Geotechnical Engineering Circular (GEC) No. 8
Design And Construction Of Continuous Flight Auger Piles
Final
April 2007

Appendix A: Comparisons Of Methods For Estimating The Static Axial Capacity Of CFA Piles

A.1 Introduction

This appendix provides background documentation used to select the recommended method for computing the static axial capacity presented in Chapter 5, and to summarize alternative methods. The methods used for estimating the static axial capacity of continuous flight auger (CFA) piles include methods developed from evaluation of load tests conducted on CFA piles and methods developed for driven piles and drilled shafts. Most of the methods developed specifically for CFA piles are modifications of the methods developed originally for drilled shafts or driven piles, which are based on in-situ soil parameters. Results from four comparative studies, which were conducted to asses the effectiveness of several of these methods, are also presented. The recommendations made in Section 5.3 are based on the main findings of these comparative studies.

The methods for estimating the static axial capacity of CFA piles are commonly separated for cohesive and cohesionless soils. In addition, the axial capacity has two components: the unit side shear capacity (f_s) and the bearing unit capacity (q_p). The ultimate shear capacity is obtained by integrating the f_s values along the pile length. The ultimate bearing capacity is obtained by multiplying the q_b by the pile tip cross sectional area.

In cohesive soils, the static axial capacity of CFA piles is commonly estimated with correlations based on the undrained shear strength (S_u) of the soil. The soil undrained shear strength is estimated in the filed using in-situ techniques (i.e., SPT or CPT) or in the laboratory (e.g., undrained triaxial tests on undisturbed samples). The ultimate side shear capacity is typically estimated using the "alpha method", in which the factor α is dependant on S_u. In general, f_s at a given depth is calculated as follows:

f_s = α S_u

The end-bearing unit capacity (q_p) is typically estimated by a bearing capacity factor multiplied by S_u.

In cohesionless soils, the static axial capacity of CFA piles is commonly estimated following an effective stress approach for side shear and in-situ test parameters for the end-bearing resistance. The unit side shear capacity at a given depth is typically calculated as follows:

f_s = β σ_v′

where β is the beta factor and σ_v′ is the vertical effective stress. With β = K_s tan δ, where K_s is the coefficient of lateral earth pressure and δ is the pile-to-soil friction angle. Beta factors and end-bearing capacity parameters are correlated with SPT-N values.

Silty soils that cannot be easily classified as cohesive or cohesionless require judgment to evaluate which method is more appropriate for selecting a resistance parameter. If the silty soil is expected to behave in an undrained manner, methods for cohesive soils should be used. Conversely, if the silty soil is expected to behave in a drained manner, methods for cohesionless soils should be used.

Several methods exist that use correlations with in-situ test parameters to estimate side shear and end-bearing capacities in a variety of soil types. These methods are presented in the following sections.

A.2 Methods For Axial Capacity Of CFA Piles

The methods presented are introduced in the chronological order they were published.

A.2.1 Wright and Reese (1979) Method

Wright and Reese (1979) developed a method for predicting the ultimate capacity of drilled shafts and CFA piles in sand. In this method, the average ultimate side shear resistance, f_{s ave}, is calculated as:

f_{s ave} = σ′_{v ave} K_s tan φ ≤ 0.15 MPa ( ≤ 1.6 tsf)

Where σ′_{v ave} is the average vertical effective stress along the pile, K_s is the lateral earth pressure coefficient (assumed to be 1.1), and φ is the sand angle of internal friction. The value φ is a weighted average of the angle of internal friction of each sand layer along the pile length.

The ultimate end-bearing resistance is calculated as:

q_p (MPa) = 0.064 N ≤ 3.8 (MPa)

q_p (tsf) = 0.67 N ≤ 40 (tsf)

where N is the SPT-N value (blows/0.3 m [1 ft]) near the tip of the pile.

A.2.2 Douglas (1983) Method

Douglas (1983) developed a method for predicting the capacity of CFA piles based on full scale load test results of 28 CFA piles in sand in the UK and other European countries. The ultimate unite side shear resistance at a given depth is calculated as:

f_s = σ′_v K_o tan φ

Where σ′_v is the vertical effective stress, K_o is the lateral earth pressure coefficient (assumed to be 1.0), and φ is the sand angle of internal friction. The vertical effective stress is limited for depths greater than 6 or 10 pile diameters in loose and medium dense sands, respectively.

The ultimate capacity was defined occurring at a pile tip displacement of 30 mm (1.2 in.) although the bearing resistance was often observed to mobilize at displacements greater than 100 mm (4 in.). The ultimate end-bearing resistance is calculated from the Dutch Cone point resistance (q_c) as:

q_p = 0.25 q_c

A.2.3 Rizkalla (1988) Method

The Rizkalla (1988) method was developed in Germany to the axial capacity of CFA piles based on the CPT tip resistance (q_c). Rizkalla (1988) based his methods on a database of load tests on CFA piles. The method appears to provide conservative predictions for high displacement screw piles.

Cohesive Soils (0.025 ≤ S_u ≤ 0.2 MPa)

The ultimate unit side shear in cohesive soils for a given depth is calculated as:

f_s (MPa) = 0.02 + 0.2 S_u (MPa)

The ultimate unit end-bearing resistance, defined occurring at a pile tip displacement of 5% of the pile diameter, is calculated as:

q_p (MPa) = 6 S_u

S_u can be estimated from the CPT tip resistance (q_c) as:

Where σ_v is the total vertical stress at the tip of the pile, and N_c is the cone factor, which ranges from 16 to 22. Alternatively, S_u may be estimated conservatively from unconfined compressive strength (q_u), with S_u = 0.5 q_u, an approach that is commonly used in many methods currently employed in Europe.

Cohesionless Soils

In cohesionless soils, the ultimate unit side shear at a given depth is calculated as:

f_s = 0.008 q_c (MPa)

where q_c is the CPT tip resistance.

In cohesionless soils, the ultimate unit end-bearing resistance is calculated as:

q_p (MPa) = 0.12 q_c (MPa) + 0.1 for q_c ≤ 25 MPa

A.2.4 Neely (1991) Method

Neely (1991) developed correlations for sandy soils using a database of load tests on 66 CFA piles performed by various researchers. In this method, the effective overburden stress is computed at the mid-depth of the overall pile length. Only one layer of sand is assumed in this method. The average ultimate unit side shear capacity is calculated as follows:

f_s = K_s p′₀ tan δ = β p′₀ ≤ 1.4 tsf

where:

In this method, the ultimate unit end-bearing resistance for sandy soils is correlated to the SPT-N value near the pile tip. Neely (1991) used load test results from Roscoe (1983) and Van Den Elzen (1979) and equivalent SPT-N values, which were correlated with CPT values employing a the relationship developed by Robertson et al. (1983). The unit end-bearing capacity is calculated as follows:

q_p (tsf) = 1.9 N ≤ 75 tsf

where N is the SPT-N value.

A.2.5 Viggiani (1993) Method

Viggiani (1993) developed correlations for cohesionless soils based on load test results on CFA piles and CPT tip resistance, q_c. The load tests and CPTs were performed near Naples, Italy, where soils are volcanic (mostly pyroclastic). The ultimate unit side shear at a given depth is calculated as:

f_s = α q_c

With:

In these cohesionless soils, the ultimate unit end-bearing resistance is calculated as:

q_p (MPa) = q_c(ave)

where q_c(ave) is the average CPT tip resistance calculated in the depth interval of 4 pile diameters above and 4 pile diameters below the pile tip.

Figure A.1: β Factor vs. Pile Length (Neely, 1991)

A.2.6 Decourt (1993) Method

Decourt (1993) method was developed for CFA pile in residual soils (silt and silty soils) and was developed from load tests results. This method relies on the apparent shear resistance developed at the maximum torque required to twist a standard penetration test (SPT) split spoon sampler, after it has been driven into the bottom of the borehole. This method relies on the assumption that the influence of SPT dynamic penetration is eliminated during the subsequent twisting of the sampler in place. The ultimate unit side shear capacity for a given pile segment is assumed to be the same as the unit side shear developed during application of the maximum torque, or:

f_s = f_s (during maximum torque of SPT spoon)

In these soils, the ultimate unit end-bearing resistance is calculated as:

q_p = 0.5 K′ N_eq q_c ≤ 25 (MPa)

where

A.2.7 Clemente et al. (2000) Method

Three CFA test piles were installed in clay soils that ranged from very stiff to hard at three sites in Louisiana, Mississippi, and Texas. Alpha values (α) were back-calculated from load test side shear data and compared with S_u values. This relationship verified the common belief of a decreasing α with increasing S_u, as shown below in Figure A.2.

Figure A.2: α vs. Undrained Shear Strength - Clayey Soils (Clemente et al., 2000)

A.2.8 Frizzi and Meyer (2000) Method

Frizzi and Meyer (2000) published capacity relationships for CFA piles obtained from over 60 load testes on drilled shafts and CFA piles in the Miami and Fort Thompson limestone formations found in South Florida (Broward and Miami-Dade Counties). Procedures for this method are detailed in Chapter 5.3.

A.2.9 Zelada and Stephenson (2000) Method

Zelada and Stephenson (2000) studied the results of 43 full-scale compression load tests, five of which were fully instrumented, and ten pull-out tests of CFA piles in sandy soils. They compared the adequacy of various capacity prediction methods. Results of the comparative study by Zelada and Stephenson are detailed in Section A.6.2.

Their study recommended modifications to the β factor and unit tip values of the FHWA 1999 method (O'Neill and Reese, 1999) to reflect their observations of decreased side shear and increased end-bearing in CFA piles when compared to those for drilled shafts. Their method estimates less side shear resistance than the FHWA method and thereby recommends a reduced β factor, only 0.8 of that of the FHWA. Figure A.3 shows the β factor from Zelada and Stephensons (2000) and from the FHWA 1999 method for comparison. The reduction in the β factor with respect to that of the FHWA method accounts for the reduced soil stress that tend to occur due to soil mining during CFA pile installation. This effect may be more pronounced in clean sands. As discussed in the main text of this document, this effect may be reduced or eliminated by preventing auger over-rotation relative to its penetration rate.

Note that the β factor is also reduced for SPT-N values less than 15, as it is with the FHWA 1999 method. In this method, the expression for β as a function of pile depth is:

for N ≥ 15 bpf

β = 1.2 - 0.11 Z^0.5 (ft)

β = 1.2 - 0.2 Z^0.5 (m)

for N < 15 bpf

The range of β in the Zelada and Stephenson (2000) method is 0.20 ≤ β ≤ 0.96, which takes into account the reduction factor of 0.8.

In their analyses, the ultimate end-bearing resistance was defined at a pile tip displacement of 10% of the pile diameter. The ultimate unit end-bearing resistance in this method, which is based on test data from instrumented CFA load tests in cohesionless soils, is correlated to SPT-N values near the tip of the pile, as follows:

q_p (tsf) = 1.7 N ≤ 75 tsf

q_p (MPa) = 0.16 N ≤ 7.2 MPa

Note that this correlation gives an ultimate end-bearing resistance that is 2.8 times greater than that estimated with the FHWA method for the same SPT-N value. Figure A.4 shows the recommended q_p value as a function of SPT-N values and determined from instrumented CFA load tests. The correlation from the FHWA 1999 method for drilled shafts (i.e., "Reese & O'Neill"), the Meyerhof method for driven piles, and the Neely method (1991) for CFA piles are shown for comparison.

Figure A.3: β Factor vs. depth - Zelada and Stephenson (2000) and FHWA 1999 Methods

Source: Zelada and Stephenson (2000)

A.2.10 Coleman and Arcement (2002) Method

Coleman and Arcement (2002) compared the results of load tests conducted on 32 CFA piles in mixed soil conditions and evaluated the adequacy of various capacity prediction methods. Their study recommended modifications to the α-factor for cohesive soils and β factor for cohesionless soils contained in the FHWA 1999 method. Results of the comparison study conducted by Colman and Arcement of several different commonly used methods are detailed in Section A.6.3 of this Appendix and their recommended modifications to the α and β factors of the FHWA 1999 method are included in Chapter 5.3. Coleman and Arcement did not propose any modifications / methods to estimate the end-bearing resistance.

A.2.11 O'Neill et al. (2002) Method

O'Neill et al. (2002) developed a method by analyzing a database on CFA piles data and data from the three fully instrumented load tests performed on CFA piles in geologically diverse sites of coastal Texas in 1999 (O'Neill et al., 1999). These sites are designated as "UH" (over consolidated clay), "Baytown" (mixed soil conditions), and "Rosenburg" (sands). In these three sites, the nominal diameter of the CFA piles was 0.46 m (18 in.) for all piles while the embedment depths were 15.2 m (50 ft), 15.2 m (50 ft), and 9.1 m (30 ft), respectively, for each of these sites.

Figure A.4: Ultimate Unit End-Bearing Resistance vs. SPT-N values - Zelada and Stephenson (2000) and Other Methods

Source: Zelada and Stephenson (2000)

Based on the load tests mentioned above, O'Neill et al. (2002) presented a normalized load-settlement curve shown in Figure A.5, which may be suitable for use with CFA piles with similar aspect ratios and in similar geologic formations. Given the ultimate total load (Q_t), Figure A.3 may be used to predict the total load (Q) at a given displacement ratio of pile displacement (w) to pile diameter (B).

A.3 Methods For Axial Capacity Of Drilled Shafts Applicable CFA Piles

A.3.1 TxDOT Method (1972)

Cohesive Soils

For cohesive soils, this method (developed by Texas Department of Transportation, 1972) uses the undrained shear strength to estimate the side shear resistance and the blow count obtained from the TxDOT Dynamic Cone Penetrometer test to estimate the end-bearing resistance. Although the TxDOT Dynamic Cone Penetrometer is a relatively uncommon test, this method yielded good results for CFA piles in clays (O'Neill et al., 2002). Therefore, this method was recommended as an alternate way of predicting capacities in cohesive soils and the procedures are detailed in Section 5.3.

Figure A.5: Normalized Load-Settlement Relationship for Design of CFA Piles - Clay Soils of Texas Gulf Coast (O'Neill et al., 2002)

Cohesionless Soils

For cohesionless soils, this method estimates the ultimate unit side-shear resistance using the blow count from a TxDOT Dynamic Cone Penetrometer (N_TxDOT) according to the following Equation:

The ultimate end-bearing resistance in cohesionless soils is also determined using the blow count from the TxDOT Dynamic Cone Penetrometer (N_TxDOT) as follows:

If the pile diameter is less than 0.6 m (24 in.), the ultimate end-bearing resistance is selected as 4 tsf for cohesionless soils.

A.3.2 FHWA 1999 Method

The FHWA Method, originally proposed by Reese and O'Neill (1988) and latter modified by O'Neill and Reese (1999), has become perhaps the most recognized method for prediction of drilled shaft capacities, and has often been used to determine axial capacity of CFA piles. The comparison studies summarized in Section A.6 of this Appendix consistently show that FHWA method is reliable in estimating CFA pile capacities in both cohesive and cohesionless soils. This method is presented as the recommended method for estimating CFA pile capacity in both cohesive and cohesionless soils. Procedures for this method are presented in Section 5.3.

A.4 Methods For Driven Piles Applicable To CFA Piles

A.4.1 Coyle and Castello (1981) Method

Cohesive Soils

Coyle and Castello (1981) recommended the use of Tomlinson's Method (1957) to obtain the average unit side shear capacity of the pile using Equation A-20:

f_Sa = α S_ua

where:

The ultimate unit end-bearing resistance in cohesive soils is calculated using the following:

q_p = 9 S_u

Cohesionless Soil

Coyle and Castello (1981) also presented estimates of the ultimate side shear and end-bearing capacities for driven piles in sand based on the sand angle of internal friction and the ratio of pile embedded depth (L) to pile diameter/width (D).

In this method, the ultimate side shear resistance (i.e., maximum skin friction in the original method) can be determined from Figure A.7a, while the ultimate end-bearing resistance (i.e., maximum toe resistance in the original method) can be determined from Figure A.7b,. The ultimate end-bearing resistance was limited to 100 tsf for driven piles tipped into sand.

Coyle and Castello (1981) also recommended that the angle of internal friction of the soil (f) be obtained from Figure A.8, which presents the relationship between SPT N values and f. Friction angles in this Figure are based on Peck et al. (1974). For the case of silty sands below the water table, and for SPT N values greater than 15, Coyle and Castello (1981) recommended that SPT N values be corrected according to the following expression:

N′ = 15 + 0.5 ( N - 15 )

The corrected N value (i.e., N′) can be utilized in the relationship between SPT-N values and f as well as the pile capacity design charts shown in Figure A.7.

A.4.2 American Petroleum Institute (API) (1993) Method

Cohesive Soils

The ultimate unit side shear resistance for a pile segment in cohesive soils is calculated using an α factor as follows:

f_s = α S_u

where:

Where ψ = S_u/σ_v′, and σ_v′ = vertical effective stress of the soil at the depth of interest.

For cohesive soils, the ultimate unit end-bearing resistance is calculated as follows:

q_p = 9 S_u

Figure A.6: α vs. Average Undrained Shear Strength along Pile Length (Coyle and Castello, 1981)

Figure A.7. (a) Unit Side-Shear Capacity and (b) End-Bearing Capacity in Cohesionless Soils (Coyle and Castello, 1981)

Figure A.8. Relationship between SPT N Values and f (Coyle and Castello, 1981)

Cohesionless Soils

The ultimate unit side shear resistance for a pile segment in cohesionless soils is calculated as follows:

f_s = K σ′_v tan δ

where:

The value for K in is assumed at: a) 0.8 for non-displacement piles and open-ended driven piles; and b) 1.0 for driven piles that have plugged. Therefore, assuming a K value of 1.0 would be most appropriate for CFA. Table A.1 lists limiting unit side shear resistance (i.e. skin friction), as well as typical values of soil-pile friction angles, which may be used if no other values are available.

The ultimate unit end-bearing resistance in cohesionless soils is calculated as follow:

q_p = σ′_v N_q

where:

The unit end-bearing resistance for cohesionless soils is limited in the Coyle and Castello (1981) and the API (1993) methods. For driven piles, the soil below the driven pile tip is compacted and highly stressed after driving. It is likely that the limiting unit end-bearing values for CFA is smaller because the effect of overstressing is lacking in CFA piles.

A.5 Method For Drilled Shafts And Driven Piles Applicable To CFA Piles

Bustamonte and Gianeselli (1981, 1982) reported that the Laboratorie Des Ponts et Chausse (LPC) in France developed design procedures for use in the design of both driven piles and drilled shafts. This method is referred to as the LPC method. In this method, correlations were developed for cohesive and cohesionless soils based on CPT tip resistance measurements. While the original method is uses an average ultimate unit side shear resistance for the entire pile, the technique has also been applied to layered soils where the design charts are utilized for average CPT tip resistances measured at different depths along the pile length. The comparative studies summarized in Section A.6 of this Appendix show that this method is reliable in estimating capacities of CFA pile in both cohesive and cohesionless soils. This method is presented as an alternative for estimating CFA pile capacity in soils when CPT q_c data is available. The procedures of this method are presented in Section 5.3.

A.6 Comparison Of Design Methods For Axial Capacity Of CFA Piles

Several studies have been conducted in the last few years, between 1994 and 2002, to compare load tests results on CFA piles to capacities predicted by various static analysis methods. These comparisons were made using existing load testing databases and, in some cases, were augmented with additional load test programs. The comparative studies report results according to soil type and for normal construction of CFA piles (i.e., drilled displacement piles were not considered in these studies).

A.6.1 Comparison of Five Methods in Predominantly Sandy Soils (McVay et al., 1994)

McVay et al. (1994) studied the results of 17 full-scale compression load tests and four pullout tests of CFA piles located at 21 different sites throughout Florida. The subsurface soils consisted of sandy soils in for 19 test piles, while clay was predominant in two test piles (both compression tests). The ultimate resistance was evaluated in their study at displacements defined by: (1) the Davisson Criteria; (2) a load corresponding to displacement equal to 2% of the pile diameter; and (3) load corresponding to displacement equivalent to five percent of the pile diameter.

Five methods were evaluated: Wright and Reese (1979), Neely (1991), LPC (1981), original FHWA (Reese and O'Neill, 1988); and Coyle and Castello (1981). In the comparison, the mean and standard deviations of the ratio of the measured (Q_{t meas}) to predicted (Q_{t pred}) resistances developed in the McVay et al. (1994) study. These quantities are presented in Table A.2. The study concluded that all methods provided reasonably accurate estimates of total capacity when failure was defined as 5% of the pile diameter displacement. The results for the 5% criteria are illustrated in Figure A.9. This study concluded that the original FHWA method (Reese and O'Neill, 1988) for drilled shafts provided the best prediction for ultimate total pile capacity in sandy soils and this is followed closely by the Wright and Reese Method (1979) for CFA piles in sand.

A.6.2 Comparison of Eight Methods in Predominantly Sandy Soils (Zelada and Stephenson, 2000)

Zelada and Stephenson (2000) studied the results of 43 full-scale compression load tests and ten pullout tests of CFA piles located at 28 sites locations in the United States and Europe. Subsurface soils ranged from medium dense to dense, silty to fine sands. Installed piles that had more than 25 percent of clay along their length were not included in this study.

The ultimate resistance was defined at displacements of 5 to 10% of the pile diameter. The pile diameters ranged from 0.3 to 0.6 m (12 to 24 in.), with the majority within 0.40 to 0.45 m (16 to 18 in.) diameter range. The pile lengths ranged from 7.5 to 21 m (25 to 70 ft), with the majority between 9 and 12 m (30 and 40 ft). All piles were constructed with sand-cement grout, except those reported by Roscoe (1983), which were constructed using concrete.

Eight methods for static capacity were evaluated using this database: Viggiani (1993), Reese and O'Neill (1988), LPC (1981), Wright and Reese (1979), Rizkalla (1988), Coyle and Castello (1981), Douglas (1983), and Neely (1991). The results of this evaluation for the 5 and 10% of pile diameter as displacement criteria are shown in Table A.3. A summary of results for all tests (both compression and tension) at a pile displacement of 10% of the respective pile diameter is provided in Figure A.10. The Reese and O'Neill (1988) method for drilled shafts provided the best correlation for ultimate total pile capacity in these sandy soils, regardless whether the 5 or 10% of the pile diameter displacement failure criteria was used.

The accuracy of the eight methods for a given pile diameter was also evaluated. The results from the Zelada and Stephenson (2000) study are presented in Table A.4. These results indicate that, in most cases, the accuracy of the method increased with increasing pile diameter and there was a tendency for better correlation with higher pile length-to-pile diameter ratios. Similar to the findings in the McVay et al. (1994) study, the original FHWA method for drilled shafts provided the best estimate of ultimate total pile capacity, regardless of pile diameter.

Results comparing side shear resistance from Zelada and Stephenson (2000) are provided in Table A.5. Again, according to this study, the Reese and O'Neill (1988) method for drilled shafts provides the best correlation for side shear resistance in these sandy soils, with the LPC (1981) and the Wright and Reese (1979) methods also giving reasonably accurate results.

Results comparing end-bearing unit resistances from the Zelada and Stephenson (2000) study are provided in Table A.6. The comparison indicates that the Viggiani (1993) method overestimated the end-bearing resistance by approximately 50 percent at displacements of 5 and 10% of the pile diameter. The Neely (1991) method generally provided the best predictions, slightly overestimating and underestimating the end-bearing resistance at displacements of 5 and 10%, respectively. The Coyle and Castello (1981) method overestimated the end-bearing resistance, but the estimates from this method are close to measured values for a displacement of 10% of pile diameter. The remaining five methods underestimated the end-bearing resistance by more than a factor of two, with the exception that the Douglas (1983) method at 5% of the pile diameter.

A.6.3 Comparison of Four Methods in Mixed Soil Conditions (Coleman and Arcement, 2002)

Coleman and Arcement (2002) studied the results of compression load tests conducted on 32 auger-cast piles in mixed soil conditions and performed in accordance with ASTM D 1143. These load tests were conducted at 19 different sites throughout Mississippi and Louisiana during the period of 1985 to 2001.

Several methods for defining CFA failure were considered. It was concluded that while displacements of 5% of the pile diameter was a suitable criterion for large diameter shafts, displacements of 10% of the pile diameter was more appropriate for smaller diameter shafts such as those diameters typical for CFA piles. Therefore, Coleman and Arcement (2002) defined ultimate failure occurs at displacements of 10% of the pile diameter or at the plunging failure, whichever occurred first. In the event that the load was not increased to a level sufficient to cause plunging failure or displacements of ten percent of the pile diameter, the final load obtained was considered ultimate. The pile diameters ranged from 0.36 to 0.6 m (14 to 24 in.) and the pile lengths ranged from 6.1 to 32 m (20 to 105 ft).

Four design methods were evaluated: API (1993), Neely (1991), Coyle and Castello (1981) and FHWA 1999 methods. The results for these design methods are provided in Figure A.11. The API method (1993) gave the best correlation for ultimate total pile capacity in mixed soil conditions and was slightly conservative (i.e., the average ratio of the measured to predicted capacity was 1.03). However, when considering the sites with sandy profiles only, the Neely method (1991) was slightly unconservative (i.e., the average ratio of the measured to predicted capacity was 0.93) and it gave the smallest standard deviation. The FHWA 1999 method (i.e., equivalent to Reese and O'Neill, 1988) was the most conservative and showed a relatively small standard deviation. The Coyle and Castello method (1981) was slightly unconservative and had the highest standard deviation.

Based on this comparison, Coleman and Arcement (2002) recommended a modified FHWA 1999 method because it is widely used and employs factors of side shear resistance (i.e., α and β factors) that can easily be modified. Coleman and Arcement reported that the FHWA 1999 method, although conservative, generally appeared reasonable for predicting the end-bearing component for cohesionless soils. Coleman and Arcement did not make recommendations for modifications to the end-bearing resistance provided by the FHWA method, as they believed there was insufficient data to formulate such modifications.

Figure A.9. Summary of Total Resistance - Results from Five Methods (McVay et al., 1994)

Figure A.10. Summary of Total Resistance - Results from Eight Methods (Zelada and Stephenson, 2000)

O'Neill et al. (1999) studied the results of 43 load tests performed on CFA piles in coastal Texas. Their study was for mixed soil conditions. Because the available data from CFA piles for sands in coastal Texas were limited, their study was augmented with the database developed by McVay et al (1994) from load tests performed on CFA piles in Florida sands. In addition, three CFA test piles were instrumented and load tested in geologically diverse sites in coastal Texas (O'Neill et al., 2002). These sites were designated as "UH" (over consolidated clay), "Baytown"(mixed soil conditions), and "Rosenburg" (sands). These three sites had CFA piles with a nominal diameter of 0.46 m (18 in.) and had embedment depths of 15.2, 15.2, and 9.1 m (50, 50, and 30 ft), respectively.

A.6.4 Comparison of Seven Methods - Sandy and Clay Soils (O'Neill et al., 1999)

To investigate the effect of CFA construction on the earth pressure at a given depth, an effective stress cell was placed near the surface at each of the three instrumented CFA piles to monitor the changes in lateral earth pressure during drilling and grouting. Figure A.14 shows the test arrangement and the measured horizontal earth pressures during a typical installation. The increase and the subsequent drop of horizontal earth pressure from the in situ value as the auger tip moves is part of the cell location during the drilling process. The horizontal earth pressure then increased again during grouting to a residual 20 kPa (0.21 tsf) above the in situ value; however, the residual increase in lateral earth pressure at completion was small compared to the approximately 1,500 kPa (15 tsf) of grout pressure that was maintained during auger extraction.

As part of this study, seven methods were evaluated against this database to evaluate their effectiveness in predicting the CFA capacities. These methods include: Wright and Reese (1979), Neely (1991), TxDOT (1972), FHWA 1999 (O'Neill and Reese, 1999), Coyle and Castello (1981), API method (1993), and LPC method (1981). Additionally, O'Neill et al. (2002) later published a normalized load-settlement curve suitable for design of similar diameter, single CFA piles in the geologic formations of the Texas Gulf Coast (as is presented in Section A.2.11 of this Appendix). In their investigation of load-settlement response, O'Neill et al. (2002) reported that although predictions were generally good, they may have been an artifact of compensating errors in the prediction of the side shear resistance. The measured side shear resistances were generally larger than those predicted for the surficial layer to depths of approximately 4 m (13 ft) but they were generally smaller than the values predicted for deeper strata.

For clay soils, the LPC (1981) and the TxDOT (1972) methods gave the closest correlation for ultimate total pile capacity. Their predictions produced average ratios of the measured to measured capacity of 0.98 and 0.86, and had standard deviations of this ratio of 0.35 and 0.33, respectively. The remaining three methods compared similarly, with the FHWA 1999 method being the most conservative (average ratio of the measured to predicted capacity of 1.52, and a standard deviation of this ratio 0.56). The results with the LPC (1981) and the FHWA methods are illustrated in Figure A.15 (a) and (b), respectively.

Figure A.11. Summary of Total Resistance - Results from Four Methods (Coleman and Arcement, 2002)

For sandy soils, the LPC (1981) and FHWA 1999 method s gave the best correlation for ultimate total pile capacity; both methods produced an average measured to predicted capacity ratio of 1.02 and had standard deviations of this ratio of 0.34 and 0.27, respectively. The results of the LPC (1981) and the FHWA methods are illustrated in Figure A.15 (c) and (d), respectively.

A comparison of various prediction methods of total capacity for cohesive soils is presented in Figure A.16. In this Figure, the mean and standard deviation of the ratio of the measured total capacity to the predicted total capacity is presented for the various methods. The comparison study conducted by O'Neill et al. (1999) is the only study with profiles consisting of entirely cohesive soils. The results for mixed soil conditions from the O'Neill et al. (1999) and Coleman and Arcement (2002) studies are also shown in Figure A.16.

The results from O'Neill et al. (1999) show that the best method for cohesive soils is the LPC method (1981), followed by the TxDOT (1972) method. However, the LPC Method (1981) requires CPT soundings while the TxDOT (1972) method requires a Dynamic Cone Penetrometer (uncommon to most regions of the U.S.). The remaining three methods all appear to have the same trend, with the FHWA 1999 method being the most conservative. The Coyle and Castello (1981) method generally has the average ratio of measured to predicted total capacity closest to 1.0, but also has the lowest mean of this ratio (i.e., it is most unconservative) for mixed soil conditions (Coleman and Arcement, 2002). In all three methods, the most conservative predictions of total capacity occur for clay soils and mixed soil conditions.

Figure A.12. Total Capacity - Results From FHWA 1999 Method (Coleman and Arcement, 2002)

Figure A.13. Total Capacity Results - Coleman and Arcement (2002) Method

A.6.5 Summary of Comparison Study Results for Cohesive Soils

Figure A.14: Effective Lateral Earth Pressure near a CFA Pile during Construction (O'Neill et al., 2002)

A.6.6 Summary of Comparison Study Results for Cohesionless Soils

A comparison of the mean and range of the measured-to-predicted total capacity ratios for cohesionless soils using various prediction methods is presented in Figure A.17. The mean and the standard deviation of this ratio are presented. The results for mixed soil conditions from Coleman and Arcement (2002) are also presented.

The FHWA 1999 method consistently provides the most accurate method for cohesionless soils in all comparison studies. O'Neill et al. (1999) for sands and Coleman and Arcement (2002) for mixed soils report good average mean ratios for the API method (1993), but also a greater standard deviation than for the FHWA 1999 method. McVay et al. (1994), Zelada and Stephenson (2000), and O'Neill et al. (1999) also found that the Wright and Reese (1979) and LPC (1981) methods provide good results; however, the LPC (1981) method may be occasionally unconservative.

The FHWA 1999 method requires SPT borings for estimating the end-bearing resistance and the pile depth for the side shear component. The LPC Method (1981) requires CPT soundings for estimating both components of resistance. The Wright and Reese (1979) and API (1993) methods require estimation of the lateral earth pressure coefficient (K). The Wright and Reese (1979) method requires the internal soil friction angle (f) and the API (1993) method requires the pile-to-soil interface friction angle (d).

Figure A.15. Total Resistance - Results from Four Methods

Source: O'Neill et al. (2002)

The study by Zelada and Stephenson (2000) for cohesionless soils concluded that the FHWA 1999 method had the best correlation for the side shear capacity, although this method is somewhat unconservative. This study concluded that the FHWA method was conservative for end-bearing resistance. Zelada and Stephenson (2000) recommended modifications to the FHWA 1999 method to reflect the tendency for CFA piles to exhibit less side shear and greater end-bearing resistance in cohesionless soils than that predicted with the FHWA 1999 method. The side shear capacity predicted was reduced by a factor of 0.8, and the end-bearing resistance was increased by a factor of 2.8.

Coleman and Arcement (2002) recommended modifications to the β factor of the FHWA 1999 method for side shear estimation in cohesionless soils (either sandy or silty soils). This was done to reflect the tendency for CFA piles to exhibit greater side shear resistance than the values predicted near the surface and less side shear resistance than that predicted at greater depths.

Figure A.16. Comparison of Study Results - Axial Capacity in Cohesive Soils

Figure A.17: Comparison of Study Results - Axial Capacity in Cohesionless Soils