Rock Mass Characterization and Support Design for Underground Additional Surge Pool Cavern — A Case Study , India

For better rock mass characterization and support design, 3D engineering geological mapping was carried for the heading portion of the under construction 200.00 m long, 68.75 m high and 20.20 m wide underground additional surge pool cavern of a Pranahitha-Chevella Sujala Sravanthi lift irrigation scheme package 8, India. To study cavern behavior, 3D geologic mapping of heading portion is very important for large cavern for predicting geologic conditions in benching down up to invert level, planning support system, selecting inclination for best location of supplemental rock bolt and choosing strategic locations for various types of instrumentation. The assessment of Tunnel Quality Index “Q” and Geomechanics classification for the granitic rock mass was done based on the information available of the rock joints and their nature and 3D geological logging. Hoek-Brown parameters were also determined by the statistical analysis of the results of a set of triaxial tests on core samples. On basis of geological characteristics and NMT Q-system chart, support system is recommended which includes rock bolt, steel fibre reinforced shotcrete and grouting. To evaluate the efficacy of the proposed support system, the capacity of support system is determined.


Introduction
Analytical, observational, and empirical are the main design approach for excavations in rock.In this paper, empirical approach for support design of additional surge pool cavern of a Pranahitha-Chevella Sujala Sravanthi lift irrigation scheme package 8 (PCSSLIS-P8) is discussed.Rock mass classifications as practiced in civil and mining engineering form an integral part of the empirical design methods, which is the most predominant design approach [1].The main objectives of the rock mass classifications are to identify the most significant parameters influencing the behavior of a rock mass, divide area into rock mass classes of varying quality and provide quantitative data for engineering design purpose.Rock mass classifications have played an important role in estimating the strength and deformability of rock masses and in assessing the stability of rock slopes.They were also shown to have special uses for serving as an index to rock rippability, dredgeability, excavatability, cuttability, and cavability.For underground excavation, stable empirical approaches are developed based on the evaluation of a large number of case studies.
The major components of the PCSSLIS-P8 are: 4.133 km long and 10.00 m finished diameter "D" shaped twin tunnels, old surge pool (350 m long × 20 m width × 54 m height), 58 m long five numbers of draft tube tunnels, one pump house (215 m long × 25 m width × 54 m height) and five numbers, 50 m long horizontal and 150 m vertical shaft having 5.0 m finished diameter pressure mains, 80 m long delivery cistern and 5.85 km long gravity canal from delivery cistern to join flood flow canal.Lift height is about 126 m and five numbers of pump will be installed in the pump house cavity having 130 MW capacities each.The reengineering of the project was done and because of this additional surge pool is being constructed for increased discharge from 419 to 624 cumecs.Summary of input data of additional surge pool cavern used for support design as provided by sponsoring agency are given in Table 1.Sufficient lateral rock cover is available, and the vertical cover is more than 1D i.e. >70 m above the surge pool.
For the underground cavern rock mass characterization was done based on 3D geologic mapping and laboratory test results.On basis of geological characteristics and NMT Q-system chart, support system is recommended and its efficacy is evaluated.

3D Geological Mapping
3D engineering geological mapping was done in 1:100 scale so that closely spaced geological discontinuities can be mapped (Figure 1).Geologic logging provides a permanent record of all geologic defects exposed on the walls and crown of an underground excavation.Rock type mapped was pink granite belongs to the Peninsular Gneissic Complex of Archaean age [2] [3].Granite was coarse grained, hard and jointed in nature.The granite was generally fresh in nature.It was interpreted that same rock will be present during the benching of additional surge pool up to its invert level.
The details of the joint characteristics are given in Table 2. Joints are generally Figure 1.3D Geological map of the heading portion.

Laboratory Testing
Selected rock core samples were tested for their physico-mechanical properties and test results as provided by MEIL are summarized in Table 3.The compressive strength of core specimens is ranging from 132 to 238 MPa and density varies between 2645 to 2695 kg/m 3 .According to strength classification criterion for rock substance, the rocks are of very high strength [4] and density of material is high.

Tunnelling Quality Index (Q)
The Q-system was developed at NGI between 1971 and 1974 on the basis of approximately 200 case histories of tunnels and caverns [5].They presented a useful correlation between the amount and type of permanent support and the Q with respect to tunnel stability.There has been a significant advance within The Q-value gives a description of the rock mass stability of an underground opening in jointed rock masses.High Q-values indicates good stability and low values means poor stability.The numerical value of the index Q varies on a logarithmic scale from 0.001 to a maximum of 1000 and is defined by six parameters (Equation ( 1)).Q-value 0.001 is generally for exceptionally poor quality squeezing ground, while 1000 is for exceptionally good quality rock which is practically unjointed [5].
where RQD is Rock Quality Designation (degree of jointing), n J is Number of joint sets, r J is Joint roughness number, a J is Joint alteration number, w J is Joint water reduction factor and SRF is Stress Reduction Factor For the heading portion of additional surge pool the individual parameters were determined during geological mapping using tables that give numerical values to be assigned to a described situation.For the calculation of Q-values all the discontinuities per 5 m length and circumference were taken into consideration.An average piece size or block size can be determined using the same data i.e. discontinuities per 5 m length and circumference.The assessment of Q-values for the granitic rock mass, based on the information available of the rock joints and their nature and 3D geological logging, is tabulated in Table 4.
The grade of rock mass based on the rock joints characteristics has the Q-values varying from 4.17 to 16.33, and it comes under fair to good rock mass category.Total 59 percent of area comes under fair rock mass category while 41 percent under good rock mass category.The average Q-value calculated is 9.58.The low Q-values are because of intersection of more joint sets in the excavated span of 5 m and joints surface characteristics.

Geomechanics Classification
The Geomechanics Classification, also known as the Rock Mass Rating system, was developed by Bieniawski during 1972-1973 on the basis of 49 case histories [8].It was modified over the years as more case histories become available and to conform with international standards and procedures [9].In 1984, 62 coal mining case histories were added and a further 78 tunneling and mining case histories collected by 1987.Last time it was modified in 1989 by Bieniawski amounting to 351 case histories.Since then it is being used in tunnels, chambers, mines, slopes and foundations projects.Most of the applications have been in the field of tunneling.This classification is one of the most commonly used rock mass classification system.This is based on the collection of field data and strength parameter.The six parameters which are used to classify a rock mass using RMR system are: uniaxial compressive strength of rock material (UCS), rock quality designation (RQD), spacing of discontinuities (SD), condition of discontinuities (CD), groundwater conditions (GW) and orientation of discontinuities (OD) (Equation ( 2)).

RMR UCS RQD SD CD GW OD
In order to apply the RMR classification, the rock mass has to be divided into a number of structural regions such that certain features are more or less uniform within each region.Rock Mass Rating technique has been found to be quite useful due to the ease with which it can be practiced and its effectiveness in interpreting stability and recommending control measures.The RMR classification parameters are easily obtained either from borehole data or underground/ surface mapping [10] [11] [12].Average stand-up time for an arched roof, cohesion and angle of internal friction, modulus of deformation, allowable bearing pressure, shear strength of rock mass and estimation of support pressure of rock mass may be obtained using RMR.For the heading portion of additional surge pool, RMR values are determined at every 5 m interval (Table 5).The grade of rock mass based on the 3D geological mapping and strength characteristics, has the RMR values varying from 53 to 71, and it comes under fair to good rock category.The average RMR-value calculated is 62.

Hoek-Brown Parameters
In order to use the Hoek-Brown criterion for estimating the strength and deformability of jointed rock masses, the value of the Geological Strength Index (GSI) for the rock mass, the uniaxial compressive strength ( ci σ ) of the intact rock pieces, and the value of Hoek-Brown constant ( i m ) for these intact rock pieces have been estimated.Geological Strength Index (GSI) was introduced by Hoek and Brown (1997) to provide a system for estimating the reduction in the rock mass strength for different geological conditions.The GSI can be related to the rock mass rating (RMR) or the modified rock-mass quality index (Q').Modified rock-mass quality index is defined as (Equation ( 3)): where RQD is the rock quality designation, n J is the joint set number, r J is the joint roughness number, a J is the joint alteration number.
Hoek and Brown [13] suggested that GSI can be related to Q' and RMR by following equations (Equation (4) and Equation ( 5)).Bieniawski's RMR classification should be used for estimating GSI values for better rock masses (GSI > 25) and should not be used for poor quality rock masses.
For the additional surge pool GSI is calculated from Q′ , RMR and Hoek and Brown [13] chart.Hoek and Brown chart is based on geological description of the rock mass i.e. on the basis of interlocking and joint alteration.Minimum, maximum, average and mean values of Q, RMR and GSI are given in Table 6.
The values of ci σ and i m were determined by the statistical analysis of the results of a set of triaxial tests on core samples.After obtaining the test results, they were analysed to determine the uniaxial compressive strength ( ci σ ) of the intact rock pieces, and the value of Hoek-Brown constant ( i m ) as described by Hoek and Brown [14].A spreadsheet for the analysis of triaxial test data is given in Table 7.
For each sample the uniaxial compressive strength ( ci σ ), the constant ( i m ) and coefficient of determination ( 2 r ) are calculated from Equations ( 6)-( 8) re- spectively and values are given in Table 8.The Hoek-Brown parameters that describe the rock mass strength characteristics can be derived from GSI (Equation (9)).
where b m is the value of the Hoek-Brown constant m for the rock mass and mi is the Hoek-Brown constant for the intact rock.
Hoek-Brown constants " s " and " a " are depend upon the rock mass characteristics.For GSI > 25, i.e. rock masses of good to reasonable quality, the original Hoek-Brown criterion is applied with (Equation (10) and Equation ( 11)): and The rock mass strength can be characterized by a GSI value of 55 (fair category), which was used to establish the parameters ( , , b m s a etc.) required for the Hoek-Brown failure criterion.The constants " s " and " a " calculated are 0.0067 and 0.5 respectively.For average/fair category rock masses Hoek and Brown [13] assumed that post failure deformation occurs at a constant stress level, defined by the compressive strength of the broken rock mass.The reduction of the rock mass strength from the in situ to the broken state corresponds to the strain softening behaviour.Martin and Maybee [15] assumed that the failed rock behaves as a cohesionless frictional material.These values can be used for modelling because in the rock masses there are a sufficient number of closely spaced discontinuities with almost similar surface characteristics.

Estimation of Support Pressure and Ground Squeezing Condition
The rock mass quality (Q) is related with the ultimate support pressure requirement.An empirical equation relating rock mass quality Q and permanent support pressure was given by Barton et al. [5] which based on case records (Equation ( 12)).In this equation importance is given to joint roughness number.Better qualities of rock mass have their improved Q values from the dilatent property of interlocked non-planar rock joints, while the poorer qualities are dominated by more or less non-dilatent clay filled joints [5].An improved empirical fit (Equation ( 13)) by incorporating number of joint sets ( n J ) in Equation (12)   is further suggested by Barton et al. [5].When rock mass is intersected by three joint sets ( 9 n J = ) Equation ( 12) and Equation (13) will give an identical esti- mate of roof support pressure.When there are less than three joint sets Equation (13) will give a lower estimate of support pressure than Equation ( 12), and a higher estimate when there are more than three joint sets.When the number of joint sets falls below three, the degree of freedom for block movement is greatly reduced since three joint sets or two plus random is the limiting case for three-dimensional rock blocks.In those equations size of opening does not fig- ure in the support pressure prediction.Singh et al. [16] also studied the effect of tunnel size, span ranging from 2 to 22 m on support pressure and inferred that they are independent.
In this study roof support and wall support pressure was estimated as per Equations (( 14) and ( 15)), which is applicable for the non-squeezing ground condition [16] [17].Grimstad and Barton [6] also agreed on the overburden correction factor from Equation ( 13).
( ) where roof P is permanent/ultimate roof support pressure in kg/cm 2 , Where wall P is ultimate wall support pressure in kg/cm 2 , r J is joint roughness number, Q is rock mass quality, w Q is wall quality/factor equal to 5Q for better quali- ties rock mass ( 10 Q > ) and 2.5Q for intermediate qualities ( 0.1 10 Q < < ), n J is joint set number and f is correction factor for overburden.Correction factor for overburden can be estimated from Equation (16).
where H is the height of overburden above crown in metres Singh et al. [16] suggested an empirical approach (Equation ( 17)) based on case histories and by collecting Barton et al. [5] "Q" data and overburden (H) for the estimation of non-squeezing ground condition.Minimum Q-value is used for the estimation of ground squeezing condition.Above additional surge pool cavern maximum cover is 70 m hence ground condition is non-squeezing.The required support pressure for crown is be varying from 7.89 t/m 2 to 12.43 t/m 2 and for wall 4.61 t/m 2 to 9.16 t/m 2 (Table 9).

Design of Supports
As per hydraulic design, the additional surge pool is having an excavated width   The rock mass quality Q was developed after making a consistent relationship between Q, the excavation dimension, and the support actually used.The permanent support estimate is based on the rock mass quality Q, the support pressure, and the equivalent dimension and purpose of the excavation.The Equivalent Dimension (De) is applied by dividing the span or height (m) by the Excavation Support Ratio (ESR).The ESR for surge pool cavity as given in the ESR updated classification standard of NMT Q-system is applied to 1.0 [19].
Bolt lengths depend on the dimensions of excavations and the length of rock bolts can be estimated from the excavation span (B) or height (H) and the excavation support ratio (ESR) [5] [20].Lengths used in the roof arch are usually related to the span (Equation ( 18)), while lengths used in the walls are usually related to the height of excavations (Equation ( 19)).
where, roof walls L are bolt length in metres for roof and walls, B is span in me- tres, H is excavation height in metres and ESR is the excavation support ra- tio.
By applying the above formula, the length of rock bolt for the crown and walls is calculated to be 5.03 m and 10.78 m respectively.The value of NMT Q-system chart proposed is 5.0 -6.0 m and 11.50 -13.0 m for crown and surge pit walls respectively.
The Norwegian Institute for Rock Blasting Technique has proposed a formula to estimate the length of the bolts in the central section of the opening [18].By applying this, the length of rock bolt for crown of pump house is calculated to be 5.12 m (Equation ( 20)).
1.40 0.184 where B is the span of the opening in metres The thickness of steel fibre reinforced shotcrete can be estimated as per equation (Equation ( 21)) from the ultimate support pressure ( roof P ) and size of opening ( B ) [21] [22] [23].The thickness of SFRS for crown and surge pool walls is calculated from the average Q-value to be 104 mm and 222 mm respectively.The value of NMT Q-system chart proposed is 80 -100 mm and 120 -140 mm for crown and surge pit walls respectively.
where, fsc t is thickness of SFRS lining, roof P is ultimate roof/wall support pressure, B is size of opening, fsc F is mobilization factor for shotcrete (0.6 ± 0.05) and fsc q is shear strength of fibre reinforced shotcrete (550 t/m 2 ) The rock support arrangement includes steel fibre reinforced shotcrete, rock bolt, grouting and drainage holes provisions (Figure 2, Table 10).

Estimation of Support System Capacity
The capacity of support system consisting of SFRS, rock bolt and grouted arch/ rock column for surge pool cavern is determine using the integrated approach given by Singh et al. [21], Singh and Goel [22] and IS: 15026 [23].The total support pressure ( roof wall u p + ) will be equal to the sum of capacities of support system (Equation ( 22)).It is assumed that the fibre reinforced shotcrete is intimately in contact with the rock mass and having the tendency to fail by shearing.Before putting shotcrete, the exposed surface should be properly cleaned and scaled because the strong bond between shotcrete and rock mass is the key to success in stabilizing a cavern The capacity of SFRS as estimated (Equation ( 23)) for roof and walls is 13.61 t/m 2 and 6.27 t/m 2 respectively.
The capacity of rock bolt is estimated (Equation ( 24)) and the minimum capacity for roof and surge pit walls calculated is 1.577 t/m 2 , and 0.349 t/m 2 respectively.Singh et al. [21] proposed mobilization factors after back analysis of Barton et al. [5] support systems case studies.From 120 case histories, Thakur [24] where, l = length of bolt (6 m for roof and 7 m for walls).FAL = fixed anchor length (2.5 m).

Conclusion
3D geologic mapping of heading portion using pilot and side slashing is very important for large cavern for predicting geologic conditions in benching down up to invert level.Geologic logging data were used for rock mass characterization and for support pressure estimation.Logging data were also used in planning tunnel support system and selecting best location and inclination of supplemental rock bolt.Support design empirical approaches are used.Empirical approaches are the best way for support design which is backed by a systematic approach to rock mass classification and providing a quantitative assessment of rock mass conditions.For structural stability, the rock support arrangement includes steel fibre reinforced shotcrete (SFRS), rock bolt, grouting and drainage hole provisions.Geologic logging data will also be very useful for choosing strategic locations for various types of instrumentation to study tunnel behavior.This cavern will be one of the biggest caverns in the world, so it is recommended that the support requirements may be re-evaluated in the light of the rock mass conditions revealed during the benching down of the cavern and the instrumentation data.
continuous and persistent, smooth-planar with unaltered to slightly altered joint walls.Staining has been recorded along the joint surfaces where the joints are tight and where opening is up to 3.0 mm, clay filling has been recorded.In general, the rock mass is characterized by dry condition or minor inflow i.e. <5.0 l/min.

of 20 .
20 m and length 200 m.The bottom level of surge pool is kept at EL 181.50 m and crown level is kept at EL 250.25 m.The maximum upsurge level of surge works out to EL 239.90 m and minimum downsurge level works out to EL 214.80 m.As per design 300 mm thick concrete lined is proposed at the invert level of surge pool.For structural stability of surge pool segment above concrete lined portion, rock support arrangements were recommended based on rock mass quality Q and site geological condition.The objective of reinforcement system was to minimize deformations induced by the dead weight of loosened rock mass, as well as those induced by stress redistribution in the rock surrounding an excavation[18].
On the basis of geological mapping of the heading portion additional rock bolts of 6 m length is recommended at the centre of each grid between Ch 125 m and Ch 180 m (3 m on either side of centre line) and at Ch. 193 m (3 m on either side of centre line).

Figure 2 .
Figure 2. Support system of the surge pool cavern.

BF
reinforced rock mass(18.38 and 41.09 t/m 2 for roof and 10.34 and 23.11 t/m 2 for walls) (Equation25) l′ = thickness of reinforced rock arch/rock column (5.125 m for roof and 4.00 m for walls) (Equations ((26) and (27))) = size of opening (20.20 m-roof; 58.50 m-pump pit wall) s = mobilization factor for rock bolts bolt (1.5 m for roof and 2 m for walls).rockS=average spacing of joints (0.750 m).d = depth of damaged rock due to blasting in walls (av.2.0 m).u = seepage pressure in the rock mass (0.00 t/m 2 ).

aJ
= joint alteration number.
roof p = roof support pressure (varying from 7.89 to 12.43 t/m 2

Table 1 .
Summary of input data.
15Thickness of concrete lined bottom portion 300 mm 16 Rock ledge between old and new surge pool 100 m

Table 2 .
Joint sets recorded in coarse grained pink granite.

Table 3 .
Results of lab tests to rock samples.

Table 4 .
Q-values recorded from the heading portion of the additional surge pool.
RQD = Rock Quality Designation, Jn = Joint Set Number, Jr = Joint Roughness Number, Ja = Joint Alteration Number, Jw = Joint Water Reduction Factor and SRF = Stress Reduction Factor.

Table 6 .
Rock mass classification of granite.

Table 7 .
Spreadsheet for the calculation of σ ci and mi from triaxial test data.

Table 8 .
Rock mass properties for granite.

Table 9 .
Support pressure for the roof and walls.

Table 10 .
Details of rock support arrangement.: Where additional support capacity is required to support local areas of weaker rock, bolts placed at the centre of each grid square will suffice. Note = roof support pressure (varying from 7.89 to 12.43 t/m 2 ).
roof p con- firmed these design criteria.For rock bolt mobilization factors ( s F ) are calcu- lated from Equations 28 and 29 for roof and walls respectively.For roof s F

Table 11 .
Total capacity of support system for roof and walls calculated at different Chainage is given in Table11.Capacity of support system for the roof and walls.