Soft Computing and Multi-Scal Model Technics to Predict the Early Age Compression Strength of Cement Paste as a Function of Polymer Contents, Water-to-Cement-Ratio, and Curing Ages


 In this study, the effect of two water reducer polymers with smooth and rough surfaces on the compression strength of Ordinary Portland cement (OPC) was investigated. Three different initial ratios between water and cement (w/c) 0.5, 0.6, and 1 were used in this study. The amount of polymer contents varied from 0 to 0.06 % (%wt) for the cement paste with initial w/c of 0.5 and the polymer contents ranged between 0 to 0.16% (%wt) for the cement paste with initial w/c of 0.6 and 1 were investigated. SEM test was conducted to identify the impact of polymers on the behavior of cement paste. The compression strength of OPC cement was increased significantly with increasing the polymer contents. Because of a fiber net (netting) around cement paste particle was developed when the polymers were added to the cement paste which leads to decrease the void between the particles, binding the cement particles, therefore, increased the viscosity and compression strength of the cement rapidly. In this analysis, the hardness of cement paste with polymer contents has been evaluated and modeled using four different model technics. More environmentally sustainable construction, and lower cost than conventional building materials and early age strengths of the cement. To overcome the mentioned matter, this study aims to establish systematic multiscale models to predict the compression strength of cement paste containing polymers and to be used by the construction industry with no theoretical restrictions. For that purpose, a wide data a total of 280 tested cement paste modified with polymers, has been conducted, analyzed, and modeled. Linear, Nonlinear regression, M5P-tree, and Artificial Neural Network (ANN) technical approaches were used for the qualifications. In the modeling process, the most relevant parameters affecting the strength of cement paste, i.e. polymer incorporation ratio (0-0.16% of cement's mass), water-to-cement ratio (0.5-1), and curing ages (1 to 28 days). According to the correlation coefficient (R), mean absolute error and the root means a square error, the compression strength of cement paste can be well predicted in terms of w/c, polymer content, and curing time using four various simulation techniques. Among the used approaches and based on the training data set, the model made based on the Non-linear regression, ANN, and M5P-tree models seem to be the most reliable models. The sensitivity investigation concludes that the polymer content is the most dominating parameter for the prediction of the compression strength of cement paste with this data set.


51
Cement plays an adhesive role in binding materials used in Construction Engineering. Cement is 52 widely used in construction and well-cemented oil fields. Cement alone (neat) can be used as  used. The chemical and mineralogical components of the cement used are shown in Fig. 1.

104
Based on the SEM test the cement particle size was varied between 14.4 m to 42 m as shown 105 in Fig. 2(a).  Table 1. From the SEM test, the polymer 1 had a rough or fibers 110 surface ( Fig. 2(b)) while the polymer 2 had a smooth surface (Fig. 2(c)). The polymer particles 111 have an attractive property to each other and the particles of different materials. Analysis of the composition of chemical substances of cement at 25°C was conducted for X-ray 114 diffraction (XRD). The XRD pattern of particles was obtained by a Siemens D5000 X-ray the cement included C 3 S, C 2 S, C 3 A, C 4 AF, and Quartz, (SiO 2 ) (see Fig. 1). 120

121
An SEM quantum 400 from FEI Company was used in this paper. It is a high-resolution field 122 emission gun scanning electron microscope appropriate for imaging and analysis of nano-scale 123 size. The samples were prepared and put the specimen on the surface of the stamp. The  The purpose of this study is to determine the minimum water mix to measure the initial chemical 135 reaction between water and cement. Cement is one of the products that require the right amount  The cement paste after mixing is lined with cubic molds with a height of (4x 4x 16) cm 3 . The 141 cement paste put into the mold in one layer. After that, the mold is leveled and covered with a   Where: w/c is a water-to-cement ratio, t is curing age (days), and the Polymer content (P, %), 164 respectively, and the parameters of the model are a, b, c, and d.  evaluating each attribute at that node is chosen for node division. Due to the branching method, 190 child node data (subtree or smaller nodes) have less StDev. value. Parent nodes (greater nodes).

191
After reviewing all possible structures, select a system with the highest potential error reduction.

192
This division also creates a large tree-like structure that leads to overfitting. In the second step, (6) 230 yi = tested data; xi = predicted data; ̅ =mean value of yi; and N is the data set. The results for the OPC, polymers, and cement paste, modified with 0,06 percent polymers, were 234 tested using a Scanning Electron Microscope (SEM) at seven days of curing (Fig. 5). Based on 235 the SEM test analysis, the polymers samples were amorphous [63]. Fig. 2(a) showed that the 236 cement particle size has varied particle sizes ranging between 14.4 m to 42 m. Fig.   237 2(b) showed that most of the polymer 1 particles reposed of near-spherical with rough or fibers surfaces. Most of polymer 2 particles also consisted of near-spherical but were very smooth as 239 compared to polymer 1 (Fig. 2(c)).

241
Consistency is called the required amount of water combined to cause the chemical initial 242 reaction between water and cement. The use of polycarboxylate polymer on cement paste limited 243 the water that was needed to reach the required fluidity using mini-slump cone test outcomes.

244
The application of polymers (1&2) and cement decreased the needed water by 3 to 14 percent,  reduced the water-cement ratio by 25% to 32.5% depending on the types of polymer, water-257 cement-ratio, and fluidity (Fig. 3 a). Based on the results of mini-slump cone tests, the addition 258 of polycarboxylate polymer to cement decreased the water needed for the fluidity necessary. The 259 addition of 0.12% polymer decreased the w/c by 38% to 56% based on the types of polymers, 260 w/c, and fluidity shown in Fig. 3(b). can be seen in Fig. 12 Focusing on model parameters Eq.9 and Eq. 10, the maximum influence of polymer contents on 313 increasing compression strength relative to other cement paste compositions can be obtained.

314
Also based on the model parameters in the Eq.9 and Eq. 10 (d=3.37 and d= 3.21), the polymer 1 315 was also more productive than the polymer 2 for enhance compression strength of the cement 316 paste the similar observation was made in the experimental work. Eq. 9 and Eq.10 have also 317 been validated using the testing dataset (Fig. 13). are in ± 30% error lines (Fig. 15). Furthermore, the output of this model is more reliable than The model parameters (a, b, c and d), are listed in Table 3 and based on the linear tree 333 registration function (LM num:) the model variables will be selected. The models (Eq. 11 and 334 Eq. 12) have also been evaluated using the testing dataset as can be seen in Fig. 16.

336
The network was equipped with the training data set, accompanied by the test data to predict the 337 compression strength values for the correct input parameters (Fig. 4). A sensitivity test for the 338 model predictions was also carried out for the cement paste modified with P1 and P2 (Fig. 17).  with the testing dataset in Fig. 17 and Fig.18. Overall, based on the results shown in Table 2   testing. The best performing model is selected for the sensitivity analysis. In this study, the 366 ANN-based model is used for sensitivity analysis. Results obtained from Table 4 indicate that 367 the polymer content is the most influencing parameter for the prediction of compression strength 368 using the M5Ptree-based model.    No data, models, or codes were generated or used during the study.