An Evaluation of the Shaft Resistance of Piles Embedded in Gneissic Rock

In the design of piles drilled in rock, the following questions arise: (i) at what point of the soil/weathered rock/sound rock profile should the pile socket be designed; (ii) is the contribution of residual soil to be disregarded; (iii) how much consideration should be given to the pile boring method (rotary or hammering). Furthermore, usual design methods consider only the side shear capacity of the socket, which is evaluated through empirical expressions that require the uniaxial compressive strength of the intact rock (qu or UCS). And, quite often in practice, a comprehensive test program is not available, and only boring logs are available. This paper examines data from a BRT (Bus Rapid Transit) project in Rio de Janeiro, with 8 bridges, in which some 30 dynamic tests were performed on piles partly embedded in residual soil and partly in rock a gneiss. These tests produced profiles of mobilized side shear. For the evaluation of the mobilized side shear, a series of laboratory tests were performed on rock samples with different RQDs, taken from borings at the pile sites. A relation between qu (uniaxial compressive strength of the intact rock) and RQD could be established for gneissic rocks of Rio. Values of mobilized side shear are compared to qu derived from the RQD correlation. Finally, an expression for the prediction of mobilized pile shaft capacity is put forward.


Introduction
Pile foundations drilled in rock are required for heavy structures and/or when rock occurs at relatively shallow depths.Although a straightforward solution, the design engineer often faces questions such as (i) at what point of the soil/weathered rock/sound rock profile should the pile socket be designed; (ii) is the contribution of residual soil to be disregarded; (iii) how to proceed in the -not uncommon -situation in which only boring data, such as rock classification and RQD, is available; (iv) how much consideration should be given to the boring method.The last question arises from the fact that most common boring methods are either (a) water assisted ("wet") rotary drilling or (b) compressed-air assisted ("dry") down-the-hole hammer drilling.
Usual design methods consider only the shaft capacity of the rock socket -through side shear or skin resistance -which is evaluated through empirical expressions.Such expressions have as starting point the uniaxial compressive strength of the intact rock (q u or UCS).However, it is not uncommon in practice that a laboratory test program, which covers the extension of the work, is not available and design has to be based on boring data (rock classification and RQD).
Several bridges had to be built for a BRT project in Rio de Janeiro, and most of the foundations were piles in rock.In the Rio de Janeiro region, the usual rock is gneiss, with occasional granite occurrences.The city is known for its rock outcrops, and, in most of the flatter areas, rock is found under a cover of weathered material, usually a few meters thick.In this project, as usual in the city, piles were embedded partly in residual soil and partly in rock.As part of the quality control of the project, 30 dynamic pile tests were performed.
The dynamic tests produced a profile of mobilized side shear for each pile.For the evaluation of the mobilized side shear (as a function of the uniaxial compressive strenght), a series of laboratory tests were performed on rock samples with different RQDs, taken from borings at the pile sites.A relation between uniaxial compressive strength of the intact rock (q u ) and RQD could be established for gneissic rocks in the Rio de Janeiro region.Values of mobilized side shear are compared with the uniaxial compressive strengths of the intact rock (q u ) derived from the RQD correlation.Finally, an expression for the prediction of mobilized pile shaft capacity is put forward.clearing of the bottom of the borehole is not ensured, pile base contribution (Q b ) is not considered.Furthermore, since the displacement necessary for full mobilization of the pile shaft shear resistance in rock (t rock ) is much smaller than that for the overlying soil (t soil ), soil contribution (Q s,soil ) is considered with restriction.Therefore, rock-socketed piles derive most of their bearing capacity from the shaft shear resistance in rock (Q s,rock ).
According to Goodman (1989), when concrete is poured against a drilled rock surface, it develops a strong bond, which can carry shear stresses up to the shear strength of rock or of the concrete, whichever is smaller.
Most design methods derive from a comparison of the rock compressive strength (q u ) measured in laboratory tests with the pile shaft shear resistance (t max ) observed in field or model tests.Laboratory tests are performed on rock specimens that are homogeneous and have no joints and, therefore, some correction has to be made to extend the laboratory measured strength to the strength of the rock mass.
The pile/rock interface (or side shear) resistance is usually predicted from the rock compressive strength through an expression such as: Eq. 1 can be expressed in normalized form by dividing both unit side shear resistance and compressive strength by atmospheric pressure (p atm = 0.1013 MPa): where a (and a') and b (and b') are empirical factors, while q u is the uniaxial compressive strength of the intact rock.When using Eq. 1 in MPa, values for a in the range 0.2-0.8(upper limit for very rough or grooved boring surfaces) and for b in the range 0.6-0.8 have been suggested by Horvath (1978), Meigh & Wolski (1979), Pells et al. (1980), Rowe & Armitage (1987), Zhang (1997), Zhang & Einstein (1998).In the case of Eq. 2, values for a' in the range 0.8-2.5 (upper limit for very rough or grooved surfaces) and b' in the range 1.0-2.5 have been suggested by Rosenberg & Journeaux (1976), Horvath & Kenney (1979), Mcvay et al. (1992).
As mentioned earlier in this item, the pile shaft shear resistance is limited by the shear resistance of the concrete, which can be estimated as: where f ck is the characteristic (compressive) strength of the concrete.
The above expression is based on laboratory tests of concrete joints conducted at the Federal University of Rio de Janeiro.The strength envelope indicated that the shear strength at a concrete joint under no compressive stress is very close to the tensile strength of the concrete.In the upper part of the pile, normal stresses are low and the assumption of these conditions for the whole shaft is on the safe side.

BRT Project in Rio de Janeiro
A new BRT (Bus Rapid Transit) line, named Transcarioca, connects the borough of Barra da Tijuca to Rio de Janeiro International Airport (Maestro Tom Jobim).As it approaches the airport, a series of 6 bridges/viaducts were required, as shown in Fig. 2. Figures 3 and 4 show OAE-02/03 and OAE-06 in perspective.
The region, as typical of Rio de Janeiro, presents gneisses at relatively shallow depths.Therefore, foundations for most bridge columns were cast-in-place piles excavated through soil and penetrating rock.
With the aim of both quality control and of acquiring a better knowledge of rock-socketed pile behaviour, 30 (high strain) dynamic load tests were carried out.
Piles partly embedded in rock are commonly installed by driving a steel casing down to the top of the bedrock and then boring the socket.The pile is completed by placing the reinforcement (usually a "cage") and pouring a self-compacting concrete.The steel casing is driven by a hammer or forced down by a rotary equipment that has a pull-down force and the interior of the casing is cleared from soil (usually washed or removed by auger) before introducing the boring tool.Boring methods usually fall in two categories: (a) rotary drilling, with rock debris removed by water circulation (sometimes referred to as "wet" or Wirth drilling), and (b) use of down-the-hole (DTH) hammer, with debris removed by compressed-air ("dry" boring).The steel casing is usually lost and becomes part of the pile (as in this work) but, in some situations, the casing can be retrieved with the help of a vibrator.
Piles types for the BRT Project, their service (or working) loads and rock drilling equipment can be seen in Table 1.
The "dry" hammer boring proved much more efficient than the "wet" rotary drilling.Since compressed-air lifting of rock debris is restricted to a limited depth (typically 15 m), "dry" hammer boring was used on land and the "wet" rotary drilling on water.Blows for pile diameter 50 cm (socket diam.39 cm) were delivered by a 40 kN weight, falling from heights that varied from 15 cm to 75 cm.For pile diameter 80 cm (socket diam.70 cm), a 61.2 kN weight, falling from 15 cm to 160 cm, was used.In the case of the 80 cm pile, the hammer is somewhat lighter than what would be desirable (Rausche et al., 2006, suggest a ram weight of at least 1% of the test load for piles in rock).It should be mentioned that the dynamic tests were hired by the Contractor without interference from the authors.

Criteria followed in the definition of pile length in the design stage
In the definition of pile length in the Design Stage, the following criteria were followed: (i) the pile socket was considered starting at the point where, in a mixed boring, percussion driving (SPT procedure) was not possible and had to change to rotary driving; at this point material classification conventionally changes from residual soil to weathered rock; (ii) the contribution of residual soil to pile capacity was disregarded (only the rock socket was considered); (iii) in the lack of laboratory tests, rock resistance was estimated from boring data, available at each bridge column, basically rock classification and RQD.
The rock -always a gneiss -unconfined compression resistance (q u ) was estimated as: This relation, adopted by the design firm, was intended to be on the safe side for use in preliminary design, and indicates a maximum q u of 45 MPa for 100% RQD.
For the design of the lengths of the pile sockets, Eq. 2 was used with a' = 0.3 and b' = 0.7 (values in the center of the literature's range).This led to rock socket lengths in the range 4 -8 m (6 m being the typical length), adopted in the initial Design Stage.As in usual design practice, pile point resistance was not considered.

An Investigation into the Gneiss Resistance as Related to RQD
There is an argument as to whether RQD is an indication of the degree of rock fracturing only or is also an indication of rock weathering.In the case of the Rio de Janeiro gneiss, which is not a much fractured rock, RQD has a direct relation with weathering.Thus, a correlation between RQD and the uniaxial compressive strength (q u ) -obtained from the large pieces of the sample -was sought.In order to establish a relation between the uniaxial compressive strength of the Rio de Janeiro gneiss and RQD, a set of samples was taken from boring with 4 RQD intervals, as if representing a rock weathering profile (as proposed by Deere, 1969).Figure 8 shows the set of selected samples, all obtained by rotary coring with NX double tube barrels (diameter 54.7 mm).
Rock samples (diameter 54.7 mm and height 109.4 mm) were cut and trimmed to ASTM D4543 ( 2001) specifications, and subject to axial compression in a Shimadzu testing machine, Model UH-F 1000 kN (Fig. 9a), of the Structures Laboratory of COPPE (Graduate School of Engineering), Federal University of Rio de Janeiro.Loading rate was 0.1 mm/min and vertical displacements were measured with a LVDT system, as shown in Fig. 9b.
Typical failure modes can be seen in Fig. 10.

Test results and analysis
The results of laboratory tests are summarized in Table 2 (samples obtained from the BRT Project), which presents individual results (5 specimens) and average compressive strength, together with average RQD values for each set.
Test results of Table 2 (average values) are plotted in Fig. 11, which also displays a straight trend line and the line from the Design Stage Eq. 4. As can be seen in this figure, the Design Stage relation is well on the safe side.
The authors had access to a set of boring logs and laboratory test results also performed on gneiss in the Rio de Janeiro region.The Harbour of Rio de Janeiro is undergoing improvement works, more precisely at the cruiser's terminal (Pier Maua).At this location, rotary borings were conducted by Geodrill, and 13 samples, of various RQDs, were tested at the Rio de Janeiro State University (UERJ).Table 3 presents the results of individual tests.
Test results from both sets of data (Tables 2 and 3) are plotted in Fig. 12, also with a trend line.From this figure, it can be concluded that an average relation -close to the trend line -would be: In projects for which a comprehensive investigation, that includes laboratory tests, is not available for design, the authors suggest a safer equation:   also shown in Fig. 12 -as current proposal.

Mobilized Side Shear and an Evaluation of Current Design Methods
An evaluation of the mobilized side shear of the pile was carried out by the analysis of the results of the 30 (high-strain) dynamic load tests (ASTM standard D4945).The blows with the highest energy were analyzed with CAPWAP, which can produce the mobilized side shear distribution with depth.Although subject to some discussion as to the uniqueness of its results (e.g., Danziger et al., 1996), CAPWAP results are accepted in practice, for both distinguishing base and shaft loads and for the assessment of side shear distribution.
Boring logs -with SPT (in soil) and RQD (in rock)were drawn alongside with the mobilized side shear revealed in CAPWAP analyses.All 30 ground and side shear profiles can be seen in Juvencio (2015), and a set of 5 typical profiles are shown in Figs. 13 to 17.In these figures, the following situations can be distinguished: i) the soil overlying the rock is weak, and side shear is developed only along the rock socket: Fig. 13; ii) soil and rock quality varies with depth, and side shear is very variable down to the pile tip: Fig. 14;  Figure 17 is in agreement with remarks by Wyllie (1999) that, in high quality rocks, load is carried by the upper portion of the socket.Wyllie suggests that this early transfer of shear stresses is favored by the fact that the rock has a higher Young's modulus (E R ) than the concrete.In the case of the Rio de Janeiro gneissic rock, when RQD was higher than 75%, E R exceeded 45 GPa.On the other hand, the Young's modulus of the concrete is 30 -35 GPa.In the case of Fig. 17, the rock had a RQD of 78%.
Figure 18 presents typical maximum displacement (DMX in Pile Dynamics software notation) vs. mobilized pile (total) resistance, in a series of blows with increasing energy, for two piles.This plot gives an indication of how close the test came to reach the pile maximum resistance (or capacity).Figure 18a, with results of pile E111, Block 16, Bridge OAE 8, presents a straight line (with maximum displacement of approximately 2 mm), which indicates that the maximum mobilized load was far from failure. Figure 18b, with results of pile E04, Support 2, Bridge OAE 6, presents a somewhat curved line (with a maximum displacement of approximately 5 mm), which indicates that the mobilized load was high.It should be noted that all tests mobilized loads that were at least twice the service load.Figure 19 (data from Table 4) presents a plot of the average mobilized shear stress obtained in dynamic tests against the average q u estimated from average RQD along the pile socket with Eq. ( 6).
This figure indicates that mobilized shear stress values were lower than the side shear resistances predicted by the more commonly used methods, such as Horvath (1978), Williams & Pells (1981)  = .; t ult u q = 05 06 0 5 . . . ) and Rowe & Armitage (1987).This can be explained by the fact that maximum shear resistance was not reached, mainly due to limited energy in the dynamic tests.Another contributing factor is that the above mentioned methods were developed for sedimentary rocks, the maximum resistances of which are more easily reached in tests.
Figure 19 also shows that the boring method -either wet drilling, represented by circles, or dry hammering, represented by squares -seems to have little effect on the mobilized shear stress.

A Proposal for the Evaluation of Side Resistance of Pile Sockets in Gneiss
Since failure was not reached, the use of the data from the previous section (basically average mobilized side shear) to estimate the side shear capacity of a pile would be clearly on the safe side.For a more reasonable prediction of pile capacity it would be better to correlate the maximum mobilized side shear obtained at each pile segment in the    dynamic tests (CAPWAP results) with the corresponding q u at the same level.In this case, q u were obtained from RQD (at the same level) through Eq. 6.The results are presented in Fig. 20.
An interpretation of Fig. 20 leads to the following proposal for a prediction of the available side shear of pile sockets in gneiss of the Rio de Janeiro region: with a = 0.2 and b = 0.5 (very close to those suggested by Horvath, 1978).
Since the set of data points exhibits scattering, Eq. 7 can be understood as leading to reasonable values for design purposes.

Conclusions
The paper initially presents the results of laboratory tests performed on gneiss specimens with different RQDs.If RQD is accepted as an indication not only of rock fracturing but also of rock weathering, a correlation between the uniaxial compressive strength (q u ) and RQD can be sought.Such correlation was obtained through laboratory tests for the gneissic rocks of the Rio de Janeiro region.RQDq u correlations have a practical importance due to the fact that, in most projects, a comprehensive laboratory test program is not available and design has to be based on boring data.
A series of dynamic tests, performed on piles of a BRT project in Rio de Janeiro were analyzed.Results show that the contribution to pile capacity of the residual soil overlying the rock is significant, although usually disregarded in design (there are some studies correlating SPT results with side shear in residual soils, a subject outside the scope of this paper).Although pile sockets were bored by two different methods -water assisted rotary drilling and compressed-air assisted down-the-hole hammering -the boring method did not seem to have a significant effect on shear stress mobilization.Test data also provide a relation between maximum mobilized side shear (t max,mob ) and q u , the latter obtained from RQD (correlation mentioned above).An expression for the prediction of pile shaft capacity of piles bored in gneiss, based on the inferred maximum mobilized side shear, is put forward (Eq.7).

Figure 1 -
Figure1-A rock-socketed pile, with its main resistance components.
Figure 2 -Location of bridges of the BRT Project close to Rio de Janeiro International Airport.

Figure 5 -
Figure 5 -Rotary equipment for rock drilling, with rock bit diameter 70 cm, sitting on top of an already impact driven steel casing, diameter 80 cm.

Figure 6 -
Figure 6 -Equipment used to drive (by turning and pushing down) a steel casing, diameter 50 cm, and then lower a DTH hammer, 39 cm diameter.

Figure 7 -
Figure 7 -Equipment used to drive (by turning and pushing down) a steel casing, diameter 80 cm, and then lower a DTH hammer, 70 cm diameter.

)
Figure 8 -Idealized rock profile with a set of 5 samples for each RQD interval.

Figure 11 -
Figure 11 -Relation q u vs. RQD from laboratory tests for the BRT Project (tests at the Federal University of Rio de Janeiro) and line from the Design Stage equation.
Figure 12 -Relation q u vs. RQD from laboratory tests for the BRT Project (tests at Federal University of Rio de Janeiro) and for the Harbour Improvement (tests at Rio de Janeiro State University), including Design Stage relation, and current proposal.
and b 1 06

Figure 19 -
Figure19-Average mobilized shear stress from dynamic tests vs. q u (estimated from average RQD along the pile socket).

Figure 20 -
Figure 20 -Maximum mobilized shear stress in dynamic tests vs. q u (estimated from average RQD along the pile socket).

Table 1 -
Characteristics of the BRT Project piles.

Table 2 -
Values of RQD and q u , BRT Project (5 specimens tested per RQD interval).