FIBER OPTICS MONITORING SOLUTION FOR CANAL DYKES

Transcription

1 FIBER OPTICS MONITORING SOLUTION FOR CANAL DYKES by O. Artières 1, M. Galiana 2, P. Royet 3, Y.-L. Beck 4, P. Cunat 5, J.-R. Courivaud 6, J.-J. Fry 7, Y.H. Faure 8,C. Guidoux 9 ABSTRACT A fibre optics textile composite monitoring solution has been developed to detect and localize both leaks and early signs of failure (erosion, blocking, breaches, sliding, and settlements) of hydraulic works, and specially wet dikes of canals and harbours. This unique solution can detect at the same time changes in temperature and strain. It has been installed in several experimental and real sites. The performance of the GeoDetect solution results in the combination of the high sensitiveness of the textile composite sensor connected to the relevant instrumentation to assess the soil properties changes, and the powerful mathematical model to analyse the measured data. Variations as small as 0.02% in soil strain and leaks as low as 0.1 l/min/m have been measured to detect and localise the early signs of failure. 1. INTRODUCTION Owners of waterways are managing a wide number of works with a large variety of ages and stages. The problem facing the canal owner today is to limit the impact of this aging phenomenon. The most serious task is the safety by eliminating the risk of failure. In the last decade a large number of dam and dike failures occurred. Management of the water resource is the second important topic: leakage control becomes more and more important. For this purpose, a technical solution that detects and localizes both malfunctions, precursors of failure (erosion, blocking, breaches, sliding, settlements), and leaks was developed. It is a reliable early localization and warning system for both gradual and catastrophic dam and dike failure. 2. GEODETECT : THE GEOTEXTILE FIBER OPTICS MONITORING SOLUTION This development was handled through a partnership of companies and public research institutes within the Eureka labialized project SafeDike. The use of fibre optics in structural health monitoring systems for civil engineering applications have been widely used for many years. By integrating fibre optic sensing into a geotextile fabric as shown in Fig. 1, GeoDetect is the first system designed specifically for geotechnical applications. It embodies a geocomposite fabric, fibre optics and instrumentation to provide a clearly innovative solution for the multi-functional requirements of a geotechnical application e.g. in-plane drainage capability, anchoring interface with the soil, protection of the fibre, reinforcement and data capture. It uses stimulated Brillouin or Raman scattering technology in single mode or multi-mode fibres to measure strain and/or temperature. A specific analysis of the raw data measured by the monitoring system often increases the accuracy and the speed of the early detection (Beck et al, 2009; Cunat et al., 2009a; Cunat et al., 2009b). GeoDetect is designed to detect the first steps of internal erosion processes and hydraulic works instability. The detection of the leaks, which is the early stage of the internal erosion process, is 1 Global technology manager GeoDetect solution, TenCate Geosynthetics, France, [email protected] 2 Engineer, Cetmef, France, [email protected] 3 Senior expert, Cemagref, France, [email protected] 4 Project Manager, EDF-DTG, France, [email protected] 5 PhD student, LTHE/University Joseph Fourier Grenoble, France, [email protected] 6 Engineer Specialist, EDF-CIH, France, [email protected] 7 Senior expert, EDF-CIH, France, [email protected] 8 Professor, LTHE/University Joseph Fourier Grenoble, France, [email protected] 9 Fiber optics development manager, GeoPhyconsult, France, [email protected] 1 of 11

2 assessed through the measurement of temperature changes using the passive method or the active heat pulse method (Radzicki et al., 2009). First stages of dike settlement or sliding are detected by strain measurement. In comparison to existing detection systems, the GeoDetect solution is a distributed and continuous measurement along the whole canal length, which increases the accuracy and the speed of response, both crucial parameters to prevent collapse. It can provide a leak and deformation location with a spatial resolution of 1 meter, or even 0.5 m in some cases. The system is able to monitor several tenths of kilometres. The GeoDetect solution has already been validated on several 1:1 scale experimental works and real dikes in-use. Different monitoring strategies may be designed, for example temporary monitoring, or continuous monitoring to be used as an early warning system. Figure 1: View of the GeoDetect S-BR sensor with the embedded coloured optical cables 3. LEAK DETECTION THROUGH SOIL DYKES 3.1 The PEERINE experimental basin in Aix-en-Provence (France) To test and validate the GeoDetect solution, as well as other equipment and methods dedicated to detecting and quantifying leaks from a real structure, an experimental basin was built during the second quarter of 2006 on the Cemagref site in Aix-en-Provence, France. 2 of 11

3 Figure 2: View of experimental PEERINE basin in Aix-en-Provence The main characteristics of the experimental basin (Fig. 2) are as follows: Volume of materials: 1200 m 3 of clayey materials (permeability at saturation of m/s). Perimeter: 118 m at the foot, 78 m at the head. Coating: upstream geomembrane lining system Approximate water volume: 200 m 3 The GeoDetect system was laid at three levels (referred to as T2, M2 and B2, respectively for top, middle and bottom) along the downstream embankment below a gravel abutment layer (Fig. 3). There are artificial leaks situated at two different heights. These local soil heterogeneities are made of coarse gravel. This provides a maximum flow rate of approximately 10 l/min. The main result from the experimental program is today the possibility to detect leaks with the order of magnitude of 1 l/min./m from an analysis of the residuals and the sum square error (SSE) (Fig. 4). Further results are given in Artières et al. (2007). GeoDetect PANEL Artificial leaks Figure 3: Section of one dyke of the basin PEERINE basin with the artificial leaks Leaks Figure 4: Example of the results from the EDF/Cemagref temperature analysis model. M2 is the strip at the middle of the slope, B2 is the strip at the bottom of the slope. 3 of 11

4 3.2 The Ijkdijk / piping experimental project : detection of internal erosion Four experimental dyke were built within the Dutch IJkDijk project (Smart Calibration Dyke), Piping part, to test the ability of sensoring systems to detect the first signs of internal erosion. This project funded by the Dutch government and a consortium of several companies and institutes. The IJkdijk piping project was carried out the second half of the year 2009 in North part of The Netherlands (IJkdijk, 2009). To investigate "piping", a test dike was built specially, containing advanced monitoring equipment. The phenomenon was initiated in a controlled way, resulting in the collapse of the test dike. The experiment and the measurements show that piping is a failure mechanism that needs to be taken seriously. The experiment took place in a section 4 m deep, 40 m long and 25 m wide. The subsurface in the section consisted of sand. Over the width of the section, there was a dike with high water on one side (more than 2.5 m) and low water on the other (0.1 m). The experimental dykes are about 15 m long and 4 m high, as shown on Fig. 5. The experiment was unique: this was the first time that the mechanism had been simulated and observed on this scale in a controlled setting. For the first time, it was also possible to demonstrate that there is a link between piping and the actual failure of a dike. The water head was progressively increased up to 2.6 m high up to the dyke collapses (Fig. 6). On the test number 4, six GeoDetect S-BR strips were installed lengthwise the dyke closed to the interface between the sand and the dike, at a depth of 10 cm into the sand. Despite this depth from the interface were the flow channel occurs, the change of the temperature profile indicate the start of the piping channel (Fig. 7). Without data interpretation, the piping channel was detected 17 hours before collapse. With deeper data analysis carried out by EDF has shown an earlier detection threshold at least one or 2 days before the dyke collapses. Figure 5: The IJkdijk piping dyke during the test 4 of 11

5 Figure 6: The IJkdijk Piping test after collapse due to internal erosion at the bottom interface Figure 7: The GeoDetect temperature profiles analysed with the EDF/Geophyconsult model at different time, resp. 17 hours, 11 hours, 5 hours before failure and at the failure, from the top left to the bottom right. 5 of 11

6 4. LEAK DETECTION THROUGH THIN LINING SYSTEM An experimental study was carried out in the PEERINE basin at the Cemagref in Aix-en-Provence, as previously described, to test the GeoDetect performance to detect leaks through thin lining system, such as geomembrane, applied on the upstream face of the dike. Two configurations were tested, as described in Fig. 8: one slope covered with a granular drainage layer, one slope with a geocomposite drainage layer (geonet). These configurations represent very common design of canal lining. In this case, the purpose of GeoDetect is the assessment of the amount of leakage through the liner in the perspective of a better control of the water resource. Artificial leaks were placed at different location through the geomembrane liner, as indicated in Fig. 9, able to control leaks as low as 0.2 l/min. These leaks were seen with the GeoDetect system (Fig. 10). Concrete slabs (0,5 x 0,5 x 0,05) 2,5 m Geomembrane Artificial leaks Protection geotextile 30 cm 2 1 Drainage pipe 100 mm Location of the GeoDetect strips Granular drainage layer Filtration geotextile Protection geotextile Concrete slabs (0,5 x 0,5 x 0,05) Geomembrane Protection geotextile 2,5 m Top soil, 0,5 m 2 GWB 1 GeoDetect strips Filtration geotextile Drainage geocomposite Drainage pipe 100 mm Figure 8: Two configuration on the use of GeoDetect to detect leaks through a geomembrane lining system, with a granular drainage layer (top) and a geocomposite drainage layer (bottom). 6 of 11

7 TOP Vue VIEW en Plan E 36 m Location of the leaks through Position des fuites à travers la membrane the geomembrane N 34,5 m 2 % (Sol naturel) 27 m 24 m A 14 m GEB GEM1 GEM2 14 m GEH 11 m 21,5 m 23 m 1 m Vidange de fond GWB GWH GWM2 GWM1 Clôture Caniveau périphérique W O Figure 9: Location of the artificial leaks through the geomembrane of the PEERINE basin (green bars) Figure 10: Detection of the artificial leak (black arrows) with the GeoDetect system for the strip installed at the toe of the slope (blue line) and at the middle of the slope (green line), data analysed by EDF and Cemagref. 7 of 11

8 5. SOIL STRAIN DETECTION A first experimental dyke was built within the Dutch IJkDijk/Macrostability project in 2008 to test the ability of sensoring systems to detect the failure due to internal instability. The 100 m long and 6 m high dyke was made of an internal sandy core with a 1.5 to 2.4 m thick external clay revetment, common structure of the dikes along the Dutch canals. The GeoDetect solution was installed inside the dyke, under the revetment, to measure the strain of the embankment at 4 different locations (Fig 11). During the test, the global factor of safety of the dyke was decreased step by step by digging a trench at the downstream toe and by increasing the internal water pressure till failure by slippage of an area of the downstream face (Figure 12). The FOS based monitoring system worked perfectly as it was able to detect and to localize the instable zone inside the dyke body before the failure occurred (Artières, Koelewijn et al, 2009). Soil strains as small as 0.02 % were measured indicating the first steps of the failure more than two days before the dyke collapses (Fig. 13). Figure 11: Installation of the GeoDetect S-BR strip at the top of the IJkdijk/Macrostability dyke. Figure 12: The experimental IJkdijk/Macrostability dyke before failure on the left and three different views during collapse. 8 of 11

9 Figure 13: The experimental dyke after failure and the corresponding measurement of the strain several hours before failure. 6. VALIDATION ON REAL CANAL SITES Several real sites are now equipped with the GeoDetect monitoring solution. On earthworks and geotechnical applications, reinforced vertical walls and embankments over cavities are monitored for more than 5 years to measure any change of soil strain. In canals, a test section of about 100 m long and 3 m high along the canal from the Marne to the Rhine owned by the Voies Navigables de France (VNF) in the eastern part of France was monitored since 2008 (Fig. 14). Another section is monitored along the Rhine canal near Kembs (F) : this dyke owned by EDF is about 3 m high is built with permeable aggregates. The measurements are on-going on all of these real sites. 9 of 11

10 Figure 14: The four GeoDetect strips installed on a section of the Marne to Rhine canal. 7. CONCLUSION The GeoDetect solution is an innovation that combines the benefits of geosynthetics materials with the latest sensing and measurement technologies This customizable solution provides objective, highly precise, and timely in-situ performance information, allowing the designers and owners to understand system performance in addition to providing alerts for negative geo-events (subsidence) and other potentially deleterious events. The GeoDetect monitoring solution has already proved on experimental dikes the early detection and localisation of both leakage and settlement. 8. REFERENCES Artières O., Bonelli S., Fabre J.P., Guidoux C., Radzicki K., Royet P., Vedrenne Ch. (2007). Active and passive defences against internal erosion. 7 th ICOLD European Club Dam Symposium, Fresing, Germany September Artières O., A.R. Koelewijn A.R.,Fry J.J., Royet P (2009). Early detection and localization of a failure zone in a dyke with a fiber optics monitoring system. Presented at the ICOLD Conference, Brasilia, Brazil May Beck Y.L., Cunat P., Johanson S., Dornstädter J., Aufleger M. and Goltz M. (2009), Use of fibre optics in leakage detection: a review, European Working Group in Internal Erosion, St Petersburg, Russia. Cunat P., Beck Y.L., Fry J.J., Courivaud J.R., Fabre J.P., Faure Y.H. and Radzicki K. (2009a), Leakage detection based on temperature measurement with fibre optic : methods and results, HYDRO2009, Lyon, France. Cunat P., Beck Y.L., Fry J.J., Courivaud J.R., Fabre J.P. and Faure Y.H. (2009b), Surveillance of dyke ageing, 2nd International conference on Long Term Behaviour of Dams, Graz, Austria. IJkdijk, (2009). 4 pages. Available on Radzicki K., Bonelli S., Beck Y.L. and Cunat P. (2009), Leakage and erosion processes identification by temperature measurements, in upstream part of earth hydraulic works using the impulse response function analysis method, European Working Group in Internal Erosion, St Petersburg, Russia. 10 of 11

11 9. ADRESSES OF AUTHORS Olivier Artières, global technology manager GeoDetect solution TenCate Geosynthetics, 9 rue Marcel Paul, Bezons Cedex, France Tel / Fax [email protected] Mathieu Galiana, Engineer Centre d'etudes Techniques Maritimes et Fluviales 2, Boulevard Gambetta BP Compiègne Cedex France Tel : Fax : [email protected] Paul Royet, Senior Engineer Cemagref, CS 40061, Aix-en-Provence Cedex 5, France Tel Fax [email protected] Jean-Jacques Fry, Senior expert EDF / CIH - Savoie Technolac F Le Bourget du Lac Cedex France Tel : Fax : [email protected] Jean-Robert Courivaud, Engineer specialist EDF / CIH - Savoie Technolac F Le Bourget du Lac Cedex France Tel : Fax : [email protected] Yves-Laurent BECK, Project manager EDF / DTG - 21 avenue de l'europe - BP41 F Grenoble Tel : [email protected] Pierre Cunat, PhD student UJF / LTHE - B.P. 53 F Grenoble Cedex 9 France Tel : [email protected] Yves-Henri Faure, Professor UJF / LTHE - B.P. 53 F Grenoble Cedex 9 France Tel : [email protected] Cyril Guidoux, Fiber optics development manager, GeoPhyconsult GeophyConsult - 12, allée du Lac de Garde - BP F Le Bourget-du-Lac cedex Tél. : +33 (0) [email protected] 11 of 11