Investigation of The Effect of Mineral Additives On Concrete Strength Using ANN
Investigation of The Effect of Mineral Additives On Concrete Strength Using ANN
Investigation of The Effect of Mineral Additives On Concrete Strength Using ANN
https://doi.org/10.1007/s42107-022-00431-1
ORIGINAL PAPER
Received: 11 February 2022 / Accepted: 21 February 2022 / Published online: 11 March 2022
© The Author(s) 2022
Abstract
In addition to cement, sand, gravel, and water, the current investigation of the influence of additives on the compressive
strength of concrete at 28 days includes fly ash, silica fume, and slag. 315 concrete compositions with various amounts of
additives are trained and tested using an artificial neural network. Concrete strength is largely affected by the specific gravity
of cement and the specific gravity of fine and coarse particles, according to the studies. For greater compressive strength, it
is preferable to use materials with a higher specific gravity. Compressive strength has grown as the amount of silica fumes
has increased. Increased amounts of slag or superplasticizer resulted in the same behavior. When the amount of fly ash was
increased, the compressive strength of the material decreased.
13
Vol.:(0123456789)
406 Asian Journal of Civil Engineering (2022) 23:405–414
quality. It is frequently important to determine an aggregate's Equation (1) was used in this study for a variable limited
porosity. Pore size is expressed as a percentage of the total to minimum ( pmin) and maximum ( pmax) input values given
material size divided by the size of the pores. It is impossible in Table 4.4, with:
to calculate the volume of pores in any material. The specific
gravity can be used to determine the relationship between the
Δ = pmax − pmin . (2)
pores and bulk solids. The volume of the mixture increases Any new input data should be scaled before being submit-
when cement is replaced by a mineral additive with a lower ted to the network, and the associated predicted values should
density based on mass-to-mass. The specific gravity of be unscaled before being used.
mineral additive is lower than that of cement, which must
be taken into account while proportioning the mix. If the Back‑propagation algorithm
strength and durability qualities of the concrete are main-
tained, the addition of cementations may actually result in a The BPNN is trained using the back-propagation algorithm
reduction in the number of cementations (in terms of mass) (back-propagation neural network). Using the gradient descent
per unit volume of concrete. This is significant in terms of method, this program searches for the smallest error function
making the best use of cementation ingredients in concrete in weight space. A solution to the learning issue is considered
(Jayant, 2013). The efficiency factor for mineral admixture to be a combination of weights that minimizes the error func-
is defined as the quantity of cement in the concrete mixture tion. Hudson et al. (2021) and Hagan et al., (1996) define the
that can be replaced by one part mineral admixture with- algorithm in the following steps:
out affecting the attribute being studied, which is usually
compressive strength. In many technical applications, the 1. It calculates the input to the hidden layer once the input
back-propagation neural network and neural networks with vector is supplied to the input layer., hH , as:
j
genetic algorithm approach has been applied (Iranmanesh & NI
Kaveh, 1999; Kaveh & Iranmanesh, 1998; Kaveh & Rahimi
∑
hH
j = 𝜃j + wji pi , (3)
Bondarabady, 2004; Kaveh and Raiessi Dehkordi, 2003; i=1
Kaveh & Servati, 2001; Kaveh et al. 2001a, 2001b; Kaveh
et al., 2008; Rofooei et al., 2011). where pi reflects the value of the input parameter, 𝜃j represents
the hidden layer's bias function, NI represents the number of
Aims and purposes neuron in the input layer and wji indicates the difference in
weight between the input and hidden layers.
This paper's goals and objectives are to: Each neuron in the hidden layer takes its input, passes it
through a function, and creates an output, which is determined
• Clarify and investigate the influence of the most impor- by
tant parameter on concrete compressive strength. YjH = f (hH ). (4)
• Investigate the impact of substituting additives for cement j
on the compressive strength of concrete. The input to the output layer's neurons, h0k , is now calcu-
• Determine the impact of the concrete mix's specific grav- lated as
ity on concrete strength.
NH
∑
hok = 𝜃k + wkj YjH , (5)
Data preparation for the neural network j=1
(p − pmin )0.8
pnormlize = + 0.1. (1)
Δ
13
Asian Journal of Civil Engineering (2022) 23:405–414 407
Learning an ANN into two groups: the artificial neural network ANN was
trained using 283 concrete mixtures. The accuracy of the
Previous research (Atiş, 2005; Atan & Awang, 2011; Bhik- training mixtures was 94%, as shown in Fig. 1a. ANN
shma & Florence, 2013; Burden, 2006; Duval & Kadri, was put to the test with 32 different concrete composi-
1998; Gesoğlu et al., 2009; Kesharwani et al., 2017; Ren tions. Figures 1 and 2 show that the compressive strength
& Wang, 2014; Babu & Yerramala, 2012; Bhanjaa & Sen- produced from the ANN was 93 percent accurate when
gupta, 2005; Bharatkumar et al., 2005; Lam et al., 1998; compared to earlier experimental research (Fig. 1b). The
Liu, 2010; Özcan et al., 2009; Uygunolu et al., 2012; back-propagation approach is used to train Perceptron
Yazici, 2008; Yazici et al., 2012) as stated in Table 1, the Multilayer Networks. The current study uses the multi-
selection factors were used in the creation of the concrete layer feed-forward back-propagation technique to create
mixtures. The three hundred and fifteen mixes were sorted and train the neural network. The sigmoid transform func-
tion was also utilized, as explained in Al-Gburi, (2015),
Table 1 List of parameters and Parameter Sample Min Max Parametric Unit
their values variation in the studya
experimental work of concrete
mixes and parametric study Cement C 50 600 350 kg/m3
Specific gravity of cement SGC 3.07 3.16 3.16 –
Fly ash FA 0 392 120 kg/m3
Specific gravity of flay ash SGFA 1.99 3.15 2.2 –
Silica fume SF 0 165 80 kg/m3
Specific gravity of silica fume SGSF 1.21 2.32 2.2 –
Slag Slag 0 270 135 kg/m3
Specific gravity of slag SGSlag 2.79 2.95 2.79 –
Super plasticizer SP 0 30.25 10 L/m3
Water W 105 387 200 kg/m3
Fine aggregate S 280 1235 600 kg/m3
Specific gravity of fine aggregate SGFAgg 2.24 2.7 2.6 –
Coarse aggregate G 113 1475 900 kg/m3
Specific gravity of coarse aggregate SGCAgg 2.0 2.85 2.65 –
Max aggregate size MAS 10 25 20 mm
a
Examine the effect of each parameter on its own with increasing cement levels. On the other hand, all of
the other parameters are dependent on the values of this column
13
408 Asian Journal of Civil Engineering (2022) 23:405–414
Al-Gburi et al., (2012), Yousif and Al-Jurmaa (2010). The the findings in the study provided in Table 1. The cement
ANN model was built with one hidden layer. There were content C and the SGCAgg are followed by an SGFAgg and
fifteen input parameters and one output parameter in the then the cement content C and the SGCAgg. The rest of
ANN model. The available data sets were used for the the additive factors are in low effect ratios, as indicated in
training, testing, and validation processes. Fig. 2. Concrete strength is influenced mainly by the SGC
and aggregates (fine and coarse). As a result, the remain-
ing mineral additions have a minimal effect on compressive
Relative importance strength. Additives can be used to replace high-cost cement
without affecting the strength of the concrete.
Although we can use techniques like multiple regression or
discriminant analysis to evaluate neural network prediction
or classification success rates, no method that analyzes the Results
relative relevance of the input parameters employed by the
network to reach its conclusions has been accepted. One of Effect of specific gravity of cement
the most advantages of ANN-based models is that they make
sensitivity analysis easier to analyze the relative importance As seen in Fig. 3, increasing the cement's specific gravity
of their input variables (Kim & Ahna, 2009). One of the increased the concrete's compressive strength. Depending
most interesting properties of ANN-based models is that on the amount of moisture in the cement, the specific grav-
they make sensitivity analysis of the relative relevance of ity might rise or fall. The greater the specific gravity, the
their input variables more easier (Kim & Ahna, 2009). To more water is required. It will result in a lower quantity of
determine the most significant input–output relationship water in the cement, enhancing the compressive strength.
that has been manually carried out, sensitivity studies use Furthermore, the higher cement concentration resulted in an
a "leave one out" technique. It is used to offer information increase in concrete compression strength, which is consist-
on the relative importance of the input factors in each geo- ent with Salem and Pandey's findings (2015).
metrical dimension's parameter on compressive strength, see
(Bharatkumar et al., 2005; Lam et al., 1998; Mustafa, 2018).
In this work, the relative importance of the input parameters Effect of specific gravity of fine aggregate
is determined using the partitioning weights approach estab-
lished by Garson (1991) and later adapted by Goh (1995). As illustrated in Fig. 4, increasing the specific gravity of
The weights of the connections between ANN layers are fine particles reduced concrete strength. Maina et al. (2018).
used in this method to calculate the maximum change in out- discovered the similar perception in behavior. The strength
put caused by changes in specific input variables. To deter- behavior changed as the cement content was increased to
mine the relative importance of the various input parameters, over 450 kg/m3. With an increase in the specific weight of
this approach is interested in the connection weight between fine aggregate, the compressive strength increased.
ANN layers.
The SGC is the most influential factor in the sensitiv-
ity to the compressive strength of concrete, according to
13
Asian Journal of Civil Engineering (2022) 23:405–414 409
The concrete strength is influenced by the specific grav- When silica fume is added to a concrete mix with a little
ity of coarse material. When the specific gravity of the amount of cement, the compressive strength is reduced (by
coarse aggregate was the highest, the compressive strength a significant amount). Figure 6 shows how the influence
increased, as shown in Fig. 5. Al-Oraimi et al., 2006; Ryu & of silica fume changes as the cement content is increased,
Monteiro, 2002 both found the same indication. resulting in increased concrete strength. Atiş (2005), Babu
13
410 Asian Journal of Civil Engineering (2022) 23:405–414
and Yerramala (2012), Tanyildizia and Evik (2010), and of fly ash, on the other hand, increased concrete strength, as
Uygunolu et al. (2012) demonstrate the same indication shown in Fig. 9. This behavior is consistent with the find-
of the impacts of silica fume alone on the concrete mix. ings of Jayant (2013), Bhanjaa and Sengupta (2005), Yazici
The specific gravity of silica, on the other hand, is lower (2008) and Atiş (2005).
than the specific gravity of cement. As a result, replacing
the same amount of cement with the same amount of silica Effect of slag
reduces the strength of the concrete. Previous research has
concentrated on the effect of varied amounts of silica fume As seen in Fig. 10, adding more slag to the mix improved
on concrete strength while ignoring the effect of silica fume the concrete's strength. These findings are consistent with
specific gravity. The specific gravity of silica fume is the Akcaozoglu and Atis (2011), Bharatkumar et al., (2005).
most influential, as shown in Figs. 2 and 7. Concrete strength Most standards limit the amount of BFS added to slag
increased as the specific weight of the concrete increased. cement to 70%, see (Jayant, 2013). Furthermore, increasing
the fineness of GGBS might result in higher compressive
Effect fly ash (FA) strength.
13
Asian Journal of Civil Engineering (2022) 23:405–414 411
13
412 Asian Journal of Civil Engineering (2022) 23:405–414
resulting in increased compressive strength. The artificial Al-Gburi, M. (2015). Restraint effects in early age concrete structures.
neural network's results were found to match the analysis of Doctoral Thesis, Luleå University of Technology, Luleå, Sweden
(p. 201). http://www.diva-portal.org/smash/get/diva2:991716/
previous experimental results. As a result, using an artificial FULLTEXT01.pdf
neural network to understand the behavior of complicated Al-Oraimi, S. K., Taha, R., & Hassan, H. F. (2006). The effect of the
data without having to do costly practical testing is critical. mineralogy of coarse aggregate on the mechanical properties of
high-strength concrete. Construction and Building Materials,
20(7), 499–503.
Alsadey, S. (2012). Influence of superplasticizer on strength of con-
Funding Open access funding provided by Lulea University of Tech- crete. International Journal of Research in Engineering and
nology. The authors have not disclosed any funding. Technology (IJRET), 1(3), 164–166. https://www.researchgate.
net/publication/285816594_Influence_of_super plasticizer_on_
Declarations strength_of_concrete
ASTM C128-15. (2015). Standard test method for relative density (spe-
Conflict of interest The authors have not disclosed any competing in- cific gravity) and absorption of fine aggregate. https://www.astm.
terests. org/Standards/C128
Atan, M.N., & Awang, H. (2011). The compressive and flexural
strengths of self-compacting concrete using raw rice husk ash.
Open Access This article is licensed under a Creative Commons Attri- Journal of Engineering Science and Technology, 6(6), 720–732.
bution 4.0 International License, which permits use, sharing, adapta- https://jestec.t aylors.edu.my/Vol%206%20Issue%206%20Dec
tion, distribution and reproduction in any medium or format, as long ember%2011/Vol_6_6_720_732_MD%20NOR%20ATAN.pdf
as you give appropriate credit to the original author(s) and the source, Atiş, C. D. (2005). Strength properties of high-volume fly ash roller
provide a link to the Creative Commons licence, and indicate if changes compacted and workable concrete, and influence of curing condi-
were made. The images or other third party material in this article are tion. Cement and Concrete Research, 35(6), 1112–1121.
included in the article's Creative Commons licence, unless indicated Babu, K. G., & Yerramala, A. (2012). Strength properties of high
otherwise in a credit line to the material. If material is not included in workable and high volume fly ash roller compacted concrete.
the article's Creative Commons licence and your intended use is not Journal of Engineering Research and Studies JERS, 3(3),
permitted by statutory regulation or exceeds the permitted use, you will 11–17.
need to obtain permission directly from the copyright holder. To view a Bhanjaa, S., & Sengupta, B. (2005). Influence of silica fume on the
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. tensile strength of concrete. Cement and Concrete Research,
35(4), 743–747.
Bharatkumar, B. H., Raghuprasad, B. K., Ramachandramurthy, D.
S., Narayanan, R., & Gopalakrishnan, S. (2005). Effect of fly
ash and slag on the fracture characteristics of high performance
References concrete. Material and Structures, 38, 63–72. https://doi.org/
10.1007/BF02480576
Akcaozoglu, S., & Atis, C. D. (2011). Effect of granulated blast furnace Bhikshma, G., & Florence, A. (2013). Studies on effect of maximum
slag and fly ash addition on the strength properties of lightweight size of aggregate in higher grade concrete with high volume
mortars containing waste pet aggregates. Construction and Build- fly ash. Asian Journal Of Civil Engineering (BHRC), 14(3),
ing Materials, 25(10), 4052–4058. 101–109. https://www.researchgate.net/publication/266501960_
Al-Gburi, M., Jonasson, J.-E., Yousif, S.T., & Nilsson, M. (2012). Sim- Studies_on_effect_of_maximum_size_of_aggregate_in_higher_
plified methods for crack risk analyses of early age concrete part grade_concrete_with_high_volume_fly_ash
2: restraint factors for typical case wall-on-slab. Nordic Concrete Burden, D. (2006). The durability of concrete containing high levels
Research, 46(2), 39–58. https://www.diva-portal.org/smash/get/ of fly ash. MSc Thesis, The University of 24 New Brunswick,
diva2:975530/FULLTEXT01.pdf New Brunswick. https://trid.trb.org/view/795859
13
Asian Journal of Civil Engineering (2022) 23:405–414 413
Dumne, S. M. (2014). Effect of superplasticizer on fresh and hard- Kesharwani, K. C., Biswas, A. K., Chaurasiya, A., & Rabbani, A.
ened properties of self-compacting concrete containing fly (2017). Experimental study on use of fly ash in concrete. Inter-
ash. American Journal of Engineering Research (AJER), 3(3), national Research Journal of Engineering and Technology, 4(9),
205–211. https://www.ajer.org/papers/v3(3)/ZA33205211.pdf 1527–1530. https://www.researchgate.net/publication/32085
Duval, R., & Kadri, E. H. (1998). Influence of silica fume on the 3091_Experimental_Study_on_Use_of_Fly_Ash_in_Concrete
workability and the compressive strength of high-performance Kim, J., & Ahna, H. (2009). New perspective for neural networks:
concretes. Cement and Concrete Research, 28(4), 533–547. Application to a marketing management problem. Journal of
Garson, G. D. (1991). Interpreting neural network connection Information Science and Engineering, 25, 1605–1616. https://
weights. Artificial Intelligence, 6, 47–51. https://www.seman www.researchgate.net/publication/220587367_A_New_Persp
ticscholar.org/paper/Inter preting-neural-network-connection- ective_for_Neural_Networks_A pplic ation_to_a_M arket ing_
weights-Garson/d5e392035d5f4b1ae37027cabfd1bfdf6733015b Management_Problem
Gesoğlu, M., Güneyisi, E., & Özbay, E. (2009). Properties of self- Lam, L., Wong, Y. L., & Poon, C. S. (1998). Effect of fly ash and silica
compacting concretes made with binary, ternary, and quaternary fume on compressive and fracture behaviors of concrete. Cement
cementitious blends of fly ash, blast furnace slag, and silica and Concrete Research, 28(2), 271–283.
fume. Construction and Building Materials, 23(5), 847–185. Liu, M. (2010). Self-compacting concrete with different levels of pul-
Goh, A. T. C. (1995). Back-propagation neural networks for mod- verized fuel ash. Construction and Building Materials, 24(7),
eling complex systems. Artificial Intelligence in Engineering, 1245–1252.
9(3), 143–151. Maina, K., Gwaya, A. O., & Koteng, D. O. (2018). Effects of different
Hagan, M. T., Demuth, H. B., & Beale, M. H. (1996). Neural net- fine aggregate on concrete strength. International Journal of Soft
work design (2nd edn). Boston, MA: PWS Publishing. https:// Computing and Engineering, 8(2), 16–21. https://www.ijsce.org/
hagan.okstate.edu/NNDesign.pdf wp-content/uploads/papers/v8i2/B3133058218.pdf
Hudson, B., Hagan, M., & Demuth, H. (2021). Neural Network Tool- Mustafa, T. (2018). Cement and concrete mineral admixtures (p. 318).
box for Use with MATLAB, User’s Guide, The Math works. https://www.adlibris.com/se/bok/cement-and-concrete-mineral-
https://se.mathworks.com/help/pdf_doc/nnet/nnet_ug.pdf/ admixtures-9780367028008
Iranmanesh, A., & Kaveh, A. (1999). Structural optimization by gra- Özcan, F., Atiş, C. D., Karahan, O., Uncuoğlu, E., & Tanyildizi, H.
dient base neural networks. International Journal of Numerical (2009). Comparison of artificial neural network and fuzzy logic
Methods in Engineering, 46, 297–311. models for prediction of long-term compressive strength of
Jayant D. B. (2013). Mineral admixtures in cement and concrete (1st silica fume concrete. Advances in Engineering Software, 40(9),
edn, p. 309). https://www.routledge.com/Cement-and-Concrete- 856–863.
Mineral-Admixtures/Tokyay/p/book/9780367028008#:~:text= Ren, Z., & Wang, L. (2014) Research of high fly ash content in con-
Cement%2 0and%2 0Conc rete%2 0Mine ral%2 0Admi xture s%2 0con crete with dipy construction formwork. In International Work-
centrates%20mostly%20on%20natural%20pozzolans,and%20sha shop on Sustainable Development and Concrete Technology (pp.
les%20are%20also%20discussed 347–359). https://citeseerx.ist.psu.edu/viewdoc/download?doi=
Kaveh, A., & Servati, H. (2001). Design of double layer grids using 10.1.1.475.6788&rep=rep1&type=pdf
back-propagation neural networks. Computers and Structures, Rofooei, F. R., Kaveh, A., & Masteri Farahani, F. (2011). Estimating
79, 1561–1568. https://www.infona.pl/resource/bwmeta1.eleme the vulnerability of concrete moment resisting frame structures
nt.elsevier-e2fafe86-2f88-3a43-b17a-a7c7b088a052 using artificial neural networks. International Journal of Opera-
Kaveh, A., Elmieh, R., & Servati, H. (2001a). Prediction of moment- tional Research, 1(3), 433–448. https://w ww.s emant icsch olar.o rg/
rotation characteristic for semi-rigid connections using BP neu- paper/E STIMA TING-T HE-V
ULNER ABILI TY-O F-T HE-C ONCR
ral networks. Asian Journal of Civil Engineering, 2(2), 131–142. ETE-MOMENT-Rofooei-Kaveh/faa4dcd631c3eaef1e912e6b0
https://www.sid.ir/en/Journal/ViewPaper.aspx?ID=40312 77992b8022ac29d
Kaveh, A., Fazel-Dehkordi, D., & Servati, H. (2001b). Prediction of Ryu, J.-S., & Monteiro, P. J. M. (2002). Effect of specific gravity of
moment-rotation characteristic for saddle-like connections using coarse aggregate on interfacial transition zone, permeability, and
BP neural networks. Asian Journal of Civil Engineering, 2(1), strength of concrete. British Ceramic Transactions, 101(1), 30–34.
11–30.https://www.sid.ir/en/Journal/ViewPaper.aspx?ID=39305 https://doi.org/10.1179/096797801225000806
Kaveh, A., Rahimi Bondarabady, H. A. (2004). Wavefront reduction Salem, M. A., & Pandey, R. K. (2015). Influence of specific gravity
using graphs, neural networks and genetic algorithm. Inter- on weight of proportions of concrete. International Journal of
national Journal for Numerical Methods in Engineering, 60, Engineering Research & Technology (IJERT), 4(02), 319–320.
1803–1815. https://www.academia.edu/16715990/Wavefront_ https://www.ijert.org/research/influence-of-specific-gravity-on-
reduct ion_u sing_g raphs_n eural_n etwor ks_a nd_g eneti c_a lgori thm weight-of-proportions-of-concrete-IJERTV4IS020317.pdf
Kaveh, A., Gholipour, Y., & Rahami, H. (2008). Optimal design of Siddique, R. (2003). Effect of fine aggregate replacement with class
transmission towers using genetic algorithm and neural networks. f fly ash on the mechanical properties of concrete. Cement and
International Journal of Space Structures, 23(1), 1–19. https:// Concrete Research, 33(4), 539–547.
doi.org/10.1260/026635108785342073 Tanyildizia, H., & Çevik, A. (2010). Modeling mechanical perfor-
Kaveh, A., & Iranmanesh, A. (1998). Comparative study of backpropa- mance of lightweight concrete containing silica fume exposed to
gation and improved counterpropagation neural nets in structural high temperature using genetic programming. Construction and
analysis and optimization. International Journal of Space Struc- Building Materials, 24(12), 2612–3261.
tures, 13, 177–185. https://doi.org/10.1177/026635119801300 Uygunoğlu, T., Topcu, I. B., Gencel, O., & Brostow, W. (2012). The
401m effect of fly ash content and types of aggregates on the properties
Kaveh, A., & Khalegi, H. A. (2000). Prediction of strength for concrete of pre-fabricated concrete interlocking blocks (PCIBs). Construc-
specimens using artificial neural network. Asian Journal of Civil tion and Building Materials, 30, 180–187.
Engineering, 2(2), 1–13. Yazici, S., & Arel, H. S. (2012). Effects of fly ash fineness on the
Kaveh, A., & RaiessiDehkordi, M. (2003). RBF and BP neural net- mechanical properties of concrete. Indian Academy of Sciences,
works for the analysis and design of domes. International Jour- 37(3), 389–403. https://www.ias.ac.in/article/fulltext/sadh/037/
nal of Space Structures, 18(3), 181–194. https://doi.org/10.1260/ 03/0389-0403
026635103322437463
13
414 Asian Journal of Civil Engineering (2022) 23:405–414
Yazici, H. (2008). The effect of silica fume and high-volume Class C C.-b eams-w ith-F RP-A l/a 836b7 a3ed6 8704b b431f 834db f6ad
fly ash on mechanical properties, chloride penetration and freeze– 7d0048167a
thaw resistance of self-compacting concrete. Construction and
Building Materials, 22(4), 456–462. Publisher's Note Springer Nature remains neutral with regard to
Yousif, S. T., & Al-Jurmaa, M. A. (2010). Modeling of ultimate load jurisdictional claims in published maps and institutional affiliations.
for RC beams strengthened with carbon FRP using artificial neural
networks. Al-Rafidain Engineering, 18(6), 28–41. https://www.
semanticscholar.org/paper/Modeling-of-ultimate-load-for-R.
13