BULLETIN OF CANADIAN PETROLEUM GEOLOGY

VOL. 50, NO. 3 (SEPTEMBER, 2002), P. 363-369

Visualizing the internal physical characteristics of carbonate sediments

with magnetic resonance imaging and petrography

MURRAY K. GINGRAS BRYCE MACMILLAN BRUCE J. BALCOM

Department of Geology Department of Physics, MRI Centre Department of Physics, MRI Centre
University of New Brunswick University of New Brunswick University of New Brunswick
Fredericton, NB E3B 5A3 Fredericton, NB E3B 5A3 Fredericton, NB E3B 5A3

ABSTRACT

Magnetic resonance (MR) images are analyzed in conjunction with petrographic data to evaluate the textural character-
istics of rocks dominated by fabric-selective dolomitization. The magnetic resonance imaging (MRI) measurements reveal
the three-dimensional geometry of the physical sedimentary structures and the trace fossils that influenced dolomitization
and porosity development.
Because MRI images represent composites of nuclear spin density and MR relaxation times, each of which can be
related to pore size, stronger MRI image intensity must be calibrated to known porous zones by integrating petrological
data with MR data. Pairing of MR images with petrography helps map the distribution of porosity in diagenetically
altered rock.
The data presented herein show the potential of a new class of MRI technique as an imaging tool for low-porosity
rocks. The results demonstrate that MRI technology can significantly enhance petrological studies. Notable results
include 1) the successful resolution of the porosity distribution in carbonate rocks characterized by low porosity
(generally less than 6%); 2) the successful acquisition of the three-dimensional data required to model the porous
network; and 3) recognition that the complex distribution of porosity and its relationship to the matrix show that this
fabric represents a dual porosity/permeability system and may reduce the resource quality of similarly burrowed
carbonate rocks.

RÉSUMÉ

Des images de résonance magnétique (IRM) ont été analysées en conjonction avec des données pétrographiques pour
évaluer les caractéristiques texturales des roches dominées par la dolomitisation à texture préférentielle. Les mesures
des IRM montrent une géométrie tridimensionnelle des structures sédimentaires physiques et les traces fossiles qui ont
influencé la dolomitisation et le développement de la porosité.
Parce que les images IRM représentent des composites de la densité de spin nucléaire et des temps de relaxation RM,
chacune peut être mise en relation avec la dimension des pores, et l’intensité la plus forte des IRM doit être calibrée par
rapport à des zones poreuses connues en intégrant des données pétrologiques avec les données de RM. Corréler les images
RM avec la pétrographie permet de cartographier la distribution de la porosité dans la roche altérée par la diagenèse.
Les données présentées ici montrent le potentiel d’une nouvelle classe de technique IRM en tant qu’outil d’imagerie
pour les roches de faible porosité. Les résultats démontrent que la technologie IRM peut significativement augmenter les
études pétrologiques. Les résultats les plus notables incluent: 1) le succès de la résolution de la distribution de la porosité
dans les roches carbonatées caractérisées par une faible porosité d’ensemble (généralement de moins de 6%); 2)
le succès de l’acquisition des données tridimensionnelles nécessaires pour modéliser un réseau poreux, et 3) la
reconnaissance que la distribution complexe de la porosité et la relation avec la matrice montre que la texture représente
un système à couple porosité/perméabilité qui pourrait réduire la qualité de la ressource dans des roches carbonatées qui
sont taraudées de façon similaire.
Traduit par Lynn Gagnon

363
364 M.K. GINGRAS, B. MacMILLAN and B.J. BALCOM

INTRODUCTION 2000), show that the SPRITE technique is capable of providing

high-resolution MR images where standard spin-echo MRI
Visualizing the internal structure of carbonate rock by non- techniques fail.
destructive means is inhibited by the lack of density
contrast between the primary carbonate-rock-forming min- R
OCKS STUDIED
erals, calcite, aragonite and dolomite. Visualizing the porosity
distribution is also difficult, because only mesoscale vugs are This paper focuses on the textural heterogeneity of two
easily detected with standard techniques, such as X-ray and limestone-dolomite samples taken from outcrop. They are
X-ray computed tomography (CT) scans. Also, if the porosity from the Palliser Formation near Cadomin, Alberta (Fig. 1),
distribution is determined by such techniques, one cannot where the rock is exposed in thrust sheets associated with the
determine if the observed porosity represents effective (con- Front Ranges. Both samples, selected for their reservoir
nected) or ineffective (isolated) pores. characteristics, have a low permeability, limestone matrix
An advanced magnetic resonance imaging (MRI) tech- with negligible effective porosity. The limestone encom-
nique, single-point ramped imaging with T1 enhancement passes chaotically distributed, fabric-selective dolomite,
(SPRITE, Balcom et al., 1996), has the potential to improve which has a higher effective porosity of about 5–6% (Fig. 2).
our understanding of porosity distribution in carbonate rocks. The dolomite is interpreted as representing diagenetic alter-
Like the aforementioned techniques, SPRITE does not ation around physical sedimentary structures and biogenic
resolve mineralogical heterogeneity. It is exceptionally structures (ichnofossils).
useful, however, for the mapping of effective pore space.
Because the porosity must be imbibed with a fluid to provide METHODS
a measurable MRI signal, the images generated by these The SPRITE method represents a modified version of
techniques generally show the effective porosity (to that the single point imaging (SPI) method that has been used
fluid). This technology may also be used to study petrophys- to image porous concrete material (Beyea et al., 1998; Cross
ical aspects of a medium and the macroscopic flow properties et al., 1998). SPRITE employs ramped, phase-encode, mag-
of fluids (Kenyon, 1992). netic field gradients, with a radio frequency (RF) pulse
applied at each gradient step (Balcom et al., 1996). Single
PREVIOUS WORK data points are acquired at each gradient step. Sample-based
Only standard spin-echo MRI techniques have previously characteristics that influence the signal intensity include the
been applied to the study of sedimentary rocks. Although this local proton (hydrogen) density and the local MR relaxation
was initiated with high expectations and produced some times. Experimental parameters are the signal encoding times
excellent results (Chen et al., 1996; Hoffman et al., 1996; (tp ), the time between RF pulses (repetition time, TR) and
Brancolini et al., 1997), the research gradually lost momen- the RF flip angle (α). Given these variables, image interpre-
tum as some of the limitations regarding traditional MRI, and tation is complex.
its application to rocks, became more apparent. The key lim- Ongoing work is directed at increasing the fundamental sen-
itation was the sensitivity of the MR method applied, and the sitivity of these techniques, while simplifying the image con-
minimum observable porosity. Clinical MRI instruments, for trast and thereby the image interpretation. Equation 1 relates
example, are not especially useful when the effective porosity the signal intensity at any point in a SPRITE image to the pro-
falls below 20%. In material applications, MRI technology ton (water) density ρ0, the experimental parameters tp , TR and
has tended towards the observation and spatial encoding of α, as well as the MRI signal relaxation times T2* and T1. The
shorter MR-lifetime signals, thereby increasing the tech- T2* relaxation time is the effective spin-spin relaxation time. It
nique’s sensitivity. In recent work, we showed that short echo governs the exponential decay of the MR signal following an
time spin-echo techniques might permit observation of minor RF pulse. The T1 relaxation time is the spin-lattice relaxation
variations in porosity, and pore geometry can be resolved. time, which governs the recovery of the sample magnetization
Gingras et al. (2002) used this method to observe millimetre- during the recovery period (TR), following perturbation by an
scale porosity conduits, interpreted as representing the fossil RF pulse,
passages of worms, which could be mapped throughout a
sandstone sample. Notably, the difference in porosity between S = ρ0 e–tp/T2* • [1–e–TR/T1] sin α (1)
the matrix and the conduits was less than 2% (based on the 1–cos α e–TR/T1
total volume).
The SPRITE technique has previously been applied to the Two fragments of Palliser Formation were prepared for analy-
study of sediments or sedimentary rocks. Rack et al. (1998) sis. Sample A, with a crude wedge shape, measured 9 x 12 x
were able to visualize the internal architecture of Pleistocene– 16 cm and had fresh (broken) surfaces on all sides. Sample B
Holocene lake sediments from Lake Winnipeg in central measured 16 x 12 cm, had an almost rectangular shape, and
Canada. The application of this technique and turbo spin-echo was cut to form a slab 2 cm thick. To prepare for acquisition of
single-point imaging (turboSPI) in other heterogeneous materi- the SPRITE images, the two samples were saturated by placing
als, such as concrete and Tepexel aggregate (Beyea et al., in about 25 cm of distilled, standing water for about 24 hours.
MAGNETIC RESONANCE IMAGING AND PETROGRAPHY OF CARBONATE SEDIMENTS
Fig. 1. Location maps. A. Cadomin in west-central Alberta. B. Sample location. The samples were taken along a railway cut that exposes tens
of metres of Palliser Formation, adjacent to the quarry.
The samples were removed from the water 20 minutes before
being subjected to MRI.
The SPRITE technique was implemented on a Nalorac 2.4
T, 32 cm horizontal bore, superconducting magnet. Imaging
experiments were conducted at ambient temperature with a
Tecmag Libra S-16 console. The RF probe was a homebuilt
birdcage coil. The RF amplifier was a 2 kW AMT Model
3445. Magnetic field gradients were driven by Techron 7780
amplifiers. For each image (in three orthogonal directions),
the gradient was ramped in 64 steps with 2 ms step lengths.
The delay between successive gradient ramps was 300 ms.
Under these experimental conditions, the acquired images
have substantial T1 weighting (desirable with the SPRITE
technique). Signals from rock regions with T1 relaxation
times longer than several milliseconds will have reduced
image intensity.
The image data matrix has 643 data points, which were
acquired in a conservative overnight experiment, with 32 sig-
nal averages. The image field of view was 17 cm isotropi-
cally, with a phase encode time of 80 µs, and an RF pulse flip
angle of 16°. The nominal image resolution was 2.5 mm
isotropically. Measurement of the bulk MR relaxation time
for carbonate samples A and B revealed T2* signal lifetimes
of hundreds of microseconds with multiexponential T1 relax-
ation with short lifetime components of tens of milliseconds
and long lifetime components of hundreds of milliseconds.
Relaxation times in these ranges are ideally suited to the
SPRITE technique.
365
RESULTS AND INTERPRETATION
PETROGRAPHY
Both samples used in this study are petrographically similar.
The rock comprises lime mudstone that envelops planar and
tubular finely crystalline, euhedral, equant, and locally sucrosic
dolomitized zones. Although little variation exists in either
cement, small allochems (mostly crinoids and small bra-
chiopods) are scattered throughout. Dolomite appears to have
replaced the calcite wherever it is present, the strongest evi-
dence of which is locally common, partially dissolved calcite
crystals scattered amongst the dolomite crystals. The limestone
has a variable (point count) porosity, between 1 and 2%.
Dolomitized zones commonly exceed 5% porosity. The
absence of dissolution textures, replacive nature of the
dolomite, its pervasive (and penetrative) distribution, and its
similarity to subsurface examples from the Wabamun
Formation suggest that the dolomite is not a surficial weather-
ing product.
MR IMAGES
Sample A
Sample A was chosen for study based on the chaotic distri-
bution of dolomite patches seen on the rock’s surface (Fig. 3A).
As expected, selected ‘slices’ through the image, which are
bedding parallel, show that internal heterogeneity is remark-
ably high (Fig. 3B–E). Calibration of high MR signals (bright
366 M.K. GINGRAS, B. MacMILLAN and B.J. BALCOM

A B
Fig. 2. Diagenetic and sedimentary textures in the samples studied (Sample B shown). A. Polished slab of Palliser Formation. The location of
dolomitized limestone is indicated by white arrows. Dolomitized sections are mostly circular in this view and are generally associated with trace
fossils. Primary horizontal lamination and well developed stylolites are indicated by black arrows. B. Close-up view showing lime mudstone (ct)
adjacent to dolomitized zones (dt). The lime mud is micritic, whereas the dolomite cement crystals are fine grained, equant and locally sucrosic.

zones) to the petrographic data suggests that the strongest sig-
nals are generated from portions of the rock where the porosity
exceeds 4% and approximately coincide with the presence of
dolomite cement. Presumably, because the sample was satu-
rated with water, only the effective porosity is shown.
Higher MR signals define approximately tubular, dolo-
mitized ichnofossils that locally branch and interpenetrate
(Fig. 3A). Each of these zones has a diameter of between
0.5 and 1.0 cm. Higher resolution images may reveal smaller-
diameter loci of comparatively high porosity. The burrows
in Sample A are mostly parallel to bedding but some vertical,
branching tubes are present (Fig. 3B, C). The trace assem-
blage is composed of the ichnogenera Thalassinoides and
Planolites; diagenetic cements inhibit a more detailed taxo-
nomic evaluation.
Higher MR signals associated with the highest porosity in
Sample A occupy 18% of the rock volume. This together
with burrow-affected zones, which show a moderate MR sig-
nal, constitute 57% of the rock volume. A surface-rendered,
3-D image of high intensity regions associated with the
inferred porous zones (Fig. 3F) show that the porosity net-
work is connected and potentially provides a tortuous flow
medium for fluids.
Sample B
Sample B is different from A in that some of the primary
sedimentary structure is preserved, mostly as wavy-parallel
lamination. In addition, three stylolites crosscut the rock fabric
(Fig. 2A). These are approximately bedding parallel (Fig. 4A).
Like Sample A, high MR signals are associated with areas of
the rock that have approximately 4% porosity or more (Fig. 4B,
C). The stylolites, or the stylolite-affected rock, are also asso-
ciated with a strong MR signal. This seems to be related to
higher porosity rock adjacent to the stylolites (confirmed pet-
rographically), but the resistate minerals preserved within the
stylolites may also contribute to the MR signal.
Higher porosity zones are associated with horizontal,
planar elements, and tubular structures that are dominantly
perpendicular or oblique to the planar structures. The planar
porosity streaks correspond to locations where planar
MAGNETIC RESONANCE IMAGING AND PETROGRAPHY OF CARBONATE SEDIMENTS 367

Fig. 3. Various images of Sample A. A. Photograph of the sample, showing the

mineralogical segregation: dolostone is light gray, and the limestone is dark. B–E.
MRI sections that reflect the internal heterogeneity of the sample. F. Three-dimensional
reconstruction of zones that exceed about 4% porosity (extrapolated from thin section).
High signal intensity is white, low signal is black. The images are blurry due to the
resolution of the technique.

lamination has been identified and to areas affected by the
stylolites. Tubular structures, which are mostly about 1 cm
in diameter, branch locally, and probably represent
Thalassinoides. Close inspection of the rock shows that
smaller Planolites and Chondrites are not resolved in these
experiments (Figs. 2A and 4A).
Areas characterized by higher MR signals occupy 24% of
the rock volume. Areas that show a moderate MR signal, or
higher, represent 49% of the sample volume. The 3D surface
rendering of the high intensity regions (Fig. 4D, E) suggest that
the porosity network locally connects planar elements and
provides a tortuous means of vertical communication between
horizontal, parallel beds.
DISCUSSION
The MR images clearly show the internal heterogeneity
present in the Palliser Formation (Figs. 3 and 4). Bright spots
in the image that are related to higher porosity are consistently
associated with diagenetic cements. Those cements are textu-
rally selective and are focused typically in and around physical
and biogenic sedimentary structures.
368 M.K. GINGRAS, B. MacMILLAN and B.J. BALCOM

Fig. 4. Various images of Sample B (also shown in Fig. 2). A. Vertical section through the
slab. Horizontal elements are much more common in this sample (compared to Sample A).
Vertical heterogeneities, interpreted as occurring around trace fossils, are quite common
(white arrow). Black arrow shows the location of porosity adjacent to stylolites. B. Similar to
4A, but represents a different ‘slice’ within the rock. White arrow shows the location of a ver-
tical trace fossil. C. Three-dimensional model, with a porosity cutoff of approximately 4%. D.
Three-dimensional model, with a porosity cutoff of 5-6%. The two discrete, horizontal porosity
zones are conceivably linked vertically by burrow-related dolomite. High signal intensity is
white, low signal is black. The images are blurry due to the resolution of the technique.

Textural relationships and the selective dolomitization
observed in the Palliser Formation are important, as they
generate heterogeneous porosity, the flow properties of
which, if understood three-dimensionally, are tractable
(Gingras et al., 1999). Similar fabrics are common, espe-
cially in Ordovician and Devonian strata. Excellent examples
include the Ordovician Bighorn Dolomite in Wyoming
(Zenger, 1996); the Ordovician Red River Formation in
Saskatchewan and Manitoba, Canada (Kendall, 1977); a
Devonian example in the Camsell Formation, Northwest
Territories (Morrow, 1990); and the Palliser’s subsurface
equivalent, the Wabamun Formation.
Although hydrocarbon extraction is common from the Red
River and the Wabamun formations, the tortuous, hetero-
geneous fabrics, which locally constitute a burrow-hosted
reservoir, have received very little attention. This is partly due
to difficulties in deriving three-dimensional data for input into
modelling software. Quantifying the impact of heterogeneous
fabric on reservoir and resource quality is an exceptionally
important next step, if accurate estimates of deliverability and
recoverability are desired.
Such a clear example of patchy porosity distribution shows
that 1) fluid-transport interaction between the flow media
and the tighter limestone matrix are probably extensive;
2) patchy dolomite and its associated porosity, or lack of
porosity, are conceptually similar to fractured media; and 3)
these fabrics are more complex than fractured media.
Therefore, the potential presence of a complex reservoir fab-
ric must be considered, if enhanced recovery techniques are
planned for a reservoir.
MRI-SPRITE methods are an ideal tool for producing raw
data for flow modelling. The MRI technique has great potential
in flow modelling as it is passively sensitive to the effective
porosity present in a rock. Dynamic tests should be possible
with these techniques; such tests would enable researchers to
determine important parameters, such as bulk permeability,
MAGNETIC RESONANCE IMAGING AND PETROGRAPHY OF CARBONATE SEDIMENTS
hydrodynamical permeability, and dispersivity, using real
rocks. Although scaling issues are considerable, all of these
parameters, except dispersivity, are scalable. Dispersivity is
scale-dependant, but is a function of rock heterogeneity, which
can be assessed with MRI.
CONCLUSIONS
The data presented herein show the potential of MRI-
SPRITE as an imaging tool for low-porosity rocks. The
improved results offered by this technique demonstrate that the
technology to advance petrological studies with MRI tech-
niques is available. The results are notable because 1) they suc-
cessfully resolve the porosity distribution in a rock that is
characterized by a low overall porosity (generally less than
6%); 2) the three-dimensional data required to model the
porous network is accessible through the MRI technique; and
3) although the technique doesn’t measure the pore size and
pore distribution directly, these data can be retroactively calcu-
lated from the image if petrographical analyses are conducted
in conjunction with MRI analyses. More sophisticated MRI
experiments would probably provide detailed porosity data.
The link between petrographic evaluation and the interpretation
of the MR images will ultimately support nuclear magnetic res-
onance data obtained from subsurface techniques and should
improve interpretations derived from such data. Therefore,
future research might focus on dynamic MRI applications, and
to the effect mineralogy, pore character and pore distribution
have on the MR data.
ACKNOWLEDGMENTS
Tom Saunders, Floyd Henk, Arjun Keswani, Rosalia Pak
and Gladys Fong made extremely valuable contributions to
this manuscript, especially where burrow-related permeabil-
ity was concerned. Richard Evoy, William Martindale and
J.-P. Zonneveld improved the manuscript with their critical
reviews. The MR images were made in the MRI laboratory at
the University of New Brunswick. The operation of that lab-
oratory is made possible through a major facilities grant from
the Natural Science and Engineering Research Council
(NSERC) of Canada, an NSERC operating research grant,
Manuscript received: October 1, 2001
and Steacie Fellowship (2000–2002) to BJB. Sampling and
petrographic work was funded through an NSERC research Revised manuscript accepted: December 27, 2001
grant to MKG.
369
REFERENCES
Balcom, B.J., MacGregor, R.P., Beyea, S.D., Green, D.P., Armstrong, R.L. and
Bremner, T.W. 1996. Single-point ramped imaging with T1 enhancement
(SPRITE): Journal of Magnetic Resonance, v. A123, p. 131-134.
Beyea, S.D., Balcom, B.J., Bremner, T.W., Prado, P.J., Green, D.P., Armstrong,
R.L. and Grattan-Bellew, P.E. 1998. Magnetic resonance imaging and
moisture content profiles of drying concrete. Cement and Concrete
Research, v. 28(3), p. 453-463.
_______, _______, Mastikhin, I.V., Bremner, T.W., Armstrong, R.L. and
Grattan-Bellew, P.E. 2000. Imaging of heterogeneous materials with a
turbo spin echo single-point imaging technique. Journal of Magnetic
Resonance, v. A144, p. 255-265.
Brancolini A., Cominelli, A., Kulkarni, R. and Watson, A.T. 1997. Spatial dis-
tribution of petrophysical parameters on a core scale using magnetic reso-
nance imaging. Transactions of the Society of Professional Well Log
Analysts (SPWLA). Symposium, no. 38.
Chen S., X. Yao, J. Qiao and Watson, A.T. 1996. MRI investigations of frac-
tures and multiphase flow in fractured media. Journal of the American
Institute of Chemical Engineers, v. 42, p. 820-828.
Cross, A.R., Armstrong, R.L. and Grattan-Bellew, P.E. 1998. The influence of
shrinkage-cracking on the drying behaviour of White Portland cement
using Single-Point Imaging (SPI). Solid State Nuclear Magnetic
Resonance, v. 13, p. 93-100.
Gingras, M.K., Pemberton, S.G., Mendoza, C. and Henk, F. 1999. Modeling
fluid flow in trace fossils; assessing the anisotropic permeability of
Glossifungites surfaces. Petroleum Geoscience, v. 5, p. 349-357.
_______, MacMillan, B., Balcom, B.J., Saunders, T.D.A. and Pemberton, S.G.
2002. Using magnetic resonance imaging and petrographic techniques to
understand the textural attributes and porosity distribution in
Macaronichnus-burrowed sandstone. Journal of Sedimentary Research,
v. 72, no. 4, p. 552-558.
Hoffman, F., Ronen, D. and Pearl, Z. 1996. Evaluation of flow characteristics
of a sand column using magnetic resonance imaging. Journal of
Contaminant Hydrology, v. 22, p. 95-107.
Kendall, A.C. 1977. Origin of dolomite mottling in Ordovician limestones
from Saskatchewan and Manitoba. Bulletin of Canadian Petroleum
Geology, v. 25, p. 480-504.
Kenyon, W.E. 1992. Nuclear magnetic resonance as a petrophysical measure-
ment. Nuclear Geophysics, v. 2, p. 153-171.
Morrow, D.W. 1990. Synsedimentary dolospar cementation: a possible
Devonian example in the Camsell Formation, Northwest Territories,
Canada. Sedimentology, v. 37, p. 763-773.
Rack, F.R., Balcom, B.J., MacGregor, R.P. and Armstrong, R.L. 1998.
Magnetic resonance imaging of the Lake Agassiz-Lake Winnipeg transi-
tion. Journal of Paleolimnology. v. 19, p. 255-264.
Zenger, D.H. 1996. Dolomitization patterns in widespread “Bighorn Facies”
(Upper Ordovician), western craton, U.S.A. Carbonates and Evaporites,
v. 11, p. 219-225.