A model and sounding rocket simulation tool with

Mathematica R

Tiago Miguel Oliveira Pinto

[email protected]

Instituto Superior Técnico, Lisboa, Portugal

November 2015

Abstract
This work aims to simulate trajectories from model to sounding rockets taking into account the wind
and the atmospheric boundary layer. The developed tool is also capable to forecast the most probable
landing region and perform rocket design optimization using Monte Carlo methods. A study of the
atmospheric boundary layer is carried out and are considered three wind profiles that take into account
the atmospheric conditions and the terrain type in the launch site. Presenting the developed tool’s
flexibility, a Monte Carlo simulation is performed in order to deduce the dispersion of the landing sites
due to the wind parameters’ uncertainties. Owing to the Mathematica R
’s potentialities, the developed
tool allows to define any number of stages, with external or internal boosters in the first stage, and to
easily modify the required models to compute the trajectory or to implement new ones in the database.
A test simulation using a two-stage model rocket was performed and the results showed a large
dependence of the trajectory on the wind profile. Considering the defined uncertainties from the wind
speed and direction, as well from the surface type, they presented a strong influence on the dispersion
of the landing sites.

Keywords: Trajectory Simulation, Model Rocketry, Atmospheric Boundary Layer, Monte Carlo.

1. Introduction missiles); flexibility to change easily a certain input

This paper focuses on passively controlled rockets, in a given model and figure out its influence in the
whose trajectories are influenced by the wind that trajectory; freedom to edit or implement new func-
changes according the land surface [1] and time of tions and values into the simulator’s database; and
the day [2]. Thus, if the wind profile is foreseen un- process the resulting data in order to obtain fur-
der given conditions, the simulations’ errors can be ther conclusions. Besides these, are also embedded
minimized and simulations in different conditions other basic features such as multi-staging, boosters,
can provide to users important information to help launch direction and atmospheric, wind and drag
selecting the site and day of the launch to achieve coefficient models. The simulator does not include
successful flights. Sounding rockets consist of one a modeling interface, however, it allows to define
or more stages (propelled by solid or liquid fuel) a versatile rocket model that is functional to ev-
carrying a scientific payload in order to study the ery type of rockets. This is accomplished by us-
atmosphere and carry out microgravity research be- ing a default scheme of lists structured according
tween 40 km to 2000 km height [3]. From 40 km to the number of stages and boosters. Therefore, this
200 km, these rockets are specially interesting be- structure of lists, allied to the simulator code, allows
cause neither balloons nor low-earth satellites can the rocket model to have any number of stages, al-
reach these altitudes. Hence, allowing the devel- though this scheme of lists must always be respected
oped tool to simulate trajectories of fin-stabilized so that the program runs properly. As this tool is
sounding rockets up to these heights, brings addi- developed under a modular programming philoso-
tional useful applications to our project. phy, the necessary inputs to the trajectory function
Although the simulations from the developed tool are independent from each other. Hence, all of them
may cost a larger processing time due to the inter- can be easily modified without worrying about the
preted language of Mathematica R
[4], it presents other models. This allows to run several simulations
valuable features as: simulation of trajectories from under very different conditions and compare their
micro to sounding rockets (and even any type of impact in the trajectory. Two databases are already

1
implemented as default, one for the rocket’s mo- w
tors and another for the atmospheric models. The
v
databases are defined in a way that simplifies the vt α
access to the data and their properties. Besides, an-
other models can be joined in the databases to ex-
pand the options to use during the simulations. The Figure 1: Air speed components.
developed simulator also allows the user to compute
trajectories in loop, changing a determined group
2.1. Wind
of parameters. This procedure employs the Monte
The lowest layer of the troposphere is directly influ-
Carlo methods enabling the formulation of further
enced by the ground surface characteristics. There-
studies. These analysis may consist of calculations
fore, a wind speed profile is worth to be imple-
to determine a probable landing region giving cer-
mented since its effect will be noticed, mainly, in the
tain parameters’ uncertainties or a rocket design op-
trajectory of small and medium model rockets. This
timization evaluating the combination of possible
layer is known as the Atmospheric Boundary Layer
values for the components that achieve the desired
(ABL) and its thickness may change from about a
goal in the trajectory.
hundred meters to a few kilometers varying in time
2. Theoretical background and with geographic region [6]. The bottom 10%
A rocket is stable if its Center of Pressure (CP) is of the ABL is named the Surface Layer (SL) with,
behind the Center of Mass (CM) (seeing from the in average, 100 m height [7] which makes the ABL
nose to the tail of the rocket) [5]. The CP is the about 1 km thick.
point where acts the resultant aerodynamic force Above the ABL stays the free atmosphere, a
produced by the air pressure and moves toward the more stable region where the frictional influences
nose with increasing angle of attack (α). This force, of the surface can be ignored and the wind is nearly
when there is an angle of attack, may be decom- geostrophic [6]. This wind, moving parallel to the
posed as an axial and normal force where the latter isobars, results from the balance between the Cori-
creates a moment about the CM of the rocket. In olis and the pressure gradient forces [8]. Consider-
stable conditions, this moment produces a damped ing only geostrophic balance, which is the case of
oscillating movement between the air flow direction a barotropic atmosphere, the wind is constant with
until the rocket flies without angle of attack and the height and its intensity can be assumed as equal to
normal force vanishes [5]. The arm of the moment the one at the top of the ABL.
is the length between the CM and the CP, called Regarding atmospheric stability, the ABL is clas-
static margin. Decreasing the static margin until sified in three different classes: neutral, stable and
the CP overtakes the CM, makes the rocket unsta- unstable. A neutral atmosphere implies an adia-
ble because the resulting moment reverses its direc- batic lapse rate and no convection which is the case
tion. In this situation, if a perturbation deviates of a partially or highly cloudy atmosphere that may
the rocket from the air flow direction, the aerody- reduce the insolation at the surface [6]. Stable con-
namic moment will amplify the disturbance causing ditions occur mostly at night but can also appear
the rocket to spin and crash. With a larger static when the ground surface is colder than the sur-
margin, however, the rocket may become overstable rounding air. An unstable atmosphere is formed in
and reaches a lower apogee because it turns sooner clear weather during the day when there is high ra-
to the wind [5]. This effect, called weathercocking, diation from the sun causing ascending heat trans-
is due to the contributions of the rocket and wind’s fer. These conditions influence the wind profile and
velocities (v and w, respectively) that define the must be taken into account due to convective ef-
direction and intensity of the true air speed as fects.
From measurements of the wind speed, the litera-
ture [9] presents the following wind profiles modeled
vt = v − w . (1)
considering the entire ABL and its stability:
"  
u∗0
 
The angle of attack is maximum when the rocket z 4.7z z
u(z) = ln + 1− +
leaves the guiding rods and is found by (see Fig.(1)) κ z0 L 2h
 #
v · vt z z z
α = arccos . (2) + − , (3)
kvkkvt k LM BL h 2LM BL

for stable conditions,

Note that v may describe other angles in respect   
to the ground, so that Figure 1 no longer depicts a 
u∗

right triangle. u(z) = κ ln zz0 +
0 z
LM BL − z
h
z
2LM BL , (4)

2
for neutral conditions, and Obukhov length interval (m) Stability class
"   10 ≤ L ≤ 50 Very stable
u∗0
 
z z z
u(z) = ln −ψ + − 50 ≤ L ≤ 200 Stable
κ z0 L LM BL
 # 200 ≤ L ≤ 500 Near stable
z z |L| ≥ 500 Neutral
− , (5)
h 2LM BL −500 ≤ L ≤ −200 Near unstable
for unstable conditions, where h is the ABL depth, −200 ≤ L ≤ −100 Unstable
z0 is the surface roughness length, κ is the von −100 ≤ L ≤ −50 Very unstable
Kármán constant, LM BL is the length scale in the
middle of the ABL for its respective conditions, u∗0 Table 2: Atmospheric stability classes according to
is the local surface friction velocity near the ground, intervals of Obukhov length, L [9].
L is the Obukhov length and ψ is a stability correc-
tion for the SL.
The surface type influences z0 which is quantified ao is the linear acceleration of this frame given as
and can be obtained from Table 1. Ω × (Ω × Ro ), where Ro is the position of the lo-
cal frame’s origin relative to the Earth’s center. The
Surface type z0 (cm) rocket’s velocity seen from an observer at the launch
2

Sand 0.01...0.1 site is v = ∂∂tr , where the partial derivative stands

for the time derivative of r with reference to the
Cut grass (∼ 0.01 m) 0.1...1
local frame. The term dΩ dt × r is the angular accel-
Small grass, steppe 1...4 eration of the moving frame, 2Ω × ∂r ∂t is the Coriolis
Uncultivated land 2...3 acceleration and Ω × (Ω × r) is the centripetal ac-
High grass 4...10 celeration. In order to get the trajectory seen by
Coniferous forest 90...100 an observer at the launch site, we must write these
vectors in the local frame coordinates.
Uptown, suburbs 20...40
After using the momentum conservation (tak-
Downtown 35...45 ing into account the loss of mass and the external
Big towns 60...80 forces) and applying (6), the trajectory equation
results in
Table 1: Surface roughness values for different land
surfaces [1]. ∂2r dΩ ∂r
m = −mao − m × r − 2mΩ × −
∂t2 dt ∂t
The Obukhov length (L) represents the thickness −mΩ × (Ω × r) + T + D + W , (7)
near the surface in which the shear stress dominates
over the buoyancy effects in generating turbulence. where m is the rocket’s mass, T is the thrust, D
This parameter is hard to estimate since we get con- is the drag and W is the weight. Solving (7), the
siderable errors even for measurements taken hun- position of the rocket seen from the launch site can
dreds of meter above the surface. Nevertheless, if be obtained.
the atmospheric stability is deduced qualitatively The thrust is always aligned with the rocket’s lon-
from the weather conditions and time of day (as gitudinal axis as the rocket is not steerable. Since
described before in the section), we get to know the rocket may fly with angle of attack until it sta-
which wind profile to use and a value for L can be bilizes and the thrust becomes collinear with the
estimated from Table 2. true speed vector, the effective propulsive force may
be lower in this direction. Therefore, the adopted
2.2. Trajectory model to the thrust force is
The launch site is a non-inertial topocentric-horizon
coordinate frame with the x axis pointing south- vt
T = Tp cos α (8)
ward and the y eastward. Thus, as this is a right- kvt k
handed coordinate frame, the z axis points up-
where Tp is the propulsion from the motors’ thrust
ward collinear with the Earth’s radius direction
profile. The aerodynamic force is assumed to be
(or zenith). The total acceleration, in the Earth-
merely the drag since the rocket only flies with an-
centered inertial coordinate frame, is given by [10]
gle of attack during a short interval of time after the
2
a = ao + ∂∂t2r + dΩ × r + 2Ω × ∂r
+ Ω × (Ω × r) , (6) launch (or after any perturbation) and it describes
dt ∂t
a low amplitude oscillation. The lift that is pro-
where Ω is the Earth’s angular velocity, r is the po- duced during these moments contributes essentially
sition of the rocket relative to the local frame and to stabilize the rocket, its average is considered to

3
be zero, and its effect can be implemented as de- rocket is rotating [11]. During the rotation, the an-
scribed in Section 2.3. Thus, the aerodynamic force gle of attack of each part of the rocket changes due
comes to the tangential velocity of this motion. Like the
1 stabilizing moment, only the normal force of this
D = ρkvt k SCD vt , (9) additional resistance contributes to dampen the ro-
2
tation. Thus, since the aerodynamic damping mo-
where ρ is the air density, S is the reference area ment (Mda ) is determined by summing all the ele-
(usually chosen as the maximum sectional area of mental moments along the rocket [11], we get [12]
the rocket or, after the apogee, the reference area of n
the recovery device) and CD is the drag coefficient. 1 X
Mda = ρV S α̇ CNαi (xcpi − xcm )2 , (14)
The weight points to the Earth’s center and is found 2 i=1
by
where n is the total number of components of the
µ⊕ m
W=− z , (10) rocket that contribute to this moment.
(R0 + z)2
The other contribution to the damping moment
where µ⊕ = 3.986 × 10 km s 5 3 −2
is the Earth’s comes from the Coriolis acceleration due to the
gravitational parameter and z is the unit vector of change of the gas flow through the nozzle, also
the vertical coordinate. called jet damping, given by [11]
During the launch, when the rocket is constrained Mdj = α̇ṁ(xnozzle − xcm )2 , (15)
by the guiding rods, the drag and weight normal
to the launch direction are counterbalanced by the where xnozzle is the nozzle distance from the refer-
rod’s reaction and only the tangential forces to the ence point and ṁ is the rocket’s burning rate.
rods contribute to the launch. Thus, the last three The governing equation for the motion of the an-
terms in (7) become gle of attack, assuming the rotation is two dimen-
sional (in the plane formed by the rocket’s velocity
T + D + W = − 12 ρkvk SCD v + (Tp + W sin El)el , (11) and wind vectors), is found by the angular momen-
tum knowing that the stabilizing and damping mo-
where el is the unit vector of the rods’ direction de-
ments counteract the rotation. Thus, we get
termined by the launch elevation (El) and azimuth
(Az) as −Ms − Mda − Mdj = I˙α̇ + I α̈ , (16)
el = (cos El cos Az, cos El sin Az, sin El) . (12) where I is the transversal moment of inertia relative
to the CM. Applying (13), (14) and (15) in (16), we
2.3. Dynamic stability
get a second-order differential equation that repre-
After leaving the platform, and when a distur-
sents a damped harmonic oscillator (considering the
bance occurs, the rocket acquires an angle of attack
rocket is stable) giving the angle of attack.
that will gradually decrease and force it to oscil-
late about the true air speed direction. To generate 3. Rocket trajectory simulator
this damped oscillating movement, we conclude the The simulator’s main function, which integrates the
rocket is subjected to two moments [11]: a stabiliz- trajectory equations, receives all the required in-
ing (or restoring) moment and a damping moment. puts and calls the function that computes every pa-
The stabilizing moment comes from the normal rameter of the rocket over time. These properties
aerodynamic force acting on the CP which causes are determined by combining the data of the sev-
the rocket to rotate about the CM since these points eral stages regarding the developed structure that
must distance from each other by the static margin. gathers the data of the rocket and considering the
Therefore, assuming small angles of attack, the sta- instants of the ignitions and ejections. The most
bilizing moment is found by relevant inputs are the rocket, density, wind and
1 CD models that, since they are external parame-
Ms = ρV 2 SCNα α(xcp − xcm ) , (13) ters, the user is allowed to define or change them
2
according to his needs. The outputs from the sim-
where xcm and xcp are, respectively, the rocket’s ulation are three functions giving the position (one
CM and CP distances from the reference point (of- for each cartesian coordinate) plus a function giving
ten considered as the nose tip, which we also con- the angle of attack. These are computed in three
sider throughout this work), V is the flow veloc- steps: one for the constrained trajectory due to the
ity and CNα is the rocket’s normal force coefficient guiding rod, another for the climbing phase and the
derivative . last one for the recovery trajectory. The angle of
One source of the damping moment results from attack starts to be computed only in the second
the aerodynamic resistance of the air while the phase, since it is when the rocket suffers the first

4
perturbation from the wind. After this process, the the stages and the recovery system plus the motor
three parts of the trajectory are merged together cluster in the first stage (see Figure 3). The stages
and, having these time dependent functions, it is list contains as many lists as the number of stages.
possible to determine its derivatives and get other Each one of these lists holds other lists respecting
flight data such as total velocity, total acceleration, the stages components plus a list to indicate the
Mach and flight path angle. time intervals between the discharges. All of this
data (dimensions, mass, CM, inertia, CP and CNα )
3.1. Rocket assembly can be defined manually by the user or, in some
The proposed structure for a general multi-stage cases, estimated by default functions implemented
rocket developed in the present thesis is shown in in the simulator.
Figure 2. Although the figure represents a three-
3.2. Developed models
The atmospheric model stored in the database is
the CIRA-86 which gives pressure and temperature
data up to 120 km (the latter is also dependent on
the latitude and the month of the year). This or
other atmospheric models can take into account
3rd Stage 2nd Stage 1st Stage the conditions at the launch site using developed
functions that match the measured pressure and
Figure 2: Multi-stage rocket design. temperature at the corresponding altitude with the
values from the model at higher altitudes (see Fig-
stage rocket, this structure can be extended to any ure 4a). One of this functions also determines the
number of stages and was developed with the goal of density from the equation of state.
modeling several types of rockets, specially the most The developed wind model uses the ABL pro-
common. For each stage the fundamental compo- files (taking into account the atmospheric stability)
nents are: a connector (or nose, in the case of the that approximate the wind data also provided in the
last stage), a body tube and fins. Internally, it is CIRA-86 model at higher altitudes. If desired, the
only considered the motors since data from addi- wind may be constant above the ABL. This model
tional components can be joined with the body tube requires local wind speed, its respective measure-
properties. In the first stage the rocket can also be ment altitude, the terrain’s specification and the
supplied with external boosters in order to achieve downwind direction. Wind gusts can also be imple-
parallel staging. mented providing their intensity and the instants of
the respective perturbations as another input of the
Rocket
Assembly List trajectory function. These disturbances are simu-
Boosters List
lated considering they change instantly the angle of
Number
Position attack to another value determined by the rocket
Thrust Data List and wind gust’s velocities at the perturbation’s in-
Structure Data List
Delay Booster
stant. After that moment, the simulator computes
Stages List new oscillations for the angle of attack (see example
Stage 1 List
in Figure 4b).
Stage 2 List

α (º)
Stage i List
Temperature (K)
Motor List 10
300 CIRA-86
Body List CIRA-86 with local conditions 5
280
Fins List
260 t (s)
Connector/Nose List 1 2 3 4 5 6 7
240 -5
Delays List
Miscellany List 220 -10
Altitude (km)
Motor Cluster List 5 10 15 20 25

(a) Temperature profiles (b) Angle of attack until

Number Motors
Motor List
for the CIRA-86 model. the apogee with a wind
Recovery List
Reference Area gust perturbation.
Drag Coefficient
Figure 4: Temperature and angle of attack models.
Figure 3: Rocket assembly list structure.
The simulator also provides a drag coefficient
The defined structure of lists enables to access model that takes the structure of lists respecting
easily the rocket’s data and to be quickly processed the rocket’s data and determines the drag coeffi-
by the simulator. This structure has three major cients for each one of the external components de-
lists concerning, respectively, the external boosters, pending on Mach and Rex . Then it gathers all the

5
coefficients to compute the rocket’s CD in respect To check the drag effect, we can make a terminal
to its reference area. Using this drag coefficient velocity test for a given rocket in free fall after the
model, we are considering that all the components apogee. From an equilibrium between the gravita-
contribute to the skin friction drag but the pressure tional force and the drag force, we can find the ter-
drag comes only from the nose, connectors and fins minal velocity and conclude that if the rocket loses
(due to the normal surface facing the flow). In this mass, it must reach a slower velocity. In Fig.(7)
model the rocket’s nose is the only wave drag source it is represented this quasi-static process computed
whose effect is implemented by a fitted function in by the simulator. As we can see, the rocket tends
the transonic regime under M = 1 and by an em- asymptotically to a new terminal velocity, as ex-
pirical function above this Mach. The final drag pected.
coefficient function for the conical nose, including
Velocity (m/s)
the wave drag, is represented in Fig.(5).
-10
CD
-20
0.38
0.37 -30

0.36 -40
0.35
-50
0.34
t (s)
0.33 15 20 25 30 35 40

0.32

Mach
Figure 7: Descent velocity from a vertical launch in
0.0 0.5 1.0 1.5 2.0 2.5 3.0
which the rocket loses mass at 25 s.
Figure 5: Conical nose drag coefficient with Mach.
4. Test results
3.3. Validation of the tool In Fig.(8) are presented the results from the as-
The developed tool requires validation in order to sembling of a two-stage model rocket. Since model
assure the simulator meets the specifications and rockets reach lower velocities and apogees, they are
correctly computes the expected output. This may Thrust (N) Mass (g)

be performed by comparing our results with a 30

110
25 100
known analytic solution from a specific launch. 20 90

One of developed tests consists in determine the 15 80

70
10
error presented between the altitude from a verti- 5
60
50
cal sounding rocket launch and the tool’s output 1 2 3 4 5
t (s) t (s)
1 2 3 4 5
under the same conditions. The analytical solution
(a) Thrust. (b) Mass.
requires that these conditions correspond to a case
without atmosphere and the effects of the Earth’s CM (cm) CP (cm)
40 40
rotation, constant thrust and constant acceleration
of gravity. The obtained relative error over time 35 35

(until the burnout) is represented in Fig.(6). It can

30 30

Relative Error (%)

25 25
t (s) t (s)
0 1 2 3 4 5 0 1 2 3 4 5
3. × 10-6
(c) CM relative to the nose (d) CP relative to the nose
2. × 10-6 tip. tip.
1. × 10-6
Mom. Inertia (g.m3 )
Ref. Area (cm2 )

t (s) 12
5 10 15 20 25 30 4.0

10
3.5
Figure 6: Altitude’s relative error between analyti- 8

cal and numerical solution until the burnout. 3.0

6
t (s)
1 2 3 4 5
t (s)
1 2 3 4 5

be observed that the error presents values in the (e) Transversal moment of
(f) Reference area.
order of magnitude of 10−6 %. Although this error inertia relative to the CM.
gives reasonable results under the mentioned con-
ditions, more tests should be performed in order Figure 8: Model rocket properties’ functions.
to increase the confidence in the solutions from the
developed simulator. suitable to observe the effects of the wind and the

6
ABL on the trajectory simulation. Due to the first descends toward the downwind direction. Hence, it
discontinuity shown in Fig.(8b), Fig.(8b), Fig.(8c), surpasses the launch site and lands 380 m north and
Fig.(8d) and Fig.(8e), we can see that the first 382 m east away from that place (coordinates taken
stage’s structure (not considering the connector) is from the simulator). The influence of the ABL is
discarded at the first burnout. Also in these fig- also observable during the recovery phase in which
ures, during the coasting phase (lasting 1 s), we see the trajectory describes a smooth curvature since
the connector’s discharge represented by the second the wind speed is decreasing as the rocket falls. If
discontinuity. Except from these sharp changes, the wind was constant with altitude, the recovery
the rocket maintains its properties during coasting, trajectory would describe a linear path.
leading the mentioned figures to match with Fig- In Fig.(10) the flight data from this launch is rep-
ure 8a when there is no thrust between the burn-
Altitude (m) Velocity (m/s)
ings. The reference area is defined as the largest 120
500
cross section of the rocket (without considering 400
100

80
boosters, for other cases) at each instant. This def- 300 60

inition agrees with Figure 8f, where it shows the 200 40

100 20
rocket’s reference area changing nearly 2 s after the t (s) t (s)
20 40 60 80 100 120 2 4 6 8 10 12 14
launch, which is the instant when the connector is
discarded and the rocket loses its largest sectional (a) Profile of altitude vs (b) Profile of vertical veloc-
time. ity vs time.
area. Comparing Figures 8c and 8d, we conclude
this rocket is always stable since the CM, instead
Accel.(/g)
Mach
of the CP, is closer to the nose all the time. Hence, 0.4
20
the rocket is well assembled and able to be used in
10 0.3
a trajectory simulation. t (s)
2 4 6 8 10 12 0.2

4.1. Trajectory simulation -10

0.1
-20
From the previous rocket, we get the trajectory de- 2 4 6 8 10 12 14
t (s)

picted in Fig.(9). This rocket is launched vertically (c) Longitudinal accelera-

(in order to be easily observed the deviation from tion vs time. (d) Mach vs time.
the launch direction) under a neutral atmosphere
Flight Path (º)
with a constant wind speed above the ABL, blow- α (º) 100
15
ing northeastward in a high grass terrain. 50
10

t (s)
5 20 40 60 80 100 120

t (s) -50
Up (m) 0.5 1.0 1.5 2.0 2.5 3.0

-5
500

400 (f) Flight path angle vs

300 (e) Angle of attack vs time. time until the landing.
200

100
CD
0.9
-400 -200 0 200
Ground Track (m)
0.8

(b) Trajectory profile. 0.7

0.6
(a) 3D trajectory.
0.5

Figure 9: Model rocket trajectory with a vertical 0 2 4 6 8

t (s)

launch and a neutral ABL; Parachute ejection at (g) Drag coefficient vs time
the apogee. until the apogee.

In this trajectory the weathercock effect is ob- Figure 10: Vertical model rocket launch flight data.
servable. Since the wind blows northeastward, the
rocket climbs in the opposite direction (southwest- resented. From Fig.(10a) we deduce that the rocket
ward) to follow the true air speed. Although this reaches the apogee at about 560 m of altitude and
effect increases smoothly with altitude due to the lands 120 s after the lift off. Observing when the
ABL, we can see a slightly discontinuity in the up- vertical velocity turns negative, in Fig.(10b), we
ward flight path around 200 m of altitude. This conclude that the apogee is reached at 9.5 s, ap-
happens because the weathercock is enhanced dur- proximately. As already mentioned, we also see
ing the coasting time, between the first burnout and that the rocket decreases velocity due to the coast-
the second ignition, as the rocket decreases its ve- ing between stages. This is also observable from the
locity. During the recovery phase, the rocket slowly Mach number, in Fig.(10d), which remains subsonic

7
throughout the entire flight. bution for each input parameter from which a set of
During the burnings, in Fig.(10c), the accelera- data is randomly generated. Then, all the possible
tion describes a profile similar to the thrust (see combinations between the created data correspond
Fig.(8a)). However, it is decreasing over time as to the inputs given in each trajectory simulation,
the rocket gains velocity, which increases the drag which highly increases the number of simulations
opposing the thrust force. In contrast, when the if we are studying many parameters with a large
rocket is not burning, the acceleration is negative sample. Therefore, since we do not own such great
and increases over time (decreases in absolute value) computational resources to minimize the simulation
because the velocity is decreasing. The peak accel- time, a better approach to perform what if simula-
eration at 9.5 s results from the parachute ejection. tions in which it is analyzed the outcome from the
The angle of attack, in Fig.(10e), shows the uncertainties of only two or three parameters with
rocket rapidly damps its oscillations and follows the smaller samples.
true air speed during the majority of the ascending In the following simulations we use a single-stage
flight. During the first instants the angle of attack high power rocket in order to present results from
is not computed because its model receives the im- a more powerful rocket.
pulse to start the oscillations only when the rocket
leaves the guiding rod. 5.1. Landing site uncertainties
From Fig.(10f), we can see the flight path angle After studying the uncertainties presented in the
as 90◦ at the ignition since the launch is vertical. landing site due to the parameters required to de-
This angle decreases until the apogee due to the termine the atmospheric and CD models, we con-
weathercock effect that causes a curvature in the cluded that only the wind speed and direction mea-
trajectory. Reaching the apogee, the flight path sured at the launch site and the surface roughness
angle becomes negative because the rocket starts to length (terrain type) presented a large uncertainty
descend. There is a peak value of about −90◦ since in the results under the defined conditions. There-
the rocket reverses its direction and stabilizes in a fore, these are the three parameters to taking into
30◦ slope, approximately, toward the ground due account in the Monte Carlo simulation. The wind
to the drag force on the parachute. As the rocket speed and direction data are randomly generated
descends, the flight path decreases, which means from a normal distribution, creating a sample of two
the velocity is becoming steeper. This fact agrees hundred points for each parameter. On the other
with the curvature described by the trajectory due hand, only the worst case scenarios for the rough-
to the ABL, observed in Fig.(9). ness length are considered (for a high grass terrain).
The drag coefficient, in Fig.(10g), is dependent Also, in order to reduce the simulation time, and
on the Reynolds and the Mach numbers. There- since it is not expected to impact the results much,
fore, when the velocity is increasing, the drag co- a mean CD is determined from a single simulation
efficient decreases its value due to the skin friction under the same conditions. The resulting landing
drag. Inversely, as the velocity decreases, the drag sites dispersion is represented in Fig.(11). Since
coefficient starts to increase. When the first stage
is discarded, the CD decreases instantaneously be-
cause the rocket loses a great source of drag. On
the other hand, when the connector is discarded,
although it is lost another source of drag, the CD
increases significantly. This happens because this is
the instant when the reference area becomes smaller
and the CD has to change in order to be given in
respect to another section.
5. Stochastic methods
Many of the parameters required to determine the
trajectory cannot be known (or well defined, at
least) and the measured data may vary randomly, Figure 11: Landing positions and respective confi-
presenting some uncertainties. Therefore, in order dence ellipses from the Monte Carlo simulation.
to analyze the outcome of several uncertain scenar-
ios, it lead us to perform simulations that rely on re- the rocket follows the wind speed direction, it is
peated random sampling and statistical analysis to expected the uncertainties from the local wind ve-
compute the results, known as Monte Carlo simula- locity and the surface roughness length (both only
tions [14]. To get the outcome due to the uncertain- influencing the wind intensity) change the landing
ties, we must identify a suitable probability distri- site along this direction, describing a straight line.

8
Analogously, the changes in the wind direction, due 5.2. Rocket design optimization
to its uncertainties, make the landing points to de- Taking the previous simulation from a different ap-
scribe an arc of a circumference that gets larger as proach, we get another tool that allows us to per-
the distance to the launch site increases. form optimization. Thus, after defining an opti-
The dispersion of the landing points, in Fig.(11), mization criteria (maximization or minimization of
clearly shows the influence of the normal distribu- an output) given an universe of input variables, we
tions used to generate the local wind direction and can deduce the best alternative that matches our
speed applied in each iteration. Since a mean of goal.
45◦ was defined for the wind direction’s normal dis- Changing the rocket parameters in each Monte
tribution, the orange dots are heavily concentrated Carlo iteration (opposing to the previous simula-
along the northeast direction. In the same way, the tion, where the rocket properties are kept constant),
rocket lands less often apart from the northeast di- we can optimize the rocket characteristics. For ex-
rection as there are fewer inputs that deviate more ample, we are aiming to determine the best body
from the mean. Also due to this principle, the wind tube’s length and diameter combination so that the
speed uncertainties contribute to the lower density rocket reaches the highest apogee keeping constant
of points near the closest and furthest distances the other properties. To this example a windless
from the launch site. This pattern can also be seen atmosphere is defined and three options for each
in Fig.(12) where the probability density function tube’s dimension, which are selected randomly from
an uniform distribution. The minimum and maxi-
mum length limits (constraints) are 1 m and 2.5 m,
respectively, and for the diameter are 7.5 cm and
13 cm, respectively.
0.06 In Fig.(13) the altitude over time is depicted (un-
til the highest apogee’s instant) of these nine con-
0.04
figurations. . As shown, the configuration with the

Altitude (km) l=1.686m d=10.1cm

0.02
l=1.686m d=8.4cm
5
l=1.686m d=12.2cm
0 4 l=1.215m d=10.1cm

3 l=1.215m d=8.4cm
l=1.215m d=12.2cm
2
l=2.266m d=10.1cm
1 l=2.266m d=8.4cm
t (s) l=2.266m d=12.2cm
Figure 12: Probability density function of the land- 5 10 15 20 25
ing coordinates resulting from the Monte Carlo sim-
ulation. Figure 13: Altitude over time from nine launches
with different rocket’s length and diameter config-
of the landing sites is represented. Furthermore, urations.
this figure enables us to distinguish better the most
probable landing area since Fig.(11) gets saturated smaller dimensions (1.215 m and 8.4 cm) reaches the
in this region due to the great amount of trajecto- highest apogee, as opposed to the rocket with the
ries simulated. largest dimensions that reach the lowest apogee.
The confidence ellipses in Fig.(11) are rather ec- These facts are expected due to the major influ-
centric. Their major axes (aligned northeastward) ence of the rocket’s dimensions on the drag coeffi-
measure 10.4 km, for a confidence of 95%, and cient and the reference area. The longer and larger
9.1 km, for 90% of confidence. As the furthest land- the rocket is, the greater is the drag that decreases
ing site, collinear with the major axes, is 14.6 km the velocity and forces the rocket to reach sooner
away from the launch location, the major axis from a lower apogee. From this result, we may say this
the lowest confidence level represents almost two method works properly but a more extensive opti-
thirds of the furthest distance. Thus, we conclude mization should be performed in the future since
the measured wind speed, allied with the surface the geometric parameters of the rocket also influ-
roughness length range, brings a strong uncertainty ence the mass of the components.
to the landing site estimation under the specified Even though this is a simple and expected exam-
conditions. Regarding the minor axes compared to ple, it gives a glimpse of the many possibilities at
their distances from the launch site, we see that the our disposal after developing a Monte Carlo simula-
wind direction also induces a considerable uncer- tion to our tool. From this starting point, and hav-
tainty. ing enough computational resources, more complex

9
optimizations may be processed. Besides, other effi- showed expected results.
cient optimization methods (not considered in this
work) may be developed and applied to the pre- References
sented tool so that the expected result can be fast [1] Degeratu, M., Georgescu, A. M., Bandoc, G.,
and directly found, saving resources finding the best Alboiu, N. I., Cosoiu, C. I., and Golumbeanu,
alternative from the outputs. M. Atmospheric Boundary Layer Modelling as
Mean Velocity Profile Used for Wind Tunnel
6. Conclusions Tests on Contaminant Dispersion in the At-
The present work focused on the development of mosphere. Journal of Environmental Protec-
a trajectory simulator tool for fin-stabilized model tion and Ecology, 14(1):22–28, 2013.
and sounding rockets addressing the effects of an
[2] Thuillier, R. H. and Lappe, U. Wind and Tem-
ABL profile. This tool was developed under Math-
perature Profile Characteristics from Observa-
ematica R
and supports from model to sounding
tions on a 1400 ft Tower. Journal of Applied
rockets with any number of stages. This work pre-
Meteorology, 3:299–306, June 1964.
sented the tool’s flexibility to customize the de-
fault models and to compute the outcome from the [3] Seibert, G. The History of Sounding Rockets
inputs’ uncertainties (or perform an optimization) and Their Contribution to European Space Re-
from a Monte Carlo simulation. search. ESA Publications Division, November
The developed structure to describe the rocket 2006.
was presented and is adaptable to any type of com-
mon rockets. The developed atmospheric model [4] Wolfram Research, Inc., Mathematica, Version
takes into account the local conditions at the launch 10.2, Champaign, IL (2015).
site which tend to the CIRA-86 profile at a speci- [5] Barrowman, J. Stability of a Model Rocket
fied rate. Likewise, above the ABL, the wind model in Flight. Technical Information Report 30,
approximates the values given in the CIRA-86 ta- Centuri Engineering Company, 1970.
bles. Within the ABL, as the wind plays an impor-
tant role in the model rockets trajectory, three pro- [6] Stull, R. An Introduction to Boundary Layer
files were implemented considering the atmospheric Meteorology. Kluwer Academic Publishers,
stability, which can be deduced from the Obukhov 1988.
length. As this parameter is hard to determine with [7] Tennekes, H. The Logarithmic Wind Pro-
low errors using common measuring instruments, a file. Journal of the Atmospheric Sciences,
qualitative alternative was presented to figure out 30(2):234–238, March 1973.
the atmospheric stability from the weather condi-
tions. Besides the local atmospheric data, the ABL [8] Arya, S. P. Introduction to Micrometeorology.
profiles also take into account the type of the sur- Academic Press, Inc., 1988.
face. In order to improve the simulations consider-
[9] Gryning, S. E., Batchvarova, E., Brümmer,
ing the rocket’s attitude, as our tool possesses three
B., Jørgensen, H., and Larsen, S. On the ex-
degrees of freedom, we developed a model to com-
tension of the wind profile over homogeneous
pute the angle of attack over time that can be in-
terrain beyond the surface boundary layer.
fluenced by wind gusts appearing along the wind
Boundary-Layer Meteorology, 124(2):251–268,
direction. A drag coefficient model was also imple-
August 2007.
mented depending on the Mach, the Reynolds and
the configuration of the rocket. [10] Tewari, A. Atmospheric and Space Flight Dy-
A simulation from a model rocket launch was per- namics. Birkhauser, 2007.
formed in order to present examples from the tool’s
[11] Feodosiev, V. I. and Siniarev, G. B. Intro-
outputs. The results revealed a large impact of
duction to Rocket Technology. Academic Press
the wind in the trajectories (due to the weather-
Inc., 1959.
cock effect), which leads us to reassure the impor-
tance to forecast the atmospheric stability and to [12] Milligan, T. Basics of Dynamic Flight Analysis
model a suitable wind profile. Exploring the poten- (Part 3) - The Damping Moment Coefficient.
tialities from the tool, from a Monte Carlo simula- Peak of Flight Newsletter, (195), October 2007.
tion, we also conclude that the wind profiles must
always concern the local conditions at the launch [13] Curtis, H. Orbital Mechanics for Engineering
since, for the defined uncertainties, they impose a Students. Elsevier, 2nd edition, 2010.
strong influence in estimating the landing site. Us- [14] Raychaudhuri, S. Introduction to Monte Carlo
ing the same simulation process, we also presented a simulation. In Proceedings of the 2008 Winter
method to optimize the rocket characteristics which Simulation Conference, pages 91–100, 2008.

10