Vous êtes sur la page 1sur 10

Proceedings of the ASME 2016 International Mechanical Engineering Congress and Exposition

IMECE2016
November 11-17, 2016, Phoenix, Arizona, USA

IMECE2016-65624

3D DYNAMIC SIMULATION OF A FLOW FORCE COMPENSATED PRESSURE RELIEF VALVE

Giorgio Altare Massimo Rundo Micaela Olivetti


Politecnico di Torino Politecnico di Torino OMIQ s.r.l.
Energy Department Energy Department Milan, 20135, Italy
Turin, 10129, Italy Turin, 10129, Italy olivetti@omiq.it
giorgio.altare@polito.it massimo.rundo@polito.it

ABSTRACT Fc opening force on the cone due to the pressure


The paper deals with the 3D and 0D simulation of a Fjet flow force
conical popped pressure relief valve with flow force F0 spring preload
compensation. The commercial CFD code PumpLinx® was Fp resultant of the pressure forces
used to create a dynamic model of the valve and the interaction k spring stiffness
between the poppet dynamics and pressure field was taken into kjet equivalent stiffness of the flow forces
account. The model is able to determine the equilibrium m poppet mass
position of the poppet in order to estimate the regulated p regulated pressure
pressure as function of the flow rate. A good agreement with p* cracking pressure
the experimental data was found in the evaluation of the effect Q volumetric flow rate
of the flow force compensation. Once validated, the CFD code x poppet lift
was used to study the influence of the deflector geometry on the  fluid velocity
opening force. Moreover it was also used for determining some
 cone half-angle
proper data to be supplied as input to a lumped parameters
model of the valve. The tuning of the 0D model involved the  flow number
discharge coefficient and the flow force. For the evaluation of p pressure drop
the flow force compensation, a lookup table was calculated by  kinematic viscosity
the CFD code and then interpolated in the 0D model as  fluid density
function of the poppet displacement and of the flow rate. g gas density
NOMENCLATURE l liquid density
A flow area v vapour density
c damping coefficient
Cd discharge coefficient INTRODUCTION
Cd,max maximum discharge coefficient In a hydraulic valve the steady-state performances are
d diameter of the seat and of the damper strongly affected by the flow forces, which are generated by
dh hydraulic diameter the change of fluid momentum inside the valve. The flow
dr diameter of the poppet rod forces are constituted by three terms: a steady-state component,
f correction factor for the flow force proportional to the flow rate and to the fluid velocity, and two
fg gas mass fraction unsteady terms, functions of the flow rate derivative and of the
fv vapour mass fraction spool acceleration respectively. The first term is by far the most
important and tends always to reduce the flow area. It can be

1 Copyright © 2016 by ASME


demonstrated that, with a constant pressure drop across the However there is no evidence in the open literature of a study
valve, the steady-state force is proportional to the spool carried out with PumpLinx on a conical poppet valve.
displacement, hence it behaves as a virtual spring. Moreover such type of valve is not present among the built-in
In the electro-actuated direction control valves the presence of templates available in the software.
the flow force entails the use of large solenoids, in order to In this paper the procedure for modelling conical poppet
increase the activation force required to move and to keep in pressure relief valves with flow force compensation in
position the spool. In the pressure control valves the PumpLinx is described and the reliability of the simulation is
consequence is a worsening of the precision, since the demonstrated through experimental tests.
controlled pressure becomes highly influenced by the valve
opening and therefore by the flow rate. In particular in a VALVE DESCRIPTION
pressure relief valve the undesired effect is the increment of the Figure 1 shows a section view of the relief valve used for
regulated pressure as the flow rate discharged by the valve the present study. The movable element is constituted by a
increases. conical poppet provided with a damper, which can slide within
Some techniques exist for the compensation of the flow forces, a fixed sleeve with 4 radial holes. The valve is maintained
most of them are applied to spool valves [1, 2]. A method of normally closed by an adjustable spring, while the opening
compensation in conical poppet valves consists in deflecting force is generated by the inlet pressure at port P acting on the
the outgoing flow in order to reduce the change in the axial left surface of the cylindrical damper, thanks to the radial
momentum [3]. This can be done by shaping the rear edge of clearance with the sleeve.
the poppet in order to obtain a radial outlet. Moreover if the jet
is deflected backwards it is possible to obtain a net resulting
opening force, which can also compensate the closing force
due to the increment of the spring compression. However a risk
of overcompensation or of instability exists.
Although the qualitative effect of the force compensation can
be described analytically by applying to a proper control
volume the 2nd Newton’s law, the quantitative evaluation of
the net force acting on the poppet can be performed only using
CFD analyses. In fact the axial component of the opening force
is strongly dependent on the outgoing jet angle, which cannot
be determined a priori simply based on the poppet geometry. FIGURE 1: SECTION VIEW OF THE RELIEF VALVE
While several papers are available on the flow force studies in
spool valves, the literature focusing specifically on the CFD A flow deflector is used to compensate the steady-state
analysis of the flow force compensation in conical poppet component of the flow force, which arises on the conical
valves is quite limited. poppet due to the increment of the fluid velocity upstream of
The first work was carried out by Vaughan et al. [4]. The the minimum flow area. The aim of the deflector is to generate
authors studied with 2D simulations by means of StarCD® the an opening force by means of the deviation of the fluid jet.
force on the poppet with different geometries and for different The main geometric quantities of the valve under study are
fixed valve openings. listed in Tab. 1.
Sørensen [5] used CFD analyses for studying two methods for TABLE 1: MAIN GEOMETRIC PARAMETERS OF THE VALVE
the flow force compensation.
Quantity Value Unit
Andersen et al. [6] related the geometric parameters to the flow
force of a conical poppet with a rim on the rear edge. seat/damper diameter d 8 mm
A valve provided of a plate able to deflect backwards the jet cone half-angle  20 degrees
was studied by Lisowski and Domagala [7] using ANSYS
spring stiffness k 42 N/mm
Fluent®. The interaction between the poppet position and the
pressure field was studied in 2D by Huguet [8]. max cracking pressure p* 75 bar
Finally a complete 3D dynamic model with mesh deformation
developed with Fluent is described in [9]. THEORETICAL ANALYSIS
Recently some analyses of valves have been carried out with With reference to Fig. 2, where a qualitative section view
PumpLinx®. For example Ni et al. [10] studied the flow force of a conical poppet valve is shown, it is evident that the angle
compensation on a spool type pressure relief valve used in of the jet leaving the valve can be approximated with the half-
aeronautic applications. Frosina et al. [11] used the same angle of the cone . Due to the fact that the jet angle at the
simulation environment for a two-port flow control valve. outlet differs from the angle at the inlet, a flow force Fjet arises:

2 Copyright © 2016 by ASME


F jet = Q cos  (1) In this way the only non-ideality of the valve is due to the
stiffness of the spring. However it is possible to go further and
Based on the Bernoulli’s law, Eq. (1) can be also rewritten as: to generate an opening force, which is able also to compensate
F jet = 2C d  A  x   p  cos  (2) the incremental force of the spring, in order to obtain a quasi-
ideal valve characteristic. The principle is to deviate backwards
To a first approximation the flow area, function of the poppet the jet with an angle  greater than 90° (Fig. 3b).
displacement x, can be expressed as: The difficulty during the design phase is due to the fact that the
A  x  =   d  x  sin  (3) jet angle could not coincide with the trailing edge angle of the
deflector. Moreover the intensity of the force is also function of
the fluid velocity where the jet leaves the deflector, hence it
depends on how much the jet spreads. In addition if the angle
 tends to 180°, the jet is partially reflected inwards by the seat
of the poppet and a vortex is generated.
Therefore to predict with a good accuracy the valve behaviour
a CFD simulation is mandatory.

3D TRANSIENT MODEL
Mesh construction
Starting from the CAD model of the valve, the wet
surfaces of the fluid volume were extracted in STL format. The
generation of the mesh in PumpLinx is usually performed
FIGURE 2: SCHEMATIC OF A NON-COMPENSATED POPPET VALVE through specific templates for several types of pumps and
some kinds of valves. However, due to the lack of a dedicate
Since the aim of a pressure relief valve is to maintain constant template for conical poppets, to simulate the motion of the
the upstream pressure, it can be assumed that, when the valve movable element and the consequent mesh deformation, a
regulates, the pressure across the metering edge is almost manual approach was used. Hence the generic module axial
constant and equal to the cracking pressure p* of the valve. valve was employed and the computational domain was split in
The consequence is that the flow force is proportional to the subdomains with different types of grids: fixed, deformable
poppet displacement x and can be considered as the elastic and sliding. In Fig. 4 the volumes subdivision is reported.
force of a virtual spring with stiffness:
k jet = 2C d    d  p  sin   cos  (4)

For high values of the cracking pressure the stiffness expressed


by the Eq. (4) can be comparable or higher than the elastic
constant of the spring. This makes the regulated pressure
strongly dependent on the poppet displacement and therefore
on the discharged flow rate.
The most common method to compensate the flow force is to
add a lip at the rear edge of the poppet, in order to obtain a
radial outlet (Fig. 3a).

FIGURE 4: VOLUMES SUBDIVISION


As far as the sliding volumes are concerned, the cells simply
move along the poppet axis maintaining their original shape.
The displacement is calculated by solving at each time step the
equation of motion (5):
mx·· + cx· + kx + F 0 = F p (5)
FIGURE 3: EXAMPLES OF FLOW FORCE COMPENSATION
where m is the poppet mass, c is the damping coefficient, F0

3 Copyright © 2016 by ASME


and k are the spring preload and stiffness respectively and Fp Model refinement
the resultant of the forces acting on all surfaces due to the An important aspect regarding the mesh is the cell density
pressure distribution. around the metering edge and along the jet path downstream
Instead for the deformable volumes, the cells are stretched from the metering edge up to the deflector surface.
along the motion direction. To build this type of mesh, three The former zone has a significant effect on the calculation of
different surfaces must be set: the pressure field along the cone and therefore on the flow
• Valve end, identifies the fixed surfaces, hence the nodes force. The latter region affects the calculation of the force
lying on them are locked. produced by the deflector.
• Valve, identifies the mobile surfaces, therefore at each time In Fig. 5 the regions with different cell refinements are shown.
step the nodes belonging to them shift by a quantity calcu-
lated by the Eq. (5), being anchored to the surfaces.
• Cylinder, whose axis identifies the direction of the transla-
tion of the Valve elements and at each time step the nodes
slide along the cylindrical surface.
All volumes are meshed separately, hence must be connected
afterwards by means of mismatched grid interfaces (MGI)
[12]. In this way the overlapped surfaces are identified as flow
areas. In order to reduce the number of cells, only the portion
near the deflector of the spring chamber was meshed, and to
simplify the mesh the spring volume was excluded. Finally the
grid in the damper gap was obtained using the built-in surface
mesh generator module, which permits to generate a regular
mesh in very narrow gaps. Since the aim of this study was to
obtain the steady-state curves, it was preferred to increase the
gap size from the original value of 0.05 mm to 0.3 mm and to FIGURE 5: HIGH DENSITY CELLS ZONES
set in the translational dynamic module a damping coefficient
to be tuned in order to reduce the transient time. To determine the proper number of cells in the region C of the
minimum flow area, a grid independence analysis was carried
Physical models and numerical scheme out by means of steady-state simulations at fixed poppet lift (1
The governing equations are discretized with the finite mm) and with a flow rate of 50 L/min. In Fig. 6 the pressure
volume method. The model used to simulate the cavitation and drop P-T is plotted as function of the number of cells along the
aeration phenomena is the Equilibrium Dissolved Gas, based radial direction in correspondence of the metering edge. It is
on the model developed by Singhal et al. [13], where the evident that the discriminating number is around 15.
density of the mixture is evaluated as:
80
f fg 1 – fv – fg
--1- = ----v- + ----- + ----------------------- (6)
pressure drop [bar]

  v  g l
75
being fg and fv the mass fraction of the gas (air) and of the oil
vapour respectively, while v, g and l are the densities of the
70
vapour, of the gas and of the liquid. The total mass of air is
maintained constant during the simulation, while dissolved and
separated fractions are calculated as function of the 65
equilibrium value given by the Henry's law and the local 0 10 20 30 40 50

pressure. For the present study it was found that the cavitation number of cells in the radial direction
model must be activated, since in some regions downstream
from the metering edge the local pressure can assume negative FIGURE 6: MESH INFLUENCE IN THE METERING EDGE
values. As far as the mesh geometry along the jet path is concerned,
To avoid this unphysical behaviour, which can modify the some dynamic simulations were performed to determine the
direction of the jet and as a consequence the force on the equilibrium position of the poppet for different refinement
deflector, the aeration and evaporation phenomena must be levels of the grid. Figure 7 reports the regulated pressure at the
simulated in order to limit the minimum pressure. port P as function of the total number of cells.
The standard turbulence model k-was used. Finally the first 6
In particular the point 1 with about 1.5  10 cells was obtained
order upwind interpolation scheme was applied.
with a coarse grid in both regions A and B, points 2, 3 and 4

4 Copyright © 2016 by ASME


with a higher refinement in the region A and three different EXPERIMENTAL TESTS
levels in the zone B. Finally the point 5 with about 6 million Test rig layout
cells was calculated by applying also to the region A the The valve was tested in order to determine its steady-state
maximum refinement used in the region B for the point 4. The flow-pressure characteristic with three different cracking
6
mesh independence was reached with about 2.8  10 cells pressures. The hydraulic scheme of the test rig is shown in Fig.
(configuration shown in Fig. 5). 8. The flow rate, supplied by a variable displacement pump
provided with an absolute pressure limiter, is controlled by a
80 two-port flow control valve. A turbine flow meter KEM-
1 KUPPERS HM11E with a measuring range 6-60 L/min
inlet pressure [bar]

78
measured the volumetric flow rate at the valve inlet. Two
miniature absolute pressure transducers GS XPM were
2 mounted directly on the valve. Both transducers have a
76
linearity error of  0.25% , while the measuring range is
3 4 5 respectively 0-100 and 0-20 absolute bar.
74

72
1 2 3 4 5 6
total number of cells 6
×10

FIGURE 7: INFLUENCE OF THE TOTAL NUMBER OF CELLS

Parameters setting
As far as the oil properties are concerned, in Tab. 2 the
values used in the model are reported. However it was verified
that an amount of air of 9% in volume leads to the same results.
TABLE 2: PARAMETERS SET IN THE MODEL

property set value


density 866 [kg/m3]
dynamic viscosity 0.0415 [Pa s]
bulk modulus 1.7 109 [Pa] FIGURE 8: HYDRAULIC CIRCUIT OF THE TEST RIG
total gas fraction 2% in volume A photo of the test rig is shown in Fig. 9, where it is possible to
dissolved gas ref. pressure a 4
2.2 10 [Pa] see from right to left the flow meter, the valve under test, the
filter and the heat exchanger. The working fluid used for the
a. according to the Henry’s law, this value corresponds to the test was an ISO VG 46 DTE25 oil.
equilibrium condition of 2% gas volume in mineral oil.

A flow rate was imposed as boundary condition at the inlet


port, while the atmospheric pressure was set on the external
surface of the cartridge. The preload of the spring was set in
order to obtain the desired cracking pressure as follows:
2
d p- (7)
F 0 = --------------
4

The valve characteristic was constructed as a sequence of


steady-state pressure drop P-T values for different imposed
flow rates. To reduce the CPU time and to avoid the excessive
stretching of the cells in the region B’ (see Fig. 5), a simulation
with a fixed opening close to the expected value of the
equilibrium position was calculated for each steady-state point.
Then such solution was used as initial condition for the
transient simulation. FIGURE 9: PHOTO OF THE VALVE ON THE TEST RIG

5 Copyright © 2016 by ASME


Test description If the curves with no compensation are considered, it is evident
Tests were performed with an oil temperature in the range that the regulated pressure increases significantly as the flow
35-40 °C. To heat the oil, the flow control valve was rate is incremented. At the maximum flow rate the deviation of
maintained fully open, so that the pump could work at the the pressure imposed by the valve with respect to the
maximum displacement, in order to supply the maximum flow theoretical value is more than 20 bar.
rate. During this warm-up procedure, the oil temperature The valve behaviour is substantially modified by the deflector.
increased due to the power dissipation in the pressure relief It can be observed that with a low cracking pressure the
valve under test. Once reached the desired temperature, the regulated pressure is still increasing with flow rate, even if the
flow rate was reduced to very low values of the order of 1 L/ change of the slope is already evident. This means that the
min and the cracking pressure of the valve was set by reading force generated by the deflector is not enough to compensate
the measurement of the pressure transducer P1. After that the completely the increment of the force due to the spring
flow rate was increased with steps of about 5-8 L/min and in stiffness and to the flow force. In fact the valve works with
each step after about 10 s the flow and pressure values P1 and high openings and the consequence is a low flow velocity
P2 were recorded. The time required to obtain a complete against the deflector.
flow-pressure curve was about 1-1.5 minutes for each value of On the contrary as the cracking pressure increases the
the cracking pressure. To cool the oil for the next test, the compensation effect becomes more and more effective. In fact,
piloted relief valve PRV was vented, so that the oil could pass due to the higher pressure, at equal flow rate the poppet lift is
through the heat exchanger. Several tests were preliminary lower and as a consequence the fluid velocity in the vena
made to assess the repeatability, with increasing and decreasing contracta will be higher. In this way more kinetic energy will
flow rate and also with the valve discharging directly to the be converted in pressure on the surface of the deflector. For the
atmosphere, i.e. without any back-pressure induced by the analysed geometry, at the cracking pressure of 75 bar the flow
filter F2. Overall, the maximum variation in terms of measured force compensation is even excessive, since the slope of the
pressure drop across the valve was about 1 bar at equal flow. characteristic becomes negative (overcompensation). The
maximum error in the evaluation of the regulated pressure,
MODEL VALIDATION among all analysed operating conditions, is about 2.3 bar.
For the validation of the model, the experimental flow- In Fig. 11 the velocity field at 75 bar and 40 L/min is shown:
pressure curves at three cracking pressures of the valve (14, 46 the deviation of the flow is evident, even if the mean jet angle
and 75 bar) have been contrasted with some simulated steady- is slightly smaller than the trailing edge angle of the deflector
state points in Fig. 10. The maximum flow rate was limited by due to the inertia of the fluid. For lower flow rates or lower
the volumetric efficiency of the pump, decreasing with the cracking pressures the jet angle tends to be equal to the
pressure. In the same figure the simulations performed without geometric angle of the deflector.
the deflector (with the legend “no def”) are also reported. In
this case a comparison with the experimental data is not
possible, since the deflector works also as spring seat.

60

14 bar 46 bar 75 bar


50
flow rate [L/min]

40

FIGURE 11: VELOCITY FIELD AT 75 BAR AND 40 L/MIN


30
The effectiveness of the flow deflector can be visualised in Fig.
12. The conversion of the kinetic energy of the jet induces a
20
pressure increment on the surface of the deflector up to 10 bar.
In the same figure the locations of the pressure transducers are
experim.
10
simulated
also indicated. The regions where cavitation and air release
sim. no def occur are clearly visible in Fig. 13. The largest region is
0 located in the middle of the recirculation vortex. Moreover
10 20 30 40 50 60 70 80 90 100
another zone is detected along the profile of the poppet in
pressure difference P-T [bar] correspondence of the change of the cone angle. It was checked
that without the cavitation model the pressure in that region
FIGURE 10: FLOW-PRESSURE CHARACTERISTICS can reach an unphysical value up to -20 bar, which alters the

6 Copyright © 2016 by ASME


velocity field. It was found that in this case the deviation of the valve without the cavitation module enabled even for high
calculated pressure with respect to the experimental value values of fluid velocity. In Fig. 15 the variation of the regulated
increases of about 3 bar. pressure difference P-T with respect to the cracking pressure at
For lower jet velocities, for instance with the lower cracking 40 L/min is plotted as function of the deflector angle. A
pressure, the recirculation vortex is not generated and the positive value of the pressure variation implies that the opening
pressure remains positive even if the cavitation module is force is not enough to balance completely the flow force and
deactivated. the spring force increment, while negative values indicate an
overcompensation.

a) b)

 = 135°  = 90°

FIGURE 14: VELOCITY FIELD WITH DIFFERENT DEFLECTOR ANGLES


It can be noticed that the pressure variation is almost linear
with the deflector angle and higher flow deflections are more
FIGURE 12: PRESSURE FIELD AT 75 BAR AND 40 L/MIN effective at higher pressure levels. This is justified by the
increment of the jet velocity due to a lower flow area.

15
pressure drop P-T [bar]

10 46 bar
75 bar
5

-5
80 100 120 140 160 180

deflector angle θ [deg]


FIGURE 13: GAS FRACTION AT 75 BAR AND 40 L/MIN
FIGURE 15: INFLUENCE OF THE DEFLECTOR ANGLE
Influence of the deflector geometry The jet angle is also highly influenced by the depth of the
Since the flow force compensation is generated by the deflector. In Fig. 16a the fluid path is shown when the deflector
deflection of the jet, the deflector effectiveness is influenced by depth is the half of the original value (0.75 mm against 1.5
the shape of the lip. In order to assess the variation of the mm). In this case the flow force compensation is less effective,
opening force, the angle  in Fig. 3 was reduced from the since the jet leaves the deflector with an angle comparable to
original value of 170° up to 90°. The simulations were the geometry of Fig. 14a.
performed at the maximum cracking pressure and with a flow At equal depth, the fluid velocity where the jet leaves the
rate of 40 L/min, since in this operating condition the force on deflector can be slightly incremented with a proper shape. In
the deflector achieves the highest value. Figure 14 shows the Fig. 16b an arc of circle is used to connect the conical surface
velocity field of the jet for two different geometries. It is with the trailing edge. However the variation in terms of
evident that in these cases the jet angle is almost coincident regulated pressure with respect to the original solution is less
with the deflector angle. In case of flat geometry (Fig. 14b) it than 1 bar, therefore it can be concluded that the internal shape
was found that the regions where the pressure falls below the of the deflector is not so influential.
atmospheric pressure are very limited, due to the absence of
the recirculation vortex. In this case it is also possible study the

7 Copyright © 2016 by ASME


being p the pressure at port P, dr the diameter of the poppet rod
and Fjet is calculated with the Eq. (2) using the exact
expression of the flow area as function of the poppet lift.
The coefficient f is a correction factor used for tuning the flow
force. The pressure drop across the radial holes in the sleeve
has been neglected.

a) b) Tuning of the discharge coefficient


The discharge coefficient Cd is function of the flow
number defined as [3]:
d h 2p
FIGURE 16: INFLUENCE OF THE DEFLECTOR SHAPE  = ----- ---------- (9)
 

0D MODEL where dh is the hydraulic diameter and  the kinematic


Description viscosity. The discharge coefficient increases with the flow
Although the 3D model was proved to be reliable in number and tends asymptotically to a maximum value Cd,max.
predicting the real regulated pressure, many practical The relationship between Cd and  can be approximated
applications require a very fast lumped parameters model to be analytically by the Eq. (10):
used inside a fluid power system, with the possibility to modify 2
some parameters, first of all the spring characteristic. For this C d = C d max tanh  ------ (10)
 c 
purpose a 0D model in LMS Imagine.Lab Amesim® was
developed. This model uses a lookup table, derived from the where  c is the critical flow number, which represents the
3D simulations, to calculate the force acting on the deflector. In transition between the laminar and turbulent regime. To define
this view the CFD model becomes a tool for tuning the 0D the Eq. (10) the value pair Cd,max and  c must be determined.
model of the valve. Moreover the CFD code can be useful for
correcting the theoretical Eq. (4) for a better evaluation of the In Fig. 18 the discharge coefficients calculated in the operating
closing force due to the pressure distribution along the cone. points of Fig. 10 are shown.
Finally, another use of the 3D model is the evaluation of the It can be noticed that for critical flow numbers greater than
discharge coefficient, which has an influence of the flow rate about 600, the discharge coefficient remains constant with
through the valve. These last two uses make the 3D model small differences between low and high cracking pressures,
useful even in valves without the flow force compensation. corresponding to high and low poppet lifts respectively.
The model in the LMS Amesim environment is shown in Fig.
17. 0.8
discharge coefficient [-]

0.75

0.7

3D model - 75 bar
3D model - 46 bar
0.65 3D model - 14 bar
0D model

0.6
0 200 400 600 800 1000
FIGURE 17: LMS AMESIM MODEL flow number λ [-]

The flow rate is imposed through an ideal flow source, whose FIGURE 18: TUNING OF THE DISCHARGE COEFFICIENT
value, along with the poppet position, is used as input for a 2D
table interpolation block for the evaluation of the deflector In the same figure the analytic function (10) with Cd,max = 0.76
force. In the conical poppet submodel the opening force is and  c = 300 is also superimposed. Since the influence of the
evaluated as: discharge coefficient is more important for small values of the
d –
2 2
dr
flow area, it was decided to fit the points corresponding to the
F c = p -----------------  – f  F jet (8) maximum pressure. In fact a variation of the discharge
4
coefficient from 0.76 to 0.74 leads to a modification of the inlet

8 Copyright © 2016 by ASME


pressure of more than 1 bar at 75 bar, while at 14 bar the
variation is one order of magnitude smaller.

Tuning of the flow force


For determining the correction factor f to be used in the
Eq. (8) the flow force was estimated by means of the 3D model
as function of the poppet lift for the three cracking pressure
values. It was evaluated as difference between the closing force
acting on the right surface of the damper and the opening force
acting on the conical surface of the poppet.
Figure 19 shows the results of the 3D model along with the
flow force calculated through the lumped parameters model
with the indication of the gain value f used to best fit the CFD
data. It was found that at medium-high setting a minimal
FIGURE 20: 3D MAP OF THE FORCE ON THE DEFLECTOR
correction is needed, while at low cracking pressure the
theoretical model slightly overestimates the flow force, hence a
Steady-state simulations
gain of 0.9 must be applied.
After the tuning, the 0D model was able to reproduce the
steady-state curves with negligible CPU time. In Fig. 21 the
100
f = 0.95 simulated flow-pressure curves are contrasted with the
3D model - 75 bar
3D model - 46 bar f = 0.97 experimental data. The simulations were performed by
80
3D model - 14 bar imposing a slope of the flow rate at the inlet of the valve from 0
to 50 L/min.
flow force [N]

0D model
60
60
40
50
flow rate [L/min]

20
f = 0.9 40

0 30
0 0.5 1 1.5 2
poppet lift [mm] 20 exp - 14 bar
exp - 46 bar
exp - 75 bar
FIGURE 19: TUNING OF THE FLOW FORCES 10 0D model

Flow force compensation 0


0 10 20 30 40 50 60 70 80
To simulate the effect of the deflector, the force acting on pressure difference P-T [bar]
the poppet was supplied as a lookup table to the 0D model as
function of the poppet position and of the flow rate. FIGURE 21: FLOW-PRESSURE CURVES WITH THE 0D MODEL
A 3D map, shown in Fig. 20, was created in MATLAB® by The precision obtained is largely satisfactory for a 0D model.
interpolating with a polynomial surface the steady-state points The accuracy is slightly worsened if the correction factor f for
of Fig. 10, indicated with a dot on the graph. Two additional the flow force is maintained constant and equal to 1, however
points, in the origin and at maximum lift, were added to impose the maximum difference between the experimental and the
that the force is null when the flow rate is zero. simulated pressure increases of just 1 bar.
The surface was generated with a third-degree polynomial Once tuned, the 0D model can be used to simulate the valve in
interpolation with respect to the poppet lift and with a different operating conditions with results comparable with a
quadratic polynomial with respect to the flow rate. Then a 3D simulation, as long as the geometry of the deflector remains
squared matrix 20x20 was created and supplied to LMS the same. For example with a different preload of the spring
Amesim to be interpolated linearly. corresponding to a cracking pressure of 30 bar and a flow rate
30 L/min the difference between 3D and 0D models is 0.33 bar.

9 Copyright © 2016 by ASME


CONCLUSION [4] Vaughan, N.D., Johnston, D.N. and Edge, K.A., 1992,
The assessment of the effectiveness of the flow force “Numerical simulation of fluid flow in poppet valves”,
compensation in poppet hydraulic valves requires the Proceedings of the Institution of Mechanical Engineers,
development of a CFD model with deformable mesh able to Part C: Journal of Mechanical Engineering Science,
calculate the poppet dynamics. 206(2): 119-127.
This work reports a quite novel application of PumpLinx for [5] Sørensen, H.L., 1999, “Experimental and numerical
the evaluation of the steady-state flow-pressure characteristic analysis of flow force compensation methods for hydraulic
of flow force compensated conical poppet relief valves. Some seat valve”, The sixth Scandinavian international
critical aspects for the construction of the model have been conference on fluid power”, Tampere, Finland.
solved.
[6] Andersen, T. O., Hansen, M. R. and Sørensen, H. L., 2003,
The correct evaluation of the deflector force can be obtained
“Using CFD to Establish a Correlation between Design
only using a good cell refinement along the entire path of the
Parameters and Performance Characteristics for Seat
jet downstream from the metering edge. In the operating
valves”, Proceedings of 1st International Conference on
conditions with a high fluid velocity, the cavitation model must
Computational Methods in Fluid Power Technology,
be activated to avoid negative pressure values in the middle of
Melbourne, Australia.
the recirculation vortex, which can alter the calculation of the
force on the deflector. [7] Lisowski, E. and Domagala, M., 2003, “Determination of
As far as the comparison with the experimental results is Relief Valve Characteristics by the Use of CAD System
concerned, in the analysed operating range the maximum error and CFD Tools”, Proceedings of the 1st International
in the evaluation of the regulated pressure is about 2.5 bar at Conference on Computational Method in Fluid Power
the maximum flow rate. Technology, Melbourne, Australia.
Since the force compensation is originated by the deviation of [8] Huguet, D., 2004, “Dynamic mesh modelling of a direct
the fluid jet, the geometry of the deflector lip plays a acting relief valve”, Proceedings of the 3rd FPNI - PhD
fundamental role. The maximum force is obtained when the jet Symposium on Fluid Power, Terrassa, Spain.
is deflected backwards and this effect can be attained with a
high geometric angle of the lip combined with a sufficient [9] Finesso, R. and Rundo, M., 2016, “Numerical and
depth of the deflector. On the contrary the link between the experimental investigation on a conical poppet relief valve
conical poppet and the deflector has a negligible influence. with flow force compensation”, International Journal of
For the construction of the lumped parameters model a crucial Fluid Power.
aspect is the generation of the 2D map for the reproduction of [10] Ni, W. B., Heitz, S., Bartholme, D. and Cass, M., 2011,
the deflector force. To cover the entire working range with a “Compensation Force CFD Analysis of Pressure
quite limited number of CFD simulations, it is convenient to Regulating Valve Applied in FMU of Engine and System
create a polynomial interpolation surface with a proper degree. Controls”, SAE Technical paper 2011-01-2641.
Above all at high cracking pressures the results are very
sensitive to the poppet position, therefore it is necessary a very [11] Frosina, E., Senatore, A., Buono, D. et at., 2014,
good interpolation of the force values. “Improving the performance of a two way flow control
Being the CFD model able to simulate the valve dynamics, valve, using a 3D CFD modeling”, Proceedings of the
future work could evaluate its possible application for stability ASME 2014 International Mechanical Engineering
analyses. Congress and Exposition, Montreal, Canada.
[12] Ding, H., Visser, F. C., Jiang Y. and Furmanczyk, M.,
ACKNOWLEDGMENTS 2009, “Demonstration and validation of a 3D CFD
Authors wish to thank Hui Ding (Simerics Inc.) for the simulation tool predicting pump performance and
crucial support. cavitation for industrial applications”, Proceedings of the
ASME Fluids Engineering Division Summer Meeting,
REFERENCES Vail, USA.
[1] Blackburn, J.F., Reethof, G. and Shearer, J.L., 1960, “Fluid [13] Singhal, A., Athavale, H. and Jiang, Y., 2002,
Power Control”, Technology Press of M.I.T, New York - “Mathematical Basis and Validation of the Full Cavitation
Wiley, London. Model”, Journal of Fluids Engineering, 124: 617-624.
[2] Merritt, H.E., 1967, “Hydraulic Control Systems”, John
Wiley & Sons, London.
[3] McCloy, D. and Martin, H.R., 1980, “Control of Fluid
Power: Analysis and Design”, John Wiley & Sons.

10 Copyright © 2016 by ASME

Vous aimerez peut-être aussi