e
IADCISPE
lADC/SPE 17228
Development and Field Testing of a High-Accuracy Full-Bore
Return Flow Meter
by H,K. Johnsen, Petreco AfS; P, Skalle, * U. of Trondheim; A.L. Podio, * U. of Texas;
and G. Sirevaag and A. Vigen, Statoil A/S
SPE Membsrs
-ml 1988, lADC/SPE Drilling Conference
This paper was prepared for presentation at the 1988 lADC/SPE Drilf@ Conference held in Dallas, Texas, February 28-March 21988.
This paper was selected for presentation by an lADC/SPE Program Committee following review of information contained in an abalracl submitted by
the author(a). Contents of the Paper, aa presented, have not been review@ bY the SWietY of petroleum En9ineere or international A==ialion of Drill
ing Contractor and are subject to correction by the author(a), Tha material, as presented, does not naceaaarily raflect any position of tha IADC or
SPE, Ila officere, or members. Papera presented at lADC/SPE mwfinga are subject to publication review by Editorial Committals of the IADC and
SPE. Permission to copy ie restricted to an abstract of not more than 200 words. Illuatrationa may not be copied. The abstract should contain conspicuous acknowledgment of where and by whom the paper is presented. Writ6 Publications Manager, SPE, P.O. Box 83383S, Richardson, TX 75083.3836.
Telex, 730989 SPEDAL.
ABSTRACT
Measurement
of return mud flow is the single most
diagnostic indicator of potential well control and lost
circulation
date,
problems
except
flowmeter,
during drilling
for
operations.
electromagnetic
To
Results indicate that the flowmeter
has performed
successfully
conditions
under
adverse
extended
periods
of time
extensive
modification
field
without
the
of the existing
for
need
of
mud flow
system,
induction
which can only be applied for conductive
INTRODUCTION
drilling fluids, it has not been possible to have access
to accurate
measurements
over a wide range of drilling
adverse conditions
It is essential for the safety cf the drilling operation
of return mud flow rate
fluid properties
that any influx
and
below or losses of drilling fluid are detected promptly
such as drilling through gumbo
and recorded accurately.
shale, gas cut mud and other non ideal conditions.
flowmeter
This paper describes
Up to this date, however,
have not been able to accurately record
and
the flow rate of the returning mud. The main reason
field testing of a full bore mass flowmeter based on
for this difficulty is that mud is a dirty, sticky and non-
the measurement
the design, development
of gas or liquid from the formations
of forces generated due to change
in momentum
of the
return
accomplished
by attaching
flow
stream.This
a short
linear-viscous
is
liquid, and measured parameters such
as mud level inside the return pipe, surface velocity
J-shaped
of the mud, drag forces on immersed bodies, etc. are
extension to the return flow line just before the shale
not directly convertible to flow rate over The practical
shaker.
range of mud properties.
This extension
sensors that measure
is instrumented
forces
with force
caused by the flowing
Presently there are two exceptions;
fluid.
electromagnetic
The first generation
prototype has been extensively
ultrasonic
flcwt meter
doppler
meter
one being the
and the second an
2. Two important
tested in the laboratory and in the field over a range
disadvantages
severely
of flow rates up to 1700 GPM and over wide ranges
electromagnetic
flowmeter:
of mud weights.
based mud bev~ me the fluid is non-conductive,
limit
application
of the
it does not work with oil
and
since one of the requirements for correct operation is
a liquid-filled pipe, the return flow line needs to be of
References and illustrations at end of paper
a U- tube shape.
435
DEVEUl?W1 ANDFIEU IW1ING OF A HIOH ACLWWY FULL EIXE RJNUPNklUD~
IADC/SPE I172?B
Also, If the mud is gas cut to a certain extent the
The momentum force created by the fiuid flowing in
electromagnetic
the bend of the pipe, is recorded by a load cell, as
flowmeter
will
give
erroneous
shown
readings.
in fig. 1.
The momentum
force
can be
expressed as shown in equation. (1);
The
newly developed
full bore flow mete~
,which
FM.
works both with water and oil mud, is based on two
measurements:
the liquid level in the return flow line
where
and the velocity of the mud at a fixed distance from
the bottom of the flow line. From these the flow rate is
estimated
assuming
a fked
velocity
profile.
This
meter is therefore subject to possible inaccuracies
the
actuai
velocity
profile
deviates
from
if
the
Klp
=
momentum force
K1
constant
velocity
cross sectional area
In 1983 STATOIL initiated a project aiming towards
recording
and testing
mud
flow
rate,
of a new principle
for
an idea
by
patented
(1)
FM
theoretical flow description.
the evaluation
V2A
= density
=
filled portion of J-pipe
In equation. (1) the liquid mass flow rats may in itself
be expressed as
PETRECO. The principle can be applied to any flow
of either gas, liquid or solids,
and application
return mud flow measurements
requires only minor
modifications
of the fiow line. Preliminary laboratory
tests
1984-85
during
were promising
and it was
decided to build a prototype, test it and instail it on
the semi submersible
ROSS ISLE for further field
tests in the North Sea during 1986-87.
The principle upon which the fiowmeter is based is
the measurement
of forces generated by changes in
the
of the
mud caused
(2)
m which A is the cross section of the J-pipe,
In order
to know the amount of mass (pA%) at any time in the
J-pipe, the degree to which the pipe section is filled
with mud ( A k) must be recorded, This is done by
means of a weight sensor , and the resulting force is
Fw
flowing
VAk
estimated in equation, (3)
DESIGN PRINCIPLE
momentum
to
where
by
changing the direction of flow through a bend in the
K2gp
LAX
acceleration due to gravity
K2
constant
length of J-pipe
(3)
pipe. In the specific case of the return mud flowmeter
the last 6 feet of the flow iine is separated from the
rest by a flexible universal joint. This last part of the
flow line is formed like a J.
The density of the fluid must be known or measured
either with a gamma ray densitometer or with a mud
balance. Because of the low density of the gas in gas
cut mud gas will not cause and error in
the flow
readings if a pressurized density value is fed into the
computer.
The change of velocity in the x-direction,
AVX, depends on the angle of the J and the velocity
itself as shown in fig. 1.
Based on these
built
with
performance
the
principles
objective
information
the first prototype
was
to
and
provide
data
so that the following tasks
could be undertaken:
a)
to build a fuil scale unit to be installed
on a floating drilling rig;
IAIx/SPE 17228
c)
to
11.1:. JOKNSEN,P. 61WIU,
modify
and
mechanical
improve
construction
both
the
and
the
A. I,. IUXO,
G. ISIREVAAG,ANDA. VIGIN
The mechanical design of the J-meter !s very simple,
However, the development
and testir,g has shown
electronic unit of the prototype, in order
that the repeatability and accuracy is grestly aff~cted
to
by some vital details.
achieve
highor
reliability
and
accuracy.
The mounting of the J-meter should in essence be
horizontal, however, slopes of 1-10
FLOWMEIER DESIGN AND OPERATION
degrees will not
alter its accuracy wiih any significance,
The unit described
extensive
here is a version resulting from
tests and ~xperience
iab and the field.
development
gained both in the
The tests
of a completely
have
resulted
in
The first version of the electronic unit was found to be
new electric unit and
unsuited for field use during the preliminary
major changes of the mechanical design.
Any
changes
(universal
joint,
procedure,
A typical
These
J-meter
are the
is composed
load
cells,
of five main parts.
the
flange
end,
reprogrammed;
the
electronics.
The dimension of the
, the universal joint that connects
Two of the blade springs connect the flange
furthe to the J-pipe, The springs are mounted so that
plane
through
two
of
the
blades
and some of
gained from a shallow gas blowout,
It should be possible to receive
least 6 analog iilput signals.
3,
Changes in signal processing should
be carried out via the computer
program.
4.
Continuous
memory is required in
case the electric power on the rig
should fail.
5.
A printer for continuous storing and
dumping of signals is required.
6.
For ease of interpretation the flow
data should also be displayed as
histograms showing the difference
between flow in and out versus time.
7,
A flight recorder function should
store the last 48 hrs of data and
automatically dump the last 1/2 hours
of flow rate data to a printer in case of
alarms.
is
their wiring, and also extends the base of the load
cells to the flange end. The cover is connected to the
flange end with three bolts.
Two forces are recorded by strain gage load cells
of O - 500
N (1 12.4 Ibf) for the
momentum force and O -2000
weight,
manual
2,
The steel cover protects the J-pipe, the load cells and
a range
many
be
operational
be carried
type
out
at
to ihe direction of each of the forces;
gravity and momentum force.
with
too
to
All changes
of
should
possibly
through software.
end with the ring, and the other two connect the ring
perpendicular
a calibration
software
1.
springs ,a flexible seal and a universal link or chain
each
besides,
after
package were identified.
the J-pipe with the flange end consists of four blade
ring.
the
tests.
hardware
some important design principles for the electronic
ROSS ISLE drilling rig.
fig. 3
cells)
Based on this experience
the experience
J-pipe is 12 ID, which matches the flow line on the
As shown in
load
required
of the
operations were necessary during adjustments in the
universal joint with flexible seal, the cover and finally
the J-pipe as shown in fig. 2.
or adjustments
N (449.6 Ibf) for the
This enables the measurement
flow rate of approximately
of a water
5000 l/rein (1321 GPM),
and for mud of density 2.0 kg/1 \l 6.7 PPG) a flow rate
of 2500 I/rein (660 GPM).
Since the sensors should
An overview of the electronic system is shown in
not be exposed to excessive forces the instrument is
fig.
4. The load cells, filters and amplifiers are standard
equipped with over-load safety bolts which block the
equipment
J-pipe when the full scale force is reached.
(Bofors Electrortik).
The load cells are
connected to the amplifiers via a six conductor cable
protected with Zener-barriers.
437
Amplifiers and Zener-
DEWUWQ2W ANDFIELD mIWJ
barriers
are mounted
in a separate
explosion
hazard Zone 1.
amplifiers
is 4-20
OF A HIGH AaRWFULLI
box outside
The output
from the
mA, which corresponds
XXIEBEIURNl~
The Operation
ItW/SPE
menu lists the operations
that the
user most commonly undertakes:
to the
set time constant
lowest and highest forcg on each of the two load
cells. The two signals are collected by the computer
set alarm
AWACS (Analog Warning And Control System).
set alarm delay
The front and back panel of the computer are shown
print
in
stop print
fig. 5.
terminals
On the back panel several
are seen.
multi-pin
AWACS receives the signals
from the load cells, the densitometer
and the pump
flowmeter and converts them to digital values.
AWACS
is designed
easily be expanded
in a modular
The
form and may
to other purposes
by building
The giet time co -
may be chosen as a vaha
t,etween 1 and 100 seconds and indicates the time
period over which the signal representing
difference
should
be averaged.
the flow
If the difference
or just plugging
between the return flow and the pump rate is larger
more external units to it as indicated in fig. 4 and fig.
than the aiarm Iim& then the alarm activates after a
more modules
5.
into the computer
The AWACS
back-up
battery
is equipped
package
with a long lusting
in order to maintain
its
preset delay.
The alarm
cause a {emporary
continuous memory,
de! ~
changing the pump rate.
is important
when
This change will normally
large difference between flow in
and out, and without the delay the alarm would be
When the AWACS
is turned
on the main menu
triggered unnecessarily.
appears, displaying the following sub menus:
All data concerning
Alarm reset
flow in and out of the well over
the last 48 hours are logged twice every minute and
Operation menu
stored in the AWACS.
By choosing ~,
Display menu
will be printed out .
If an alarm condition
should
occur, this also causes the flow information
for the
System menu 1
these data
last half an hour to be printed out automatically.
System menu 2
After selecting the Display
It takes about 15 minutes for a new user to become
between
familiar with the use of AWACS,
delta-flow
The same goes for
menu one may choose
four ways of graphically
presenting
the
vs. time histogram.
calibration of the J-meter on location, This is done b:
setting the appropriate constants into AWACS
using
the menus. A menu or a function in the sub menu is
chosen by
moving the cursor with the arrow keys
Using System menu 1 all the constants necessary to
calibrate
the pump rate and the fluid
density are set. Through !he sub menu, Autozerow,
the
and pressing ENTER when appropriate.
the J-meter,
load
cells
on
the
J-meter
are
zeroed
automatically.
&j
choosing
Alarm reset all the alarms are shut Off.
The alarms are activated when a preset difference
between pump rate (flow rate in) and return flow rate
is exceeded.
Via the relay outputs, different alarms
may be selected
such as horns or flashing
The word ALARM is simultaneously displayed.
lights.
in System
menu 2 it is possible to choose between
two serial or parallel type printers, to set the time and
to display
the four analog signals
histogram
presentation
overlaying
of delta flow.
The latter
function is very handy when ~hecking the functioning
of sensors
and electrical
connations,
drift
variations, and also for adjusting the J-meter.
the
and
17228
H,K. JUI!NWX, P. SKAWE, A,L. FQDXOi G, WJW3Wf3, ANDA. YI.Cj~
IADc/WE 17228.
TESTING OF THE J=ME1ER
However, since this caused no changes in the flow,
In order to test the J-meter after proving the feasibility
the J-meter
of the.concept, it was decided to put it out in the field
personnel again visited the rig. The technical side of
as early
the problem was quickly solved by changing the link,
as possible,
in this
gearing
environment,
and gaining general field experience.
tests during
the tough
the
quickly
To make prelimina~
towards
way
development
offshore
tests of the prototype and later
the progress
of the instrument,
and
the
was just
left alone
administrative
side
until
was
PETRECO
sohmd
after
ir,volving the mud logging company on board Ross
Isle in the evaluation study.
two
laboratory flow loops were built one of which, in fact,
was a full scale copy of the return flow line on ROSS
ISLE.
Calibration
and testing:
After field installation of the
J-meter the next task was to accurately calibrate the
ins~ument,
Initial measurements
on the rig gave
results that were not of the same quality as those
observed in the lab. The inaccuracy was found to be
flow ~
caused by the different
flow path and liquid fill-up
A sketch of the flow loop is shown in fig, 6. The mud
inside the J-pipe compared to when installed in the
pump is a centrifugal pump with a maximum flow rate
laboratory
The flow characteristics
flow loop.
will
of 6434 l/rein (1700 GPM) at 2 bar (30 psi).
To
vary slightly as a function of slope of the flow line,
control
type of bend, the number of branches,
the
electromagnetic
function
of
the
J-meter
flowmeter working on a filled section
The slope difference
of the pipe was used. Variations in the returning flow
installation
rate sometimes
cause
simulated
occurring
by regulating
on a floating
rig were
the pump flow rate with a
computer controlled air operated butterfly valve.
injecting
batches
of compressed
By
air into the loop
ahead of the surge tank it was possible to simulate
valves etc.
between the rig and the lab
of only a few degrees was sufficient to
different
results.
This caused
no futther
problems as the flow rate could easily be calibrated
against the mud pump and/or an electromagnetic
flowmeter on the stand pipe. The field installation
shown schematically
is
in fig. 7.
gas bubbles migrating through the mud and pushing
Due to rig motions
out large volumes of mud down the flowline.
and the absence of filtering of the
raw signal from the load cells prior to the calibration,
The laboratory tests were performed with flow rates
from O to 6434 l/rein (O to 1700 GPM), and with mud
weights from 1.0 to 1.5 kg/1 ( 8.5 to 12.5 PPG).
it was a time consuming process to adjust the zero
values of the transducers.
The
gave valuable
Iaboratoty tests showed that the J-meter measured
generation
the flow rate with an absolute accuracy of:: 50 l/rein
Hence, this operation
input to the design
AWACS
which
of the second
includes
the
filtering
capability.
(13.2 GPM) over the whole range of flow rates. This
is of the order of
After calibrating the J-meter it was tested for a variety
A 1 %.
of flow rates, and
I?~
ard
The rig: The first prototype J-meter was installed on
the semisubmersible
during the next
ROSS ISLE in 1985. However,
Variation
In the beginning of the
Mechanical
led to
breakdown
of the
section
of
fig.
9.
a stable
fig. 10,
Here it is
~will
suppress
high frequency
variations.
alarms caused by these
In
fig, 77
a similar
comparison as in fig. 8 is made for the paddle type
failures, such as a broken link between J-pipe and
cell
) is interpreted in
flow lag time through the well and the flow line. Sf#l
because no person on board the rig was assigned to
load
(fig. 9
indicated that the main r~ason for delta flow is the
first test period practical problems were encountered
the J-meter.
J-meter are
in flow out of the well during
which is an enlarged
of the return flow line was slightly altered to allow far
the job of maintaining
typical readings from the
drilling period
operated using an off the shelf PC. The last section
the extra curve in the J-pipe.
fig. 9
fig. 8
shown and compared to the mud flow into the well.
because of problems with the AWACS it was then
the
rrud weights
months with both oil and water based muds. In
return fIowmeter.
meter.
439
Given
the
equipment,
crews
typical
skepticism
to
new
of gumbo clay was being produced and accumulated
two different tests were performed to test
the metets sensitivity and accuracy.
on top of the shale shakers to such an extent that the
During the first
flow also went backwards and in between the cover
test the pump speed was increased by an unknown
and J-pipe. This was not cleaned properly at first and
number of SPMS,
the flowmeter showed erroneous flow values.
The object of the test was to
determine the pump speed by means of the J-meter.
Based on the J-meter return flow and the volume per
Problems
stroke, the pump speed was calculated
periods with heave, pitch ( 1.5 degree ) and roll, the
to be 224
Related to Offshore
instantaneous
SPM. Readings on the drill floor showed 226 SPM.
significantly
flow
Operations:
rate was observed
During
to vary
since the flowline axis is perpendicular
The second test was performed after simultaneously
to the rigs longitudinal axis.
starting two stop watches, one
of the average flow rate was unaffected due to the
on the drill floor the
other in the mud logging unit. The pump speed was
However, the accuracy
use of filtering and averaging of the raw signal,
!O be changed at some unknown time from 112 SPM
to 114SPM,
This change
was detected
Summary:
by the J-
found
meter down in the mud logging unit 2 min and 17 s
After almost two years of field testing was
that
the
after time zero. The actual change had taken place at
necessary
to
1 min and 40 seconds.
especially
with
Considering
the small flow
was
seconds,
filtered
this
with
behavior
a time
was
constant
found
of 30
problems:
bent.
After having made a new link
The
load
standardized,
maintenance.
4.
Better over-load safety system should
be designed.
Stainless
steel
construction
is
increase
necessary
to
the
instruments life and reduce corrosion
related problems.
5.
between the load cell and the pipe from a chain as
well as adjusting the over-load
safety bolts,
tile
instrument,
reduction
3,
The wire connecting the
load cell to the J-pipe broke off and one load ceii was
permanently
the
The J-pipe must be protec?ed against
mud accumulations
on the outside
due to splashing,
had to be subjected to what could be
called severe mistreatment.
better
to
2.
and the flow
line often gets plugged. To clean out the clay, once
the J-meter
respect
The weight of the J-pipe should be
minimized to further reduce sensitivity
to heave, pitch, and roll.
In the Central North Sea,
gumbo shales are often encountered,
implement
were
1,
to be quite
impressive.
Mechanical
modifications
of
mechanical problems.
rate change of only 27 l/rein ( 7 GPM) and that the
signal
following
J-
meter was tested again. Even though the momentum
cells
shouiii
be
resulting
in easier
cell was permanently bent, showing a force of 105 N
These modifications
at zero load, this offset was
zeroed out and the J-
second generation prototype which is currently being
meter was back in operation
after only 30 minutes of
work.
have been implemented
in a
field tested in the North Sea.
The overload safety bolts were then tested by
having a man standing on top of the J-meter.
Further applications:
At one time the J-meter showed gradually increasing
Other
potential
applications
flow rate, caused by a slow build up of dried mud on
besides detecting volumetric
the outside
system,
of the J-pipe.
This problem
has now
include
keeping
for the
instabilities
track
flowmeter,
in the flow
of the mud return
been solved by using an extended cover to protect
volume during a cement job. Volumetric
the J-pipe from splashing mud.
during a normal primary cement job is inadequate
No serious build up
control
since currently the return flow rate is not monitored.
of material has been observed inside the pipe.
It
is difficult to detect if some of the cement slurry is lost
After the modifications,
the J-meter functioned well
even with large amounts
through.
of gumbo
clay coming
At one occasion, however, large quantities
440
due to fracturing of the formation.
This is especially
the case when cementing
With an accurate
liners.
.,
IAW/SPE 172%
ILK. JQHNWN, p. WWJll
N+
flowmeter on the return line, the quality of the cgment
W!!?,
G* &UWY&lG, WA.
YWN
REFERENCES
job could be better evaluated.
10
CONCLUSIONS
A new flowmeter
for measuring and recording
return mud in an atmospheric
developed.
Speers, J.M. and Gehrig, G, F,: Delta flow:
An accurate,
The existing
system for detecting
kicks and loss of circulation
the
during drilling,
SPE/lADC 13496 (1985).
flow line has been
flowmeter
reliable
is now of an
2,
Orban, J, J,, Zanner,
K,J. and Orbart, A, E.:
industrially acceptable standard and can be installed
New flowmeter
for kick and loss detection
on existing
during drilling,
SPE 16665 (Dallas,
mud return flowlines
modifications.
with only minor
Extended field tests have proven the
1987).
meter to be adequately rugged and easy to maintain.
The
overall
accuracy
k 1
approximately
confirm
flowmeter
is
of the full range flow rate, the
for
mud weights in
1.0 and 1.5 kg/1 (8.5 and
between
12.5 PPG).
the
1700 GPM.
full range being
the range
?40
of
This accuracy,
on a platform
however,
where
is hard to
the reference
flow
readings in some cases are of uncertain quality.
Field tests
have shown
that ?he instrument
has
adequate sensitivity to detect small changes in return
mud flow associated
with normal drilling operaticms
such as making connections
and reciprocating
pipe as well as small variations
d:l
due to changes
pumping speed.
After
field
calibration,
the
instrument
readings
compare favorably with measurements obtained from
electromagnetic
flowmeter installed at the standpipe.
Signal filtering and averaging has proven effective in
reducing
the sensitivity
of the
meter
to vessel
motions.
ACKNOWLEDGEMENT
The Authors
supporting
express
this
their gratitude to STATOIL for
research
and for
permission
to
publish this information.
441
Sept.
Y////h
f
sad
WEIGWT CELL
cdl
,44J-*Z!=?Q==
s
Momentum
force,
Fm
Volume
y/
~
r
UNIVERSAL JOShT
MOMENTUM
CELL
1
-------
Universal joint
------OVERLOAD
SAPETY BOL
Load cdl
FLANGE ZND
SUPPORT PRAME
ns.2-vw-tkd pm)utim 0sm9J.nKAulh-ir?4wPP01s8mduvMdm
t0th9mummuS?9mSn9.
Universal joint
:.
,
load cell
I Zmw
Barriers
momentum cell
Amplifiers
COVER
w
11
Am
21A
UNIVERSAL JOJNTRING
Stt
set
dansss
constanls
calculates
----T-
F.j. -lllWwdltlN
unlvudj0mt.
h--l
Printer
PIOtter
4-=7
E
-w
Iw
N
Fig. 6--TM 19b.n3twf
m
dllD
low.
m.
FLOWIN (ELMAGN.;
-
$W.
13
16
14
17
10
18
lCQ
L-
/L-~ /
REFUtW
L00Si140
,00
TM
Lwom-
Mow a#lN
L r
13
14
15
AFLOWM.
Ie
IN
17
TIME (HRS)
,MmN@n@mllilnud
FhI.s-CQM@WIOFK@fiW~(
WHY
bcl circvmlOm
to
nLw10ut@fn91u)wlwl
nt13-1.
18
. .,
SE 17228
4
1.5f,o 0.5 -
J - meter
(flow
0.0-
out)
1.51.2-
0.90.60,3 _
0.0
0
0
0
0
m
..
TIME (HRS)
Flo. 9-~y@cd
tlow rocwdl~
by tltolna
rlow mtu)
(~
Of HW
dwlng
.hlOtW
on
bald
HOS2 12LE,com~cd
with tha PUMPfbW tic
9tabk dtllllllg pwkwr.
PUMP SiiurOFF
luMhuNGDOWN
OF ONE
VOLUME
Flow in (ol,~.)
3
DR3LL PIPE, 0,11 m3
2
--
W02JT
FLow
r
d
=0.35m dn
0
0
590xd
20
40
80
ao
\
VOIm dunge= 1/2 35 0,35 n 0,1@5m
Flow WI (p3ddle)
2
MAK3NG CONNECTION
1
Cm+ WITIi ONS JOINT BEFORE
CONNECTING KELLY
0b
o
20
40
60
80
100
ST,\R , PUMP
3
\~
- out
~.
t
INCREASING
TIME
Fig. 10-Enlm-gad portion of J+ootw tlow, hkon trom
Fig. 0 at
pproxlmtely t2:20 hours.
- ~
444
t 00