Vous êtes sur la page 1sur 6

A risk index approach for risk identification in hard rock tunnelling

D. Stematiu Technical University of Civil Engineering Bucharest I. Zamfir ISPCF Consulting Company R. Cojoc ISPH Consulting company, Romania

ABSTRACT: An equivalent numerical scale for the tunnelling risk quantitative values is defined starting from the risk factors. The risk factors include the accuracy of the geological investigations, the design capabilities to handle the unexpected difficult situations and the construction aspects as contractor experience, site engineering etc. Each factor or main criterion is quantified by a numerical index evaluated on the basis of number of points attributed to some partial criteria into which the main criterion is subdivided. The combination of the numerical indices characterizing the main criteria allows for the definition of a global index named tunnelling risk. The procedure capabilities are validated on the basis of two selected case studies characterized by significant non-homogeneity of the s ite condition. risk is adopted: "a divergence from what is predicted or expected". In our case the predicted/expected target is tunnel performance in compliance with the initial contractual requirements, particularly with respect to deadlines, level of performance and overall costs. The risks of non-compliance with the expected target are associated with unforeseen natural conditions, inadequate design or quality of execution. The divergence from what was expected could result in loss of life, overrunning of the construction time, increase in the work quantities, immobilization of equipment and personnel during the holdup caused by the event, total cost increase etc. Quantitative evaluation of all the consequences is imposs ible. Taking into account the above considerations, the term risk is attributed to the probability of occu rrence of an adverse event that in its turn leads to a serious divergence from main tunnelling targets. The numerical evaluation of such a probability is still a very difficult task requiring probabilistic definition of many parameters, most of them unknown. Consequently a different approach is proposed. The probability of occurrence of an adverse event is expressed by a numerical risk index (Tunnel Risk Index - TRI). The risk index TRI is evaluated in terms of three numerical indices characterizing the main risk factors: geological conditions (G), tunnel design (D) and tunnel construction (C). The numerical values of G, D and C indices are obtained by a simple addition of the number of points attributed to each subcriterion into which the main criterion (factor) is subdivided. Each of the three numerical indices has the maximum value (100 points) for the best tunnel-

1 INTRODUCTION One of the main concerns in tunnelling is to guarantee that the design, construction and commissioning of a tunnel comply with a determined schedule and cost and will attain the expected performance and quality levels. The assessment of the risks associated with execution of the project is of paramount importance for the owner, investors and their bankers. The term risk in its generally accepted formulation is the product of the probability of an event occurring (technical accidents, failure, and project abandonment) and the consequences of the event, should it occur. The paper deals with the procedure developed in order to quantify the occurrence of severe accidents in tunnelling. The proposed procedure is based on risk factors and a system of numerical indices that characterize each of the factors. The procedure capabilities are validated on the basis of two selected case studies in Romania. 2. THE PROCEDURE FOR RISK EVALUATION 2.1. Tunnelling risk index The application of probabilistic risk analysis in its full concept in order to assess risk numerical values is a very difficult task (if not impossible) for a tunnelling process where many unknown parameters are involved. Consequently, a simpler definition of

ling conditions regarding geological investigations, design and construction and is gradually diminished when the requirements are not met. In order to express a probability, the TRI index is defined as:
TRI = 1 1 1 C

2.2.1. Geological and geotechnical investigations (G - index). Various geological and geotechnical parameters are required for a proper design and construction work, essentially regarding the stability conditions . Amongst them one can recall mechanical behavior, discontinuities, alterability, swelling, underground hydrology, natural stresses. The important role of accurate engineering geological data in the design and construction of tunnels requires little demonstration. The first stage in geological characterization of the rock mass along the future tunnel is the use of ordinary geological maps. Existing geological and geotechnical data from maps, documents, aerial photographs and so on are assembled into maps and sections to show the expected tunnelling condition. The next stage is based on ground exploration. Boreholes are the main source of quantitative information available for tunnel design purposes since

G D

Considering the whole range of values for the numerical indices G, D and C, the tunnel risk index may take values in the range of 10-6 to 9 x 10-4. Having in mind that for large civil projects the probability of failure have an average value of 10 -6...10-5, the TRI index may be used to appreciate the acceptability of risk involved and if it is the case to pe rform an adequate risk management. 2.2. Risk factors

Table 2.1Geological and geotechnical investigations criterion (G - index)


Subcrit eria Design stage Geological maps; aerial photographs; global geological sections 5 Grounds explorations by boreholes; geophysical surveys 15 In situ tests, laboratory tests; direct measurements 30 Direct observations in pilot excavation; survey of joints and ground water condition 40 Construction stage Description of rock masses 10 Rock mass class ification systems 20 Systematic monitoring 40 Updating the rock parameters as constru ction progresses. Observations and tests. 60

Points

Points

Points

Points

they provide measurements of the thickness and character of stratigraphic units, make it possible to observe the water conditions and give an opportunity to carry out in situ tests and measurements. The boreholes also provide a source of material for laboratory tests. During this stage, surface geological, geotechnical and geophysical surveys might also be carried out. The third stage of geological investigations takes place during the construction of a pilot tunnel. Direct observations of the tunnelling medium are possible. Tests performed on the material, surveys of joints, measurements of the ground water conditions all contribute to the completion of geotechnical data.

The requirements concerning the geological data are different at the design stage and during the effective tunnelling. Consequently, the criterion "geological investigations" was subdivided into two subcriteria, namely geological investigations for design stage and geological investigation for construction stage. The number of points attributed to each subcriterion is shown in table 2.1. 2.2.2. Tunnel design (D - index) Whether to support a tunnel and if so, how to proportion the support was always amongst tunnelling's most controversial and difficult problems. The new state of equilibrium can be endangered by several causes:

- the removal of any natural support to the ground around the excavation by the advancement of the excavation face of the tunnel; - the time dependent behavior of the ground around the tunnel; - the time dependent reaction to changes in the ground water regime if the tunnel acts as a drain. Consequently, the tunnel designer can solve the problem using different approaches from baldly empirical rules to the most sophisticated analytical
Subcrit eria Preliminary design Based on rough em pirical rules 5 Based on a support s election system 10 Based on support selection system and numerical analysis

techniques. In any case, a correct final decisions has to be made by the exercise of experience and engineering judgement, aided by the conclusions drawn from the computational studies. The best approach requires a close monitoring of the tunnelling process (excavation, support installation, ground reactions) that allows continuos design improvement. The closer the designer can get physically to the ground, to observe it, prod and work it, the better the decisions are likely to be.

Table 2.2 Tunnel design (D - criterion)


Final decision Build the tunnel before it could be designed 0 Engineering judg ement 10 Experience, engineering judgement, performance of existing tunnels, cost and effect on site organization 30 Design improvement during tunnelling None 0 Adaptive to the new ground data 20 Based on continuos data acquisition concerning ground characteristics 40 Based on relevant data revealed by geological investigations and monitoring 50 -

Points

Points

Points 20

Points

2.2.3. Tunnel construction (C - index) It is well known that success of a tunnelling project depends essentially on the contractor experience and the personnel qualification. Construction aspects are and must be among tunneller's earliest consideration. The tunnelling procedure may have significant impact on support system selection. The contractor and the site engineer have to know the questionable effect of rock bolts in severely fractured rocks, the difficulties of shotcreting on irregular surfaces, the time consuming in wedging and packing of steel ribs. Personnel training is of primary importance. As an example, in shotcreting the qualification of operators in charge is essential for obtaining satisfactory results. Even in the case of qualified contractors experience shows that as the job is done negligence by the personnel may cause sever incidents. Type of contract has its implications in the risk level. During the job completion the contractor have to provide a proper answer to the question if the risks are worth the benefits. According to a proper contract the contractor - owner relationship has to be devoted to job completion rather than disagreements on claims.

Based on the above considerations, the main criterion "design" was divided into three subcriteria, with the corresponding number of points as shown in table 2.2 Site engineering if properly done has at least two beneficial results. Primary the tunnel is built in compliance with contractual requirements with respect to overall costs. Secondly the quality of work is observed by checks and supervision of project implementation and by prompt decisions concerning the adaptive design. Finally, the financial support of tunnelling operation is also a key factor concerning the risk. Based on the above considerations, the main criterion "tunnel construction" was divided into four subcriteria that are shown in table 2.3 together with the corresponding number of points. 3. CASE STUDIES In order to verify the capabilities of the proposed procedure to evaluate the risk of occurrence of severe accidents that lead to overrunning of the con-

struction time and total cost increase, two case

studies are briefly presented in the followings.

Table 2.3 Tunnel construction (C - criterion)


Subcrit eria Contractor experience and qualific ation U n e x p e r i e n c e d c ontractor; lack of qualified personnel 5 P r o f e s s i o n a l n e g ligence Type of contract Site Enginee ring Adaptive technical specifications 10 Provides early recognition of ground condition; substantial changes to the design solutions 20 Permanent monitoring of the quality of work with continuos adjustment of execution technology 30 Financial Aspects Temporary suspension of work due to lack of funds 0 Overdue payments

Points

Turnkey contract for a lump-sum price 5 Contract prices on the basis of quantity measurements 10

Points 10 Personnel training; Qualification is checked during the job 30

5 Regular payments and covers of additional costs 15

Points

Target cost contract with checks and supervision of project implementation 15

Points

Experienced contra ctor; know -how Combined wit h proper eme rgency Training 40

3.1. Mestecanis railroad tunnel In order to double a very crowded railroad line in northern part of Romania a new tunnel was required. The new tunnel with a length of 1670 m was performed parallel to the existing one, actually a very old tunnel built in 1912 with tunnelling methods of that time. The interax distance between the new and the old tunnel was 28 to 35 m. Some of the geological data available from the old tunnel excavation were extended to the new site. Supplementary investigations based on boreholes have revealed some new data concerning the stratigraphic units. The rook mass consists mainly of crystalline shiest that were divided into three main categories. Due to tectonic movements the rock mass was affected by several faults. Two systems of fissures were associated to the faults. Significant water circulation with high rates of water flow were related to existing discontinuities. High rock pressures and rock swelling potential were predicted. The new tunnel cross section area was some 50 m2. The excavation method was drill-and-blast. In order to overcome the difficulties related to the rock large deformation potential and the adverse effects of wa-

ter presence that could be stimulated by the new drain created by the tunnel excavation, the designer selected the NATM (New Austrian Tunnelling Method) as support system. Shotcrete and rock bolts were specified, but shotcrete was the main component. Being applied to the rock at high pressures it forces its way into rock discontinuities and prevents the water access. It also gains strength rapidly, permitting it to begin functioning quickly as a membrane. The lining was conceived as a conventional freestanding arch. The inner reinforced concrete arch with a thickness of 40 cm had to be cast in place in the final stage. The owner had assigned the responsibility to the contractor for excavating and stabilizing the ground. Unfortunately the contractor did not have experience concerning the NATM and at the request of the site personnel decided to change the excavation and support technology. The Belgian method was selected and the lining structure was adapted to the new support system. The first outer reinforced concrete ring was the primary lining. It was also part of the support. The crown heading was excavated and concreted in the first stage. The next excavation stage was performed under the protection of the concrete vault. The two lining ring concept was preserved.

The inner reinforced concrete lining was performed in the final stage, following the contractual dimensions and details requirements. During the tunnel construction the unprotected rock surface exposed by excavation (up to the outer ring concreting) exhibited a wide range of weathering. The water inflow and the afflux of water have contributed significantly to the difficult situation. Due to very large rock pressures the concrete arch of the primary lining was heavily fissured and the vault springs were displaced towards the rock contour. In some zones the total settlement at the roof has reached 1...1,20 m. The widening of the arch springs combined with a heavy reinforced continuos beam and 4 m long anchors where provided to control the support movements. Even in that case the 6month delay between the wall concreting and the invert installation has lead to lateral inner displacements and even bottom swellings where the water inflow was significant. Along several zones the floor was rebuilt. The whole range of incidents and difficulties, mainly due to the unproper excavation technology has lead to a one year and six month delay in tunnel commissioning and a cost increase of some 30%. The associated tunnel risk index was evaluated on the basis of the proposed procedure. As far as the geological investigation criterion is concerned, for the design stage the borehole exploration and the existing data have provided an acceptable level of knowledge. During the construction stage only the traditional geological survey was conducted. As a consequence, one attributes 15+10 = 25 points to G index. The original tunnel design was based on a support selection system and very simple numerical analysis (freestanding arch). The final decision concerning the support and the lining was based on engineering judgement. During the tunnel execution the design was adapted to the new conditions imposed by the excavation and support technology but is hard to say it was a design improvement. As a consequence, the D criterion may be quantified as 15+10+0 = 25 points. The tunnel construction was entrusted to a contractor without experience in NATM application. The payments were regular and based on quantity measurements covering the whole range of extra work. The owner provided the site engineering but the personnel qualification was low. The recognition of difficult ground conditions and the changes of the design solutions were much behind the actual needs on site. As a consequence, the

C index corresponding to tunnel construction has 5+10+5+15 = 45 points. The tunn el risk index results with the value of TRI = 3,5 x 10-5 and can be rated as unacceptable. The change of the support system without realizing the consequences imposed by the natural rock conditions was the main source of risk. The second significant contribution was the poor site engineering that from the first place should provide the required modifications in the excavation and support technology. 3.2. Runcu-Firiza headrace The Runcu-Firiza tunnel is the headrace of the hydropower development Firiza, under construction with an installed power of 16,4 MW and an average energy output of 56 GWh/year. The tunnel length is 2,78 km, the excavation diameter is 3.20 m and after the tunnel lining the inner diameter will be 2-80 m. At the design stage, the geological data based on geological maps, direct observation on outcrops and several boreholes have shown that the tunnel will be excavated in andesites with amphiboles and pyroxenes. Frequent pegmatite injections were encountered. Weathering phenomena of hydrothermal origin were expected. The tectonic movements have induced numerous discontinuities in the rock mass, faults and fissure systems being encountered during the excavation. Along the discontinuities the water circulation created significant inflows. The support design was based on a support selection system. Two characteristic support types were included in the design, one corresponding to rock categories I and II and the other to rock categories III and IV. The first support arrangement consists of a 6 cm thick shotcrete layer and rock bolts with a length of 1,50 m, one for each square meter. The second support arrangement consists of a 10 cm shotcrete layer, 2 m rock bolts and steel ribs at 1 m intervals. It was predicted that the second support system will be required locally, where rock condition is severely affected by the main discontinuities or weathering. The tunnel construction was entrusted to an experienced contractor and the site engineering was provided by the geologists from the design company. The tunnel excavation started in 1990 from the both ends and the break through was ready in 1993. Due to lack of funds the tunnel work was temporarily suspended. The lining concreting started again in 1996, after three years of rock exposure.

During the tunnel excavation the rock condition was very good except several zones around the faults or affected by severe weathering. The detailed geological survey of the tunnel has classified the rock into category I for 86% of the tunnel length, and into categories III and IV for 9% of the length. In order to increase the advance rate, the contractor, encouraged by the rock quality, did not observed the technical specifications and the support was not systematically installed, especially the shotcrete. The support omission was done even in zones where the site geologist pointed out the rock category III or IV. The rate of rock weathering was much faster than expected, being associated to weathering phenomena of hydrothermal origin. The exposure time of the unprotected rock was also much longer than scheduled. Consequently the new geological survey of the tunnel after three years since the tunnel excavation was finished has shown that 80% of the tunnel length was now located in rocks rated as belonging to the category III or IV. When the lining concreting was started caving closed a whole zone of the tunnel with a length of 890 m. The tunnel zone had to be re-excavated using techniques adequate for soft ground: forepoles, steel ribs and precast concrete lagging. Local reinforcement was required in some other zones. The technical incidents and accidents induced by the long rock exposure and by the rapid rate of rock weathering have lead to a total cost overrun of 70% and at least one-year delay in tunnel commissioning. The tunnel risk index was evaluated for Runcu Firiza tunnel on the basis of the proposed procedure. The geological investigations and the geological data were based on geological maps, direct observations on outcrops and several boreholes. During the construction stage a system of rock mass description was used in order to select the support system. As a consequence, the G index, corresponding to geological investigations criterion has 15+10 = 25 points. The preliminary design was based on a support selection system and the final decision concerning the support arrangements and the permanent lining was substantiated by engineering judgement. During the tunnelling process the design had allow the selection of the support based on new ground data. Finally a new design was required to re-excavate the cave in zone. As a consequence, the D index, corresponding to tunnel design criterion has 10+10+20 = 40 points. The contractor was experienced but did not find the proper answer to the question if the ris ks are worth the benefits. Besides, in order to increase the advance rate the contractor has shown professional negligence. The contractual payments were based on quantity measurements. The site engineering was

practically inexistent once the technical specifications were not observed. The temporary suspension of the tunnel works due to owner lack of funds was one of the main causes of the accidents. As a consequence, the C index, corresponding to tunnel construction criterion has 10+10+0+0 = 20 points. Us ing the above indices for the main factors, the tunnel risk index results TRI = 5 x 10-5. The risk value can be rated as unacceptable. The long time exposure of the unprotected rock surface was the main source of risk. The second major contribution was the unexpected rapid rate of rock weathering associated to the original weathering phenomena. 4. CONCLUDING REMARKS In the present work, the term risk is attributed to the probability of occurrence of an adverse event that in its turn leads to a serious divergence from main tunnelling targets - deadlines, level of performance and overall costs. The application of probabilistic risk analysis in its full concept in order to assess risk numerical values is a very difficult task for a tunnelling process. Consequently, a different approach is proposed. The probability of occurrence is replaced by a numerical index that is proportionate to the probability and an equivalent numerical scale is adopted in order to estimate the risk level. The risk index is evaluated on the basis of three criteria and several subcriteria that are quantitatively expressed by a certain number of points. The selection of criteria and especially of subcriteria is still a matter of analysis but the approach is to be judged and not it's actual accuracy. The proposed procedure is only an attempt and is far from being perfect. However, its application to a sample of tunnels for the risk evaluation has lead to encouraging results. The two case studies presented in the paper refer to tunnels where the overrunning of the construction time and the total cost increase are equally due to less expected geological conditions and to contractor inexperience and/or negligence. Financial aspects of the contract management are also implied. The risk indices for both tunnels were rated outside the tolerable range rendering evident the need for a better risk management.

Vous aimerez peut-être aussi