Signal test for acoustic fibre optics for leakage detection of water bottoms

(1)

Signal test for acoustic fibre

optics for leakage detection of

water bottoms

(2)

Signal test for acoustic fibre optics

for leakage detection of water

bottoms

Pauline Kruiver

Edwin Obando Hernandez Manos Pefkos

(3)

elbt r

s

Title

Signal test for acoustic fibre optics for leakage detection of water bottoms

Client

Rijkswaterstaat

Water, Verkeer en Leefomgeving

Project 11203670-000 Attribute 11203670-000-BGS-0001 Pages 54 Keywords

Signal test, Distributed Acoustic Sensing, leakage

Summary

See management summary (English) & managementsamenvatting (Dutch)

References

2019 8003 Waterbodems: Directe meting met de kabel.

Version Date Author Initials Review Initials Approval Initials

0.7 28 Nov 2019 Pauline Kruiver

Wk

Victor Hopman

8

Maaike Blauw

%

Edwin Obando Hernandez Manos Pefkos Chris Bremmer Status final

(4)

11203670-000-BGS-0001, November 28, 2019, final

Signal test for acoustic fibre optics for leakage detection of water bottoms i

1 Management summary

An analysis has been performed in the framework of the KPP-project “2019 BO03 ‘Waterbodem: Directe meting met kabel’” of experiments with a fibre optics cable which have been executed at Deltares in November 2018 and February 2019. The goal of this project is to analyse the results of the two large laboratory tests in order to assess whether distributed fibre optic sensing can detect the presence or absence of a clay layer in a water bottom. It is about a ‘signal test’ which must show that variations in lithology of the water bottom have an effect in the measured signal.

The experiments for which the distributed fibre optic measurements have been performed consisted of experiments at a scale of 18 m x 5.5 m surface area and 2.5 m depth and for which fibre optics cable for acoustic and temperature measurements has been placed in a send bed of 80 cm thickness and covered with a clay layer of 10 cm. Three situations have been analysed:

1. An experiment without the flow of water and with a continuous clay cover.

2. An experiment without the flow of water and with a clay cover with a hole of 1m x 1m. 3. An experiment with a clay cover with a hole of 1m x 1m through which hot water

infiltrates.

Two types of fibre optic measurements were performed:

1. Acoustic fibre optic measurements with the Silixa iDAS-system (iDAS = intelligent Distributed Acoustic Sensing). For this a fibre optic cable has been wrapped around a pole which has been buried in the sand bed. A Knudsen 3.5 kHz Pinger has been used as an acoustic source.

2. Temperature measurements with a fibre optic cable (DTS = Distributed Temperature Sensing). A fibre optic cable has been buried in the sand bed with several loops from one side of the tank to the other and back.

The experiments show that the iDAS-system gives a reproducible signal when seismic sources are positioned at different locations and for different shots at one location. This means that the measurement can be considered representative for the configuration of the experiment and is not influenced by factors attributable to the source. The iDAS-measurements have been analysed in the time- as well as in the frequency-domain. In both analyses, a clear signal has been found at the location of the hole. This signal differs from the measurements done for the situation without a hole. From this it is concluded that the iDAS-measurement is sensitive to changes in the presence of a hole in the clay layer. However, the measured amplitude is not unique: variations in amplitude also exist outside of the hole. This means that – at least with the current analyses – a hole in a clay layer can be detected when the fibre optic measurements is used as monitoring device (anomaly detection), but not as a mapping device (reference-situation).

For the flow experiment, warm water was injected in the hole and he water was drained from the sand bed. In this way the iDAS-signal can be compared with an independently obtained DTS-measurement. A correlation was found between the amplitude of the seismic signal measured with the iDAS, and the temperature measured with the DTS. A cluster analysis of the two signals confirms the correlation between temperature and the amplitude of the acoustic

(6)

Signal test for acoustic fibre optics for leakage detection of water bottoms 2 signal. A raise in temperature of the iDAS cable results in a constant offset of the signal. For a gradually increasing temperature, there is a drift in the data. We do not expect that this slow drift will affect the registration of the high-frequency acoustic vibrations.

We conclude that fibre optics can be further developed for the purpose of monitoring of variations in water bottoms. In order to have it applied as a proven technology, several development steps must be taken:

• Define more specifically possible applications and model the response of the iDAS-system to different conditions based on the experimental results obtained so far. • Based on the modelling results, experiments are to be set up to verify the sensitivity of

the iDAS to various lithological conditions, primarily reflecting naturally occurring transitions (gradual instead of abrupt), the position of the cable on the water bottom instead of under the clay cover, and controlling sources disturbing the iDAS-signal during flow conditions, such as pumps and high temperatures.

• Based on these results a field experiment will need to be executed in order to validate the expected performance of the iDAS-system.

(7)

2 Managementsamenvatting

In het kader van KPP-project “2019 BO3 ‘Waterbodem: directe meting met kabel’” is een analyse gedaan van experimenten die in november 2018 en Februari 2019 uitgevoerd zijn met een glasvezelkabel bij Deltares. Het doel van het project was om te onderzoeken of de resultaten inzicht kunnen geven in het kunnen detecteren van de aan-/afwezigheid van een kleilaag in een waterbodem. Het gaat daarbij om een signaaltest waaruit moet blijken of variaties in de opbouw van een waterbodem doorwerken in het gemeten signaal.

De experimenten waarvoor de glasvezelmetingen zijn geanalyseerd bestonden uit schaalexperimenten van 18 m x 5.5 m oppervlakte en 2.5 m diepte waarbij een glasvezelkabel voor akoestische metingen en voor temperatuurmetingen was ingegraven in een zandlaag van ongeveer 80 cm dikte met daarboven een kleilaag van ongeveer 10 cm dikte. Drie verschillende situaties zijn nagebootst:

1. Een experiment zonder waterstroming en met een continue kleilaag.

2. Een experiment zonder waterstroming met een kleilaag waarin zich een gat bevindt van 1 m x 1 m.

3. Een experiment met een kleilaag waarin zich een gat bevindt van 1m x 1m en waardoorheen warm water stroomt.

Twee typen glasvezelmetingen zijn uitgevoerd:

1. Akoestische glasvezelmetingen met het Silixa iDAS-systeem (iDAS = intelligent Acoustic Distributed Sensing). Hiervoor is een paal omwikkeld met een glasvezelkabel in het zand ingegraven. Als bron is een Knudsen 3.5 kHz Pinger gebruikt.

2. Temperatuur metingen met een glasvezelkabel (DTS = Distributed Temperature Sensing). Hiervoor is een glasvezelkabel slingerend in het zandbed gelegd.

De experimenten laten zien dat het iDAS-systeem een reproduceerbaar signaal geeft als de akoestische bron zich op verschillende posities bevindt en bij opeenvolgende schoten op één bronpositie. Dit betekent dat de meting representatief is voor de configuratie van het experiment en niet beïnvloedt wordt door factoren die met de bron te maken hebben. De iDAS-metingen zijn zowel in het tijdsdomein als in het frequentiedomein geanalyseerd. In beide analyses wordt op de positie van het gat in de kleilaag een duidelijk signaal gevonden dat afwijkt ten opzichte van de meting zonder gat. Daaruit wordt geconcludeerd dat de meting gevoelig is voor veranderingen in aanwezigheid van een gat, in de vorm van een onderbreking, in de kleilaag. De gemeten amplitude is echter niet uniek: variaties in amplitude treden ook op buiten het gat. Dit betekent – althans met de huidige analyse – dat een gat in de kleilaag waargenomen kan worden als de glasvezelkabel als monitoringinstrument wordt gebruikt en variaties in amplitude onderscheidend zijn (verschilmeting of anomaliedetectie) maar niet als kartering (nul situatie). Voor het stromingsexperiment is gebruik gemaakt van injectie met warm water. Op deze wijze kan het iDAS-signaal met een onafhankelijke meting vergeleken worden. Er is een correlatie gevonden tussen de amplitude van het akoestische signaal dat met de iDAS gemeten is en de temperatuur die met de DTS is gemeten. Een statische analyse (clusteranalyse) bevestigt de correlatie tussen temperatuur en amplitude van het akoestische signaal. Het temperatuurseffect op de iDAS kabel is dat er bij een hogere temperatuur een constante

(8)

Signal test for acoustic fibre optics for leakage detection of water bottoms 4 verschuiving in de waarde optreedt. Bij een geleidelijke toename van temperatuur zal er een drift in de waarde optreden. Dit is echter niet van invloed op het hoogfrequente deel van het signaal. We verwachten dat deze langzame drift geen invloed heeft op de registratie van de hoogfrequente akoestische trillingen.

We concluderen dat de glasvezelkabel als monitoringinstrument voor het meten van variaties in waterbodem ontwikkeld kan worden. Om tot een praktijktoepassing te komen zullen nog verschillende ontwikkelstappen gezet moeten worden. Hierbij zijn de volgende stappen in ieder geval noodzakelijk:

• Het in meer detail definiëren van praktische toepassingen en op basis van specificaties en eisen die aan een monitoringsysteem gesteld worden en gebruik makend van de in deze experimenten opgedane kennis, een modelstudie te doen naar de gevoeligheid van een monitoringsysteem op basis van glasvezelmetingen.

• Het op basis van deze analyse uitvoeren van experimenten waarbij in ieder geval de gevoeligheid van het iDAS-systeem getest voor het detecteren van meer graduele overgangen in lithologie en kleibedekking getest wordt, getest wordt wat het effect is van de iDAS-kabel op de waterbodem in plaats van onder het kleidek en het uitvoeren van stromingsexperimenten waarbij de invloed van pompen en temperatuur op de metingen zoveel mogelijk uitgesloten kunnen worden.

• Het uitvoeren van een veldtest onder gecontroleerde omstandigheden voor het valideren van de metingen.

(9)

3 Introduction

3.1 Background

This report describes the results of KPP-project 2019 BO03 ‘Waterbodem: directe meting met de kabel’. This KPP project is part of the “Kavel BO” Soil and subsurface (reference 2018 BO03 and 2019 BO03).

The project is meant to test the opportunity to measure and/or monitor properties of water bottoms over large stretches (100 m – 1 km) by applying two types of distributed fibre optical sensing: Distributed Acoustic Sensing (DAS) and Distributed Temperature Sensing (DTS). The underlying question of RWS is their wish to determine the effects of groundwater flow on civil engineering works independently of external influences. Usually, the permeability of water bottoms is estimated using models. The level of uncertainty of the model results is rather high, because of uncertainties in the soil parameters. Direct measurements of the properties of the water bottom could reduce uncertainties, leading to a reduced risk of groundwater related failures and hence to lower costs and shorter durations of projects. It is expected that the new knowledge on distributed fibre optics will become applicable in feasibility studies, planning and construction works.

3.2 Scope of the signal test

Earlier preliminary tests performed in a small tank (2 m x 2 m x 2 m) showed that there were changes in the acoustic signal registered by Distributed Acoustic Sensing (DAS) indicative of lithological variations. These positive results were the start of designing a large laboratory test in a tank of 18 m length, 5.5 m width and a depth of 2.5 m to investigate the effect in lithological variation of the water bottom on the registration of the acoustic signal by the DAS system. The goal of this project is to analyse the results of the two large laboratory tests, which were performed in November 2018 and February 2019, in order to assess whether distributed fibre optic sensing can detect the presence or absence of a clay layer in a water bottom. It is about a ‘signal test’ which must show that variations in lithology of the water bottom have an effect in the measured signal.

3.3 Acknowledgements

The application of DAS to shallow subsurface applications is new. We collaborate with the English company Silixa, who provided the DAS system (iDAS). Athena Chalari and Stoyan Nikolov provided a workshop on the iDAS technique. Francesco Ciocca and Simon Dathan assisted in operating the system in the laboratory. Rufat Aghayev and Simon Dathan are acknowledged for their help in interpreting the data.

The experimental set-up and the performance of the experiments were facilitated by our colleagues Marcel Busink, Marcel Grootenboer, Edvard Ahlrichs, Mike van der Werf, Marco de Kleine, Rik Noorlandt, Sven Schilke and independent contractor Maarten Malingré. We express our thanks for their help.

(10)

3.4 Reading guide

The use of fibre optics as a sensor is described in chapter 4. The experimental set-up and the different experiments we performed are described in chapter 5. A more extensive overview of the experiments with sketches and photos is provided in appendix B. A description of the processing steps of the DAS and DTS data is included in chapter 6. The results of the signal test are described in chapter 7. This includes an analysis of the iDAS, DTS and combined iDAS/DTS data sets for the situation with and without a hole in a clay layer, and a situation with and without flow. Discussion on the applicability of the results and other issues is provided in chapter 8. The conclusions and recommendations are given in chapter 9.

(11)

4 Fibre optics as a sensor

DAS is an emerging technique which has been developed over recent years. The basic idea is that acoustic waves deform the fibre optic cable and this deformation is measured using the backscatter of a light pulse. In the oil industry, DAS has been applied for several years (e.g. Li et al., 2015). Geothermal applications are emerging, e.g. Mondanos and Coleman (2019). Applications of the technique in the shallow subsurface are being investigated (e.g. Jousset et al., 2018. The laboratory test set-up was also used in the KPP project “new techniques to monitor salt intrusion in waterways” (Kruiver et al., 2019).

4.1 Distributed acoustic sensing

The DAS system that we used is the iDAS (intelligent Distributed Acoustic Sensor), provided by Silixa. Figure 4.1 shows the principle of iDAS operation where the acoustic field interacts with the backscattered light generated along a continuous fibre optic cable. By analysing the backscattered light and measuring the time between the laser pulse being launched and the signal being received, the iDAS can measure the acoustic signal at all points along the fibre. The iDAS natively measures the rate of change of strain in the fibre or strain rate. This strain rate is averaged over a length of fibre equal to the gauge length of the system, which is equal to 10 m in the system used. This means that the light that is backscattered over a cable length of 10 m is averaged into one data point. However, the spatial resolution of the fibre optic cable is much finer, because the cable can be interrogated at intervals varying from 0.25 to 2.0 m (moving-average type of sampling over the gauge length). This results in one data point for every 0.25 m along the cable for the finest spatial sampling.

(12)

Signal test for acoustic fibre optics for leakage detection of water bottoms 8 The length of the cable, the duration of the pulse, the speed of light and data-transfer rate pose a limit to the sampling frequency of the recorded signal. For the experimental set-up that we used in the experiments, the maximum achievable frequency varied between 40 kHz and 60 kHz.

The ability of the system to record signals with certain frequencies is linked to the sampling frequency. According to Nyquist theory, the sampling frequency needs to be at least twice as high as the source frequency. Ideally, the sampling frequency should be higher. Because of the limit in sampling frequency of the iDAS system of 40-60 kHz, the choice of usable frequencies of the sources is limited as well. Preliminary tests were performed using a variety of acoustic sources with different frequencies. The final tests were performed using a 3.5 kHz Pinger source since this source performed well and complies with the conditions set by the Nyquist-theorem.

4.2 Distributed temperature sensing (DTS)

The DTS system used in these experiments is the Ultima DTS (Distributed Temperature Sensor), developed by Silixa. The principle of operation is similar to the iDAS in the sense that the analysed signal is the returned backscattered light from the initial emitted pulse. In this case however, different properties of this backscattered light are measured (Figure 4.2). Temperature differences can locally change the characteristics of light transmission inside a fibre, leading to scattering effects which can be detected due to the resulting spectral shift,

Figure 4.2 Principle of operation of a DTS system: the backscattered light from the original emitted laser pulse is returned to the system, analysed, recorded and displayed.

which contains three components. Of these, the Stokes and anti-Stokes components (Figure 4.3) are the used to obtain a temperature reading, with the former being temperature-dependent and the latter temperature-intemperature-dependent. By taking the ratio of these two components, the local temperature of the fibre cable is derived.

In this way, DTS allows the instantaneous measurement of temperature along an optical fibre. The Silixa Ultima can measure with a spatial resolution as fine as 25cm and a temporal resolution up to 1s, over a maximum distance of 1.8km (without splicing the cable). In optimal working conditions, temperature can be resolved with a precision of 0.01˚C. As with the iDAS,

(13)

Signal test for acoustic fibre optics for leakage detection of water bottoms 9 DTS systems also make use of a gauge length, or a pre-defined distance over which temperature is averaged into one data point, resulting in multiple temperature readings over the fibre length at 25cm intervals. For these experiments, the gauge length of the DTS unit was 0.7m.

(14)

5 Experimental set-up

5.1 Set-up of the tank

The experimental set-up was situated in the facilities of Deltares in Delft. The tank has a size of 18 m length, 5.5 m width and a depth of 2.5 m (Figure 5.1).

Figure 5.1 The experimental set-up in the Deltares facilities in Delft.

The building-up of the water bottom consisted of different phases (Fig. 5.2): 1 Construction of injection and drainage system (Figure 5.2a)

2 Base layer of sand with a thickness of approximately 80 cm. with fibre optic cables embedded and a densely wrapped pole horizontally placed on top (Figure 5.2b and c and Figure 5.4).

3 Coverage of the sand layer with a membrane (Figure 5.2d)

4 Coverage of the sand with a clay layer of approximately 10 cm., covering the densely wrapped pole (Figure 5.2e).

(15)

Signal test for acoustic fibre optics for leakage detection of water bottoms 11 The schematic side view is shown in Figure 5.3. After finishing each of the three set-ups, acoustic measurements were performed. The relative positions of the iDAS fibre optic cables and the DTS cables is shown in Figure 5.5.

A more complete overview of sketches and photos is included in Appendix B.

(16)

Signal test for acoustic fibre optics for leakage detection of water bottoms 12 Figure 5.3 Schematic representation of different phases of the tank lay out (side view). Poles A, B and C

(redlines) represent iDAS fibre optic cables, which are densely wrapped around a PVC pipe (Figure B.9 in appendix B). Pole C is the main iDAS cable for this project. Poles A and B were used in another KPP project.

(17)

Signal test for acoustic fibre optics for leakage detection of water bottoms 13 Figure 5.4 The position of the densely wrapped pole C in the top of the sand bed.

Figure 5.5 Schematic representation (top view) of the various fibre optic cables (zoom). Pole C is lying in between two DTS cables. The DTS cables are embedded in the sand. The pole with the iDAS cable wrapped around it is embedded in the top of the sand bed. The position of the hole in the clay during the experiments on 13 and 14 February 2019 is indicated. For clarity, the cables are drawn on top, but in reality, they are situated below in the (top of the) sand.

(18)

5.2 Experiments

Measurements were performed in different rounds, corresponding to the different phases of the set-up. The different test days are summarised in Table 5.1. The tests in November 2018 were performed using a range of different sources. Based on the results and conditions set by sampling frequency, the Knudsen 3.5 kHz Pinger (Figure 5.6) was selected for the experiments in February 2019. The source in the clay cover experiments was the Knudsen 3.5 kHz Pinger which was fully immersed in the water. This is not the only suitable source. In this case, the sampling frequency of the iDAS system is the limiting factor. The source signal must be sampled with enough temporal resolution in order to avoid aliasing. Currently, and in this laboratory set-up, the maximum sampling frequency of the iDAS was 50-60 kHz. This frequency depends on e.g. the fibre cable length and the iDAS unit. Future systems might have better specifications, in our case meaning a higher sampling frequency. Due to the current maximum sampling frequency, the maximum source frequency is 20-30 kHz. The 3.5 kHz Pinger proved to be a suitable source and we demonstrated that signals could be retrieved with this source. In Kruiver et al. (2019) the Edgetech X-star Sb-424 with a sweep from 4 to 24 kHz performed well. Some other sources, e.g. a 33 kHz dual frequency echosounder did not produce any usable data on the iDAS cable in this experimental set-up. Any source with a sufficiently low frequency could be suitable in combination with the iDAS recording. The choice is not limited to the 3.5 kHz Pinger or the X-star.

Table 5.1 Summary of test days.

Date Set-up Purpose Acoustic sources Flow iDAS/DTS

21, 22, 23 November 2018 Sand bed only Testing different acoustic sources and basic signal test Knudsen 3.5 kHz Pinger, Knudsen dual frequency echosounder (33 & 210 kHz), Hammer, Seismic tube (experimental source)

Yes (cold) iDAS/DTS

6 February 2019 Continuous clay cover Reference of clay cover. Knudsen 3.5 kHz Pinger No iDAS 13, 14 February 2019 Clay cover with holes Response of holes in clay cover Knudsen 3.5 kHz Pinger Yes (warm) iDAS/DTS

(19)

Signal test for acoustic fibre optics for leakage detection of water bottoms 15 Figure 5.6 (A) Knudsen 3.5 kHz Pinger source used in the tests. In some cases, the source as pointing down as in classical operation. In other cases, the source was tilted to vertical, directing the acoustic waves more horizontally through the water tank.(B) Source positions (red line is position of the pole; green rectangle is position of hole). DTS data were gathered on 21, 22 and 23 November 2018 and on 13 and 14 February 2019. For the experiments on the 21st_{, 22}nd_{and 23}rd_{of November, a square wooden box of} dimensions 1m x 1m was placed on the sand bed, topped with a lid containing a hole through which the hose was placed for the injection of water. Flow was induced by turning on the pump in the corner box of the tank. This results in extraction of water via the drainage system in the sand bed in three compartments. At the same time, water was injected in the square wooden box on top of the sand bed (indentation visible in Figure B.4 in appendix B). The injected water was not heated on 23 November 2018. We therefore expect that no significant changes in temperature occur during the flow experiment on 23 November 2018.

On the 6th_{February 2019, the tank was then prepared for the introduction of the continuous} clay layer over the existing sand layer by placing a non-permeable, PVC membrane cover on top of the sand to have a sufficient sharp interface between the sand and the clay cover. This membrane has a thickness of 1 mm and therefore we do not expect an influence on the acoustic signal retrieved. Three wooden frames were then placed and glued on top of the PVC: one on top of pole C and two others. These frames were present on all the experiments carried out in February 2019. The clay layer had a thickness of approximately 10cm and was applied both inside and outside the wooden frames. Tests were performed using only iDAS for this situation in which the clay cover was continuous and temperature constant.

(20)

Signal test for acoustic fibre optics for leakage detection of water bottoms 16 After the tests for a continuous clay cover, the clay was taken out from the wooden frames to create a situation with a discontinuous clay cover for the experiments on 13th_{and 14}th_February.

Next, a large box was placed on top of the wooden frame on the 14th_{February 2019, like a} diver’s bell for the consecutive flow experiments. A hole was made for the hose on the top of the box, as well as some holes for the air to escape. With the holes now present in the clay layer, the flow was performed in a different way. Tanks of 700-800L capacity were heated up to 55-60 degrees Celsius, next to the large tank in which the experiments were performed. The water was pumped from the two storage tanks and injected in the box + frame combination on top of pole C. The experiment was done with the hole that was situated on top of pole C. In this case, a wooden box was inserted in the tank that fitted on top wooden rim of the hole. Hot water was pumped into this box and at the same time, the pump in the corner of the tank was operating. In this way, hot water was flowing through the hole from the top down into the sand bed and the water was extracted via the drainage system. We expect that temperature differences will be detectable in the DTS data. Two portions of the DTS cable and one portion of iDAS cable are situated below the hole (Figure 5.5).

(21)

6 Processing of the DAS-data

In this chapter we will give a summary of the processing steps of the iDAS-measurements. We will first summarize the steps taken for quality control and after that describe the processing performed in the time domain and the frequency domain1_.

6.1 Raw data visualisation and Quality control

A proper understanding of the spectral features of all collected iDAS acoustic signals allows the selection of appropriate processing parameters in time and frequency domain. The main advantage of iDAS acquisition system relies on the high sampling rate and very fine spatial resolution, offering a unique opportunity of mapping variations or anomalies in both, temporal and spatial domain simultaneously. The data were collected in a linear geometry along the length of pole C that is buried within a sand and clayey layer. An example of the shotgathers2 recorded at all source positions are displayed in Figure 6.1.

a) b)

c) d)

Figure 6.1 Collected raw record along pole C for shot-positions for a) source A, b) source B, c) source C and d) source F (see Figure 5.6).

1_{Time Domain refers to the analysis of a signal with respect to time. Frequency Domain refers to the analysis of a} signal with respect to the frequency content of the signal. A time-domain graph shows how a signal changes over time whereas a frequency domain graph shows how much of each frequency contributes to the final signal.

(22)

Signal test for acoustic fibre optics for leakage detection of water bottoms 18 The prominent wave arrivals observed in the recorded shot-gathers are indeed consistent with the actual source positions. The signals along Pole C, are contained within trace numbers 1,400 – 4,500. High amplitude waveforms before 1,400 and after 4,500 trace numbers correspond to portions of fibber cable that were not attached to Pole C.

A first assessment consists of computing the Fourier Amplitude Spectrum (FAS) as a function of trace number, the highest spectral energy trend can be identified (Figure 6.2a). It appears that most of the energy is concentrated in a frequency band between 1,400 Hz to 5,000 Hz and predominantly at 3,500 Hz which is below the valid maximum frequency limit of 25,000 Hz imposed by Nyquist criterion which means that the data is sampled at a high enough temporal resolution.

a)

b)

Figure 6.2 a) Frequency response of collected traces as a function traces number, b) Fourier amplitude spectrum of individual traces of a single shotgather (red line represent the average spectrum).

The frequency response appears to be consistent in all collected traces, which is observed in the average FAS displayed in Figure 6.2b. A filtering of the collected signals isolating the main energy trend will improve the signal-to-noise ratio, enhancing the main amplitudes of the acoustic signals of interest.

(23)

Signal test for acoustic fibre optics for leakage detection of water bottoms 19 Figure 6.3 a) Single time domain acoustic signal, b) Fourier amplitude spectrum, c) Phase angle spectrum. A more detailed analysis can be made through the description of the spectral characteristics of a representative single trace (Figure 6.3) of the collected shot-gathers displayed in Figure 6.1. The main signal (highlighted in blue square) depicts clear and coherent cycles that resembles the main acoustic response of the measured wavefield (Figure 6.3a). A second part of the signal (highlighted in red) shows a smaller amplitude and incoherent wave pattern with very marked influence of high frequency noise above the 10,000 Hz as observed in Figure 6.3b. Figure 6.3c portraits the typical shape of phase angle [-180, 180] (in degrees) of the full raw signal.

6.2 Time domain analysis

The data analysis is performed using the envelope of the time domain signals, comprised by the absolute values of the analytical signal computed by the Hilbert transform. The envelope of the time domain signal is utilized as a measure of the maximum energy of each trace. An example of the computed envelope of one of the collected acoustic signals displayed in Figure 6.1, is presented in Figure 6.4.

(24)

Signal test for acoustic fibre optics for leakage detection of water bottoms 20 To enhance the prominent amplitudes of all collected shotgathers and prior to the envelope computation, the data are filtered in a frequency band where the strongest energy occurs as observed in Figure 6.2. Hence, the filtering is performed in a frequency limit of 1,400 Hz - 4,800 Hz. The benefits of the bandpass filtering applied, is shown in Figure 6.5b. It is observed that after applying the filtering the resulting time domain signal depicts a smoother and clearer oscillatory pattern, which enhances the shape of the computed envelope. The computed envelope after filtering the signal shows a clearer wave trend reducing the effect of the high frequency noise as observed when comparing Figure 6.5a and Figure 6.5b.

Figure 6.5 a) envelope of Raw signal, b) envelope of filtered signal, and c) Fourier Amplitude Spectrum for raw and filtered signal.

Figure 6.6 Computed envelopes as function of traces number (Envelopgram) for a collected shotgather. Red colouring indicates high amplitudes.

(25)

Signal test for acoustic fibre optics for leakage detection of water bottoms 21 An example of a computed “Envelopgram3_{” as a function of trace number using the computed}

envelopes is displayed in Figure 6.6. The observed dark-red patterns represent the enhanced highest amplitudes of the envelopes, that can be used for “imaging” all reflections and/or resonances associated to potential anomalies.

As a final step, the envelopes that are computed for at least 100 individual shots per source positions A, B, C, and F (see Figure 5.6 for shot positions). As a result, at each position an average envelope (Envelopgram) is computed. If the resonances caused by the potential anomalies are independent of the source positions, then the average images (Envelopgrams) of the four sources can be summed to enhance the common features. This procedure is performed for both, the hole and no-hole situations.

Since the shot positions A and B were collected using a sampling frequency of 15 kHz those are not considered for this analysis.

6.3 Frequency domain analysis

Frequency analysis was performed by Silixa in an early stage of the project. The data for clay with and without the hole were acquired using a different sample frequency. The amplitudes were corrected for that. Data per trace in time domain were converted to data per trace in frequency domain using Fourier transformation. Because of the good repeatability of the waveform, the data of numerous shots were stacked channel by channel to increase the signal to noise ratio. In order to enhance the effect of the hole, the ratio of amplitudes between the data of clay with hole to the data of clay without the hole is computed.

Deltares performed additional frequency domain analysis by the computation of density spectra for all source positions for both the hole and no-hole cases. For the no-hole case only source positions C and F were considered. Finally, the spectrograms computed for all source positions were averaged aimed at enhancing existing anomalies which should be consistent independent of the source position.

(26)

7 Results signal test

In this chapter we will describe the results of the analysis of the experiments. We will start with an analysis of the processed results of the experiments which were performed under no flow condition, both with a continuous clay cover and with a clay cover with a hole in it. In the second paragraph we will treat the experiments where flow was conducted, treating both the iDAS and DTS measurements separately as well as combined.

The first round of experiments on 21, 22 and 23 November 2019 were performed with several different acoustic sources. The response of the iDAS cable to the different sources is reported in Kruiver et al. (2019). From these tests, the Knudsen 3.5 kHz Pinger was selected as the best source for the situation with the clay cover and the clay cover with a hole since it complies with the maximum sampling frequency. This source was then used in the experiments of 6, 13 and 14 February.

7.1 No flow Experiments

The first sets of experiments were performed in a static situation. There was no flow of water. Measurements were conducted with the sand bed, the clay bed and a hole in the clay as explained in chapter 5. The acoustic results were analysed in time domain (section 7.1.1) and in frequency domain (7.1.2). The results from the two domains are combined in section 7.1.3.

7.1.1 Acoustic results in time domain

An example of a collected shot-gather (normalized) for source position C for the hole and no-hole conditions are displayed in Figure 7.1. Despite that the amplitudes are normalized, by visual inspection, it is not immediately clear how the wavefield is modified when the hole is present.

○

a) No hole

b) Hole

(27)

Signal test for acoustic fibre optics for leakage detection of water bottoms 23 The existence of potential anomalies can be observed by means of the Envelopgrams as described in section 6.1.2. The computed average envelopes for both source positions C and F, when no hole is present, do not show a common prominent amplitude pattern (Figure 7.2). Despite the common strong amplitudes close to trace number 1,500 and the strong amplitude in source position C, the remaining traces show no marked amplitude indicating that no anomaly is imaged in the computed average envelopes.

Figure 7.2 Average envelope (Envelopgram) when the hole was not present, for source positions C (left) and F (right). Red indicates high amplitude.

On the other hand, for the case when the hole was present (Figure 7.3), the source positions A, B, C, and F consistently display a common prominent amplitude pattern between trace number 3000 and 3400. Despite that the data were collected using four different shot positions the prominent amplitudes appear to be consistently occurring at the same trace numbers.

(28)

Signal test for acoustic fibre optics for leakage detection of water bottoms 24 Figure 7.3 Average envelope when the hole was present, for source positions A, B, C, and F. Red indicates high amplitude.

Given that the location of the anomalies is independent of the source location, the averaged images from all source positions are summed together, so the signal-to-noise ratio is increased enhancing existing common features. The peak values extracted from summed source-averaged envelopes for both, the no-hole (Figure 7.4a) and hole (Figure 7.4b) conditions are displayed in Figure 7.4c. In general, both curves appear to be very similar, except at traces 3,000 and 3,400 where prominent resonance peaks occur. The position of the peaks fits the dimensions of the hole that was, indeed, present during the data acquisition.

(29)

a) b)

c)

Figure 7.4 a) Average envelope of No-hole condition, b) hole condition, and c) Peak amplitude per trace for hole (red) and no-hole (green) conditions.

7.1.2 Acoustic results in frequency domain

A preliminary analysis in frequency domain was performed by Silixa in an early stage of the project. Data from the set-up with sand only were not analysed in frequency domain, because the recorded signals were clipped. This was because of the powerful acoustic source, in combination with pole C being directly exposed to the acoustic signal. As a result, the recorded waveforms were distorted, and no meaningful frequency spectra could be retrieved.

Data from the test day of 6 February 2019 (clay without holes) and from the test day of 13 February 2019 (clay with holes) were analysed. Because of the clay cover, the acoustic signal was attenuated, and the source power was adjusted. The data on these two days were recorded with a different sample frequency: 15 kHz (source positions A, B, C, D, and E) and 50 kHz (source positions C and F) on 6 February 2019 and 50 kHz on 13 February 2019. This results in differences in acoustic amplitudes. A frequency-based correction is applied to compensate for the difference in amplitude.

The frequency response between shots is very similar and the shots were triggered. Therefore, the shots from the same shot position were stacked to improve the signal to noise ratio. Figure

(30)

Signal test for acoustic fibre optics for leakage detection of water bottoms 26 7.5 shows the stacked data for source position A in time domain and in frequency domain. Averaged over all channels (Figure 7.6), the spectral content of the clay data with and without holes is very similar. Analysing the frequency content per trace (Figure 7.5), however, shows that there is an anomaly between traces 3000 and 3400. This anomaly is even more pronounced when the ratio clay data with and without holes of the stacked acoustic signals is plotted for different shot positions (Figure 7.7). The anomaly of the hole is present between traces ~3000 and ~3400.

Figure 7.5 Source position A, average of 100 shots. Top: shot gather data with clay cover without holes (left) and with holes (right). Bottom: frequency spectra versus trace of data with clay cover without holes (left) and with holes (right). Figure from Silixa.

(31)

Signal test for acoustic fibre optics for leakage detection of water bottoms 27 Figure 7.6 Source position A average power spectrum over all channels. Figure from Silixa.

Source position A Source position C Source position F

Figure 7.7 Ratio of clay data with holes to clay data without holes. Top: in time domain, bottom in frequency domain. From left to right for different shot positions. Figure from Silixa.

7.1.3 Combined time domain and frequency domain analysis

In order to compare the time domain analysis and frequency domain analysis, Deltares repeated the frequency domain analysis using the 50 kHz sampling frequency data only. The source-averaged time domain Envelopgrams and Spectrograms computed from sources

(32)

Signal test for acoustic fibre optics for leakage detection of water bottoms 28 positions A, B, C, and F are displayed in Figure 7.8. Both average images show a consistent high energy anomaly between traces 3000 and 3400.

a)

b)

Figure 7.8 Average time and frequency domain spectrograms for hole detection.

The computed peak values as a function of trace is presented in Figure 7.9. When comparing both, time and frequency domain curves, the strong peaks at traces 3000 and 3400 is clearly a common feature, as observed in Figure 7.9a. In time domain analysis the amplitudes appear to be noisier compared to those computed in frequency domain that show more clear prominent peaks. The fact that the anomaly is observed in both analyses, indicates that the spikes indeed correspond to the strong resonances caused, most likely by the boundaries of the hole. A clearer picture can be obtained after normalizing and convolving both curves in time and frequency domain. In Figure 7.9b the resulting curve enhances the existing 3 prominent peaks. The left peak corresponds to strong amplitudes outside pole C, while the two spikes between traces 3000 and 3400 are strong resonances most likely caused by the presence of the hole in the clayed layer.

(33)

a)

b)

Figure 7.9 a) Peak average amplitude density spectrum and envelope for all source positions, b) convolved time and frequency domain response.

7.2 Flow experiments

Several experiments with the flow of water were performed (see Chapter 5). During the first tests in November 2018, we had a rough idea about how to perform a flow test. Water was simply drained from below using the drainage system and injected on a selected location on the sand bed. No difference in water temperature was applied. The performance was optimised in the last experiment. Hot water was used to infiltrate and could be traced by DTS cables, thus confirming flow conditions. Section 7.2.1 describes the results of the first trial. In section 7.2.2 till 7.2.5 the processing and analysis of the data of the last experiment, when hot water is injected in the hole, is described.

7.2.1 Sand bed – DTS data

During the first tests in November 2018, one flow experiment was performed. The position of the injection was not co-located with the iDAS pole C and acoustic data were clipped due to the strong acoustic source. Therefore, no acoustic data were analysed.

(34)

Signal test for acoustic fibre optics for leakage detection of water bottoms 30 The DTS data were visualised along chosen distances along the fibre cable and time intervals. In the produced plots, the colour denotes temperature in degrees Celsius. In all generated plots, a mirroring effect was seen, leading to areas where data is duplicated and areas where there is no valuable data. By investigating the entire dataset over all lengths, it was found that the area of interest to us is between 65m and 150m length along the fibre. The fibre lengths between 60-65m and from 150m onwards clearly has no useful data which has to do with the fact that these sctions were outside flume.

Figure 7.10 represents the temperatures measured on the 23rd_{of November between 14:00} and 14:45. On this day, flow experiments were performed, starting at 14:18. The water from the basin was used in the test, with no temperature difference. By observing Figure 7.10, it is evident that the flow experiment had no impact on the measured temperature, which essentially remains constant along the fibre section of interest throughout this measurement period. Some stripes of slightly higher temperature (around 14.5°𝐶, compared to 13.5°𝐶 zones surrounding the stripes) can be seen at 2m intervals, particularly between 90m and 120m, although the temperature along them does not change significantly with time.

Figure 7.10 Temperatures measured with the DTS system on the 23rd_{of November 2018. Note that the flow}

(35)

Signal test for acoustic fibre optics for leakage detection of water bottoms 31 7.2.2 Clay with hole – iDAS data

Figure 7.11 displays the magnitude of the amplitude envelope of signals coming from the fibre cable wrapped around pole C from the experiment on the 14th_{February 2019 (containing the} hole in the clay layer). The edges of the hole are clearly distinguished: two sections of the cable seem to be associated with higher amplitudes over the entire duration of the flow experiment. The first is roughly situated around ~ 2950 and contains the highest amplitudes measured during this experiment, with a noticeable sharp contrast in amplitude envelope magnitude compared to adjacent sections. The second section is approximately located around trace number ~3320. It also contains relatively high amplitudes albeit lower than the previous section and displays a sharp contrast in amplitude envelope magnitude compared to the fibre cable sections surrounding it. These sections coincide with the edges of the hole dug into the clay layer, which seems to be characterised by high amplitude envelopes during the injection of warm water.

Figure 7.11 The magnitude of the amplitude envelope of acoustic signals coming from the fibre cable wrapped around pole C and situated underneath the wooden box. The colours indicate the amplitude intensity (blue = low and red = high intensity). Trace selection is zoomed in on the hole. ‘14’ on the x-axis indicates the date. The other numbers indicate time. The pump wat turned on at 9:00:53 (green line) and the injection of hot water started at 9:03:05 (red line) and ended (second red line).

Some other features that can be identified in Figure 7.11. Pumping started at 9:00:53. The signal is much noisier from that moment on. This is most visible in the high amplitude section between trace numbers 2950 and 3000. The injection of warm water, which is initiated around 09:03:05 leaves a more distinct imprint on the signal. From that time on, there is a marked difference in magnitudes of the envelope inside the box (cyan colours) and outside the box (blue colours). Between 09:05:00 and 09:15:00, noise increases inside the high amplitude envelope sections, as seen by the discontinuous and oscillating values. This can be associated to the continuous injection of warm water, which is sensed by the instrument.

7.2.3 Clay with hole – DTS data

Figure 7.12 displays the acquired temperatures on the 13th_{of February 2019, before hot water} was added to the tank. The recorded temperatures along specific fibre lengths remain constant with time, indicating that the experimental conditions have not impacted the DTS temperature readings. A region between 105m and 110m seems to have a very low temperature (around 10 degrees Celsius) throughout the measurement period compared to its surroundings. Since the cable is looped, this region refers to the same regions as at about 65m – 70m and 145m – 150m and refer to the boundary of the flume.

(36)

Signal test for acoustic fibre optics for leakage detection of water bottoms 32 Figure 7.12: Temperatures measured with the DTS system on the 13th_{February, no flow. Times DTS referenced}

(local machine time), this time stamp is one minute behind relative to the iDAS reference which is UTC.

In Figure 7.13, the temperature data belonging to the experiment of the 14th_{February 2019 are} plotted. The injection of warm water began at 09:01:56 and stopped at 09:35:05, with the pumps being shut off at 09:57:19 (local machine time). The first detection of an increase in temperature is focused in a stripe-shaped region centred around 115m at about 09:15, thus at an approximately 13-minute delay compared to the start of the warm water injection phase. Another heated region is seen centred around 100m, in which the temperature increase due to the injection of warm water is delayed slightly a few minutes more than the previous one. These are the only distances along the fibre cable on which an increase in temperature was sensed and which coincides with the hole where the injection of warm water takes place.

Both regions have a thickness of around 2.5 meters respectively and are about 8m apart. They represent the sections of the fibre cable which are adjacent to the box placed over the hole (Figure 5.5). The distance between them is the section of the fibre cable which loops towards the edge of the tank before returning to the box, and this area once again shows the lowest temperatures (around 10 degrees), similarly to Figure 7.12. We observe two temperature anomalies related to the two sections of DTS fibre that coincide with the box. We also observed

two anomalies in the iDAS data, but these are related to the two edges of the hole recorded at

different locations in one section of pole C that coincides with the hole.

The fact that the length of the sections showing a temperature increase (around 2.5m) is larger than the length of the fibre below to the frame that marks the hole (1 m) is further highlighted in Figure 7.14. In contrast, the cessation of warm water injection, which occurred at 09:35:05, can

(37)

Signal test for acoustic fibre optics for leakage detection of water bottoms 33 be detected in our data at the correct time, as the temperature signals around 115m show a clear decrease in temperature following that time. The smearing of the elevated temperatures over a larger section of the fibre length is probably related to the gauge length of the DTS fibre (2 m), which is somewhat larger than the spatial resolution (0.25 m).

Figure 7.13 Temperatures measured by the DTS system on the 14th_{February 2019, during the pumping test and}

the addition of warm water into the tank. Times DTS referenced (local machine time), this time stamp is one minute behind relative to the iDAS reference which is UTC.

(38)

Signal test for acoustic fibre optics for leakage detection of water bottoms 34 Figure 7.14 Temperature evolution with distance along the fibre cable for 3 selected times: before turning the pumps on (blue), during injection of the warm water (orange), and after the warm water injection has ceased (green).

7.2.4 Clay with hole – combined iDAS and DTS data

In this section, we combine the iDAS and DTS data from during the flow experiment. Figure 7.15a is essentially a zoomed-in version of Figure 7.13, showing the temperatures recorded by the DTS system on the experiment of the 14th_{February 2019. In this case, the temperatures} along only one side of the box are shown. It must be noted that the distance along the fibre cable has been converted to trace number on the y-axis. This was done to enable the precise comparison and identification of common features between the DTS and iDAS data. To achieve this, an interpolation was performed along both the spatial and time dimensions of the DTS data to match the data ‘co-ordinates’ of the iDAS data. The datasets were then visualised using common time and distance axes.

The onset of a high amplitude envelope section and of a sharp increase in temperature seem to be separated by 3 minutes, with the former starting at around 09:14 in the iDAS data and the latter at approximately 09:17 in the DTS data. A delay was also observed in the response of the DTS data to the injection of warm water (Section 7.2.1). This delay can be explained by the fact that, for the warm water injection to be detected by the DTS cable, the warm water must first mix with the colder water in the tank and then propagate to the DTS cable to be sensed. This process clearly takes longer than that of the acoustic signature of warm water injection reaching the iDAS cable.

(39)

Signal test for acoustic fibre optics for leakage detection of water bottoms 35 Furthermore, the upper and lower bounds in terms of trace number for the high amplitude envelope magnitude section (iDAS) and high temperature region (DTS) seem to match quite well, with the former being between 2050 and 4150 and the latter between 1900 and 4200. One difference between the two datasets is that the DTS data between the edges of the hole (given by trace numbers ~1900-4200) coincide with the highest measured temperatures after warm water injection, while the interior of the whole measured by the iDAS data (between traces 2050 and 4150) is associated with overall low amplitudes reaching progressively higher amplitudes as the experiment continues and particularly after 09:30.

a)

b)

Figure 7.15 (a) The DTS-derived temperatures corresponding to the section of the fibre adjacent to one side of the wooden box (zoom in on Figure 7.13), distance on y-axis was converted to trace number to be able to compare with the iDAS data. DTS time is corrected to iDAS reference. (b) The magnitude of the amplitude envelope of acoustic signals coming from the fibre cable wrapped around pole C and situated underneath the wooden box.

Figure 7.16 is a plot of the maximum temperature of all traces of the DTS data combined with the maximum magnitude of the amplitude envelope of the iDAS data across the entire duration of the experiment. A strong correlation in the increase of maximum temperature and maximum amplitude can be seen at around 09:14 in the iDAS data, and 09:16 in the DTS data. This agrees with our previous observations, but we can now also see that the rate of increase is also very similar. After the maximum temperature and amplitude is reached, both datasets show a slight decreasing trend. The lag between the rising limbs of the iDAS-signal and DTS-signal can be attributed to the fact that the acoustic vibrations due to the injection of the warm

(40)

Signal test for acoustic fibre optics for leakage detection of water bottoms 36 water travels faster than the warm water front itself, which is limited by the combined effect of flow velocity and diffusion. This time lag is further increased due to the fact that the iDAS-pole is located at a shallower depth and at midpoint whereas the DTS-cable lies deeper and at the adges of the hole through which infiltration takes place.

Figure 7.16 Maximum temperature from the DTS data (blue) plotted against the maximum magnitude of the amplitude envelope of the iDAS data (red) across the entire duration of the experiment on the 14th_{February 2019.}

X-axis represents the time in seconds elapsed since the start of the experiment at 08:58:00.

7.2.5 Clustering temperature and amplitude envelope data

To analyse the effect of temperature in the acoustic signals, a clustering analysis is performed taking the prominent amplitudes (related to one edge of the hole) of iDAS data that are concentrated in the yellow/red pattern between trace numbers 2900 and 3000 (Figure 7.15b) and the entire DTS section that provides the temperature variability nearby the hole before and after pumping hot water. To perform the clustering analysis, an equivalent temporal domain is set for both iDAS and DTS data (Figure 7.17). The clustering analysis is aimed at determining to what extent changes in temperature influences the wave amplitudes of the selected acoustic signals in the spatial and temporal domain.

(41)

Signal test for acoustic fibre optics for leakage detection of water bottoms 37 Figure 7.17 Selected DTS (top) and iDAS (bottom) for clustering analysis.

The clustering analysis is performed using the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) method. The method finds those groupings of measurements with the highest density observations which are then used to select clusters that may describe most of the correlation. The distribution of both selected set of data is presented in the histograms of Figure 7.18a. Figure 7.18b show the distribution of 77 samples with high density from which the same number of clusters is estimated. The coloured points are the computed clusters, while the black thin points are the samples considered noise. For this analysis, we select three clusters that are more dominant according to the histogram of Figure 7.19.

(42)

a)

b)

Figure 7.18 a) Histogram of selected temperature and amplitude data, b) identified clusters based on Density distribution.

(43)

Signal test for acoustic fibre optics for leakage detection of water bottoms 39 Figure 7.19 Histogram with computed classes.

Figure 7.20a show classes 1 (red), 2 (green), and 3 (blue) displayed as a function of temperature and amplitude envelope, which are selected based on dominant classes observed in the histogram of Figure 7.19. The spatial and temporal distribution of the three classes identified are extracted and displayed in Figure 7.20b. The Class 1 confirms the correlation between high temperature between 25 and 35 °C and the highest amplitude of the envelopes, which take place after 09:15 until the end of the measurements. Classes 2 and 3 correlates the lower and medium amplitude values (that is dominant before 9:15 am) with low temperature between 10 and 15 °C. Small increments in envelope amplitudes before injection of hot water takes place seem to be caused by the water flowing.

(44)

a)

b)

Figure 7.20 a) Selected clusters according to density distribution, b) Class distribution as function of trace number and time.

(45)

8 Discussion

Acoustic sources

The source in the clay cover experiments was the Knudsen 3.5 kHz Pinger. This is not the only suitable source. In this case, the sampling frequency of the iDAS system is the limiting factor. The source signal must be sampled with enough temporal resolution in order to avoid aliasing. Currently, and in this laboratory set-up, the maximum sampling frequency of the iDAS was 50-60 kHz. This frequency depends on e.g. the fibre cable length and the iDAS unit. Future systems might have better specifications, in our case meaning a higher sampling frequency. Due to the current maximum sampling frequency, the maximum source frequency is 20-30 kHz. The 3.5 kHz Pinger proved to be a suitable source and we demonstrated that signals could be retrieved with this source. In Kruiver et al. (2019) the Edgetech X-star Sb-424 with a sweep from 4 to 24 kHz performed well. Some other sources, e.g. a 33 kHz dual frequency echosounder did not produce any usable data on the iDAS cable in this experimental set-up. Any source with a sufficiently low frequency could be suitable in combination with the iDAS recording. The choice is not limited to the 3.5 kHz Pinger or the X-star.

Applicability of the iDAS for detecting variations in lithology

One of the goals of this study was to assess whether the iDAS measured significantly different signals under a given experimental set-up between a situation with and without a continuous clay cover. The comparison of results between the situation with and without the hole shows that there is a clear difference, both in time domain and in frequency domain at the position of the hole. The position of the hole could be deduced from the iDAS-measurements. However, the variation of the signal along the cable is rather high (Figure 8.1). One can deduct that the location of the hole can be simply obtained from the ratio between the hole to no-hole curve. However, in real world conditions a “no-hole curve” does not exist, so the location of the anomaly associated to the hole itself must be deducted based on a single multiple spikes curve. The single multiple-spikes approach in turn will be challenging. Without the comparison with the green line, we cannot tell where the hole is in the red line. Data to the left of the hole look just as variable in amplitude as inside the hole.

Figure 8.1 Peak amplitude with respect to distance for average envelopes. Red line is the situation with hole; green line is the situation with continuous clay cover.

(46)

Signal test for acoustic fibre optics for leakage detection of water bottoms 42 In a real-life situation in for example canal bottoms and in a monitoring set-up, we expect to be able to derive changes in clay cover thickness by changes in the iDAS response, given a similar measurement set-up, i.e. having the iDAS-cable buried underneath the clay cover. It must be said that in practice, the cable will be positioned atop of the water bottom, thus resulting in a different measurement set-up.

Secondly, the experimental results have given confidence in the measurement of a hole with straight edges in a clay cover. In reality, the variations will be smoother. With the current experiment we have not yet been able to assess whether smooth gradients in lithology will give sufficient differences in measured signals. Insights in the sensitivity to these changes in relation to sources of noise can be gained by a modelling exercise.

Sensitivity to holes in the vicinity of the iDAS cable

Based on the results of this experiment, when convolving the time and frequency domain curves, the anomalies associated to the presence of the hole can be clearly identified. However, the hole was constructed sharp enough so that it caused reflections or resonances located on both sides. In real-world situations the geometry of a hole can be very variable and sometimes so smooth that one could not be able to properly identify reflections or resonances. The fibre optics, however, possess a wide range of sampling resolution, so it has the potential of identifying small and large anomalies by just varying the spatial sampling of the shot-gather. In order to assess the ability of the iDAS to detect buried anomalies in more complex situations a sensitivity-analysis must be performed.

Flow measurements

The experiments show that both temperature as well as acoustic signals are well retrieved from the DTS and iDAS systems during flow conditions. The flow experiment with hot water through the hole showed that the onset of the different stages (turn on pump, injection of hot water, etc) could be identified in the acoustic data. For example, the data is very stable prior to turning on the pump. At that moment, the data contains much noisier. Next, the acoustic energy increased in the hole, while there was hardly any change next to the hole. This shows that the water flow through the hole has a different signature. The patterns in iDAS data and DTS data are consistent.

Correlation of iDAS and DTS in the flow experiment

We find a correlation between temperature and amplitude which relates to the injection of hot water through the hole. However, up to now, it is challenging to assess the exact nature of this correlation since various sources might contribute to this correlation. The increase in amplitude might be due to vibrations of the pump or the changing elastic properties of the medium due to the temperature increase. The increase in temperature induces a strain in the fibre, but this is constant. It will therefore give rise to a constant shift in iDAS readings. We do not expect that this will affect the high-frequency acoustic vibrations. Still, a further analysis of these influences must be carried out to assess more definitely the sensitivity of the iDAS-system to flow.

Signal test for acoustic fibre optics for leakage detection of water bottoms

Signal test for acoustic fibre

optics for leakage detection of

water bottoms

Signal test for acoustic fibre optics

for leakage detection of water

bottoms

elbt r

s

Wk

8

%

Contents

1

Management summary

2 Managementsamenvatting

3 Introduction

4 Fibre optics as a sensor

5 Experimental set-up

6 Processing of the DAS-data

a)

b)

7 Results signal test

a) No hole

b) Hole

c)

a)

b)

a)

b)

a)

b)

a)

b)

a)

b)

8 Discussion