GB2243908A - Distributed fibre optic sensor - Google Patents

Abstract

A sensor comprises two light sources ( lambda 1 lambda 2) of different frequencies, transmitting a series of short pulses along a monomode birefringent optical fibre (3). Each pulse possesses two polarization eigenmodes whose relative velocity depends upon an external parameter to be measured. Light which is backscattered from the fibre enters the optical system (5, 6, 7) causing the components of the light to optically interfere on a square low photodetector (8). An integrator (9) and a frequency meter (1) allow the distribution of the parameter along the fibre to be displayed on a screen (11). The external parameter may be temperature, strain, pressure or an electric or magnetic field. <IMAGE>

Description

OPTICAL FIBRE POINT AND DISTRIBUTED MEASUREMENT APPARATUS AND METHOD This invention relates to the measurement of parameters (measurands) using optical fibres.

Optical-fibre methods of measurement sensing offer many important advantages for industrial use: the fibre is a flexibly, insulating, dielectric medium which can readily be installed in industrial plant without significant disturbance of the measurement environment; the range of measurands which is accessible to measurement by opticalfibre techniques is very large, since the propagation'of light within an optical fibre is sensitive to a wide variety of physical influences external to it.

Optical-fibre distributed measurement sensing is a technique which utilizes the one-dimensional nature of the optical fibre as a distinct measurement feature. It is possible, in principle, to determine the value of a wanted measurand continuously as a function of position along the length of a suitably-configured optical fibre, with arbitrarily large spatial resolution. The normal temporal variation of the distribution is determined simultaneously.

Such sensing systems are normally referred to a fullydistributed systems, to distinguish them from the quasi distributed systems, which possess the capability of sensing the measurand only at a number of discrete, predetermined points.

The fully-distributed facility opens up an enormous number of possibilities for industrial application. For example, it would allow the spatial and temporal strain distributions in large critical structures such as multi storey bui buildings, c;i bridges, dams, aircraft, pressure vessels, electrical generators, etc. to be monitored continuously. It would allow the temperature distributions in boilers, power transformers, power cables, aerofoils, office blocks, etc. to be determined, and thus heat flows to be computed. Electric and magnetic field distributions could be mapped in space so that electromagnetic design problems would be eased and sources of electromagnetic interference would be quickly identifiable.

There are two important definable reasons for requiring to obtain the information afforded by distributed optical fibre measurement sensors. The first is that of providing continuous monitoring so as to obtain advance warning of any potentially damaging condition in a structure, and thus to allow alleviative action to be taken in good time. The second is that this spatial and temporal information allows a much deeper understanding of the behaviour of large (or even quite small) structures, with many implications for improvements in their basic design.

Conventional industrial measurement sensor technology does not provide this facility. When measurand distributions of any kind are vital in a given situation the solution usually is to festoon the structure with a multitude thermocouples, or strain gauges, or whatever. This then presents problems of multiplexing, logging and calibration and, in any case, relies on the choice of positions for each of the many sensors being the correct one - a choice which cannot properly be made without a prior knowledge of the very distribution one is seeking to measure. This solution is thus expensive, tedious and usually broadly inadequate.

The optical fibre can be readily installed in industrial plant (retrospectively if necessary) produces minimal disturbance of the measurement environment, is cheap, passive and electrically insulating, acts as its own telemetering channel, can easily be re-arranged in accordance with acquired knowledge and allows a choice of any or all measurement points along its length within the limits of the spatial resolution interval. If such technique can be made to work satisfactorily for a number of measurands, a new dimension appears in the field of industrial measurement.

The value of the measurand is determined by allowing it to affect the light propagation properties of a fibre in some way and thus causing it to affect one or more of the characterising properties of light propagating in the fibre, such as its amplitude, intensity, phase, polarisation state or frequency. One of the difficulties faced in trying to design sensor systems based on these ideas is that of ensuring that the observed change in the light parameter is dependent only on the measurand and not on one or more of a variety of other unwanted effects; this difficulty is often met by introducing some degree of referencing with respect to a i light s i signal which is influenced by the unwanted effects, but not by the measurand (eg. light at a different wavelength).

We have devised a method which provides automatic referencing, at the same time as providing an indicating signal in a convenient, accurately-measurable form. It can also be used in a configuration which allows the measurement processing to be performed in a much longer time than is normal for distributed optical fibre measurement systems, and by some orders of magnitude.

According to the present invention, there is provided a method of measuring a measurand comprising: subjecting a region of birefringent monomode optical fibre to the influence of said measurand; launching radiation into said fibre; causing two polarization eigenmode components of radiation scattered from said region to interfere; deriving an electrical signal representative of the interference frequency between those two components; and determining, from the time dependence of the frequency of the resulting derived signal, the spatial distribution of said measurand, in which said radiation is so launched into the fibre at two differing launch frequencies simultaneously.

In order to promote a fuller understanding of the above, and other aspects of the invention, the background of the invention will now be described, and some embodiments will also be described, by way of example only, with reference to the accompanying drawings in which: Figure 1 shows a schematic diagram of an optical fibre system for measuring a measurand, and Figure 2 shows a similar diagram of an embodiment of the invention.

Highly birefringent fibre, known as hi-bi' fibre, is available commercially, and one form of such fibre allows the propagation of two orthogonal linearly polarised eigenmodes with significantly differing respective velocities. The birefringence properties are normally characterised in terms of the fibre length ('beat length") over which a phase difference of 2 w is introduced between the modes for a specified optical wavelength. Beat lengths of less than lmm are presently available, for 633nm wavelength radiation.

Suppose that a short optical pulse is launched into this fibre, with equal energies in each of the two eigenmodes.

Provided that the polarisation effects of all external agencies are small compared with the intrinsic polarisation properties of the fibre (and this will almost certainly be so for beat lengths as small as lmm) there will be negligible coupling of energy between the modes, even over several hundred metres of fibre. Backscattered radiation from the propagating pulse will return to the launch end of the fibre, and its emerging polarisation state will vary, with a period of the order of the time which light takes to traverse one half beat length, from one to ten picoseconds, say. This is too rapid a variation for practical detection and any detector with response time significantly greater than this would interpret the light as essentially unpol ari sed.

The physical basis of this invention is to allow the measurand to vary the birefringence, and thus the relative velocity, of the two modes. Provided that the quantitative relationship between the measurand and the birefringence is known, a mapping of the velocity difference as a function of position along the fibre allows the corresponding distribution of the measurand to be determined.

To do this we allow the two modes, for example, re-emerging at the launch end of the fibre after backscatter, to interfere optically, on the surface of a "square-law" photodetector. (This will require that one of the modes first has its polarisation direction rotated through 900, or that each is resolved into a common direction at +450 to the other). We may write the electric fields for the two modes in the form: E1 = E0 cos (wt + l) (la) E2= Eo cos (wt + +2) (ib) (where o is the angular frequency of the optical wave, E0 is the peak amplitude of the fields and #1, #2 are the phase angles of the modes).

Now for a distance s along the fibre we have for constant velocity, c; # = ws (Taking # = 0 at s = 0) c For variable c we must write, however:

where +(s) is the function of phase angle with distance s along the fibre and c(s) is the function of velocity with distance along the fibre.

Hence we have for the two modes:

Since the two modes possess different velocities c11c2, in "hi-bi" fibre.

Now it may be arranged that the external measurand varies either one or both of these velocities in a deterministic way. The action, for example, of pressure, strain, temperature, electric field, magnetic field via the elastooptic thermo-optic, electro-optic and magneto-optic effects on the refractive index of a material medium are well known. With or without the action of a measurand the difference between c1 and c2 will, however, always be small.

Thus we may write: c1(s) = c - v(s) 2 v(s) c c2(s) = c + v(s) 2 where v is a velocity difference parameter.

Thus from equa tion (2) we see that:

If we now consider the phase difference, A+, between the two re-emerging modes after backscatter from a point s in the fibre, then we must multiply by two, hence:

If, now, the two modes are caused to interfere on the surface of a "square-law" detector, sum and difference frequencies will be generated (amongst others) and, referring back to equations (1), we see that this implies the presence of an electronic difference term: E02 cos 2(#1-#2) = E02 cos ## (4) Now as the pulse propagates down the fibre the s dependence of v will generate a time dependence of A (equation (3).

The term given in equation (4) thus becomes time-variable with "instantaneous" angular frequency given by:

From equation (3) we see that

Hence,

where FD(t) is the function of the derived frequency with time showing that a measurement of the function FD(t) provides a measurement of v(s) and thus also of the spatial distribution of any measurand which affects v(s) in a known way. One advantage of this is that frequency is a relatively accurately measurable quantity.

A practical difficulty is that the frequency to be measured is usually large. Its value is given by FD = b where b is the beat length of the high-birefringence fibre.

For a beat length of 2 mm (of the order required for polarization maintenance) we have FD =100GHz.

This difficulty can be resolved in accordance with the invention by employing, simultaneously, two optical pulses of differing wavelengths.

In view of the wavelength dispersion of the birefringence in the fibre, this will lead to a different beat length for each pulse. This in turn leads to two signals each at a different electrical frequency FD and FD. say. These two signals also mix on the surface of the square-law photodetector to give an electrical signal of frequency Fu - FD, which can be low if FD is close to FD. In the example above, if the two wavelengths are chosen to give a 1% difference in b, then Fu.F0 --1GHz. Such values of frequency are much more easily measurable in practice.

The length of the optical pulse used n these techniques must be small compared with the beat length of the fibre if the instantaneous phase is not to be averaged (and thus smeared) over more than one cycle of birefringence. For small beat lengths this implies very short pulses (e.g. 2mm = 10 pico-seconds). It is difficult to provide large amounts of energy in such small durations without producing peak optical powers which would drive the propagation into the non-linear optical regime (and thus perturb the performance) and without necessitating sophisticated, expensive, optical sources. One solution to this difficulty is to use a series of short pulses from a modelocked laser.Another solution is to use longer pulses at two optical wavelengths which interfere to give rise to an interference pattern with an envelope within which the spacing between maxima is equal to a fibre beat length, each maximum then effectively comprising a short measurement pulse. Each short pulse can then have a width less than the fibre beat length; while the total energy will be the sum of all the pulses in the series or in the longer pulse, which in each case will define the spatial resolution of the system.

Figure 1 of the drawings shows, in schematic form, an optical fibre apparatus for measuring the spatial variation of a measurand.

Referring to the drawing, a light source (1) fc9ds a pulse of light into a "hi-bi" monomode optical fibre (3) via a beamsplitter (2). The pulse energy is distributed equally between the two eigenmodes of the fibre. As the light propagates in the two eigenmodes the birefringence inserts a phase delay between them. Light which is backscattered from each mode in the fibre enters the polarisation beamsplitter (5) via the beamsplitter (2). The two linearly polarised components of the light which emerge from (5) are controlled by the polarisation optical components (6) and the mirrors (7) in such a way as to cause them to interfere optically on a 'square-law" photodetector (8). An integrator (9) and a frequency meter (1) allow the distribution of the measurand field along the fibre to be conveniently displayed on a screen (11).

In contrast to this, an alternative arrangement employing two similar length sets of short pulses (indicated at 2a and 20b) of differing optical wavelengths is shown in fig (2). In this Figure where items are the same as or serve the same purpose as those of Figure 1, they are given the same reference numbers.

Each pulse in each set 20a or 20b acts as described with resoect to Figure 1 for the previous arrangement. The pulses in one set lead to a derived frequency Fda which mixes or heterodynes with the derived frequency Fdb produced by the pulses from the other set. Figure 2a shows schematically the pulse waveform of each such set of pulses. The individual short pulses have a length indicated at 21 which is short enough to maintain their energy to an adequately low level, while giving an envelope indicated at 22 which is of higher total energy and giving an effective measurement pulse length which determines, as mentioned above, the resolution of the system. The result is a measurement signal of a frequency which is of easilymeasurable value, whose time variation maps the distribution of the birefringence, with a spatial resolution equal to the length if the sets of pulses.The distribution of any measurand field capable of influencing the birefringence is then mapped correspondingly.

Many other types of arrangement are possible, using the same basic ideas, but differing in detail. One such arrangement which may be mentioned is that which uses the interaction between propagating light and propagating acoustic waves in the optical fibre. This arrangement has the advantage of increasing the processing time for the output measurement indication. In this case light is launched, as two sets of pulses of slightly differing frequencies, into just one of the two eigenmodes of the fibre. An acoustic pulse is launched into the other end of the fibre to propagate in the opposite direction down the fibre; and transfers some of the light to the other eigenmode as it so propagates.The optical interferenca in this case, which is observed as discussed above with reference to Figure 2, corresponds instantaneously to the position of the acoustic pulse, which travels many orders of magnitude more slowly along the fibre than the light, and thus allows the detection system a much longer time to recognise and to interpret the output light signal. In a further arrangement, the above principle is used for point measurements. In this method, the effect of the measurand field is integrated over a length of "hi-bi" monomode fibre.In this case either the backscattered or forward radiation's eigenmode components are allowed to interfere optically to give a difference frequency on the "squarelaw" photodetector whose value should be averaged over the fibre propagation time, in order to give the required integrated value of the measurand over the full fibre length.

A further alternative arrangement of invention, again employing pulses of differing optical wavelengths, offers enhanced sensitivity to a measurand. The effect of a measurand on each polarisation beatlength associated with each optical wavelength may be very similar when both wavelengths are travelling in a single optical fibre. This would limit the excursion of the beat frequency at the detector. One way to avoid this limitation would be to separate the two wavelengths so that each travels in a separate fibre, one subject to the measurand and forming a sensing path, and the other shielded from the measurand, thus forming a reference path. The full change in the birefringence due to the measurand would then be observable. Another advantage of this two-fibre arrangement is that there can be finer control of the derived beat frequency via, for example, strain or temperature tuning of the reference fibre.

Claims (13)

1. A method of measuring a measurand comprising: subjecting a region of birefringent monomode optical fibre to the influence of said measurand; launching radiation into said fibre; causing two polarization eigenmode components of radiation scattered from said region to interfere; deriving an electrical signal representative of the interference frequency between those two componerts; and determining, from the time dependence of the frequency of the resulting derived signal, the spatial distribution of said measurand1 in which said radiation is so launched into the fibre at two differing launch frequencies simultaneously.

2. A method as claimed in Claim 1, in which said radiation is so launched at each said launch frequency as sets of short pulses, and said electrical signal is derived from the mixing of the interference frequencies resulting from the interference between the scattered eigenmode components at each said launch frequency.

3. A method as claimed in Claim 1 or 2, further comprising deriving from the temporal variation of said electrical signal the spatial variation of said measurand along the fibre.

3. A method as claimed in Claim 3, further comprising integrating said electrical signal over a time substantially equal to the time of propagation of said radiation in said fibre.

4. A method as claimed in any one of the preceding claims, in which said electrical signal is derived from backscattered radiation.

5. A method as claimed in any one of Claims 1 to 3, in which said electrical signal is derived from forwardscattered radiation.

6. A method as claimed in any one of Claims 1 to 3, in which said radiation is so launched into one eigenmode component at one end of said fibre, and is coupled to the other eigenmode component of said fibre by an acoustic wave launched into said fibre.

7. A method as claimed in Claim 6, in which said acoustic wave is launched into the other end of said fibre.

8. A method as claimed in Claim 1 or 2, in which the radiation of each said launch frequency is so launched into a respective such fibre, one of which is caused to experience said measurand and the other of which is substantially isolated from said measurand.

9. A method as claimed in any one of claims 2 to 8, in which said radiation is produced by means of a mode locked laser.

10. A method as claimed in any one of Claims 2 to 8, in which said sets of pulses of launch frequency are produced by the interference of pulses at two close optical wavelengths to give an interference pattern with an envelope having maxima the spacing of which corresponds to the beat length of said fibre or fibres.

11. A method of measuring a measurand through the effect of that measurand on a birefringent monomode optical fibre, substantially as herein described with reference to Figure 2 of the accompanying drawings, or as described elsewhere herein.

12. Apparatus for measuring the spatial variation of a measurand comprising a source of radiation, a birefringent optical fibre having a region sensitive to variations of said measurand, means for launching radiation at two launch frequencies from said source into said fibre, means for deriving an electrical signal from orthogonally-polarised components of radiation scattered from said region, the temporal variation the frequency of which provides an indication of the spatial variation of said measurand in the vicinity of said region.

13. Apparatus for measuring a measurand by means of its effect on a monomode birefringent optical fibre substantially as herein described with reference to Figure 2 of the accompanying drawings, or as described elsewhere herein.