Page 273 - Biomedical Engineering and Design Handbook Volume 1, Fundamentals

250 BIOMECHANICS OF THE HUMAN BODY

a few major ones (An et al., 1991). In a two-dimensional model of the wrist or even in a 3D model,
a reasonably accurate model may be produced by only adding a few ligament attachments per carpal
bone. Ligament attachment sites may also be approximated as a first step, although point attachments
such as might be made by a single spring are cause for concern.
As noted in Sec. 10.2, a 3D model of a joint will be more accurate than a 2D model. If one is
evaluating a long bone model in isolation, it must be decided whether the joint area is of importance
and whether the model is to be constructed for static or dynamic evaluation. In a static case, if one
is concerned with structures away from the articular cartilage, a simple loading or constraint system
may be used successfully to represent the joint. A fully dynamic model or one that examines the joint
itself will require a more complex 3D model with fixed constraints and loadings further from the
joint surfaces.
Soft tissue analysis requires an evaluation of insertion geometries and paths for complete evalu-
ation. As tendons travel from muscle to bone they often travel through sheaths and are partially
constrained. While these issues are generally not of importance to general modeling of bone inter-
actions and muscle forces, they may be needed if the desired result is a clinically accurate model
of a given tendon.
Blood flow requires a knowledge of both the velocity and pressure conditions and the time-
based changes in these parameters. While venous flow is basically steady state, it is readily apparent
that arterial flow is pulsatile. Even in arterial flows, a mean flow is considered reasonable as a first
modeling mechanism. Higher and lower steady flow velocities and pressures begin to produce a
fuller picture of behavior. In the more complex models, a fully pulsatile flow is desirable, although
generally at a large computational expense (hence the use of supercomputers for many of these
models). As noted in Sec. 10.3, vessel wall behavior is an important boundary condition for blood
flow.
Airways flow has a boundary condition of diffusion only at the terminal level within the alveolar
sacs. While air is moved in and out in the upper respiratory system, diffusion is the means by which
material is shifted near the lung-blood interfaces. A flow model of the lower respiratory system
should be such that air does not physically move in and out of the alveolar sacs.

10.5 CASE STUDIES

The first study is an evaluation of an ankle-foot orthosis (AFO) or brace such as is used for patients
with “drop foot” (Abu-Hasaballah et al., 1997). Although the orthosis is made of lightweight ther-
moplastic, many patients still find it heavy and uncomfortable, especially in the summer when the
orthosis may become sweat covered.
A 3D model with appropriate geometries and material properties was developed from a physical
brace (Fig. 10.2). After verifying the model by comparison to actual AFO behavior, a series of design
modifications was made, by element removal, to reduce weight while retaining structural integrity.
Figure 10.3 presents one of the final versions, with segments along the calf removed. The weight
reduction was approximately 30 percent, and the large openings along the calf would reduce sweat
buildup in this region.
It should be noted that a single AFO cost $200 or more, so FEA is an inexpensive means by which
many design changes may be evaluated before any actual brace has to be built. Computational times
were reasonably low, from a few minutes to a few hours on a PC.
As a second example of the process that may be followed when modeling the human body with
FEA, let us consider blood flow through the carotid artery. This is the main blood supply to the brain
(through the internal carotid artery) and is the site of stenoses. Plaque formation (atherosclerosis)
narrows the common carotid artery at the origin of the external carotid artery (which flows toward
the face) until blood flow is reduced to the brain.
To first evaluate this flow, many simplifications will be made to reduce the opportunities for error
in modeling. Each model result should be evaluated based on available bench and clinical data prior