A theoretical and experimental
study was performed to
develop a dynamic model for gaspassage
performance of a 1.5-in.,
nitrogen-charged, bellows-operated
gas-lift valve. Performance curves
were obtained by using air for 0.25-
and 0.50-in. ports with flow rates
reaching 2.5 MMscflD. Internal pressures
and temperatures were measured
during flow-performance tests
to develop a dynamic model for both
orifice and throttling flow.
A theoretical and experimental
study was performed to
develop a dynamic model for gaspassage
performance of a 1.5-in.,
nitrogen-charged, bellows-operated
gas-lift valve. Performance curves
were obtained by using air for 0.25-
and 0.50-in. ports with flow rates
reaching 2.5 MMscflD. Internal pressures
and temperatures were measured
during flow-performance tests
to develop a dynamic model for both
orifice and throttling flow.
A theoretical and experimental
study was performed to
develop a dynamic model for gaspassage
performance of a 1.5-in.,
nitrogen-charged, bellows-operated
gas-lift valve. Performance curves
were obtained by using air for 0.25-
and 0.50-in. ports with flow rates
reaching 2.5 MMscflD. Internal pressures
and temperatures were measured
during flow-performance tests
to develop a dynamic model for both
orifice and throttling flow.
perimental study was performed to develop a dynamic model for gas- passage performance of a 1.5-in., nitrogen-charged, bellows-operated gas-lift valve. Performance curves were obtained by using air for 0.25- and 0.50-in. ports with flow rates reaching 2.5 MMscflD. Internal pres- sures and temperatures were meas- ured during flow-performance tests to develop a dynamic model for both orifice and throttling flow. 576 Dynamic Model of Gas- Lift Valve Performance Gokhan Hepguler, SPE, Union Pacific Resources Co., and Zellmlr Schmidt, SPE, R.N. Blais, SPE, and D.R. Doty, SPE, U. of Tulsa Introduction To design an efficient gas-lift installation, the production engineer needs reliable in- formation on the performance of all system components, from the outer boundary of the reservoir to the separator. One critical com- ponent is the gas-lift valve. In a producing system, the gas-lift valve controls the point of entry of compressed gas into the produc- tion string and acts as a pressure regulator while the injection gas is controlled at the surface choke. During the unloading proc- ess, the behavior of gas-lift valves becomes the primary factor for reaching optimum single-point gas injection depth. Injection- pressure-operated valves are the most com- monly used continuous-flow gas-lift valves. They consist of a nitrogen-charged dome and bellows assembly connected to a stem and ball that seat on a port (Fig. 1). The performance curves of injection- pressure-sensitive valves show two distinct flow regions (Fig. 2). In the orifice flow region, at a constant injection pressure, the flow rate increases as downstream pressure decreases during subcritical flow, but even- tually critical flow occurs, where flow rate remains constant despite further decreases in downstream pressure. On the other hand, in the throttling flow region, at a constant injection pressure, the flow rate increases with decreasing downstream pressure until it reaches a maximum and then decreases with decreasing downstream pressure. For a given port size, the occurrence of orifice or throttling flow depends mainly on the relative magnitudes of the nitrogen pressure in the dome and the injection pressure. One way to obtain reliable data in orifice and throttling flow regions is to perform flow-performance tests on the gas-lift valves currently available with the valve treated as a black box and volumetric flow rates reported as a function of valve-setting pa- rameters and the differential pressure across the valve. This data-acquisition method is extremely time-consuming because of the combination of parameters affecting gas- passage performance of a valve. Modeling the valve on physics principles allows a sig- nificant reduction in the number of tests needed to characterize valve performance. Copyright 1993 Society of Petroleum Engineers This study investigates pressure and tem- perature distribution within the valve, inter- nal valve mechanism, and forces acting on internal elements of the valve. The paper ex- plains the nature of the experimental data and results obtained, defines the important parameters that affect valve performance, and provides a model for both orifice and throttling flow regions. Experimental Methods The experimental study was performed with a 1.5-in. nitrogen-charged, injection- pressure-operated valve. The valve was custom-made so that required temperature and pressure measurements were possible; the general structure of the valve, however, was unaltered. Three sets of dynamic flow experiments were conducted with the Tul- sa U. Artificial-Lift Project's (TUALP) test facility, which is capable of simulating downhole conditions (Fig. 3). In the initial set of experiments, the valve flow perform- ance and the pressure and temperature dis- tributions within the valve were determined. Table 1 gives a sample data set. Seven pres- sures and three temperatures were record- ed during steady flow with pressure transducers, dial gauges, and thermocouples (Fig. 1). For the second and third set of ex- periments, a valve stem was manufactured so that it could be connected to an adjustable screw assembly, which replaced the valve's nitrogen dome and bellows. These experi- ments revealed discharge coefficients and an effective force around the ball as functions of stem displacement. A micrometer meas- ured the displacement of the stem from its closed position to within O.OOI in. In effective-load tests, the valve stem incorpo- rated a miniature load cell having an accura- cy of O.5Ibf (Fig. 4). In all experiments, air was the flowing fluid. A separate experimental setup was used for static probe tests where the flow system acted only as the air supply. To validate the data gathered on the instrumented valve in the dynamic model test facility, another group of flow-performance tests was con- ducted in a separate test facility with an un- modified 1.5-in. valve that was the same type and model as the custom-made valve. Extensive data were gathered during the experimental study.! Refs. 1 and 2 give June 1993 JPT TABLE 1-GAS-LiFT VALVE FLOW-PERFORMANCE DATA WITH INTERNAL PRESSURES AND TEMPERATURES RECORDED DURING STEADY FLOW Valve N-15R (instrumented, 0.25-ln. port) p(1=636 psig at 79F Pvc =635 psig at 79F Ptro =685 pslg at 79"F P Inj = 700 pslg Pd Td P1 Pp Data (psig) (oF) (psig) (psig) -- 1 645.2 67.9 708.7 592,0 2 646.1 67.4 710.9 532.4 3 644.6 67.0 704.5 465.0 4 644.3 67.2 705.1 395,8 5 643.2 67.0 705.2 374.8 6 642.3 66.3 706.1 374.8 7 641.5 65.8 705.3 370.3 8 641.9 65.7 704.8 373.2 9 641.6 65.3 705.5 372.7 10 642.3 68.0 705.3 376.8 11 641.3 64.9 706.6 373.5 significant details on the experimental fa- cility. Preliminary Analysis Under operating conditions, when there is no flow through the valve, the forces act- ing on the valve just before it opens are given by3 Pinj (AB-Ap)+PprAp =PdAB' ..... (1) The valve will begin to open as AB Ap Pinj =Pd----Ppr--- ... (2) AB-A p AB-A p The closing pressure of the valve, Pvc' is defined as the pressure at which the valve closes when the injection and production pressures are equal. Test-rack opening pres- sure, Ptro' is the pressure at which the valve opens when the production pressure is atmospheric-Le., Ptro =Pinj in Eq. 1 when Ppr=O-psig gauge. However, the force bal- ance given by Eq. 1 and pressure values ob- tained from it are valid only under static conditions or when the flow rate is ap- proaching zero. Under dynamic flow con- ditions, the gas-lift valve is a variable venturi device with a flow area controlled by both upstream and downstream pressures. To develop a model for gas-lift valve per- formance, the dynamics of the compressi- ble flow is applied at each stage of the analysis. Gas flow through the valve is assumed to be an adiabatic frictionless (isen- tropic) process, and the deviation of the ac- tual performance of the valve from theoretical results is accounted for by a dis- charge coefficient. Analysis and Results for the Orifice Flow Region In the orifice flow region of the gas-lift valve performance, the combined force resulting from the injection pressure and the pressure around the ball on the stem is sufficient to keep the valve fully open. The nitrogen charge pressure does not apply enough JPT June 1993 Tp P3 Ppr T inj q (oF) (psig) (psig) (OF) (Mscf/D) 64.0 634.2 630.0 70,0 556.9 60.0 583.4 580.0 69.0 661.6 57.4 540.0 540.0 69.0 717.5 55.6 487.8 485.0 69.0 756.3 54.5 452.4 455.0 69.0 768.0 53.1 382.3 380.0 70.0 773.0 52.9 321.5 325.0 70,0 767.2 52.6 246.3 255.0 70.0 767.2 52.3 182.4 190.0 70.0 767.2 54.2 155.7 125.0 71.0 766.4 52.0 135.7 90.0 71.0 766.4 downward force to cause the valve stem and ball to restrict the flow around the seat. The flow area is equal to the port area of the valve; therefore, analysis of the nitrogen dome and bellows assembly connected to the stem does not enter into the model for this flow region. The experimental study is performed with a O.25-in. port with a O.3125-in. ball and a O.50-in. port with a O.6250-in. ball. The dynamic model predicts valve performance in three steps: calculation of the flow rate by applying the energy equation to the con- ditions upstream of the valve and at the p0l1; prediction of the pressure gain downstream of the port before the check valve by apply- ing momentum and energy balances and the law of the conservation of mass; and com- putation of the pressure drop through the check valve at high flow rates. Calculation of the Flow Rate. The energy equation is applied to determine the volu- metric flow rate. For our purposes, the sys- tem consists of the mass of air in a control volume bounded by the valve inlet ports and the lower end of the valve seat port. The equation to compute the volumetric flow rate becomes 1 q=I,240.3A t Y r;-;:;;;:- . -. --. ,,1-/34 TjZII'g ................... (3) Eq. 3 provides the theoretical flow rate based on /:"P between the injection and port pressures. As expected, the experimental or actual flow rate is always less than the value given by Eq. 3. Deviation from the theo- retical rate results mainly from the neglect of the frictional losses and of the possible effects of the valve geometry. To account for this deviation, a discharge coefficient, the ratio of the actual and theoretical flow rates, must be included in Eq. 3. This dis- charge coefficient, Cd' is calculated with the experimental flow rate measured by the orifice meter run. "To design an efficient gas-11ft installation, the production engineer needs reliable Information on the performance of all system components, from the outer boundary of the reservoir to the separator." otT--'"TRClGEN DOME P, >----14-- STEM !NLETPORT, Pinj,Tlnj
VALVE Fig- 1-Schematic of an injection- pressure-operated valve showing location of pressure and temperature meas- urements. --ORIFICE FLOW --THROTTLING FLOW CRITICAL FLOW-t-..... SUBCRITICAL FLOW Fig- 2-Gas-lift valve performance curves. 577 7 1)1------' 4 I AIR COMPRESSOR 2 PRESSURE VESSEL 3 HEAT EXCHANGER 4 SAFETY VALVE 5 PRESSURE REGULATOR 6 FLOW CONTROL VALVE 9 7 SURGE TANK 8 ORIFICE FLOW METER 9 DYNAMIC TEST BLOCK 10 EXHAUST AND MUFFLER II STATIC TEST BLOCK Fig. 3-Diagram of gas-lift valve test facility. As Ref. 4 noted, the product of the dis- charge coefficient and the expansion factor, Cd Y, for a given geometry is a function of the acoustic ratio, (t:.plpl)/k, and the Rey- nolds number. However, the Reynolds num- bers for this particular case are about 10 7 and have no effect on determining CdY. The plot of (t:.pIPI)/k vs. CdY indicates a linear relationship for both port sizes. A least-squares linear regression yields C d Y=(0.53t3 -0. 7l)[t:.pl(Pinl)] -0.2413 +0.84 ....................... (4) The final form of the e::juation to calcu- late the volumetric flow rate through the valve is CdY t:.PPinj q=1,240.3A t . 'Y 1-13 4 TinjZinj'Y g ................... (5) Eq. 5 is valid for subsonic and transonic flow. Once critical flow is achieved, t:.p must be the difference between the injection and critical pressures, and there is only one value of Cd for a given port size. Any low- er pressure downstream of the minimum flow area, which is the valve port area in an orifice flow region, has no effect on flow rate. Temperature at the Valve Port. The gas temperature at the valve port must be known so that density can be calculated to deter- mine the Mach number, N Ma' at port con- ditions before the pressure gain downstream of the port can be computed. For each flow rate, a temperature value was measured with a thermocouple probe. The measured values, however, do not reflect the actual static temperature of the air flowing through the port. Hottel and Kalitinsky5 state that, when an air stream is partially stopped by an inserted temperature probe, the temper- ature increase caused by conversion of ki- netic energy affects the probe reading. The ideal static temperature can be observed only if the measuring instrument is moving together with the gas. The relation between the stagnation and true temperatures of a flowing gas is given by Ts-Tt=v2/2gc]cp' ............. (6) The static temperature of the flowing gas at the port must be determined theoretically with the isentropic-process assumption and the definition of density as ( Pp )(k-l)/k Tpzp= -.-. Tlz l ......... (7) Pm] Pressure Gain Downstream of the Valve Port. Our experimental study showed that there is a pressure gain downstream of the valve port before the check valve during subcritical orifice flow. From the valve port area, the gas expands to a larger area before the check valve, and this sudden expansion results in a pressure gain during subcritical flow. Once critical flow is reached, the flow ADJUSTABLE SCREW ASSEMBLY ... , . -4+-- STEM Fig. 4-Diagram of valve used in effective-load tests. rate and pressure at the port remain constant even though the downstream pressure con- tinues to decrease. The pressure increase resulting from an abrupt enlargement is analyzed by applying conservation of mass, conservation of momentum (assuming free shear flow), and conservation of energy (assuming isentrop- ic process). These conservation equations are applied to the flow stations at the valve port and downstream of the port. The conservation of energy equation may be written with Mach numbers and temper- atures:
2 ' 2+(k-I)N Map ........... (8) where N Ma = v ;.,J kg cpl p. . ......... (9) Combining the three conservation princi- ples with the perfect -gas equation of state (EOS) and Eq. 9 yields 6
1 N Map
----'--------'--. .. ..... (10) +[(I-cp)/cp] Rearranging Eq. 10 results in a quadratic equation in the form ax 4 +bx 2 +c=O, "Our experimental study showed that there is a pressure TABLE 2-THEORETICAL AND EXPERIMENTAL RESULTS IN ORIFICE FLOW REGION gain downstream of the valve port before the check valve during sub critical orifice flow." 578 Valve Instrumented N-15R Unmodified N-15R Instrumented N-15R Port (in.) 0.25 0.25 0.50 'E 3 [1 OO(q expotlmental - q tn_al )jiq expotlmantal E* (%) -3.5 -1.1 8.5 Absolute E' (oAl) 4.1 3.2 8.5 Standard Deviation (oAl) 4.0 4.1 8.8 June 1993 JPT y An"lean t t t O'.... Aseat I I Fig. 5-Flow chart of the computational procedure for orifice flow. Fig. 6-Flow geometry and active pressures around the ball during throttling flow. with the coefficients a=A2k2 + 1-k, b=2kA 2 -2, and c=A, where N Map --J2+(k-1)Nltap A= . 1 +kNltap + [(l-c/lcP] The positive root between zero and one is the desired Mach number, N Ma3 , which allows computation of the static pressure ratio: Pp 0.97N Ma3
P3 c/>N Map
.................. (11) 110.00 100.00 90.00 80.00 Pvc = 446 psi 9 @ 70 F 70.00 60.00 50.00 "- <10.00 " "- 30.00 .m 20.00 "- 10.00 0.0 -10.00 Eqs. 10 and 11 incorporate the non- recoverable losses in the abrupt expansion, although a simple loss coefficient cannot be presented as it can in incompressible-flow analysis. A coefficient of 0.97 eliminates the discrepancies between the experimentally and theoretically obtained pressure gain downstream of the port. Pressure Drop Through the Check Valve. At flow rates exceeding 1.5 MMscflD, the pressure drop through the check valve be- comes significant. The downstream pressure after the check valve (Le., the production pressure) can be determined with the flow rate calculated by Eq. 5 and the pressure and 1.10 temperature before the check valve. Eq. 5 is rewritten as ........ (12) where t:.p = pressure difference between P3 and production pressure. The check-valve coefficients are calculat- ed with the experimental values of flow rate and pressure before the check valve. The plot of C v vs. (.J t:.p/P3 )/k shows a linear relationship given by C v = -7.36(..jt;p /kp3) +0.33 ..... (13) 0.25 in. port I. FULLY OPEN 1.00 2. 0.096 In. 3 0.076 In. 0.0 4. 0.046 in. 5. 0.039 in. 6. 0.025 in. 0.8
" 0.7 u 0.8 0.5
-6 -20.00 0.00 P hYS = -14.5 psi a ,*P.i = Pair ."'os,----;l0'-;.100 --ot-.",,--o'o-'o. .30 o. ,"'0-""0'-;. '"'s-'/o. 40 [[PIn) - Pp) / PIn II I k DISPLACEMENT, in. Fig. 7-Pressure differential vs. stem displacement from static probe tests. JPT June 1993 Fig. a-Product of discharge coefficient and expansion factor as a function of acoustic ratio for different stem displacements. 579 Initialize data Assume Pp C.lcul.t. the co.fflcl.nt. ., b, 0, d, of Eq. 24 Solve Eq.24 for \!item dil!iplacement Calculate A min by Eq.14 Calculate Cd' Y from Eq. 25 Calculate flow rate by Eq. 5 Calculate temperature at valve seat. Eq.7 Find In-situ gas velocity using the flow rate from Eq.5 and compute NMap from Eq.9 Solve Eq.10 for NMa a Calculate pressure drop before the check valve, p 3 by Eq.11 Nco Flow Area Calculate pressure drop and p". by Eq.12 & Eq.13 Fig. 9-Flow chart of the computational procedure for throttling flow. 580 Simultaneous solution of Eqs. 12 and 13 provides the pressure drop and production pressure. Fig. 5 is flow chart of the procedure for analysis of the orifice flow region. The theo- retical results obtained from the dynamic model are in very good agreement with the results of the experimental study. Table 2 shows the percent absolute and average errors. The dynamic model successfully pre- dicts the flow performance of the 1.5-in. nitrogen-charged, bellows-operated gas-lift valve in the orifice flow region. This model can be applied to all port sizes and other 1.5-in. valves with similar geometry by per- forming a limited number of tests to verify or adjust the discharge coefficient curve presented. Analysis and Results for the Throttling Flow Region In the throttling flow region of the gas-lift valve performance, the downward force on the stem and bellows assembly resulting from the nitrogen pressure in the dome be- comes sufficient to depress the stem, thus restricting gas passage through the valve. The throttling flow performance curves in- dicate that, as the production pressure decreases, there is sufficient flow area ini- tially to allow an increase in flow rate. After a maximum flow rate is reached, the flow rate declines approximately linearly as a function of production pressure. In the throttling flow region, critical flow occurs after the maximum point on the flow performance curve. Once critical flow is achieved, with a constant upstream pressure, the flow rate cannot be increased unless the minimum flow area is increased. Because the flow area decreases as production pres- sure decreases, critical flow must start at a production pressure on the downward slope of the throttling flow performance curve. Critical pressure occurs at the minimum flow area just above the seat. The flow re- gime at the lower end of the port remains subcritical after a normal shock. From the standpoint of force-balance anal- ysis, the area open for flow corresponding to an equilibrium position of the stem and ball assembly (stem displacement) is a func- tion of (1) the combined upward force re- sulting from the injection pressure and the effective pressure around the ball and (2) the combined downward force applied by the nitrogen pressure acting on the bellows area and the resistance of the bellows. Analysis of the throttling flow region fo- cuses mainly on the parameters affecting stem displacement from the closed position. Once the stem displacement for given flow conditions is determined, the area open for flow and the discharge coefficient can be cal- culated. The remainder of the model for the throttling flow region then uses the concepts and equations discussed for the orifice flow region. The minimum flow area is represented by the lateral surface area of a truncated right circular cone (Fig. 6). The base of the cone is the edge of the valve seat ring. The top June 1993 JPT 700r-__ ______________________________________________ . ttl r-
ex: 800 500 400 300 Port = 0.25 in. P (experimentaL)=-610 psig vc Pvc z: 605 psig THEORETICAL Pvc = 610 psig --'-- T (P ) =79 F vc
o -1 lL EXPERIMENTAL ex:
200 100 Pvc:: 615 pslg o
PRODUCTION PRESSURE, PSIA Fig. 1 O-Experimental flow performance curve and theoretical curves showing sensitivity of the model to valve closing pressure. of the ring is determined by the position of the ball on the stem. This minimum flow area is defined as A min = 7f(r p +Xmin)-v' (Ymin _Yp)2 +(r p -Xmin)2 . .................. (14) Although there is an area between Points 1 and 2 in Fig. 6 smaller than the "mini- mum" flow area given by Eq. 14, that smaller area must not be used. The analy- sis presumes ID compressible flow; there- fore, the fluid streamlines must be perpendicular to the flow surface that deter- mines the flow area. The location of the point (xmin, Ymin), ob- tained from the equation of the circle, is de- fined by Ymin-h=--v'r}-Xfrun ......... (15) J r2r2 and xmin = b p . .. (16) (Yp- rb-Ys)2+r} During steady flow, stem displacement is a function of the constant injection pressure, the nitrogen pressure in the dome, the bel- lows load rate, and the upward effective force that results from a mean pressure value acting around the ball. Although the value of the constant injection pressure is known, the remaining parameters need to be deter- mined by vigorous theoretical and/or ex- perimental study. The following sections JPT June 1993 explain the analytical and experimental methods applied to describe these pa- rameters. Prediction of the Nitrogen Dome Pres- sure. The nitrogen pressure in the dome un- der dynamic flow conditions is calculated as a function of the stem displacement, valve closing pressure (Le., the initial nitrogen charge pressure at flowing gas temperature), dome volume, and bellows area. The ana- lytical method involves taking the pressure differential with respect to volume using the EOS for real gasses and transforming volume to displacement 7 : A BYsiV d Bellows Load Rate. The literature contains contradicting definitions of the bellows load rate on gas-lift valves. 7-9 For this analysis, the bellows load rate is defined as the in- cremental pressure difference between the gas opening pressure and the nitrogen pres- sure acting on the entire bellows area re- quired for 1 in. of stem travel. Starting with the force balance, PgIAB-PdAB=ksYs+hys, ...... (18) and defining LB =ksiAB as the load rate and the pressure constant that reflects hysteresis as Phys =fhysIAB, Pg-Pd=LBYs+Phys . ........... (19) The load rate in Eq. 19 is determined from static probe tests during which sufficient time is allowed between measurements to ensure the precise equilibrium position of the stem, though the valve reacted rapidly. The temperatures of both the nitrogen and the air remained equal to the laboratory tem- perature. Fig. 7 shows a plot of the pressure differ- ence between the nitrogen and air pressure applied in opposite directions over the en- tire bellows area vs. the displacement of the stem and ball from the closed position. The steep slope after a stem displacement of 0.240 in. indicates that the valve is fully open. The significant result of the static probe tests is that the response of the bel- lows assembly depends on the direction of the change in pressure differential. This hys- teresis is reflected by Phys in Eq. 19. Phys is a function of the amount and direction of change in the net force applied on the entire bellows area. Ideally, for a given net force resulting from the pressure applied across the bellows area, there is only one stem displacement, regardless of the previous lower or higher net force, and the valve should be closed when that net force is zero. The test data, however, have shown that the stem displacement would be great- er if the net force is applied in decreasing order, exhibiting hysteresis. 581 Effective Force on the Stem and Ball. The upward opening force resulting from total gas pressure acting on the bellows assem- bly is not a directly measurable quantity dur- ing dynamic flow tests. In the force-balance equations (Eqs. 18 and 19), P g acts on the entire bellows area, A B , under static con- ditions. Under dynamic flow conditions, the pressure equivalent to P g that acts on A B is a function of an effective pressure, P e , act- ing on some projected area, A mean , around the ball and the injection pressure acting on the remaining area. The value of P e is bounded by the injection pressure and the pressure at the valve port: 1 r Pp Pe=--J pdA . ............ (20) Ameanpinj The difficulty in solving Eq. 20 is decid- ing on and justifying how the pressure pro- file changes between the injection point and the valve port. Effective load tests are con- ducted to find a solution. In these tests, the load cell measures tension caused by a dif- ferential load given by M=(Pe-Pinj)Amean, ........... (21) where Pinj>Pe>Pp' Analysis of the stem displacement shows that A mean can be any area greater than the projected area of the ball and less than or equal to the bellows area. Analysis of the load cell data results in a relatively simple relationship for effective pressure as a function of stem displacement, injection pressure, and valve port pressure: A seat (Pmin +Pinj)(Ab -A seat) Pe=P p --+ A mean 2Amean (AB-A b ) +Pinj ............. (22) A mean where A seat = r X ~ i n ............. (23) Pmin at the point (xmin,ymin) is assumed to be equal to Pp during subcritical flow; how- ever, at critical flow, Pmin must be equal to the critical pressure. Also, test results con- clude that the differential load corresponding to a pressure difference between the injec- tion point and the valve port for a fixed stem displacement is independent of the flow rate, and the pressures of interest are static rather than stagnation pressures. Stem Displacement. Because all the param- eters that effect the stem displacement can be expressed analytically as a function of measurable variables, direct calculation of stem displacement becomes possible. With Eqs. 15 through 23 combined, the stem dis- placement takes the form I yff+ayl+bY;+CYs+d=O, ...... (24) where a= (-C 4 -2L B A B C 3 -LBV di -A b C 2 )ILsAB, b=(2C 3 C 4 +LBABC} +LBABr) +2L B C 3 V di +C 2 Vdi +2ABC2C3)ILBAB' c=( -ABC I -C 4 C} -C 4 rJ -LBVdiC} 582 -LBVdirJ -2C 2 C 3 V di -A B C 2 C} -A B C 2 rJ)IL B A B , and d=(C I Vdi+C2C}Vdi+C2 VdirJ)ILBA B , and where -Pdi.-Phys' C 3 =yp- rb' and C 4 =A b Pdd(1- Pdi Oz). zdi op Eq. 24 is solved by the algebraic proce- dure given in Ref. 10 for the solution of a quartic equation and its resolvent cubic equation. Two of the four roots of Eq. 24 are complex, and only one of the two re- maining roots physically corresponds to the desired stem displacement. Discharge Coefficient. In the throttling flow region, stem position is the main factor de- termining the flow geometry and flow area. CdY is expected to be a function of stem displacement. Discharge-coefficient tests are performed to determine the discharge coefficient as a function of stem displacement from the closed position and of acoustic ratio, (ApIPinj)/k. The expansion factor is calcu- lated with the minimum flow area for the fixed stem displacement. Fig. 8 is a plot of CdYvs. acoustic ratio as a function of stem position for 0.25-in. port. Linear regression provides Eq. 25 to be used in Eq. 5 for the throttling flow region: C d Y=6.02ys Apipi -1.23 Apipi k k -2.34ys + 1.05 ............ (25) Fig. 9 shows a detailed flow chart of the computational procedure for the throttling flow region. The dynamic model is very sen- sitive to small changes in stem displacement. But the stem displacement, in turn, is very sensitive to small variations in injection pres- sure, valve closing pressure, and bellows behavior. Therefore, the predicted flow per- formance in throttling flow is significantly altered by small inaccuracies in measured values of the injection pressure, valve clos- ing pressure, bellows load rate, and the Phys observed during bellows-load-rate tests. Because Pinj and Pvc are measured with an experimental uncertainty of 5 psi, the theoretical performance curves are generated to examine how that uncertainty is reflected in the model. As an example, Fig. 10 shows the effect of PvC" Theoretical curves form a band when 5 psi is added to and subtracted from the measured pressure values. The lower curve coincides with the experimental performance curve. The theoretical model predicts the gas- passage performance of the valve in throttling flow within the uncertainties of the experimental study. All parameters that ac- count for the complicated flow geometry are analyzed theoretically and experimentally. This model sets guidelines to obtain a more general solution in throttling flow region for gas-lift valves with distinct flow geometry. Conclusions and Recommendations 1. The dynamic model accurately de- scribes valve flow performance in the orifice flow region. It significantly reduces the volume of data required to characterize orifice flow; one simply needs to obtain dis- charge coefficients by performing a limited number of tests. 2. The model is general enough to be ap- plied to all l.5-in. gas-lift valves that exhibit orifice flow if discharge coefficients are ad- justed to account for geometric factors. 3. For throttling flow, the model predicts valve flow performance by taking into ac- count all parameters that affect the compli- cated flow geometry. 4. The unique theoretical model, tuned by empirical parameters, sets guidelines and defines procedures to obtain a general mechanistic model for different types of gas- lift valves. 5. The consistency of the behavior of the bellows under dynamic flow conditions and over a time period must be investigated fur- ther. Further analysis of the response of the bellows, which is connected to the valve stem and ball assembly, is necessary to understand why small variations in individu- al parameters significantly affect valve per- formance. 6. Application of the model for I-in. nitrogen-charged valves needs to be inves- tigated. 7. The equations that constitute the model are convenient to program for both flow regions and can be incorporated into a de- sign procedure. The complete model was transformed into a FORTRAN program in Ref. 1. 8. Most of the continuous-flow gas-lift in- stallation designs use bellows-charged injection-pressure-sensitive valves with a small port. These valves function as back- pressure regulators to control casing pres- sure. Although the procedure given for the throttling flow region may not always ap- ply, the computational procedure given for the orifice flow region can be incorporated directly into the design of most continuous- flow gas-lift installations with injection- pressure-operated valves. Nomenclature A = area, L2, in.2 Amean = area that P eff applies on, L2, in. 2 Amin = minimum flow area, L2, in. 2 A seat = area the port pressure applies on during throttling flow, L2, in. 2 At = throat or port area, L2, in. 2 June 1993 JPT C p = specific heat at constant pressure, m(Ll/t 2 )/mT, Btu/lbm-oR Cd = discharge coefficient C v = valve flow coefficient E = average error f hys = force constant that reflects hysteresis, mLlt 2 , Ibf g c = proportionality constant h = y coordinate of center of circle, L, in. 1 = mechanical equivalent of heat k = ratio of specific heats, dimensionless ks = spring rate, m1t2, Ibf/in. L = load, mLlt 2 , Ibf LB = bellows load rate, m/Llt 2 , psi/in. N Ma = Mach number, dimensionless P = pressure, m/Lt 2 psia unless stated otherwise Pdi = initial dome pressure, m/Lt 2 psia P e = effective pressure on the stem and ball, m/Lt 2 psia P g = gas pressure that applies on entire area of bellows under static conditions, m/Lt 2 , psia Phys = pressure constant that reflects hysteresis, m/Lt2, psia Pinj = injection gas pressure, m/Lt2, psia Prnin = pressure at minimum flow area, m/Lt2, psia Ppr = production pressure, m/Lt 2 , psia Ptro = test-rack opening pressure, m/Lt2, psia Pvc = valve closing pressure, m/Lt2, psia q = volumetric flow rate, L3 It, MscflD r = radius, L, in. R = universal gas constant, mLl It 2 T, psi_ft3 Ilbm-mol-oR T = temperature, T, oR unless stated otherwise Tinj = injection gas temperature, T, oR Ts = stagnation temperature, T, of T t = true static temperature, T, OF v = velocity, Lit, ft/sec V = volume, L3, in. 3 V di = initial dome volume, L3, in. 3 Xmin = x coordinate for minimum flow area, L, in. Yrnin = Y coordinate for minimum flow area, L, in. Y s = stem displacement from closed position, L, in. Y = expansion factor, dimensionless z = gas compressibility factor, dimensionless Zinj = gas compressibility factor at injection pressure and temperature, dimensionless () = ratio of square root of minimum flow area to square root of total upstream flow area 'Y g = gas specific gravity, dimensionless t:.p = pressure differential, m/Lt2, psi JPT June 1993 p = density, m/L3, Ibm/ft 3 <J> = area ratio, A p lA3 Subscripts b = ball on stem B = bellows d = nitrogen dome P = gas-lift valve port 1 = upstream 2 = downstream 3 = downstream of valve port before check valve Acknowledgments The support provided by the member com- panies of TUALP is gratefully ac- knowledged. Teledyne Merla deserves special thanks for fabricating significant components of the test facility. We also thank Camco for providing a test facility. References 1. Hepguler, G.: "Dynamic Model of Gas-Lift Valve Performance," MS thesis, U. ofTul- sa, Tulsa, OK (1988). 2. Hepguler, G. et al.: "Instrumentation for the Dynamical Modeling of Gas-Lift Valves," paper ISA 88-0768 presented at the 1988 Inti. Instrumentation Symposium, Albuquerque, May 2-6. 3. Brown, K.E.: The Technology of ArtificiaLLift Methods, Petroleum Publishing Co., Tulsa (1980) 2a, 96-109. 4. Fluid Meters-Their Theory and Application, sixth edition, report, ASME Research Com- mittee on Fluid Meters, ASME, New York City (1971). 5. Hottel, H.C. and Kalitinsky, A.: "Temper- ature Measurements in High-Velocity Air Streams," J. Applied Mech. (March 1945) A25-A32. 6. Benedict, R.P. et al.: "Generalized Flow Across an Abrupt Enlargement," J. Eng. for Power (July 1976) 327-34. 7. Decker, L.A.: "Analytical Methods for De- termining Pressure Response of Bellows Operated Valves," paper SPE 6215 available at SPE, Richardson, TX. 8. API Gas Lift Manual, Vocational Training Se- ries, API, Dallas, TX (1984) 6. 9. Winkler, H.W. and Camp, G.F.: "Dynamic Performance Testing of Single-Element Un- balanced Gas-Lift Valves," SPEPE (Aug. 1987) 183-90. 10. Beyer, W.H.: CRC Standard Mathematical Tables, twenty-sixth edition, CRC Press Inc., Boca Raton, FL (1981) 9-12. 51 Metric Conversion Factors ft3 x 2.831 685 E-02 m 3 OF (OF-32)/1.8 C in. x 2.54* E+OO cm Ibm x 4.535 924 E-01 kg psi x 6.894 757 E+OO kPa * Conversion factor is exact. Provenance Original SPE manuscript, Dynamic Model of Gas-Lift Valve Performance, received for review April 7, 1991. Revised manu- script received Sept. 21, 1992. Paper accept- ed for publication April 1, 1993. Paper (SPE 21637) first presented at the 1991 SPE Pro- duction Operations Symposium held in Ok- lahoma City, April 7-9. JPT Authors Blais Doty Gokh ... Hepguler Is a petroleum en glneer at Union Pacific Resources Co. in Fort Worth. Before Joining Union Pa- cific In 1989, be worked at Baker 011 Tools In Houston. His teChnlcallntereste Include artificial-11ft methods, reservoir simulation, multlphase flow, and com- puter systems and networks for,engl- nearing applications. Hepguler holds as and MS degrees in petroleum engineer. Ing from the U. of Tulsa. ZeUmlr Schmidt Is professor of petroleum en- gineering at the U. of Tulsa and direc- tor of TUALP. He spent 10 years as a production engineer with INANaftaplin in Yugoslavia and has served as a con- sultant to various companles. Schmidt was a 1987 Distinguished Lecturer and is a member of the Editorial Review Committee. He holds an engineering degree from the U. of Zagreb and MS and PhD degrees in petroleum engineer- Ing from the U. of Tulsa. Roger BlaiS, interim provost and ecedemlc vice presi- dent at the U. of Tulsa, eamed,a BA degree from the U. 01 Minnesota and a PhD degree from the U. of Oklahoma, both In physics. He has been at the U. of Tulsa since 1977 and has been associ ate director of TUALP since Its founding in 1983. He chaired the Physics Dept. during 1986-88. Dale R. Doty is associ- ate professor of mathematical sciences at the U. of Tulsa and 8ssoclate director of TUALP. He Joined the U. of Tulsa In 1975 and hes been Involved In research and consulting In petroleum production. He holds BS. MS, and PhD degrees In mathematics from Michigan State U. 583