Engineer's Perspective A Review of Braided Fluvial Hydrocarbon Reservoirs: The Petroleum

Geological Society, London, Special Publications
A review of braided fluvial hydrocarbon reservoirs: the petroleum

engineer’s perspective
John H. Martin
Geological Society, London, Special Publications 1993; v. 75; p. 333-367

doi:10.1144/GSL.SP.1993.075.01.20
Email alerting click here to receive free email alerts when new articles cite this
service article
Permission click here to seek permission to re-use all or part of this article
request
Subscribe click here to subscribe to Geological Society, London, Special
Publications or the Lyell Collection
Notes
Downloaded by Gary James Hampson on 6 December 2007
© 1993 Geological Society of

London
A review of braided fluvial hydrocarbon reservoirs:
the petroleum engineer's perspective
JOHN H. MARTIN
Department of Geology, Imperial College of Science, Technology and Medicine,
London SW7 2BP, UK
Present address: Reservoir Geological Consultant, 150 Croxted Road,
London SE21 8NW, UK
Abstract: Braided fluvial reservoirs form some of the world's giant oilfields and are found in
many petroleum provinces. At best, they have excellent characteristics, but where non-net
intervals (shales and siltstones) form higher proportions, they are more difficult to appraise
and develop. Geological characterization (lithological correlatability, vertical sequence and
porosity/permeability variability) must be communicated effectively to the reservoir
engineer, whose role is to formulate a development plan and predict the production profile
for a field, and to monitor and optimise day-to-day field performance. He has at his disposal
other techniques: cased hole neutron logging to monitor contact levels; well testing to
determine reservoir flow capacity, boundary and layering effects; production logging to
assess well inflow performance or injectivity; wireline pressure profiling, field pressure and
production history analysis to define vertical permeability and to reveal reservoir
compartmentalization and/or communication with aquifer or injected fluids.
Computer reservoir simulation modelling, now an essential tool for management of most
major fields, has provided the impetus for advances in geological reservoir characterization.
The question is: how can we adequately represent braided fluvial reservoirs in such models?
At one end of the spectrum, sheet sands (although laterally variable) can be adequately
mapped and modelled as 'layered' systems based on conventional approaches. At the other
are multiple low-sinuosity channel sand reservoirs which, being uncorrelatable at well
spacings typical during appraisal of offshore fields, may be suitable targets for stochastic or
network - type approaches.
Although the behaviour of a hydrocarbon reservoir must be controlled at least in part by
its sedimentological characteristics, the engineer's view of a field may be somewhat
different to that of the geologist. This is as true for braided fluvial reservoirs as for those of
other environmental origin. The geologist and reservoir engineer must be fully aware of
each others' data, analytical methods and objectives in order to avoid wasted effort. In the
future, despite greater computing power and better modelling software, specialist technical
disciplines must not loose sight of the overall objectives of field development.
Small-scale geological variability affects the viability of enhanced recovery processes such
as miscible flooding. However, at the development planning stage, more generalized
parameter assessments may be sufficient. Reservoir characterization at the small scale, such
as geostatistical outcrop analogue studies, should be balanced by more geological
involvement in the planning and interpretation of well tests, which often provide the only
dynamic data available during the appraisal and development planning of offshore fields.
Greater cross-disciplinary understanding of reservoir characteristics could in turn ensure
that an engineer would not develop an incorrect view of a braided fluvial reservoir, perhaps
by correlating shaley intervals which just happen to be present at about the same location in
the section.
This p a p e r discusses some e n g i n e e r i n g aspects of m e t h o d s in d e v e l o p m e n t planning and

b r a i d e d fluvial reservoirs at the appraisal and m a n a g e m e n t , focusing w h e r e possible
d e v e l o p m e n t planning stage and during on b r a i d e d and low-sinuosity channel
d e v e l o p m e n t . Its objectives are: reservoirs.
(i) to summarize the reservoir characteristics It addresses the n e e d for collaboration b e t w e e n
of b r a i d e d fluvial sands; the geological and p e t r o l e u m e n g i n e e r i n g pro-
(ii) to illustrate the use of e n g i n e e r i n g data in fessions, and indicates s o m e possible future
m o n i t o r i n g field behaviour, with particular trends. It is a i m e d at readers unfamiliar with
reference to fluvial reservoirs; p e t r o l e u m e n g i n e e r i n g practise.
(iii) to discuss t h e use of reservoir simulation
FromBest, J. L. & Bristow, C. S. (eds), 1993, BraidedRivers,Geological Society 333

Special Publication No. 75, pp. 333-367.
334 J . H . MARTIN
Q 0il and i a a Norlh Rankin " ~'"# - ~
Fig. 1. Location of braided fluvial reservoirs. See Tables 1 and 2 for details of fields.
Reservoirs about 4% of world cumulative oil production.

Remaining proven gas reserves are more diffi-
Published data provides basis for review cult to quantify but may be at least 40 TSCF,
Braided fluvial and related reservoirs indicated comparable to around 2% of world cumulative
on Fig. 1 are summarized in Tables 1 and 2. In gas production (extrapolated from Tiratsoo
the space available it is not possible to discuss 1990).
individual fields in reservoir trends. Information The largest braided fluvial reservoir, and the
on many fields is, in any case, not freely avail- fifteenth largest of all oil fields, is the supergiant
able. From English language publications it Prudhoe Bay Field (Wadman et al. 1978; Atkin-
is currently impossible to make confident son et al. 1988,1990) recently producing at a rate
interpretations of even the most significant equivalent to around 70% of that of the entire
fields within the former Soviet Union, which United Kingdom Continental Shelf and supply-
doubtless include a number of braided fluvial ing up to 15% of the oil consumed in North
reservoirs. Screening of the many hundreds America. The second largest, the Sarir C-Main
of onshore oilfields in North America proved Field (Sanford 1970) has contributed approxi-
impractical, as in the petroleum engineering mately 0.2% of the world's cumulative oil pro-
literature which formed the main data source for duction, while the nearby Messla and Bu Attifel
this review, environmental interpretations of Fields (Table 2) are just two of some 16 separate
many individual fields are either vague or omit- fluvial Sarir Group pools in the Libyan Sirte
ted entirely. Basin. In the Arabian peninsula, the Alif Sand
forms a prolific recently discovered reservoir
trend; the 700 MMSTB Alif Field has produced
Braided fluvial sands: high quality with at a peak rate approaching 200 MSTBD.
major reserves Although little is published, its smaller neigh-
bour, the Azal Field (Table 2) has the same
Significance. Developed or appraised oil fields reservoir. In the Cooper Basin of Australia
in which major reservoir units are of braided Permian braided fluvial sands flow gas. A
fluvial origin probably contain remaining proven braided fluvial sand also produced the highest
recoverable oil reserves of at least well over gas test rate recorded in South Australia
30 000 MMSTB in petroleum provinces includ- (Malavazos & McDonough 1991). More signifi-
ing Alaska, the North Sea and the Sirte Basin of cantly, braided fluvial facies contribute much
Libya. (Tables 1 and 2; Fig. 1). This is approxi- of the natural gas liquids produced from
mately equal to original UK total oil reserves, or Australia's North West Shelf (Harris 1981).
PETROLEUM ENGINEER'S PERSPECTIVE 335
Some of the J-Sand reservoirs in the Malay Basin duction behaviour of the Morecambe Field
are likewise interpreted to be of braided fluvial (Woodward & Curtis 1987). Although sedi-
origin (J. Stiles, pers. comm.). mentological studies indicated considerable
In the UK the Sherwood Sandstone reservoir depositional heterogeneity, for field manage-
of the Wytch Farm Field (Dranfield et al. 1987; ment purposes a much simpler model has so far
Bowman et al. 1993), now producing at some 60 been adequate.
MSTBD, makes it the largest onshore oil field in (ii) The low expected recovery of the Aracas
Western Europe. The same reservoir produces Field (Nascimento et al. 1982) is probably partly
gas in the offshore East Irish Sea Basin (Bushell a result of poor aquifer pressure support which
1986; Stuart & Cowan 1991). Bunter sands has not been compensated by sufficient injected
(Table 1) are significant producing gas reservoirs water.
in the southern North Sea (Ketter 1991; Cooke- (iii) Partial barriers between the oil leg and
Yarborough 1991) whereas development of aquifer in the Prudhoe Bay and some Sarir
Westphalian channel gas sands (Green & Slatt Group Fields are caused by geochemical effects
1992; Ritchie & Pratsides 1993) is scheduled. rather than sedimentary characteristics.
(iv) The production performance of the Amal
Characteristics. Braided fluvial reservoirs are Field in Libya and Buchan in the Moray Firth
usually sheet-like with high net/gross ratios, (Table 1; Edwards 1991; Benzagouta & Turner
porosity and permeability. Oil recovery factors 1992) is controlled by natural fracturing, which
are generally high (up to more than 50% of is responsible for increasing the productivity of
STOOIP) either by natural depletion (which otherwise low permeability rock matrix caused
typically takes advantage of strong regional by deep burial and/or diagenesis.
aquifer support) or by 'engineered' recovery Thus it appears that many braided fluvial
using fluid injection (Tables 1 and 2). Reservoirs reservoirs, although laterally variable, can be
mainly are in internal pressure communication adequately mapped and modelled as 'layered'
with common field-wide contacts (e.g. Azal, systems (reservoirs which consist of vertically
many Sarir Group fields). Separate fault block stacked sheets of different properties) based
compartments are observed in only a few of the on conventional geological approaches (Weber
fields listed. In many, pressure drops due to & Van Geuns 1990) although stochastic
production propagate across intra-field faults, approaches may be useful for modelling dis-
indicating their non-sealing or only partially continuous shales (see below). Simple con-
sealing nature. Vertically separated reservoir ceptual models (Richardson et al. 1987b) may
'flow units' (that is, distinct horizons whose be appropriate for engineering purposes.
production characteristics differ from those of
other intervals) occur only if laterally continuous
non-fluvial shales are interbedded (Geehan et al. Alluvial f a n reservoirs: m o r e variable
1986). Most low permeability layers within
channel sands are laterally discontinuous and The distinction between 'fluvial' and 'fan'
effective vertical permeability is generally good models is somewhat arbitrary (e.g. Esmond
(e.g. Bu Attifel Field; Table 2). In the Pulai Complex; Table 1) so few reservoirs can be
Field of the Malay Basin sandy intervals are reliably assigned to this category. The Barrancas
separated by layers of continuous shales such Formation of the Vacas Muertas Filed in
that separate development of the reservoirs is Argentina (Table 2) is poorly sorted with a high
required (J. Stiles pers. comm.). Although the clay matrix but nevertheless has performed well,
sands have high net/gross ratios there are shaley whereas the Haima Group of south Oman
streaks and it would be easy to assume that (poorly described in the literature; parts may
vertical flow might be restricted. However, the have been deposited on an extensive fan) con-
combination of short permeability barriers tains thick stacked sequences of fine-grained,
within the braided fluvial section and more laminated or occasionally channelized sands.
continuous shales between sands provides effec- Although oil-bearing in a number of major fields
tive gravity drainage. Oil drains vertically including Marmul, well productivities are
through the reservoir sands and collects on the typically modest as a result of high fines content.
shales from where it continues to drain down-
structure to the producers. Transitional channel types:
There are exceptions to these generalizations
c o m p l e x reservoirs
but most result from non-sedimentological
features. Several fluvial reservoirs which have a higher
(i) Differential diagenesis above and below a proportion of non-net silts and shales are
palaeo gas-water contact controls the pro- interpreted as being composed of partly discrete
336 J.H. MARTIN
Table 1. Selected braided fluvial reservoirs, Northwest Europe

Esmond
Field Brent Buchan Caister B Caister C Complex Heidrun
Location Offshore UK Offshore UK Offshore UK Offshore UK Offshore UK Offshore

Norway
Age U. Triassic- Devono- Triassic Carboniferous Triassic M. Jurassic
L. Jurassic Carboniferous
Group U. Old Red Bacton Coal Measures Bacton Fangst
Sandstone
Formation Statfjord -- Bunter Sst -- Bunter Sst Garn
Status Developed Developed Appraisal Appraisal Developed Appraisal
Type Gas & oil Oil Gas Gas Gas Oil & gas
Drive Edge water Depletion/ Depletion Depletion Depletion Gas cap
mechanism water expansion?
Area (sq km) 39 14 7.5 15 39 40
STOOIP 580 466 -- -- n/a
(MMSTB)
Estimated oil 54% 18% 'high'
recovery
Max. prodn 200 28 'high'
(MSTB/D)
GIIP (BSCF) 3241 w 156 230 n]a 1000
Estimated gas 1215 BSCF 75% 81% 533 BSCF n/a
recovery
Max. prodn n/a 280t 280t 200 n/a
(MMSCF/D)
No. of wells 46 12 n/a 3 21 8
Well spacing 750-1500 700-1500 n/a n/a 300-2000 1000-3000
(m)
Engineered Flank water Water injection?
recovery drive/
miscible gas
Net/gross ratio 0.35-0.9 n/a 0.84-1.00 0.67-0.80 0.70-0.95 n/a
Porosity 0.16-0.29 0.07-0.11 0.11-0.30 0.06-0.15 0.09-0.24 0.27-0.35
(fraction)
Permeability 20-10 000 0.1-2 (matrix) 1-1000 0.1-400 n/a 9000
(mD)
Interpretation Braided/ Braided Channel/ Braided/ Braided/ Braided/
meandering sheetflood low sinuosity alluvial fan meandering
Fluvial facies Dominant Dominant Dominant Dominant Dominant Subordinate
importance
References Johnson & Benzagouta & Ritchie & Ritchie & Ketter (1991) Harris (1989)
Krol (1984) Turner (1992) Pratisides Pratisides
Struijk& Edwards (1991) (1993) (1993)
Green (1991)
Keijzer &
Kortekaas
(1990)
Description is specific to braided fluvial part of mixed reservoirs.

Well spacing q u o t e d is that at date of most significant published information.
Oil and gas in place values are those assigned to fluvial parts of reservoir only.
All figures are approximate.
Interpretation is based on published accounts.
* Gas storage, scheme i m p l e m e n t e d .
t Figures not separately assigned.
n/a: data not available (or cannot be assigned to fluvial part of reservoir).
- - not present, not appropriate or not applicable.
s a n d b o d i e s . T h e s e c o r r e s p o n d to t h e ' l a b y r i n t h ' using c o n v e n t i o n a l geological t e c h n i q u e s . T h e

r e s e r v o i r t y p e o f W e b e r & V a n G e u n s (1990); S t a t f j o r d F o r m a t i o n ( T a b l e 1) is t h e b e s t - k n o w n
t h a t is, a r e s e r v o i r in w h i c h flow p a t h s a r e N o r t h Sea e x a m p l e : as a result of t h e relative
extremely tortuous and probably unpredictable c o m p l e x i t y o f its i n t e r n a l a r c h i t e c t u r e , o f t e n
Hewett Morecambe Murdoch Snorre Snorre Statfjord Wytch Farm
Offshore UK Offshore UK Offshore UK Offshore Offshore Offshore UK/ Onshore UK

Norway Norway Norway
Triassic Triassic Carboniferous L. Triassic L. Jurassic U. Triassic- Triassic
L. Jurassic
Bacton -- Coal Measures -- --
Bunter Sst Sherwood Sst -- Lunde Statfjord Statfjord Sherwood Sst

Developed Developed Appraisal Appraisal Appraisal Developed Developed
Gas Gas Gas Oil Oil Oil Oil
Depletion Depletion -- Water (weak) Water (weak) Edge water Water (weak)
(weak)
190 170 n/a 50 30 35 34
-- -- -- 2080* 2080r 1500 350
-- -- -- 21-41% 32-42% 42% n/a
250 60
n/a 6750 ~a
'very good' 80% ~a
n/a 1220 280?

30 17 (40) ~a 11 11 30 49
300-500 500-4000 n/a 2000-4000 2000-4000 1000-2000 n/a
Water injection Water injection Miscible gas Water injection

(Water
injection)
0.88-98 0.56-0.85 n/a 0.35-0.68 0.4-0.63 0.50-0.72 n]a
0.19-0.23 0.14-0.15 n/a 0.19-0.29 0.19-0.29 0.18-0.23 0.16
250-1000 1-100 1-400 320-535 1300-2000 1500 5O
Alluvial plain Braided Braided Braided Braided/ Braided/ Braided

low sinuosity meandering?
Dominant Subordinate Dominant Dominant Dominant Dominant Dominant
Cooke- Bushell (1986) Green & Stanley et al. Stanley et ai. Buza & Dranfield et al.
Yarborough Slatt (1992) (1990) (1990) Unneberg (1987)
(1991) Nybrfiten et al. Nybrfiten et al. (1987) Bowman et al.
(1990) (1990) Roberts et al. (1993)
Hollander Hollander (1987)
(1987) (1987) Haugen et al.
(1988)
uncorrelatable at well spacings typical during whereas in o t h e r fields a low sinuosity to braided
appraisal of offshore fields, it has b e e n the sub- interpretation is favoured (e.g. Martin e t al.
ject of recent studies which aim to predict sand 1988; Buza & U n n e b e r g 1987). W h a t e v e r its
b o d y connectivity and fluid flow characteristics interpretation, g o o d reservoir m a n a g e m e n t is
using u n c o n v e n t i o n a l m e t h o d s (Martin e t al. r e q u i r e d to avoid p o o r reservoir sweep (poor
1988; Keijzer & Kortekaas 1990; Stanley e t al. displacement of oil by aquifer influx or injected
1990). I n t e r p r e t a t i o n of the formation, which fluids) and/or p r e m a t u r e aquifer or injected
forms an i m p o r t a n t reservoir in two of the water b r e a k t h r o u g h . Nevertheless, m o d e r a t e
largest N o r t h Sea fields, is the subject of s o m e recoveries (at least 2 0 - 4 0 % of S T O O I P ) should
debate: J o h n s o n & Krol (1984) suggested a be possible. In fact the Statfjord reservoir of the
m e a n d e r i n g c h a n n e l origin in B r e n t Field, Statfjord Field is used for t e m p o r a r y storage of
338 J . H . MARTIN
Table 2. Selected braided fluvial reservoirs, rest of world
Field Aracas Azal Bu Attifel Katnook Messla
Location Onshore Brazil Onshore Yemen Onshore L i b y a Onshore Australia Onshore Libya
Age U. J u r a s s i c Cretaceous L. Cretaceous n/a L. Cretaceous
Group -- -- Sarir n/a Sarir
Formation Sergi Alif -- Pretty Hill Sst --
Status Developed Developed Developed Developed Developed
Type Oil Oil & gas Oil Gas Oil
Drive mechanism n/a Gas cap expansion/ n/a Depletion Water
water
Area (sq km) 8 13 55 4 230
STOOIP 113 n/a n/a -- 3000
(MMSTB)
Estimated oil 8% 142 MMSTB 15% to date 50%
recovery
Max. prodn n/a n/a n/a 100 (per well)
(MSTB/D)
GIIP (BSCF) n/a n/a
Estimated gas 5.7BSCF m n/a
recovery
Max. dailyprodn -- n/a n/a
(MMSCF/D)
No. of wells 37 20 40 3 100
Well spacing (m) n/a 500-1000 1000-2000 n/a 2000-4000
Engineered Water injection Gas injection Waterinjection
recovery
Net/gross ratio 0.95 n/a n/a n/a n/a

Porosity (fraction) n/a 0.16-0.18 0.80-0.16 n/a 0.17
Permeability n/a 500-1200 1-1000 11-250 500
(mD)
Interpretation Braided Braidplain/ Braided Braided Braided
channel fill
Fluvialf a c i e s Dominant Dominant Dominant Dominant Dominant
importance
References et al.
Nascimento Huurdemanetal. ErbaetaI. (pers. Malavazos& Cliffordetal. (1979)
(1982) (1991) comm.) McDonough
(1991)
Notes: see Table 1.
produced gas from the Brent Group (Buza and M i x e d reservoirs: c o m p l e x b u t ultimately
U n n e b e r g 1987). Gas storage is only possible in
predictable?
good quality reservoirs.
The Peco Field (Table 2) is interpreted as a Other fields produce from sequences which
braided fluvial system (Gardiner et al. 1990) but comprise a variety of depositional environ-
its elongate sand body geometry could perhaps ments, and thus correspond to the 'jigsaw
favour a single composite channel. Unusually puzzle' reservoir type of W e b e r & Van Guens;
low net to gross ratios (0.3) and estimated that is, different shaped units of different
recovery factor of only 10% of STOOIP does properties interleaved in a complex, but
not correspond to the general pattern. Poor ultimately predictable, fashion. The fluvial Garn
reservoir performance is due to severe dia- Formation of Heidrun (Table 1), currently
genesis. While stimulation can increase the awaiting development, is the best reservoir in
production from individual wells by increasing the field. Interpretation is controversial (Harris
the effective permeability in the wellbore area, 1989): some workers favour a meandering
the permeability in the bulk of the reservoir, fluvial origin although from the engineering
which controls the displacement process, point of view, this distinction may be irrelevant
remains low. as the formation is extensive. The mature South
North Rankin Peco Prudhoe Bay South Belridge Sarir C-Main Vacas Muertas
Offshore Australia Onshore Canada Onshore Alaska Onshore USA Onshore Libya Onshore Argentina
Trias-L. Jurassic Cretaceous Permo-Trias Pleistocene L. Cretaceous Cretaceous
n/a -- Sadlerochit -- Sarir
n/a Belly River Ivishak Sst Tulare -- Barancas
Developed Developed Developed Developed Developed Developed
Gas condensate Oil & gas Oil & gas Oil Oil Oil
Depletion Solution gas/ Gravity drainage/ Solution gas/ Bottom water Depletion/water
gas cap gas cap gravity drainage
expansion expansion
50 19 1140 50 800 n/a
-- 34 215000 n/a 8000 400
-- 10% 56% 1200 MMSTB 25% to date 38.5% to date
-- 4 1450 170 350
n/a n/a 46 500 -- m
7700 BSCF n/a 7000 BSCF --

(todate)
n/a n/a n/a --
14 (50 planned) 61 900 6100 300 n/a

500-4000 400 569-805 61-152 1000 n/a
Gas recycling Fraccing Water injection Steam flooding Water injection Water injection/
Water alternating (suspended) ? caustic flooding
gas
0.75-0.83 0.3 0.87 0.3-0.7 0.5-0.95 n/a
0.17-0.21 0.06-0.13 0.22 0.32-0.42 0.13-0.17 17.5
500-1500 1-150 400 100-10 000 30-600 217
Braided Braided/single Braided Braided Braided Alluvial fan

channel
Dominant Dominant Dominant Subordinate Dominant Dominant
Barker & Vincent Gardiner et al. Atkinson et al. Gates & Brewer Sanford (1970) Simlote et al. (1985)
(1988) (1990) (1990) (1975)
Beston (1986) Atkinson etal. Miller etal. (1990)
Harris (1981) (1988)
Wadman et al. (1978)
Belridge Field (Table 2) comprises s e p a r a t e p e r m e a b i l i t y distribution (e.g. J o h n s o n &

a c c u m u l a t i o n s within stacked fluvio-deltaic Stewart 1985; V a n d e r G r a a f & E a l e y 1989 a n d
sands (some b r a i d e d ) w h o s e distribution a n d r e f e r e n c e s therein). This p r o d u c e s a static
depositional origin is i n t e r p r e t e d f r o m the m o d e l . E n g i n e e r i n g t e c h n i q u e s , on the o t h e r
e x t r e m e l y d e n s e well control, dating back to h a n d , p r e d i c t and m o n i t o r d y n a m i c (pressure
1911. R e c o v e r y of the viscous oil is p o o r , but a n d fluid flow) b e h a v i o u r ( A r c h e r & Wall 1986).
shallow burial and high p e r m e a b i l i t y have m a d e Since the 1970s oil c o m p a n i e s have u n d e r s t o o d
the field an attractive c a n d i d a t e for E O R s t e a m that the two disciplines m u s t a d o p t a synergistic
injection projects, w h i c h d e m a n d g o o d u n d e r - a p p r o a c h ( H a l b o u t y 1976; R i c h a r d s o n et al.
standing of the g e o m e t r y of target zones. 1977; Harris & H e w i t t 1977; W a d m a n et al.
A r c h e r 1983).
T h e p e t r o l e u m e n g i n e e r ' s role
Field appraisal focuses on reservoir
Dynamic modelling requires synergy continuity
T h e reservoir geologist describes a h y d r o c a r b o n A n oil c o m p a n y requires cash flow, which in turn
reservoir's external g e o m e t r y , lithological relies on field d e v e l o p m e n t (Corrigan et al.
correlatability, vertical s e q u e n c e and porosity/ 1990). F u t u r e i n c o m e g e n e r a t e d f r o m a field is a
340 J . H . MARTIN
Fig. 2. Idealized production profile (modified from Dake 1978). The initial part of the profile showing rapid
build-up to a production plateau is typical of offshore development, whereas the long decline period with
secondary or enhanced recovery is more typical of onshore fields.
function of its production profile (Fig. 2) pre-

dicted by the petroleum reservoir engineer from
a reservoir model. Engineering analysis carried
out on a new discovery must establish its size and
commercial viability. The engineer must estab-
lish the likely reserves (Garb & Smith 1989) and
production rate of the field, its depletion
mechanism (the processes which allow the pro-
duction of hydrocarbons) and well numbers
required for efficient production. At this stage
reservoir continuity is one of the main causes
for uncertainties: the main problem is charac-
terizing inter-well space from a limited number
of observation points (Fig. 3).
In the earliest stages a 'rule of thumb'
recovery factor may be applied to determine an
order of magnitude for the potential reserve
based on knowledge of similar situations (Arps
1967; Fielding & Crane 1987; Garb & Smith
1989). This will be refined as appraisal/develop-
ment drilling and reservoir performance data
become available. The geometry and the
geology of the reservoir are of the first
importance, and the engineer depends very Fig. 3. Location and spacing of appraisal wells on the
heavily on the accuracy of reservoir description Snorre Field, offshore Norway (after Nybrhten et al.
1991) compared with a map of part of central and east
given by the geophysicist and geologist (e.g. London, showing motorways and trunk roads. As a
Barker & Vincent 1988). In many situations the crude analogy, predicting the geometry of individual
engineer has available more computational channel sands in the reservoir from limited well data
power than can be used effectively, the may be likened to an attempt to draw this road map
limitation in models is often the physical based on observations made at a few random
description of the reservoir in terms of corre- locations.
lation and continuity of sand and shale layers

(e.g. Nybr~ten et al. 1990). Monte Carlo 0 Producer ~
techniques are frequently used at this stage
to gain a feeling for possible uncertainties 0- ObservationWell
(Haldorsen & Damsleth 1990).
9 0 9
.,~ iojecto, .~K -I- ~
Development planning often requires
'engineered recovery'
Field development is a plan to optimize
economic recovery from a reservoir or reser-
250m
voirs. The development plan defines the capital
investment required for the field. This includes
the number of wells, the processing require-
ments and requirements for fluid injection. In
the past, development of an onshore oilfield
could be divided into three phases.
Primary: in which oil recovery relies on the
natural drive mechanism of the reservoir,
including natural aquifer influx or gas cap expan-
sion etc. (e.g. Sanford 1970; Exploration Staff of
the Arabian Gulf Oil Company 1980). J/ ~L/
Secondary: in which production is boosted by
pressure maintenance schemes; mainly water
flooding (e.g. Yusun et al. 1985, 1989; Zhiwu Fig. 4. Well locations of an EOR pilot (Martin &
et al. 1989) or gas injection. Cooper 1984) superimposed on a detailed street plan
Tertiary: in which declining production is of the Piccadilly Circus area, London. The geologist
countered by the use of enhanced oil recovery and engineer will have much data with which to
(EOR) methods such as steam flooding (e.g. evaluate connectivity between injectors and
Gates & Brewer 1975). Onshore fields typically producers, but the flow pathways must be
show a long decline period, during which time characterized in detail.
EOR projects may be designed to recover (rela-
tively small amounts of) incremental oil (e.g.
Rupp et al. 1984; Simlote et al. 1985). Well particularly because of contract requirements
spacings associated with EOR schemes (particu- which may impose penalty clauses in the case of
larly pilot projects) may be very small (Fig. 4). non-achievement of agreed sales volumes. Most
However, modern practice (particularly off- development offshore NW Europe has been
shore where rapid return on investment is restricted to those reservoirs with good
demanded) has blurred these distinctions. These properties and high well productivity (Bushell
days 'engineered recovery' schemes requiring 1986; Stuart & Cowan 1991). Many have been
some form of fluid injection may be scheduled considered to be homogeneous 'tanks' which
very soon after field production startup (e.g. allowed development wells to be clustered at
Johnson & Krol 1984; Buza & Unneberg 1987). structural culminations, and completed high
Development offshore is much more sensitive to above potentially troublesome gas-water
uncertainty than development onshore, particu- contacts (e.g. Esmond, Forbes and Gordon
larly when development wells are 'pre-drilled' Fields; Ketter 1991). One result of this is lack of
through a subsea template prior to platform geological control on field flanks: in such cases it
installation, to shorten the build-up to plateau may be more accurate to estimate gas initially in
production. The development plan must be place (GIIP) from material balance (Dake 1978,
determined without the benefit of substantial p. 25) than from geological maps, especially
dynamic data (in contrast to onshore fields, in where water influx is absent or limited.
which pilot production from the first wells can Problems arise, however, during equity deter-
provide information in order to fine-tune the mination of fields which straddle licence block
overall development). boundaries as flank structure controls the areal
distribution of GIIP, and thus the share of
Gas field development. Gas reservoirs pose reserves and production allocated to each
different problems (Simpson & Weber 1986) owner. Furthermore, lack of peripheral wells
342 J . H . MARTIN
can also leave no way to monitor contact move- simulation of model test responses. Well tests
ments on the flanks (see below). can be used to estimate bulk reservoir properties
(which control well and field behaviour) because
they are relatively insensitive to local hetero-
Production optimization maximizes geneities. Test analysis is an inverse problem:
economic recovery model parameters are inferred by analysing
model response (pressure transient) to a given
After initial development, the engineer's task is input (flow rate transient). The main drawback,
to monitor reservoir performance and to however, is that a model may act like the reser-
maximise economic recovery through planning voir even though physical assumptions are
of infill wells (wells drilled to decrease the over- invalid.
all grid spacing of producers), recompletions A powerful analytic tool made possible by
(modifications to perforated intervals), secon- computer-aided interpretation is non-linear
dary and advanced recovery methods. As reser- regression, or automated type-curve matching.
voir data become available, understanding of This differs from graphical techniques in that
reservoir continuity increases, and supplements it uses a mathematical algorithm to match data
stratigraphic and lithogical information to a chosen reservoir model. Matching is
(Knutson 1976; Lorenz et al. 1975; Chancellor achieved by changing the values of unknown
& Johnson 1988). Offshore, this may be too late reservoir parameters until the model and the
for major investment decisions, but will control data fit as closely as possible. However, non-
redrilling and recompletion plans. linear regression analysis must be comple-
mented by visual diagnosis of the data so that the
Engineering tools engineer can select the most appropriate reser-
voir model. Geological input is critical.
Well test analysis helps reservoir
characterization Test design. The test period has to be sufficient
to reach that part of the reservoir response which
Background. Drill stem tests (DSTs), effectively is of interest. The analyst must determine
temporary completions of exploration and whether the maximum flow rate attainable will
appraisal wells; and production tests of perma- provide sufficient pressure change over the part
nent completion intervals in development wells of the reservoir most diagnostic of the unknown
have for many years been used to determine reservoir parameters. One of the best approaches
fluid type, well productivity rate, productivity to design is to perform a computer simulation of
index and identify wellbore damage (Dake 1978; the test with prospective values of the reservoir
Archer & Wall 1986). Boundary and layering parameters, and then examine the simulated
effects can be determined by analysis of pressure data to see if it is able to provide valid estimates
response (Gringarten 1987; Ehlig-Economides of the required parameters. Tests can be
1987; Horne 1990; Ayestaran & Karakas 1990). designed to validate the depositional geometry
Many methods consider the pressure build-up of a reservoir interval interpreted by the
following a flow period normally of a few hours geologist, or alternatively to help differentiate
or days duration. between two possibilities.
Major recent advances in well test design and Sequential DSTs in which the perforated
interpretation include: interval is successively increased can be very
(i) higher precision, high frequency data, useful if planned and interpreted with care
(Schlumberger 1992a). Unfortunately in older
both for flow rate and pressure;
(ii) software data presentation and interpret- fields a high proportion of tests is commonly
uninterpretable due to technical problems,
ation using graphical presentations to
identify different characteristic flow insufficient pressure response or co-mingled
production from several perforated invervals.
periods (Matthews 1986) during a test;
(iii) use of pressure derivative plots, which
provide simultaneous presentation of Application. Despite sedimentological com-
plexity as observed in core or outcrop
log Ap versus log At and Atlog Op/Ot versus
log At (Fig. 5); (Rudkiewicz et al. 1990; Miall 1988) test
(iv) non-linear regression methods (automated response from some braided fluvial reservoirs
type curve matching). shows that the engineer may consider them to be
essentially homogeneous within the region of
Well tests are increasingly used for reservoir investigation (Fig. 5).
characterization, through computer analysis and This behaviour, however, differs from
(a)
I,,,.
co
03
i
13_
I o i~
03
03 .. i ~ i .......................
c
0
03
c
E
a
-1
0 1 2 3 z
Dimensionless Time
(b) 3520
oOaatz~m
3500
(o
13_
v
3480
(1)
11.,.
03
03
3460
12.
3440
3420
0.1 1.0 10.0
dt (hours)
Fig. 5. (a) Pressure build-up test analysis from a braided fluvial reservoir, showing features typical of a
homogeneous reservoir, with a radial flow pattern. 'Dimensionless variables' (Horne 1990) are used in test
analysis to reduce the number of unknowns, and provide solutions that are independent of any particular units
system. Dimensionless pressure and time are linear functions of actual pressure and time. The shape of the
derivative curve can be used to infer the best model from which to seek the best interpreted solution.
(b) Logarithmic plot of pressure versus build-up time for a less typical test in the same field. The diagnostic
change in slope is interpreted as the effects of heterogeneity (a homogeneous reservoir would yield a single
straight line on this type of plot).
responses from low sinuosity channels within a reservoirs (Martin et al. 1988). Well tests w e r e
Statfjord F o r m a t i o n interval in which pressure used to estimate characteristic channel width
response c o n f o r m e d to a linear inflow pattern, (below 100m, for 3 m thick channels). This
rather than radial flow typical of extensive p r o v i d e d i n d e p e n d e n t support for the geological
344 J . H . MARTIN
a) GEOLOGICAL SCHEMATIC interpretation. However, single channel models

were not sufficient to account for pressure
response during the latest period of build-up,
which indicated a degree of interconnectedness
(Fig. 6). Although more than one model could
account for this pattern, an 'expanding flow'
network of channels was geologically defensible
(Figs 7 & 8). Well test analysis software was used
to establish typical 'repeat lengths' between
b) EQUIVALENT NETWORK
I." INTERSECTIOI~
'
WELL LOCATION
POINT j successive expansions.
Well tests have also been used to estimate the
dimensions of southern North Sea Carbon-
iferous channel reservoirs (P. Haynes, pers.
comm.) and in Australia (Malavazos &
McDonough 1991). Wadsley et al. (1990)
describe a successful match of a long term
(6 month) production test followed by a 3 month
build-up using computer test analysis combined
with stochastic channel modelling (see below).
Fig. 6. Abstraction of a braided fluvial system for well Well tests can also determine the effective
test analysis and simulation: plan view (after Martin radius of a non-correlatable permeability barrier
et al. 1988).
observed in a well (Fig. 9). Some testing tech-
niques have more general applicability: let us
imagine that there are two sandy intervals
FULL WELL MODEL separated by a shale whose lateral extent is in
question. One can investigate this by perforating
L 226m the sand with the lower permeability • height
product (the 'flow capacity' of an interval is the
8W 3W W -0- 3W 9W
product of permeability, k, and interval height
I_ .,6~, LI h). This kh term, expressed in mD ft or mD m,
I_ e75m ~
r -L
~ e76m ml
is a fundamental property of a perforated
interval which can be determined independently
~ e75m ~ = 675m ~ by appropriate test analysis. Perforations can
WELL L O C A T I O N then be added in the other sand, with the initial
perforation still open. The well would again be
Fig. 7. 'Expanding flow' model for well test analysis tested, and the k h again interpreted. If the two
using computer methods; derived from Fig. 6 (after k h values are the same, it is quite likely the
Martin etal. 1988). The value for W was derived from
conventional test analysis. The repeat length was shaley interval is not laterally continuous.
modified iteratively until a satisfactory match was Conversely, different kh values would suggest
obtained between measured and predicted pressure. that it was continuous for some distance.
The offset well location with respect to the first Well tests can also be used to estimate vertical
expansion was also suggested by iterative pressure permeability. This is done by perforating the
matching.
INITIAL PRESSURE - SIMULATION

INITIAL PRESSURE - M[AaUIIIED
%oo
=o
o MEASURED DATA
9 SIMULATION RESULTS
O O
"'~ 0 IAR
i l oi 1o;o ~oloo i
10000
T -+- ~t
ZXt
Fig. 8. Excellent pressure match obtained between model (Fig. 7) and raw data. (After Martin et al. 1988).
(a)
5000 ft Layer
Gas / Oil contact
Shale barrier
190 ft
Perforations
Shale barrier
(b)
r
O_
0
~) R = 150fl
a) ............ ...=**==.=.=.~176176176
""~'"F;'nai match R=250fl

4==m
.>- 102
~._,,~176176176176176176176176 ]
~176176176 ~176176176
~176176176176176176
if)
/
ca. 101 R = 150~'"..-- "'".. R = 250ff
r-
ff)
e~
~
Or) 100
O3
I I I I
10 .3 10 .2 10 -1 10 ~ 101
Elapsed time (hrs)
Fig. 9. Use of non-linear regression techniques for estimating the length of a permeability barrier penetrated by
a well (after Ayestaran et al. 1989). (a) Well test simulation model. (b) Sensitivity analysis on length of barrier.
poorer section of an interval and running a contribution from the other interval. If it does,
welltest for a long period, maybe a day or two there are analytical techniques to calculate
(depending on the permeability level). The vertical permeability from such a test.
shape of the pressure plot will differ depending However, one must remember that the
whether or not the pressure transient 'feels' the volume of reservoir 'interrogated' by a test
346 J . H . MARTIN
depends on factors including its duration, and vertical communication between units. The
the permeability of the reservoir. Most tests only technique has particular value in planning E O R
provide information on average reservoir projects, in which it is critical to establish
characteristics up to several hundred metres whether producers and injectors communicate.
away from the wellbore. Nevertheless, properly Pressure changes at a distance from the producer
conducted and interpreted tests provide much are much smaller than at the well itself, so inter-
independent data with which to characterize ference tests require sensitive pressure recorders
dynamic reservoir behaviour. and long duration. The technique is not appro-
priate to all reservoirs: the fluid system should
have low compressibility, well spacing should
Interference testing can establish be small, and pressure perturbations from
surrounding areas should be minimal. Although
reservoir c o m m u n i c a t i o n
no readily accessible account of interference
In an interference (or pulse) test one well is testing in a braided fluvial reservoir is in the
produced and pressure is observed in a different public domain, such reservoirs are attractive
well or wells, to evaluate either the degree of candidates for interference testing as their high
lateral communication (Britt et al. 1991) or the permeability will allow shorter flow periods.
Fig. 10. Flowmeter log showing restriction of inj ectivity to highly permeable intervals, Prudhoe Bay Field. Track
A shows original flow profile dominated by high permeability thief zone which takes 95% of injected water.
Following profile modification with sand pack, injection profile is more uniform. (After Atkinson et al. 1990).
Production logging demonstrates flow tester ( R F r ; Schlumberger 1981) and similar

contribution of individual zones tools allow the engineer or geologist to investi-
gate pressure communication in partially
Background. Cased hole wireline flowmeter depleted reservoirs (Stewart & Ayestaran 1982;
surveys (Schlumberger 1989) quantify the rela- Gunter & Moore 1987; Dickey 1989). The tool is
tive proportion of total flow entering the casing positioned at successive points across a reservoir
from specific perforated intervals. Other interval and direct fluid pressure readings
measurements allow the interpreter to deter- obtained. Offsets in fluid pressure gradients in
mine the nature of the fluid in the wellbore. infill wells in partly depleted reservoirs indicate
Flowmeters and/or temperature logs can also be permeability barriers or partial 'baffles' (Fig.
run in injector wells to check on where the fluid 12). Correlation schemes must take into account
is entering the reservoir, and can thus form RFT surveys, as they will indicate the presence
part of a surveillance package for waterflood of separate flow compartments (although RFTs
operations (Thakur 1991). must be interpreted with care, as the pressure
profile will reflect production: if pressure
Application. Flowmeter logs are often directly decline in two adjacent but non-communicating
related to medium scale reservoir heterogeneity
observed from cores or logs. Interpreted flow- reservoirs is balanced, no differential will be
observed). Although now standard in the North
meter surveys from braided fluvial reservoirs
(Figs 10 & 11) show high permeability zones Sea since its introduction in the mid-1970s, in
many operating areas the RFT is not used as
dominating flow. The kh term (defined above)
much as it should be.
is proportional to the amount of fluid flowing out
of (or into) an interval. The reservoir geologist
is concerned with predicting which intervals will Application. RFT responses in braided fluvial
dominate flow. Unfortunately many geologists reservoirs (Fig. 13) show good vertical com-
display measured or predicted permeability vs. munication in most sand bodies, with only
depth plots on semi-logarithmic scales. How- localised evidence for compartmentalisation,
ever, linear plots are more useful as they give a particularly in sequences with higher pro-
much better representation of flow capacity portions of overbank deposits. Normally, fine-
(Fig. 11). grained intervals do not form pressure barriers
within individual sands (as most shales are
Wireline formation pressure profiling relatively short, except for intercalations of
establishes differential depletion lacustrine facies between channel sands). The
RFT is the best way of identifying any con-
Background. The wireline repeat formation tinuous shales.
Fig. 11. Comparison of permeability profiles (from cores and logs) with productivity of perforated intervals
(from flowmeter) in a braided fluvial reservoir. Note restriction of flow to high permeability intervals.
348 J . H . MARTIN
PRESSURE
\ gradi~ ~t
measurements
A C
Fig. 12. Schematic RFT pressure profiles in infill wells from partially depleted fields. (a) Effect of two
permeability barriers; (b) 'baffle zone'- (c) fully communicating (after Schlumberger 1981).
Cased hole nuclear logging monitors 1985). The thermal decay time (TDT) tool
(Schlumberger 1989), and other pulsed neutron
contact m o v e m e n t
tools are through-casing chlorine detectors:
Background. The engineer needs to know if the provided that formation water is salty, and reser-
oil-water contact (OWC) in a reservoir moves voir quality good, sequential runs of this tool will
over time (Westaway et al. 1979; Libson et al. allow changes of O W C with production to be
Fig. 13. RFT pressure profile from braided fluvial reservoir. The main reservoir zone shows good quality partly
depleted sands in pressure communication. At the top there is an isolated sand. Note the pressure offset between
the oil leg and aquifer due to permeability reduction by a geochemical barrier. Aquifer pressure gradient
indicates disequilibrium possibly associated with the effect of a baffle zone caused by discontinuous low
permeability streaks (channel abandonment drapes and overbank deposits) observed in the permeability profile.
monitored. This involves logging wells at

different times and overlaying the logs to detect
differences. Today a wide range of cased hole
nuclear tools (Schlumberger 1989) is used to
monitor OWC and gas-oil contact (GOC)
behaviour.
Application. The TDT log of Fig. 14 shows a

typical response of a braided fluvial reservoir
with an extensive aquifer. Despite sedi-
mentological variability, in this field OWC rises
predictably in line with production. This is
typical of a bottom water drive reservoir. Only
limited evidence of bypassed oil (Felder 1988) is
found (in any case, bypassed oil observed on a
TDT log could actually be producible from an
up-dip well location). In contrast, isolated
channel sand reservoirs may show evidence of
lateral water encroachment due to unfavourable
permeability profile leading to poor vertical
sweep efficiency (Fig. 15). In fact the permea-
bility profiles associated with different
depositional environments play a great role in
determining recovery u n d e r different drive
mechanisms: this can be demonstrated by the
use of simple simulation models (Poston &
Gross 1984).
Johnson & Scanlon (1991) describe the use of
pulsed neutron logging in the Esmond Complex.
Haugen et al. (1988) used case hole neutron logs
to monitor the growth of a secondary gas cap in
the Statfjord Formation of the Statfjord Field.
Logging techniques to monitor the movement of
Fig. 14. Two TDT logs run in a braided fluvial gas caps are also used routinely in the Prudhoe
reservoir showing evidence for bottom water rise: the Bay Field, where high pressure drawdowns have
second log, run four years after the first, does not led to gas 'cusping' beneath extensive shale
overlay it across the lower part of the reservoir. barriers (Fig. 17). More typical of braided
sequences, however, are discontinuous barriers
Fig. 15. Water encroachment at base of fining-upward channel sand.

350 J. H. MARTIN
Fig. 16. Gas coning. Arrows indicate perforated intervals, which act as a pressure sink. A shale protects the
perforations from downward gas movement. An analogous situation may occur at an oil-water contact, where
wells may be completed above tight streaks to prevent upward water coning.
which do not extend between production well field-wide pressure surveys made while wells are
spacings. Such barriers can nevertheless be of not producing (shut-in) will reveal reservoir
extreme importance in limiting either gas or compartmentalisation or communication with
water 'coning' (Fig. 16; Richardson et al. 1987a) aquifer or injected fluids. Many braided fluvial
even if they extend only relatively short dis- reservoirs are in good pressure communication
tances: well completion design should take (Fig. 18) as a result of good rock quality, the low
advantage of any such intervals. permeability streaks not forming significant
pressure barriers. The geologist is, however,
cautioned against over-interpretation of field-
Pressure and production history trends wide or single well pressure v. time plots:
mechanical problems such as communication
help reservoir understanding between reservoir and shallow aquifers (recog-
Review of field pressure history since pro- nized or unrecognized) can cause pressure
duction start-up, based on regular measure- anomalies and equipment failure or tran-
ments of pressure from dedicated observation scription errors can sometimes not be excluded.
wells (completed across a reservoir sand but Close co-operation with the reservoir engineer
used neither for production nor for injection) or is required.
Fig. 17. Gas cusping and water tongues in the Prudhoe Bay Field resulting from extensive shales (lacustrine
intervals). Note how early gas and water breakthrough occurs despite perforations some distance from OWC and
GOC. (After Haldorsen & Chang 1986).
d~
Aberrant~
points: casing I
.... .... leaks.
tO
tO
Main reservoir:
D. good aquifer
support.
Rapid initial
pressure decline
due to partial
500 psi I Low pressures in I "" isolation.
stratigraphically Later increase in
separated sand. pressure results
from short period of
water injection
coupled with lower
5 years field production rate.
Time since commencement of production
Fig. 18. Pressure trends in a braided fluvial field.
Production decline curve behaviour of 1985; Archer & Wall 1986; Fisher 1989; Coats
individual wells can be investigated to determine 1989; Dalton & Mattax 1990 and references
remaining oil reserves or productive life (Garb therein).
& Smith 1989). Cumulative production and time Grid points or nodes are each assigned a value
are selected as independent variables. Of the for permeability, thickness, porosity, fluid
many dependent variables that can be con- content, elevation and pressure at pre-
sidered, the rate of production is most appro- production conditions. Initial rates for produced
priate when production is not restricted. If fluid phases are assigned to the wells in the field.
wellbore and mechanical conditions are satis- Then, for a finite time difference new saturations
factory, a declining production trend can be and pressures for each of the grid nodes are
extrapolated to predict remaining recoverable calculated, which in turn allows new rates for
reserves if there is a long production history each of the producing wells to be estimated. This
and the drive mechanism is unlikely to change. process is repeated for a number of time steps, to
However, the technique is inappropriate if calculate rate and pressure histories for each
production/injection distribution or drive well in the system, as well as saturation and
mechanism changes and although the engineer pressure histories for each grid node.
will inspect production data to monitor prob- A reservoir simulator allows the engineer to
lems caused by water or gas breakthrough into investigate various operating schemes, select the
perforated intervals (e.g. Chancellor & Johnson optimum from the cases which were run and
1988), computer reservoir simulation models are produce answers which could not have been
now more regularly used to predict future arrived at intuitively:
performance.
(i) determine field performance under
depletion or fluid injection operations;
(ii) predict the effect of well spacing, flood
Reservoir simulation: pattern or producing rate on recovery;
conventional approach (iii) calculate critical well rates for water or
gas coning;
Why simulate reservoirs? (iv) evaluate the benefits of workovers, stimu-
lation or gas lift.
Computer reservoir simulation incorporates
multi-dimensional, multiphase reservoir Simulation modelling has been a significant
analysis to investigate the behaviour of discrete advance in field development planning, and
field regions over time, taking into account allows sensitivity analysis (evaluation of un-
changes in saturation resulting from pressure certainties in input parameters) and prediction
drops induced by production (Arnovsky et al. which would not otherwise be possible. How-
352 J. H. MARTIN
ever, it is simply a 'tool' for use alongside other Despite the value of simulation, many
geological and engineering methods. It is too engineers use hand calculations and analytical
easy to get carried away by the numerical and techniques to obtain answers to many reservoir
graphical output of a reservoir simulator: mis- questions. In general one should use the simplest
takes have been made as a result of faith in model possible to solve a particular problem and
simulator output, without critical evaluation of only go on to more complex models if required.
uncertainties in key input data. For example, in braided stream reservoirs, gas
or water coning (defined below) can be a critical
question and single well models may be
Different model types address specific adequate early on. If the reservoir is layered by
the presence of continuous shales, cross-section
problems models may give answers to most of the major
Single well models have radial geometry con- questions. Thankfully, one does not necessarily
sisting either of a single sector or a cylinder need a full field simulation model to solve every
extending away from the well bore. Cell radii problem.
decrease logarithmically or geometrically
towards the well bore to give greater detail close
to the perforations. Single well models are
usually employed to investigate coning and Steps in performing a simulation study
cusping phenomena and to design production Geological input is critical
rates to limit encroachment of undesirable fluid
phases; also to design and simulate well tests Layering. Fluid movement during drainage is
or determine optimum completion policy. The principally controlled by permeability; thus the
required geological reservoir description is correct distribution of permeability is the major
location-specific and more detailed than for factor in successful modelling. Ideally, the actual
other model types. permeability profile from each well should be
used in modelling, but except in finely gridded
Cross-section models allow the representation of single well or cross-sectional models, this is
detailed layering effects and structural dip of the impracticable.
reservoir, to: The reservoir is instead broken down into a
manageable number of layers, each of which
(i) calculate and validate averaged permea- must be provided with the key parameters of the
bilities for application in coarse grid reservoir, not only at the position of the wells but
models, along with a simplified vertical extrapolated between them. The process of
grid description; layering the reservoir is purely for the con-
(ii) investigate mechanisms of pressure support venience of engineering analysis and inevitably
(both natural and assisted) and the influ- leads to loss of detail. The reservoir geologist
ence on the vertical efficiency of the dis- must try to ensure that all critical parameters are
placement process caused by layer and retained during this process. Layers may follow
inter-layer properties; a sedimentary facies or be purely artificial (e.g.
(iii) evaluate completion method. in a complex, uncorrelatable reservoir in which
They require careful interwell, layer and inter- 'slice mapping' may be adopted). The vertical
layer reservoir description in terms of capillary dimension of layers must be determined with
pressure, relative permeability and vertical/ regard to the nature of the reservoir and of the
horizontal permeability. model required. For example in a cross-sectional
model it may be possible to explicitly represent
3D models. Sector or full field models (Barnes layers on a metre scale, whereas in full-field
1989) are usually based on rectilinear or dis- models layer thicknesses may typically be on a
torted grids and can combine results from more 10 m scale.
detailed models into a coarse areal and vertical The layering process requires permeability to
description, which includes all interwell effects. be averaged across specific zones. How do we
Full field models are used to determine overall generate these averages? The first step is to
field performance (Starley et al. 1991; Wood & ensure that permeabilities derived from well
Young 1991) but may not be detailed enough to tests are broadly in line with those determined
adequately represent small-scale effects. They from cores or predicted using some combination
should, however, be re-usable throughout of log response. In braided fluvial sequences
development and subsequent production of a without fracturing or delicate clay mineral mor-
field. phologies this should not be a major problem
bearing in mind that results are unlikely to be 1986; Van der Graaf & Ealey 1989; Slatt &
exactly equal due to differences of scale or possi- Hopkins 1990).
ble sample bias.
We then have to determine which type of
average is most appropriate. For modelling hori-
H o w to use the m o d e l
zontal flow in a layered system, an arithmetic
average will probably be most suitable. For History matching. Once a model is constructed
modelling vertical flow, a harmonic average, and running it is necessary, wherever possible,
which produces a result biased towards the to validate assumptions and input data. History
lowest values, may be chosen. However, this matching is the procedure of adjusting reservoir
model assumes lateral continuity of low permea- description (as given in a simulation model) to fit
bility beds: clearly inappropriate in the case observed data (pressure, field or well per-
of most braided fluvial sands containing un- formance). Historical information is available as
correlatable shales (Martin & Cooper 1984). In well rates of individual phases and flowing/static
such reservoirs it is most appropriate to choose pressures. With the advent of the RFT tool layer
layer boundaries at any continuous (lacustrine?) pressures in new wells are also available: these
shales which form pressure barriers. Within the make very good layer-specific match criteria,
layers effective vertical permeability (on the whilst flowmeter data can assist in ascribing
scale of the simulator unit) can be very difficult flowrates and phase contributions from indi-
for the geologist and engineer to specify. It is vidual reservoir layers. Obviously, history
unlikely to be equal to vertical permeability matching cannot be performed before pro-
determined from core plug analysis. Many duction commences, which is why simulation
methods for determining vertical permeabilities modelling is less reliable at the field appraisal
have been proposed: some rely on statistical stage.
techniques discussed below (e.g. Begg & King History matching gives the engineer the
1985; Begg et al. 1985; King 1990). However, opportunity to adjust some of the least well-
most engineers still rely on the simple method of defined parameters against actual performance
assigning one or more vertical/horizontal before predictions are made. As more reservoir
permeability ratios to the layers within a model. history becomes available, so simulation models
The sensitivity of simulation results to changes become more reliable representations. History
in vertical/horizontal permeability ratio is matching is not a unique process since several
normally evaluated. variables could usually be modified to obtain a
match. Each modification must be both geo-
Mapping. For 3D simulation, structural and logically and in engineering terms both reason-
reservoir property maps are required for each of able and defensible: history matching should be
the simulation layers. However, the simulation collaborative.
model may at best only be able to handle a Neither the engineer nor the geologist should
fraction of the complexity of a full reservoir be fooled into believing that assumptions made
description. The most important map which is have given a unique solution: matches can often
required is that of permeability. A plot of be made by several combinations of reasonable
porosity against the logarithm of permeability is data. Even a 'good' history match does not
often used to assign permeabilities in uncored guarantee a successful prediction. History
wells. However, where significant differences matching can be one of the most contentious
exist between core- and test-derived estimates, issues during an engineering study, and con-
this method gives an entirely false precision. siderable man- and computer-time can be
Note that the facies of high permeability zones expended in trying to achieve an acceptable
may have a different orientation compared to match. Experience shows that accurate reservoir
other permeability layers, and permeability description, and co-operation between all
itself may be directional, especially in channel relevant technical specialists will speed this
sands. process.
The account above, based largely on Corrigan
et al. (1990) is only a brief introduction; further Prediction. This encompasses the simulation of
descriptions of geological support required for individual well and total field production pro-
simulation studies are given by Harris (1975), files with corresponding hydrocarbon recovery.
Archer (1983, 1987) and Weber & Van Geuns It will usually include alternative development
(1990). A repeated theme is that of scaling: of strategies. The most attractive strategies would
geological heterogeneity and of geological, be refined with further runs incorporating fine
petrophysical and engineering tools (Haldorsen tuning adjustments. The optimum exploitation
354 J . H . MARTIN
strategy would be selected on the basis of

economic evaluation of a few cases.
Application to braided fluvial reservoirs

Evaluation o f well performance
Huurdeman et al. (1991) use single well models
to investigate how halite-cemented layers in a
sand control water and gas coning (Fig. 19).
Layer rock properties were derived from logs
and cores from several wells, while different
possible configurations of reduced permeability Fig. 19. Single well model (after Huurdeman et al.
halitic zones were tested against observed 1991). The diagram shows the radial grid construction,
pressure, gas-oil ratio (relative volumes of gas plus the fine layering adopted to model fluid flow in
and oil flowing) and watercut (water production the region of the perforated interval.
expressed as a percentage of total liquid flow-
ing). Martin et al. (1988) used single well models
to attempt to match DST results in several
appraisal wells: these models used a standard connected nature of the reservoir both areally
radial and concentric grid, but stochastic and vertically (Fig. 20): since confirmed by
techniques (see below) were used to assign pressure surveys.
values for rock properties.
Full-field models of fluvial reservoirs adopt a
more generalized approach. A more recent
Scaling-up flow effects model of the Brent Group and Statfjord Forma-
Tollas & McKinney (1991) used cross-section tion in the Brent Field (Tollas & McKinney
models to define 'pseudofunctions' for use in a 1991) comprised 34000 grid blocks (30 east/
full-field model (see below). Further discussion west and 54 north/south; dimensions 150 •
of pseudofunctions is outside the scope of this 300m within the oil leg, with larger aquifer
paper: 'rock curves' of relative permeability and blocks). The Statfjord Formation was divided
capillary pressure (important properties which into four layers (pressure history match of one of
control fluid flow and displacement) derived these layers is shown in Fig. 21a). Tollas &
from core plug measurements are often inappro- McKinney (1991) show clearly how the results of
priate for use in field calculations (Dake 1978, the simulation (e.g. Fig. 21b) were used to
ch. 10) so they may be replaced by values calcu- optimise future field performance. It became
lated to represent gravity and scale effects. clear that several of the platforms would be
constrained by gas handling capabilities for
some time: upgrades were put in hand following
3D models to predict field performance
the study.
Sector models. Johnson & Krol (1984) describe Buza & Unneberg (1987) review the results of
the integrated geological and engineering the first and second post-production simulation
approach necessary to achieve a satisfactory exercises of the Statfjord reservoir of the
reservoir simulation of the Statfjord Formation Statfjord Field, which matched production
during the early stage of development of the periods of 21 and 44 months respectively. Gas-
Brent Field. The main problem was to ade- oil ratio, water cut and RFT pressure gradients
quately represent the contrast between channel in infill wells were used as matching parameters.
sands and overbank shales. Grid block thick- The models correctly predicted the time of first
nesses (4.5-9m) were chosen to be represen- gas breakthrough. Haugen et al. (1988) used a
tative of individual channel dimensions deter- later model (11000 active grid-blocks in seven
mined by sedimentological study. Although by layers) to ensure that oil production was maxi-
modern standards the model was small (4000 mized, and overpressuring by gas injection
grid blocks) it had a relatively fine vertical avoided. Full field modelling of the Prudhoe Bay
definition (12 layers) which allowed the inactive Field was particularly difficult because of the
channel and overbank shales, which together high number of wells. In such cases, a more
form (largely discontinuous) permeability generalized approach to history matching well
barriers, to be incorporated into the simulation performance is often adopted (O'Brien et al.
in some detail. The model predicted the inter- 1984).
Fig. 20. Sector model: (a) cross-section and areal view illustrating development strategy investigated by the
model. (b) Cross-section illustrating predicted fluid saturations following production period (Johnson & Krol
1984). Reproduced with permission of the Society of Petroleum Engineers.
Non-conventional reservoir has been found that the stochastic model gives
only a starting point for history matching,
characterization and simulation although it can be speeded by the initial
Stochastic models can quantify uncertainty stochastic approach.
Two points must be stressed.
Background. Since Haldorsen & Lake (1984) (i) There is no justification for using stochastic
suggested that stochastic modelling could be an techniques unless the geologist is unable
effective tool in reservoir modelling, techniques to derive a convincing deterministic corre-
have been refined particularly to estimate effec- lation. Successful models must in fact be
tive vertical permeability in reservoirs contain- hybrid: based on any overall deterministic
ing discontinuous shales (Martin & Cooper framework, and honour (i.e. be 'con-
1984; Begg & King 1985; Haldorsen & Chang ditioned' to) observed well data.
1986; Haldorsen et al. 1987). 2D (or more (ii) A realization is not 'reality'. Early models
recently 3D) 'realizations' are created, based on suffered from lack of computing power,
geological input data including length distri- time and lack of an interface between the
bution statistics derived from outcrop stochastic model and the simulator. For
analogues, and observations of permeability this reason, insufficient realizations were
barrier location and thickness from cores. constructed or tested. However, recent
Having created one or more realization(s) the advances have made it much simpler to
detailed model(s) are translated into a reservoir create and simulate more realizations, to
simulation. This has been attempted using obtain a better idea of the uncertainty
simple 'building blocks' (Martin & Cooper 1984) range of the predictions.
or shale frequency maps and statistical pro-
cedures (Begg & King 1985; Begg et al. 1985). Shale distribution models for fluvial reservoirs.
Where production history has been available it Many braided fluvial reservoirs are basically
356 J. H. MARTIN
(b) 100 12000

~'..: .., ,',., ~','. ....
_,
,....,,,~
!, :../'/
6000
600
l< HISTORY MATCH ~l_qI PREDICTION ~ I
..,= WATE RFLOOD ,,.- I _. DEPRESSURIZATION ,,,._ I
LU
Q: - 2500
C
- 2O011 ~ ~
~m
8~ 300- gO
15o0 m~
.-I-10
~oo
'1000 O
0
c
0
- 500
z
0 i i i 0
1975 1980 1985 1990 1;95 21:~ 2005
(a)
62OO
=o..
56oo. 0 RFT MEASUREMENT
~lb SIMULATOR BLOCK
G)
Fig. 21. Full field simulation: Brent field. (a) History
SIMULATOR AVERAGE
540084
match against pressure; Statfjord Formation Unit 1;
1978 1980 1982 1984 1986 (b) prediction of field performance, Statfjord and
YEAR Brent reservoirs (after Tollas & McKinney 1991).
sheet sands with intercalated permeability C h a n n e l models. Partly isolated low sinuosity
barriers, many of which are discontinuous and ribbon sands developed in non-reservoir matrix
uncorrelatable on a well spacing of several form the other end of the spectrum. Develop-
hundred to thousands of metres. This is the basic ment planning of the Statfjord and Lunde
model followed in studies of the Prudhoe Bay Formations (and distributary sands in the Ness
Field (Geehan et al. 1986; Haldorsen & Chang Formation of the Brent Group) in offshore
1986) Dranfield et aI. (1987) built a more prospects with well spacing of several thousands
complex hybrid model, which incorporated the of metres has been difficult. This led to con-
distribution and properties of reservoir facies as siderable interest, particularly by the Norwegian
well as shales, during appraisal of the Sherwood oil industry, in integrated stochastic channel
Sandstone reservoir of the Wytch Farm Field sand software modelling packages.
(Fig. 22). With development drilling, however, (i) 'SISABOSA' (Augedal et al. 1986; Stanley
it has become clear that the overall continuity et al. 1990) which creates a reservoir by placing
of reservoir facies is higher than originally 3D parallelepipeds of permeable sand into a
envisaged, and a more deterministic model has volume of impermeable mudstone until a pre-
recently been introduced (Bowman et al. 1993). scribed sand content and sand content variation
INPUT AVERAGING SCALE OUTPUT

Whole Core
---T'T'--,
Core-plug poroperm PDF's
for each rock type
Identify rock types

I
I Rock Type 2
KH KV
which have different I

poroperm characteristics
L~ Rock Type 3
/ L KHE and K VE
for each rock type
{
f f e ctiv e p er m e a bl'iltle:~kr Te~Pch~ o ck
type
i| i i
Upper B
.....
"~ 2 9 -2 -
Proportion of each KHE and KVE

rock type In each layer for sand part of
-/- each layer
z
2
Middle
Determine effective permeabllitles for the sand
part of each layer
ShalepDF.slength
Shale length , ~ ~ .
ades | ~ U P P E R B~'--"--~,
. . . .
~._.__= UPPER C ~ '
Upper A I Overall
F~ I ~: ~m- ;> , v . KHE and KVE
for each
9 0
Shale length layer
> 0
~ PDF
Facies 10 ' Lower
Determine overall effective permeabllltles for

each layer
Fig. 22. Stages in construction of stochastic model for the Sherwood sandstone in the Wytch Farm Field
(Dranfield et el. 1987). PDF, probability density function; KH, horizontal permeability; Kv, vertical
permeability; KE, effective permeability. Reproduced with permission of Graham and Trotman.
358 J . H . MARTIN
Fig. 23. Stochastic model of channel sands: schematic representation (after Nybrhten et al. 1990).
is achieved (Fig. 23). Sandbody length is con- high degree of detail to a simulation grid?
sidered infinite; thicknesses are determined by Wadsley et al. (1990) offer one solution.
a probability distribution; widths from a con- (iv) Shell's M O N A R C H 3D reservoir model-
ditioned probability distribution that may be ling package, which primarily allows the
tied to thickness, while orientation is also operator to interactively create a deterministic
determined probabilisticall~r sand architecture scheme, also has stochastic
(ii) 'FLUREMO' (Clemetsen et al. 1990): capabilities (Keijzer & Kortekaas 1990).
a development of SISABOSA which allows Caution must be exercised before applying
individual sand bodies to be clustered within such models to braided fluvial systems. Input
channel belts. It calculates measures of inter- requirements are summarized by Martin et al.
connectedness and continuity ratio for describ- 1988: in my experience two of the most import-
ing the reservoir potential of each realisation, ant parameters controlling model behaviour:
and is able to transfer reservoir realisations to (a) channel width (or width-thickness relation-
a grid form suitable for input to reservoir ships) and (b) channel sinuosity are often hugely
simulators. uncertain.
(iii) 'SESIMERA' (applied to a fluvial system There are to date few published examples of
by Gundes0 & Egeland 1990) which uses high- the practical use of such approaches in field scale
resolution 3D grids for hybrid modelling of simulation for development planning: a major
heterogeneous reservoirs. As in all approaches a study of the Snorre Field by Nybr~ten et al.
problem remains; how to effectively reduce the (1990) and a briefer account of a revision to part
of the Brent Field model by Keijzer & Kortekaas studies is to develop workable practical tech-
(1990) which indicates that the stochastic model niques which may allow the geologist and
gives a better history match owing to a more engineer to predict subsurface lithofacies (or
realistic shale distribution. more importantly, permeability) distribution
between isolated well control points.
N e t w o r k models incorporate dynamic data Outcrop minipermeameter studies show that
within even sands considered relatively
Stochastic modelling is not the only channel homogeneous, horizontal correlation lengths for
modelling technique which deserves mention. permeability are normally only a few metres,
Martin et al. (1988) attempted unsuccessfully and less than a metre vertically. At present this
to validate single well stochastic models against level of heterogeneity can not be incorporated
observed test response, before matching test within workable field-scale simulators, although
pressure histories against an idealized 'expand- research interest in techniques for scaling up
ing flow' network (see above). This concept was such detail commenced with demonstrations
then used to create a very simple 3-layer areal some six years ago (e.g. Lasseter et al. 1986).
model (each layer, e.g. Fig. 24, was again An exception has been the application of geo-
matched against well-test response) to construct statistical models and fractal theory to estab-
a simple simulation model of a low sinuosity lishing realistic levels of heterogeneity within
channel sand reservoir, with which to investigate 2D models for simulation of E O R processes
sensitivity of a planned water injection project to (Hewett & Behrens 1988; Mathews et al. 1989;
different well spacings. The key was to use well Payne et al. 1991; summarized in Schlumberger
test results to independently verify sector model 1992a, b). Changing correlation lengths of
characteristics, in the absence of established permeability domains in a simulation model
production history, prior to prediction. The affects the performance of a miscible flood such
approach was geared towards reduction of as CO2 injection (Fayers 1991). Small scale
uncertainty, rather than quantification of heterogeneity clearly does have greater (detri-
uncertainty. mental) effects on miscible flooding than on
other recovery processes (Khataniar & Peters
Geostatistics, conditional simulation 1992): this has implications for E O R in Prudhoe
Bay and other Alaskan fields. Fingering of flood
and fractals: current research topics
fronts in reservoirs which contain discontinuous
Rudkiewitcz et al. (1990) apply the HERESIM shales is especially difficult to model using large
software system to reproduce 2D macro-scale grid blocks without introducing explicit permea-
heterogeneity of fluviodeltaic outcrops using bility barriers. Interestingly, the presence of
variograrn-based geostatistical methods. As small shales now appears to have little detri-
with all the other techniques discussed above, mental effect on at least some conventional oil
the objective of geostatistic outcrop-based recovery processes (Thomas 1990).
I I
.a w.oT._I00m
~ ! ...
I
b : BARRIER WIDTH 13m !i
I
a c : GAP WIDTH - 50m I t
I I
d : REPEAT LENGTH - 600m i= d =I
"////////1 _ WELL L O C A T I O N
,,
Fig. 24. Plan view of a simple element model to represent interconnecting channels derived from geological
representation in Fig. 6. Parameters which gave a satisfactory match are shown (Martin et al. 1988).
360 J . H . MARTIN
Discussion (4) Be aware of the limitations of subsurface

geological techniques. The main task of the
Geologists and reservoir engineers need geologist is to predict the 3D geometry of a
to c o m m u n i c a t e m o r e reservoir. We believe that the best way to do this
is to interpret its depositional environment b y
Studies reviewed above illustrate the synergy vertical sequence analysis of core. Yet how does
which should exist between the reservoir the geologist approach an outcrop study? With-
geologist and engineer during any field evalu- out doubt he would examine all 3D exposures
ation study. However, day-to-day relationships before feeling confident about his interpreta-
between the two may not always be optimal. Of tion. This is particularly important in the assess-
the work undertaken by the specialist sedi- ment of channel type. Thus there is an element
mentologist (particularly core description and of circular argument in geological work.
interpretation) many engineers believe, rightly (5) Predict large-scale permeability trends in
or wrongly, that only a limited amount is of the reservoir. Directional permeability is often
practical use: related to palaeocurrent trends. Yet with few
But the truth is, we engineers can't use a lot exceptions (Stanley et al. 1991; Green & Slatt
of information that geologists provide. 1992; Cowan et al. 1993) we are not good at
Geologists do a great service in explaining predicting palaeocurrent direction before
the nature of the rocks, the depositional drilling development wells, or in advance of
environment, the petrology - information problems (e.g. preferential water breakthrough)
that we don't use. Conversely, a number of for which anisotropic permeability is respon-
things we really do want to know, geologists sible. The new generation of high-resolution
can't or don't routinely provide. So our job dipmeter tools should be beneficial, especially
is to educate them. We need to make sure when combined with interactive systems for
that the earth scientists we work with know selecting features for dip computation rather
enough about reservoir simulation inputs, than relying on automatic correlation tech-
operations and outputs - - not in tremen- niques.
dous detail - - to know which of their tools The petroleum engineer should, in turn
will be most useful. (D. N. Meehan in undertake the following.
Anon. 1990, p. 27). (1) Inspect cores (or at least, core photo-
Whether or not one agrees with Meehan's view- graphs) as this is a very effective method for
point, he demonstrates the need to more effec- gaining a 'feel' for the reservoir. Brief visits to
tively 'sell' the implications of geological efforts appropriate outcrops will also usually give a
to a wider audience. With this aim, I list below better understanding of heterogeneity than
several practical tips. simple models described in geological text-
(1) Make geological results accessible to non- books.
specialists. The main results of development (2) Give the geologist specific guidelines for
geology studies must be expressed simply, information that will be critical and agree
clearly and briefly. mutally on the level of detail that will be
(2) Environmental classification is interpreta- required.
tion, not fact. A non-genetic facies scheme (e.g. (3) Ensure that assumptions made in
Miall 1978) is preferable at least during the early analytical studies are reasonable and consider
stages of studies. This should avoid scepticism also the uncertainty range.
when an environmental 'label' is changed as a (4) Accept that geological studies are bene-
result of additional information, or simply fical. Although it is reasonable to expect the
disagreed. geologist to quantify his observations and pre-
(3) Do not overstress idealised models. dictions, the study of geology relies on infer-
Except within the pages of basic textbooks ences made from limited data. It is unwise to be
fluvial sequences can not be 'pigeonholed' into sceptical of the value of geological studies
neat categories. Despite generalizations, exact because two or more geologists disagree
correlations between depositional environment (engineering calculation methods and data
and reservoir quality, geometry and (especially) themselves do not always give unequivocal
size are rarely observed. The geologist should answers!). Different or uncertain interpreta-
carefully explain the relationships between local tions often point to the requirement for further
tectonics, sea level fluctuations and sedimen- data to be acquired specifically to resolve the
tation rates which control the preserved rock problem, or for further studies to be under-
record. taken.
Future trends favour the technological size and complexity: a million grid block model
approach should be possible within five years. In parallel,
mathematical techniques are now established
Forecasting trends is always dangerous, and it is which (a) remove the need for grid blocks to be
enlightening to compare yesterday's predictions constrained to cartesian geometry, allowing
(e.g. Chierci 1985) with today's reality. During much better areal representation of geological
the early 1990s the structure of the oil industry detail, and (b) allow greater fexibility in order
changed dramatically. 'Down-sizing' and 'out- to optimise model runs, by refinement of model
sourcing' have been dominant trends. Oil structure in near-well regions, or gathering of
company reservoir engineers and geologists now blocks in non-critical areas. It should become
act more to co-ordinate, commission and possible to 'embed' detailed single well models
analyse results of specialist studies performed by within more generalized 3D models, and switch
subcontractors. Geologists, especially, must be seamlessly between the two to investigate
very clearly aware of the cost/benefit of the work different aspects of production performance.
which they recommend to be performed.
Reservoir surveillance. Integration of flowmeter
Reservoir description. Ideally, the petroleum data with reservoir description is a very powerful
engineer would like a seismic tomographic reservoir surveillance technique. One should
technique which will allow the reservoir to be run flowmeters to define the distribution of
visualised on a metre scale (Haldorsen & production and injection in a layered reservoir.
Damsleth 1991; Hill-van Beck 1991). This would In addition one should have a knowledge of the
allow him to concentrate on recovering hydro- permeability distribution in each well. Then,
carbons from a perfectly constrained rock with a knowledge of the pressure distribution the
volume. Despite advances in geophysical tech- expected flow distribution can be calculated,
niques this is over-optimistic. More achievable based on the permeability model. This should
would be further reduction of uncertainty then be compared with flowmeter results. Large
through the application of integrated geological, differences can identify intervals which are
petrophysical and fluid dynamic studies; in damaged and need to be stimulated. However,
particular, better planned and executed well we need improvements in technology which will
testing (including long term production testing) allow the engineer to accurately define the
with accurate pressure measurement and down- distribution of flow in highly deviated or hori-
hole fluid flow monitoring. zontal wells producing multiple fluids. Currently
most fowmeters in horizontal wells are useless
Geological modelling. Geological work once water breakthrough occurs (J. Stiles pers.
practices will change in line with new engineer- comm.)
ing capabilities. During the 1980s the transfer of
data (maps, correlations etc.) from the geologist Horizontal drilling. Horizontal wells have
to the engineer was often a bottleneck. Current become a viable alternative to conventional
integrated software packages which should vertical production wells (Freyss & Burgess
make this transfer virtually 'invisible' have some 1991). Annual activity is increasing at some
drawbacks: the next generation (e.g. Van et al. 250%, (although this represents only about 1%
1991) should allow common data sharing of total drilling world-wide). Costs of drilling
between disciplines. In the mid-1980s geologists horizontally have reduced to some 1.5 times that
realized that the computing power of reservoir of a conventional well. In clastic sequences, one
simulators was starting to outstrip our capacity of the main benefits of drilling horizontally is
to provide quantified input data (particularly of to reduce gas or water coning or cusping, for
permeability distributions): this led to efforts to example in the Prudhoe Bay Field, in which over
quantify and model heterogeneity levels from 20 horizontal wells have been drilled (Sherrard
outcrops. In the next five years we shall see how, et al. 1987; Wilkinson 1988). There are many
and if, this information can be used in practical other potential applications in braided fluvial
day-to-day field management. Despite interest reservoirs.
in their application the day-to-day use of expert
systems (e.g. Maher 1991) will probably take
Conclusions
longer.
Computer reservoir simulation modelling, now
Reservoir simulation. As computing power an essential tool for the engineer, has provided
becomes cheaper, simulation models increase in the impetus for advances in geological reservoir
362 J . H . MARTIN
characterization. The question is: how can we such fields could then be used in a likely range of
adequately represent braided fluvial reservoirs key reservoir description parameters to be used
in such models? At one end of the spectrum, in models to assess development requirements
sheet sands (although laterally variable) can be for a new field and help predict its likely
adequately mapped and modelled as 'layered' performance.
systems based on conventional geological Finally, despite ever-increasing use of
approaches. For many engineering purposes, sophisticated computer modelling in both the
typical braided fluvial reservoirs can be con- engineering and geological fields, it is still
sidered to fall within this category. At the other essential to have a sound understanding of basic
end of the spectrum we encounter reservoirs geological techniques, and also of the ultimate
composed of discrete but overlapping and inter- economic constraints governing field develop-
communicating sand bodies. Where the scale of ment.
this intercommunication is smaller than average
well spacing, stochastic modelling techniques Discussions with R. Lind and S. Berczi of Heinemann
may be effective. Oil Technology provided the starting points for several
The geologist must not forget that, despite the aspects of this synthesis. Sections on field develop-
recent interest in small scale reservoir charac- ment and geological input to simulation are partly
abbreviated from course notes originally compiled in
terization, large-scale features will dominate
A. F. Corrigan, while contributions from J. Stiles
many recovery processes. Greater involvement improved the discussion of well test analysis and
by the geologist in improved dynamic under- reservoir monitoring. A. Miall, A. Brayshaw and
standing through the careful application of well C. Braithwaite are also thanked for their helpful
testing, pressure profiling, production logging reviews of an earlier draft.
and reservoir monitoring should be beneficial.
This will be one way of addressing the problem
Principal abbreviations and units
that, despite the amount of reservoir charac-
terization research which is currently being B billion
pursued we cannot demonstrate that we are able DST drill stem test
to consistently economically recover signifi- EOR enhanced oil recovery
Cantly higher proportions of hydrocarbons-in- GIIP gas initially in place
place than would have been possible ten years M thousand
ago (Haldorsen & Damsleth 1991). However, mD millidarcy
we have a wealth of information from past MM million
performance of fields all over the world and RFT repeat formation tester
need to learn to use it more effectively. SCF standard cubic feet
Engineers have a vision of first identifying the SCF/D standard cubic feet per day
likely depositional environment of a reservoir STB stock tank barrels
and then identifying analogous fields drilled on STB/D stock tank barrels per day
close spacing to define sand/shale continuity) STOOIP stock tank oil originally in place
with good core control and well-documented T trillion
pressure - production history. Information from TDT thermal decay time
References
ANON. 1990. Round table discussion on reservoir generation. Oil and Gas Journal, November,
simulation. Oilfield Review, 2, 24--33. 75-80.
ABBOTS,I. L. (ed.) 1991. United Kingdom Oiland Gas ARPS, J. J. 1967. A statistical study o f recovery
Fields, 25 Years Commemorative Volume. efficiency. American Petroleum Institute
Geological Society, London. Memoir 14. Bulletin, D14.
ARCHER, J. S. 1983. Reservoir definition and charac- ATKINSON,C. D., McGowEN, J., BLOCH,S., LUNDELL,
terisation for analysis and simulation. Proceed- L. L. & TRUMBLY,P. N. 1990. Braidplain and
ings o f the 11th World Petroleum Congress, deltaic reservoir, Prudhoe Bay Field, Alaska. In:
PD6(1), 65-87. BARWIS, J. H., MCPHERSON,J. G. & STUDLICK,
- - 1 9 8 7 . Some factors influencing fluid flow in J. R. (eds) q.v., 7-29.
petroleum reservoir models. In: KLEPPE,J. et al. , TRUMBLY, P. N. & KREMER, M. C. 1988.
(eds). Sedimentology and depo-environments of the
--& WALL, C. G. 1986. Petroleum Engineering Ivishak sandstone, Prudhoe Bay Field, North
Principles and Practice. Graham and Trotman. Slope Alaska. In: Giant oil and gas fields: a core
ARONOFSKY, J. S., Cox, T. F. & LANE, S.M. 1985. workshop. SEPM Core Workshop No. 12,
Reservoir simulation moves into third Houston, 561-613.
AUGEDAL,H. O., OMRE, H. & STANLEY,K. O. 1986. BULLER, A. T., BERG, E., HJELMELAND,O., KLEPPE,
SISABOSA: a program for stochastic modelling J., TORS~TER, O. • AASEN, J. O. (eds) 1990.
and evaluation of reservoir geology. In: North Sea Oil and Gas Reservoirs - H. Proceed-
NITTEBERG, J. (ed.) Proceedings of a conference ings of the 2nd North Sea Oil and Gas Reservoirs
on reservoir description and simulation with Conference, Trondheim, Norway 1989. Graham
emphasis on EOR, Institute for Energy Tech- and Trotman.
nology. BUSHELL, T. P. 1986. Reservoir geology of the
AYESTARAN,L. & KARAKAS,M. 1990. Layered reser- Morecambe Field. In: BROOKS, J., GOFF, J. &
voir testing. Middle East Well Evaluation VAN HOORNE, B. (eds) Habitat of Palaeozoic Gas
Review, 9, 22-47. Schlumberger Technical in N.W. Europe. Geological Society, London,
Services, Dubai. Special Publications, 23, 189-208.
, NURMI,R. D., SHEHAB,G. A. K, & EL SISI,W. S. BUZA, J. W. & UNNEBERG,A. 1987. Geological and
1989. Well test design and final interpretation reservoir engineering aspects of the Statfjord
improved by integrated well testing and geo- Field. In: KLEPPE,J. et al. (ed.) q.v., 23-38.
logical efforts. Society of Petroleum Engineers CHANCELLOR,R. E. t~ JOHNSON,R. C. 1988. Geologic
Paper 17945 presented at Middle East Oil and engineering implications of production
Technical Conference, Manama. history from five Mesaverde wells in Central
BARKER, G. & VINCENT, P. 1988. Geological and Piceance Creek Basin, Northwest Colorado.
reservoir engineering considerations for the Society of Petroleum Engineers Formation
N. Rankin gas recycling project. Australian Evaluation, 3,307-314.
Petroleum Exploration Association Journal, 28, CHIERICI, G. 1985. Petroleum reservoir engineering in
54-66. the year 2000. Energy Exploration & Exploita-
BARNES, A. L. 1989. Approaches to full field simu- tion, 3, 173-193.
lation. Society of Petroleum Engineers Paper CLEMENTSEN, M. R., HURST, A. R., KNARUD, R. &
19041 presented at Middle East Oil Technical OMRE, H. 1990. A computer program for evalua-
Conference, Manama. tion of fluvial reservoirs. In: BtrLLER, A. T. et al.
BARWlS, J. H., MCPHERSON,J. G. & STUDLICK,J. R. (eds) q.v., 373-386.
(eds) 1990. Sandstone Petroleum Reservoirs. CLIFFORD, H. J., GRUND, R. & MUSRATI, H. 1979.
Springer Verlag, New York. Geology of a stratigraphic giant: Messla oil field,
BEGG, S. H. & KING, P. R. 1985. Modelling the effects Libya. American Association of Petroleum
of shales on reservoir performance: calculation Geologists Memoir, 30,507-524.
of effective vertical permeability. Society of COATS, K. 1989. Reservoir simulation. In: BRADLEY,
Petroleum Engineers Paper 13529 presented at N. B. (ed.) q.v., 48/1-48/20
Reservoir Simulation Symposium, Dallas. COOKE-YARBOROUGH, P. 1991. The Hewett Field,
- - , CHANG, D. M. & HALDORSEN,H. H. 1985. A Blocks 48/28-29-30, 52/4a-Sa, UK North Sea.
simple statistical method for calculating the In: ABBOTS, I. L. (ed.) q.v., 433-442.
effective permeability of a reservoir containing CORRIGAN,A. F., DAVIES,N., GUEST, W., KAYE, L. &
discontinuous shales. Society of Petroleum MARTIN, J. H. 1990. Advance reservoir
Engineers Paper 14271. geology: a North Sea perspective. Joint
BENZAGOUTA, S. ~z TURNER, B. 1992. Heirachial Association for Petroleum Exploration Courses,
analysis of reservoir quality in the Buchan Oil- 101.
field (Abstract). Braided rivers: form, process COWAN, G., OTrESEN, C. & STUART, I. A. 1993. The
and economic applications; a two-day meeting use of dipmeter logs in the structural interpreta-
held at the Geological Society of London. tion and palaeocurrent analysis of Morecambe
BESTON, N. B. 1986. Reservoir geological modelling of Fields, East Irish Sea Basin. In: PARKER, J. R.
the North Rankin Field, Northwest Australia. (ed.) Petroleum Geology of NW Europe: Pro-
Australian Petroleum Exploration Association ceedings of the 4th Conference, Geological
Journal, 26,375-388. Society, 867-882.
BOWMAN, M., MCCLURE,N. M. & WILKINSON,D. W. DAKE, L. P. 1978. Fundamentals of Reservoir
1993. Wytch Farm oilfield: deterministic reser- Engineering. Elsevier.
voir description of the Triassic Sherwood DALTON, R. L. & MATTAX, C. C. 1990. Reservoir
Sandstone. In: PARKER. J. R. (ed.) Petroleum simulation. Journal of Petroleum Technology,
Geology of NW Europe. Proceedings of the 4th 42,692-699.
Conference. Geological Society, 1513-1518. DICKEY, P. A. 1989. Application of pressure measure-
BRADLEY, N. B. (ed.) 1989. Petroleum Engineering ments to development geology. In: MASON,J. F.
Handbook. Society of Petroleum Engineers, & DICKEY,P. A. (eds) q.v., 105-108.
Richardson. DRANFIELD, P., BEGG, S. H. & CARTER, R. R. 1987.
BRITT, L. K., JONES, J. R., PARDINI, R. E. & PLUM, Wytch Farm Oilfield: reservoir characterisation
G. L. 1991. Reservoir description by interference of the Triassic Sherwood Sandstone for input to
testing of the Clayton Field. Journal of Petroleum reservoir simulation studies. In: BROOKS, J. &
Technology, 43,575-577. GLENNIE, K. W. (eds) q.v., 149-160.
BROOKS, J. & GLENNIE, K. W. (eds) 1987. Petroleum EDWARDS, C. W. 1991. The Buchan Field, Blocks
Geology of North West Europe. Graham and 20/5a and 21/la. UK North Sea. In: ABBo'rrs,
Trotman. I. L. (ed.) q.v. 253-259.
364 J.H. MARTIN
EHLIG-ECONOMIDES, C. A. 1987. Well testing in GUNDESO, R. & EGELAND, O. 1990. SESIMIRA

layered reservoirs. In: North Sea Oil and Gas a new geological tool for 3D modelling of
Reservoirs, Norwegian Institute of Technology heterogeneous reservoirs. In: BULLER, A. T.
(Graham and Trotman), 269-276. et al. (eds) q.v., 363-372.
EXPLORATION STAFF OF THE ARABIAN GULF OIL GUNTER, J. M. & MOORE, C. V. 1987. Improved use of
COMPANY 1980. Geology of a stratigraphic wireline tests for reservoir evaluation. Journal of
giant--the Messlah oil field. In: Second Petroleum Technology, 39,635-644.
Symposium on the Geology of Libya, Tripoli, HALBOUTY, M. 1976. Needed: more coordination
Sept 16-21, 1978, 521-536. between earth scientists and petroleum
FAYERS, F. J. 1991. The need for continued reservoir engineers. Society of Petroleum Engineers Paper
description and simulation for advanced 6107 presented at the Annual Fall Conference,
recovery technologies in large fields. Advances New Orleans.
in Reservoir Technology: Characterisation, HALDORSEN,H. H. 1986. Simulator parameter assign-
Modelling and Management, Edinburgh, ment and the problem of scale in reservoir
February 1991. Unpublished conference engineering. In: LAKE, L. W. • CARROLL,H. B.
proceedings. (eds) q.v., 293-340.
FEEDER, R. D. 1988. Cased-hole logging for evaluating --& CHANG, D. M. 1986. Notes on stochastic
bypassed reserves. Journal of Petroleum shales; from outcrop to simulation model.
Technology, 40, 969-974. In; LAKE, L. W. & CARROLL, H. B. (eds) q.v.,
FIELDING, C. R. t~ CRANE, R. C. 1987. An application 445-486.
of statistical modelling to the prediction of --& DAMSLEana, E. 1990. Stochastic modelling.
hydrocarbon recovery factors in fluvial reservoir Journal of Petroleum Technology, 42, 404-412.
sequences. In: ETHRIDGE, F. G. et al. (eds) --& -- 1991. Challenges in reservoir charac-
Recent developments in fluvial sedimentology. terisation. In: Advances in Reservoir Tech-
Society of Economic Paleontologists and nology: Characterisation, Modelling and
Mineralogists Special Publications, 39,321-327. Management, Proceedings of a conference held
FISHER, W. G. 1989. Reservoir simulation. In: MASON, on 21/22 February, Edinburgh. Petroleum
J. F. & DICKEY,P. A. (eds) q.v., 141-158. Science and Technology Institute.
FREYSS,I-I. P. & BURGESS,K. 1991. Overcoming lateral & LAKE, L. W. 1984. A new approach to shale
reservoir heterogeneities via horizontal wells. management in field scale simulation models.
Paper presented at the 6th European symposium Society of Petroleum Engineers Journal, August,
on improved oil recovery, Stavanger, May. 447-457.
GARB, F. J. & SMITH, G. L. 1989. Estimation ofoil and --, CHANG, D. M. & BEGG, S. H. 1987. Dis-
gas reserves. In: BRADLEY, N. I . (ed.). continuous vertical permeability barriers: a
Petroleum Engineering Handbook. Society of challenge to engineers and geologists. In: KLEPPE
Petroleum Engineers, Richardson. 40/1-40/38. et al. (eds) q.v., 127-152.
GARDINER, S., THOMAS, D. V., BOWERING, E. D. & HARRIS, D. G. 1975. The role of geology in reservoir
MCMINN, L. S. 1990. A braided fluvial reservoir, studies. Journal of Petroleum Technology, May,
Peco Field, Alberta, Canada. In: BARWIS,J. H., 625-632.
McPHERSON, J. G. & STUDLICK,J. R. (eds) q.v., -- & HEwn'r, C. H. 1977. Synergism in reservoir
31-56. management: the geologic perspective. Journal
GATES, C. F. & BREWER, S. W. 1975. Steam injection of Petroleum Technology, July, 761-770.
into the D and E zone, Tulare Formation, South HARRIS, E. I. 1981. Production geology of the North
Belridge Field, Kern County, California. Journal Rankin gas field. In: Energy Resources of the
of Petroleum Technology, March, 343-348. Pacific region. American Association of
GEEHAN, G. W., LAWTON,T. F., SAKURAI,S., KEOB, Petroleum Geologists, Studies in Geology, 12,
H., CLIFTON,T. R., INMAN,K. F. & NITZBERG, 273-283.
K. E. 1986. Geologic prediction of shale con- HARRIS, N. B. 1989. Reservoir geology of the Fangst
tinuity, Prudhoe Bay Field. In: LAKE, L. W. & Group (Middle Jurassic) Heidrun Field, offshore
CARROLL, H. B. (eds) q.v., 63-82. Mid-Norway. American Association of
GIEEESPIE,S. t~ SANDFORD,R. M. 1970. The geology of Petroleum Geologists Bulletin, 73, 1415-1435.
the Sarir Oil Field. In: Proceedings of the 7th HAUGEN, S. A., LUND, O. & HOYLAND,L. A. (1988).
World Petroleum Congress, Volume II, Statfjord Field. development strategy and
181-197. reservoir management. Journal of Petroleum
GREEN, C. & SLAT'f,R. M. 1992. Complex braided Technology, 40, 863-873.
stream depositional model for the Murdoch Field HEWETr, T. A. & BEHRENS, R. A. 1988. Conditional
Block 44/22 UK Southern North Sea. Braided simulation of reservoir heterogeneity with
rivers: form, process and economic applications; fractals. Society of Petroleum Engineers Paper
a two-day meeting held at the Geological Society 18326 presented at the 63rd Annual Technical
of London. (Abstract). Conference, Houston.
GRINGARTEN, A. C. 1987. Type curve analysis: what HILL-VAN BEE~, F. 1991. Reservoir technology
it can and cannot do. Journal of Petroleum development in the next decade. In: Advances
Technology, January, 11-13. in Reservoir Technology: Characterisation,
Modelling and Management, Proceedings o f a LASSETER, T. J., WAGGONER, J. R. & LAKE, L. W.

conference held on 21/22 February, Edinburgh. 1986. Reservoir heterogeneities and their
Petroleum Science and Technology Institute. influence on ultimate recovery. In: LAKE, L. W.
HINDERAKER, L., BYGDEVOLL,J., BU, T., NYaRATEN, t~ CARROLL,H. B. Jr (eds) q.v., 545-559.
G. t~ KRAKSTAD, O. 1992. IOR resource LIBSON, T. E., VACCA,H. L. & MEEHAN,D. N. 1985.
potential of Norwegian North Sea sandstone Stratton Field, Texas Gulf Coast: A successful
reservoirs. Journal of Petroleum Science and cased hole re-evaluation of an old field to deter-
Engineering, 7, 3-14. mine remaining reserves and to increase pro-
HOLLANDER,N. B. 1987. Snorre. In: SPENCER,A. M. duction level. Journal of Petroleum Technology,
et al. (eds) Geology of the Norwegian Oil and Gas 37,105-124.
Fields. Graham and Trotman, 307-318. LORENZ, J. C., HEINZE,D. M., CLARK,J. A. & SEARLS,
HORNE, R. N. 1990. Modern Well Test Analysis: a C. A. 1985. Determination of meander-belt
Computer-aided Approach. Petroway Inc., Palo sandstone reservoirs from vertical downhole
Alto. data, Mesaverde Group, Piceance Creek Basin,
HUURDEMAN,A . J. M . , BREUNESE, J. N . , AL-ASBAHI, Colorado. American Association of Petroleum
A. M. S., LUTGERT,J. E. & FLORIS,F. J. T. 1991. Geologists Bulletin, 69,710-721.
Assessment of halite-cemented reservoir zones. MAHER, C. E. 1991. The future for expert systems in
Journal of Petroleum Technology, 43,518-523. applying reservoir technology. In: Advances
JOHNSON, A. • SCANLON,M. E. 1991. Pulsed-neutron in Reservoir Technology: Characterisation,
monitoring in the Bunter sands of the Esmond Modelling and Management, Proceedings of a
complex. Society of Petroleum Engineers conference held on 21/22 February, Edinburgh.
Formation Evaluation, 6, 327-333. Petroleum Science and Technology Institute.
JOHNSON, H. D. & KROL, D. E. 1984. Geological MALAVAZOS, M. & MCDONOUGH, R. C. M. 1991.
modeling of a heterogeneous sandstone reser- Pressure transient response in compart-
voir: Lower Jurassic Statfjord Formation, Brent mentalised gas reservoirs: a South Australian
Field. Society of Petroleum Engineers Paper field example. Society of Petroleum Engineers
13050 presented at Annual Fall Conference, Paper 23009 presented at Society of Petroleum
Houston. Engineers Asia-Pacific Conference, Perth.
--t~ STEWART, D. J. 1985. Role of clastic sedi- MARTIN, J. H. & COOPER, J. A. 1984. An integrated
mentology in the exploration and production of approach to the modelling of permeability
oil and gas in the North Sea. In: BRENCHLEY,P. J. barrier distribution in a sedimentologically
8s WILLIAMS, B. P. J. (eds) Sedimentology: complex reservoir. Society of Petroleum
Recent Developments and Applied Aspects. Engineers Paper 13051 presented at Annual Fall
Geological Society, London, Special Publi- Meeting, Houston.
cations, 18, 249-310. , EVANS, A. J. & RAPER, J. K. 1988. Reservoir
KEIJZER, J. H. & KORTEKAAS,R. F. M. 1990. Com- modelling of low sinuosity channel sands: a net-
parison of deterministic and probabilistic work approach. Society of Petroleum Engineers
simulation models of channel sands in the Paper 18364 presented at European Petroleum
Statfjord reservoir, Brent Field. Society of Conference, London.
Petroleum Engineers Paper 20947 presented at MASON, J. F. & DICKEY, P. A. (eds) 1989. Oil Field
EUROPEC 90, The Hague. Development Techniques: Proceedings of the
KETFER, F. J. 1991. The Esmond, Forbes and Gordon Daqing International Meeting, 1982. American
Fields, Blocks 43/8a, 43/13a, 43/15a, 43/20a, Association of Petroleum Geologists Studies in
UK North Sea. In: ABBOTrS, I. L. (ed.) q.v., Geology, 28.
425-432. MATTHEWS, C. S. 1986. Transient, semisteady state
KHANTANIAR, S. & PETERS, E. J. 1992. Reservoir and steady-state flow. Journal of Petroleum
heterogeneity and unstable displacements. Technology, 38, 385-387.
Journal of Petroleum Science and Engineering, MATHEWS, J. L., EMANUEL,A. S. & EDWARDS,K. A.
7,263-281. 1989. Fractal methods improve Mitsue miscible
KING, P. R. 1990. The connectivity and conductivity predictions. Journal of Petroleum Technology,
of overlapping sand bodies. In: BULLER, A. T. 41, 1136--1142.
et al. (eds) q.v., 353-362. MIALL, A. D. 1978. Lithofacies types and vertical
KLEPPE,J., BERG,E. W., BULLER,A. Z., HJELMELAND, profile models in braided fluvial deposits: a
O. 8Z TORS.~TER, O. (eds) 1987. North Sea Oil summary. In: MIALL, A. D. (ed.) Fluvial
and Gas Reservoirs. Norwegian Institute of Sedimentology. Canadian Society of Petroleum
Technology (Graham and Trotman). Geologists Memoirs, 5, 597-604.
KNUTSON, C. F. 1976. Modelling of non-continuous - - 1988. Reservoir heterogeneities in fluvial sand-
Fort Union and Mesaverde sandstone reservoirs, stones: lessons from outcrop studies. American
Piceance Basin, Northwestern Colorado. Society Association of Petroleum Geologists Bulletin, 72,
of Petroleum Engineers Journal, August, 682-697.
175-188. MILLER, D. D., MCPHERSON,J. G. & COVINGTON,T. E.
LAKE, L. W. t~ CARROLL, H. B. Jr (eds) 1986. 1990. Fluviodeltaic reservoir, South Belridge
Reservoir Characterisation. Academic Press, Field, San Joaquim Valley, California. In:
Orlando. BARWIS, J. H. et al. (eds) q.v., 109-130.
366 J . H . MARTIN
MURPHY, P. J. 1990. Performance of horizontal wells Giant Petroleum Fields. Memoir of the American
in the Helder Field. Journal of Petroleum Association of Petroleum Geologists, 14, 449-
Technology, 42,792-800. 476.
NASCIMENTO,O. S., BORNEMANN,E., JOBIM,L. D. C., SCHLUMBERrER 1981. RFT Essentials of Pressure Test
CARVALHO, M. D., PIMENTEL, A. M., BONET, Interpretation.
E. J., RODRIGUES, E. B., SANDOVAL,J. R. L., -- 1989. Cased Hole Log Interpretation Principles/
LASSANDRO, V., RODRIGUES, T. C. & HocoTr, Applications.
C. R. 1982. Aracas Field-reservoir hetero- Reservoir characterisation using expert
- - 1 9 9 2 a .
geneities and secondary recovery performance knowledge, data and statistics. Oilfield Review,
(abstract). American Association of Petroleum 4, 25-39.
Geologists Bulletin, 66,612. - -1992b. Trends in reservoir management. Oilfield
NYBRATEN, G., SKOLEM, E. & OSTaY, K. 1990. Review, 4, 8-24.
II Reservoir simulation of the Snorre Field. In: SHERRARD, D. W., BRICE, B. W. & MACDONALD,
BULLER, A. T. etal. (eds) q.v., 103-114. D. G. 1987. Application of horizontal wells at
O'BRIEN, D. G., BROWN, M. E. & LEDERER, M. C. Prudhoe Bay. Journal of Petroleum Technology,
1984. Use of reservoir performance analysis in 39, 1417-1426.
the simulation of Prudhoe Bay. Society of SIMLOTE, V. N., EBANKS,W. J. Jr & ESLINGER,E. V.
Petroleum Engineers Paper 13217 presented at 1985. Synergistic evaluation of a complex
Annual Fall Conference, Houston. conglomerate reservoir for EOR, Barrancas
PAYNE, D. V. et al. 1991. Examples of reservoir Formation, Argentina. Journal of Petroleum
simulation studies utilising geostatistical models Technology, 37,295-305.
of reservoir heterogeneity. In: LAKE, L. W., SIMPSON, R. E. t~ WEBER, A. G. 1986. Gasfield
CARROLL, H. B. Jr & WESSON, Z. C. (eds) development - reservoir and productions
Reservoir Characterisation H. Academic Press, operations planning. Journal of Petroleum
San Diego, California, 497-523. Technology, 38, 217-226.
POSTON, S. W. & GROSS, S. J. 1984. Numerical SLATI', R. M. & HOPKINS, X. 1990. Scaling geologic
simulation of reservoir sandstone models. reservoir description to engineering needs.
Society of Petroleum Engineers Paper 13135 Journal of Petroleum Technology, 42,202-212.
presented at Annual Fall Conference, Houston. STANLEY, K. O., JORDER, K. R/ESTAD, N. & STOCK-
RICHARDSON, J. G. et al. 1977. Synergy in reservoir BRIDGE, C. 1990. I. Stochastic modelling of
studies. Society of Petroleum Engineers Paper reservoir sand bodies for input to reservoir
6700 presented at the Annual Fall Technical simulations, Snorre Field, northern North Sea,
Conference, Denver. Norway. In: BULLER, A. Z. et al. (eds) q.v.,
~, HARRIS, D. G., ROSSEN, R. H. & VAN HEE, G. 91-102.
1978. The effect of small, discontinuous shales on STARLEY, G. P., MASINO, W. H., WEISS, J. L. &
oil recovery. Journal of Petroleum Technology, BOLLING, J. D. 1991. Full field simulation for
November, 1531-1537. development planning and reservoir manage-
~, SANGREE, J. B. & SNEIDER, R. M. 1987a). ment at Kuparuk River Field. Journal of
Coning. Journal of Petroleum Technology, 39, Petroleum Technology, 43,974-982.
883-885. STEWART, G. & AYESTARAN,L. 1982. The interpreta-
- -& -- 1987b. Braided stream reservoirs. tion of vertical pressure gradients measured at
Journal of Petroleum Technology, 39, 1499- observation wells in developed reservoirs.
1500. Society of Petroleum Engineers Paper 11132
RITCHIE, J. S. & PRATSlDES, P. 1993. The Caister presented at the Annual Fall Meeting, New
Fields, Block 44/23a UK North Sea. In: PARKER, Orleans.
J. R. (ed.) Petroleum Geology of NW Europe: STRUUK,A. P. & GREEN, R. T. 1991. The Brent Field,
Proceedings of the 4th Conference. Geological Block 211/29, UK North Sea. In: ABaoq-'rs, I. L.
Society, 759-770. (ed.) q.v., 63-72.
ROBERTS, J. D., MATHIESON,A. S. & HAMPSON,J. M. STUART, I. A. • COWAN, G. 1991. The South
1987. Statfjord. In: SPENCER, A. M. (ed.). Morecambe Field, Blocks 110/2a, 110/3a, 110/8a,
Geology of the Norwegian Oil and Gas Fields. UK East Irish Sea. In: ABBOa~rs, I. L. (ed.) q.v.,
Graham and Trotman, 307-318. 527-541.
RUDKIEWICZ, J. L., GUI~RILLOT, D., GALLI, A. & THAKUR, G. C. 1991. Waterflood surveillance
GROUP HERESI 1990. An integrated software for techniques-- a reservoir management approach.
stochastic modelling of reservoir lithology and Journal of Petroleum Technology, 43, 1180--
property with an example from the Yorkshire 1192.
Middle Jurassic. In: BULLER, A. T. et al. (eds) THOMAS, J. M. D. 1990. The movement of oil initially
q.v., 399-406. bypassed behind stochastic shale barriers. In:
RuPP, K. A. etal. 1984. Design and implementation of BULLER, A. T. et al. (eds) q.v., 437-444.
a miscible water alternating gas flood at Prudhoe TIRATSOO, E. N. 1990. World oilfields - past, present
Bay. Society of Petroleum Engineers Paper 13272 and future. In: ALA, M., HATAMIAN, H.,
presented at Annual Fall Meeting, Houston. HOBSON, G. D., KING, M. S. & WILLIAMSON,I.
SANFORD, R. M. 1970. Sarir Oil Field, Libya - desert (eds) Seventy-Five Years of Progess in Oil Field
suprise. In: HALBOUTY, M. T. (ed.) Geology of Science and Technology. Balkema, Rotterdam,
39-54.
TOLLAS, J. M. & MCKINNEY,X. 1991. Brent Field 3-D monitoring. Society of Petroleum Engineers
reservoir simulation. Journal of Petroleum Paper 7776 presented at Middle East Oil
Technology, 43, 589-596. Technical Conference, Manama.
VAN. B. T., PAJON, J.-L., JOSEPH, P. • CHAUTRA, WILKINSON,J. P. 1988. Horizontal drilling techniques
J.-M. 1991.3D reservoir visualisation. Journal of at Prudhoe Bay, Alaska. Journal of Petroleum
Petroleum Technology, 43, 1310-1314. Technology, 40, 1445-1451.
VAN DER GRAAF, W. J. E. & EALEY, P. J. 1989. WOOD, A. R. O. & YOUNG, M. S. 1991. The role of
Geological modelling for simulation studies. reservoir simulation in the development of some
American Association of Petroleum Geologists major North Sea Fields. Society of Petroleum
Bulletin, 73, 1436--1444. Engineers Paper 17613 presented at the Inter-
WADMAN,D. H., LAMPRECHT,D. E. & MROSOVSKY,L. national Meeting on Petroleum Engineering,
1978. Reservoir description through joint Tianjin, China, November.
geologic-engineering analysis. Society of WOODWARD, K. & CURalS, C. D. 1987. Predictive
Petroleum Engineers Paper 7531 presented at modelling of the distribution of production
Annual Fall Conference, Houston. constraining illites - Morecambe Gas Field, Irish
WADSLEY,A. W., ERLANDSON,S. • GOEMANS,H. W. Sea, Offshore UK. In: BROOKS, J. & GLENNIE,
1990. HEX - a tool for integrated fluvial archi- K. W. (eds) q.v., 205-215.
tecture modelling and numerical simulation of --, WANLI, Y. & ZHIWU, W. 1989. Exploitation of
recovery processes. In: BULLER,A. T. et al. (eds) multizones by water flooding in the Daqinq Oil
q. v., 387-398. Field. In: MASON, J. E. & DICKEY, P. A. (eds)
WEBER, K. J. & VAN GEUNS,L. C. 1990. Framework q.v., 63-87.
for constructing clastic reservoir simulation ZHIWU, W., QIMING,W., BOHU, L., CHENGJING,P. &
models. Journal of Petroleum Technology, 42, XIANGZHONG,L. 1989. Ways to improve develop-
1248-1253, 1296-1297. ment efficiency of Daqing Oil Field by flooding.
WESTAWAY,P., WITTMAN,M. & ROCHETrE, P. 1979. In: MASON, J. F. & DICKEY, P. A. (eds) q.v.,
Application of nuclear techniques to reservoir 49-62.

Engineer's Perspective A Review of Braided Fluvial Hydrocarbon Reservoirs: The Petroleum

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Engineer's Perspective A Review of Braided Fluvial Hydrocarbon Reservoirs: The Petroleum

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Engineer's Perspective A Review of Braided Fluvial Hydrocarbon Reservoirs: The Petroleum

Uploaded by

Copyright:

Available Formats

Geological Society, London, Special Publications

A review of braided fluvial hydrocarbon reservoirs: the petroleum

Geological Society, London, Special Publications 1993; v. 75; p. 333-367

Downloaded by Gary James Hampson on 6 December 2007

© 1993 Geological Society of

This p a p e r discusses some e n g i n e e r i n g aspects of m e t h o d s in d e v e l o p m e n t planning and

FromBest, J. L. & Bristow, C. S. (eds), 1993, BraidedRivers,Geological Society 333

Q 0il and i a a Norlh Rankin " ~'"# - ~

Reservoirs about 4% of world cumulative oil production.

Table 1. Selected braided fluvial reservoirs, Northwest Europe

Location Offshore UK Offshore UK Offshore UK Offshore UK Offshore UK Offshore

Description is specific to braided fluvial part of mixed reservoirs.

s a n d b o d i e s . T h e s e c o r r e s p o n d to t h e ' l a b y r i n t h ' using c o n v e n t i o n a l geological t e c h n i q u e s . T h e

Hewett Morecambe Murdoch Snorre Snorre Statfjord Wytch Farm

Offshore UK Offshore UK Offshore UK Offshore Offshore Offshore UK/ Onshore UK

Bunter Sst Sherwood Sst -- Lunde Statfjord Statfjord Sherwood Sst

-- -- -- 21-41% 32-42% 42% n/a

n/a 1220 280?

Water injection Water injection Miscible gas Water injection

250-1000 1-100 1-400 320-535 1300-2000 1500 5O

Alluvial plain Braided Braided Braided Braided/ Braided/ Braided

Table 2. Selected braided fluvial reservoirs, rest of world

Field Aracas Azal Bu Attifel Katnook Messla

Net/gross ratio 0.95 n/a n/a n/a n/a

Notes: see Table 1.

-- 10% 56% 1200 MMSTB 25% to date 38.5% to date

-- 4 1450 170 350

n/a n/a 46 500 -- m

7700 BSCF n/a 7000 BSCF --

14 (50 planned) 61 900 6100 300 n/a

Braided Braided/single Braided Braided Braided Alluvial fan

Belridge Field (Table 2) comprises s e p a r a t e p e r m e a b i l i t y distribution (e.g. J o h n s o n &

function of its production profile (Fig. 2) pre-

lation and continuity of sand and shale layers

a) GEOLOGICAL SCHEMATIC interpretation. However, single channel models

INITIAL PRESSURE - SIMULATION

Gas / Oil contact

""~'"F;'nai match R=250fl

Production logging demonstrates flow tester ( R F r ; Schlumberger 1981) and similar

monitored. This involves logging wells at

Application. The TDT log of Fig. 14 shows a

Fig. 15. Water encroachment at base of fining-upward channel sand.

strategy would be selected on the basis of

Application to braided fluvial reservoirs

(b) 100 12000

INPUT AVERAGING SCALE OUTPUT

Identify rock types

which have different I

Proportion of each KHE and KVE

Facies 10 ' Lower

Determine overall effective permeabllltles for

Discussion (4) Be aware of the limitations of subsurface

EHLIG-ECONOMIDES, C. A. 1987. Well testing in GUNDESO, R. & EGELAND, O. 1990. SESIMIRA

Modelling and Management, Proceedings o f a LASSETER, T. J., WAGGONER, J. R. & LAKE, L. W.

You might also like