You are on page 1of 8

ARMA/USRMS 05-767

Evaluation of Capacity of Rock Foundation Sockets


Kulhawy, Fred H.
Cornell University, Ithaca, New York, USA

Prakoso, Widjojo A.
University of Indonesia, Depok, Indonesia

Akbas, Sami O.
Cornell University, Ithaca, New York, USA
Copyright 2005, ARMA, American Rock Mechanics Association This paper was prepared for presentation at Alaska Rocks 2005, The 40th U.S. Symposium on Rock Mechanics (USRMS): Rock Mechanics for Energy, Mineral and Infrastructure Development in the Northern Regions, held in Anchorage, Alaska, June 25-29, 2005. This paper was selected for presentation by a USRMS Program Committee following review of information contained in an abstract submitted earlier by the author(s). Contents of the paper, as presented, have not been reviewed by ARMA/USRMS and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of USRMS, ARMA, their officers, or members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: Drilled foundations often are socketed into rock to increase the foundation capacity. However, procedures to quantify the side resistance capacity of sockets vary considerably. This paper reviews many of the proposed methods to predict this capacity and critically assesses them. One method then is recommended, based on the currently available data. Statistics for this method are presented, and design implications are noted.

1. INTRODUCTION Drilled shafts are a common foundation selection for all types of structures. When the structure loads are relatively large or where the soil is of relatively poor quality, the shafts often are drilled through the soil to the underlying rock mass. These shafts then could be founded or seated on the surface of the rock mass, or they could be drilled into the rock mass to create a rock socket, as shown in Figure 1. In this figure, the load or stress applied at the butt is supported by the socket through both tip and side resistances, assuming for illustration that the soil is non-contributory. How the loads are distributed between the tip and side is a function of the loading magnitude, problem geometry, elastic properties of the rock mass and shaft concrete, ultimate bearing capacity of the tip, and the side resistance of the socket. A complete discussion of all of these issues is well beyond the scope of this paper. Herein the focus is on the socket side resistance. In this paper, the basics of socket side resistance are described first. Then early approaches to evaluating sockets are discussed briefly. Following then are discussions of pertinent papers that trace the evolution of methods to evaluate socket side resistance, leading to the most current thoughts on the subject.

Fig. 1. Illustrative rock socket.

The paper concludes with final recommendations for use in design. 2. EARLY APPROACHES In early literature on the subject, the socket side resistance was commonly called the socket bond or the bond stress, using the common analogy of bond between concrete and reinforcing bars. This analogy is useful conceptually, but it is not strictly correct and it negatively influenced some later developments, as will be discussed.

Examination of earlier foundations texts indicates a relative lack of sophistication in addressing rock socket design, although it is frequently mentioned that load tests can and should be done because of this lack of knowledge. In the 1961 Chellis text [1], there is an illustration and discussion of a test method for rock bond. But he also states that, on work designed and built to date, a bond stress of 200 psi (1.38 MN/m2) has been used on the socket wall. Considering his calculation example, this value likely comes from 0.05 times f 'c, for 4,000 psi (27.6 MN/m2) concrete. In his 1962 text, Teng [2] differentiates between hard and soft rocks. In hard rock, he suggests taking the side resistance as the bond value between concrete and reinforcing bars. In soft rock, he says the side resistance "is governed by the shear strength of the rock which must be determined by test of rock samples". In their 1972 text, Woodward, et al. [3] stress that strength reduction factors () have not been developed for rock and the side resistance can only be determined from load tests. They also note, from a sampling of typical design practice, that the side resistance is seldom more than 1/5 or less than 1/10 the allowable tip resistance. Values they tabulated from practice were 7.5 tsf (104 psi or 720 kN/m2), except in one locale where values to 250 psi (1.72 MN/m2) were used. They further noted that, in some areas of practice, the side resistance in relatively strong and sound rock is considered to be controlled by the strength of the shaft concrete, using 0.04 to 0.05 f 'c but not exceeding 200 psi (1.38 MN/m2). In their 1974 text, Peck, et al. [4] do not discuss the design of rock sockets. It is clear that, as of the mid-1970s, our collective knowledge of the behavior of rock sockets was limited, and our ability to predict the side resistance was rudimentary. 3. GENERALIZED SOCKET BEHAVIOR Figure 2 depicts the generalized load-displacement behavior of drilled shafts under axial load. This general pattern holds in both soil [5] and rock [6], as shown in many load tests that were carefully conducted and well-documented. There is essentially a linear response from the origin to L1, followed by a nonlinear transition region to L2, after which there is a final linear region. In rock masses, these

Fig. 2. Generalized load-displacement behavior [5, 6]

regions correspond to initial linear elastic behavior, followed by bond breakage and progressive slip, and then full frictional slip with dilation. The same general pattern holds for both compression and uplift tests, although the relative sizes and importance of the regions differ somewhat. In all cases, the occurrence of a clearly defined peak to the curve is infrequent. With nonlinear curves such as these, there is always a major question about how to define the foundation "capacity" for subsequent design use. Examination of the literature [5] reveals at least 41 different methods for the interpretation of axial load tests, including displacement limits (absolute and percent of diameter), graphical constructions, and mathematical functions. These also reflect a mix of what actually are both ultimate limit state and serviceability limit state criteria. Our detailed studies [5, 7, and many others] indicate that a consistent and reasonable method for defining the "interpreted failure load" is to use QL2, which is the load at L2. The L1 and L2 points are determined graphically from a plot at a scale similar to that of Figure 2. [Note that L1 represents the "elastic limit".] The QL2 value always follows the nonlinearity, sometimes represents the actual curve peak where there is little or no dilation, and can be evaluated from virtually all quality test data. Once the "capacity" is defined, then the side resistance (QsL2) can be evaluated from measurements that separate the tip and side resistances in compression tests of full sockets. In uplift tests, and in compression tests with a void or frangible material beneath the tip (i.e., shear tests), the evaluation is

straightforward and only requires consideration of the shaft weight. Using this side resistance and the actual as-built socket side area (As), the average or unit side resistance (f) can be computed as follows: f = QsL2 / As (1) This value then is most often compared to one of the simpler rock material indices, such as the uniaxial compressive strength (qu). The qu tests should all be done in accordance with proper test procedures, such as those given by ASTM, ISRM , or others. Estimating qu from simpler tests such as point load index, Schmidt hammer, or others, is inappropriate. Strictly speaking, any comparison also should be with the average qu over the depth of the socket. Most of the studies conducted to date have not met these criteria, based on the documentation presented or stated. This statement is not intended to fault the authors, who undoubtedly presented the best information they could. It is intended to point out that we are frequently dealing with imperfect and sometimes poor data, and therefore our expectations should be tempered acccordingly. 4. SOCKET SIDE RESISTANCE MODELS Beginning in the mid-1970s, a number of models were proposed to compute the socket side resistance. To compare these models, they have been rewritten in consistent form as follows: f / pa = C (qu / pa)n (2) in which pa = atmospheric pressure in the desired units (1 atm = 1.058 tsf = 101.3 kN/m2 = 14.7 psi) to make the relationships dimensionless, C = constant, and n = exponent. 4.1. Rosenberg and Journeaux, 1976 Perhaps the first relationship for socket side resistance was given by Rosenberg and Journeaux [8]. They presented a relationship between f and qu that can be approximated by: f / pa = 1.09 (qu / pa)0.52 (3) Unfortunately, their plot was based on only six data points, with qu / pa between 5 and 340. Of these six points, two were not conducted to failure, and two only had estimated qu values. There were no apparent consistencies in evaluating the capacity and qu.

4.2. Horvath, 1978 The first systematic attempt to assess the socket side resistance was by Horvath [9] and was subsequently described by Horvath and Kenney [10]. Their database included large and small scale drilled shafts in the field, rock anchors in the field, and small scale shafts in the laboratory, with multiple tests at some sites. Of the 87 field tests reported [10], 75 were in sedimentary rock (with 50 in the shale family). Apparently, the capacity was defined as the maximum applied test load, while qu was reported as given in the original source or was estimated. They also suggested that the weaker of the concrete or rock would control the side resistance, so therefore their property range was for the lesser of f 'c / pa or qu / pa between 1 and 400. It should be noted that no other researcher has adopted this convention. All others use qu /pa. Horvath [9] plotted his data as shown in Figure 3.

Fig. 3. Bond strength for shafts and anchors from Horvath [9]

The regression equation given by Horvath [9] for all of his data is as follows (f = bond): f / pa = 1.04 (lesser of f'c / pa or qu / pa)0.5 (4) For larger field scale drilled shafts (B > 400 mm), Horvath and Kenney [10] gave the following: f / pa = 0.65 to 0.78 (lesser of f'c / pa or qu / pa)0.5(5) and then recommended using C = 0.65. Horvath, et al. [11] subsequently discussed the improvement of shaft capacity by roughening the socket. This technique can increase the capacity significantly in softer rock. However, a detailed discussion on quantifying roughness effects is beyond the scope of this paper. 4.3. Meigh and Wolski, 1979 Meigh and Wolski [12] reviewed the Rosenberg and Journeaux and the Horvath and Kenney relationships, and they compared them to some 13 drilled

shaft tests. Of these 13 tests, several were used in the prior relationships, and about half were considered to have "uncertain" data. The range of qu / pa was from 2 to 200. They suggested a lower bound for weak rock (qu / pa = 7 to 125) that can be approximated as follows: f / pa = 0.55 (qu / pa)0.6 (6) For qu / pa between 4 and 7, they recommended a constant lower bound at f = 0.25 qu. 4.4. Williams, et al., 1980 Williams, et al. [13] focused on some 18 field load tests they conducted at four sites in Melbourne mudstone. They supplemented their data with results of some 18 tests by others at several sites in the same mudstone and in Sydney shale. The range of qu / pa was from 5 to 800. For their tests, they achieved peak values of side resistance from the load-displacement curves. However, the qu values were determined from correlations between the insitu water content and the drained strength parameters. The resulting relationship they developed can be approximated by: f / pa = 1.84 (qu / pa)0.37 (7)
Fig. 4. Side resistance from Rowe and Armitage [14]

They also addressed many issues of socket roughness. 4.5. Rowe and Armitage, 1984 Another comprehensive summary was done by Rowe and Armitage [14, 15], with more than 80 tests from over 20 sites. The range of qu / pa was from over 4 to under 400. Foundation capacity was largely as defined by the original authors, as was the rock strength. The resulting data plot is given in Figure 4, which also shows that a substantial percentage of the tests did not reach failure. From these data, the following suggested correlation was given (f = max): f / pa = 1.42 (qu / pa)0.50 (8) They further suggested a higher value of C for roughened sockets. 4.6. Carter and Kulhawy, 1988 Carter and Kulhawy [6, 16, 17] examined the Rowe and Armitage data further and noted that there is an approximate lower bound to these data that is given conveniently by: f / pa = 0.63 (qu / pa)0.50 (9)

They also conducted a more sophisticated evaluation of 12 of the field load tests, with the results shown in Figure 5. Basically these analyses confirmed Eq. (8) for the suggested correlation and Eq. (9) for the lower bound.

Fig. 5. Load test evaluations by Carter and Kulhawy [6, 17]

After examination of these data, they also made two important design check recommendations. First, values of f in excess of 0.15 qu, over the full range of expected values, should be used only when they are demonstrated to be reasonable by a load test, local experience, or adequate in-situ testing. And second, after obtaining the design value of f, typically from Eq. (8), and applying a factor of safety to this value, a check should be made against the concrete bond value of 0.05 f 'c. The lower value should be used unless load test data show otherwise. 4.7. Reese and O'Neill, 1988 and 1999 Reese and O'Neill [18] used some of the relationships described previously and have suggested design recommendations based on them for geomaterials they define as rock, with qu / pa > 17. For qu / pa > 19, they recommend the Horvath and Kenney Eq. (8) with C = 0.65. This recommendation is very conservative since others have shown this to be a lower bound value. For qu / pa from 17 to 19, they recommend taking the Carter and Kulhawy design check of f versus 0.15 qu and turning it into a design recommendation to evaluate f = 0.15 qu. This unintended usage actually gives values that are even lower than the lower bound by Meigh and Wolski. In 1999, O'Neill and Reese [19] revised the above recommendations. Rock is now defined by qu / pa > 50, and they still recommend the Horvath and Kenney Eq. (8) with C = 0.65. 4.8. Kulhawy and Phoon, 1993 Kulhawy and Phoon [20] used the database developed by Rowe and Armitage [14], as described previously, and a database for drilled shafts in Florida limerocks developed by Bloomquist and Townsend [21] and McVay et al. [22]. For the Florida data, there were 47 tests from 23 sites. These data had essentially the same limitations as the other data. Figure 6 shows these integrated results, along with those for shafts in clay. Figure 6a shows all of the data, and Figure 6b shows the data averaged per site. The second case essentially eliminates the site bias caused by multiple tests at one site. In both of these figures, there are regression lines shown (for a, r2 = 0.46 and standard deviation = 0.25, and for b, r2 = 0.71 and standard deviation = 0.17).

Fig. 6. Side resistances by Kulhawy and Phoon [20]

These results were the first to demonstrate the importance of eliminating site bias and were among the few to use regression analyses for ther data. The resulting interpretation of the data suggested using the Rowe and Armitage Eq. (8), which is appropriate since these data dominate. 4.9. Zhang and Einstein, 1998, 1999 Zhang and Einstein [23, 24] also looked at the available data and proposed relationships. Their initial assessment [23] suggested the following: f / pa = 1.26 (qu / pa)0.50 (10) However, their subsequent assessment [24] gave the same as the Carter and Kulhawy lower bound, given by Eq. (9), with C = 0.63, which is 1/2 Eq. (10). 4.10. Prakoso, 2002 More recently, Prakoso [25] re-examined the data available and attempted to evaluate them in a more consistent manner. First, the only data used were those that had load-displacement curves to failure so that the "interpreted failure load" could be determined for all the data. Therefore, at least all the load test "capacities" were evaluated in a consistent

manner. However, it was not possible to reevaluate the qu data to ensure consistency in test conduct and averaging over the shaft depth. The results were presented in terms of the side resistance factor (r), which is given by: r = qL2 / qu = f / qu (11) in which qL2 is the stress at L2. To link the prior formats with this one, note that the equations for these formats are given by: log10 r = A - B log10 (qu / pa) or or or r = 10A (qu / pa)-B f / pa = 10A (qu / pa)B log10 (f / pa) = A + B log10 (qu / pa) (12) (13) (14) (15)
Side Resistance Factor, r

0.1 0

0 0.01

0.001 0

log10 r = 0.24 - 0.67 log10(qu / pa) m = 52, r2 = 0.69, S.D. = 0.30


I. Intrusive I. Extrusive I. Pyroclastic S. Clastic (fine) S. Clastic (coarse) S. Chemical M. Non-Foliated Man-Made

0.0001 0 1 10 100 1000 10000

Figure 7 shows the results for all of the data, including multiple tests at the same site and results for (a) shafts in natural and man-made rocks, (b) grouted piles in natural rocks, and (c) rock anchors in natural rocks. The regression line is given by: f / pa = 2.00 (qu / pa)0.69 (16) Figure 8 shows the results of the data averaged per test site. The regression line corresponds to: f / pa = 1.74 (qu / pa)0.67 (17) Careful examination of these results indicates that the rock anchor data are clustered in the lower portions of the figure, especially in the lower right. Setting these data aside gives the results for drilled shafts and grouted piles as shown in Figure 9 by the
1

Uniaxial Compressive Strength, qu / pa

Fig. 8. r vs. qu for all data, averaged per site [25]


1

Side Resistance Factor, r

0 0.1

0 0.01
log10 r = - 0.01 - 0.50 log10(qu / pa) m = 41, r2 = 0.51, S.D. = 0.31

0.001 0

I. Intrusive S. Clastic (coarse) I. Extrusive S. Chemical I. Pyroclastic M. Non-Foliated S. Clastic (fine) Man-Made Regression Line for Data with Rock Anchors

0.0001 0 1 10 100 1000 10000

Uniaxial Compressive Strength, qu / pa


Side Resistance Factor, r 0 0.1

Fig. 9. r vs. qu for drilled shafts and grouted piles, averaged per site [25]

solid line. The regression line corresponds to:


0 0.01

f / pa = 0.98 (qu / pa)0.50


log10 r = 0.30 - 0.69 log10(qu / pa) m = 104, r2 = 0.72, S.D. = 0.29
I. Intrusive I. Extrusive I. Pyroclastic S. Clastic (fine) S. Clastic (coarse) S. Chemical M. Non-Foliated Man-Made

(18) (19)

which can be conveniently rounded to f / pa = (qu / pa)0.50 This value is on the order of about 70% of the value from Eq. (8), which was the recommended value when the capacity definitions were unknown or uncontrolled. Now, when the "interpreted failure load" is given at L2, Eq. (19) is the more appropriate one to use. The lower bound C value of 0.63 that was

0.001 0

0.0001 0 1 10 100 1000 10000

Uniaxial Compressive Strength, qu / pa

Fig. 7. r vs. qu for all data [25]

cited previously actually represents the lower bound for 90% of the data in Figure 9. To capture 100% of the data points, the absolute lower bound would be about 0.25. It should be noted in Figure 9 that the regression is altered significantly when the rock anchor data are included. Clearly these data constitute a separate population. 5. LOCALIZED RELATIONSHIPS In addition to the general relationships described above, there have been a number of studies that have focused exclusively on localized rock units, such as the chalks of southern England and the limerocks of Florida. These studies are of local importance and are too specialized to be discussed herein. When these are addressed, they should be considered within the broad framework described above. 6. ROCK SOCKET SIDE RESISTANCE AND CONCRETE BOND STENGTH Carter and Kulhawy [6] made a design check recommendation to compare the allowable side resistance of the rock socket (f / FS) to the concrete bond strength, given by 0.05 f 'c. The lower value would control, unless field testing showed otherwise. By using typical safety factors of 2 and 3, the ultimate side resistance can be compared with the factored concrete bond strength, as given in Figure 10. Typical ranges of concrete strength, f 'c / pa = 200 - 400, were used for comparison. As can be seen, most side resistances are below the limiting concrete values. The percentages are given in Table 1, which shows that there are more cases of sockets exceeding the concrete bond strength with
100 Side Resistance, f / pa 80 60 40 20 0 1 10 100 1000 10000
I. Intrusive I. Extrusive I. Pyroclastic S. Clastic (fine) 0.05 FSlim (fc' / pa) fc' / pa = 400 2 3 fc' / pa = 200 2 S. Clastic (coarse) S. Chemical M. Non-Foliated Man-Made FSlim = 3

lower concrete strength and factor of safety. It must be noted that all of these cases showed acceptable behavior when the concrete bond strength was exceeded. Clearly the concrete behaves better when it is confined in a socket and reinforced than when it is unconfined and unreinforced.
Table 1. Socket side resistance exceeding concrete bond

concrete f 'c / pa 200 400

% socket fallow > 0.05 f 'c FSlim = 2 FSlim = 3 16 2 4 1

7. SUMMARY AND CONCLUSIONS The behavior of normal rock sockets for drilled shafts has been reviewed, and a survey has been conducted of various proposals to estimate the socket side resistance. In general, the side resistance can be estimated from the following: f / pa = C (qu / pa)n in which C = constant and n = exponent. The databases used by different authors have varied widely, as have the controls on the data type and quality. However, for all practical purposes, nearly all authors have shown that n = 0.50. On the other hand, values of C have varied depending on the author and database. The most recent evaluation, apparently the only one in which all the load test data were interpreted in the same manner, gave a mean value of C equal to 0.98, which is conveniently rounded to 1.0. Therefore, the recommended equation for predicting the side resistance of normal rock sockets for drilled shafts is as follows: f / pa = (qu / pa)0.50 (19) For a lower bound to 90% of the data, the value of C is equal to 0.63. For roughened sockets, C will be larger than 1, but this evaluation is beyond the scope of this paper. Finally, the data show that, in a small percentage of the cases, the allowable socket side resistance is larger than the concrete bond strength. This point illustrates that the concrete behaves better when it is confined in a socket and reinforced than when it is unconfined and unreinforced. (2)

Uniaxial Compressive Strength, qu / pa

Fig. 10. Socket side resistance versus concrete bond strength

REFERENCES
1. 2. 3. Chellis, R.D. 1961. Pile Foundations. 2nd ed. New York: McGraw-Hill. Teng, W.C. 1962. Foundation Design. Englewood Cliffs: Prentice-Hall. Woodward, R.J., Jr., W.S. Gardner & D.M. Greer. 1972. Drilled Pier Foundations. New York: McGrawHill. Peck, R.B., W.E. Hanson & T.H. Thornburn. 1974. Foundation Engineering. 2nd ed. New York: Wiley. Hirany, A. & F.H. Kulhawy. 1988. Conduct & interpretation of load tests on drilled shafts. Report EL5915. Palo Alto: Electric Power Research Institute. Carter, J.P. & F.H. Kulhawy. 1988. Analysis and design of drilled shaft foundations socketed into rock. Report EL-5918. Palo Alto: Electric Power Research Institute. Hirany, A. & F.H. Kulhawy. 2002. On the interpretation of drilled foundation load tests. In Deep Foundations 2002 (GSP116), ed. M.W. O'Neill & F.C. Townsend, 1018-1028. Reston: ASCE. Rosenberg, P. & N.L. Journeaux. 1976. Friction and end bearing tests on bedrock for high capacity socket design. Canadian Geotech. J. 13(3): 324-333. Horvath, R.G. 1978. Field load test data on concrete-torock bond strength for drilled pier foundations. Publication 78-07, Toronto: Univ. of Toronto.

17. Kulhawy, F.H. & J.P. Carter. 1992. Socketed foundations in rock masses. In Engineering in Rock Masses, ed. F.G. Bell, 509-529. Oxford: ButterworthHeinemann. 18. Reese, L.C. & M.W. O'Neill. 1988. Drilled shafts: construction procedures and design methods. Report FHWA-HI-88-042. McLean: Federal Highway Administration. 19. O'Neill, M.W. and L.C.Reese. 1999. Drilled shafts: construction procedures and design methods. Report FHWA-IF-99-025. McLean: Federal Highway Administration. 20. Kulhawy, F.H. & K.K. Phoon. 1993. Drilled shaft side resistance in clay soil to rock. In Design and Performance of Deep Foundations: Piles and Piers in Soil and Soft Rock (GSP 38), ed. P.P. Nelson, T.D. Smith & E.C. Clukey, 172-183. New York: ASCE. 21. Bloomquist, D. & F.C. Townsend. 1991. Development of insitu equipment for capacity determinations of deep foundations in Florida limestone. Report to Florida Dept. of Transportation. Gainesville: University of Florida. 22. McVay, M.C., F.C. Townsend & R.C. Williams. 1992. Design of socketed drilled shafts in limestone. J. Geotech. Eng.(ASCE). 118(10):1626-1637. 23. Zhang, L. & H.H. Einstein. 1998. End bearing capacity of drilled shafts in rock. J. Geotech. Eng.(ASCE). 124(7):574-584. 24. Zhang, L. & H.H. Einstein. 1999. Closure to "end bearing capacity of drilled shafts in rock". J. Geotech. Eng.(ASCE). 125(12):1109-1110. 25. Prakoso, W.A. 2002. Reliability-based design of foundations in rock masses. PhD Dissertation. Ithaca: Cornell University.

4. 5.

6.

7.

8.

9.

10. Horvath, R.G. & T.C. Kenney. 1979. Shaft resistance of rock-socketed drilled piers. In Symposium on Deep Foundations, Atlanta, Oct. 1979, ed. F.M. Fuller, 182214. New York: ASCE. 11. Horvath, R.G., T.C. Kenney & P. Kozicki. 1983. Methods of improving the performance of drilled piers in weak rock. Canadian Geotech. J. 20(4): 758-772. 12. Meigh, A.C. & W. Wolski. 1979. Design parameters for weak rock. In Proceedings, 7th European Conference on Soil Mechanics and Foundation Engineering, Brighton, Sep. 1979, 5: 59-79. London: British Geotechnical Society. 13. Williams, A.F., I.W. Johnston & I.B. Donald. 1980. Design of socketed piles in weak rock. In Structural Foundations on Rock, ed. P.J.N. Pells, 327-347. Rotterdam: Balkema. 14. Rowe, R.K. & H.H. Armitage. 1984. Design of piles socketed into weak rock. Report GEOT-11-84. London: Univ. of Western Ontario. 15. Rowe, R.K. & H.H. Armitage. 1987. A design method for drilled piers in soft rock. Canadian Geotech. J. 24(1): 126-142. 16. Kulhawy, F.H. & J.P. Carter. 1992. Settlement and bearing capacity of foundations on rock masses. In Engineering in Rock Masses, ed. F.G. Bell, 231-245. Oxford: Butterworth-Heinemann.

You might also like