Virtual Drilling for Liquid Hydrocarbons

When sound passes through a liquid, something interesting happens. The different frequencies which make up the pulse start to separate: a well-known physical phenomenon from photonics that also manifests as rainbows in the sky. When light shines on tiny rainwater droplets, these act as prisms and cause white light’s frequencies to separate, forming beautiful colored rainbows. This effect is referred to as dispersion and occurs due to photons exhibiting frequency-dependent velocities through the water. While legend has it that there is a pot of gold at every rainbow’s end, rainbows remain merely admired for enhancing the scenery with their spectacular colors.

This prism effect in photonics has its equivalent in acoustics. When sound pulses (analogous to white light) penetrate a liquid-filled reservoir (equivalent to the water droplets) at different velocities, the shape of the sound pulses is altered (as with the different colors of the rainbow). However, this effect, compared to reflection-based attributes, is very minute and therefore not easy to identify, but while these acoustic rainbow equivalents in seismic data may be difficult to see, they offer real black gold – oil – when found. 

The ability to measure the dispersion effect allows for the identification of liquids in the ground. And since dispersion is a function of reservoir thickness, porosity/permeability and viscosity, identifying strong dispersion events clearly correlates to identifying liquid hydrocarbon-filled reservoirs, while conversely, the absence of dispersion signals in the seismic data suggests a low chance of finding an oil-filled reservoir.

So why are more companies not using dispersion as the powerful Direct Hydrocarbon Indicator (DHI) tool it is? As it turns out, measuring dispersion is rather complex and requires some novel and refined approaches to seismic data processing and analysis.

The Need for an Improved and Broadened Usage of DHIs.

Despite modern computer power, better seismic quality, enhanced data processing and improved seismic analytical methodologies, global exploration drilling success rates remain surprisingly low, while discovery sizes are simultaneously shrinking. The seismic industry has worked hard over recent decades to refine seismic reflection and refraction attribute interpretation, achieving higher resolution imaging for improved identification of sediments, facies changes and structures. The arrival of machine learning and artificial intelligence offers exciting and promising new approaches to high multidimensional analyses. The use of DHIs is growing; yet their success is still only moderate, with some notable exceptions such as the successful use of amplitude versus offset (AVO) in the Gulf of Mexico. There is no doubt that there is a significant need for an improved and broadened usage of DHIs.

A detailed understanding about dispersion requires the ability to study minute shifts in frequencies. Luckily, access to both computer power and new mathematical models for spectral decomposition are now available. What is possible today in the world of spectral decomposition for oil exploration purposes was not practically feasible even 15 years ago – we are at the inflection point of a paradigm shift. Utilizing a hybrid of spectral decomposition methods, combined with high-performing multi-core computers, permits minute dispersion signals to be identified and analyzed.

Frequency-based seismic attribute analysis is already an integral part in many companies’ exploration de-risking toolbox. However, shortcomings in seismic data processing, unsuitable selection of spectral decomposition methods, lack of understanding of spectral decomposition method limitations and the very high computational requirements, have held back the extent of the implementation of spectral decomposition in oil exploration. Red-Green-Blue (RGB) frequency blending is a tool popular for channel sands identification, yet it says little about the actual presence of liquid hydrocarbons.

Using Frequency Decomposition and Dispersion as a DHI

A successful approach to identifying liquid hydrocarbons via dispersion requires the use of a hybrid of frequency decomposition technologies. This approach grants high time/frequency resolution, within the constraints and the quality of the original seismic dataset. Once the computationally intense time-frequency decomposition hybrid method has been selected and the data processed, dispersion analysis can be performed.

Frequency dependency with liquid content in a reservoir was demonstrated in the lab as early as 2002 in a SEG paper by Goloshubin, Korneev and Vingalov. Velocity-dependent dispersion needs to be distinguished from frequency-dependent amplitude attenuation, although the two phenomena may at times correlate. For example, a study by Carcione and Picotti (2006) showed that gas saturation and porosity are the main factors influencing frequency-dependent attenuation. There are numerous studies in which frequency-dependent AVO shows its usefulness for seismic interpretation (Wilson et. al., 2009; Wilson, 2010). Several papers suggest the authors have studied dispersions events, when in fact they make no reference to dispersion but instead to frequency-dependent amplitude attenuation events. This distinction is important, since thin beds of gas can cause frequency-dependent attenuation and therefore identifying oil based on such anomalies may be challenging, whereas dispersion signals are less influenced by gas presence.

Velocity Dispersion Analysis

A more precise analytical tool for liquid hydrocarbon detection is therefore velocity dispersion analysis, since it is less influenced by the presence of thin layers of dry gas. A high-dispersion response has empirically been seen to strongly correlate with the presence of liquid hydrocarbons.

By studying velocity-based dispersion, information beyond frequency-related attenuation can be analyzed, thereby increasing the overall geophysical understanding derived from the seismic data. One approach to measuring attenuation and possible dispersion correlation was to analyze reflection-coefficients. A more advanced approach encompasses very detailed frequency analyses in combination with the study of resonance waves. With seismic dispersion signals being inherently weak, high requirements are put on the seismic data quality and in particular, its processing. Any post-stack frequency alteration such as spectral bluing or whitening runs the risk of altering the sought-after frequency information. Too aggressively boosting low frequencies in broadband data may introduce frequency artefacts across the entire seismic spectrum.

In 2009, Rex Technology Management started studying resonance effects in seismic data using the RVD technology. Resonance occurs in any elastic acoustic environment and is known and basically understood, yet greatly understudied. The idea was to use resonance effects to show information about lithologies and the presence of oil, which would otherwise not be seen. The technology had been highly successful, predicting several dry wells in assets that were offered on the farm-in market. In 2015 the technology was fundamentally revamped to combine resonance with advanced dispersion studies to improve the prediction of liquid hydrocarbon reservoirs. RVD today is significantly more accurate and has a stronger ability to identify not only dry wells, but also liquid hydrocarbon reservoirs.

Rex Virtual Drilling: Successful Applications

RVD has already been successfully used by a number of companies to find oil and to avoid drilling dry wells. In 2014, the technology contributed to the first ever offshore oil discovery in south-eastern Oman, the GA South in Block 50 by Masirah Oil Ltd, as illustrated in the figure above. This discovery opened up a new play and also led to the identification of a new source rock in Oman.

The technology has also been successfully used in the Norwegian North Sea, contributing to the Rolvsnes basement reservoir discovery, in which Lime Petroleum AS participated. The technology adds entirely new information, which contributes to a richer geological understanding: for example, the figure above shows strong dispersive events in the well-known Balder field in the Norwegian North Sea with its complicated injectite reservoirs. The figure below shows a narrow seismic section where identifying the magnitude of the dispersion event adds new and relevant information to be put into a geological context.

This new and bold technology has been extensively tested on a number of wells around the world, successfully finding oil and predicting dry wells. RVD shows that more information can be extracted from conventional seismic data than previously known and that conventional wisdom can be challenged, with remarkable results.