A Novel Fiber-Optic Fluid Interface Sensor

We describe a fiber-optic sensor that detects the position of the interface between a clear and an opaque fluid (e.g., crude oil and water or crude oil and air) and two clear liquids of sufficiently different refractive index (e.g., water or clear oil and air). In each case the sensor's structure is the same, although the two forms of detection rely on different principles. As an optical sensor, the device is immune to EMI and will not create sparks in a potentially explosive environment.

The device is an outgrowth of a fiber-optic position sensor described earlier in this publication ("A Fluorescent Long-Line Fiber-Optic Position Sensor") and it has been patented by Sandia National Laboratories. Interface sensing is needed in numerous applications, in the oil industry and elsewhere. The interface does not necessarily need to be between two fluids; in an agricultural application, for example, it could be between grain and air. While the laboratory prototype has yielded the proof-of-principle discussed here, the sensor must still be engineered and field-tested for a particular application. Sandia welcomes collaboration with an industrial partner to achieve that end and to license this technology.

Principle of Operation
Clear and Opaque Fluids. In its most basic form (Figure 1), with an opaque fluid floating atop a clear one, the sensor uses optical fibers that are doped with chemicals called fluorescers, which emit light over a particular spectral range when excited by light of a shorter wavelength. The fiber on the left, called the primary fiber, is shown fluorescing green when excited by blue light provided by a pump beam. The fiber on the right, called the secondary fiber, fluoresces in the red region of the spectrum when excited by green light. Clear fibers carry signals to and from an optical detector and an optical source that, as electrical components, can be remotely located.

Figure 1. Basic principle of sensing the interface between clear and opaque fluids

The basic principle of operation is simple. The fluorescent light is emitted isotropically, so most of the light is emitted from the fiber rather than being guided along it. In the case of the primary fiber, the unguided fraction is transmitted across the clear gap to the secondary fiber, where it excites secondary fluorescence. A fraction of that optical power is guided by the secondary fiber to the photodetector. If the medium between the two fibers is clear over the spectral range of interest, the optical coupling between them is high, resulting in a strong signal. If the medium between the fibers is opaque, the optical coupling is low and the signal is weak. Therefore, the output of the photodetector monotonically diminishes from a maximum value to zero as the interface between crude oil and water moves from the top of the sensor down to the bottom.

There are some additional points to note. First, the diagram shows mirrors placed at the far end of each of the fibers and these serve to increase the signal, whatever the position of the interface. The first mirror reflects pump light that's made it to the far end of the primary fiber back into the fiber, generating more fluorescence. The second mirror reverses the direction of the secondary fluorescence that is moving away from the detector. In terms of the coupling efficiency between the fibers, the configuration shown in Figure 1 is very inefficient because so little of the primary fluorescence impinges on the secondary fiber. We can enhance the coupling efficiency by using reflectors and multiple secondary fibers surrounding the primary. On the other hand, we can enhance the coupling efficiency with only two fibers if they are placed at the foci of an elliptically shaped, reflective enclosure. Finally, let's consider the mechanism of light production. We used inexpensive plastic fluorescent fibers made of polystyrene obtained from Saint-Gobain. However, so-called side-emitting fibers are also available from companies such as Intelite. These silica fibers contain defects that cause light to scatter out of the core without experiencing any shift in wavelength. Conversely, they can scatter light into the fiber (and, unfortunately, partially back out again). Thus, a pair of side-emitting fibers could be used in place of the two fluorescent fibers, reducing the concern over the choice of wavelength. Mixing the two types of fiber in a given sensor may also be desirable.

Two Clear Fluids of Different Refractive Index. Detecting this kind of interface uses the increased out-scattering of light from the primary fiber as determined by the degree of matching that exists between the refractive index (n) of the primary fiber and the medium separating it from the secondary fiber (Figure 2). Primary fluorescence is emitted over an array of angles within a material assumed to have a refractive index of 1.5. If the medium between the primary and secondary fibers is air, then the primary fluorescence is reflected back into the primary fiber for angles greater than the critical angle (θc), 41.8°. If the medium is water, with a refractive index of 1.333, then the critical angle is 62.7°. For oil, with a refractive index of 1.47, it is 78.5°. Because the optical power coupled out of the primary fiber increases with the index of the intervening medium, if our two liquids are clear oil and water, the signal will montonically increase as the interface moves from the top of the sensor to the bottom.

Figure 2. Basic principle of sensing the interface between two clear fluids of differing refractive index

Experimental Arrangement
We built and tested a laboratory prototype to determine how the sensor responded under ideal conditions. For simplicity, we attached all sources and detectors directly to the sensor and we simulated the opaque liquid using a movable barrier, which produced perfect opacity and a perfect interface. The clear fluid was air. The prototype (Figure 3) consisted of two aluminum bars, about 1.5 m long, with long rectangular grooves machined into them. The primary fiber was placed in one groove and the secondary in the other—both fibers were 5 mm dia.—and they were encapsulated in optical epoxy for protection and mechanical stability. The two bars were bolted together to produce optical coupling between the two fibers. A shallow groove machined in each bar allowed the movable barrier to slide between them. It is not clear that rectangular grooves produce the best optical coupling between the two fibers; they were, however, the simplest to machine; V grooves or sections of an ellipse may prove to be preferable. Except in one case of using a green laser pointer, the light source was a Luxeon LED made by Phillips Lumileds Lighting Company and the silicon photodetectors were made by OSI Optoelectronics.

Figure 3. Schematic of the laboratory prototype. All data presented here were obtained with two mirrors. In all cases, x, the position of the interface, is measured from the far end of the sensor

Characteristics of the Fluorescent Fibers
The fluorescent fibers are characterized by both an absorption spectrum and an emission spectrum. The fiber absorbs light over a range of wavelengths (the absorption spectrum) and emits light over a range of wavelengths (the emission spectrum). As shown in Figure 4, all four spectra are quite wide, and in this figure only the relative values within a given spectrum are of quantitative significance. The green fluorescent fiber absorbs heavily in the blue spectrum and emits heavily in the green. In addition, the red fluorescent fiber absorbs over a wide range of wavelengths, including the green, and emits light predominantly in the red. There is a noticeable overlap between the emission and absorption spectra of this fiber from 550–670 nm. This means that much of the emitted light within this spectral range will self-absorb and disappear within a short distance from the point of generation.

Figure 4. Emission and absorption spectra of the primary and secondary fluorescent fibers. Data obtained from Saint-Gobain

Experimental Results
Interface Between Clear and Opaque Regions. In the early stages of experimentation we discovered that the spectral range of the optical source and the nature of the secondary fluorescence spectrum are important in determining the sensor's response. Fortunately, we could use standard optical components to tailor the spectra to produce a desirable response curve for the sensor.

Given the nature of the absorption spectrum of the green fluorescent fiber, we first chose LEDs emitting at peak wavelengths of 455 nm and 470 nm as pump sources because we thought that they might produce sufficient fluorescence power for the current purpose. These wavelengths are near the extreme long-wavelength end of the fiber's absorption spectrum, but they produce measurable optical power at 20 nm less than the LED's peak wavelength. The results are shown in Figure 5 for reflectors at the far end of both fibers. Unexpectedly, the absorption was so efficient that essentially all of the pump power was absorbed over the first 5–10 cm of the sensor. It is difficult to see in the figure, but this is less true of the 470 nm LED. (Apparently, the standard concentration of fluorescers is too high for this application, but fibers of low concentration would provide a broadened response and these are available on special order.) Using LEDs with 505 nm and 530 nm wavelengths provided progressively more favorable results. However, these wavelengths are well within the fiber's emission spectrum. Apparently, there is enough residual overlap between the fiber's emission and absorption spectra that a 530 nm LED is useful for the current purpose.

Figure 5. Normalized signal vs. position of opaque-clear interface using LEDs having peak wavelengths of 455 nm, 470 nm, 505 nm, and 530 nm, respectively

We decided to continue this process by progressively filtering out more of the short-wavelength region of the 530 nm LED spectrum using long-pass absorbing filters from Schott Glass. In addition, we filtered the secondary fluorescence spectrum with a similar kind of long-pass filter to eliminate the overlap region between the emission and absorption spectra. We also obtained data using a green laser pointer (emitting sharply at 532 nm) without additional filtering. The results are shown in Figure 6. The denotation "OG550" refers to a long-pass optical filter with its 50% transmission point at 550 nm and rising from 0%–98% transmission over about 60 nm. Filters OG515, OG530, and OG570 were also placed in front of the source but did not produce a significant difference in the normalized transmission. The filter RG665, which was placed in front of the detector, has its 50% point at 655 nm and a similar rise and this filter produced the most dramatic improvement in the response.

Figure 6. Normalized signal vs. position of opaque-clear interface using the 530 nm LED with and without optical filters, and using a green laser pointer

When we performed similar experiments with the barrier moving through water, which is a better simulation of the interface between crude oil and water, the shape of the sensor response was essentially unchanged.

Interface Between Two Clear Fluids (Water and Air). Figure 7 shows the results for nominally pure water (n = 1.333) below air, using the same filters. The shape of the sensor response is unchanged, but the magnitude of the signal has increased by a factor of 2.25 when water completely fills the space between the two fibers. This increase is caused by the effect discussed earlier. A similar measurement with sugar water (n =1.444) produced a multiplication factor of 3.25. Both of these numbers can be calculated quite accurately using the associated values of the critical angle.

Figure 7. Normalized signal vs. position of water-air interface, using 530 nm light emitting diode and both optical filters

Spatially Resolved Fluorescence Measurements. Overall, the sensor's response is reasonably linear over at least 80% of its range. However, even after optical filtering, the sensor response still contains a rapid increase over the last few centimeters of the sensor's length and a weak echo of that increase over its first few centimeters. These characteristics could be useful in certain situations. This effect is a consequence of the spatial dependence of the fluorescence output from the primary fiber combined with that from the secondary fiber. These dependences are graphed in Figure 8, along with a cartoon showing how they were obtained. The spatially resolved measurement of the output from the primary fiber is shaped like a bathtub curve, with the wall several times higher near x = 152 cm than x = 0 cm. However, it is not entirely clear why it has this shape, as opposed to one that more closely resembles a decaying exponential.

Figure 8. Spatially resolved measurements of the primary and secondary fluorescence, along with a cartoon showing how the measurements were obtained. Both the LED and detector were filtered as in Figure 6

We have demonstrated the behavior of a laboratory prototype optical interface sensor that is expected to have many applications in industry. It is ready to be engineered for any one of them.