Você está na página 1de 8

Nuclear Engineering and Design 72 (1982) 197-204 North-Holland Publishing Company

197

AN ANALYTICAL SAFETY VALVE Avtar SINGH,

STUDY OF THE DYNAMICS

AND STABILITY OF A SPRING

LOADED

EPRL Palo Alto, California, USA

Received April 1982

Spring loaded self-actuating safety valves are employed as part of the overpressure protection systems in various industrial applications. In order to design and predict their performance it is necessary to study the dynamic behavior of the valve over a range of fluid and system conditions. A one-dimensional model has been developed to study the effects of different valve parameters such as the spring-mass characteristics, geometry of internal parts, adjustment ring settings, bellows etc. which influence the dynamic behavior and stability of the valve. Analytical results for steam flow conditions are presented to demonstrate the relative effects of these parameters on the valve opening time, maximum lift, blowdown (upstream pressure differential between the valve opening and closing) and any oscillations of the valve stem. If the valve is not properly backpressure compensated, it may become unstable as the stagnation pressure at the valve inlet decreases. Lowering of the guide adjustment ring position or raising the nozzle adjustment ring generally results in improved stability, shorter valve opening time, higher lift and longer blowdown. The effect of damping on the valve stability is also demonstrated. The model can be used to evaluate the design of safety valves and damping devices to eliminate unstable valve dynamic behavior.

1. Introduction
High pressure vessels in various industries are protected against over-pressure by self-actuating springloaded safety valves. They open when the upstream pipeline or vessel pressure exceeds a certain allowable value, thus releasing the fluid to maintain the system pressure below a certain design value. In order to properly design the overpressure protection system for various transients, it is necessary to understand the dynamic behavior of these safety valves. For example, a slowly opening valve, may not release a sufficient quantity of fluid to control the rise in system pressure and maintain it under the design value. On the other hand, quick opening or cycling may cause substantial dynamic loadings on the discharge piping. A prolonged or oscillatory dynamic behavior can inflict damage to the valve itself. To date, not much work has been done to analyze the dynamic behavior of a safety valve and its interaction with the protected system. In most cases it is assumed that these valves open within a certain opening time when the upstream pressure exceeds a certain value called the Set Pressure. When the valve is fully open, it is expected to allow a desired flow rate and when the pressure upstream of the valve decreases by a designed 0029-5493/82/0000-0000/$02.75

value called the Blowdown the valve is assumed to close. In order to design and evaluate the performance characteristics of a safety valve it is necessary to understand its dynamic behavior under different fluid and thermodynamic conditions. The objective of this study is to develop a coupled thermal-hydraulic and spring mass systems model to be utilized as part of the overall system interactive model. Most of the existing theoretical work related to valves involves analysis of steady state mass flow rate, [1-3] vent pipe sizing [4] and transient discharge pipe loads [5]. Fowler et al. [6] attempted to simulate the safety valve dynamics considering only the mass-spring effect on pressure surges in a heat exchanger. Funk [7] considered the interaction between the inlet pipe fluid dynamics and the spring mass system of a poppet valve to analyze the valve dynamic stability. Ray [8] formulated a non-linear semi-empirical model of a safety valve and dynamic equations were derived from fundamental principles of rigid body motion and fluid dynamics. The analytical model developed in the present study is based on a similar approach where the empiricism has been minimized by representing the fluid dynamic force acting on the valve disc in the form of an

1982 N o r t h - H o l l a n d

198

A. Singh / A spring loaded safe(v valve

explicit analytical expression. Effects of pressure distribution in the huddling chamber (volumetric region between the valve disc and seat) and the backpressure in the body bowl on the stem dynamics have also been included and shown to be quite significant.

Bellows

/ , ~ - ~ Spring

Disc Seat ~ I
....... I-< ......

2. Theory The cross section of a typical safety valve is shown in fig. 1. Functional internal components include a nozzle, seat, disc, piston rod, spring, bellows and adjusting rings. The bellows are designed to perform two functions - firstly to isolate the back of the disc from the backpressure existing in the body bowl and secondly to prevent leakage of fluid to the atmosphere. It should be noted that the bellows do not cover the entire area at the back of the disc, hence a part of this area located around the outer periphery of the disc may be exposed to the backpressure, as shown in the simplified sketch of fig. 2. Adjustment rings can be moved up or down vertically and are used to adjust opening and closing characteristics, lift and blowdown settings. Governing equations for one-dimensional fluid and valve stem motion have been derived for the control volume shown by dotted lines in fig. 2. Fig. 3 shows the

Guide ad;ustment ring

,\

\ \

~"""~

NOZzle adjustment ring

\\ \

\\

I\'

Flow

Fig. 2. Schematic diagram of valve components.

forces acting on the control volume in the vertical direction; F D representing the reaction force exerted from the valve disc onto the fluid. Ps denotes average pressure distribution over the seat area A s and is approximated by the critical pressure Pc based on the assumption that during the opening and closing of the valve, choking occurs at the inner radius of the seat. Applying the m o m e n t u m equation, in the x direction, to the control volume of fig. 3 the following can be derived:
Fsprmg

Vs,~e--~
disc
-~

LJ.LL

MIT FD
Bellows Guide ring -

Valve stem I displacement (x)

m[

F~

ttttttP,
Fig. 1. Cross section of a typical safety valve. Fig. 3. Force distribution in the valve.

A. Singh / A spring loaded safety valve PIAt + P s A s - FD = ~ f f f~.vPVx dv + (~ PVxV:, dA


c.s.

199

where the first term on the right hand side represents the rate of change of momentum inside the control volume. Because of the small geometric distance inside the valve and the time scales involved during the valve motion, this term is negligible; however, it has been approximated by assuming negligible pressure and density gradients inside the valve. The second term denotes the rate of momentum effhix from the control surface. Expanding and simplifying the above equation, the following can be obtained for the force exerted by the fluid on the disc for frictionless compressible flow:

the spring force, the third denotes the force due to backpressure and the last term is the weight of the disc and other moving parts inside the valve. The valve stroke x is assumed to be zero when it is closed and the maximum valve lift Xmax is limited by a mechanical stop. At either limit the stem velocity must be zero, i.e.

dx/dt=O,

i f x : O o r x = X m ~ X.

3. Applications and results


A typical safety valve installation may consist of a high pressure vessel (to be protected against over-pressurization) connected to the safety valve through a certain piping configuration upstream of the valve. Downstream of the valve there may also be a connection to some piping in order to discharge the effluent at a certain location. The set point pressure at which the valve opens is designed to be higher than the normal operating pressure in the vessel by a certain design margin. In a typical safety valve operation transient, the following sequence of events may occur. Pressure in the vessel may start to rise at a certain rate depending upon the rate of increase in total energy caused by some normal or abnormal system conditions. As the pressure upstream of the safety valve reaches the set point pressure, the valve starts to open at some finite rate. As the valve opening proceeds, the discharge flow rate through the valve increases which may cause the vessel pressure to drop after it has reached some peak value. The design of the valve opening time and discharge capacity are expected to be such that a maximum or peak pressure reached in the vessel is below the allowable design pressure of the vessel. As the vessel depressurizes, sometimes it is also required that the valve be designed to close to limit the pressure drop to a certain design value called at Blowdown. On the downstream side, as the valve opens, the pressure in the valve body bowl and any connected piping increases with time resulting in some dynamic structural loading. The transient fluid and thermodynamic conditions upstream and downstream of the valve will not only depend upon the valve performance characteristics but will also be a strong function of the upstream and downstream system configurations. The model described in this study does not include the analysis of the upstream system and its interaction with the valve. In order to study the effect of different upstream pressure conditions expected to occur during a typical valve operation transient, a time varying pressure boundary

F D = PiAi + PsA s + AtpiV ~ -- plAl (L+x)-d-f+


( .dV,

vdx) l dt

m~VecosO
gc

(1)

where the last term in the above equation approximates the x component of the exit flux from the valve. the is calculated by assuming that ideal gas choke flow occurs at the minimum flow area, which is defined as follows:

A* =2~rRlx
A* = rrR~ and,

forO<x~R~/2Ri,
for x ~ R2N/2R1,

(2)

mo= CoA*eo

T~i I

(3)

where for steam k = 1.3. F represents the exit velocity, which is approximated by the average fluid velocity between the seat and disc. Fluid properties inside the valve are calculated by applying the ideal compressible gas flow equations. Choked mass flow rate through the valve is corrected by applying a discharge coefficient (assumed to be 0.9). 0 represents the effective angle at which the fluid discharges as it comes out of the flow area between the guide and nozzle adjustment rings. 0 can be altered by moving the two adjustment rings up or down. The equation of motion for the valve disc may be written as follows considering the forces acting on the disc as shown in fig. 3:

MD:i + c~ = F D -- ksp~i,s( x + Xo) -- VaA B - MDg.

(4)

Here the second term on the right hand side represents

200

A. Singh / A spring loaded safe(y valve

8=30

c=OO 2600 ksp,,~g = 2 233 105 Ib/fi AB= 00

2500

~ 2400

23OO

22oo

I
0

~ i

[ J J i
1

~ I

~ ~ I
3

~ i

i
4

2 Time (s)

Fig. 4. Prescribed stagnation pressure time history at the valve inlet.

condition has been prescribed at the valve inlet as shown in fig. 4. The upstream reservoir pressure is assumed to rise at a constant rate of 300 p s i / s starting from the set point value of 2500 psia. After a peak value of 2600 psia is reached, the upstream pressure is prescribed to drop at a constant rate of 100 p s i / s until the valve closes. Upstream stagnation temperature is also prescribed as remanining constant at 650F. Backpressure downstream of the valve is approximated by assuming the downstream piping system to be a lumped capacitor volume which is fed by the valve discharge flow at one end and assumed to be open to the atomsphere at the other end (choke flow is assumed to exist at the open end). For the calculations presented

in this study the downstream pipe is assumed to be of 6" inside diameter and 12 ft long. The valve design and range of parameters considered in this study are typical of a safety valve used for high pressure application. Since the objective of this study is to demonstrate the analytical capabilities of the proposed model, the numerical results presented here are typical only of the selected valve type and design and prescribed boundary conditions. However, the model can be used to evaluate the design and stability of other existing valves or for designing a new valve. The calculated valve stem position as a function of time is shown in fig. 5 for the following values of various parameters: mass of the moving parts, m = 110 pounds; spring constant, kspring z 2.233 105 p o u n d s / f o o t : effective area for back pressure, A B = 0.0: damping coefficient, C = 0; maximum lift = 0.044 feet; discharge angle, 0 = 30 degrees; effective seat area, A S = 20% of the area ~'R~. The valve opens to a stable full lift in 90 ms and stays

full open against the maximum lift stops until the upstream stagnation pressure drops to about 2520 psia, at which time the fluid force F D acting on the disc falls
below the downward force and the valve starts to close. The fluid force F D on the disc is shown in fig. 6a. Before the valve opens, F o is equal to the valve upstream

pressure multiplied by the disc area ~rR2 exposed to the inlet pressure. As the valve begins to open, fluid flows
through the opening between the disc and seat, hence an additional area A s gets exposed to the inlet pressure

giving rise to an increase in upstream force [estimted by


the term

P~A~in

eq. (1)]. This results in a faster opening

005

27,500

e= c=00 004

30 25,000 --

8=30 c = 00 kspr,ng = 2 2 3 3 1051b/ft Ae = 00

ksDr,ng = 2 233 X 105 Ib/ft A 8 = 00 ~=

003

g
--> 2 0 0 0 0 --

~5 E o~ 0 0 2

g
o 17,500 --

O01

15,000

--

0 O0 1 2 Time (s) 3 4

2 Time (s}

Fig. 5. Valve stem position as a function of time.

Fig. 6a. Upward fluid force acting on valve disc and stem.

A. Singh / A spring loaded safety valve


2000

201

8=30' c=O0 15OO kspr,ng = 2 2 3 3 X 105 Ib/ft AB= 00 A s = 0 2 X nR 2


ta

,~ 10o0 t5

fully closed position. The results described above will be considered as the base case valve performance. In the following calculations, sensitivity results are presented to show the effect on valve opening, closing characteristics and stability, by varying some of the valve parameters listed above. Prescribed inlet pressure and temperature conditions are kept unchanged.

500

--

4.
0 --

Effect

of

backpressure

-I

I I I I I I I I I [ I I I I I I I I I [ I
1 2 Time (s) 3 4

Fig. 6b. Net upward force acting on the valve disc and stem as a function of time.

or popping action of the valve. The net upward force on the disc is shown in fig. 6b. As the valve opening increases, the mass flow rate through the valve also increases as shown in fig. 7, giving further rise to the disc force due to increased momentum flux. As the upstream stagnation pressure decreases, the mass flow rate and F o decrease and the valve starts to close due to the unbalanced compressive spring force. As the valve closes, the mass flow rate decreases resulting in a rapid decrease in the momentum component of the disc force F o. Finally, the valve fully closes at a lower stagnation pressure than the set point (about 10% below the opening pressure) due to an increased area exposed to the upstream pressure during the open position versus the

When the valve is fully closed the backpressure in the body bowl is approximately equal to the atmospheric pressure. As the valve opens, the backpressure builds up as the flow increases. If the valve is not fully backpressure compensated, some area A B at the back of the disc is exposed to the pressure downstream of the disc in the valve body bowl. This causes a downward force PBAa on the disc opposing the fluid force F D and reinforcing the compressive spring force tending the close the valve. Fig. 8 shows the calculated valve stem displacement time history for an effective backpressure area A n equal to 0.10 of the disc area (erRS) exposed to the inlet pressure when the valve is fully closed. Due to a decrease in the net upward force on the disc, the valve becomes slightly unstable during opening until the inlet stagnation pressure rises to a sufficiently high value (approximately 70 psia above the set point) when the valve can be kept in stable fully open position. As the upstream pressure decreases, the valve starts to close earlier and starts oscillating during closing due to feedback between the forces acting on the upward and downward direction on the disc which excites the spring

150

0=30 c=00

k$0rlng = 2 2 3 3 X 105 Ib/fl ~ 100 -A8 = 00 s = A


~ 0.03

kswing = 2 233 X 1051bill A B = 0 1 X ~R 2 A s = 0 2 ~R 2

E
I~ 50 --

002 i
0.01 0 --

i
0

1
1

I
3

I
4

I
0.00

2 Time {s)

Fig. 7. Steam mass flow rate through the valve as a function of time.

2 Time (s)

Fig. 8. Effect of back pressure on valve stem motion.

202

A. Singh / A spring loaded safety valve


005 0 = 30 k c=O0 = goq'~ w O= 20

ln~b/fl
004 -A 1=

c=O0 ksprm ~ = 2 2 3 3 105 Ib/ft A8= O0 As = 0 2 X ~rR21

= ~5 2

1000

o= = o

i
o
E m -1000

003

002

--

001

--

Illl~llll~lllllJlfl~li
0 1 2 Time (s) 3 4

o.oo -I
0

I ~ I I I I I J
1 2 Time (s) 3

i i
4

Fig. 9. Net upward force on the valve disc under the influence of back pressure. mass system. Fig. 9 shows the total forces acting o n the valve disc. As the u p s t r e a m pressure approaches 10% below the opening pressure, valve stem oscillations decrease until the valve disc reseats.

Fig. 11. Effect of lowering the guide ring on valve stem motion.

5. Effect of adjustment ring settings


As discussed earlier, the effective angle of discharge O c a n be varied by moving the a d j u s t m e n t rings up or down. M o v i n g the guide a d j u s t m e n t ring up will increase the discharge angle 8, resulting in a reduction in the fluid force on the disc F D in eq. (1). Fig. 10 shows the opening and closing characteristics of the valve for 8 - - 4 0 a n d no backpressure effects. As expected, due to a net decrease in the total u p w a r d force on the disc,
0 05

the valve attains a stable fully open position a n d starts to close at a higher inlet stagnation pressure than the base case (8 = 30 , fig. 5). It should be noted that the complete closure of the valve still occurs at a b o u t 10% (blowdown) below the opening stagnation inlet pressure a n d remains unaffected by the change in 8. Lowering of the guide a d j u s t m e n t ring will decrease the effective angle of discharge 8, resulting in an increased m o m e n t u m force on the disc. Hence, as shown in fig. 11, for 8 - - 2 0 the valve opens at a faster rate reaching a stable full lift in 15 ms a n d does not start to close until at a b o u t 1.9 s when the inlet stagnation pressure has d r o p p e d approximately 50 psia below the o p e n i n g or set p o i n t pressure. Again, the complete closure occurs at the same blowdown, i.e., 10%.

0 05

0=40 004 -c=O0 kspr,ng = 2 2 3 3 105 ~b/ft A B = O0 = 5003 A s = 0 2 X ~R12 -003


004

8=30 c=O0

kspring

= 2.233 X 105 Ib/fi

A 8 = 0.0

= ol

. R ,2

o=

$
002
--

0 02

001

001

ooo-~

I I
1 2 Time (s) 3

i J
4

0.00

I i = J I I J I I I I I"llk f I I I I = I t
0 1 2 T i m e (s) 3 4

Fig. 10. Effect of raising the guide adjustment ring or increasing the angle 0 on valve stem motion.

Fig. 12. Effect or reduced effective seat area or lowering of the nozzle ring on valve stem motion.

A. Singh / A spring loaded safety valve


Lowering of the nozzle adjustment ring primarily decreases the effective seat area As, which reduces the total upward seat force component acting on the valve disc. The effect of reducing the effective seat area by 50% from the base case is shown in fig. 12. Due to the decreased upward fluid force F D in eq. (1), the valve becomes unstable during opening causing it to flutter during opening and closing. The valve does not achieve a stable fully open position and starts to close at a much higher inlet pressure (approximately 2580 psia). Note that the valve completely closes at about 2375 psia corresponding to a 5% blowdown.
0.05

203

0.04 &-

0=30* c = 0.0 klmting = 2.0 X 105 Ib,'It AB = 0.0

i
>

0O3
=

~002

0.01

000

__ 1 2 Time (s) 3 4

6. Effect of spring stiffness


For a stiffer spring (ksp,~,8 = 2.5 105 lb/ft), the stiffness constant ksprin$ is higher which causes a bigger downward spring force for a given deflection or valve stem position. This causes the valve t o become less stable in the absence of any damping as shown in fig. 13a. Also, as expected, the valve does not remain open in the full position and starts to close at much higher inlet stagnation pressures (2600 psia). For a softer spring (ksp,~,g = 2 . 0 105 lb/ft), fig. 13b, the valve opens to a stable fully open position faster (valve opening time = 14 ms) due to the smaller spring force, and does not start to close until the inlet stagnation pressure drops to 2420 psia, about 3% below the opening set pressure. Finally, the valve completely closes at an inlet pressure of 2250 psia.

Fig. 13b. Effect of decreasing spring stiffness on the valve stem motion.

7. Effect of damping
In the calculations presented above, the friction at the sliding surfaces of moving parts was assumed to be negligible. In a real valve, due to a slight asymmetry between the surfaces of the moving and fixed parts, some friction force may exist opposing the valve stern motion. In certain applications where safety valves are observed to be unstable causing fluttering or chattering, it may be possible to employ a damper device to the valve stem to subdue the valve stem oscillations. In fig. 8, it was shown that due to a higher backpressure, the valve may become unstable and start to oscillate at lower upstream stagnation pressures. These oscillations can be eliminated by providing a damping device to damp the valve stem motion as shown in fig. 14,
0 05

k~

= 2.50 X 10 s Ib/ft AB = 0 0 = . n 2

0.04

--

= , 0 fb-s/ft kit)ring = 2233 X 105lb/ft

003

~
002 ~5

003

~0.02

0.01 ~ ' - I

0.01

000 0

ItllllllLtllll
1 ;2 Time ($) 3

I 000 "1 I I 1 [ 11 1 I I [ I 2 Time (s) I [ 11 3 I I~1"% [ 4 I

Fig. 13a. Effect of increasing spring stiffness on the valve stem motion.

Fig. 14. Effect of damping on valve stem motion.

204

,.t. Smgh / A ,wring loaded safe(v valve

where the valve dynamic response was calculated under the same conditions as for fig. 8, except a damping coefficient c = 500 lb s / f t (approximately one third of the critical damping) was used in eq. (4).

AB
m~;

A*
C

8. Conclusions and discussion


A one-dimensional valve dynamic model has been developed considering the effects of various valve design parameters that can be used to study the valve dynamic response and stability under different upstream and downstream conditions. Sample calculations made for steam flow through a typical high-pressure application safety valve demonstrate the analytical capabilities of the model and the relative effects of different parameters. Valve stability and duration of full opening in general can be increased by a lowering of the guide adjustment ring (smaller discharge angle 0), a softer spring or smaller backpressure effects, on the other hand, a stiffer spring, higher backpressure or a lowering of the nozzle adjustment ring (lower blowdown) may lead to instabilities or fluttering of the valve. A damping device applied to the valve stern may help to eliminate valve dynamic instabilities or oscillations. Valve dynamic performance depends strongly upon the upstream and downstream thermal-hydraulic conditions, hence for a given safety valve application, coupled interactions between the valve and upstream and downstream system must be considered to evaluate the design adequacy of the overpressure protection system. The model presented here could be used in conjunction with the transient simulation of other components of the system such as the vessel, nozzles, upstream and downstream piping with elbows, bends and other flow restrictions. Further work needs to be done to calibrate the model against test data on actual valves.

CD FD g k
kspring

L rn MD Pi PB R R1 Ry Vj

vo
X Xo

Xmax Ot

area of disc subjected to body bowl backpressure; seat area; nozzle or orifice area; damping coefficient for valve stem motion; valve discharge flow coefficient; upward fluid dynamic force on valve disc: gravitational constant; isentropic exponent: spring constant; length of valve nozzle; valve mass flow rate; mass of valve disc and moving parts; static pressure at the inner seat radius; back pressure in the body bowl; gas constant; inner seat radius; nozzle radius; upward fluid velocity at the inner seat radius: exit fluid velocity at rings; valve stem displacement; initial spring compression; maximum valve stem displacement; fluid density at the inner seat radius.

References
[1] L. Thompson and O.E. Buxton, Maximum isentropic flow of dry saturated steam through pressure relief valves, presented at the Pressure Vessel and Piping Conf. Montreal. Quebec, Canada (July 1978). [2] D.W. Sailer, The flow of liquids and gases through pressure relief valves, presented at the Third National Congr. on Pressure Vessel and Piping, San Francisco, California (June 24-29, 1979). [3] J.W. Sale, Safety valve mass flow rate calculations and correction factors, presented at the Third National Congr. on Pressure Vessel and Piping, San Francisco, California (June 24-29, 1979). [4] G.S. Liao, Analysis of power plant safety and relief valve vent stacks, ASME Trans., J. Engrg. Power (October, 1975) pp. 484-494. [5] F.J. Moody, A.J. Wheeler and M.G. Ward, The role of various parameters on safety and relief valve pipe forces, presented at the Third National Congr. on Pressure Vessels and Piping, San Francisco, California (June 24-29, 1979). [6] D.W. Fowler, T.R. Herdon and R.C. Wahrmund, An analysis of potential overpressure of a heat exchanger shell due to a ruptured tube, ASME Petroleum Engineering Conf. Preprints (September, 1968). [7] J.E. Funk, Poppet valve stability, Trans. AMSE, J. Basic Engrg., paper number 62-WA-160. [8] Ashok Ray, Dynamic modelling and simulation of a relief valve, J. Simulation (November, 1978) pp. 167-172.

Acknowledgement
This study was performed as part of the E P R I / P W R Safety and Relief Valve Test Program sponsored by Electric Power Research Institute and participating P W R utilities. The author is thankful to Dr John Carey of EPRI for initiating this work and to Dr Alan J. Bilanin of Continuum Dynamics for helpful discussions. Computing and graphics help from Mr Glen Snyder of E P R I is also gratefully acknowledged.

Nomenclature
A~ seat area at inner radius R~;

Você também pode gostar