GB2441552A - Measurement of attenuation in distributed optical fibre sensing - Google Patents

Abstract

Attenuation in the deployed fibre of a distributed optical fibre sensing system, such as a distributed temperature sensing system, is assessed by calculating a signal impairment function SIF based on a function calculated from measured Stokes and anti-Stokes backscatter signals and a constant B that gives a ratio of the Stokes and anti-Stokes signal a minimum sensitivity to temperature of the deployed fibre. This function is adjusted for the characteristics of a pristine optical fibre of the same type as the deployed fibre to give the SIF, which for a non-degraded fibre has a constant value of zero. A plot of SIF along the length of the fibre can then be analyzed for any deviations from zero, which indicate a change in the fibre attenuation. The resulting data can be used for qualitative and quantitative assessment of the fibre attenuation, and for correcting temperature profiles obtained from measured anti-Stokes signals.

Description

METHOD OF ASSESSING ATTENUATION iN A DISTRIBUTED OPTICAL

FIBER SENSING SYSTEM

BACKGROUND OF THE INVENTION

Field of the Invention

100011 The present invention relates to a method of assessing attenuation in a distributed optical fiber sensing system.

Description of Related Art

100021 Optical fibers used in distributed temperature sensing (DTS) systems are often deployed in harsh conditions, such as at high temperatures and in an aggressive chemical environment, such as within an oil well. Under these circumstances, the optical attenuation of the fiber can be degraded over time, which affects both the accuracy and resolution of DTS temperature measurements. Under extreme circumstances, the fiber can break.

100031 It is therefore desirable to have a technique for remotely monitoring the attenuation and integrity of the deployed fiber. It is useful if the monitoring can make usc of the optical sigvals employed by the DTS system to measure tempcrature so that temperature may be measured and fiber attenuation may be monitored simultaneously.

Also, the attenuation measurement should be unaffected by variations in the temperature of the fiber, both along its length and over time.

100041 All DTS systems should incorporate some way of compensating for the unavoidable attenuation of the optical fiber. Conveniently, this can be done using a reference measurement made at the same time as the temperature measurement For example, a Reman scattering-based DTS system may measure both the scattered anti-Stokes (temperature-sensitive) and Stokes (assumed to be temperature-insensitive) wavelengths. The temperature may be calculated from the ratio of the scattering at the -2--two wavelengths. The accuracy of the compensation depends on the difference in the attenuation at the Stokes and anti- Stokes wavelengths remaining unchanged.

100051 The Stokes wavelength scattering is only commonly used as a measure of fiber attenuation. If expressed in decibels (with reference to some arbitraiy level) it will decrease linearly with distance from the DTS system light source, if the fiber attenuation is constant Consequently, deviations from linearity indicate the presence of excess attenuation, either distributed over the length of the fiber, or, if the deviations take the form of steps, concentrated at specific points Such Stokes measurements are commonly employed as quality indicators for the accompanying temperature measurements derived from the anti-Stokes measurement Assuming the Stokes scattering is tnily temperature-independent, the actual attenuation (loss per unit distance) of the fiber is proportional to the slope of the plot of the scattering (in dB) with distance along the fiber. Hence, the measured curve may be differentiated numerically to provide an estimate of the variation of fiber attenuation with distance.

However, numerical differentiation of noisy data is known to exaggerate the effect of the noise, and such plots can be difficult to interpret as a result 100061 Also, the Stokes scattering is not, in fact, entirely independent of temperature.

As a result, temperature variations along the fiber may be misinterpreted as changes in attenuation.

(00071 It is possible to use a DTS system in double-ended mode, in which light is alternately launched into and detected from both ends of the fiber. As is well known, this allows temperature measurements to be compensated for changes in the fiber attenuation at both Stokes and anti-Stokes wavelengths. Since the Stokes scattering is measured from each end separately, the rate of change with distance of the ratio of the first end and second end Stokes scattering may be used as a measure of fiber attenuation that is almost insensitive to temperature. However, the technique requires access to both ends of the fiber, which may be inconvenient or impossible.

100081 Another method that may be used to determine attenuation is to measure the backscattered signal at the wavelength of the incident light launched into the fiber. -3--

Since this light has not undergone any Raman shift, the signal is tnily temperature-independent, and may be processed to give attenuation information in the same way as for conventional optical time domain reflectometry (OTDR). However, it has the disadvantage of requiring an additional optical receiver channel in the DTS system, or intemiption of the temperature measurement while the attenuation is measured.

BRIEF SUMMARY OF THE INVENTION

[00091 Accordingly, a first aspect of the present invention is directed to a method of assessing attenuation in a distributed optical fiber sensing system comprising: launching a pulse of probe light into an optical fiber deployed in a sensing environment; detecting backscattercd light returned from the optical fiber at a Raman Stokes wavelength to obtain a signal NIS(z), where z is distance along the optical fiber; detecting backscattered light returned from the optical fiber at a Raman anti-Stokes wavelength to obtain a signal TI(z); calculating a signal impairment function SIF(z) using ( (N7(z)' (77(z) log10 -B*1og10 1. 1 m(o)) (ris(o) SIF(z)=lO.aN.

aN-B.(aN+DLC) lO where B is a dimensionless constant having a value that gives the function K = lO.1og10(Nm(z))-B.lO.1og10(77S(z)) a minimum sensitivity to the temperature of the deployed optical fiber, a,, and aT are values of attenuation at the Stokes wavelength and the anti-Stokes wavelength in a pristine optical fiber, and DLC = (ar -aN); such signal impairment function having the property that for an ideal, pristine optical fiber SIF(z) =0 for all z; and analyzing a plot of SIF(z) for at least part of the deployed optical fiber to identify any deviations from S!F(z) = 0, such deviations indicating a difference in attenuation between the deployed optical fiber and the pristine optical fiber.

[00101 A suitable choice of the constant B in the determination of the signal impairment factor removes or greatly reduces the dependence on temperature of the Stokes signal N7, so that an assessment of attenuation free from the effects of temperature can be obtained. Even a very simple analysis of 8fF can yield valuable information about attenuation, and more detailed information including a quantitative measurement of the effective attenuation at the Stokes wavelength can be obtained from 8fF via straightforward processing or calculations. The method is advantageous in that it uses only signals that are already routinely measured in distributed sensing systems, so extra complexity, such as additional measurement channels or longer acquisition times, is avoided. Further, the method can be used with the prefewed single-ended distributed sensing systems, which are simpler to install and operate than double-ended systems, and in some cases are the only usable system. Also, the attenuation and temperature data is acquired simultaneously.

[00111 The pristine optical fiber may be a fiber of the same specification as the deployed optical fiber. The more closely matched are the characteristics of the two fibers, the more accurate will be the results of the method. However, fibers which are merely similar, rather than exactly the same, will still give useful results.

100121 The method may further comprise using information obtained by analyzing the plot of SJF(z) to detemiine a measurement of the effective attenuation in the deployed optical fiber at the Stokes wavelength.

BRIEF DESCRIPTION OF THE DRAWINGS

100131 For a better understanding of the invention and to show how the same may be carned into effect reference is now made by way of example to the accompanying drawings in which: [00141 Figure 1 shows a schematic representation of a simple distributed optical fiber sensing system to which the present invention may be applied; -5-- [00151 Figure 2 shows an example plot of a distiibuted temperature profile obtained from an optical fiber distributed sensing system such as that of Figure 1; (00161 Figure 3 shows a linear plot of measured backscattcred Stokes (NTh) and anti-Stokes (TTh) signals used to obtain the temperature profile of Figure 2; (00171 Figure 4 shows a logarithmic (decibel) plot of the NTh and 773' signals of Figure3; (00181 Figure 5 shows a plot of a temperature-independent function K' derived from the NTh' signal of Figure 4; and 100191 Figure 6 shows a plot of a signal impairment function SW derived from K' from Figure 3 in accordance with an embodiment of the invention, from which the state of the optical fiber of the distributed sensing system can be assessed.

DETAILED DESCRIPTION OF THE INVENTION

(0020J A distributed temperature sensing (DTS) system utilizes an optical fiber deployed in an environment of interest, such as down an oil well, to obtain a measurement of temperature along the length of the fiber. This is achieved by launching into a first end of the fiber a pulse of interrogating probe light, and detecting, at the same end, light which has undergone a Raman shift in the fiber and which is backscattered along the fiber. The Raman shifting produces light at a shorter wavelength than the probe light, called the anti-Stokes wavelength. The amount of light produced at the anti-Stokes wavelength is dependent on the temperature of the fiber at the point where the anti-Stokes light is produced, so that a measurement of the intensity of the returning anti-Stokes backscattering with time can be used to determine a profile of temperature with distance along the fiber. The Ramn shifting also produces Stokes light, at a longer wavelength than the probe wavelength; the intensity of the Stokes scattering is largely temperature-independent The backscattering at both the Stokes and the anti-Stokes wavelengths experiences attenuation as it propagates along the fiber; if ignored this will produce errors in the derived temperature profile. Hence, it is known to measure the temperature-independent Stokes backscatter and use it as a reference signal to compensate the temperature-dependent anti-Stokes backscatter for the effects of attenuation, as indicated in the introduction. If the attenuation of the fiber at the Stokes and anti-Stokes wavelengths is the same, the compensation will be exact 100211 In practice, the two attenuations will be different, and it is neccssaiy to correct for this difference when calculating the temperature, as is well known. If it is possible to access both ends of the installed fiber, the well-known double-ended technique can be used to make this correction. Alternatively, it is known to measure the actual difference in attenuation, either on a sample of the fiber, or on the installed fiber, before starting to make temperature measurements Provided the difference in attenuation remains constant over time, the correction will remain valid. Since this method requires access to only one end of the fiber, it is often preferred in practice.

100221 Figure 1 shows a schematic representation of a simple DTS system. The system 10 comprises an optical source 12 such as a laser operable to produce short pulses of light 14 at a probe wavelength. An input optical fiber 16 canies the pulses 14 from the optical source 12 to a first port of an optical circulator 18. The circulator 18 delivers the pulses 14 to its second port, to which is connected a launch end of an optical fiber 20 that is deployed in a sensing region, i.e. an environment of interest in which it is desired to make a temperature measurement. Each pulse propagates along the optical fiber 20, experiencing Ram2n shifting and backscatter at all points.

(00231 The backscattered light 22, comprising direct backscattered light at the probe wavelength in addition to the Stokes and anti-Stokes backscatter, returns to the launch end of the deployed optical fiber 20 from all points along the fiber, and enters the optical circulator 18 via its second port. The optical circulator 18 then delivers the backscattered light to an output fiber 23, which canies the light to a photodetector 24 for detection. The output fiber 23 is likely, in reality, to include various filters and amplifiers to enhance the optical signal and remove any wavelengths that are not of interest The photodetector 24 produces electrical signals representing the intensity of the light at the wavelength or wavelengths required for determining the temperature profile. Hence, the photodetector 24 will probably be a multi-channel device, operable to divide and detect the incoming light according to wavelength. The electrical signals are then sent to a processor 26, which perfonns signal processing to determine the temperature profile from the signals. The processor 26 may be a local processor, to produce the temperature information on site at the location of the launch end of the optical fiber 20. Alternatively, the signals may be transmitted in real time to a remote processor, or stored locally and transmitted to the remote processor at a later time.

100241 The backscattered Stokes light is not in actuality truly independent of temperature although it is much less temperature-dependent than the anti-Stokes backscatter. Hence, any attenuation information derived from the Stokes signal alone may be incorrect, as the Stokes signal may show variation that is caused by either attenuation or temperature, the two effects being indistinguishable. The present invention addresses this issue by using the more temperature-dependent anti-Stokes scattenng to cancel out the residual temperature dependence of the Stokes scattering.

The attenuation information carried by the Stokes signal can thereby be extracted more precisely and used to give an improved measurement of attenuation at the Stokes wavelength.

100251 As described above, it is necessary to compensate for the difference in fiber attenuation at the Stokes and anti-Stokes wavelengths when calculating fiber temperature. In a hostile environment, such as an oil well, the DTS sensing fiber may degrade over time, and show both distributed and localized increases in attenuation caused by the environment in which it is deployed, which may affect the Stokes and anti-Stokes wavelengths differently. Hence it may not be sufficient to compensate for attenuation using loss values known for the fiber when it was in a pristine condition before being installed for measurement The present invention provides measurements of attenuation at the Stokes wavelength during normal temperature measurement. By monitoring changes in the attenuation measurement, fiber degradation which may affect the accuracy of the temperature measurement can be detected. If a significant -8--change in attenuation is observed, the temperature measurement may be interrupted, and the degraded fiber may be replaced, or the fiber loss values at both Stokes and anti-Stokes wavelengths may be re-measured, using known techniques such as multi-wavelength optical time domain reflectometiy (OTDR).

100261 The operation of the invention is described by reference to a DTS system using Rmn scattering. As described above, such a DTS system launches a pulse of light into a deployed fiber, and measures the backscattered signals at two wavelengths corresponding to the maximum intensities of the Stokes and anti-Stokes scattering.

These are referred to as the non-temperature sensitive wavelength 2N, and the temperature-sensitive wavelength, respectively. The measurements of the intensity of the received backscattered light at the two wavelengths are denoted NTh and 77S respectively.

100271 Without loss of generality, it is assumed that the attenuation per unit length of the fiber at the two wavelengths are constant values of aN and aT (nepers/m).

[0028J It is known that the temperature of the fiber at any point may be determined from the ratio of the intensities of the Raman scattering wavelengths, 7/IN, at that point However, these intensities are not directly measurable. Instead, the N7S and 77S measurements are made at the location of the DTS (at the launch end of the optical fiber) and are subject to losses in the fiber at the two wavelengths over the propagation distance from the point of interest back to the launch end: rz, -exp(-a z) /N exp(-aN. z) (1) =1% .exp(-(aT-aN).z) where z is the distance along the fiber from the DTS launch end to the point of scattering (point of interest). In other words, in order to determine 1/IN from the measured ratio of uS and N75, a correction factor dependent on fiber length must be applied. The quantity (ar -aN) is the difference in attenuation per unit length of the fiber at the two wavelengths, expressed in nepe1/m. The equivalent in dB/km is referred to as the differential loss correction factor (DLC), which may be used to calculate the required correction as a function of length.

[00291 Several techniques for determining the DLC are known, of which the simplest and most widely applicable is to use a constant value, derived from calibration measurements on a sample of the temperature measurement fiber.

[00301 Both Stokes and anti-Stokes scattering increase with increasing temperature, and as indicated above, the anti-Stokes scattering is much more temperature sensitive, but the sensitivity of the Stokes scattering is not negligible.

[00311 To take account of this, the present invention calculates a signal impairment factor (SIF) based on a measured NTS signal that has been corrected for temperature dependence. From this, attenuation information that is not modified by any inherent temperature dependence can readily be extracted by a simple analysis.

2] Calculation of the 5fF begins by forming a combination of the measured NTh' and Th signals that has a minimum sensitivity to temperature. Considering the measurements over the length of the fiber, where z is the distance along the fiber, the combination used is: N7S(z) (2) 77(z)' where B is a constant that gives the desired minimum temperature insensitivity. It is convenient to calculate loss (attenuation), and also 5fF, in logarithmic units such as decibels. In decibels, the above expression becomes: K(z) =10* log10 (NIS(z))-B lO* log10 (77S(z)) dB (3) [0033] The value of B is chosen to make K have a minimum sensitivity to fiber temperature. The optimum value of B depends on, among other factors, the operating wavelengths of the DTS system, the composition of the optical fiber, and the temperature of the fiber. The temperature dependence is residual and minor, and a single value of B may be used over the typical temperature measuring range of the DTS system without significant deviation from the optimum.

100341 The optimum value of B may be found by measuring a length of pristine fiber, known to have unifomi attenuation along its length, different sections of which are maintained at different temperatures within the operating range of the DTS; calculating the SIF using a chosen value of B; and adjusting the value of B until the calculated SIF is as nearly as possible constant over the length of the fiber.

[00351 As an example, for a DTS system with operating wavelengths 2j = 1115 nm, 2= 1016 nm, a Ge02-doped graded-index multimode fiber, and a temperature measurement range of 0 C to 220 C, a value B =0.184 was found to be suitable. As described above, for other systems, the value of B may be found experimentally by measurement at a range of temperatures.

100361 Assuming constant loss, the value of K will decrease with distance along the fIber (increasing z), much like its component signals NTh and 775. Being a logarithmic function, the intrinsic exponential decrease in optical signal levels caused by the loss gives rise to a straight-line variation of K with distance, for constant fiber loss.

100371 Since a fraction of the 275 signal has been removed (by inclusion of the constant B that is less than unity), the value of K, in a fiber of constant attenuation, reduces more slowly with distance than either N75 or 77S. This is corrected using measured values of aN and DLC (or equivalently measured values of aN and ar) from a sample of pristine fiber (i.e. new fiber that has not to been subjected to the rigors of the measurement environment and therefore does not show a degraded performance in terms of attenuation) give an effective signal' K': K(z) K(z)=a. (4) N aN-B.(aN+DLC) [00381 In the case of a pristine fiber, the signal K' will decrease at exactly the same rate as the N7S signal for that fiber would, were the fiber at constant, uniform temperature. In effect, therefore, K' is N7S corrected for temperature. If the fiber attenuation increases at some point along an actual fiber, because the fiber is no longer pristine, the plot of K' for that fiber will deviate from the straight line corresponding to the pristine fiber from that point on. Hence any changes in attenuation that arise after a fiber is deployed can be determined from an analysis of K'(z), with the knowledge that any deviations in a plot of K (z) are due to attenuation changes only, and not temperature effects.

100391 Fiber degradation is known to cause an increase in loss at both AN and 2r above the pristine fiber level. For a DTS system based on Raman scattering, 2T is less than 2N, and both wavelengths are preferably chosen to be in the low-loss wavelength range, and not subject to excess absorption from hydrogen or hydroxyl contamination of the fiber. Consequently, the initial loss at Ar is always greater than that at Aw, i.e. DLC is always positive.

[00401 However, in the presence of fiber degradation, losses may increase by different aznOufltsatANafldAT. Provided the loss increase atAwis greaterthanli times the loss increase at 2r, the plot of K' with distance for a degraded fiber will have a steeper slope than for a pristine fiber. Since B is generally much less than I, and most commonly-encountered loss mechanisms cause similar loss increases at AN and Ar, this is almost always the case.

[0041J The final step in determining SIF is to remove the underlying linear trend coiresponding to the case of a pristine fiber. This represents the expected or normal losses for the fiber; any degradation which increases losses will cause SIF to fall below its value at distance zero. Since this value is arbitrary, it is subtracted from the calculated SIF so that the value of SIF is 0 dB at distance z =0. Since subtraction of the logarithm of a quantity is the same as division by that quantity, the expression for SIP can be written as: (N7S(z) (77S(z) log10 -Blog10 Im(o)J (jm(o) SIF(z)=lO.aN. .. (5) -B.(a,, + DLC) lO [0042J In this expression, the units of z are meters, and the units of a,, and DLC are dB/km. These are the conventionally used units for these quantities.

100431 SIF represents the amount by which an ideal' temperature-independent NTS signal from a deployed fiber would fall short of its expected level at any point along a uniform, pristine fiber with attenuation aN. A pristine fiber would give 5fF = 0 dB at all points (i.e. for all z) -a horizontal line. Thus, any deviations from the 0 dB straight line indicate a change in attenuation since the fiber was deployed, and the nature of a deviation (step change, change in slope) yields information about the type of degradation that has caused the attenuation change.

(00441 Plotted out, SJF will generally be a non-increasing function of distance.

Downward steps in SJF correspond to point losses such as connectors, while downward-sloping regions correspond to fiber sections of higher loss. Also possible are upward or downward steps, occurring at the junction of fibers of slightly different characteristics. The locations of such junctions will in general be known, and their effects may therefore be ignored.

[00451 Slight mismatches between the assumed values of aN and DLC for the pristine fiber, and the actual values for the deployed fiber in its un-degraded state may cause a slight upward or downward slope to be present. Generally, any such slope is much less than the effects of fiber degradation, and may easily be ignored. The best results will be obtained if the pristine fiber and the actual fiber used for deployment are as closely matched as possible. Preferably, therefore the pristine fiber and the deployed fiber have an identical specification. However, similar specifications of fiber will still give useful results.

(00461 Once calculated, 5fF can be plotted out over part or all of the fiber length (some or all values of z), and analyzed to reveal any deviations from the horizontal line at 0 dB. These deviations correspond to changes in attenuation that have occurred to the fiber during or after installation in the sensing region, in other words, that are caused by damage to or degradation of the fiber. This altered attenuation should be taken into account when obtRining distributed temperature pmtiles from 775' measurements, or any other measurements derived from detected light returning from the fiber. Also, the assessment of attenuation can be used to determine when a fiber has become so degraded that it is no longer fit for purpose and should be replaced.

(0047) SIF plots may be used to detect lengths of degraded fiber, by looking for regions of the plot that slope downwards from the horizontal, where the steepness of the slope depends on the degree of attenuation and hence the amount of degradation.

The slope can be monitored over time to assess the rate of degradation. Steps in a SIF plot correspond to point losses at features such as connectors, splices, bends, and pressure seals, so can be identified. The position of known components can be taken into account when analyzing the plot, and the corresponding step changes ignored.

However, a newly appearing step change indicates a recent development such as localized damage to the fiber, which can then be investigated if necessaly. An increase in the size of an existing known step may indicate a deterioration in or problem with the corresponding component which can be similarly investigated. A breakage in the fiber can be identified by looking for a reduction in the SIP level with time beyond the point of the breakage.

100481 The deviations in the SIF plot can be interpreted qualitatively, to obtain the location and possible nature of a drop in the transmission quality of the fiber. Such an event can be investigated immediately, or the location monitored over time, and further changes to the SIP profile can be assessed to determine when an investigation becomes appropriate.

(00491 Alternatively or additionally, the deviations can be interpreted quantitatively to obtain a measure of the changed attenuation over all or part of the fiber. This information can then be used to compensate temperature measurements obtained from a contemporaneous 77Y signal, by removing the effects of attenuation from the measured 2'Th' intensity. Such quantitative interpretation depends on detailed knowledge of the fiber degradation mechanism in the specific environment in which it is deployed, so that changes in loss at the anti-Stokes wavelength may be calculated from the changes in loss at the Stokes wavelength, as measured by the present invention.

100501 Returning to Figure 1, the calculations required to determine SIF(z) can be carried out by the processor 26 of the DTS system 10, as can the subsequent analysis to identify deviations from the 0 dB level and the possible numeiical calculations to determine the value of the attenuation. This gives an in situ assessment of the state of the deployed fiber. Further, if the processor 26 is also operable to determine the temperature profile from a measurement of the I7S signal, and provided a detailed model of the fiber degradation mechanism is known, the processor 26 can fwther apply the results of the attenuation calculation to compensate the temperature measurement for attenuation. Alternatively, all or some of these functions can be canied out remote from the DTS system, either in real time if the relevant 77S and NTh' data is transmitted from the DTS system to a remote processor as it is collected, or at a later time if the data is stored in memory comprised within the DTS system and transferred to a remote processor after some delay, perhaps at periodic intervals. 100511 In any arrangement, the processing can be implemented using

hardware software or a combination of the two. The processor may be linked to a visual display unit that it can instruct to display a plot of the SIF, so that a visual inspection and inteipretation can be performed by a user. This may supplement or replace any automated analysis of the SIF data carried out by the processor.

(00521 Figures 2-6 show graphs of various parameters to illustrate the operation of the invention by way of example only.

100531 Figure 2 shows a plot of an example DTS system temperature measurement, shown as temperature (vertical axis) against distance along the fiber (horizontal axis).

A poition of the fiber (from about 2850 m from the DTS system end onwards) is at a high temperature. In this example, the high temperature corresponds to the interior of a steam injection well. The large step change in temperature at the wdllhead, where the fiber enters the well, may induce excess fiber loss through the required pressure seal.

It is desirable to be able to assess this loss to determine whether the fiber is degraded excessively in this region.

[00541 Figure 3 shows plots of the measured NTh' and 77 signals used to create the temperature profile of Figure 2. The curve 30 is the N7S signal and the curve 32 is the iTS signal, both shown as signal level in aEbitraiy linear units (vertical axis) against distance along the fiber (horizontal axis). Both signals show a step change at the location of the wellhead, although that for the 7Th' signal is much greater. Both NTh and 773' signals are known to increase with temperature, so the step changes are expected. However, the veiy small increase in the NTh' signal may indicate that there is loss (attenuation) in the fiber at the welihead location which is making the temperature-induced signal increases smaller than they would otherwise be.

f0055J Figure 4 shows plots of the same measured P/73' and 773' signals shown in Figure 3, but on a decibel scale. Hence the curves 30 (Ni'S) and 32 (TTh) are plotted as signal level in dB (vertical axis) against distance along the fiber (horizontal axis).

The conversion to decibels produces a plot of NTh' that decreases almost linearly with distance up to the location of the wellhead. Beyond the wellhead, the decrease becomes much steeper, which indicates the presence of additional losses in the hot portion of the fiber.

(00561 Figure 5 shows a plot of the scaled K' signal calculated from the decibel versions of the P173' and 773' signals of Figure 4, using Equation (4); this is the line 34.

Also shown for comparison is the decibel plot of the P173' signal 30 from Figure 4, and the theoretical expected Ni'S signal from a pristine fiber (i.e. no additional attenuation from degradation) at constant temperature, shown as line 36. All plots are shown as a signal level in dB (vertical axis) against distance along the fiber (horizontal axis). The K' plot 34 has been moved vertically to have the same value as the Ni'S signal 30 at the start of the fiber (distance = 0). From K', which is temperature-independent; a point loss at the welihead, caused by the pressure seal, is clearly visible as a vertical drop in K', as is the additional loss beyond, indicated by the large downward deviation of K' from the slope prior to the wellhead.

[0057J Figure 6 shows a plot of SIP calculated from the K' plot of Figure 5, as a value in dB (vertical axis) against distance along the fiber (horizontal axis). The virtually pristine condition of the initial part of the fiber is readily apparent, since SIF deviates little from the 0 dB line up to the wellhead location. At this point, the step loss caused by the pressure seal is clearly distinguishable, and beyond the pressure seal, an increasing additional attenuation (beyond the inherent attenuation of the fiber when in pristine condition) is indicated by the downward slope of the curve away from the 0 dB line.

[00581 This demonstrates that a qualitative assessment of the state of a deployed optical fiber can be immediately and readily made by a simple inspection of a plot of SIF(z) to identify any deviations from the 0 dB horizontal line. A visual inspection could be carried out, but a computer assessment is straightfoiward to implement, such as the use ofa simple algorithm to search for values of 5fF that differ from 0. This qualitative information can be used to monitor the state of the fiber over time, allowing problems with the condition of the fiber or its integral components to be addressed promptly. Further, the amount by which a 5fF plot is found to deviate from o can be used to determine a quantitative measure of the fiber attenuation. If a detailed model of the fiber degradation mechanism is known, this can be used to compensate 77 measurements for attenuation, to improve the accuracy of temperature profiles. Also, numerical values can be monitored over the long term to identify when a fiber performance has degraded beyond a pre-determined acceptable level.

100591 Although the invention has been described with respect to a distributed temperature sensing system, it may also be utilized in conjunction with distributed optical fiber sensing systems directed to the measurement or monitoring of other parameters. Fiber attenuation and degradation is of interest and concern in virtually any fiber sensing system, so the ability of the method of the present invention to provide qualitative and quantitative information about the state of the fiber is widely applicable.

Claims (5)

1. -17--

   What is claimed is: 1. A method of assessing attenuation in a distributed optical fiber sensing system comprising: launching a pulse of probe light into an optical fiber deployed in a sensing environment; detecting backscattered light returned from the optical fiber at a Raman Stokes wavelength to obtain a signal NTS(z), where z is distance along the optical fiber, detecting backscattered light returned from the optical fiber at a Raman anti-Stokes wavelength to obtain a signal 77S(z); calculating a signal impairment function SIF(z) using ( (Ni(z) (flS(z) ilog i i-B*log I oyi(o) S1F(z)=10.aN. ±aN -B. (aN + DLC) iO where B is a dimensionless constant having a value that gives the function K =10. log10 (JV7S(z)) -B* 1O log10 (77(z)) a minimum sensitivity to the temperature of the deployed optical fiber, aN and ar are values of attenuation at the Stokes wavelength and the anti-Stokes wavelength in a pristine optical fiber, and DLC = (ar -aN); such signal impairment function having the property that for an ideal, pristine optical fiber SIF(z) =0 for all z; and analyzing a plot of SIF(z) for at least part of the deployed optical fiber to identify any deviations from SIF(z) = 0, such deviations indicating a difference in attenuation between the deployed optical fiber and the pristine optical fiber.
2. 2. A method according to claim 1, in which the pristine optical fiber is a fiber of the same specification as the deployed optical fiber.

   -18--
3. 3. A method according to claim!, further comprising using information obtained by analyzing the plot of SIF(z) to determine a measurement of the effective attenuation in the deployed optical fiber at the Stokes wavelength.
4. 4. A method according to claim 3, wherein the distributed optical fiber sensing system measures temperature, and further comprising using the measurement of effective attenuation in the deployed optical fiber to compensate the temperature measurement for attenuation.
5. 5. A method according to claim 3, further comprising plotting S1F(z) at a second time and comparing such plot with the plot of SIF(z) to monitor the state of the optical fiber over time.