Advances in Characterizing

Natural-Fracture Permeability From

Mud-Log Data
C.G. Dyke, SPE, BP Exploration (Alaska) Inc" Bailin Wu" and David Milton-Tayler"" BP Exploration

Summary More than half of these losses occurred in hydrocarbon reservoirs

Small drilling-fluid losses provide an accurate means of detecting or potential reservoir sections. Within the North Sea, conductive
conductive natural fractures and pinpointing potentially productive natural fractures were less important in causing major mud losses
zones. We outline improved interpretation techniques that allow but still accounted for 47% of all incidences between 1984 and
losses as small as 0,5 bbl to be used reliably as fracture permeability 1988. In total, 16% of all BP wells suffered major losses at depth as
indicators. We compare results from four case histories with con- a result of conductive natural fractures.
ventional fracture-detection methods. Because massive mud losses are an excellent indicator of natural-
fracture permeability, we assessed the potential to detect far smaller
losses indicative of in-situ permeability. This paper details the re-
Introduction
sulting technique for analyzing mud-logging records, together with
Natural fracture permeability can be both a help and a hindrance to several field examples of fracture-permeability chacterization.
reservoir performance. For instance, a conductive fracture system The idea of using mud losses to help detect natural-fracture
may provide the necessary permeability to drain a high-porosity but permeability is not new. Drummond,4 in discussing the fractured
low-permeability oil-saturated matrix, or it may lead to early break- carbonates of Iraq, reports that "losses led directly to the discovery
through of water and gas within production wells. Which scenario of the Ain Zalah field, where it was found that future well productiv-
applies to any individual reservoir depends on the characteristics of ity could be estimated roughly by the amount of mud lost at drilling
the fracture system and how that permeability is in communication in. High losses were indicative of high future production and no ap-
with the matrix, parent loss was an ominous sign." Advances within electrical log-
Natural fracture permeability is usually very variable spatially ging caused the technique to become less used, because it was per-
and often difficult to characterize. Of critical importance during res- ceived to be imprecise and unable to distinguish small losses into
ervoir appraisal is the ability to distinguish induced fractures from natural fractures from other causes of fluctuating mud volumes.
natural fractures. Even more important, however, is the ability to However, the use of mud losses to detect natural fracture permeabil-
distinguish conductive natural fractures from ones that do not con- ity has six important advantages over other fracture-detection tech-
tribute to rock mass permeability. However, most natural fracture niques.
detection techniques do not clearly differentiate between fractures I. Unlike log- or core-based methods, reservoir fluid-flow prop-
that allow fluid flow and those that do not because they do not mea- erties are measured directly by monitoring mud flow.
sure fluid-flow properties directly. Core description, for example, 2. Only conductive fractures are detected, and hence the problem
can characterize fracture porosity but is unable to distinguish rela- of differentiating unconnected fracture porosity from fracture
tively isolated porosity from porosity creating high permeabilities. permeability disappears.
Detection and characterization of fracture permeability is further 3. Fluid-flow properties are measured within a larger volume of
complicated by its susceptibility to damage from the invasion of drill- rock than is the case with core-based or borehole wall-logging tech-
ing-fluid solids. I During well testing, fracture permeability often re- niques. Results are therefore more representative of overall reser-
mains plugged by these solids, and thus it is often impossible to distin- voir properties because the scale effects of fracture inhomogeneities
guish damaged conductive fractures from impermeable ones. are reduced.
For these reasons, naturally fractured reservoirs are frequently re- 4. Damaged plugged-up conductive fractures can be distin-
garded as troublesome to appraise, and misconceptions abound guished from impermeable ones into which no mud is lost. This aids
about their character. One such view is that sandstones of moderate stimulation design,
matrix permeability (10 to 30 md) do not, at depth, contain signifi- 5. The technique is inexpensive and is suitable for attempting to
cant natural-fracture permeability, because it is thought that any detect fracture permeability within reservoirs where it is considered
fluid flow within a fracture system would be dwarfed by that unlikely to exist. Fractured reservoirs, therefore, do not remain un-
through the matrix, This means that certain naturally fractured pros- diagnosed.
pects may never be considered as such, and conductive fractures go 6. The technique is carried out at an early stage of reservoir ap-
unobserved because their detection is not requested. Nevertheless, praisal---during drilling in. This allows maximum flexibility for fur-
production levels of thousands of barrels a day can flow from very ther appraisal and the fine tuning of completion design to take ad-
short naturally fractured intervals. 2.3 vantage of any natural fracture permeability.

Downhole Mud Losses: An Indicator Characterizing Natural Fracture Permeability

of In-Situ Permeability From Mud-Loss Data
Massive losses of drilling mud flowing from the wellbore into the Drilling mud is a non-Newtonian shear-thinning fluid containing 10
surrounding formations are good indicators of the existence of natu- to 15 wt % of particles up to 40,um in diameter. Mud typically pos-
ral fracture permeability. In a review of mud losses occurring at sesses a Bingham plastic viscosity of 10 to 40 cp and a yield point
depth within BP wells drilled during 1988, conductive natural frac- of 10 to 30 Ibml100ft2 . Ifleft static for a few seconds, the mud devel-
tures caused 76% of losses large enough to cause drilling problems. ops a gel strength that must be broken before the fluid will resume
flowing. It is these nonidealarities in fluid behavior that are used to
'Now with CSIRO.
help detect and characterize fracture permeability, as outlined later.
"Now with FracTech Ltd. While making a new hole, mud is circulated from the mud tanks
Copyright 1995 Society of Petroleum Engineers
down the drillstring, through the bit, up the hole annulus, and back
into the tanks. The system is effectively closed, except for downhole
Original SPE manuscript received for review Jan. 1, 1993. Revised manuscript received May
5. 1994. Paper peer approved June 10, 1994. Paper (SPE 25022) first presented at the 1992
losses, retention of mud on drill cuttings, and a small amount of
SPE European Petroleum Conference held in Cannes, Nov. 16-18. evaporation. Mud levels within the tanks are continuously moni-

160 SPE Formation Evaluation, September 1995

 TABLE 1-CAUSES OF FLUCTUATING MUD-TANK LEVELS
Downhole losses through matrix permeability.
Downhole losses into induced fractures.
Downhoie losses into natural fractures.
Fracture Volume change owing to temperature and pressure effects.
Filling and draining of surface lines.
Addition of water and mud chemicals.
Change in circulating rate.
Porous Rock Change in bottom hole lithology.
Change in cuttings retention.
Well Bore Change in rate of spurt loss.
Change in rate of penetration.
A. Drilling fluid invasion of fracture. Hole collapse and enlargement.
Blinding of shale shakers.
Heave on a floating rig.
Surface mud losses.
Gradual decline in
loss rate as Even though losses are often dominated by a single highly perme-
TIME /" fracture plugs up
¥ with mud solids able fractured zone, additional losses at greater depths can usually

t ~apid initial loss rate

be distinguished reliably from further losses into the same overlying
zone. This judgement is based on a number of factors, such as past
behavior of the shallower loss zone, how quickly the additional loss
initiates, and the response of the well during tripping drill pipe in and
out of the hole.
Pit Level_ Besides the detection of natural fracture permeability, other frac-
ture characteristics can also be deduced from the way a loss prog-
B. Characteristic shape of mud-tank level response. resses with time. For instance, the ability of a damaged fracture to
backflow on drawdown can be determined. During tripping of drill-
Fig. 1-Mud losses into natural fractures. pipe in and out of the hole, pressure swabs and surges are established
around the drill bit. These are caused by the drill bit acting as a piston
tored by fluid-level sensors; ultrasonic reflectance or float devices within the well bore and frequently exceed 800 psi in magnitude
are the most common types. Alternatively, high-resolution flowme- within 8.S-in. diameter hole sections. This pressure transient can
ters can measure the rate of fluid flow into and out of the well bore. momentarily loosen a lightly compacted blockage within a fracture,
Besides fluid levels, several other parameters are monitored contin- allowing a further small loss to occur within that zone. This response
uously; this allows specific drilling operations that might affect flow can signify that the damaged natural fracture will backflow and pro-
rates or fluid levels to be determined at any point of time from data duce on drawdown without stimulation if no further damage occurs
records. These engineering parameters include mud weight, mud during well completion. In-situ fracture apertures can also be esti-
temperature, pump strokes, standpipe pressure, rate of penetration. mated as outlined later.
depth of bit, weight of bit, hook load, hook height, and torque.
Fig. 1 shows the typical response on drilling through a permeable Distinguishing Losses Into Natural Fractures From Other
natural fracture. Initially, mud flows into the undamaged fracture at Causes of Fluctuating Mud-Tank Levels. Apart from downhole
a high rate. Leakoff of mud filtrate through the walls of the fractures mud losses into natural fractures, a number of other factors influ-
tends to concentrate mud solids within the fracture, thereby locally ence mud levels and flow rates during drilling. These include addi-
increasing the mud viscosity. As the fracture becomes plugged with tional downhole losses through matrix permeability and induced
mud solids, its permeability becomes damaged and the rate of mud fractures, as well as a number of operational and geological factors
loss slows down. This gives rise to the typical response in mud-tank outlined in Table 1.
level, as Fig. IB shows. The characteristics are a rapid initial loss Losses through matrix permeability or into induced hydraulic
rate and a gradual decline as the fracture plugs up. For a narrow frac- fractures can be distinguished from those into natural fractures by
ture open by approximately 500 .um, the initial loss rate may be up the characteristics of the loss (Fig. 2). While making new hole,
to 60 bblJhr, with a total loss volume of 2 to 3 bbl occurring over no losses through pores initiate slowly and gradually build up in rate as
more than a couple of minutes. A large part of the analysis involves additional penetration into the highly permeable stratum is achieved
searching for suitably shaped changes in the mud-tank levels while (Fig. 2a). Because the width of an induced hydraulic fracture is ex-
new hole is being made. This method allows the detection of down- tremely sensitive to small changes in fluid pressure, loss rate is simi-
hole losses far smaller in volume than previously thought possible larly sensitive to such a change. As downhole pressures are fre-
to observe and for these losses to be confirmed as flowing into natu- quently changed during drilling operations, frequent checks can be
ral fracture permeability. The technique can be regarded as continu- made as to the nature of a loss. For instance, if mud-circulating rate
ously carrying out a rather rough-and-ready permeability test at the is increased, downhole pressure also increases because of increased
drill bit as new hole is made. friction losses within the annulus. Likewise, if the mud is allowed

(a) PORES (b) NATURAL FRACTURES (c) INDUCED FRACTURES

TIME

Loss can occur

Sudden start,
on increase in
exponential
decline
~~~~~~~ps
Gradual build up
in loss rate with
penetration

PIT LEVEL
.. PIT LEVEL
~
PIT LEVEL
~

Fig. 2-Determining the type of loss zone from the shape of the mud-loggers' pit-level trace.

SPE Formation Evaluation. September 1995 161

 1000 •• • to be pinpointed and characterized. This subset of permeable frac-
Cause of Loss
~
..
a: 800
• • Induced Fractures
• Natural Fractures At Uft-off
tures can then be further characterized from core and logs. Addition-
ally, the technique allows an estimation of fracture apertures under
co
co ... Natural Fractures
.3 600 in-situ conditions. This is estimated from analysis of how a fracture
.
.5 • becomes blocked with mud solids and from how a loss progresses
".
en 400
with time.

.......
.s: Back analysis of production, log, and core data from naturally
(,) 200
... fractured reservoirs, together with laboratory investigations into the
a blockage of fractures by mud solids, has allowed estimates of in-situ
a 10 20 30 40
fracture apertures to be gained from the interpretation of mud losses
% Change in Overbalance Pressure
(Table 2). Fig. 4 summarizes results offracture blockage by a water-
Fig. 3-Changes in downhole mud-loss rate with change in over- based mud for various fracture apertures with and without filtrate
balance pressure. leaking into a permeable matrix.
Similar laboratory studies reported in the literature confirm these
to remain static for several seconds, a pressure surge is created on
findings. Kasperskii and Panov 8 showed that fracture apertures of
regaining circulation as the gel strength of the fluid is broken down.
up to 150 to 300 ,urn, depending on mud type, could be blocked by
This happens when mud pumps are turned off and on, and if a loss
muds with little or no penetration of the fluid along the fracture. In
into an induced fracture is occurring at this time, its rate will be ac-
the case of larger apertures, mud could flow freely into the fracture,
celerated because of the pressure (Fig. 2c).
with the depth of penetration depending on the overbalance pres-
Fig. 3 is a compilation of variations in loss rate with variation in
sure. Experiments conducted by Morita et at. 9 revealed that fracture
overbalance pressure gathered from BP wells. Losses into hydraulic
widths of between 250 to 380,um could be sealed by mud solids
fractures are highly sensitive to small changes in pressure, whereas
bridging across the fracture. With the aid of mud filtrate leaking into
losses into natural fractures are much less so. For natural fractures
a permeable matrix, fracture widths between 250 to 630,um could
!t might be expected that the ratio of change in loss rate to change
be sealed by dehydrated plugs of mud.
m overbalance should be approximately unity. However, this ratio
Massive losses of mud, at rates of up to several hundred barrels
is greater because (I) the fluid is non-Newtonian and particulate in
an hour, are caused by fracture apertures wider than 500,um. These
nature, (2) temporary fracture blockages are broken down by the
higher fluid pressure, and (3) new areas of the fracture walls are ex- fractures generally require blocking by lost circulation materials
posed to leakoff, causing changes in rate of filtrate loss and mud vis- (LCM). By knowing the bridging behavior of a particular LCM,
cosity. These factors all work in unison to cause any changes in loss how quickly a particular loss was cured and the subsequent behavior
rate to be greater than the change in overbalance pressure, especially of these large fracture apertures can be determined. One commonly
for low-rate losses occurring at low overbalances. used product, coarse cellulose flakes of up to 1,300-,um diameter,
Of the operational factors influencing mud-loss detection, heave can seal fractures open by up to 3,000,um (Fig. 5).
of semi submersible rigs caused by wave action is the most trouble- In comparing mud-loss derived apertures with those obtained
some. In rough seas, mud levels often fluctuate by 3 bbl. With care- from other sources, one must remember that fracture apertures de-
ful analysis though, losses as small as 1 to 2 bbl can still be observed termined in this way are an estimate of the fracture opening at the
under these conditions because the heave period is far shorter than narrowest constriction along the flow path. This is the most likely
the duration of the mud loss. location for any blockage.
Losses into natural fractures are not obscured by other factors, These apertures are generally larger than those commonly per-
such as thermal expansion of the mud. For example, a temperature ceived to exist, at depth, under high in-situ stress conditions. Howev-
increase averaging 20°F throughout an active mud volume of 800 er, such values are substantiated by results from recent advances in
bbl would cause an expansion of 3.5 bbl. This is a change of similar downhole logging, such as the Formation Micro-Scanner (FMS) and
magnitude to a mud loss but it would occur over several tens of min- Stone ley wave tool. lO.11 Mechanisms allowing natural fractures to
utes, a different time scale from a similarly sized loss into natural possess these large apertures at great depth and under high-effective
fractures. stresses include partial cementation and dissolution. 12- 14 These allow
fractures to remain "locked open" by partial cement in-fill.
Detection of Losses. With standard mud-logging systems on land
?gs, whole mud losses as small is 0.5 bbl can be detected flowing Examples
mto natural fracture permeability. Offshore, on semisubmersible The following examples, in which the mud-loss technique was run
rigs, the detection limit is 2 bbl because of rig heave and wave ac- in conjunction with core and fracture-detection logs, show the value
tion, unless modifications are made to the mud-circulating system. of detecting minor mud losses for the detection and characterization
By use of high resolution flow-in and flow-out meters such as a "}" of natural fracture permeability. In particular, the losses highlight
meterS or an electromagnetic flowmeter,6,7 detection limits can be which zones to complete and stimulate and distinguish the conduc-
further reduced. On a recent slimhole well with electromagnetic tive natural fractures from the nonconductive ones.
flowmeters, losses as small as 0.1 bbl were detected flowing into
permeable natural fractures. Alternatively, a similar resolution can Well A: Fractured Carbonate. To test the reliability of using mud
be obtained by modifying the volume and geometry of the mud losses for detecting natural fracture permeability, we performed an
tanks. initial trial on a fractured carbonate well. Natural fractures had been
characterized by core, FMS, and Stoneley wave logs, leading to the
In-Situ Natural Fracture Apertures completion of eight intervals. The mud-loss analysis was carried out
The technique of detecting open fractures from small losses of drill- independently of log data and immediately after well testing before
ing fluid allows the individual fractures responsible for fluid flow any results were released. The well had been drilled by a semi sub-

TABLE 2-DETERMINATION OF FRACTURE APERTURES FROM MUD LOSSES

Fracture Aperture
(um) Blockage Mechanism
oto 150-250 Mud will block fracture near well bore without any detectable mud loss.
150-250 to 500-750 Mud will block fracture at a distance from wellbore with detectable loss.
>500-750 Mud alone will not seal fracture, LCM required to stop losses.

162 SPE Formation Evaluation, September 1995

 No Detectable Detectable

350
Mud Loss Mud Loss
.. LCM Required To Cure Loss

WBM&LCM,No
20

WBM,No ~ Laakoff • ,
.2 ,, ,,
Leakoff ,

~I
".
300
250 .
, ,
,
, :a
15

,, ,,
Q.

~S :is.
200 , o· 10
':; 8, 150 ,, oJ!:!
0

S~
0

sm
" 0
100
"L!':k~it ,/ WBL~at~~;/.o 5
oJ 50
"." a,Jl " o_-----a'
0~0--~~~20~0~~----40~0--~L--6~0-0----~-8~0-0------~1000 0
0 0.04 0.08 0.12 0.16 0.20
Fracture Aperature(u m) Maximum Bridgeable Fracture Aperture (in.)

Fig. 5-Blockage of natural fractures by LCM. (Tests were con-

Fig. 4-Blockage of natural fractures by water-based mud. ducted in water-based mud under dynamic flow conditions at
(Tests were carried out in an artificial fracture 1.5 in. wide by 9.6 overbalances of up to 1,000 psi without leakoff through the frac-
in. long within 50-md sandstone. The mass of mud entering the ture faces).
fracture before blockage at 200-psi overbalance was measured
with and without leakoff. LCM was a fine-grained cellulose fiber
which appear to link with highly conductive fractures away from the
at 15 ppb).
wellbore, as interpreted in Fig. 7.
mersible rig in fairly rough seas which caused a background fluctua-
tion in mud levels of 2 to 4 bbl. Within a 656-ft section of chalk, five Well B: Fractured Carbonate. Within a partially cored 115-ft-
losses ranging in volume from 2 to 5 bbl were detected (Fig. 6). thick interval of chalk, four separate downhole mud losses were re-
Before acid fracturing, the majority of production flowed from the corded (Fig. 8). After a small acid stimulation, a downhole spinner
upper two completed intervals. Narrow conductive natural fractures, survey revealed high productivity from the two uppermost mud-loss
into which little or no drilling mud flowed, appear to be responsible zones with no production emanating from the lower two zones. One
for this permeability. On performing the acid stimulation, a lower of these nonproductive loss zones correlated with open fractures in
zone opened and dominated production, which increased from < 400 core, whereas the other was just below the cored interval. Because
to more than 6,000 BID. Two small mud losses, each of 2 bbl, were the acid stimulation was carried out with no diverting agents, all the
detected within this 13-ft long interval. These delineate two minor acid flowed into the top two zones, leaving the lower zones unpro-
conductive natural fractures, plugged with mud solids during drilling, ductive and blocked with mud solids. This demonstrates an impor-
tant aspect of natural fracture permeability; drilling mud is highly
damaging to fracture permeability and frequently causes it to remain
,,'" ~o~ ~o~
;§ §-'b' blocked with mud solids during production. Consequently, natural
1} 91'" §-'b'
~ ~'6 r.'d ~~,i. fracture permeability can exist in-situ but remain hidden without
~
~o<i)
0 ~eRe~
,,0
f.J~0~ :-,:t:fJ«,0
<i~' .;f' (.0,0 ~~~ q<'''' qO'" contributing to fluid flow because of drilling damage .

7.546 Well C: Fractured Sandstone. A 2,OOO-ft oil column in sandstone

0 55 % 0 3% was drilled. Because parts of the sequence came in tighter than an-
ticipated, interest was expressed in the possibility of natural fracture
-4bbls permeability. Mud-loss analysis was the only available fracture
detection technique and was carried out. More than 40 losses into
0 25 % 0 4% natural fractures were detected, ranging in volume from 0.5 to more
Pre-stimulation: no flow
Post-stimulation: >6000 bbls/day High skin
00% 01%

00% 00%

7.874 0 16%

[=:JO% [=:J 0%
c:=J 0%
-5bbls

-2bbls
-2bbls 0 0% 0 92 %
Characteristics: Small mud loss volumes (4bbls)
Total Flow Total Flow Plugging of near wellbere fractures only
<400 bbls/day <6000 bbls/day High positive skins, even when not damaged by mud
High far field permeability
Very high potential procuction

_5bbls Stimulation requirements: Significant widening of near wellbore fluid pathways to

reduce skins

ft Fig. 7-Fracture characteristics of highly productive interval,

Fig. 6-Mud losses and production, Well A. WeIlA.

SPE Formation Evaluation, September 1995 163

 Depth Natural Fractures Fracture Opening Mud Losses Comments
(It) (no. per foot) ( in. per fool) (bbls)
POST ACID 02468
10,000 +--'--'---'---J
° 0.008 0.Q16 0.024

PRODUCTION LOG
WHOLE
MUD
LOSSES
:t= :5
t= ; Oil Stained Fracture
Vugular Fracture
10,050

+- 075 Open & Vugu[ar Fracture

10,100

_1

10,150

:...---1----+ ....... 10bbl 10,200

+;3bbl
-,
15 bbl

~+

I/tl----+----+-----l- ...., 6 bbl

Fig. 9-Correlation of mud losses with open fractures observed
.......
I ..
in core, Well C.

......, 6bbl
tervals produced, into each of which drilling fluid was lost (Fig. 10).
~ .. Out of the 15 mud losses recorded within completed intervals, 13
indicated productive zones. Whether the other two losses delineated
plugged fracture permeability or were incorrectly diagnosed is un-
clear.
The response of the Stoneley wave and FMI logs (Fig. 11), show-
7,218
ing indications of conductive fractures over a considerable propor-
tion of the interval, is typical of many fractured wells. This demon-
strates that an integrated approach to characterization that uses data
DEPTH from several fracture indicators is necessary to eliminate spurious
(ft) Perfs miliiiiiimJ
log responses and successfully complete naturally fractured wells.
With regard to fracture density, the majority of production surpris-
Fig. 8-Detection of natural fracture permeability, Well B. ingly originates from intervals with fairly low fracture densities.
This implies that a subset of wider fractures is more important in
controlling fluid flow than the remainder of the fractures.
than 250 bbl. A high fracture gradient and relative insensitivity of
loss rate to change in overbalance pressure confirmed that the major
losses flowed into natural fractures. From the behavior of lost cir- Discussion and Conclusions
culation material in curing the losses, fracture apertures of up to 1. Downhole mud losses as small as 0.5 bbl can be reliably de-
2,000 f.1m were estimated. tected with conventional mud-tank monitoring systems. Changing
Substantial parts of this hole had enlarged from 8.5 in. diameter the geometry of the mud tanks or using high-resolution flowmeters
to more than 20 in. Downhole losses may have been confused with allows even smaller downhole losses to be detected. By applying the
cavings coming over the shakers. To confirm that this was not the outlined methodology, losses into natural fracture permeability can
case and that minor losses did indeed flow into natural fracture be distinguished from losses into other types of in-situ permeability,
permeability, data from the short cored interval were compared with as well as from other "apparent losses" caused by changing drilling
the mud-loss data (Fig. 9). Eight of the 10 losses in this interval cor- operations. Analysis of small changes in downhole pressure, frac-
related with open fractures within the core. The two smallest losses ture gradients, and production data confirms that these losses are oc-
do not correlate with any observed open fractures and may have curring at pressures far below that required to overcome the mini-
been misidentified. However, core recovery in this interval was less mum in-situ stress and induce fractures.
than 100%. 2. Both minor losses as well as massive losses into natural frac-
tures provide information crucial for optimum completion design.
Well D: Fractured Limestone. A full suite of natural fracture Minor losses indicate good zones for completion, whereas massive
detection logs, including Formaton Micro Imager (FMI) and Stone- losses delineate zones that may either produce substantial oil or may
ley wave, were run within a 1,000-ft fractured limestone interval. quickly turn to water or gas. Which scenario applies to any particu-
Matrix permeability was less than 1 md. A small acid stimulation lar situation depends on the standoff from the fluid contacts, matrix!
was carried out to bypass near-wellbore formation damage. Four in- fracture recharge, and the continuity of the fracture system.

164 SPE Formation Evaluation, September 1995

 Fracture density Hydraulic
Production Instantaneous Depth
(F radure! ft) aperture
Completed Stoneley FMI <in.>
Depth Completed log mud loss ftBRT intervals wave
ftBRT intervals (%) (bbls)
o 100 0 4

8,530 - r.::7t
~

r:?1
9,416
I
8,694 - U

9,481

8,858 - I':":":"'l
~

~
~

m -
w --
9,022 - 9,547

~
~

m
w
9,186 - Fig. 11-Natural fracture characterization from mud losses and
logs, Well D.

the fractures, In this way, it becomes possible to gather the infonna-

tion necessary to predict the occurrence and distribution of those frac-
n
w tures dominating fluid flow, rather than that of all fractures grouped
together,
9,350 -
n
w Acknowledgments
We thank the British Petroleum Co, for permission and encourage-
ment to publish this paper, Special thanks to all those with whom we
have worked on naturally fractured reservoirs,
IL
9,514 - ~ls'n5mins References
~ I, Crawford, D,L., Coulter, A,W., and Osborn, F.E.: "Removal of Well bore
Damage From Highly Permeable Formations and Naturally Fractured
Reservoirs," paper SPE 8796 presented at the 1980 SPE Formation
Fig. 1G-Mud losses and production, Well D.
Damage Control Symposium, Bakersfield, Jan, 28-29.
2. Daniel, EJ.: "Fractured Reservoirs of Middle East," AAPG Bull. (I 954)
3. Because offonnation damage to natural fracture penneability, 38,774,
stimulation is often required to bypass fractures blocked with mud 3. Missman, R.A. and Jameson, J.: "An Evolving Description of a Frac-
solids. Losses as small as 2 bbl can obscure fracture permeability tured Carbonate Reservoir: The Lisburne Field, Prudhoe Bay, Alaska,"
representing significant production capability. If 6,000 BID of pro- paper SPE 22161 presented at the 1991 SPE IntI. Arctic Technology
duction is considered sufficient to delineate an economic resource, Conference, Anchorage, May 29-3\.
Well A reveals that a 4 bbl mud loss can "hide" a prospective reser- 4. Drummond, J.M.: "An Appraisal of Fracture Porosity," Bull. Cdn. Pet.
Geology (1964) 12,226.
voir. Generally, only a short induced fracture is required to connect
5. Johnsen, H.K. et al,: "Development and Field Testing of a High Accura-
the wellbore hydrologically with natural fracture penneability be-
cy Full-Bore Return Flowmeter," paper SPE 17228 presented at the
yond the zone of damage. 1988 SPElIADC Drilling Conference, Dallas, Feb. 28-March \.
4. Comparison of results from the detection of mud losses with 6. Maus, L.D., Peters. B.A., and Meador, DJ.: "Sensitive Delta Flow
those from more traditional fracture characterization techniques re- Method Detects Kicks or Lost Returns," Oil & Gas J. (1979) 77, 125.
veals that only a small proportion of open fractures are hydrologi- 7. Schaffer, D.M. et al.: "An Evaluation of F10wmeters for the Detection
cally active. It is these occasional irregularly distributed fractures of Kicks and Lost Circulation During Drilling," paper SPE 23935 pres-
that dominate rock mass permeability and control fluid flow. This ented at the 1992 SPElIADC Drilling Conference. New Orleans, Feb.
suggests that the total number of conductive fractures within a rock 26-27.
mass is not the major control on fracture permeability and that the 8. Kasperskii, B.Y. and Panov. B.D.: "Investigations on Plugging By
Weighted Drilling Muds on Slotted Models," Bllrenie (1971) 28.
characteristics of individual fractures are more important. This is
9. Morita, N .• Black, A.D., and Fuh, O.F.: "Theory of Lost Circulation,"
clearly demonstrated by the fourth example well. paper SPE 20409 presented at the 1990 SPE Annual Technical Confer-
5. By combining the mud-loss fracture detection technique with ence and Exhibition, New Orleans, Sept. 23-26.
core and log techniques, the characteristics and orientation of the frac- 10. Cheung, P.S.Y. and Heliot, D.: "Work-Station Based Fracture Evalua-
tures dominating fluid flow can be separated from the remainder of tion Using Borehole Images and Wireline Logs," paper SPE 20573 pres-

SPE Formation Evaluation. September 1995 165

 ented at the 1990 SPE Annual Technical Conference and Exhibition, Chris Dyke is a petroleum engineer with BP Exploration (Alaska)
New Orleans, Sept. 23-26. Inc. in Anchorage responsible for a variety of production
II. Hornby, B.E. and Luthi, S.M.: "An Integrated Interpretation of Fracture technology projects. Previously, he worked at BP Research on
Apertures Computed From Electrical Borehole Scans and Reflected rock mechanics and well-productivity issues. He holds a PhD de-
Stoneley Waves," Ceol. Appl. of Wire line Logs IJ, Geol. Soc. of London gree in rock mechanics from Imperial C., London. Bailin Wu is a
Spec. Pub I. (1992) No. 65. research scientist with the CSIRO Div. of Pertroleum Resources,
12. Laubach, S.E.: "Subsurface Fractures and Their Relationship to Stress in Australia. Previously, he worked for BP Research, primarily on
History in East Texas Basin Sandstone," Tectonophysics (1988) 156,37. rock mechanics, related drilling, and completion R&D. He holds
13. Nelson R.A.: "Geologic Analysis of Naturally Fractured Reservoirs," a PhD degree in rock mechanics from Imperial C. Dave Milton-
Gulf Publishing Co., Houston (1985). Tayler is the U.K. technology manager for Frac Tech Ltd. Pre-
14. Dyke, e.G.: "How Sensitive Is Natural Fracture Permeability at Depth viously, he worked on a variety of fracturing and rock mechan-
To Variation in Effective Stress?" Proc., IntI. ISRM Symposium on ics issues in the Well Stimulation Group of BP. He holds a PhD
Fractured and Jointed Rock Masses, CA (1992). degree in petroleum engineering from Imperial C.

51 Metric Conversion Factors

bbl x 1.589 873 E-Ol =m 3
cp x 1.0* E - 03 = Pa ' s
'F ('F-32)/1.8 = 'c
ft x 3.048* E-Ol =m
ft3 x2.831 685 E-02 =m 3
in. x2.54* E+OO =cm Dyke Wu Milton·Tayler
Ibf x 4.448 222 E + 00 =N
psi x 6.894 757 E + 00 = kPa
'Conversion factor is exact. SPEFE

166 SPE Formation Evaluation, September 1995