COMBAT GROSS WEIGHT 64,000 LBS.
WING AREA 1,225 SQ FT. LENGTH 77'9.65"
SPAN 50' HEIGHT 21'.3"
are a number of relatively unconventional features
on the Arrow and a reasonably detailed appraisal of
these might easily fill a volume of 500 to 600 pages.
Therefore, I intend to pick out a few of the highlights
and present a broad-brush picture of the design philosophy
The RCAF had established a requirement for a two-place, twin-engined
aircraft. Their preference for a crew of two was partly based on the complexity
of the newer fire control systems and the fact that, while the chosen system
was intended to be entirely automatic during the midcourse and terminal phases
of the attack, it was the intention to press home an attack on the
basis of a manual mode, in the event of the failure of the automatic
The choice of two engines was based on a combination of circumstances,
the advantages of two engines being obvious in reduced attrition, especially
during training. One of the most important factors, however, was the fact that
with the very large weapon package required as payload, and the large amount
of fuel carried for the range requirements, the size of the aircraft was obviously
going to be such that there was no single engine large enough to power it.
The configuration of the basic fuselage was determined almost entirely
by the two-seat, two-engine arrangement and the large armament bay. I will deal
more specifically with these items later.
OF WING DESIGN
the time we laid down the design of the CF-105, there
was a somewhat emotional controversy going on in the
United States on the relative merits of the delta plan
form versus the straight wing for supersonic aircraft.
We tried very carefully not to become inhibited by association with
either side and our choice of a tailless delta was based mainly on the compromise
of attempting to achieve structural and acroelastic efficiency, with a very thin
wing, and yet at the same time, achieving the large internal fuel capacity required
for the specified range.
This established our delta plan form and the lack of a tail can be
attributed almost entirely to our desire not to have to face the problem of putting
a tail on top of an extremely thin fin out of the effects of wing downwash, or,
otherwise, having to put it so low, again out of the downwash region, that our
landing angles would be impossible. We felt that the problems associated with
a tailless delta were more predictable and manageable.
We were also very conscious of the problems that tailed deltas were
having at that time, where a large increase in downwash at the stall made the
tail strongly destabilizing, so that the stalling characteristics became objectionable.
It was obvious from the outset that to give the RCAF an aircraft
with flexibility of development, the aerodynamic characteristics should be such
that they would not limit the speed to less than the structural limitations.
The aluminum alloy structure which we favoured was good for speeds greater than
a Mach number of 2, and we therefore felt that our aerodyna- mics should be at
least as good as this.
To achieve this we had to go to the thinnest possible t/c ratio,
and started out with a 3 per cent t/c wing throughout the span, but aileron reversal
forced us to go to a thicker and stiffer wing section, and we compromised at
3.5 per cent at the wing root, and 3.8 per cent at the tip. The structural advantages
of the delta made the achievement of a thin wing section possible without a large
So for us, the tailless delta had distinct virtues, with the added
advantage that Avro Manchester had, by that time, done considerable flying with
the 707 delta research aircraft, prior to the design of the Vulcan tailless delta
bomber, and this information was, of course, available to us.
However, the delta, like everything else, also had its vices. For
instance, aeroelasties were obviously to play a very big part in our design,
due to the extremely thin wing and fin sections and, in calculating the aeroelastic
and flutter characteristics of a delta wing the standard semi-empirical methods
of analysis would have produced a prohibitively heavy structure if we had used
them indiscriminately. We had to examine all types of aeroelastic and flutter
problems from first principles, and we repeated these as the design progressed.
The establishment of the structural matrix was a very laborious process, most
of which had to be done on our digital computers.
Due to the short elevator arm we were in trouble with trim drag.
The high elevator angles required to trim at high altitude increase the elevator
drag considerably. We investigated means of reducing this and the most promising
appeared to be the introduction of negative wing camber. Camber has the effect
of building in some elevator angle without the excessive
control surface drag. The amount of camber chosen, which was 3/4 per
cent negative, was that which would give a good compromise between the
positive angles to trim at low altitude, and the negative angles required
at high altitude.
EDGE NOTCH AND EXTENSION
in the design stages modifications were made to the
original clean wing. These were the addition of leading
the semi-span notch with outer wing chord extension.
These modifications were made as a result of wind tunnel tests at Comell
Laboratories in Buffalo on a 3 per cent complete model, sting mounted.
The approximate Reynolds number used during the tests was between one
and two million. These tests showed that we were getting a pitch-up or
non-linearity in the Cm-a curve at moderate angles of attack. This phenomenon
is not peculiar to deltas, being common to all swept wing aircraft. In
flight it could cause a tightening in the turn.
Crudely, the condition appears to be caused by vortices which start
at the tip and move to the apex of the swept wing. Low pressure air is collected
from the fuselage and causes a breakaway outboard of the area covered by the
vortex, which is mainly at the trailing edge (Figs. 3&4). This causes the
FIGURE 3. Vortex pattx plain wing.FIGURE 4. Vortex p of wing with notch
and ext leading edge.
aerodynamic centre to shift forward, giving a "pitch-up" or
an abrupt change in moment curve.
While the pitch-up appeared on test to be of small magnitude, since
very moderate amounts of pitch-up can be embarrassing to the pilot, attempts
were made eliminate it.
We were aware of the work that had been done by N.A.C.A. and the
R.A.E. and the fact that a number of other aircraft which had exhibited this
tendency used either notches in the leading edge at about semi span, or extensions
of the wing leading edge outboard, in an attempt to prevent the flow separation.
The notch had been used, for instance, on the English Electric F-23, and the
leading edge extension had been installed on a Grumman F9F9, and a Chance Vought
aircraft. The notch has a somewhat similar effect to a fence causes the disturbing
vortices to move away from apex of the swept wing toward the notch, which is
at semi-span, and reduces the area of disturbed flow over the wing. The notch,
however, produces these effects by air flow rather than as a physical barrier.
It was our opinion that the effects of the notch are present over the whole speed
range, whereas a fence is usually only effective over smaller speed ranges, and
the notch was expected to increase the drag by a lesser amount than a fence.
We did find, however, in our tests that with the notch alone the
test results were not repeatable; in other words, the same results could not
be got in subsequent tests. When the leading edge extension was installed in
addition to the notch the results were far more repeatable. Eight different notches
and three extended leading edges were tried in various combinations. The depth
of the notch appeared to be the most critical parameter, and here again we had
to bear in mind that we could not have too deep a notch because of structural
Figure 5 shows the effect of the 5 per cent notch, 10 per cent extension
of the local chord on the outer wing, which was finally adopted, against the
unmodified 31/2 per cent wing at M=0.9, and at an elevator angle of - 20deg.