In this article, the forced lubrication system of an 18-cylinder
engine is studied by means of a fluid dynamic model of the
lubrication channels. Particular attention is paid to the pressure
trend and behavior in the oil channels and in the vicinity of the
big-end bearing (BEB). The purpose of the study is to evaluate
the effect of a change in the bearing design on the surrounding
Due to the nature of the system, a traditional finite volume CFD
approach would be unfeasible due to the many moving parts
and the complexity of their motion. Therefore, a meshless CFD
approach based on the moving particle simulation (MPS) method
Moving particle simulation (MPS) method
The moving particle simulation (MPS) method, originally called
moving particle semi-implicit method, was conceived by Prof.
Koshizuka of Tokyo University in 1996 . It is a meshless method
for resolving fluid motion by solving Navier-Stokes equations in
incompressible conditions. It discretizes the fluid domain using
particle elements, and each particle becomes a computational
node carrying information about position, velocity, temperature,
and all variables of interest. Since the core of the computation
is the particles, the Navier-Stokes equations are written in a
Lagrangian framework, in contrast to traditional mesh-based
methods, which use the Eulerian approach. In the Lagrangian
approach, particles are tracked and followed during their motion,
whereas in the Eulerian approach the viewpoint is fixed to the
computational grid (see Fig. 1).
Fig. 1 - Difference between the Eulerian approach (a) and the Lagrangian approach (b).
Its meshless nature makes the MPS approach highly flexible and
well-suited to applications involving moving parts where mesh
generation and deformation using a mesh-based method can be
In this work, the Particleworks MPS solver developed by
Prometech Software Inc. is used to analyze the oil flow in the
engine lubrication system.
Model description and boundary conditions
The complete engine consists of 18 pistons with connecting rods,
mounted on a crankshaft. The crankshaft is supported by ten main
bearings (MB) with their respective main bearing caps; internal
channels running through the engine components distribute the
lubricating oil to all the bearings (see Fig. 2). The bearings are
of the hydrodynamic type. The crankshaft rotates at a constant
speed, converting the reciprocating motion of the pistons into
Oil enters the system through the main bearing caps, lifting the main
bearings; then it is delivered to the big-end bearings (BEB) via the
crankshaft oil channels. Finally, it reaches the small-end bearings
(SEB) through the connecting rod oil channels. Figs. 2 and 3 show
the oil path and diagram of the internal lubrication system.
Fig. 4 shows a cross-section of the crankpin and its oil channel,
the lower part and stem of the connecting rod, and the BEB that is
The lubricating oil from the crankpin oil channel reaches the SEB
by flowing through the clearance between the pin and BEB, its
groove, and then the oil channels inside the connecting rod. The
rotation of the crankshaft, together with the alternating movement
of the connecting rod and the engine oil pump, ensure sufficient
oil flow through the channels that feed all the bearings. The
moving particle simulation only takes into account the first three
turns of the crank, six pistons, and four main bearing caps. An
inflow with the prescribed volumetric flow rate is applied to the
base of each main bearing caps, and outflow rate conditions
are applied in the bearing regions to simulate oil leakages that
will then be collected in the oil sump. Fig. 5 summarizes the oil
sources and leaks.
Fig. 5 - Distribution of inflow and outflow boundary conditions.
The same simulation method is applied to two different BEB
geometries. The crankshaft rotation speed, inlet and outflow rates
are kept constant throughout the simulation.
The particle size and the integration time step are the two main
parameters that play a key role in the simulation. The particle size
must be small enough to address the smallest gap of interest in
the geometry. A 2mm particle size is chosen in the MPS model to
capture the smallest passages in the channels and grooves. The
film of lubricating oil between the bearings and crankshaft pins
is not the scope of the current work. Particle size affects solution
accuracy and simulation time (smaller particles increase solution
accuracy and computation time).
The second key parameter is the integration time step,
which must be set to ensure the numerical stability of the
simulation. Its value depends on the maximum velocity
in the system, and simulation of the two geometries was
in the range of 8E-6s. An adaptive time-step is used to
allow the solver to adjust according to the solution.
An implicit pressure solver is used to improve
the accuracy of the pressure field calculation. The
simulation was performed on two CPU cores (Intel
Xeon Silver 4114) along with an NVIDIA V100 GPU.
The benefit of using a graphic processing unit (GPU)
is the remarkably high performance-to-cost ratio due to
the large number of computing cores in the GPU card.
For example, an NVIDIA V100 has 5120 CUDA cores
and is capable of 15.7 TFLOPS in single precision, with
performance comparable to 48-64 CPU cores.
The goal of the activity is to compare two BEB designs, with
attention to the pressure distribution in the oil channels and the
pressure behavior in the vicinity of the bearings in question.
During the transient simulation, the pressure values in different
areas of the channels are monitored, focusing on the two central
connecting rods to avoid boundary effects.
An overview view of the oil flow in the full domain shows that
the highest pressure values are found in the lower part of the
connecting rod (see Fig. 6). Due to the motion of the power
system parts, pressure waves generated in the system are periodic
and are continuously reflected from the thrust side and the antithrust
side of the BEB.
Fig. 7 shows the evolution of the pressure field around the BEB of
the central pair of pistons (marked with an asterisk in Fig. 6). The
pressure field is smooth and the pressure waves are well captured
by the solver. The left corner of the C-shaped channel is reached
periodically by a pressure peak (red values in frames 4 and 6),
which is followed by low pressure values.
Low-pressure values are predicted near the corners of the
C-shaped tube and other areas of the BEB. This may result in
an improper lubrication. The predicted pressure trends were
confirmed by experimental investigations.
To measure the magnitude of the pressure shocks, ten control
regions are defined in the connecting rod channels: three on the
thrust side of the connecting rod; three on the anti-thrust side of
the connecting rod; one in each corner of the C-shaped tube; and
two in the rising pipe of the connecting rod. Fig. 8 highlights the
locations of these control regions.
Along with the pressure measurement, quantitative indices are
defined to compare the different BEB design configurations.
Specifically, the following quantities are monitored: average
pressure and standard deviation, peak pressure, time at low
pressure, and pressure derivative at peak occurrence.
A high average pressure with a low standard deviation is desirable;
it means that the flow is uniform and the risk of inadequate
lubrication is low. Pressure peaks, time at low pressure, and
pressure derivative should be as low as possible to ensure
longevity of the engine components.
Fig. 9 shows an example of the pressure signal measured in the
control regions described above. These signals reveal spikes, and
some control regions have long intervals at extremely low pressure
followed by sudden spikes. This behavior can be observed in
particular in control regions arc0_L, arc1_L, and elbow_1(L).
A thorough analysis of the previous indices over the entire domain
shows that the most critical regions are the lower regions of the
bearing (arc0_L, arc0_R), the left middle section (arc1_L), and
both corners of the C-shaped channel (Fig. 10).
These regions all feature long periods at low pressure followed by
steep spikes with high pressure derivatives.
The locations of the most critical regions suggest a potential
improvement in the bearing design. From the simulation data,
it is clear that the new bearing design (see Fig. 11b) has two
substantial advantages: it reduces pressure spikes and provides
pressure continuity on both sides of the bearing.
Comparing the simulation results for the two configurations shows
an overall improvement in all quality indices. The oil has a higher
average pressure overall and a lower standard deviation; pressure
peaks and pressure derivatives are reduced by 20-30%. As a
result, the internal channels are more filled and the oil flow is
Fig. 12 shows the comparison of pressure distribution between
the original (a) and improved (b) designs. The new bearing design
helps balance the pressure in the system, which is more uniform
overall than the original design.
Figs. 13 and 14 show the comparison of the pressure spikes and
derivative signals measured for the two configurations. The grey
lines represent the peak values of the original design, while the
black lines refer to the improved design.
The blue and orange signal lines refer to the new geometry. Most
regions depicted in the images show good improvement (lower
pressure peaks and lower pressure derivatives).
Fig. 13 - Pressure measurement comparison between the original and improved
designs. The black dashed line represents the peak values of the improved design; the
grey line indicates the peak values of the original design. The blue signal line is for the
improved BEB configuration.
Fig. 14 - Pressure derivative comparison between the original and improved designs.
The black dashed line represents the derivative values of the improved design; the pink
line indicates the derivative values of the original design. The orange signal line refers
to the improved BEB configuration.
The lubrication system of a large bore engine was analyzed using
Particleworks moving particle simulation (MPS) software. MPS
is a meshless method of solving the Navier-Stokes equations
which easily accommodates complex moving geometries, such
as crankshafts, connecting rods, and pistons.
The simulation method was applied to two different big-end
bearing geometries. The pressure behavior for both designs
was compared in terms of pressure peaks, average values, and
minimum values. The first big-end bearing design showed higher
pressure peaks, higher pressure derivatives, and longer periods at
low pressure. This combination is associated with higher pressure
shocks and potentially inadequate lubrication.
The modified big-end bearing design allows pressure waves
to propagate between the thrust and anti-thrust sides of the
connecting rod, helps reduce spikes and stabilizes overall oil
The predicted pressure trends and behavior of both designs were
confirmed by experimental investigations.
 Koshizuka, S., Oka, Y. (1996). Moving-particle semi-implicit method for
fragmentation of incompressible fluid. Nuclear science and engineering,
123. Pag. 421-434
Newsletter EnginSoft Year 18 n°4
By Luciano Perinel, Alessio Cherini, Irene Gallici | Wärtsilä Italy
Gianluca Parma | EnginSoft
Wärtsilä is a global leader in innovative technologies and lifecycle
solutions for the marine and energy markets. We emphasise
innovation in sustainable technology and services to help our
customers continuously improve their environmental and economic
Our dedicated and passionate team of 17,500 professionals in 200
locations in more than 70 countries shape the decarbonisation
transformation of our industries across the globe. In 2020
Wärtsilä’s net sales totalled EUR 4.6 billion. Wärtsilä is listed on