Effect of Pile Section Shape on Bearing Capacity under Dry and Saturated Conditions

1. Introduction

Soil properties around the pile are crucial for assessing its geotechnical capacity, as this capacity comprises two main aspects: load transfer capacity, which directly relates to the interaction between the pile and the soil, and structural capacity, which depends on the pile's material properties. Under soaking conditions, the cementation bonds between soil particles break down, increasing the potential for collapse and reducing the large void ratio, leading to rapid settlement and significant deformation in collapsible soil structures.^1,2

Collapsible soils are problematic soils that exhibit considerable strength when dry but lose strength significantly upon saturation, which can lead to settlement problems. Gypseous soil is especially prone to collapse, influenced by factors like loading, moisture, soil density, and immersion conditions.³

Pile foundations are the most common type of foundation used in geotechnical engineering because of their remarkable carrying capacity, low settlement, high flexibility, and structural stability.⁴

The physical and mechanical properties of gypseous soil, including specific gravity and solubility (2-2.5 g/l), are directly affected by high gypsum concentration, as gypseous soil is a type of collapsed soil.⁵

Samarra Tourist Hotel, Karbala High Water Tank, and the deterioration and collapse under the weight of many facilities in Mosul and Tikrit were among the most significant soil-collapsing issues in Iraq.⁶

Because a down-drag force developed around the pile shaft embedded in collapsible soil, the structures were damaged during saturation.⁷ Determining the degree of gypseous soil severity based on the collapse potential value is essential for assessing the possible damage in the region.⁸

The pile is typically designed using theoretical equations provided by various codes. To verify the pile's load capacity, a pile load test must be conducted. The ultimate load capacity, also known as the load that causes rapid settlement due to a significant or moderate increase in applied load, is the sum of skin friction and end bearing resistance. According to ASTM D-3689(1995), ASTM D 3966-07, and ASTM D-1143, there are four main types of pile load tests: pull-out load, dynamic load, lateral load, and axial compression load. Four primary methods are used to apply axial compression loads in axial compression load tests.⁹

These techniques include the Slow Maintained Load Test (SMT), the Constant Rate of Penetration Test (CPR), and the Swedish Cyclic Test (SC). Any of these approaches has requirements and limits. The CRP method is recommended because, in some methods, a significant load is needed to reach the pile’s plunging failure load.¹⁰

Various criteria have been suggested for determining the ultimate load capacity using graphical methods based on load-settlement curve results. Hansen (1963), Chin-Kondner (1970), Fuller and Hoy (1970), Mazurkiewicz (1972), Buller and Hoy (1977), De Beer and Wallays (1989), Corps of Engineers (1991), Decourt (1999), and others proposed these graphical techniques. In accordance with ASTM 1143, the analysis was conducted using various criteria to determine the capacity of the board pile in Baghdad soil based on the pile load test. Hansen (1963), the Chin-Kondner estimation, and the Log-Log techniques all provided high estimates of the pile’s ultimate load capacity. Due to its simplicity, the Terzaghi method was used to predict the ultimate pile capacity. This method considers the load value corresponding to a settlement of about 10% of the pile diameter.¹¹

The ultimate pile load test for the pile placement in gypseous ground was examined through an experimental simulation. When comparing the load-settlement graph data with the theoretical calculations, several failure criteria were considered, and the most appropriate one was selected. Results obtained using Shen's Approach (1980) were suitable, but the procedures developed by Brinch Hansen in 1963, Decourt extrapolation, and Chin-Konder extrapolation yielded higher values.¹²

Each technique may accurately estimate the ultimate load if the tested load is high and close to the ultimate load limit, but at lower loads, the results tend to be overstated. Davisson and De Beer's techniques cannot be used for non-failed piles in this study; however, Chin, Mazurkiewicz, and Decourt's methods are suitable. As a result, it is challenging to propose a precise strategy for determining the final pile capacity.¹³

Nine techniques were used to evaluate the bearing capacity of driven piles at various locations in Iraq. The most effective methods are De Beer, ChinKonder, and Vander Veen. Although Vander Veen's method is time-consuming, these techniques yield the highest bearing capacity, which is linked to minimal pile settlement. Meanwhile, the Bulter, Hoy, and Brinch Hansen's 90% techniques provided satisfactory capacity with limited settlement, while Fuller, Hoy, and Davison approaches produced acceptable results with sustainable settlement. The results indicated that the Terzaghi method (10% of pile diameter) was overestimated, and a sufficient criterion for this research is 4% of pile diameter.¹⁴

When calculating the ultimate load capacity using a mathematical formula or a computer program, several factors should be considered, including the specific pile type, the applied load, its dimensions, and the material used in its construction. A comparison of the results obtained through the computer program method was conducted using the Brinch-Hansen, Decourt, and Chin-Kondner approaches. Although the Decourt and Chin-Kondner procedures produced results roughly comparable to those of the Brinch-Hansen and software methods, the study suggested that none of these procedures should be used until a settlement failure occurs.¹⁵

The most accurate field-experiment data were obtained using Bazaar and Luciano Decourt’s empirical analysis method. Test results were interpreted with the methods of Chin, Lastiasih, and Mazurkiewicz. Based on the conclusions from these empirical techniques and the interpretation of the field test’s ultimate bearing capacity, pile load data were analyzed using the finite element approach. According to pile loading tests at failure, the study’s findings indicated that the finite element methods by Lastiasih and Luciano Decourt are among the most precise for determining ultimate load capacity.¹⁶

Various alternative interpretation methods were employed to determine pile capacity based on results from pile load tests conducted in three separate areas in Nasiriyah, south of Iraq. Considering that Chin-Kondner's ultimate load for the 22 pile load tests is 22% higher than Hansen's ultimate load, these methods cannot be used to predict pile capacity. The approaches of DeBeer, Decourt, and Mazurkiewicz produced the closest average failure load, but Buttler-Hoy's method showed the smallest failure capacity.¹⁷

Although several research studies have explored the behavior of piles in gypseous and other collapsible soils, the combined influence of pile geometry (shape) and saturation, and its analysis using multiple graphical criteria for ultimate capacity, has not been comprehensively assessed. The research aims to examine the impact of pile geometry and saturation using multiple interpretation criteria that have not been fully explored. The three pile geometries (square, circular, and rectangular) were chosen to investigate how cross-sectional shape affects shaft resistance while maintaining a constant area, despite differing perimeter-to-area ratios. This structural analysis provides a systematic approach, while the physical model allows for direct observation of pile-soil interaction under controlled boundary conditions. Comparing dry and saturated conditions is essential because wet gypseous soil behaves differently mechanically. Gypsum bond dissolution and apparent cohesiveness influence pile performance in both environments, aiding in assessing pile geometry failure susceptibility and establishing safe design parameters. Twelve load-settlement interpretation methods were employed to improve scientific accuracy and minimize judgment bias.

2. Soil and pile properties

The floating bored piles (square 14 mm, circular 16 mm, and rectangular 20×10 mm) used in this study are hollow aluminum, 1.3 mm thick and 250 mm long.

Rigid-type friction piles were used in this study; therefore, hollow piles were selected to minimize the impact on the bases. The lateral roughness was standardized across the three pile types to eliminate it as a variable, allowing the study to focus on the effect of pile geometry on load-bearing capacity. Table 1 presents the geometric properties of the model piles and details of their cross-sectional areas. They are placed in the soil inside the experimental steel box.

The values indicate that the cross-sectional areas are nearly equal. The perimeter increases from circular to square and rectangular piles, which may affect the available shaft surface for skin friction mobilization. The perimeter-to-area ratio is a key dimensionless parameter that may directly influence pile behavior, but shear and loading mechanisms also play significant roles. Air-dried, collapsible, gypseous soil containing 58% gypsum, with a loose dry density of 11.4 kN/m³, was used. It was placed in a steel container measuring 37×80×80 cm with a 6 mm wall thickness to accurately simulate pile behavior on site and maintain boundary conditions.

As shown in Figure 1, sandpaper was used to roughen the pile shafts to generate friction along the embedded length of the piles in gypseous soil.¹⁸ Based on the soil sample's physical characteristics (a mixture of fine to medium-sized particles with a high gypsum concentration, which imparted some cohesiveness), medium-grit sandpaper (80-120 grit) was selected.¹⁹

Figure 1. Shows the aluminum pile models with different geometries and a steel laboratory box container.

The floating bored piles were installed using a steel frame that maintains their vertical position, with an embedded length of 200 mm. The surface roughness of each pile was treated with medium-grit sandpaper to simulate pile shaft friction. Table 2 lists the soil characteristics.

Note that the 250 mm pile length reflects a realistic L/D ratio for a small laboratory model of a rigid pile. This minimizes the influence of test box boundaries and aligns with laboratory measurements of gypseous soil. It has all the features needed to simulate a high-gypsum friction pile. Due to the loose, dry soil, a relative density of 30% effectively mimics natural gypseous soil in certain areas and clearly demonstrates soil settlement from load and saturation. A highly compressible soil with 58% gypsum enables the observation of saturation effects.

Analyzing full-scale pile behavior under specific field conditions requires a detailed model, which is often more costly and time-consuming. The practicality of using a scaled-down model, despite its limitations, must be acknowledged, including stress distributions and pile movements, which can be affected by the soil container's boundaries and by the friction between the container wall and soil particles. The ratio of the pile diameter to mean grain size (D50) should be greater than 35 for vertical loading.²⁰

The distance below the pile tip to ensure it is a friction pile should be about 8d, which is 8(16) = 128 mm. However, the actual distance was approximately 400 mm, and the distance between piles exceeded 7d to ensure a single isolated pile and prevent group effects.

To reduce the internal scale effect between the pile and the tested soil²¹:

3. Pile load test

The Constant Rates of Penetration Testing Method, as defined in ASTM D-1143, was employed in this study by gradually increasing the pressure load to push the pile through a collapsible gypseous soil sample at a steady rate of 1mm per minute. Various criteria can be used to interpret the test's load-settling curve. The soil model was designed to represent three different pile geometries installed in a dry, loose gypseous soil sample.

The rain technique was employed to achieve a relative density of 30%, with a consistent drop height of 25 cm to ensure uniform particle deposition. The resulting dry density of 11.4 kN/m³ was confirmed through three density measurements of the container after deposition. Six layers of gypsum soil were placed in the steel box model, each 10 cm thick, as shown in Figure 2.

Figure 2. Represents the raining technique.

Figure 3 shows that a mechanical jack can be used to apply axial compression force to the pile after soil preparation and pile installation. The load cell, shaped like an 'S,' recorded the load transmitted to the pile, which differs from the ultimate load. Ensure that the weight is transmitted as a central pressure over the pile head; each pile has an aluminum plate cover measuring 10 cm × 10 cm × 2 cm. The load cell was calibrated with known weights, while the LVDTs were calibrated using a micrometer. The recommended static loading rate of 1 mm/min is advised for slowly applied loads in collapsible soils to minimize sudden collapse surges and prevent pore pressure buildup during saturation. Both the LVDT and the load cell are connected to the data logger, which records the LVDT measurements of the pile's settlement. The soil must be saturated for a full day to ensure complete saturation. A tank measuring 45×45×45 cm was constructed and positioned beside the laboratory model box. Water is pumped from the tank to three openings at the base of the model, allowing the soil layers to become saturated from below over 24 hours. The rise of water through the surface layers was observed through a side glass window integrated into the laboratory model. When the readings stabilized and water flooded the upper soil layers, it indicated full saturation.

Figure 3. Illustrates the preparation of the piles and the components for the laboratory test.

4. Pile load settlement test results

A load-settlement curve shows the results of the pile load test for both saturated and unsaturated samples. This study considers three pile shapes: square, round, and rectangular, with equivalent cross-sectional areas. In loose, dry soil with a relative density of 30%, the piles were installed as bored piles. Figure 4 illustrates a simulation of a floating pile on gypseous soil under unsaturated conditions, and Figure 5 presents a soaked test.

Figure 4. Dry load-settlement curves for floating piles (circular, rectangular, and square) with 25 cm length embedded in loose gypseous soil with y dry = 11.4 kN/m3.

Figure 5. Soaked pile load test curves for floating piles (circular, rectangular, and square) with 25 cm length embedded in loose gypseous soil with y dry = 11.4 kN/m3.

Figures 4 and 5 display the measured load-settlement response of the three pile shapes under dry and saturated gypseous soil conditions, respectively. The curves show an initially stiff, nearly linear reaction, followed by gradual nonlinearity and progressive degradation in stiffness. These plots serve as the basis for the 12 graphical and semi-empirical interpretation methods used in this study.

The ultimate load capacity of the piles under the specified conditions has been determined from the provided load-settlement curves. As discussed in the next section (part 5), many factors have been taken into account to complete this review.

5. Failure criteria

Several criteria were used to evaluate the results of the pile-loading experiment for each pile geometry, with a dry unit weight of 11.4 kN/m³ and a loose relative density of 30%, under both wet and non-soaked conditions. The following conditions were applied:

6. Conclusions

The geometry of the pile clearly affects its bearing behavior and the load transfer to the surrounding layer. This effect becomes more pronounced under different soil-moisture conditions. This study uses various failure criteria methods under both dry and saturated conditions to determine the pile's bearing capacity: