You are on page 1of 29

LOFTING PROBLEMS OF STREAMLINE BODIES

ROY A. LIMING
Head of Engineering-Lofting Mathematics
(NORTH AMERICAN AVIATION)

Basic Projective-Analytic Principles of Conic Lofting

The ability to determine the “fairness” of a curve by eye, aided principally with “spline and duck,” is
developed only after years of extensive practical experience. This same experience, however, only
confirms the fact that no two individuals “see” a curve exactly the same, regardless of how extensive
such experience may be. “Trial and error” procedure of this type signifies just what the phrase
implies, namely, lack of control and specifically directed effort in the lofting process.

Fairing transverse sections in the body plan of a streamline body constantly imposes the necessity of
establishing a “fair” curve defined by two point-slopes and a control (shoulder) point. Practically,
there is only one “fair” curve which may be fitted to such a set of conditions, and that is a conic
section.

The P-51 “Mustang,” world’s fastest of the complete application of lines which are so evident ore based propeller driven airplane,
represents the first instance in aircraft history conic sections in the streamlining of a fuselage shape. The sleek upon the conic controls
illustrated by the perspective view of fig. 92

The tremendous speed of the “Mustang” - more than 450 mph - is attributed in considerable measure to the small frontal
area of the fuselage, so evident in this view

1
This plan view of North American’s P-51 fighter shows the compactness of the controlling lines of the fuselage shape

2
Fig 92 Preliminary to conic lofting of fuselage lines the maximum depth and half breadth at each of five critical control
stations are established from basic design and aerodynamic data. The upper and lower centerlines are thus each defined
by five points, through which a conic may be developed by the projective method shown in figs. 93 and 94

Any other curve, regardless of construction methods, is only an empirical approximation of the
indicated curve, i.e., the particular conic section defined by the given conditions. The highly skilled

3
loftsman will, with the “fairing-by-eye” technique, finally approach this conic section quite closely
with his “trial and error” procedures.

Obviously, then, any layout system other than one embodying the application of conic sections in
such a development tends to be circuitous and clumsily indirect in application. On the other hand,
with the use of conic sections the engineer has available a set of projective-analytic techniques, the
application of which results in a geometrically control-led, direct, definite, and immediate solution of
the problem.

No premium is placed upon long years of experience with “spline and duck.” With analytic
calculation or projective development of a sufficient number of points, consistently uniform,
accurate, and easily checked results are assured in the development of the curve. Both the analytic
and projective techniques are relatively simple. Experience has proved that mechanical mastery of
either or both can be established with a few hours practice.

The overall dimensional specifications of a streamline body such as an aircraft fuselage are based
upon considerations involving such heterogeneous factors as type of design, whether fighter, bomber
or transport; type of propulsion, whether internal combustion, jet or rocket propulsion; location and
number of subordinate developments, such as nacelles, scoops, turrets, enclosures, etc.; space
requirements for members of the crew.
In general, when the body plan section is considered, at least three major factors must be weighed for
any chosen type of design:

1) Clearance of essential items of equipment and installations related to powerplant, and


provision of space for personnel.
2) Stressing of the fuselage to withstand bending loads.
3) Streamlining of the fuselage to secure maximum aerodynamic efficiency.

The interplay of these three factors will determine the exact nature of the transverse section in any
given fuselage station plane. With aircraft speeds approaching and exceeding sonic velocities, the
third of these factors naturally increases in significance and relative importance as a definitive
criterion.

As a preliminary step in conic lofting of fuselage lines, the maximum depth and maximum half-
breadth at each of five critical control stations are established from basic design and aerodynamic
data (fig. 92). The upper and lower centerlines are thus each defined by five points, through which a
conic may be developed by the projective method illustrated in figs. 93 and 94.

This preliminary analysis may be conducted through the use of some convenient reduced scale. If it
is assumed that the maximum half-breadth curve is located in a specific plane, e.g., a waterplane,
then the same convenient procedure applies in the development of this curve.

In many developments, however, particularly in fighter design, this simplified treatment of the
maximum half-breadth is not practicable. The maximum half-breadth may, of necessity, be curved in
both the side and plan views in order to affect a satisfactory compromise with respect to the three
factors mentioned above.

4
Figure 93

5
6
Figure 94
The recommended order for the establishment of construction lines is based np&n the following two considerations:
1. New points will lie upon the solid red construction lines. Therefore, these lines may be used to directly “spot” the
approximate location desired for each new point. If Pascal lines (broken red) are drawn first, control over the
locations of newly developed points will not be as direct. In actual layout, only two short segments of the solid red
ray need be drawn, one defining the intersection of the ray with the construction lines drawn through the control
point, the other located in the immediate vicinity of the desired new point. A few hours’ practice will enable one to
reduce the actual length of construction line segments to the very minimum.
2. 2. The order of establishing the necessary construction lines indicated results in a maxim urn of accuracy in the
projective development of new points on the conic. This conclusion is immediately fustifièd upon even the most
cursory examination of the mechanics of the procedure described.

It is obvious that a slight layout tolerance may occur in the construction of a Pascal line (broken red) through the
intersection 01 a ray (solid red) with the construction lines (solid black) drawn through the control point. This tolerance,
however, is proportionately less at the intersection of the Pascal line to which the broken black ray is drawn, which
means that the new point has a minimum displacement as a result of the layout tolerance. If the order of construction of
the rays and Pascal lines is reversed, a maximum error in location of the new point affected will result for any given
layout tolerance.

With the maximum depth and width thus established, the slope conditions in the body plan view may
now be decided upon. In this view, the simplest arrangement involves right-angle relations between
the tangents and the reference planes (fig. 95).

Figure 95

An alternative arrangement (fig. 96) is to establish the common tangent to the upper and lower frame
contours at other than a right angle with the horizontal tangents. This angle may be either fixed or
variable, depending upon the requirements imposed in the basic design of the aircraft. If the angle is
fixed, it may be more practical to return to the more convenient relations of fig. 95 by selecting the
maximum half-breadth as the point of common tangency (fig. 96).

7
Figure 96
The precise definition of the control or shoulder curves is the final step in the formulation of the
basic fuselage lines. The procedure indicated here involves returning to the five control sections (fig.
92). In each of these sections, critical conditions affecting the given section are examined, and conics
are fitted to various clearance points until a satisfactory contour for that station is established.

This phase of the lines development is likewise featured by the utilization of conic techniques. In this
case the familiar projective method of establishing a conic, given two point slopes and control point
is employed.

When this analysis is complete for all five control sections, a shoulder plane making a convenient
angle with the profile (or horizontal) reference plane may be passed through these sections. A conic
may then be fitted to the five points of intersection thus obtained. This conic is most conveniently
developed in a true view of the shoulder plane.

With the establishment of this line, the body plan of the fuselage may be completed, showing any
desired fuselage station. All such transverse sections are, of course, projectively developed from
analytically determined control points calculated from the equations of the longitudinal control lines.
Basic principles governing such equations have been discussed in detail in previous articles in this
series.

This brings us to the projective discriminant as a conic control factor.

Certain projective relationships exist with respect to the intersection point a conic section with a line
joining intersection of tangents and the mid-point of the common chord through the points of
tangency (fig. 96).

Thus, with reference to fig. 97, when the ratio ( k ) of A to B is equal to unity, the conic is a
parabola; when the ratio ( k ) of A to B is less than unity, the conic is elliptical; and when the ratio
( k ) of A to B is greater than unity, the conic is hyperbolic.

8
Figure 97

Since this ratio enables one to discriminate with respect to the nature of the conic section, and since
the geometric basis for this ratio is projective character, the phrase projective discriminant (*) may
well be adopted to describe the ratio.

*For the projective derivation of the projective discriminant in complete detail, see Practical Analytic Geometry with
Applications to Aircraft - Liming (Macmillan - 1944 ) - Chapter XIII - “Conic Lofting” pp. 252 ff.

Profile view of the P-51 revealing the modifications of the basic control lines required to integrate the cockpit enclosure
and scoop developments. These subsidiary conic developments blend with characteristic flexibility and ease with the
basic conics of the fuselage shape

9
The student of projective and analytic geometry will readily recognize the need for this geometric
concept from the tabulation below, which shows identifying characteristics of the conics from the
viewpoint of projective-analytic geometry as well as from the more familiar metric viewpoint:

The analytic coefficient ( R ) – above - is identified with the general equation of a conic expressed in
the form y  Px  Q  Rx 2  Sx  T , and is observed to he the coefficient of the square under the
radical. This coefficient has likewise been analyzed in complete detail relative to its meaning as a
discriminant.

Body-Plan Development

Once the basic longitudinal control lines have been established analytically, the lofting of the
fuselage shape may then be consummated in the body-plan view.

Any given transverse section may be found by calculating the intersection of the plane of the section
with the basic lines. These analytically determined frame control points may be used to govern the
projective development of the frame contour to any desired scale, full or reduced, in orthogonal or
perspective views.

It is highly desirable that these projective-analytic lofting controls be imposed in the preliminary
design stage. In this way, systematic control of successive stages of design is assured. There is a
minimum of duplication of developments in the transition from experimental to production
engineering, and a maximum of precisely coordinated information available at every step in the
progressive evolution of the project. The designer may secure, on a few hours’ notice, complete
analytic description of all transverse sections located at regularly spaced intervals (e.g., 5’’ intervals),
as well as special frame sections located in an intermediate position.

There is no need to rush completion of actual full-scale lofting of such lines in the early design stage.
As a matter of fact, such practice is to be strongly condemned from the viewpoint of economy of
time and materials as well as protection of quality in finished lines layouts. In other words, analytic
description should largely supplant such layouts in early design development; actual full-scale layout
should proceed at a pace which will insure an orderly, accurate, and timely formulation of the full-
scale lines release, with a minimum of changes due to the lack of decisive definition of the design
inherent at the outset of every new design project.

From ordinary metric geometry, we have acquired familiarity with certain characteristics of the conic
sections. We have observed that the conic is basically defined from that viewpoint as the intersection
of a plane with a right circular cone; furthermore, that an algebraic definition of the conic, based
upon the simultaneous solution of the equation of this intersecting plane with the equation of the
cone, results in an equation of the second-degree for the curve of intersection.

A conic section may, therefore, be defined as a second-degree curve, since it is represented


algebraically by a quadratic equation.

The general equation of the second- degree in x and y can be written in the form

10
Ax 2  Bxy  Cy 2  Dx  Ey  F  0

Nevertheless, while this equation has six constants, since any one of these may be divided out, there
are actually only five independent coefficients. Therefore, five independent conditions determine the
equation of a conic.

We are thus in a position to examine the procedure for finding the equation of a conic based upon
five given conditions. In particular, let us consider the equation of a conic through five points, no
three of which are collinear.

In order to write the equation of a conic through five given points, we shall find it convenient to
write the equation of the pencil of conies passing through four of the points, and then introduce a
parameter so that the conic which passes through the fifth point may be uniquely determined.

Let us now assume that we have a pencil of conies consisting of all the point conies which pass
through the four points, A, B, C, and D (see fig. 98). This pencil we define as the totality of conies
passing through these four points. Since each two of the conics intersect in these points, any two
distinct conics of the pencil may he employed as the base conies of the pencil.

Figure 98

As we pointed out in the first article of this series (Lofting Problems of Streamline Bodies Part l - Aero Digest -
Dec. 1941), the most convenient method for developing the equation of a conic based upon five given
conditions is to employ the two degenerate conics as the base conics of the pencil. We are reminded
at this point that a point conic is degenerate if and only if it consists of two straight lines, distinct or
coincident

Referring again to fig. 98, let us assume the following equations:

Line AD :   0
Line BC:   0
Line AB :   0
Line CD :   0

Then   0 , and   0 are the equations of the two degenerate conics which we are using as our
base conics for the pencil, and

  k  0 (141)

11
is an equation of the pencil consisting of all the conics passing through the four points, A, B, C, and
D. The value of the parameter k is determined by imposing the fifth condition upon the conic, i.e.,
by evaluating eq. (141) for the fifth point (point E of fig. 98) through which the conic is to pass.

Obviously,


k  (142)

The equation of the pencil is thus simply a linear combination, i.e., a combination of the equations of
the straight lines constituting the degenerate base conics of the pencil; it is also an equation of the
second degree. Furthermore, the four points, A. B, C, and D, can be shown to lie on the locus of the
equation.

Since the practical use of the conic section in aircraft design and lofting procedures has revolved
almost exclusively about fitting the conic to the five conditions which include three points and the
slopes at two, we are immediately concerned with applying the general analytic theory developed
above to this particular case.

At this point we find it relatively easy to derive the equation for the conic tangent to each of two
lines at a given point and passing through a third point.

In fig. 99, each two conics have double contact at A and D. In other words, points A and B and
points C and D (see fig. 98) coincide in this case. Thus, all the conics of the pencil have the same
tangents t1 and t 2 at A and D respectively; conversely, every conic which is tangent to t1 at A and to
t 2 , at D is uniquely determined by a third point and therefore belongs to the pencil.

Figure 99

There are two distinct degenerate conics in the pencil. The tangents t1 and t 2 form the one, and the
chord of contact taken twice constitutes the other.

12
Let t1  0 and t2  0 represent the equations of the two tangents; let l  0 represent the equation of
the chord AD. Then the equation of the pencil, from eq. (141), becomes

t1t2  kl 2  0 (143)

Solving for the parameter k , we find that

t1t2
k  (144)
l2
To develop the equation of any particular conic of the pencil, it is necessary to specify a third point,
known as the control point. The value of the parameter k is determined by evaluating eq. (144) for
the coordinates of the control point. When this value is substituted for k in eq. (143), that expression
becomes the equation of the one and only conic (of the pencil which passes through the given control
point.

We have outlined in considerable detail the procedure by which we may arrive directly and
conveniently at an analytic expression for the general conic. However, this expression is established
in the form f ( x, y)  0 . Practical experience shows that this form of the equation is of little value,
inasmuch as we are primarily concerned with establishing arbitrary points on a required curve.

An equation in the form y  f ( x) [or x  f ( y ) ] in which y represents an ordinate corresponding


to a given arbitrary abscissa (x-value) is clearly indicated. There are two possible procedures:

First, we may develop the equation constants for the f ( x, y)  0 form, then from the values thus
established secure the equation constants for the y  f ( x) and x  f ( y ) forms. Second, we may
proceed directly from basic conditions to the equation constants for the y  f ( x) form of the
equation. The x  f ( y ) equation constants, if needed, may be secured by direct transformation of
the y  f ( x) constants.

What are the criteria which should govern our choice of procedure? The following suggest
themselves as of fundamental importance in industrial applications:

(1) Number of chances for error;


(2) Mechanical speed of development through adaptation to available calculator equipment;
(3) Maximum utility of developed data;
(4) Directness of procedure in calculation from original basic data to required data;
(5) Simplicity of checking procedures;
(6) Conformity to design procedures.

Let us consider the two procedures in the light of these criteria. First, how do the chances for error
compare? The original basic conditions are points; coordinates of these points are the truly original
basic data postulated by the designer. Logically, the shortest calculation route between these original
coordinates and the necessary equation constants should produce the fewest chances for error. This
route is lengthened, for example, by the use of slope-intercept description of the degenerate conics.
An extra checking procedure must be initiated to accompany this step.

13
Four Checking Procedures Required

If we proceed from this step to the development of the constants for the f ( x, y)  0 form of the
equations, we must institute another checking procedure to verify these values. From this second step
we are then required to move on to the development of constants for either the y  f ( x) form or
the x  f ( y ) form, or both, each of which compels a check for accuracy. We have thus elected a
procedure which requires four distinct and independent checking procedures, compared with two
such procedures if we choose to calculate the y  f ( x) and the x  f ( y ) equation constants
directly from the basic coordinate conditions.

Given certain basic conditions, we are interested in developing specific data defining the curve, viz:
arbitrary points on the curve, slope and radius of curvature values at specified points, area under the
curve, location of the centroid, etc. The development of data which are not directly essential to the
achievement of these ends tends to defeat the practical character of the whole idea of analytic
control. All these desired data may be developed from the equation in the y  f ( x) [or x  f ( y ) ]
form.

Of tremendous importance in the discussion of the advantages of an analytic system of calculation is


the factor of speed in the development of a comprehensive body of accurate data. Indeed, this factor
ranks at the in the final practical analysis. In other words, analytic control can justify itself in the
teeth of competition with other techniques only insofar as superior efficiency in the utilization of
man-hours, materials, and equipment is demonstrated.

Streamlining calculation procedures in order to conserve the time element to the absolute maximum
is not merely desirable, it is absolutely imperative in competition for survival. Evaluation of the
comparative of an analytic system of technique as against other methods must inevitably face the
acid test of this criterion.

Conflicting views have been expressed regarding the practicability of calculation within the body
plan view itself. This conflict of opinion does not revolve about the necessity for determination of the
control which are used to govern the projective development of the contour of a transverse section.
The overwhelming advantages of this system are generally conceded.

The question which arises concerns the need for additional calculated points on the curve, such as
conventional waterplane and buttockplane intersections with the contour, intersections of basic
structural planes with the contour, or the calculation of projected and true views of canted fuselage
bulkheads. Shall such information await the full-scale loft layout and the scaling of templates, or
shall provision be made for calculation, in whole or in part, of such data?

A practical solution to this problem has been successfully formulated, as follows:

(1) The designer and loftsman are equipped with the correct layout procedure for the projective
development of the body-plan conic from analytically established control point.
The projective principle involved is illustrated by fig. 100. The correct layout procedure for
maximum speed and projective accuracy in the development of points on the desired conic is shown
in fig. 101. This projective technique is employed up to the limits of physical convenience. For
transverse sections of fuselage shapes too large for convenient projective development, the conic is
established through the plotting of analytically determined points.
(2) In either case, equations of the body-plan contours are established. A projective development can
therefore be paralleled with or entirely replaced, at any time, by calculation. Any point on the surface

14
of the fuselage can be secured on a moment’s notice by calculation. Scaling of layouts, with all the
evils attendant upon such practice, is thereby virtually eliminated.

Figure 100

15
Figure 101

Time Element Paramount

The calculation of such coordinate data requires the establishment of the equation constants of the
body-plan contours. Obviously, the time element is paramount in determining the feasibility of
entering the body plan with such analytic controls. If a streamline calculation procedure is available
which takes full advantage of any possible simplifications, proceeding directly from the calculated
control point coordinates to the required equation constants, then such analytic control, always highly
desirable, becomes equally practicable.

Let us consider a typical body-plan development. In this view (see fig. 102), the tangents and
coordinate reference lines form a mutually perpendicular system. We are quite likely however, to
occasionally encounter modifications involving an obtuse or acute angle rather than a right angle
between the tangents (see fig. 103).

16
Figure 102

17
Figure 103

We shall, therefore, generalize our body-plan approach to include these variations. There is no
increase in complexity as a result of this step. Referring to the curve conditions shown in fig. 103, let
us assume as coordinates for the points O, B. C, and D the following:

O (0,0)
B ( a, b)
C (0, c )
D ( d , e)

Applying the analytic principle which we have just presented, we find the following equation for the
conic governed by the various conditions shown in fig. 103:

y  Px  Rx 2  S

in which

1  2bk
P
2ak
1  4ck
R 1+2bk
(2ak ) 2
c
S
ak

The value of the parameter k is given by the expression:

18
(ae  ac  cd  bd )d
k
(ae  bd )2

This formula applies universally to conies in body plan and any other view as long as one tangent is
vertical through the origin of the coordinate system of the curve.

Consider the simplicity of the calculation chart required to develop the equation constants of body-
plan conic sections using the standard automatic electric-calculating machine. This analytic
procedure has been found far superior to any other approach when evaluated in the light of the
criteria discussed herein. Chart 10 may be completed by reading the basic control coordinates
directly from the original tabulation, eliminating the danger of miscopying the coordinate data.

It should be noted that when the two tangents to the conic are perpendicular, c  b , and the value of
k simplifies to

(ae  ab)d
k
F
This simplification involving the mental substitution of b for c is actually employed in the example
in chart 10, although the k -column should show the heading

(ae  ac  cd  bd )d
F
which

F  (ae  bd )2 .

and c should replace b in the ( R ) and ( S ) columns to cover the general case. Thus, one chart form
suffices for all cases.

In either event, the numerator can be quickly obtained in one continuous operation on a calculator
featuring automatic cumulative addition and subtraction of products.

We are now concerned with defining the absolute minimum effort needed to establish the “necessary
and sufficient” condition for verification of the accuracy of our equation constants. Basically, we
have established a conic through five points; therefore, we must prove that our conic passes through
these five points.

Chart 10

19
Actually included in our five points are two pairs of coincident points, since we have dealt with the
special case of two-point slopes and a control point as the basic conditions for the conic. Therefore,
our checking procedure reduces to substituting the coordinates of the two points of tangency and the
control point in our equation, a purely mechanical step adapted to the high-speed calculation
operations illustrated by chart. 11.

Chart 11

Attention should be focused at this point upon the relative ease of completing this check. Since we
have employed coordinates directly in our calculations, we are not required to check the accuracy of
the numerical values of the two tangents to the conic; visual inspection of the coordinates of the point
of intersection of the tangents before the calculation of the equation constants is actually started is
sufficient to cover this point.

On the other hand, if we use slope-description of the base-degenerate conics we are required to add
the calculation of the first derivative values at the two points of tangency in order to completely
establish the “necessary and sufficient” condition for verification of the equation constants. This
necessity immediately completes the bogging down indicated by the f ( x, y)  0 procedure, and
impractical the analytic control of the body plan by this approach.

Analytic calculation controls of the type recommended herein have proved of definite practical value
in the following specific applications:

(1) Calculation of canted fuselage sections of all types.


(2) Calculation of intersections of structural planes with the fuselage molded surface.
(3) Calculation of body plan sections (both normal and canted) of extreme-large fuselage shapes.
(4) Calculation of critical mold-line conditions affecting the design of structure closely coordinated
with molded surfaces of the fuselage. This calculation is possible long before projective development
of the layout, and replaces the need for the latter in most cases.
(5) Completion of the concentration of responsibility for dimensional coordination in the hands of
one group, from the preliminary design to the production tooling stage.

It should be pointed out that we consistently employ one general basic equation in establishing the
specific equation of a body-plan contour. This same general equation, developed in the preliminary
part of this article, is basic to our whole approach to analytic analysis of the general conic.

We may well conclude on the basis of this discussion that in order to capitalize upon the flexibility
which is so inherently characteristic of the conic section, we must preserve the same quality in our
choice of techniques and methods of application.

20
The Projective Discriminant as a Conic Lofting Control Factor

One of the basic conic control factors which of itself contributes so heavily to the realization of the
advantages described above is the projective discriminant, derived and identified by this new and
original phrase earlier in this series.*

We may proceed to a restatement of the definition of this concept by referring to the conditions
illustrated in fig. 104. This figure shows what we have previously pointed out as the most commonly
recurring set of conditions in actual engineering layout problems, namely, two point-slopes and a
control point governing the required curve. Let us examine the triangle OBC formed by the two
degenerate base conies of the pencil of conies, i.e., the pencil governed by the two tangents t1 and t 2
and the two points of tangency, O and B, respectively.

Figure 104

Each conic of the pencil intersects the median CE in a point, such as D. It is the ratio of the segments
of this which has been designated as the projective discriminant of the conic. For reasons already
cited, we have defined this ratio ( k ) as follows: if k  ED / DC  1 , the conic is an ellipse; if
k  ED / DC  1 , the conic is a parabola; and if k  ED / DC  1 , the conic is a hyperbola. By
means of this extremely useful concept, we are enabled to directly control the choice of the conic
section best adapted to a given layout situation. This is achieved by simply varying the position of
the control point along the median.

Even when it is necessary to choose a control point elsewhere than on this median line, the
intersection of the resulting conic with this line may be used thereafter as the control point of the
conic. This follows from the fact that any point on a conic may be used as a control point for the
projective or analytic development of the curve.

The Projective Discriminant

The use of the projective discriminant in the fairing problems illustrated by figs. 105 and 106
indicates special applications of parabolic sections to meet certain typical contour requirements.

21
The parabola affords a maximum degree of simplicity from an analytic viewpoint, since its equation
may be expressed in the form y  p x  qx . Integration of such an equation to obtain the area
under curve, a procedure associated with airfoil tip fairings, is thus relatively simple.

Figure 105

22
Figure 106

Figure 107

The projective discriminant affords a convenient alternative to the use of the shoulder plane in the
fairing of a streamline shape such as an aircraft fuselage.

Two basic considerations will influence the choice of the projective discriminant in a given fairing
development, viz:

23
(1) In a large fuselage development in which the designer is more or less free of the need to consider
special structural clearances in certain areas, one specific value can be applied to the projective
discriminant to control the complete body-plan development.
(2) At sonic and supersonic speeds, a more positive control is required over radius of curvature
conditions. The designer is compelled to reduce the deviation range in this particular variable to the
absolute minimum. Control of the body-plan development through the use of the projective
discriminant will facilitate adherence to the limits imposed by the aerodynamicist for high-speed
performance, and still permit a flexible adjustment to other design factors.

In order to grasp fully the basic procedures required to “fair” the fuselage shape by the use of the
projective discriminant, let us return to the point at which several critically located control section
contours have been determined from basic aerodynamic and design data.

At this point, instead of passing a shoulder plane through these basic section, and subsequently
fitting a conic to the resulting points of intersection, let us construct our triangle of the degenerate
conics in each section, and establish as closely as possible the projective discriminant value for each
conic section. From a comparative analysis of these values, we shall be able to impose a uniform
projective discriminant value upon the whole development.

If our analysis of the basic sections shows too great a range in the value of this ratio, a
reconsideration of the basic centerlines, both upper and lower, and the maximum half-breadth
conditions may be required in order to reconcile our discriminant values. However, in large-scale
fuselage developments this situation will rarely develop.

Three-Dimensional Control with the Projective Discriminant

The projective-analytic development of the streamline shape of the fuselage as practiced at North
American eliminates the need for waterplane and buttock plane fairing since the developed shape is
known to be automatically “fair” when its development is based upon such procedures. Only the
body plan view consisting of the basic structural frame contours is developed in full scale lofting
layout practice.

The use of the projective discriminant as a constant to define the “fairing” of a streamline shape
offers among other advantages from an analytic viewpoint the opportunity of reducing the
establishment of the shoulder or control point to a simple calculation exercise in linear algebra.

Earlier in this series we predicated the definition of the shoulder point upon the quadratic algebra of
the conic section. We shall also indicate later that the projective discriminant may be introduced into
a “fairing” problem as a variable involving quadratic as well as linear variation.

The advantage of this simplified control over other systems, e.g., the one employing the familiar
“two-circle” or “synthetic” ellipse control in the body plan development, is immediately apparent.
Not only do the procedures involved in the projective discriminant approach afford complete three-
dimensional calculation control, they also afford maximum simplicity in effecting the modifications
of basic lines which are required in certain types of subsidiary developments.

In other words, a highly flexible condition is maintained to serve as the foundation for all subsequent
modifications, with the resulting “fairing” still amenable to convenient projective-analytic controls.

24
Analytic procedures based upon the projective discriminant.

Let us proceed to examine the typical body plan conditions of a fuselage shape. In fig. 108, we
observe a normal transverse section with the coordinate xy-system of the upper contour
superimposed upon the basic fuselage coordinate system.

The coordinates of the points of intersection of this plane with the basic centerlines, upper and lower,
and the maximum half-breadth conic control lines are made available in our projective-analytic
system from previously established analytic data.

Let us define our xy-system for the upper contour conic as follows:

O (0,0)
B ( a, b)
C (0, c)

In the triangle formed by the degenerate conics of the indicated pencil of conics, let us indicate the
control or shoulder point located on the median drawn from the vertex C, as D(d , e) .

This point divides the median into two segments, A and B, the ratio of which, expressed as A / B
we have designated as the projective discriminant ( k ) of the conic through the point C.

Derivation of Formulas

Our next step is to derive formulas for the coordinates (d , e) of the control point D in terms of the
given coordinate data, Le, the centerline and maximum half-breadth points defined above in our xy-
systems as (0,0) and ( a, b) .

Let us denote the coordinates of the midpoint of OB as (a / 2, b / 2) .

Then, from elementary geometry,

d  a / 2  kd
or

d (1  k )  a / 2

whence

a
d (145)
2(1  k )

Similarly,

b / 2  e  k (e  b)

or

25
e(1  k )  b(k  .5)

whence

b(k  .5)
e (146)
1 k
Eq. (145) and (146) are the required formulas.

Let us assume the following numerical coordinate data for the transverse section shown in fig. 108:

O( X f , Z f ) : O  0, 55.156 
B( X f , Z f ) : B  27.889, 0 
C ( X f , Z f ) : C  27.889, 55.156 

26
Figure 108

Translating to the xy-system, we have a = 55.156 and b = 27.889. Assume that the desired contour is
de fined by a projective discriminant (k) value of 0.8694.

Then, by substitution in eq. (145) and (146),

d = 14.752
e = 20.430

which, translated to the fuselage reference system, becomes

D( X f , Z f ) : D(20.430, 40.404)

Thus, when a constant value is assigned to the projective discriminant, we are able by means of
elementary linear algebra to define the coordinates of the control or shoulder point of the contour in a
given transverse section or fuselage station plane.

We observe the following specific advantages accruing from the use of the projective discriminant in
such streamline fairing applications:

(1) Positive control over the radius of curvature conditions along the axis of longitudinal airflow.
(2) Complete three-dimensional control over the streamline shape for lofting and tooling
developments.
(3) Maximum degree of simplicity in calculation procedures.
(4) Highly selective control over the type of contour in the body plan development.
(5) Minimum time required in the preliminary design stage of formulating the basic shape.

27
Fig. 109 - The case involving the development of a conic through two point-slopes and a control point in which the point-
slopes are parallel is of interest in that such parallel elements are assumed to intersect in a point at infinity. Therefore, all
Pascal lines through this point of intersection are drawn parallel to the given point – slopes.
A glance at the conditions defining the projective discriminant of such a curve immediately verifies our conclusion that
all conics governed by such conditions are elliptical, since obviously the value of the projective discriminant ratio will
always be less than unity.

28
Fig. 110 - Layout procedure in the application of projective techniques is determined by the necessity for attaining a
maximum degree of accuracy in the definition of the curve. In general, the guiding principle for layout procedure may be
stated thus: points on the desired curve should result from a construction order which establishes these points as
intersections of converging rays. The engineer skilled in layout work will immediately recognize from the physical
conditions of the required development the proper order for establishing the construction lines in order to reduce the
effect of layout tolerances to a minimum

29

You might also like