Revisiting Axial Load Testing of Helical Piles in Glaciolacustrine Clay
- deepexcavation
- 3 days ago
- 8 min read
Performance of the Helical Piles P-2 and P-3 Using DeepFND
1. Introduction
This article reviews the axial performance of two single helical piles installed in a glaciolacustrine clay deposit in Edmonton, Canada, as documented by Lanyi-Bennett and Deng (2019). The field results provide a reliable basis for calibrating DeepFND simulations involving double-helix piles in cohesive soils. The study focuses on piles P-2 and P-3, examining their measured behaviour and comparing it with simulations carried out within DeepFND using soil and geometric parameters consistent with the field conditions.
2. Pile Properties
Piles P-2 and P-3 share the same geometry and installation method. Their essential characteristics and measured capacities are summarised in Table 1.
Table 1- Properties of the Helical Pile (adapted from Lanyi-Bennett and Deng, 2019).
Property | P-2 | P-3 |
Test time, ts | 12 | 15 |
Interhelix spacing ratio, sh/D | 3 | 3 |
Shaft diameter, d (mm) | 73 | 73 |
Helix diameter, D (mm) | 305 | 305 |
Interhelix spacing, sh (mm) | 914 | 914 |
Total pile length, L (m) | 6.10 | 6.10 |
Ultimate load, Qu (kN) | 101.5 | 89.3 |
Torque factor, Kt (m⁻¹) | 23.4 | 23.9 |
Both piles failed by individual bearing at the helices, making them suitable reference cases for validating DeepFND's helical pile module.
3. Soil Stratigraphy and Properties
The subsurface profile at the site consists of layered glaciolacustrine deposits with generally consistent undrained shear strength in the main clay layer. Table 2 summarises the stratigraphy interpreted from CPTs, sampling, and laboratory testing.
Figure 1- Representation of the soil stratigraphy and soil characterisation, and their results of su (adapted from Lanyi-Bennett and Deng, 2019).
Table 2. Summary of Soil Stratigraphy and Strength Parameters
Depth (m) | Soil Layer | Key Properties |
0.0 – 0.7 | Topsoil | Organic, low stiffness |
0.7 – 1.5 | Silty Clay Crust (MH) | LL = 27%; PL = 17%; γsat ≈ 17.7 kN/m³; su ≈ 100 kPa |
1.5 – 6.0 | Stiff Clay (CH) | w ≈ 34%; γsat ≈ 18.2 kN/m³; su ≈ 55–77 kPa (UCS + CPT) |
6.0 – 7.5 | Silty Clay with sand seams | Transitional stiffness |
7.5 – 9.5 | Silty Sand with clay lenses | Higher stiffness, granular behaviour |
≥ 9.5 | Till | Very stiff/hard till |
Groundwater varied between 3.0 and 4.0 m during the testing period.
Figure 2 presents the stratigraphy and model for pile P-2 modelled in DeepFND.
Figure 2- Representation of the soil stratigraphy and pile layout in DeepFND.
4. DeepFND Modelling Methodology
The simulation in DeepFND reproduced the geometry, soil layering, and installation conditions from the field study. An su-based undrained model was applied to the cohesive layers (Table 3).
Table 3- Key DeepFND Input Parameters.
Model Component | Assigned Input |
Soil model | Undrained (su - based) |
Shear strength profile | Layered su values from Table 2 |
Helix diameters | 305 mm (two helices) |
Helix spacing | 914 mm (sh/D = 3) |
Shaft condition | Plugged pipe |
Bearing coefficient, Nt | 7–10 (calibrated range from field data) |
Adhesion coefficient, α | Based on cohesive soil and installation disturbance |
Loading procedure | Incremental compression, displacement-tracked |
5. Project Details
The test site at the University of Alberta included multiple 2×2 pile groups and four single piles. P-2 and P-3 served as the primary reference tests after sufficient setup time, ensuring that excess pore pressures had dissipated.
Table 4- Relevant Project Information.
Parameter | Value |
Site geology | Glaciolacustrine clay over sand/till |
Test pile type | Double-helix screw piles |
Groundwater depth | 3 to 4 m |
Installation torque (general range) | Recorded every 0.3 m during installation |
Failure definition | Plunging load (single piles) |
6. Results
The measured load–settlement curves for P-2 and P-3 show a similar progression: an initially stiff response, progressive softening, and a sharp transition into plunging failure. The ultimate capacities fall within the expected range predicted by the torque–capacity relationship Qu = Kt · T, confirming the reliability of the torque-based estimation.
Interpretation of the Load–Settlement Comparison
The comparison between the measured load–settlement curves for piles P-2 and P-3 and the corresponding DeepFND predictions shows very good agreement across the full loading range. For both piles, the initial response is captured accurately, with DeepFND replicating the shallow, gradual settlement observed under working loads. As the applied load increases towards the non-linear region, the numerical curves follow the measured trends closely, reflecting the progressive mobilisation of helix bearing resistance.
For P-2, DeepFND slightly overpredicts stiffness in the mid-range of loading but matches the onset of rapid settlement well, reproducing the characteristic transition to plunging. For P-3, the measured and predicted curves are even more closely aligned, with only minor deviations appearing near failure where natural variability in soil stiffness becomes more influential. In both cases, the predicted ultimate capacities fall within the expected margin of the field values, and the settlement at failure is reproduced with strong consistency.
Figure 3- Comparison between measured and DeepFND-predicted load–settlement responses for single helical piles P-2 and P-3 installed in glaciolacustrine clay: (a) P-2, (b) P-3.
Overall, the graphical comparison shows the DeepFND’s capability to model the axial response of double-helix piles in cohesive soils with high fidelity, as long as the soil strength profile and helix geometry are defined accurately. The software captures both stiffness and failure behaviour, reinforcing the reliability of the modelling approach adopted in this study.
DeepFND simulation output for pile P-2 illustrating the soil profile, helix geometry, estimated installation torque, and the decomposition of axial resistance into shaft, cylindrical shear, and tip components (Figure 4). The right-hand curve compares the computed load–settlement response with the measured field data. The model replicates the observed stiffness and the onset of non-linear behaviour, with a predicted ultimate capacity of approximately 109 kN at a settlement of 3.3 cm, closely matching the field performance. The figure also displays structural and geotechnical design capacities, installation torque correlation, and the governing failure mode, confirming that the pile behaves in individual bearing at the helices with strong agreement between numerical predictions and test measurements.
Figure 4 - DeepFND axial capacity and settlement analysis for pile P-2.
DeepFND simulation output for pile P-3 showing the soil profile, helix configuration, estimated installation torque, and computed axial resistance components (Figure 5). The right-hand plot compares the predicted load–settlement curve with the measured field response. The model captures the initial stiffness and the transition into non-linear settlement, with a predicted ultimate capacity of approximately 100 kN at a settlement of 3.28 cm, consistent with the field load test. The diagram also illustrates shaft and helix contributions, cylindrical shear resistance, tip stress parameters, and the governing design capacities (geotechnical and structural), confirming individual bearing behaviour at the helices and strong agreement between DeepFND predictions and test results.
Figure 5- DeepFND axial capacity and settlement analysis for pile P-3.
To evaluate the accuracy of the numerical simulations, the measured behaviour of piles P-2 and P-3 was compared directly with the corresponding DeepFND predictions. The comparison focuses on ultimate capacity, stiffness characteristics, failure mechanism, and settlement at the point of capacity mobilisation. These parameters reflect the primary aspects of axial helical pile response and provide a clear basis for assessing how well the numerical model reproduces the field performance. A summary of this comparison is presented in Table 5.
Table 5- Comparison of Measured and DeepFND-Predicted Behaviour.
Behavioural Aspect | Field Observation | DeepFND Prediction |
Ultimate capacity | 89–102 kN (P-2: 101.5 kN; P-3: 89.3 kN) | P-2: ~109 kN; P-3: ~100 kN. Slightly higher but within acceptable engineering agreement when using calibrated su and Nₜ. |
Load–settlement stiffness | Moderate initial stiffness; non-linear transition after ~40–60 kN | Very similar stiffness profile using sᵤ-based layers; numerical curves closely follow measured behaviour up to failure. |
Failure mechanism | Individual bearing at helices | IBM reproduces helix-controlled bearing failure accurately, showing the same transition to plunging. |
Settlement at Qᵤ | 20–30 mm | P-2: 3.3 cm; P-3: 3.28 cm. Nearly identical to measured settlements at failure. |
7. Discussion
The comparison between the measured behaviour of piles P-2 and P-3 and the corresponding DeepFND simulations demonstrates that the software reliably reproduces the key features of axial performance in stiff glaciolacustrine clay. The predicted ultimate capacities—for P-2 (~109 kN) and P-3 (~100 kN)—closely match the field values of 101.5 kN and 89.3 kN, respectively, with deviations remaining within an acceptable engineering margin. This alignment indicates that DeepFND’s capacity formulation, when calibrated with realistic undrained shear strength values and appropriate bearing coefficients, is capable of capturing the fundamental bearing mechanisms governing helical pile behaviour.
The predicted load–settlement curves also show strong agreement with the field tests. In both piles, DeepFND replicates the moderate initial stiffness observed under working loads and mirrors the non-linear transition into rapid settlement as the helices approach full mobilisation. The settlements at failure predicted by DeepFND—3.3 cm for both P-2 and P-3—are nearly identical to the field-measured settlements, reinforcing the accuracy of the numerical stiffness representation.
Field data show that the lower helix consistently carries a greater proportion of the axial load, with measured bearing coefficients Nt ranging from approximately 7.7 to 10.4, while the upper helix, affected by installation disturbance, exhibits slightly reduced capacity. DeepFND reflects this distribution well when realistic Nt values and effective shaft lengths are applied. The close agreement between simulated and measured behaviour confirms that the Individual Bearing Method is appropriate for analysing double-helix piles in cohesive soils where individual helix failure governs the response.
Minor discrepancies between measured and simulated curves are attributed primarily to natural stratigraphic variability and the simplifications inherent in layered undrained modelling. These differences, however, remain small and do not alter the overall reliability of the predictions. The results demonstrate that DeepFND provides an accurate and practical analytical framework for interpreting and predicting the axial response of helical piles in stiff clays.
8. Conclusions
The field performance of piles P-2 and P-3 highlights the defining characteristics of helical pile behaviour in glaciolacustrine clay, notably non-linear load–settlement response, individual bearing at the helices, and clear sensitivity to the undrained shear strength profile. DeepFND successfully reproduces these mechanisms when supplied with accurate geotechnical characterisation and carefully defined helix geometry.
The close agreement between measured and simulated ultimate capacities, settlement at failure, and failure mode confirms that DeepFND is well suited for analysing helical piles in cohesive soils. Accurate modelling requires appropriate selection of undrained shear strengths, realistic assignment of bearing coefficients (Nt), and proper representation of helix spacing and diameter. When these conditions are met, DeepFND provides a robust platform for evaluating axial capacity, load transfer behaviour, and settlement performance, supporting both design and interpretation of load tests.
Overall, the consistency between field observations and DeepFND predictions reinforces the software’s capability as a reliable tool for engineering practice involving helical pile foundations in stiff clays.
References
Lanyi-Bennett, S.A., & Deng, L. (2019). Axial load testing of helical pile groups in glaciolacustrine clay. Canadian Geotechnical Journal, 56(2), 187–197. https://doi.org/10.1139/cgj-2017-0425
ASTM International. (2013). Standard Test Methods for Deep Foundations Under Axial Compressive Load (D1143/D1143M-07). ASTM International, West Conshohocken, PA.
Elkasabgy, M., & El Naggar, M.H. (2015). Axial compressive response of large-capacity helical and driven steel piles in cohesive soil. Canadian Geotechnical Journal, 52(2), 224–243.
Elsherbiny, Z.H., & El Naggar, M.H. (2013). Axial compressive capacity of helical piles from field tests and numerical study. Canadian Geotechnical Journal, 50(12), 1191–1203.
Hoyt, R.M., & Clemence, S.P. (1989). Uplift capacity of helical anchors in soil. Proceedings of the 12th International Conference on Soil Mechanics and Foundation Engineering, Rio de Janeiro, 1019–1022.
Perko, H.A. (2009). Helical Piles: A Practical Guide to Design and Installation. Wiley, Hoboken, NJ.
Poulos, H.G., & Davis, E.H. (1980). Pile Foundation Analysis and Design. Wiley, New York.
Randolph, M.F., & Wroth, C.P. (1979). An analytical solution for consolidation around a driven pile. International Journal for Numerical and Analytical Methods in Geomechanics, 3, 217–229.
Robertson, P.K., & Cabal, K.L. (2015). Guide to Cone Penetration Testing for Geotechnical Engineering. Gregg Drilling & Testing Inc., California.
Tappenden, K.M. (2007). Predicting the axial capacity of screw piles installed in western Canadian soils. MSc thesis, University of Alberta, Edmonton, Canada.
Zhang, D. (1999). Predicting capacity of helical screw piles in Alberta soils. MSc thesis, University of Alberta, Edmonton, Canada.
Let us show you how to reduce your design time by up to 90%!