Académique Documents
Professionnel Documents
Culture Documents
Contents
1 Introduction
2 Measurement
2.1 Build . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Gathering Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
2
3
3 Data
4 Analysis
4.1 Elementary Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Lagrangian Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
5
6
9
9
9
6 Conclusion
10
12
MEASUREMENT
Introduction
The trebuchet is a complicated mechanical weapon of warfare that was common in the Middle Ages. It
has been built in various shapes and sizes; however, every trebuchet shares a common launching mechanism.
Trebuchets use a gravitational pull on a large object and a fulcrum to launch a smaller projectile. This
project examines the behavior of the trebuchet with two different methods of analysis. . By categorizing the
trebuchet as a double pendulum, Lagrangian mechanics can be used to analyze its complicated behavior.
Shown below is a time lapse photo of a trebuchet firing from a software simulation program [4].
As one can see, unique behavior of a trebuchet that makes it more complicated than other catapults is
the behavior of the hanging masses shown above. The doubled coupled pendulum allows the user to optimize
stall points [5], which allows extremely efficient energy transfer to the projectile. The youtube video link
posted has an excellent simulation of a coupled pendulum and explanation of stall points. The idea here is
to analyze the system and see if we can take the behavior of the double pendulum in our analysis to calculate
the speed the projectile leaves - and its horizontal range - more accurately.
Measurement
This section explores how the team setup the experiment and gathered data. Data was gathered by
building a miniature trebuchet and launching small projectiles in doors. The following sections discuss the
construction of the trebuchet and how we gathered the data for our analysis.
2.1
Build
In order to launch projectiles, we had to construct a trebuchet. We chose a simple model that was sturdy
and capable of being fired in a systematic manner. The model is found on the instructables site [3]. Below
shows the process of glueing the wooden dowel rods together. We chose to increase the size of the model by
50% to increase durability.
2.2
Gathering Data
MEASUREMENT
We fixated a nail into the throwing arm and made a pouch out of an old t-shirt to hold the small projectile.
2.2
Gathering Data
To gather the data we used an empty lab room and placed the trebuchet on flat ground. The team placed
the trebuchet on a rubber mat and weighed it down to minimize precession and the effects of sliding caused
by the counterweight.
After building and firing the trebuchet, the team was ready for data collection. We shot the projectiles
down range after coating them in chalk so that we could precisely measure where they landed. We made
DATA
sure to measure their weight after applying the chalk for precision. From the picture one may notice the
measuring tape on the floor, we used this tape to measure the range of the projectile after it left its mark
on a dark towel (sliding of the towel should be considered negligible).
Data
This section tabulates and explores the data for the entire experiment, from weight to range of projectiles.
Each firing required precise measurement of distances as well as specific documentation of the length and
angle of the trebuchet components. We fired seven small glass marbles in two trials and recorded the
distances.
Weights Because the masses that we measured were all within .1 grams of each measurement we can
simply assume that they are invariant.
Type
Counterweight
Projectile
Mass (kg)
0.5006
0.0022
Dimensions
Type
Throwing Arm R
Sling r
Counter Weight Arm L
Counter Weight Swing l
length (cm)
24.2
11.2
4.3
3.2
Trial 1
Projectile
1
2
3
4
5
6
7
Range (meters)
7.176
7.143
7.026
6.988
6.912
6.874
6.645
Range (meters)
7.522
7.318
6.998
6.950
6.902
6.956
6.598
ANALYSIS
Analysis
Because of the trebuchet?s design, the launching process can be analyzed by both elementary Newtonian
mechanics and Lagrangian mechanics. Because we are firing very small projectiles at low speeds, it is rather
safe to assume that the drag coefficient for spherical-like objects in air at low speeds is negligible [1].
4.1
Elementary Mechanics
The key to the analysis is the launch angle, if we can accurately measure the launch angle then we can
separate the velocity into components and calculate the range. Luckily we obtained a picture that describes
that angle well.
We measured the angle of launch to be 42 degrees. To calculate the theoretical max range we must take the
energy perspective:
E = T + U = constant
1 2
m~r = M gh
2
where h is the height the counterweight fell, g the gravity, M the mass of the counterweight, m the mass
of the projectile, and ~r the velocity magnitude of the projectile. We are interested in the horizontal range
which is simply found by calculating ~r and separating the components using standard kinematics.
s
r
2(0.5006kg)(9.8 sm2 )(0.075m)
2M
gh
m
~r =
=
= 18.289
m
0.0022kg
s
m
m
r~y = sin 42 18.289 = 12.238
s
s
2r~y
time of f light = tf =
= 2.498s
g
m
Range = r~x tf = cos 42 18.289 2.498s = 33.945m
s
4.2
4.2
Lagrangian Mechanics
ANALYSIS
Lagrangian Mechanics
The goal of this section is to analyze the trebuchet as a coupled double pendulum. The prospect of
an analytically closed solution is not really reachable, so the hope is that we can use this method to more
accurately characterize the trebuchet and its behavior. The following section sets up the equations of motion
and given the initial conditions attempts to calculate the velocity at release. For an extremely thorough
discussion of all the different types of trebuchets and their analyses please consider glancing at [2], which
our analysis is based off of. Below is the mathematical setup, we chose to use the pivot point of the beam
on the stands as the origin and because of the mass of our projectile being of less mass than our beam, the
beam should be considered in our system.
L=
sin ( + ) + M gL sin
M gl( + )
1
1
+ mb (L2 LR + R2 ) 2 + mb g(L R) cos
6
2
4.2
Lagrangian Mechanics
ANALYSIS
1
2 2rR( )
cos cos (2 ) + R2 2
m r2 ( )
2
1
2 2lL( + )
cos cos () + L2 2
+ M l2 ( + )
2
sin ( ) + mgR sin M gl( + )
sin ( + ) + M gL sin
mgr( )
1
1
+ mb (L2 LR + R2 ) 2 + mb g(L R) cos
6
2
L=
The differential equations can be obtained from the Euler-Lagrange Formalism...We notice that r = 0, R = 0,
l = 0, and L = 0 so we need not worry about those equations of motion.
L
d L
=
dt
d L
L
=
dt
d L
L
=
dt
After fitting the algorithm to our needs - graciously found in [2] - We have the following plots for the
angles of interest with respect to radians vs time.
psi@tD
2.0
1.5
1.0
0.5
0.1
0.2
0.3
0.4
0.5
-0.5
-1.0
phi@tD
8
0.1
0.2
0.3
0.4
0.5
4.2
Lagrangian Mechanics
ANALYSIS
This algorithm, setting the center of mass of the beam to be zero maintains the following maximum
range:
The maximum range is 11.164m at time 0.1328 s
We may also consider adding the mass of the beam into the equations in some way, unfortunately the
numerical methods do not appreciate the value of the mass of the beam we have of 0.057kg as it renders a
boundary issue. However we multiply the projectile mass by two (which is rather reasonable as the center
of mass is located on the long arm of the beam and the mass is comparable to that of the projectile from a
torque standpoint), we obtain the following results.
psi@tD
2.0
1.5
1.0
0.5
0.1
0.2
0.3
0.4
0.5
-0.5
-1.0
phi@tD
8
0.1
0.2
0.3
0.4
0.5
This algorithm, setting the center of mass of the beam to be zero, but multiplying the projectiles mass
by two maintains the following maximum range:
The maximum range is 7.870m at time 0.1492 s
This section discusses the results and the error of the theoretical range to what was observed in our
experiment. It should be mentioned that the error in our theoretical methods were omitted numerically,
but that in reality there is some error involved with the theoretical techniques. The elementary analysis
error would be rather easy to use the tangential error method, however an estimation for the error in the
Lagrangian formalism would be an academic report in itself so the error is omitted. Again, the real goal was
to compare Elementary and Lagrangian methods in order to find which one better describes our system.
5.1
Elementary Mechanics
As we can see, the elementary analysis is quite ambitious! Thirty-four meters is a very long distance with
our current setup. If we compare to the overall average range found in the experiment we are over a factor
of four off! This is most likely due to the complexity of the setup. The coupled double pendulum must be
tuned to optimize the range and this is clearly the maximum range possible, but this analysis doesnt take
into account the energy dissipation in the system such as moving the beam, translating the counterweight,
flinging the sling, and so forth.
5.2
Lagrangian Mechanics
The Lagrangian method proved far more mathematically taxing than the elementary analysis, but it
seemed to have paid off. The Euler-Lagrange equations were analytically derived from the Lagrangian using
Mathematica and the generous script found from [2]. The algorithm predicted a maximum range of 11
meters without using a mass for the beam, which is about 60% off value of what was observed.
When we added the effect of the beam mass in the algorithm I was unable to calculate because of
computer error - we therefore added the center of mass by just doubling the projectiles mass. The resulting
maximum range was extremely pleasing. We obtained 7.870m which is only a percent difference of 12% thus
the extra mass of the beam was really important in the analysis. The range given by the algorithm also
assumes a 45 degree angle of launch which we were very close to obtaining, but we where still slightly off which means that this approach was even closer than expected. In comparison with the elementary analysis,
we were able to obtain results that were much closer to the experimental results.
If given more time, and perhaps a time lapse camera, we could detect the exact angle of release and trace
that to a time stamp to compute the velocity of the projectile at release to extreme precision and accuracy.
The numerical method approach to the trebuchet seem to be paramount to analyzing it effectively.
CONCLUSION
Conclusion
This project explored the mechanical nature of a trebuchet. Trebuchets use a gravitational pull on a large
object and a fulcrum to launch a smaller projectile, and are often categorized by complicated mechanical
additions such as slings. The type of trebuchet analyzed here was a type with a hanging counterweight and
a hanging projectile. The system can be categorized mathematically as a coupled double pendulum.
The team gathered the experimental data after tuning the trebuchet by shooting chalk covered glass beads
and measuring the range of the projectiles. The data showed that the hand-built system was systematic
with an average of 7.000 meters and standard deviation of 0.232 meters.
Two different techniques were used to analyze the trebuchet and compare theoretical to experimental
results. The first technique was simple elementary techniques that used conservation of energy and kinematics
to calculate the range of the projectile based on the launch angle. This analysis proved far too ambitious,
predicting a value around 30 meters.
The second technique used the Euler-Lagrange formalism, which is complicated using the double coupled
pendulum model. The prospect of a closed form analytical solution was lost and the team had to find a
numerical method to generate and solve the equations of motion. The team used a Mathematica algorithm
from [2] and found that the algorithm was sufficient in analyzing the trebuchet. Omitting the mass of the
beam that coupled the slings produced a predicted value of 11.164 meters and adding the mass of the beam
by doubling the projectiles mass produced a range of 7.870 meters which was far closer to the experimental
results.
Both analyses of the trebuchet certainly had error involved but the team was more concerned with fitting
the data from the experiment with an accurate model. One may compute the tangential error method for
the elementary analysis and do a numerical error analysis of the Euler-Lagrange method. Again, the real
goal was to compare Elementary and Lagrangian methods in order to find which one better describes our
system.
The Euler-Lagrange numerical method produced results that were closer to the experimental data. Thus
the coupled double pendulum is sufficient in describing the trebuchet. The inference to best explanation
of the error in the elementary analysis is due to tuning parameters. Although the trebuchet had a rather
impressive range, from conservation of energy, it should be expected that one could tune our system further to
reach the theoretical limit. So the efficiency of the trebuchet was rather low. With more time and equipment
one could analyze the trebuchet and optimize parameters based on results of the numerical analysis. The
project was successful in establishing a mathematical model and analyzing results from a trebuchet.
10
REFERENCES
REFERENCES
References
[1] NASA, Editor: Tom Benson, NASA Official: Tom Benson, http://www.grc.nasa.gov/WWW/k12/airplane/shaped.html Last Updated: June 12, 2014.
[2] Donald B. Siano,
Trebuchet Mechanics,
content/uploads/2013/09/trebmath35.pdf
March
28,
2001.
http://asme.usu.edu/wp-
11
12
th=.;phi=.;psi=.;l1=.;l2=.;l3=.;l4=.;
m1=.;m2=.;m3=.;mb=.;g=.
cn={l1,l2,l3,l4,m1,m2,mb};
x1[th_]:=l1 Sin[th];
y1[th_]:=-l1 Cos[th];
x2[th_]:=-l2 Sin[th];
y2[th_]:=l2 Cos[th];
x4[th_,phi_]:=l1 Sin[th]-l4 Sin[phi+th];
y4[th_,phi_]:=-l1 Cos[th]+l4 Cos[phi+th];
x3[th_,psi_]:=-(l3 Sin[psi-th]+l2 Sin[th]);
y3[th_,psi_]:=-(l3 Cos[psi-th]-l2 Cos[th]);
V[th_,phi_,psi_]:=m1 g y4[th,phi]+m2 g y3[th,psi]-mb g (l1 l2)Cos[th]/2
T[th_,phi_,psi_]:=(m1/2)((Dt[x4[th,phi],t,Constants->cn])^2+
(Dt[y4[th,phi],t,Constants->cn])^2)+(m2/2)
((Dt[x3[th,psi],t,Constants->cn])^2+(Dt[y3[th,psi],t,Constants>cn])^2)+(mb/6) (l2-l1 l2+l1) Dt[th,t,Constants->cn];
Lag[th_,phi_, psi_]:=T[th,phi,psi]-V[th,phi,psi];
ltrr=Lag[th,phi,psi]/.{Dt[th,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]>thd,Dt[phi,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]>phid,Dt[psi,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]->psid};
The lagrangian:
0.-0.02156 (-0.112 Cos[psi-th]+0.242 Cos[th])-4.90588 (-0.043 Cos[th]
+0.032 Cos[phi+th])+0.0011 ((-0.112 Cos[psi-th] (Dt[psi,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]-Dt[th,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}])-0.242 Cos[th]
Dt[th,t,Constants->{0,0.0022,0.032,0.043,0.112,0.242,0.5006}])2+(0.112
(Dt[psi,t,Constants->{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]Dt[th,t,Constants->{0,0.0022,0.032,0.043,0.112,0.242,0.5006}])
Sin[psi-th]-0.242 Dt[th,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}] Sin[th])2)+0.2503 ((0.043
Cos[th] Dt[th,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]-0.032 Cos[phi+th]
(Dt[phi,t,Constants->{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]
+Dt[th,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]))2+(0.043
Dt[th,t,Constants->{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]
Sin[th]-0.032 (Dt[phi,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]+Dt[th,t,Constants>{0,0.0022,0.032,0.043,0.112,0.242,0.5006}]) Sin[phi+th])2)
ltrr=Lag[th,phi,psi]/.{Dt[th,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]>thd,Dt[phi,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]-
13
>phid,Dt[psi,t,Constants->{l1,l2,l3,l4,m1,m2,mb}]->psid};
eqbig=Simplify[{Dt[D[ltrr,thd],t]-D[ltrr,th]0,Dt[D[ltrr,phid],t]D[ltrr,phi]0,Dt[D[ltrr,psid],t]-D[ltrr,psi]0}/.{Dt[l1,t]->0,Dt[l2,t]>0,Dt[l3,t]->0,Dt[l4,t]->0,Dt[mb,t]->0,Dt[m1,t]->0,Dt[m2,t]>0,Dt[g,t]->0,Dt[th,t]->thd,Dt[phi,t]->phid,Dt[psi,t]->psid,Dt[thd,t]>thdd,Dt[phid,t]->phidd,Dt[psid,t]->psidd}];
(*input the parameters here*)
m1=0.5006;
m2= 0.0022;
mb=0;
l1= 0.043(*0.141076;*);
l2= 0.242(*0.793963;*);
l3= 0.112(*0.367454;*);
l4= 0.032(*0.104987;*);
g=9.8;
l5=l4/Sqrt[2];
(psis=ths-Pi/2;);
ths=(3*Pi)/4;
phis=-ths+Pi;
(*set up the equations*)
eqs=eqbig/.{th->th[t],thd->Derivative[1][th][t],thdd->Derivative[1]
[Derivative[1][th]][t],phi->phi[t],phid->Derivative[1][phi][t],phidd>Derivative[1][Derivative[1][phi]][t],psi->psi[t],psid->Derivative[1]
[psi][t],psidd->Derivative[1][Derivative[1][psi]][t]};
(*Solve the DE:*)
solslcw=NDSolve[{eqs,th[0]ths,phi[0]phis,psi[0]psis,th'[0]0,phi'[0]0,p
si'[0]0},{th[t],phi[t],psi[t]},{t,0.,1}];
(*get th,psi and the velocity from this solution*)
thint[t_]=Chop[th[t]/.Flatten[solslcw][[1]]];
psiint[t_]=Chop[psi[t]/.Flatten[solslcw][[3]]];
Plot[thint[t],{t,0,.5},PlotLabel->"psi[t]"]
Plot[psiint[t],{t,0,.5},PlotLabel->"phi[t]"]
(*now to get the range,we first need to get the velocity of the
projectile*)vx3[t_]:=Chop[-l3 Cos[psiint[t]-thint[t]]*(psiint'[t]thint'[t])-l2*Cos[thint[t]]*thint'[t]];
vy3[t_]:=Chop[l3*(psiint'[t]-thint'[t])*Sin[psiint[t]-thint[t]]l2*thint'[t]*Sin[thint[t]]];
range[t_]=2 vx3[t]*vy3[t]/g;
Plot[range[t],{t,0,.4},PlotLabel->"range[t]"]
14
15