Performance of Bored Piles Constructed Using Polymer Fluids: Lessons from European Experience

Since the pioneering work of Veder (1953), bored piles and diaphragm walls around the world have been regularly constructed using bentonite clay slurries to stabilize the excavations prior to concrete placement. A vast database of knowledge and experience with bentonite is now available and industry guidance and specifications exist [Federation of Piling Specialists (FPS) 2006]. However, since bentonite is a natural and finite resource and its use requires bulky ancillary plant for mixing and cleaning, foundation engineers have been in search of a more sustainable and better alternative for many years. Since the 1970s, solutions of natural polymers have been used sporadically in several countries but with mixed results partly due to their biodegradability. Natural polymers are still used today but are limited mainly to the construction of deep drainage trenches and permeable reactive barriers where biodegradation of the fluid is desirable (Day et al. 1999).

For the construction of deep foundations, the breakthrough came in the early 1990s when partially hydrolyzed polyacrylamides (PHPAs) were introduced to the construction industry by some material suppliers in the United States and their use has now spread from North America to other parts of the world. PHPAs are water-soluble synthetic polymers which carry a negative charge on the molecules. PHPAs now used in the foundation industry have a high molecular weight, so that when dissolved in water they form a non-Newtonian solution which may be used in replacement of bentonite slurry for excavation support. Unlike bentonite, polymer fluids do not form a gel when left undisturbed (nonthixotropic) and have negligible yield stress, although they can still have very high viscosity, up to ${10}^5\mathrm{M}\mathrm{P}\mathrm{a}·\mathrm{s}$ at low shear rates (Lam et al. 2015). Polymer fluids have been found to offer many benefits such as smaller site footprint, ease of fluid mixing, and better concrete–sand interface resistance (Lennon et al. 2006; Lam et al. 2014a). However, since polymer fluids are very different from their bentonite counterparts both physically and chemically, their methods of use are also very different (Jefferis and Lam 2013; Lam et al. 2014b) and workmanship issues have arisen on some projects in the past (Berkovitz and Long 1995; Institution of Civil Engineers 2007). Therefore, despite the potential benefits many practicing engineers still remain wary of polymer use (Wheeler 2003). The current situation is also compounded by a lack of industry guidance and specifications. For example, ICE (2007) in the United Kingdom is silent on the required properties of polymer fluids although it specifies bentonite in detail. Similarly, the European standards on the construction of bored piles and diaphragm walls do not provide any guidance on the use of polymers although they are mentioned as a possible medium for excavation support [BS EN 1536 (BSI 2010a); BS EN 1538 (BSI 2010b)]. The same can be said for many other countries. In the United States, AASHTO (2010) gives some required properties for polymer fluids, specifying ranges for density, viscosity, pH, and sand content but the suggested lower and upper limits appear to have been directly transferred from material suppliers’ recommendations. It is considered in this paper that these could be further developed.

To help develop guidance on polymer use, it is important to garner the experience available from published case histories, which are scattered through the literature and may also be written in languages other than English. To achieve this objective, a critical reappraisal of six selected European case histories has been conducted and the results summarized in this paper. The choice of this limited geographic area makes the task more manageable. Only case histories that contain a reasonable amount of site and construction information are included. In the subsequent sections, six different cases from the United Kingdom, Portugal, Italy, and Germany are discussed individually. During the discussions, the writers’ views are offered to explain some of the highlighted (non)performance issues. To aid the subsequent discussions, Fig. 1 shows the soil profiles of the test sites that will be discussed.

Polymer fluids were used to construct the bored piles supporting the Vasco da Gama Bridge across the Tagus River in Lisbon. This project was documented by several teams of researchers including Bustamante et al. (1998), Sêco e Pinto and Oliveira (1998), Manuel Correia and Sêco e Pinto (1999), and Guadagnini (2001). Mr. G. Guadagnini provided additional information to the writers. As shown in Fig. 1, the ground conditions below the river bed consisted of 35 m of very soft silty clay (Layer 1), which was underlain by 9 m of medium dense to dense silty sands (Layer 2), 6 m of hard clay (Layer 3), and then at least 7 m of very dense sandy gravel with pebbles (Layer 4). The polymer used for this project was a high-molecular-weight PHPA marketed as Geomud-15. Brackish water from the Tagus River was used to mix the polymer at a concentration of $2{\mathrm{kg}/\mathrm{m}}^3$. The Marsh funnel viscosity of the fluid was 40 s. Although this viscosity value is rather low for PHPA polymers, it is within the expected range if brackish river water was used to prepare the fluids. The Marsh funnel efflux time for clean water is $26\pm 0.5\mathrm{s}$.

Bustamante et al. (1998) and Guadagnini (2001) presented the load test result for a 1.2-m diameter instrumented pile constructed at the Pylon South (PS) location. This pile had a total length of 60.8 m, of which 52 m was embedded into the river bed. To evaluate the behavior of base (toe) grouting, after the first load test to 17.5 MN, the base of the pile was grouted in three stages (total injection volume 2,076 L) and the pile was tested again. Fig. 8 shows the load–movement curve. Base grouting caused in a much stiffer response and also probably a higher ultimate capacity. Stiff pile response was also observed in a trial in Taiwan where the pile was also constructed using polymer fluids and was base grouted in several stages (Duann et al. 2004).

From the load test results, ultimate unit shaft resistance ($f_s$) values were derived for each of the three soil layers and presented by Bustamante et al. (1998). By comparing these $f_s$ values with those obtained from other sites, the original researchers concluded that polymer fluids did not lead to any reduction in shaft resistance, a finding that corresponds with the conclusions from the United Kingdom case histories discussed previously. Because of the satisfactory performance of the test pile, polymer fluids were allowed to be used for the construction of 278 working piles; these had larger diameters ranging from 2.0 to 2.2 m. At the time this paper was written, these piles are still some of the largest ever formed under polymer fluids. Another 548 working piles were also constructed using bentonite slurry by a different contractor. Of those piles formed under polymer fluids, three were found by sonic logging tests to have defects at the pile toes. This was equivalent to a defective rate of only 1%. This appears to suggest that polymer fluids do not adversely affect the structural integrity of piles.

Despite the encouraging experiences reported previously, a major problem was also encountered on this project; two of the working piles supported by the Geomud-15 fluid had to be redrilled using CDP polymer fluids. This was because the Geomud-15 fluid became contaminated with the in situ soil during excavation and lost its properties. The supplier of the CDP polymer noted that the Geomud-15 fluid was contaminated by the large amount of calcium and magnesium ions in the ground due to a nearby salt flats (KB Technologies 2000). The adverse effect of dissolved salts (cations) on the viscosity of PHPA polymer fluid was also observed by Schwarz and Lange (2004) in another piling project in Benin and by Jefferis and Lam (2013) in a laboratory trial. In the writers’ opinion, the two collapses might have been prevented if (1) the Geomud-15 polymer fluid had been mixed at a higher concentration, or (2) potable water had been used rather than brackish waters to prevent the adverse effect of dissolved ions in the water; Point 2 also applies to bentonite slurry because dissolved salts can inhibit the dispersion of the bentonite (clay) particles (FPS 2006). To summarize, to avoid potential problems caused by fluid contamination, it is recommended that the salt contents of soils and pore waters are checked in advance when polymer fluids are proposed, if saline conditions are suspected. While on site, it would also be prudent to run the fluid at a concentration (viscosity) higher than would otherwise be required to create a buffer.

Bustamante et al. (1998) reported on the piling works of the Naples–Rome section of the Italian high-speed train [Treno Alta Velocità (TAV)] project, for which a total of 1,732 bored piles were required to support the proposed viaducts. About half of piles were formed using bentonite slurries with the remainder under Geomud-15 polymer fluids. The polymer fluids were mixed at a concentration of $0.6\mathrm{kg}/{\mathrm{m}}^3$ which gave a Marsh funnel viscosity of 48 s. Compared to the Vasco da Gama Bridge project, the dosage of the polymer was lower but the viscosity was higher. This was probably due to the use of potable rather than brackish waters for mixing.

Two static pile tests were carried out [at the (1) Peccia-1, and (2) Cassino South-2 sites], where pyroclastic soil consisting of pumice and lapilli were present (Fig. 1). Pile 1 at Peccia-1 had a diameter of 1.2 m and a total length of 33 m. Fig. 9 shows the load test result. Due to the limited reaction load, the pile was tested only to 13.8 MN although the deduced ultimate capacity was over 20 MN. The derived unit shaft resistance ($f_s$) curves showed that, except near the pile top, the shaft resistance of this pile was not fully mobilized during the test so that the ultimate (highest possible) values were not achieved during the test. Pile 2, which was located at the Cassino South-2 site, also did not reach its ultimate resistance during the load test. Nevertheless, by comparing the mobilized $f_s$ values with those obtained for continuous-flight-auger (CFA) piles from similar sites, the original researchers concluded that the polymer fluid did not lead to any reduction in shaft resistance in the pyroclastic soils. This conclusion is in line with the results from the United Kingdom and Portuguese cases described previously.

Following the satisfactory load test results, polymer fluids were used to construct the working piles at eight viaduct sites, all of which were sited on pyroclastic grounds. A total of 951 piles with a diameter of 1.2 m and up to 42-m deep were constructed. However, it was reported that at one site (Pisciarello) bentonite fluids had to be used instead due to the markedly cohesionless nature of the soil. In the writers’ opinion, this was probably due to the relatively low Marsh funnel time of the polymer fluids used (48 s) as for PHPA polymer a funnel time of 60 s is now the norm. If the fluid at the Pisciarello site had been mixed to a higher viscosity, the use of bentonite might not have been necessary.

Based on the results of structural integrity tests conducted on the completed working piles, it was revealed that about 10% of the piles had minor defects at the toes, although none required any repair work. This defective rate is significantly higher than that reported in the Vasco da Gama Bridge project although the defects were more minor. Although the original researchers did not explain the cause of the defects, the writers consider that the defects possibly may have been due to sediment accumulating at the base of the bores over the period from the end of excavation to the first pour of concrete. Unlike bentonite, polymer fluids have a negligible yield stress so that they cannot effectively hold soil particles in suspension. Careful cleaning of the used fluids and of the pile bases thus becomes extremely important when polymer fluids are used. Any sediment that is left at the base of a pile bore may not be displaced during casting. To further analyze the development of defects, it would have been useful to have had more information on the base and fluid cleaning procedures adopted by the contractors.

Lesemann (2010) reported a field trial near the Munich Airport in Germany. The ground conditions consist of sand and gravel mixtures with an average permeability ($k$) of approximately $5\times {10}^{-3}\mathrm{m}/\mathrm{s}$. The ground water table was located near the surface (Fig. 1). The trial was conducted to assess the performance of polymer fluids in highly permeable coarse grounds. To this end, six 0.6-m diameter bored piles (Piles P1–P6) with a length of 10 m were constructed. Three types of polymer, namely (1) polyacrylamide (PAA), (2) carboxymethyl cellulose (CAM), and (3) XAN xanthan gum (XAM), were used for the trial. Bentonite slurry was also used to construct a pile for comparison. Table 1 summarizes the support fluid information for each pile.

To assess the stability of the pile bores, profiling of the side walls was carried out using the ultrasonic technique after excavation. Negligible difference was found between the side-wall profiles of the piles. This finding disproves the common perception that only bentonite slurry can stabilize excavations in gravely soil due to its ability to seal the surface by forming a layer of filter cake. As mentioned previously, polymer fluids were also successfully used to stabilize a sandy gravel layer with pebbles for the Vasco da Gama Bridge project. These experiences demonstrate that a layer of filter cake is unlikely to be a prerequisite for a stable fluid-supported excavation. For polymer fluids, the key is probably sufficient viscosity coupled with clogging of the coarser soil pores with the finer excavated materials. If fluid viscosity is not maintained due to salt contamination or other reasons, the stability of an excavation may deteriorate as previously shown by the bore collapse at the Vasco da Gama Bridge site.

After the construction of the test piles, Piles P3 (bentonite at $50\mathrm{kg}/{\mathrm{m}}^3$) and P4 (XAN at $2\mathrm{kg}/{\mathrm{m}}^3$) were load tested in compression using the other four piles to provide the reaction (tension) forces. Fig. 10 shows the load–movement curves for the six piles. Although the performance of both compression piles exceeded expectations, Pile P3 settled 10 mm less than Pile P4 did under the maximum load of about 5 MN. This finding is contrary to those obtained from the other case histories discussed previously, which all showed better pile performance when polymer fluids were used. The reason why this was not the case was possibly due to the low polymer concentration used for Pile P4. Typically, for foundation drilling a concentration of $3\mathrm{kg}/{\mathrm{m}}^3$ would be used for a natural xanthan polymer (Beresford et al. 1989) whereas only $2\mathrm{kg}/{\mathrm{m}}^3$ was used for Pile P4. As a result, the Marsh funnel viscosity for Pile P4, which was 38 s, was the lowest among all the piles constructed under polymers (Table 1). The effect of polymer concentration can also be seen by comparing the load–movement curves of Piles P4 (tested in compression) and P6 (tested in tension), both of which were excavated under xanthan fluids but at different concentrations (2 and $4\mathrm{kg}/{\mathrm{m}}^3$, respectively). When the applied head load is small, the effect of pile base resistance is often small or negligible so the response of these two piles can be compared. Fig. 10 shows that the initial load–movement response of Pile P6 was much stiffer than that of Pile P4. This suggests better excavation support for the former, although the original researchers also cited a flaw in Pile P4 as a reason for its inferior performance.

Piles P1 and P2 are another pair that is worthy of discussion, as they are also excavated under the same type of polymer fluid (i.e., PAA) but at different concentrations (6 and $2\mathrm{kg}/{\mathrm{m}}^3$). As shown in Fig. 10, pile P1 showed much larger movement than Pile P2 although the former was excavated under higher concentration (viscosity) fluid. The inferior performance of Pile P1 can be explained by the fact that concreting of this pile was only successful at the third attempt when the support fluid was replaced with water. During the first two failed attempts, the pile bore was supported by high-viscosity polymer fluids (funnel viscosity, 190–288 s) and the tremied concrete stiffened prematurely. The original researchers postulated several reasons for the failed attempts including chemical interaction between the concrete and the high-viscosity fluid. In the light of this experience, it may be prudent not to use polymer fluids prepared at a very high concentration or viscosity. Jones and Holt (2004) also noted that polymer fluids with a viscosity of 60 s appeared to have less detrimental effect on rebar–concrete bond strength than polymer fluids with a viscosity of 100 s. To conclude, this German case history highlights the importance of selecting a suitable concentration for the particular type of polymer used. Too low or too high a polymer dosage is likely to cause problems or unsatisfactory performance.

The following conclusions can be drawn from the case histories examined in this paper.

Pile bore stability. The field experience gained in Norfolk, United Kingdom, suggests that water cannot provide adequate support to excavation in very weak chalk as shown by the irregular side-wall profile and large concrete overbreak. On the other hand, polymer fluids can stabilize such soils without problems. The German case history demonstrates that polymer fluids, despite having no gel and no ability to form filter cakes, can be successfully used in gravel–sand mixtures with an average permeability of $5\times {10}^{-3}\mathrm{m}/\mathrm{s}$. This finding challenges the common perception that only bentonite slurry can stabilize very coarse soil. The Italian case history also confirmed that polymer fluids can be used to stabilize coarse pyroclastic soils. In terms of excavation size, polymer fluids have been successfully used to support pile bores with a diameter of up to 2.2 m for the Vasco da Gama Bridge project in Portugal. However, collapses have also occurred twice there due to polymer contamination by salts in the ground. Together, these experiences show that polymer fluids are capable of supporting excavations in a wide range of ground conditions but care has to be taken to maintain the properties (including viscosity) of the fluids.

Marsh funnel viscosity. The previous discussions indicate that the chosen polymer concentration, and thus the viscosity of the polymer fluid, plays a substantial role in the successes and failures seen at the various sites. For the Portuguese and Italian sites, the PHPA (Geomud-15) polymer fluids were run at a Marsh funnel viscosity of 40 and 48 s, respectively, without major problems, although in the writers’ opinion higher viscosity could have helped to prevent the collapses at the Portuguese site near the salt flats and avoided the use of bentonite at the Italian Pisciarello site. Equally, very high viscosity should also be avoided to prevent problems during concreting as shown in the German case. In the light of these experiences, until further evidence becomes available, contractors are advised to run PHPA polymer fluids a viscosity of at least 60 s but no higher than 100 s. This represents a tighter requirement than that currently recommended in AASHTO (2010) which specifies a viscosity range of between 32 and 135 s.

Concrete quality and structural integrity. The London Stratford case history has shown that there is little difference between bentonite and polymer fluids in terms of their effect on the quality of hardened concrete. Piles formed under polymer fluids have occasionally been found to have minor defects at the pile toes as shown by the Portuguese and the Italian cases. The writers note that defects can be formed if sediment is allowed to accumulate at the base of a pile during the period between end of excavation and the first pour of concrete, a period that can be up to several hours.

Pile performance. The experiences gained in Norfolk (chalk), East London (stiff clay and dense sand), Lisbon (mixed geology), and Italy (pyroclastic soil) all show that polymer fluids gave excellent load-movement characteristics to the completed piles. In addition, the two London case histories independently confirm that increasing the construction time to 26–37 h has negligible effect on the performance of the completed piles. This is possible because of the ability of polymer molecules to prevent the swelling of the clay soils. The German case history, however, shows that a pile formed under xanthan gum fluids settled slightly more than a similar pile formed under bentonite. This was believed to be caused by the low polymer concentration used ($2\mathrm{kg}/{\mathrm{m}}^3$) as the typical dosage for this polymer type is around $3\mathrm{kg}/{\mathrm{m}}^3$.

The case histories discussed in this paper were studied as part of research projects funded by the U.K. Engineering and Physical Sciences Research Council (EPSRC) under Grants EP/H50026X/1 (Knowledge Transfer Secondment) and EP/K503782/1 (Impact Acceleration Account). Mr. Giancarlo Guadagnini (ENSER S.r.l.) provided additional information on the Vasco Da Gama Bridge project. Messrs. Duncan Nicholson (Arup) and John Crack (Canary Wharf Contractors Ltd.) permitted use of previously unpublished data collected for the Canary Wharf BP1 project. Constructive comments on the draft were given by peer reviewers.