Goto

Layer | Fill | Outline |
---|

Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|

Fill | Stroke |
---|---|

**Abstract**

Determination of Unconfined Compressive Strength of rock is important in Geotechnical Engineering. Although laboratory test is the most direct way of rock compressive strength estimation, but UCS determination in laboratory is problematic if the rock masses are weathered because obtaining proper core segments is difficult. Hence, using index testing as an alternative for UCS prediction is investigated by researchers. It is well established that Indirect/Brazilian Tensile Strength is related to UCS. In this paper, to develop a correlation between UCS and BTS, collected data of laboratory tests on dry limestone specimen including 20 Unconfined Compression Tests and 20 Brazilian Tests have been used. Then, to apply Artificial Neural Networks, a Radial Basis Network is developed to reach a relationship between BTS and UCS. Based on the low Mean Squared Error of the network, a new correlation is introduced for prediction of the UCS of limestone core samples from BTS data.

**1 Introduction**

Unconfined Compressive Strength (UCS) of rock is considered as an essential parameter in analysis of geotechnical problems such as rock blasting and tunneling. Although laboratory test is the most reliable and direct method for estimating UCS, direct determination of UCS in laboratory is time-consuming and expensive. In addition, in direct method of UCS determination, having sufficient number of high quality rock samples is a prerequisite. However, it is not always possible to extract proper cores for sampling purpose in highly weathered rocks. Therefore, the use of various correlations for UCS prediction has been highlighted in the literatures. These correlations often relate other rock index parameters such as point load index, rebound number of Schmidt hammer and indirect tensile strength of the rock to UCS. Implementing such correlations is of interest, mainly due to the fact that rock index tests have the advantages of being relatively fast and economical. Brazilian Test (BT) is used for indirect determination of tensile strength of rock samples. It is established that Brazilian tensile strength is related to UCS. One of the most agreed correlations between UCS and indirect tensile strength or Brazilian Tensile Strength (BTS) of the rock is highlighted in the study by Sheorey (1997). According to his study, the compressive strength of the rock is approximately 10 times its tensile strength. Nevertheless, Sheorey`s strength ratio variation is high (Cai 2006) and consequently cannot be generalized due to the fact that rock behavior varies from place to place and is site specific. This paper proposes a new correlation between UCS and BTS of specific type of rock i.e. limestone as the relationship between compressive and tensile strength of rock depends on rock type (Brook 1993).

Artificial Intelligence, artificial neural network, Brazilian tensile strength, compression test, compressive strength, correlation, determination, estimation, indirect tensile strength data, laboratory test, Limestone Rock, machine learning, neural network, neuron, prediction, Reservoir Characterization, rock sample, strength, strength test, tensile strength, uniaxial compressive strength, Upstream Oil & Gas

Country:

- Europe (0.47)
- Asia > Middle East (0.47)

SPE Disciplines:

Technology: Information Technology > Artificial Intelligence > Machine Learning > Neural Networks (1.00)

Hisatake, M. (Kinki University, Osaka) | Cording, E.J. (University of Illinois, Ill) | Ito, T. (Osaka Institute of Technology, Osaka) | Sakurai, S. (Kobe University, Kobe) | Phien-Weja, N. (Asian Institute of Technology, Bangkok)

Stresses and displacements in a rock surrounding a tunnel are fundamentally important in planning the tunnel and support systems, and which depend on stress-strain relationships, ground failure criteria, initial stresses of the ground and executive conditions. There are many literatures for calculation of ground response, most of which give closed form solutions to problems with hydrostatic initial stresses and circular geometry but some use numerical approaches such as a finite element method (FEM), a boundary element method (BEM) and a coupling method of FEM and REM, to solve problems involving more complex two or three dimensional tunnel geometry and stress fields[l-5]. Generally, stress-strain relationships for a rock show non-linearity, and mechanical parameters involved in the stress-strain relationships are affected by confining pressure. Also, under a high initial stress state, the strength of the rock surrounding the tunnel decreases from its peak strength to residual one and volumetric change in residual strength region gives big influence on tunnel displacements and support system. Experimental results show that the rock failure criterion can be expressed by stresses but there is a unique relation on failure strains. Yoshinaka and Yamabe[6] showed that the failure of rocks occurs nearly at the same strain regardless confining pressure. The failure condition of a rock surrounding a tunnel should satisfy the both experimental results on failure stresses and failure strains. If a linear stress-strain relationship in the peak strength region is used in the tunnel analysis, the residual strength region is determined only by using stress conditions at the boundary between the peak and the residual regions. In other words, the failure strains at this boundary is automatically determined from the stress conditions, and which do not coincide with experimental failure strains. Namely in this linear analysis, there is no way to satisfy the both conditions on failure stresses and failure strains. In order to make tunnel movements clear by taking into account the realistically mechanical behavior mentioned above and to contribute to practical tunnel engineering, this paper presents solutions to a simple axisymmetric tunnel problem. The influence of the non-linear stress-strain relationships, non-linear criteria, brittle stress reduction, internal pressure on tunnel movements is investigated.

2.1 Stress-Strain Behavior

application, Artificial Intelligence, boundary, criteria, displacement, equation, failure strain, internal pressure, peak strength region, Reservoir Characterization, reservoir geomechanics, residual region, residual strength region, strength, Strength Criterion, Strength Reduction, strength region, tunnel, tunnel movement, Upstream Oil & Gas

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Reservoir geomechanics (0.48)

Kahraman, S. (Nigde University, Mining Engineering Department) | Gunaydin, O. (Nigde University, Geological Engineering Department) | Fener, M. (Nigde University, Geological Engineering Department)

Artificial Intelligence, correlation, dry uc, equation, machine learning, marble, moisture content, MPa, point load strength, Reservoir Characterization, reservoir geomechanics, Rock Mech, sandstone, strength, strength loss, strength value, tensile strength, uniaxial compressive strength, Upstream Oil & Gas, water saturation

SPE Disciplines:

Technology: Information Technology > Artificial Intelligence > Machine Learning > Statistical Learning > Regression (0.34)

Based on the strength distribution of individual links, a chain cable strength model which conforms to an asymptotic extreme value distribution is formulated. The subject of probablistic chain breaking strength is investigated and it is found that the variability of link strength is of crucial importance. Based on the formulated chain strength model, the mooring system reliability analysis is subsequently performed. It is concluded that increased variability or reduced mean of link strength will significantly increase the line failure probability and the likelihood of failure occurring in lower sea states. The study also finds that the strength of K4 chains may not necessary be superior to that of ORQ chains if the strength of K4 links has a larger spread.

Approximately 85\ of the current semisubmersiblesdeployed in the North Sea are equipped with chain mooring systems. A typical chain mooring line consists of a few thousand links and the strength of the chain is only as strong as its weakest link. However; this fact is usually ignored in conventional mooring analysis and design, and the fitness of the mooring system is assessed based on a nominal chain breaking strength specified by manufacturers and classification societies. This alone could significantly undermine the mooring system reliability.

To improve the understanding of the subject and stimulate interest in this area, an investigation of probabilistic chain cable strength and its impact on mooring systems is carried out. The paper focuses on the following main issues:

- Probabilistic prediction of chain strength,
- Parametric effects of links,
- Impact on mooring reliability,
- ORQ vs K4 chains.

A typical mooring chain is over 1000m long and has a few thousand links. Assuming that the strength of individual links is independent and follows theGaussian distribution, the probability density function of link strength can be expressed as:

(available in full paper)

The corresponding cumulative distribution function is given by:

(available in full paper)

where n and 0 are the mean and the standard deviation of the link strength. The probability distribution function of the strength of the weakest link out of n links is given by:

(available in full paper)

and the probability density function follows:

(available in full paper)

where y is the chain strength variable. Though Eq.(3) and Eq.(4) are exact solutions, the above integrations have to be performed numerically, i.e. no closed form solutions are available. However, as n 00, the above distribution converges to the type I asymptotic extreme value distribution. The probability distribution function can be expressed as [1):

(available in full paper)

Artificial Intelligence, BTL, chain strength, coefficient, density function, distribution function, etrength, line failure, link strength, link strength coy, mooring system, orq chain, probabilistic chain cable strength, probability, reliability, strength, strength distribution, subsea system, tension response, Upstream Oil & Gas, variation

Country:

SPE Disciplines: Facilities Design, Construction and Operation > Offshore Facilities and Subsea Systems > Mooring systems (1.00)

Buocz, I. (Budapest University of Technology and Economics) | Rozgonyi-Boissinot, N. (Budapest University of Technology and Economics) | Torok, A. (Budapest University of Technology and Economics)

The shear strength of rocks along discontinuities is one of the key parameters for the determination of rock slope stability, the stability of rocks during underground space development and tunneling. Its value is influenced by numerous factors including surface roughness, which is one of the most widely investigated discontinuity property. The present paper introduces a simple graphical methodology for the classification of the surface roughness of rocks, based on the example of two different rock types, Bátaapáti Granites (Hungary), and Mont Terri Opalinus Claystones (Switzerland). The 3D surface of 24 rock samples was digitized using a photogrammetric surface detection method with the help of the ShapeMetrix3D software. The plane of each rock surface was defined by fitting a linear regression plane to the surface data. The distance between the data points of the surface roughness model and the regression plane was measured, and cumulative frequency diagrams of the measured distance values were constructed. This procedure allowed to define three surface roughness categories. The methodology proposed represents a promising new approach to surface roughness quantification, which could improve shear strength estimation.

Shear strength of rocks is one of the key input parameters for stability analyses of rock masses. The design of appropriate supporting systems (type and strength) used to ensure the ideal degree of safety for people interacting with any engineered rock surface depends on the results of these analyses. However, the value of the shear strength is influenced by several factors, such as the mechanical properties of the intact rock and the discontinuities, as well as the laboratory testing methods and testing machines (Barton, 1973, 2013; Grasselli 2001; Buocz, 2016, 2017a; Dzugala et al., 2017). Therefore, the exact calculation of this parameter is very challenging. Among others, surface roughness is one of the most widely investigated discontinuity property with significant influence on the direct shear strength of rocks. Forty years ago, Barton presented for the first time 10 typical 2D surface roughness profiles, which defined as many value intervals for the Joint Roughness Coefficient (JRC). Once implemented into his rock shear strength model, these values helped to provide an appropriate estimate of the shear strength (Barton and Choubey, 1977). With the fast development of technology and the increasing precision of methodologies for surface detection, i.e., laser scanning or photogrammetry (Gaich et al. 2006), the 3D analysis of surface roughness gained again a central attention (Ge et al. 2015). Different theories were elaborated for the quantification of surface roughness and the determination of values for JRC in three dimensions (Zhao, 1997; Grasselli 2001, Bae et al., 2011). However, due to its simplicity, the well-accepted 2D surface roughness determination still remains in use in practice.

Artificial Intelligence, bátaapáti granite, direct shear strength test, moderately rough surface, mont terri opalinus claystone, peak shear strength value, regression plane, Reservoir Characterization, rough surface, sample surface, shear strength, shear strength value, strength, structural geology, surface roughness, Upstream Oil & Gas

Country:

- Europe > Hungary (0.37)
- Europe > Switzerland (0.36)

SPE Disciplines: Reservoir Description and Dynamics > Reservoir Characterization > Exploration, development, structural geology (0.68)

Thank you!