lecture 1210

Upload: yalew-tamene

Post on 05-Apr-2018

233 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 Lecture 1210

    1/20

    ME 563

    Mechanical VibrationsLecture #12

    Multiple Degree of Freedom

    Free Response + MATLAB

  • 8/2/2019 Lecture 1210

    2/20

    Free Response1

    We can solve for the homogeneous solution to a coupled set of

    equations in a multiple degree of freedom linear system by:

    -Identifying the initial conditions on all the states-Assuming a solution of the form {x(t)}={A}estWhat does this last assumption imply about the response?

  • 8/2/2019 Lecture 1210

    3/20

    Two DOF System2

    Consider the two degree of freedom system of equations:

    If we make a solution of the form, , as we did forthe single DOF case, we obtain:

    Non-trivial solutions satisfy:

    Ms2+ 2Cs+ 2K( ) Ms2 +Cs+K( ) Cs+K( )

    2

    = 0

  • 8/2/2019 Lecture 1210

    4/20

    Characteristic Equation3

    There are four solutions that satisfy the characteristic

    equation and these solutions are expressed as follows when the

    modal frequencies are underdamped:

    The solution to the homogeneous equation is then written asfollows (for the first two complex conjugate roots):

    s1=1+j1, s2=1-j1, s3=2+j2, and s4=2-j2

    Real part: decay rate Imaginary part: oscillation frequency

  • 8/2/2019 Lecture 1210

    5/20

    Free Response Form4

    The free response is usually written in the following form for

    a multiple degree of freedom system:

    Four constants Four initial conditions are required.

    Modal vector

    (can be scaled)

    Decaying co-sinusoid

    (common to both degrees of freedom)

  • 8/2/2019 Lecture 1210

    6/20

    Example Solution5

    For K=1 N/m, C=0.1 Ns/m, and M=1 kg, the solution is given

    by:

  • 8/2/2019 Lecture 1210

    7/20

    MATLAB Solution6

    To obtain solutions for the free response in MATLAB, the

    procedure explained before for a single DOF system is used:

    Or

  • 8/2/2019 Lecture 1210

    8/20

    MATLAB Solution7

    We use the following formulation to define the outputs, {y}, if

    we are only interested in the displacement variables:

    Or:

    y{ }21

    = C[ ]24

    q{ }41

    + D[ ]22

    f(t){ }21

  • 8/2/2019 Lecture 1210

    9/20

    Free Response (Eigen-Analysis)8

    We can also solve the homogeneous equations of motion by:

    -Identifying the initial conditions on all the states-Identifying the modal frequencies, s, and vectors, {X}, usingeigen-analysis.

    Do you remember what an eigenvalue problem looks like?

  • 8/2/2019 Lecture 1210

    10/20

    Eigenvalue Problem9

    Consider the case when we have no damping:

    We can write this set of equations of motion as follows:

    Now assume a solution of the form {x(t)}={A}est

  • 8/2/2019 Lecture 1210

    11/20

    Eigenvalue Problem10

    Consider the case when we have no damping:

    This last equation is the standard eigenvalue problem.Matlab solves this equation with the command eig(A):

    [v,d]= eig(A);

  • 8/2/2019 Lecture 1210

    12/20

    Matlab eig11

    How do we interpret what Matlab gives us with this

    command?

    is a diagonal matrix of -s1,22

    is a matrix of modal vectors

    How do we find the modal frequencies and vectors?

    d

    v

    d =

    2.6180 00 0.3820

    v =

    -0.8507 -0.52570.5257 -0.8507

  • 8/2/2019 Lecture 1210

    13/20

    Matlab eig12

    First, we take the negative square root of the diagonal entries

    of the dmatrix providing the modal frequencies in rad/s:

    Second, we can scale the columns of v to produce easy to

    interpret modal vectors:

    -d =

    0+1.6180i 00 0+0.6181i

    v =

    1.0000 0.6180-0.6180 1.0000

  • 8/2/2019 Lecture 1210

    14/20

    Matlab Form of Response13

    The form of the free response can then be written in the form:

    In summary, the eigenvalue problem for an undamped system

    is given by:

    Where does the damping fit?

  • 8/2/2019 Lecture 1210

    15/20

    Damped Eigenvalue Problem14

    To obtain solutions for the free response in a damped system,

    the state variable form of the equations of motion are used:

    and then the eigenvalues and eigenvectors of the state matrix

    are calculated using eig. This approach works because theassumed solution {q}estis also used for the 1st order system:

    [v,d]= eig(A);

  • 8/2/2019 Lecture 1210

    16/20

    Matlab eig15

    How do we interpret what Matlab gives us with this

    command in the first order form of the state variable model?

    is a diagonal matrix of modal frequencies s1,2and s3,4

    (1:2,1:4) is a matrix of modal vectors {X}1 and {X*}1

    d

    v

    diag(d) =

    -0.1309 + 1.6127i-0.1309 - 1.6127i

    -0.0191 + 0.6177i

    -0.0191 - 0.6177i

    s3,4

    s1,2

    {X}2 and {X*}2

  • 8/2/2019 Lecture 1210

    17/20

    Matlab eig15

    Eigenvectors are not the modal vectors of the 2nd order system

    eigenvectors involve both position and velocity states. We

    look to the position states to obtain the scaled modal vectors.

    v = [

    -0.0362 - 0.4457i -0.0362 + 0.4457i

    0.0224 + 0.2755i 0.0224 - 0.2755i

    0.7236 0.7236

    -0.4472 + 0.0000i -0.4472 - 0.0000i

    -0.4472 + 0.0000i -0.4472 - 0.0000i

    -0.7236 -0.7236

    0.0085 - 0.2763i 0.0085 + 0.2763i

    0.0138 - 0.4470i 0.0138 + 0.4470i

  • 8/2/2019 Lecture 1210

    18/20

    Damping in MDOF Systems16

    Note that for the previous example, the damping matrix[C]

    corresponded to a proportionally viscously damped system:

    For this type of damping, note that the modal vectors were

    entirely real, i.e., the phase between DOFs was either 0 or 180o.

  • 8/2/2019 Lecture 1210

    19/20

    Damping in MDOF Systems16

    What if the system is not proportionally damped? The modal

    vectors in this case are complex but the response is not:

    For this type of damping, note that the modal vectors are

    complex, so the phase between the DOFs is not 0 or 180o.

    What does this mean?

  • 8/2/2019 Lecture 1210

    20/20

    Damping in MDOF Systems17

    What if the system is not proportionally damped? The modal

    vectors in this case are complex but the response is not:

    Real

    Imag

    Mode #2

    Mode #1

    2nd order EOMS

    cannot be uncoupled

    using real modal

    vectors!