Technical papers
The Thrust Optimised Parabolic nozzle
Introduction
In the early 1950’s, rocketeers attempted to devise ‘the perfect nozzle’: one that would cause
the least thrust losses. They used the semi-algebraic Method of Characteristics (of supersonic
flow) to devise such a nozzle, for whatever nozzle expansion ratio ϵ was required.
Unfortunately, at large expansion ratios, this nozzle was far too long, resulting in excess
nozzle mass.
Rao in America and Shmyglevsky in Russia found a way to modify this method in order to
produce an optimum nozzle that was much shorter: it resembled a church bell and was hence
known as a ‘bell nozzle’.
For ease of use, Rao measured the lengths of his resulting nozzles as fractions of the length
of a standard 15 degree half-angle conical nozzle which has length:
(√∈−1) 𝑅𝑡
𝐿𝑁_𝑐𝑜𝑛𝑒 = equ. 1 where Rt is the radius of the throat.
tan(15)
So an ‘80% bell’ would have a length of 0.8 times this length.
Rao was clearly also schooled in traditional geometry. He was able to spot that the bell part of
his bell nozzles could be approximated very closely by a skewed parabola allowing us to
quickly sketch his nozzles with negligible loss of thrust performance.
These sketched approximations are known as Thrust Optimised Parabolic (TOP) nozzles,
and have found use on a variety of actual launch vehicles because they perform better when
over-expanded at ground-level altitude than the actual optimised bell nozzle (flow separation
from the TOP nozzle wall is delayed at high back-pressure).
Rao’s parabolas are known in Europe as quadratic Bézier curves, after French car bodywork
draughtsman Pierre Bézier. Bézier curves are now used extensively in computer graphics.
The shape of the bell nozzle changes only minutely with the propellants used (varying ratio of
specific heats γ) so one TOP nozzle methodology fits all propellants and is described below.
Author: Rick Newlands 1 updated: 18/04/17
Technical papers
The geometric approach
Construction
The ‘Rao nozzle’ starts with Rao’s preferred throat geometry, presumably a Rockedyne
(where he worked) optimisation.
Two circular arcs are drawn:
The first curve, of radius 1.5 Rt, is drawn from an angle of say, -135 degrees, to the throat at
-90 degrees (angles measured from the arc’s origin). Then the second curve of radius 0.382
Rt is drawn from this angle of -90 degrees to an angle of (𝜃𝑛 - 90) at inflection point N.
(Rt is the throat radius, Re is the exit radius.)
Then a skewed parabola is drawn from point N to nozzle exit point E, tangent to the throat
curve, and starting at an angle of 𝜃𝑛 and ending at an angle of 𝜃𝑒 .
The radius of the nozzle exit: 𝑅𝑒 = √𝜖 𝑅𝑡 equ. 2
(√∈−1) 𝑅𝑡
and nozzle length 𝐿𝑁 = 0.8 ( ) equ. 3 for an 80% bell from equ. 1
tan(15)
Angles 𝜃𝑛 and 𝜃𝑒 were pre-calculated by Roe to match his bell nozzle, and presented as
graphical data from which the following chart is reproduced for various percent lengths (from
Ref. 1):
Author: Rick Newlands 2 updated: 18/04/17
Technical papers
(The data for expansion ratios greater than 50 is extrapolated).
A parabola is then sketched-out using an ancient geometrical method for drawing a parabola
detailed in Ref. 2.
Straight lines are drawn at angles 𝜃𝑛 from point N, and 𝜃𝑒 back from point E, terminating
where these lines cross at point Q.
Next, both of these lines are divided into an equal number of divisions (four in this example)
These are labelled a,b,c and e,f,g.
Author: Rick Newlands 3 updated: 18/04/17
Technical papers
A straight line is then drawn from point a to point e, then from b to f, and c to g. These form a
mesh, the edge of which gives the parabola outline. The parabola is also tangent to the lines
QN and QE.
Using many more divisions, on a CAD package for example, gives a sharper contour.
Removing most of the mesh gives a series of straight-line segments: joining the midpoint of
each segment with a smooth curve such as a CAD spline gives the nozzle contour.
Alternately, interpolating along the construction lines gives the same points. In our example
with four divisions and (4-1) construction lines, the parabola is defined at 1/4 the distance
along line a-e, 2/4 along line b-f, and 3/4 along c-f.
Efficiency
At a length ratio of 85% bell, a nozzle efficiency of 99% is reached, and only 0.2% of
additional performance can be gained by increasing the length ratio to 100%. For this reason,
85% is often taken as upper bound. At length ratios below 70%, nozzle efficiency suffers. For
these reasons, the 80% bell parabola is often chosen.
Example
An 80% bell nozzle with an area ratio of 70 (a typical upper stage nozzle) has angles 𝜃𝑛 = 33º
and 𝜃𝑒 = 7º from the above chart. Drawn on a CAD package it looks like this (mesh removed
for clarity):
Author: Rick Newlands 4 updated: 18/04/17
Technical papers
The mathematical approach
The throat
The equations of the above circular arcs defining the throat are defined trigonometrically,
defining the origin of the coordinates as the centre of the narrowest part of the throat:
For the entrant section:
𝑥 = 1.5 𝑅𝑡 cos 𝜃
𝑦 = 1.5 𝑅𝑡 sin 𝜃 + 1.5 𝑅𝑡 + 𝑅𝑡 equ.s 4
where: −135 ≤ 𝜃 ≤ −90
(The initial angle isn’t defined and is up to the
combustion chamber designer, -135 degrees is
typical.)
For the exit section:
𝑥 = 0.382 𝑅𝑡 cos 𝜃
𝑦 = 0.382 𝑅𝑡 sin 𝜃 + 0.382 𝑅𝑡 + 𝑅𝑡 equ.s 5
where: −90 ≤ 𝜃 ≤ (𝜃𝑛 − 90)
The bell
The bell is a quadratic Bézier curve, which has equations (see Wikipedia):
𝑥(𝑡) = (1 − 𝑡)2 𝑁𝑥 + 2(1 − 𝑡)𝑡 𝑄𝑥 + 𝑡 2 𝐸𝑥 0≤𝑡≤1
𝑦(𝑡) = (1 − 𝑡)2 𝑁𝑦 + 2(1 − 𝑡)𝑡 𝑄𝑦 + 𝑡 2 𝐸𝑦 0≤𝑡≤1 equ.s 6
Selecting equally spaced divisions between 0 and 1 produces the points described earlier in
the graphical method, for example 0.25, 0.5, and 0.75.
Equations 6 are defined by points N, Q, and E (see the graphical method earlier for the
locations of these points).
Point N is defined by equations 5 setting the angle to (θn – 90).
Coordinate Ex is defined by equation 3, and coordinate Ey is defined by equation 2.
Point Q is the intersection of the lines: ⃗⃗⃗⃗⃗⃗
𝑁𝑄 = 𝑚1 𝑥 + 𝐶1 and: ⃗⃗⃗⃗⃗
𝑄𝐸 = 𝑚2 𝑥 + 𝐶2 equ.s 7
where: gradient 𝑚1 = tan(𝜃𝑛 ) , gradient 𝑚2 = tan(𝜃𝑒 ) equ.s 8
and: intercept 𝐶1 = 𝑁𝑦 − 𝑚1 𝑁𝑥 , intercept 𝐶2 = 𝐸𝑦 − 𝑚2 𝐸𝑥 equ.s 9
The intersection of these two lines (at point Q) is given by:
(𝐶2 −𝐶1 ) (𝑚1 𝐶2 −𝑚2 𝐶1 )
𝑄𝑥 = , 𝑄𝑦 = equ.s 10
(𝑚1 −𝑚2 ) (𝑚1 −𝑚2 )
Author: Rick Newlands 5 updated: 18/04/17
Technical papers
References
1. “Liquid Rocket Engine Nozzles”, NASA SP-8120, July 1976
2. “Exhaust Nozzle Contour for Optimum Thrust,” G.V.R. Rao, Jet Propulsion 28, 377
(1958)
3. www.wikihow.com/Draw-a-Parabolic-Curve-(a-Curve-with-Straight-Lines)
4. Wikipedia: quadratic Bezier curve
5. www.waters.to/blog/rocket-nozzles-part-1-the-math/
6. Wikipedia: line-line intersection
7. www.wikihow.com/Find-the-Equation-of-a-Line
Author: Rick Newlands 6 updated: 18/04/17
