Nature-inspired optimization of tubular joints for metal 3D printing

Table 3 Topology optimization cases (The percentages refer to the volume fractions of the design space—DS)

Objective Volume Stress constraints Loading condition Load (kN)

Minimum Weighted Compliance Vol. fraction = 50% DS (DS-R) σ Von Mises ≤ σ yielding 1.1 Full compression 350
Vol. fraction = 40% DS (DS-M) (380 MPa) for SS 316 L-1.4404 1.2 Full tension
Vol. fraction = 30% DS (DS-L) 1.3 Asymmetric (1)
1.4 Asymmetric (2)

a Structure under permanent loads b Typical joint

Fig. 13 Structural analysis to decide the design load for the topology optimization

FEA analysis, the compression-tension (1) combination pro- Moreover, some local points at the level of the intersection zone
duced the highest stress values. Therefore, we chose this case reached plastic strains. Compared with the benchmark and DS-
for the non-linear analysis. The plastic strain curves were ob- L shapes, the maximum stress values are much lower, and the
tained from the Ramberg-Osgood criteria (Ramberg and stress distribution is smoother in DS-M and DS-R solutions.
Osgood 1934) for true stress-strain curves and respectively con- The DS-R and DS-M joints reduced the stresses, respectively,
verted into effective plastic strain (Total strain–Elastic strain). by 73% and 51%, compared with the unstiffened joint.
The benchmark and light joints (from both symmetric and Figure 22 and Fig. 24 display the stress and strain contour plots
asymmetric groups) displayed the same plasticity scenarios. of all joints. Only for the non-linear analysis of the symmetrical
The central part of the stiffeners first exceeded the characteristic cases, the intermediate volume fraction DS-M (40%DS) was
strain at rupture values for the benchmark and light shapes. introduced. This slight additional weight stiffed the joint, and it

Fig. 14 Thickness and smoothing control tool on Grasshopper

 <. Kanyilmaz et al.

Fig. 15 Density scale of colours

Red Color Density = 1 Loading bearing elements - Permanent Blue Color Density = 0 Non-Loading elements - Can be
Neglected

Fig. 16 TO results for symmetric (S) Benchmark (S) DS-R (50% DS) (S) DS-M (40% DS) (S) DS-L (30% DS)
joint case (S)

Weight = 44.7 kg Weight = 54.6 kg Weight = 48.3 kg Weight = 43.9 kg

reduced the strain values with respect to the benchmark case, as benchmark case). On the other hand, the strain at rupture for
it is shown in Fig. 22. SS-316 L is higher than conventional steel (30% vs 25%). For
Metal additive manufacturing (MAM) alloys present lower this reason, DS-L case achieved slightly higher plastic strain
ultimate strength than the conventional steel (used in the value and a stress reduction, with respect to the benchmark

Fig. 17 TO results for (A) Benchmark (A) DS-R (50% DS) (A) DS-M (40% DS) DS-L (30% DS)
asymmetric joint case (A)

Weight = 37.2 kg Weight = 47.4 kg Weight = 44.2 kg Weight = 38.2 kg

Nature-inspired optimization of tubular joints for metal 3D printing

Fig. 18 FEA results for Unstiffened Benchmark (S) DS-R (50% DS) (S) DS-L (30% DS)
symmetric joint case

case. Table 5 shows a comparison of non-linear performance DS-R joint (V = 50%) required much higher load values
between the benchmark solution and the optimized joints for to reach the collapse, besides having the highest initial
the symmetric case. The analyzed values belong to the refer- stiffness values. Under a load value of 1700 kN, the
ence point (RP) when the joint was loaded at 1700 kN; at this unstiffened benchmark case has already reached the ul-
value the unstiffened case reaches the ultimate strain. timate strain and the highest values of displacements.
Figure 23a shows the force versus displacement The inclusion of stiffeners reduced the displacements,
curves of all the symmetric cases. Figure 23b zooms and higher load values can be applied. The DS-M and
in the linear portion of the force displacement graph DS-L solutions (V = 40% and V = 30% DS) leave the
to compare the initial stiffness in different cases. The linear region and reach the ultimate strain at a very
optimized joints were able to develop higher resistance close loading value with the benchmark case Fig. 24.
and stiffness and stiffness values thanks to the extra The asymmetric cases had a similar trend as well
material distributed as a sort of “natural” stiffeners. (Table 6). In DS-L, there is a reduction of stresses around
The “robustness” property emerges in the case of DS- 49% compared with 36% of the benchmark case. The
R joint, which is optimized accepting a slightly heavier strains in the (A) DS-L solution are three times lower than
weight to have a margin for exceptional loading. The the benchmark joint. It presents an improved non-linear

Fig. 19 FEA results for Unstiffened Benchmark (S) DS-R (50% DS) (S) DS-L (30% DS)
symmetric joint case

55 350

50 300

250
45
MPa
kg

200
40
150
35 100

30 50
Weight PP Stress(RP) PP
 <. Kanyilmaz et al.

Fig. 20 FEA results for Unstiffened Benchmark (A) DS-R (50% DS) (A) DS-L (30% DS)
asymmetric joint case

behaviour and similar percentages of stress decrease with & A joint shape can be optimized to provide the same struc-
the ones achieved by the symmetric cases. For the non- tural performance with the conventionally stiffened joint
linear analysis of the asymmetric joints, it was not neces- without the need for cutting and internal welding opera-
sary to introduce the intermediate volume fraction DS-M tions (“DS-L” joint proves this).
(40%DS) because the lighter solution DS-L (30%DS) pre- & With a slight increase of the joint weight, much more stiff,
sented an enhanced non-linear performance with respect to resistant and robust joints can be obtained (“DS-M”, and
the benchmark case. The (A) DS-R joint shows a signifi- “DS-R” solutions prove this).
cant reduction of stresses and strains, in the order of 73%
and 100%. The natural stiffeners along the horizontal and
vertical direction decrease the plastic strain values sharply.
Figure 25a shows the force versus displacement curves of 4 Simulation of additive manufacturing
all the asymmetric cases and Fig. 25b zooms in the initial process and preliminary cost estimations
linear portion. The same trend of the behaviour as the
symmetric cases is observed. These analyses highlight two A simulation of the additive manufacturing process has been
general conclusions: performed using the software Amphyon (Additive Works

Fig. 21 FEA results for Unstiffened Benchmark (A) DS-R (50% DS) (A) DS-L (30% DS)
asymmetric joint case

55 350

50 300

250
45
MPa
kg

200
40
150
35 100

30 50
Weight PP Stress (RP) PP
Nature-inspired optimization of tubular joints for metal 3D printing

Table 4 Symmetric and

asymmetric joints: stress Linear analysis (S) Symmetric (A) Asymmetric
reduction with respect to the
unstiffened benchmark Cases Stress (MPa) % red./unstiffened Stress (MPa) % red./unstiffened

Unstiffened 308.30 - 316.20 -

Benchmark 96.20 − 68.80 109.50 − 64.48
V = 50% DS 57.60 − 81.32 55.40 − 82.03
V = 30% DS 72.60 − 76.45 68.80 − 77.68

GmbH 2018). The powder bed fusion (PBF) process is con- This is done by normalizing the value for the needed
sidered. The main objective was to understand how the 3D volume of supports for each orientation.
printing stages can affect the mechanical properties of the During the MAM process, high residual stresses are in-
connection. The selected case was the DS-R solution-symmet- duced. For this reason, stress relief treatments are added
ric with a volume fraction = 50% DS. The first three steps of most of the time. The simulations were performed under
the workflow include importing a tessellation file (STL), the the assumptions of homogeneity of the material properties
orientation selection and the support generation (Fig. 26). The and thermal field. The base plate and the MAM part are
typical problems during the build-up of metal powder layers assumed to be made of the same material. The thermal
have been studied, such as residual stresses and distortions. expansion and volume variation occur simultaneously
Post-process treatments can largely eliminate the re- along the part during the phase changes. For this reason,
sidual stresses formed during the process. The orienta- the plastic deformation after the annealing process is not
tion assessment is essential to evaluate the feasibility of affected by the expansion and contraction induced by ad-
a specific part to be printed within the build chamber. ditional stresses. Due to insufficient experimental data, the
This process considers the deformation tendency and material models of Amphyon are restricted to be isotropic
possible cracks that could be generated from a specific and ideally plastic. In the AM process, the yield strength
orientation (Additive Works GmbH 2018). In this case, restricts the residual stress, defining its maximum value
a maximum overhang angle of 45° was selected. It re- and the deformation of the printed part after being released.
quires a minimal amount of support density of 40% and Furthermore, an ideal plasticity mechanism is considered
10% under overhanging and down faces, respectively. to describe the material behaviour (Additive Works GmbH

Unstiffened Benchmark (S) DS-R (50% DS) (S) DS-M (40% DS) (S) DS-L (30% DS)

0 350 700
Von Mises Stress [MPa]

0 0,3
Deformaons and Eq. plasc strains

Fig. 22 Non-linear FEA results for symmetric joint cases