Hello. I'm Lee Esch. I'm a Senior Geoscience Advisor at Exxon Mobil's
Upstream Integrated Solutions Company in the Subsurface Stratigraphy Area. My
specialty over the years has been reservoir quality analysis. And that's
something I'm actually quite passionate about. And I want to thank AAPG and the
AAPG Foundation for allowing me to talk about some of the aspects of my work
and hopefully to get many of you interested in this as well.
Where we're at today, there's actually been a significant evolution in reservoir
quality analysis and prediction over the years. And we can actually see an
evolution in the types of reservoirs that we're chasing as well. And what I
want to do is go through the history of that today with you and talk about
where things are going and why there's actually a need to become better at
evaluating, predicting reservoir quality in diagenetically complex reservoirs.
So the first thing that I think we really need to talk about before we go
any further is, what is reservoir quality? Reservoir quality is the capacity of
a reservoir to store and produce hydrocarbons. This is the way that we define
reservoir quality. And reservoir quality is really measured by porosity and
permeability. These are both aspects of pore systems that occur within
sedimentary rocks. And that is where we actually store and produce the
The interesting thing is that not all pore systems are created equally. If
we look at the thin section image in the background, if you look at the upper
central portion of this image, you'll see some blue dyed epoxy filling pore
space in a fairly complex sandstone here. We can see that there are grains
present. And we can see in cases there are cement overgrowths on tops of these
But the actual reservoir pore space is that blue area. Not all this porosity
is equal. If you look at the lower portion of this image, below the word
systems, you'll see a patch of light kaolinite cement. But if you notice, there
is a little bit of blue tint to that. And that is actually our blue dyed epoxy
again but this time in a much smaller style of pore space, something that we
So this can be very variable in one sandstone to another. And it's important
to actually understand which of these pore spaces is actually giving you the
better reservoir quality versus poor. And so really, fundamentally, what a
reservoir quality specialist does is characterize and locate favorable pore
systems in the subsurface.
Key in this is the locating part. That's really asking for prediction. And
prediction is really where this type of work has the greatest impact on the
activities that various oil and gas companies are involved in. And we'll talk
more about how that impacts the industry and at what stages and what that looks
like in a moment.
So as you can imagine, we're going to talk about how pore space in a
sedimentary rock evolves and in particular in sandstones here. You have to
think about a number of things that actually control the characteristics of
those sediments. And what we really recognize is that reservoir quality is
determined by the initial sediment characteristics and then the subsequent
impact of burial diagenesis.
And burial diagenesis includes all physical and chemical processes that
modify a sediment after deposition. So if we think about that for a moment, if
you look at the diagram in the lower right portion of the slide, we have a
burial curve there that's marked from A to B to C. And if we looked at a
typical sediment, say, a very quartzose rigid grain sediment that was
deposited, what that would look like at the surface is more like in the image
that we have in A there, where the grains are there, and they appear to be
floating in our friendly blue dyed epoxy.
And all the space between those grains is something that we would term
intergranular volume. And this, in essence, is the porosity that you have
between those grains at that point in the burial. If you bury that rock to a
greater depth, say, to point B where you're approaching almost four kilometers,
what we find is that sediments actually come to full compaction between two and
three kilometers. This was well established by Paxton and others in the '90s
and published in the early 2000s.
A rock that's been compacted will go to a intergranular volume of approximately
26% for a well sorted, rigid grain type of sandstone. And then if we have
subsequent overprints of cements, those will fill that remaining pore space.
I'll note that cements conform any time in a burial history. So you could form
them at A near the surface, in which case, they may support a very large
intergranular volume, or late in that sequence, in which case, they would
reflect the fully compacted value.
So if we think about all the things that control reservoir quality, we
really have to think about what is controlling the initial sediment
characteristics to start with. So this typically starts with an analysis of the
hinterland mythologies and the climate in that area. This really controls what
the mineralogy is of the initially generated sediment. And there is also a
textural control at this point.
The weathered sediments subsequently are mobilized into the transport
system. And so things such as transport distance, sediment storage, and climate
along that transport system also affect the composition of those sediments. And
typically what we find is that, in wetter climates, more reactive minerals tend
to be destroyed and altered into clays. And that the more resistant materials,
such as quartz, tend to be more resistant and make it all the way to the basin
with little alteration. So again, climate is a very important piece to consider
in that initial sediment generation system.
Once the sediments reach a basin, they're subject to a number of different
processes. So as the mobile fluids that are carrying the sediments lose their
competency to carry the sediments, they'll drop out different grain size
fractions. In other words, we'll see texturally different sandstones that
become finer and finer as you move into the basin. But there may actually be a
mineralogical segregation in that process as well. And so that is one of the
things that we need to think about initially.
Once those sediments are deposited, there a number of seafloor processes
that we want to concern ourselves with. And this includes such things as
bioturbation or, in other words, the add mixing of different sediments by
burrowing organisms or the production of early cements due to early diagenetic
processes in many cases which are driven by microbial processes. These typically
end up in the production of such things as carbonate concretionary cements.
And then as we start burying the sample, we need to think about its time
temperature history. And of course, temperature plays a key role in controlling
the rates of reactions. And then we also need to concern ourselves with the
presence of mobile fluids in a basin.
In other words, as you bury sediment package, it compacts. It expels fluids.
You may have fluids driven in the subsurface by other processes, such as thermo
hailing, convection. Many of these fluids may have different compositions. They
can all end up having a different type of overprint on the rocks. So as you can
see, it's actually a fairly complex array of things that we think about. But
there are actually systematic ways to approach it.
Not surprisingly, we've actually, over the years, developed a number of
reservoir quality analysis and prediction workflows. And the basic technology
itself, in all cases, is really done in three stages. One is a characterization
stage. The second is where we integrate the data from that characterization
stage into some type of geological framework. This is done, of course, with
other disciplines, such as stratigraphy, geophysics, structure, and so on. And
then lastly, we engage in some type of RQ predictive effort.
We really recognize two different types of projects. And these include
projects that occur in data-poor areas. And these are typically exploration
projects. And then projects that occur in data-rich areas, and these typically
tend to be on the production end of things.
In the production end of things, we generally have abundant data. And so we
generally have a good understanding of the types of sandstone that we're
dealing with. And typically at this stage, there is something very specific
that has occurred, such as a cement occurring in a field that was not expected,
or it is more prevalent than thought and may impact reserves numbers or
Another type of project that we commonly encounter is one where the actual
producibility is not well understood. In other words, you may drill several
wells. They may all look identical on the well logs. But in fact, some may
produce at economic rates and others not. And so there are typically questions
that need to be answered there. So generally, we take the data that are at hand
in that type of a study, basically use the geological framework as it's been
defined, try to understand what's controlling the presence of cements or
influencing the producibility of a particular reservoir, and then build some
kind of prediction from there that the production company would then take
advantage of to refine their operations.
On the exploration and/or the data-poor end, in many cases, we'll be
operating in frontier basins where really there is no offset data. So we don't
have a good sense of the sediment composition and characteristics. And so at
this stage, we do a basic analysis of the system to understand what the
hinterland was like, what the contributing bedrock types were like.
We try to understand the EOD and stratigraphic architecture, of course, the
sediment transport portion of the system. And we end up coming up with a series
of initial rock properties. And these are what we term inherited control, in
other words, the inherited sediment properties.
And that generally is something that constrains the initial texture of a
sediment in a particular EOD and the mineralogical composition. This, in turn,
gets built into some type of RQ modeling tool. And in general, our typical
industry modeling tool is a compaction core cementation model. So this is where
we get our texture and mineralogical control from.
The burial side of things is generally addressed through basin modeling
efforts. We work closely with our basin modeling colleagues to come up with a
time temperature history that then is used within the RQ modeling tool. And
ultimately, we make a prediction for porosity and permeability.
And so these are the basic workflows that are used on almost every project
that we're involved with. So this is really the state of art for industry right
now for typical types of reservoir quality analyses. How does it impact things?
So on the exploration level, it really provides key input to play in prospect
assessments in terms of constraining, porosity, and permeability. This also
helps us predict locations of reservoir quality sweet spots. And this is
important if you're trying to locate a well in the most optimal position in an
At the development stage, we really try to identify key RQ issues that might
impact the development plan. Commonly, we don't get deeply involved unless
there is some type of cement overprint. And once that is identified, we
actually work with the geological modelers to develop a way to distribute the
cements within their geological models then work with them until they obtain a
decent history matches during the actual simulation stage.
At the production scale, it's almost always some question around field and
prospect level RQ variability. I worked a number of projects over the years
where it was not clear what was controlling producibility in the field. And we
would go in and work these types of projects in detail.
And in general, at the end of the day, you end up having some kind of field
scale prediction for presence of cement or distribution of porosity or things
along those lines. We may also work a little bit outside of the field if there
is a wildcat type of effort in that particular area. We also consult on well
and completions designs at this stage. And that's a fairly common thing to do.
And at all stages, we help with the design and execution of core evaluation
programs. And in all cases, and we've actually gone back and done look back
valuation studies over the years to understand what our impact is. And that can
commonly be in the range of tens to hundreds of millions of dollars, depending
on the scale of the problem.
Right now, I think it's a good point in time to talk about where the
technology has actually come from over the years and where it currently is and
then what the emerging trends are. In the emerging trends, that's really key to
what we need to be thinking about in future efforts in terms of research and
how we actually want to impact the business. We look at where the technology
has actually come from. In the early '80s, and prior to that time, most of the
work was really just basic characterisation.
So an RQ specialist would get samples from a field. Thin sections would be
made. There may be a description of the rock types. Or we may actually have
collected quantitative data at that point in time. That would be summarized in
a report and handed off to various teams that might use those particular data.
Unfortunately, what we found is that those data were not always very
effectively integrated into those studies. And the real value that they might
provide was not realized. And so by about the late '80s, there was an effort to
really start developing genetic concepts for sandstone evolution that would
help us get to a more predictive way of using the work. And then there was a
very significant effort on integrating this type of work into the more regular
prospect and play evaluation type of efforts that occur in oil and gas
And so that was a very busy time. It was actually a step change in terms of
the way we did our work and the way we integrated with project teams and was in
general a very good event. It was also realized in the early '90s with the
emergence of cheaper and cheaper computing power that there was actually a way
to actually build quantitative types of models to help us establish porosity
and permeability in different areas.
And at this time, Rob Lander and Olav Walderhaug developed concepts and the
ideas for building the first quartz cement compaction model. And that showed up
in about 1995 as an application called Exemplar. And then later on, Rob formed
his own consortium and put out a commercial application called Touchstone. That
was so successful in terms of constraining reservoir quality analysis and
prediction that it is in broad application by all majors and many mid-sized
companies at this point in time.
It was also recognized in an Exxon production research just prior to merger
in 2000 with Mobil that there was value in building predictive tools to
constrain the initial sediment composition in these systems. And Bill Heins and
Suzanne Kairo worked on this from 2000 to about 2007 and basically developed
the first effective model for doing texture and mineralogical predictions for
initial sediment composition in sandstones. This was called Sand GEM and was
patented at that time, and it remains in use within Exxon Mobil.
Along with all this work, in '95 when I started at Exxon Mobil in the Exxon
Production Research, I had a great interest in working on other segments
besides quartz cement. And that was one thing in the existing technology that
is not really addressed on a genetic basis. Anything that is along the lines of
a carbonate cement, a clay cement, or zeolite cement, for example, are
basically incorporated into those models based on an understanding of the
paragenetic sequence. So it's a non-genetic type of approach.
In areas where you are data-poor, from a predictability type of perspective,
you really need to have a genetic model to address reservoirs under those
conditions where you don't have analog data to work from or analog data that
you're confident in. And so I started, at that time, working on genetic models
and more or less as skunkworks but as part of specific production oriented
projects where there were field scale issues that required a chemical approach
to potentially understand the cementing question. That was worked over the
years up until about 2016 where we had a sanctioned research project, which
I'll talk about a little bit later.
But we're at the stage, at this point, where we might actually be able to
build an effective genetic chemical model for all of the diagenetic minerals.
The last thing I want to mention on this is that we're at a stage where we're
really recognizing that, in many of our reservoirs, that we have increasing
diagenetic complexity in those reservoirs. And this is something that, if we
look at the evolution over time, when I joined Exxon Production Research in
1995, there were many fields left in our inventory that had very complex
diagenetic overprints. And I had an opportunity to work on those.
But that was also about the same time that all of the majors, in fact, really
stepped off into chasing deepwater plays. And the interesting thing about many
of the early deepwater plays is that they were mineralogically simple. They
tended to be very quartzose sandstones. And they were not buried very deeply.
So they had relatively little in the way of a diagenetic overprint.
And that has changed. In the past 10 years, it's very clear that what we're
looking at today in deepwater settings are increasingly complex in terms of
their diagenesis. If we look at what these emerging trends actually are, if you
look at the QFR plot, the Quartz Feldspar Rock Fragment plot in the upper
right, I've outlined a light yellow area that really outlines the composition
of the early deepwater plays. These tended to be very quartzose sediments and,
as I mentioned, with little propensity for burial diagenesis under cool
So the image in the photo micrograph on the left on the upper left shows a
typical example of one of these shallowly buried quartzose sediments. It has
lots of good pore space left. And it has very little in the way of cements. If
we look at the current tranche of opportunities that many companies are looking
at, they tend to be much more feldspathic or much more rich in terms of their
rock fragment composition or the lithic component. This is the area on the QFR
plot outlined in orange.
So if we look at the rocks that are basically a representative of that, if
we look at the two lower photo micrographs, we can see two here that are not
uncommon. We see significant carbonate cement in cases, as you see in the right
image in that lower pair. Or we see evidence of grains, unstable grains that
have been dissolved and have left secondary minerals, such as clays. And that's
quite apparent in the image to the left. So these are much more complex rocks
than we had been dealing with.
The other factor that really complicates this is the fact that the plays are
becoming deeper and hotter and older. And this is really driven by the advances
in technology in deepwater drilling. So it's a new day. And what we really need
are advanced approaches, concepts, and tools to help us chase these types of
So I want to dive off into that. But first, I think it's worth actually
talking about what some of our existing tools are. And there are actually some
very good tools out there. And there's actually an established strategy that
you can use to address complex chemical diagenetic problems.
And so we'll go through the tools. We'll look at the strategy. And then
we'll actually walk through an example, so you can understand how the tools are
applied in the typical project.
So what are our tools? Well, we still have our petrographic microscopes and
scanning electron microscopes is really the basic tools in our tool box for
doing the petrographic characterization. And these tools really help us
identify what the key minerals are in a reservoir and what some of the key
reactions might be that are tied to the presence or absence of certain types of
And then we also, at this stage, tend to collect quantitative data, which
are used to help us make sure that we're looking at the right minerals and
reactions. It's interesting. When you look at thin sections, there are times
that you get focused on something that is not important. And the quantitative
data and a little bit of statistical analysis really help focus you back on the
things that are important in those types of projects. So this is something we
try to always do.
All of the things that we're talking about, in terms of chemical diagenesis,
are really complex water rock interaction problems. And the basic theory behind
all this is captured in aqueous chemistry and hydrogeology. So we can look at
different fluids and how they interact with different types of minerals. And
then we can think about processes that control fluid flow in basins as well.
And so those are key components to the analysis that we do.
But something that's even a more substantial development that's occurred
over the years is the fact that we actually have quantitative tools that are
available to us to help understand these types of systems. And this includes
anything from simple speciation programs, such as SOLMINEQ.88 to very complex
one, two, and 3D reactive transport models, such as TOUGHREACT or X1t or X2t in
Geochemist's Workbench. So these are all things that are available to us.
And why is this important? I mentioned earlier that prediction is really the
key thing that we're asked to do in many of these studies. That's really what
people are after. What are my rocks like where I don't have any control?
And a very nice statement that was written years ago by Ron Surdam, 1989,
was this. And it's, "casting the diagenetic history of the sandstone into
a process-oriented model helps us move from a descriptive mode to a predictive
mode of analysis." And that's exactly where we want to be. And that's why
these tools are needed when we engage in this type of work.
So how do you do this? There is a general strategy in workflow. And it's
really based on constraining the problem within the hydrogeological framework
using our petrological data.
The actual workflow is one of following a sequence of asking, what dissolved
or precipitated? When did it dissolve or precipitate? Where did it dissolve or
precipitate? And why? And at the why stage, we build hypotheses. We apply some
of our geochemical tools to test those hypotheses and then use the supported
hypotheses to make our genetic predictions away from control.
So let's use an example to walk through this what, when, where, and why
concept and see how that actually helps us answer some very complex questions.
This is an example where I was approached in 2002 to look at a field that our
production company had that they had been drilling in, and the center portions
of the field had very excellent reservoir quality. No problems were
However, when they started drilling out to the south, they started
encountering pervasive carbonate cements. And then when they started drilling
to the north, again, they encountered very extensive carbonate cementation.
They had not drilled much of the southeastern or eastern blocks. And so the
question rose, what are we to expect there, because this potentially had some
serious implications for reserves in place and how we might actually develop
the field in terms of infrastructure. And infrastructure is expensive.
So we start out working this process by doing the what dissolved or
precipitated. And when you actually work through the samples, and there are
three thin section photo micrographs here that really illustrate some of this
diagenetic overprint we're talking about, well, we went through this whole
phase of looking at a whole series of thin sections from about 15 different
wells in the field and asking the question, what is impacting archaea? Well,
you could clearly see that you had early calcite grain dissolution in portions
of the field.
You also had the production of early dolomite cements and replacements in
portions of the field. And then you had the production of late calcite cements
and replacements in other parts of the field. And these are nicely illustrated
here in the thin section photo micrographs.
And from this particular part of analysis, we were able to actually write
some of the key reactions that were involved here, such as calcite dissolution.
Then we also have the production of dolomite cements. We have actual
dolomitization of pre-existing calcite grains. And then we actually have
production of new calcite cements in other portions of the field.
We also constrained, with the petrographic data at this point, whether or
not these were early or late cements. What we found is that the dolomite
cements held intergranular volumes in the 38% to 40% range. This clearly
indicated an early occurrence of the cements. And then we also saw that with
some of the late calcite segments that their IGVs were down around 26%,
indicating cements that occurred after the sediments had reached full
compaction. This effort here is one that actually lets you understand, well, my
early cements, maybe I need to be thinking about some of the shallow
groundwaters that were involved in the system. And the light cements, I need to
be thinking about some late deep basinal fluid.
Then we go through the when. And this is actually doing a part of the work
that's known as paragenetic study and determining the paragenesis. And in this,
you use relative timing criteria that really involve looking at cement
stratigraphy or cross cutting relations. And application of these simple
concepts let's you come up with a relative timing for each cement.
I probably looked at about four to five different wells and up to about 15
different thin sections to really determine the paragenetic sequence here. And
once that work was done, what we recognized is that we're 17 distinct chemical
diagenetic events across the field. This ranged from a series of early dolomite
cements that altered from iron-poor to iron-rich dolomites, production of
kaolinite cements. We saw dissolution event. There is an event where we see
fracturing. And then we see some of the late calcite cementation events.
And so again, this is part of the effort that helps us, again, focus into
the system and understand what type of hydrogeological or chemical framework I
need to be looking at this particular problem from. So the next step in this is
really looking at the distribution cements. And this is a question of, where
did it dissolve or precipitate? And in this case, we're really talking about,
what is the cement distribution?
And so we work closely with our colleagues in petrophysics, stratigraphy,
and structure to help better understand the whole of the system. With our
petrophysicists, we're able to develop a cement curve that we're able to
correlate and understand where the cements were coming from and how they
actually were distributed within the field, at least on a cross section basis.
We're also able to place the sequence or elements of the sequence
stratigraphic framework onto this to understand how that sat within some of the
early deposition system. And then our stratigrapher also developed a very good
series of paleogeographic maps. And here, we have one off of a particular
interval of interest where he was able to show that, basically, from the EOD
interpretations from core description, we're able to come up with a
paleogeographic model that showed that this was a braided fluvial deltaic
Within that system, what you could show is that in the proximal portions or
essentially the more fluvial dominated portions of that system, you had
abundant dolomite cements. And there were only traces of the stridal limestone
components that we knew were in the system in a more distal position. So this
is a relationship that we'll come back to a little bit later. In the more
distal areas of the field our light calcite cements appear. But these actually
were associated with elements of the fault framework in this particular field.
So this really helps us understand, again, what some of the early fluids
might be like. And if we look at our paleogeographic map, one of the first
things that you might start thinking about is, well, gee, you're transitioning
from a fluvial system to a marine system. So there's definitely a change in our
fluid chemistry across that particular depositional system.
So this brings us really to the fourth stage, the why. And this is where we
build our hypotheses. And I won't go into each one. But the one that was tested
and rose to the top was for the early dolomite cements was the marginal marine
mixing zone process. And I'll show you how that was arrived at. I won't talk
about the late stage calcite. That is a great story for another day.
What we did was basically build a cartoon of the hydrogeological framework.
And that's in the lower portion of this image. And what we have is a system in
which our fluvial deltaic system actually cut across the pre-existing carbonate
And so we have essentially classic fluvial sediments that have incised into
a carbonate shelf. And then this deltaic system that deposited out on the
margin of that. If we think about that, that's really an environment where you
transition from fully fluvial to fully marine types of pore fluids. And so we
can actually apply a certain type of geochemical analysis to that where we look
at the mineral stability within a system like that.
We use a parameter called the saturation index, which is plotted on the
vertical axis on the above plot that is separated into areas that I call
supersaturated, saturated, and undersaturated. And the heavy line is the
saturation line. So the basic way to read this is that anything above that line
is supersaturated and can have a tendency to precipitate and has no tendency to
dissolve. Below that line, it will not have a tendency to precipitate, and it
has a tendency to dissolve. Along that line, it has no tendency in either
So if we look at a couple of lines here, the red line describes this
saturation behavior for dolomite. And the green line describes the saturation
behavior for calcite. As you can see, the curves are shaped similarly, but they
cross the saturation point at different locations in the mixing system. Calcite
crosses at about a 0.6 mixing fraction. Dolomite crosses at about 0.8.
This lets us really define the three fields that I have there in the yellow
boxes. In the most distal portion of the system, both phases are saturated. So
we would not expect there is a tendency for any dissolution of either phase
there. And in fact, there's a tendency that those phases, our dolomite and our
calcite could precipitate.
On most proximal end, both phases are undersaturated. So neither would have
a tendency to precipitate and, in fact, would dissolve in that system. There is
this interesting intermediate area between 0.6 and 0.8 where our calcite is
undersaturated, and the dolomite is supersaturated.
And so this area here actually reflects what we actually see in our petrography.
Recall that we looked at the more proximal portion of the system. And in our
petrographic data, we could show that that's where the abundant dolomite
cements form. But in fact, we were dissolving our limestone class in that
portion of the system.
In fact, in core, which I probably should have included an image here, but
in core, you could actually see where a whole class of this carbonate were
dissolved away. Yet, this existed within a heavily dolomite cemented framework.
So this model, in fact, exactly describes the tendencies of the different
diagenetic phases within this system.
We also looked a bit at some of the clays in the system. And they were also
consistent with what you would think regarding their occurrence in this type of
a mixing profile. So that helped us establish the actual mechanism behind the
cement production in the field, especially for the dolomites.
We then took this to the stage where you ask yourself, OK, so how does this
apply in my hydrogeological framework? So how do we extend that? So this part,
we tried to understand where this mixing zone went to, where we did not have
control, and then made our predictions based on that type of an understanding.
And so we built this cement risk map, and that is what had been used over the
I'll mention that, at the end of the project, a well was sidetracked over to
the area we made the prediction. And it basically confirmed the presence of the
cements and the models. This was actually a big success because this ended up
helping the production company modify the development plan for the field, save
significant money. It also resulted a bit in a reserves write down, but it was
much better to do this work and understand the kind of investment you needed to
actually produce the field in the most economic way.
We have other tools that are available. And these include temperature
activity and activity activity diagrams. And these help us understand the
tendencies for mineral precipitation in the subsurface.
Just as an example, on the image at the left, we see a series of formation
water data plotted on this plot of silica activity versus temperature. And at
temperatures below about 100 degrees C, we see that these formation waters are
typically supersaturated with respect to quartz, but that, above that
temperature, they start tracking the quartz stability boundary, suggesting that
you're actually entering a point in which you have effective buffering of the
This is actually the reason that the current quartz cement model and the
kinetic model that's used in industry works so well because it makes the basic
assumption that your fluids are typically supersaturated and capable of
precipitating quart cements. And this really establishes that the thermodynamic
drive for that is there, and that is a good assumption.
The plot in the center is one that lets us look at the stability of
different types of sodium, aluminum silicates at different temperatures. And as
we can see, at low temperatures, we find that sodium clinoptilolite, a zeolite,
is favored in many systems. Whereas at higher temperatures, albite is favored.
And in fact, this is something that we typically see in sediments such as those
found in the Gulf of Mexico sands that contain minor volcaniclastic components.
And then typically, in most reservoirs, we look at elevated temperatures,
albite is definitely a favored diagenetic phase under those conditions.
We can use activity activity plots such as the one for barite in the upper
right to show that barite's typically buffering natural formation waters and,
in fact, most careful thin section analysis, especially if you're using the SEM
will generally turn up a little bit of diagenetic barite. So that tendency's
nicely illustrated on this diagram. And then another diagram showing mineral
stabilities at the bottom for calcite versus dolomite and the fact that those
two, in this system, appear to be buffering your fluid compositions as well. So
all of those plots help you understand the geochemical tendencies under
changing conditions if you use those plots in an effective way.
Another plot, if we come back to this whole saturation index concept that we
used for our hydrogeological model and the geochemical model for our previous
example with our carbonate cements and the dolomite cement in particular, we
can look at that saturation index for many other mineral phases. And here's an
example from a current study that I'm working on where we're trying to
understand early diagenesis and early chemical sedimentation trends in evaporative
In this case, you can take natural data from systems. And this is an example
from an East Africa riff lake that's in literature. And you can look at the
saturation trends for different minerals. So you can see, from this diagram,
basically as you evaporatively concentrate the fluids, in other words, as the
TDS goes up, you first saturate with respect to calcite, then dolomite, and
then sodium stevensite, a magnesium rich smectite clay. So this has actually
helped us understand some of the patterns of diagenesis in certain plays.
We also occasionally get involved with questions regarding scaling potential
for mixing of different formation waters. And here's an example on the lower
right where we had a norphlet well and a portioned norphlet that was somewhat
drawn down. And there was a risk for having smackover water's drawn into the
norphlet. And the question was, is there a potential for barite scale and this
was an easily addressed question again using the saturation index and again a
mixing model here where we look at mixing the two end member waters. So again,
it's a valuable tool to help you understand these types of trends.
So one of the more advanced tools that we have in our toolboxes, a reactive
transport model. And there are various styles of these that go from anywhere from
one dimensional type of models illustrated on this page up to three dimensional
types of models. And the idea here is that, basically, you can define a
sediment with a certain mineralogical composition. And you can define fluids
that are reacting with that and define fluxes.
This actually lets us look at systems in which we have mobile fluids
involved. And there are cases in the subsurface that we do need to understand
these types of processes. And so here, I'm giving you an example where we have
a sediment made up of quartz, andesine, and K-feldspar as some of the primary
mineral constituents and how they might react with a particular fluid moving
So if we look at-- we can look at it from two different perspectives. One
is, what does that diagenetic or paragenetic sequence look like along the flow
path of a single cell or what does the diagenetic overprint look like across
the whole flow path, which is illustrated in the left hand diagram. So these
types of models help us understand two things.
You can look at a paragenesis at a point, or you can look at actually
patterns of diagenesis and understand that in a diagenetic system that includes
mobile fluids, the character of that overprint may not be the same everywhere
along the flow path. And that's something that is quite evident in the diagram
on the right. So again, this is a very valuable tool for understanding this
type of a system.
Here's an example where I ran a two-dimensional model. And it shows,
basically, a reactive transport pathway in two dimensions where we represent a
sediment package that had a higher permeability sediments at the base of the
package and lower permeability sediments at the top of the package. Fluids were
introduced through a fault plane on the left hand portion of this diagram. And
the flow path defined off to the right is about a kilometer long. And what we
can see is that, with this particular system, that near the fault, we actually
end up precipitating large volumes of calcite cement. But as you move away from
the fault, these calcite volumes drop off quite substantially to only a few
This model was actually built based on some of our understanding out of the
previous model for the late calcite cements, we mentioned before when we
started looking across other areas of the basin and trying to understand
diagenetic overprints in other areas that we were interested in. When our
exploration teams first saw the presence of these cements in some wells, their
approach was going to be to take a blanket deduct across all of the strata
wherever it existed to represent the amount of reservoir destruction that they
thought they might have from the presence of these calcite cements.
I began to work with them. And what we were able to do is show them that
these cements likely were constrained at least where they were present in high
volumes just to within 100 meters or so of the faults that were bringing the
fluids into this particular system. This has proved to be a correct assessment
over time. And in fact, in our effort to chase this play in the first place, it
made us realize that there were actually much larger volumes involved in this
particular play than was being recognized at least during the initial stages of
the exploration. So again, these simple models and the concepts, if you
understand how to extrapolate that into the larger geological framework, can be
very valuable tools.
So that really defines what we have in terms of tools in our current tool
box. And the question going forward is, what do we need in the future? Well,
I'll reiterate a few things here.
We're now seeing this challenge where we have emerging hydrocarbon plays
with complex diagenetic overprints. And it would really be a powerful thing to
have a good predictive tool for those types of problems. We don't currently
have that. As I mentioned previously, the existing models really rely on analog
data to constrain those. And without a genetic concept, those types of
approaches can be very wrong.
So we need to have some type of new tool to help us really more effectively
evaluate these types of overprints. And when I was really posed with this
problem, I thought there were actually ways to take some of our current
reactive transport models and apply them in a way that could actually be built
in to some of the existing technology.
So we kicked off this chemical diagenesis research project a few years ago.
And initially, my idea was to approach really the early diagenesis component
because that was really the easiest thing to do. And it's easier primarily
because defining the initial fluids and flux rates, that's not hard to do
because you can use modern analogs for fluid compositions and flow rates. So
that was, in my opinion, a more straightforward thing.
When I was presenting the research proposal initially, I had also mentioned
that I had some ideas around burial diagenesis. And at the end of the meeting,
my research manager said, that's great. I want you to work on both. So I did.
And so at this stage, we went ahead, and I had another colleague was
assigned to work with me. And I had him work the early diagenesis component.
And then I started working this part regarding how you would actually approach
the burial diagenesis side. This has not really been done effectively before.
So it took some substantial thinking at the time in terms of how to apply it.
So I looked at the existing reactive transport models. And one in particular
had had a recent upgrade. And this was in Geochemist Workbench, the X1t and X2T
applications. Craig Bethke had recently added the ability to model multiple
stages, so you could actually have a chance of representing a rock being buried
and being exposed to higher and higher temperature conditions and different
fluxes of different types of fluids.
And this is exactly the kind of thing that was needed. The problem is that,
at this point, you can't really compact any of these models. So that would be a
step where we take this technology and then meld it with the current forward
modeling technologies. That's a future step.
The real effort here was to essentially do a proof of concept type of study
and, first of all, demonstrate that you could actually replicate diagenetic
trends in the subsurface. And this had challenges because our RTMs come under
the criticism at times that you can get any answer you want if you tweak the
kinetics adequately. So one of the goals upfront was to establish a set of
really standardized kinetic equations to use in the model and then ensure that
those worked over a broad range of different sandstone types.
And so that was the approach that I took. The basic idea, if you look in the
upper left corner, there's a cartoon there that I'm calling an RTM reference
volume model. And in that type of model, you define a cell that has a certain
starting mineralogical composition and porosity, and then you take that and
vary it through time sequentially. And that's reflected in the image at the
very lower right where we have a burial history. And I've segmented that and
have basically illustrated that you can, at each step, define the fluids, the
temperature, and the fluid flux through that cell, and then essentially model a
rock as it is buried in the subsurface.
So I did this with four very different types of reservoirs and actually had
good success. I think we've actually accomplished proof of concept at this
point in time. And this work is actually documented in my paper that just came
out this month in a AAPG Bulletin. So I invite you to read it and get a sense
of where some of the efforts may need to go in the future.
So I've given you a synopsis of where reservoir quality has been in the
past, present, and future. Basically, it's evolved from a basic rock
characterization effort to a predictive integrated science rooted in the
understanding of the genetic controls on the evolution of sedimentary rocks.
And at this stage, we've clearly established, over the years, that we provide
significant economic value to industry that ultimately benefits the consumer.
And then lastly, we're at a stage where there really are a series of
emerging opportunities for renewed research in the field of predictive
diagenesis. There are many, many areas that need to be addressed at this stage
to really do this kind of effectively. And obviously, that's a fertile ground for
a lot of good research opportunities. And these range anywhere from things
looking at the geological controls on reactive surface area of grains and
minerals to a better constraint of fluid flow and fluid flow pathways in the
So again, I'd like to give thanks to AAPG and the AAPG Foundation for
allowing me to talk about something that I'm quite passionate about. And then
I'd like to thank Exxon Mobil for letting me talk about actually some pretty
interesting projects over the years, and then my colleagues as well who
supported my work and efforts over the years. Thank you.