Correlating blast vibrations and geomechanical properties to determine damage proﬁ les and improve wall conditions in open pit mining

.


INTRODUCTION
In the current mining economy, some obstacles to high productivity stand out: the lower value of metals, the high cost of dilution and the expansion of dependence on low-grade mineral deposits.Such facts require the search for an optimized mineral exploitation in which each operation is well controlled, including the control of the damages generated by the rock excavations, subject of the present study.
Among all the factors to be considered while mining, the slopes stability and the fulfi llment of the planned pit have signifi cant importance and impact at an effi cient and safe mineral exploitation.The increase in the slope angle provides an increase in the strip ratio and consequently the higher cost of mineral extraction.This statement is in accordance with Adamson & Scherpenisse (1998), who state that the development of an open pit mine operations depends, among other factors, on the creation of stable walls.
In order to overcome the obstacles mentioned before, it is extremely important to understand a factor that is usually ignored in the blast design: the rock is heterogeneous, with unique and local characteristics (Singh 2001).Therefore, it is important to consider the geo-mechanical parameters and rock mass structures.
Regarding wall protection, classifi cations of rock masses such as RMR (Bieniawski 1973), Q (Barton et al. 1974) and GSI (Hoek 1994) are very important, for both geotechnics and rock blasting.In this context, Hoek (2012) introduces the blast damage factor D, which describes the damage caused by the blasting in the rock mass, thus affecting its GSI classifi cation, in particular Bulk Modulus.In addition, it correlates this damage to the bench height, according to parameters used in blast design such as buffer rows and confinement.McKenzie (2016) points out the three main mechanisms of damage caused by blasting: induced vibrations (responsible for creating new fractures), extreme gas pressure (responsible for the expansion of fractures) and release of load (responsible for the creation of fractures parallel to the rock movement).Those are very important factors while addressing pit slope stability.
In this context, many studies have discussed the relationship between blasting, induced vibrations and damage: Ouchterlony et al. (1996), Adamson & Scherpenisse (1998) McKenzie & Holley (2004), McKenzie (2012), Wahyudi et al. (2011) and Onederra et al. (2013).Among these studies, it is important to highlight Adamson & Scherpenisse (1998) and McKenzie (2016), who discussed the effects related to vibration according to Hooke's law and explain the concept of critical peak particle velocity (PPVc), assuming the elastic behavior of the intact rock.According to this concept, PPVc refers to the level of agitation that can be supported by the rock before tensile damage occurs and this value depends on the tensile strength, Young's modulus and primary wave propagation velocity.In these studies, correlations between the vibration peaks and the extent of fracturing are shown.Thus, critical PPV is defi ned as: (1)

Where
is the tensile strength of intact rock (MPa), Vp the primary wave velocity, and E is the Young's modulus.
Furthermore, another very important factor for wall strength preservation is to understand how vibrations and consequently damage are attenuated with distance.Several authors, such as Devine et al. (1966), Blair (1999), Hizen (1998), Silva-Castro (2012), McKenzie (2012), andMcKenzie (2013) have studied the behavior of the shock wave generated by the one hole charged with explosive, the seed wave, determining its characteristics (length, velocity, amplitude, frequency) and understanding the relationship between the generation and attenuation of induced vibrations.
This study seeks to assist the staff of an open pit copper mine in Brazil to tackle the challenge of obtaining more stable walls facing blasting induced damage to build a safer and more productive mining operation.The aim is to correlate blasting induced vibrations to damage and build a representative model in order to determine the best blast design to improve wall stability with optimized muck pile´s size distribution.Figure 1 illustrates some slopes of this mine and depicts the challenges of this project: geometry bench issues, such as the loss of crests, the generation of large boulders, and the high degree of damage induced by the actual blasting practice.Figure 1 also shows the slope parameters at this lithological domain.
Although, the actual blast performance regarding muck pile´s size distribution and shape meet the requirements for reasonable plant effi ciency.Nevertheless, wall stability and geometry have become an issue due to the high powder factor applied at this domain.
The biotitite domain was chosen for this study, which according to Miranda et al. ( 2018) is characterized by the intercalation of biotite schist and amphibole schist, and contains boudinaged amphibolite lenses, muscovite schists containing sericite, as well as hydrothermalized gneiss biotite.This domain was classifi ed into RMR rock mass types II and III.Furthermore, discontinuities sets are well defined, and the influence of hydrothermal alteration can be observed.

METHODOLOGY
This paper discusses the study conducted to generate a vibration attenuation and seed wave model for the biotitite domain of a copper mine used in order to correlate blasting induced vibrations and damage to the rock mass.
As a first step, a cross-hole test, using geophones and small explosive charges, was performed to capture seismic data in near field, characterizing the dynamic interaction between rock mass and the explosive released energy, providing full characterization of waves generated.The geophones'used are shown in fi gure 2 and their characteristics are described by table I. Mrel's Data Trap II was used to record the information measured with the geophones.
The second step was to generate the equation which characterize the waves' amplitude attenuation due to distance and rock mass heterogeneity.Once wave attenuation was characterized, the second step was to defi ne the limits where vibrations cause damage in this specifi c rock mass based on the reference values proposed by Adamson & Scherpenisse (1998).These two information enable to correlate Peak Particle Velocity (PPV), damage and distance from the blast and create a model.
This study simulated different blast designs, contemplating two blasts tests to evaluate model effectiveness and it was characterized the damage of these blasts through the inspection of test holes drilled in the close vicinity of the blasts.Figure 3 illustrates the methodology applied in this study.Cienc (2022) 94(4) e20211080 4 | 14

Field test and seismic characterization
The fi rst step of this study was to gather the geomechanical parameters that characterize the biotitite.Table II summarizes this information collected from the mine's geology team.Another important information to be considered is the geometric parameters of the expected slope, as shown in Table III, provided by the mine planning team.
Then, a fi eld test was carried out to capture seismic data at near distances, characterizing the dynamic interaction between the rock mass and the detonated explosive charges.Eight holes were drilled, with a diameter of 171 mm and an average depth of 10.4 m, 5m away from each other.The geophones were cemented in the holes G1 and G2, as shown in fi gure 4. The other holes, Q1-Q6 were charged with bulk emulsion, density 1.15 g/cm 3 and VoD 5400 m/s.All holes were charged with the same explosive charge, 53 kg per hole.
Once the holes were detonated and in possession of vibration information generated in relation to time, it was possible to determine factors such as: propagation velocity P waves (Vp), wave geometric parameters (length, amplitude, period).Figure 5 illustrates the wave form generated by hole Q6 and acquired with geophone 1.

Vibration attenuation and damage model
There are several models and equations from different authors, who sought ways to characterize the phenomenon of attenuation of vibrations with distance.Devine et al. (1966), proposed the model used in this study, due to its easy applicability and representativeness, based on the stepped distance illustrated in equation ( 2): (2 Where PPV is the peak particle velocity measured in the geophones, K is the constant which characterizes the interaction between rock mass and explosive, W is the maximum charge blast instantaneously and n is the ground attenuation constant.K and n are determined empirically using a data base containing peak particle velocity and the amount of explosive  charge which generated it and its respectively distance from the source.This data allowed the calculation of the vibration attenuation equation in order to characterize this phenomenon.The aim is to defi ne the seismic waves´ parameters and how they attenuate with distance. This equation was used to calculate de vibration amplitude in different positions upstream the blast.The resultant PPV is gathered from the addition of seed waves (measured in the fi eld test) considering timing and sequencing and scaling charge and distance, using equation 2, as proposed by Blair (1999), generating the vibration attenuation and seed wave model.

Correlation between vibration and damage
According to Adamson & Scherpenisse (1998), there is a close relationship between fracturing and PPV and it is described as shown in Table IV.Therefore, once vibration attenuation phenomenon was characterized, the next step was to defi ne which vibration thresholds cause damage at the rock mass, the critical PPV.
The geomechanical properties of the rock mass gathered with geology mining team and the P-wave velocity acquired during the fi eld tests enabled to calculate the critical PPV.
The author connected the vibration attenuation model and the expected vibration levels with the effect (damage) as a function of the critical PPV.This combination allowed creating a model which generated vibration heat maps to analyze the damage extension behind the blasts, according to each blast design tested.T able II.Biotitite geomechanical properties.

Bench Angle 70º
Berm width 7 m

Ramp width 35 m
Inter-ramp angle 45º

Simulations
The next step was to conduct simulations to understand damage propagation and defi ne the blast design to be utilized in the blast tests.In accordance with operational restrictions, such as borehole diameter, explosives and bench height, different simulation scenarios were defi ned as described by table V. Scenarios H and I take into account buffer lines which were dimensioned to reduce blast induced vibration on the slope and, consequently, reduce the damage.Table VI describes the buffer's design.
Vibration heat maps were generated for each scenario where the expected induced vibration was analyzed and compared with the PPV c of the studied lithological domain.These heat maps permitted to determine which blast design would provide a better slope condition (reducing the damage), thus permitting to defi ne, within the mine operational restrictions, which drilling and blasting confi guration should be selected for testing to compare with the currently adopted at the mine site (A).

Blast tests
Two blast tests were conducted nearby in order to evaluate the simulations's fi ndings.One of the test adopted the scenario A and the second adopted scenario H.They were executed at the same domain of the fi eld test aiming to have similar conditions.A borehole camera was used to inspect the drillholes specially made upstream of the two blast tests in order to confi rm the damage indications of the vibration heat maps, and the effectiveness of the proposed blast design in reducing blasting damage.These images were used to identify the degree of fracturing of the rock mass and counting it before and after the blasts, which allowed correlating the modeled levels of vibration with the fracturing increase, therefore permitting to validate the proposed model.

RESULTS
The gathered data in the cross-hole testing enabled to obtain an average value of Vp (primary wave propagation velocity) of 4926 m/s.This value, which is considerably high, indicates high competence of the rock mass, corroborated by its geo-mechanical information, gathered from the mine staff: UCS of 118 MPa and Young Modulus of 51 GPa.Thus, using Equation 1, it was possible to determine the critical PPV characteristic of this rock mass.The value found for biotitite in site conditions was 1139 mm/s.
The data obtained in the fi eld was suffi cient to generate a vibration attenuation model of high correlation.Equation 2 highlights the applicable equation, according to the reference work of Devine et al. (1966).Thus, the K value found is 725 and the n value is -1.81, with a correlation coeffi cient (R²) of 0.92. (2)

BLAST VIBRATION AND GEOMECHANICAL PROPERTIES
An Acad Bras Cienc (2022) 94(4) e20211080 8 | 14 In view of the expected damage as a result of PPVc, heat maps that correlate vibration and damage were generated for each of the considered scenarios.The results of each scenario were compared with scenario A (commonly used in the mine) to determine, within the blast design parameters range, those that contribute most to the rock mass blastinduced damage.
Table VII presents the relationship between damage extension and bench height, as proposed by Hoek (2012).The damage extension was measured from the slope until the border of the zone where the PPV level was smaller than a quarter of PPVc.It is very important to consider this damage extension as the change of the rock bulk modulus and consider the impacts of this change at the rock mass quality and its effect at slope stability.
These findings were the fundamentals for the definition of an optimized but feasible blast design considering the existent drilling equipment, however being safer and delivering higher productivity.
The damage extension of each scenario was calculated based on the damage model being compared with the bench height, as suggested by Hoek (2012).The values found are in accordance with Hoek´s statements, varying only for scenario G, with a bench height of 20 m and, according to the model, corresponding just to a 23% damage increase in comparison to the base scenario A. Thus, the ratio of bench height and damage extension is 1.1, the smallest value simulated.The results in Table II show that adequate timing and sequencing and the use of buffer rows promote a strong impact on damage, corroborating the results stated by McKenzie (2016).
According to the modeling results, scenarios A and H were chosen to be tested in the mine: A is the one that is already in use, while H was chosen due to the considerable reduction of expected damage in contrast with the relatively minimal efforts for its application (since the use of more than one buffer rows would considerably increase the number of drilled holes).
The Figure 6 shows the area of both tests.It illustrates the location of the blasts and the main geological structures of the area.It brings to attention the proximity of both blasts and its similar conditions, presenting the same predominant rock mass and main structures.
Figure 7 illustrates the first blast test, with the blast design referring to scenario H, and its respective heat map, showing the location of the inspection holes (holes upstream of the polygon) and in black the slope after the blast, indicating a good geometrical adherence between modeling and reality.The inspected holes are 2m away from each other and the fi rst inspected hole was drilled 7m away from the last row of boreholes.
Figure 8 shows images captured while inspecting the holes.The inspection of the other two holes suggests a signifi cant reduction in the damage level, illustrated by fi gure 8b and c.
These results corroborate the indications of the damage model, which suggest the more intense appearance of fractures in the first inspection hole and considerable reduction in the second and third holes.Thus, according to these fi ndings the real damage extension is around 12-13 m due to the reduction of fracturing in the last hole inspected.Table VIII shows the details of the fracturing intensity pre and post blast test 1.
The same procedure was repeated in the second blast test using the design described for scenario A and due to operational limitations just two holes were drilled for inspection.Figure 9 shows the heat map of this blast test, highlighting the remaining slope in black, showing a large discrepancy between the planned and the executed.
This time, due to the non-use of buffer rows and appropriated timing and sequencing, very extensive damage was observed causing the collapse of the fi rst hole inspected and very intense fracturing at the second one.Figure 10 shows images of the holes inspected, with a high degree of damage and the collapsed hole (5C).Table IX shows the details of the fracturing intensity pre and post blast test 2.

DISCUSSION
The first blast test, where an appropriate buffer row, timing and sequencing were used, only the first third of the inspected holes did not fulfilled the modeling prediction, since it was observed a strong disturbance of the hole shape.It is believed that this fact is a side effect of excessive subdrilling used in the blast of the upper bench, which it has caused a significant damage at the top part of the blast test one.Thus, this is supported by the high vibration profile observed at the berms in the heat maps.
In the second blast test, with no buffer rows and where the direction of the blast was parallel to the slope, strong adherence between modeling and reality was observed, since high vibration levels were expected in the first inspection hole.The collapse of the closest inspected hole highlights this fact and strengthens the representativeness of the damage modeling indications.
The results observed in field for these two blasts tests showed a strong correlation with the modeling results, as registered in the images obtained with the borehole camera.This fact proves that this methodology is a useful tool for building a blast design to reduce wall damage.Thus, the results of simulations and the blast test has shown the importance of the use of buffer rows and appropriated timing and sequencing, since it was observed significant reduction in the number of fracturing and respectively damage where this technique was applied.
Furthermore, the reduction of the vibration levels by using buffer rows and appropriated timing reduced the impact of blasting at the rock mass, not affecting the bulk modulus as much as blasting without any control.Due to this fact, the use of this technique helped to improve slope stability as the damage zone extension is smaller.The model helped to identify the changes of damage zones according to the blast design providing guidance to address the challenge of more stable walls.Although the number of fractures pre and post blast were surveyed, it was not possible to estimate RQD due to equipment limitations (the borehole camera did not have meter count).

CONCLUSION
This study has presented a feasible methodology to achieve more stable walls taking into account site particularities.The results were in accordance with the studies presented in the introduction section.The Blast Damage Factor is a great guide to geotechnical teams at mines, which can improve their modeling using this methodology to consider site-specific assessments.
The results obtained in the simulations and blast tests showed a strong correlation between the modeling results and the field observations, validating the values presented in Table II.
This methodology can help to implement new practices to promote wall control blasting and consequently more stable slopes.This is because the model can be used to analyze * Hole d was collapsed and it was impossible to determine the increase of the number of fractures properly.
the extension and level of damage zones (consequently rock bulk modulus degradation) created behind the blasts and the influence of each blast design parameters.Furthermore, the key factors which influences most fracturing and consequently damage at pit walls were presented in respect of the studied domain.Upon the review of literature and as stated by authors, it has been proven significant influence of blast design factors, such as sequencing and buffer rows, at the PPV levels and consequently induced damage.
As a further work, it is proposed to conduct more blast tests in order to increase the data set to confirm the findings and also install a meter count in the camera to estimate RQD.Furthermore, the same methodology should be replicated to other domains in this mine site and other mine sites to support the findings presented here.

Figure 1 .
Figure 1.Slope conditions at the mine site and geometrical parameters according to mining staff.A shows geometry issues, B big blocks formed by excessive blast damage and C crest loss.

Figure 2 .
Figure 2. Geophone used in the fi eld test.

Figure
Figure 3. Chart illustrating the methodology steps.

Figure 4 .
Figure 4. Test confi guration where G1 and G2 are the two geophones installed and Q1-12 are the holes with explosive charges.

Figure 6 .
Figure 6.Area of the blast tests and their direction and dip direction of the region's main discontinuity sets.

Figure 7 .
Figure 7. Vibration heat map showing the inspected holes A B C in gray, expected induced damage based on PPVc and the actual slope in black for blast test 1.Fracturing is expected throughout the yellow zone and the distance between the last row of boreholes until its border is around 13 m.

Figure 8 .Figure 9 .Figure 10 .
Figure 8. Images from the inspected holes of blast test 1.A has been made in the upper third of hole a, while b and c show the reduction of fracturing of holes b and c, both represented in Figure 3.

Table I .
Geophones technical specifi cations.

Table IV .
Relationship between damage and PPV.

Table V .
Blast designs simulated using the model.

Table VI .
Buffer design details.

Table VII .
Comparison between the expected damage extensions for the different scenarios.

Table IX .
Fracturing intensity pre and post blast test 2.