Beruflich Dokumente
Kultur Dokumente
net/publication/289633576
CITATIONS READS
0 1,232
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Faruk Selimovic on 31 August 2016.
NOMENCLATURE
Water hammer is an important load in many pipe systems, for example, in the process
industry, district heating and power generation plants. It is a pressure surge or wave
resulting when a fluid in motion is forced to stop or change direction. If pressure surge
are induced by check valve they could result in pressure peaks and oscillations causing
fatigue damages and could lead to pipe break and support deformation. Check valves
should close so that the amount of fluid flowing to wrong direction is as small as
possible.
There are different types of check valves: swing, lift, tilting disk, duo/double disk, stop
and nozzle. But most common because of its simple design, low pressure drop, price and
wide range of sizes are Swing Check Valves (SCV). Unfortunately, information about
the SCV:s characteristic is usually limited. This makes it difficult to perform precise
calculations of the pressure and forces caused by the water hammer. Hence, safety factors
should be used if the valve characteristic is not known which can lead to oversize of the
entire system.
One difficulty with describing check valves lies in the fact that the characteristics depend
on the circumstances as the pressure drop across the disk is depending on its speed. It is
relatively easy to produce a loss characteristic for a stationary disk, but because of the
fact that the disk movement is dependent on the pressure loss across it, there is no
definition for a general loss characteristic. Sometimes a process engineer could face the
fact that there is not any specification given of the loss characteristics regarding
stationary disk. Instead, a curve from a similar valve is used. When there is no
sufficiently similarity between valves with known and unknown characteristic two
alternatives could be the solution. One is to perform experiments with a fixed disk, which
is quite expensive and often difficult. The second is using three-dimensional flow
calculations, Computational Fluid Dynamics (CFD), producing a curve for a stationary
disk. This is also a quite expensive alternative, but quite common today in industry.
In this work a CFD tool was applied to analyse the flow field with the objective of
investigating the pressure drop over the disk and calculating the hydraulic force acting on
the disk. These investigations were performed for both stationary and moving disk.
Moving disk simulations give better understanding of the closing sequence which when
transferred to a one dimensional model can predict forces more close to reality.
The goal of this work is to perform three-dimensional flow calculations with CFD tool
FLUENT, Ref [7] for both a stationary and movable disk. A one dimensional pressure
surge tool (in house coded) has been applied for water hammer calculations with more
than 30 years of experience and verifications by field tests. The results of this work
expect to give improvements for a SCV model when integrated into in house code. The
results are also compared with measured and calculated results from the literature.
The motion of the valve disk can be described by Newton’s second law, an angular
momentum conservation law. The momentum conservation equation can be traditionally
written as:
(1)
(2)
Where is the hydrodynamic torque, is the torque because of weight. The weight
of moved parts is corrected by the buoyancy force of steel mass in the fluid. Additionally
one should add a torque because of friction axis. The friction torque stands for the
friction of the lever arm at the axis of rotation. It is the sum of a steady friction and a
moving friction. Typically this value is negligible comparing the other values for the total
torque and we are not taking it into account here.
Left side of the Eq. (1) and (2) terms the total moment of inertia: I, the moment of inertia
of the disk assembly with respect to the valve shaft axis. If an external weight is applied
on a valve then additional term for torque and inertia in Eq. (2) should be added. This
was not subject in this work. In the dynamic CFD simulation the hydraulic torque
comprises also an acceleration part which is object of the added mass model. Therefore
no added mass must be introduced into the moment of inertia in CFD simulation. Further
the hydraulic torque is defined by:
(3)
(4)
(5)
where . We introduced the relative velocity on the bases of a field test of pump
stops during 1980.
It is quite common to refer to the velocity to the seat area. Sometimes the velocity is
referred to the gap area, e.g. DRAKO- Ref [8], and RELAP5, Ref [9], models.
The impingement of the flow jet on the valve disk may result in an inclination of the
resulting force acting on the disk. Here the total hydraulic force acting on the disk is:
(6)
KSB swing check valve Staal40 AKKS DN400 PN40 has been selected in such purpose
that large diameter makes it easier to perform grid generation and also increase reverse
velocity at higher velocity gradients, Ref [1]. Valve data for simulations can be found in
Table 1.
The direction of hydraulic force deviates from the normal to the disk by an angle .The
calculated or actual lever arm of the hydraulic force is represented by Lx which is
displaced to the centre of the disk and is calculated by:
(7)
12
10
8
Deviation Angle (°)
0
0 10 20 30 40 50 60
Fig. 4: The deviation angle delta of the hydraulic force of the opened valve deviates
from the normal to the disk by about 12° and decreases to negligible values for
less than 30°
In the partly opened valve, 30° to 60°, the actual lever arm of the hydraulic force exceeds
the geometrical lever arm up to 8,5%. The negative offset between about 15° and 30 °
means that the actual lever arm of the hydraulic force becomes shorter than the
geometrical lever arm.
Offset , Distance of the Acting Point of the Hydraulic Force to the Centre of Flap
25
20
15
Offset (mm)
10
0
0 10 20 30 40 50 60
-5
Valve Angle (°)
10000
1000
Torque (Nm)
100
10
0 10 20 30 40 50 60
Valve Angle (°)
The fluid velocity in the seat is kept constant to 1.97 m/s in the simulation and is lower
than the critical velocity, which is needed to fully open the SCV which in the actual
application is about 2.45 m/s.
The hydraulic torque results from the local pressure loss at the disk:
(8)
Inversely the local pressure loss is calculated with the knowledge of the hydraulic torque
as result of FLUENT simulation and herewith the local loss factor referred to the fluid
velocity in the seat can be evaluated:
(9)
For general use the pressure loss factor K is often transformed to the flow coefficient Cv
which is unlimited for low K therefore it seems preferable to transfer K to the so called
flow factor alpha defined by:
(10)
The force normal to the disk Fn differs slightly from the value of the hydraulic force Fhyd,
Fig. 7 which is a result of the simulation. Following to Ref [2] and [5], an effective force
Feff acting normal to the disk center was introduced:
(11)
(12)
with
(13)
700
600
500
400
300
200
100
0
30 35 40 45 50 55 60
Valve Angle (°)
With the knowledge of the force factor Ff , Fig. 8, and the local loss factor Kloc the static
hydraulic torque for some other swing check valves follows in static simulations from:
(14)
With this definition the effects of the offset of the acting point as well the deviation from
the normal direction of the hydraulic force are included.
This definition differs to the non-dimensional torque coefficient introduced by Ref [2]
and [5]:
(15)
A different non-dimensional factor for the static torque was introduced by Ref [3] with:
(16)
0.9
0.8
Force Factor (_)
0.7
0.6
0.5
0.4
0 10 20 30 40 50 60
Valve Angle (°)
alfa-5D alfa-loc
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Valve Angle (°)
Fig 9: Local and total flow number alpha ( ) for different valve angles, (K=1/ ²-1)
The effect of higher losses in the partly opened valve can be seen in the growing gap of
the alpha numbers for local and total losses.
Tuerk, Ref [4] performed static measurement of pressure loss at a variety of swing check
valves of different diameter including reverse flow. The presented characteristic is made
up of mean values, Fig.10. Remarkable in work of Tuerk is the lower pressure loss
(higher alpha number) for low valve opening.
The simulated KSB valve differs in its design constructively to the Westark, Ref [6]
valve at the outlet. The difference between both valves is reduced for low valve angles,
Fig. 10.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Valve Lift normed (-)
A set of simulations was performed which differ in the inlet-velocity and the angular-
velocity of the disk. The angular-velocity of the disk was kept constant respectively
depending on the acting forces and moments. The initial inlet-velocity was increased to
2.75 m/s compared to 2.0 m/s in the stationary simulations in purpose to start with a
stable disk position.
70 0
60 -0.2
50 -0.4
40 -0.6
30 -0.8
20 -1
10 -1.2
0 -1.4
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Time (s)
For the simulation of valve closure accelerated by own weight, a mean inlet velocity in
time was: v = 2.75 – 2*t m/s. The valve angle together with the angular velocity is shown
over time in Fig. 12. The simulation stopped at about 6 ° because of numerical grid
problems (the minimum valve angle is 2.51 °). The valve closure starts at a fluid velocity
of about 2.45 m/s when the weight torque exceeds the hydraulic torque. Remarkable is
the gradient of the angular velocity which shows acceleration at the begin and
deceleration near the end of the valve closure. The numerical derivation of the angular
velocity, Fig 13 shows two inflection points. The first inflection point at about 0.247 s
and nearly still 60 ° opening angle can probably be explained by an adjustment of the
growing dynamic friction. From Fig. 13 it could be seen that the added mass effect,
assumed as proportional to the acceleration, increases considerably at starting of the
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6
-0.8 -0.8
-1 -1
-1.2 -1.2
-1.4 -1.4
-1.6 -1.6
time (s)
Fig. 13: Angular velocity and numerical its derivation together with trend line
Torque, Fluid Velocity in Seat and Relative Velocity between Flap and Fluid
3 140
2.5 120
2 100
1.5 80
1 60
0.5 40
0 20
0 10 20 30 40 50 60
-0.5 0
Valve Angle (°)
Fig. 14: Hydraulic torque, fluid velocity in seat and relative velocity
The mean velocity in the seat becomes negative below 18°,Fig 14 but the hydraulic
torque is still positive. This effect is explained by the still positive relative velocity
between fluid and disc. The contour plot, Fig. 15 of the total pressure for a valve angle of
10.4° shows qualitatively a higher pressure backward than forward of the disk. The still
positive pressure difference over the disk is responsible for a positive torque for the
global back flow situation. The relative velocity comprises a mean velocity value of the
disk only i.e. the inner end of the disk (near the axle) moves slower than the outer end of
the disk. The definition of the relative velocity refers to the velocity of the centre of the
Fig. 15: The contour plot of the total pressure for a valve angle of 10.4 °
(7.8 ° opening angle)
The loss factor at the disk, Fig. 16 demonstrates once more the relevance of usage of the
relative velocity definition. Beginning at about 18 ° the velocity in the seat becomes
negative whereas the relative velocity is still positive.
Local Loss Factor K as Function of the Velocity in the Seat and of the Relative Velocity
K-loc Kloc_Vrel/dyn
100000
10000
1000
Loss Factor (-)
100
10
1
0 10 20 30 40 50 60
0.1
Valve Angle (°)
Fig. 16: Local loss factor and relative velocity as function of valve angle
alfa-loc_Vrel alfa-loc_static
0.8
0.7
0.6
0.5
Alfa (-)
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70
Valve Angle (°)
Fig. 17: Flow number (alpha) for static and dynamic simulations
11
5
Deviation (°)
-1 0 10 20 30 40 50 60
-3
-5
-7
Valve Angle (°)
The deviation, Fig. 18 and deflection, Fig 19 of the hydraulic force become also modified
in the dynamic valve closure. The deviation of the hydraulic force falls fast down and
becomes negative below a valve angle of about 37 °. In the stationary disk case the
deviation tends to zero but does not become negative.
25
20
15
Offset (mm)
10
0
0 10 20 30 40 50 60
-5
-10
Valve Angle (°)
In the dynamic case the offset is always positive and varies between 4 % and 8.7 %. The
offset falls down to a relative minimum at about 35 ° and grows versus 10 °. In the static
case the deviation starts for the open valve at the same value and becomes negative
beyond 31 °.
The force factor Ff differs for static and dynamic simulations. The difference reveals
once more the remarkable influence of the valve motion. The non-dimensional factors
derived from static analyses cannot be applied to dynamic analyses without
modifications. The increased drag and dynamic hydraulic torque together with the altered
deviation and offset of the forces also a variable added mass and possibly further other
effects are responsible for the non applicability of the force factor in the dynamic case as
demonstrated by Fig. 20.
Ffac_Vr-dynamic Ffac/static
0.9
0.8
Force factor Ffac (-)
0.7
0.6
0.5
0.4
0 10 20 30 40 50 60
Valve Angle (°)
From the dynamic CFD simulation follows the total dynamic hydraulic torque. The
difference of the total hydraulic torque and the static hydraulic torque was formed to
evaluate the factor CHR. All three variables are shown in Fig. 21. The rotational
hydraulic coefficient CHR becomes really high for the starting closure of the disc and
tends to low values near the end of valve closure where the difference Tdiff tends to zero.
250 120
100
200
80
60
100
40
50
20
0 0
0 10 20 30 40 50 60
Valve Angle (°)
Fig 21: Rotational hydraulic coefficient CHR, total dynamic hydraulic torque
Torque-dyn and the difference Tdiff to the static hydraulic torque
CONCLUSIONS
The application of the CFD-simulation results to the simulation of a swing check valve in
1D flow simulation with finite difference methods is quite straight forward by solving the
equation of movement Eq.(1). The transient fluid velocity at the entry to the valve
follows from the fluid code. With the force factor Ff, Eq. (12) follows the effective force
Feff and further with Eq. (11) directly the hydraulic torque to the disk.
The force factor Ff as well the loss factor Kloc are evaluated by this CFD simulation for
steady flow and also for a transient flow with an gradient of 2.0 (m/s²) of the inlet
velocity. These results are transferable to other geometries as the parameters are non-
dimensional. Conclusions can further been summarized into following:
1. The known effects of deviation and offset of the hydraulic forces acting onto the
disc of a SCV valve are quantified for a static application. The modification of
these effects in the dynamic case is demonstrated.
2. A non dimensional force factor applicable to the static one dimensional simulation
of related SCV is quantified. On the bases of the force factor the critical velocity
which is needed to fully open the SCV can be worked out reliably.
REFERENCES