Behavior of Reinjectable and Prestressed Anchors in Soil Masses : Construction Case Study in Congonhas-Brazil

. The present work evaluates the behavior of 26 reinjectable and prestressed anchors executed during the construction of a ground anchored wall in the city of Congonhas, Minas Gerais, with the purpose of stabilizing a slope near a railway of MRS Logística. It refers to a 4.5-m high retaining wall, 30 m in length, and wall thickness of 0.25 m, with three anchors’ lines, totalizing 36 anchors. The anchors were installed in a soil primarily composed of sandy silt with ore. The anchored length was 6.8 m. During the acceptance tests, significant differences in load capacity were verified, despite the geotechnical characteristics being alike. This work aims to analyze the behavior of anchorages, as well as the resistance gain through the increase of injection numbers, and finally the possible causes of the divergence in the results.


Introduction
The use of a retaining wall system for stabilization of slopes is presented as a viable and economical technique for cases in which the re-sloping is impractical.Among the retaining wall techniques, the construction of reinforced concrete walls integrated with reinjectable and prestressed anchors is presented as the most suitable solution for great heights and restricted areas as hillside roads, canals and bridge abutments (Das et al., 2016).This is due to its convenience, rapidity of execution, versatility, and safety.It ensures small deformation compared to other containment techniques, for example, soil nailing.
Despite the widespread use of this technique in Brazil, little research has been done on its implementation method.The knowledge and improvement of this technique mainly come from the experience of contractors, through the implementation and monitoring of the works (Porto, 2015).
Understanding the anchor-soil interface behavior is essential in order to determine the anchor pullout resistance and predict the deformation of the prestressed anchor system in working load terms (Liang & Feng, 2002).It is known, therefore, that the amplitude of anchor load is the factor that governs the performance of the retaining structure (Wang et al., 2016).
Through experimental studies conducted in 2014 by Porto (2015), this research aims to analyze the behavior of anchorages performed in the construction of a retaining wall in Congonhas city, compare the results of acceptance tests on some anchors, and identify possible factors responsible for differences between readings of anchor load capacity.

Ground Anchored Wall
As established by NBR 5629 (ABNT, 2006), the execution process of an anchor wall starts by the vertical cut of the ground, followed by the execution of the inclined hole.The anchor system is installed and grout is inserted into the hole for its stabilization, from the deep end to the entry.Then, the first grout injection phase is performed and, after its cure, the second phase is injected and so on.In this work, the gravity grouting will be referred as "sheath" and the terms "first", "second" and "third injections" refer to pressurized injection phases as shown in Fig. 1.In the sequence of the executive process, the reinforced concrete wall is made.After concrete curing, the anchors are subject to proving, suitability or acceptance tests, according to NBR 5629 (ABNT, 2006), with the purpose of approving or rejecting the anchors through the evaluation of their geotechnical load capacity.

Acceptance Tests
In accordance to NBR 5629 (ABNT, 2006), acceptance tests were performed to control the anchors' behavior and load capacity.The test consists of measuring the head displacement of the anchors, using dial or strain gages while applying tensile loads measured by hydraulic pumpjack-manometer.This test should be done mandatorily in all anchors of a retaining wall.The acceptance test type A should be performed at least on 10% of anchors of a construction and type B on the rest.Test type A is performed on permanent anchors, whose stages of loading and unloading are: F 0 ; 0.3 F t ; 0.6 F t ; 0.8 F t ; 1.0 F t ; 1.2 F t ; 1.4 F t ; 1.6 F t and 1.75 F t , where F 0 is the initial load applied to the anchor through the hydraulic pump-jack-manometer set and corre-sponding to 10% of anchor yield strength.F t is the working load or maximum load that can be applied for which the anchor provides the necessary safety against tendon yielding, anchored length pullout and creep deformations.
In other words, the working load corresponds to the geotechnical load to which the anchor is submitted and under which it should work.After performing full loading and unloading to F 0 , the anchor is directly loaded from F 0 to 1.0 F t .Test type B is performed on permanent anchors too, but stages of loading and unloading are F 0 ; 0.3 F t ; 0.6 F t ; 0.8 F t ; 1.0 F t ; 1.2 F t and 1.4 F t .The displacements that occur due to minor loads F 0 are not measured.Each load increment can only be applied after the stabilization of pressure reading obtained by the manometer (ABNT, 2006).
According to Xiao & Guo (2017), the anchor load is mainly influenced by six factors: the internal friction angle, soil cohesion, wall friction angle, surcharge on the ground surface, anchor dip angle, and depth from the anchor force action point on the wall to the ground surface.
In the acceptance test, the total stabilized displacement must be checked for the maximum load to which the anchor is submitted.Part of this displacement is attributed to elastic elongation of the anchor, in conjunction with per-manent elongation.The elastic displacement is considered only from the unbonded part, while the permanent elongation comes from the bulb elongation.
The displacements limits are presented by NBR 5629 (ABNT, 2006) and Fig. 2 represents a typical test result of the acceptance type A. The lines are calculated using the following variables: displacement, acting force on the test, initial load, unbonded part length, grout bulb length, steel elasticity modulus and steel cross-sectional area.

Geotechnical Features and Design
The work at Congonhas corresponded to the construction of a ground anchored retaining wall for a slope.A railway is located at the slope crest, therefore the need for the retaining wall to provide greater stability to the ground (Fig. 3a).The phases of construction were: terrain cleaning, foundation, excavation, installation of formwork, reinforcement, and wall concreting.For the foundation, root piles were performed, with 2 m spacing and 15 cm diameter.Figure 3b shows the finished ground anchor wall.For this construction, the "sheath" was made by gravity, despite the injection of pressurized grout to increase the pullout resis-  tance by compression of soil due to cavity expansion (Seo et al., 2012).
After the execution of the foundation, the ground was drilled and tendons were installed (CA 50 steel bars with 25°inclination), according to their location planned during design (Figs. 4,5,6 and 7).Then the sheath was made to replace the excavated soil.The next day, the first grout layer was injected (f ck = 25 MPa and water/cement ratio = 0.5) with sufficient pressure to rupture the sheath in order to form the anchor bulb.After the minimum period of seven days after the final injection phase, the anchors were prestressed and tested by acceptance tests and, soon after, the anchor heads were concreted.The lengths of the free and anchored parts were hypothetically defined, using the technical expertise of the design engineer.From these lengths, and results of SPT tests, it was possible to estimate the average N SPT for the location of the anchored part of each tendon.Consequently, the ground anchored wall was divided into three modules, each with three levels of anchors, as shown in Figs. 4, 5, 6 and 7.
The determination of the N SPT at the bulb location is needed to calculate the rupture load of the anchors, which was obtained using mathematical extrapolation from the load data and displacement reached at the acceptance tests.
Four Standard Penetration Test boreholes were made near the slope, as illustrated in Fig. 8. Boreholes one and two had a soil composed basically of clay layers and sandy silt filled with coarse gravel and ore, fine to medium sand, moderately compact and coarse gravel with ore.The soil in the first two layers of boreholes three and four are similar to the ones in boreholes one and two, except for the third hole which is composed of sandy silt with ore, instead of gravel.The geotechnical soil profile and the traces are shown in Figs. 9 and 10.Once the bulbs of this construction had the correct length, it was necessary to evaluate the average N SPT at the geometric center of the anchor bulb, in order to estimate the anchor load capacity (Table 1).
Table 2 presents some additional information on the anchors in the study.To estimate the bulb diameter it was considered an increase of 60% in the hole diameter.(Porto, 2015).
Figure 4 -Overall view of the ground anchored wall (Porto, 2015).

Test Methodology Used in Congonhas
According to Lee et al. (2012), the effect of pressurized grouting is more prominent in soil with N SPT value less than 50 for an increase of bulb diameter and pullout resistance.To ensure this resistance gain, it is necessary that the injection procedure is from bottom to the surface (hole entry).However, the anchors in Congonhas were executed by pressurized injection from the entry (surface) to the bottom, hampering the control of injection pressure by the headline valve.
In total, 36 anchors were installed of which 26 were tested.The test load, according to NBR 5629 (ABNT, 2006) was 600kN and the working load was 340 kN.In the absence of physical rupture, a mathematical extrapolation was made, employing Van der Veen (1953) method.Technical control of the test anchors is shown in Table 3.
However, for unreported reasons, the tests did not reload the anchors to the working load during the last stage of the acceptance test.Consequently, the incorporation of the load was made at a later time.

Results and Discussion
The results of load capacity in the acceptance test were analyzed in terms of the following parameters: number of grout injections and quantity of cement bags per anchor.
Due to the innumerable variables that influence the anchors' behavior, such as anchor tilting angle, injected grout volume, bulb injection pressure, soil variability, an-   chor depth, among others, the comparative analysis was performed in terms of load capacity gain as a function of the average values obtained, since a unit analysis is often not representative from the statistical point of view regarding the average behavior of the anchors analyzed.

Physical rupture
Only three anchors were tested until physical rupture.Figure 11 shows the rupture load evolution as a function of the injections number.Briefly, there is a gain in load capac-  (Porto, 2015).ity as the grout injections number increases, as showed by average values presented in Fig. 11.As discussed earlier, an increase in the geotechnical rupture load is observed as the number of injections increases (for the analyzed anchors).However, this increase does not show a trend.The A12 and A35 anchors presented a resistance gain of 35% from the sheath phase to the first injection phase.Nevertheless, the A27 anchor in the same phase achieved a 74% load gain, showing the great dispersion of results for the same soil type.

Theoretical rupture (mathematical extrapolation)
a) Anchors A02, A03, A05, A16, A17, A18, A19, A31 and A32 (first injection phase) Geotechnical load capacity and the quantity of cement utilized in the first line in the first injection phase are shown in Fig. 12.For this group of tests, 896 kN was the average rupture load obtained, with 24% variability coefficient, 215 kN standard deviation and an average of 72 kN.(Porto, 2015).
Despite the geotechnical characteristics of the ground being very similar, there was a significant dispersion among the results.The load capacity data were obtained by mathematical extrapolation, from the acceptance tests results, as the anchors were not loaded to the limit of the geotechnical load.b) Anchors A12, A27 and A35 (sheath) In Fig. 13, the geotechnical load capacity and the quantity of cement utilized in the anchors' third line at the sheath stage are shown.It was obtained an average rupture load of 485 kN, with 24% variability coefficient, 116 kN standard deviation and an average of 67 kN.
It is important to note that the coefficient of variation is considered low, and therefore acceptable, when less or equal to 30% in a set of homogeneous data (Nogueira et al., 2014), as achieved by our results.c) Anchors A12, A27, A35 and A26 (first injection phase) Figure 14 presents the geotechnical load capacity and the quantity of cement utilized in the anchors' third line, in the first injection phase.For this group of tests, an average rupture load of 1044 kN was obtained, with a 44% variability coefficient, 456 kN standard deviation and an average of 228 kN.Such results show a great dispersion in the measurements, especially for A27, despite the geotechnical ground conditions being very alike.d) Anchors A11, A12, A14, A15, A26, A28, A29 and A35 (second injection phase) Figure 15 presents the geotechnical load capacity and the quantity of cement utilized in the anchors' third line, in the second injection phase.For this group of tests, 1274 kN was the average rupture load obtained, with 44% variability coefficient, 561 kN standard deviation and average of 198 kN.It is noticed that there was a significant dispersion between the results obtained, especially for A35. Figure 11 -Analysis of the influence of the injections number in load capacity (Porto, 2015).

Final discussion
Evaluating the average load capacity values presented in Figs. 13, 14 and 15, it is possible to conclude that there is a significant gain in geotechnical load capacity in the first injection phase (125%) and a new gain of load capacity in the second injection phase, less significant than that of the first phase (50%).
In addition, analyzing the average load capacity shown in Figs. 12 and 14, it was observed an increase in the load capacity of 16.5% in the anchors' third line, in relation to the first line.This fact is possibly due to the fact that the anchors are anchored in a more resistant region of the geotechnical mass.
An increase in the geotechnical load capacity was expected as the amount of cement bag per anchor increased, since, in theory, the diameter of the bulb would increase and, therefore, there would be an increase in the contact between the soil and the bulb.This expectation was not verified in most of the analyzed cases, Figs. 12, 13, 14 e 15.This fact has a number of plausible justifications, for instance: (a) error in the measurement of the quantity of cement bags in the work.Anchorage works are quite dynamic, and the pumping of grout is quite fast.A slight lack of attention of the operator may cause errors in the quantity injected; (b) Loss of grout through soil voids.This fact is common in soft soils or regions with fractured rocks or boulders.Thus, the results presented in terms of quantity of cement bags in the work are inconclusive.However, it is expected that this work will be used as an analysis guideline for future works.

Conclusion
Through the analysis of the influence of grout injection numbers on load capacity, this study evidenced the dispersion of results in terms of load capacity for the same soil type.
To explain these divergent results, it is considered that the bulb diameters have different values, due to the lack of systematic injection control by headline valve and consequently, poor bulb structural control.This fact explains the different load capacity readings.Another reason for the difference of load capacity between the A35 (Fig. 15), for instance, and the other anchors of the third line in the second injection phase would be, probably, the fact that the A35 anchor reached a soil region more resistant than those reached by other anchors (such as boulder or fractured rock).The same occurred with anchor A27 (Fig. 14), whose result was discrepant in relation to anchors A12, A26, and A35 in the first injection phase.
Therefore, it is advisable that the anchorages should be performed with injection control by headline valve, injecting grout from the bottom to the hole entry, thereby the   (Porto, 2015).
bulbs tend to be more uniform.Further research must be done in order to study the anchors' behavior, as well as to further analyze the factors that influence their load capacity in order to obtain a more comprehensive database to facilitate the implementation and improve the quality control of retaining wall construction, providing a more economical and safe construction.
Figure 3 -(a) View of the slope to be retained; (b) View of finished ground anchored wall(Porto, 2015).

Figure 12 -
Figure 12 -First injection phase of the anchors' first line: a) Graph load capacity x tested anchor, b) Number of cement bags (60 kg) used(Porto, 2015).

Figure 13 -
Figure 13 -Sheath of the anchors' third line: a) Graph load capacity x tested anchor, b) Number of cement bags (60 kg) used(Porto, 2015).
Figure 14 -First injection phase of the anchors' third line: a) Graph load capacity x tested anchor, b) Number of cement bags (60 kg) used(Porto, 2015).

Figure 15 -
Figure 15 -Second injection phase of the anchors' third line: a) Graph Load Capacity x Tested Anchor, b) Number of cement bags (60 kg) used(Porto, 2015).

Table 1 -
(Porto, 2015)hip between the N SPT average of the bulb and the type of soil(Porto, 2015).