You are on page 1of 10

Trebuchet Truss Design and Analysis

Using the Finite Element Method


Tal Bar-Or, Samuel H. Gilbert, Lucas G. Hartman, & Wilson R. Pulling
University of Pennsylvania; Department of Mechanical Engineering and Applied Mechanics
229 Towne Building | 220 S. 33rd Street | Philadelphia, PA 19104-6315

“Simplicity is the ultimate sophistication.”


–Leonardo Da Vinci
INTRODUCTION
Team Carchidi designed a trebuchet arm to be constructed with acrylic truss elements. Load computations were
performed with a supplied MATLAB script to maximize the strength-to-weight ratio of the structure. The
motivations behind this project were to gain valuable design experience and to expand our understanding of load
levels throughout truss members.

GOALS
This project was aimed at optimizing the strength-to-weight ratio of a truss structure (defined as the failure load of
the truss divided by its total weight) while adhering to a list of dimensional and strength specifications. Each
requirement is detailed below.

Structure Size and Supports


The truss had to be 50 𝑐𝑚 long, supported at the left end by two points A and G at least 5 𝑐𝑚 apart and no more
than 10 𝑐𝑚 apart. Truss elements were not to cross each other. Truss elements that met had to end, and be
connected by pin elements that could transmit loads.

Material
The truss was designed to be constructed of acrylic, with a density of 1200 𝑘𝑔 ∙ 𝑚−3 .

Element Area
The cross-sectional area, 𝐴# , of each truss member could vary but had to be a minimum of 4 × 10−5 𝑚2 .

Load Requirements
The truss had a design load of 𝐹𝐶 = 120 𝑁. The load was to be applied vertically at the right end of the truss.

Factor of Safety
The truss had to have a factor of safety of at least 2.0, meaning it had to resist at least load of up to twice the design
load without failure, per the criteria listed below:

Yield: The magnitude of stress could not exceed |𝐹# | ⁄ 𝐴# = 15 𝑀𝑃𝑎 = 15 × 106 𝑁 ∙ 𝑚−2 in
any member.

Buckling: No truss member in compression can have the magnitude of its load, 𝐹# , exceed the
value 𝜋 2 𝐸𝐴# 2 ⁄12𝐿# 2 , where 𝐿# is the member’s length, and 𝐸 is the Young’s Modulus
of each member (𝐸 = 2.5 𝐺𝑃𝑎 = 2.5 × 109 𝑁 ∙ 𝑚−2 ). This formula assumes a square
cross-section. Buckling will not occur in tension.

Stiffness
The truss structure as a whole had to have a minimum stiffness of 5 𝑘𝑁 ∙ 𝑚−1 . This was calculated at the node
where the maximum load was applied. The stiffness was calculated as the ratio of the applied force (design load) to
the magnitude of the vertical deflection at the point of application.

METHOD
The truss script was downloaded off of Blackboard and saved into a working directory. The design optimization
process consisted of multiple iterations. Each iteration started with the design of a new and/or improved truss on

1
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

paper. Using a text editor, the input files for the Matlab script were edited to reflect each new truss structure to be
analyzed. The first iteration of each unique design was performed using the minimum weight possible for that
structure (meaning all members had the minimum allowable cross-sectional area). The script was then executed,
followed by mathematical post-processing of its output data, which was done in Excel. The calculations performed
will be outlined in Appendices section of this report.

Each iteration was based off of the previous design’s failures and successes. For example, when a member in an
intermediate design was found to have a stress within the allowable tolerance, its cross sectional area was gradually
decreased until its stress was just below the tolerance. This saved weight without compromising the structure’s
factor of safety. Similarly, when a specific member of the truss structure would have failed by buckling, its cross-
sectional area was gradually increased up until it met the required specifications to not fail. By addressing each
failure point individually, it was possible to meet all specifications while maximizing the strength-to-weight ratio.

The opening quote by Da Vinci exemplifies the approach used to designing the truss. To this end, the very first
iteration was a simple triangular arrangement of three truss members, as depicted below in Figure 1. Building on
this design, several iterations of varying complexity took place to yield the design of the final truss.

Figure 1 – Truss Iteration A

Figure 2 - Deformation and Distribution of Loads of Truss Iteration A Under Load of 𝟐𝟒𝟎 𝑵

Undeformed
0.2 Tensile stress
Compressive stress

0.15

0.1

0.05
Y

-0.05

-0.1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
X

2
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

Table 1 – Truss Iteration A: Node Coordinates and Displacements Under Load of 𝟐𝟒𝟎 𝑵
Horizontal Vertical
Node X-coordinate (m) Y-coordinate (m) Displacement @ Displacement @
240 N(m) 240 N (m)
A 0.0 0.1 0.00000 0.00000
B 0.5 0.1 0.00200 -0.01350
C 0.0 0.0 0.00000 -0.00024

Table 2 – Truss Iteration A: Member Properties: Cross-sectional Area, Length, Young’s Modulus, Stress
Needed to Cause Yield, and Force Needed to Cause Buckling for Each Member
Cross-sectional Young’s Stress Needed Compressive Force
2
Member area (m ) Length (m) Modulus (GPa) to Yield (MPa) to Buckle (kN)
1 0.000120 0.500000 2.5 15 -
2 0.000395 0.509902 2.5 15 1.233898259
3 0.000040 0.100000 2.5 15 -

Table 3 – Truss Iteration A: Stress and Buckling Force Experienced by Each Member at Load of 𝟐𝟒𝟎 𝑵
Stress in Member Force in Member
Member @ 240 N (MPa) @ 240 N (kN)
1 10.00 1.20
2 3.10 -1.22
3 6.00 0.24

The data in Tables 1-3 was processed to reveal that this truss had a strength-to-weight ratio of 76.81. Its stiffness
was 17.82 𝑘𝑁 ∙ 𝑚−1 . The truss’ total mass was 0.318 𝑘𝑔; its weight was 3.12 𝑁. Its volume was 2.65 ×
10−4 𝑚3 . The maximum load the structure could handle before failure was 240 𝑁. At this load, member 2 (as
depicted in Figure 1) failed by buckling due to compression. Thus, the truss’ factor of safety was 2. While this
initial iteration did fulfill all of the required parameters, further steps were taken to drastically improve the truss.

A second major iteration that was undertaken during the lengthy trial and error period is depicted in Figure 3:

Figure 3 – Truss Iteration B


240
NN

“UNLOADED”

“UNLOADED”

“UNLOADED”

“UNLOADED”

3
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

Figure 4 - Deformation and Distribution of Loads of Truss Iteration B Under Load of 𝟐𝟒𝟎 𝑵

Undeformed
0.2 Tensile stress
Compressive stress

0.15

0.1

0.05
Y

-0.05

-0.1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
X

Table 4 – Truss Iteration B: Node Coordinates and Displacements at Load of 𝟐𝟒𝟎 𝑵


Horizontal Vertical
Displacement Displacement
X- Y- @ @
Node coordinate (m) coordinate (m) 240 N (m) 240 N (m)
A 0.0 0.10 0.00000 0.00000
B 0.5 0.10 0.00300 -0.03081
C 0.4 0.08 -0.00051 -0.01017
D 0.3 0.06 -0.00077 -0.00576
E 0.2 0.04 -0.00078 -0.00259
F 0.1 0.02 -0.00054 -0.00067
G 0.0 0.00 0.00000 -0.00024

Table 5 – Truss Iteration B: Member Properties: Cross-sectional Area, Length, Young’s Modulus, Stress
Needed to Cause Yield, and Force Needed to Cause Buckling for Each Member
Cross-sectional Young’s Stress Needed Compressive Force
2
Member area (m ) Length (m) Modulus (GPa) to Yield (MPa) to Buckle (kN)
1 0.0000804 0.50000 2.5 15 -
2 0.0000820 0.10198 2.5 15 -
3 0.0000820 0.10198 2.5 15 1.3294
4 0.0000820 0.10198 2.5 15 1.3294
5 0.0000820 0.10198 2.5 15 1.3294

4
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

6 0.0000820 0.10198 2.5 15 1.3294


7 0.0000400 0.10000 2.5 15 -
8 0.0000400 0.40050 2.5 15 -
9 0.0000400 0.30266 2.5 15 -
10 0.0000400 0.20881 2.5 15 -
11 0.0000400 0.12806 2.5 15 -

Table 6 – Truss Iteration B: Stress and Buckling Force Experienced by Each Member Load of 𝟐𝟒𝟎 𝑵
Stress in Member Force in Member
Member @ 240 N (MPa) @ 240 N (kN)
1 14.9876 1.2050
2 14.9861 1.2289
3 14.9861 -1.2289
4 14.9861 -1.2289
5 14.9861 -1.2289
6 14.9861 -1.2289
7 6.0250 0.2410
8 0.0000 0.0000
9 0.0000 0.0000
10 0.0000 0.0000
11 0.0000 0.0000

This next design also fell within all of the required parameters, however its was not optimized. Since the previous
iteration, several zero-force truss members were added to help prevent buckling. It was noted that members carrying
no load could be used to prevent buckling by breaking a long member into several shorter ones connected linearly
by multiple nodes. Simple mathematical calculations revealed that the strength benefits of these zero force members
far outweighed their weight toll. The data in Tables 4-6 revealed that the truss depicted in Figure 3 had a strength-to-
weight ratio of 161.09. Its stiffness was 7.85 𝑘𝑁 ∙ 𝑚−1 . The truss’ total mass was 0.1532 𝑘𝑔; its weight was
1.50 𝑁. Its volume was 1.27 × 10−4 𝑚3 . The maximum load the structure could handle before failure was 241 𝑁.
Immediately beyond this load, members 1-6 (as depicted in Figure 3) failed by yielding. Thus, the truss’ factor of
safety was also 2.

To further optimize the truss, the thickness of the truss members was carefully increased. Through meticulous
optimization of cross-sectional areas, the maximum strength-to-weight ratio was achieved. The final truss design is
outlined below.

RESULTS
The final design fell within all of the required parameters. The truss’ total mass was 0.4661 𝑘𝑔; and its weight was
4.5720 𝑁. Its volume was 3.8838 × 10−4 𝑚3 . The maximum load the structure could handle before failure was
1000 𝑁. The optimized truss had a strength-to-weight ratio of 218.72. Its stiffness was 32.09 𝑘𝑁 ∙ 𝑚−1 .
Immediately beyond this load, members 1-7 (as depicted in Figure 5) failed by yielding. Thus, the truss’ factor of
safety is 8.33. Calculations for all of the above values can be found in the Appendices.

5
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

Figure 5 – The Final Truss Design

“UNLOADED”

“UNLOADED”

“UNLOADED”

“UNLOADED”

Tables 7-9 contain data utilized to calculate the aforementioned characteristics.

Table 7 – Node Coordinates and Displacements at Max. Load of 𝟏𝟎𝟎𝟎 𝑵 and Design Load of 𝟏𝟐𝟎 𝑵
Horizontal Horizontal Vertical Vertical
Displacement Displacement Displacement Displacement
X- Y- @ @ @ @
Node coordinate (m) coordinate (m) 1000 N (m) 120 N(m) 1000 N (m) 120 N (m)
A 0.00 0.10 0.00000 0.00000 0.00000 0.00000
B 0.50 0.10 0.00300 0.00036 -0.03119 -0.00374
C 0.40 0.08 -0.00052 -0.00006 -0.01046 -0.00126
D 0.30 0.06 -0.00080 -0.00010 -0.00598 -0.00072
E 0.20 0.04 -0.00082 -0.00010 -0.00274 -0.00033
F 0.10 0.02 -0.00060 -0.00007 -0.00074 -0.00009
G 0.00 0.00 0.00000 0.00000 -0.00060 -0.00007

Table 8 – Member Properties: Cross-sectional Area, Length, Young’s Modulus, Stress Needed to Cause
Yield, and Force Needed to Cause Buckling for Each Member
Cross-sectional Young’s Stress Needed Compressive Force
2
Member area (m ) Length (m) Modulus (GPa) to Yield (MPa) to Buckle (kN)
1 0.00033350 0.50000 2.5 15 -
2 0.00033998 0.10198 2.5 15 -
3 0.00033998 0.10198 2.5 15 22.8524
4 0.00033998 0.10198 2.5 15 22.8524
5 0.00033998 0.10198 2.5 15 22.8524
6 0.00033998 0.10198 2.5 15 22.8524
7 0.00006670 0.10000 2.5 15 -
8 0.00004000 0.40050 2.5 15 -

6
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

9 0.00004000 0.30266 2.5 15 -


10 0.00004000 0.20881 2.5 15 -
11 0.00004000 0.12806 2.5 15 -

Table 9 – Stress and Buckling Force Experienced by Each Member at Max. Load of 𝟏𝟎𝟎𝟎 𝑵 and Design Load
of 𝟏𝟐𝟎 𝑵
Stress in Member Stress in Member Force in Member Force in Member
Member @ 1000 N (MPa) @ 120 N (MPa) @ 1000 N (kN) @ 120 N (kN)
1 14.9925 1.7991 5.0000 0.6000
2 14.9980 1.7998 5.0990 -0.6119
3 14.9980 1.7998 -5.0990 -0.6119
4 14.9980 1.7998 -5.0990 -0.6119
5 14.9980 1.7998 -5.0990 -0.6119
6 14.9980 1.7998 -5.0990 -0.6119
7 14.9925 1.7991 1.0000 0.1200
8 0.0000 0.0000 0.0000 0.0000
9 0.0000 0.0000 0.0000 0.0000
10 0.0000 0.0000 0.0000 0.0000
11 0.0000 0.0000 0.0000 0.0000

Figures 6 and 7 show the graphic output of the FEA performed on the truss structure using Matlab.

Figure 6 – Deformation and Distribution of Loads of the Final Truss Design Under Max. Load of 𝟏𝟎𝟎𝟎 𝑵

Undeformed
0.2 Tensile stress
Compressive stress

0.15

0.1

0.05
Y

-0.05

-0.1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
X

7
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

Figure 7 – Deformation and Distribution of Loads of the Final Truss Design Under Design of 𝟏𝟐𝟎 𝑵

Undeformed
0.2 Tensile stress
Compressive stress

0.15

0.1

0.05
Y

-0.05

-0.1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
X

DISCUSSION
A loaded truss structure contains two-force members that experience either tensile or compressive forces. In static
equilibrium, the sum of all forces (applied and reactive) acting on the truss must equal zero. By extension, the sum
of the forces experienced at each node must be zero. The sum of all moments about any point must also be zero.

A Matlab script was used to model the truss design. The location of the nodes and cross-sectional areas of the
members, the young’s modulus, and the external forces were inputted into the script in order to generate the forces
in the members and the corresponding displacement. The accuracy of the Matlab script was verified by manual
calculation of the forces in each member. According to the post processing of the script’s output data, all seven
load-bearing members experienced a stress of approximately 14.99 𝑀𝑃𝑎 (Table 9) at the maximum load. This is
just under the maximum allowable stress of 15 𝑀𝑃𝑎, and is an indication that the design was optimized. The
compressive force experienced by members 3-6, 5.0990 𝑘𝑁, is substantially smaller than the force which would
cause compressive buckling, 22.8524 𝑘𝑁 (Table 8). The compressive force to buckle is inversely proportional to
the length of the member squared. Had a sole member been used to connect nodes B and G, as was the case in one
of the first iterations, buckling of this member would occur at a load 25 times smaller. This is because the potential
member would have been five times as long as members 2-6 were. The stiffness of the truss is 32.09 𝑘𝑁 ∙ 𝑚−1 , far
exceeding the requirement of 5 𝑘𝑁 ∙ 𝑚−1 . According to post-processing of the script’s output data, the truss’ factor
of safety, 8.33 is well above the minimum requirement of 2. Thus, given the validity of the Matlab script data, it is
likely that the truss will not fail when it is manufactured and utilized as part of a trebuchet.

The biggest weakness of the final truss design is that it has four zero force members. As previously discussed, the
strength contributed by these members far exceeds their weight. In other words, a sole member extending from B to
G in lieu of members 2-6 would have resulted in the removal of the non-load carrying members (members 8-11).
However, to meet strength requirements, this member would have to have a large cross sectional area and thus
would be far heavier. Another major weakness of the design is that if at any moment the instantaneous load on the
truss exceeds 1000 𝑁, it could fail. It is important to note that the Matlab script is not perfectly accurate. In the real
world, the truss would react very differently and possibly fail sooner than expected. In real-life situations, the zero-
force members may not actually strengthen the truss.

8
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

Had the truss been designed using CAD and FEM software, the design would differ slightly. Members that were far
from failure at the design and maximum loads under the current constraints could have small holes drilled through
them to reduce their weight. These holes would be sized and placed to further improve the strength-to-weight ratio
of the truss.

CONCLUSION
Designing the truss offered insight into the complexities of real-world engineering. By far, the most striking
realization was that the process of optimizing a design while meeting given criteria can be tedious and time-
consuming, but also gratifying. Specifically, small changes in structural properties can significantly alter the
behavior and safety of a particular design. Even zero-force members can affect the behavior of a truss, increasing its
strength-to-weight ratio.

APPENDICES
Variable Nomenclature
𝑀 is the mass of the truss. 𝐷is the density of the acrylic. 𝑉 is the volume of the entire truss. 𝑊 is the weight of the
truss. 𝑔 is the acceleration caused by gravity. 𝐹𝐵 is the magnitude of the maximum load applied to the structure at
node B to put the truss on the verge of failure. 𝐹𝐷 is the design load of the structure, 120 𝑁. 𝜇𝐵 is the magnitude of
node B’s vertical displacement at the design load. 𝐸 is the Young’s Modulus of each member. 𝐴# is the cross
sectional area of member number #. 𝐿# is the length of member number #. 𝜎# is the stress experienced by member
number #. 𝐹# is the force experienced member number #.

Calculations for Final Truss Design


Mass and Weight:

𝑀 = 𝐷 𝑉 = 1200 𝑘𝑔 ∙ 𝑚−3 3.8838 × 10−4 𝑚3 = 0.4661 𝑘𝑔


𝑊 = 𝑀 𝑔 = 0.4661 𝑘𝑔 9.81 𝑚 ∙ 𝑠 −2 = 4.5720 𝑁

Strength-to-Weight Ratio:
𝐹𝐵 1.000 𝑘𝑁
= = 218.72
𝑊 0.0045720 𝑘𝑁

Stiffness:
𝐹𝐷 0.120 𝑘𝑁
𝑆𝑡𝑖𝑓𝑓𝑛𝑒𝑠𝑠 = = = 32.09 𝑘𝑁 ∙ 𝑚−1
𝜇𝐵 −0.00374 𝑚

Factor of Safety:
𝐹𝐵 1.000 𝑘𝑁
= = 8.33
𝐹𝐷 0.1200 𝑘𝑁

Compressive Force to Cause Buckling for Members 2-5:

𝜋 2 𝐸𝐴2 2 𝜋 2 (2.5 𝐺𝑃𝑎)(0.00033998 m2 )2


= = 22.8524 𝑘𝑁
12𝐿2 2 12(0.10198 𝑚)2

Stress in Each Member at Max. Load of 1000 N:

𝐹1 5.0000 kN
𝜎1 = = = 14.9925 𝑀𝑃𝑎
𝐴1 0.00033350 m2

𝐹2 5.0990 kN
𝜎2 = = = 14.9980 𝑀𝑃𝑎
𝐴2 0.00033998 m2

9
Trebuchet Truss Design and Analysis Using the Finite Element Method
Bar-Or, Gilbert, Hartman, & Pulling

𝐹3 −5.0990 kN
𝜎3 = 𝜎4 = 𝜎5 = 𝜎6 = = = 14.9980 𝑀𝑃𝑎
𝐴3 0.00033998 m2

𝐹7 1.0000 kN
𝜎7 = = = 14.9925 𝑀𝑃𝑎
𝐴7 0.00006670 m2

𝐹8 0.0000 kN
𝜎8 = 𝜎9 = 𝜎10 = 𝜎11 = = = 0.00 𝑀𝑃𝑎
𝐴8 0.0000400 m2
Matlab Script Input Files
node.dat: elem.dat:
1.0 0.0 0.1 1.0 1.0 2.0 3.335E-4 2.5
2.0 0.5 0.1 2.0 2.0 3.0 3.3998E-4 2.5
3.0 0.4 0.08 3.0 3.0 4.0 3.3998E-4 2.5
4.0 0.3 0.06 4.0 4.0 5.0 3.3998E-4 2.5
5.0 0.2 0.04 5.0 5.0 6.0 3.3998E-4 2.5
6.0 0.1 0.02 6.0 6.0 7.0 3.3998E-4 2.5
7.0 0.0 0.0 7.0 1.0 7.0 6.67E-5 2.5
8.0 1.0 3.0 4.0E-5 2.5
9.0 1.0 4.0 4.0E-5 2.5
10.0 1.0 5.0 4.0E-5 2.5
11.0 1.0 6.0 4.0E-5 2.5

forces.dat: scalefac.dat: dispbc.dat


1.0 2.0 2.0 -1.0 1 1.0 1.0 2.0
2.0 1.0 1.0
3.0 7.0 1.0

Matlab Script Output Files


displ_out.dat force_out.dat
0.0 0.0 5.0
0.0029985007 -0.031190122 5.0990195
-5.2312139E-4 -0.010462428 -5.0990195
-7.9667609E-4 -0.0059750707 -5.0990195
-8.206641E-4 -0.002735547 -5.0990195
-5.9508541E-4 -7.4385677E-4 -5.0990195
0.0 -5.9970015E-4 1.0
6.6418088E-15
1.0638028E-15
9.7983472E-16
1.8619006E-16

REFERENCES
“Simplicity is the ultimate sophistication. By Leonardo da Vinci.” Famous Quotes at QuoteDB -
Interactive Database of Famous Quotations. 19 Oct. 2008
<http://www.quotedb.com/quotes/3069>.

10

You might also like