Paulo Cella, BVP Geotecnia e Hidrotecnia, Brazil
Carlos Santos, BVP Geotecnia e Hidrotecnia, Brazil
Felipe Neiva, BVP Geotecnia e Hidrotecnia, Brazil
Laura Rodríguez, Ex-BVP, Brazil
Abstract
This work aims to understand the dominant displacement mechanism that resulted in a set of longitudinal cracks, a few tens of meters long, that affected the Itabiruçu Dam, Conceição Mine, Itabira Complex, Minas Gerais, Brazil The dam is currently 71 meters high and has a crest length of around 920 meters. The appearance of cracks caused the fourth dam raise to be halted for over 2.5 years, starting in mid-2019. A significant number of field and laboratory investigations, as well as a comprehensive monitoring program implemented after construction was halted, produced a rich data set, which in this article is subjected to a careful analysis of the actual displacements caused by differential settlements of the foundation beneath the downstream shell of the dam. In addition to looking into the numerical model results (on which this paper is not specifically focused) as a basis for the monitoring evaluation, a comparison was also made with a finite element classical study (Clough and Woodward, 1967), in which the construction of an idealized dam in several stages is appraised by assigning material deformability within a range of contrasting embankment/foundation stiffness ratios. In the case study presented herein, the vertical-to-horizontal displacement ratio was found to be more likely to correspond to the pattern of components associated with the gravitational dam settlement as an isolated mechanism, rather than a potential underlying foundation shear band development.
Introduction
Loading effects in compacted soil embankments can lead to anomalies that affect the physical stability of a facility during operation and decommissioning. According to Silvera (2006), construction-related issues can develop depending on the rate at which the embankment is built on a saturated foundation or differential stiffnesses occurrence contrasts between the embankment and the foundation.
Clough and Woodward (1967) studied the pattern of displacements induced along the embankmentfoundation interface of an earth dam in idealized gravity settlement models, based on different stiffness ratios between these materials, considering incremental and single-stage loading. The results showed that the distribution of horizontal and shear stresses at the embankment-foundation interface, as well as vertical and horizontal displacements, are significantly affected by the stiffness contrast between embankment and foundation.
Poulos and Davis (1972) also arrived at similar conclusions, by normalizing the displacements based on the displacement influence factor concept proposed by Poulos et. al (1972), which differs according to construction stages.
This paper presents the case of the Itabiruçu Dam (Figure 1). The dam has had a history of surface cracks. The study attempted to investigate whether the cracks were due to differential gravitational settlement, or horizontal displacement along a potential deep shear band, at the embankment-foundation interface.
Key facts about the most recent raise of the dam
Itabiruçu Dam was built at Conceição Mine, Itabira Complex, Minas Gerais, Brazil, to store iron tailings generated and provide water for ore processing. The starter dyke was built circa 1980/1981 to a crest elevation of 812.0 m. During operations, the dam has undergone four downstream raises: in 2004, to elevation 817.5 m; in 2011, to elevation 833.0 m; in 2016, to elevation 836.0 m; and in 2023, to elevation 841.0 m.
The last-mentioned raise was started in 2018, and was planned to reach an elevation of 850.0 m. However, in July 2019, after the crest reached an elevation of 819 m, construction was preventively halted following the appearance of centimetric lengthwise cracks, close to the contact between the previous raise and the new one (see Figure 2), as well as cross cracks at the left abutment. Construction resumed in May 2022 after a comprehensive field investigation program was carried out, monitoring instruments installed, and numerical simulations performed to understand the displacement mechanism. Construction was completed in July 2023, to a modified design crest elevation of 841 m, and guidelines were set out to control the rate of rise, and to establish a threshold settlement for the structure.
A comprehensive field investigation program was carried out, and monitoring instruments installed. This allowed for enhanced knowledge on both foundation conditions and behavior, and evolution and gradual stabilization of facility movements. Among the studies was the numerical simulation of the dam with undrained foundation, in which constitutive model selection, mesh sizes, and medium discretization refinement, as well as important distinctions of geotechnical foundation units sectorization, were thoroughly discussed and enhanced over more than two years of work, with various professionals having been engaged, including designers, EoR, client, Public Prosecutor’s Office auditors. Stiffness, compressibility, and hydraulic conductivity parameters of the embankments were adjusted in several stages, in order to replicate the decreasing settlement rates and the piezometric levels established. The numerical model was accurately calibrated, hence data value and reliability for interpretation of dominant deformational mechanism and facility behavior trend in the medium term, as long as no significant changes occur in the patterns identified to date.
Geological-geotechnical foundation condition
Geological-geotechnical data on Itabiruçu dam is based on several site investigation programs performed during dam design and raise phases (Figure 3).
The dam footprint comprises weathered rocks and soils from Nova Lima Supergroup schists and Guanhães Complex migmatite gneisses. Schists occur predominantly at the dam toe and on the left abutment. They can include mafic schist, steatite, and serpentinite (talc-schist), usually as saprolite, saprolite and residual soil, generally clayey. Gneiss, on the other hand, outcrops mainly as saprolite soil along the right abutment and central portion of the dam. Within this area, large saprolite thicknesses were encountered in the boreholes, underlying clayey-sandy to sandy-clayey saprolite soil, with felsic bands. In total, there are about 7500 meters of boreholes, five CPTs, and over fifty in situ? strength tests.
Geological-geotechnical units
The geotechnical units under the Itabiruçu Dam footprint were carefully characterized following the assessment of the site investigation data. Two gneiss saprolite units were identified: a saprolite soil / young residual soil, referred to as “SAC” underlain by a well-structured saprolite soil, referred to as “SSG”. SAC is a more compressible and less permeable unit than SSG. It has lower strength parameters and develops excess pore pressure during shear up to the maximum deviatoric stress. SSG has been shown to dissipate partly or totally the excess pore pressures in undrained triaxial tests.
These two subunits were differentiated by applying double corrections to the SPT blow count (NSPT) obtained from drilling programs, especially from 2013 onwards. The corrections were required to account for the significant increase in confinement (due to placement of 40 to 45 m of embankment) compared to in situ testing carried out in 1977, 1983, and 2004, which were drilled on native ground or minor embankments, and also to account for energy losses due to much longer string lengths. Correction due to confining stress used the methodology proposed by Boulanger (2003), specific to low-plasticity silty soils (correction index lower than sand index), in order to obtain NSPT values in conditions comparable to those of the boreholes drilled in earlier investigation phases. On top of these two corrections, an additional factor of 0,7 concerning the energy efficiency. For SAC, NSPT < 10. For SSG, 10 ≤ NSPT ≤ 35.
The prominent heterogeneity of saprolite horizons, with varying textures (from clayey to sandy soils) and colors, favors the adoption of a correction parameter for NSPT, at the expense of reduced assertiveness in tactile-visual differentiation from the drill cores.
The section in Figure 4 identifies the two units, SAC (in brown) overlying the less compressible soils SSG (in green). The thickness of the SAC soil ranges from 5 to 25 m, whereas the SSG ranges from 7 to 30 m. A slightly altered gneiss saprolite underlies the saprolite layer. The thickness of the saprolite ranges from 5 to 15 m. The other foundation materials are also presented.
Settlement basin
The settlements large-scale pattern in the downstream shell, as measured from the prisms/survey monuments, was shaped like a bow (or basin). The center of the basin coincided with the lengthwise crack observed at the contact between the existing embankment and the new raise. Figure 5 plots the vertical displacements over the monitoring period (from 2019 to 2023). Ranges of accumulated vertical displacement (15 to 100 mm; 100 to 300 mm, and greater than 300 mm) are shown in different colors, for reference. A maximum settlement of 420 mm was reached during the stabilization phase, before construction was resumed, and over 500 mm was reached during the final raise to the crest at elevation 841m, starting from the downstream plateau, halted in 2019 at elevation 819 m. As a result, a 38 m wide berm was shaped along the downstream slope (Figure 4).
The data shows a gradual slowdown in the settlement rate during the first few years of monitoring. The decreasing settlement rate was used to calibrate the numerical model. Note that monitoring only started about two months after construction was stopped, in 2019.
Figure 6 plots contours of accumulated settlement measured from August to October 2019. The arrows plot the direction and magnitude of the horizontal displacement component at each instrument location. The contours and arrow pattern indicate the existence of a settlement basin. Figure 7 shows the settlement basin six months after construction was completed, in July 2023. The instruments MS-19 and MS-20 located at the center of the basin show vertical displacements of 548 mm and 523 mm, respectively, with full stabilization yet to be achieved.
Theoretical reference model
Clough and Woodward (1967) developed a theorical model to evaluate the effects of embankment construction rate. They carried out finite element simulations of an idealized dam under construction in stages (or in a single-stage), assuming foundation to embankment stiffness ratios of 1:1, 1:2, and 1:5 (i.e., the embankment is 1×, 2× or 5× stiffer than a homogeneous foundation). For displacement pattern configuration, the theoretical study assumed a single valued embankment deformability. The dam load was applied as a gravity load in ten steps, up to the final height of the structure.
Figure 8 plots the resulting vertical and horizontal displacements at the base of the dam for the three stiffness ratio cases.
To apply the Clough and Woodward (1967) model to Itabiruçu Dam, the vertical displacements were divided by the horizontal displacements, as shown in Figure 9.
The next section compares the model and actual vertical and horizontal displacements obtained from instruments installed near the embankment foundation interface.
Comparison between field measurement and theoretical model
Inclinometer and settlement gauge data was used to compare field measurements with the model.
Monitoring evaluation mainly focused on checking the displacement profiles over instrument operation history, aiming at identifying any signs of raise-induced shear zone formation and mass movement. The instrumentation used to observe horizontal and vertical displacement behavior at the
embankment/foundation contact is shown in Figure 10 for the typical section.
Also, four inclinometers anchored in competent rock are shown, from dam crest to toe, as well as three settlement gauges (RA) near the embankment/foundation interface more directly subjected to final works loading.
Figure 11 shows the horizontal displacement (Figure 11 (A)) and vertical displacement (Figure 11 (B)) profile indicated by the instrumentation, whose maximum and minimum distribution and distance from the crest is similar to Figure 9 of the theoretical reference modeling.
It is worth noting that the profiles on Figure 11 correspond to different times during the final dam raise works, making it possible to evaluate the effects of the increased load with the actual embankment and foundation stiffness contrast, conditioned in the actual case by the high SAC unit compressibility against the higher embankment stiffness. Thus, embankment/foundation stiffness relationships addressed in the idealized model are inverted.
In order to check whether the displacement pattern identified in the selected section replicates the idealized model, but with stiffness ratio inverted, the analysis included monitoring data on vertical and horizontal displacements from instruments installed in a farther section, also at depths corresponding to the embankment/foundation interface. Figure 12 shows the plot plan for additional instruments (an inclinometer and a pressure gauge) included in the comparison basis.
We believe that the considerable distance of the additional instruments from the most instrumented section is strongly significant for a typical downstream shell deformational behavior to be established, in a considerably more complex foundation condition from a geological-geotechnical standpoint.
Figure 13 shows vertical and horizontal displacement ratios recorded by the instruments considered. The relationship between vertical and horizontal displacements in the actual case (dashed red curve) quite reasonably follows idealized model ratio pattern, but it is on the opposite side to the equivalent stiffness curve (blue curve). This is clearly compatible with the actual case of the foundation being more flexible than the embankment, according to the calibrated numerical model.
The shape of the normalized curve between vertical/horizontal displacements corresponding to the actual measurements shows a similar pattern to the normalized reference model curves, which suggests the actual case behavior, in terms of settlement and horizontal displacement, is in line with deformations associated with the gravitational facility settlement, in which the actual foundation is around 2 to 2.5 times more flexible than the embankment stiffness.
Table 1 shows the E50 modules of the Hardening Soil model, Undrained module ‘A’ applied to the foundation, on the off-the-shelf program PLAXIS. These values led to the best result of comparison between the numerical Itabiruçu Dam simulation and the monitoring data, especially regarding construction pore pressure dissipation rate, as shown in Figure 14, with numerical model and settlement results over time for survey monuments 19 and 20.
According to this analysis, the actual facility behavior, as per synthetic vertical and horizontal displacement ratios along the embankment/foundation interface downstream the dam axis (Table 1), fits the facility/foundation deformation mechanism, in which foundation is 2-2.5 times more flexible than the embankment, due to self-weight settlement.
Highlight of the discussion
Despite the presence of a highly compressible layer in the foundation as compared to the embankment stiffness, the ratio of horizontal to vertical displacements has shown full compatibility concerning the shape of the function captured in the Theoretical Reference Model in which the foundation was repeatedly stiffer than landfill in all cases modeled by the authors. This similarity suggests the dam would have experienced just gravitational settlement as the single deformational mechanism.
Conclusion
Typical horizontal and vertical displacement behavior evolution, as the downstream raise of the analyzed dam developed, has shown a similar pattern to that of the theoretical reference model of Clough and Woodward (1967), with various relative embankment/foundation stiffness ratios being considered. In the actual case, as in the model, the horizontal displacements measured near the dam axis tend to be minimal or cancel out where vertical displacements reach their peak value, whatever the relative embankment/foundation stiffness may be. At about one-third of the distance from dam toe to crest, the horizontal displacements are more significant and decrease both upstream and downstream. Itabiruçu Dam measurements replicate with excellent adherence the idealized model pattern of deformations induced by gravitational settlement.
The curve shape of normalized displacements along the embankment/foundation interface underneath the downstream shell in the actual case is compatible with the normalized curve pattern of the theoretical self-weight settlement model, and is even coherently located on the opposite side to the theoretical curves, since the typical Itabiruçu Dam foundation stiffness turned out to be around 2 to 2.5 times lower than that of the embankment.
The main value of this finding is that the origin of the dam cracks is confirmed by differential settlement given its occurrence as pockets of most compressible soil unit (SAC), with the lowest SPT blow counts of the program, usually less than 10. To date, there have been no unequivocal findings of horizontal displacements in inclinometer logging, regardless of the depth below the facility/foundation interface, nor occurrence of cracks in drainage channel concrete downstream of the dam. Horizontal displacements recorded by the instrumentation have been proven to correspond to settlement components of SAC deformation, mainly. Over almost four years of continuous monitoring, low-angle shear bands in the facility foundation embankment have been ruled out. The most important aspect in analyses of this nature is the actual facility/foundation behavior, over the long period of instrumentation monitoring, which points to gradual and consistent settlement stabilization over time, mainly due to the dissipation of excess construction pore pressures induced in the SAC unit.
As for horizontal displacements, no behavior was recorded during the period of continuous measurements for almost four years, which could raise suspicions of a second facility movement mechanism, possibly associated with strain softening of the foundation under large deformations, as a result of potential mobilizations of foundation strength in the post-peak regime. The observations of the entire set of monitoring instruments on the dam scale show convergence with the mechanical stress-strain response in dozens of undrained and drained triaxial tests, carried out both in the SAC and in the SSG unit, which showed maximum losses of peak strength of only 10% in some tests, despite the development of positive pore pressures in the SAC up to close to the maximum deviator strength.
Finally, settlement recurrence along the reference section as raise works resumed was at lower magnitudes than those recorded immediately after work was halted, due to a maximum height loading of 17 meters in the 2022 stage, compared to a 40-meter loading between 2018 and 2019. On the other hand, the decreasing settlement rates from the recent construction completion are similar to those of the previous long period, in our opinion caused by gravitational settlement.
References
Boulanger, R.W. 2003. High overburden stress effects in liquefaction analyses. Journal of Geotechnical and Geoenvironmental Engineering 129(12):1071–1082.
BVP Geotecnia e Hidrotecnia. 2019a. Relatório Técnico RL-1850CC-G-31335 – Consolidação preliminar dos dados – Anexo. Belo Horizonte.
BVP Geotecnia e Hidrotecnia. 2019b. Relatório Técnico RL-1850CC-G-31337 – Consolidação Final Dos Dados – Fase 2 Modelo Adensamento – Anexo. Belo Horizonte.
BVP Geotecnia e Hidrotecnia. 2023a. Relatório técnico RL-1850CC-X-33700 – Parecer e análise crítica – Estudo tensão x deformação alteamento até 2022. Belo Horizonte.
BVP Geotecnia e Hidrotecnia. 2023b. Relatório Técnico RL-1850CC-X-33059 – Análise de tensão e deformação do alteamento utilizando Plaxis 2D. Belo Horizonte,.
BVP Geotecnia e Hidrotecnia. 2023c. Relatório Técnico RL-1850CC-X-33585 – Revisão Periódica de Segurança de Barragens. Belo Horizonte.
Clough, R. and Woodward, R. 1967. Analysis of embankment stresses and deformations. Journal of Soil Mechanics and Foundations Division 4: 529–549.
Poulos, H.G. and Davis, E.H. 1972. Elastic Solution for Soil and Rock Mechanics. New York: John Wiley & Sons.
Poulos, H.G., Booker, J.R. and Ring, G.J. 1972. Simplified calculation of embankment deformations. Soils and Foundations 12(4): 1–17.
Silveira, J.F.A. 2006. Instrumentation and safety of earth and rockfill dams. São Paulo: Oficina de Textos.