Engineering geology
molasses tanks settlement

Analysis of results of molasses tanks settlement testing


During 2009, two molasses tanks were built as part of an alcohol plant in Sladorana Ltd. in Županja, Croatia. These circular steel tanks are 24 m in diameter and 14 m high (Fig. 1). The smallest clear span between the tanks is 3.4 m. These tanks do not differ significantly from tanks designed for other fluids, but molasses is not as environmentally hazardous as oil or chemicals.

Fig. 1. Constructed molasses tanks.

Calculations based on geotechnical investigations estimated large and long-term settlements which would result in prolonged construction time and unwanted deformations due to the sensitivity of the structure itself. Three different options were considered for foundations during the design phase: shallow foundations, deep foundations using either jet grouting or stone columns, or drilled piles (Marić et al., 1996). Because drilled piles and, more recently, stone columns tend to result in a desirable small settlement, they are standard practice for tanks on weak subsoil in Croatia. At the time, however, there was no drilling rig available which could finish construction on time. The client request was to be operational in September, and not to have to invest in the ground replacement of the top soil layer. To accelerate consolidation, installing prefabricated vertical drains was considered the best option (Cascone and Biondi, 2013, Chu et al., 2004, Chu et al., 2006). Instead of preloading with soil, the design solution was to use a hydro test with equal loading for both tanks. With this drain accelerated consolidation during hydro test, bigger differential settlements between the tanks during exploration were avoided. Controlling the differential settlements was possible by filling the tanks. The solution in this project was more cost effective because it involved a less rigid structure which allowed the tanks to settle more than usual, and also allowed the tight control of critical differential settlements.

The settlement of tanks was measures as part of the monitoring program. Because the tank structure is relatively soft compared to the soil foundation, any foundation soil deformation significantly affects the behaviour of the tank structure, and differential settlements compromise functionality and even the bearing capacity of the tanks. One criteria was to achieve a differential settlement of less than 10 cm between the points along the circumference of the tank bottom.

An analysis of deformation was performed in Plaxis 2D V8 (Brinkgrave et al., 2010). The drained analysis in Plaxis enabled variations in the tank loading. The 2D model shown in Fig. 2 was used to simulate the longitudinal cross section through two tanks. This allowed the additional analysis of the situation in November 2009, when there was critical difference in loading of 32%. Generally, the resultant effective stress on the soil-slab contact induced by different loading in the tanks was used to model spring stiffness for the 3D model of reinforced slab (Radimpex, 2009) in Tower 5. Fig. 3 shows the full render of that model. The iterative design of the slab by adopting springs stiffness based on Plaxis analysis gave good results. In Plaxis, both two-dimensional finite element and elastoplastic models were used for modelling soil behaviour. The analysis is conducted using the Hardening Soil, HS, model that describes the nonlinear ratio of stress and strain, including the soil strength. The model parameters were obtained by interpreting the results of the geotechnical investigation. A secondary analysis of the consolidation processes was performed in GGU-Consolidate (Buß, 2012, Das, 2008). This software allows for a consolidation analysis with the subsequent installation of vertical drains.

Fig. 2. Differential settlements in Plaxis model (left tank 44% load, right 12% load).
Fig. 3. 3D model of reinforced concrete foundation in Tower 5.

Geotechnical investigation

Geological and geotechnical investigations on the location of the tanks in Županja were carried out in 2009 (Grget, 2009a, Grget, 2009b).

According to the Basic geological map (Brkić et al., 1985) the structure location is in an area covered by quaternary sediments (b) – the older marsh sediments: silty sand, silts and clayey silt. Fig. 4 depicts an excerpt from the Basic geological map and shows the corresponding legend and location of the construction site.

Fig. 4. Basic geological map.

Geotechnical investigations were carried out according to recommendations set in the API Standard 650 (1998) taking into account the diameter of the tanks, the types of soil, the settlement criteria and the design of the foundations.

Two geotechnical boreholes were drilled under tanks: one was 20 m deep and the other was 40 m deep. In addition to geotechnical drilling, one static penetration probe (CPTU) was performed. Fig. 5 shows some overall results of the laboratory and in-situ tests, the generalized soil profile for only three layers, the physical and mechanical soil properties and the tip resistance qc obtained from the CPTU probe. Investigation work established soil layering on location (Grget, 2009a, Grget, 2009b): they were classified in accordance with their properties and the depth at which they appear. The initial visual soil classification gave six different types of soil. Categories for Field classifications were made by referring to the categories provided by Professor Nonveiller (1990). The density state for incoherent materials was determined in accordance with EN ISO14688-2 (2004). The fill up to depth of 3 m consists of a mixture of clay, silt, sand and gravel with brick fragments. The clay (CI, CH) found in the layers up to a depth of 5 has a low to medium plasticity, a semi-solid consistency, and has a medium firmness (c = 17 kN/m2; φ = 23°; Eoed(σ=100kN/m2) = 5.5 MN/m2; k = 10−8 cm/s; PI = 9%).

Fig. 5. Geotechnical soil profile.

The clay (CH-OH/Pt) found in the layer at depths of approximately 7 m had a medium to high plasticity, and the consistency of the organic clay is medium plastic, and is of medium firmness. The peat has a granular to fibrous structure (c = 19 kN/m2; φ = 18°; qu = 62 kN/m2; Eoed(σ=100kN/m2) = 1.9–3.6 MN/m2; k = 10−8–10−9 cm/s; PI = 32%).

Clayey sand (SC, SC/CL, CL/SC) found in the layer at depths of approximately 10 m (consisting of 59% sand, 41% silt and clay) is a small grain sand, with loose to very loose density. The clay (CL, CI) found in the layer at a depth of 12.6 m is of a low to medium plasticity, with a soft to medium plastic consistency. The sand is a clayey type with mostly small grains with some medium-coarse and coarse grains. At some locations there are the soil is clearly clayey in thinner intercalations, and is medium dense to dense at depths up to 16.0 m. The soil layer at depths between 18.6 and 22.8 m contains 32% fine gravel. This layer is traced at depths between 12.3 and 30.7 m and below the layer of gravely sand to sandy gravel, from 39.2 m down to the bottom of the borehole. Th ecoarse gravely sand to fine sandy gravel (SC) found at a depth of 39.2 m is poorly graded and dense. The blow count in a standard penetration test is 16, and the permeability coefficient, k, is 10–2 cm/s.

Design conditions

Load analysis

The maximum load on the foundation soil happens in the exploitation phase when both tanks are filled with molasses with a density of 1.43 t/m3. The load on the foundation slab by the full tanks amounts to 228.3 kN/m2: this includes the load from tank filled with molasses, as well as the load from sand fill located between tank steel bottom and concrete foundation slab. The influence of the geological load reduction due to excavation and material replacement below the foundation slab is taken into account and is included in the load analysis model through the excavation phase, the foundation slab weight and the difference in the bulk density of clay and replacement gravel.

Settlement criteria

The Croatian norm Foundations rulebook (1990) recommends locating the tank steel bottom at a height not less than 40 cm above ground level after the final settlement is reached. Laying the tank bottom at this level allows for making the drainage system, so that the tank bottom is always on dry surface. At the same time, it compensates for the small tank bottom settlement. The criteria for tank settlement, both for total and differential settlements, were, according to Marr et al. (1982), determined as set in API Standard 650 (1998). The different types of settlement and their significance in relation to the construction of tanks structure is then elaborated on further. It was shown by the main designer that as long as the total estimated settlement is under 40 cm, there are no undesirable deformations of the tank connections, the tank does not sink into the ground, and settlements does not result in any undesirable deformation of the drainage system below the tank bottom plate. The allowable differential settlement of the edge and the centre is 24 cm. This is the height of the cone at the center of the tank (Fig. 6) that allows all the fluid to be pumped out. These settlements do not result in any undesirable deformation of the drainage system below the tank bottom plate and do not create issues with drainage from the centre to the edge of the tank.

Fig. 6. Characteristic cross section of the foundation structure.

Differential settlement less than 10 cm between the points along the tank bottom circumference do not result in undesirable stress in the tank shell or at the joint between the tank bottom plate and tank shell. When the differential settlement between the points of sudden changes in the loading or stiffness is less than 5.7 cm, undesirable bottom plate deformation does not take place.

Description of foundation

The foundations for the two molasses tanks (Grget, 2009c, Grget, 2009d) are shown in Fig. 6. The foundations supporting the tank structures are on foundations are in the form of a reinforced concrete slab with a thickness of 130 cm. The surface fill layer was completely removed down to the level where a solid soil made of plastic clay with better geotechnical parameters appears. The foundation soil was replaced with gravel up to the elevation of top of reinforced concrete slab. Following the replacement of the foundation soil, prefabricated vertical drains were installed in the entire area in order to improve the characteristics of the foundation soil. After the RC slab and the concrete foundation ring were constructed, the subgrade layer was laid, and the bottom plate of the steel tank was placed above this. The subgrade layer consists of gravel and sand material complete with the geomembrane, and has a total thickness of 140 cm, measured at the edge of the concrete foundation ring, while towards the centre of the foundation ring the subgrade is inclined at 2% angle, resulting with the chamber at the centre, at 24 cm.

Foundation soil improvement

The steel structure of the tanks is susceptible to deformation in the foundation soil. That is, the tanks structure can be significantly compromised in terms of functionality due to differential settlement, with the risk of a considerable change in the bearing capacity of the tank. In order to accelerate the consolidation of the foundation soil and the occurrence of the expected ground soil settlement, prefabricated vertical drains were installed. In the design process, different grids of the installed prefabricated vertical drains were analysed. Grids of 1 × 1 m to 3 × 3 m and 2 × 2 m were chosen because it allowed acceptable acceleration of consolidation and the costs were acceptable. The acceleration of consolidation due to installation of vertical drains is calculated in the main project (Grget, 2009c) using the familiar theoretical solution for a cylindrical column of loaded soil with a central drain (Barron, 1948). An analysis of consolidation was carried out using the GGU-Consolidate computer program (Buß, 2012, Das, 2008) on the soil model with and without the vertical drains located on the 2×2 m grid. An analysis of the model using vertical drains showed a clear reduction in the consolidation time. The time period required to reach 90% of the final settlement was reduced from 1390 to 330 days. Table 1 shows a comparison of the foundation soil consolidation time with and without vertical drains for a constant full tank load. The soil characteristic are shown in Chapter 3, and the vertical drain is described with 5 cm radius and 2 × 2 m grid. Measured settlements, as well as the rate of consolidation settlement, depend on the soil permeability and the regime of filling and emptying the tanks.

Table 1. Estimate of settlement and the consolidation time with and without the use of vertical drains.

Acceleration of consolidation was performed by installing prefabricated vertical drains “Membradrain” MD88 (Rijn, 2001). This type prefabricated drain strip is highly suitable for water transportation. The flexible core is made of high quality polypropylene. Both sides of the core have grooves for unhindered water flow. Because the core is wrapped in strong and durable geotextile filter, the filtration properties are excellent, and the free passage of pore water into the drain is enabled. At the same time, this filter prevents the passage of fine soil particles from the surrounding soil and prevents clogging. The factors affecting the functionality of the drains were established by Chai and Miura (1999).

This type of vertical drain is well suited to highly compressible soils, where large deformations can occur. The advantages of the system MebraDrain are the following: minimal possibility of disturbance in the surrounding soil, guaranteed drainage, the possibility of modifications of the core and filter with respect to the applicable soil conditions, fast installation (4000–30,000 m’/day, an average of 8000 m’/day), adjustable spacing of the drains, no requirement for water for installation, the possibility of installing drains at depths of up to 65 m and the easy control of the installation using monitoring equipment. In order to avoid the damage of the drain, a rectangular steel tube is used during installation. This tube has small cross section dimensions for easier penetration into the soil and in order to cause as little disturbance of the soil as possible in the proximity of the drain. The drain is temporarily anchored into the anchoring plate at the bottom of the tube. The tube with the drain is penetrated into the soil using a force of 50–200 kN. When the desired depth is reached, the tube is retracted, and the drain with the anchoring plate remains in the soil. After the tube is completely pulled out from the soil, the drain is cut off and fastened to the anchoring plate of the following drain.

During the exploitation period, the achieved settlements and their trend of subsidence have been established based on the monitoring of installed monitoring equipment, using horizontal inclinometers and geodetic points. It was concluded that 90% of the settlement is achieved through two cycles of tanks filling and emptying, which represents the time equivalent to the design estimated period of one year at full load. The speed of tank settlement and its dependence on the loading regime is shown on Fig. 11. The regime of filling and emptying the tank as defined in the design was also taken into account during the exploitation period. The main design requirement was that 30% difference in loads between two tanks should not be performed before 90% of consolidation is achieved, or before a I year time period has elapsed. In addition, during the exploitation phase, the difference in height between the contents in the two tanks should not be greater than 50%.

Reinforced concrete foundation slab

A joint reinforced concrete foundation slab was built below the tanks. After a surface soil layer with a thickness of 3.1 m was removed, geotextile was installed on the foundation soil. On such arranged base soil, a gravel layer with a thickness of 1.1 m was installed. A reinforced concrete slab was poured on top of the gravel layer. The height of the foundation slab is 130 cm. Vertical drains were installed starting from the top surface of the gravel layer following the grid 2 × 2 m down to the depth of 14 m, i.e. to the layer of sand as detected by geotechnical investigations at a depth of 14 m.

Reinforced concrete foundation ring

The tank shell is supported by the stiff reinforced concrete foundation ring (Fig. 6). The foundation ring has a rectangular cross-section, b/h = 100/140 cm, and is made as a composite structure with the foundation slab. Outside ring diameter is D = 25 m. Steel bottom plate and the steel tank shell are supported by the foundation ring. A layer of stiff asphalt, with a thickness, t, of 20 mm, is laid on top of the RC foundation ring. The asphalt layer provides insulation and serves as an expansion joint. The steel tank is set directly on the layer of asphalt.

Setting of tank bottom directly onto the arranged subgrade

The subgrade is a multi-layered filled structure that represents the base for supporting the steel plates on the bottom of the tanks, and is installed on top of the foundation slab following the completion of the RC foundation ring.

Horizontal inclinometer tubes were installed within this subgrade layer and lean concrete was poured over them.

A layer of gravel with a grain size of 0–64 mm was placed directly onto the RC foundation slab, which has a thickness of 60 cm and the camber at the tank centre on top surface is 24 cm with respect to the tank edge. At the joint location with the foundation ring, the geomembrane was raised along the foundation wall and gaskets at the height of 15 cm below the top elevation of the foundation ring. Natural sand was installed on top of the geomembrane in layers with a total thickness of 80 cm. The final surface of this subgrade was performed with the same camber as the gravel layer underneath.

This layer forms the final layer of the subgrade before installing the steel bottom plates. Special attention was paid to maintaining the quality of the surface in terms of flatness, the required camber and compactness.

Geotechnical analyses

Stress and strain analyses were conducted using the finite element analysis software Plaxis 2D V8. Because the rigid rectangular concrete slab provided uniform load distribution on the ground, the analysis was done using a 2D plane model, even though the tanks are cylindrical. The soil is modelled using the Hardening Soil (HS) model. This elastic-plastic model with isotropic hardening was formulated by Shanz et al. (1999).

The main feature of the HS model is the correlation between the soil stiffness and the stress state (1). Soil stiffness has a non-linear correlation with stress, characterized with Eoedref, which represents a referent stiffness corresponding to referent stress pref, and the average (median) stress exponent m that regulates the correlation between stiffness and stress. In the HS model, plastic deformations occur due to shear and compression stresses. Another feature of this model is that the development of plastic deformation shifts the yielding plane. When subjected to primary shear stresses, the soil shows a decrease in strength, and at the same time, has a tendency to develop irreversible plastic deformation. In the case of primary normal stresses, the soil shows an increase in strength and becomes compacted. In the HS model, soil failure occurs in accordance with the Mohr-Coulomb law. This is an advanced soil model that describes the behaviour of stiff and soft soils (Brinkgrave et al., 2010). Details of the HS model and its behaviour can be found in the paper published by Szavits-Nossan and Sokolić (2011).

In the geotechnical analyses for molasses tank foundations, the soil parameters are determined according to the results of geotechnical investigations, from consulting the literature and existing experience reported for the Sava river embankments. The first geotechnical investigation included only one borehole, which was 20 m deep. However, after understanding the complexity of the structure and considering the issues which emerged, it was clear that the conducted investigation was insufficient. During the site investigation and when visiting the site, it was established that several tanks had been constructed at this location, and each had issues with the foundations, subsequent settlements and cracks in the foundation structures. Based on these findings and considering the importance of the structure, additional geotechnical investigation works were proposed and carried out. Another exploration borehole 40.0 m deep was made, in addition to another CPTU to a depth of 20 m. The importance of field investigations was confirmed in this case. In many cases, investigations are designed without sufficient regard to the complexity of the actual structure and geotechnical issues, and often even before the concept of structure has been made. In accordance to Eurocode 7 (EN 1997-1, 2004), it is necessary to conduct a preliminary investigation to evaluate location suitability and, most importantly, to prepare a programme of geotechnical investigations for different design phases. This, unfortunately, is not often the case in Croatian practice and investigations are often deemed complete even though the preliminary investigations are carried prior to the preliminary design phase.

CPT probe testing was done in the first fill layer of 3 m in thickness. This fill layer consists of waste material and carbocalk and is deemed unsuitable for use in foundations. The parameters of the fill were estimated conservatively, and because the replacement of this fill with gravel was done in the first stage of the analysis, it does not affect the amount of settlement. The strength parameters for clay were obtained from the direct shear test. The modulus of compressibility of clay was obtained by comparing the three oedometer tests with the values obtained from the CPT (Lunne et al., 1997). Undisturbed samples for oedometer tests were taken from the top part of the clay layer, even though it was evident from the CPT results that part of the sandy clay has a higher compressibility modulus. For that reason, a characteristic value of the oedometer modulus was selected as a value higher than the one obtained in the laboratory. The modulus of compressibility, Eoed, in sand intercalation was determined from the SPT blow count (Callanan and Kullhawy, 1985) according to the recommended correlations in the literature. Experience shows that oedometer modulus of compressibility (Eoed) obtained in this way gives a conservative value, or a value that has been proven too low, and needs to be corrected to a higher design value. The correlation between the SPT blow count and geotechnical material parameters for coherent soils have not been used because undisturbed soil samples were taken from the clay layer which were used for more elaborate laboratory tests. Table 2 shows oedometer modules obtained through field and laboratory investigations.

Table 2. Oedometer modulus (obtained from investigation work).

The field investigation showed that the clay layer up to 12 m is generally intersected with layers of sand, silty sand and sandy silt. Oedometer tests were conducted on clay samples and the low compressibility modules found are in good agreement with the results of the CPT in clay layers. However, field investigation showed an increase in the modulus of compressibility in sand intercalations. These thin layers of sand were not modelled in the analysis, but a higher compressibility modulus of the clay layer was chosen for this very reason.

The coefficient of permeability of clay was selected from oedometer tests. The mean value of modulus of compressibility of sand obtained from CPT corresponds to the value obtained by correlating SPT (Callanan and Kullhawy, 1985). The coefficient of permeability is determined based on the material grading curve. For replacement gravel material, the parameters for well compacted and well graded gravel were taken (Bowles, 1997). Table 3 shows the design values of the soil parameter for the selected geotechnical soil model.

Table 3. Design soil parameters.

The monitoring and analysis evaluation

The aim of the monitoring process was to measure the settlement of the tanks during their uneven filling and emptying. Their steel structure is susceptible to settlement and also to the differential settlement of the foundation soil.

Monitoring involved the use of a system of horizontal inclinometers, placed on the reinforced concrete foundation slab, and permanent geodetic points aligned along the outer edge of the RC foundation ring and on the manholes of the horizontal inclinometer pipes. Six permanent geodetic points were located on the outside perimeter of each of the concrete foundation rings, while four points were located on the manholes of the horizontal inclinometer pipes. Two horizontal inclinometer pipes were installed below each tank. All pipes run directly below the tank centre, as shown in Fig. 7.

Fig. 7. Monitoring equipment.

Access to and the protection of the inclination pipes openings is secured using manholes that provide manipulative space for settlement measurement using horizontal inclinometers. Inclination manholes were located approximately 50 cm from the outer edge of the reinforced concrete foundation slab. Given that all pipes run directly below the tank centre, two pipes were installed at different depths, which ensured easy overlapping. The pipes were placed on the RC foundation slab together with the plates to prevent rotation (stabilization plates), placed 6 m apart, and lean concrete was poured around the pipes, followed by backfilling with gravel material up to the final elevation level, and compaction to the desired modulus of compressibility. The verticality of the grooves on the pipes serve as the guide for horizontal inclinometers, and was constantly monitored during the installation. The perviosity of the joints was also checked using the blind probe. Control measuring was done after the installation of horizontal inclinometer pipes, as shown in Fig. 8.

Fig. 8. Installation of monitoring equipment and settlement testing in the inclination manhole.

A series of 12 measurements were carried out from August 2009 until January 2012. Of these, the first measurement, performed on August 24th, 2009 was taken as a reference value. During this period, the tanks were filled and emptied several times. Even though the levels of molasses in the tanks were different, the difference between the molasses levels in the tanks was always less than that allowed by the main design provisions, which list this value at half the height of the molasses level. Table 4 shows the measured settlement in the time period stated above and the molasses loads in the tanks during this period. Molasses loading is given as a percentage of the full tanks loading on the reinforced concrete slab in the amount of 228.3 kN/m2.

Table 4. Loads and measured settlement values.

The extreme loads and tank settlement values of tanks R-2 and R-3 are shown in Fig. 9. The legend shows the tank identification, the loading percentage compared to the tank maximum, and the date when measurements were taken. Measurement values are taken for the three representative phases: at the initial filling, the first emptying and the second filling. These indicate the settlement values along the longitudinal section through both tanks, in addition to the mutual influence of settlement of the tanks due to non-uniform filling and emptying of each tank. The maximum settlement value was observed at the maximum reached loading and the tank tilting was recorded due to the first filling of only one tank. The difference between the tanks load reached 32% and was more than the allowable 30% in the design, and tank R-3 started leaning towards tank R-2 in November 2009. Over the period of a few days, the difference decreased to under 30% without any permanent consequences. The maximum differential settlement of tank centres tended to occur at the time of the sixth measurement (Table 4), when the largest difference in tank loading occurs.

Fig. 9. Settlement along the longitudinal section through both tanks at extreme loads.

Fig. 10. shows the history of filling and emptying the tanks along with the diagram of the estimated and measured settlement values in tanks centre points for two cycles of filling and emptying. The progress of settlement during the time required for filling and emptying the tanks can be observed. Settlement values obtained by calculations have been proven greater than the actual values obtained through measuring. The most significant difference between the settlement values obtained by calculations and those measured is noticeable during the first emptying and second filling. The greater plastic deformation when the tank was filled for the first time can likely be attributed to the lower estimated value for the unload-reload oedometer.

Fig. 10. The comparison of estimated settlement values obtained by calculations and those measured in tank centre points with the filling and emptying load curves.
Fig. 11. Rate of settlement diagram including the filling and emptying load curve.

Fig. 11. shows the diagrams of settlement velocity for both tanks along with the filling and emptying curve during the monitoring period. Velocities marked Vtot represent secant velocities for current settlement values during the monitoring period. As opposed to those, Vrel represent the tangential velocities for interval tank settlement during the monitoring period. The Vtot chart shows the largest velocity growth rate in the period between days 91 and 101. During this period, the tanks were filled up to 55% and 65% of the maximum tank load. As the filling of the first cycle progressed to 87% of the maximum load, the values of Vtot remained approximately constant, and thereafter gradually decreased until the last measurement.

The oscillations of tangential velocity (Vrel) are greater and can register negative values due to the elastic uplift deformations of the ground soil during tank unloading.

Because the tank load did not surpass the value of 200 kN/m2 during the first three loading cycles, this value was taken as a reference value for the analysis of measured consolidation settlements. The analysis of the settlement values during the first three cycles showed that over 80% of settlement occurred during the first loading cycle. This settlement during the first filling cycle consisted of primary settlement and partially of secondary consolidation settlement. During the second loading cycle, only the part corresponding to consolidation settlement was analysed: it was found to amount to approximately 14% of total settlement. In the third loading cycle, consolidation settlement totalled approximately 5% of total settlement.

Fig. 12. shows the difference between the estimated and measured settlement values. The estimated values were found to be higher than values obtained by measurements. The observed difference appeared because the clay modulus of compressibility was underestimated. Even though the laboratory tests on clay samples gave low values for the clay modulus of compressibility, the influence of sand intercalations in the layer of clay on the increasing modulus of compressibility was considerably greater than expected. The chosen value for the oedometer modulus was 6.0 MN/m2, which represents a value higher than the one obtained from oedometer testing, and is equal to the one obtained from the CPT. While the CPT showed an increase in the local oedometer modulus to 15.0 MN/m2 in sand intercalations, this was neglected in the numerical model. Because thin layers with significantly different characteristics often result in errors in the numerical modelling software, they are usually not included in the models. An increase in the modulus of compressibility of clay due to the presence of sand intercalations could be determined by performing the back-analysis. Considering the design analyses and the insufficient data from inclinometers for the settlement through each individual layer, only a modulus of compressibility for the complete set of soil material can be obtained, i.e. clay with sand intercalations.

Fig. 12. Correlation between settlement and load at tank centre points.

The analysis shows a larger settlement is obtained in the second loading cycle for approximately the same loading values, which indicates an occurrence of consolidation of foundation soil. Although the two tanks had different loading and unloading regimes, their settlements are quite uniform at maximum reached loading and unloading of both tanks. The inclination of the curve during loading and unloading represents the soil modulus of compressibility and unload-reload modulus of compressibility of the foundation soil (Fig. 12).


Prefabricated vertical drains were shown to allow for a marked improvement in the stability of circular tanks 24 m in diameter and 14 m high supported on shallow joint foundation slab on foundation soil. Because the steel structure of tanks was shown to be susceptible to deformations of the foundation soil, the settlement and differential settlement of the foundation soil significantly affects the behaviour of the tanks structure and compromises the functionality and even the tanks bearing capacity. Settlement below the tank bottom plates, as well as the settlement along the concrete ring foundation was measured. These settlement values can be used for back-analysis of clay modulus of compressibility.

The results of the measured settlements using installed horizontal inclinometers and geodetic points, and the settlement velocities calculated from those values confirmed the design estimate: that is, 90% of total settlement was achieved within the one year period. Considering the specific molasses manufacturing process, this load duration of one year was carried out through two filling cycles. The measurements have confirmed the trend in settlements subsidence in the predicted time.

The analysis of measurements results also confirmed the design assumptions related to the regime of differential filling and emptying of the tanks. The design limited unloading, or emptying, of one of the tanks to a maximum value of 30% compared to the second tank, should not take place before 90% of consolidation of foundation soil has been achieved, or before a one year period has elapsed. During the exploitation phase, the difference between the molasses level between the two tanks must be less than 50%. Settlement measurements confirmed the stability of the tanks and allowed differential settlement for differential filling and emptying within the given limits.

Evaluation shows the time course of settlement was in the first 2.4 years of exploitation: the load curve over the same period also supports this. Another point of interest was the settlement curve as a function of the filling and emptying regime. The settlement values obtained by numerical analyses were found to be greater than the actual measured settlements. The largest difference in the settlement values was measured during the first emptying and second filling cycle. The comparison of the estimated and measured settlement values showed that the actual oedometer modulus and unload-reload oedometer modulus are greater than the values estimated in the design, while the design permeability coefficients correspond to actual values.

Source: Analysis of results of molasses tanks settlement testing

Authors: G.Grget, K.Ravnjak, A.Szavits-Nossan

Leave a Reply

%d bloggers like this: