Flowing and Gas-lift Well Performance

ABSTRACT The paper describes the tao-phase vertical-lift lunctlon, explains the hydraulics of natural flow, flow through orifices, sumnlarlzes outlines t ~ o - p h a s e methods for estimating individual well capabilities, and includes approximations for solutlon of naturalflow and gas-lltt problenls for tubing of the 1.66-, 1.90-, 2.375-, 2.875- and 3.50-ln. API sizes, and crude 011s in the gravlty range from 25 t o 40 API. INTRODUCTION
Advances I n knowledge of the diHerent lifting methods do not lend theniselves to evaluation qulclily or i n slmple econoniic ternis. I n the aggregate, however, they constitute the necessary b a s ~ sfor in~provedl l f t ~ n g o l i c ~ e sand profitabilities, wherp ever oil is r a ~ s e d . Production by natural flow rightly tops the llst of llftlng methods, inasmuch a s it produces more 011 than all other methods con~blned. It proceeds with n ~ l n i n ~ ucost In relatlve absence of operating diffin~ s culties; and 1 relinquished finally In an atnlosphere charged with regret, and supercharged with explet i v e s intended to fortify the conclusion that the stoppage 1s an irreversible act of Providence. Nevertheless, production men have been haunted for years by the thought that a more defin~teknowledge of flowlng performance would suggest means of resuming flow after premature stoppages, permit more effective well control, more appropriate How-string selections, and serve In general to Increase the proportion of oil quantities economically recoverable by natural flow. bevelopment of organized information on v e r t ~ c a l flow has been s o far a matter of slow growth. A presentatlon in 1930 of the baslc theory by the late Professor boctor J. Versluys' prov~dedan initial impetus for current developments, but h a s been applied only to a liniited extent because of practical difficulties in evaluating factors which appear in the Versluys differential. An Interesting attempt to solve the problem of t two-phase vertical l ~ f by testing flow through short (67-ft) tubes was reported in 1931 by 1. V. Moore and H, 1). Uilde.' Failure of this project to provide the deslred generalization seems attr~butableto use of tube lengths s o short that representative condltions were not attained. kemler and Poole,' in a pawells presented before the American per on flow~ng Petroleunl Institute I n 1936, developed a limited correlation between gas-llquid ratlo and pressure drop per unit of tublng length, and explained a niethod of estinratlng flowing Ilfe. The work of C. J. May and A. Laird4" resulted in vertical-l~ft peneralizations well adapted to predlct results within a restricted range of condltlons. The interesting paper by Poettn~an and Carpenter6 appeared subsequent to the tinie of derivation of the material here presented. "Gas-Llft Principles and Practices" by S. E'. Shaw,' the pioneer consultant on vertlcal flow, provides an lnterestlng discussion of gas-lift history and methods with correlations which, though limited in scope, were none the l e s s useful. Shaw's observation, that power lunctions may be applied in approximat~ng relationships between nlininiuni gasthe lift Intake pressures and glven liquid production rates, has been used here. The excellent paper by L. C. Babsone added considerably to knowledge of vertical Bow, particularly in the range lor gas-llquld ratios greater than 2.0 Mcf per bbl. T o a large extent the present paper i s a result of reviewing Babson's data and work after addlng a considerable fund of depth-pressure information involving gas-liquid ratios l e s s then 2.0 blcf per bbl. Thus, Shah, Babson, and the late L. N. R,lerrill mentloned by Babson, provided the prior hork nlainly used In the appended correlations of vertlcal flow. It will be noted that no distinction is made here between gas-lift and natural flow. In the gas-lift How and annular flow range covered by Babson, n ~ i s t predominate and no perceptible diHerences are to be expected. hhere foam How exists, one would be led to expect son~ewhatsteeper gradients for natural

* N. V . De Bataafsche Petroleum
t

'

Maatsch~pp~, The Hague, The Netherlands Presented at the sprlng meet~ngof the P a c ~ f i c Coast D ~ s t r ~ c t , D ~ v l s ~ o n Product~on,Los Angeles, May 1954. of References are at the end of the paper.

FLOWING AND G A S L I F T WELL PERFORMANCE flow than for gas-lift wlth the same total gas-l~quid ratios, because for natural flow more of the g a s would be in solution, and the ratios are small enough that the g a s in solution could be an appreciable part 01 the total available. I n any c a s e , i t may only be sald that no fimi differences between gas-lift and two-phase natural flow have been observed; and the results here presented involving foam flow are derived from flowing-well data and would therefore tend to represent foam-flow g a s requirements conservatively. The present purposes are to explain flowing and gas-lift well performance i n the light of what is now known of vertical flow, and to provide procedures and empirical correlations which have been usefully applied over a period of years i n solution of practical problems. It i s believed that the p a p h i c a l procedures suggested are ~ l d e l yadaptable. I'he gradient data, however, should be used with care and are oflered in the hope that they may encourage development of more specific data in forms readily adaptable to field use.

127

A PRIhlARY DISTINCTION
T o develop an understanding of the behavior of flowing and gas-lift wells, l t is essential first to recognize that there i s one s e t of conditions affecting f l o ~ gas-liquid mixtures into a bore hole, an enof tirely different s e t of conditions affecting flow of mixtures from the bottom to the top of the well, and a third s e t affecting flow through beans at the s u r face. W may designate the first a s "inflow performe ance, 1 9 the second a s "vert~cal-lift perforn~ance," and the third a s "bean performance." Inflow Perfornlance Knowledge of individual well inflow performance i s a basic necessity in equipping and operating 011 wells for n~aximumprofit under any s e t of imposed conditions. Rigorous determination of the inflow performance of a well at any s t a t e in i t s producing llfe would involve: 1. Measurement of the static pressure in the well at the midpoint of the producing interval. 2. Measurement of the mass rate of inflow of each fluid phase corrected to midpoint conditions for each of a s e r i e s of steady operating pressures measured at the midpoint. Fortunately, such detailed deternlinatlons are seldom necessary; but one or more very complete determinations in a particular field can prove helpful in de-

ciding what short-cuts and approximations are locally practicable. The liquld inflow performance of a well at any stage in i t s decline may be defined by ~ t static s pressure (in pounds per square inch) and maxirnunl inflow rate when the increment of inflow per unit of pressure or productivity index (PI, measured in barr e l s per day per pound per square inch) i s a constant. The gas-liquid ratio (GLR measured in thousand cublc feet per barrel)may be, for practical p u r poses, constant. Under such simple conditions the inflow perforn~ancecan be depicted a s shown in Fig. 1. Individual-well inflow-performance trends may be depicted a s shown in Fig. 2. T h i s latter type of graph i s useful in predicting both the timing and character of lifting changes needed to mantain production and implement the local reservoir policy. The relation between the midpoint pressure and the liquld inflow rate (inflow performance relationship, or IPH, a s mentioned later) i s not always a straight line a s shown in Fig. 1, but may be concave to the origin a s shown in Fig. 3. Even l f this curvature i s marked, it is posslble by study of a group of such curves from a given field tp develop a locally applicable nrethod of predicting the IPH from the well's static pressure and a single drawdown deterniination. Gas-liquid ratio control i s a principal factor i n development of reservoir policy. Hatios in practice are affected by cumulated withdrawals; and, at any one stage of withdrawal, they are altected by production rates, a s has been pointed out by H. J. S u l l i ~ a n . Generally, of all factors bearing upon ~ future well performance, the gas-liquid ratio is the least predictable.

Lu

i? 5
Lu

2000

z Z 0 4 z
LU cd
I -

w3
a

1000

2
200 400

boo

800

PRODUCTION IN BID

Fig. 1-Diagram of lnflaw Performance

128

W. E. GILBERT
of watercut curve shown in F'lg. 1 generally indic a t e s water influx predon~inantlyfrom a relatively low-pressure source, and the cut curve in Fig. 3 a predominantly high-pressure source. Cut-cumulative curves such a s the one shown in Fig. 2, are of well-known value in estimating future oil recoveries fronr wet wells. In selecting the best operating gross rate for an helpful to plot individual wet well, it IS sometin~es the approximate water IPH from the gross IFH and two or three cuts a s indicated in E'ig. 4. However, if s p e c i d t e s t s are necessary for this purpose, it i s desirable to remember that temporary use of abnornlally high inflow rates can induce a permanent in@ GROSSLIOUID IPR
DETERMINATIONS

@ )

STATIC PRESSURE MAXIMUM GROSS INFLOW R T AE

GASILIOUID RATIO CURVE)

@(CUT Cz% ti

.
0

CUMULATED OIL WITHDRAWALS

Fig. 2-Graph

of Individual-well Inflow-performance Trends

There have been instances of gas-oil ratio reduct~on following a change in the tubing setting depth In flowing wells. However, the practice i s not recommended, especially i f it involves killing the well with water, inasmuch a s the change In tubing depth can have only a very minor effect of the drawdown opposite gas s a n d s for a given liquid rate; and if the water reduces gas production, i t i s likely, at the same time, to reduce the well's future oil potential. hater can be a useful agent In an oil reservoir; but on reaching the well bore, i t s usual tendency i s to mininilze flowing life, ultimate recoveries, and total profits. The operator i s fortunate when water production can be economically excluded. The type
Fig. 4-Procedure
MOO
GASlLlOUlD RATIO = 1 5 MCFIBBL

3 CORRESPONDING WATER RATES

DEFINING W T R IPR AE

MAXIMUM GROSS

LIQUID INFLOW RATE

for Outlining the Water IPR

2000

1OOO

PRODUCTION IN BID

F i g . 3-Diagram

of Inflow Performance with Curved IPR

crease in water productivity under some circunistances, e.g., if the water shut-off i s insecure, or if there is a wide disparity between water and oil viscosities, gentle formation dips, and edge water close to the bore. Lilfierential depletion i s progressive during sustained flowing periods wherever the ratio of lateral permeabilities to vertical permeabilities i s large; and interflow through the bore between producing layers takes place durlng any subsequent periods of shutdown unless a suitable mud i s spotted in the producing interval. Thus any water inflow from a relatively high-pressure source tends to seek out and enter the more depleted oil layers during a shutdown period, and with permanent injury in some fields to effective oil pernleabilities. After long pe-

FLOWING AND GAS-LIFT W E L L P E R F O R M A N C E

129

n o d s of flow, a two- or three-week shutdown can cause from 20 to 40 percent permanent reduction of the inflow capacity of a well tapping a depletiontype reservoir even if the water cut i s no greater than 5 percent. Thus, differential depletion i s a factor requiring close consideration in many fields if the operator's equity in operating wells i s to be protected. Generally, the rate of liquid inflow increases a s the operating pressure in a well i s reduced, and the absolute maximum'liquid inflow would result if zero absolute pressure could be maintained at the bottom of the well. T h i s condition, of course, cannot be attained, and l e a s t of all in a flowing well, because a pressure drop in the tubing from the bottom to the top of the well is necessary to sustain verti:a1 flow. A large pressure at the bottom of a well facilitates vertical outflow but discourages lateral inflow. In this sense, the pressure requirements for inflow and those for vertical lift are opposed to one another, and, in particular c a s e s , when an effective compromise can no longer be made by adjustment of controllable factors, flow must cease. For any steady flowing condition, the sum of 1 , the effective pressure dfop from the drainage radius to the bore, 2, the pressure drop i n the vertical column, and 3, the pressure drop across tbe bean (or orifice) at the surface is substantially equal to the difference between the well's static pressure and the flow-line pressure. Vertical-lift Performance All we need know about two-phase vertical flow i s how much pressure i s required to lift the well liquid a t a given rate from a given depth with a give n gas-liquid ratio through tubing of a given size. T h i s problem i s more complicated than the problem of single-phase flow in surface pipelines because the flow of a gas-liquid mixture, we are dealing w ~ t h and because the input pressure must be sufficient not only to overcome flow resistance in the pipe and the bean a t tbe surface, but must in addition be sufficient to support the total weight of the compressible mixture in the pipe. In single-phase horizontal flow, the total pressure drop for a given flow rate can be represented a s so. many pounds per square inch per thousand feet of length. No such convenient can be used for vertical two-phase flow because the pressure drop per unit of length i s not constant, but increases with depth. For this reason, in proceeding to systematize field information on vertical flow for pressures l e s s than the bubble-point pressure, we find there i s a different

depth-pressure gradient for each s i z e of pipe, each rate of liquid flow, and each gas-liquid ratio.*
Depth-Pressure G r a d ~ e n t s

The depth-pressure gradient i s the basic unit of two-phase vertical flow, and solution of individual well problems i s largely a matter of having available a representative family of gradient curves covering suitable ranges of tubing s i z e s , rates of liquid flow, and gas-liquid ratios. h e are not concerned here with any detailed analy s e s of the physical phenomena which cause the pressure gradients observed or derived from practical field information. However, it is of interest to know that a single gradient curve represents a sequence of different types of flow. I hus, starting a t the upper end of a gradient like that for 200 bbl per day in 2.875-111. tubing w ~ t h1.0 Mcf per bbl gasliquid ratio, a s shown at A in Fig. 5, mist flow will predominate a t the lowest pressures, modified progressively by an upwardly moving oil film which clings to the inside surface of the pipe and increases in thickness with depth. T h i s film, combined with mist flow in the center of the pipe, has been described a s annular flow. A s relative depths Increase, the film becomes s o thick and wavy that it occasionally bridges across the section, resulting In slug f l o ~ . still greater depths, slug flow merges At into foam flow, and this finally merges into singlephase flow a t the pressure beyond which all of the gas is in solution. Fig. 5 illustrates the use of a gradient curve in determining the tubing pressure from the intake pressure, or the intake pressure from the tubing pressure, for a given production rate and ratio for a well of any depth. For instance, as shown a t t) in Fig. 5, 5,000ft wells with 2.875-in. tubing producing 200 bbl per day with 1.0 Mcf per bbl will have an intake pressure of about 440 psi if the tubing pressure at the surface is one atmosphere (zero gage and, a s shown a t C, they will have an intake pressure of about 1,750 psi if the tubing pressure i s 800 psi. Similarly, as shown a t D in Fig. 5, an 8,000-ft well for the same conditions having an intake pressure of
*Grad~entspresumably are a l s o affected by many other factors lncludlng 11qu1d surface tenslon, v ~ s c o s ~ tand gravlty, flowy lng temperatures, g a s gravlty, and gas-l~quids o l u b ~ l ~ t ~ e s . However, there 1s a reasonably c l o s e correspondence between results whlch have been obta~ned the 11ght-011 (25 t o 40 API) In fields of Long Beach, Santa F e . Domlnguez, Ventura, Canal, and T e n S e c t ~ o n , and several forelgn fields, vv~thoutad~ustlng for such factors. Also, d has not been found necessary to correct grad~entsfor water cuts. However, the grad~entsare Inadequate to predlct the effects of emulsions.

130

W. E. GILBERT
27, the t ~ o - ~ h a s e funct~onfor any particular depth and tubing-outlet pressure may be constructed. For to example, confining attent~on 8,000-ft wells, 2.875in. tubing, and zero gage tubing pressure, the function may be represented either in terms of intake pressure and gas-liquid ratio, a s shown at A in Fig. 6, in terms of Intake pressure and production rate, a s shown at B in Fig. 6, or in terms of intake pressure, production rate, and gas-liqu~dratio, a s shown a t C in Fig. 6. All three q a p h s represent equivalent information: At A i s the type of graph used by Babson; the fornl of illustration B is interesting because the ordinates and abcissae are the same a s those for the inflow performance (1PH)of a well, and the two types of data may thus be superimposed. They are cons~dered following in connection with estimates of the duration of flowing life. Several general characteristics of two-phase vertical flow may be observed at C In Fig. 6. In particular, it may be noted that: 1. For any constant gas-liquid ratlo there i s a rate of flow which requires minimum intake pressure. Also, this rate of Bow for minimum pressure and the minimum pressure itself both increase a s the gasliquid ratio i s decreased a s indicated by curve 1. (These observations are of Interest in connection with flowing wells because of the tendency of flowing wells to have a more or l e s s constant gas-liquid ratio a t any one tlme.) 2. For any constant rate of flow, there i s a gasliquid ratio which prov~desminimum intake pressure. 1his minimum intake pressure i s directly related to the rate of flow, while the gas-liquid ratio s for minimum intake pressure 1 inversely related to by the rate of flow a s ~ndicated curve 2. ( I h e s e o b servations are of interest in connection w ~ t hgaslift w h ~ c h permits control of gas-liquid ratios.) The form of the two-phase function i s largely the result of the Interaction of flow resistance *and s l ~ p page of g a s through the oil, the resistance factor being least important when slippage i s greatest and vlce versa. It i s more or l e s s obvious that the column pressure, being the result of the weight of the mixture, i s greatest at low gas-liquid ratios. However, it i s l e s s obvious that for any gas-liquid ratio and depth there i s a rate which requires minimum lifting pressure with lower rates requiring more lifting pressure because of slippage, and h ~ g h e rrates requiring more lifting pressure because of resistance.
$

2
4

6
8
10 12 14

PRESSURE

PRESSURE IN HUNDREDS OF P S 1

Fig. 5-Use of Gradient Curves

1,390 psi will have a tubing pressure of 200 psi. Thus by using the gradient for the desired production rate and ratlo and interpolating when necessary, we nlay estimate either the tublng pressure or the intake pressure in a well of any depth when one of these pressures i s known. Empir~calgradient curves for 2.875-111. tubing are glven in Fig. 23 through 27,*t and slmilar families of g a d i e n t curves for 1.66-, 1.90-, 2.375- and 3.50in. tubing are given in Fig. 28, 30, 31, 32, and 33.*$ l'he 2.875-1n. gradients are based upon correlation of data with gas-liquid ratios ranglng from 0.4 to 2.0 Mcf per bbl from 8,000-ft wells in Ten Section F ~ e l d , combined with Babson's generalizations in the gaslift range. Errors attributable to the 2.875-in. curves for 100,200,400, and 600 bbl per day should not exceed 15 percent. The gradients for 50 bbl per day in the 2.875-in. s l z e and for 50, 100, 200,400, and 600 bbl per day In the other s i z e s are based upon information which is far l e s s con~plete. However, they are offered In the belief that they reliably indicate the relative characteristics of each s i z e and are not likely to lead to nl~sapplications if conserthat vatively employed. Further, it may only be s a ~ d they represent a fair correlation and interpretation of the 'lnformation readily available.
The Two-phase Vertical-lift Functlon

'

Using the process outlined in Fig. 2, and a s e t s, of g r a d ~ e n t . c u ~ esuch a s those on Fig. 23 through
*In each plate, group A includes gradients for gas-llquld ratios whlch a r e ' l e s s than the optimum, and group B includes gradients for ratlos greater than the optimum. The gradient for the optlmum gas-llquid ratlo (called optimum because it prcvldes the lowest pressure for the given rate of flow) 1s at the bottom of group A and I S marked wlth an m o w pointing t o the ratio. The grad~ents have been dlvided Into two groups slmply to avold crossing of Ilnes.

t See p.

145-149, ~ n c l .

$ See p. 150, 154-157. incl.

FLOWING AND G A S L I F T WELL P E R F O R M A N C E

13 1

0
1

GAS I LIOUID RATIO IN M C F 1 B

-LlOUlD PRODUCTION IN B I D - - a

Fig. 6-The Two-phase Vertical-lift Function for 2.875-in. Tubing Set (Tubing Pressure = Zero P S I Gage)

at

8,000 F t

132

W. E. GILBERT

For the reader who finds difficulty with this s e t of facts, the following explanation may prove helpful: a. Imagine for this purpose a 2,000-ft length of 2.875-in. API tubing mounted vertically on the face of a precipice in the Hocky Mountains with s a wind-swept platform at ~ t upper end where the reader may observe results, and a special-type pump station a t i t s lower end capable of continuously injecting oil of, say, 0.85 gravlty and g a s with a constant ratlo of, say, 1.2 hlcf per bbl. l b. Taklng the i n ~ t l a rate a s 0.01 bbl per day of oil with the 1.2 Mcf per bbl ratlo, the liquid level would appear a t the upper platforn~after about 115 days (2,000 0.01 x 172.7 = 115.8). Out of the oil column the observer would s e e g a s bubbling a t the rate of about cu in. per s e c (1,200 x 0.01 x 1,728 + 1,440 x 60 = 0.24); and the pump a t the lower end would be operating at graan input pressure approximating the l ~ q u i d dient pressure, amounting to 736 psi (2,000 x 0.433 x 0.85 = 736.1). kiost of the gas punlped Into the tubing durlng the 115-day period would have slipped out of the column by the time the level of the mixture reached the observation platform. c. Now, I pumplng of the same mixture were maintained a t 400 bbl per day, because the l ~ q u l d alone would fill the pipe in l e s s than 45 min, (2,000 x 1,440 400 x 172.7 = 41.7+), there would be very llttle tlme for slippage of g a s through the oil and the 480 hlcf per day of g a s (400 x 1.2 = 480) accompanying the oil would soon produce an oil mlst around the top of the pipe, the gas-oil volun~eratio at that point belng about 213 volun~esper volume (480,000/400 x 5.614). The input pressure would be about 150 psi. d. Last, with an i n j e c t ~ o nrate of 4,000 bbl per day, slippage would be obviated by extreme turbulence and the input pressure, increased by resistance, would be about 230 p s ~ . Similarly, if we hold the liquid rate constant and increase the gas-llquid ratlo, the intake pressure decreases, from a starting point with zero g a s when welght and the the pressure 1s the sum of the l l q u ~ d l i q u ~ dflow r e s ~ s t a n c e ,to a mlnlmum, and then increases steadily for greater ratios due to increasing resistance accon~panyinghigher fluid v e l o c i t ~ e s . ' he observations sun~marizedin Par. 1 and 2 ap1 ply In general to two-phase v e r t ~ c a lflow from any

Y1

rm 200 m LIQUID PRODUCTION IN B / D

I
400

F i g . 7-comparison from

of T u b i n g S i z e s for V e r t i c a l - l i f t

5,000

F t w i t h Z e r o T u b i n g Pressure

depth wlth an educter of any s l z e or type and any given outlet pressure. l h e relative effects of the difierent tubing s l z e s are indicated In Fig. 7 (which was constructed from the gradlent curves of Fig. 23-28,30-33). In general, the smaller tubing s u e s offer the advantage of lower Intake pressures at comparatively low rates of flow, of and therefore tend to prolong the flowing l ~ f e low gas-l~quidratio wells. The smaller s i z e s , however, limit rates of Bow, especially for the higher gasliquid ratios. Annular flow i s not treated here. However, it may be mentioned that the poorer an annulus i s as a flow section for slngle-phase flow, the more effective it will be for two-phase flow In minimlzlng gas s l ~ p p a ~ e and in improving g a s utilization in the slippage range a s compared with a circular flow string of equal s e c t l o n d area.

Two-phase Bean Performance "Bean" i s the 011-country term for the orifice used on the tublng outlet to control the production rate men are accustomed to of a flowlng well. Product~on

FLOWING AND G A S L I F T WELL P E R F O R M A N C E

selecting bean s i z e s for particular wells on a trialand-error b a s i s and no correlation for two-phase flow through beans has been generally applied. The following approximation i s derived from regularly reported daily individual well production data:

435 r 0 5 4 6 B
=

1.89

wherern: P = tubing outlet pressure, in psig r = gas-liquid ratio, in klcf per bbl = gross liquid, i n bbl per day 5 = bean s i z e , in sixty-fourths of an inch ' Approximate solutions for any one of the lour variables when the other three are known, may be made by use of Fig. 29. Directions for use of this chart with examples are given on p. 151. The constant (435) used in the formula i s based upon Ten Section data with beans of the types shown in Fig. 29, the results of one sanipling from the Ten Section Field being shown in Fig. 8. An error of '/,,,-in. in bean s i z e can effect an error of 5 to 20 percent in pressure estimates. In many c a s e s , gas-liquid ratios are reported bnly to the nearest 50 or 100 cu ft per bbl and are frequently difficult to determine because of fluctuations which occur In many wells. Spot readings of pressure at heading wells are not representative of daily averages. For reasons of thls kind, no fornrula could be expected to maintain a 100 percent correspondence with observed data at individual wells even though it correctly correlated the variables involved. As an exan~pleinterpreting the forn~ula,the p e r formance of ''/,-in. beans 1 in. in length i s illustrated
13

LlOUlD PRODUCTION IN B I D

Fig. 9-Performance Pressures More than

of

''4, -in.

Bean for Tubing

70

Percent Greater than the

Lead-l ine Pressure

15

11

18

ACCURACY

7 / 40

20

60

80

100

PERCENT OF READINGS

Fig. 8-Indicated

Range of Accuracy of

Bean-performance Formula (Ten Sect~on Data)

in Fig. 9. In the type of formula used, it i s assumed that actual mixture velocities through the bean exceed the speed of sound, for which condition the downstream, or flow-line, pressure h a s no eflect upon the tubing outlet pressure (i.e., pressure on the upstream side of the bean). 1hus, the formula applies for tubing pressures at l e a s t 70 percent greater than the line pressure. Fo; tublng pressures l e s s than 70 percent greater than the line pressure, the bean s i z e indicated by the formula will be too small for the given conditions. fio study has been made of two-phase bean performance for tubing pres-. sures in the range from zero to 70 percent greater than line pressure. Field men usually try to avoid operation in this range because fluctuations of line pressure affect the well's operation. Flowing-well Performance Individual well problems in natural flow may be analyzed quite simply by graphical means, a form which has general application being shown in Fig. 10. In this figure, curve A represents the inflow p e r formance of a well with tubinglntake pressures plotted against liquid production rates. The gradient curves B for 2.0 hlcf per bbl gas-liquid ratio were then plotted starting with the known intake pressures for 0 , 50, 100, 200, and 400 bbl per day. The intersections of these gradients at the surface establish the tubing pressures which the well will sustain, and the tubing-outlet performance may then be plotted a s shown by curve C . The depth vs. pressure diagram of Fig. 10 illustrates the part

134

W. E. GILBERT

played by gradients in flowlng-bell perforniance. liowever, curve C may be obtpned dlrectly iron, curve A by subtracting the total gradlent pressure for each of several rates of flow. Supenniposlng the perfornlance curve I) for a ''/,,-in. bean, we conclude that the well should flow a t about 96 bbl per day with a tublng pressure of about 850 psi. Cons ~ d e r i n gcurves C and L), a s shown i n Fig. 11, it will be seen that there are p o s s ~ b l eequllibr~un~
eaamAL W
U DATA

PRODUCTION IN BID

Fig. 11-Diagram

S I A K FUESlR # PSI = ZYI) INROW RATE IN BID = YX) W l W U D RATK) IN M C F I B B L = 2 UOUDWGT INPSIPER 100 Fl = 3 8

nu

illustrating Equilibrium Conditions a t the Well Head (For same w e l l as shown I n Fig. 10)

ratlo. Both the tubing outlet curve C and the bean curve 13 change with each change in the gas-liquid ratio after the manner shown a t B in F'ig. 12, but for each ratio there IS a stable equil~briumpoint. The relationship between bean s i z e and production rates i s of the type shown at A in Fig. 12. Two similar curves derived from wells where several bean s i z e s were used over a short period of time (6 months or l e s s ) are shown In Fig. 13. Neither of these wells could be expected to maintain steady flow a t rates of l e s s than 50 bbl per day, although they both would produce satisfactorily a t h ~ g h e r rates. The reason for t h i s limitation l i e s in the fact that the differential pressure, available to increase product~onwhen the rate of flow temporarily drops below the stable equilibrium rate, progressively dlmlnishes a s the bean s i z e i s reduced below the bean s i z e which provides maximum tubing pressure. Thus, considering the bean-performance intersections with the tubing-pressure curve for 1,600 cu ft per bbl at B In Fig. 12, a ''/,,-in. bean provides
Fig. 10-Diagram of Flowing-well Performance

points, 1 and 2. Intersection 1 provides a stable flowing condition because, if a tendency develops to increase the flow rate, the bean imposes more pressure resistance than the well can sustain; and if a tendency develops to reduce the flow rate, the well develops a higher tubing pressure than the bean requires. In each of these c a s e s a temporary displacement of the flowing rate generates a press sure differential which returns the well to ~ t equilibriumproducing condit~on. Intersection 2 i s an unstable equilibrium point because any temporary displacement brings into action a pressure d i f f e r ential which increases the displacement, and causes the well either to flow faster or to die, depending upon the direction of the initial displacenlent. T h i s limited explanation assumes a constant gas-liquid

100

200

300

400

PRODUCTION IN B I D

Fig. 12-Effect

of Gas-liquid Ratio on E q u i l i b r i m Production Rates

FLOWING AND G A S L I F T WELL PERFORMANCE

1 6 1 6
, ~

135

yzlJ

12

N ~ Z Ggz a z<*
I + u (5 0
Ln

a
4

0 0 100 200 300

PRODUCTION IN 0 I D

0 0 I00 200 300

PRODUCTION IN 0 I D

rates. h e e d l e s s to say, the most popular procedure in handllng wells with allotnlents too small to support flow with the existing tubing s i z e i s to install nlechanical lift. Nhether or not this IS the best procedure depends upon the well's future capabilities a s judged from data summarized in the form indicated in Fig. 2.

Fig. 13-Relationship Between Production and Bean Sizes

FLOWING LIFE
If future IPH's estimated froni a graph such a s Fig. 2 are plotted for successive s t a g e s of oil withdrawal a s shown in Fig. 14, and curves such a s those shown at B in Fig. 6 are superimposed, an estiniate can be made of flowing life. A procedure for this purpose was developed by h. J. hoodward.* In Fig. 14, prepared by hlr. hoodward, the IPR's (shown a s straight lines on the left side of the figure) and the gas-liquid ratio against cumulative withdrawals (shown a t A in the figure) provide the essential well data. A number of curves for the Intake pressures necessary to sustaln lift with 200 psi tubing pressure for a s e r i e s of gas-liquid ratios *Shell 011 Company

maximum tubing pressure; and for smaller beans, large difl'erential pressures are available to prevent sustained increases of production rate, but the differential available to prevent decreases of production rate becomes s n ~ a l l e ra s the bean s l z e i s reduced. For a 6/,-in. bean, the tubing and bean perforniance curves nearly coincide and reductions in rate of flow cannot be prevented. If thls s i z e of bean were applied to the well represented, the well would "head up and die." Flow rates, below the practical minimum for one tublng s i z e , can only be maintained by use of smaller tubing, or by flow~ngthe well intermittently for short periods a t relatively high
-

2 875" TUBING AT 8000'

200

PS I

TUBING PRESSURE

Fig. 14-Prediction of Flowing L i f e

136

W. E. GILBERT

Headrng Cycle: 1. Because of the bypassing of gas Into the annulus, the llquid level IS slowly belng lowered, annulus o i l IS belng displaced Into the tub~ng. 2. The w e l l 1s st111 producing at a low rate and the tublng column IS heavy because gas 1s beingdlverted into annulus and o i l from the annulus is belng dlverted Into the tubing. 3. The welght of the tubing column IS belng reduced because no further gas can be stored In the annulus. Th~s further reduces the Intake pressure, and 4. Allows gas from the annulus t o "blow around." For a short tlme the well 1s gas-lifted w ~ t h annulus gas a t a hlgh rate, and thls reduces the Intake pressure t o ~ t s lowest value. The w e l l gas is NOT belng used t o best advantage. 5. The extra annulus gas has been dissipated and fluid, because of the low Intake pressure, IS flowlng at a high rate Into both the tublng and the annulus. The tubing column IS becomlng heavler and the outflow rate IS d i m i n ~ s h i n ~ . 6. F l u l d IS st111 flowlng Into the well at a faster rate than ~t will flow out of the w e l l w i t h the exlsting Intake pressure. 7. The rate of outflow IS again In balance wlth the rate of inflow, and bypassing of gas into the annulus starts repetltlon of the cycle. Observatrons. T h l s IS not an efficient type of flow because i t produces a large proportion of the o i l wlth a deficient supply of gas and a small proportion of the o i l w ~ t h excess supply of gas. an type of surging is not effiThe usual practice of beanlng a well back t o a low rate of flow t o avoid t h ~ s c ~ e n t ,and In many Instances results in avoidable reduction of Income. T h l s type of flow IS c h a r a c G r ~ s t ~ c the latter part of the flowlng l ~ f e wells i n most areas, but i s also of of characterlst~cof many relatively new low g a s - l ~ q u ~ d ratlo wells.

Heading actlon of t h l s klnd can be mlnirnlzed by use of tubing-casing packers, but where packers are not already ~nstalled,use of casing-actuated intermitters may be preferable.
Fig. 15-Unsteady Natural Flow

Headrng Cycle: 1. A t the start of the flowlng perlod, the tublng IS opened by the r i s l n g caslng pressure whlch actuates the motor valve. The column of gas whlch has collected I n the upper part of the tubing i s produced, and the consequent r e d u c t ~ o n pressure ensures flow of the fluld mixture In the tubing below t h l s gas colof umn. 2. The tubing pressure trends downward whlle fluld IS being displaced out of the annulus, and then 3. Rlses as annulus gas starts t o break around the foot of the tubing. 4. When the caslng pressure reaches the predetermined mlnlmum, the motor valve closes the tubing outlet, flow lng Into the but flow Into the we1 l continues w l t h very l l t t l e decrease I n rate, both gas and l ~ q u l d annulus. Tublng pressures contlnue t o r i s e . The caslng pressure, whlch IS d ~ r e c t l yrelated t o the amount of gas stored I n the annulus, a l s o Increases I n response t o gas and llquld enterlng the well. When the casing pressure reaches the predetermined maxlmum, r e p e t ~ t i o n of the c y c l e IS started by opening of the motor valve. Observatrons: B y regularizing flow, caslng-actuated lntermitters can be used t o Increase the rate of flow and extend the flowlng I l f e of w e l l s when they reach the heading stage. lntermltters are not t o be recommended for w e l l s whlch will produce more on the pump, but they are recommendable: a. T o Increase the rate of flow of w e l l s whlch have been beaned back t o avold heading. b. T o regularize and increase t h e flow of new low-ratlo w e l l s u n t l l Increase of gas r a t l o s may permlt steady flow a t deslred rates. lntermltters have been used t o flow w e l l s whlch would not flow wrthout an ~ntermltter, wells w ~ t h water cuts exceeding 50 percent and w e l l s at rates exceeding 500 bbl per day. lntermitters have been misapplied, and mostly by inexpert selection of bean slze and casing-pressure ranges. The bean should preferably be as large as necessary t o ensure a continuous drop of casing pressure during the on-perlod (1 t o end of 3) and n o larger. T h e casing-pressure range must be large enough t o ensure a flow of gas around the foot of the tublng, s l g n a l ~ z e d a steady buildup In tublng pressure. A longer range IS unnecessary and by often undes lrable.

Fig. 16-Intermitter Control of Natural F l o w

138

W. E. GILBERT
Heading of the second type (sometimes called annulus headingw) occurs when I, bubbles of free enough to escape entraing a s at the intake are b ~ g the ment ~ i t h liquid entering the intake, and 2, the gas-liquid ratio i s materially s n ~ a l l e rthan the gaslift optimum for the average producing rate of the well. Fig. 15 provides diagrams illustrating heading of this type, and Eig. 16 illustrates the control function of an intermitter in this connection. T h i s type of intermitting does produce oil from the well head by "jerks," but it regularizes inflow, and with suitable adjustment, it reduces the range of velocities through the liner screen a s compared with unregulated heading. A normally closed motor valve is preferable for operating a well at a low p e r centage of i t s full flowing capacity and a normally operopen valve should be used for n~axin~um-rate ation. Casingtubing packers obviate annulus heading i f installed at the intake, but do not serve the function of an intermitter in regularizing production a t rates below the minimum stable-flow rate; and i f not installed when the well is completed, the danger of damaging the well by killing it with mud or water to install a packer may make alternative use of an intermitter more attractive. Incidentally, the only functlon of a packer in this case IS to gulde bubbles into the tubing, which fact may suggest new forms which are equally eflective and still permlt ready means of circulating to kill the well, if necessary, without moving the tubing.
66

were superimposed over the IPR lines and the well's flowing progress was interpolated using the gasliquid ratio data given. Rhen the well's progress curve becomes tangent to the IPR, flow must cease. By constructing sinlilar graphs for other tubing s i z e s , it was concluded that this well would flow 120,000 bbl in 415 days with use of 3.5-in. tubing; 135,000 bbl i n 505 days with 2.875-in. tubing; and 265,000 bbl in 1,370 days with 2.375-111. tubing. Any serious errors in tubingsize selection can be avoided by applications of this procedure. In connection with Howing life, it may be menhoned that observed rates of decline during natural flow are usually unreliable a s indices of the decline of well inflow capacities. Consequently, misapplications can result from assuming that flowing decline i s representative of the decline to be expected on mechanical Ilft. T h e reason for this l i e s in the fact that flowing i s a high-rate lifting method and requires greater lifting pressures as rates are reduced. Examples can easily be drawn from practice. Iiowever, in the hypothetical case of Fig. 14, it will be seen that the well starts out flowing about 57 p e r cent (625~100/1,100) of its initial maximum-rate capacity, and at the end of i t s flowing life, it i s only producing about 14 percent (75 x 100/540) of i t s residual maximum-rate capacity.

UNSTEADY FLOW
A working knowledge of unsteady flow i s a necessary tool in maintaining desired production rates and in avoiding unnecessary stoppages, particularly in the latter s t a g e s of flowing life. There are two principal sources of unsteady flow: 1, segregation of free g a s from liquid in the rising fluid column, and 2, segregation of free g a s from liquid at the tubing intake. Fornlatlon heading, which is evident in L a P a z F'ield, Venezuela, n ~ a y excluded from usual conbe sideration inasmuch a s it cannot occur unless the well i s tapping a fissured or cavernous reservoir. Heading of the first type i s observable even in settled pumping wells operating wlth low-liquid volumetric-pump efficiencies and i s a relatively unimportant phenomenon. It causes relatively small, and often irregular, pressure changes of short cyclical duration and has little effect on the continuity of production except in very weak flowing wells. It may accentuate unsteadiness of the second type. Also, its presence makes tubing pressures inferior to casing pressures both a s indices of operation, and a s means of flow control.

CASING AND TUBING PRESSURES

hhen g a s bubbles are large enough to escape entrainment with liquid entering the tubing, the annulus fills with gas and the casing pressure becomes a sensitive indicator of flowing performance. An empirical formula for estimating intake pressures from casing pressures i s given in Fig. 17. Bubble s i z e s a t the intake, of course, cannot be measured directly. However, an engineer armed with simultaneous measurements of casing and intake pressures can easily determine whether or not the casingpressures of the wells in llis area are useful in estimating intake pressures. Gas g a d i e n t s are most likely to exist when gas-liquid ratios are on the high side and productivit i e s are small. Two-phase g a d i e n t s are helpful in this connection, but are often l e s s accurate than casing-pressure data because they depend upon gasliquid ratios which are l e s s reliable than pressure measurements.

FLOWING AND GAS-LIFT WELL P E R F O R M A N C E

139

= TUBING

IN P S l A DEPTH I N THOUSANDS

1 20

110

and there is much to be said In favor of accurate reporting and control of bean sizes. Range-indicating pressure gages are beginning to be usefully applied, the thought being that flowing wells do not require gaging or other individual attention except when a change occurs i n the normal range of variation of thelr tubing and casing pressures. Hange readings are often conclusive a s ind i c e s for operating control, and spot readings are not conclusive.

RESTARTING NATURAL FLOW

1.oo 0

DEPTH I N THOUSANDS O F FEET

Fig. 17-Approximate Gas-gradient Relationship Between Casing Pressure a t the Well Head and Intake Pressure at Lower End of the Tubing

In many fields the IFH of a well may be estimated from knowledge of i t s static pressure plus a representative casing pressure and the corresponding liquid-production rate, a s indicated in Fig. 18. For heading wells, the average casing pressure may be used. h s t i m a t e s of this type are in error on the conservative side ~f the annulus contains liquid (~.e., the true IPR will be larger than the estimated IPH). Thus, if casing pressures and static pressures have been carefully measured and recorded, they can often be used to outline IPH trends in terms of withdrawals. Generally, tubing and casing pressures merit attention both from engineers and operating personnel,
GIVEN AVG CSG P = 1709 P S I A 250 BID R T STATIC PR = M O O D31 GLR = 0 7 M C F I B TUBING. 2 375 INCH S T A 8000 FT E T WELL DEPTH ezw n ESTIMATE THE I P R PROCEDURE 1 FROM FIG 17 P = 1 226 P q e = 1 226 (1709 P* = 2115 PSlA
VI
0.

+ 15)

2000

2 CONNECT INTAKE P = 2100 PS l G R TO STATIC P MAKING I P R ESTIMATE R

TUBING INTAKE) 200 400 600 800

PRODUCTION-RATE

IN B ID

Fig. 18-Method of Estimating IPR's from the Static Pressure and a Casing Pressure

Flowing wells are not improved by periods of shutdown; and when they die, they are candidates for immediate attention. Even though flowing i s no longer the optlmum method, it i s sometimes desirable to restart a well a s a protective measure until finallift equipment i s ready for operation. Strong high-pressure high-productivity wells restart when the flow valve is opened; and a t the other end of the s c a l e , there are weak wells which can only be restarted by a procedure tailored to the requirements of the individual well and applied without unnecessary l o s s of time. In connection with restarting, the following facts are often significant: 1. The gas-llquid ratio typical of the well may require two or three days of operation for full development after restarting because, under shutdown conditions, liquid produced into the well bore frequently invades the more active gas-producing layers and temporarily reduces effective g a s p e r meabilities. 2. For wells without tubing-casing packers, i t is usually necessary to fill the annulus with gas at the desired intake pressure before stable flow can be re-established. 3. If the well d i e s a s a result of heading action, delay may diminish the possibility of successful restarting. Also, in critical c a s e s , use of smaller tubing or u s e of a casing pressure-actuated i n t e r mitter may be necessary if i t is desired to prolong the well's flowing life. Swabbing i s definitely inferior to use of a g a s compressor for difficult restarting jobs because, with swabbing, the well ordinarily restarts with a considerable column of liquid still remaining in the annulus; and by the time the production crew h a s moved to another location, bypassing of g a s p a s t the intake loads the tubing with a heavy mixture from the annulus and the well dies. Incidentally, a s a well declines, reduction of the bean s i z e i s necessary to maintain stable flow. However, changes

140

W. E. GILBERT

must be made in small increments at weak wells because for each increase I n Intake pressure an easil y computed extra quantity of g a s must be stored in the annulus before the well's full gas-l~quidratio can be effectlve In the tubing for lifting at the lower rate. Ignorance of this fact and lack of a readily available starting compressor result in premature relinquishment of natural flow In many cases.

GAS-LIFT APPLlCATIONS
It i s still true that no one method.of lift will provide optimunl results in all wells, and due process of rating wells and methods will always be d e s i r able to niainta~n profit margins e s s e n t ~ a in meetthe l ing demands for oil.
100

200

400

600

loo0

PRODUCTION IN B I D

p $ I

Fig. 20-Use of Log-log Paper for Extrapolations of the Intake Pressures and Gas-liquid Ratios of Maximum-rate Gas L i f t

TOTAL G L R

PERFORMANCE FROM 5000 F T WlTH 50 PSI TUBING PRESSURE

R T WITH AE 2 875'

(ii)

\I

PRODUCTION IN B I D

pressure, it i s only necessary to plot the intake pressure for each of a s e r i e s of rates. In each c a s e , start on the optimum gradient (the one gradient in each g a p h marked with an arrow in Fig. 23-28, 3033J at the given tubing pressure, measure down from this point the depth of the well, and read the Intake pressure. A s shown in E'lg. 19, the resulting curve, A, outlines the highest-rate g a s lift possible w ~ t h the given tubing s i z e , tubing pressure, and depth; the and the intersect~onof this curve w ~ t h IPH gives rate and lowest intake pressure attainthe h ~ g h e s t able under the given c o n d i t ~ o n sin the particular well. By plotting the gas-liquid ratio for each gradient, an estimate, B, of the required total gas-liquid ratlo i s obtained from which the well's gas-liqu~d

Fig. 19-Illustration of the Procedure Used for Maximum-rate Gas-lift Estimates

Gas 11ft i s primarily a h~gh-ratemethod and can be a final-lift method in wells tapping strong waterd r ~ v e reservoirs, as illustrated by experience in some Louisiana fields. It can have useful applicasmaller than tions when allotted rates are n~aterially inflow capacities. Use of g a s a s a means of liquid 11ft 1s always attended by impos~tionof some back . pressure against the formation, even when packers are used, and even when standing valves and concentric or eccentric induction tubes are used. but i t must also be observed that formation back pressures are often unavoidable with the other forms of lift and especially when g a s accompanies the oil produced. For this reason g a s lift can be expected to find applications where the bore of the well is s o small a s to prevent effective gas-anchor a c ~ i o n .It i s also applicable when depth limits the relative capacities of other methods. T o estimate the maximum-rate gas-lift possibilit i e s of a well with a given IPR at a given tubing

PRESSURE H PS I

GLR IN M C F l B
1 -

e m
z

Ya

7-

g g"
Z

lm

STIUNCSOF sun I N W ~

Fig. 21-Diagram Illustrating Some Flow and Gas-lift Alternatives (Gas-lift t u b ~ n g pressure =

100 psi)

ratio i s subtracted to obtain the net gas-liquid ratio needed. Use of a higher or lower total gas-l~quld ratlo will increase the intake pressure and reduce the liquld-production rate. Both maxlmum-rate curves, ratlo such a s A in Fig. 19, and optimum gas-l~quid curves such a s the one shown in the figure, may be considered to be straight llnes on log-log paper when extrapolations are necessary, as shown in E'lg. 20. T h i s observation does not hold for other g r a d ~ e n t sand their ratios; and, of course, any extrapolation of enlpirical data must be made advisedly. Flg. 21 illustrates some of the many gas-lift and flow alternatives adaptable to a given well, and i s introduced a s an illustration of the type of graph~cal ' analysis which may be used. Such a well with a full string of 2.875-in. tubing would be nearlng the end of i t s flowing life, a s indicated by the smaller of the two tubing-pressure curves in graph C. It would flow strongly with use of a full string of 1.66-in. tubing using a compressor for starting.Its maximumrate gas-lift performance with a full string of each of three s i z e s i s also shown in g a p h C and correspending gas requirements are shown in graph U. The input casing pressure for gas 11ft from 5,000 ft would be between the intake pressure and the static gasgradient pressure shown. The well's maximum-rate gas-lift perfornlance with a 1,000-, 2,000-, 3,000-, or 4,000-ft string of 1.66-in. tubing inside 2.875-in. tubing i s 'shown in graph B. Graph A shows 2.875in. gradients for 0.4 Mcf per bbl.

Solution of well problems uslng combination strlngs can be made by adding gradients, using the same ratio for natural Bow or gas lift through the different s i z e s , and using the same pressure at the depth at which a juncture between s i z e s takes place. Any adequate discussion of flow valves would be beyond the intended scope of this paper. Flow valves have contributed greatly to the practical utility of gas lift. They are indispensable for starting against the high well pressures and for automatic restarting. Intermitter types with standing valves have been rates of flow than those used for maintaining h ~ g h e r p o s s ~ b l ewith straight gas lift In particular wells. The predictability of gas lift with flow valves i s not solely the responsibility of the valve manufacturer, but depends q e a t l y upon the deference the operator has pald to determining well data of the type indicated in Fig. 2. However, a s better gradient data become available, improvements may be made in valve spacing, and both installation costs and the proportion of total horsepower needed for startlng can be reduced. Incidentally, ~t seems likely that, with use of improved gradient data, a demand will develop for accurate determination of pressure l o s s e s through flow valves of the different types; and the exemplary data provided by one manufacturer for flow of gas through surface beans do not necessarily apply without modification to other types of orifices used in a tubing string. In order to compare g a s lift with other methods of lift In particular c a s e s , the oil-country engineer

T a k ~ n gk = 7.25 for wet gas, t h ~ s formula may be written:

The gas horsepower required by the gas-11ft user is:

wherern. M = Mcf per day at 14.7 PSI. It 1 here sugs gested that thls 1 the horsepower upon which quos tatlons should be based w ~ t h the glven 7, pressure ratio; 2, Input gas temperature; and 3, Input pressure, and may be from 20 to 40 percent less than manufacturer's brake-horsepower ratings depending upon the dev~ation from Boyle's law, the auxiliaries used, and the overall plant effic~ency atta~ned.

Fig. 22-Chart Suggested for Estimating Gas-horsepower Requirements for Gas Lifting

142

W. E. GILBERT
he oil industry IS accustomed to thinking of conservation in terms of barrels of liquids or material recovered, yet often fails to realize that the conserhe vation of energy i s equally important. 1 conservation of energy leads to a conservation of petroleum and increased recovery. 1herefore, the conservation of nature's energy in producing oil becomes as important a s any other phase of conservation or economies. I h e author briefly mentions the use of the "inflow performance relationship," or IPH, which corresponds to the well-knonn "productivity index" or PI. In actual practice, the mass rate of inflow of s s each phase i s usually replaced by the PI. Th'1 1 a good approximat~onwhere the water cut is not large ( l e s s than 25 percent) or the gas-oil ratio i s not excessive. What constitutes an excessive ratio for calculations in which P I i s substituted depends upon temperature, pressure, and fluid characteristics. Certainly, P I i s meaningless for calculations involving retrograde condensate wells which make significant volumes of water. Cne question which occurred to nle is: Ahat effect has viscosity on the relationships developed7 It i s realized the Ten S e c t ~ o nField crude has low viscosity, in the range of 0.5 to 5 centlpolses within the flow string, depending upon temperature and composition. I should like to ask the author ~f he has any data showing the effect of viscosity on the charts prepared. \hat effect does viscosity have on the equation for the bean s i ~ e - p r e s s u r e ;elation? hould the rapid change of slope in the viscositypressure curve at lower pressures explain the sensitivity of flow volume to changes in bean s i z e at these pressures? Perhaps the onussion of a viscosity function in the bean sue-pressure equation would explain the author's point that the equation cannot be used at pressures l e s s than 70 percent greater than l ~ n e pressures. It w ~ l lbe r e c o g n i ~ e d that this i s the pressure range where bean s i ~ ies most critical-a slight error often results in killing the well. l'he author s t a t e s ". . rates of decline during natural flow are usually unreliable a s indices of the decline of well inflow capacities." l h i s difTiculty nlay be overcome by observing the decline in potential or productive capacity with time. Students of this will agree with the author i that t h ~ s s "a phase of production operations deserving wider attention than it has been accorded In the past." l h l s paper represents; a pra~seworthy cqntribution to the industry.

stands in need of a ready means oi estimating horsepower requirements. P o s s ~ b a ~formula of the type l given In Fig. 22 can be adapted to this purpose. Although n~anufacturersmay regard this suggestion a s an over-simplification, they also may be Inclined to agree that some simplificat~on would serve to promote conlpressor s a l e s for gas lift In competition ~ith other lifting means. 1he temperature of power gas, because of ~ t low specific heat, i s usually s controlled by well temperatures at points of application, and usually the gas-rate requ~rementi s estimated at standard conditions.
~

CONCLUSION

1he material presented here is ofleredas an i n t e r

in) report on a phase of production operations deserving uider attention than i t has been accorded in the past. It i s desired to thank the Shell Oil Company for making the presentation possible, and particularly since this outline was con~pletedduring the w r ~ t e r ' sstay in their California area.

REFERENCES
'Versluys, J. klathematlcal Development of the Theory of Flowing b e l l s , Trans Am. Znst. Mtntng Met Engrs. (Petroleum Developnlent and Technology) 192 (1930). kloore, T . V. and Wllde, H. D: Experimental Measurements of Shppage In Flow through Vertical P l p e s , Trans Anr. Znst Mrntng Met Engrs. (Petroleum Development and Technology) 296 (193 1). Kemler, Emory, and Poole, G. A. A Prellmlnary Investlgatlon of Flowlng PJells,Dnllrng and Productton Practrce, 140 (1936). May, C. J and Laird, A. The Efficiency of Flowlng h e l l s , J . Znst. P e t Tech., 20, 214 (1934). k a y , C. J Efficiency of F'lowlng 'Aells, Trans Am Znst Mrnrng Met Engrs (Petroleun~Developnlent and Technology) 114, 99 (1935). Poettmann, Fred H. and Carpenter, Paul G. The blultiphase Flow of Gas, 011, and Uater through Vertlcal Flow S t r ~ n g swlth Appllcatlon t o the Deslgn of Gas-lift Installatlons,Drtllrng and Productton Practrce, 257 (1952). Shaw, S. F Gas-lrft Prrncrples and Practrces, Gulf Publishing Company, Houston, T e x a s , 1939. 'Babson, E. C T h e Range of Appllcatlon of Gas-llft Methods, Drtllrnk and Productron Practrce, 266 (1939). Sullivan, R. J. Gas-011 Ratlo Control In Flowing R'ells, Drrlltnh and Productron Practtce, 103 (1937).

'

DISCUSSION

'A. C. Sheldon ( 1 he Ohio Oil Company, L o s Ange. . a very useful les)(written): 1 he author has tool for the solution of problems relative to flow of 011, gas, and water in a tubing string. In s o doing, he has assembled many pertinent facts, a s well a s helpful graphs. l ' h e s e should prove especially helpful to the engineer who i s confronted with the problem of design for a tubing string in a flowing orgaslift well.

Rlr. Gilbert: T h e I P R and the P I a r e not equivalents. I h e IPR, a s u s e d in t h i s paper, i s t h e re. . lationship between intake pressure and liquid Inflow rate. T h e P I i s the first differential of t h e I P R in the s p e c i a l c a s e when t h e latter is a straight line, or i s s o nearly straight t h a t i t s curvature may b e ignored. Very roughly, depth-pressure gradients some 15 percent heavier than t h o s e In t h e paper were found n e c e s s a r y for gas-011 r a t i o s up t o 1 Mcf per bbl in one field; but in t h i s c a s e the oil v i s c o s i t y i s a t l e a s t 1 0 0 times greater than that of T e n Section crude. I t i s believed that v i s c o s i t y e f f e c t s on flow through b e a n s will prove very minor b e c a u s e of t h e e x i s t e n c e of i n t e n s e turbulence. T h e r e i s a wellknown formula, available in a n y standard summary of flow of g a s e s through nozzles, which may b e shown a s follows:

a n d a l s o provide means of calculating p r e s s u r e s and volumes of g a s n e c e s s a r y for optimum gas-lift operation. T h e author points out that t h e gradient curves should b e u s e d with care. T h e s e a r e apparently t h e r e s u l t of t h e correlation of d a t a from many t e s t s In flowing and gas-lift w e l l s modified and extended with information of other authors. k h e t h e r they a r e completely reliable or not, they certainly provide a p i d e as t o t h e magnitude of p r e s s u r e s and volumes of g a s required t o lift oil. T h e y further provide a convenient pattern t o a n operator or engineer for accumulating and presenting the r e s u l t s of h i s own well t e s t s . I have s e v e r a l q u e s t i o n s t h a t I should like t o a s k Mr. Gilbert.

1. In F i g . 2-Individual Well Inflow Performance Trend, w a s t h e productivity index omitted intentionally?

-

k being t h e a d i a b a t i c constant, and p l / p o , the minimum ratlo for which flow through n o z a l e s i s unaffecte d by po, t h e downstream pressure. If k e q u a l s 1.25, the critical ratio i s about 1.8, compared with 1.7 u s e d in the report. T h u s , s u g g e s t e d u s e of upstream p r e s s u r e s exceeding the downstream p r e s s u r e s by a t l e a s t 7 0 percent i s b a s e d upon observation of field d a t a and rough consideration ot critical r a t i o s (values of k for oil-laden g a s were not determined). T h e d e c l i n e of t h e production r a t e of a flowing well is greater than the decline of t h e well's inflow capacity, and t h e disparity i n c r e a s e s as flow continues. A s Mr. Sheldon points out, i t i s preferable t o follow t h e well's inflow capacity, which may b e plotted in terms of time, or, better s t i l l , in terms of oil withdrawals, as indicated by c u r v e s 1 a n d 2 of Fig. 2.

2. It i s s t a t e d that flowing w e l l s a r e not improved by periods of shutdown. Where restarting is not a problem and the well i s clean, d o you believe some damage may r e s u l t ?

3. In t h i s study, the location of t h e tublng intake with r e s p e c t t o the perforations must have had t o b e considered many times. Was there a n y general conclusion indicated as to t h e optimum tubing location, i.e., a t the bottom of the perforations, top of the perforations, or a t some point above? I h e author i s t o b e commended for a n excellent paper. I belleve h e h a s had niuch of t h i s d a t a for some time, and I am g l a d h i s company provided the time for him t o a s s e m b l e it.
M. Gilbert: Answering Mr. Acres' questions:

T. H. A c r e s (Sunray Oil Corp., L o s Angeles) (written): l h i s paper is a n e x c e l l e n t a n a l y s i s and presentation of t h e f a c t o r s affecting t h e behavior of flowing and gas-lift wells. I t i s t h e first treatment I have s e e n that r e c o g n i z e s n o distinction between g a s lift and natural flow; and, b y s o doing, i t ena b l e s one t o better understand why natural Bow c e a s e s and how, by g a s lifting, t h e energy n e c e s s a r y for the well t o flow is augmented. T h e g a p h i c a l methods and gradient d a t a presented provide means of predicting when the flowing life will end

1. Yes, t h e P I w a s intentionally omitted from Fig. 2 in recognition of the f a c t t h a t IPR's frequently a r e curved. However, some engineers may prefer t o plot t h e PI i n s t e a d of t h e maximum inflow r a t e (curve 2) if t h e well's I P R may b e taken t o b e a straight line. 2. Yes, there a r e some p o s s i b i l i t i e s that damage may result in shutting down e v e n a c l e a n well which may b e r e s t a r t e d e a s i l y . T h e dangers for short periods of shutdown a r e u s u a l l y negligible. However, incipient c a s i n g l e a k s may become active under shutdown conditions ( l e a k s , incidentally, which were s e a l e d off when t h e c a s i n g w a s t e s t e d with mud in t h e well). Also, under shutdown conditions interflow will occur i n long perforated

144

W. E. GILBERT
should contribute to further development in t h i s long-neglected area. Mr. Gilbert: I wlsh t o thank Dr. Kemler for h ~ s kind comments. If natural flow is in a s e n s e t h e "orphan child" among lifting methods, there may b e some significance in t h e f a c t that flowing ret q u i r e s very little s p e c i a l e q u ~ p m e n and consequentl y h a s lacked t h e extra stimulus of englneering attention accorded by t h e supply industry t o other methods of lift. Diminishment of well I P R ' s , and change of g a s liquid r a t i o s in terms of withdrawal, a r e f a c t o r s tending to limit the practical utility of tapered strings. However, Insertion of a length of smaller l tubing i n s i d e 3.5- or 2.875-in. tublng w ~ l frequentl y s e r v e to extend natural-flow production of a n allotted rate. T h e methods of Poettmann and Carpenter a r e proving valuable in developing lmproved depth-press u r e d a t a in readily u s a b l e form, although some difficulty h a s been experienced with e x t e n s i o n s into the pressure range below s a y 500-400 psi. Also, the a n a l y s i s of two-phase flow through b e a n s (chokes) given in the paper d o u b t l e s s c a n be improved. In the i n t e r e s t s of t h e oil Industry, it i s t o b e hoped that reference material concerning the dynami c s and economics of natural flow will b e considera b l y augmented over t h e next decade.

lengths, with some degradation of ~ o t e n t i a lrecovery. T h e s e a r e f a c t o r s b e s t judged under l o c a l conditions.

3. In general, whenever tubing-flow gradients a r e lighter than g a d i e n t s in the liner and c a s i n g for the same r a t e s , i t is preferable t o s e t t h e tubing a s low a s i s c o n s i s t e n t with safety, b e c a u s e the w e l l will flow longer and a t higher r a t e s than would b e t h e obtainable with a shallower setting. However, in clean w e l l s with undersaturated crude there IS no s p e c i a l point in s e t t l n g t h e tubing below the depth where a l l the g a s remains in solution over the full range of desired operating rates.
I t is d e s i r e d t o thank Mr. Sheldon and Mr. A c r e s for their comments.

E.N. Kemler (University of Minnesota, Minneapolis) (written):* Flowing-well performance h a s not been given t h e attention which i t s importance d e s e r v e s . 1 h i s 1 indicated In part by the very meager biblios graphy included in connection with hlr. Gilbert's paper. Development of better understanding of reservoir performance, together with more r e c e n t b a s i c investigations on multiphase flow should make i t p o s s l b l e t o arrive a t a more rational approach t o t h e s t u d y of flowing wells. T h e low c o s t of production by flowing, together with the a d v a n t a g e s of postponing artificial lift a s long a s p o s s i b l e t o permit more a c c u r a t e evaluation of requirements and capacity of pumping equipment a s well as postponement of the accompanying investment, would a l l contribute to making s t u d ~ e swhich would prolong flowing life most desirable. Mr. Gilbert's paper i s a n outstanding contribution to t h l s field and should l e a d to further s t u d i e s relating t o a rational approach t o the study of flowing wells. Tubing and choke ~ n s t a l l a t i o n sIn flowing w e l l s a r e deserving with of t h e s a m e c o n s ~ d e r a t ~ o n r e s p e c t t o d e s i g n a s that given t o design of sucker-rod strings. It should b e expected t h a t u s e of tapered tubing s t r i n g s , for example, would give better flow conditions than a uniform string.
T h i s paper r e p r e s e n t s a n outstanding contribution t o the literature of flowing-well performance. It
*Prepared following presentation of the paper.

E.C. Babson (Union Oil Co. of C a l ~ f o r n i a , C a l g a r y , Alberta, Canada)(written):* A s one who b a s had a l i t t l e experience wlth the empirical approach t o gas11ft problems, I am impressed by t h e lnvestigatlon which i s summarized in hlr. Gilbert's paper. T h ~ s paper c a r r i e s t h e e m p ~ r i c a l n a l y s i s of g a s l i f t pera formance far beyond any prevlous work, and I believe i t may well prove t o b e t h e definitive work on the s s u b j e c t . It 1 probable that further refinements and e x t e n s i o n s of correlations c a n b e made, but t h e paper c o v e r s almost completely t h e p r ~ n c l p l e sinvolve d In applying t h e s e d a t a t o p r a c t i c a l production problems h%. Gilbert: I value highly both Mr. 13abson's opinion of the paper and the part h ~ earlier work s played in making it possible.

FLOWING AND G A S L I F T WELL PERFORMANCE

145

Fig. 23-Approximate Depth-pressure Gradients for 2.875-in. Tubing

W. E. GILBERT
GRADIENT PRESSURE IN PSI

Fig. 24-Approximate Depth-pressure Gradients for 2.875-in. Tubing

FLOWING AND G A S L I ~ T ' WELL P E R F O R M A N C E

GRADIENT PRESSURE I N P S I

Fig. 25-Approximate Depth-Pressure Gradients for 2.875-in. Tubing

W. E. GILBERT

Fig. 26-Approximate Depth-pressure Gradients for 2.875-in. Tubing

FLOWING AND G A S L I F T WELL P E R F O R M A N C E

GRADIENT PRESSURE IN PSI

Fig. 27-Approximate Depth-pressure Gradients for 2.875-in. Tubing

FLOWING AND GASLIFT

WELL PERFORMANCE

USE O F BEAN-PERFORMANCE CHART A g a p h i c a l means of e s t ~ m a t i n g the bean performance is given in F i g . 29. F o u r v a r i a b l e s are considered, a n d when three of t h e s e v a r i a b l e s are known, the fourth may b e estimated. In construction, P a r t I of the chart r e p r e s e n t s the performance b e a n and P a r t I1 i s simply a means of of a ''/,-in. correcting r e s u l t s for other bean s i ~ e s .In solving for a n y one of t h e four variables, enter P a r t I if both the barrels per d a y and t h e g a s l i q u i d ratlo a r e known; or enter P a r t I1 of t h e chart if t h e tubing pressure and bean s i z e a r e known. I ' h e further procedure in solving for e a c h variable may b e b e s t explalned by examples.

Solutzon: Enter P a r t I1 a t t h e intersection of the 40-bean and 750-psi l i n e s and g o vertically to t h e 1 0 line; then g o horitontally to 1.4 hlcf per bblthe estimated production rate being found t o b e 1,530 bbl per day.

3. E s t i m a t e Tubing P r e s s u r e
E s t a b l i s h t h e performance of a 13 bean for 0.8 hlcf per bbl. Solutzon: Slnce the formula i n d i c a t e s that there i s a straight-line relationship between tubing press u r e and barrels per day, only one point need be established. T h u s , going h o r i ~ o n t a l l y from the intersection in P a r t I of the 0.8 ratio and, s a y , 200 bbl per day, t o the 10-bean line in P a r t I1 and thence vertically to the 13-bean line, the pressure 1 found t o b e 605 p s l and for any production rate s with the 1 3 bean and 0.8 ratlo, the tubing pressure i n p s i is equal t o the barrels per d a y multiplied by 605/200 or 3.03. 4. E s t i m a t e B e a n S i z e A well h a s been operating for a n extended period a t 200 bbl per d a y and 4.0 Mcf per bbl ratio. It is desired to reduce the rate t o 1 0 0 bbl per d a y and the tubing outlet performance curve Indicates t h a t t h e tubing pressure a t t h i s rate will be 1,800 psi. What s i z e bean should be u s e d ?

1. E s t i m a t e Gas-liquid R a t i o or t h e G a s R a t e
Suppose we have a well equipped with a '%,-in. bean operating with a tubing p r e s s u r e of 7 2 0 p s i a t 200 bL1 per day. No differential recording meter h a s been i n s t a l l e d a n d a n e s t i m a t e of the g a s production i s desired. Solutzon. F i r s t find in P a r t I1 of the chart, the intersection of the line for the 15 bean a t 7 2 0 psi; g o vertically t o the s l a n t i n g 10-bean line and then horizontally t o t h e v e r t i c a l 200 bbl per day line in P a r t I. At t h i s point the gas-liquid ratlo i s about 1.8 Mcf per bbl, from which t h e total g a s i s 360 Mcf per d a y (200 x 1.8).

2. E s t i m a t e Production R a t e
After a bean change, a new well which showed a ratio of 1,400 c u f t per bbl on t h e prevlous gage, i s flowing through a 4 0 bean with 7 5 0 p s i tubing pressure. What i s the probable production r a t e ?

Sohtzon: Enter P a r t I and g o horizontally from

t h e intersection of the 1 0 0 bbl per d a y and 4.0 r a t i o l i n e s t o the 10-bean line in P a r t I1 and thence 4vertically t o the horizontal 1,800-psi line. An ' , in. bean should b e used.

152

W. E. GILBERT
PRODUCTION IN BID

Fig. 29-Bean-performance Chart

gMPTH SCALE IN THOUSANDS

l l l l l l l l f

OF FE@ WE S ET G DW I ~

DEPTH SCALE I THOUSANDS N

a
A

OF FEET-@ G R A D I ~ T S
U

cm

8 -.

DEPTH SCALE N THOUSANDS OF RET-@

GRADIENTS

DEPTH SCALE IN THOUSANDS OF FEET-@ GRADIENTS

U
a
A

DEPTH SCALE N T H O U S A N M O F FEET-@ GRADIENTS

DEPTH SCALE IN THOUSANDS OF FEET-@ GRADIENTS

U
d

DEPTH SCALE IN T H O U S A N D S O F FEET-@

d d

GRADIEN~S
0

~ " " ' " ' " " " 1 ' " ' 1

"

'

'

"

'

e
0

I %
X

3 3
(0

r!

B
m
() I

:
E (D

?
0

k 3.
() I

:
u
3

4
n
3

Y' -.

W F ' -.
?
a

2
z
cP

<

! . 0

C L a

C
DEPTH SCALE I N THOUSANDS Of FEET-@ GRADIENTS
A A

B : :

" I

VI

DEPTH SCALE H THOUSANDS OF FEET-@ GUADlENTS