Wing Truss Optimization

Asa Berger, David Hopperton, Sachit Rao

University of Cincinnati, Cincinnati, OH 45219

November 30, 2009

Executive Summary
[Type text] [Type text] Page |2

We were to optimize the two feasible trusses that were designed in our last project. Our
first design is a modified Pratt truss and our second design is a modified Warren truss. The cost
function that was used to determine the optimal truss design took into account the compressive
and tensile forces in each member, the length of each member, the number of members, and the
average applied load for the truss. Eight variations were made for each truss design by
increasing the number of sections in each truss. The value of the cost function was found for
each variation and then plotted against the number of sections. The lowest cost function value is
the most optimal design and this was easily seen by the curve formed by the plots of the cost
function values. The optimal truss design was the modified Warren truss with two sections,
which had a cost function value of .82785. The lowest cost function value for the modified Pratt
truss was the four section with a value of .88481.

Introduction

The objective was to optimize the two wing trusses that were deemed feasible under the
following conditions: a rectangular wing 2.5 m in half-span and 1.4 m in chord comprised of
NACA 5410 sections attached to an aircraft with a fuselage diameter of 1.2 m; an altitude and
speed of 14,000 m and Mach 0.31 respectively; the maximum compression and tension for the
truss is 32 kN. It is assumed that the only force acting on the wing truss is the lift produced by
these conditions and the force are only acting at the joint. In order to optimize the two wing truss
designs the cost function was used with eight different variations of each design, which were
made by increasing the number of sections.
The following report gives technical background on the problem that is to be solved and
the process of solving it. It then describes the truss designs that were used for the project. Next,
plots of our collected data are given along with a discussion of the results and our
recommendation. References are then given. Lastly, there is an appendix of sample
calculations.

Technical Background

To calculate the forces acting at each joint in the truss, we needed to calculate the
distribution of lift across the span of the wing. This was done by using a modified version of a
MATLAB program that was written by Krista Kirievich and Cory Kunkel.
[Type text] [Type text] Page |3

Once the elliptical lift distribution is found it is integrated to find the lift force at specific joints.
For example, to find the lift force at a joint that is 0.5 m from the center of the span the lift
distribution equation must be integrated from 0.25 m to 0.75 m and this value for lift is then
applied to the joint at 0.5 m. The figure above illustrates this idea. The solid lines are the
locations of joints, the dotted lines are the boundaries of integration, and the arrows show where
the force within the boundaries is applied.

When a truss is designed and all the forces are known, the forces within the truss must be
solved in order to determine whether the truss is within working limits or not. This is done using
method of joints. When using the method of joints, the x and y components are found for all of
the forces and then the forces are set in equilibrium by adding the forces in the x-direction and in
the y-direction and setting them equal to zero. These equations can be solved for the unknowns
and then the next joint can be solved for. This process is done until all of the unknowns are
solved for the entire truss.

Once all of the forces are known within a truss, the truss can then be optimized. The
following cost function was used to optimize each truss and its variations:
[Type text] [Type text] Page |4

ω1 = 0.6 ω2 = 0.3 ω3 = 0.1

The ω values in the cost function (CF) are weights to give an order of importance of each
value to be considered within the function. The weighted values of F1, F2, and F3 were all plotted
with bar graphs against the number of sections in order to compare this value within the sections.
The CF values were also plotted with bar graphs on one plot against the number of sections so
that the most optimized wing truss design could be determined. The lowest value for CF
represents the most optimized wing truss design.

Proposed Trusses

The following figures show the 1 section design for the two different trusses that were
optimized:

Figure 1: Design 1
[Type text] [Type text] Page |5

Figure 2: Design 2

Design 1, which is represented by figure 1, was based off of the Pratt truss while design
2, represented by figure 2, was based off of the Warren truss. For each design, eight variations
were done by increasing the number of sections of the truss with the maximum being eight.
These variations helped to show the trend that the cost function had within each truss design.

Results and Recommendations

Based on our calculations we observed that F1 followed a generally parabolic path for
truss 1 while truss 2 had an exponential curve. The F1 values for truss 1 started at 1.07 and
decreased until it reaches the 6 section and then increased after that. The F1 values for truss 2
were much more erratic, starting at 1.71 decreasing to .81 at the 4 section design.

The values for F2 followed a much more linear trend because F2 was based solely on the
number of members for a truss, the fewer the number of members, the lower the value of F2. The
smallest value for truss 1 was .2222 and the smallest value for truss 2 was .1852.

Our values for F3 followed the same kind of general trend that F1 followed. That is, a
parabolic trend for truss 1 and an exponential path for truss 2. F3 was based on the average
applied load to each member, and because the maximum load to each member seemed to level
off for the number of sections we used, the average applied load would decrease for the sections
we used.

Using this data combined with the cost function mentioned earlier we concluded that the
2 section design for truss 1 (CF = .8279) and the 4 section design for truss 2 (CF = .8848) are the
lowest values for the cost function.
[Type text] [Type text] Page |6

References

Beer, Ferdinand P., Jr. ,, Elliot R. Eisenberg, William E. Clausen, David Mazurek, and Phillip J.
Cornwell. Vector Mechanics for Engineers Statics and Dynamics. 8th ed. New York:
McGraw-Hill Science/Engineering/Math, 2006. Print.

Kirievich, Krista, Cory Kunkel and Christopher Knowles. Wing Force Calculation Applying
Lifting Line Theory. University of Cincinnati. 2008

Sutton, Matthew G. “Wintruss 1.0.” University of Arkansas. 1999

Appendix: Sample Calculations

Truss 1 Section 2 Calculations:

Solving for Reaction Forces

Reaction at B in x direction
MA=0
0.6053*0.625+0.5405*1.25+0.4089*1.875+0.1025*2.5+(RBX*0.14)=0
RBX=-14.835 kN
Reaction at A in x direction
MB=0
0.6053*0.625+0.5405*1.25+0.4089*1.875+0.1025*2.5-(RBX*0.14)=0
RAX=14.835 kN
Reaction at B in y direction
Fy=0
0.3127+0.6053+0.5405+0.4089+0.1025+(RBY)=0
RBY=-1.97 kN

Solving for Truss members by method of joints

Joint F FG

FE

Fy=0
0.1.025+FG=0
FG= -0.1025 kN
[Type text] [Type text] Page |7

Fx=0
EH=0
Joint G
Fy=0
-0.1025+EGsin(12.63°)=0
EG=.469 kN
Fx=0
GH+.469cos12.63°=0
FG=-.457 kN
Joint E
Fy=0
⁡
EHsin12.63°+0.4089+ .469sin(12.63)=0
EH= -2.34 kN
Fx=0
-2.34cos12.63-2.34cos12.63°+ED=0
EF=2.74 kN
Joint D
Fy=0
.5405+DH=0
DH=-.5405 kN
Fx=0
DC-2.74=0
DE=2.74 kN

Joint H
Fy=0
⁡
HCsin12.63°-.5405-2.34sin(12.63)=0
HC= 4.812 kN
Fx=0
⁡
.457+4.812cos12.63°+HA+2.34cos(12.63)=0
HA=-7.436 kN
Joint C
Fy=0
⁡
.6053+4.812sin12.63°+ CA sin(12.63)=0
AC=-7.58 kN

Cost Function Calculation for Truss 1 Section 4

Finding F1:

F1= memberslength50kN∙m∙43*Load Compression or (Load Tension)

F1=.14m50kN∙m∙43*.1025kN+.1.25m50kN∙m∙43*.458
kN+.14m50kN∙m∙1.657kN+1.25m50kN∙m∙43*7.436kN+.14m50kN∙m∙43*.5405kN+.625m50kN∙
m∙2.741kN+0.64m50kN∙m∙.469kN+0.64m50kN∙m∙43*2.34 kN+0.64m50kN∙m∙43*7.582
kN+0.625m50kN∙m∙14.835kN+0.625m50kN∙m∙2.741kN +0.64m50kN∙m∙4.812kN
=0.7611

Finding F2:
F2= # of members27
[Type text] [Type text] Page |8

F2= 1327=.4815

Finding F3:
F3=Average Applied Load1.67kN
F3=4.158 kN1.67 kN=2.4896

Cost Function:
CF=ω1F1+ω2F2+ω3F3
CF=0.60.7611+0.3.4815+0.12.4896=0.82785

Calculation of Lift Force acting at each joint

L(at 0.5m from center span)=0.250.75ρV∞*2V∞SCLbπ*1-2xb2

L(at 0.5m from center span)=0.250.750.1665(91.48)*291.487(0.7732)(5)π*1-2x52

Lat 0.5m from center span=0.4906 kN

Work Split Report

Truss Calculations/Sample Calculations: Asa Berger

Presentation: Sachit Rao, David Hopperton, Asa Berger
Excel sheet: David Hopperton
Executive Summary: Sachit Rao
Introduction: Sachit Rao
Technical Background: Asa Berger
Proposed Trusses: Asa Berger
Results: David Hopperton
Recommendations: David Hopperton