3.1. Data from the study used as data base to

The work used as data base do apply the AHP was developed by ^{Pereira (2011)} using the following variables:

During the experimental program it was worried to maintain the same conditions of the specimens producing them in the same range of workability, varying in the interval of 260 mm and 300 mm. It was used cylindrical mortar specimens with dimensions of 50 mm, in diameter, and 100 mm, in high, to execute the complementary tests. For the accelerated corrosion test it was made prismatic specimens (60 x 80 x 25 mm) with two steel bars with 5 mm in diameter, 100 mm of length and 10 mm of cover.

After properly dried and cured in a laboratory environment for 15 days, the prismatic specimens were subjected to a corrosion test that consists of half cycles of partial drying and wetting in an aqueous solution with 5% of NaCl, for three months. The half cycles occurred for 5 days and the specimens stayed inside the storage, after 2 days they were submerged on the solution cited.

To execute the accelerated corrosion test it was needed the following equipment: a reference electrode, an electrical junction device, electric wires and a high impedance voltmeter. It was connected the positive pole to the steel bar and the negative pole to the reference electrode to measure the corrosion potential. In this way, the voltmeter indicates the potential difference intensity between the cells, representing a parameter to determine the occurrence of steel corrosion (^{PEREIRA, 2011}).

It is necessary a conductive interface between the reference electrode and the steel bar to close the circuit, and it was made using a saturated sponge placed in electrical contact between the reference electrode and concrete surface. Figure 1 illustrates the configuration of the test for determining the corrosion potential (^{MEDEIROS et al., 2013}).

Figure 1 Circuit configuration for determining the corrosion potential (^{ROCHA, 2012}; ^{MEDEIROS et al., 2013}) 

The aim of the method is not quantify the corrosion of steel, since the corrosion rate is not determined by the procedure. According to ^{Medeiros, Balbinot and Helene (2010)}, the potential for corrosion identifies part in the study sites where the thermodynamic conditions are favorable to the start of the corrosion process, although the armor does not show signs in the concrete surface that surrounds it.

For the assessment of the likelihood of corrosion was used the procedure recommended in the ^{ASTM C 876 (2009)} , and for relating the corrosion potential of the reference electrode copper/ copper sulphate (E_corr) of less than -350 mV , the probability corrosion is higher than 90%. In addition, for E_corr greater than -200 mV, the likelihood of corrosion is less than 10%.

In addition to the accelerated aging test was held the testing of immersion absorption (^{NBR 9778: 1987}), wicking (^{NBR 9779: 1995}) and compressive strength (^{NBR 7215: 1997}) . The author's intention to conduct the testing was to assist in the understanding and interpretation of the main trial.

3.2. Analytic Hierarchy Process

Decision-making for complex problems requires treatment of simplified data, however systematic in order to get results fast, reliable and are inexpensive.

Taking into account factors of importance to the realization of a particular choice, Saaty developed in 1971 the technique of hierarchical analysis (Analytic Hierarchy Process - AHP). The AHP consists of a multi-criteria decision tool addressing factors that are organized in a hierarchical structure (^{SAATY, 1990}).

In civil engineering, the potential of hierarchical analysis has been little explored, but it is possible to cite examples of studies using this tool as a decision making tool, such as: 1 - ^{Marchezetti et al. (2011)}, in the treatment of household waste; 2 - ^{Silva and Souza (2011)}, the check-gatherer solid waste compactor trucks; 3 - ^{Lisboa e Waisman (2006)} and ^{Zayed et al. (2008)}, in decisions related to highway project area; 4 - ^{Pan (2008)}, in the selection of bridge construction methods; 5 - ^{Lai et al. (2008)}, the public works project; 6 - ^{Pereira, Medeiros and Levy (2012)}, for selecting replacement percentage construction waste as aggregate in concrete dosage; 7 - ^{Mattana et al. (2012)}, the choice of coating mortars; 8 - ^{Pereira et al. (2014)}, the choice of adhesive mortars for use in floor system on floor.

The method AHP develops three main operations, which correspond to build a hierarchy of priorities analysis and consistency checks. The construction of the hierarchy is to tease out the component parts of the problem and organize them into levels. As for the priority analyzes carried out a comparison of the elements of each level based on the knowledge and experience of the decision maker. Finally, it is executed consistency checks to ensure consistency among comparisons (^{HO, 2008}).

Thus, before you start implementing the AHP technique it is necessary to define the objective , which in this work is to choose a type of cement a water/cement ratio (w/c) and a time of healing , as previously specified , able to better protect the concrete for corrosion of the reinforcement under the action of chloride ions . So the alternative of choice for this work are shown in Table 1. We analyzed each situation seeking to obtain and compare the best type of cement for each w/c ratio and healing time.

After the definition of choice, shown in Table 1, it has established the criteria to be considered in assessing the performance of alternatives. ^{Pereira (2011)} analyzed the results of three major trials being water absorption, compressive strength and accelerated corrosion, which are the primary level of the hierarchy. The water absorption test includes three other trials arranged on the secondary level of the hierarchy, and accelerated corrosion test also involves test data evolution of the corrosion potential, based on data obtained from the variation of mass of the specimens.

Figure 2 is a tree showing the hierarchical levels analyzed and considered criteria.

Figure 2 Hierarchy tree with the criteria and levels of analysis 

Table 3 shows the values of the first hierarchical level decision matrix indicating the weights of each parameter. As the water absorption and compressive strength tests was considered slightly less important than the accelerated corrosion test by comparing these with the last received values for standardizing the inverse of the data trend.

In preparing the second hierarchical level decision matrix was decided to consider all the criteria considered as equally important. This practice was carried out because it is very difficult and subjective to assign different levels of importance for each of the three results of water absorption. In Table 4 are willing values of the decision matrix of the second hierarchical level.

After developing matrices, decision with the weights of each criterion was calculated consistency AHP through the reasoned consistency test by Saaty (2000) and pointed out by (2002). Thus, we determined the ratio of consistency, it should be a value less than or equal to 0,1. In the case of this study, because of the calculation resulted in 0 (zero) for both developed arrays, being less than the limit. Thus, the AHP is mounted considered perfectly consistent.

In this work, all the data used in the comparison analysis of quantitative criteria are requisites, that is, we used the numerical values extracted from the results of tests performed by ^{Pereira (2011)}, as follows in Table 5.

Table 5 General data for the preparation of AHP 

The test evolution of the corrosion potential was performed with 12 cycles of drying and wetting (in 5% NaCl solution) and total duration of 84 days. For the cement performance evaluation considered the average of all the results only of the phases of wetting, that the presence of water presents the most representative values in relation to the likelihood of corrosion.

4.1. Choice of cement to w/c ratio = 0.4 and curing time = 7 days

After meeting the criteria considered in AHP and establish the weighting factors for comparison of these properties. the appraiser must perform the normalization of data for performance verification. For the calculation divides the value of each item of the line by the larger value. so that all criteria exhibit the same magnitude of values between 0 and 1.

The values of the tests relating to water absorption and corrosion potential needed to be reversed. since the results were the smaller. the better the performance. When converting the data it is possible to standardize analysis by greater values getting coherence in the assessment. In Table 6 are standardized and normalized data of the water absorption test (immersion absorption and capillary absorptivity). The compressive strength and accelerated corrosion (potential corrosion).

The normalized value of each item (Table 6) multiplied by the relative importance of each criterion (Table 4). and also by 100 (percentage). it was possible to achieve the performance levels. An example: Equation (1) describes the procedure for calculating the performance index of the immersion absorption criterion to the cement type CP II, w/c ratio 0.4 and 7 days.

1

Where.

After that. we add up all the performance indices of comparison criteria corresponding to water absorption. The performance indices of the compressive strength criteria and evolution of the corrosion potential are not added for each criterion it is an isolated test. Table 7 shows the performance indices of the criteria.

Finally. the final evaluation is the sum of the performance index multiplied by the relative importance (Table 3) of each property analyzed. The calculation procedure for the final evaluation of the cement type CP II is exemplified by Equation (2).

2

Where.

Table 8 shows the evaluation of each cement type. The best result among the combinations of cement types. w/c ratio and cure times presents the highest value in the final assessment.

For the case study. seeking the best cement type to protect the reinforced concrete to corrosion in chloride ion environment. to w/c equal to 0.4 and cure time of 7 days. the option that presented the best performance using the Analytic Hierarchy Process (AHP) was the cement type CPIII-40. followed by the CPII-Z-32 and CPIV-32.

4.2. Choice of cement to other combinations of w/c ratio and curing time

The calculation procedures stated above for the choice of the type of cement. with w/c ratio equal to 0.4 and 7 days of curing. were the same as those used for the other combinations of w/c ratio and curing time. Thus. in order to simplify the presentation of results in Tables 9. ¹⁰ and ¹¹ are placed data from performance indicators and the final evaluation of other matrices analyzed.

Individually analyzing each test. it was found that the cement CPIII-40 stands out with respect to water absorption performing better as identified with w/c ratio equal to 0.4 with a cure time of 28 days. This result is due to the improvement of Portland cement composite microstructure caused by the extension of the curing period. In relation to the w/c ratio. it influences the amount of voids in mortar. that controls the porosity influencing the properties related to transport mechanisms in concrete (^{MENDES et al.. 2010}).

Using Analytic Hierarchy Process (AHP). regarding the compressive strength it was confirmed which is expected: the compression strength tends to increase with decreasing the water/cement ratio (^{MEHTA e MONTEIRO. 2008}).

It was noted that the CPIII-40 cement showed the highest compressive strength values because of the very high blast slag furnace quantity. ^{Pereira (2011)} showed that such cement has a greater fineness in relation to the other studied. which results in a higher reactivity of the cement.

Evaluating final results of accelerated corrosion tests is noticed again that in all matrices analyzed to choose the best cement type corresponded to CPIII-40. followed by the CPII-Z-32 and CPIV-32. This is in accordance with the statements of ^{Tumidajski and Chan (1996)}. that concretes incorporating blast furnace slag. in partial replacement of Portland cement. are more efficient in preventing the penetration of chloride ions compared to concretes proportionated with ordinary cements.

Finally. it was found that for the samples made with CPIII-40. which was chosen in all matrices analyzed. water/cement ratio values resulted in relatively similar effects. and in favor of the concrete protection by chloride attack. For curing. prolonged time also improved the CPIII-40 in concrete protection.

This work was carried out taking as a database ^{Pereira (2011)} and it should be noted that the AHP application would be more complete if the database includes corrosion rate data and even another indication of rebar corrosion as occurrence of corrosion spots on the surface of the concrete. However. this study shows that AHP is an open system to include other parameters. which can enrich the result of choice.