Landslide Hazard Evaluation of a Large Waste Landfill in Bogotá City

Abstract

The Doña Juana waste landfill receives solid urban waste from about nine million people living in Bogotá and several nearby municipalities. This article describes the characteristics of this landfill which has suffered some important landslides with serious environmental effects, shows the main geotechnical studies carried out for the characterization of the materials, analyzes the operation and monitoring processes, especially the pore pressures generated by leachate and the results of the landslide hazard evaluation in terms of the probability of failure for static and pseudo-static conditions.

1 Introduction

The waste amount produced by cities is directly proportional to the size of its population, and it is also related to economic development and social consumption patterns. The increasing waste production represents a significant problem for many cities in the world because it is very difficult to find suitable storage places since people are not willing to have sanitary landfills near their homes. For that reason, many of the existing waste landfills, also known as municipal solid waste (MSW), tend to prolong their useful life several years by increasing areas and heights of the fills. MSW are based on waste disposal in cells that are mechanically compacted and covered daily and whose leachate and gases are captured or recirculated by different procedures such as gravity drains, chimneys and forced pumping.

Doña Juana Landfill is the largest landfill in Colombia, occupying nearly 500 ha. It receives around 6300 tons (7250 m³) of waste per day from Bogotá city and many nearby municipalities, serving around nine million people.

Doña Juana Landfill suffered a great landslide on September 27, 1997, in which more than 1,200,000 m³ of garbage was mobilized, generating very serious environmental problems such as contamination of the Tunjuelo River, bad odours perceived from more than 10 km away, the proliferation of flies and, in general, damage to the surrounding communities. Since that moment, many studies have been carried out to understand the complex stability behaviour of this waste landfill. In 2015, a second landslide of 600,000 m³ blocked the pipes of the evacuation leachate system, and in 2020, a third landslide of about 80,000 m³ occurred with no significant impacts.

Doña Juana waste landfill started to operate in 1988, and it was designed to provide a useful life of nearly 25 years. However, it has not been possible to find a new place to dispose of this great amount of daily garbage, and the useful life of the landfill has had to be extended several times, creating new waste cells above the old ones. This represents a great challenge from the technical and environmental point of view and implies a detailed and continuous monitoring and landslide hazard evaluation under different operational scenarios. The monitoring results, the operational process and hazard analysis are discussed in this article, emphasizing mechanical parameters, pore pressures and stability calculations based on probabilistic analysis.

2 General Description of Doña Juana Waste Landfill and Previous Instability Problems

Doña Juana MSW covers an area of 500 ha, and its waste height varies from 30 to 120 m. Geologically, the area corresponds to a basin of sedimentary rocks, mainly claystone and limestone from the Bogotá Formation and Quaternary Deposits. It is in the Southwest of Bogotá city, near the Tunjuelo River, as shown in Fig. 1. The operational area has been divided into 14 internal sectors and each sector in turn divided into several sub-sectors or operational stages. Figure 2 presents a planned view of the study area with the historical landslides and all the Doña Juana Landfill areas.

Fig. 1

General location of the study area

Fig. 2

General view of the Doña Juana MSW indicating the location of past landslides and the study area

There have been three important landslides in the more than 30 years of operation. The first, and the largest one, was a flow-type landslide that occurred on September 27, 1997, in which more than 1.2 million m³ were removed and travelled about 1500 m in 20 min. The removed material clogged the Tunjuelo River (Caicedo et al. 2002). It caused severe environmental affectation, such as contamination of the river water and propagation of bad odours more than 10 km away. On October 2, 2015, a second landslide of approximately 600,000 m³ occurred in the area called Terrace 1 (current area of operation). The slipped waste clogged a micro tunnel, which worked as an auxiliary leachate evacuation system because the main pipe had suffered frequent clogging. The third landslide occurred on April 28, 2020, removing about 80,000 tons of waste, but the situation was promptly controlled with no further affectations. One remarkable detail about the last event was the lack of an available disposal zone, the reason why the operator had to overfill the removed area.

From the first landside, many geotechnical studies have been carried out, including field and laboratory tests, stability analysis and modelling of the complex physical, chemical, and biological decomposition processes (Caicedo et al. 2002; González and Espinosa 2003a). The main conclusions from these studies are:

• (a) Large changes in the material properties occur over time, due to the chemical and biological reactions that result in the conversion of solid materials into liquids or gases.

• (b) One of the main causes of the first landslide was the recirculation of leachate in the waste cells, a method that is used in many industrialized countries to reduce the contaminant load of fluids.

• (c) Pore pressure is severely affected by the gas pressure generated during the decomposition of the waste and for that reason, geotechnical analysis can be addressed by means of the unsaturated soil behaviour theories.

• (d) Mechanical parameters were measured directly on the waste material using field shear tests, Mennard-type pressurometer and piezocone field tests (Caicedo et al. 2002).

• (e) Field instrumentation and monitoring were recommended to avoid new instability problems in order to perform more precise stability models.

3 Operational Problems Related to Leachate and Gas Evacuation

The initial design of the drainage system contemplated filters, pipes, ponding, and reinjection of leachate, as shown schematically in Fig. 3 (Collazos 1998). Leachate reinjection has been used in many MSW and according to Espinosa and González (2003a), some of the reported benefits are: reduction of wastewater treatment, temporary storage for leachate, with which its disposition differs in time, part of the leachate may evaporate during recirculation, settlements acceleration due to a more efficient biological degradation, etc. and some of the most important disadvantages are deformations of the internal facilities due to the high settlements, possible biological pipe obstruction and filter clogging, lack of experience related to the design and operation and a dangerous increase of pore pressures.

Fig. 3

Initial leachate management design, including gravity-based drainage and reinjection process. Adapted from (Collazos 1998)

Leachate reinjection has been considered the main cause of the 1997 Doña Juana Landslide (Caicedo et al. 2002; González and Espinosa 2003a) because pore pressures increased significantly due to the inefficient leachate evacuation system. This problem increased with the second landslide (2015) because a micro-tunnel that served as a gravity-base leachate evacuation system was clogged so it was necessary to adopt contingent measurements for the leachate evacuation through the construction of smaller diameter pipes and pumping wells, as shown in Fig. 4 and in Fig. 5. However, the new system has not been efficient, because gas and leachate packets are frequently detected. Due to this condition, the National Environmental Licensing Agency (ANLA) imposed a preventive measure filed in which the temporary disposal of material in one of the terraces was prohibited, based on the increase in pressures of leachate and gases.

Fig. 4

Actual leachate management design

Fig. 5

(a) Pressurized output of leachate through chimneys, (b) chimneys with extraction of leachate by pumping output of leachate through chimneys

Another operation problem that has been reported is related to deficiencies in the compaction process and, for that reason, considerable settlements have occurred, affecting the gas and leachate extraction wells. According to a recent geophysical investigation, it is estimated that around seven million m³ of non-extracted leachate are in the waste disposal zone (UAESP, UT INTER DJ,, and Saicon Ingenieria SAS 2023).

4 Geotechnical Characterization and Monitoring

Doña Juana MSW has been the subject of several geotechnical studies. The first one with the objective of determining the causes and legal responsibilities of the 1997 landslide; In this case, a total of 14 piezocone tests with an average depth of 20 m. were performed. Also 4 in-situ pressure tests, waste sampling, large-size in-situ shear tests and many laboratory tests (Gonzalez and Espinosa 2003; Caicedo et al. 2002).

From 1998 to today, a system of vibrating wire piezometers was installed in old areas and sectors of the current operation, also 26 inclinometers were installed in the contacts between the waste cells. There are more than 300 piezometers in operation, but several of them are recent and have few records. Figure 6 shows the location of 195 piezometers in the current operating area, called the Optimization Fase II and its adjacent disposal areas. This figure also shows the sites where piezocone tests (CPTu) were carried out. In 16 of these test points, excess pore pressure dissipation tests were also carried out, stopping the advance of the cone and continuously recording the change in pressure until equilibrium was reached.

Fig. 6

Location of piezometer monitoring and points of CPTu tests

Another monitoring process consists of daily topographic controls. Initially, these controls were done using conventional topography, yet more recently, they have been done using drones. Geophysical monitoring, such as electrical resistivity tomography, is also carried out with the purpose of identifying the presence of leachate and gas pockets.

The classification of the materials from the CPTu tests was made based on the classification index (CI) methodology proposed by Robertson (2016).

$$ CI={\left[{\left(3.47-\log \left(\frac{q_t}{P_a}\right)\right)}^2+{\left(\log (Rf)+1.22\right)}^2\right]}^{0.5} $$

Where, q_t is the tip resistance corrected for probe dimensions (kPa), P_a is the atmospheric pressure (100 kPa), and Rf is the friction ratio. The results of these tests are shown in Fig. 7. These results are consistent with those obtained in other sanitary landfills in Brazil, China, the United States, India and Portugal, presented by (Ramaiah et al. 2017), where the largest amount of reported data falls in zones 4 and 5, which are behaviours of silt mixtures and sand mixtures respectively.

Fig. 7

Classification index of Doña Juana parameters in the Robertson (2016)

The resistance parameters were calculated conventionally as:

$$ \tau ={\sigma_v}^{\prime}\tan \left({\varphi}^{\prime}\right) $$

Where, σ_v^′ is the effective vertical stress, τ is the equivalent shear stress and φ^′ the friction angle. Internal effective, which was estimated using the correlation proposed by Kulhawy and Mayne (Kulhawy and Mayne 1990), given by

$$ {\varphi}^{\prime }={17.6}^{{}^{\circ}}+11\ast \log \left({q}_{t1}\right) $$

Where, q_t1 is the tip resistance normalized by stress, which is defined by:

$$ {q}_{t1}=\frac{q_t-{\sigma}_{vo}}{{\sigma_v}^{\prime }} $$

Where, q_t is the tip resistance corrected for probe dimensions, σ_vo total vertical stress, and ${\sigma}_v^{\prime }$ effective vertical stress.

On the other hand, cohesion is not interpreted as the intercept of the line with the axis of the ordinates, in this case the criterion established by (González and Espinosa 2003b) is used, where the intercept is assumed as cohesion. of the envelope for an equivalent stress of 19 kPa (almost the pressure at 2 m deep).

Figure 8 shows the results of the cohesive resistance and friction angle parameters. It is observed that there is an inverse-proportional relationship between cohesion and friction. The greater the angle of friction, the less cohesion. Thus, a potential relationship between the resistance parameters is projected in which the cohesion value fluctuates between 11 kPa and 21 kPa for variable friction angles between 7° and 48°.

Fig. 8

Relationship between Cohesion and Friction Angle in Doña Juana MSW

Figure 9 shows the results of the friction angle based on the relationship of Kulhawy and Mayne (1990) of all the CPTu tests with depth, highlighting the average in bold. This mean friction angle is used to calculate the cohesion behavior by the correlation of Fig. 8.

Fig. 9

Behavior of the friction angle with depth

A decrease in the angle of internal friction with depth is observed as well as an increase in the cohesion with deep (Fig. 10), which is consistent with what was found by González and Espinosa 2003b) for the Doña Juana landfill, where a decrease in the friction angle with age is established, considering that the deeper the waste the older the material.

Fig. 10

Left, variation of friction angle with depth. Right, variation of cohesion with depth in Doña Juana MSW

The statistical behaviour of the parameters obtained is consistent with the residue resistance parameter database created by (Daciolo 2020). It is observed in Fig. 11 that the distribution of the friction angle in Doña Juana material follows a normal distribution, and the value of cohesion follows a log-normal distribution. The following authors were revised to prepare a more complete database and the results (See Fig. 11).

Fig. 11

Strength parameter histograms of Doña Juana MSW compared with strength parameters of similar MSW reported by (Daciolo 2020)

5 Analysis of Pore Pressure and Efficiency of the Pumping Systems

Figure 12 shows pore pressure behaviour with depth within the current operating area, and Fig. 13 shows the results in areas outside the current operating polygon. As observed in the graphs, the pore pressure values in the adjacent external areas are lower than those in the current operating zone. This is explained because in the area of operation, the biodegradation of the garbage is in its early stages, and there is a higher leachate generation rate than in the adjacent areas.

Fig. 12

Pore pressure variation in depth in the actual operation zone

Fig. 13

Pore pressure variation in Depth in zones adjacent areas to the actual operation zone

For the stability analysis model, the Ru parameter is used, which is defined by:

$$ Ru={U}_f/\gamma H $$

Where, Uf is the pore pressure at a point in space, γ the unit weight of the material (in this case, the garbage), and H is the height or depth of the measurement point. In the case of garbage from the Doña Juana sanitary landfill, a unit weight value of 1.12 t/m³ is adopted, according to the mean value reported by the operator based on many waste density tests. The behavior indicated in Figs. 12 and 13, combined with the previous expression, is summarized in Table 1.

Table 1 (a) Ru vs. depth in the actual operation area (Fase II area) and (b) in adjacent waste disposal zones (next to Fase II)

Figure 14 shows the distribution of areas of equal pressure, according to the values recorded in April 2022. The red colours indicate high Ru values, above 0.7, and the orange colours, Ru values between 0, 5, to 0.7. According to the analysis made by the MSW operator, Ru values greater than 0.5 are considered to have a high potential to generate instability problems in the landfill.

Fig. 14

Map of equal pore pressure according to the piezometric readings in April 2022

Figure 15 compares the proportion of areas of high pore pressure (Ru > 0.5) with the precipitation values that occurred in different months between 2021 and 2022. As expected, areas of high-pressure level tend to increase as rainfall increases and decreases with rainfall reductions. Still, there is a delay in the response of 1 to 1.5 months.

Fig. 15

Comparison between monthly precipitation and areas of high pore pressure (Ru > 0.5)

To reduce pore pressure ratios, leachate extraction is made by forced pumping. In Fig. 16, the monthly pumping hours are compared with the areas of high pore pressure. It is observed that the operator tends to increase pumping hours when the rain increases. However, in many cases, pumping is not enough to reduce the areas of high pore pressure significantly and this is an important factor that may affect stability conditions.

Fig. 16

Comparison between monthly forced pumping and areas of high pore pressure (Ru > 0.5)

6 Stability Models

For the construction of the geotechnical model, in the first place, a friction angle value was estimated at the surface level based on the interpreted data. Then an extrapolation of the friction values in depth was made using the relationship presented in Fig. 9. The analysis profiles are divided based on the relationship of the friction angle with the depth and cohesion (Fig. 8). Average friction and cohesion values were assigned as they correspond to the average depths of the stratum, and finally, the values were entered. Ru data is based on the mean values at depth according to what is reported in Table 1. It should be noted that, the methodology described was suggested by Saicon Engineering and Union Temporal INTER DJ companies which performed the CPTu essays.

In total, 6 profiles were made for stability analysis, as shown in Fig. 17. As part of the discretization, layers almost parallel to the surface are created and on which there are variations of the properties using the previously explained methodology. Figure 18 shows one of the analysis profiles. It is important to clarify that the different layers indicated there represent a differentiation of mechanical properties and do not represent a variation in the typology of materials, as is usually done in conventional geotechnical models, since it is a very heterogeneous material that makes up the waste.

Fig. 17

Analysis profiles in the study area

Fig. 18

Two-dimensional geotechnical model of profile 2

For the threat analysis, stability modelling was carried out regarding the probability of failure, using the Monte Carlo method in the Slide software (Rocscience, 2005). For this modelling, the statistical values of the input parameters are needed, such as the mean, standard deviation, data distribution, and maximum and minimum values.

Table 2 shows the cohesion and friction values entered, which, together with Table 1, constitute the properties of the discretization of the model. Stability calculations were made by the simplified Bishop and Morgersten-Price methods for static and pseudo-static conditions. Table 3 summarizes the results of these analyses regarding the safety factor and the corresponding probability of failure.

Table 2 Parameters entered to each layer

Table 3 Results of the Fs calculation and the probability of failure of the profiles analyzed

7 Hazard Analysis

Based on the stability analysis results, it is possible to make a hazard map in terms of the probability of failure (Pf). In this case, based on the methodological guide for studies of hazard, vulnerability, and risk of the Colombian Geological Service (Ávila et al. 2016), a high hazard level corresponds to an annual failure probability greater than 16%; a medium hazard level to a probability between 0.1% and 16%, and a low hazard level if the probability of failure is less than 0.1%. Figure 19 shows the hazard zoning map for static conditions. It can be seen on this map that there are two large areas in medium hazard level, and the rest is in low hazard level. One of the medium hazard level zones is near the Biogas treatment plant, which is one of the internal facilities of the sanitary landfill, located in a sector of high topographic slopes, and the other, in the sector of maximum pore pressure values, close to a neighbourhood called Mochuelo Alto. The stability models are very sensitive to slope and pressure, dictating these areas with the highest probability of failure.

Fig. 19

Landslide Hazard map based on the probability of failure under static conditions

Under pseudo-static conditions, the situation may be critical because, as shown in Table 3, predominates high-level hazard conditions with failure probabilities greater than 16%, showing the importance of a cautious operation in keeping the pore pressures as low as possible.

8 Conclusions

The Doña Juana Landfill has been forced to expand its useful life, forcing waste cells to be generated on top of older ones to optimize space. These changes have modified the original designs; therefore, continuous monitoring and opportune decision-making are required to guarantee stability.

Results of piezocone tests on the waste material showed that as depth increases, friction angle decreases and cohesion increases. This behaviour is consistent with the variation of the parameters in terms of the age of the waste material proposed by González and Espinosa (2003), who establish that the older a residue (more depth), the lower the friction angle and the greater cohesion.

The variable that most controls the stability of the waste landfill is the fluid pressure, for which an efficient extraction of leachate and gas is essential.

The leachate evacuation system by gravity suffered a general collapse and clogging, which made it necessary to adopt a forced extraction system by pumping. This system is not sufficient for an efficient leachate evacuation in a large sanitary landfill such as Doña Juana and although the hazard levels are in medium and low conditions, a failure or delay in the pumping could cause drastic increases in pressure and increase the hazard condition. It is also observed that with the current pressures, in the event of an earthquake, a high-hazard situation can be generated in a large proportion of the landfill.

The elements with the highest landslide exposure levels are Mochuelo Alto neighbourhood, where several low-income families live, and the internal biogas treatment plant.