Specialized Applications of Noise LoggingR.M. McKinley, SPE-AIME, Exxon Production Research Co.F.M. Bower, Exxon Production Rese,arch Co.

IntroductionA 1973 paper1 described how the noise logger candetect flow through poor-quality cement behindpipe. This idea is illustrated in Fig. 1. Turbulencegenerated by fluid moving from Sand A to Sand Ccreates within the tubing a sound field whose in­tensity is greater than the ambient noise level in thewellbore. The logging sonde, which is simply amicrophone, transmits this sound level to the sur­face, where it is decomposed into frequenciescharacteristic of the type of flow. A depth record ofnoise level will reveal peaks at those locations wherethe fluid rapidly changes velocity. For example, seeentry point (A), constriction (B), and exit point (C) inFig. 1.

For this tyPe of application, the noise log com­plements the temperature log very well. But this toolhas other uses, too. We have found that the noise logis a valuable aid to logging methods that track fluidmovement in the wellbore. The purpose of thispaper, therefore, is to extend noise logging'technology into the general area of flow insidecasing. Specifically, we discuss how to calibrate thesonde for use in the following situations: (1) axialflow past the sonde, (2) flow from perforations, (3)liquid production from gas perforations, (4) sandproduction from perforations. For each case, theparticular calibration coefficients refer to the

0149-2136/79/0011-6784$00.25©1979 Society of Petroleum Engineers of AIME

standard detector sensitivity and load described byMcKinley et 0/. 1

Noise From Axial Flow Past SondeSingle-Phase FlowSuppo~e that the logging sonde is hanging in a

. wellbore down which water is being injected./Thevelocity increase acquired by the water while flowingpast the sonde will generate turbulence. This can bedetected as noise. Recall the familiar hiss fromoverhead pipes in' steam-heated buildings. We canexpect a similar sound from flow past the loggingtool. The experiments described in Ref. 1 show that asingle-phase fluid accelerating across a constrictionradiates a noise intensity directly proportional to thepump work required to move the fluid. The sameconcept applies here. If Ap is the pressure differentialrequired to flow a volumetric rate, q, past the sonde,then the resulting noise level should be proportionalto the product Apq.

This is, in fact, the case, as Fig. 2 shows. Data inFig. 2 were measured in a flow loop with a verticaltest section whose diameter varied over the rangeindicated. Flow rates varied from 0.1 to 30 Mcf/D.From the data correlation, we have N*6f.X) = A X t:.pq:Jwhere A is a constant and N*6f.X) is the noise level (atstandard sensitivity) above 600 Hz. For turbulentflow, Ap=Bp(q/A.s) 2, where B is a drag coefficient,p is the fluid density, and As is the cross-sectionalarea for flow between the pipe wall and sonde.

This paper describes the use of the noise logging technique to monitor flow insidecasing. Calibrations for the following flow situations are shown: axial flow past thesonde, flow from perforations, liquid production from gas-zone perforations, andsand production from perforations. The forms of the correlating equations are in­dependent of specific tool design.,

NOVEMBER 1979 1387

~I--II

I-

III/--IIII-­IIII­IIII-III ---------1

NOISE LEVEL~

Fig. 1 - How crossflow b~hind tUbing affects the nois~level inside tubing.

Combining these two relationships givespq3

N*600 = C-2-, (1)As

where C is a constant analogous to a drag coefficient.As happens to the latter, the value of C decreaseswith increasing Reynold's number until the flowbecomes turbulent. For Reynold's number greaterthan 6,000, we find that. for gas or liquid,C=(4::1:: 1.5) x 10-6 • Thus,

3 '

N*600 =4 X 10-6 PQ2' •••••••••••••••••••• (2)As

Inverting this,A 2M Y3

q=63 ( s P 600), (3)

where q is in units of thousand cubic feet per day atflowing temperature and pressure, p is in pounds percubic foot, As is in square feet, and the noise levelabove 600 Hz is in standardized peak-to-peak (P-p)millivolts. For the SI metric system of units, thenumerical coefficient of Eq. 3 becomes 0.25.

These are the calibrating equations. Significantly,they show that noise level is a function of flow rate orvelocity cubed. This is the advantage the tool hasover linear meters. The noise level will be a sensitiveindicator of small changes in an otherwise high flow­rate situation, as illustrated in Fig. 3.

The 600-Hz noise level measurements were ob­tained from a well on injection at 25,000 BID water.A spinner survey had shown that a lower perforatedinterval (below 8,900 ft) was taking most of thewater, While only about 1,000 BID appeared to beentering the perforations shown in Fig. 3. This 1,000

1388

010000

AIR.-0 •Z A WAtERI • DIESEL

N% DETECTOR,.. OIAMEiER - i /iiltl"i5

PIPE SIZES - 17/8" _ 21,-0Y.I

>0a::Ic(... 100Y.I

>Y.I...IW

"".0z'".......0>::;...i

10~c(IoUG.

0....100.1 1

~ MCFc( a p x q - PSI x --IoU DAYG.

Fig. 2 - Noise level generated by flow past a detector.

BID is 40/0 of the total flow, which is at theresolution of the spinner. On the other hand, thechange in noise level across the interval is 190/0.Furthermore, the noise level indicates water is goinginto the top half of the perforated interval. Toascertain the amount of injection, we apply Eq. 3 tothe noise levels above and below the interval. Abovethe interval, N 600 = 1,610 RMS mV. Before this valuecan be used in Eq. 3, it first must be converted fromRMS millivolts to peak-to-peak millivolts. Thisentails multiplying the value by 2.83. Next, the valuemust be corrected for attenuation caused by theloging cable and for the specific gain of the particularlogging sonde. These factors (available from eachlogging company offering the service) are 1.6 and2.0, respectively. The total correction factor is 9.06.Therefore, N*600 = 14,490 p-p mV, while below theinterval, N 600 =1,310 andN*600= 11,773.

The 7-in. wellbore pipe has a 6.37-in. ID while the00 of the tool is 1.69 in. Therefore,

A =_11"__(6_.3_7_+_1_.6_9)_<6_.3_7_-_1_.6_9)s 4· 144

=0.206 sq ft (0.0191m2),

and

A~=0.0424(£1)4 (3.65 x 10-4m4).

At the injection temperature, P = 61 Ibmlcu ft (977kg/m3). Substituting these values into Eq. 3, weobtain a flow rate above the perforations of

_ "( 0.0424 x 14,490) Y3q- 63 61

, = 136 McflD = 24,230 BID (0.044 m31$),

JOURNAL OF PETROLEUM TECHNOLOGY

8700F-'---t---H~-t---_+_-_+-____i

NOISE LOGGER SENSITIVE TO SMALL RATE CHANGES

8900L---....l.-_-..l-__.l....-_~_ _.L._ __I

1100 1300 1500 1700NOISE LEVEL - RMS MILLIVOLTS

-N° _

600 -

8100 r----....,----r---,....---,I--"T'-----,

C p q J

As 2

PIPE SIZE· 7 "

8300 r--_-t-_-+__t-T_O_O_L-rD_IA_-_l_11.,..~-6-"----i

1-~-+---t--+-----:fIII-/-f---l~ /V1-..,. -+---I-t""'---+----j.---I-----4

~

:;)v

'"I---+-----!-+-z---t---t---t----ioo0()

~ 8500~a.1&01Q

(0.0424 X2)11

q=63, 61

The net injection into the interval, therefore, is~q=24,230-22,610= 1,620 BID (0.003 m3Is),which illustrates the tool resolution for this type ofapplication.

However, the same cubic relationship that ensuresthis type of sensitivity also means that the tool willnot detect low· flow rates. At 600 Hz, the ambientnoise level in a typical well is N*600 = 2 p-p mV, whichmeans that for 7-in. casing, the minimum detectablerate is .

which compares closely with the spinner total in­jection of 25,000 BID. Below the perforated interval,the flow rate is

(11,773 ) VJ

q=24,230x14,490

=22,610 BID (0.042 m3 Is).

=7.03 Mcf/D= 1,250 BID (0.0023 m3 Is). Fig. 3- Noise profile across a perforated interval, taking4% of total injection.

This situation could be improved by adding tur­bulence generators to the tool body.

In the above example, the agreement between thespinner rate and that estimated from the noise log iscloser than is usually the case. For water injectionrates down to 5,000 BID in 7-in. casing, the noise-logderived rates have been within 25070 of the flowmeterrates, provided, of course, that the injectionequipment itself is not the dominant noise source.Nearby injection pumps can create mechanicalvibrations that carry to total depth in a well. Fur­thermore, at low flow rates, the noise level at aparticular depth may be the result of carry-awayfrom noise sources located at other depths within thewell. This -is the situation, for example, on the log inFig. 4 from a flowing oil well. The noise levels abovethe topmost perforations are much too large to resultfrom the 1,400-B/D flow past the sonde. This highresidual noise results from 'the high values at theperforations.

,Finally, here are a few comments on the use of the6OO-Hz level. A complete spectrum for this type of'axial flow would show a peak amplitude at about 600Hz for the lower flow rates; consequently, this cutwill pick up these lower rates better than a higherfrequency cut. Moreover, this frequency is suf­ficiently low so that for liquids the numerical valuefor the' coefficient C should be independent of pipesize. Our field data verify this. For gases, however,some dependence of the coefficient C on pipe sizeoccurs. The value given seems to be correct fortubing up to 2 Y8 in., which is the limit of our fielddata.

GaslLiquid FlowWe also have calibrated the sonde for gaslliquid two­phase axial flow, but the results are not very useful.

NOISE LOGGING FOR PERFORATION FLOW

13,450r-------r------r----~

500t------+-

55011------j.-

~ 600~y.

y. ........I '"~

:::J: 0~ :::J:a.. '".....c 650 C'l

700t:r-----+-- ::::::::!;::~;----I

750tt-----+-- "-l-.i-;;;;::-----i

2000 Hz13,800L..

1----.....10-----10...0----1.....J

000P - P MV NOISE

Fig. 4 - Noise log from an oil well flowing 1,400 BID.

NOVEMBER 1979 1389

NOISE LOGGING WITH GAs- LIQUID FLOW

61110III~ I~ ~II

N

I:J:

\~• I•••

--~6190 •~ •&.l- •I •

:J: •~ •~ •1&1Q •

6200

where D p. is the perforation diameter and C' is anacousticcil orifice coefficient. A jet stream from aperforation actually transports the turbulent field tothe sonde, a process whose effectiveness depends onthe jet cohesiveness. The latter in turn is inverselyproportional to the kinematic viscosity, p,/p, of thefluid in the pipe. Hence, we need pq/p, rather than qin Eq. 4. This substitution gives

................... (5)

................... (6)

1390

159 MSCF IDAY GAS26 BPD OIL

185 BPD WATER

where qp (Mcf/D) is the volumetric flow rate fromthe perforation, p (lb/cu ft) is the fluid density, p, (cp)is the fluid viscosity, and DJ! (in.) is the perforationdiameter. For the 81 metric system of units, thecoefficient in Eq. 9 becomes 7234 instead of 35. Theflow rates appearing in Fig. 6 range from 0.1 to 20Mcf/D. Once more, the nearly, cubic dependence of

JOURNAL OF PETROLEUM TECHNOLOGY

To test this relationship, we constructed a core holdersimilar to those used in standard perforationevaluation2 , with the exception that the holder couldbe fastened to and could admit flow into a section ofcasing. Perforations were drilled into the core sec­tions along with equal-sized holes in the casing wall.In this fashion, we ran flow tests on cores whosepermeabilities ranged from 1 to 3,000 md, withperforations of diameter ranging from 1/16 to Y2 in.and depths from 1 to lOin. The size dependencygiven by Eq. 6 is needed to correlate the resultingdata in its entirety.- But in a 'field situation, onehardly ever will know the perforation diameter, Dp.'with any degree of precision. For gases, the same istrue for the density, p, at bottomhole conditions. Wetherefore simplify Eq. 6 to

N600 = C"X~, .~ (7)

where Xp =pqpIp.l)p.This accomplishes two things. First, it removes

some of the pretentiousness from estimating theinfluence of perforation diameter. But more im­portant, the need for estimating downhole density ofgases is eliminated because pq= (pq) se' Use ofdensity ,at standard conditions gives the volumetricflow rate at these same conditions. An expression likeEq. 7 is satisfactory for perforation with diameters inthe range of ~ to Y2 in., as Fig. 6 demonstrates. Thecomposite data are represented by Eq. 8:

N*600 =3.84 x 10 -5X~·86" (8)

which inverts to

so, pq3

N*600 =C '4' (4)Dp

62101...--__.....1.-__---'- -'--__'""":'

0.1 10 100 1000PEAK MV NOISE

Fig. 5 - Noise log from a well flowing gas and liquid at2,160 psi and 180°F bottomhole conditions. ,

More often than not, the state of the gas/liquidmixture is determined by the tangential fluid velocityat the perforati'ons rathe'r than by axial velocity in thetubing. The log in Fig. 5, taken over the perforatedinterval of a well producing considerable gas andwater with a lesser amount of oil, illustrates theinfluence of the perforations. The loss in noise levelabove 6,194 ft results from foam generation at thispoint with the subsequent decoupling of sonde fromfluid. The lower frequency cuts lose character. Thisfoam persisted all the way up the wellbore until thepressure dropped to a point where expansion brokethe foam. This is not uncommon. Consequently, we

'will not pursue the flow-loop correlations. Rather,we proceed to the topic of flow from perforations.

Flow From PerforationsSingle-Phase Flow .

Consider a perforation at least 6 in. deep and 0.5 in.in diameter. Even at formation porosities as low as5070, the porous surface area of the perforationtunnel is still several times greater than the cross­sectional area. This means that the velocity of fluidentering the perforation tunnel is less than that of thefluid leaving the hole in the casing. Becausemaximum velocity determines noise level, the hole inthe casing wall acts as an isolated noise source, just'asany orifice does. This hole is acoustically in­dependent of the formation. Thus, we again maywrite N 600 =A' X (!:I.pq) , where A' is a constant ofproportionality. For an orifice with coefficient B': '

, q2 16B' pq2!ip=B P2 = -2- --4-'

A p '7r Dp".

TABLE 1-NOISE LEV.ELS AT 12 PERFORATIONSIN A GAS WELL FLOWING 5.9 MscflD (Ref. 3)

noise level on flow rate assures that the, most prolificperforations will stand out.

An excellent illustration of this is given byRobinson3 in his Fig. 15. Robinson shows a log froma flowing gas well producing 5.9 MMscf/D dry gas.The perforations are 12 single shots spaced suf­ficiently far apart so that each one is recognizable onthe log. Noise levels at the perforations are given inour Table 1. Four perforations near the middle areflowing a high percentage of total gas. Some of thenoise levels are more than 10 times higher than our.data in Fig. 6. Therefore" we are extrapolating inthese instances with Eq. 9 for values ofX p:. To .obtain the total production from the summed

X's, we use the following values:

Nsoo7,150

13,32518,850

113,000139,000188,000188,00058,50042,25052,00045,5003,250

Total

Xp(Eq.9)782972

1,0972,0542,2082,4552,4551,6301,456

'1,5661,494

59318,762 (3.88 x 1'0&)

Psc =0.045Ibm/cu ft (0.721 kg/m3),p. = 0.024 cp (0.000 024 Pa· s)

Dp =0.45 in. (l1.4mm)

Then,

qsc= pDp EXpPsc

0.024xO.45 x 18,762=

0.045

=4.5 MMscf/D (1.47 m3 Is).

Gol GoIl'QQ.::a..

IIGo

><~ 100

!lIl::

••• I ,(,pq )2.16N'600 =B x I'D:

PERF SIZE: % - Yz"PERF DEPTH: 5 - 10"CORE PERMEABILITY: 1 - 3000 MDFLUIDS: AIR I

WATERDIESEL

The 25% error is within the accuracy of Eq. 9. Suchclose agreement generally will not be obtainedbecause of the tool's inability to resolve noise peaksassociated with closely spaced perforations. Theimportant point is illustrated by the magnitudevariation of the noise levels in Table 1. Four centrallylocated perforations are producing almost 500/0 ofthe gas.

For a second applic~tion of Eq. 9, we refer to thelog shown in Fig. 4 from an oil well flowing at 1,400BID. Since this well has thick-walled casing per­forated with a through-tubing gun, we would expectthe spotty production evident in the figure. There areabout 20 points of major productivity with thesecond and third intervals being the biggest con­tributors. Furthermore, the bottom peak appears toresult from a plugged perforation. The sound pitch atthis point is considerably higher than elsewhere, asthe 2000-Hz curve illustrates. Discounting this lastpeak for the moment, we find for the r-emaining 19peaks that EXp = 5,609 (1.6 x 106 ), which, with thevalues p=51 Ibmleu ft (817 kg/m3), p.=0.6 cp(6 x 10 -4 Pa·s), and D p =0.2 in. (5.08 mm) gives anoil rate of

0.6 x 0.2 x 5,609q= 51

=13 Mcf/D=2,350 RB/D (4.33 x 10-3 m3 Is),

whereas the actual rate was

q~ 1,400 x 1.25 = 1,750 RB/D.

NOVEMBER 1979

10 l::--'----L---L--l.-L.L.l..~_--l.---l.-..L..L..LW:l:l:_:,.___..L-..L-J---L..I_l:_!_LJ

10 100 1000 10,000

PEAK - PEAK MILLIVOLTS NOISE LEVEL ABOVE 600 Hz - N'6oo

Fig. 6- Noise level generated by single-phase flow froma deep perforation.

Single-Phase Flow from Plugged PerforationsWhen flowing, a gas well perforation 6 in. deepwould emit an "organ tone" with a pitch of

c 1,000f= -:::::-- =500 Hz.

4£ 4xO.5· .

This tone would not be emitted by a plugged or veryshallow perforation. We illustrate this differencewith the noise spectra in Fig. 7, where the tone at 450Hz is evident on Spectrum A, the unplugged per­foration. Therefore, we can learn to recognizeplugged perforations in gas wells from their flowingspectra. But even a loose pack in a perforation willeliminate the organ tone without adding much in­cremental pressure drop along the perforation. Theimportant question is: How much flowing pressureloss results from the plug? To help answer thisquestion, we have prepared the correlation in Fig. 8,which shows the incremental pressure loss per unitrate in 6-in.-deep perforations as a function of the4000-Hz noise cut at the perforation. The form of thecorrelation occurs as follows:

N4000 =~x (Apqp),

thus,

Ap =A' 4,0002 'qp qp

1391

TABLE 2- ESTIMATES OF PERFORATION PRESSURELOSS FROM PLUGGING FOR THE OIL WELL IN FIG. 4

Depth tip/qp tip~ NZooo NZooo/c#, (Fig. 8) (psi)13,486 15.4 64 0 013,585 22.0 91 3 013,620 36.3 150 18 913,762 60.0 249 70 3413,754 72.0 298 100 4913,770 203.0 841 260 127

1750 x5.615-2-0-X-1000- - 0.49 t-:1cflD/perf

101----------11---

0;1 '--_......L---.J.--.J.--l--L..I-L..L-l--_--l-_.l-.-.J..-J-.l...J...l-.1..J

100 1000 10,000

FREQUENCY - Hz

Fig. 7 - Flowing noise spectra for open and pluggedperforations.

->: ~ 1001--------1,...--------1-.;,yylL----j.-----+---::I~v~

I

a.1 Cl.<111'

Fig. 8 - Correlation of 4000-Hz noise level with in­cremental pressure drop caused by perforationplugging.

1392

where Ap (psi) and qp .(Mcf/D) are at bottomholeconditions. We return now to the oil well in Fig. 4.Table 2 summarizes values for N"'4000 at selecteddepths along with the incremental values 'of Apobtained from Fig. 8 (Y4-in. liquid curve). Theestimated pressure loss across the plugged per­foration at 13,770 ft is 127 psi. This represents nearlyone-half the total flowing drawdown for this par­ticular well. Thus, the perforation is flowing at onlyabout one-half its potential rate.

Liquid Production_From Perforations in ·Gas ZoneIn some wells, a perforation flowing gas and liquidproduces the liquid as discrete droplets that arecarried. away. .In other wells, the liquid may beproduced into a foamy environment. In eithersituation, the liquid incident on the sonde enrichesthe noise with higher frequency components, giving alog similar to that opposite the plugged perforationin Fig. 4. We have found that for an open per­foration flowing dry gas,

M2~ s3 x 10- 5 ; (3 x 10- 12)(dry gas), .... (10)Xp

where Xp=(pqplpDp) gas. As 'one would expect,increasing values above the Eq.-l0 threshold reflectincreasing concentrations of water in the jet stream.Our data give a relationship approximated by

NioooqIiq = 1.5Dp .JX; , (11)

where qIiQ. is the liquid flow rate in barrels per dayandDp is In inches. For the 51 metric system of units,the constant becomes 1.56 x 10- 3 , rather than 1.5.

For example, a log from a gas well flowing 1MMscflD showed that 20 perforations producedmost of the gas. The well also was producing 200BID salt water. We estimate that

(pqp)gas= 0.054 x 1,000 =150,

p. 0.018 x20

150X p = -- = 333(6.88 X 104).

0.45At all entry points except one,

M2~ sIx 10 -5(1 X 10 -,12),Xp

while at this one perforation the ratio increased to19x 10- 5 • At this location, Niooo =7,050 p-p mY.These values, when substituted into Eq. 11 withDp =0.45 in., give

1.5 x 0.45 x 7,050qIiq = J333

= 261 BID (4.80 x 10 -4 m3Is).

Consequently, this perforation probably was pro­ducing most of the water. A workover verified this.

JOURNAL OF PETROLEUM TECHNOLOGY

a'/JNioooFig. 9 - Volume percent of gas in perforation jet stream

vs noise ratio.

100.01 2 3.. 6 80.1 2 3 .. 6 8 I 2 3.. 6 810 2 30

qGAS - MCF/DAY - PERF @ BOTTOMHOLE CONDITIONS

Fig. 10A - Low-frequency noise level as a function ofgas/water flow rate from perforations.

10.01 2 3.. 6 8 0.1 2 3.. 6 8 1 2 3.. 6 8 10 2 30

qGAS - MCF/DAY • PERF @ BOTTOMHOLE CONDITIONS

Fig. 108 - High·frequency noise level as a function ofgas/water flow rate from perforations.

110

qli,g(B/D)

10035

100

235

qgas(Mcf/D)·

0.100.150.10

0.35

Ygas(Fig. '9)

~224020

Niooo8470

126

365633378

Depth(ft)

6,1986,1966,194

TABLE 3 - ANALYSIS OF RESULTS FROMTHE NOISE LOG IN FIG. 5

f- -

3 I ~I

Q- r~:I I 1-

~~pr1---1-- q; . -'" - ()I---I--h~. I ,01---1-- Q ~ , T::l~ e,0 •

" I

3-,o~ D .,..--0.'.(

8~~)

~GAS - WATER FLOWPERF DIAMETER

RANGE0.30 • 0.60 INCHES

I

~~-~~;~G1

.

~1~~=t1;{-~,,~u;r-8t~~~ o.~f~ .

It~ GAS· WATER FLOWPERF DIAMETER

RANGE0.30 - 0.60 INCHES

10

6000

10008

6

~ ..o 3ooN...>o 100~ 8... 6

'"(5 ..Z 3

>~

II. 10I 8o 6oo ..

~ 3

...'"(5 1000Z 8> 6~

oo

.... 100Z 8I 6o~

ZII

<J

~

>-" 401-----1--

~ 601---1---1---1-

Totals

Gas-Liquid Flow From PerforationsWhenever liquid is the continuous phase in thewellbore, a perforation jet stream with gas createssuch a high level of noise inside the casing that thecondition of the perforation itself is of minor con­sequence. However, the noise contains information

. about the jet stream. As McKinley et 01. 1 explained,the noise in 200- to 600-Hz band is proportional togas rate, while Eq. II shows that liquid content of thejet is related to the higher frequency noise. Thecorrelation in Fig. 9 uses these ideas. This figure, onwhich the actual data points are omitted, showspercent by volume, Ygas' of gas in the jet streamas a function of the ratio of noise levels,~•I IN*2000' where ~. = Nioo - N*6(J() is the noise levelin the 200- to 600-Hz frequency band. The gaspercentage could not be correlated totally in­dependent of liquid production rate. This is the-reason .for the parametric values shown on thecurves. Also, note the gas rate restriction (qgas sO.6Mcf/D/perf) that results from the tendency of thefluids to become foamy.

Fig. 9 helps us to diagnose what is happening in thewell illustrated in Fig. 5. We also have a density logand a spinner survey from this well. The formershowed gas production from the entire perforatedinterval. This production and the resulting wellboreturbulence complicates the spinner interpretation.

Looking at the noise log in Fig. 5, we see the gasflow from the lower perforations. This flow isevident from the separation between the 200- and600-Hz curves. Energy in this band is greater thaneverything above 1000 Hz. We have mentionedalready the foam production at 6,194 ft, whichsuggests this depth is the point of oil entry.Altogether, there are six major entry points. We thenmay estimate the liquid rate per entry point as

185+26qIiq = 6 = 35 B/D/perf.

From the water curves in Fig. 9, we find the jet­stream gas concentration shown in Col. 4 of Table 3.

Ignore for the moment the last two columns inTable 3. The estimated gas percentages, Ygas ' showrelatively lower gas concentrations at the oottom ofthe interval (6,198 ft) and at 6,194 ft. This furthersuggests that these depths may be the water and oilentry points, respectively. To confirm this, we willneed flow rates from these locations. The noise logwill help estimate these rates.

Figs. lOA and lOB show correlations of gas flowrates at downhole conditions as a function of ~. andNiooo, respectively, with liquid rates as parameters.These charts are for gas/water flow from per­forations with the indicated diameter range. We havesimilar charts for smaller perforations as well as forgas/oil flow. These are not needed here because theliquid in our example is almost all water.

From Fig. lOA, we see that the relationship be­tween ~. and q&.as is not very sensitive to liquid rate.This rate is renected more strongly in the Niooovalues in Fig. lOB. This figure then is sufficient toestablish values for both qgas and qIiq' The charts are

NOVEMBER 1979 1393

used in the following manner: All possible solutionsconsistent with the measured value of A· first areread by projecting this value horizontally across Fig.lOA. These values of qgas and qIiq then are trans­ferred to Fig. lOB to give a set of predicted values forNiooo. The correct solution is the one with Niooonearest the measured value. For example, we seefrom Table 3 that at 6,198 ft, A* =365 p-p mV,andN*2000 = 84 p-p mV. Entering the first of the valuesonto Fig. lOA gives the possible solutions shown inCols. 1 and 2 below.

qIiq qgas Mooo(Fig. lOA) (Fig. lOB)

0 0.034 3120 0.060 3650 0.10 32

100 0.10 100200 0.094 230

If these values now are transferred to Fig. lOB, wefind the numbers for Niooo in the last column above.Comparing these numbers with the measured value,N*2000 = 84, gives the solution at 6,198 ft of

(

qgas =0.10 Mcf/D

qIiq =100 BID

In a like manner, we establish the other two entries inthe last two columns in Table 3. At 6,196 ft, a liquidrate of either 20 BID or 50 BID is consistent; thus,the average is listed. We observe from Table 3 thatcomparable amounts of liquid are produced at both6,198 and 6,194 ft. Therefore, the 6,198-ft depth isnot the only point of water entry. A squeeze andreperforate workover was tried without success.

Table 3 shows the deepest three entry pointsproducing the observed total liquid (211 BID) alongwith an amount of gas equal at standard conditionsto

q=0.35 x 130=45.5 MscflD (0.015 m3 Is).

Much of the gas must, therefore, enter at the topthree peaks. Dry-gas jetting into a foamy en­vironment is similar to single-phase flow. Our datashow that for dry-gas flow into foam, the gas jet­stream group is related to noise level by

xp =47 X (Niooo)0.35 , ••••••..•••••••••• (12)

where the l000-Hz cut is used rather than the 600-Hzcut. For the SI metric system, the constant 47 in Eq.12 should be replaced by 9714. Using Eq. 12 at the

"top three entries, we find EXp =670 (1.38 x 105).This, in turn, gives

0.018 x 0.45 x 670. qsc= 0.054

= 100 Mscf/D (0.033 m 3 Is).

This amount added to that produced from the lower·

1394

perforations accounts for the observed gas rate of159 Mscf/D.

The individual rate estimates as above can beconsiderably in error for several reasons. First, theindividual perforations may be resolved poorlybecause of (1) either the distance between stopsduring the logging operation or (2) the interferencefrom adjacent perforations. Secondly, at higher flowrates, noise level becomes only a weak function offlow rate, as shown in Figs. lOA and lOB. This Js theresult of foam· production in the wellbore. We have,therefore, found that the estimated rates are moreuseful in the relative sense, illustrated by the previousexample.

Sand Production From PerforationsAbove 4000 Hz, noise created by fluid movement..,.even water droplets impacting on the tool- decaysvery rapidly as frequ~ncy rises. This results fromfluid viscosity. Such is not the case when sand grainshit the sonde or the casing wall. These impactsproduce a very broadband noise spectrum whosemajor peaks ~e the result of tool· resonances. Therate at which noise level decreases with frequency is,therefore, a good indicator of sand in a perforationjet stream. If we de~ne a decay ratio as

N*4000R= . , (13)N*4000 - N*6000

then a jet stream with entrained sand has thefollowing threshold values in our experimentalchamber.

Single-Phase Flow

R>2.5 (14)

Gas!LiquidFlow

[

3.5; Xp < 100(2.0 X 194)R>

210g X p - 0.5;Xp~100, (15)

where

Xp=X",. +Xp .Yllq gas

These relationships, which assume a jet noise level atleast four times ambient values, provide criteria fordetecting sand production.

We also can relate noise level to sand con­centration in the jet stream:

C,and =B[:7 -1 X10-5] •••••••••••• (16)

where

IbmlDsandCsand = (17)

('!!!.e)II- fluid

The coefficient B, which equals 3.5 x 104 for 16-20mesh sand, varies inversely with mesh size. In the SImetric system, the coefficient B e~uals 3.1 x 108 ,while the threshold value 1 x 10- becomes 1 x

JOURNAL OF PETROLEUM TEr:HNOLOGY

Original manuscript received in Society of Petroleum Engineers office Sept.15, 1977. Paper accepted for publication April 21, 1978. Revised manuscriptreceived Aug. 24, 1979. Paper (SPE 6784) first presented at the SPE·AIME 52ndAnnual Fall Technical Conference and Exhibition, held In Denver, Oct. 9-12,1977.

References1. McKinley, R.M., Bower, F.M., and Rumble, R.C.: "The

Structure and Interpretation of Noise From Flow BehindCemented Casing," J. Pet. Tech. (March 1973) 329-338.

2. API Recommended Procedure 43, 2nd ed., AmericanPetroleum Inst., Dallas (Nov. 1971).

3. Robinson, W.S.: "Field Results from the Noise LoggingTechnique," J. Pet. Tech. (Nov. 1976) 1370-1376.

qIiq =liquid flow rate·, B/D (m3 /s)qp = volumetric flow rate, from a single per-

foration, Mcf/D/perf (m3 Is) .qsc = volumetric flow rate, at standard conditions,

Mscf/D (m3 Is)X p = jet-stream group defined by Eq. 9

(dimensionless in consistent units)Ygas = percent gas, ·by volume, in perforation jet

streamd * ::I Nioo - N*600 = noise level in 200- to 600-Hz

bandp. =fluid viscosity at bottomhole conditions, cp

(Pa·s)p =fluid density at bottomhole ..conditions,

Ibm/cu ft (kg/m3)

10 - 12. The sand rate then is in kilograms per secondrather than pounds per day. All the abovecorrelations come from laboratory data and have notbeen field tested.

ConclusionsWe have illustrated how the noise log can becalibrated for flow inside casing. The forms of thecalibration equations are more important than theparticular numerical coefficients which will change asthe tools evolve. In general, for single-phase flow,noise level increases as the cube· of the volumetricflow rate and decreases as the square of the cross­sectional area normal to flow. Gas/liquid flows aremore complex.

Specifically, we show how to use noise levels toestimate (1) axial flow rate past the sonde for single­phase flow, (2) flow rate from .perforations forsingle-phase or gas/liquid flow, (3) pressure dropacross a plugged perforation, (4) liquid flow rate·from perforations in a gas well, and (5) sandproduction rate from perforations. -

Those correlations using jet-stream noise atperforations are especially useful when standardmethods fail to give a clear diagnosis. We do notwish, however, to leave the impression that the noiselog is a substitute for standard flowmeter surveys.Instead, this device should be viewed as a relativelyinexpensive complement that provides estimates offlow rates. The accuracy of these estimates willdepend on how well a particular tool is calibrated fora particular problem.

NomenclatureAs = cross-sectional area, normal to flow, sq ft

(m2)Csand = sand concentration defined by Eq. 17

Dp = perforation diameter, in. (m)f = frequency, Hz

Nj = noise level, peak-to-peak millivolts, abovefrequency f at standardized sensitivity

p = pressure, psi (Pa)~ = pressure difference, psi (Pa)

q =volumetric flow rate, ~t flowing temperatureand pressure, Mcf/D (m3 /s)

SI Metric Conversion FactorsB/D x 1.589 873 E-Ol =cp x 1.000* E-03 =cycles/sec x 1.000* E +00OF (OF - 32)/1.8ft x 3.048* E -01in. x 2.540· E-02Ibm/cu ft x 1.601 846 E-02McflD x 2.863 640 E-02psi x 6.894 757 E +00sefiD x 2.863 640 E - 02sq ft x 9.290 304 E-02 =

·Converslon factor Is exact.

m3 /dPa·sHz°Cmmkg/m3

103m3/skPastd m3 /sm2

NOVEMBER 1979 1395