Distributed sensing in harsh environments

Raman distributed temperature sensing (DTS) systems are optoelectronic devices that detect scattered light within optical fibers to measure temperatures. DTS systems have been commercially available for several decades and have been successfully used in the oil and gas industry for pipeline, transmission line, and downhole applications. To achieve reliable and more accurate temperature measurements, or to simultaneously measure other parameters (e.g., strain and vibration), multiple fibers are required. The fibers are therefore often bundled together in the same space-limited installation cables.^{1, 2} This, however, gives rise to several problems, such as congestion of the conduit and errors associated with fiber length mismatches.

Multicore fibers (MCFs) have been proposed, and successfully demonstrated, in telecommunications applications to solve the problem of fiber congestion in narrow conduits.^{3, 4} These MCFs are typically coated with acrylate, however, and are unsuitable for situations where high temperatures and harsh environments may be encountered. It is also a challenge to connect the MCF to individual fibers so that independent signals can be coupled to the input and output MCF signals. This is typically accomplished using a tapered fiber bundle that is specifically designed to match the geometry of the MCF. A critical design factor for an MCF is therefore to achieve easy connections with the existing fibers.

We have manufactured an MCF that has a silicone coating, an ethylene tetrafluoroethylene (ETFE) buffer for high-temperature applications (up to 125°C), and good chemical and abrasion resistance.⁵ The ETFE buffer has a diameter of 600μm, which is small enough to fit into very small spaces, e.g., the 0.125″ (∼3.2mm) metal tubes that are often used to deliver high-temperature sensors. We have also enlarged the outer glass cladding of the MCF to give a diameter of 250μm, so that the space between cores is increased and core-to-core cross talk is minimized. This provides additional reliability as the fiber break strength is increased by about four times, compared with traditional 125μm-diameter clad optical fibers.

A cross section through the end of our MCF is shown in Figure 1. Our design contains three multimode (MM) cores and one single-mode (SM) core. The MM cores each have a numerical aperture (NA) of 0.20 and a diameter of 50μm. These MM cores are compatible with the 50/125 graded-index (GI) fibers that are typically used in DTS systems. The SM in our MCF has a core size of 8μm and an NA of 0.12. We used an optical time-domain reflectometer to measure the fiber attenuation of the MCF. The MM cores experienced a signal loss of 2.4dB/km at 850nm and 1.1dB/km at 1300nm, whereas the SM had a loss of 0.54dB/km at 13,100nm and 0.7dB/km at 1550nm.

Figure 1. The end of the multicore fiber, illuminated at both ends and imaged with an optical microscope. The three multimode cores (each with a diameter of 50μm) are labeled 1–3, and the single-mode core (diameter of 8μm) is labeled 4.

The distance between adjacent cores in our MCF design is 85μm. Thus we are able to minimize the amount of core-to-core cross talk. We tested the cross-talk level by using a 50/125 fiber to launch light toward Core 2 and then scanning the optical power at the output end of the MCF with a separate 50/125 fiber. The cross talk between Core 2 and Core 1, or between Core 2 and Core 3, is less than 40dB with 400m of MCF coiled on a 10″ (25.4cm) spool, at a wavelength of 1064nm (i.e., within the Raman band that is of most interest in DTS measurement systems).

The four cores of our MCF are arranged in a square geometry (see Figure 1) set on a circle of radius 62.5μm that is concentric to the cladding. This structure enables easy mechanical connection to standard optical fibers, which we tested in two different configurations. First, we used two spatially opposed 50μm cores, together with Core 1 and Core 3. We found that it was practical to couple the opposing MM cores of interest to the two traditional 50/125 optical fibers, which were aligned and confined to a ferrule with a 250μm through-hole diameter. With this connection state we achieved a sufficiently low insertion loss to complete the experimentation. We also conducted a second connection experiment, in which we demonstrated mechanical connection with reasonable insertion loss when all four MCF cores were connected to individual non-MCF fibers (3 × 50/85μm, 1 × 8/80μm) that are commercially available.

We have produced a low-loss MCF that is suitable for distributed sensing applications in harsh environments. The MCF design includes four separate optical waveguide cores in a very small cross section and with low levels of cross talk between the cores. We have tested the MCF and have demonstrated that mechanical interconnects can be used to achieve reasonable insertion losses between the MCF cores and traditional, commercially available optical fibers. We now plan to measure the temperature distribution in a double-ended configuration using the MCF and compare the results with those that are obtained using traditional fibers.

We thank Paula Fournier for conducting measurements of the fibers and Pat Vannasouk for making the multifiber connectors.