Вы находитесь на странице: 1из 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271963978

Modelling and Simulation of Gerotor Gearing in Lubricating Oil Pumps

Article  in  SAE Technical Papers · March 1999


DOI: 10.4271/1999-01-0626

CITATIONS READS

50 291

4 authors, including:

Massimo Rundo Nicola Nervegna


Politecnico di Torino Politecnico di Torino
36 PUBLICATIONS   290 CITATIONS    40 PUBLICATIONS   337 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Massimo Rundo on 08 February 2015.

The user has requested enhancement of the downloaded file.


1999-01-0626

Modelling and Simulation of Gerotor Gearing


in Lubricating Oil Pumps
M. Fabiani, S. Mancò, N. Nervegna, M. Rundo
Politecnico di Torino

G. Armenio, C. Pachetti, R. Trichilo


Magneti Marelli

Copyright © 1999 Society of Automotive Engineers, Inc.

ABSTRACT nisms. This paper is instead focused on gearing kinematics


and also provides updates in simulated and experimental
The paper presents geometric and kinematic aspects that results.
constitute a premise to the modelling and simulation of
gerotor lubricating oil pumps. With reference to a commer- GEARS KINEMATICS AND PROFILES
cial oil pump two different modelling approaches of the
pumping elements are addressed: the classical integral- The gerotor pump consists of an internal gear having one
derivative approach and the new derivative-integral tooth less than the external. Each tooth of the inner ele-
approach. The latter, based on volumes swept by vector ment is in continuous sliding contact with the surface of the
rays, is easier to implement and requires less computer outer element. This keeps an oil seal between them and
time at equal accuracy. Two approaches to modelling are prevents backward slippage as oil is displaced through the
also detailed that feature different reticulations of the pump pump.
and consequently involve a different number of ordinary dif- The external gear profile is composed of N circular arcs
ferential equations (ODE). Depending on the extent and with radius S (Fig. 1) with centres at a distance K from the
detail of expected informations, either 4 or N+2 ODE must gear centre O1. The internal and external gearing must
be solved, N being the number of variable volume cham-
bers in the pump. Finally, numerical results of the simula-
tion code, developed in the AMESim environment, have
been compared with experimental results presented else-
where [4].

INTRODUCTION

A consistent effort in current automotive research is ori-


ented to identify areas where a potential exists for fuel con-
sumption and emissions reduction. Lubricating oil pump
systems can make a contribution in this direction and,
depending on the oil pump type, the potential of fuel reduc-
tion amounts up to 3% in the European MVEG-Cycle [1].
However, the development of the oil pump is increasingly
becoming a special discipline [2] that, due to the high costs Figure 1. The pump gearings
of test stand fixtures and prototypes evolution, extensively
involves CAE and virtual modelling tools for simulating and engage and therefore are conjugate. In meshing, pitch cir-
optimizing the pump. Performance comparisons of various cles roll without slip (in C, Fig. 2) about their respective cen-
pumping systems have been presented and several varia- tres O1 and O2 at angular velocities ω1 and ω2 with a gear
tions introduced with different tooth forms [3]. In particular ratio of (N-1)/N. Different methods of generation of cycloi-
in this paper, an internal-external gear pump configuration dal gearings are known [8]. One of these relies on the basic
(Gerotor) will be considered that has no crescent divider. A law of gearing which states: ‘Normal to the profiles of mat-
recent paper [4] dealt with the modelling, simulation and ing teeth must, at all points of contact, pass through a fixed
testing of a gerotor unit emphasizing flow losses mecha- point located on the line of centres’, and identifying point C

989
Figure 3: Pump schematic

Figure 2: profiles generation Selecting appropriate control volumes, internal pressure


gradients must be determined according to the following
(Fig. 2) as the instantaneous centre of relative motion (pitch paradigm equation:
point), equations of the contact line, in the reference frame
integral with the external gear, can be determined: β
= ----------- ⎛ ∑ Q e – ∑ Q u – ω 1 ⎞
dp dV
(4)
dα Vω 1 ⎝ d α⎠
S
⎧ x ( α ) = K cos α – ---- ( K cos α + R 1 )
⎪ m
Two approaches exist in selecting control volumes. One
⎨ (1)
⎪ y ( α ) = K sin α – ---
S
-K sin α involves N variable chambers volumes and two constant
⎩ m volumes (inlet and delivery ducts) leading to a system of
2 2 N+2 ordinary differential equations similar to eq. (4). The
where: m = R 1 + K + 2R 1 K cos α . In a reference frame cen- second considers only four volumes (see Fig. 4): (i) delivery
tred in O2 and integral with the internal gear, the contact
line (eq. (1)) describes its profile:

α
K cos ⎛ -------------⎞ + e cos ⎛ ------------- α⎞ –
N
⎝ N – 1⎠ ⎝N – 1 ⎠
xi ( α ) =
α
---- K cos ⎛ -------------⎞ + R 1 cos ⎛ ------------- α⎞
S N
m ⎝ N – 1⎠ ⎝N – 1 ⎠
(2)
α
– K sin ⎛ -------------⎞ – e sin ⎛ ------------- α⎞ +
N
⎝ N – 1⎠ ⎝N – 1 ⎠
yi ( α ) =
α
---- K sin ⎛ -------------⎞ + R 1 sin ⎛ ------------- α⎞
S N
m ⎝ N – 1⎠ ⎝N – 1 ⎠

Selection of this frame is equivalent to impress on gears an


angular velocity - ω2 so that, while the internal stands still,
the external rotates at a velocity: Figure 4: Control volumes

N ω1 Vm where connected chambers are lumped together and


ω 1 – ω 2 = ω 1 – ------------- ω 1 = – ------------
- (3)
N–1 N–1 with the delivery duct; (ii) suction Va where connected
while its centre O1 follows an orbital motion about O2 with chambers are lumped together and with the inlet duct; (iii)
an angular velocity - ω2. As a consequence, if in the abso- minimum trapped volume Vimin ; (iv) maximum trapped vol-
lute reference frame the external gear makes one anticlock- ume Vimax .
wise revolution, in the reference integral with the internal Regardless of the approach, knowledge of volumes, their
gear it will make 1/(N-1) clockwise revolutions and eq. (2) derivatives and flow passage areas is a prerequisite for
will only describe one tooth; to obtain the entire gear pro- model simulation. Specifically, the hydraulic model for a sin-
file, α must vary between 0 and 2π ( N – 1 ) . gle chamber is shown in Fig. 5. Volume V j ( α ) represents
the variable volume of the j-th chamber, Vcm and Vca are
MATHEMATICAL MODEL - FIRST APPROACH constant volumes associated with the delivery and inlet
ducts respectively while variable restrictors areaa(α) and
A schematic of the pump gearings and casing is shown in aream(α) represent flow passage areas. Load is modelled
Fig. 3.
990
tally to coincide with the external gear frame. Internal gear
profile equations become:
S
x i ( ϑ ) = K cos ϑ + [ cos ( Nϑ ) – 1 ]e – ---- [ K cos ϑ + R 1 cos ( Nϑ ) ]
m
(6)
S
y i ( ϑ ) = K sin ϑ + e sin ( Nϑ ) – ---- [ K sin ϑ + R 1 sin ( Nϑ ) ]
m

2 2
where: m = R 1 + K + 2R 1 K cos ( N – 1 )ϑ
α
and: ϑ = – ------------
-
N–1
Figure 5: Equivalent hydraulic model for a single chamber
To determine the external profile, the following constants
with a restrictor of section Am. are defined (see Fig. 7):
In this instance we have:

⎧ dp
β ⎛ dV
⎪ j = ------------ - Q – Q u, j – ω 1 j⎞ j = 1, 2, …, N
⎪dα V j ω 1 ⎝ e, j dα ⎠

⎪ N
β ⎛
⎪ dp m = ---------------- ⎞
V cm ω 1 ⎝ ∑ u, j
⎨dα - ⎜ Q – Q m⎟ (5)
⎪ 1


⎪ dp a β ⎛
N

⎪ = ---------------- ⎜ – ∑ Q e, j + Q a⎟
⎪dα V ca ω 1 ⎝ ⎠
⎩ 1

Figure 7: Tooth parameters of the external gear

yo
σ = atan ⎛ -----⎞ yo
τ = π – atan ⎛ ---------------⎞
⎝ x o⎠ ⎝ K – x o⎠
(7)
2 2 2
K +G –S 2 2
x o = -------------------------------- yo = G – xo
2K

Setting Λ = i ⋅ ∆α ( i = 0, 1, …, N – 1 ) the external profile


equations follow:

⎧ x e ( θ ) = K cos Λ + S cos ( π + Λ – θ )
⎨ 0≤θ<π–τ
⎩ y e ( θ ) = K sin Λ + S sin ( π + Λ – θ )

⎧ x e ( θ ) = G cos ( Λ + ξ ) π
⎨ σ ≤ ξ < 2 ---- – σ
Figure 6: Reference position and numbering of chambers ⎩ y e ( θ ) = G sin ( Λ + ξ ) N
(8)

⎪ x e ( θ ) = K cos ⎛⎝ Λ + ------⎞⎠ + S cos 2π ⎛⎝ -------------⎞⎠ + Λ – τ – θ
2π N+1
ANALYSIS OF VOLUME VJ AND ITS VARIATION ⎪ N N
⎨ 0≤θ<π–τ
⎪ ⎛ 2π⎞ ⎛ N + 1⎞
the integral - derivative approach ⎪ y e ( θ ) = K sin ⎝ Λ + -----
N⎠
- + S sin 2π ------------- + Λ – τ – θ
⎝ N ⎠

Volume Vj can be obtained by integration of equations
Equation (6) and eq. (8) determine gears profiles in the ref-
describing gears profiles and subsequent difference erence set forth in Fig. 6. For any value of α, rotation formu-
between the two integrals. A precondition is that (a) para- las must be applied to eq. (6) and eq. (8):
metric equations of the two profiles must be written in the
same reference frame and (b) profiles, as their parameter ⎧ x i ( α ) = x i cos α – y i sin α ⎧ x e ( α ) = x e cos α – y e sin α
changes, be generated in the same sense of rotation start- ⎨ ⎨ (9)
⎩ y i ( α ) = x i sin α + y i cos α ⎩ y e ( α ) = x e sin α + y e cos α
ing from point P1. In this respect, a reference condition at
α = 0° is set for the engaging gears and shown in Fig. 6. To perform integration polar coordinates ρ ( α ), ψ ( α ) are
Accordingly, the frame centred in O2 is displaced horizon- appropriate where:

991
2 2 through eq. (9) and eq. (11) the angles ψ ;
ρ(α) = x(α) + y(α) (10)
• for each gear, when ψ 1 ≤ ψ ≤ ψ 2 , ρ ( α ) is determined
and through eq. (10) and used in eq. (15).
dV
⎧ ( a = 0 ) if ( x ≥ 0 ) ( y ≥ 0 ) Once V j ( α ) is known, numerical differentiation yields ---------j .

y(α) ⎪
ψ ( α ) = atan ⎛ -----------⎞ + aπ where ⎨ ( a = 1 ) if ( x < 0 ) (11)
⎝ x ( α )⎠

⎩ ( a = 2 ) if ( x > 0 ) ( y < 0 ) the derivative- integral approach (vector rays)

The area subtended by a curve in polar coordinates is: This alternate approach allows direct evaluation of volume
variation and requires numerical integration to determine
1 ψ2 2
A = --- ∫ ρ dψ (12) the chamber volume. This volume is delimited by gears
2 ψ1 profiles and bounded by two contact points (e.g. A and B in
where ψ 1 and ψ 2 are angles that locate vectors of contact Fig. 9). Its variation can be determined through analysis of
points P1 and P2 delimiting chamber 1 (see Fig. 8). Angle (i) two control volumes1 (see Fig. 9):
• Vi relative to the internal gear and bounded by the
ψ 1 ( α ) is obtained applying eq. (11) to eq. (1). Since, after
polyline AO2B and the line of contacts;
rotation of the external gear equal to one angular pitch ∆α ,
• Ve relative to the external gear and bounded by the arc
point P1 moves to the position that point P2 occupied prior
CD, the line of contacts comprised between A and B
to rotation, angle ψ 2 ( α ) can always be obtained from eq.
and the two segments BC and AD;
(1) through a phase shift:
ψ 2 ( α ) = ψ 1 ( α + ∆α ) (13)
In general, the j-th contact point stems from a shift of j-1
angular pitches:

⎧ ρ j ( α ) = ρ 1 ( α + ( j – 1 )∆α )
⎨ (14)
⎩ ψ j ( α ) = ψ 1 ( α + ( j – 1 )∆α )

By substitution of eq. (9) into eq. (10) and application of eq.


(12), chamber 1 volume can be evaluated as the difference
of two integrals:
Figure 9: Control volumes
1 ψ2 2 1 ψ2 2
V 1 ( α ) = --- H ∫ ρ e dψ – --- H ∫ ρ i dψ (15) and (ii) four vector rays identified as follows:
2 ψ1 2 ψ1
• ρi1 and ρi2 originated at O2 and ended at contact points
where ρ i and ρ e refer to the internal and external gears A and B;
respectively. For the j-th chamber one simply obtains: • ρe1 and ρe2 originated at O1 and ended at D and C.
V j ( α ) = V 1 ( α + ( j – 1 )∆α ) . In fact, if a rotation is impressed on gears, Vi and Ve change
due to volumes swept by vector rays ρi1 and ρi2 for the inner
gear and by segments AD and BC (belonging to ρe1 and
ρe2) for the outer gear. However, it must be observed that
contact points as well as vector rays rotate a different angle
than gears so that the gearing solid volume comprised in
control volumes is variable. Accounting for this it is possible
to evaluate the net volume change of chambers.
Vi - Considering an infinitesimal rotation dα of the external
N
gear, the internal will rotate of an angle dϕ = ------------- ⋅ dα
N–1
while contact point A undergoes a rotation dψ i1 (see Fig.
10). Neglecting to a first approximation its variation in
2 1
Figure 8: Profile integration (internal gear) length, vector ray ρi1 will sweep a volume --- Hρ i1 dψ i1 ,
2
However, ρ ( ψ ) is not known explicitly and use of eq. (15) thereby reducing control volume Vi . Since the internal gear
rotates faster than point A (i.e. dϕ > dψ i1 ) the small solid
requires the following:
• evaluate through eq. (1), eq. (11), eq. (14) angles
ψ 1 ( α ) and ψ 2 ( α ) of points of contact;
• vary parameter ϑ in eq. (6) and eq. (8) and evaluate 1.for a better understanding reference is made to a 6-7 gearing couple

992
1 2
slice O2A’A’’’ with volume --- Hρ i1 dγ i1 will enter control vol-
2
ume Vi . Consequently, the net volume variation will be:
1 2
--- Hρ i1 ⋅ ( dψ i1 – dγ i1 ) . This, being dγ i1 = dϕ – dψ i1 , can also
2
1 2
be written as --- Hρ i1 dϕ . The same happens with contact
2
point B and therefore:
1 2 2
dV i = --- H ⋅ ( ρ i2 – ρ i1 ) ⋅ dϕ (16)
2

Ve - Considering again a rotation dα of the external gear


(see Fig. 11) the ray vector ρe1 will rotate of an angle dψ e1
about O1 and its portion AD will reduce control volume Ve
1 2 2
of the quantity --- H ( G – ρ e1 ) ⋅ dψ e1 . In this case, the external
2
gear is slower than the contact point (i.e. dα < dψ e1 ) and the
1 2 Figure 11: External gear volume variation
small slice A”D”A’’’D’’’ with volume --- Hρ e1 dγ e1 exits the con-
2
1 2 2 2 2
trol volume. Therefore, the net volume variation is: dV c = --- H [ ( ρ e1 – ρ e2 )dα + ( ρ i2 – ρ i1 )dϕ ] (20)
2
1 2 2 1 2 2
--- H ( G – ρ e1 )dψ e1 – --- H ( G – ρ e1 )dγ e1 . (17) From eq. (20) the derivative of chamber volume follows:
2 2

This, with dγ e1 = dψ e1 – dα , can be written as dV c 1 2 2 2 2 N


= --- H ( ρ e1 – ρ e2 ) + ( ρ i2 – ρ i1 ) ------------- (21)
2
dα 2 N–1
1 2
--- H ( G – ρ e1 )dα . By analogous reasoning on contact point B
2 and, through integration:
the variation of volume Ve is:
α
dV c
1 2 2 V1 ( α ) = V1 ( 0 ) + ∫ dα (22)
dV e = --- H ⋅ ( ρ e1 – ρ e2 ) ⋅ dα (18) dα
2 0

A critical comparison of the two approaches in terms of


accuracy and time to solution is reported in the Appendix.

FLOW PASSAGE AREAS

As gears rotate the isolated chamber shown in Fig. 12 will

Figure 10: Internal gear volume variation

It is now possible to determine the total chamber volume


variation as :
dV c = ( dV e + dV i ) (19) Figure 12: Gearing and porting plate

Substituting expressions previously obtained: eventually cross the rim of the porting plate towards deliv-
ery. A restricted flow passage is formed delimited by con-
tact point P2 and that rim (see Fig. 13a). In turn, also the

993
other contact point P1 goes beyond the boundary and the
flow passage becomes unrestricted and coincident with the
chamber plane surface (Fig. 13b). With reference to Fig. 14

Figure 13: Restricted and unrestricted flow passages

port angles ψ bd, ψ ed, ψ bi, ψ ei are recognized by intersec-


tions of the line of contacts with almost radial edges of the
porting plate.

Figure 15: Vector rays tracking

Reasoning for passage flow area aream is analogous. As


characteristic parameters of the porting plate, with refer-
ence to Fig. 16, the following angles χ bd , χ ei and dis-
tances ded e dbi are set. Port angles are then evaluated as
functions of these parameters. A plot of aream and areaa is

Figure 14: Port angles

To identify flow passages of chamber 1 it can be expedient


to plot vector rays (1 and 2) angles ψ 1 ( α ) and ψ 2 ( α )
through use of eq. (1), eq. (11) and eq. (14) and to include
the port angles (see Fig. 15).
Figure 16: Porting plate parameters
the integral - derivative approach
reported in Fig. 17.
If a restricted flow passage exists (vector rays internal to
either ∆ψ′ or ∆ψ″ , Fig. 14), the flow area can be calculated the derivative - integral approach (vector rays)
through eq. (15), one of the limits of integration being
replaced with the appropriate angle ψ ; if the flow is unre- Earlier in this paper vector rays have been used to evaluate
stricted then the flow passage coincides with the plane sur- chamber volume variations and volumes themselves. In
face. that case the problem was simple since the chamber was
For section areaa with reference to Fig. 15 we obtain: always confined by two consecutive points of contact of the
mating gears. It was therefore sufficient to use the line of
⎧ 1 ψ″ 2 1 ψ″ 2 contacts through its parametric equations (eq. (1)). In the
⎪ areaa ( α ) = --- ∫ ρ e dψ – --- ∫ ρ i dψ α 2bi ≤ α ≤ α 1ei present instance the problem is more complex due to the
⎨ 2 ψ′ 2 ψ′ (23)
⎪ areaa ( α ) = 0 α < α 2bi α > α 1ei fact that in situations where a chamber is restricted (by far

the most interesting) the flow passage to be determined is
where: confined within one point of contact and one of the porting
plate rims. This requires that the approach and what was
⎧ ψ′ = ψ bi α ≤ α 1bi ⎧ ψ″ = ψ 2 ( α ) α < α 2ei implied in eq. (21) be validated in the present situation
⎨ if ⎨ if (24) where two vector rays stay fixed while the other two follow
⎩ ψ′ = ψ 1 ( α ) α > α 1bi ⎩ ψ″ = ψ ei α ≥ α 2ei
the contact point. An additional issue is the attainment of
994
To evaluate vector rays ρi1 and ρe1 between gears centres
1bd 1ed 1bi 1ei
80 and points Ai and Ae (Fig. 18), the determination of gears
2ed 2bi 2ei 2bd
70 profile in polar coordinates is necessary.
External gear - Beside parametric equations (eq. (8)) a cor-
60 DUHDP DUHDD
relation between the auxiliary angle θ and the anomaly ξ
50 must be found.
DUHD>PP @


40
Correlation between parameters ξ and θ
30

20 For the first portion of the tooth, we introduce:


10
ξ′ = ξ (25)
0
0 40 80 120 160 200 240 280 320 360 While for the second:
α>GHJ@

ξ′ = ξ – ------ (26)
Figure 17: Passages aream and areaa N

profile equations for both mating gears (see ahead). So, Looking at the triangle O1BC in Fig. 7 it is possible to write
when the chamber is restricted, vector rays ρe2 and ρi2
K S
track points belonging to the line of contacts, whereas ρe1 --------------------------- = ------------
sin ( θ + ξ′ ) sin ξ′
(27)
and ρi1 identify intersections of the external (point Ae) and
that may be further expressed as:
internal (point Ai) gear profiles with the porting plate rims.
This implies that, as gears rotate, the length of vector rays K
tan ξ′ cos θ + sin θ – ---- tan ξ′ = 0 (28)
S

To solve eq. (28) with respect to θ we use the following


π
substitutions in cases where θ ≠ --- :
2

2
2t 1–t θ
sin θ = ------------- cos θ = ------------- t = tan --- (29)
2 2 2
1+t 1+t

π S
instead if θ = --- from eq. (28) we obtain: tan ξ′ = ---- .
2 K
Introducing eq. (29) into eq. (28) we have a quadratic equa-
tion in the variable t that has two solutions: when ξ′ = 0
then θ = 0 whereas if ξ′ ≠ 0 we have:

K 2
Figure 18: displacement of points delimiting control volumes 1− + 1 – ( tan ξ′ ) ⎛⎝ ----⎞⎠ – 1
S
t = ------------------------------------------------------------------- θ(ξ′) = 2 atan ( t ) (30)
( tan ξ′ ) ⎛ ---- + 1⎞
ρe1 and ρi1 varies, while their angular position is invariant K
⎝S ⎠
(in Fig. 18 point Ae will be displaced to Ae’’’ while point Ai
will move on to Ai’’’). From Fig. 7 it is evident that at a given ξ two intersections
In particular, for an infinitesimal rotation, the two hatched exist with the circle having radius S. However, since the
areas in Fig. 18 identify (account being taken of gears π
tooth extends for an interval --- ≤ τ ≤ π (see Fig. 7), that
thickness) volumes aggregated to Vi and, respectively, Ve. 2
1 2 1 2 2 closer to the origin is significant.
Their values of --- ⋅ Hρ i1 dϕ (internal gear) and --- H ( G – ρ e1 )dα Internal gear - What is needed is a polar equation of the
2 2
(external gear) can then be used in assessing volumes form ρ = ρ(ψ) where:
variations dVi and dVe. As can be easily verified, these are
y i(α)
expressed exactly as in eq. (16) and in eq. (17). Further- ψ = atan ------------ ρ = x2 + y2 (31)
x i(α)
more, the infinitesimal variation of chamber volume is again
written as in eq. (19) and by appropriate substitution the Accordingly, given the anomaly ψ , it would suffice to obtain
same result stated in eq. (21) is obtained. This validates the
from the first relation in eq. (31) α = α(ψ) and use it in the
approach.
second to evaluate ρ. This has not been possible. The sim-
plest viable structure found for the first relation in eq. (31)
GEARS PROFILE IN POLAR COORDINATES
is, [5]:

995
α B sin α
ψ = ------------- + atan --------------------------- (32)
N–1 A + B cos α 80
1bd 1ed 1bi 1ei
2ed 2bi 2ei 2b
where A and B are also functions of α. No further simplifica- 70

tion of eq. (32) was attainable, so an iterative numerical 60


solution was adopted. Fig. 19 shows the plot of ψ as a func-

DUHD>PP @
50


tion of α. By so doing, for each gears position, vector rays 40
length have been calculated and this also when one of the
30
boundaries is formed by the rim of either the discharge or
suction ports. 20

10
360
0
0 40 80 120 160 200 240 280 320 360
α>GHJ@
270

Figure 21: Flow passage and chamber plane area


ψ>GHJ@

180

90

0
0 720 1440 2160 2880 3600

α >GHJ@

Figure 19: Plot of ψ as a function of α

As shown, the vector ray method directly yields the deriva-


tive of the flow passage areas (see Fig. 20) and through
subsequent integration the evaluation of areas themselves. Figure 22: Equivalent hydraulic models
There is then a need of establishing constants of integra-
tion. However, it can be noticed that the minimum value of and delivery volume when they become connected. When
flow areas both at inlet and discharge is obviously zero and trapped volumes are undefined (do not exist), their pres-
so is the constant of integration in either case. In this sure derivatives are set equal to those at delivery for the
respect Fig. 20 shows derivatives of the flow passage and min volume and at suction for the maximum. Model equa-
of the plane areas; these are indeed superimposed with the tions are shown hereafter.
exception of situations where the chamber is restricted.
⎧ dp m β dV m⎞
The same happens with integral curves (see Fig. 21 and ⎪ = --------------- ⎛ – Q m + Q r – ω 1 delivery volume
d α ω V ⎝ dα ⎠
compare it with Fig. 17). ⎪ 1 m

⎪ dp a β dV a⎞
⎪ = ------------- ⎛ Q a – ω 1 inlet volume
400
⎪ d α ω 1 aV ⎝ dα ⎠
1bd 1ed 1bi 1ei
⎨ (33)
300 ⎪ dp imin β dV
= --------------------- ⎛ – ω 1 ⎞ min. trapped volume
imin

2ei
ω 1 V imin ⎝ ⎠
2ed 2bd
200 2bi
⎪dα dα

>PP UDG@

100 ⎪ dp imax β dV imax⎞


⎪ = ---------------------- ⎛ – Q r – ω 1 max. trapped volume
ω 1 V imax ⎝ ⎠


0 ⎩dα dα

-100
ANALYSIS OF VOLUMES AND THEIR VARIATIONS
-200

-300 To identify chambers connected to inlet and delivery, it is


0 40 80 120 160 200 240 280 320 360 feasible to extend to all vector rays the plot shown in Fig. 15
α>GHJ@ and owing to cyclic repetitions, limit the analysis to one
Figure 20: Derivative of flow passage and of plane area pitch angle ∆α . Phase angles α bi, α ei, α bd, α ed presented
in Fig. 17 within the interval [0 - ∆α ] indicate the external
MATHEMATICAL MODEL - SECOND APPROACH gear rotation that must be applied so that a generic ray vec-
tor crosses, in due sequence, the angles ψ bi, ψ ei, ψ bd, ψ ed .
This second approach involves four control volumes only. In this respect vector rays delimiting the start points of dif-
Synthetic hydraulic models of such volumes are collected ferent control volumes are (see Fig. 23):
in Fig. 22. Qr is the flow rate between max trapped volume • begin delivery:

996
Delivery volume

For the analysis of volumes it is necessary to take into


account the variation of vector rays consequent to their
motion, but also of volume variations consequent to either
aggregation or separation of a chamber from either the inlet
or delivery volumes. When α=0 chambers numbered from 1
to (N-1)/2 are connected to delivery and therefore the con-
stant of integration V m0 for the delivery volume is:

V1 ( 0 ) + … + V N – 1 ( 0 ) =
-------------
2

V m0 = N–3
------------- (39)
2
= V 1 ( 0 ) + … + V 1 ⎛ ------------- ∆α⎞ =
N–3
⎝ 2 ⎠ ∑ V1 ( j ⋅ ∆α )
j=0

When α = α bd the delivery volume undergoes the addition


of the chamber N volume and it is therefore necessary to
account for this discontinuity V m01 :
Figure 23: Angles of contact point vector rays V m01 = V N ( α bd ) = V 1 ( α bd + ( N – 1 )∆α ) (40)

⎧ ρ1 ( α ) ψ 1 < ψ bd When α = α ed the delivery volume undergoes the removal


ρ bd ( α ) = ⎨ if (34)
⎩ ρ N ( α ) = ρ 1 ( α – ∆α ) ψ 1 ≥ ψ bd of chamber (N-1)/2 and therefore:

V m02 = V N – 1 ( α ed ) = V 1 ⎛ α ed + ------------- ∆α⎞


• end delivery: N–3
(41)
------------- ⎝ 2 ⎠
2
⎧ρ ⎛ N–1 ⎞ ψ N – 1 < ψ ed
N + 1 = ρ 1 ⎝ α + ------------
⎪ ------------ 2
- ∆α

-------------
2
⎪ 2 - The delivery volume can then be expressed as:
ρ ed ( α ) = ⎨ if (35)
⎪ ⎛ N–3 ⎞
⎪ ρN
------------- 2
- ∆α
– 1 = ρ 1 ⎝ α + ------------ ⎠ ψ N – 1 ≥ ψ ed
------------- V m ( α )=
⎩ 2 2

⎛ 0 ≤ α < α bd ⎞
• begin inlet: ⎧ α dV m ⎜ ⎟
⎪ V cm + V m0 +
⎪ ∫0 d α dα ⎜



⎧ρ ⎛ N+1 ⎞ ψ N + 3 < ψ bi ⎪
+ 3 = ρ 1 ⎝ α + ------------
-∆α ------------- ⎜ ⎟ (42)
⎪ N 2 ⎠ 2 ⎪ α dV m ⎜α ≤ α < α ⎟
⎪ ------------
2
- = ⎨V + V + V
m01 ∫ d α
+ d α ⎜ bd ed⎟
ρ bi ( α ) = ⎨ if (36) ⎪ cm m0
0 ⎜ ⎟
⎪ ⎛ N–1 ⎞ ⎪ ⎜ ⎟
⎪ ρN 2
- ∆α
+ 1 = ρ 1 ⎝ α + ------------ ⎠ ψ N + 3 ≥ ψ bi ⎪ α dV m ⎜ ⎟
m02 ∫ d α α
-------------
⎩ 2 -------------
2 ⎪ cm
V + V m0 + V m01 – V + d ⎜ ⎟
⎩ 0
⎝ α ed ≤ α < ∆α ⎠
• end inlet:
being Vcm the delivery duct volume.
⎧ ρ1 ( α ) ψ N < ψ ei
ρ ei ( α ) = ⎨ if (37)
⎩ ρ N = ρ 1 ( α – ∆α ) ψ N ≥ ψ ei Inlet volume

Substituting in eq. (21) the expressions found for vector Due to symmetry, the constant of integration for α=0 is the
rays it is possible to determine variations of the control vol- same found for the delivery volume:
umes:
V a0 = V m0 (43)
⎧ dV m 1 2 2 2 2 N
⎪ = --- H ( ρ ebd – ρ eed ) + ( ρ ied – ρ ibd ) ------------- When α = α bi the chamber (N+1)/2 is being added to the
⎪dα 2 N–1
⎪ inlet volume:
⎪ dV a 1 2 2 2 2 N
⎪ = --- H ( ρ ebi – ρ eei ) + ( ρ iei – ρ ibi ) -------------
⎪ d α 2 N –1
V a01 = V N + 1 ( α bi ) = V 1 ⎛ α bi + ------------- ∆α⎞
N–1
⎨ (38) ⎝ ⎠
(44)
------------- 2
⎪ dV imin 1 2 2 2 2 N 2
⎪ = --- H ( ρ eed – ρ ebi ) + ( ρ ibi – ρ ied ) -------------
⎪dα 2 N–1
When α = α ei the chamber N is being removed:

⎪ dV imax 1 2 2 2 2 N
⎪ = --- H ( ρ eei – ρ ebd ) + ( ρ ibd – ρ iei ) -------------
⎩ d α 2 N–1
V a02 = V N ( α ei ) = V 1 ( α ei + ( N – 1 )∆α ) (45)

997
and finally: from delivery; further, since the pump works at low pres-
sures the flow through the restrictor will be negligibly small.
Then, it is necessary to intervene on the definition laws of
Va ( α ) =
volumes and respective derivatives through a switch im:
⎛ 0 ≤ α < α bi ⎞ α > α bd p m – p imax > ∆p ε
⎧ α dV a ⎜ ⎟ im = 1 if and
⎪ V ca + V a0 +
⎪ ∫0 d α dα ⎜


⎟ im = 0 otherwise
(51)
⎪ ⎜ ⎟
⎪ α dV a ⎜α ≤ α < α ⎟
(46)
In eq. (51) ∆p ε represents a discriminating pressure drop;
a01 ∫ d α
= ⎨V + V + V + d α ⎜ bi ei⎟
⎪ ca a0
0 ⎜ ⎟
⎪ lower values indicate equal pressures in the two volumes.
⎜ ⎟
⎪ α dV a ⎜ ⎟ Chambers aggregation takes place after a flow passage
⎪ V ca + V a0 + V a01 – V a02 + ∫ dα ⎜ ⎟
⎩ 0 d α opening occurring at angle α col (when im shifts from 1 to
⎝ α ei ≤ α < ∆α⎠
0). We must distinguish between two cases, i.e. if the con-
where Vca is the inlet duct volume. nection is established before or after the ending of the
angular pitch under consideration.
Minimum trapped volume

At reference conditions (α=0) it is determined by chamber


(N+1)/2:

V imin0 = V N + 1 ( 0 ) = V 1 ⎛ ------------- ∆α⎞


N–1
⎝ 2 ⎠
(47)
-------------
2

whereas in the interval α bi ≤ α < α ed it is not defined,


accordingly:

⎧ α dV imin
( 0 ≤ α < α bi ) ( α ed ≤ α < ∆α )
⎪ V imin0 +
V imin ( α ) = ⎨ ∫0 d α d α
(48)

⎩ 0 α bi ≤ α < α ed
Figure 24: Possible locations of angle α col

Maximum trapped volume


In the case shown in Fig. 24a the onset of the connection
At α = α ei it is determined by chamber N: corresponds to an angle α col included in the interval [ α bd -
∆α ], and therefore the vector ray relative to the beginning
V imax0 = V N ( α ei ) = V 1 ( α ei + ( N – 1 )∆α ) (49)
of delivery continues to be number 1 up to α col and
whereas in the interval 0 ≤ α < α ei and α bd ≤ α < ∆α it is not becomes vector ray N thereafter. Instead, in Fig. 24b, α col
defined, accordingly: falls in the interval [0 - α bd ] due to the fact that the preced-


ing angular pitch ended with im=1 and therefore up to α col
α dV imax α ei ≤ α < α bd
⎪V
V imax ( α ) = ⎨ imin0 ∫0 d α
+ dα the begin delivery vector ray is number 2 and then number
(50)
⎪ 0 ≤ α < α ei α bd ≤ α < ∆α 1. The equation that expresses the begin delivery vector
⎩ 0
ray (eq. (34)) must be substituted with eq. (52), while those
pertinent to delivery volume (eq. (42)) become eq. (53), eq.
Some refinements (54), eq. (55) and eq. (56). In these equations Vcm repre-
sents chamber 1 volume when it becomes aggregated to
In this second approach, the aggregation and separation of delivery (im shifts from 1 to 0 and α = α col ).
chambers are assumed instantaneous and coincident with
the opening or closing of either the discharge or intake ψ 1 > ψ bd
⎧ ρ N ( α ) = ρ 1 ( α – ∆α ) im = 0
ports. This implies that aggregation to delivery of the maxi- ⎪
ρ bd ( α ) = ⎨ ρ 2 ( α ) = ρ 1 ( α + ∆α ) ψ 1 < ψ bd im = 1 (52)
mum trapped volume originates a sudden levelling of pres-

sures; on the contrary, when a port begins to open (variable ⎩ ρ1 ( α ) otherwise
restrictor), a flow exists that progressively makes pressures
even. It is only at equilibrium pressure conditions that the α dV m α < α bd
originally separated volumes can be considered an aggre- V m ( α ) = V cm + V m0 – V 1 ( 0 ) + ∫ dα (53)
0 dα im = 1
gated entity. However, this difference can be neglected at
aggregation of minimum trapped volume to inlet. Since ded α dV m α > α bd
and dbi are equal (see Fig. 16) the minimum trapped vol- V m ( α ) = V cm + V m0 + V 1 ( α col ) + ∫ dα (54)
0 dα im = 0
ume pressure will be equal to what it was at separation
998
α dV m bd ed
V m ( α ) = V cm + V m0 – V m02 + ∫ dα α > α ed (55) 2400 -1000
0 dα
Vm02

9ROXPH9DULDWLRQ>PP UDG@
Vm01
α dV m


V m ( α ) = V cm + V m0 + ∫ dα otherwise (56)

9ROXPH>PP @

0 dα
1400 -1200
The maximum trapped volume, see eq. (50), must also be
rewritten:

α dV imax
V imax ( α ) = V 1 ( 0 ) + ∫ dα α < α ei im = 1 (57)
0 dα
400 -1400
0 32.5 65
α dV imax
V imax ( α ) = V imax0 + ∫ dα se α > α ei (58) α >GHJ@
α fa d α
Figure 26: Delivery volume and volume variation
V imax ( α ) = 0 otherwise (59)
bi ei
2900 1500
With regard to flow rates it should be clear that Qr differs

9ROXPH9DULDWLRQ>PP UDG@
from zero only when the chamber that was trapped is being


connected to delivery but the two pressures still differ one Va02

9ROXPH>PP @

another; this implies that the switch im=1. Va01

1900 1350
AN APPLICATION CASE-STUDY

Building on the second modelling approach, results are


presented relative to a specific commercial pump with non-
900 1200
symmetrical porting and gerotor gearing, possessing the
0 32.5 65
following characteristics:
α >GHJ@
• Number of chambers: N = 11
• Eccentricity e = 2.2 mm Figure 27: Inlet volume and volume variation
• Radius of circular arcs: S = 7.5 mm
eq. (41) and eq. (42) for delivery, and in eq. (45) and eq.
• External diameter of the internal gear DeRi = 58.9 mm
(46) for inlet) are observed when a chamber either leaves
• Radius of circle to complete external gear: G = 32 mm
or enters these volumes. Fig. 28 details plots of trapped
• Gears thickness: H = 10 mm
volumes where their cyclic yet sharply discontinuous exist-
• χbd =28˚ ence is brought to evidence. These quantitative evaluations
• =8˚ χ ei
• ded = 5 mm bi ei bd ed
37.0 780

0D[LPXQ7UDSSHG9ROXPH>PP @
0LQLPXQ7UDSSHG9ROXPH>PP @




• dbi = 5 mm 0D[LPXP
9ROXPH
Volume variations and volumes themselves are shown for
chamber 1 (Fig. 25). The same, within approximately two
36.0 770
α9PLQ α9PD[ 0LQLPXP
800 400 9ROXPH
9ROXPHYDULDWLRQ>PP UDG@

600 200

9ROXPH>PP @


35.0 760
0 32.5 65
400 0
α >GHJ@

Figure 28: Trapped volumes


200 -200

do answer the needs expressed with reference to eq. (4).


0 -400 Yet, in the foregoing two modelling approaches were pro-
0 40 80 120 160 200 240 280 320 360 posed, inspired by a different reticulation of the pump.
α >GHJ@ Some comments seem appropriate. In the first approach
Figure 25: Chamber 1 volume and volume variation
there is clear evidence that more ODEs must be solved
than in the second. However, despite the larger number,
angular pitches, is done for delivery (Fig. 26) and inlet vol- the (N+2) equations do not involve discontinuities. In the
umes (Fig. 27). Quite sharp discontinuities (anticipated in case of 4 equations only, we mentioned, beside other

999
points, the fact that trapped volumes are not always in
Vmax ei bd
existence and as the integrator negotiates these regions a 25
lot of step size reductions and Jacobian re-evaluations will
be necessary. Direct experience has shown that, at equal 24
@
conditions, the (N+2) model takes about half the time to Q
L
P
completion than the 4 model. Further drawbacks of the 4 
/ 23
>

volumes model can be identified in: (i) a difficult handling of H
W
trapped volume aggregation to delivery being this depen- D
U
LGHDO
 22
Z
dent on pressure rather than geometry (see Some refine- R
OI
ments); (ii) a difficult handling of volumes and volumes 21
derivatives equations in that more integration constants UHDO
must be evaluated; (iii) based on the reticulation it is not 20
possible to monitor pressures within each chamber along a 0 32.7 65.4 98.1
complete cycle. An advantage lies instead in the fact that
α [ GHJ]
there is no need for the evaluation of flow passages. As a
final comment both models predict an identical pressure Figure 30: Ideal and real flow rate at 3000 rpm and 35 cSt
ripple of the pump.
Based on these remarks and wishing to cope with the absence of leakage. Fig. 31 contrasts experimental and
degree of detail needed and expected in research and pro-
40
totype development, the choice has been that of adopting 5000 rpm
the (N+2) model.
The AMESim1 environment was used to implement the the-
ory detailed in this paper, test the submodels, tune the sys- 4000 rpm
@Q 30
tem to experimental results and perform predictive L
simulation analyses. Fig. 29 presents the AMESim system P
/
>H 3000 rpm
WD 20
U
Z
R
OI 2000 rpm
OL
R
10

0
0 1 2 3 4 5 6 7
GHOLYHU\SUHVVXUH>EDU@
Figure 31: Experimental vs simulated flow-pressure characteristic

simulated flow pressure characteristics of the unit, inclusive


of the pressure relief valve, for a number of angular veloci-
Figure 29: The AMESim system
ties of the internal combustion engine. Finally, Fig. 32 pre-
where the icon ‘gears’ shrink wraps to a single ‘supercom-
ponent’ the entire pump unit. At this extent an efficient 1

determination of the displacement (7.45 cm3/rev) and flow


ripple is possible; In fact, Fig. 30 shows the instantaneous 0.8
\
F
flow rate under ideal (no leakage) and real conditions. In Q
H
LF
the ideal case, when the volume of a chamber starts to LI 0.6
diminish, pressure within rises sharply since that chamber IH
F
is in the maximum trapped volume situation with no leak LU
W
H 0.4
paths and no contribution to the delivered flow. Yet, when P
X
Experimental

the connection with delivery starts (α > αbd) a high flow O


R
Y Simulated

peak is observed. Thereafter, when chamber pressure lev- 0.2


els up with that of delivery, the flow rate follows variations
proper to the delivery volume. In the real case, when cham- 0
ber volume diminishes, a flow out of the chamber is already 0 1 2 3 4 5 6 7
existent through leakage paths (α < αbd). This, contributing GHOLYHU\SUHVVXUH>EDU@
to delivered flow, cancels off the flow peak observed in
Figure 32: Volumetric efficiency @ 4000 rpm

1.a Trademark of Imagine, Roanne, France

1000
sents simulated and experimental data on pump volumetric dα infinitesimal rotation of external gear
efficiency @ 4000 rpm. ∆α angular pitch
The predictive capability of simulation is clearly evident, an β bulk modulus
exception being made at high velocities. This is consequent dγ see Fig. 10 and Fig. 11
to aeration phenomena that go beyond the scope of the
ϑ auxiliary variable eq. (6)
present paper. However, further details on experimental
conditions and phenomena that are not ‘captured’ by the θ auxiliary variable Fig. 7
simulation model can be found in [4]. ξ auxiliary variable Fig. 7
ρ(α) vector ray modulus
σ tooth parameter eq. (7)
CONCLUSIONS
τ tooth parameter eq. (7)
ϕ angle of rotation of internal gear
At a considerable detail this paper delved into the geomet-
dϕ elementary rotation of internal gear
ric and kinematic modelling of gerotor gearing pumps. Two
χ porting plate parameter
methods have been proposed to achieve a correct descrip- ψ port angle
tion of chamber volumes and their derivatives as well as of
dψ infinitesimal rotation of vector rays
flow passage areas. Quantitative informations have been
∆ψ' see Fig. 14
supplied for both methods and the superiority of the “vector
ray” approach demonstrated on grounds of accuracy and ∆ψ″ see Fig. 14
time to solution. Moreover, two approaches to modelling ψ’ see eq. (24)
have been considered based on different reticulations of ψ“ see eq. (24)
ψ(α) vector ray angle
the pump. A choice in favour of the model involving N+2
ω1 angular velocity of external gear
ODEs has been expressed and supporting arguments pro-
ω2 angular velocity of internal gear
vided. Finally, an application of modelling and simulation to
a real gerotor pump unit has been discussed and delivered
results contrasted with those stemming from experimental subscripts
studies on the same unit.
a,e inlet
bd begin delivery
NOMENCLATURE bi begin inlet
c control volume
A,B see eq. (32) ca inlet duct
Am section of load restrictor cm delivery duct
areaa inlet flow passage area d suction
aream delivery flow passage area e external
d porting plate parameter ebd external, begin delivery
DeRi external diameter of internal gear ebi external, begin inlet
e eccentricity ed end delivery
G radius of circle to complete external gear eed external, end delivery
H gears thickness eei external, end inlet
i counter ei end inlet
K distance of the centres of circular arcs from O1 i internal
m see eq. (1) ibd internal, begin delivery
N no. of circular arcs, teeth of external gear, chambers ibi internal, begin inlet
O1 centre of external gear ied internal, end delivery
O2 centre of internal gear iei internal, end inlet
p pressure imin minimum trapped volume
P1,P2 contact points imax maximum trapped volume
m,u delivery
Q volumetric flow rate
re external gear
R1 radius of pitch circle (external gear)
ri internal gear
R2 radius of pitch circle (internal gear) Vmax maximum chamber volume
S radius of circular arcs (external gear) Vmin minimum chamber volume
t auxiliary variable eq. (29)
V volume ACKNOWLEDGMENT

greek letters This research work was performed at the Politecnico of


Turin, Italy under Contract with Magneti Marelli S.p.A., Divi-
α angle of rotation of external gear sione Componenti Meccanici. Authors acknowledge per-
2N – 1
αVmax α Vmax = ---------------- π mission of publishing the present material granted by
N
Magneti Marelli S.p.A.
N–1
αVmin α Vmin = ------------- π
N
CONTACT

1001
N. Nervegna: niner@polito.it reach its true value. In particular, chamber 1 volume, in the
S. Mancò: saman@polito.it reference condition (α=0˚), approaches the limit value
Web address www.polito.it/fluidpower V1(0)true = 764.27 mm3 for dθ=0.001˚. Fig. 33 shows a plot

REFERENCES
764.5 0.17
1. Koch, F. Maassen, F., Geiger, U.: “ Development of Modern
764.2 0.13
Engine Lubrication Systems”, SAE paper 970922.

YROXPH>PP@
2. Eisenmann, S., Härle, C., Screiber, B.: “Comparison of differ-

HUURU
763.9 0.09
ent lubricating oil pump systems for combustion engines”, Lubri-
cation...
763.6 0.05
3. Kluger, M..A., Fussner, D.,R., Roethle, B.: “A Performance
Comparison of Various Automatic Transmission Pumping Sys-
763.3 0.01
tems” SAE paper 960424.
4. Mancò, S., Nervegna, N., Rundo, M. et al. : “Gerotor Lubricat-
763.0 -0.03
ing Oil Pump for IC Engines”, SAE paper 982689.
5. Teppati, G: “Comunicazioni private”, Dipartimento di Matem- 0 0.1 0.2 0.3 0.4
atica, Politecnico di Torino, 1996. Gθ>GHJ@
6. Fabiani, M.: “Studi teorici e sperimentali su pompe di lubrifi-
Figure 33: Chamber volume as a function of dθ
cazione gerotor”, Tesi di Laurea, Politecnico di Torino, 1997.
7. Bortoluzzi, E.:”Sviluppo di un modello matematico per la sim-
of V1(0) calculated with eq. (15) and the error percentage
ulazione e lo studio delle prestazioni di pompe oleodinamiche ad
ingranaggi esterni”, Tesi di Laurea, Politecnico di Torino, 1987. referred to the limit value V1(0)true expressed as functions
8. Litvin, F.,L., Feng, P.,H.: “Computerized design and genera- of the increment dθ. In detail:
tion of cycloidal gearings”, Mech. Mach. Theory, Vol. 31, No. 7, pp.
891-911, 1996. V 1 ( 0 ) – V 1 ( 0 ) true
9. Adams, G.P. - Beard, J.e.:”Comparison of helical and skewed error ( dθ ) = ----------------------------------------------
-
V 1 ( 0 ) true
axis gerotor pumps”, Mechanism and Machine Theory, Vol 32 No.
6, 1997, pp. 729-742.
10. Ansdale, R. F.:”Mathematical analysis of the NSU Wankel RC Derivative - integral method (vector rays)
Engine”, Iliffe, London, 1968.
11. Bearnd, J. E., Pennock, G.R., Stanisic, M. M.:”The effects of
0.05
the design parameters on the generated curvature and displace-
ment of epitrochoidal gerotor pumps”, SAE Transactions 1989, 0.04
No. 891831.
12. Colbourne, J. R.:”The geometry of trochoid envelopes and
HUURU

0.03
their application in rotary pumps”, Mechanism and Machine The-
ory, Vol. 9, 1974, pp. 421-459. 0.02
13. Cornell, R.W.: “Compliance and stress sensitivity of spur gear
teeth”, Journal of Mechanical Design, Transactions of the ASME, 0.01
Vol. 103, No 2, 1981, pp. 447-459.
0
14. Maiti, R.: “Contact stresses at the active contacts of epitro-
0 40 80 120 160 200 240 280 320 360
choid generated rotor-stator set in ‘ROPIMA’ type ‘LHST’ hydrau-
lic motor”, Indian Institute of Technology, Kharagpur, India. α>GHJ@
15. Maiti, R., Sinha, G. L.: “Kinematics of active contact in modi-
Figure 34: Error percentage in volumes with vector rays
fied epitrochoid generated rotary piston machines”, Mechanism
and Machine Theory, Vol. 23, 1988, pp. 39-45.
16. Robinson, F. J., Lyon, J. R.: “An analysis of epitrochoidal pro- In this case the error is bound to the integration step dα
files with constant difference modification suitable for rotary used in eq. (22). To determine the accuracy of this method
expanders and pumps”, Journal of Engineering for Industry, reference is made to the plot of V1(α)true obtained with the
Transactions of the ASME, 1976, pp. 161-165. integral-derivative method (eq. (15)) with an increment step
17. Saegusa, Y. et al.: “Development of oil-pump rotors with a tro-
dθ=0.001˚. Fig. 34 shows the relative error percentage vs
choidal tooth shape”, SAE Transactions 1984, No 840454.
gears position that is made in calculating V1(α) through eq.
APPENDIX (22) with an increment dα = 1˚:
V 1 ( α ) – V 1 ( α ) true
error ( α ) = ------------------------------------------------
-
DETERMINATION OF CHAMBER VOLUME V 1 ( α ) true
The maximum error (0.045%) occurs at minimum volume
Integral-derivative method condition; Precisely, at α = αVmin, Fig. 35 shows the error in
volume calculations as a function of the integration step dα:
In the calculation of the area subtended by gear profiles an
error exists that depends on the increment dθ used in
V 1 ( α Vmin ) – V 1 ( α Vmin ) true
assessing the integration step dψ in eq. (15). By progres- error ( dα ) = ------------------------------------------------------------------------
-
sively reducing dθ, the chamber volume will, in the limit, V 1 ( α Vmin ) true

1002
0.8 0.1

0.6 0.08
HUURU

0.06

HUURU
0.4

0.2 0.04

0 0.02
0 1 2 3 4 5
0
Gα>GHJ@
0 1 2 3 4 5
Figure 35: Error percentage in volumes vs integration step
Gα>GHJ@

DETERMINATION OF CHAMBER VOLUME VARIATION Figure 37: Error in volumes derivative vs integration step.

Derivative - integral method (vector rays) REQUIRED CPU TIME TO SOLUTION

Being the derivative of chamber volume expressed by an To obtain with the derivative - integral method a complete
analytic function (eq. (21)) there is no error dependence on plot of chamber volume and of its derivative (i.e. a complete
the integration step. revolution of the external gear) with an integration step dα =
0.5˚ (i.e. 720 points per rev) the required CPU time is 0.16 s
Integral-derivative method on a Sun Ultra 10 Workstation. To obtain the same number
of points via the integral-derivative method with a step
The error is in this case consequent to the increment dα increment dθ = 0.1˚ the required time is 101.7 s (i.e. 635
adopted in the numerical evaluation of the derivative of eq. times as much!)
(15). Taking as reference the exact volume yielded by vec-
tor rays, Fig. 36 shows, as function of gears position, the

0.1

0.08
HUURU

0.06

0.04

0.02

0
0 80 160 240 320
α>GHJ@
Figure 36: Error percentage in derivatives of volumes

relative error percentage that is made in evaluating the


derivative of chamber 1 volume V’1(α), using a step incre-
ment dα = 1˚; the reference in this case is the exact plot of
V’1(α)true calculated with vector rays (eq. (21)):

V' 1 ( α ) – V' 1 ( α ) true


error ( α ) = --------------------------------------------------
-
V' 1 ( α ) true
Given an angular position (e.g. α = 130˚) the error
increases with the angular step dα (see Fig. 37):

V' 1 ( 130° ) – V' 1 ( 130° ) true


error ( dα ) = ------------------------------------------------------------------
-
V' 1 ( 130° ) true

1003
View publication stats

Вам также может понравиться