Case study: the Casappa LVP140 piston pump
The case study was Casappa’s new LVP140 piston pump, a
swashplate piston pump with a maximum displacement of 140
cc/rev (see Fig. 1), capable of working in mobile and stationary
applications, with an open circuit and a hydraulic pressure of up
to 420 bar.
In order to simulate the main fluid dynamic aspects that are
involved in a working cycle, it is necessary to consider the main
volumes inside the pump (Fig. 3), including:
- the variable volume inside each cylinder, called “piston
chamber”
- high- and low-pressure ducts inside the pump housing
- valve plate connection: kidneys, grooves and orifices, whose
shapes and positions define the pump’s timing
Fig. 3 - Main volumes inside the pump
Special attention was paid to the valve plate design because
the transition between the low-pressure and high-pressure
connections and vice versa affects the pump’s performance in
terms of efficiency and noise generation.
In addition to the volumes described above, LVP140 is
characterized by the presence of a noise-reduction feature, called
“pre-compression volume”. This feature makes this pump ideal
for industrial applications where, due to the presence of electric
motors, the pump becomes one of the first sources of noise.
The pre-compression volume is an isolated chamber realized in
the rear cover of the pump: it is connected to the outside only
through an orifice in the valve plate, located in the transition
region from low pressure to high pressure.
Its function is to partially pressurize the fluid inside each piston
before it is connected to the high-pressure line, in order to ensure
a smoother pressure transition. In this way, flow ripple (the
most important source of noise) can be minimized, by reducing
backflows from the delivery to the suction line, and peak flow rates
in the delivery.
The CFD model
The computational fluid dynamics (CFD) model of the pump was
developed using the commercial software Ansys CFX (Fig. 4).
The simulation was time dependent to account for the rotational
and reciprocating movements of the pistons. Fig. 4 shows the
simulated working condition.
Modeling the entire pump, in particular the pre-compression
volume, was especially challenging because of the high pressure
drops (in the order of hundreds of bar) and the small transition
zones: special attention was paid to the meshing phase, using
high mesh refinement in the transition zones in the valve plate,
such as the grooves and orifices, where the highest fluid velocities
are reached.
The rotating and reciprocating motions of the pistons was
managed using the domain motion and mesh deformation features
within CFX. The rotary motion was obtained by setting each piston
domain as rotating, while the reciprocating motion was achieved
through mesh deformation, which involves a displacement
diffusion model with constant mesh stiffness.
Fig. 4 - CFD model
Results and validation
The model was validated by comparing the predicted pressures
and flow rates with those obtained from a well-established 0D
circuital model developed in the Simcenter Amesim environment
(Fig. 5). The lumped parameter model, which has been developed
over the years in collaboration with the Polytechnic of Turin and
validated through experimental tests, is frequently used in the
design phase of Casappa’s components.
In just a few minutes of simulation, the 0D model estimates pump
performance such as piston pressure peaks,
internal backflows, flow ripples, and forces
between internal components. In addition, it can
predict cavitation conditions by simulating the
dynamics of aeration (air release and dissolution
in oil) and of vapor generation. However, since
the fluid was considered as monophase in the
3D model, Amesim’s full cavitation model was
disabled in this study to obtain a consistent
comparison between the two approaches.
Fig. 6 shows the comparison of the pressure
inside a piston chamber and its corresponding
flowrate. The results are perfectly comparable
in all the phases. Owing to the absence of the
gas phase, a non-physical negative pressure is
achieved inside the pump chamber.
Fig. 7 shows the flowrate exchanged between the pre-compression
volume and one piston. As soon as the transition area is uncovered,
oil flows from the pre-compression volume to the piston chamber,
which is at lower pressure (1).
Due to the high pressure drop, the oil jet is particularly intense
and, in a reduced angular phase, is able to pressurize the piston
chamber (2). Therefore, when the piston is connected to the
delivery line, a re-charge flow is established from the delivery to
the pre-compression volume (3-4).
The comparisons reported show an
excellent correspondence between the
two models. Although the 0D model
remains the best tool in terms of
computation time and ease of reaching
convergence, a three-dimensional
approach is the most advanced and
accurate method. A CFD model can
predict the spatial distribution of
pressure and velocity within the pump
allowing the following to be explained:
filling-emptyng dynamics, flow field
distribution, identifying vorticity and
stagnation areas (Fig. 8).
It was also useful in understanding the
erosion phenomena that was observed
on the test bench under extreme working
conditions.
For example, Fig. 9 shows the velocity
contour and its corresponding normalized
vectors within the piston chambers while
it begins opening the connection with
the pre-compression volume as it moves
from the suction region.
Due to the high pressure drop, a highspeed
oil jet with a component of nearaxial
velocity hits the surface of the cylindrical piston seat. Only
later, when the piston intercepts the delivery groove, the flow is
partially diverted.
The oil jet reaches high speeds, therefore the impact against the
cylinder surface probably generates the erosion visible on the
surface of the actual component subjected to the endurance test.
Conclusions
In this work, a CFD model of an axial piston pump, operating
at high pressure, was created to investigate the fluid dynamic
aspects involving the main volumes inside the pump.
The model was validated using a quantitative comparison of
predicted pressure and flow rates with those obtained from a
validated lumped parameter model; the comparison shows an
excellent correspondence between the two models.
Although the 0D model remains the best tool in terms of
computation time and ease of reaching convergence, the 3D
model proved to be the most advanced and accurate method to
investigate the fill-empty dynamics, flow field distribution, and the
erosion phenomena observed on the test bench under extreme
working conditions.
The activity does not end here; the next steps will include the
simulation of the most critical working conditions, i.e. the highest
speed and pressure, and the introduction of a full cavitation
model that will consider the release of air and steam, in order to
investigate the spatial distribution of gas inside the pump.
