I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue.

Parameter Optimization of Tube Hydroforming

Edina Karabegović 1, Miran Brezočnik 2
1
University of Bihać, Faculty of Technical Eng ineering of Bihać,
2
University of Maribor, Facu lty of Mechanical Eng ineering, Maribor

Abstract:
Tube hydroforming is mostly applied in automotive industry. In this respect, necessity for the procedure
improvement of flu id forming is constant. One of the reasons of its improvement is the procedure performance in optimal
conditions. The process parameters have the direct in fluen ce on quality and optimal of forming procedure. This paper
provides an example of the fluid p ressure optimization in T-shape tube hydroforming. Three types of material have been
analysed, with three wall thickness and three course levels of axial printers. For the optimizat ion, the evolutional method
with applied genetic algorith m (GA) was utilized. The applicat ion of GA is significant in solving of many problems in
engineering practice. The simp licity and adaptability of the genetic algorithm to the engine ering problem results with the
increasing volume of applications in a research work. In this paper we investigated interactions of the internal parameters
of the T tube hydroforming process, towards achieving the GA model fo r the optimal internal pressure, necessary for
hydroforming.

Keywords: Hydroforming, tube, modelling, optimization, parameter,genetic algorithm, T-shape,

1. Introduction
A significant development of unconventional procedure processing has brought up by the market demands for
the rapid part changes in automotive industry, whereas economic and market demands couldn't be satisfied with the
conventional method of automotive part manufacturing. The demand that lays for nowadays products is a very high level
of quality which imp lies product manufacturing, with lesser number of constituent elements and lesser quantity of
materials (thin-walled elements). The automotive manufacturers are faced with this problem, where apart fro m functional,
the securing and ergonomic product demands lays. It justifies the increasing application of hydroforming in automotive and
aviation industry. Hydroforming in that area has showed as satisfying, as besides technological process advantages
(possibilities for amplification, narrowing, tube calibration), th eir advantage is significant in different material applicat ion,
large d imension parts and complex shape. One of the procedure improving methods is the parameter optimizat ion of the
tube hydroforming process. Fro m previous researches, it is observed that th e parameter process analyses of element tube
hydroforming have been carried out by many other researchers. Investigations were related on modelling and the parameter
optimization and considered as the most effective on element tube forming. Fo r modelling a nd optimization, there are
applied analytic methods, numerical methods, evolutional methods, artificial neural networks, finite neural method and
alike. A group of researchers, Giuseppe Ingaro and the others (2009) have been investigating and optimizing th e internal
pressure functioning and printer-counter in Y string hydroforming process in the surging zone, considering the tube wall
thickness change. The applied methods are numerical and experimental. Researchers Nader Abedrabo and the others
(2009) are analysing and optimizing the internal flu id pressure and axial shift in symmetrical tube amp lificat ion, using the
fin ite element method and genetic algorith m. Researchers Zhan Yong and the others (2009) apply FE simu lation and GS
method for the internal pressure optimizat ion and axial shift in element tube hydroforming. The results of all investigations
have influence on the improvement and development of the hydroforming process. This paper provides an example of the
internal flu id pressure optimisation in T-shape tube hydroforming, with three wall th ickness levels (s=1-2-3mm), three
materials (σ0.2 = 164-290-412 N/ mm2 ) and three axial shifts (10-15-20mm). The genetic algorith m method has been applied
(1-3,10 - 15).

2. Tube Hydroforming Process

Almost all recent technological processes are based on utilization of the unconventional procedure of metal
processing with plastic forming. It is significant that about 10% of manufacturing in automotive industry includes fluid
plastic forming of a tube. The demands are corresponding to the techno-economic justification and hydroforming process
utilizat ion. So me of the reasons of the utilizat ion: (4, 7, 8, 10):satisfactory quality of the obtained tube elements ,a single
tube part fabrication, the welding utilization red ucing, possibility of different shape tube forming and wall
thickness,abrication possibility of the high repetition percentage products .

||Issn 2250-3005(online) || |December| 2012 Page 273

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

Some of the examp les of element tube forming are given in Figure 1.

p p
p

Figure 1: Tube elements obtained by hydroforming

2. 1. T tube Hydroforming
In T tube forming, without drain narrowing, the deformation of the drain terminate in the initial forming stage,
while rising of the drain height is obtained by plastic tube deformat ion, i.e. material inflow fro m th e tube to the drain.
F2

d´

p d
F1 F1

Figure 2: T –shape hydroforming tube

A successful tube forming doesn't depend only on dimensions and contact stress distribution, i.e. the stress -strain
states in certain zones of plastic forming part. It also depends on th e requisite forces for the operation achievement and the
other parameters like: material characteristics, tribological conditions and geometry setup. With defining of the block
scheme in process of the input/output sizes, the mutual dependence of the influ ential parameters of the hydroforming
process is recognized (4, 7, 8. 10).

si , h i, l (mm) Hidroforming of tube

0.2 (N/mm2 )
p uc (bar)
d, d‫׳‬, l0, (mm)

Legend:
 pressure of the cylinder, p c (N/mm2 )
C = const.
 stress of the flow, 0.2 (N/mm2 )
 diameter of the tube, d i (mm)
 a device(machine tool)
 diameter of the bulge, d  (mm)
 a tool,
 height of the bulge, h 1 (mm)
 the fluid,
 wall thickness, si (mm)
 the outer diameter d v=const.
 the shift, l (mm)
 an abrasion.
 length of the tube, l0 (mm)

Figure 3: Inner/outer scheme block o f tube hydroforming process sizes

The internal fluid tube pressure (p uc) via T-shape tube is obtained and represents a cost function for modelling and
optimisation of parameter forming process.

2. 2. Tube hydroforming device

Researches of tube hydroforming have been done on the tube hydroforming mach ine in the laboratory of the
Faculty of Technical Engineering of Bihać, Figure 4.

||Issn 2250-3005(online) || |December| 2012 Page 274

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

Figure 4: Experimental analysis device of tube hydroforming process

The working pressure, obtained during the forming process has reached the valuation to 3200 (bar).

2. 3. Measuring equi pment for the internal flui d pressure

The measuring amplifier device, Spider 8 fro m Hottinger Saldwin Messtechnick (HBM ), Germany (10), has been
utilized for the internal fluid p ressure measurement of T -shape tube hydroforming.

Figure 5: Measuring amp lifier device, Spider 8, P3M B – Flu id pressure sensor in the tube

The device contains eight independent measuring channels, where various sensors can be connected, based on principle of
the electronic dimension change. The fluid pressure sensor P3MB, differs with small dimensions, and has been fitted at the
flu id output of the multip lier (10).

2.4. Hydroforming tubes

For the analysis there has been used tubes made of three types of material (alu min iu m alloy, brass and steel), with
the outer diameter d v=20 (mm) and length lo= 80 (mm).

Figure 6: Init ial tube shapes for the aluminiu m alloy, brass and steel

In Table 1 there are the basic material characteristics of which the initial tube shapes have been made.
Table 1. Properties of Materials
TYPES OF MATERIALS
Aluminium alloy Brass Steel
AlMgSi0.5 Cu63Zn Ck10 (soft annealed)
Mechanical Chemical Mechanical properties Chemical Mechanical Chemical
Properties properties properties Properties properties
σ0.2 σm Al 98.5% σ0.2 σm Cu 63% σ0.2 σm C=0.10%
164 175 Mg≤1 164 164 Zn 37% 164 164 Si≤0.10%
N/mm2 N/mm2 Si=0.5 N/mm2 N/mm2 N/mm2 N/mm2 Mn≤0.30%
P≤0.035%
S≤0.035%

||Issn 2250-3005(online) || |December| 2012 Page 275

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

2. 5. Genetic algori thm modelling

Modelling represents a mathematical law description of the parameter process change in certain t ime and space. In
engineering practice problem solving, the genetic algorithm, with its simplicity and repeatability has given solutions for
better problem solving. The genetic algorith m (GA ) is adjusted to the evolutional organism previsions, and as a modelling
shape is necessary for solution improving of cost functions. The classic GA works with the coded variables. The GA seeks
the solution by the dot population (not only by one dot). The linear model has been utilized for the modelling of the first
line (general shape) for the triple -factor interaction:
y  b0  b1s0  b2 0.2  b3 l  b4 s0 0.2  b5 0.2 l  b6 s0 l  b7 s0 0.2 l (1)
The modelling aim is to optimize coefficients b0 , b1, ...,b 7, in the equation 1.
A single organism coding has been derived (9):
   O1
    O2
  O m 
     (2)
  b0 , b1, ... b7 ,  b0 , b1, ... b7 , ..., b0 , b1, ... b7 
    
    
The population P(t) in a generating time t (Figure 5 ), is an organism set (O1 , O2 ,…., Om ), and the single organism is the
dot (solution) in a mu ltid imensional area. A single gene is a coordinate of th e dot (3, 5, 6, 9). The init ial stochastic
generated population in our concrete examp le consists of the m organisms. Each organis m is made of eight genes (see
equation 1):
Oi = b 0 , b 1 , …, b 7 , (3)
Where (i) is a single organism index, i.e. the mathematical model considering the equation 1. The coefficient evolution of
the model fro m equation 1, i.e. overall mode, is carried out in a way g iven in the pseudocode, Figure 7.

Evolutionary algorithm
t : 0
Generate the initial random population of the solution P(t)
evaluate P(t)
repet
P  Pt 1
edit t 
P
evaluate t 1
t : t  1
Till the criteria for the evaluation abruption is not achieved

Figure 7: Pseudo code evolutionary algorith m

In the evaluation phase of population P(t), in the generation time t, an absolute deviation D( i,t) of the individual model
(organism) i for all measurements has to be calculated:
n
Di,t    E( j )  Pi, j  (4)
j 1
Where:
 E(j)-experimental values for j measuring
 P(i,j) prediction value of individual model i, for j measuring
 n- a maximu m nu mber of measurements
The equation (4) is a raw adaptive measure. The aim of the optimizat ion by GA is to achieve the equation solution (4) with
the lack of deviation. But the smallest given absolute solution value doesn't indicate that the smallest percentage model
deviation is achieved. Therefore, the average percentage deviation, for all measurements and for individual model i was
introduced:

i  
 
D i,t
 100% (5)
E f  n

The equation (5) was not utilized as an adequate measure for the evaluation population, but is utilized in order to obtain th e
best organism in the population after the settled optimisation.

||Issn 2250-3005(online) || |December| 2012 Page 276

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

3. The Measurement Results

3. 1. Experimental results
The table shows the values of obtained experimental results of the internal fluid pressure (puc) in the tube, upon T-
string hydroforming.

Table 2: Experimental measurement results of the internal fluid pressure

Number Input values of Output

of parameters values
exp. s0 0.2 l p uc
Nj 2
mm N/mm mm bar
1 1 164 10 432

2 3 164 10 1220

3 1 412 10 1120

4 3 412 10 2940

5 1 164 20 477

6 3 164 20 1298

7 1 412 20 1210

8 3 412 20 3090

9 2 290 15 1603.75

The diagram of experimental values for the fluid press ure of T tube forming, made of steel, wall thickness s0 =2 mm,
obtained by measuring on Spider 8 device.
3 00 0

2 50 0

2 00 0
1m m
1 50 0
2m m

1 00 0 3m m

50 0

0
Al Cu Ck1 0
Figure 8: Fluid pressure in the tube by experimental measurement,fluid pressure in the tube

The flu id pressure in the tube for three types of material has been given in Figure 8.

3. 2. GA model of flui d pressure in the tube

The analysis of the experimental values of the fluid pressure in the tube, with the GA application has been
obtained. The population carried the 100 organisms, and the number of the system motions for GA was 10. The maximu m
number of the generations was 1000.
The best mathematical model obtained by GA for the fluid pressure in the tube has a shape:

||Issn 2250-3005(online) || |December| 2012 Page 277

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

p uc  - 81,1683  43,9205 s 0  0,559208 0.2  0,59908 l  2,02987 s 0 0.2

 0,0132628 0.2 l  0,860965 s 0 l  0,00514024 s 0 0.2 l
(6)
The figure 9 shows the best evolution solution (i.e. the best mathematical model for the fluid pressure in the tube), over the
generations. The solution imp rovement in the in itial phase was rapidly, and after the generation of 100, relatively s mall
solution imp roving has been perceived.

30

25

20
Greška [%]

15

10

0
1 10 100 1000 10000
Generacija

Figure 9: Improvement of the best generation solutions

3. 3. Comparati ve results of the experi mental values by the GA model

Experimental researches of the T tube hydroforming process parameters, the values for the defined outer
parameter are obtained, i.e. tube fluid pressure (p uc). These values are compared with the values (p uc) obtained by the GA
model, Table 3.

Table 3: Co mparat ive data of the experimental and obtained GA model values

Number
of Input values of Output values GA model
exp. parameters

Nj

s0  l p uc p uc

mm N/mm2 mm bar bar

1 1 164 10 432 432,1424

2 3 164 10 1220 1219,8601
3 1 412 10 1120 1119,8733
4 3 412 10 2940 2939,9021
5 1 164 20 477 476,9239
6 3 164 20 1298 1298,7208
7 1 412 20 1210 1210,2943
8 3 412 20 3090 3089,8979
9 2 290 15 1603.750 1483,3960

The mean percentage deviation within experimental and prediction values, obtained by the GA model is Δi = 0.85148%.

||Issn 2250-3005(online) || |December| 2012 Page 278

 I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 2 Issue. 8

4. Conclusion
Plastic forming with the fluid application has been known since the last century, and researches in this area are
significant for the process improvement. With the process parameter optimization of the plastic forming, the techno-
economic justification of the process is achieved.With the mathematical modelling and optimization of the experimental
values for the fluid pressure in the tube, with the applied genetic algorith m (GA), the mathematical model equation is
obtained, with the percentage deviation Δi = 0.85148%. The mathematical model for the fluid pressure in the tube refers to
the optimal work area, describ ing the solution regression for the derived experiment. The solutions in this area ensure the
forming efficacy, with the optimal internal press ure for the generated dimension lines: s 0  1mm  3mm ,
 0.2  164  412 N / mm 2 i l  10mm  20mm .

References
[1] Jurković, M.: Matematičko modeliranje in ženjerskih procesa i sistema, Mašinski faku ltet Bihać, 1999.
[2] Jurković, M ., Tu fekčić, Dž.: Tehnološki procesi projektiranje i modeliranje, Un iverzitet u Tuzli, Mašinski
fakultet, Tuzla, 2000.
[3] Brezočnik, M .: Uporaba genetskega programiran ja v inteligentnih proizvodnih sistema, Fakultet za strojništvo,
Maribor, Slovenija, Tiskarna Tehniških Faku ltet, Maribor, 2000.
[4] Karabegović, E., Jurković, M .: The Analytical Models of hydraulic Forming and Discussion, 8th International
Research/Expert Conference, Trends in theDevelopment of Machinery and Associated Technology, TMT 2004,
Neu m, 2004., p.119-122.
[5] Brezočnik, M . and Kovačić, M.: Integrated Genetic Programming and Genetic A lgorith m Approach to Predict
Surface Roughness, Materials and Manufacturing Processes Vol.18. N 0 . 3, p. 473-489, 2003.
[6] Brezočnik, M., Gusel, L. and Kovačić, M.: Co mparison Between Genetic Algorith m and Genetic Programming
Approach for Modeling the Stress Distribution, Materials and Manufacturing Processes Vol.20. N 0 . 3, p. 497-508,
2005.
[7] Karabegović, E., Jurković, M., Jurković, Z: Theoretical Analysis of Hidroforming of Tubes, Manufacturing
Engineering, vol. 5, N0 . 3/ 2007, p.58-62.
[8] Jurković, M., Karabegović, E., Jurković, Z., Mah mić, M.: Theoretical Analysis of the Tube Hydroforming Process
Parameters and a Suggestion for Experimenta, 11th International Research/Expert Conference, Trends in
theDevelopment of Machinery and Associated Technology, TMT 2007, Hammamet, Tunisia, 2007. p.83-86.
[9] Brezočnik, M., Balić, J.: System for Discovering and Optimizing of Mathematical Moldels Using
GeneticProgramming and Genetic Algorith ms, 8th International DAAAM Sy mposium: Intelligent Manufacturing
& Automation, Dubrovnik, Croatia, 1997, p. 37-38.
[10] Karabegović, E.: Modeliranje i optimizacija parametara procesa hidrooblikovanja tankostijenih cijevnih
elemenata, Do ktorska disertacija, Tehnički fakultet Bihać, ju l 2009.
[11] Altan, T. and Jirathearanat S.: Successful tube hydroforming:Watching parameters, accurately simulat ing the
process yield good results, TPJ-The Tube&Pipe Journal, 2001.
[12] Rosa Di Lo renzo, Giuseppe Ingarao, Francisco Chinesta: Integration of gradient based and response surface
methods to develop a cascade optimisation strategy for Y-shaped tube hydroforming process design, Advances in
Engineering Software, Volu me 41 , Issue 2 (February 2010), ISSN:0965-9978, p. 336-348.
[13] Giuseppe Ingarao, Rosa Di Lo renzo, Fabrizio Micari: Internal pressure and counterpunch action design in Y-
shaped tube hydroforming processes: A multi-objective optimisation approach, Computers and Structures,
Vo lu me 87 , Issue 9-10 (May 2009), ISSN:0045-7949, p. 591-602.
[14] Zhang Yong i dr.: Optimization for Loading Paths of Tube Hydroforming Using a Hybrid Method , Materials and
manufacturing processes, ISSN 1042-6914, 2009, vol. 24, n o 4-6, p. 700-708.
[15] Nader Abedrabbo i dr.: Optimizat ion of a Tube Hydroforming Process,Red Cedar Technology, AB 2027,
Rev.04.09.

||Issn 2250-3005(online) || |December| 2012 Page 279