5.

G EOMETRIC M ODELING
Types of Curves and Their Mathematical Representation
Types of Surfaces and Their Mathematical
Representation
Types of Solids and Their Mathematical Representation
CAD/CAM Data Exchange
 T YPES
C URVES AND T HEIR
OF
M ATHEMATICAL R EPRESENTATIONS
Wireframe Model ( 2D in 1960s for drafting, 3D in 1970s)
Wireframe Entities
analytic entities (points, lines, arcs and circles, fillets and chamfers, and
conics
synthetic entities (splines and Bezier curves)
methods of defining points: P(x,y,z), P(r,,z), P(x+x,y+y,z+z), , end
points of existing entity, center point, intersection of two entities.
methods of defining lines: between two points, parallel to axis, parallel or
perpendicular to a line, tangent to entity
methods of defining arcs and circles: center and radius, three points, center
and a point, a radius and tangent to a line passing through a point.
methods of defining ellipses and parabolas: ellipse (center and axes lengths,
four points, two conjugate diameters), parabola (vertex and focus, three
points).
methods of defining synthetic curves: cubic spline (a set of data points and
end slopes), Bezier curves (a set of data points), B-spline curves (interpolate
a set of data points with local control possible).
 Curve Representation: Two types of representation are parametric and
non-parametric representation. In parametric representation all variables
(i.e., coordinates) are expressed in terms of common parameters. For
example, a point can be expressed with respect to a parameter as

P(u) [ x(u), y(u), z(u)], umin u umax

Non-parametric representation is the conventional representation as

P [ x, y( x), z( x)]

Ex. Non-parametric form of a circle: x^2+y^2=r^2, parametric form: P(u)=[r

cos 2u, r sin 2u], 0u1. This form can be used to find slopes at a
certain angle for example.
 PARAMETRIC R EPRESENTATION OF
A NALYTIC C URVES
The following list shows most of the analytic curve that are used in
CAD/CAM system for part design and modeling.

Lines
Circles
Ellipses
Parabolas
Hyperbolas
Conic Curves
 L INE AND C IRCLE
A line between two points P1 and P2 can be expressed with respect to a
parameter.

P P1 u(P2 P1 )

A circle for a center and the radius can be written as

x x c R cos u
y y c R sin u, 0 u 2
z zc
 E LLIPSE

An ellipse with a center and major and minor axes of 2A and 2B can be
expressed as.

x x c A cos u
y y c B sin u, 0 u 2
z zc
PARABOLA
E XAMPLE 1
E XAMPLE 2
H YPERBOLA
E XAMPLE 1
 C ONICS
The most general form of planar quadratic curves is conic curves or conic
sections that include the previously covered curves; lines, circles,
ellipses, parabolas, and hyperbolas. The general implicit nonparametric
quadratic equation that describes the planar conic curve has five
coefficients and naturally needs five conditions to complete it.

The conic parametric equation can be described if five conditions are

specified appropriately. One case is specifying five points on the curve.

L1 0, L2 0, L3 0, L4 0

L1 L2 0, L3 L4 0

L1 L2 aL3 L4 0
E XAMPLE 1
 PARAMETRIC R EPRESENTATION OF
S YNTHETIC C URVES
Hermite Cubic Splines
Bezier Curves: cubic curve with four control points.
B-Splines: general case of Beziers curve (non-uniform)
Rational Curves (algebraic ratio of two polynomials)
NURB (non-uniform rational B-spline) curve combines all features of
previous curves
H ERMITE C UBIC S PLINES
E XAMPLE
E XAMPLE
 B EZIER C URVES
Another alternative to create curves is to use approximation techniques
which produce curves that do not pass through the given data points that
are rather used to control the shape of the curves. Approximation
techniques arc more often preferred over interpolation techniques in
curve design due to the added flexibility and the additional intuitive feel.

Bezier curves and surfaces are credited to P. Bezier of the French car firm
Regie Renault who developed (about 1962) and used them in his
software system called UNISURF which has been used by designers to
define the outer panels of several Renault cars.
E XAMPLE
E XAMPLE
 B-S PLINES
In contrast to Bezier curves, the theory of B-spline curves separates the
degree of the resulting curve from the number of the given control points.

The B-spline curve P(u) for the degree k defined by n + 1 control points
n
P(u ) B kj (u )P j , 0 u u max
j 0

partition of unity
n
B kj (u ) 1
j 0
the recursive property
u u j u j k 1 u
B j k (u ) B j k 1 (u ) B j 1k 1 (u ), k 1
u jk u j u j k 1 u j 1

1, u j u u j 1
, for j n
0, otherwise
B j 0 (u )
1, u j u u j 1 , otherwise
0, otherwise

open parametric knots

0, j k 1

u j n k 1, jn for 0 j n + k + 1
j k, otherwise

Knot Values

number of knots = number of points + degree + 1 (nk = n+1+k+1

must be satisfied)
The individual knot values are not meaningful by themselves; only the
ratios of the difference between the knot values matter.
To be able to use the parametric knots produced by the coordinates,
the values must be in increasing order. Thus, if the curve intersects
itself, this cannot be used.
The number of duplicate values is limited to no more than the degree.
Duplicate knot values make the b-spline curve less smooth. At the
extreme, a full multiplicity knot in the middle of the knot list means
there is a place on the b-spline curve that can be bent into a sharp
kink.
Summary of B-spline Basis Function Bjk
a polynomial function of degree k
nonnegative for all j and k (nonnegativity)
non-zero only on [uj,uj+k+1] (local support)
at most k+1 basis functions of degree k are non-zero on knot span
[uj,uj+1], namely, Bj-kk, Bj-k+1k,, Bjk
sum of all degree k basis functions on knot span [uj,uj+1] is one
(partition of unity)
at a knot of multiplicity m, basis function Bjk is Ck-m continuous.
a composite curve of degree k polynomials with joining knots in
[uj,uj+k+1]
n+1 control points, P0,P1, ,Pn
n-2 basis functions (i.e., cubic polynomial curve segments,
Q3,Q4,,Qn)
n-1 knot points, t3,t4,,tn+1 (any additional ones needed for
computation are set to same values as the nearest)
Bjk is
- defined over a knot interval [uj,uj+1]
- defined by four of the control points, Pj-3, ,Pj
Notable benefits of b-spline include:

Independent degree: the degree is set by user and it has nothing to

do with number of data points
a single piecewise curve of a particular degree: there is no need to
stitch together separate curves as in interpolation splines such as
natural cubic spline
E XAMPLE
MCAD P ROGRAMMING
U NIFORM Q UADRATIC B-S PLINES
U NIFORM C UBIC B-S PLINES
R ATIONAL C URVES
E XAMPLE
R ATIONAL F ORMS OF C OMMON
C URVES
E XAMPLE
 NURBS
Remember that the b-spline curve is weighted sum of its control points.
n
P(u ) P j B kj (u ), 0 u umax
j 0

Since the weights depend on the knot vector only, it is useful to add another
weight wj to every control point that can be set independently. The weights
can be added as

n w j B kj (u )
P(u ) N kj (u )P j , N kj (u ) n
, 0 u u max
j 0
w j B kj (u )
j 0
Here, Njk is NURBS basis functions. Thus, the NURBS curve of degree k is
defined as weighted sum of control points with NURBS basis functions.
Some of Valuable Properties of NURBS Curves and
Surfaces

invariant under affine as well as perspective transformations (affine

invariance and projective invariance).
single mathematical form for both analytical and free-form shapes.
flexibility to design variety of shapes.
less memory is used when storing shapes in comparison to other
methods.
numerical algorithms can evaluate them easily and quickly.
 E XAMPLE

Solution. For three control points n = 2 and for the second degree k=2. The uniform
knots can be computed by the program in the B-spline example. The following shows
result in MathCad with modified program for Bjk. It now uses the knot vector as an
input.
Shown below are various curves that vary with different weights. Note that
the last shows violation of convex hull property with negative weight for mid-
point.
 C URVE M ANIPULATIONS
The cases that require manipulation of curves include:

Displaying
Evaluating Points on Curves
Blending
Segmentation
Trimming
Intersection
Transformation
A PPLICATIONS
 D ESIGN A PPLICATIONS
Example 1. For the given state of stresses (a)
determine the principal stresses and (b) state of
stresses on a plane a-a.

Example 2. For a plane motion of a bar sliding down

A
the step, find the locus of the point C that is
located at one-third length from B.
C

B
 T YPES OF S URFACES AND T HEIR
M ATHEMATICAL R EPRESENTATIONS
Surface model is an extension of wireframe but has advantages: less
ambiguous, provide realism for display with hidden lines, mesh, and
shading.

Surface Entities
Plane surface
Ruled (lofted) surface (surface created by two curves being blended)
Surface of Revolution
 Tabulated Cylinder (surface created by a curve and a vector)
Bezier surface: only approximates the given data points permitting only
global control
B-spline: surface that can approximate and interpolate permitting local
control
Coons Patch: used to create a surface using curves that form closed
boundaries in contrast to the above surfaces that use open boundaries
or set of points.
Fillet surface: B-spline surface that blends two surfaces together
Free-form surface: formed by free-form curves that are extensions of
Bezier, B-spline, and NURB curves.
S URFACE R EPRESENTATION :
E XAMPLE
 A NALYTIC S URFACES
Plane Surface: P(u,v) = P0 + u(P1-P0) + v(P2-P0)
Ruled Surface
Surface of Revolution
Tabulated Cylinder
P LANE S URFACE
 B ILINEAR S URFACE

Bilinear surface is a linear interpolation of four corners in two different

directions (u, v).

Pu ,0 (u ) (1 u )P0,0 uP1,0
Pu ,1 (u ) (1 u )P0,1 uP1,1
P(u, v) (1 v)Pu ,0 vPu ,1
R ULED S URFACE
R EVOLVED S URFACE
TABULATED C YLINDER
PARAMETRIC R EPRESENTATION OF
S YNTHETIC S URFACES
H ERMITE B I C UBIC S URFACES
E XAMPLES
 B ILINEAR S URFACE

Bilinear surface is a linear interpolation of four corners in two different

directions (u, v).

Pu ,0 (u ) (1 u )P0,0 uP1,0
Pu ,1 (u ) (1 u )P0,1 uP1,1
P(u, v) (1 v)Pu ,0 vPu ,1
 C OON S PATCH
Bilinear surface generated by four corners has straight sides and produce
quite flat surfaces. In contrast, Coons surface uses four side curves.

PLR (u, v) (1 u )PL (v) uPR (v)

PBT (u, v) (1 v)PB (u ) vPT (u )
P(u, v) PLR (u, v) PBT (u, v)

The surface obtained as above does not produce the end curves. Thus,
evaluating the surface on boundaries and forcing it to confirm to the
boundary curves yields extra-terms that must be subtracted.
Evaluating the surface along edges one finds extra terms that must be
subtraceted.

PLR (u, v) (1 u )PL (v) uPR (v)

PBT (u, v) (1 v)PB (u ) vPT (u )
P(0, v) PLR (0, v) PBT (0, v)
PL (v) (1 v)PB (0) vPT (0)
must
PL (v) (1 v)P0,0 vP0,1 PL (v)

On the left edge, the extra terms are identified as

(1 v)P0,0 vP0,1
Similarly, the other extra terms are found along the other edges and all
surplus terms are

surplus (1 v)P0,0 vP0,1

(1 v)P1,0 vP1,1
(1 u )P0,0 uP1,0
(1 u )P0,1 uP1,1

P(u, v) PLR (u, v) PBT (u, v) surplus

Advantages:
Follows boundary curves
Limitation:
not able to control internal shape
 E XAMPLE
For given data along four edges, use cubic Bezier curves along the edges
to create Coons patch.
First find the curves along the edges. Their three components in (u,v) are

L,R,B, and T refer to left, right, bottom, and top.

The plot is shown below.

L,R,B, and T refer to left, right, bottom, and top.

The Coons patch is then obtained as below and the plot is shown with
data points.
B EZIER S URFACE
E XAMPLE
E XAMPLE [P ROGRAMMING ]
B-S PLINE AND NURBS S URFACES
 E XAMPLE [P ROGRAMMING ]
1

3 4 6 9

1

1 1 1 1

7

3 4 6 9
1 2 3 4 5
1
3 4 6 9 3 3 3 3 3 5 3 9 6 3 1 5 25 6 2
xp , yp
, zp
wt
,
1 3 4 6 9 4 4 4 4 4 4 5 2 5 4 4 5 8 6 3

1 9 9 9 5 6 5
3 4 6 9 9 9 9 4 3 6 8 6 4
O THER S URFACES
 S URFACE M ANIPULATIONS

As in the curve manipulations, the surfaces have to be manipulated for

various reasons in the CAD designs. Some of the applications are
D ESIGN AND E NGINEERING
A PPLICATIONS
 S OLIDS AND T HEIR
R EPRESENTATIONS
Solid Model is based on informationally complete (or spatial addressability),
valid, and unambiguous representation of objects and stores geometric data
as well as topological information of associated objects. This representation
permits automation and integration of tasks such as interference analysis,
mass property calculation, finite element modeling, CAPP (computer-aided
process planning), machine vision, and NC machining.

It is very easy to define an object with a solid model than other two previous
modeling techniques (curves and surfaces) because solid models do not need
individual locations as with wireframe models.
 G EOMETRY AND TOPOLOGY

The above figure illustrates the difference between geometry and topology.
The geometry that defines the object is the lengths of lines, areas of
surfaces, the angles between the lines, and the radius and the center of the
cylinder and the height. On the other hand, topology (sometimes called
combinatorial structure), is the connectivity and associativity of the object
entities. It has to do with the notion of neighborhood and determines the
relational information between object entities. From a user point of view,
geometry is visible and topology is considered to be nongraphical relational
information that is stored in solid model databases and are not visible to
users.
 S OLID E NTITIES
There are various basic building blocks, so called, primitives that can
be combined in certain boolean operations to construct complex
models. They include:
block
cylinder
cone
sphere
wedge
Torus

In the previous figure, a block and a cylinder were combined with

union (i.e., addition) and difference (i.e., subtraction). One more
available Boolean operation is intersection.
 E XAMPLE
Example 5.6.1.1. Explain how to construct the solid model of the
bearing support with primitives and boolean operations.
 S OLID R EPRESENTATION
Underlying fundamentals of solid modeling theory are geometry, topology,
geometric closure, set theory, regularization of set operations, set
membership classification, and neighborhood. Solid representation is
based on the notion that a physical object divides an n-dimensional
space, En, into two regions: interior and exterior separated by the
boundaries. A region is a portion of space En and the boundary of a
region is a closed surface.

The set in the set theory is a collection of objects and the operations
include complement, union, and intersection. Further, regularized set
operations are used to avoid irregular object created by boolean
operations. Regular set is a geometrically closed set. The set
membership classification determines if some objects intersect with a
given object.
Half-spaces are unbounded geometric entities; each one of them divides
the representation space into two infinite portions, one filled with material
and the other empty. By combining half-spaces in a building block
fashion, various solids can be constructed.

Thus, a solid model of an object can be defined as a point set S in three-

dimensional Euclidean space E3. If three sets for a region is denoted by
Si (interior set), Se (exterior set), and Sb (boundary set), then the set for
the solid model can be expressed as
The mathematical properties that the solid model should capture can be
stated as:

Rigidity
Homogeneous three-dimensionality (ie. no dangling boundaries)
Finiteness and finite describability
Closure under rigid motion and regularized boolean operations
Boundary determinism (ie. Boundary must contain the solid)

The mathematical implication of the above properties suggests that valid

solid models are bounded, closed, regular, and semi-analytic subsets of
E3. These subsets are called r-sets (ie. regularized sets), which are
curved polyhedra with well-behaved boundaries. Here, regular means
no dangling portion and semi-analytic means no oscillation in value.
 F UNDAMENTALS OF S OLID
M ODELING
Various solid representation schemes require several underlying
fundamentals of solid modeling theory. They include geometry, topology,
geometric closure, set theory, regularization of set operations, set
membership classification, and neighborhood. For detailed discussion of
these topics, the readers are encouraged to refer to advanced topics on
solid modeling.

In the following some of most popular solid modeling techniques are

discussed.
 M ODELING T ECHNIQUES
Boundary Representation (B-rep): A B-rep model is one of the two
most popular and widely used schemes and is based on the topological
notion that a physical object is bounded by a set of faces. The following
shows various faceted B-rep solids.
Constructive Solid Geometry (CSG): This is based on the
topological notion that a physical object can be divided into a set of
primitives that can be combined in a certain order following a set of rules
(i.e. Boolean operations) to form the object. The basic elements are
block, cylinder, cone, sphere, wedge, and torus and building operations
are Boolean operations. The following bearing support was constructed
with various primitives in a certain sequence by CSG technique.
Sweep Representation: This is especially useful for two-and-half
dimensional objects used most frequently for extruded solids and
revolved solids. This is based on sweeping of a section along a path that
may be linear, nonlinear, and hybrid operations. When the path is
straight, its a simple extrusion and when it is axisymmetric, it becomes
the revolution. For nonlinear path, the sweeping is done along a
nonlinear curve in space. Cutting tool path simulation is one good
applications of this technique.

The section may vary along the sweeping path. The following shows such
an example with variable section.
Analytic Solid Modeling (ASM): Historically it is closely related to
three dimensional isoparametric formulation of finite element analysis for
8- to 20-node hexahedral elements. This arose from the need to model
complex objects for finite element analysis. ASM uses the parametric
representation of an object in three-dimensional space that is a mapping
of a cubical parametric domain (so called, master domain) into a solid
described by the global coordinates (MCS).
 Other Representations:
primitive instancing: is based on notion of families of objects or
family of parts.
cell decomposition scheme: an object can be represented as the
sum of cells.
spatial enumeration scheme: a solid is represented by the sum of
spatial cells that it occupies. The cells have fixed size.
octree encoding scheme: is generalization of spatial enumeration
scheme with variable cell size.