The 13th Scandinavian International Conference on Fluid Power, SICFP2013, June 3-5, 2013, Linköping, Sweden

Modeling and verification of an excavator system – Axial Piston Pump, Kinematics and Load Sensing Flow Sharing Valve Block

Professor Dr.-Ing. Paolo Casoli, Dr.-Ing. Luca Riccò, Dr.-Ing. Dolcin Cesare*

Industrial Engineering Department, University of Parma (Italy) *Walvoil S.p.A., Reggio Emilia (Italy)

E-mail: paolo.casoli@unipr.it

Abstract

This paper presents the results of a study focused on mathematical modeling of an excavator hydraulic system, composed by: the pump gray box model, the kinematics model and the valve white box model. The kinematics model has been realized using the planar mechanical library of AMESim® and is composed by the front excavation tool: the arm, the boom and the bucket. The valve section white box model has been validate with the comparison between the numerical and experimental result obtained during the laboratory tests. The excavator is equipped by a full flow sharing valve, that is very useful in this kind of machinery when during a digging cycle all the valve sections are used at the same time. In this paper the excavator mathematical model system will be composed by the pump, the kinematics and two valve sections. The new system will be validated with the comparison between the experimental results, obtained with two sections working at the same time, and the numerical results provided by the simulation. The experimental results will be obtained in two different working conditions: standard operation condition and flow saturation condition. This will show the mathematical model capability in the study of the interaction between all the system components and could be useful in the study of alternative control strategies towards energy efficient systems and new control system designs.

Keywords: Hydraulic Excavator, Load Sensing, Mathematical Model, Energy Saving.

1 Introduction

In the last years, the simulation of the kinematics and dynamics of multi-body systems is a topic of increasing importance in many industrial branches, due to its potential for providing insight into inherent effects governing the systems’ behavior and for reducing manufacturing costs. Furthermore with the continuous petrol price rising the higher interest in energy saving is rising as well. Lots of researchers have been studying for years these topics. Focusing on mobile machineries, one of the primary needs is that of consume the less energy possible to complete a working cycle, like a standard digging cycle[1]. In other words one of the principal target being that of achieving more efficient machinery. Due to this, one of the first goals for machinery producers

is the energy saving toward the optimization of their products, the system components interaction, and also with the introduction of energy recovery systems. Study such a complex system (like an excavator) require time, know-how and lot of money. Thus mathematical models are always more used. Typically, mobile machineries hydraulic system are composed by a variable displacement axial piston pump [2] and proportional directional valves [3]. The hydraulic system controls the kinematics movements, under user needs. An excavator system is a very complex nonlinear plant, and for this reason a suitable mathematical model is needed as well as a good simulation tool. Many qualities are required from mathematical models, including accuracy, predictability for the simulated

53

system behavlow computaThe researchmathematicais under denvironmentdeveloped tothe original eThe followinpump, the floand validatemodels.

2 Physica

As already important qmodel of cpredictable olast but not leCertainly in study the firrole, but in tinteraction bpump, the computationaTaking this under develotime while behaviors preThe mathemgraybox modwhite box mo

2.1 Pump M

The pump, mvariable dispby Casappa®

Its mathemastate conditexperimental[4]. The valiis in progressin fig.1, hascomposed byits regulatorsand the flowmodel of the

2.2 Kinema

The kinematithe front exc

viors (stationation time. h reported i

al model, for developing u. The math

o be as simplelements functng sections pow sharing vaed results of

l Modeling

described in qualities requcomplex sysof transient aneast short calc

the individurst two characthe complete between all th

valve and al time becom

into accounopment, has thmaintaining edictability.

matical model del, the kinemodel.

Model

mounted on thplacement axia® S.p.A. and beatical model htions, with l and numericidation of the s. The pump ms been definey the combinas, such as thew compensatopump’s flow

atics Model

ics model, shoavation tool o

nary and trans

in this papera middle sizeusing AMEShematical mole as possibletions. present the malve, the excavthe pump m

g

the introducuired from atems are: r

nd stationary bculation time. ual componencteristics playsystem study,he componen

the kinemmes very impont, the mathehe skill of a sa satisfactory

is composedmatics model

he excavator, ial piston pumelongs to the Mhas been valid

the compacal results inmodel in tran

mathematical ed gray box ation of whitee pressure coor (FC), andcharacteristic

own in fig.2, only: arm, boo

sient) as well

r describes te excavator, thSim® modeliodel has be while ensuri

modeling of tvator kinemat

model and val

ction, the moa mathematicesult accuracbehaviors and

nts optimizatiy a predomina, focused on tnts, such as tmatics a sh

rtant. ematical modshort calculatiy accuracy a

d by: the puml and the val

is a load sensimp [2] develop

MVP series. dated, at steaarison betwen previous wonsient conditimodel, depictbecause it

e box model fmpensator (P the black b

c.

is composed om and bucket

as

the hat ing een ing

the tics lve

ost cal cy,

d at

ion ant the the

hort

del ion and

mp lve

ing ped

ady een ork ion ted

is for

PC) box

by t.

Thon

Cinpathfahybothcaeqremst[6ac[7

2.

ThWItscocoreInthduthoc

he kinematicsn the hydrauli

Figure 1: A

onsidering thentegral with arameters werhe Planar Mecacilitates the uydraulic cylinoom, arm andhe hydraulic ran be derivequations to evolute pairs

multipliers andabilization me

6]. The modecting on the bu7].

.3 Valve Sec

he valve secWalvoil® S.p.a.

s mathematiconsidering onomparison betesults obtainedn a mobile mahe use of thisuring a standahe valve sectioccur, the pu

s model is usec actuators.

MESim® Pum

e benefits of hthe hydrau

re coupled to chanics librarunderstandingnder. The drid bucket are gram actuators.ed by applya Lagrangia

have beend are calculaethod applied l is not yet ucket during t

tion Model

ction is a ful. and belongs cal model hanly one sectween the sim

d during laboraachinery, suchs kind of valvard digging cyons at the samump could w

ed to create re

mp Mathemati

having the kinulic model,

the hydraulicry of AMESig of dynamic iving joint togenerated by . The equatioying the Euan energy fun modeled ated from thto the constraset to simulathe interaction

ull flow sharito the DPX se

as been validction workinmulation and ratory tests. h an excavatorlve is very usycle [1] are al

me time. Whenwork in flo

ealistic forces

cal Model

nematic modelthe linkage

c model usingim® [5]. Thisloads on the

orques of thethe forces of

ons of motionuler–Lagrangeunction. Theas Lagrange

he Baumgarteaint equationsate the forcesn with the soil

ing valve byeries. dated in [8],

ng, with theexperimental

r under study,seful becauselmost used alln this situationow saturation

s

l e g s e e f n e e e e s s l

y

, e l

, e l n n

54

condition, i.required by t

Figure 2: AM

If this situaticontrol on dangerous, bmaintains ththe control onThe principakeeping in composed bythat of defineactuator, mathe main spoand provide needed. It’smathematicarecreate all oFor this reasodeveloped ncomplex valfig.3, able toreal valve. Tcompare by retain all thevalve sectiomathematicaall the valve

2.4 Govern

The governinteraction ba mechanicacalculates thflow rate bet

e. the pump the valve bloc

MESim® Kine

ion occur, thethe moving

but with this khe control on n their movemal working cmind that thy lots of sectie the outlet flintain the preool metering the flow shars easy und

al model for of this functionon the valve m

not like a simlve, the AMEo recreate theThis has incran ideal valv

e original woon. This is al model, able sections.

ning Equation

ning equationetween a fluid

al-geometricalhe pressures itween adjacen

flow is lessk.

ematics Mathe

e operator canparts, that

kind of valve tthe parts mo

ments velocityondition of a

he complete ions (defined low through tessure drop coarea, define ting operation erstandable the valve canning conditionmathematical mple valve, bESim® sketche working funeased the simve model but orking conditi

the plus ofto recreate th

ns

ns are descd-dynamic mol model (MGinside the chnt chambers, w

s than the flo

ematical Mode

n lose the direcould be ve

the operator sovements losiy only. a valve sectiovalve could valve block),he section to onstant betwethe LS pressucondition whthat a simpnnot be able ns. model has be

but like a deh is reportednctioning of tmulation time

has allowed ons, of the ref this kind he interaction

cribed by todel (FDM) a

GM). The FDambers and twhile the MG

ow

del

ect ery still ing

on, be

, is its

een ure hen ple to

een emi

in the

e if to

eal of of

the and DM the

GM

caits

ThThunpr

Thtefusuencowocgeco

Thcobalitcooforcasp

alculates the fos dynamics an

Figure 3: AM

he FDM is bahe pressure iniform and timressure-rise ra

he model amperature. T

unction of pummation terntering or leaonsidering the

with the consccurring throueneralized Beonditions, Eq.

he user sets oefficient of asis of experimterature, suchoefficient of df Reynolds nur fully turbulalculates the inpool using New

orces acting ond the flow are

MESim® ValvMod

ased on a lumnside each come dependentate equation:

1assumes a cThe fluid depressure as m represents

aving the vole contributionsidered volumugh the orificernoulli’s equ(2):

an appropriatdischarge of mental data o

h as [10]; thdischarge valueumber, to accoent conditionnstantaneous pwton’s second

on the spool aneas.

ve Section Matdel

mped parameteontrol volumt, and is deter

constant valensity is eva

described is the net malume. This isn of all orificme. The maces is calculauation under

2|∆ |ate saturated vf each connecor using valuehereafter the ue is evaluatedount for partians. The mechposition and vd law.

nd determines

thematical

er framework.me is assumed

rmined by the

(1)

lue of fluidaluated as ain [9]. Theass flow rates obtained byces connectedass exchangeated using the

quasi-steady

(2)

value for thection, on thees reported ininstantaneous

d as a functionally developedhanical modelvelocity of the

s

. d e

d a e e y d e e y

e e n s n d l e

55

The forces aspring forceStatic and dyof the KarnStribeck effeare assumed model the hy

3 Experim

Figure 4 reprsetup that hfunctioning athe features o

Sensor

M

P

P1 – P8

Q1 – Q2

T Tor

Θ Incr

LVDT L

diffe

Table 1: Feaappa

acting on the e, friction foynamic frictionopp friction ect; static and

constant; in ydrodynamic f

mental Setu

resents the ISOhas been usedas a double seof the used sen

Type

Prime mover

Pump

Strain gage

Flow meter

rque/speed meter

remental encoder

Linear variable erential transducer

atures of sensaratus used in

spool are: hyorces, hydrodon forces are e

model and dynamic frictthis simplifiedforces are neg

up

O schematic od to verify t

ections. In tablnsors.

MainABB®, 4-quad

7

CASAPPA®

Trafag®, Scaleac

VSE® VS1, S0.3% measur

HBM® T, ScaAccu

HEIDENHAIimp./r, 4000 r/

1/20 pe

r Magnet Sc

sors and main n the present r

Figure 4: Te

drostatic forcdynamic forcevaluated by uconsidering ttion coefficiend version of tlected.

of the test benthe valve blole 1 are report

n features drant electric mot75 kW

® MVP60, 84 cm3

0..400 bar, 0.3%ccuracy

Scale 0.05..80 l/mred value accuracale: 0..500 Nm, 0.uracy Class IN® ERN120, 36/min Limit velociriod accuracy

chuz AVAX 015

elements of thresearch

est setup for L

ces, ces. use the nts the

nch ock ted

tor,

3/r

% FS

min, cy .05

00 ity,

he

ThcovaThdiFiwpospFiex

Load Sensing F

he tests weromposed by ariable displache pump has irectly connecigure 5 illustra

with the pressuositions of thpool and the pigure 6 show txperimental te

Figure 5: Va

Flow Sharing

e carried outwo valve s

cement pump.been instrumeted with the sates the cross ure mapping

he two LVDTressure compethe test bench

est.

alve with LVDMapping

Valve Block

ut using the sections coup. ented by an a

swash plate. s section of th

points and tT’s connectedensator. h picture befor

DT mounted ang Points

valve block,pled with the

angular sensor

he valve alongthe mounting

d to the main

re starting the

nd Pressure

, e

r

g g n

e

56

3.1 Experim

As mentionecarried out:

- Dosa

- Dosa

The first opstandard conas double prhas a constaload. The twoThe pump spspools openinpump margitemperature a

Fig

The second oflow saturatcircuit is higsections loadhas been set for the two pump margitemperature m

Figure 6: Te

mental Settin

ed before, two

ouble pressureaturation; ouble pressur

aturation. perating condindition, i.e. wiressure compeant load and o loads pressu

peed has been ng position hain has been at 50 °C.

gure 7: Sectio

operation condtion, i.e. the gher than the pd are represenat 1000 rpm amain spools

in has been maintained co

est Bench Layo

ng condition

o different typ

e compensatio

re compensat

ition has beethout flow satensation, whethe section 2

ure are displayset at 1500 rp

as been reportset at 20 b

ons Imposed L

dition has beerequired flow

pump’s outletnted in fig.9, tand the openinare depicted set at 15 ba

onstant at 50 °

out

pe of tests we

on, without flo

tion, with flo

en performed turation, definre the section

2 has a variaby in fig.7. pm and the mated in fig.8. T

bar and the

Loads

en performed iw rate by tht flow. The twhe pump speeng position lain fig.10. Th

ar and the o°C.

ere

ow

ow

in ned n 1 ble

ain The oil

in he

wo ed aw he oil

Thtofeinabto

Figure

he transient po the test beneedback controntroduce problble recreate tho prove the mo

Figu

Figur

e 8: Main Spoo

pressure peaksnch pressure ol system settlem, rather if his actual conodel accuracy.

ure 9: Sections

re 10: Main Sp

ools Displacem

s, visible in fproportional

ting. The transthe mathemat

ndition, will b.

ns Imposed Lo

pools Displac

ment

fig. 9, are duel relief valvesient does nottical model is

be very useful

ad

cement

e e t s l

57

4 Comparsimulati

Figure 11 decomprising olayout mount

4.1 Double saturati

In this paraexperimentaltest.

rison betwion result

epicts the AMof the pump ated on the test

pressure ion

agraph are col result obtain

Figure 12

ween exper

MEsim® sketcand valve usedt bench, as sho

compensati

ompared the ned during th

2: Inlet Pressu

Figure 11: T

imental an

h of the systed to recreate town in fig.4.

on, no flo

numerical ahe first type

ure

Test setup for L

nd

em the

ow

and of

Load Sensing

Fig

Figure 1

Flow Sharing

gure 13: Load

14: Intermedia

g Valve Block

Sensing Pres

ate Pressure S

k

sure

Section 1

58

Figur

Figure 1

Figure 1

re 15: Interme

16: Compensa

17: Compensa

ediate Pressur

ator Displacem

ator Displacem

re Section 2

ment Section 1

ment Section 2

1

2

Figur

Figur

Figur

re 18: Outlet F

re 19: Outlet F

re 20: Pump Sw

Flow Rate Sec

Flow Rate Sec

Swash Plate Po

ction 1

ction 2

osition

59

Figures 12, 1load sensing for all the vaFigure 14, 1pressures (se1 and 2 pshowed in figFigures 18, experimentalsection 1 andAs can be seall the pressu2% if compamathematicarates with a ma simplified mIt’s possible capability tposition. Figure 20 repthe simulatioaccurately.

4.2 Double saturati

In this paraexperimentaltest. Figureschambers in common forpressures. Firates compabetween 30s 1 decreases open.

Figure 25 dunfortunatelysection 1 prLVDT absenalready seenwithout flowvalve section

13 show the copressure. Thi

alve sections. 15 refer to see fig. 5). Thepressure comgures 16 and1

19 depict thl and simulad the section 2een the model ures inside theare with the eal model is abmaximum erromodel. to observe th

to calculate

ports the pumpon data matc

pressure ion

agraph are col result obtains 21 – 24 shthe valve blor the valve igures 26, 27 arison. When

and 70s, the until the sec

Figure 21

epicts the disy is not posressure compnce when this n in the doublw saturation cn has the capa

omparison of is two pressur

ection 1 ande comparison

mpensators di7. he compariso

ation outlet f2.

has the capabe chambers wexperimental dble to recreateor among 10%

hat the model the actual

mp swash platech the experi

compensatio

ompared the ned during thehow the presck: the inlet ablock and thshow the sect

the flow saoutlet flow ra

ction 2 main

1: Inlet Pressu

splacement ofssible compa

pensator posittest has been

le pressure cocondition, theability to perf

the inlet and tres are comm

2 intermediabetween sectisplacement a

on between tflow rates fro

bility to recreawith errors unddata; further te the outlet flo%, acceptable f

has also a gocompensato

position, whemental positi

on, with flo

numerical ae second type ssure inside tand LS pressuhe intermediations outlet floaturation occate from secti

n spool remai

ure

f the section are it with ttion due to tn carried out. Aompensation te model of tform good res

the mon

ate ion are

the om

ate der the ow for

ood ors

ere ion

ow

and of

the ure, ate ow

cur, ion ins

2, the the As

test the sult

foananm

or the two wond the pressurn acceptable

match the exper

Fig

Figure 2

Figure 2

orking conditire compensatoway by the rimental one s

gure 22: Load

23: Intermedia

24: Intermedia

ions. The outor position ar

model and satisfactorily.

Sensing Pres

ate Pressure S

ate Pressure S

let flow ratese recreated inthe pressures

sure

Section 1

Section 2

s n s

60

Figure 2

Fig

Fig

5 Excavat

The excavatodepicted in fthe engine po

25: Compensa

gure 26: Outle

gure 27: Outle

tor Model a

or under studyfig 28. Its opeower is 46.5 k

ator Displacem

et Flow rate S

et Flow rate S

and cycle s

y is a middleerating weightkW at 2200 rpm

ment Section 2

Section 1

Section 2

imulation

size one andt is 9500 kg am.

2

d is and

BprmThhaThde15LSinpomFifigdethfoim

Figure

ased on the revious sectio

models are caphus to widenas been includhe complete sescribed in fi500 rpm. FiguS pressures. T

n fig 31. Figuosition during

model capabiliigure 33 depigures in thevelopers an he pump and thor studying mprovements i

Figure

e 28: Photogra

analysis of n, it is evidenpable of repron the model cded as illustratsystem was sug 29. The puure 30 shows The sections inure 32 reportsg the cycle anty to recreatects the pump is section pinsight into the valve. Thenew control in the design o

e 29: Valve se

raph of the exc

results presnt that the pumoducing actuacapabilities thted in fig 34. ubjected to a ump speed has the simulatenlet flow ratess the section nd it’s possible the compen

swash plate provides to the internal fu

ese informatiostrategies o

of these comp

ection spools p

cavator

sented in themp and valveal conditions.he kinematics

duty cycle asas been set atd system ands are depictedcompensatorse observe the

nsating mode.position. The

the systemfunctioning ofon are needfulor proposing

ponents.

position

e e . s

s t d d s e . e

m f l g

61

Figurre 30: System

Figure 31: Se

Figur

and Load Sen

ections inlet fl

re 34: AMESi

nsing pressure

flow rate

m® sketch of t

e

the front exca

Figure 3

Figu

vator tool and

32: Sections co

ure 33: Pump

d hydraulic cir

ompensator d

swash plate p

rcuit.

displacement

position

62

6 Conclusion

This paper has presented the analysis of an excavator system. A nonlinear mathematical model of an excavator has been developed using the AMESim® modeling environment to replicate real operating conditions. The model is described by hydraulic model comprising of a load sensing pump, a flow sharing valve model and a 2D kinematic model to simulate the excavator front excavation tool. This approach has enabled the study of the dynamic behavior and interaction of the pump and valve with a completely developed kinematic model of the excavator. The detailed hydraulic model described regard a load sensing flow sharing valve block, composed by two valve sections model working at the same time. The mathematical model of the valve has been validated against experimental results in different conditions: standard operating without flow saturation and flow sharing condition, i.e. with flow saturation. The pump and valve models and their verification, as represented in this paper, offer a great deal of confidence in the individual models capabilities of recreating the functioning of an excavator. The last section of the paper provides an overall view of the hydraulic model capability coupled with the kinematics of the system, with the simulation of an arm-bucket movement. The results presented show the paramount of advantages that this model possess in aiding a pump and valve designers in analyzing/assessing the behavior of the components as a part of a complex system. The mathematical model of the excavator could be useful in future study of both new control system and new energy saving solutions.

Acknowledgments

The authors would like to acknowledge the active support of this research by Casappa S.p.A., and Walvoil S.p.A (ITALY).

Nomenclature

A flow area [m2] Cd discharge coefficient -

i volume index -

mass flow rate [kg/s ]

p pressure [bar]

V volume [m3]

t time [s]

Greek Letters

β bulk modulus [Pa]

ρ density [kg/m3]

Acronyms

FC Flow compensator PC Pressure compensator

PS System Pressure

PLS Load Sensed Pressure

References

[1] JCMAS H 020 – Earth Moving machinery – Test methods for energy consumption – Hydraulic excavators, 2010.

[2] J Ivantysyn , M Ivantysynova. Hydrostatic Pumps and Motors, Tech Books International, New Delhi,2003, India. ISBN 81-88305-08-

[3] Blakburn J. F., Reethof G. & Shearer J. L. (1966) “Fluid Power Control”. USA: MIT Press, 1966. ISBN 0262520044

[4] P. Casoli, A. Anthony (2013) “Gray box modeling of an excavator’s variable displacement hydraulic pump for fast simulation of excavation cycles” Control Engineering Practice 21 (2013) pp.483-494. Elsevier Ltd. http:/dx.doi.org/10.1016/j.conengprac.2012.11.011

[5] LMS Imagine – AMESim® Reference manual 2011

[6] Shih-Tin Lin, Jiann-Nan Huang. (2002) .Stabilization of Baumgarte’s Method Using the Runge-Kutta Approach. In ASME Journal of Mechanical Design, vol. 124, n.4, pp. 633-641. December 2002.

[7] Grzesikiewicz, W. (1998). Dynamics of a ground digging tool. XI scientific Conference Problems of Work Machines, konferencja Naukova Problem Maszyn Roboczych) Polska: Zakopane, 1998.

[8] P. Casoli, A. Anthony, L. Riccò (2012) “Modeling of an Excavator System – Load sensing flow sharing valve model” SAE 2012 Commercial Vehicle Engineering Congress, Rosemont, Illinois, USA, 13-14 September 2012. doi:10.4271/2012-01-2042H

[9] E Merritt. Hydraulic Control Systems. John Wileys and Sons, Cincinnati, Ohio, 1967. ISBN 0-471-59617-5.

[10] McCloy, D., Martin, H. M. (1973). “The control of Fluid Power”, Longman, London, 1973.

63