Society For Industrial and Applied Mathematics
Society For Industrial and Applied Mathematics
Society For Industrial and Applied Mathematics
3 (Sep., 1992), pp. 482-491 Published by: Society for Industrial and Applied Mathematics Stable URL: http://www.jstor.org/stable/2132606 . Accessed: 18/04/2012 14:57
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Society for Industrial and Applied Mathematics is collaborating with JSTOR to digitize, preserve and extend access to SIAM Review.
http://www.jstor.org
c> 0
the of byusing method vanderPol.Theharmonic solutions those are periodic solutions with frequency theexcitation, Thisequation the of w. models forced the oscillation a of mass to single subjected anelastic force restoring andlinear damping. Notethe following cases[1]: special
* F = 0, freeoscillation, * c = 0, undampedoscillation, * of =0, linearoscillation, * of3> 0, hardspring, * of < 0, softspring.
The first invanderPol'smethod to assume step is that 3,F, andw - woare all c, small ofthesameorder, 0(6). Thatis,we areintheneighborhoodthelinear, and say of In free undamped, vibration. what follows, w = 0(1). wo, Thesecond crucial is theassume and step that (1.2)
s(t) = b1(t)sin(wt) b2(t)cos(wt), +
where b2 areslowly functions. varying is Periodicityenforced requiring b2 to by bl, bl, be constant. w Thatis to say,s(t) is essentially oscillation frequencywith an of slowly varying phaseandamplitude. express analyticallysetting We this by bi= 0(1), bi = 0(E), b6= 0(E2) Weassume possible that harmonics O(E2). are higher
(1.3) (i
=
1,2).
*Receivedbytheeditors September 1991; acceptedforpublication revised 1, (in form)January 1992. 31, tSchool of Engineering and Science, Hartford Graduate Center,275 WindsorStreet,Hartford, Connecticut 06120.
482
CLASSROOMNOTES
483
Hence, have we
(1.4) s(t) = b1sin(wt) b2cos(wt) wb1 + + cos(wt) wb2 sin(wt), = 2wb1 - 2wb2 s(t) cos(wt) sin(wt) -w2b1sin(wt)- w2b2 cos(wt) + b1sin(wt) b2cos(wt), + 2 + b2)(b, sin(wt)+ b2cos(wt)) + terms frequency s3(t) = 3(b of 3w.
than coefficients ofsin(wt) cos(wt) and results the in first-order one,andequating system
2b1 + [, oww (1.5)
+ 3b]
4 w
-2b2 + [Wo W + 4 w
b]bl-cb2 J
O,
b2 where = bl + b2. The qualitative nature thesolutions this of of varies system greatly depending upon thevaluesoftheparametersand 3. Hence,thissystem besttreated separate c is by considerationthefollowing of cases: * Case I: c = 0, of = 0, * Case II: c = 0, of > 0, *CaseIII: c= 0, < 0, * Case IV: c > 0, of = 0, * Case V:c > 0, of > 0, * Case VI: c > 0, o < 0.
In what we A to follows, define = 2(wo- w), i.e.,A is proportional thedetuning (w0- w). It thenfollows from - w) = O(E) that (w0 (1.6)
w2 _ -w2 =
(wo-w)(wo+ w) = 2w(wo-w)
0(62),
andhence(w2 - w2)/w = A to within we the first-order terms. Consequently, rewrite abovesystem as 2b,+ [A+ 4 b] b2+ cb1= F
-2b2 +
A+ 4
bi - cb2 =
0.
Sincewe areprimarily in that we interested harmonic solutions, require b1,b2be conto of stant, b1 = b2= 0. Equation reduces a system algebraic i.e., (1.7) then equations for response the the of curves. Thisrestrictionconstant b2alsoallows application to b1, the method Andronow Witt. of and That the solutions Duffing's of is, harmonic equation areidentified thesingular with differof points theabovefirst-order ofordinary system ential criteria for is determined utilizing equations (ODEs). Stabilitythen Poincare's by theclassificationsingularitiessuchsystems of of [2]. we to Below, adhere thenotation (1.8)
dn A(+Bn dq CA +DB
484
CLASSROOMNOTES
2. Case I: c = 0,of = 0. In this we the linear section, analyze caseofan undamped, oscillator. Define (2.1)
oa=
A, x=bl,
y=b2,
r2 =x2+y2,
T=t,
F1=-F
FIG. 1.
that Letting = x2 + y2,wededuce p F1 =0 (free vibration) either = 0 ora =0, X> p * F1 0 (forced ?> vibration) p = F12/o2. ,
2.2. Stability.
(2.4)
(2.5) x = xo+,
-
ax dy dx F1-coy' Y= Yo +=,
(B-C)2
(2.7)
are sincelinearization notnecessary obtain Hence,all singular was to points centers (2.6). Alloscillations stable this in are case.
CLASSROOMNOTES
485
hard 3. Case II: c = 0,of > 0. In this the we section, analyze caseofan undamped, Define = 1/ (3131). Then(1.7) canbe rewritten as spring. aO (3.1)
2b, + -2b2+
?2
aO_
b2 =F
Fw
A ?0,
bi =
where ambiguous is chosen coincide that 3.Letting the to with of sign (3.2)
o
=
aO
bi y =-,b2
r2
~2
aO
T =-
F, =
F
(wao)'
theabovesystem becomes
dx + (a?r2)y=
(3.3) d
F1,
-dy +(o?r2)x=0.
and in of Theresponse curves this for caseappear Fig.2. Theregions stability instability aretherein depicted.
CASE 11: C=O0 BETO>0
STABLE
UNSTABLE
STABLE
FIG. 2.
C~~~~~~~~~~~~~~~~~~~~~~r
P)yo
F1,
* F1 = 0 (freevibration) either = 0 or p = Ta, =X p * F1 0 (forced vibration) p = F12/(( ? p)2. =X Hence, foro > 0 we have (3.5) p(a + p 2 = F2
small p Owehave F/oa2. ForFsmallandp and ForF12 p wehavep, the on we depending whether areaboveorbelow linep = -o-.
486
CLASSROOMNOTES
p) + p[2(T + p) (d+)]
= 0,
do,
dp dy dx
(a Fi-(a?
i
1
3
r2)x
r2)y'
2
y=yO +7 rr
p+2xof +2yor= X
d< (B
The region abovethelinep = -o/3, butbelow linep = -a is seentobe comthe and a posedofsaddles ishence region instability. of the However, remaining where area, (B - C)2 + 4AD < 0,posesa bitofa problem. Thisis duetothefact B that + C = 0 there, hence and breaks down. Thatis,wecannot Poincare's that theory conclude a centerofthelinearized to of system We corresponds a center thenonlinear system. are, that however, guaranteed itcorresponds either center a spiral. to a or Tosettle issue, must this we onceagain of singular study inthe (3.9) neighborhoodthe but time must point this we retain terms. yields This quadratic (3.13)
dr
(p_ 2yo_q -
-[uq ? (3pr+
+
yo(2+
+ 43 +
n2)
+ q3)] 42 + 2,
3yoq2 + q(2
1 ((d 1 (d
dr7) )
dt
dt
dr _1(yf 3+ dr 1 +Y3
dO 1 dO
2(Tp)
,_
? 4 + (+4 + +423(?3yo) 4)
2q
The following observations should made. The case that are trying study, be we to i.e.,a + p andaT+ 3pofthesamesign, seentobe a center. is Thisfollows from directly
CLASSROOMNOTES
487
thefact (3.13)is oddin( andhence that cannot havea spiral an isolated as singularity. Moreover, previous our that conclusion theregion bounded p = -o andp = -of/3 by wascomposed saddles againconfirmed. this In of is dr/dt oddin is case,as ingeneral,
(, whileforsmallr we have (3.15)
df
p) + n (o ?
3p)
dO/dt Hence, changes sign, of suggestive a saddle. In summary, * above p = -o =X centers(stable), * belowp = -oj13 =X centers(stable), * betweenp = -oj3 and p = -o =X saddles (unstable). Notethatp = -a, corresponding thefree to vibration, unstable is beingon the of and It that locusofverboundary a stable an unstable region. should noted be the ticaltangencies, = -o/3, represents transition stability instabilityyo p the from to as in increases magnitude through negative values. 4. Case III: c = 0,of< 0. In this section, analyze caseofan undamped, we the soft A of spring. quick perusal Case II will convince that us a of Case III ismerely reflection Case II aboutthep-axis. Thissituation instark is contrast thetraditional to treatment oftheDuffing and a equation representsconsiderable over advantage suchtreatments. Theresponse for caseappear Fig.3. Theregions stability instability in curves this and of aretherein depicted.
CASEIII:C=0. BETA(B
STABLE
UNSTABLE
STABLE
FIG.3.
oscillator. Letting A
(5.1)
or=
=bl,
y=b2,
~~~~2 r =x 2 +y,2
T=-
ct 2
F1=-,
WC'
488
CLASSROOMNOTES
CASE IV, C>G, BETA=e
FIG. 4.
Yo =
0,
* F1 = 0 (freevibration) p = 0, =X
5.2. Stability.
(5.4)
dy
ax - y
dx
X=Xo+(,
d7
F1-ay-x
Y=YO+?7?*, (af) + (- 1)
(5.5)
(5.6)
(B-C)2
(5.7)
to Hence, all theoscillations correspond stablespirals. we 6. Case V: c > 0, ,3 > 0. In thissection, analyzethe case of a damped,hard spring. Letting F A ct r2 =2 2 2 c =WC aO aO equation(1.7) becomes
dx+
(fa r2)y+ X = F1,
(6.2)
dy
-dy +(fa
and of for The responsecurves thiscase appearin Fig. 5. The regions stability instability are therein depicted.
dT
2)x
y=.
CLASSROOMNOTES
CASE V CM0. BETh>0
489
\ %s
UNSTABLE
STABLE
FIG. 5.
point(xo,yo), 6.1. Response. At a singular (6.3) ()( ((a?+p)yo +xo = F1, + p)xO-yo = o,
= 0=,
(6.6)
[a+P)2
+ 1] + p[2(
? p)
(dT 1
?,
d = ?X
u/3,
=? da {P[(aJp) ?1]}
(6.10) (6.11) dPj[(rp)2 +1] + p [2(J?p)(1 p either = 0,
=,
dU)] or p = a, = 0,
dP = 0o
490
CLASSROOMNOTES
dx
dy
_ -
y=yo+71=r2
*df
(a + (2x2 + p))( + (-1I+ 2xoyo)7 (-1 T 2Xoyo)( + (-af F (2y2 + p))r7 + 4AD = -4(a + 3p)(oa p), + AD-BC = _(a2 + 4ap + 3p2 + 1), B + C = -2.
(6.15)
(B-C)2
the Hence, following situation prevails: * above p = -a =X stablespirals, * below p = -af/3=X stablespirals, * betweenp = -af/3and p = -a but"outside"hyperbola stablenodes, =X * betweenp = -af/3and p = -a but"inside"hyperbola saddles (unstable). =X Notethat, onceagain, locusofvertical the tangencies corresponds transition to from to stability instability. 7. Case VI: c > 0, Q < 0. In this we the soft section, analyze case ofa damped, As this to of spring. before, caseis seenuponinspection be thereflection Case V about thep-axis. The response curves this for case appearinFig.6. The regions stability of andinstability therein are depicted.
CASE VI: C>X, BETA(0
ISTABLE
FIG. 6.
8. Conclusion. hasbeenshown Duffing's It that can anaequation be successfully in detail themethod vanderPol andthemethod Andronow of of lyzed great by and Witt. wererequired assume We to that certain in parameters thedifferential equation weresmall that detuning also small.Theserequirements infact and the was are more than restrictive thoseforalternative methods analysis. of However, within stated the this has attractive The features. first simplicity, of is assumptions, method two very this a course of taste. second thesymmetric played The is roles being largelymatter personal reduces overall the burden. by,3> 0 and,3< 0, which significantly analytical
CLASSROOMNOTES
491
Acknowledgments. author The wouldliketo thank Professor J. Stoker the J. of Courant Institute Mathematical of in Sciences particular suggesting topicfor for this study ingeneral hisinspirational and for teaching thesubject nonlinear of of vibrations.
REFERENCES [1] J.J. STOKER, Nonlinear Vibrations inMechanical Electrical and Systems, Interscience, York, New 1950. [2] W. HuREwicz, Ordinary Differential Equations,BrownUniversity, Providence, 1943. RI,