Full-Load Testing of Large Gearboxes Using Closed-Loop Power Circulation

R.IB. Frost University of IN'ew S'outh Walles .Kensington, AustraHa

'lR. C'ross
J'ohn We'lsb Ply. Ltd. M'elbourne .Austr,a'iia
Summary: Thi . method of testing large geareleoiently mas ive, unless further supplementary gearing is provided ..One way of doing this, of
C~lU"

boxes or, indeed. any power transmission

ment. has numerous advantages, and offers 'the possibility of large avings in time. energy, and plant. if the overall. situation is conducive to its use. This, u ually requires that several such units need to be tested. and that they can be conveniently connected toeach other in such a way as to form a clo ed-loop drive train ..No power ink

e, is to test two gearboxes simuttaneously

in series, as shown in Fig. 1.0.. This, however. does nothingto avoid the necessity fora large source and sink.

(a)

IPowerll1
Source
III

is required. and the drive input system has only to make up power losses. The level of circulating power i controlled by the torque, which i the drive (b) applied statjcaUy during rotation,and

Gear Box Under

Test
I I I

,Power Sink

I
Sink

peed. Principles. advantage, and limitations are described. together witb. recent experiences the only known large-scale nique in Australia, in usage of this tech-

lP,owerl

Ie

,I FIrst Gear Box Under

Test

Source

IS~:~d,
IBox Under
I

I Po;-'t'sr

II
I

Test
I

Introduction
Full-load testing of large gearboxes and other power tran mission modules or components has traditionally been carried out. with 'Ille tesung linear
OT

Fig. ~ - The traditional, schemal:icaJI.y open arrangement of gearbox test rigs: (0) the . implesl arrangement; (b) a tandem anangememt which tests two gearboxe simllllnneolisly and which avoids 8.massive torque at a low speed al the power sink. The above situation can be markedly im.-

drive train arranged in .3. schematically open manner, a

hown in fig. lao This necessi.3.

tates the use of a power source and a power sink. or brake, each with that ofthe required capacity at lea t equal 10 test further. all energy test, ink. units

proved by the use of a closed-loop arrangement of the test drive train. Such an arrangement allows the return to the test system ofthe output
power from

IR.B. Fras1
is Senior Lec/urer. School 01Mechanical &. Industrial EngineerinB. University olNew Soulh Wales,

each gearbox under test, so that.

ource only has to provide make-

which passes through the gearboxunder which can. be considerable. This 13 t factosiscornpounded need to be tested, Iflhe is wasted inlhe unumerous

no power sink is required; .' the power

lip power equal to the sum of system losses:

.' several gearboxes are 'tested simultaneously. Obviously. large savings can accrue from the avoidance of the necessity for
iii

rated power of the

T.R. Cross
is MalUIging Director. John Wt'/~h ,Pty. LId. in fIilelbouf"1!l!. Australia.
18111

gearbox under test is large.and the output speed is low, the torque at the sink can be inconve-

power sink and!

SEPTE,IIBERIQCTQ!lER

35

nom thelarge reduction

capacit.y of the power source. If numerous units need to be

tested, large savings in time and energy are

m 'the required

input stages, chen four can be coupled as shown in Fig, 3. For generality, and because of recent experience with such a ' erup, the remainder of this article is written as relating to a setup
lIS

possible. Disadvantages or Iirmtancas on the use of such a test method are that

shown in Fig. 3. The three additional. components above can be incorporated referred to

OOI1-backdriving gearboxes. suchas high-sarin

worm drives, cannot. be tested in this manner; a somewhat larger floor area is required for the test. setup;

in any convenient manner or location in the loop, The drive input

can be incorporated ina coupling shaft as a 'belt or chain drive or as a separate gearbox with a double-ended output shaft, or it can be attached directly to a shan extension on one oflhe units under test, If one tis available. Alternatively.

the power level in each gearbox. in a particular setup' win be slightly different due to losses
around die drive train loop; . ' the method is best suited to situations where at least as many gearboxe need to be tested as are required to make up the drive 'train loop;

if

the motor speed corresponds to the speed 01 the high- peed shafts, and the motor has a doubleended shaft, it can be incorporated one of the high-speed directly in a shaft lines, The torque

if the number 'of units to. be tested is not. an

integral multiple of die ollmber required to make

application coupling is a device which join

up the loop, some will need to undergo two tests

to allow fonnation ofthe final loop at the end of the series of tests. While such multiple simultaneous testing is not uncommon in Europe. it has not been used to any great extent in AllJstralia. The Drive Traio. Loop This consists of several of the limits to be tested, coupling shafts, and three other compo

pair of adjacent. shan ends and which applie

equal and opposite torques to them whilethey are borh rotating. and in the presence of some degree uf relative rctatian between them, It may be thought of in practice as a hydraulic motor, supplied and drained by two lines via a rotatable hydrauli.c coupling, with :its shaft connected to one of the two shaft ends referred to above. and its body connected to the other.

nents: a motion input dnve,a torque application

,coupling, and a torque transducer. Such a loop can be achieved in several convenient ways ..Ifthe gearboxes are parallel-shafted, two can be coupled as shown in Fig. 2; if they involve shafts at right angles, e.g., as wwth bevel

The Action .of the Test Rig

'This is best described by considering tbe rig at rest while the torque application coupling is loaded by the application of hydraulic pressure. The whole drive train loop is thereby loaded torsionally so thateach tors.ionaI equilibrium, section of it. being in has equal and opposite

tor1lues at its two ends: i.e., an actioa and a reaction ..If the motion input :isnow started. the whole drive train loop w:iUrotate in the direction

in which it is driven. Fora particular loop 'ocGEA.R

IBOX 11 tion, the applied torque and the rotation will be in. the same direction at one end, where power wi II be entering thadoop sec tion, and ill oppos ite directions atthe other end, where power will be II ~eavinglh3!t loop section. This applies to each loop section, so that power eirculaees continuously around the drive train loop in a dieecticn
_

~,i~

10....-

DRIVE IN'PUT

which is determined by the combinanon application,

of the

directioas of the drive rotation and the torque The level of circulating power is, detennined by the combination of the speed of the drive rotation and the level of the torque application.

Fig. 2 - A elo .ed-loop arrangement rorr testing two parallel-shafted gearboxes simultaneously and IJtiJizingcirculating power, thereby avoiding thenecessity for a 'power sink.

,361 GEAR

TECHNOLOGY

With the exception of the Iosse due 110 the inefficiencies lin the gearboxe power does not leave the drive train loop; therefore, it doe '11101 ne d to be provided by the drive input, except tor the losses, regardless of the level ofthe circularing power. Power flows and 110 ses are shown in Fig ..3. If P I is the input power to the gearbox IOn the "downstream" sideof the drive input (in the power-flow en e) and P:sis the power delivered to the drive input by theadjacent gearbox on the "up ITeam" ide, then

P'

lOSS2

+ (1-'1) P'

Pi-

Loss!

(1-11) P

where Tl is the efficiency of each individual gearbox. 0 that the total power [0 is given by

The input power fraction, which is the ratio of input power required when 'l.Isinga clo ed-loop rig to thaI. required when using a rig of the type shown in Fig .. I, where input power must equal test power, is given by f
p

P. .
mpul

P
4

I -11 =1'\3

(3)

where P , which is the lowest input power level 4 [0 any gearbox in the [001'. is taken robe the agreed power level for 'the test. This agreement could be made differently. of course. in which case, f will be defined differently. but this will p make little difference tD fp . The input energy fraction :for testing a series of gear boxe in this way, rather than a in Fig, 1, is given by f 1 ~ 114 fE ==....:::: -3 N Nil
c -

(4)

where N is the number of gearbox units tested simaltoneously ..In the situation considered here.
N=4.

Similarly. the overload factor on Gearbox I, in order to achieve P4 into Gearbox 4, is given by
(5)

Fig.], - Similar to Fig.2. but to suit four right-angle-drivegearboxc are hewn, boxes, Clearly. large avings ioinstalled capacity. time and energy can be realized, provided the overload OR some units can be tolerated for the duration of the test. This overload i , of course. significantly reduced if PI' P2' orP 3 can be agreed upon as the power level for the test, rather than P4' Consideration needs to be given to the "handing" of gearboxe and the resulting compatibilily of rotationaround the drive train loop. Regardle of this. however, it i nece arily true that one diagonally oppo ire pa'ir of boxes will be operating in the chematicaUy forward (powerflow) sense as ' peed reducers, bearing on the forward flanks of the teeth, while the other diagonally opposite pair are operating in the schematically backward sense as speed increaser , bearing on the rear flIanks. PDr spiral bevel gearing the difference in geometry and force levels between thesetwo ca e are lgmficant, Whether this is seen as a problem. a in the necessity to prove every box in every mode, or as all,opportunity whereby each of these two. ca es can be demon trated simultaneously, depends IOnthe way in wnich test requirements are written. At most, however, this problem would only involve a reversal ofrotation and/or torque in the , rune setup once testing is under way, Recent Experience With Closed-Loop Testing John Welsh Ply. Ltd. of Melbourne recently upplied 2~ gearboxes, each rated at 610 kW,
and losses

. Power flows

Fig,4 hows \la\l.1esof fpi fE' and Fo fm a range of T\ which is realistic t:or a large range of gear-

\485 RPM input speed .. I7.76: \ redaction ratio

with a spiral bevel input stage and a helical

SE PT E!.IBE RIO

cr e B E R

1991

37

1..5, ,..---""""T----,~I----,.---.------,I ,05

1

appro imarely anticipated,

31.5" for angular

assembly last in the flanges on

of the flanges of this coupling,

which. it was

would be assembled

drive train loop with its mating however,

the adjacent coupl ing shaft ends. In tile event,

1.3 I--~-.,...j-~'r-I~--+--""""""---il
I

0.3
~~~

it was found that the torque transholes in its flanges to final closure application and conneccoupling ini-

ducer had sufficient allow convenient with the torque tiaUy set of testing.
.0

1.2 I----II---~---"'rl"'-'---t__--t- 02

tion 'of the drive train

loop at its flanges. 7SOtl of

that it had approximately

its stroke available The Lebow was located

ill 'Ihe reqaired direction

L-~L-~L-~L-~~~o
091.
1)
F~g.4 - Variation of energy fraction fE' power fraction fp' and overload factor Fo far a range of efficiency .:1fIl. for the setup shown in Fig. 3. output stage, for coal conveyor drives. These were te ted four-at-a-time in an arrangement similar to that shown in Fig. 3, at 50% of full load and then at HXJ% of full load, in each case until temperature and thereafter to be taken. equilibrium was reached to allow noise measurements

1641- lOOK torque transducer

ill the other high-speed

of testing, was

shaft.

096

098

10

In eachinstance driving the rig was coupling application

the motor

tarted while the torque ubject to a small to in

required minimum pressure of approximately 5 bar. This gave rise to the necessity overcome the lightly was nota some degree of initial stiction torsionally real problem, pre-loaded however,

rig. Thi and once pressure to

the rig broke the stiction and started to move, it ran very freely. The.hydraulic the torque progressively application increa coupling was then

Drive input. was via a multiple V -belt drive

of the appropriate of the high-speed ratio from an induction shafts connecting which the gearmotor rated at 220 kWal 960. RPM. onto one

ed, thereby increasing reading readspecifi-

the circulating
the motor was achieved ing of course, cally

power Ievel, the losses. lind en the torque meter. Thi had to be calculated

load, until the required

boxes ..This
application etary item,

haft also contained the torque coupling, model was a proprifrom

to

in view of the pcsition

1.11 Ihe drive for the

VM2-4000-100 It provided in each direction and thereby

train loop occupied testing haved procedure. in a stable

by the torque transducer. In all cases manner the rig be-

Glyco in Germany. hydraulic a compact total relative tic torsional calculated the torque chambers. design, cilitate tile production

two active

The rig was then run as required

faa

of a large torque from provided

and the testing

proceeded

without problems. ConcilUlsi.on res ting of mulltiple gearboxes rig arramgementis a very testing method are satisfied. On

rotation of on I.)!'WO. The elasdeflection around the drive !rain torque, 9

E ,

Slmaltaneous convenient

loop at 1.15% of futl-Iead to be approximately application tive rotation at the adjacent located and tile backlash was estimated

was

using a elosed-looptest

31.so of relawas to be 5 ..5" at this this

and co t-effective

shaft ends where

where the conditions favoring such a method. as outlined in the introduction.

eration of this method.

coupling

in the gearing, ,(JSL'

such occasions. (he authors recommend con sid-

to be approximately

I.
0'1Engineers
0/1

loeation. For loading in both direction,

application coupling of

required a total relative rotation of the torque

Acknowledgemen't: lia, Filth Intemattona!

Th.is aniele was originally ..AUSlrQ1990.

with Conference

presented at The }n.r/jlmiO'n (6)

turing Engineering, NSW,

Manufac-

July
Australia.

1113,
Reprinted

W,ollongong,

which was within

its range and which left

permission.

,38,

GEAR

TECHNOLQGV