Effect of hydrogen gas on FBG-based optical fiber sensors for downhole pressure and temperature monitoring

: The influence of hydrogen gas on Fiber Bragg Grating (FBG)-based optical fiber sensors has been validated experimentally. More in particular, the focus was on FBGs written in the so-called Butterfly Micro Structured Fiber that targets simultaneous pressure and temperature monitoring with a minimum in cross-sensitivity to be used in, for example, downhole applications for the oil and gas market. The hydrogen-induced pressure and temperature errors from this type of sensor have been quantified as a function of the partial hydrogen pressure. The induced errors can be related to the diffusion of the hydrogen into the microstructure and to refractive index changes due to the presence of the hydrogen in the micro holes and penetration of it into the fiberglass. Furthermore, we have also shown that the hydrogen-induced errors scale with the partial hydrogen pressure.


Introduction
Fiber optic sensing has demonstrated its benefits for sensing of different critical parameters in downhole applications [1][2][3][4][5].With respect to downhole temperature monitoring, Distributed Temperature Sensing (DTS) is a well-known intensity-based detection method within this industry [3,6,7].DTS allows picturing the well temperature as a function of depth down to several kilometers with meter sized resolution.The technique relies on the encoding of temperature information contained in the Raman back scattered signal (Stokes and Anti-Stokes components).Another method relies on the use of Fiber Bragg Gratings (FBG) [4,8,9].This technology yields quasi-distributed sensing: FBGs with different Bragg wavelengths can be inscribed along the same sensing fiber but the number of FBGs that can be multiplexed is rather limited when compared to DTS.Each Bragg resonance is intrinsically sensitive to temperature and strain and thus by eliminating strain effects, they can be equally employed to capture temperature profiles down an oil well.
A common characteristic for downhole environments is the presence of hydrogen [10].Intensity-based sensing systems, like DTS, are affected by hydrogen as it diffuses into the optical fiber and thereby induces photo-darkening which increases the optical attenuation in the sensing fiber [5,11].Researchers have proposed different mitigation techniques, such as reducing the penetration rate of hydrogen by means of a hermetic fiber coating [12] or by using self-calibrating methods involving either dual wavelength scanning or exploiting two sensing fibers [13][14][15].For FBG wavelength-based sensing systems, the attenuation itself is less of an issu [16].This dri silica and the the temperatu More rece have also bee The effect of in an addition [25,27,28].Th bar of hydrog shift of ~100 Recently, More particul fibers for sim such specialty reported in th pressure moni based sensors diffusion attri the disappeara reason for the load the senso induced error hydrogen indu pressure.Bas hydrogen on quantified w environment.applications, e

Gas load
To create a commonly us inscription [32 ue but the pres ft of the Bragg formation of G ure [17][18][19][20][21][22].ently, FBG-bas en considered the presence o nal wavelengt he induced wa gen pressure at °C, considerin we have also larly, the usage multaneous mea y sensors when his document.itoring with the s.The modific ibuted to the m ance of pressu e apparent pres ors with hydrog rs for each se uced effects ha sed on the exp MS-FBG sen ith respect to With that, th especially to th ental set-up ding facility Bragg wavelen e refractive ind en pressure lev optical fiber (M ole applications studied [25,26] in the micro m after exposu a fictitious tem nd 10 pm/°C.downhole app utterfly micro-s 9-31].The be was studied an nce of tempera other conventio glass due to h ft.For pressure ppears to be the utoclave facilit a and the analy ogen.In sectio e with partial h ection, the infl induced erro ng temperatur G sensor in d aluated. nment.ng autoclave d) fibers prior t r with an inner ngth shift dex of the vel and on MS-FBG) s [23,24].] to result structure ure to 200 mperature plications.structured havior of nd will be ature and onal FBG hydrogen e sensing, e primary ty used to ysis of the on 4, the hydrogen fluence of or can be re in the downhole which is to grating cavity of 20 cm in height and 13 cm diameter.It is equipped with four epoxy-based fiber feed-throughs with FC-connectors, which allow for online monitoring of samples during gas loading.The temperature of the autoclave can be controlled by an external oven.The epoxy-based fiber feed-throughs limit the maximum operating pressure and temperature to 80 bar and 80 °C.The temperature and pressure in the autoclave are continuously monitored with reference gauges.Figure 1 shows a schematic arrangement of the autoclave set-up.

FBG-based optical fiber sensors
We experimented with three types of FBG-based sensors.The first is an FBG in a Polyimide coated pure silica fiber written with a femtosecond (fs) laser by means of a phase mask.This sample will further be referred to as the Femto Second Grating or 'FSG'.The second is a Draw Tower Grating or 'DTG' with Ormocer coating written with a UV laser using a Talbot interferometer.The sample was provided by FBGS International.The third is an FBG in the Butterfly Micro-Structured fiber (MS-FBG) written again with a fs-laser and phase mask.A typical reflection spectrum of the Butterfly MS-FBG sensor features two Bragg resonances due to the birefringent nature of this fiber.One peak corresponds to light polarized along the fast axis, whilst the other to light polarized along the slow axis.More information about the typical reflection spectrum of the Butterfly MS-FBG sensor and the cross section image of the Butterfly MSF can be found in [29][30][31].The unique ability of this sensor is that pressure changes appear as changes in the peak separation (i.e. the wavelength difference between the fast and slow axes) while temperature can be monitored by tracking the change of the individual Bragg wavelengths.Since temperature appears as common mode for both Bragg wavelengths, it becomes cancelled in the peak separation and in this way, most of the pressure-temperature cross-sensitivity can be eliminated [29][30][31].We used two such MS-FBG sensors.The first has its fiber end sealed by collapsing the micro-holes with an electrical arc of a fusion splicer in order to be sensitive to environmental pressure.Since this sealing process is done under atmospheric pressure, sealed MS-FBG sensors will sense the differential pressure i.e. the difference in pressure with respect to the atmospheric pressure that is present in the micro-holes.The other sample has its fiber end untreated and so with open micro-holes.This sample is not sensitive to pressure changes because the pressure inside the micro holes will always be the same as the pressure outside.The Bragg resonances of each FBG sensor are tracked using a Micron Optics SM125-500 interrogator combined with an HP 11896A polarization scrambler.All sensors are spliced to a FC/APC pigtail with a lead-in cable spooled on a metal rim.The sensing region of the MS-FBG sensors was kept straight by taping it before and after the FBG to the inner wall of the autoclave.

Nitrogen loading
The aim of this first experiment was primarily to verify whether both the autoclave and the sealed MS-FBG sensor are properly sealed.We examine this by loading nitrogen gas into the autoclave since this is not explosive.The experiment was carried out in the following stages: (1) pressurization of the autoclave with nitrogen from atmospheric pressure to 20 bar and then to 80 bar sequentially, (2) heating up the autoclave to a temperature of 80 °C with a holding time of a few hours, (3) leave the autoclave to cool down to room temperature (RT) and (4) release the pressure in the autoclave down to atmospheric pressure.Notice that the heating step also causes the pressure to increase since the autoclave should be regarded as a closed volume.
Figure 2(a) shows the evolution of the Bragg resonances of the open MS-FBG sensor during nitrogen loading, together with the reference pressure and temperature readings.The recorded wavelength shifts for both the slow and fast axes have been normalized for clarity.The figure also indicates the different stages (1) to (4) explained above.A few transients in the wavelength shifts for both Bragg resonances can be observed during stage (1), which actually correspond to the moments when the pressure is increased in the autoclave.The sudden change in pressure in the closed volume of the autoclave induces a sudden change in temperature, which quickly dissipates again over time.Since the individual wavelengths are sensitive to temperature changes, this explains the transients.Apart from that, both Bragg resonances shift first to longer (shorter) wavelength as the nitrogen pressure increases (decreases) during the warming up (cooling down) period and we can see a good correspondence between the wavelength shift and the pressure reading.However, the magnitude of the shift for the fast axis Bragg resonance is larger than that for the slow axis.The wavelength correspondence to pressure and the change in peak separation can both be related to the change in refractive index that occurs due to the presence of the pressurized nitrogen gas in the micro holes.As the pressure increases, the refractive index will change accordingly.This also causes the Bragg wavelength to increase since the light from both modes overlaps partly with the air holes.However, the light guided along the fast-axis has more overlap with the micro-holes and therefore it experiences a larger change in effective refractive index (and wavelength) compared to the slow-axis and hence there is a net change in peak separation: a pressure increase corresponds to a drop in peak separation and vice versa, see Fig. 3.The peak separation decreases with 42 pm with a pressure increase of 80 bar and returns to its initial value when the autoclave pressure has been released.As stated before, this initial loading test with nitrogen was primarily intended as a sealing check of the autoclave and sealed senor.But it also nicely illustrates the effect of the gas in the microholes on the refractive index and wavelengths.This effect will also be present during the hydrogen loading tests, see next section.
Similarly, Fig. 2(b) shows the evolution of the Bragg resonances of the sealed MS-FBG sensor.Both Bragg resonances respond to pressure in the way it is understood for sealed samples of this fiber type [29,30]: the slow axis Bragg resonance features a positive and larger pressure sensitivity, whilst the fast axis Bragg resonance has a negative and lower pressure sensitivity during stages (1) and ( 2).The sealed MS-FBG sensor operates as an actual pressure sensor as expected.The sensing functionality of the sealed MS-FBG becomes obvious when considering the peak separation as shown in Fig. 3, which follows the reference pressure reading.The influence of temperature changes on the pressure measurements is negligible since both Bragg resonances experience the temperature variation as a common mode effect that is cancelled out by taking the difference.We obtain a pressure sensitivity of the sealed MS-FBG sensor of 2.8 pm/bar (a total peak separation change of 229 pm over a pressure difference of 81.7 bar).This proves that the sealed sensor is indeed properly sealed.
The FSG and DTG samples behave similarly to standard FBG based sensors during pressure cycling.At stage (1), a typical pressure sensitivity of around −0.2 pm/bar was observed.Afterward, the wavelength shifts are mainly dominated by the change in temperature.The net wavelength changes are around + 0.746 nm and + 0.651 nm for the DTG and FSG sensor respectively.
The findings of the nitrogen loading tests can be summarized as follows.The sealed MS-FBG acts as a pressure sensor as the change of hydrostatic pressure can be encoded into the change of peak separation and a positive pressure sensitivity in peak separation was observed.On the other hand, the open MS-FBG sensor is acting as a refractive index sensor since the pressure changes will predominantly cause the refractive index of the nitrogen in the micro holes to change and this in turns induces wavelength shifts in the light guided in the fast and slow axes.Due to the difference in overlap, the effect is larger in the fast axis and this causes a net negative change in peak separation.

Hydroge
After the ini hydrogen load since it is kn conducted as monitoring th autoclave up t in step), (3) heating of th completely sta   ), the sealed M of hydrogen is side.This is re on to −0.

Diffusin
In stage (5), t process.Figu The hydrogen as the autocla necessary for temperature c expected that effect.2).These equations will be used to calculate the pressure and temperature errors by comparing the calculated values with the reference measurements.
For the open MS-FBG sensor, the above equations and coefficients are still used to calculate the corresponding temperature and pressure.The wavelength changes of the open MS-FBG sensor are mainly originating from the refractive index modifications when hydrogen is present in the air holes and in the fiber glass.The open MS-FBG sensor can quickly reveal the response of both Bragg wavelengths when hydrogen diffuses through the open channels.Doing so yields the presented in Fig. 10(a).The temperature error for the open MS-FBG sensor at 80 °C and 100 bar of hydrogen is around 74 °C, which comes purely from the refractive index changes from the hydrogen in the MS-FBG.Similarly, the refractive index induced drop in peak separation leads to an apparent pressure drop of −4.6 bar, causing the overall pressure error to increase from 100 to 104.6 bar.
For the sealed MS-FBG sensor, the influence on temperature monitoring is rather limited in the beginning when the hydrogen diffusion rate is low, as shown in Fig. 10(b).The temperature difference however starts to grow when the hydrogen starts to diffuse into the air holes and through the glass, yielding eventually to a total induced temperature error of around 75 °C.We thus find good agreement between the two MS-FBG sensors at the end of the hydrogen diffusing in step.In terms of pressure monitoring, the pressure reading in the sealed MS-FBG sensor gradually drops to zero as the hydrogen diffuses into the air holes.One could expect that the pressure difference becomes zero from the moment when the hydrogen pressure in the air holes is the same as in the autoclave.But due to the difference in effective refractive index induced by the hydrogen in the air holes and glass, the calculated pressure for the sealed MS-FBG sensor will be approximately 4 bar lower.Again, good agreement between open and sealed MS-FBG sensors could be found.
When the hydrogen is removed from the autoclave, the sealed sample initially gives a similar temperature difference as just before the hydrogen removal.The pressure error becomes negative because the pressure of the trapped hydrogen is larger than the pressure in the autoclave and hence the net pressure difference is negative.During the out-diffusion, both the temperature as well as the pressure error gradually restore to zero.
For the conventional FBG sensors, the response of the temperature changes can be approximated to the response of the wavelength variations with a linear relationship.The used linear coefficients are 10.25 pm/°C and 11.6 pm/°C for the FSG and for the DTG sensor, respectively.Figure 11 shows the calculated temperature during the hydrogen loading test for the FSG and DTG.Due to the diffusion of hydrogen, the wavelength shifts will be translated into apparent temperature shifts.Eventually, the temperature differences are around 58 °C and 49 °C for the FSG and DTG sensor, respectively.These apparent temperature shifts will vanish again once all hydrogen has diffused out from the fiber.
In these initial experiments, the pressure and temperature errors have been quantified under a pure hydrogen gas environment.In the next step, we will study the influence on the same set of FBG sensors in case of hydrogen-nitrogen gas mixtures.

Experime
For gas mixtu sum of the in are concerne hydrogen pre experimented MS-FBG sen used the FSG

Nitrogen
We follow a two gases.(    ensor, the xture has f the order s starts to leased.A nsor when nds to the hydrogen nsors was downhole particular ured fiber st to pure ples were .For the sors was rs coming from the hydrogen could be quantified.From the pure hydrogen loading tests, it was found that the responses from the sealed samples evolved towards those of the open samples and eventually, their response was found to be identical.This indicated that the hydrogen diffuses through the glass into the microstructure for the sealed samples.Because the diffusion time is different for the closed compared to the open samples, we could clearly separate the two main mechanism behind the hydrogen induced errors: (1) the first being the disappearance of the (partial) hydrogen pressure from the pressure reading due to the diffusion of the hydrogen through the glass into the microstructure and (2) the second being a change of the individual Bragg wavelengths and also of the peak separation due to a change in the refractive index originating from the hydrogen entering the micro structure and penetrating the fiber glass.The changes in the individual wavelengths are causing apparent temperature shifts whereas a change in the peak separation causes an apparent pressure shift.
The tests with the hydrogen-nitrogen mixture was done to confirm that the driving parameter behind this process is the partial hydrogen pressure.This could be confirmed and it indicates basically that the size of the hydrogen induced errors scale almost linearly with the partial hydrogen pressure.Although the pressure and temperature errors shown in this work may seem relatively large, it should be kept in mind that these are for relatively large partial hydrogen pressures (80 bar and 20 bar).In practice however, the hydrogen level in a downhole environment is typically a few order of magnitude smaller.Therefore, the hydrogen induced errors will scale accordingly and therefore can be expected to be much smaller.The adaptability of the MS-FBG sensor for temperature and pressure monitoring in downhole applications is feasible if the hydrogen level is of the order of the present work.Some mitigation techniques would be required for MS-FBG sensor if the hydrogen induced error is above the required measurement accuracy.

Figure 4
Figure 4 show for the open M both Bragg atmospheric p Fig. 5 FBG s Figure 6 s stages (1) to ( pressure with slight increase room temper beginning of and the hydr wavelength sh Note that the loading.We diffusion in se necessary neit hydrogen pre 0.594 nm at r shifts are com Fig. 6 hydro Fig. 7 sensor Similarly, the out-diffus complementar out of the mic in a decrease almost 150 ho sealed MS-FB Fig. 1 during Fig. 1 MS-F The result pressure error stage (2).Fro −1.4 bar but a diffusion is n autoclave to hydrogen and temperature is the continuati approximately total of 27.3 from the sens with 20 bar of (4) is n gas is still al 60 bar of ni rom 20°C to 48 change of the xt 144 hours b ences for the open n gas mixture.