Composition Fractionation During Petroleum Migration

By Zhiyong He, ZetaWare, Inc.

One of the goals of petroleum system modeling and analysis is to predict fluid composition and properties (GOR, API gravity etc.). However, most of the work in the past has been focused on the generation process, with compositional kinetics, etc. Below I will try to show that the petroleum under goes significant changes in composition and properties along the migration pathways due a number of secondary processes not well understood yet.  Most people are familiar with the Gussow (1954) migration model in the figure below. The trap closest to the kitchen would receive the latest, and most mature and therefore lighter fluid, which displaces less mature fluid to traps up dip. 

Fig. 1. Differential entrapment of petroleum along migration path (Gussow, 1954). Late forming gas displaces oil to up dip traps. 

Even without forming gas caps, the later fluid tends to reach the crest of the trap because it is lighter and more buoyant. This pattern is generally true in most basins. Oils with lower gas oil ratios, and lower API gravities are found further away from the generation kitchen. Closer to the kitchen, lighter fluids, sometimes gas condensates are found. 

There are a couple of other factors not obvious from the Gussow model. When the migrating fluid reaches bubble point, a separate gas (vapor) phase starts to form, as shown in the trap in the middle. The gas in the gas cap selectively dissolves the lightest fraction of the liquid as condensate. The remaining oil in the leg retains the heavier part of the incoming fluid. The physical properties in a dual phase trap would over time equilibrate to profiles shown in the figure below. 

Fig. 2. Properties of fluid in a dual phase trap under thermodynamic equilibrium. Red and green lines are reservoir pressure of the gas phase, and oil phase respectively. The blue dashed lines show both phases are undersaturated away from the gas oil contact. GOR and API gravity both decrease with depth, in both phases. 

The dashed lines are bubble point pressure (Pb) and dew point (Pd) pressures. The oil near the oil water contact is always the least saturated with gas, lowest in GOR and heaviest in gravity, as is the the oil that spills from the trap to the next. If the trap is leaking from the crest, the next trap above will receive a gas with lowest condensate content. 

Phase separation happens due to pressure drop below saturation pressure, so it happens along the migration path as well as in traps. If it happens along the migration path, the gas would gradually "bubble" out from the migrating oil phase and either get stuck along the migration pathway as residual saturation (migration losses), because relative permeability for the minor phase is much lower or zero, or trapped in small traps below seismic resolution, along with the light ends of the oil fraction (condensate) dissolved in the lost gas.  The remaining oil will have less solution gas, and become heavier, gradually.  

Even in single phase traps, not only the late arriving lighter fluid goes to the top and displaces the heaver fluid to the flank due to gravity (charge disequilibrium). Gravity segregation and thermal equilibrium may enhance or alter the composition profiles. The figure below shows some observed GOR and API gravity profiles in single phase reservoirs in different basins. 

Fig. 3. Fluid property (API gravity and GOR) profiles in single phase reservoirs, plotted against depth below crest of the trap. Both API gravity and GOR decreases toward the oil water contact.

Significant grading occurs in near critical fluids, as show in figure (c) on the right. These profiles are controlled by complex migration and filling process and PT history and some not well understood thermodynamic processes. Therefore we do not yet have the ability to predict the composition and properties of the fluids quantitatively during the migration process. There are also other secondary processes such as mixing, methane diffusion, water washing, stripping by non-HC gases, biodegradation that can significantly alter the fluid properties. 

In the previous post below. I discussed an alternative "top down" approach to provide a probabilistic estimate of the fluid type and properties in a given prospect.   

Select references:

Gussow, W.C., 1954. Differential entrapment of oil and gas - a fundamental principle. American Association of Petroleum Geologists, Bulletin 38, 816-853

Zhiyong He, and Andrew Murray, 2020.  Migration loss, Lag and fractionation: Implications for fluid property prediction and charge risk. AAPG annual conference, Houston Texas, Sept 28-30, 2020.

Gas Oil Ratio Trends In Sedimentary Basins & PVT Behavior

By: Zhiyong He, ZetaWare, Inc.

Gas oil ratios of oil and gas fields plotted against depth show interesting trends as shown below. The figure on the left is from large global datasets, and the one on the right is from an area in the North Sea. What are the reasons we may ask?

We have recently talked about this in several presentations (see references below). We concluded that this is a result of PVT behavior during migration. At shallower depth, the pressure is lower, and oil cannot dissolve as much solution gas as it can at a deeper depth. Likewise, gas can not dissolve much liquid at shallow depth.

This is supported by the relationship between saturation pressure (Psat) and GOR relationship based on some large PVT databases.

In this figure above, the blue shaded area are based on thousands of saturation pressure (bubble point on the left and dew point on the right) measurements. As HC generation windows are typically deeper than the blue band, migrating fluid toward shallow depth will reach saturation pressure at different depth depending on initial GOR of the fluid coming from the source rock. But once the Psat is reached, GOR will be limited by Psat, and follow the trend of the Psat-GOR relationship, resulting the distribution in figure 1.

This process not only changes GOR significantly from the initial fluid expelled from the source rock, it will also change the composition and API gravity. Below is a local example from a gas condensate system.

The initial fluid is found in a deeper reservoir (1) close to the kitchen, it is undersaturated as reservoir pressure is higher than initial Psat. In shallow trap along the migration path, the fluid is separated as a gas cap and an oil leg (2) and (3), with very different GOR and liquid API gravity. The fluid can further fractionate depending on whether migration is vertical or lateral, (4) and (5). Please also note that the saturation pressure itself is also modified by the same process, and becomes lower at shallow depths.

The same happens without a trap, or in between traps, along the migration pathway. In a gas condensate system like the above, the liquid phase that drops out is a) the heaviest fraction of the liquid first, and b) as droplets that are unable to form a continuous phase to migration along with the main gas phase.

Similarly, if the generated fluid is mainly oil, the GOR of the oil will follow the bubble point side of figure 2. Gas bubbles gradually drop out, or trapped in small traps that spill the liquid, reducing GOR along the way. As the gas phase that was dropped out retains the lightest ends extracted from the oil, the API gravity of the remaining oil decreases approaching shallower traps.

In any given trap, the HC fluid composition and therefore properties are not only a function of the initial generated fluid, but also on the pressure history and the complexity of the migration paths. The self regulating process of changing composition and in turn Psat itself, is much too complex to model at this time. Attempt to predict reservoir fluid properties solely based on source rock kinetics, as you may find in recent basin modeling literature, is misdirected in our opinion. A top down approach based on analyzing observed fluid properties in traps and trends in the geological context (Top Down PSA) is recommended.

Select References:

He, Zhiyong, and Andrew Murray, 2019, Top Down Petroleum Systems Analysis and Geospatial Patterns of Petroleum Phase and Properties. Celebrating the life of Chris Cornford (1948-2017): Petroleum Systems Analysis ‘Science or Art?’ The Geological Society, 24 - 25 April 2019

He, Zhiyong, and Andrew Murray, 2019, Top Down Petroleum System Analysis, Exploiting Geospatial Patterns of Petroleum Phase and Properties. AAPG Annual Convention, San Antonio, May 19-21, 2019 Download pdf from search and discovery

Murray, Andrew, and Zhiyong He, 2019, Oil vs. Gas: What are the Limits to Prospect-Level Hydrocarbon Phase Prediction? AAPG Hedberg Conference, The Evolution of Petroleum Systems Analysis, Houston, Texas, March 4-6, 2019 Download pdf from search and discovery

Zhiyong He, and Andrew Murray, 2020.  Migration loss, Lag and fractionation: Implications for fluid property prediction and charge risk. AAPG annual conference, Houston Texas, Sept 28-30, 2020.

Biodegradation Much? Common Wisdom vs Statistics

By: Zhiyong He, ZetaWare, Inc.

Many studies have shown that biodegradation can have significant impact on oil quality (eg. Larter et al, 2006, Yu et al 2002, Wilhelms, et al 2001). Peak degradation rates are around 30-40 degrees Celsius, which is roughly at about 1000 meters below mudline on average. How much is the risk (% probability) of finding heavy oil if we have a prospect at this depth? For practical purposes, lets say heavy oil means an API gravity lower than 20 API. This question was posted on LinkedIn as a poll, and the answers are anywhere between 10 to 90%, and the mode is around 70%. See the original post here  and many thanks for all who participated. 

Many of you know Andrew Murray and I have been working on examples and methods for Top Down PSA for the last few years. While looking for field/fluid data, I came across a paper titled "Properties of crude oils in Eastern Hemisphere" by Kraemer and Lane 1937. Having read the papers on biodegradation and developed a tool for modeling biodegradation in Trinity a while back, my first thought was that by 1930s the wells were probably very shallow and that many of them would be heavy oils. I was only right about the depths. It was very surprising that out of the 142 fields, less than 10% (13) had an API gravity of less than 20, as shown in the figure below.

The next paper I found was McKinney et al. 1966, which included fluid properties of 546 oil fields in the United States. The API gravity depth plot on the left shows the typical trend, that API in general decrease to shallower depth (Similar to figure 2, Larter et al, 2006). The figure on the right is the 359 fields shallower than 2000 meters. Only 18 (5%) of those are below 20. You can see most of the heavy oils are from California. Most of them are probably sourced by the well known Monterey Fm, which belongs to organo-facies A, perhaps that is (at least partly) the reason for the low gravity (and often high sulfur). Texas and Louisianan have a lot of shallow fields but have no oils below 20. Reservoirs formation of some of these outcrop at surface not too far from the fields.   

Next I plotted a global data set of ~16,000 fields that are less than 2000 meters deep. 14% of the top 1000 meters are less than 20 API, and only 7% of those between 1000 to 2000 meters. The figure on the right include the deeper fields as well.

So what does this all mean? Globally the base rate of heavy oil at shallow depth where biodegradation is a concern is only 10%. If we were to only rely on a basin model that includes the biodegradation process, we are much more likely to predict a heavy oil at these depths.

It is possible that such field databases may not include some discoveries where oil is too heavy to be produced (therefore not counted).  But I don't believe that is a significant enough number to change the statistics because this is such a large dataset, and if it is a prevalent problem there would have been a lot of literature on it. Note that some of the large heavy oil pools are included such as the Athabaska, Orinoco, Rubiales and Kern River.

We are aware of other factors that may prevent biodegradation - such as OWC configuration, nutrient supply,  paleo-pasteurization, and timing of charge (duration of oil in reservoir), etc. Most of these are very hard to determine. The statistics above would imply the possibility that one or some of these factors are very prevalent in most basins. My own suspicion is that in in vast majority of cases/basins charging of shallow reservoirs are active at present day, due to migration lag regardless of when generation occurred.

The most important take away from this is that we should always check base rate (Bayesian analogs) when using basin modeling (bottom up) to predict fluid properties in prospects. Our models only include a small fraction of physical/chemical processes that happen in nature and much of the input of these models are assumptions due to lack of data, and lack of understanding. 

Select references:

Wilhelms, A., S. R. Larter, I. Head, P. Farrimond, R. di Primio, and C. Zwach, 2001, Biodegradation of oil in uplifted basins prevented by deep-burial sterilization: Nature (London), v. 411, p. 1034– 1037.

Yu, A., G. Cole, G. Grubitz, and F. Peel, 2002, How to predict biodegradation risk and reservoir fluid quality: World Oil, April, p. 1– 5.

Larter, S. R. et al. 2006, The controls on the composition of biodegraded oils in the deep subsurface: Part II-Geological controls on subsurface biodegradation fluxes and constraints on reservoir-fluid property prediction. AAPG Bulletin, v. 90, no. 6 (June 2006), pp. 921–938.

Kraemer A. J. and E. C. Lane, 1937, Properties of typical crude oils from the fields of the eastern hemisphere. Department of the Interior. United States Government Printing Office. 

McKinney C. M. E. P. Ferrero, and W. J. Wenger, 1966. Analysis of crude oils from 546 oilfields in the United States. Bureau of Mines. Untied States Department of the Interior.