Académique Documents
Professionnel Documents
Culture Documents
INTRODUCTION In cases where the line of action of the load is offset from
BOLTED CONNECTIONS of various types are widely used in the bolt axis, additional bending effects will occur in the
mechanical engineering. The type here considered is that in bolt and these will influence the choice of pretension.
which the load is applied to the connection along an axis The use of a torque spanner is by far the most common
parallel to the axis of the bolts. The behaviour of such a method for obtaining a specified bolt pretension, although it
joint has been previously studied (I)t (2) and where the suffers from the disadvantage that its accuracy depends
axis of the load coincides with the centre line of the bolt upon the coefficient of friction between the sliding surfaces
and no bending action is present, the general behaviour is remaining constant and equal to the assumed value.
shown in Fig. 1. Provided that the applied load is not
sufficient to cause separation between the two halves of the
joint, the relation between bolt load A and applied load W Notation
is given by A Load in the bolt.
A=A,+kW . .
. * (1) A0 Pretension load in the bolt.
where A , is the initial pretension in the bolt and k is a d Mean diameter of screw thread.
constant dependent upon the relative stiffness of the bolt r Mean radius of screw thread.
and the abutments. rm Mean radius of nut seating.
The two quantities which will be of principal interest to T Total torque applied to the nut.
the designer of such a joint will be the clamping force on t Number of threads per unit length.
the joint under the action of applied load, which is given by W Applied load.
A-W=A,-(l-k)W . . .
(2) a: Semi-angle of thread form.
and the ability of the bolt to sustain the maximum load A
B Helix angle of thread.
P Coefficient of friction between screw threads.
under either static or repeated load conditions. Coefficient of friction between nut and washer.
P
Subject to the overriding condition that A/W is greater
than 1, both the clamping force and the maximum bolt load
will increase linearly with A,, the value of the pretension, RELATION BETWEEN APPLIED TORQUE
and thus the two requirements for the joint are to some ANDBOLTLOAD
extent conflicting. Normally, however, a suitable value of By considering the nut as a block on an inclined rough
A , can be chosen which will satisfy both conditions. plane, the relation between the applied torque and bolt
The M S . of this article was first received a t the Institution on 29th load can be expressed as
November 1961 and in its revised form, as accepted by the Council
for publication, on 13th September 1962. T = Ao(hr+prm) . . - (3)
* Department of Mechanical Engineering, University College,London. tan p+p sec CI
Associate Member of the Institution. where A=
t A numerical list of references is giwen in the Appendix. 1-p sec a tanp
JOURNAL MECHANICAL E N G I N E E R I N G SCIENCE Vo14 No 4 1962
402 T. H. LAMBERT
COMMUNICATIONS
The following Communication relates to the paper Matrix Again for the benefit of those whose prime interest in
techniques for determination of fundamental mode shape and torsional oscillation was practical, a note should be made
frequency parameter of torsional oscillations in engines, of the degree of arithmetical precision that was necessary.
by K. N . GUPTA, which was published in 1962, vol. 4, For frequencies, three significant figures were ample; for
No. 2, p . 156. relative amplitudes of oscillation, two were more than ample.
For an engine with n cylinders (n having any value from
Professor W.A. Tuplin (Member) wrote that this paper 4 to 10 inclusive) the s u m of the relative amplitudes at the
was of great interest in demonstrating a possible application cylinder-lines for the one-node vibration form was given
of matrix techniques but, for the benefit of newcomers to with adequate accuracy by
the study of torsional oscillation it should be made clear 2 (I/Jc/I/Jl) = 0-59n+0.47
that practical necessities in calculation were covered by The individual relative amplitudes might be derived
simple routine in arithmetic. from a normal elastic curve whose end-points were first
Determination of the frequency equation of any straight- fixed by means of the relation
line system was straightforward and quick as had been
#F/I/JI = -(0*59n+0*47)a
demonstratedin Enginem-ng of 18th May 1934. An improve-
ment on this technique was that given by Crossley and The curve itself might be drawn freehand as a circular
Germen in the Journal of Applied Mechanics of 27th June arc of height
1960, No. 2350. Ten-inch slide-rule working was more than (0-27- 0.6/n)9hl
adequate in determining the frequency equation and in but modified near cylinder No. 1 so as to be parallel to the
solving it. axis of the crankshaft at cylinder No. 1.
The simple case considered by Mr Gupta was important
because of its frequent occurrence, and it had been covered AUTHORS R EP LY
in Engineering of 19th February 1937 in an article that gave Mr K. N. Gupta wrote that he was grateful to Professor
a formula for the lowest natural frequency for any system Tuplin for his suggestions, but he wished to make it clear
with four to ten cylinders, values of a from 0 to 0.04 and that though the determination of frequency equation for a
values of from 0.8 to 1.2. straight torsional system was straightforward and quick, yet
the solution of a polynomial equation was time-consuming. be the point of contact of the (badly named) pitch circles
Further, it did not yield directly the relative amplitudes of and so could not move backwards and forwards along the
different masses. Both frequency and mode shape were of common tangent to the pitch circles or radially. T h e
great interest to those engaged in the study of torsional natural way to define the transmission error of a pair of
oscillation. Because of that fact tabular methods had come gears was to speci@ the departures frpm equality of their
to much prominence. Matrix techniques not only fulfilled circumferential displacements at the co-axial contacting
the above requirements but also displayed a distinct circles that divided the line of centres in the ratio of the
advantage in that they reduced repeated approximations numbers of teeth. (These departures could be estimated
involved in the tabular methods to straightforward from pitch errors and profile errors measured in ordinary
computation. ways.)
Further, the author wished to add a word to the degree If t h i s failure had in fact been due to resonant rotational
of precision for frequency and relative amplitudes laid vibration, it justified critical investigation, as a resisting
down in the communication. In s m a ll engines where the torque that varied with speed so markedly as did that of a
ratio Kcyl/lcylwas generally quite high, at least four signi- rotary compressor was usually regarded as a strong vibra-
ficant figures were required for frequency parameter (6) tion-damping influence in addition to the unidentified ones
to have better estimation of frequency. For relative ampli- that sufficed to hold magnification factors down to finite
tudes two significant figures were no doubt sufficient for values.
estimating the amplitude of oscillation at mass No. 1, but
for calculating the vibratory stress at nodal point or at any
section of the shaft a still higher degree of accuracy was
needed.
Finally, the empiricalformulae set out inthe communica-
tion for the estimation of the individual relative amplitudes
were of interest to newcomers to the study of torsional
oscillation, but they involved the laborious process of
drawing the elastic m e . Some results for extreme cases
were compared with the correct values in Table C1 to
give an idea of the dgree of accuracy involved in them. The
author was of the opinion that individual relative amplitudes
could easily and with greater accuracy be obtained from the
graph shown in Fig. 9 of the paper under discussion.
Resonance was a recognized scapegoat for strange mech- the centre of the latter were moved from its actual position to
anical failures, but in this case it still remained to consider the position which it would have were it perfectly mounted;
why the shaft should fail while the gears did not; in any during this process it would suffer a slight rotation if
ordinary geared transmission, overload showed itself first contact were maintained at the teeth.
at the gear teeth. Whether or not failure occurred at a gear depended upon
the amplitude of the fluctuating torque at that gear. This in
turn depended, not only on the violence of the oscillation
AUTHORS REPLY
(as Professor Tuplin evidently believed), but also upon the
Professor R. E. D. Bishop wrote in reply that he was position of the gear in the resonant mode. If it was placed
particularly interested in Professor Tuplins communica- near an anti-node it might support only a small oscillatory
tion as he was aware that there is a serious gap between the torque (in fact the actual gear teeth did not entirely escape
work of vibration analysts and those who specialize in damage in the system concerned). It would be noted that
gears. Each type of engineer had his own technical language. the difference between the torques sustained by the shaft
The problem in question was easier to discuss in the and pinion was equal to the inertia torque of the pinion at
language of the former, though Professor Tuplin wanted the frequency in question. That differencehad an amplitude
to use that of the gearing specialist. of about 220 1bf.ft.
Suppose a perfect gear was mounted eccentrically. The In introducing the question of back-lash, Professor
gear was then no longer perfect but had a sinusoidally Tuplin might have forgotten that there was a rigid body
distributed spacing error of the teeth. If it mated with a mode which, in this case, accounted for a power transfer
second gear which was both perfect and perfectly mounted of about 2500 hp. In any event, detuning by back-lash was
then there would be fluctuating velocity ratio. When he had by no means a well-documented phenomenon.
used the words pitch point in the paper, he had done so There seemed to be no difficulty in demonstrating that
loosely. To be more precise, he should have referred to the violent torsional vibration could be set up in the manner
point of intersection of the common normal to the mating explained. Fig. C1 showed a torsion pendulum A whose
teeth with the line of centres. It was this which wandered upper end was driven by a gear B which meshed with a
backwards and forwards along the line of centres, thereby pinion C driven by a variable speed motor. B and C were
causing variation in the velocity ratio. He assumed that class A gears and C was mounted on an adjustable eccen-
Professor Tuplin would not deny that a fluctuating velocity tric, the amount of the eccentricity being shown by adial
ratio could cause a torsional vibration. gauge. An annular container of oil D could be offered up to
The argument could be put another way. Suppose the the skirt of A so as to increase the damping of the system.
perfectly-mounted gear was held fixed while that with When the speed of C equalled the first non-zero frequency
the eccentric mounting was held in mesh with it. Suppose of the rotating system, the oscillation of A became violent.