A Brief Visit To 3D Rotations

A Brief Visit to 3D Rotations
Rate This
Bulent Alemdar Thu, Jun 11 2009 10:30 PM
Bulent N. Alemdar, Ph.D. SR Software Product Research Engineer
Three-dimensional rotations are one of the oldest topics of classical mechanics of rigid bodies. Early studies go back to 16th century at it was mostly regarded as an intellectually challenging endeavor for many people to determine positions of moving celestial objects. Even today, it plays a vital role in many disciplines, including but not limited to physics simulations, computer games, stability analysis of engineering structures, satellites antennas in space, etc...
One of the earliest theories for rigid body rotations is attributed to Euler's work (Leonhard Euler, 1007-1783), which is stated as
"the attitude of a body after having undergone any sequence of rotations is equivalent to a single rotation of that body through an angle around vector n."
In other words, if one takes a bar, having pinned at one end but free at the other, then arbitrarily move the free end, the final orientation of the bar can be uniquely determined by two pieces of information: its aggregated rotated axis (n) and the magnitude of the rotation around n.
Figure 1 portrays a different interpretation of 3D rotations graphically. The vector ro is the initial position of a vector, and let's say it is rotated around the vector axis n to its new position r. The relationship between ro and r is uniquely defined with a tensor, T:
where
is a skew-symmetric matrix and constructed from n ( or from = 1 2 3 ).
It should be noted that 1-3 are not rotations around any axis. They are just rotation components of . And also note that is not a true vector (i.e., it is a pseudo -vector) since it does not commute : final 1+ 2. A graphical representation of T is given below:
Figure 2 demonstrates non-commutative property of 3D rotations. In the figure, a cube is subjected to a series of rotations, namely rotations around fixed axes. In the top 4 figures, the cube rotated one at a time (from left to right). Note that the painted surface is still pointing to right after 3 rotations. At the bottom 4 figures, the order of rotations are changed (even though they are the same rotations around the same fixed axes), and finally found that the painted surface is pointing upward. This is, again, another manifestation of non-commutative nature of 3D rotations.
It should be noted that rotations are commutative if and only if rotation axis does not change. This is illustrated in Figure 3, where lines AB, AC and AD are on the indicated plane, and the rotation axis (n) is perpendicular to the plane. Thus, the final rotation is final= 1+ 2 . A typical example is 2D frame analysis where all frame members are on the same plane, and then one can say that rotations are commutative.
In his seminal work, Argyris (1982) recognized the complex nature of 3D rotations, and he coined the concept of "semi-tangential moments". But most of his work relied on the observations and balancing equilibrium equations of non-collinear frame assemblies. Real understanding of the 3D rotations based on a more rigorous mathematical background dates to the work of first Simo and Vu-Quoc (1986) and then later to the work of Cardon & Geradin (1988). Finally Ibrahimbegovic et. al. (1995) extensively studied the topic. In a typical nonlinear solution strategy, equilibrium equations are incrementally\iteratively solved. In other words, equilibrium equations are linearized and each linearized solution is nothing but simply a solution tangent to equilibrium curve. This is portrayed with blue lines in Fig.4. It is very important to note that correct forms of linearized equilibrium equations play a crucial role in capturing accurate nonlinear effects (like geometric nonlinearity). Otherwise solution may not converge or it converges to a wrong result (an example is miscalculating incorrect buckling or strength limit of frame members or wrong assessment of building strengths). Thus, it is an essential that equilibrium equations should be consistently and correctly linearized.
Equation 1 shows the rotation tensor T, it is well known that T belongs to a special orthogonal (Lie) group (a nonlinear manifold), which is uniquely defined with the following characteristics:
And it is found that linearized forms of 3D rotations (i.e., finite rotations) are actually located on a different plane at each iteration. This is explained with Fig. 5. The nonlinear manifold T is represented with blue curved surface and green planes are tangent to T at points I, A and B. One can visualize these points as tangent points to equilibrium curve (like blue lines in Fig. 4). During an incremental-iterative numerical solution, finite rotations are located on different planes (for instance, iteration 1: 1 is on plane so(3):2 and iteration 2: 2 is on the plane so(3):3). And this is the very same reason that one cannot easily sum up rotations to find final rotation (i.e. final 1+ 2) because 1 and 2 are on different tangent planes.
In conclusion, complex nature of 3D rotations is briefly revisited and it is demonstrated that these types of rotations are not commutative. Any finite element formulation development should handle these rotations properly. 3D rotations play a crucial role in computational mechanics, especially for 3D frames undergoing large displacements and rotations. Many building design codes emphasize need of proper assessment of nonlinear effects in analysis and design of 3D framing systems. For structural engineering, these nonlinear effects are simply geometric nonlinear effects (i.e., P-large delta). To this end, it is fair to say that accurate assessment of frame stability and strength issues depends on rigorous methods to capture 3D rotations effects.
REFERENCES: Argyris, J., (1982), "An Excursion into Large Rotations", Comput. Methods Appl. Mech. Engrg. 32 , pp. 85-155. Cardona, A., and Geradin, M. (1988), "A Beam Finite Element Non-Linear Theory with Finite Rotations", International Journal for Numerical Methods in Engineering, 26, pp. 2403-2438 Ibrahimbegovic, A., Frey, F., and Kozar, I., (1995), "Computational Aspects of VectorLike Parameterization of Three-Dimensional Finite Rotations", International Journal of Numerical Methods in Engineering, Vol. 38, pp. 3653-3673 Simo, J. , Vu-Quoc, L. (1986), "A Three-Dimensional Finite-Strain Rod Model. Part II: Computational Aspects", Computer Methods in Applied Mechanics and Engineering, 58, pp. 79-116.

A Brief Visit To 3D Rotations

Uploaded by

Copyright:

Available Formats

A Brief Visit To 3D Rotations

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Brief Visit To 3D Rotations

Uploaded by

Copyright:

Available Formats

A Brief Visit to 3D Rotations

Bulent Alemdar Thu, Jun 11 2009 10:30 PM

Bulent N. Alemdar, Ph.D. SR Software Product Research Engineer

is a skew-symmetric matrix and constructed from n ( or from = 1 2 3 ).

You might also like