International Journal of Fluid Power 10 (2009) No. 2 pp.

57-72

MECHANICAL MODELLING OF A BENT AXIS PUMP

Alessandro Roccatello and Nicola Nervegna

Politecnico di Torino - The Fluid Power Research Laboratory (FPRL)

Corso Duca degli Abruzzi 24, 10129 Torino, Italy
[Link]@[Link], [Link]@[Link]

Abstract

In the technical literature numerous studies are found focused on the mathematical modelling of mechanical aspects
of swash plate axial piston pumps. Instead, bent axis pumps are rarely considered despite their widespread use in mobile
and fixed applications. This research paper presents the mechanical model of a bent axis pump that simulates the dy-
namic behaviour of the main measurable quantities (e.g. the shaft torque) and the mutual forces in interacting compo-
nents. The model is parametric and thus apt in predicting the influence of geometric design variables on pumps me-
chanical characteristics. A number of simulation analyses grounded on the presented model and on an ADAMS multi-
body approach are considered and contrasted one another and with experimental torque data for validation purposes.

Keywords: bent axis pump, modeling, simulation

1 Introduction
Over the last two decades the Fluid Power Research scalar relations based on decomposition of forces and
Laboratory (FPRL) has developed and validated simu- kinematic quantities along cartesian axes. In Osama et
lation models for axial piston pumps and motors al. (2002) a partial description of piston kinematics for
(Mancò et al., 2002), external and internal gear (gero- a BAP pump is described followed by studies on
tor) pumps (Fabiani et al., 1999), radial pistons (Caretto pump displacement controls. In Manring and Dong
et al., 1996) as well as variable and fixed displacement (2004) rotational matrices are used to express coordi-
vane pumps (Mancò et al., 2004). All have generally nate systems applied in the development of kinematic
evolved in AMESim, elaborating proprietary libraries analysis of a swash plate pump; the analysis considers
leading to an accurate prediction of the main hydraulic the existence of a secondary axis of rotation for the
and mechanical quantities; recently, a multibody soft- swash plate. Subsequently, results provided by kine-
ware code has also been proposed for the analysis of matics are applied: piston velocity and acceleration
axial piston pumps (Roccatello et al., 2007). For this are not attained through integration of the equation of
last pump family, models have been specifically devel- motion but rather through time derivatives of analyti-
oped for the swash plate category; this paper, instead, cal relations expressing its position. An analogous
addresses modelling aspects of bent axis pumps (BAP). approach is followed in the present paper where
Since the hydraulic modelling has not required substan- kinematics is analysed first and all unknown reaction
tial modifications, being only adapted to the new pump forces are determined thereafter. For this reason ex-
topology (e.g. flow leakage between slippers and swash changed forces do not influence either position or
plate is absent), the present study will purposely focus velocity to account, for example, of microscale piston
on the mechanical modelling of the pump. The perti- motion within the cylinder. Such an approach is de-
nent technical literature does not provide numerous tailed in Wieczorek and Ivantysynova (2002) where it
resources about this topic: in Ivantysyn and Ivanty- is oriented to the study of tribologic phenomena. The
synova (2000) a description of the kinematics of BAP present paper proposes a compact vector algebra ap-
pumps is reported (considering various manufacturing proach; kinematics is initially described, stressing its
solutions) that, in turn, supports the analysis of forces higher complexity when compared to swash plate
exchanged among components. These studies rely on units. Subsequently, the mechanical modelling of the
three principal components is discussed (piston, cyl-
This manuscript was received on 11 December 2008 and was ac-
cepted after revision for publication on 8 June 2009
inder block, shaft) to evidence reciprocal forces. Fur-

© 2009 TuTech 57
 Alessandro Roccatello and Nicola Nervegna

thermore, the dynamic behaviour of interacting forces simple to express since pistons undergo rotation about
is documented and analysed trying to provide an ex- the shaft’s axis and axial translation determined by
planation of existent relations between forces and swash plate’s tilt (see (Roccatello et al., 2007)). In-
piston kinematics. As to the shaft torque a comparison stead, for a bent axis unit, pistons are constrained by
is shown that confronts simulated results with ex- spherical joints integral with the shaft and by the
perimentally obtained data. collinearity of elastic ring centres with the cylinder
axis. Consequently, pistons axes do not remain paral-
lel but rather orbit in space in a more complex man-
ner. Kinematic analysis has been grounded on four
coordinate systems: geometric points of interest and
coordinate systems being described by vectors and
matrices, respectively.
Generally (Litvin et al., 2004), point M is repre-
sented in coordinate system Sm(xm, ym, zm) by the posi-
tion vector:
T
m
r = [ xm ym zm 1] (1)
The same point M can be determined in coordinate
system Sn(xn, yn, zn) by the position vector:
Fig. 1: Section view of the bent axis pump, coordinate T
systems and main geometric quantities
n
r = [ xn yn zn 1] (2)
with the matrix equation (position vectors being
repre-sented with homogeneous coordinates):
n
2 Pump Description and Operation r = M nm m r (3)
Matrix Mnm is represented by:
Generally, BAP pumps, when compared with Om ) ⎤
swash plate units, are considered (Ivantysyn and Ivan- ⎡ ( in im ) ( in jm ) ( in km ) xn(
⎢ ⎥
Om )
tysynova, 2000) more expensive and less compact, of ⎢( j i ) ( jn jm ) ( jn km ) yn( ⎥
M nm =⎢ n m (4)
( Om ) ⎥
more complex manufacturing and less adaptable to the
various control strategies; nonetheless these units ⎢( k n im ) ( k n jm ) ( k n k m ) zn ⎥
usually have a better total efficiency, are less sensitive ⎢ 0 0 0 1 ⎥⎦
⎣
to fluid contamination and allow a higher rotational
speed. Different types of bent axis pumps are com- Subscript “nm” in the designation Mnm indicates
mercially available; for fixed displacement units, that the coordinate transformation is performed from
distinct techniques are accomplished to transfer shaft Sm to Sn. Here, (in, jn, kn) are the unit vectors of the
rotary motion to the cylinder block: cardan joints, axes of Sn; (im, jm, km) are the unit vectors of the axes
connecting rodspistons and bevel gears. This last of Sm; (xn, yn, zn) represent the coordinates of the ori-
solution will be considered henceforth since it allows gin Om of Sm in coordinate system Sn (origin On). Dot
a larger tilt of the cylinder block (up to 45°, (Ivanty- products in matrix Mnm (e.g. inkm) can be expressed
syn and Ivantysynova, 2000)), thus enhancing pump through direction cosines or as an indexed sum of
displacement; in addition, this solution is adopted by their components (e.g. in,xkm,x+in,ykm,y+in,zkm,z).
numerous manufacturers. Figure 1 shows a section The inverse coordinate transformation that deter-
view of the pump (Casappa Strada-BAP 63): a prime mines the coordinates (xm, ym, zm) taking as given
mover provides shaft rotation. The large cylindrical coordinates (xn, yn, zn) can be written as :
m
shaft boundary houses in spherical joints pistons that r = M nm m r (5)
are, in turn, lodged within cylinders in the cylinder
block. This is tilted of an angle a and rotates (guided
by a cylindrical pin) at shaft angular velocity due to 3.1 Coordinate Systems, Coordinate Transforma-
the bevel gears coupling. Variable volume chambers tions and Matrices
are isolated from pump casing through elastic rings In concert with general principles recalled above,
that slip onto cylinders faces due to the influence of the analysis of pump kinematics, as presented hereafter,
fluid pressure. considers four coordinate systems that prove expedient
in the development phase of governing equations. The
chosen systems are identified as S1, S2, S3 and S4. Coor-
3 Pump Kinematics dinate transformations and related matrices will now be
introduced:
For bent axis pumps, kinematic analysis is signifi-
cantly more complex than for swash plate units. By 3.1.1 Coordinate Systems S1 and S2; Matrix M12
way of example think of the piston centre of mass Coordinate systems S1 (x1, y1, z1) and S2 (x2, y2, z2)
(CM): for swash plate pumps analytical relations are fixed in space and indicated in Fig. 1. Worth of
describing coordinates (x,y,z) of CM are relatively notice is the fact that their origins are separated by the

58 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

distance b. The coordinate transformation from S2 to S1 1

rB = [ −Ra sin ϑ Ra cos ϑ 0 1]
T
(9)
is based on the matrix equation:
1
r = M12 2r (6) In turn position vector rB of point B in coordinate 1

system S1 can be formally expressed by the matrix

It is straightforward to write matrix and its inverse equation:
as follows: 1
rB = M12 2 rB (10)
⎡1 0 0 0⎤
⎢0 cos α sin α 0 ⎥⎥ where:
M12 = ⎢
⎢0 − sin α cos α b⎥ ⎡ − Rd sin ϑ ⎤
⎢ ⎥ ⎢ Rd cos ϑ ⎥
⎢⎣0 0 0 1 ⎥⎦ 2 ⎢
rB = 2 ⎥ (11)
(7) ⎢ z + 2z − a ⎥
⎡1 0 0 0 ⎤ ⎢ A A 1⎥
⎢0 cos α − sin α b sin α ⎥⎥ ⎣⎢ 1 ⎦⎥
M 21 = ⎢
⎢0 sin α cos α −b cos α ⎥ being:
⎢ ⎥ 2 2
⎣⎢0 0 0 1 ⎦⎥
a1 = ( 2 xB − 2 xA ) + ( 2 yB − 2 yA ) + 2 zA − b12
2
(12)
The coordinate transformation in transition from S2
to S1 is represented by the equations: Having identified the position of points A and B, it
is possible to express unit vectors of S3 in S1; through
x1 =x2 Eq. 9 and 11, the unit vector of axis z3 in S1 is:
y1 = y2 · cos α + z2 · sin α (8)
z1 = - y2 · sin α + z2 · cos α + b 1 ( 1rB − 1rA )
k3 = 1 (13)
rB − 1rA
3.1.2 Coordinate System S3; Matrix M13
Coordinate system S3, movable in space, has the
origin at point A, centre of the spherical piston joint and
axis z3 directed along the piston’s axis from A to B,
centre of the elastic ring (see Fig. 2).
Consider now point A that, constrained to rotate about
the shaft’s axis, describes in S1 a circumference of radius
Ra. Position vector 1rA is then written as (see Fig. 3).

Fig. 4: Coordinate systems S3 and S1

1
j k = j
3
1

3
1

3,x
1
k 3,x
+ j 1

3,y
k
1

3,y
+ j
1

3,z
k
1

3,z
=0 (14)
The unit vector of axis y3
Fig. 2: Coordinate systems S3 and S1 ( j = ⎡⎣ j
1
j 1
j ⎤⎦ ) has the following properties:
1 1
T

3 3,x 3,y 3,z

(a) is normal to k3; (b) lays on a plane parallel to plane

1

(y1, z1); (c) has a unitary module. While (a) is equivalent

to the following scalar relation (dot product between 1j3
and 1k3 equal to zero) property (b) sets to zero the compo-
nent of the unit vector 1j3 along the x axis:
1
j3, x = 0 (15)
From (c) immediately follows:
1
j3, x2 + 1j3, y2 + 1j3, z2 = 1 (16)
Equations 14 to 16 in three unknowns ( j3,x, j3,y,
1 1
1
j3,z) lead to:

Fig. 3: Approach to pump kinematics

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 59

 Alessandro Roccatello and Nicola Nervegna

⎛ k 1
⎞ 3.1.3 Coordinate System S4; Matrix M14
1
j =⎜− 3,z
⎟⎟ j
1

3,y
⎜ k 1 3,z

A last fixed coordinate system S4 has been intro-

⎝ 3,y ⎠
duced O4 (x4, y4, z4), to identify the axis z4 that cor-
1 (17)
1
j =± responds to the line of action of the force exchanged
k
3,z 2
1

1+ 3,z
in the bevel gears mating: this being convenient while
k
1 2

3,y
writing equilibrium equations and more specifically
1
j =0
the reaction force Rcs between cylinder block and
3,x
shaft through the mating gears. In the left portion of
where the positive sign is selected to define the positive Fig. 5 it can be observed that the driving gear is inte-
direction of 1j3. The unit vector of the x3 axis is derived gral with the shaft while the driven with the cylinder
from knowledge of the other two: 1i3 = 1j3 ^ 1k3 Matrix block. Axes of rotation are tilted of an angle α, gears
M13, is written as follows: feature the same number of teeth and have equal pitch
O3 ) ⎤ cones angles (δa = δt). Under the hypotheses that (i)
⎡ ( i1i3 ) ( i1 j3 ) ( i1k3 ) x1(
⎢ ⎥
O3 )
the condition of meshing involves only one pair of
⎢( j i ) ( j1 j3 ) ( j1k3 ) y1( ⎥ teeth (Jacazio and Piombo, 1997), (ii) that gears are in
M13 = ⎢ 1 3
( O3 ) ⎥ point contact and (iii) that the exchanged force Rcs is
⎢( k1i3 ) ( k1 j3 ) ( k1k3 ) z1 ⎥
applied at O4, midpoint of teeth faces on the conical
⎢ 0 0 0 1 ⎥⎦
⎣ (18) pitch surface, the position vector of the origin O4 in
⎡1i3,x 1
j3,x 1
k3,x 1
rA,x ⎤ coordinate system S1 follows:
⎢1 1 1 1 ⎥ T
i j3,y k3,y rA,y ⎥ ⎡ mZ mZ ⎤
= ⎢ 1 3,y
1
1 1 1
rO4 = ⎢0 - cot δ a + b 1⎥ → (19)
⎢ i3,z j3,z k3,z rA,z ⎥ ⎣ 2 2 ⎦
⎢ ⎥
⎣⎢ 0 0 0 1 ⎦⎥ being the pitch cone diameter at the contact point d =
mZ. It is then possible to write:
where (i1, j1, k1) are the unit vectors of the axes of S1;
(i3, j3, k3) are the unit vectors of the axes of S3; (x1, y1,
2
rO4 = M21 1rO4 (20)
z1) represent the coordinates of the origin O3 of S3 in Knowledge of rO4 allows to express the three com-
2

coordinate system S1 (origin O1). ponents of unit vectors x4, y4 and z4 as follows:
- The unit vector of x4 in S1 (11i) is oriented from
O4 to O2 and consequently:
2
2 rO4 1
i4 =- 2
→ i4 =M12 2i4 (21)
rO4
- The unit vector of z4 in S1 (1k4) has a tilt equal to
the pressure angle (αp) and can be defined as follows:
⎡ − cos α p sign (ω ) ⎤
⎢ ⎥
k4 = ⎢sin α p cos δ sign (ω ) ⎥
1
(22)
⎢ sin α p sin δ sign (ω ) ⎥
⎣ ⎦
Fig. 5: Coordinate systems S4 and S1 where the operator sign (ω) accounts for the possibility
of reversing pump rotational speed.
In the analysis N coordinate systems (S3k, k = - The unit vector of y4 in S1 (1j4) is, by definition:
1...N) are effectively considered each integral with the 1
j4 = 1k4 ^ 1i4 Matrix M14, is written as follows: (cfr. (4)):
corresponding piston k (see Fig. 2) featuring points Ak
⎡ (i i ) (i j ) (i k ) x( )⎤
O4

and Bk: the aim being that of expressing with relative ⎢

1 4 1 4 1 4 1
⎥
ease some involved forces; e.g., that exchanged be- ⎢( j i ) (j j ) (jk ) y( ⎥ )
O4

M =⎢ 1 4 1 4 1 4 1

tween piston and shaft and directed along the piston’s ( 4) ⎥

⎢( k i ) (k j ) (k k )
14

z ⎥
O

axis movable in space. Coordinate systems are identi- 1 4 1 4 1 4 1

fied by their different angular position: ϑk = ϑ – (k – ⎢ 0 0 0 1 ⎥⎦

⎣
1)·2π / N1. Matrix M13 presented above (18) is ex- ⎡ 1
i 1
j k
1
0 ⎤ (23)
⎢ ⎥
4,x 4,x 4,x
panded in analytical relations that, owing to their
considerable length, are here omitted; however, a ⎢ mZ ⎥
1
i 1
j k
1

Matlab code is provided in the Appendix that allows ⎢ 4,y 4,y 4,y
2 ⎥
=⎢ ⎥
the generation of the complete symbolic expression. mZ
⎢ 1
i 1
j 1
k − cot δ + b ⎥
⎢ 2 ⎥
4,z 4,z 4,z a

⎢ 0 0 0 1 ⎥
⎣ ⎦
1
To simplify notations, while expressing points A, B and matrix
M13, subscript k has been deliberately omitted. In (9) and (11) co-
This also leads to: M24 = M21M14
ordinate ϑ should, in fact, read ϑk, thus identifiying N couples of
points A and B as well as N coordinate systems S3. Pistons are
numbered sequentially in the CCW direction: hence, piston k + 1
follows k if the pump rotates clockwise.

60 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

4 Mechanical Modelling Note that in the N coordinate systems S3, with origins
in the spherical piston joints, reaction (3Rs,k) has a single
component along piston axis (k3). Position vector 2Rs,k can
4.1 Piston Model be written in S3 as 3Rs,k through matrix M23k = M21 M13k.
- Rc,k: reaction force on piston (point Ck) from cylinder
Figure 6 shows a generic piston k and applied ex- block; the assumption is here made that the piston may
ternal forces. contact the internal cylinder face in a generic point be-
longing to the circumference with centre Ck (see Fig. 6).
In this respect it should be noticed that the elastic ring is
not integral with the piston and, as a consequence, the
latter may lean onto the cylinder in a point that differs
from the centre of the sphere defining the external surface
of the ring (at point Bk)2. In coordinate system S2 only two
unknown reaction forces exist since no contribution is to
be accounted along the z2 axis3.
T
2
R
c,k
= ⎡⎣ R
2

c, kx
2
R c, ky
0 ⎤⎦ (30)

Fig. 6: Free body diagram of the piston

4.1.3 Equilibrium Equations
Piston translational and rotational (about O2) equi-
4.1.1 Known Forces Acting on Piston librium equations in coordinate system S2 are written,
- Fp,k: force originated by fluid pressure acting in vector notation, as follows4:
2
along axis z2: Fp,k + 2 Fpf,k + 2 Pp,k + 2 FiA,k + ...
(31)
... + 2 FiB,k + 2 Rs,k + 2 Rc,k = 0
T

⎡ πD ⎤ 2

F = ⎢0 0 −p ⎥
2 pi
(24)
4 ⎥⎦
p,k k

⎢⎣ 2
rB ∧ 2 Fp,k + 2 rB ∧ 2 Fpf,k + 2 rGpi ∧ 2 Pp,k + ...
- Fpf,k: friction force: ... + 2 rA ∧ 2 FinA,k + 2 rB ∧ 2 FinB,k + ... (32)
sign ( v )⎤⎦
T
2
F
pf,k
= ⎡⎣0 0 − F pf,k
2

Bk, z
(25) 2 2
... + rA ∧ Rs,k + rCk ∧ Rc,k = 0 2 2

in general F pf,k
depends on point Bk velocity. From Eq. 31 and 32 a linear system of five equa-
tions in five unknowns (2Rs,kx 2Rs,ky 2Rs,kz 2Rc,kx 2Rc,ky)
- Pp, k: piston weight: is obtained:
T
1
P = ⎡⎣0 − m g 0 ⎤⎦
p,k pi
Ap Xp = Bp (33)
(26)
2
P =M P
p,k 12
1

p,k
4.2 Shaft Model
- FiA,k and FiB,k: inertia forces. Two contributions are
considered since piston mass is, by hypothesis, lumped in Figure 7 shows the shaft and applied external
points Ak and Bk. This assumption avoids calculations of forces. Equilibrium equations will be written in coordi-
nate system S1. The shaft and the driving bevel gear
terms dependent on angular acceleration (ω ) and of will be considered as a single rigid body.
correspondent inertial contributions. In the Appendix it is
shown that this simplification does not give rise to signifi-
cant differences in attained results. Thus, piston mass mpi
is divided, into two generally different portions: mA at
point Ak and mB at point Bk as follows:
mA + mB = mpi (27)
mA = χ mpi
Taking the time derivatives of Eq. 9 and 11, veloc-
ity and acceleration of points Ak e Bk are determined.
Consequently:
2
FiA,k = mA 2aAk 2FiB, k = - mB - 2aBk (28)
Fig. 7: Shaft free body diagram
4.1.2 Unknown Reactions on Piston
- Rs,k: reaction force on piston (point Ak) from shaft; 2
The elastic ring seals the variable volume chamber while the piston
mating of the two components has been modelled with contacts the cylinder in a point of the circumference centred in Ck.
a spherical joint that removes three translational DOF. 3
Friction, expressed previously, is here considered as a known
Three reaction forces are then unknown: force.
T
2
R
s,k
= ⎡⎣ R 2

s,kx
2
R s,ky
2
R s,kz
⎤⎦ (29) 4
To simplify notation Ak and Bk are written as A and B

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 61

 Alessandro Roccatello and Nicola Nervegna

4.2.1 Known Forces Acting on Shaft 4.3 Cylinder Block Model

- Fm: spring force; in coordinate system S2 has a Figure 8 shows the cylinder block and applied ex-
single component (along z2): ternal forces. Equilibrium equations will be written in
2 2 T 1 coordi-nate system S2.
Fm = ⎡⎣0 0 Fm,z ⎤⎦ Fm = M12 2 Fm (34)
where 2Fm,z is the spring force magnitude, determined
from elastic and geometric properties.
- 2Rs,k: this reaction has already been obtained in S1
in the piston model; hence it is now a known force:
T

− R 2

s,k
= ⎡⎣ − R 2

s,kx
− R 2

s,ky
− R 2

s,kz
⎤⎦
(35)
→ R 1

s,k
=M 12
2
R s,k

Fig. 8: cylinder block free body diagram

- Rcs: reaction force on the shaft from the cylinder
block through the bevel gears mating. This reaction
will be evaluated in the cylinder block model and can
be considered here as a known force with three compo- 4.3.1 Known Forces Acting on Cylinder Block
nents in coordinate system S1: - Fm: spring force; in coordinate system S2 has a
T single component (along z2):
− R = ⎡⎣ − R
1 1
−R 1
−R 1
⎤⎦ (36)
cs cs, x cs, y cs, z

2 2 T
Fm = ⎡⎣0 0 Fm,z ⎤⎦ (45)
- Ps: shaft weight:
1
Ps = [0 -msg 0]T (37) - Fpf,k: friction force (see (25)).
- Tsf: friction torque on the shaft. Two parameters (Cs) - Pc: cylinder block weight. Active along the y1 axis:
Pc = [ 0 −mc g 0]
T 2
and (Ct) are introduced to consider viscous torque losses 1
Pc = M 21 1 Pc (46)
and losses proportional to torque Td required by the pump
to keep the shaft turning at constant angular velocity: - Tcf: friction torque on cylinder block. By account-
T
ing for the viscous component only:
Tsf = ⎡⎣ 0 0 −Td (1 − Ct ) − Csω ⎤⎦
1
(38)
Tcf = [ 0 0 -Ccω ]
2 T
(47)

4.2.2 Unknown Reactions on Shaft - 2Ftp,k: force, originated by fluid pressure within the
cylinder, pushing the cylinder block against the pump
- Rb1: reaction force (point C1) from tapered roller cover; it acts along axis z2 at point Htp,k:
bearing. In S1 we will generally observe three components: T

T
2
F tp,k
= ⎡⎣0 0 2
F tp,kz
⎤⎦ (48)
1
R = ⎡⎣ R 1 1
R 1
R ⎤⎦ (39)
where Ftp,kz = Ac pk and Ac being shown in Fig. 8.
2
b1 b1, x b1, y b1, z

- Rb2: reaction force (point C2) from cylindrical - 2Fth: force, originated by fluid pressure, pulling
roller bearing. In this case the component along z1 is the cylinder block away from the cover. The assump-
missing, therefore: tion is here made that the pressure field, consequent to
T fluid pressure in cylinders, evolves linearly on sealing
1
R = ⎡⎣ R
b2
1

b2, x
1
R b2, y
0 ⎤⎦ (40) lips as specifically indicated in Fig. 8. This determines
- Tex: drive torque required from prime mover: (2Fth,z) and its point of application (Hth).
T
1
Tex = ⎡⎣0 0 1
Td ⎤⎦
T
(41)
2
F = ⎡⎣0 0
th
2
F ⎤⎦th,z
(49)
- Rc,k: force on cylinder block (point Ck) from pis-
4.2.3 Equilibrium Equations ton. A known force already evaluated (see(30)):
T

Shaft translational and rotational (about O1) equi- − R

2

c,k
= ⎡⎣ − R2

c, kx
− R 2

c, ky
0 ⎤⎦ (50)
librium equations in coordinate system S1 are written,
in vector notation, as follows:
4.3.2 Unknown Reactions on Cylinder Block
Fm + ∑ ( − 1 Rs, k ) − 1 Rcs + 1 Ps + 1 Rb1 + 1 Rb2 = 0
1
(42)
k - Fh: hydrodynamic force, originated by fluid veloc-
ity and pressure, that pulls the cylinder block away from
1
r02 ∧ 1Fm + ∑ 1rAk ∧ ( − 1 Rs, k ) + 1r04 ∧ ( − 1 Rcs ) + ... the cover; by assumption Fh acts along axis z2 and arbi-
k trates the axial equilibrium of the cylinder block:
1
rGs ∧ 1 Ps + 1Tsf + 1Tex + ... (43) T
2
F = ⎡⎣0 0 2
F ⎤⎦ (51)
1 1 1 1 h h,z

... + rc2 ∧ Rb2 + rc1 ∧ Rb2 = 0

However, it should be observed that such a force is
The system takes the following matrix notation (six not applied on the cylinder block axis but rather in a
unknowns 1Rb1,x 1Rb1,y 1Rb1,z 1Rb2,x 1Rb2,y 1Td): point of coordinates (x2, y2) = (xh, yh) to be so identi-
Aa Xa = Ba (44) fied to guarantee its equilibrium also in reference to

62 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

tilting moments about axes x2 and y2. instead the hydraulic modelling of the unit. Testing condi-
Beside Fh two unknown moments should then be tions for simulations are set at a delivery pressure of p* and
considered. a rotational speed of 1500 rpm, the working fluid (a mineral
- Th: moment of force Fh due to its offset position oil) is at constant tem-perature of 60°C. Table 5.1 collects
from axis z2: further informations on pump and fluid characteristics. Fig.
T 10 shows portplate timing and the instaneous pressure
T = ⎡⎣ T
2

h
2

h,x
2
T h,y
0 ⎤⎦ (52) within a cylinder attained from hydraulic simulation. This,
the following equations correlate involved quantities: along with the complete pressure distribution within pump
cylinders, neglecting friction and inertia, beside providing
T
2

needed informations for the mechanical analysis, represents

xh =
h,y

− F 2

h,z
the basic source for the onset of mutually exchanged forces
(53) within the pump unit.
2
T
yh = h,x

2
F h,z
Table 5.1: Informations on pump and fluid characteristics
Pump type Casappa Strada-BAP 32.63
- Rcn: reaction force exerted by the pin; active in the
radial direction at the mean contact point (Hcn): Pump dis- 63.7 [cm3/rev]
placement
T
2
R = ⎡⎣ R
cn
2

cn, x
2
R cn, y
0 ⎤⎦ (54) Max. angular 1600 rpm (@ pmax = 350 bar)
velocity 2350 rpm (@ 0 bar)
- Rcs: reaction force from the bevel gear integral with Number of 5
the shaft and acting along the line of action of the mat- pistons
ing gears. Its orientation coincides with axis z4 in S4 Fluid Agip Arnica 46
and its point of application with O4. Kinematic viscosity (@ p*r, 60°C)
R = ⎡⎣0 0
4 4
R ⎤⎦
T
= 41.3 cSt
cs cs,z
(55) Portplate
→ R =M 4

cs 24
4
R =M M
cs 21 14
4
R cs integral with
pump cover
4.3.3 Equilibrium Equations Flow rate @ 95.6 [l/min]
1500 rpm
Cylinder block translational and rotational (about (ideal value)
O2) equilibrium equations in coordinate system S2 are Torque @ p*r 278.8 [Nm]
written, in vector notation, as follows: (ideal value)
2
Fm + ∑ 2 Fpf,k + ∑ 2 Ftp,k + ∑ ( − 2 Rc, k ) + ...
k k k (56)
... + Pc + Fth + Rcs + Fh + 2 Rcn = 0
2 2 2 2

∑ r ∧ (− F )+∑
2
2

Bk
2

pf, k
2
rHtp
+ F tp,k
+ ...
k k

... + ∑ r ∧ ( − R
2

Ck
2

c, k )+ 2
r ∧ P + r
Gt
2

c
2

Hth
∧ F ... (57)
2

th

... + T + T + r
2

cf
2

h
2

Hcn
∧ R + r ∧ R =0
2

cn
2

O4
2

cs

The system takes the following matrix format (six un-

knowns 2Fh, z, 2Th, y, 2Th, x, 2Rcn, y, 2Rcn, y, 4Rcs, z): At Xt = Bt

Fig. 9: Bent Axis Pump Library

5 Simulation
notches
The aforesaid mechanical model, coded in Fortran, gave
1.2
origin to a dedicated library of specific submodels (see Fig. suction
delivery

9) in the AMESim simulation environment. As to the hy- 1

draulic modelling, use has been made of previous research
piston 1 pressure
work at FPRL on axial piston pumps (Monacò et al., 2002). 0.8
p / p* r [ − ]

Figure 10 reports the complete AMESim sketch assessed

0.6
for the attainment of simulation results proposed ahead.
Basically, the lower portion of the sketch shows icons of 0.4
mechanical submodels (cylinder block, piston, shaft and
spring). Those at the right side permit the assignment of 0.2
geometric parameters of the portplate and of other pump
components. Those to the left evaluate, in turn, flow areas 0
0 50 100 150 200 250 300 350
for the suction and delivery side as well as chambers vol- theta [deg]

umes variations. The upper part of the sketch illustrates Fig. 10: Pressure on piston 1

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 63

 Alessandro Roccatello and Nicola Nervegna

Fig. 11: Bent axis pump modelling

5.1 Reactions on Piston

Reaction forces on piston originate from the inter- w
action with the shaft (via the spherical joints) and with
the cylinder block. Components of these reaction O2
forces will be considered in coordinate system S2. x2

5.2 Reactions between Piston and Shaft

Figure 12 shows, in a complete shaft revolution, the
three components of the reaction force Rs on piston 15:
the most significant contribution is along axis z2 with
oscillations about a mean value of 11000 N (delivery delivery
suction
phase) that clearly reproduce instantaneous pressure in
cylinder 1. It can further be noticed that components a) Rs z2
along the other two axes, namely x2 and y2, though
sensibly smaller, reach nonetheless values that cannot be
ignored as both tend to tip the cylinder block. In more
Fig. 13: Reaction Rs: plane x2-z2 and plane y2-z2
detail, Fig. 13(a) demonstrates that when the piston is at
the left of axis z2 (180-360°) Rs has a negative compo-
nent along axis x2 since the piston is so tilted to give
always rise to a negative Rs (x2). Along axis y2 the cor- 5.3 Reactions between Piston and Cylinder Block
respond-ing component is instead negative in the angular Figure 14(a) shows components of the reaction
interval 180° to 270° being the piston axis tilted as r1 in force acting on piston 1 from the cylinder block; during
Fig. 13(b); subsequently (from 270° to 360°) it becomes suction Rc essentially balances piston inertia effects. In
positive ( piston axis tilted as r2). the delivery phase higher values exist (up to 700 N) as
12000
Rs(z2)
already was the case dealing with shaft reactions. In
Fig. 14(b) force Fp and the two reactions Rs and Rc are
10000
indicated: in plane x2 - z2, their composition is also
8000 shown with values corresponding to ϑ ≈ 280° and ne-
glecting inertia.
Force [N]

6000
800
F
Fp
Rc(z2) = 0
Rc(x2)
600
4000 w
400

2000 O2
Rs(y2) 200 x2
Force [N]

0 0

Rs(x2) −200

−2000 Rc(y2) Fp
0 50 100 150 200 250 300 350 −400
theta [deg] Rc
Rs
−600
b) Rs
Fig. 12: Rs: piston-shaft reaction force −800
z2
Rc
0 50 100 150 200 250 300 350
theta [deg]

5
Piston 1 is so identified: at time t = 0, it has ϑ = 0 i.e. its point A is Fig. 14: Reactions Rc (components) and Rs (plane x2-z2)
on the axis y1 (see Fig. 1, top right).

64 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

5.4 Reactions on Shaft 5.4.5 Pump Torque

Components of these reaction forces will be consid- Figure 17 shows the instantaneous torque 1Td re-
ered in coordinate system S1. quired to keep the pump running at constant speed, see
Eq. 41. At steady state and in one shaft revolution, a
5.4.4 Reactions from Bearings number of oscillations equal to the number of pistons is
detected; their extent being correlated with delivery
Figure 15 shows, in continuous lines, plots of compo-
pressure (dashed lines, bottom) and, in turn, with the
nents in S1 of the reaction force Rb1 (tapered roller bear- continuously changing pressure inside variable volume
ing). Prevailing components act along axes y1 and z1 and
chambers (piston 1, full line, bottom)7. Figure 18 dis-
exhibit comparable magnitude (the shaft is subjected to plays a comparison of experimental torque data with
forces Rs from pistons (Fig. 16) that have an approximate those predicted by the present model. Both are obtained
tilt of 41 degrees with the shaft axis6). In addition and with as averages of torque signals sampled over a given time
dashed lines Fig. 15 also reports the two components of window (1 shaft revolution in simulation and at least
the reaction force Rb2 (cylindrical bearing). 1 s in the test rig). Experimental data acquisition is
4
x 10 performed, upon reaching steady-state (1500 rpm), for
4
Rb1 (y1) a number of discrete pump loading conditions (delivery
3 pressure range: 0 - 350 bar).
2 320
Force [N]

Rb1 (z1)
1 300

Torque [Nm]
Rb1 (x1) 280
−1
260
−2
0 50 100 150 200 250 300 350
4
x 10 240
1
Rb2 (x1)
0.5 220
0 50 100 150 200 250 300 350
0
Force [N]

−0.5 1.1
delivery pressure
−1 1.05
p / p* [ − ]

−1.5 1
Rb2 (y1)
r

−2 0.95
0 50 100 150 200 250 300 350
theta [deg] 0.9

Fig. 15: Rb1 and Rb2 reactions (components) on shaft 0.85

piston 1
pressure
0.8
0 50 100 150 200 250 300 350
y1 theta [deg]

Fig. 17: Pump torque (top) and pressures (bottom)

Rb1

400

350
Measured

z1 300 Simulation
Rs
que [Nm]

O1 C1 C2
250

Rb2 200
Torq

150

Fig. 16: Reactions on shaft from bearings 100

6 50
The model and ensuing simulation take into account the fact that
piston axis is not parallel with that of the cylinder.
0
7
Piston pressure coincides with the variable chamber pressure. Delivery 0 50 100 150 200 250 300 350 40
pressure is instead evaluated in the fixed capacity used to model the deliv-
ery volume. In Fig.17 the two traces seem to overlap. In effect the former pressure [bar]
is higher than the latter since a pressure drop occurs as fluid flows out of
the piston chamber through the cylinder block kidney and the portplate. Fig. 18: Measured and simulated torque

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 65

 Alessandro Roccatello and Nicola Nervegna

1000
Rc(x2) 1500rpm; p*
i
0

Force [N]
800
−500 1500rpm
p*r
600
1500rpm Rs(x2)
p* −1000
r 1500rpm; p*i
Force [N]

400 0 180 360 540 720 900 1080

1000
Rs(y2) 1500rpm; p*
i

200 500

Force [N]
0 1500rpm
0
p* r

−500
−200
0 180 360 540 720 900 1080
theta [deg] 0 180 360 540 720 900 1080
theta [deg]
Fig. 19: Rc reaction, component x2
Fig. 21: Rs reaction, components x2 and y2

1000
Rc(y2) 1500rpm; p* 18000
800
i
Rs(z2) 1500rpm; p*i
16000
600 1500rpm
p*r 14000
400
12000
200
Force [N]

10000
Force [N]

0 1500rpm
8000 p* r
−200
6000
−400
4000
−600
2000
−800
0
−1000
0 180 360 540 720 900 1080 −2000
theta [deg] 0 180 360 540 720 900 1080
theta [deg]
Fig. 20: Rc reaction, component y2
Fig. 22: Rs reaction, component z2

6 Influence of Load on Internal Forces

As anticipated, the model easily allows quantita-
tive knowledge of exchanged forces in the pump
when different loading conditions are examined. In
this respect, a situation is analysed whereby, at con-
stant speed, the unit is operated at peak load (p*i).
Figure 21 to 24 report predicted results from the pre-
sent model at 1500 rpm and p*i, set against the previ-
ously shown case at 1500 rpm and p*. All forces that
were considered formerly are now detailed over three
shaft turns and, as expected, all increase with load.
Worth of notice is the fact that the tapered roller bear-
ing undergoes cyclic component forces (see Fig. 23)
surpassing, respectively, 14000 N (Rb1x), 35000 N
(Rb1z), and 50000 N (Rb1y). It can further be ob-
served (Fig. 19) that, during suction (see pointing
arrow), the reaction Rc(x2) initially rises and then
decreases as a consequence of piston acceleration and
related inertial effects.

Fig. 23: Tapered roller bearing reactions (Rb1)

66 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

Figure 30 collects results relative to the evaluation

of the instantaneous torque in Adams and in the present
model. In the same figure are also reported torque mean
values collected from experimental data and predicted
in Adams and in the mechanical model. The Adams
model, where friction between pistons and cylinder
block is neglected, yields a slightly lower mean torque
value than the present model. Further, it is also interest-
ing to investigate the torque required from the shaft to
transfer rotary motion to the cylinder block through the
bevel gears. In this respect, simulations indicate that
this originates from two different sources:
• friction effects between the cylinder block and (i)
guide pin, (ii) portplate (pump cover), (iii) working
fluid in the pump casing. These contributions are
modelled through a viscous friction coefficient C,
see (47);
• onset of a periodic torque featuring a change in
sign and consequent to the balancing effect of reac-
tion forces that pistons exert on the cylinder block.
Fig. 24: Cylindrical bearing reactions (Rb2)
Figure 32 (top, full line, present model) shows a
plot of reaction 4Rcs, z (see Eq. 55) while (bottom, full
line, Adams model) shows the torque to be applied to
the cylinder block inline joint, where the same shaft
7 ADAMS Virtual Pump Model velocity is imposed9. In both cases an equal viscous
friction coefficient value is adopted for the cylinder
In lack of challenging experimental verifications of block: C = 0.0054 Nm/(rev/min) that, at 1500 rpm,
the internal forces predicted by the present model one originates a constant resistant torque of a 8.1 Nm. Also
additional investigation is here detailed. This has re- the situation where this coefficient is supposed to be
quired the deployment of a full 3-D virtual pump in the zero is deliberately considered; though this is not realis-
ADAMS multibody environment (see Fig. 35). The tic (the and negative sign (the barrel therefore either
con-straints assigned are coherent with those used in brakes or barrel rotates fully immersed in a viscous
the mechanical model: i.e. spherical joints (pistons- fluid and a accelerates the shaft). It will be demon-
shaft; tapered roller bearing-shaft), and inline joints strated that this is lubricated gap exists with the fixed
(pistons-cylinder block; cylindrical bearing-shaft). portplate) still it consequent to the behaviour of inter-
External actions on the pump are exerted as follows: vening reactions Rc serves the purpose of highlighting
• the instantaneous pressure within variable volume the fact that the shaft-between pistons and barrel. In
chambers attained in the hydraulic model (Fig. 11) fact, if this contribution barrel exchanged torque exhib-
fitted with a spline as shown in Fig. 25, is put to its an alternate positive were absent the situation de-
use in ADAMS. picted in the same Fig. 32 with dash lines would be
• an angular speed of 1500 rpm is assigned to the shaft. obtained respectively for the force (top) and torque
(bottom). Due to the periodic change in sign, both
The ADAMS simulation will ultimately yield the
models bring to evidence this specific aspect. In greater
torque required to keep the pump running at steady-
detail consider Fig. 31 showing a sufficient to multiply
state. However, the multibody environment also allows
force times the torque front view of the cylinder block:
quantitative knowledge, through appropriate measures,
reactions Rc from pistons onto the cylinder bock are
of the intervening internal forces. Strip charts of such
forces can be monitored while the pump is running and indicated for the angular position ϑ = 0° (circular
the data stored for appropriate post-processing analysis. marker shown in Fig. 32).
Figures 26 to 29 provide a visual cue of ADAMS re-
sults contrasted with those achieved utilizing the me-
chanical model exposed in this paper. Plots of the dif-
ferent reaction forces are nearly identical for the two
approaches. Figures 26 and 27 appear, at first sight, as
being one the mirror of the other this being false due to
into a triangle of forces. Then, in cited figures, the presence of iner-
the effects of piston inertia. If this contribution were tial effects motivate the small but existing difference in compo-
neglected then reactions Rc, Rs and Fp would be in nents of Rc and Rs.
equilibrium as shown previously in a simplified scheme 9
To verify if values of the present model are in agreement with
in Fig. 14b, where the component along axis x2 of Rc those provided by the multibody approach it is sufficient to multi-
equals that of Rs8. ply force R times the torque arm; as an example, at ϑ = 0 the
4
cs, z

present model leads to 329.7 [N] * 0.050023 [m] = 16.49 [Nm]

whereas ADAMS to 14.22 [Nm]. As stated, the difference may
8
In Fig. 14b the polygon of forces has been simplified by omitting well be accepted since friction between pistons and cylinder block
the contribution of piston inertia. Owing to this the polygon turns has been neglected in Adams.

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 67

 Alessandro Roccatello and Nicola Nervegna

Spline: pressure piston 1 −8000

1.2

Force [N]
−9000
1
−10000
0.8 Rb1(x1)
−11000
p/p* [−]

0.6 0 4 50 100 150 200 250 300 350

x 10
r

4
Rb1(y1)
0.4
3.5

Force [N]
0.2 3

0 2.5

2
0 4 50 100 150 200 250 300 350
0 100 200 300 400 500 600 700 x 10
theta [deg] 3
Rb1(z1)
Fig. 25: Instantaneous pressure within a variable volume

Force [N]
2.5
chamber (800 points for cubic spline, not all
shown)
2

800 1.5
______ = present model 0 50 100 150 200 250 300 350
theta [deg]
600 − − − − − − = ADAMS results
Rs(y2) Fig. 28: Comparison on reactions Rb1
400

200 ______ = present model

Rb2(x1)
Force [N]

9000
0 − − − − − − = ADAMS results

8500
Force [N]

−200

8000
−400

Rs(x2)
−600 7500

−800 7000
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
theta [deg] Rs(x2) 4
x 10
−0.6
Fig. 26: Comparison on reactions Rs Rb2(y1)
−0.8

800
Force [N]

______ = present model −1

Rc(x2)
600 − − − − − − = ADAMS results
−1.2

400
−1.4
200
−1.6
Force [N]

0 50 100 150 200 250 300 350

0 theta [deg]

−200 Fig. 29: Comparison on reactions Rb2

Rc(y2)
−400
340
1500 rpm−p*r
present model
−600
320
−800
0 50 100 150 200 250 300 350
theta [deg] 300 295.5
Torque [Nm]

Fig. 27: Comparison on reactions Rc

280

260
ADAMS results
Measured torque
240
Mean torque: present model = 294.7 Nm
Adams model = 293.4 Nm
220
0 50 100 150 200 250 300 350 400
theta [deg]

Fig. 30: Comparison of pump instantaneous and mean torque

68 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

may be fulfilled the shaft will transfer a positive torque

of (+ 6.1 Nm) to the block corresponding to a reaction
4
Rcs, z > 0.
On the contrary at ϑ = 38° Fig. 33 shows that
torque (+ 4.5 Nm) acts now in the CCW direction as is
the case for ω: equilibrium will then require that the
shaft transfers a negative torque. One supplementary
check may be obtained through the Adams model by
transferring motion from the shaft to the cylinder block
through three-dimensional contacts among bevel gears
teeth10 rather than via the inline joint: Fig. 34 shows a
plot of reaction Rcs (along axis x2, see Fig. 31) that
demonstrates how its behaviour, similar to that ob-
tained from the present model, clearly reveals the
aforementioned periodic change in sign.
250
Fig. 31: Reactions Rc, ϑ = 0 Force Rcs (x2 component)
200

150
present model result
4
400 Present model result Rcs 100
z Cc=0.0054 Nm/(rev/min)
300 50

Force [N]
0
Force [N]

200
−50

100 −100

−150
0
Cc=0 −200
ADAMS result (3D contact for bevel gear)
−100 −250
0 100 200 300 50 100 150 200 250 300 350
theta [deg]
Adams result: torque on cylinder block joint
15
Cc=0.0054 Nm Fig. 34: Reaction cylinder block-shaft (present model and
rev/min 3D contacts in ADAMS)
Torque [Nm]

10
From what has been written the conclusion may be
5 drawn that, for this type of pump, the torque required to
keep the unit turning at constant speed is rather low
0 (Ivantysyn and Ivantysynova, 2000): moments that
Cc=0
need be balanced are those stemming from friction,
−5
0 100 200 300 inertia and reactions transmitted from pistons to the
theta [deg] cylinder block.
Fig. 32: Periodic Force and Torque

8 Conclusions
This research paper has presented the mechanical
model of a fixed displacement bent axis pump. Pump
kinematics has been addressed introducing four coordi-
nate systems that were found expedient in the subse-
quent formulation of needed equations. Interesting
peculiarities of piston kinematics have been pinpointed.
Modelling phases led to the build up of a dedicated
library coded in Fortran and integrated in AMESim to
enrich capabilities that were limited to the fluid-
dynamics of bent-axis pumps. The only available ex-
perimental data were relative to the torque required to
drive the unit at constant speed under different loading
conditions. Consequently, validation was effected util-
izing these data as references and attained results were
Fig. 33: Reactions Rc, ϑ = 38 satisfactory. However, the effort required in modelling
also aimed at the evaluation of internal forces ex-
The resultant negative torque (- 6.1 Nm) acts in the
CW direction which is opposite to the shaft angular 10
The specific 3D contact algorithm of ADAMS is used.
speed (CCW); consequently in order that equilibrium

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 69

 Alessandro Roccatello and Nicola Nervegna

changed among intervening pump components. In this 110

|Rs|
respect, since specific experimental data were unavail- 100 @ 2350 rpm
able being, altogether, rather difficult or even impossi- real piston
90
ble to obtain, a full three dimensional virtual model of
the unit was assessed in the multibody code Adams. 80

This second approach was used to provide opportuni-

Force [N]
70
ties of performing cross-verifications with the original
60 2 masses
AMESim predictive results of exchanged forces. Also
in this case a fair agreement was confirmed. The 50
@ 1500 rpm real piston
AMESim model is, at this stage, more flexible and 40
complete than the ADAMS counterpart also allowing
30
the hydraulic simulation of the pump and its interac-
2 masses
tions with the circuit it is feeding. Furthermore, being 20
50 100 150 200 250 300 350
fully parametric, it permits with relative ease to gain theta [deg]
quantitative knowledge of the effects entailed by
Fig. 37: Reaction Rs (ADAMS)
changes in one or several geometric parameters on
pump hydraulic and mechanical performance.
On the contrary, the ADAMS model, tied with an
imported 3D-CAD geometry of the specific unit, defi-
nitely lacks this flexibility. Obviously, at least this Acknowledgments
restraint, can be subdued by generating anew pump
components through purposely written macros (Roc- The present work has been performed under a re-
catello et al., 2007), yet at the expense of rather marked search contract with Casappa S.p.A.. Authors acknowl-
efforts. edge permission of publishing the present material.

Appendix: Remarks on the Modelling of

Piston Inertia
Dealing with piston inertia (subheading 4.1.1) an
assumption was made: the piston mass was concen-
trated in points A and B with values mA and mB. It is
then appropriate to verify if this is acceptable or if
erroneous approximations in the evaluation of piston-
shaft (Rs) and piston-cylinder block (Rc) reaction
forces are introduced. To check this aspect use has been
made of the ADAMS multibody pump model (Fig. 35).
Two approaches were followed: the first portrays the
Fig. 35: ADAMS multibody model piston as a massless rod linking two equal masses at
points A and B. The second retains true piston geometry
70 and mass. Constraints are identical and conform with
|Rc| those already explicited in this paper: point A attaches
60
2 masses
to the shaft and the piston (spherical joint); point B
adapts to the cylinder axis (inline joint). A constant
50 @ 2350 rpm angular speed was applied to the shaft and the pump
was run in absence of loads. Fig. 36 confronts the be-
real piston
Force [N]

40
haviour of Rc at 1500 and 2350 rpm using the two
30
approaches, whereas Fig. 37 provides the same infor-
@ 1500 rpm
2 masses mations for Rs. Though differences exist (most evident
20 at max rated pump speed), observing that involved
real piston forces are at least an order of magnitude smaller than
10 other intervening forces (e.g. see Fig. 14), the hypothe-
ses set forth in the modelling of inertia effects are
0
50 100 150 200 250 300 350 deemed acceptable.
theta [deg]

Fig. 36: Reaction Rc (ADAMS)

70 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72

 Mechanical Modelling of a Bent Axis Pump

Nomenclature Rs, k Reaction on shaft from piston k

Sm Coordinate system m
Ak Centre of the spherical piston Td Drive torque required from prime
joint. A mover (component along z axis)
Ac Area on which cylinder pressure Tsf Friction torque on shaft
acts Tcf Friction torque on cylinder block
Bk Centre of the elastic ring Th Moment of force Fhy due to its
Cc Viscous friction coefficient (cyl- offset position from axis z2
inder block rotation) Tex Drive torque required from prime
Cij Direction cosine mover (vector)
Ck Point on piston where the reac- Z Number of teeth of bevel gears
tion force with the cylinder block im, j m, Unit vector of the axes of coordi-
is applied km nate system Sm
Cs Viscous friction coefficient (shaft n a Xq component along axis q of the
rotation) acceleration of point X in coordi-
Ct Torque loss coefficient (Coulomb nate system n
friction) b1 Distance of pointAk from Bk
Dpi Cylinder diameter b Distance between origins O1 and
Fh Hydrodynamic force O2
FiA, k Inertia force (mass lumped at d Bevel gears conical pitch diame-
point Ak) ter
FiB, k Inertia force (mass lumped at g Acceleration of gravity
point Bk) m Bevel gears module
Fm Spring force mA, mB Masses associated with points Ak
Fp, k Force from fluid pressure on and Bk
piston ms, mc, Shaft, cylinder block and piston
Fpf, k Piston-cylinder block friction mpi masses
force Pk Pressure in variable volume
Fth Hydrostatic component force chamber of piston k
pulling the cylinder block away
from the portplate (acts at Hth) p*r Rated pressure
Ftp, k kHydrostatic force (acts at Htp) p*I Intermittent (peak) pressure
Gpi, Gt, Piston, cylinder block and shaft nrX, q component along axis q of the
Gs centres of mass position of point X in coordinate
Mnm Matrix linking coordinate system system n
m with n nvX, q component along axis q of the
N Number of pistons velocity of point X in coordinate
O1 Intersection between shaft axis system n
and plane hosting centres of x m , y m, Cartesian axes of coordinate
spherical piston joints zm system S
O2 Intersection between shaft and α Tilt of cylinder block axis with
cylinder block axes the shaft axis
O4 Point of application of the force αp Pressure angle in bevel gears
exchanged between shaft and δa/t pitch angles in bevel gears
cylinder block through bevel ϑ, ϑk Angular position of piston 1 (ϑ)
gears and piston k
On, Om Origins of ‘new’ and ‘old’ coor- χ Mass fraction associated with
dinate systems (matrix M nm). point A
Pc ,Ps Cylinder block, shaft and piston Angular speed.
,Ppk weights k reference to the specific piston k =
Ra Distance from axis z1 of centres 1...N
of spherical piston joints […]T Matrix transponse
Rc, k Reaction on piston from cylinder
block
Rcs Reaction on shaft from cylinder
block
Rcn Reaction on pin from cylinder
block (acts at Hcn)
Rd Distance between cylinder and
cylinder block axes
Rb1 Rb2 Reactions on shaft from tapered
conical and cylindrical bearings
(points C1 and C2)

International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72 71

 Alessandro Roccatello and Nicola Nervegna

References Alessandro Roccatello

Graduated in Mechanical Engineering (2003)
from Politecnico di [Link] graduation
Caretto, R., Mancò, S., Nervegna, N. and Rundo, M. he joined the Fluid Power Research Labora-
tory of the Politecnico di Torino, focusing on
1996. Modelling, Simulation and Experimental the simulation of axial piston pumps. His
Studies on a Variable Displacement Radial Piston research fields are mainly multibody simula-
Pump Prototype for Automotive Applications. The tion, tribology, axial piston pumps and
motors modelling.
ASME International Mechanical Engineering Con-
gress and Exposition, Fluid Power Systems and
Technology, Vol. 3, pp. 1-9.
Nicola Nervegna
Fabiani, M., Mancò, S., Nervegna, N. and Rundo, M. Graduate in Nuclear Engineering (1971) from
Politecnico di Torino, Italy. He joined the
1999. Modelling and Simulation of Gerotor Gearing Politecnico staff in October 1971 and, at
in Lubricating Oil Pumps. SAE International Con- present, is Professor of Fluid Power Systems.
gress and Exposition, Detroit, March 1-4, SAE pa- His research interests lie in the broad fields of
Fluid Power with involvement in positive
per 1999-01-0626. displacement pumps as well as in the model-
ling, simulation and testing of various hy-
Gad, O., Rabie, M. G. and El-Taher, R. M. 2002. draulic components.
Prediction and Improvement of Steady-State Per-
formance of a Power Controlled Axial Piston Pump.
Journal of Dynamic Systems, Measurement, and
Control, Vol. 124, Issue 3, pp. 443-451.
Ivantysyn, J. and Ivantysynova, M. 2000. Hydrostatic
Pumps and Motors: Principles, Design, Perform-
ance, Modelling, Analysis, Control and Testing.
Akademia Books International. ISBN: 185522162.
Jacazio, G. and Piombo, B. 1997. Meccanica appli-
cata alle macchine, Vol. 2. Levrotto & Bella
(Torino). ISBN: 888218014X.
Litvin, F. L. and Fuentes, A. 2004. Gear geometry
and applied theory. Cambridge University Press.
ISBN: 0521815177.
Mancò, S., Nervegna, N., Lettini, A. and Gilardino,
L. 2002. Advances in the simulation of axial piston
pumps. Fifth Japan Fluid Power Society Interna-
tional Symposium on Fluid Power, Vol. 1, pp. 251-
258. ISBN: 4931070053.
Mancò, S., Nervegna, S., Rundo, N. and Armenio,
M. G. 2004. Modelling and Simulation of Variable
Displacement Vane Pumps for IC Engine Lubrica-
tion. SAE International Congress and Exposition,
Detroit, March 8-11, SAE paper 2004-01-1601.
Manring, N. D. and Dong, Z. 2004. The Impact of
Using Secondary Swash-Plate Angle Within an Ax-
ial Piston Pump. Journal of Dynamic Systems,
Measurement, and Control, Vol. 126, pp. 65-74.
Roccatello, A., Mancò, S. and Nervegna, N. 2007.
Modelling a Variable Displacement Axial Piston
Pump in a Multibody Simulation Environment.
Journal of Dynamic Systems, Measurement, and
Control, Vol. 129, Issue 4, pp. 456-468. ISSN:
0022-0434.
Wieczorek, U. and Ivantysynova, M. 2002. Computer
Aided Optimization of Bearing and Sealing Gaps in
Hydrostatic Machines - The simulation Tool CAS-
PAR. International Journal of Fluid Power, No. 1,
pp. 7-20.

72 International Journal of Fluid Power 10 (2009) No. 2 pp. 57-72