Introduction and scope

The city of Taranto in the south of Italy is one of the areas declared as “at high risk of environmental crisis” by the national government (Italian Law n. 349 1986) because it represents one of the most complex industrial sites in Europe, located near urban areas of high population density. All the industrial activities are responsible for the high environmental contamination, mainly due to heavy metals and organic pollutants. This is why Taranto has been recently included into the list of polluted Sites of National Interest (SIN) by the Italian Law n. 426 1998, for which the environmental remediation has been identified as a national priority (Ministerial Decree n. 468 2002).

The Mar Piccolo (literally, Small Sea) basin is located in the northern area of Taranto city (Fig. 1a). It is a semi-enclosed shallow coastal basin (total surface of 20.72 km2) with lagoon features, divided into the so-called first and second bosom. It is presently connected with the open sea (Mar Grande in the Ionian Sea) by two channels, the Navigabile Channel and the Porta Napoli Channel (Fig. 1a). As a consequence of the heavy industrialisation experienced by the Taranto city during the last 50 years, the complex ecosystem of the Mar Piccolo area started exhibiting unconfutable signs of environmental pollution enhanced by several uncontrolled discharged sewages, the activities of the Arsenal of the main naval base of the Italian Navy and the fishing-boat fleet. A number of researchers (e.g. Lerario et al. 2003; ICRAM-APAT 2007; Cardellicchio et al. 2007, 2009; ISPRA 2010; Petronio et al. 2012; Di Leo et al. 2013) have shown that also the submarine sediments in the Mar Piccolo contain high concentrations of heavy metals (e.g. Hg, Pb, Cd, Cu and Zn) and organic pollutants (PCBs, PAHs and dioxins).

Fig. 1
figure 1

a Gulf of Taranto and sampling area within the first bosom of Mar Piccolo; b location of the three boreholes (1I, 1L and 1M) and of the three samples taken by scuba divers (1I-M, 1L-M and 1M-M) within the “170-ha area”

Full size image

This experimental research is part of the multidisciplinary studies that have been funded by the Regional Agency for Environmental Protection (ARPA-Puglia), and financially supported by specific national legislative procedures (Italian Law n. 129 2012), for a preliminary selection of sustainable strategies for the remediation and management of the Mar Piccolo environmental contamination. Among the different aspects involved in the environmental studies, i.e. sediments, water and biota, this paper focuses on submarine sediments and reports and implements the results of the first geotechnical investigation carried out in the Mar Piccolo (Federico and Vitone 2014). This up-to-date knowledge contributes to the definition of a site geotechnical model of reference for the selection, design, placement and management of the potential remedy approaches (e.g. Rolling 2000; Lee 2001; USEPA 2005).

In general, the transport of contaminants can be caused by mechanisms like molecular diffusion, pore water advection and bioturbation. Pore water advection from the sediment can be driven by two mechanisms: (i) submarine groundwater discharge and (ii) consolidation of contaminated sediment. The analysis of both these mechanisms requires the knowledge of the geotechnical properties of the sediments, together with that of the in situ chemical, hydrological, hydrogeological and geological conditions. Transport of metals from capped contaminated sediments due to submarine vertical flow from the artesian aquifer at the base has been studied by Liu et al. (2001), as an example of hydrogeological conditions that cause groundwater release through the seafloor.

However, for fine-grained contaminated sediments, like those present in the Mar Piccolo, the main mechanism of pore water advection is caused by the consolidation of the contaminated sediment. As the sediments are loaded (e.g. by the placement of a cap) or unloaded (e.g. by dredging operations), due to change of effective stress they will change in volume. Since the submarine sediments are soft saturated soils, their change in volume must be due to the water seepage and so the compression of the contaminated sediments is rather like squeezing contaminated water from a soft sponge of large pores. Since the contaminated water can seep to the capped surface but also towards natural drainage layers (if any), the knowledge of both the hydrogeological and geotechnical boundary conditions within the entire volume of soil interested by the remediation solution is necessary. The sufficient time for the water to seep through the soil to permit the volume change to occur is strictly dependant on the consolidation properties of the soil. In this respect, some researchers have highlighted the relevance of the knowledge of consolidation properties of the contaminated soils in order to predict the transport of contaminants by pore water advection. For example, Lenhart et al. (2009) performed tests of one-dimensional consolidation in the laboratory to simulate in situ fluxes using sediments contaminated by polychlorinated biphenyl, dredged from the Grand Calumet River in Gary, Indiana (USA). The tests were conducted both without a cap and using three capping materials (quartz sand, activated carbon and a proprietary organoclay). Eek et al. (2007) explored, instead, how different capping materials affect the metal concentration in pore water released from contaminated soft-silty clays from the Oslo harbour.

As remarked by USEPA (2005), another relevant geotechnical issue is the bearing capacity of the sediments for the cap stability. This has to be checked both immediately after placement (short term behaviour), i.e. before any excess pore water pressure due to the weight of the cap has dissipated, and in the long term, when the excess pore pressure has been dissipated and the soil is fully consolidated and drained. This means that also both the effective strength parameters (i.e. effective cohesion, c’, and friction angle, ф’) and the undrained shear strength (c u) of sediments within the volume interested by remediation (i.e. the so-called engineering volume) are geotechnical information of paramount importance.

Therefore, the definition of an appropriate site geotechnical model is a compulsory step for the selection of potential remedy approaches. Specifically, it has to include at least the following data (USEPA 2005): (1) the soil composition and index properties; (2) the consolidation properties; (3) the strength parameters; and (4) the rates and direction of fluxes within the sediments. These data have to cover the entire engineering volume.

The aim of this study is to define a preliminary site geotechnical model, containing the essential information useful for the selection, design, placement and management of the potential mitigation solutions. The research has required an in situ investigation campaign, based on three sampling stations that have been of use also for the research groups of the National Project RITMARE for sampling the different environmental matrices (Cardellicchio et al. 2015). The geotechnical data result from an intensive investigation that has been carried out in the laboratory on both remoulded and undisturbed samples retrieved at several depths in the three stations. The research had to face with technical difficulties both in situ, for the off-shore operations that had to be carried out in a military area, and in the laboratory, for the very low consistency of the sediments from the top layer and the contaminants present, which even damaged the laboratory equipment.

The comparison of the results with others from the literature (Skempton 1970; Perret et al. 1995; Cafaro and Cotecchia 2001; Locat et al. 2002; Cotecchia et al. 2007; Levesque et al. 2007) helped in the interpretation of the site geotechnical model and shed light on some discrepancies from the expected soil mechanical response, especially with reference to the most polluted sediments within the top layer of the stratigraphic succession.

Geological setting and submarine sampling of the sediments

The geological setting of the Taranto area is characterised by the main Quaternary tectonic and sedimentary phases that involved the entire Apulian Foreland and the Bradanic Foredeep. Along the Mar Piccolo cliffs, Late Pleistocene deposits crop out with large lateral continuity (Mastronuzzi 2014; Lisco et al. 2015). In this area, the phase of extensive subsidence (Ciaranfi et al. 1988) is recorded by the thick stratum of Upper Pliocene–Lower Pleistocene Calcarenite di Gravina Formation (Azzaroli 1968; Robba 1969; Iannone and Pieri 1979) which transgressively overlies Mesozoic carbonate rocks, and, in turn, is covered by the Subapennine Clays or Argille Subappennine (Azzaroli et al. 1968). They are 10 to 65 m thick and commonly known as blue clays, being grey-blue to grey-green coloured. The subsequent uplift phase, active at least from the beginning of the Middle Pleistocene (Ciaranfi et al. 1988; Ricchetti et al. 1988; Doglioni et al. 1994), is recorded by uplifted marine and fluvial terraces (Tropeano et al. 2002; Gallicchio et al. 2014; Gioia et al. 2014; Lisco et al. 2015).

The two depressions of sub-elliptical shape that constitute the Mar Piccolo are river valleys that have been incised in a continental phase (Ashley and Sheridan 1994; Zaitlin et al. 1994) and then submerged by the sea with the Holocene marine transgression (Mastronuzzi and Sansò 1998, 2002, 2003). As outlined by Cotecchia et al. (1989) and Lisco et al. (2015), the submarine stratigraphic sequence in the Mar Piccolo, from the bottom to the top, is represented by the Late Cretaceous limestone of the Calcare di Altamura Formation, the Calcarenite di Gravina Formation that passes upward in the Argille Subappennine and, finally, the Late Pleistocene–Holocene Mar Piccolo sediments. In particular, a surficial sediment thickness map has been reported by Lisco et al. (2015), resulting from sub-bottom profiler surveys (SBP Innomar 2000; frequencies, 10–12–15 kHz). According to the authors, the thickness of the shallow sediments in the first bosom ranges, on average, from 2 to 3 m. The most relevant exception is recorded in the area between the two bosoms, where a channel has been detected and the layer of recent sediments is characterised by the highest thickness, i.e. up to about 10 m (Lisco et al. 2015). The three continuous corings were carried out in proximity of this area, into the so-called 170 ha area (Fig. 1b), that is one of those prioritised for the high levels of heavy metals and organic pollutants (ICRAM-APAT 2007; ISPRA 2010). The corings were carried out by means of a drilling apparatus that was mounted on a platform previously fixed in place. Once recovered, the cores were wrapped in plastic film and placed into catalogue boxes. The quantity of material recovered was very low due to the very soft consistency of the sediments. In particular, for the shallower sediments, one catalogue box compartment (of 1 m length) could contain up to 6.5 m of sediments (Fig. 2). Samples of sediments were taken up to depths of about 20 m from the seafloor that is about the maximum depth of engineering interest in view of the placement of any mitigation solution. To the best of our knowledge, this study is the first in the Mar Piccolo that covers depths larger than 5 m below the seafloor. In particular, coring depths of 8.5, 16.5 and 18.2 m from the seafloor were reached in boreholes 1I, 1L and 1M, respectively. The sea depth in the three stations varied from 12.25 to 13.00 m. Three undisturbed samples were taken along boreholes 1L and 1M at different depths (i.e. 11.5, 16 and 18 m from the seafloor). Moreover, in proximity of the boreholes, other three samples were taken by scuba divers using 101-mm diameters PVC core samplers that were manually inserted into the first 2 m of sediments (samples 1I-M, 1L-M and 1M-M in Fig. 1b).

Fig. 2
figure 2

Catalogue box of continuous corings 1I (a), 1L (b) and 1M (c). The length of each compartment of the catalogue boxes is 1 m. The numbers refer to metres of continuous coring below seafloor

Full size image

Geotechnical characterisation of the submarine sediments

Composition and physical properties

Figure 3a, b shows the grain-size distribution of sediment samples taken up to 6 m from the seafloor. The percentage of particles larger than 1 mm (Fig. 3a), which are mainly constituted by fragments of gastropods and bivalves, is particularly high (11–25 %) in the shallow samples taken by scuba divers (1I-M, 1L-M and 1M-M). The grading curves of the shallow samples are substantially similar, unless the highest sand fraction (SF) of the 1M-M sample. Samples 1I and 1L have similar silt fraction (MF) but higher clay fraction (CF) than the corresponding samples taken by scuba divers (i.e. 1I-M and 1L-M).

Fig. 3
figure 3

a Grading curves within the first 6 m of sediments along the boreholes 1I, 1L and 1M. b The same grading curves, without particle diameters larger than 1 mm. c Grading fractions along borehole 1L. d Grading fractions along borehole 1M. Key: CF clay fraction, MF silt fraction, SF sand fraction, GF gravel fraction, M manual sampling, U undisturbed sample

Full size image

The vertical profiles (Fig. 3c, d) of the grading fractions at various depths below the seafloor (i.e. up to 16.50 and 18.00 m in boreholes 1L and 1M, respectively) show a trend of SF decrease and CF increase. Also, the sediment composition within the first 8.5 m from the seafloor of the borehole 1I confirms this trend (Table 1). Moreover, the figures show that the sediments taken from borehole 1L exhibit, in general, higher CF (CF max = 54 %) than those sampled down borehole 1M. All these data appear to suggest a high spatial variability of the soil composition. The organic content (O) in the few samples tested is rather high (5.0 and 5.9 %) in the first metres of sediments and decreases to 0.7 % at depth (18 m below the seafloor; Table 1). The specific gravity (Gs) of the soil particles increases with depth from 2.56 to 2.74.

Table 1 Composition and physical properties of the sediments
Full size table

The plasticity properties of the sediments, determined by using distilled water (in compliance with ASTM standards), are shown in Fig. 4. The first consideration that can be made is that, for all the three boreholes, both the liquid limit (w L ) and plasticity index (PI) tend to decrease with depth (Table 1). Despite their spatial variability, the sediments from boreholes 1L and 1I and the upper part of borehole 1M can be classified, according to the Unified Soil Classification System (USCS), as inorganic clays of high plasticity (CH) (Fig. 4), whereas those from borehole 1M, are low plasticity clays (CL). Furthermore, sample 1M-M can be classified as silt of high plasticity (MH). In Fig. 4, the plasticity properties - w L and PI - of this sample are also compared with those determined by using sea water from the Mar Piccolo. The almost complete coincidence of the two determinations suggests that, at least for this sample, the plasticity properties are not appreciably affected by water salinity.

Fig. 4
figure 4

Casagrande plasticity chart. In the circle, the plasticity properties of sample 1M-M are compared with those determined by using sea water

Full size image

The activity (A) of the sediments varies between 0.5 and 1.7 (Fig. 5). In particular, the samples taken by scuba divers within the top 2 m exhibit very high activity (A max = 1.3-1.7). The less active samples are those taken at larger depth.

Fig. 5
figure 5

Activity of the soil samples

Full size image

Figure 6a, b shows that both the void ratio (e) and the consistency index (CI) of the soil samples follow similar trends when plotted against depth below sea level. The average e of the samples varies between 3.16 and 4.39, in the first 4 m below seafloor, to about 1.33 at depth (borehole 1M). With reference to the soil consistency, almost all CI values are below zero (Fig. 6b): the minimum values (about −2.2) are those of sample 1M-M and of the shallow sediments taken from 1L. It can be observed that the upper part of the stratigraphic succession, about 4 m thick, is characterised by CI values lower than −1.1 (Fig. 6b, c; Table 1). Below this top layer, the lower part of the stratigraphic succession exhibits a sharp decrease in void ratio, whereas the index properties do not vary significantly. This makes the soil consistency rise up to CI = −0.55, although remaining negative. It further increases with depth (the maximum value, CI = 0.24, is that of the undisturbed sample 1L-U), but values lower than zero have been recorded up to 18 m below the seafloor (borehole 1M).

Fig. 6
figure 6

Vertical profiles of a void ratio and b Consistency Index (CI); c photograph of sample 1M-M

Full size image

In Table 2, the results of mineralogical analyses carried out on three specimens sampled down 1M borehole are reported. The X-ray diffractometry on the total sample shows the widespread presence of phyllosilicates; minor constituents are quartz, feldspars and carbonates (calcite and dolomite, while aragonite is in traces). In particular, the phyllosilicates are interstratified illite/smectite (I/S), chlorite/smectite (Chl/S), kaolinite and chlorite. Halite is also present for the marine environment from which samples have been taken and it is abundant in the shallow sample 1M-M. Moreover, hematite is significantly present, mostly dragged by the wind from the iron ores heaps of the near steel factory. Table 2 also highlights that, except for halite and hematite, the mineralogical composition of the samples does not present significant variations with depth.

Table 2 Mineralogical composition (weight %) of the bulk sample and in the clay fraction (in parentheses)
Full size table

Compression and shearing behaviour of the soils

Nine oedometer tests have been carried out on samples of submarine sediments taken within the first metre up to 16.3 m below the seafloor (i.e. the specimen 1M-U). The testing programme, that has followed ASTM standards, is summarised in Table 3. The one-dimensional compression curves of the specimens are reported in Fig. 7 in the e - vertical effective stress (σ’v) plane.

Table 3 Loading sequence in oedometer tests
Full size table
Fig. 7
figure 7

Loading/unloading oedometer compression curves of the samples

Full size image

Samples from the top of the stratigraphic succession (i.e. the first 4 m of sediments) exhibit similar behaviour and their preconsolidation pressures (arrows in Fig. 7) seem to be either lower or equal to their overburden effective pressure (i.e. σ’p = 6–8 kPa). As expected, the undisturbed specimens 1L-U and 1M-U, taken from 11 and 16 m below the seafloor, respectively, exhibit the higher values of preconsolidation pressures. The compression index (C c) of the sediments (Fig. 8a) varies between about 0.7 and 1.6 in the first 4 m from the seafloor, whereas it becomes significantly lower (C c = 0.25–0.4) for the three deeper specimens (i.e. 1M, 1L-U and 1M-U). The irregular trend of the compression index of the specimens taken within the first meter of sediments (Fig. 8a) has to be imputed to the presence of algae and fragments of bivalves and gastropods.

Fig. 8
figure 8

a Compression index versus vertical effective stress; b swelling index versus vertical effective stress

Full size image

The oedometer test results are consistent with the composition and plasticity properties of the sediments: specimens 1I-M and 1L-M, of higher plasticity index (PI) and lower sand fraction (SF), exhibit higher initial void ratio and compressibility with respect to those from borehole 1M (1M-M and 1M-M2). In particular, the compression index of the top 0.5 m of sediments (specimens 1I-M and 1L-M) decreases in loading, becoming lower than 1 only for vertical effective stress values, σ’v, higher than 100 kPa. The swelling index (C s) varies from 0.027 (specimen 1M) to 0.092 (specimen 1L-U) (Fig. 8b).

The values of the coefficient of permeability, k, of the shallow to medium depth sediments are plotted in Fig. 9 against void ratio. They have been computed according to the consolidation theory by Terzaghi (1923), in both oedometer loading and unloading phases.

Fig. 9
figure 9

Coefficient of permeability versus void ratio (data in loading and unloading)

Full size image

For all the specimens from the top 4 m of the stratigraphic succession, k in loading varies between 8*10−11 and 7*10−9 m/s and, in general, it decreases with void ratio. Specimens from medium depth 1M and 1L-U exhibit, for the same void ratios, permeability values that are higher than those of the shallower specimens of about one order of magnitude. In unloading, k is always lower than that in loading and it seems not to vary with the unloading phase (k unl = 7*10−12 − 2*10−11 m/s; Fig. 9).

Direct shear tests (DST) have been carried out on three specimens from sample 1I-M (0–0.5 m from seafloor). They were sheared after consolidation to vertical effective pressures from 25 to 75 kPa. In Fig. 10, the state paths followed by these three specimens during consolidation before shearing are reported together with the oedometer (OED) compression curve of the same sample. The data show that the DST specimens consolidation states lie either close or just in correspondence of the normal compression line of the OED sample. During shearing, according to their consolidation states, the specimens harden and contract up to large strains, following the so-called wet behaviour of Critical State Soil Mechanics (CSSM) (Roscoe et al. 1958; Schofield and Wroth 1968). Thus, these specimens reach a pseudo-critical state, characterised by strength parameters equal to c’ = 0 kPa and φ’cs = 35° (Fig. 11).

Fig. 10
figure 10

Sediments from the top layer: consolidation states during compression in oedometer (OED) and direct shear tests (DST)

Full size image
Fig. 11
figure 11

Sample 1I-M, 0.00–0.50 m depth below seafloor: direct shear test results

Full size image

The shearing behaviour of the lower part of the stratigraphic succession has been investigated by means of DST carried out on three specimens from sample 1M-U (16.0–16.3 m below the seafloor). The three specimens were consolidated from 12 to 150 kPa before shearing (Fig. 12). The oedometer compression curve of the same sample is also plotted (Fig. 12). As for the shallow sediments, also in this case the consolidation states before shearing of the DST specimens are consistent with those followed by the OED specimen. Moreover, according to its location in the e–log σ’v plot, that is further to the left of the normal consolidation line of the 1M-U sample, the specimen sheared at lower vertical effective stress (σ’v = 12 kPa), of overconsolidation ratio (i.e. OCR = σ’p/σ’v) equal to about 10, strain-softens (Fig. 13a) and dilates (Fig. 13b) according to the so-called dry behaviour of CSSM. On the contrary, the specimen sheared at σ’v = 150 kPa exhibits a wet behaviour according to the OCR value, that is equal to 1 (Fig. 13a, b). The third specimen, that has been sheared at 50 kPa, has an intermediate behaviour, i.e. still in agreement with its OCR value (OCR = 2). The soil strength parameters at peak are c’p = 15 kPa and ф’p = 23.1°, whereas the friction angle at critical state is about ф’cs = 27.8° (Fig. 14).

Fig. 12
figure 12

Sample 1M-U, 16.00–16.30 m depth below seafloor: consolidation states during compression in oedometer (OED) and direct shear tests (DST)

Full size image
Fig. 13
figure 13

Direct shear test results (sample 1M-U, 16.00–16.30 m depth below seafloor): a shear stress versus horizontal displacement; b vertical settlement versus horizontal displacement

Full size image
Fig. 14
figure 14

Sample 1M-U, 16.00–16.30 m depth below seafloor: strength parameters from direct shear tests

Full size image

Discussion of the results

The results show that, despite their spatial variability, the sediments in the investigated area appear to be normally consolidated silty clays (or clayey silts) up to medium depths below seafloor. This means that they have never been under effective pressures greater than those existing at the present time.

Figure 4 shows that the soil samples lie within a narrow band above the A-line in the plasticity chart (Casagrande 1948), that is a relationship typical of sedimentary clays having low carbonate and organic material content. This is also confirmed by the organic content (Table 1) that is higher than 5 % only for one sample (1M-M; O = 5.9 %), whose point is located slightly below the A-line in Fig. 4.

For the investigated sediments, a rate of deposition (r) of about 1 mm/year, which is typical of Holocene clays (Skempton 1970), has been estimated (Mastronuzzi 2014). Although theoretically a clay can be fully consolidated with infinitely slow deposition, experiments suggest that 95 % consolidation can be assumed as a sufficiently close approximation to final equilibrium. Gibson (1958) showed that for this degree of consolidation to be reached in a clay, with a uniform rate of deposition, the ratio c v/rh must be lower than 10; where h is the thickness of the layer and c v is the consolidation coefficient of the clay. For r = 1 mm/year and c v  = 1 m2/year, the sediments are more than 95 % consolidated provided their thickness is less than 100 m, which is a condition widely satisfied by the present study.

Under this experimental conditions, Skempton (1970) has shown that, at any particular overburden pressure, the water content of a normally consolidated clay has to be directly related to the Atterberg limits, that reflect both the nature and amount of clay minerals. Therefore, if the water content, the effective overburden pressure and the Atterberg limits are known for an individual layer of normally consolidated clay, it is possible to reconstruct the entire sedimentation compression curve (SCC) for that clay. In the following, the consistency of some of these relevant geotechnical correlations for normally consolidated deposits is checked for the soils here of reference.

The recent sediments from the top of the stratigraphic succession

In Fig. 15 the sedimentation compression curves presented by Skempton (1970) for 20 normally consolidated clayey deposits are shown. Several of these data are from clayey sediments recently deposited on the seafloor and taken within the top 10 m. They often have clay fraction and Atterberg limits (e.g. Oslofjord, A-31 and Gosport samples in Fig. 15) similar to those of the sediments sampled from the Mar Piccolo. By analysing the void ratio-effective overburden pressure data relative to the samples taken from the first bosom of Mar Piccolo (Fig. 15), it can be concluded that only the data relative to the sediments taken below the top layer (about 4 m thick) lie about the SCC corresponding to their w L (see Table 1). On the contrary, the consolidation states of the sediments from the top of the stratigraphic succession lie on other SCCs that correspond to w L values that are between 100 and 115 % (i.e. higher than those determined). For a fixed range of effective overburden pressures, this can be due either to higher void ratios or to lower liquid limits of the sediments with respect to those expected according to the literature data. These results also explain the very low CI values that are measured especially in the top layer and that are not justified by higher consistency that is instead exhibited by the shallow sediments (see Fig. 6c). In particular, according to geotechnical standards, the measured CI should physically correspond to a totally liquid consistency not only in the top layer but also up to about 11 m below the seafloor.

Fig. 15
figure 15

Sedimentation compression curves for normally consolidated clayey sediments (after Skempton 1970, modified). Data in the table at the top of the figure are from Skempton (1970); data in the table at the bottom of the figure are from this study

Full size image

In Fig. 16, the oedometer compression curves shown in Fig. 7 are re-plotted in terms of liquidity index, LI (=1 − CI), versus vertical effective stress, σ’v. In the same figure, the narrow band identified by Skempton (1970) by plotting the same data of Fig. 15 is reported. It can be observed that only the data of the Mar Piccolo sediments below the top layer are included into this narrow band. Moreover, in the same figure three compression curves followed by samples of recent submarine (not polluted) hemipelagic sediments taken within 12 m below the seafloor from the Saguenay Fjord are also reported (Perret 1995; Perret et al. 1995; Martin et al. 2001; Locat et al. 2002; Levesque et al. 2007). In terms of composition, organic content and physical properties, these normally consolidated soils are similar to those found in the top layer of the first bosom of Mar Piccolo. However, despite these similarities, it can be observed that, differently from the soils under study, they lie within the band identified by Skempton for normally consolidated argillaceous deposits. These data seem to confirm that the top layer is characterised by either much higher water contents or much lower liquid limits than they would have been under normal consolidation conditions.

Fig. 16
figure 16

Liquidity index versus vertical effective pressure (data from this study, from Perret et al. 1995; Skempton 1970)

Full size image

Similar results have been found by Levesque et al. (2007) in the North Arm and Ha!Ha Bay in the Saguenay Fjord (Canada) in a succession of submarine sediments that have been heavily polluted due to the industrial activity in the area (Pelletier et al. 2003). The authors hypothesised that such experimental discrepancy could be due to bioturbation. In this respect, it has to be noticed that there is no significant bioturbation in the area under study (Tursi and Mastrototaro 2013), but this part of the stratigraphic succession is the most polluted since it contains the highest concentrations of organic pollutants and heavy metal cations (ICRAM-APAT 2007).

These results suggest that the changes in pore fluid chemical composition and viscosity (due to the high content of contaminants in situ) are leading both ion exchange and variations in the mechanical properties of the sediments in the upper part of the stratigraphic succession. This conclusion is supported by several laboratory test results on chemo-mechanical interactions in kaolinite, smectite and mixed clay samples prepared with solutions of different chemical compositions (e.g. Meegoda and Ratnaweera 1994; Chen et al. 2000; Calvello et al. 2005; Gajo and Maines 2007; Wahid et al. 2011a, b to cite the most recent ones). Moreover, Santamarina et al. (2001) underline that the potential of chemical–mechanical coupling is greatest in fine-grained soils at low confinement (i.e. the conditions of the sediments here of reference) and that such coupling may lead to unforeseen results because of the complexity of the clay fabric formation and the interaction between multiple internal scales.

From what above, it comes out the need of separate investigations aiming to deepen the effect of contaminants on the geotechnical properties of the sediments from the top layer.

The clayey soils from the lower part of the stratigraphic succession

The composition and plasticity properties of the soils below the top layer are similar to those of the clay samples taken from the locations B and P in Fig. 17 in the Montemesola Basin (Table 4), near the Mar Piccolo (De Marco et al. 1981; Cotecchia and Chandler 1997; Cafaro and Cotecchia 2001; Cotecchia et al. 2007). These samples belong to the stiff clay deposit of Argille Subappennine present in the Montemesola Basin.

Fig. 17
figure 17

Montemesola Basin and sampling sites

Full size image
Table 4 Index and physical properties of the clays sampled at the two sites of the Montemesola basin (data from Cotecchia et al. 2007)
Full size table

The comparison between the clay samples retrieved by the different sites in the Basin shows that their variability in composition and plasticity is similar to that recorded between the samples of this study. Their variability has to be linked to the different deposition conditions: clay sedimentation occurred in a quiet sea at site P, whereas more turbulent deposition conditions have characterised the edges of the Basin (site B). Moreover, also the mineralogy of the clays from the Montemesola Basin, that is mainly illitic with significant presence of both smectite and kaolinite (Garavelli and Nuovo 1974; Cotecchia and Chandler 1995, 1997), is quite similar to that of the Mar Piccolo sediments. However, it has to be noted that, differently from the Mar Piccolo sediments, the unloading due to erosion is the main cause of overconsolidation in the whole basin, so that the clays from Montemesola are overconsolidated clays and their consistency is higher than that of the sediments (CI = 1.01–1.05; Table 4).

In Fig. 18, the compression curves of samples 1M-U and 1M are compared with the data reported by Cotecchia et al. (2007) for two reconstituted samples (P19* and Bg*) of similar composition (Table 4). Following Burland (1990), the compression curves of the reconstituted samples represent the intrinsic compression lines (ICLs) of the clays. The figure shows that the behaviour of samples 1M-U and 1M is similar to that of samples P19* and Bg*. This would confirm that the soils from the lower part of the stratigraphic succession (i.e. below the top polluted layer) can be attributed to the Argille Subappennine. However, their compression behaviour is similar to that of the reconstituted clay since the compression curves of samples 1M-U and 1M almost coincide with the ICLs of samples P19* and Bg*.

Fig. 18
figure 18

Compression behaviour of the sediments below the top layer from the Mar Piccolo and of the Argille Subappennine from the Montemesola Basin (data from this study and from Cotecchia et al. 2007)

Full size image

Preliminary geotechnical model of the investigated area

Figure 19 reports the preliminary geotechnical model of the investigated area, containing average information about the sediment composition, physical properties, permeability, compression and shearing behaviour. The model reports to the sedimentary units by Cotecchia et al. (1989) and Lisco et al. (2015) where, from the top to the bottom, up to depths of engineering interest, the stratigraphy of the investigated area is characterised by the unit of the Mar Piccolo sediments (e1; Late Pleistocene - Holocene) and the unit of the so-called Argille Subappennine (ASP; Late Pliocene–Early Pleistocene). As shown in Fig. 19, from a geotechnical point of view, the sedimentary unit e1 has to be divided into two geotechnical units: e1-a and e1-b. The first one is likely representing an artificially modified geotechnical unit, whose physical and mechanical properties do not seem to be those expected according to the soil natural history. On the contrary, they can be explained only as the effect of the chemo-mechanical coupling induced in the natural sediments by pore fluids of altered viscosity and chemical composition. Moreover, it has to be underlined that the model is representing average values of the geotechnical data, not accounting for the heterogeneity that has been recognised between the borehole vertical 1M and the other two, 1I and 1L. The data reported in the model for the ASP unit are those taken from the deepest sample (1M; 17.2–18.0 m below seafloor).

Fig. 19
figure 19

Preliminary geotechnical model of the investigated area. (1) Sedimentary units reported in the geological model by Lisco et al. (2015). Key: CF average values of clay fraction, SF average values of sand fraction, G s specific gravity, e 0 void ratio, PI average values of plasticity index, CI consistency index, O organic content, Cc compression index, Cs swelling index, k load coefficient of permeability in loading (the range of the corresponding void ratio is specified in parentheses), ф’ cs friction angle at pseudo-critical state, nd not determined

Full size image

With respect to the in situ hydrogeological conditions, it should be noted that the mineralogical analyses have shown halite contents decreasing with depth (Table 2). This result confirms the vertical ascending flow of fresh ground water coming from the Mesozoic carbonate artesian aquifer below.

Conclusions and perspectives

The geotechnical characterisation of the submarine sediments located in the most polluted area within the first bosom of Mar Piccolo in Taranto has been carried out by means of site investigations and laboratory testing, which resulted to be complex given the off-shore sampling, the sediment very soft consistency and the high percentages of contaminants.

An original result of the research is the preliminary geotechnical model of the area investigated (Fig. 19). It covers the engineering volume and contains information about soil composition, physical properties, permeability, compression and shearing behaviour, which are relevant for the design of mitigation or reclaim strategies.

The key conclusions from this study relating to the site geotechnical characterisation are as follows: (1) despite the data spatial variability, the soils can be classified as normally consolidated either silty clays or clayey silts and their highest organic content is about 6 %; (2) up to 18 m below the seafloor, the vertical profile of the consistency index is characterised by values almost always lower than zero; (3) a 4-m thick top layer has been identified, where the polluted sediments have lower consistency index, higher void ratio and compressibility (C c is almost always higher than 1) than the soils from higher depths; (4) the shearing behaviour of the sediments seems to be consistent with Critical State Soil Mechanics; and (5) the comparison with other data from the literature has allowed us to demonstrate that the soils below the top layer are part of the Argille Subappennine, although their behaviour seems to be that of the reconstituted clay.

The soils here of reference belong to that class of normally consolidated materials, whose void ratio, at a given effective overburden pressure, should depend on the liquid limit, given that it reflects the nature and the amount of clay minerals present. However, this study puts in evidence that only the sediments below the top layer (i.e. unit e1-b in the geotechnical model in Fig. 19), that are less polluted, follow the standard correlations. On the contrary, the volumetric response of the most polluted layer of sediments (i.e. the geotechnical top unit e1-a in Fig. 19) due to gravitational compaction, is characterised by discrepant behaviour. These sediments are likely representing an artificially modified geotechnical unit, whose physical and mechanical properties do not seem to be those expected according to the soil natural history. On the contrary, they can be explained only as the effect of the chemo-mechanical coupling induced in the natural sediments by pore fluids of altered viscosity and chemical composition. Further studies should deepen this aspect of the research by focusing on the effect of pollutants on the soil geotechnical characterisation.