineering
spelling doaj-art-d0bccc313bd34439bf53cc3a16f2de9f2025-08-20T02:01:39ZengSpringerDiscover Civil Engineering2948-15462025-02-012112610.1007/s44290-025-00181-4A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabsAlireza Mahmoudian0Mussa Mahmoudi1Mohammad Yekrangnia2Nima Tajik3Mostafa Mohammadzadeh Taleshi4Department of Civil Engineering, Shahid Rajaee Teacher Training UniversityDepartment of Civil Engineering, Shahid Rajaee Teacher Training UniversityDepartment of Civil Engineering, Shahid Rajaee Teacher Training UniversityDepartment of Civil, Structural and Environmental Engineering, State University of New York at BuffaloCivil and Environmental Engineering Department, University of NevadaAbstract Punching shear failure constitutes a critical structural concern in conventional concrete slabs, prompting the development of various equations to determine their maximum shear strength. Voided concrete slabs have gained prominence for their ecological advantages and sustainability. However, the calculation of punching shear strength for voided slabs lacks a standardized approach due to limited experimental data. In this study, an innovative application of machine learning is proposed to address this challenge. Initially, a predictive model for punching shear strength in conventional slabs is established using the Random Forest Regressor (RFR). This model treats voided slabs as solid ones to generate a reduction factor, representing the reduction of punching shear strength in voided concrete slabs compared to the corresponding conventional ones, for estimating the punching shear strength of voided slabs. The coefficient is multiplied by the shear strength of the slabs based on the distance from the exterior side of the column to the first existing hole in the slab and the effective depth of the slab (d). The efficacy of the approach is showcased using the Random Forest Regressor model, finely tuned through a Grid Search technique, with performance evaluated using the R-squared coefficient and Root Mean Squared Error metrics. If the aforementioned distance is less than 0.3d, a factor of 0.55 should be applied. For distances equal to or greater than 0.3d but less than d, a factor of 0.61 should be used. Lastly, for voided concrete slabs where the distance between the first void and the column face is equal to or greater than d, a factor of 0.8 should be used. Additionally, an alternative approach utilizing a direct Random Forest model with fivefold cross-validation has been developed to enhance the robustness of the predictions. The results indicate that this model can effectively estimate the punching shear strength of voided concrete slabs, contributing to the advancement of structural engineering practices.https://doi.org/10.1007/s44290-025-00181-4Machine learningVoided concrete slabRandom forestBubble deck slabPunching shear strength
spellingShingle Alireza Mahmoudian
Mussa Mahmoudi
Mohammad Yekrangnia
Nima Tajik
Mostafa Mohammadzadeh Taleshi
A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
Discover Civil Engineering
Machine learning
Voided concrete slab
Random forest
Bubble deck slab
Punching shear strength
title A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
title_full A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
title_fullStr A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
title_full_unstemmed A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
title_short A novel machine learning-based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
title_sort novel machine learning based approach to determine the reduction factor for punching shear strength capacity of voided concrete slabs
topic Machine learning
Voided concrete slab
Random forest
Bubble deck slab
Punching shear strength
url https://doi.org/10.1007/s44290-025-00181-4
work_keys_str_mv AT alirezamahmoudian anovelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mussamahmoudi anovelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mohammadyekrangnia anovelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT nimatajik anovelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mostafamohammadzadehtaleshi anovelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT alirezamahmoudian novelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mussamahmoudi novelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mohammadyekrangnia novelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT nimatajik novelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs
AT mostafamohammadzadehtaleshi novelmachinelearningbasedapproachtodeterminethereductionfactorforpunchingshearstrengthcapacityofvoidedconcreteslabs