SPE 25344

Society of Petroleum Engineers

Gas-Lift Optimisation on Barrow Island, Western Australia

J.e. Mantecon, West Australian Petroleum Pty. Ltd.
SPE Member

Copyright 1993, Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the SPE Asia Pacific Oil & Gas Conference & Exhibition held in Singapore, 8-10 February 1993.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are sUbject to correction by the author(s). The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledg-
ment of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 163245 SPEUT.

ABSTRACT much closer well surveillance, such as faster

Improving the design and operation of gas lift
systems to obtain optimum production (within the
limits of existing facilities) is very attractive with
low investment and a corresponding
acceleration/increase in reserve recovery to
maximise profit.
This paper discusses the studies conducted and
methods used to increase oil production and
optimise the gas injectedlfluid produced ratio in
the Barrow Island Field by utilising the existing
compression capacity and field staff.
detection and diagnosis of problems, and quicker
response to failures.
The results demonstrate the potential that exists in
producing areas similar to BWI for maximising
economic return through the application of sound
engineering and operating principles. A united
effort from Operating and Engineering personnel
that is strongly supported by Management, along
with an effective communication and training
program can improve gas lift system efficiency
quicker and more consistently to minimise costly
errors and/or delays.

This multi-stage optimisation program used on

Barrow Island (BWI) can be readily adapted to INTRODUCTION
other producing areas, even those with limited
remaining field life. During the first stage, Barrow Island is located about 1,300 km north of
emphasis was placed on: (1) individual well Perth, 56 km off the Western Australia coast
optimisation; (2) improving gas lift design between Port Hedland and Onslow (Fig 1). The
techniques; (3) converting to intermittent gas lift nearest port, Onslow, is about 88 km south.
(lGL), gas assisted plunger lift (PL) or sucker BWI is the second biggest island off the Western
rod pumping when continuous gas lift (CGL) was Australia coast. Its elongated oval shape covers
no longer efficient; (4) improving system approximately 233 sq km with the highest point
diagnostics; (5) field personnel training, and; (6) rising to about 65 m above sea level.
enhancing Operations / Engineering
communication. After the first stage is Prior to oil exploratory and producing activities,
completely implemented and results evaluated, the island was uninhabited by man. BWI was
fieldwide optimisation is planned and will be declared an "A" class reserve (highest
actively pursued. classification of protection) by the W.A.
government in 1910. Wildlife has remained
Performance results indicate a very attractive undisturbed by man and alien influences which
return from a low risk investment. Significant have affected the mainland, and it. is
improvements have been obtained in: (1) environmentally unique. There is a strict ban on
individual well profitability; (2) optimising firearms, exotic plants and animals being brought
injection gas usage; (3) well monitoring, and; (4) in and all contact with vessels is carefully
the amount and quality of collected well data. controlled. Destroying or molesting wildlife, and
unnecessary damage to plants is specifically
Additional intangible benefits have come from the prohibited. Areas previously used and now idle

237
2 GAS LIFf OPTIM:ISATION ON BARROW ISLAND, WESTERN AUSTRALIA
are subject to restoration.
The first well on BWI, drilled in 1964,
discovered a small oil accumulation in the Upper
Jurassic sandstone at .±2,050 m. The fourth well
was completed as a shallower (.±650 m) oil
discovery in the Lower Cretaceous Windalia
Sandstone, which is the major hydrocarbon
accumulation discovered on BWI.
The Windalia reservoir is a silty, very fine
grained sandstone with high porosity (.±25 %),
low permeability (.±2.5 md) and high water
saturation (.±55%). Development of the
Windalia reservoir began in 1966. Development
wells were drilled on an 80 acre spacing.
Waterflooding was initiated during 1968 with a
40 acre five spot pattern at the field's
southwestern end. The flood was expanded
throughout the field by 1970, essentially to 40
acre inverted nine spot patterns. As a result of
the preferential injection water movement in an
east-west direction, a conversion of high water
cut wells to injectors was carried out in 1972.
The resulting pattern is basically line drive. In
1978, 20 acre (diagonally oriented inverted nine
spot pattern) infill drilling commenced.
Currently there are 422 producers and 194
injectors. Producing wells are hydraulically
fracture stimulated. The measured depth ranges
from 600 to 735 m and normally requires
artificial lifting equipment.
Compressors for gas lift were installed in 1970.
Currently, low pressure gas from the field
separator stations is gathered (.±12.5 MMscfpd)
and compressed at a central point to .±600 psig.
From the compressor station, gas passes through
a low temperature separation (LTS) unit where
heavy fractions are removed. The remaining
lighter fractions are then distributed through the
field for gas lift operations. Average distance
from LTS unit to well is 2 km and line sizes are
6"(trunk line), 3" and 2".
Production peaked at over 50,000 bopd early in
the field life and has since declined to .±14,100
bopd. The artificial lift optimisation program
began in October 1990. Currently, 91 (21.6%)
of the 422 producing oil wells are on gas lift
producing 4,700 bopd (33 % of total daily oil
production). Three wells are on IGL, 2 on PL
and 86 on CGL (including 6 dual completions).
238
SPE 25344
GAS LIFT OPTIMISAnON PROGRAM
The gas lift optimisation program was initiated by
collecting, collating and reviewing basic well data
including well schematics, gas lift designs, well
tests and pressure surveys (when available), as
well as system data including both gas lift system
and flowline distribution drawings.
All current GL wells were initially screened by
comparing individual well test data. Data used
that reflects the efficiency of a GL well includes:
Gas injection rate
Formation GLR
Total production GLR
Water cut
Tubing pressure
Casing pressure
Any abnormal value of the above is an indication
of producing problems that need attention.
A cost effective Gas Lift System is obtained
through the following range of activities:
Develop a plan of action.
Collect and analyse relevant information.
Identify obvious misuses of lift gas.
Improve current GL design technique
(optimise gas injection rate and depth).
Select optimum GL system (CGL, IGL
or PL).
Select wells and prioritise on profit
potential.
Modify surface piping system to reduce
excessive pressure drops.
Develop guidelines for operation,
maintenance and troubleshooting of GL
installations.
Train field personnel in gas lift
optimisation.
Monitor effectiveness of the system.
Evaluate lift system later in the well life
and switch to a more effective system as
well behaviour and conditions change
(lower reservoir pressure, increasing
water cut, etc.).
Figure 2 shows the analysis path required to
optimise the field.
Individual Well/Gas Lift Design Optimisation
The original CGL design used on BWl was an
open installation (no packer - Fig 3A) with
SPE 25344 JUAN C. MANTECON
injection pressure operated valves for unloading
and a 1/4" holed tubing collar as the operating
injection point. General guidelines for setting gas
lift valves (GLV's) were being used.
Improvements introduced to optimise the CGL
design are:
Packer installation
Reduce GLV port size (1/4" to 3/16")
Deepen gas injection point (±50 m)
Optimise tubing size (2-3/8" for rates
< 200 bfpd)
Reduce number of GLV's
Replace holed tubing collar with GLV
Remove flowline chokes
Utilise gas lift design software
Some problems associated with open installations
(Figure 3A) are:
Gas may be blown around the bottom of
the tubing, and the injection / casing
pressure may place additional back
pressure on the sand face.
Wells must be unloaded and restabilised
every time they are shut-in. With no
packer, fluid will rise during shut-in and
must be displaced from the annulus
again.
Frequent shut-ins or fluctuating injection
pressure allows the annular fluid level to
rise and fall, thereby exposing the lower
GLV's to fluid erosion.
A semi-closed installation (Figure 3B) offers
certain advantages over an open installation,
making the system more cost effective:
Suitable for both continuous and
intermittent gas lift.
Once the well has been unloaded, the
produced fluid would be unable to re-
enter the annulus.
Reduces the possibility of heading
conditions. Heading within a flowing
well tubing string usually occurs in wells
with no packer or a long packer tailpipe.
Better control of injection gas.
Makes troubleshooting of GL
239
3
installations simpler. Possible causes of
malfunctions can be detected quicker so
wells are returned to production faster.
Continuous gas lift is designed so only the lower
valve will inject gas into the tubing, and will be
as deep as the available gas pressure will permit.
By using a packer the injection gas pressure was
increased, allowing both deeper GLV setting
(optimum GLR) and a fewer number of GLV's.
The number of GLV's and mandrels per well was
reduced by 2 to 3 (> 50% in most cases), which
saves ±$A 2,400 to ,±$A 3,6oo/well and is
another step toward increasing profits by
improving system design. Deeper injection
point/optimum GLR increases the reservoir
drawdown, thereby maximising the well's
producing rate.
In summary, a semi-closed installation results in
an increase of production and more efficient gas
lift usage.
A closed installation (Figure 3C) is used in PL
and being tested in IGL wells.
Wellhead. Flowline and Compressors
Prior to GL optimisation, flowline chokes existed
in 15 of the 86 CGL wells at BWI. CGL wells
with heading problems were historically stabilised
by choking the flowline.
Also, some wellheads had an excessive amount of
elbows and bends. The typical flow path through
the wellhead into the flowline was rather
restrictive; first, through a tee and wing valve, to
another 90° elbow or choke tee and then through
at least one more 90" elbow before reaching the
flow line. This type of flowpath slows down the
slug, allowing more liquid to fall back.
During the optimisation, all 15 flowline chokes
were removed and 5 wellheads were streamlined.
Choking a CGL well is not the ideal method of
stabilisation, since bottomhole pressure (BHP)
increases and production rate decreases. Once a
well has reached a low BHP, a high surface back
pressure causes serious loss in production. High
surface back pressure decreases fluid production
and increases the injection gas required per barrel
of lifted fluid. Injection gas usage will be high
and inefficient. If heading is stabilised with a
flowline choke, the production rate may increase
but will be lower than by stabilising the well with
other means.
4 GAS LIFT OPTIMISATION ON BARROW ISLAND, WESTERN AUSTRALIA
The new stabilising methods introduced include
deepening the gas injection point, increasing the
surface orifice (injection choke), reducing the
bottomhole orifice (port size) and/or reducing
tubing size. Production can be increased by
eliminating surface back pressure and surface
restrictions.
No increase in compression capacity and/or
available injection gas was considered at the early
stages of this project. The compressors were
overhauled in July-Sept. 1992 to regain maximum
capacity of, ±12.5 MMscfpd and reduce
downtime.
IGL and PL Pilot Tests
The primary test objective of the IGL and PL
pilots was to evaluate the potential for
improvement in both lift efficiency and gas
usage. The performance of wells with a low
productivity index (PI) (+0.15 bpd/psi) is
characteristically less predictable than that of a
well with a high PI (> 1.5 bpd/psi).
CGL becomes less attractive as the produced
volumes decrease. CGL will usually be more
efficient and less expensive for wells where
continuous flow can be maintained without
excessive use of injection gas. Because of the
cyclic nature, IGL is suited to wells that produce
at relatively low rates « 100 bfpd).
Gas lift assisted plunger lift is basically a major
enhancement of IGL, in that it uses a plunger to
separate the injection gas from the liquid slugs
being lifted. The function of plunger lift
equipment is to provide a more efficient
utilisation of lift gas energy, in wells that can be
produced in a cyclic manner like IGL. The use
of plunger equipment provides the most efficient
form of IGL, by minimising liquid fallback and
eliminating possible gas penetration through the
liquid slug centre. The plunger will surface with
approximately 100% of the load, eliminating the
7% fluid loss per 1,000 feet of lift which is
characteristic of IGL wells.
The five wells selected for IGL and PL were
initially fine-tuned on CGL. After two months
on CGL, the 5 test wells were switched to IGL,
and after another 2 months the 2 PL pilot wells
were switched from IGL to PL. The PL units
were installed in BWI wells L25 (Windalia
Sandstone - medium rate well, 60 bfpd) and
240
SPE 25344
F16M (Muderong Carbonate - low rate well, 10
bfpd). Results in L25 have been promising, with
. a ±19% increase in oil rate (±15 bopd) and
21 % increase in total fluid rate. Gas injection
was reduced 20 Mcfpd by decreasing injection
time from 65 min/day (1 min-65 cycles) to 47
min/day (30 sec-93 cycles). No increase in oil
rate was obtained in FI6M, but gas injection was
reduced 80 Mcfpd by decreasing injection ,time
from 206 min/day (2 min-l03 cycles) to 20
min/day (15 seconds-78 cycles). The forward
plan is to enlarge the pilot test from 2 to 8 wells
in late 1992.
IGL wells usually require more attention than
CGL wells to maintain the maximum efficient
production rate. Also, it is more difficult and
time consuming to run a flowing pressure survey
on IGL wells as compared to CGL wells.
Production Engineers are currently looking at the
impact that individual wells have on the total gas
distribution system. IGL has periods of high
instantaneous gas injection rates followed by
periods of no gas injection. When a well uses
gas, the pressure in the injection system is pulled
down and creates problems at the compression
station since compressors are not suited to
operate under fluctuating conditions. As a result,
the volumetric capacity of the total injection
system should ideally be large enough to act as
an accumulator to help smooth out the flow
surges.
Improve System Diagnostics and Personnel
Training
The economic impact in this area is very difficult
to quantify. Intangible benefits have come from
much closer surveillance, faster detection and
diagnosis of problems and quicker response to
failures. Production Engineering worked with all
personnel concerned to plan and develop a
training program and establish
guidelines/procedures to optimise oil rate and
reduce operating costs.
The following field procedures were established
and implemented during 1991:
Running flowing pressure/temperature
(P/T) surveys in CGL wells
Unloading CGL and IGL wells
Fine tuning CGL and IGL wells
Modified scheduling and testing of GL
wells
SPE 25344 JUAN C. MANTECON
Obtaining complete three-pen surface
pressure charts
Running flowing PIT or flowing BHP
and static fluid gradient surveys in IGL
wells
Service company engineers were used to teach
schools on COL, IGL and PL. A total of 14
courses were presented, so both the engineers and
field production personnel could attend and
discuss local field examples. Field personnel
were also trained on the installation, operation
and maintenance of controllers and PL
equipment.
An extensive program of flowing PIT gradient
surveys was implemented to monitor efficiency
and identify the potential for increased
production.
The tools available to assist in evaluating gas lift
performance are:
Three-pen recorder charts and calibrated
pressure gauges
Acoustical surveys
Tagging tubing fluid level and/or bottom
of wellbore with wireline
Flowing pressure and/or temperature
surveys
These tools can be used individually or
collectively to analyse downhole performance.
Some of the above techniques have limitations
which should be fully understood, so that data
does not misrepresent the actual situation. The
flowing PIT survey is the most accurate and
expensive method of determining GL
performance. To maximise the return of
expenditures on GL diagnostic work, it is
necessary to prioritise well candidates. This was
accomplished by first using the other, less
expensive techniques mentioned above to identify
potential candidates for flowing surveys.
Enhancing Operations/Engineering
Communication
It is important to recognise that optimisation in
any area requires an ongoing team effort.
One known disadvantage of any Gas Lift System
is that continuous engineering design and
consultation is required. The role of the
Production Engineering group is to design and
select the best system for each individual well,
241
5
coordinate the previously mentioned range of
activities, monitor the effectiveness and
recommend change to a more effective system
when necessary. The downhole GL design is of
great importance and is enhanced with a
computerised GL design program.
The role of the Production Operations group is to
keep a OL well producing at optimum efficiency
and rate. Field personnel must follow the
recommended practices for operating,
maintaining and troubleshooting GL installations.
Engineers can provide assistance, but the field
operating personnel must operate the wells and
recognise/report potential problems.
A software spreadsheet database was developed
and used to:
Identify inefficient/low profit wells
Rank wells by injected gas/oil produced
ratio, GLR and/or WOR
Rank wells by difference between
previous month and current oil rates
Allow operators to efficiently allocate
available gas
Identify wells generating the highest and
lowest revenue per unit volume gas
injected
Facilitate long/short term optimisation to
minimise impact of temporary shortages
in gas supply
Detect problem wells to concentrate
troubleshooting efforts before large
production losses occur
This spreadsheet database has benefited both
Operations and Engineering, communicating
information back and forth between the field and
central office.
Fieldwide Optimisation
Fieldwide optimisation (2nd stage) is planned and
being actively pursued.
A meter run/orifice plate will be installed on
existing BWI gas lift wells and four portable
microprocessor-based gas flow meters will be
used for measuring injection gas. The portable
meters provide a more economical option, rather
than installing permanent meters on every well.
BWI currently has no gas metering equipment to
measure the injected gas per well. The present
method of injection gas rate measurement is not
6 GAS LIFT OPTIMISATION ON BARROW ISLAND, WESTERN AUSTRALIA
entirely accurate. The injected gas per well is
calculated using the flow through a choke
equation, assuming a constant 570 psig (for all
wells) upstream pressure and actually measuring
downstream pressure. Formation gas is
calculated by subtracting injection gas from the
measured total gas.
An accurate measurement of total gas injected
and produced in each well is important for both
reservoir management (allocation of formation
gas) and to optimise gas lift production/design.
System analysis through computer modelling of
the GL system is also planned, and an
optimisation / networking software was purchased
in August 1992 for that purpose.
BWI has an interlinked GL system (similar to
most fields) where all GL wells are connected to
a common compressor station through an intricate
distribution system and then produce (from
different formations) into several production
manifolds. As a result, any change in one well's
producing characteristics can affect other wells in
the system.
An increase in a well's rate is usually associated
with an increase of injection gas, which may
reduce the available pressure in the common
feeder line and reduce the gas available for other
wells. Another affect is an increase in flowline
manifold pressure, which will exert additional
back pressure on other wells sharing the common
manifold. The end result is a reduction in the
other wells' producing capacity. Due to the
problem's magnitude and complexity a computer
solution is needed to optimise injection gas
utilisation. This study is planned for early 1993.
CONCLUSIONS
I;ield results at BWI demonstrate that significant
revenue increase can be obtained through a
comprehensive gas lift optimisation program,
even in the relatively unproved area of low rate
gas lift wells. An average gain of ±11
bopd/well was achieved (Fig 4) which is a 27 %
increase in oil rate from 53 optimised gas lift
wells. Total additional oil recovered and
estimated cost through September 1992 were
±109,OOO stbo and ±$A 835,000, respectively.
This rate gain and other intangible improvements
were obtained with a united effort and by
242
SPE 25344
applying sound engineering and operating
practices. An effective communication and
training program are also of great importance for
program success. Investment was minimal <±$A
16,OOO/well) and payout time short « 8 months)
so this type of optimisation program can be
readily adapted to other producing areas, even
those with limited remaining field life. It is very
important to recognise that artificial lift
optimisation is a continuous effort that must be
maintained through the field life.
ACKNOWLEDGMENTS
I would like to thank West Australian Petroleum
Pty Limited (WAPEn management and the
participating companies, (CHEVRON, TEXACO,
SHELL and AMPOLEX) for permission to
publish this paper. Appreciation and recognition
is also extended to all WAPET personnel
involved in this optimisation program. Special
thanks to Ron J. DeBruin, Brian C. Llewellyn
and Ken J. Sample for their manuscript
suggestions.
REFERENCES
Blann, J.R., Brown, J.S., and DuFresne,
L.P.:"lmproving Gas Lift Performance in a
Large North African Oil Field," JPT (Sept. 1980)
1486-92.
Blann, J., Jacobson, L., and Faber C.:"
Production Optimization in the Provincia Field,
Colombia," SPEPE (Feb 1989) 9-14.
Campbell, I.R., Tait, A.M., and Reiser,
R.F.: "Barrow Island Oilfield, Revisited," The
APEA Journal (1984) 289-298..
DeMoss, E.E. and Tiemann, W.D.:"Gas Lift
Increases High-Volume Production From
Claymore Field," JPT (April 1982) 696-702.
Grupping, A.W., Luca, C.W.F., and Vermeulen,
F.D.:"Heading Action Analyzed for
Stabilization - Continuous Flow Gas Lift 1," Oil
& Gas J. (July 23, 1984) 47-51.
Grupping, A.W., Luca, C.W.F., and Vermeulen,
F. D.: "These Methods Can Eliminate or Control
Annulus Heading - Continuous Flow Gas Lift 2, "
Oil & Gas J. (July 30, 1984) 186-192.
SPE 25344 JUAN C. MANTECON 7

Laing, C.M.:"Gas Lift Design and Production

Optimisation Offshore Trinidad," SPEPE (May
1989) 135-141.

Redden, J.D., Shennann, T.A.G., and Blann,

J.R.:"Optimizing Gas Lift Systems," paper
SPE5150 presented at the 1974 SPE Annual
Meeting, Houston, Oct. 6-9.

Simmons, W.E.:"Optimizing Continuous Flow

Gas Lift Wells," Pet. Eng. (Aug. 1972) 46-48;
(Sept. 1972) 68-72.

Steele, R.D.:"Engineering and Economics Used

to Optimize Artificial Lift Methods," Oil & Gas
J. (Dec. 6, 1976) 107-116.

Winkler, H.W. and Smith, S.S.: Cameo Gas Lift

Manual, Cameo Inc. ,Houston (1962).

Winkler, H.W. and Eads, P.T.:" Applying the

Basic Performance Concepts of Single-Element
Unbalanced Gas Lift Valves for Installation
Design," paper SPE21636 presented at the 1991
SPE Production Operations Symposium,
Oklahoma, April 7-9.

SI Metric Conversion Factors

bbl X 1.589873 E-Ol = m3

ft X 3.048* E-Ol =m
ft3 X 2.831 685 E-03 = m3
OF eF-32)/1.8 = °C
in X 2.54* E+OO = em
miles X 1.609344* E+OO = km
psi X 6.894757 E+OO = kPa
scf/bbl X 1.801 175 E-Ql = std m3/m3

*Conversion factor is exact.

243
8 GAS LIFf OPTIMISATION ON BARROW ISLAND, WESTERN AUSTRALIA SPE 25344

~
N

BARROW ISLAND
LOCALITY MAP

DRAWINONo.

*A63511A

FIGURE 1
244
SPE 25344 JUAN C. MANTECON 9

ANALYSIS PATH
GAS LIFT SYSTEMS

NO

REVIEW SHUT·IN &

MARGINAL WELLS FOR
WORKOVEROR
RECOMPLETION POTENTIAL
AND/OR CHANGE A.L.S.

FIGURE~

TYPICAL BARROW ISLAND

GAS LIFT INSTALLATIONS

Production Out

CONTINUOUS FLOW INTERMITTING UFT

APPUCATIONS APPUCATIONS

OPEN SEMI-CLOSED CLOSED

A B c
JIIGURU
245
10 GAS LIFT OPrIMISATION ON BARROW ISLAND, WESTERN AUSTRALIA SPE 25344

PRODUCTION RESULTS OF GLOP

MONTH NO. OF WEllS OIL GAIN (BPDI MONTHLY GAIN (BPMI AV I WEll (BPDI CUM
OCT 90 1 1 1 ·18
NOV 2 12 237 6 219
DEC 2 11 327 6 646
JAN 91 2 11 326 6 872
FEB 4 42 ·382 10 480
MAR 7 113 2.476 16 2.966
APR 8 114 2,955 14 6,920
MAY 10 132 1,777 13 7.697
JUN 14 116 3.319 8 11.016
JUL 14 139 4.232 10 16.247
AUG 14 123 2.812 9 18.060
SEP 15 199 5.663 13 23,723
OCT 15 233 7.030 16 30.763
NOV 20 281 6.944 14 37.697
DEC 20 210 6,389 10 ~.086
JAN 92 23 210 6.119 9 60.206
FEB 28 190 3,995 7 64.201
MAR 28 192 3.746 7 67.946
APR 38 207 3.588 6 61.636
MAY 44 309 8,983 7 70.618
JUN 44 254 5,555 6 76.073
JUl 50 361 6,331 7 82.403
AUG 61 428 10,942 8 93.346
SEP 63 574 16.681 11 108.927

600 ··..····f·········l·..·..·..·r········· · · . ··· ·······[ "[" '!" . ···T······· .: r·········:··· 1' , r········I··········!·········,·········!"········,········1'·····..·,..······1
1 r

500
--lH ----:-
; ; ; NOO
ll•GOAFINW(EBlplOS) ++++-I+-H,········,·········j··
- , ; ; ; ; 'i ; , ····1

400
--I-+t-i-++-+++!-t-+-I--i--H--I-~-+~--I-i-~
300
i I Ii! i ! . ! . i : : ! i ! iii i . I j

200
~;-::::~~f~t;~!:::::;:~:[-l:::::r-l-r"TrtlP
100 i i ! j
r·.. 1 :v
······r······r···..r·····r..· ··r.. .··r······T·······r········!"" ······r..······j..···· ···r··..···!··..····r··..T······r····T··..··j""··..··T·····r·"T"'l
j ! ! iii ! ! i ! j I I I J

o i I i i TTl

FIGURE 4
246