Fiber optic sensors light up every inch of oil wells

Posted: September 01, 2026

The cost reductions oil and gas companies can get from AI and data analysis depend on having reliable, high-quality data to start with. But oil and gas companies have particular challenges collecting good data on their operations. Most of their work takes place under temperatures and pressures so high they can destroy sensitive sensor instruments. The work is also deep under ground or sea, making it difficult to install sensor equipment in the first place. Even then, once you install a sensor, it often can only collect data from one discrete location from along the vast distances spanned by bore holes, wells and pipelines. 

Fiber optic sensors have been helping oil and gas companies overcome all three of these challenges. Fiber optic cables can not only stand up to intense heat and pressure, they can collect data on subtle changes in that heat and pressure. What’s more, they can collect data from along the entire length of the cable so oil and gas companies can get data on every inch of their drilling and pipeline infrastructure. They’re also intrinsically safe: because they’re non-electrical, they pose no ignition risk around hydrocarbons and are immune to electromagnetic interference. With these properties, fiber optic sensors are helping companies:  

• Monitor hydraulic fracturing
• Do flow profiling and detect leaks along wellbores
• Evaluate well integrity
• And even optimize enhanced oil recovery (EOR) 

Now, new AI-driven data analysis is making the data fiber optics can collect even more valuable.

How do fiber optic sensors work? 

Optical fibers are thin glass tubes about the thickness of a human hair that can transmit light from one end to the other. They’re commonly used to transmit telecommunications data, but they also make impressive sensor instruments. 

Even minute changes in temperature, pressure, strain and movement along a fiber optic cable cause subtle changes in laser light inside it. Different kinds of changes in the light provide different sorts of information about the environment around the cable: 

Distributed acoustic sensing (DAS) isn’t really about detecting sounds, but rather about detecting any kind of movement or vibration along the length of the fiber optic cable. It relies on a property of light waves called Rayleigh scattering, which is extremely sensitive to small changes in temperature and strain at any point along the length of the cable. DAS detects how much temperature and strain are changing along with the precise locations they’re changing along the length of the cable. 

 Distributed temperature sensing (DTS) uses the property of light called Raman scattering to detect the precise temperature value at every and any point along the optic fiber. 

Distributed strain sensing (DSS) uses the phenomenon called Brillouin scattering to detect changes in strain and temperature with even greater precision and along greater distances than the Raman scattering used for DTS. 

AI and machine learning software can also distinguish between different patterns of strain and temperature change to determine whether it’s being caused by seismic activity around a bore hole, leaks or flow irregularities in pipelines, or even just the footsteps of people walking up to 30 feet away from a pipeline. 

How sensitive are fiber optic sensors? 

The techniques above can collect data along fibers up to 50 km (31 mi) long. DSS using Brillouin scattering doubles that distance to 100 km (62 mi). Even at those long distances, fiber optics can locate temperature and strain disturbances at a resolution of less than one meter and discriminate temperature with an accuracy of +/-0.01° C. 

Fiber optic signals travel literally at the speed of light, so they can provide real-time data from along the entire length of the fiber. Equinor collects data from fiber optics down its wells in the Johan Sverdrup oil field and from operations at its Kårstø processing plant. Equinor reports its system processes data equivalent to streaming 10,000 movies at a rate of 5 GB/s.

That volume of real-time data can really pay off when it’s fed into supervisory control and data acquisition systems (SCADA), which can find patterns in the incoming data to predict possible failures and optimize production. 

For example, at Equinor, the Fiber Optics Project Lead, Taber Hersum, reports that fiber optic data allowed his team to “detect a leak quickly between two annuli, which are the void spaces between casings in the well. This early detection saved Johan Sverdrup from two days of shutdown and performing additional diagnostics, resulting in savings of 40 million NOK [about $4.5M USD].”

How are oil and gas companies using fiber optic sensors? 

In addition to monitoring for faults and gathering data for predictive maintenance, oil and gas companies are using fiber optic data for everything from exploration to oil field monitoring, enhanced oil recovery and production optimization. 

For exploration, fiber optic cables can be placed down holes and used as geophones for seismic surveying. Changes in the light travelling through the fiber in response to vibrations on the surface detect voids, faults, shear zones, and other properties of the underlying geology. 

During drilling, fiber optics can detect gas kicks, when gas prematurely enters into the wellbore. Detecting those gas intrusions in real time lets drillers act quickly to prevent deadly explosions.

Standard gas-kick sensors are located just at the surface on the platform, and then thousands of feet below at the well head, leaving drillers blind to the entire length of the wellbore in between. Fiber optic sensors, by contrast, light up the entire length of the wellbore, giving drillers real-time notice of changes in fluid dynamics that indicate gas intrusions wherever they might occur.

The data arriving from the fiber optics carries information about how much gas is in the wellbore, where it is and how quickly it’s rising to the surface, which computational algorithms and AI can extract. Professor Jyotsna Sharma at Louisiana State University says, “Just the fiber data alone is not enough. You need these algorithms to come to a decision-point because how quickly the gas is coming up literally is a life and death question.” 

Fiber optics also help once extraction is underway. At bp, fiber optic sensors run the entire length of offshore wells. They collect data that help teams understand in real time where oil, gas and water are flowing so they can optimize production from one moment to the next. 

The bp head of production technology delivery, Jean-Charles Dumenil, says, “Better insights lead to better judgement and the potential to produce more barrels.” 

Fiber optics data can also help with hydraulic fracturing by mapping active fracture zones and tracking where fluids are arriving in horizontal shale wells. It can help optimize enhanced oil recovery by monitoring temperatures throughout the reservoir, detecting where steam is flashing, and also whether fluids are channeling away from oil. 

Professor Sharma also worked with Shell to explore using fiber optics installed in wellbores to identify which locations in a reservoir were producing more sand so they could shut off production at locations that were introducing excessive sand into pipelines. 

Once hydrocarbons are flowing through pipelines, fiber optics can detect very small leaks that traditional pressure gauge sensors miss. At her lab, Professor Sharma has detected pinhole leaks in pipelines as small as 0.03 gallons per minute. 

What’s the future for fiber optics in oil and gas 

It’s an old industry adage that data analytics—including AI and machine learning—are only as good as your sensor data. But, the converse is also true: sensor data is only as good as your ability to organize and analyze it. Fiber optics have already proven their ability to provide large volumes of accurate data from across large distances in extreme environments. 

What’s making fiber optics become more and more standard in the oil and gas industry is our ability to collect and analyze the terabytes of data they make available. The next step is to use those data analytics to further automize operations and make insights available to workers. Professor Sharma says, “We also use machine learning to automate the process, because ultimately, you may not have a fiber optic sensing expert who knows all the signal processing and can do all of this in real time. So, automating a lot of this so operators have an end-to-end solution is important.”