Optimization of selection process of constituent materials for high performance concrete and mortars

The paper gives coverage on approach to the optimization of selection process of constituents for high performance concrete (HPC) and mortars. There is shown the example selection of cement, mineral and chemical admixtures, as the most costly and influential on concrete, from the point of their maximum efficiency, measured by flowability and compressive strength increasing indexes. Method combines models for determination of activity indexes with simple stochastic equations obtained at mathematical treatment of experimental results.


Introduction
There exist different guidelines and methods of constituents' selection for concrete and mortars, both regulatory and advisory [1][2][3][4].
Any recommendations require preliminary laboratory batching that usually deals with selection process of constituents types and their rational proportioning.Modern production technologies of HPC and mortars include obligatory using reactive mineral and high range water reducing admixtures (HRWR) or composite admixtures based on them and require rational selection of binders as well as aggregates [5,6].Application of such admixtures aims to reduce cement consumption and viscosity of fresh mixtures, increase strength, improve durability of hardened concrete and mortar [6,7].Cost, energy and resources consumption of admixtures as well as cement are rather high, comparing to aggregates.Therefore, an important aspect of such cement-based systems is judicious choice of constituent materials aimed to obtain economical concrete and mortar [7].
In this connection, important aspect of simultaneous application of different admixtures in cement-based systems is their compatibility with each other and with cement.Testing thermokinetic parameters of cement hydration is an effective method for estimation of compatibility [8].These parameters allow prediction a series of properties for cement and concrete [9].Thus, such method along with its high accuracy requires special equipment.
There are known other parameters suggested for measuring effectiveness of mineral admixtures as well as their combinations with HRWR [10][11][12][13][14]. Thus the parameters, representing combined effect of cement and admixtures on flowability of fresh concrete and concrete strength, are the most adoptable to practical concrete manufacturing conditions [14,15].Deterministic dependencies for calculation of these parameters can be incomplete as they do not take into account some uncertainties.Whereas stochastic equations consider probabilistic distribution of factors values.
Therefore, it was suggested to use these simple deterministic equations and experimental models for estimation of efficiency of HPC and mortars constituent materials.
Comparatively widespread in Ukraine constituents were under consideration: local Portland cement, one of the most common superplasticizers S-3 of naphthalene formaldehyde type and metakaolin, obtained by dehydroxylation of kaolin clays, a highly reactive aluminosilicate mineral admixture.There are massive deposits of kaolin are located in Ukraine as well as several manufacturers of metakaolin [16].
The effect of metakaolin on high performance concrete and mortars is comparable to that of silica fume in terms of strength development rate, chloride ions permeability, drying shrinkage, frost resistance [4,17].At the same time, metakaolin has some advantages comparing to silica fume.It does not darken concrete as silica fume does.Therefore, the architects prefer it.Due to lower specific surface than silica fume, metakaolin has lower water requirement and therefore reduced dosages of high range water reducing (HRWR) admixtures [4].As Ukraine does not have any internal sources of silica fume, it makes metakaolin the only common local reactive mineral admixture.
Most of the researches prove that pozzolanic activity of metakaolin is slightly lower than that of silica fume [4], thus it corresponds to the activity of natural volcanogenic pozzolans and calcined natural pozzolans [18,19].

Aim and scope of the research
The aim of current research was the solving the problem of rational selection of main concrete components (Portland cement, mineral and chemical admixtures as components of composite dry pack modifier) for further investigation of fresh and hardened concrete properties.On a whole, binders and admixtures as reactive components are considered the most influential on concrete and mortars properties.They are also the most expensive components.Therefore, the rational selection of their type and amount is one of the key aspects of concrete and mortar technology.
The results of the research can be spread on other cement-based systems containing mineral additions and chemical admixtures (like building mortars, grouts etc.), mostly they are vital for high performance cement-based systems.

A method for determining the efficiency of cement and admixtures
Due to high water consumption, most of reactive mineral admixtures are effective only at presence of high dosages of HRWR admixtures -superplasticizers, even if they substitute cements partially [4].Therefore, a meaningful efficiency factor is strength increase, obtained by increasing the SP dosage and reduced cement content.
For determination of the efficiency of constituents there has been applied a method developed by prof.V.G.Batrakov at alias and stated in specification for composite concrete additions containing HRWR and mineral admixtures [20].The method was selected due to simplicity of its implementation and unambiguity of the interpretation of received results.
There were prepared mortars of following proportion according to the requirements of specification: 1) control -without any admixtures at weight ratio C(ement):S(and)=1:3; 2) basic -with composite modifier at weight ratio; C(ement):A(ddition):S(and)=0.9:0.1:3where C -cement, S -sand, A -composite addition, consisted of superplasticizer and metakaolin.There was fixed proportion between superplasticizer and metakaolin as 0.1:0.9 in the research.
Water-binder ratio of all basic mortars was equal to control one.Water binder ratio means ratio between water and cement (cement+admixtures) weight.
According to this method, activity indexes of composite admixture consisted of superplasticizer and metakaolin, are following: 1) coefficient of flowability increasing for basic mixtures is determined as follows: ´, where D b -spread diameter of the basic mortar, mm; D c -spread diameter of the control mortar, mm.
2) coefficient of strength increasing of basic mortar is determined as follows: 100, where R s b -compressive strength of the basic mortar, MPa; R s c -compressive strength of the control mortar, MPa.

Materials properties
During research, following materials were used.As basic components there were used local Portland Cement CEM I 52.5 DIN 11641, EN-196 and quartz sand.As mineral component of composite admixture metakaolin was used.HRWR of naphthalene formaldehyde type S-3 was used as chemical component of composite admixture.
As it has been proven by researches [21,22], the application of pozzolanic admixtures is more effective for pure Portland cement without mineral admixtures.The additions presented in cements of types II-V react themselves with clinker minerals, forming additional amount of hydrated compounds.Therefore, it may lead to reduction of the effect of reactive mineral additions like metakaolin.Portland cement.Two kinds of local cement have been under research.Their chemical and mineralogical composition is given in Tab. 1, physical and mechanical properties -in Tab. 2. Portland cement lettered as PC 1 , according to its mineralogical composition is medium alumina cement, which is typical for Ukrainian cement industry.Another Portland cement, lettered as PC 2 , is low alumina cement.Quartz sand.For preparation of mortars there was used quartz sand.Properties of sand are given in Tab. 3. Superplasticizer.According to standard classification ASTM C494 superplasticizer S-3 of naphthalene-formaldehyde type belongs to chemical admixtures of type F waterreducing, high range admixture.
Content of "active substance" in HRWR is not less than 69%, ash content -less than of aqueous solution with concentration 2.5% -7…9.There was also used superplasticizer S-3, modified by defoarming agent PTV (polyethylene vinylene) in the amount 0.01% by superplasticizer weight.Defoarming agent is added for elimination of excessive air-entraining in concrete and mortars, which caused by mineral admixture as well as by high dosages of HRWR, characteristic of high-performance concrete.Such additional compaction of the mixture leads to increasing of strength and durability parameters of concrete.Therefore, such composite admixture solves complex problem of simultaneous high water reducing or reduction of water demand at minimal air-entraining.
Superplasticizer S-3 is lettered as SP 1 , another one modified with defoamer is lettered as SP 2 .
Method of laser granulometry was used for particle-size distribution of metakaolin.Results are given in Tab. 5 and in Fig. 1.Note.d 50 and d 90 -diameters of conventional cell, through which 50% and 90% (vol.) of metakaolin sample pass.
The data above show, that metakaolin under test has particle size up to 60 µ Samples MTK 1 -MTK 3 have substantial volume of particles of 1 µm and 60 µm with medium fraction deficiency (5-10 µm).Sample MTK 4 has higher uniformity of particle size: most of the particles have diameter from 0.5 to 50 µm.As it is shown in Tab.As data reveal, increasing of specific surface of metakaolin leads to its regular increasing of normal consistency and increasing of pozzolanic activity.That is mostly the reason for opening of new active surfaces of metakaolin grains as well as possible surface amorphization during deep grinding [23].

Specimens preparation, storage and testing
Mixes have been prepared according to standard requirements adapted to such mortars with admixtures [24].Error of batching by weight was less than 1%.
Electric mixer was used for mortars preparation.Cement, modifier (where required) and 1/3 of sand were mixed during 1 min.In dry mix water was added, and then mixed during 1 min.Then 2/3 of sand was added and mortar was mixed during 3 min.
Water-binder ratio in control mortar should be sufficient for flow spread diameter equal 113-114 mm.
At the 1 st stage of research have been conducted series of comparative experiments on determining the influence of superplasticizer on air-content of the mortars.SP 1 and SP 2 , Portland cement PC 1 and 2 were used.Composite admixture content was 10% by binder weight.Superplasticizer content was fixed at the level 5, 10 and 15% by composite admixture weight (0.5, 1.0 and 1.5% correspondingly by binder weight).
There were prepared 3 batches for SP 1 and SP 2 with dosages of superplasticizer given above.Gravitational method was applied according to ASTM C 138.For each batch 2 samples were tested.
Each set included 4 batches, with specific samples of metakaolin used in each one at other constant conditions.The mixing proportions were as it was described in part 3.
For each batch spread flow was determined twice and prepared 3 mortar specimens.For each batch, three 70.7 mm edge cubes were prepared.The specimens were kept in the molds during 24 h and after that hardened during 27 days in constant climate chamber with a relative humidity about 100% and temperature of 18±20 0 C. The strength characteristics were measured using a test machine FP 500/1 with a 500 kN load capacity and accuracy d = ±1.0%.Compressive strength at the age of 28 days was tested according to standard requirements [25].

Air-entraining in mortars
Rational air-content makes positive influence on durability of concrete and mortar.According to available data, air content 10-20% resulted in improved durability without causing serious reductions in compressive strength for cement-based mortars [26].At the same time reduced air bubbles content seems to increase the workability and rheology of the mixtures [27].For some types of concrete air entraining is mandatory.In any case attention should be paid to air entraining, since certain properties of supplementary cementing materials used in HPC greatly influence air-void system stability [28].For control mortar air content was V a = 14.1%.Water binder ratio in all the batches was 0.41.
Dependences of air content on SP content can be approximated by parabolic equations: for SP 1 : for SP 2 : Index of air content change can be determined as follows: . .
Comparative dependences of air content of mortars and indexes of air content change are shown in Fig. 2.
As it can be seen, introduction of superplasticizer leads to reduction of air content comparing to control mortar due increasing of flowability of mortars at 25% and at 28% when defoamer is used.Thus increasing of superplasticizer content dramatically reduces its effect.At increasing SP 1 dosage of from 0.5 to 1.5% the air content of the mortar is almost equal to that of control mortar regardless of high increasing of flowability.That confirms known data about intensive air-entraining of this HRWR at high dosages [21,22].It is expected that application of defoaming agent allows obtaining slight decreasing of air content even when high dosages of SP 2 are used and provide higher plastifying effect and higher strength of mortars and concrete.

Activity indexes
Basic mortars had following parameters: makes negative effect on flowability (Tab.7).Two processes, contrary by its effectgrowth of pozzolanic activity and simultaneous normal consistency increasing due to higher specific surface -indemnify each other, therefore under current terms of experiment negative value of flowability coefficient is observed (Tab.7, batches No. 4, 8 and 12).At the same time, its influence on mortar strength is equal to that of MTK 3 .K 4 can be effective from the point of flowability increasing only when more effective HRWR is applied.Therefore, further selection K 1 -K 3 .Dependences of activity indexes of mortars on specific surface s of metakaolin at sÎ[1380; 1800] m 2 /kg for each set of experiments can be approximated by parabolic equations.Depending on Portland cement and superplasticizer used, the equations are following: • for experimental set #1 (PC 1 , SP 1 ): K s = -29(s•10 -3 ) 2 + 116s•10 -3 -97; (7) • for experimental set #2 (PC 2 , SP 1 ): Fig. 3 demonstrates graphical dependences of activity coefficients on specific surface of metakaolin (approximating curves 1 and 2).Comparing the values of activity coefficients for mortars specimens sets #1 and 2 (Fig. 3a, 3b) it should be mentioned that flowability coefficient values are lower for PC 2 based mortars, that is caused by higher water demand of this cement.Strength of the PC 2 based specimens is also slightly lower.That effect can be explai 3 3 in early age.Comparing the results of the experimental sets #1 and 2 it can be seen, that Portland cement PC 1 is more effective than PC 2 as it has lower normal consistency.As it can be seen from comparing results of sets #1 and 3, among two superplasticizers SP 2 is more effective one.Substitution of SP 1 for SP 2 leads to increasing of activity indexes of mortars under other constant conditions (Tab.7,Fig. 3a and c).

Selection of constituent materials
For selection of optimal constituents, there must be established specified values of activity indexes.They were set according to requirements of specifications on such composite admixtures containing HRWR and pozzolanic mineral admixture (silica fume and fly ash), that has similar specific surface range as metakaolin samples #1-3, superplasticizer and mineral admixture weight ratio is 1:9 and is the same as in current research [20].The values are as follows: K f 60%, s 10%.
Graphic approach is applied to demonstrate solution regions of inequalities' sets and determinations values for providing required activity indexes (Fig. 3, shaded areas).Thereby: • for 1 st set of experiments (PC  3a).When PC 2 and SP 1 are applied, simultaneous providing indexes values is impossible (Fig. 3b).There are observed lower activity indexes values.The obtained data meets the results of known researches, which prove that HRWR and mineral admixtures have higher activity at medium alumina cements with low normal consistency [21].Therefore, Portland cement PC 2 is not appropriate for further research.
As it is shown in Fig. 3c, when PC 1 and SP 2 are applied, required values of activity indexes is possible with the range s Î [1410;1680] m 2 /kg.Superplasticizer SP 2 has higher water reducing and strength increasing effects due to additional mortar compaction.Among metakaolin samples MTK 2 meets the requirements on specific surface (s = 1670 m 2 /kg).
Regarding the results of tests and calculations, there were selected following constituent materials: Portland cement PC 1 , superplastisizer SP 2 and metakaolin MTK 2 , as they enable to obtain required activity of composite admixture.

Conclusions
It has been determined activity indexes of composite admixture, which consists of superplasticizer and metakaolin, estimated by increasing the parameters of flowability and strength of mortar.Different factors like type and dosage of constituent materials, specific surface of mineral admixture and properties of cement affect significantly the activity of composite admixture.
By combined solving of simple deterministic and stochastic equations there have been obtained possible solution regions for different combinations of mortar constituents.As a result, cement, chemical and mineral components of composite admixture were selected to meet specified requirements as follows: medium alumina cement, superplasticizer with defoaming agent and metakaolin with specific surface 1670 m 2 /kg.
The sequence of selection procedure is following: 1) ascertainment of quality parameters required (e.g.increasing of flowability and compressive strength) and their limit values; 2) determination of factors of constituent materials (e.g.specific surface of mineral admixture, dosage etc.) and realization of experiment; 3) calculation of quality parameters and obtaining of experimental equations, describing effects of factors on them; 4) determination of optimal values of variables which meet required quality parameters (by solving the set of inequalities for K f and K s ).
The approach can be applied for different types of concrete and mortars, but mostly for high energy and resources consuming high-performance concrete and mortars.
The method, presented in the paper, considers the only two parameters: flowability and compressive strength as the most essential ones.However, similar procedure can be applied for testing frost resistance and other properties attributable to durability of concrete and mortars.As at that rate the task of optimization is multiparametric, it might be supposed, that result will be different.Eligibility of the results obtained for cement-sand mortar should be verified for High Performance Concrete.

3 1 .
In addition, PC 1 has lower value of normal consistency due to its lower specific surface.

Table 1 .
Chemical and mineralogical properties of the cements.% by weight

Table 2 .
Physical-mechanical properties of cement

Table 3 .
Physical properties of quartz sand

Table 4 .
Chemical composition of metakaolin samples

Table 5 .
Particle-size distribution of metakaolin samples

Table 6 .
Physical and chemical properties of metakaolin Sample MTK 4 has highest amount of SiO 2 and Al 2 O 3 (96%) and highest specific surface.Due to high specific surface, this sample demonstrates highest normal consistency and pozzolanic activity.
.41; spread diameter at flow table F b = 114 mm, compressive strength at the age of 28 days R c c = 47.0 MPa; • for PC 2 applied W/C = 0.42; spread diameter at flow table F b = 114 mm, compressive strength at the age of 28 days R c c = 45.6 MPa.Average activity coefficients values are given in Tab. 7. As the K f and K s values indicate, high water demand of metakaolin sample MTK 4

Table 7 .
Activity coefficients of composite admixture