Professional Documents
Culture Documents
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: Steel fibre reinforced concrete (SFRC) is widely used in the construction concrete industry as it partakes
Received 16 March 2020 an important role of evolving concrete technology. It consists of steel fibres of various shapes, sizes and
Received in revised form 23 June 2020 geometries that influence the concrete mix composition and mechanical properties. However, compared
Accepted 31 July 2020
to traditional concrete, it is difficult to design the mix proportions because more influencing variables
Available online 3 September 2020
need to be considered to optimise multiple properties including ultimate compressive strength, tensile
or flexural strength and cost. Therefore, the present study proposes an artificial intelligence based
Keywords:
multi-objective optimization model to enable an efficient method of finding the optimum mix design
Steel fibre reinforced concrete
Mixture design
for SFRC. A large dataset including 299 instances for uniaxial compressive strength (UCS) test and 269
Multi-objective optimization instances for flexural strength (FS) test were collected from previous literature. Support vector regression
Uniaxial compressive strength (SVR) model was applied to predict UCS and FS for SFRC. The hyper parameters of SVR models were tuned
Flexural strength using a firefly algorithm (FA) and a sensitivity study was conducted to understand the importance of the
inputs on the output variables for the algorithms. High correlation coefficients (0.91 for UCS and 0.85 for
FS) were achieved on the test dataset. The FA-SVR model was then applied as the objective function for a
developed multi-objective FA to search for the optimal SFRC mixture proportion. Pareto optimal solutions
were obtained and served as a design guide to determine the optimal SFRC mixtures.
Ó 2020 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.conbuildmat.2020.120457
0950-0618/Ó 2020 Elsevier Ltd. All rights reserved.
2 Y. Huang et al. / Construction and Building Materials 260 (2020) 120457
Rank the fireflies and update the Get FA-SVR model with optimum
UCS of the optimal mixtures
position hyperparameters (for UCS)
10-fold CV in training set
No
Is the stop criterion reached? Find the optimal solution using
Optimal mixtures of SFRC
MOFA
Yes
Test set (30%)
LSSVR hyperparameter tuning using FA
SVR Get FA-SVR model with optimum
FA UCS of the optimal mixtures
(for FS) hyperparameters (for FS)
Table 1
Statistics of input and output variables for the UCS dataset.
Table 2
Statistics of input and output variables for the FS dataset.
accordance to the structural risk minimisation [56], the problem out the tube. An example of a nonlinear SVR with an e-tube is
can be written as follow: shown in Fig. 3.
Xn To resolve the problems with constraints, Lagrange multipliers
1
RðwÞ ¼ kwk2 þ Lðx; y; f Þ ð4Þ can be used as follows [56]:
2 i¼1
Some errors are allowed, thus, slack variables ni and ni are used to 1 Xn
Lðw; b; n; a; lÞ ¼ kwk2 þ C ðni þ ni Þ
help with the infeasible constraints. Eq (4) can be then changed into 2 i¼1
a convex optimisation function as follow:
X
n X
n
1 Xn ai ðe þ ni yi þ w uðxi Þ þ bÞ ai ðe
minw;b;n;n RðwÞ ¼ kwk2 þ C ðni þ ni Þ i¼1 i¼1
2 i¼1 X
n
þ ni þ yi w uðxi Þ bÞ ðli ni þ li ni Þ ð6Þ
8
> yi w uðxÞ b e þ ni i¼1
>
>
< w uðxÞ þ b y e þ n
i i
s:t ð5Þ where ai 0; ai 0, li 0 and li 0 are Lagrange multipliers.
>
>
> n i 0
: When the constraint functions heavily oppose one another and
ni 0
objective function is differentiable, Karush-Kuhn-Tucker (KKT) con-
where C is the penalty parameter to determine the differences ditions are to be satisfied for each pair of primal and dual optimal
between the flatness of f ðxÞ and the penalising extent of the sample points [59] in the following manner:
4 Y. Huang et al. / Construction and Building Materials 260 (2020) 120457
8
>
>
ai ðe þ ni yi þ w uðxi Þ þ bÞ ¼ 0
>
< a ðe þ n þ y w uðx Þ bÞ ¼ 0
i i i i
ð8Þ
>
>
> ðC a i Þni ¼ 0
:
ðC ai Þni ¼ 0
The Langrage dual problem can be derived after solving the above
equations as follows:
!
1X n X n
Xn
X n
maxi ai ai aj aj xTj xj e ai ai þ yi ai ai
2 i¼1 j¼1 i¼1 i¼1
( Pn
i¼1 ai ai ¼ 0
s:t ð9Þ
ai ; ai 2 ½0; C
From Eq. (8), the weight vector can be obtained as
P
w ¼ ni¼1 ai ai uðxi Þ, and thus, the regression function can be
derived as:
X
n
f ð xÞ ¼ ai ai uðxi Þx þ b ð10Þ
i¼1
Fig. 2. Correlation matrix of input variables for UCS dataset (a) and FS dataset (b). A firefly travels towards one with higher brightness based on
the following equation [61]:
cr 2ij
xtþ1
i ¼ xti þ b0 e xtj xti þ aðrand 0:5Þ ð11Þ
where n is the total number of data samples; yi and yi are the pre-
dicted and actual outputs respectively; y and y are mean values
of the predicted and observed outputs respectively.
Solving a MOO problem is more complicated than solving a Mix parameters Expressions Lower Upper
single-objective optimisation problem [65]. In this study, the bound bound
MOFA is used to solve the SFRC mixture design problem. Cement content C(kg/m3) 300 1200
Fine aggregate content FA(kg/m3) 0 1356
2.5.1. Objective function W/C Ratio C w =C C 0.25 0.6
Steel fibre volume fraction Vf 0 0.06
As mentioned above, the FA-SVR model is used as the objective
Fine aggregate ratio C FA =ðC FA þ C CAÞ 0.3 0.6
function for UCS and FS. The third objective function (cost) is cal- Superplasticiser to cement C SP =ðC SP þ C C ) 0 0.03
culated by polynomial equations as follows: ratio
Costð=m3 Þ ¼ C c Q c þ C w Q w þ C fa Q fa þ C ca Q ca þ C SP Q SP þ C sf Q sf
ð18Þ
haps, the simplest one is to combine all the multiple objectives
where Table 3 identifies the estimated cost and unit weight of each (fk) into a composite single objective (f) using a weighted sum
variable of SFRC used in this study. Qc, Qw, Qfa, Qca and Qsf are the method as follows:
quantity (in kg/m3) for cement, water, fine and coarse aggregate,
X
K X
K
superplasticiser and steel fibre, respectively. f ¼ wk f k ; wk ¼ 1 ð21Þ
K¼1 K¼1
2.5.2. Constraints
To solve the MOO problems, constraints must be set as follows: where weights wk ¼ pKk ; pk are the random numbers generated from
a uniform distribution [0,1] and K represents the uniformly dis-
Range constraints tributed number.
For this study, the multi-objective functions using the weighted
There should be a variance amongst decision variables within a sum method are as follow:
definite range set by the datasets for SFRC mixture optimisation
J 1 ¼ w1 UCS þ w2 cost ð22Þ
which are as follow:
Dimin Di Dimax ð19Þ J 2 ¼ w1 FS þ w2 cost ð23Þ
where Di represents the ith decision variable with minimum value
(Dimin) and maximum value (Dimax). X
2
wi ¼ 1 ð24Þ
i¼1
Concrete volume constraints
Since several objectives need to be minimised simultaneously in the
The sum of the concrete must add up to one cubic meter as MOO problems, non-dominated solutions (Pareto optimal fronts)
follow: need to be obtained [66]. The Pareto optimal solutions are known
as ‘‘non-dominated” solutions [67], which means no other objective
Q c Q w Q fa Q ca Q SP Q sf
Vm ¼ þ þ þ þ þ ð20Þ functions can be enhanced without deteriorating other objective
U c U w U fa U ca U SP U sf
functions. The Pareto optimum solution is defined as follows [66]:
where Uc, Uw, Ufa, Uca, USP and Usf are the unit weight of cement, If x 2 S whereby S is the set of feasible solutions,
water, fine aggregate, coarse aggregate, superplasticiser and steel
for all i 2 f1; 2; ; kg; f i ðxÞ f i ðx Þ and ð25Þ
fibre, respectively.
Ratio constraints for at least one index j 2 f1; 2; ; kg; f j ðxÞ < f j ðx Þ ð26Þ
Table 3
Unit cost of each variables of SFRC.
Fig. 1. The three objective functions are identified as: the trained
model for UCS and FS and the cost of the SFRC mixture. The sum-
marised steps below are used to find the optimal mixture design
for SFRC:
B < A, whereas for f2, A < B. The set of feasible solutions such as As discussed above, the hyperparameters of SVR were tuned
Point C is not on the Pareto front as it can be easily dominated using FA and 10-fold cross-validation and the optimal fold obtains
by both A and B. the minimum RMSE. It is observed in Figure 8 that the 7th fold
By this weighted sum method, the FA can be extended to MOFA attains the minimum RMSE in the UCS dataset and the 3rd fold
to search for Pareto optimal solutions for multi-objective optimiza- attains the minimum RMSE in the FS dataset. The normalised RMSE
tion problems [68]. The pseudocode is summarized in Fig. 7. versus iteration in the lowest fold for UCS and FS datasets are
depicted in Fig. 9. It can be observed that it takes 11 iterations
2.5.4. Optimal mixture design of SFRC for the RMSE curve to converge for the UCS dataset and 25 itera-
The process for searching the optimal mixture design of SFRC tions on the FS dataset. This implies that the FA successfully found
using SVR and MOFA can be stated as follows and displayed in the best hyperparameters of SVR. The optimal hyperparameters of
Fig. 8. 10-fold cross validation for hyper parameter tuning on the (a) UCS dataset and (b) FS dataset.
Fig. 9. Normalised RMSE score versus iteration in the best fold for (a) UCS and (b) FS.
8 Y. Huang et al. / Construction and Building Materials 260 (2020) 120457
SVR model for UCS prediction are c is 26.89 and c is 24.23, while for
FS prediction are c is 27.45 and c is 49.51.
Fig. 11. Pareto front based on Cost and (a) UCS or (b) FS for SFRC.
Y. Huang et al. / Construction and Building Materials 260 (2020) 120457 9
indicates that an increase in strength is possible for a slight point 6 has the lowest efficiency ratio, while point 20 has the high-
increase in cost; hence, being more economical. est efficiency ratio. Thus, for UCS of SFRC, the most-cost effective
Table 6 and Table 7 show the efficiency ratio for both UCS and UCS value would be at point 6 when UCS is 41.664 MPa for the cost
FS Pareto front points. For the UCS Pareto front as listed in Table 6, of $33.4/m3. Other cost-effective points include point 16 with UCS
being 77.6 MPa and a cost of $35.10/m3 ,as the efficiency ratio is
Table 6 similar to point 6 but allows a higher UCS value for a slight increase
UCS Pareto front efficiency ratio. in cost.
No. of points UCS (MPa) Cost ($) Efficiency ratio For the FS Pareto front as presented in Table 7, point 3 has the
lowest efficiency ratio, while point 20 has the highest. The effi-
1 25.644 33.163 –
2 34.225 33.197 0.0039 ciency ratios for the FS Pareto points are comparatively high. Thus,
3 36.693 33.233 0.0147 for FS of SFRC, the most cost-effective FS value would be at point 3
4 37.687 33.314 0.0819 when FS is 9.058 MPa for the cost of $27.44/m3. If a higher FS value
5 39.560 33.443 0.0687 is sought after, point 18 when FS is 13.035 MPa for the cost of
6 41.662 33.449 0.0030
7 42.433 33.510 0.0786
$62.499/m3 could be considered.
8 43.064 33.532 0.0355 In general, it is up to the decision-maker to decide their budget
9 47.019 33.608 0.0192 and type of strength and type of mechanical property they would
10 49.555 33.858 0.0986 want to achieve, otherwise, the option of using the most cost-
11 52.604 34.190 0.1090
effective point based on the efficiency ratio is available.
12 61.328 34.426 0.0271
13 62.539 34.437 0.0086
14 69.267 34.797 0.0535
15 70.808 34.883 0.0560 3.4. Variable importance
16 77.577 35.098 0.0318
17 84.799 35.550 0.0626
The results from the sensitivity study show the measurement of
18 85.756 35.685 0.1401
19 86.965 36.294 0.5041
influences of the variables for both UCS and FS as shown in Fig. 12.
20 89.726 41.059 1.7264 The input variables for the UCS dataset are shown in Fig. 12 (a), in
which the most influential variable is the SP with the highest influ-
ence score of 1.9746. This is followed by C (1.6189), while UCS is
Table 7
FS Pareto front efficiency ratio. least sensitive to Vf (0.26). Unexpectedly, SP ranks the highest as
an influential variable compared to other cementitious materials.
No. of points FS (MPa) Cost ($) Efficiency ratio
However, this could be due to the data points inputted into the
1 6.292 27.437 - database from experimental journal-based articles. Many of the
2 6.931 27.440 0.005
journal-based articles do state that SP was added to the experi-
3 9.058 27.442 0.001
4 9.175 27.463 0.180 mental mixture design, yet the proportion amount was not given.
5 9.327 27.466 0.018 Hence, the model could potentially only understand the patterns
6 9.667 27.472 0.019 of dataset variables from the journal articles that provided SP pro-
7 9.668 27.473 0.569 portions. The model could have identified that higher UCS values
8 9.686 27.495 1.239
9 9.793 27.515 0.186
mean SP value is known but that is not necessarily the case.
10 9.863 28.258 10.638 Nonetheless, it is not unexpected that the steel fibre input vari-
11 10.069 28.609 1.713 ables (Lf, df and Vf of 0.4404, 0.3584 and 0.26, respectively) are
12 10.325 29.349 2.882 ranked the lowest as it is known that steel fibres only slightly
13 10.430 29.912 5.355
affects UCS in SFRC [70].
14 10.494 30.249 5.342
15 11.139 38.720 13.128 For FS input variables shown in Fig. 12 (b), the most influential
16 11.370 42.477 16.215 variable is the Vf with the highest influence score of 1.9410. This is
17 12.144 52.760 13.297 followed by W (1.2307), while FS is least sensitive to df (0.5776). It
18 13.035 62.499 10.928 is not unexpected that the highest variable is the Vf which has a
19 14.101 81.326 17.667
20 14.171 87.224 83.679
major influence on SFRC as steel fibre has substantial effects on
the tensile/flexural strength of concrete. However, the geometry
Fig. 12. Input variables importance measure for (a) UCS and (b) FS of SFRC mixture design.
10 Y. Huang et al. / Construction and Building Materials 260 (2020) 120457
of the steel fibres (Lf = 0.6825 and df = 0.5776) are not as influential References
as the volume of steel fibres in the concrete. Overall, this means
that the amount of steel fibre plays a more significant role in affect- [1] V. Marcos-Meson, A. Michel, A. Solgaard, G. Fischer, C. Edvardsen, T.L. Skovhus,
Corrosion resistance of steel fibre reinforced concrete - A literature review,
ing the mechanical properties of SFRC. Cem. Concr. Res. 103 (2018) 1–20.
[2] T. Ji, T. Lin, X. Lin, A concrete mix proportion design algorithm based on
artificial neural networks, Cem. Concr. Res. 36 (7) (2006) 1399–1408.
4. Conclusions [3] L. Soufeiani, S.N. Raman, M.Z.B. Jumaat, U.J. Alengaram, G. Ghadyani, P. Mendis,
Influences of the volume fraction and shape of steel fibers on fiber-reinforced
concrete subjected to dynamic loading – A review, Eng. Struct. 124 (2016)
SFRC mixture design is a fundamental yet difficult task for civil 405–417.
engineers. To help with the tough process, this study takes the [4] A.T.A. Dantas, M. Batista Leite, K. de Jesus Nagahama, Prediction of
opportunity to predict the optimised mixture design of SFRC using compressive strength of concrete containing construction and demolition
waste using artificial neural networks, Constr. Build. Mater. 38 (2013) 717–
the FA-SVR based model. This model is trained using a wide- 722.
ranging database. The following conclusions are drawn from this [5] H.-G. Ni, J.-Z. Wang, Prediction of compressive strength of concrete by neural
study: networks, Cem. Concr. Res. 30 (8) (2000) 1245–1250.
[6] M. Açikgenç, M. Ulasß, K.E. Alyamaç, Using an artificial neural network to
predict mix compositions of steel fiber-reinforced concrete, Arab. J. Sci. Eng. 40
1. FA is competent in the tuning of hyper parameters of the SVR (2) (2015) 407–419.
model. The obtained RSME versus iteration curve is able to con- [7] B.B. Adhikary, H. Mutsuyoshi, Prediction of shear strength of steel fiber RC
beams using neural networks, Constr. Build. Mater. 20 (9) (2006) 801–811.
verge within several iterations (for UCS: 11 iterations, for FS: 25 [8] N. Ahn, H. Jang, D.K. Park, Presumption of shear strength of steel fiber
iterations). reinforced concrete beam using artificial neural network model, J. Appl. Polym.
2. The FA-SVR based model can effectively and accurately predict Sci. 103 (4) (2007) 2351–2358.
[9] F. Altun, Ö. Kisßi, K. Aydin, Predicting the compressive strength of steel fiber
both UCS and FS based on performance evaluation on the test
added lightweight concrete using neural network, Comput. Mater. Sci. 42 (2)
set (for UCS, R: 0.91, for FS, R: 0.85). (2008) 259–265.
3. The proposed MOO model can obtain successful Pareto solu- [10] A. Behnood, K.P. Verian, M. Modiri Gharehveran, Evaluation of the splitting
tensile strength in plain and steel fiber-reinforced concrete based on the
tions based on the set constraints for this study but more iter-
compressive strength, Constr. Build. Mater. 98 (2015) 519–529.
ations could be done to produce more Pareto solutions. [11] J. Sun, J. Zhang, Y. Gu, Y. Huang, Y. Sun, G. Ma, Prediction of permeability and
4. The variable importance measure shows that for the most influ- unconfined compressive strength of pervious concrete using evolved support
encing variables are the volume of steel fibres and SP for FS and vector regression, Constr. Build. Mater. 207 (2019) 440–449.
[12] Y. Sun, J. Zhang, G. Li, G. Ma, Y. Huang, J. Sun, Y. Wang, B. Nener, Determination
UCS, respectively. of Young’s modulus of jet grouted coalcretes using an intelligent model, Eng.
Geol. 252 (2019) 43–53.
Overall, this is a promising study on AI application towards [13] K. Yan, C. Shi, Prediction of elastic modulus of normal and high strength
concrete by support vector machine, Constr. Build. Mater. 24 (8) (2010) 1479–
prediction of mechanical properties of SFRC and optimisation of 1485.
mixture design. It is a possible cost-effective and sustainable [14] H. Salehi, R. Burgueño, Emerging artificial intelligence methods in structural
design process that can be employed without the usage of as engineering, Eng. Struct. 171 (2018) 170–189.
[15] J.-S. Chou, C.-F. Tsai, Concrete compressive strength analysis using a combined
much time, resources and labour when compared to traditional classification and regression technique, Autom. Constr. 24 (2012) 52–60.
methods. However, it currently only has the basic input variables [16] K. Duan, S.S. Keerthi, A.N. Poo, Evaluation of simple performance measures for
and a basic dataset used for the application. In the future work, tuning SVM hyperparameters, Neurocomputing 51 (2003) 41–59.
[17] C.-W. Hsu, C.-C. Chang, C.-J. Lin, A practical guide to support vector
more input variables (e.g. types of steel fibres) should be incorpo-
classification, Taipei, 2003.
rated in the dataset to enhance the generalizability of the pro- [18] Y. Yu, C. Zhang, X. Gu, Y. Cui, Expansion prediction of alkali aggregate
posed model. Furthermore, the outcomes are based on reactivity-affected concrete structures using a hybrid soft computing method,
Neural Comput. Applic. 31 (12) (2019) 8641–8660.
experimental results from journal articles. As such, set constraints
[19] M.-Y. Cheng, D. Prayogo, Y.-W. Wu, Novel genetic algorithm-based
could be altered to make it more suitable and reasonable for real evolutionary support vector machine for optimizing high-performance
works in the civil engineering industry. Hence, more investiga- concrete mixture, J. Comput. Civ. Eng. 28 (4) (2014) 06014003, https://doi.
tions must be done to refine the application of the FA-SVR based org/10.1061/(ASCE)CP.1943-5487.0000347.
[20] J.-S. Chou, N.-T. Ngo, A.-D. Pham, Shear strength prediction in reinforced
model and the dataset. concrete deep beams using nature-inspired metaheuristic support vector
regression, J. Comput. Civ. Eng. 30 (1) (2016) 04015002, https://doi.org/
10.1061/(ASCE)CP.1943-5487.0000466.
CRediT authorship contribution statement [21] I. Fister, I. Fister Jr., X.-S. Yang, J. Brest, A comprehensive review of firefly
algorithms, Swarm Evol. Comput. 13 (2013) 34–46.
[22] A. Behnood, E.M. Golafshani, Predicting the compressive strength of silica
Yimiao Huang: Conceptualization, Methodology, Software, fume concrete using hybrid artificial neural network with multi-objective grey
Writing - original draft. Junfei Zhang: Data curation, Supervision, wolves, J. Cleaner Prod. 202 (2018) 54–64.
[23] A. Kandiri, E.M. Golafshani, A. Behnood, Estimation of the compressive
Writing - review & editing. Foo Tze Ann: Data curation, Writing-
strength of concretes containing ground granulated blast furnace slag using
original draft. Guowei Ma: Supervision, Writing - review & editing. hybridized multi-objective ANN and salp swarm algorithm, Constr. Build.
Mater. 248 (2020) 118676.
[24] E.M. Golafshani, A. Behnood, Estimating the optimal mix design of silica fume
Declaration of Competing Interest concrete using biogeography-based programming, Cem. Concr. Compos. 96
(2019) 95–105.
[25] H. Naseri, Cost optimization of no-slump concrete using genetic algorithm and
The authors declare that they have no known competing finan- particle swarm optimization, IJIMT 10 (1) (2019) 33–37.
cial interests or personal relationships that could have appeared [26] W. Abbass, M.I. Khan, S. Mourad, Evaluation of mechanical properties of steel
to influence the work reported in this paper. fiber reinforced concrete with different strengths of concrete, Constr. Build.
Mater. 168 (2018) 556–569.
[27] R.V. Balendran, F.P. Zhou, A. Nadeem, A.Y.T. Leung, Influence of steel fibres on
strength and ductility of normal and lightweight high strength concrete, Build.
Acknowledgements Environ. 37 (12) (2002) 1361–1367.
[28] A. Bhosale, M.A. Rasheed, S.S. Prakash, G. Raju, A study on the efficiency of steel
The work is financially supported by ARC Research Hub for vs. synthetic vs. hybrid fibers on fracture behavior of concrete in flexure using
acoustic emission, Constr. Build. Mater. 199 (2019) 256–268.
Nanoscience-Based Construction Material Manufacturing [29] N. Buratti, C. Mazzotti, M. Savoia, Post-cracking behaviour of steel and macro-
(IH150100006); ARC Discovery Project (DP160100119): Develop- synthetic fibre-reinforced concretes, Constr. Build. Mater. 25 (5) (2011) 2713–
ment of nano-reinforced concrete using boron nitride nanosheets. 2722.
Y. Huang et al. / Construction and Building Materials 260 (2020) 120457 11
[30] J.A. Carneiro, P.R.L. Lima, M.B. Leite, R.D. Toledo Filho, Compressive stress– [54] D.E. Farrar, R.R. Glauber, Multicollinearity in regression analysis: The problem
strain behavior of steel fiber reinforced-recycled aggregate concrete, Cem. revisited, Rev. Econ. Stat. 49 (1) (1967) 92–107.
Concr. Compos. 46 (2014) 65–72. [55] A.J. Smola, B. Schölkopf, A tutorial on support vector regression, Stat. Comput.
[31] K.-K. Choi, D.-O. Ku, Flexural behaviour of amorphous metal-fibre-reinforced 14 (3) (2004) 199–222.
concrete, Proc. Inst. Civ. Eng. Struct. Build. 168 (1) (2015) 15–25. [56] D. Basak, S. Pal, D.C. Patranabis, Support vector regression, Neural Inform.
[32] Ö. Eren, K. Marar, Effects of limestone crusher dust and steel fibers on concrete, Process. Lett. Rev. 11 (10) (2007) 203–224.
Constr. Build. Mater. 23 (2) (2009) 981–988. [57] C.J. Burges, A tutorial on support vector machines for pattern recognition, Data
[33] C. Fu, H. Ye, K. Wang, K. Zhu, C. He, Evolution of mechanical properties of steel Min. Knowl. Disc. 2 (2) (1998) 121–167.
fiber-reinforced rubberized concrete (FR-RC), Compos. B Eng. 160 (2019) 158– [58] S.K. Lahiri, K.C. Ghanta, The support vector regression with the parameter
166. tuning assisted by a differential evolution technique: Study of the critical
[34] D. Gao, L. Zhang, M. Nokken, Mechanical behavior of recycled coarse aggregate velocity of a slurry flow in a pipeline, Chem. Ind. Chem. Eng. 14 (3) (2008) 191–
concrete reinforced with steel fibers under direct shear, Cem. Concr. Compos. 203.
79 (2017) 1–8. [59] S. Boyd, L. Vandenberghe (Eds.), Convex Optimization, Cambridge University
[35] H. Hu, P. Papastergiou, H. Angelakopoulos, M. Guadagnini, K. Pilakoutas, Press, 2004.
Mechanical properties of SFRC using blended manufactured and recycled tyre [60] Y. Xin-She, Firefly Algorithms for Multimodal Optimization, arXiv.org (2010).
steel fibres, Constr. Build. Mater. 163 (2018) 376–389. [61] X.-S. Yang, M. Karamanoglu, 1 - Swarm Intelligence and Bio-Inspired
[36] S. Iqbal, A. Ali, K. Holschemacher, T.A. Bier, Mechanical properties of steel fiber Computation: An Overview, in: X.-.-S. Yang, Z. Cui, R. Xiao, A.H. Gandomi, M.
reinforced high strength lightweight self-compacting concrete (SHLSCC), Karamanoglu (Eds.), Swarm Intelligence and Bio-Inspired Computation,
Constr. Build. Mater. 98 (2015) 325–333. Elsevier, Oxford, 2013, pp. 3–23.
[37] S.J. Jang, G.Y. Jeong, M.H. Lee, K. Rokugo, H.D. Yun, Compressive strength [62] P. Cortez, M.J. Embrechts, Opening black box data mining models using
effects on flexural behavior of steel fiber reinforced concrete, Key Eng. Mater. sensitivity analysis, 2011 IEEE Symposium on Computational Intelligence and
709 (2016) 101–104. Data Mining (CIDM), IEEE, 2011, pp. 341-348.
[38] R. Lameiras, J.A.O. Barros, M. Azenha, Influence of casting condition on the [63] Yuantian Sun, Junfei Zhang, Guichen Li, Yuhang Wang, Junbo Sun, Chao Jiang,
anisotropy of the fracture properties of Steel Fibre Reinforced Self-Compacting Optimized neural network using beetle antennae search for predicting the
Concrete (SFRSCC), Cem. Concr. Compos. 59 (2015) 60–76. unconfined compressive strength of jet grouting coalcretes, Int. J. Numer. Anal.
[39] J.-H. Lee, B. Cho, E. Choi, Flexural capacity of fiber reinforced concrete with a Methods Geomech. 43 (4) (2019) 801–813.
consideration of concrete strength and fiber content, Constr. Build. Mater. 138 [64] Paulo Cortez, Mark J. Embrechts, Using sensitivity analysis and visualization
(2017) 222–231. techniques to open black box data mining models, Inf. Sci. 225 (2013) 1–17.
[40] B. Li, L. Xu, Y. Shi, Y. Chi, Q.i. Liu, C. Li, Effects of fiber type, volume fraction and [65] M.K.A. Ariyaratne, T.G.I. Fernando, S. Weerakoon, Solving systems of nonlinear
aspect ratio on the flexural and acoustic emission behaviors of steel fiber equations using a modified firefly algorithm (MODFA), Swarm Evol. Comput.
reinforced concrete, Constr. Build. Mater. 181 (2018) 474–486. 48 (2019) 72–92.
[41] M.L. Nehdi, M.F. Najjar, A.M. Soliman, T.M. Azabi, Novel steel fibre-reinforced [66] Adil Baykasoğlu, Ahmet Öztasß, Erdoğan Özbay, Prediction and multi-objective
preplaced aggregate concrete with superior mechanical performance, Cem. optimization of high-strength concrete parameters via soft computing
Concr. Compos. 82 (2017) 242–251. approaches, Expert Syst. Appl. 36 (3) (2009) 6145–6155.
[42] L. Nguyen-Minh, M. Rovňák, T. Tran-Quoc, K. Nguyenkim, Punching shear [67] Dobroslav Grygar, René Fabricius, An efficient adjustment of genetic algorithm
resistance of steel fiber reinforced concrete flat slabs, Procedia Eng. 14 (2011) for pareto front determination, Transp. Res. Procedia 40 (2019) 1335–1342.
1830–1837. [68] Xin-She Yang, Multiobjective firefly algorithm for continuous optimization,
[43] M. Nili, V. Afroughsabet, Combined effect of silica fume and steel fibers on the Eng. Comput. 29 (2) (2013) 175–184.
impact resistance and mechanical properties of concrete, Int. J. Impact Eng. 37 [69] M. Klammer, J.N. Dybowski, D. Hoffmann, C. Schaab, Pareto optimization
(8) (2010) 879–886. identifies diverse set of phosphorylation signatures predicting response to
[44] R.S. Olivito, F.A. Zuccarello, An experimental study on the tensile strength of treatment with dasatinib, PLoS One 10 (6) (2015) e0128542.
steel fiber reinforced concrete, Compos. B Eng. 41 (3) (2010) 246–255. [70] D.J. Hannant, 4.11 - Cement-based Composites, in: A. Kelly, C. Zweben (Eds.),
[45] M. Paja˛k, T. Ponikiewski, Flexural behavior of self-compacting concrete Comprehensive Composite Materials, Pergamon, Oxford, 2000, pp. 323–362.
reinforced with different types of steel fibers, Constr. Build. Mater. 47 (2013) [71] Junfei Zhang, Guowei Ma, Yimiao Huang, Farhad Aslani, Brett Nener, Modelling
397–408. uniaxial compressive strength of lightweight self-compacting concrete using
ß ahin, F. Köksal, The influences of matrix and steel fibre tensile strengths on
[46] Y. S random forest regression, Construction and Building Materials 210 (2019)
the fracture energy of high-strength concrete, Constr. Build. Mater. 25 (4) 713–719.
(2011) 1801–1806. [72] Junfei Zhang, Dong Li, Yuhang Wang, Predicting uniaxial compressive strength
[47] A. Sivakumar, M. Santhanam, Mechanical properties of high strength concrete of oil palm shell concrete using a hybrid artificial intelligence model, Journal of
reinforced with metallic and non-metallic fibres, Cem. Concr. Compos. 29 (8) Building Engineering 30 (2020).
(2007) 603–608. [73] Junfei Zhang, Dong Li, Yuhang Wang, Toward intelligent construction:
[48] D.V. Soulioti, N.M. Barkoula, A. Paipetis, T.E. Matikas, Effects of fibre geometry prediction of mechanical properties of manufactured-sand concrete using
and volume fraction on the flexural behaviour of steel-fibre reinforced tree-based models, Journal of Cleaner Production 258 (2020).
concrete, Strain 47 (s1) (2011) e535–e541. [74] Junfei Zhang, Yimiao Huang, Guowei Ma, Junbo Sun, Brett Nener, A
[49] P.R. Tadepalli, Y.L. Mo, T.T.C. Hsu, Mechanical properties of steel fibre concrete, metaheuristic-optimized multi-output model for predicting multiple
Mag. Concr. Res. 65 (8) (2013) 462–474. properties of pervious concrete, Construction and Building Materials 249
[50] M. Taylor, F.D. Lydon, B.I.G. Barr, Toughness measurements on steel fibre- (2020).
reinforced high strength concrete, Cem. Concr. Compos. 19 (4) (1997) 329– [75] Junfei Zhang, Yimiao Huang, Guowei Ma, Brett Nener, Multi-objective beetle
340. antennae search algorithm, arXiv preprint arXiv:2002.10090 (2020).
[51] D.-Y. Yoo, N. Banthia, J.-M. Yang, Y.-S. Yoon, Size effect in normal- and high- [76] Junfei Zhang, Dong Li, Yuhang Wang, Predicting tunnel squeezing using a
strength amorphous metallic and steel fiber reinforced concrete beams, hybrid classifier ensemble with incomplete data, Bulletin of Engineering
Constr. Build. Mater. 121 (2016) 676–685. Geology and the Environment (2020).
[52] D.-Y. Yoo, Y.-S. Yoon, N. Banthia, Flexural response of steel-fiber-reinforced [77] Junfei Zhang, Yimiao Huang, Farhad Aslani, Guowei Ma, Brett Nener, A hybrid
concrete beams: Effects of strength, fiber content, and strain-rate, Cem. Concr. intelligent system for designing optimal proportions of recycled aggregate
Compos. 64 (2015) 84–92. concrete, Journal of Cleaner Production (2020).
[53] X. Zheng, T. Ji, S.M. Easa, B. Zhang, Z. Jiang, Tensile basic creep behavior of [78] Junfei Zhang, Yimiao Huang, Yuhang Wang, Guowei Ma, Multi-objective
lightweight aggregate concrete reinforced with steel fiber, Constr. Build. optimization of concrete mixture proportions using machine learning and
Mater. 200 (2019) 356–367. metaheuristic algorithms, Construction and Building Materials 253 (2020).