Beruflich Dokumente
Kultur Dokumente
Mining
Mineração
Comparing blast-induced
ground vibration models
using ANN and empirical
http://dx.doi.org/10.1590/0370-44672017710097 geomechanical relationships
Wilson Siguemasa Iramina Abstract
Universidade de São Paulo - USP
Escola Politécnica Blasting remains as an economical and reliable excavation technique, but there
Departamento de Engenharia de Minas e de Petróleo are some environmental shortcomings such as the control of blast-induced vibra-
São Paulo - São Paulo - Brasil tion. The impacts of vibration over surrounding communities in a blast area have
wilsiram@usp.br, wiramina@hotmail.com been investigated for decades and researchers have been using a myriad of empirical
predictive attenuation equations. These models, however, may not have satisfac-
Eduardo Cesar Sansone tory accuracy, since parameters associated to geomechanical properties and geol-
Universidade de São Paulo - USP ogy affect the propagation of seismic waves, making vibration modeling a complex
Escola Politécnica process. This study aims for application of an Artificial Neural Network (ANN)
Departamento de Engenharia de Minas e de Petróleo method and Geomechanical parameter relationships to simulate the blast-induced
São Paulo - São Paulo - Brasil vibration for a Brazilian mining site and then compare them to the traditional ap-
esansone@usp.br proach. ANN had the best performance for this mine despite having demanded
large datasets (as much as for the traditional approach), while geomechanical pa-
Michiel Wichers rameters like RQD and GSI may be used to deliver a fair approach even without
Universidade de São Paulo - USP seismic data. Also, ANN methods may be useful in dealing with a large amount of
Escola Politécnica information to facilitate the simulation process when combined with other methods.
Departamento de Engenharia de Minas e de Petróleo Therefore, alternative prediction methods may be helpful for small budget mining
São Paulo - São Paulo - Brasil operations in planning and controlling blast-induced vibration and helping mining
michiel.wichers@usp.br in urban areas becoming a more sustainable activity.
Sugeng Wahyudi Keywords: blast-induced vibration; attenuation equation; Artificial Neural Network;
Kyushu University - Faculty of Engineering geomechanical relationships.
Department of Earth Resources Engineering
Fukuoka – Fukuoka - Japan
wahyudi_sugeng@mine.kyushu-u.ac.jp
Hideki Shimada
Kyushu University - Faculty of Engineering
Department of Earth Resources Engineering
Fukuoka – Fukuoka - Japan
shimada@mine.kyushu-u.ac.jp
Takashi Sasaoka
Kyushu University - Faculty of Engineering
Department of Earth Resources Engineering
Fukuoka – Fukuoka - Japan
sasaoka@mine.kyushu-u.ac.jp
89
REM, Int. Eng. J., Ouro Preto, 71(1), 89-95, jan. mar. | 2018
Comparing blast-induced ground vibration models using ANN and empirical geomechanical relationships
1. Introduction
Blasting is frequently the most cost tion activity. To avoid conflicts between seismograph monitoring data may not
effective technique for rock excavation. communities and blasting areas, blast- be enough for a reliable characterization
The use of explosives in mining and con- induced vibration should be controlled of the vibration behavior in one or more
struction activities is almost unanimous and monitored. Enhancements in blast directions. The question is: is it possible
and the development of new products design and the use of new products will to obtain a fair to good approach for
and new blast design studies will keep only prove sustainable if there is a precise predicting attenuation using a limited
blasting as the favorite technique in the and reliable way of knowing the impacts seismograph monitoring dataset? Is there
coming decades. Although reliable and of the ground vibration. Therefore, in the any other approach?
economical, the use of explosives must last decades, predictive attenuation mod- This study aims for the applica-
deal with some environmental shortcom- els have been proposed. However, due tion of an Artificial Neural Network
ings. Since many mining sites are sur- to their empirical nature, these models, (ANN) and relationships based on
rounded by people, some of the effects of known as predictive attenuation equa- geomechanical parameters to model the
blasting like noise and ground vibration tions, had low accuracy when several di- blast-induced vibration attenuation using
may be considered as annoying or even rections needed to be studied. A general blasting and rock mass parameters at a
a threat. In fact, blasting in urban areas prediction equation is desired but with Brazilian mining operation. It is expected
has become a challenge, since regulations different parameters of the rock mass, that the results may have better accuracy
concerning blast-induced vibration have such as geomechanical properties that when compared to some empirical meth-
become more rigid and this situation affect the seismic wave propagation in ods for feasibility purposes to help small
could severely affect the sustainability different ways for different directions. budget mines and quarries in urban areas
of the mining operations or construc- Therefore, using a limited number of to become more sustainable.
Geology and geomechanical as- velocity (PPV) and it basically depends eters are features of the rock mass
pects of the rock mass affect the on two main variables: the mass of the related to surrounding local geology
characteristics of ground vibration as detonated explosive charge and the dis- and geomechanical characteristics that
waves move away from blasting area. tance between the detonation point and have a major influence on vibrations. In
This dissipation or "geometric spread" the measuring point. This relationship massive and homogenous rock vibra-
occurs when a finite amount of vibra- can be mathematically established from tion, the waves may propagate almost
tion energy needs to fill an increasing the data obtained by seismographic equally in all directions. However, in
amount of the rock mass volume as monitoring. Among the many empiri- complex geological structures, the wave
it moves toward the farthest point of cal relationships, the most used is the propagation may vary with the direc-
detonation. Thus, there is a decay in one that correlates the peak particle tion, therefore resulting in a different
vibration amplitude with increasing velocity with “scaled distance” used attenuation requiring representation
distance from the source, similar to the to predict the effects of ground vibra- by different propagation equations. In
inverse-square law. Other effects are the tion on structures and humans. Scaled the last decades, several researchers
propagating energy losses by absorption distance (SD) is defined as the distance established empirical equations to pre-
and dispersion and the formation of from blast divided by the nth root of the dict blast-induced vibrations. Some of
surface waves. The great variation in weight (mass) of the explosive charge the most important include those from
geological conditions of rock masses used in the blast. USBM in 1959, Langefors–Kihlstrom
presents difficulties in predicting vibra- The parameters that affect the in 1963, Ambraseys-Hendron in 1968,
tion attenuation. Currently, most of the characteristics of the vibrations are Nicholls et al. in 1971, Siskind et al. in
predictions were obtained by statistical- basically the same that influence the 1980, Pal Roy in 1991 and CMRI in
empirical relationships (Attenuation results of a blasting and they are gener- 1993. These and other equations were
Equations or AE) where vibration can ally classified into two groups: control- listed by authors such as Khandelwal
be correlated with structural damage. lable and uncontrollable. Controllable and Singh (2009), Kamali and Ataei
The most used parameter to represent parameters are those related to blast (2010), Wahyudi et al. (2011), Saadat
the ground vibration is the peak particle design and the uncontrollable param- et al. (2014) and Kumar et al. (2016).
3. Vibration attenuation modeling using the ANN method and relationships based on geomechanical parameters
training and have proven generalization with sigmoid transfer functions in the threshold or reaches a desired number
capability even when dealing with noisy hidden layers and a linear transfer of iterations.
input data. The operation of an already function in the output layer. The BP In the early 21st century, numer-
trained ANN is simple and can be eas- training algorithm is applied to deter- ous researchers tried to use ANNs to
ily incorporated into design methods. mine the set of weights of an ANN by predict blast-induced ground vibration
Several algorithms can be used using a corrective–repetitive process as an alternative method, since the Em-
to train neural networks, but back- where the actual output is compared pirical Attenuation Equation (AE) may
propagation-based algorithm (BP) has with the target output. The difference not work well due to noise (Wahyudi
had preference among researchers to or error between both is processed et al., 2011). According to Kamali and
solve predicting problems (Wahyudi back through the network, updating Ataei (2010), ANN has been used in
et al., 2011; Monjezi et al., 2013; the individual weights of the synapses. mining since 1990 and more specifi-
Kamali and Ataei, 2010 and Saadat These weights are adjusted by an error cally for blast-induced ground vibration
et al., 2014). Saadat et al. (2014) have minimization technique so that a target studies since 2004. The prediction of
used, for function approximation, a output will be produced for a given PPV using up to four variables can be
feedforward ANN. These researchers input. The updating process is repeated done effectively by a backpropagation
recommend the use of a BP algorithm until the network error converges to a algorithm with good performance.
(0.3396x1.02GSIGSI1.13)0.642SD-1.463
PPV = (1)
γ
where: PPV= Peak particle velocity in (m/s) SD = Scaled Distance in (m/kg 0.5)
GSI = Geological Strength Index γ = Unit weight in (KN/m3)
And the second approach, using the RQD relationship, is quoted by Equations 2 and 3.
For RQD ≤ 75
(0.5947RQD + 0.00893RQD2)0.642SD-1.463
PPV = (2)
γ
(- 7.91562RQD + 0.12152RQD2)0.642SD-1.463
PPV = (3)
γ
where: PPV= Peak particle velocity in (m/s) SD = Scaled Distance in (m/kg 0.5)
RQD = Rock Quality Designation γ = Unit weight in (KN/m3)
91
REM, Int. Eng. J., Ouro Preto, 71(1), 89-95, jan. mar. | 2018
Comparing blast-induced ground vibration models using ANN and empirical geomechanical relationships
4. Applying ANN and GSI/RQD relationships to a Brazilian mine site vibration attenuation study
A granite quarry, surrounded by A seismographic monitoring was per- for different directions. Published field
communities located in São Paulo State formed in this quarry and four blasts blast data (Ramirez Canedo 2013 and
(Brazil), was previously studied by the were recorded. All blasts had similar Ramirez Canedo et al., 2015) were used
Laboratory of Environmental Control, features and they were performed in in this study. Figure 1 shows a satellite
Industrial Hygiene and Safety in Mining the same bench. Several scaled-distance view of the mine and the distribution of
at the University of São Paulo, Brazil. attenuation equations were estimated the sensors in Ramirez Canedo’s study.
Figure 1
Blasting sites and
distribution of sensors
(modified from Ramirez Canedo et al, 2015).
Empirical attenuation equation RQD based relationship approaches the best estimation. Figure 2 shows the
(AE), Artificial Neural Network method were performed in this study. The use performance for each method compared
(ANN), GSI based relationship and of recorded field data allowed to rank to real data.
Figure 2
Comparing AE, ANN, GSI and
RQD performances for PPV prediction.
RMS errors for AE, ANN, RQD and use field data to deliver the vibration attenu- along with the rock mass, as it may reflect
GSI methods were computed and presented ation while GSI and RQD only used mean differences in the direction of propagation.
the respective values of 3.63, 3.25, 7.81 and values of rock mass to estimate vibration. Figure 3 shows the performances of
4.36. These values indicate that ANN has Since there is a range of variation in both GSI (left) and RQD (right) relationships
the best estimation, followed by AE. This GSI and RQD values, it would be interest- for lowest and highest values obtained in
was expected considering that both methods ing to know the influence of their variation field tests.
Figure 3
Comparing performances
for lowest and highest field
values of GSI (left) and RQD (right).
The analysis of the ranging values (GSI and/or RQD) for each direction, it ic directions. PPV prediction by Kumar et
obtained for GSI and RQD show that is believed that the results could be even al.´s model was limited to scaled distances
there may be differences in the vibra- more accurate, and in addition, applying up to 80 m/kg0.5 but for the studied quarry,
tion propagation possibly related to rock ANN to study geomechanical parameters it seems to be acceptable for values up to
mass properties of the quarry. However, accurately could be a more effective solu- 120 m/kg0.5. Near-field attenuation may
comparing the prediction performance for tion. For the particular directions elected also affect the prediction for small scaled
the extremes in the scaled distance range by Ramirez Canedo (Ramirez Canedo, distance since interactions in this field are
suggests that there is a certain homoge- 2013 and Ramirez Canedo et al., 2015) not well known.
neity in terms of vibration propagation. to obtain the attenuation, the resulting Not always ANN itself may be the
Therefore, in this case both GSI and RQD equations might be preferable over other best solution for an attenuation study. Its
averages could be used to fit actual values solutions (ANN, GSI, etc.), despite the processing capacity relies on a large num-
when correlating vibration with rock relatively few number of sensors used in ber of previously obtained recorded data
mass properties. the interpolation. The way the surveys that are sometimes very difficult to get, not
If we could establish a relation- were carried out make them reliable for to mention time spent and costs for col-
ship between geomechanical parameters predicting the attenuation for those specif- lecting them (years and tens of thousands
93
REM, Int. Eng. J., Ouro Preto, 71(1), 89-95, jan. mar. | 2018
Comparing blast-induced ground vibration models using ANN and empirical geomechanical relationships
of dollars). Additionally, a large variety of when applying ANN. However, for far a large amount of information. This
input parameters may not result in better field attenuation, ANN may be a good facilitates the process itself, either for
simulation, as distance and charge per solution. As new data are collected, ANN simulation purposes or for validation of
delay are the most prevailing parameters methods may be very useful to deal with AE methods.
6. Conclusions
A Brazilian mine was studied us- able empirical model for PPV prediction rock interface. Seismograph monitoring
ing an empirical attenuation equation, since site-specific empirical models are in countries like Brazil is expensive and
Artificial Neural Network and geome- available but they cannot be general- most of the time is performed by using
chanical parameters-based relationships, ized for different sites. Mining is an few sensors (usually up to two sensors for
to obtain a more accurate blast-induced important activity in many countries each blast), which may affect the quality
vibration attenuation model. The avail- and mining in urban areas has become of the empirical attenuation equation and
ability of input parameters, the size a challenge as complexities of rock sites Artificial Neural Network approaches.
of the dataset and the purpose of the demand accurate PPV models considering On the other hand, using geomechanical
seismograph monitoring may affect the a number of effects. These effects are re- based relationships could be a useful al-
resulting simulations. lated to rock discontinuities, rock types, ternative for a satisfactory estimation of
Developing countries and small rock formation, rock joints and their ori- vibration, helping mining in urban areas
mines require the development of a suit- entation, groundwater presence and soil- to become a more sustainable activity.
Acknowledgment
The authors express their gratitude volvimento Científico e Tecnológico, of Innovation of Brazil, who has granted a
to CNPq - Conselho Nacional de Desen- the Ministry of Science, Technology and scholarship for the corresponding author.
References
95
REM, Int. Eng. J., Ouro Preto, 71(1), 89-95, jan. mar. | 2018