ArticlePDF Available

Smoothing topology optimization results using pre-built lookup tables

Authors:

Abstract and Figures

Topology optimization techniques are typically performed on a design domain discretized with finite element meshes to generate efficient and innovative structural designs. The optimized structural topologies usually exhibit zig-zag boundaries formed from straight element edges. Existing techniques to obtain smooth structural topologies are limited. Most methods are computationally expensive, as they are performed iteratively with topology optimization. Other methods, such as post-processing methods, are applied after topology optimization, but they cannot guarantee to obtain equivalent structural designs, as the volume and geometric features may be changed. This study presents a new method that uses pre-built lookup tables to transform the shape of boundary elements obtained from topology optimization to create smoothed structural topologies. The new method is developed based on the combination of the bi-directional evolutionary structural optimization (BESO) technique and marching geometries to determine structural topologies and lookup tables, respectively. An additional step is used to ensure that the generated result meets a target volume. A variety of 2D and 3D examples are presented to demonstrate the effectiveness of the new method. This research shows that the new method is highly efficient, as it can be directly added to the last step of topology optimization with a low computational cost, and the volume and geometric features can be preserved in smoothed topologies. Finite element models are also created for original and smoothed structural topologies to show that the structural stiffness can be significantly enhanced after smoothing.
Content may be subject to copyright.
"SUJDMFQVCMJTIFEJO
"EWBODFTJO&OHJOFFSJOH4PGUXBSF
IUUQTEPJPSHKBEWFOHTPGU
aKQQi?BM;iQTQHQ;vQTiBKBxiBQM`2bmHibmbBM;T`2@#mBHiHQQFmTi#H2b
w?BGB-hBM;@l2BG22-umMuQ#-uBJBMsB2-
*2Mi2`7Q`AMMQpiBp2ai`m+im`2bM/Ji2`BHb-a+?QQHQ71M;BM22`BM;-_JAhlMBp2`bBiv-J2H#Qm`M2-jyyR-
mbi`HB#K2#1M;BM22`BM;ai`m+im`2PTiBKBxiBQM_2b2`+?AMbiBimi2-LMDBM;-kRR3yy-*?BM
#bi`+i
hQTQHQ;v QTiBKBxiBQM i2+?MB[m2b `2 ivTB+HHv T2`7Q`K2/ QM /2bB;M /QKBM /Bb+`2iBx2/ rBi? }MBi2 2H2K2Mi
K2b?2b iQ ;2M2`i2 2{+B2Mi M/ BMMQpiBp2 bi`m+im`H /2bB;MbX h?2 QTiBKBx2/ bi`m+im`H iQTQHQ;B2b mbmHHv
2t?B#Bi xB;@x; #QmM/`B2b 7Q`K2/ 7`QK bi`B;?i 2H2K2Mi 2/;2bX 1tBbiBM; i2+?MB[m2b iQ Q#iBM bKQQi? bi`m+@
im`H iQTQHQ;B2b `2 HBKBi2/X JQbi K2i?Q/b `2 +QKTmiiBQMHHv 2tT2MbBp2- b i?2v `2 T2`7Q`K2/ Bi2`iBp2Hv
rBi? iQTQHQ;v QTiBKBxiBQMX Pi?2` K2i?Q/b- bm+? b TQbi@T`Q+2bbBM; K2i?Q/b- `2 TTHB2/ 7i2` iQTQHQ;v
QTiBKBxiBQM- #mi i?2v +MMQi ;m`Mi22 Q#iBMBM; 2[mBpH2Mi bi`m+im`H /2bB;Mb- b i?2 pQHmK2 M/ ;2Q@
K2i`B+ 72im`2b Kv #2 +?M;2/X h?Bb bim/v T`2b2Mib M2r K2i?Q/ i?i mb2b T`2@#mBHi HQQFmT i#H2b iQ
i`Mb7Q`K i?2 b?T2 Q7 #QmM/`v 2H2K2Mib Q#iBM2/ 7`QK iQTQHQ;v QTiBKBxiBQM iQ +`2i2 bKQQi?2/ bi`m+@
im`H iQTQHQ;B2bX h?2 M2r K2i?Q/ Bb /2p2HQT2/ #b2/ QM i?2 +QK#BMiBQM Q7 i?2 #B@/B`2+iBQMH 2pQHmiBQM`v
bi`m+im`H QTiBKBxiBQM U"1aPV i2+?MB[m2 M/ K`+?BM; ;2QK2i`B2b iQ /2i2`KBM2 bi`m+im`H iQTQHQ;B2b M/
HQQFmT i#H2b- `2bT2+iBp2HvX M //BiBQMH bi2T Bb mb2/ iQ 2Mbm`2 i?i i?2 ;2M2`i2/ `2bmHi K22ib i`;2i
pQHmK2X p`B2iv Q7 k. M/ j. 2tKTH2b `2 T`2b2Mi2/ iQ /2KQMbi`i2 i?2 2z2+iBp2M2bb Q7 i?2 M2r K2i?Q/X
h?Bb `2b2`+? b?Qrb i?i i?2 M2r K2i?Q/ Bb ?B;?Hv 2{+B2Mi- b Bi +M #2 /B`2+iHv //2/ iQ i?2 Hbi bi2T Q7
iQTQHQ;v QTiBKBxiBQM rBi? HQr +QKTmiiBQMH +Qbi- M/ i?2 pQHmK2 M/ ;2QK2i`B+ 72im`2b +M #2 T`2@
b2`p2/ BM bKQQi?2/ iQTQHQ;B2bX 6BMBi2 2H2K2Mi KQ/2Hb `2 HbQ +`2i2/ 7Q` Q`B;BMH M/ bKQQi?2/ bi`m+im`H
iQTQHQ;B2b iQ b?Qr i?i i?2 bi`m+im`H biBzM2bb +M #2 bB;MB}+MiHv 2M?M+2/ 7i2` bKQQi?BM;X
E2vrQ`/b, hQTQHQ;v QTiBKBxiBQM- aKQQi?BM;- 6BMBi2 2H2K2Mi MHvbBb- oQHmK2@T`2b2`pBM;-
`2@T`2b2`pBM;
RX AMi`Q/m+iBQM
hQTQHQ;v QTiBKBxiBQM Bb bi`m+im`H QTiBKBx@
iBQM i2+?MB[m2 iQ Q#iBM BMMQpiBp2 bi`m+im`2b i?i
TQbb2bb HB;?ir2B;?i- ?B;?@T2`7Q`KM+2 M/ +Qbi@
2{+B2Mi Ki2`BH +?`+i2`BbiB+b (R)X b bm+?-
iQTQHQ;v QTiBKBxiBQM i2+?MB[m2b ?p2 #22M rB/2Hv
/QTi2/ +`Qbb KMv /Bb+BTHBM2b- BM+Hm/BM; //BiBp2
KMm7+im`BM; (k- j- 9)- `+?Bi2+im`H TTHB+iBQMb
(8- e- d)- 7m`MBim`2 /2bB;Mb (3- N)- M/ #BQHQ;B+H K@
i2`BHb (Ry- RR)X
hQTQHQ;v QTiBKBxiBQM i2+?MB[m2b ivTB+HHv mb2
}MBi2 2H2K2Mi MHvbBb U61V iQ 2pHmi2 bi`m+@
im`H T2`7Q`KM+2X hQ i?Bb 2M/- KMv iQTQH@
Q;v QTiBKBxiBQM i2+?MB[m2b `2[mB`2 i?2 +QMiBMm@
Qmb /2bB;M /QKBM iQ #2 /Bb+`2iBx2/ BMiQ }MBi2 2H@
2K2Mi K2b?2b BM Q`/2` iQ +QM/m+i 61- `272``BM;
*Q``2bTQM/BM; mi?Q`X
1KBH //`2bb, KBF2XtB2!`KBiX2/mXm UuB JBM sB2V
iQ 2H2K2Mi@#b2/ iQTQHQ;v QTiBKBxiBQM TT`Q+?2b
(Rk- Rj)X SQTmH` 2H2K2Mi@#b2/ TT`Q+?2b BM@
+Hm/2 i?2 bQHB/ BbQi`QTB+ Ki2`BH rBi? T2MHBxiBQM
UaAJSV K2i?Q/ (R9- R8) M/ i?2 #B@/B`2+iBQMH 2pQ@
HmiBQM`v bi`m+im`H QTiBKBxiBQM U"1aPV K2i?Q/
(Re- Rd)X h?2 F2v /Bz2`2M+2 #2ir22M i?2b2 irQ
K2i?Q/b Bb i?2 /2MbBiv `2T`2b2MiiBQM Q7 i?2B` 2H@
2K2MibX h?2 aAJS K2i?Q/ HHQrb i?2 /2MbBiB2b
Q7 2H2K2Mib iQ p`v +QMiBMmQmbHv #2ir22M y M/ R
(R3)- r?BH2 i?2 "1aP K2i?Q/ QMHv HHQrb i?2 /2M@
bBiB2b Q7 2H2K2Mib iQ #2 2Bi?2` y Q` R (Re)X b
`2bmHi- QTiBKH bi`m+im`H iQTQHQ;B2b Q#iBM2/ 7`QK
i?2 aAJS K2i?Q/ Kv TQbb2bb dz;`2vǴ #QmM/`B2b
+Q``2bTQM/BM; iQ dzbQ7iǴ 2H2K2Mibc "1aP bi`m+im`H
iQTQHQ;B2b Kv TQbb2bb xB;@x; #QmM/`B2b 7Q`K2/
7`QK bi`B;?i 2H2K2Mi 2/;2b (RN)X Ai Bb rQ`i? TQBMi@
BM; Qmi i?i i?2 `2bQHmiBQM Q7 i?2b2 2H2K2Mi@#b2/
iQTQHQ;B2b +M #2 bB;MB}+MiHv BM+`2b2/ mbBM;
H`;2 MmK#2` Q7 2H2K2Mib- r?B+? +M +`2i2 }M2`
S`2T`BMi bm#KBii2/ iQ /pM+2b BM 1M;BM22`BM; aQ7ir`2 m;mbi d- kykk
;`2v M/ xB;@x; #QmM/`B2b- M/ H2/ iQ pBbmHHv
bKQQi? #QmM/`B2bX
hQ ;2M2`i2 bKQQi? bi`m+im`H iQTQHQ;B2b- ;`2v
M/ xB;@x; #QmM/`B2b ivTB+HHv `2[mB`2 7m`i?2`
KQ/B}+iBQMb (ky)X hrQ /QKBMMi bi`i2;B2b BM `2@
+2Mi bim/B2b `2 ;2QK2i`B+HHv +QMbi`BM2/ iQTQHQ;v
QTiBKBxiBQM K2i?Q/b M/ TQbi@T`Q+2bbBM; K2i?@
Q/bX h?2 7Q`K2` bi`i2;v 2KTHQvb ;2QK2i`B+ QT@
2`iBQMb /m`BM; i?2 QTiBKBxiBQM T`Q+2bb- r?2`2
bKQQi? bi`m+im`H iQTQHQ;B2b `2 ;2M2`i2/ BM 2+?
Bi2`iBQM (RN- kR- kk- kj- k9- k8- ke)X >Qr2p2`-
bm+? K2i?Q/b `2 +QKTmiiBQMHHv 2tT2MbBp2 /m2
iQ i?2 ;2M2`iBQM Q7 mMM2+2bb`v bKQQi?2/ iQTQHQ@
;B2b i?`Qm;?Qmi i?2 QTiBKBxiBQM (ky)X h?2 Hii2`
bi`i2;v 7Q+mb2b QM bKQQi?BM; i?2 ;2M2`i2/ QT@
iBKH iQTQHQ;B2b- bm+? b BbQ@bm`7+2 2ti`+iBQM H@
;Q`Bi?Kb (kd- k3- kN- jy) M/ GTH+BM bKQQi?BM;
i2+?MB[m2b (jR- jk)X >Qr2p2`- bm+? K2i?Q/b +M@
MQi +`2i2 bKQQi?2/ KQ/2Hb i?i `2 i`mHv 2[mBp@
H2Mi iQ i?2 +imH iQTQHQ;v QTiBKBxiBQM `2bmHib-
b i?2 pQHmK2 Q` ;2QK2i`B+ 72im`2b +MMQi #2 T`2@
b2`p2/ BM i?2 ;2M2`i2/ KQ/2Hb (ky)X "++BHB 2i HX
(jj) ?p2 T`QTQb2/ pQHmK2@T`2b2`p2/ bKQQi?BM;
K2i?Q/ i?i +M Q#iBM 2[mBpH2Mi bKQQi? KQ/@
2Hb- #mi i?2B` K2i?Q/ `2bmHib BM T2`7Q`KM+2
HQbb Q7 #Qmi kWX _2+2MiHv- GB 2i HX (j9) ?p2
mb2/ pQHmK2@T`2b2`p2/ bKQQi?BM; K2i?Q/ iQ Q#@
iBM bKQQi? bi`m+im`H iQTQHQ;B2bX >Qr2p2`- i?2B`
K2i?Q/ ?b MQi #22M 2ti2M/2/ iQ +QKKQMHv mb2/
B``2;mH` 2H2K2Mib bm+? b i2i`?2/`X
6Q` T`+iB+H TTHB+iBQMb- +QKTmi2`@B/2/ /2bB;M
U*.V KQ/2Hb `2 ivTB+HHv M22/2/X >Qr2p2`- i?2
bKQQi?2/ iQTQHQ;B2b 7BH iQ #2 /B`2+iHv +QMp2`i2/
BMiQ *. KQ/2Hb- 2bT2+BHHv BM j. +b2b (j8)X h?2v
`2 `2+Q;MBx2/ b MQM@KMB7QH/ KQ/2Hb BM *. bQ7i@
r`2 #2+mb2 i?2B` pQHmK2i`B+ K2b?2b BM+Hm/2 +QBM@
+B/2Mi 7+2b M/ bBM;mH`BiB2b (jj- j8)X h?2 KQbi
bi`B;?i7Q`r`/ bi`i2;v iQ Qp2`+QK2 i?Bb T`Q#H2K
Bb iQ `2#mBH/ bKQQi?2/ KMB7QH/ *. KQ/2H (j8)X
AM /QBM; bQ- i?2 ;2M2`i2/ KQ/2H Kv /Bz2` 7`QK
i?2 +imH iQTQHQ;v QTiBKBxiBQM `2bmHi `2;`/BM;
i?2 pQHmK2 M/ bi`m+im`H T2`7Q`KM+2X
h?Bb TT2` T`2b2Mib MQp2H TQbi@T`Q+2bbBM;
K2i?Q/ UQ` bKQQi?BM; K2i?Q/V iQ +`2i2 bKQQi?2/
KMB7QH/ *. KQ/2Hb 7`QK 2H2K2Mi@#b2/ iQTQH@
Q;v QTiBKBxiBQM `2bmHibX h?2 T`QTQb2/ K2i?Q/
+M i`Mb7Q`K i?2 b?T2 Q7 2H2K2Mib mbBM; T`2@#mBHi
HQQFmT i#H2b bi`B;?i 7i2` i?2 iQTQHQ;v QTiBKBx@
iBQM Bb +QKTH2i2/- r?B+? +M 2z2+iBp2Hv M/ +QMp2@
MB2MiHv bKQQi? i?2 xB;@x; #QmM/`B2bX h?2 M2r
K2i?Q/ Bb /2p2HQT2/ +QMbB/2`BM; i?2 2[mBpH2M+v
Q7 i?2 Q`B;BMH M/ bKQQi?2/ bi`m+im`H iQTQHQ@
;B2b- K2MBM; i?i i?2 ;2QK2i`B+ 72im`2b M/ pQH@
mK2 +M #2 T`2b2`p2/ 7i2` bKQQi?BM;X a2+iBQM
k }`bi #`B2~v bmKK`Bx2b i?2 K2i?Q/ Q7 iQTQHQ;v
QTiBKBxiBQM M/ /2b+`B#2b i?2 /2iBHb Q7 i?2 M2r
bKQQi?BM; K2i?Q/X a2+iBQM j /2KQMbi`i2b i?2 BK@
TH2K2MiiBQM Q7 i?2 M2r K2i?Q/ M/ +QKT`2b i?2
/Bz2`2M+2b #2ir22M Q`B;BMH M/ bKQQi?2/ bi`m+@
im`H iQTQHQ;B2bX a2+iBQM 9 /Bb+mbb2b i?2 2z2+i Q7
/Bz2`2Mi K2b? ivT2b M/ bBx2b BM TQi2MiBH T`+iB@
+H TTHB+iBQMb- 7QHHQr2/ #v +QM+HmbBQM BM a2+iBQM
8X
kX J2i?Q/QHQ;v
kXRX "1aP bi`m+im`H iQTQHQ;v QTiBKBxiBQM
h?Bb bim/v mb2b i?2 bQ7i@FBHH "1aP K2i?Q/ iQ
T2`7Q`K bi`m+im`H iQTQHQ;v QTiBKBxiBQM (Re- je)X
6Q` /Bb+`2iBx2/ /2bB;M /QKBM rBi? N2H2K2Mib-
i?2 +QKTHBM+2 KBMBKBxiBQM UQ` biBzM2bb KtB@
KBxiBQMV T`Q#H2K bii2K2Mi bm#D2+i iQ pQHmK2
+QMbi`BMi +M #2 r`Bii2M b 7QHHQrbX
Minimize :C=1
2UTKU URV
Subject to :V
N
i=1
Vixi=0 UR#V
xi=xmin Q` 1UR+V
r?2`2 CBb i?2 +QKTHBM+2 Q7 i?2 bi`m+im`2- KBb i?2
;HQ#H biBzM2bb Ki`Bt- UBb i?2 /BbTH+2K2Mi p2+@
iQ`- VBb i?2 i`;2i bi`m+im`H pQHmK2- M/ ViBb
i?2 pQHmK2 Q7 i?2 i@i? 2H2K2MiX h?2 /2bB;M p`B@
#H2- xi- /2i2`KBM2b r?2i?2` i?2 i@i? 2H2K2Mi Bb
bQHB/ Uxi=1V Q` pQB/ Uxi=xmin =0.001VX
h?2 Ki2`BH KQ/2H Q7 2H2K2Mib Bb /2}M2/ b
7mM+iBQM Q7 i?2 2H2K2Mi /2MbBivX
E(xi)k=xipE0UkV
r?2`2 kBb i?2 +m``2Mi Bi2`iBQM MmK#2`- E(xi)kBb
i?2 uQmM;Ƕb KQ/mHmb Q7 i@i? 2H2K2Mi- E0Bb i?2 /2@
bB;M uQmM;Ƕb KQ/mHmb Q7 2H2K2Mib- M/ p=3Bb i?2
T2MHiv 2tTQM2MiX
hQ `2Hi2 i?2 bi`m+im`H T2`7Q`KM+2 Q7 2H2K2Mib
iQ i?2 QTiBKBxiBQM Q#D2+iBp2- i@i? 2H2K2MiH b2MbB@
iBpBiv MmK#2`- Se(i)- +M #2 /2}M2/ b
k
Se(i)=1
p
∂C
∂xi
=
1
2uT
iK0
iuiwhen xi=1
xp1
min
2uT
iK0
iuiwhen xi=xmin
UjV
r?2`2 K0
iBb i?2 2H2K2MiH biBzM2bb Ki`Bt- M/ ui
Bb i?2 /BbTH+2K2Mi p2+iQ`X
hQ BKT`Qp2 i?2 +QMp2`;2M+2 Q7 i?2 "1aP i2+?@
MB[m2- ?BbiQ`B+H b2MbBiBpBiv MmK#2`b +M #2 p2`@
;2/- r?2`2
ˆ
Se(i)=Sk
e(i)+Sk1
e(i)
2U9V
h?2 i`;2i pQHmK2 7Q` i?2 M2ti Bi2`iBQM- Vk+1-
/2T2M/b QM i?2 +m``2Mi pQHmK2- Vk- M/ bT2+B}2/
2pQHmiBQM`v `iBQ- ert- r?2`2
Vk+1 =Vk(1 ±ert)U8V
LQi2 i?i i?`2b?QH/ b2MbBiBpBiv MmK#2`- Sth - Bb
+H+mHi2/ #b2/ QM Vk+1 mbBM; i?2 K2i?Q/ /2@
b+`B#2/ BM (je)X 1H2K2Mib i?i TQbb2bb i?2 b2MbBiBp@
Biv MmK#2` ?B;?2` M/ HQr2` i?M Sth `2 i?2M T`2@
b2`p2/ M/ `2KQp2/- `2bT2+iBp2HvX h?2 QTiBKBxiBQM
T`Q+2bb Bb `2T2i2/ mMiBH `2+?BM; i?2 i`;2i pQHmK2
M/ biBb7vBM; i?2 +QMp2`;2M+2 +`Bi2`B (Re)X
kXkX aKQQi?BM; T`2T`iBQM
PM+2 i?2 2H2K2Mi@#b2/ bi`m+im`H iQTQHQ;v Bb
Q#iBM2/ mbBM; i?2 K2i?Q/ /2b+`B#2/ BM a2+iBQM
kXR- i?2 MQ/H b2MbBiBpBiv }2H/- Sn- Bb +H+mHi2/ 7Q`
bKQQi?BM; T`2T`iBQMX h?Bb +M #2 +?B2p2/ #v
2KTHQvBM; }Hi2` QM 2H2K2Mib (jd- j3)- b b?QrM BM
6B;m`2 R- r?2`2 i?2 }Hi2` +Qp2`b +B`+mH` /QKBM
prBi? +2Mi2` Q7 MQ/2 pX h?2 }Hi2`BM; b+?2K2 Bb
;Bp2M b
Sn(p)=Nn
q=1(rmin rpq )ˆ
Se(q)
Nn
q=1(rmin rpq )UeV
r?2`2 Sn(p)Bb i?2 b2MbBiBpBiv MmK#2` Q7 MQ/2 pX
NnBb i?2 MmK#2` Q7 2H2K2Mi +2Mi2`b +Qp2`2/ #v
i?2 }Hi2`X rmin Bb i?2 }Hi2` `/Bmb- M/ rpq Bb i?2
/BbiM+2 #2ir22M i?2 +2Mi2` Q7 2H2K2Mi qM/ MQ/2
pX Ai Bb MQi2/ i?i }Hi2`BM; HH 2H2K2Mib ;Bp2b MQ/H
b2MbBiBpBiv }2H/- SnX
6BMHHv- M BbQ@pHm2- Siso- Bb bT2+B}2/ #2ir22M
i?2 KtBKmK M/ KBMBKmK Sn- i?i Bb- rBi?BM i?2
MQ/H b2MbBiBpBiv }2H/- 7Q` 7m`i?2` bKQQi?BM; T`Q+2@
/m`2b b /2b+`B#2/ #2HQrX
Mesh elements
Filter:
Element centers
Selected centers
p
:
p
min
r
6B;m`2 R, P#iBMBM; i?2 MQ/H b2MbBiBpBiv }2H/ #v 2KTHQvBM;
}Hi2` QM K2b? 2H2K2MibX h?Bb };m`2 b?Qrb i?i i?2 }Hi2`
+Qp2`b +B`+mH` /QKBM prBi? +2Mi2` Q7 MQ/2 p- r?B+?
?b `/Bmb Q7 rminX
kXjX aKQQi?BM; k. bi`m+im`H iQTQHQ;B2b
h?Bb bm#@b2+iBQM /2b+`B#2b K2i?Q/ iQ /B`2+iHv
i`Mb7Q`K i?2 b?T2 Q7 2H2K2Mib ;2M2`i2/ 7`QK
iQTQHQ;v QTiBKBxiBQM- bQ i?i bKQQi? k. bi`m+@
im`H iQTQHQ;B2b +M #2 +QMp2MB2MiHv +`2i2/X 6Q`
k. bi`m+im`H iQTQHQ;v- Bib MQ/H b2MbBiBpBiv }2H/
7Q`Kb H2p2H@b2i 7mM+iBQM (k8- jN)X aT2+B7vBM; Siso
/2i2`KBM2b i?2 bQHB/ /QKBM S- pQB/ /QKBM V
M/ #QmM/`v Γ- b b?QrM BM 6B;m`2 kU+VX 6Q` MQ/2
pBM i?2 /2bB;M /QKBM- Bib (x, y, z)TQbBiBQM- T- `2@
Hi2/ iQ Siso +M #2 /2b+`B#2/ b
pV, if Sn(p)>S
iso
pΓ, if Sn(p)=Siso
pS, if Sn(p)<S
iso
UdV
St(p)Bb BMi`Q/m+2/ iQ /2i2`KBM2 r?2i?2` Sn(p)
+Q``2bTQM/b iQ bQHB/- pQB/- Q` #QmM/`v mbBM; y M/
R QMHv- r?2`2
St(p)=0, if Sn(p)>S
iso
1, if Sn(p)Siso
U3V
h?Bb 2[miBQM `2T`2b2Mib i?i pQB/ TQBMib `2 i?Qb2
rBi? St(p)=0- M/ #Qi? bQHB/ M/ #QmM/`v TQBMib
TQbb2bb St(p)=1X
hQ i`Mb7Q`K i?2 b?T2 Q7 bi`m+im`H 2H2K2Mib-
i?Bb bim/v mb2b HQQFmT i#H2b iQ T`2@/2}M2 HH
TQbbB#H2 i`Mb7Q`KiBQM `2bmHib Q7 mMBi 2H2K2MiX
6Q` [m/`M;mH` 2H2K2Mib b?QrM BM 6B;m`2 kUV-
i?2 +Q``2bTQM/BM; HQQFmT i#H2 +M #2 #mBHi #v
KQ/B7vBM; i?2 J`+?BM; a[m`2b UJaV H;Q`Bi?K
(9y- 9R)X JQ`2 bT2+B}+HHv- i?2 BbQ@HBM2b BM i?2
j
Q`B;BMH K`+?BM; b[m`2b `2 `2TH+2/ rBi? T`2@
/2}M2/ i`BM;mH` Q` [m/`M;mH` K2b?2b iQ `2T`2@
b2Mi bQHB/ /QKBMb- b b?QrM BM 6B;m`2 kU#VX lbBM;
1[miBQM 3- 2H2K2MiH MQ/2b +M #2 bbB;M2/ rBi?
St(p)=1Q` 07Q` BMi2`MH U`2/V M/ 2ti2`MH U;`2vV
TQBMib- `2bT2+iBp2HvX //BiBQMH #QmM/`v UQ`M;2V
TQBMib `2 i?2M +`2i2/ QM i?2 KB/TQBMi #2ir22M BM@
i2`MH M/ 2ti2`MH TQBMibX Ji?2KiB+HHv- i?2`2
`2 Re TQbbB#H2 i`Mb7Q`KiBQM `2bmHib Q7 [m/`M@
;mH` 2H2K2Mi- +H+mHi2/ mbBM;
Cs(i)=
Ni
j=0
St(nj)2jUNV
r?2`2 Cs Bb i?2 +b2 MmK#2`- NiBb i?2 MmK#2` Q7
MQ/2b Q7 2H2K2Mi i- M/ St(nj)`2T`2b2Mib i?2 0/1
bii2 Q7 ji? MQ/2 BM 2H2K2Mi ic i?2 MmK#2` Q7 TQbbB@
#H2 +b2b Bb +H+mHi2/ b 2Ni=2
4= 16X >Qr2p2`-
i?2`2 `2 +imHHv R3 K`+?BM; b[m`2b HBbi2/ BM i?2
HQQFmT i#H2- b b?QrM BM 6B;m`2 kU#VX
Ai b?QmH/ #2 MQi2/ i?i i?2 +H+mHi2/ Cs pHm2
Kv +Q``2bTQM/ iQ KQ`2 i?M QM2 K`+?BM; ;2QK@
2i`v (9k)X am+? bT2+BH +b2b Q++m` r?2M i?2 /B;Q@
MH +Q`M2`b `2 /2i2`KBM2/ b BMi2`MH U`2/V TQBMib-
rBi? i?2 BMi2`K2/Bi2 `2;BQM bm``QmM/2/ #v #QmM/@
`v UQ`M;2V TQBMib #2BM; 2Bi?2` }HH2/ Q` p+Mi-
`2bmHiBM; BM +QMM2+i2/ M/ /Bb+QMM2+i2/ bQHB/ /Q@
KBMb- `2bT2+iBp2HvX AM i?2 T`2b2Mi 2tKTH2- bT2+BH
+b2b `2 Cs =5M/ Cs = 10- rBi? 2+? +QMiBMBM;
irQ TQbbB#H2 K`+?BM; b[m`2bX h?2`27Q`2- i?2 iQiH
MmK#2` Q7 TQbbB#H2 K`+?BM; b[m`2b Bb 16+ 2 = 18X
AM i?Bb bim/v- i?2 b2H2+iBQM Q7 K`+?BM; b[m`2b 7Q`
bT2+BH +b2b +M #2 bBKTHv /2i2`KBM2/ #b2/ QM
i?2 p2`;2/ 2H2K2MiH b2MbBiBpBiv- ˆ
Se(i)Ub22 1[m@
iBQM 9VX aT2+B}+HHv- ˆ
Se(i)Siso M/ ˆ
Se(i)>S
iso
+Q``2bTQM/ iQ +QMM2+i2/ M/ /Bb+QMM2+i2/ bQHB/ /Q@
KBMb- `2bT2+iBp2HvX
L2ti- HH 2H2K2Mib BM i?2 /2bB;M /QKBM `2 `2@
b?T2/ mbBM; i?2 KQ/B}2/ Ja HQQFmT i#H2X 6B@
MHHv- 7Q` 2+? #QmM/`v UQ`M;2V TQBMi t#2ir22M
M BMi2`MH U`2/V TQBMi AM/ M 2ti2`MH U;`2vV
TQBMi B- Bib (x, y, z)+QQ`/BMi2 Bb `2HQ+i2/ iQ
M2r TQbBiBQM tnew mbBM; i?2 7QHHQrBM; HBM2` BMi2`@
TQHiBQM K2i?Q/- b b?QrM BM 6B;m`2 kX
tnew =A+Siso Sn(A)
Sn(B)Sn(A)(BA)URyV
r?2`2 M/ "`2 i?2 (x, y, z)HQ+iBQMb Q7 i?2 BM@
i2`MH TQBMi AM/ 2ti2`MH TQBMi B- `2bT2+iBp2HvX
Sn(A)M/ Sn(B)`2 }Hi2`2/ MQ/H b2MbBiBpBiv MmK@
#2`b Q7 TQBMib AM/ B- `2bT2+iBp2HvX Ai Bb MQi2/
i?i `2HQ+iBM; HH #QmM/`v TQBMib ;Bp2b bKQQi?
#QmM/`vX
6m`i?2`KQ`2- i?2 BMi2`b2+iBQMb #2ir22M i?2 ;2M@
2`i2/ bi`m+im`H iQTQHQ;v M/ i?2 /2bB;M /QKBM
`2 mM+?M;2/ 7i2` bKQQi?BM;X h?Bb Bb #2+mb2 i?2
2H2K2Mib HQ+i2/ QM i?2 #QmM/`B2b Q7 i?2 /2bB;M
/QKBM TQbb2bb St =1QM i?2B` MQ/2bX
hQ bmKK`Bx2/- 1[miBQM e +`2i2b MQ/H b2M@
bBiBpBiv }2H/- Sn- 7Q` k. bi`m+im`H iQTQHQ;vX h?2
bT2+B}2/ BbQ@pHm2- Siso- /2i2`KBM2b i?2 +QM/BiBQM
Q7 2+? MQ/2- r?2`2 St =y Q` R- b /2b+`B#2/ BM
1[miBQMb 3X AM /QBM; bQ- 2H2K2Mib +M #2 `2b?T2/
mbBM; i?2 KQ/B}2/ Ja HQQFmT i#H2 #b2/ QM 1[m@
iBQM NX h?2M- i?2 b?T2b Q7 i?2 i`Mb7Q`K2/ 2H2@
K2Mib `2 7m`i?2` mT/i2/ mbBM; HBM2` BMi2`TQH@
iBQM- b b?QrM BM 1[miBQM RyX h?Bb k. bKQQi?BM;
Bb /2KQMbi`i2/ KQ`2 +H2`Hv BM 6B;m`2 kX
kX9X aKQQi?BM; j. bi`m+im`H iQTQHQ;B2b
j. bi`m+im`H iQTQHQ;B2b +M HbQ #2 bKQQi?2/
#b2/ QM /Bz2`2Mi HQQFmT i#H2b- /2T2M/BM; QM i?2
K2b? 2H2K2Mi ivT2X h?2 irQ KQbi +QKKQM HQQFmT
i#H2b `2 J`+?BM; *m#2b UJ*V (9j) M/ J`+?BM;
h2i`?2/` UJhV (99) 7Q` ?2t?2/`H M/ i2i`?2@
/`H K2b? 2H2K2Mib- `2bT2+iBp2HvX *QKT`2/ rBi?
i?2 Q`B;BMH J* M/ Jh K2i?Q/b- i?2 j. i`Mb@
7Q`KiBQM +b2b `2 T`2@/2}M2/ #v +HQb2/ bm`7+2
K2b?2b iQ `2T`2b2Mi bQHB/ /QKBMb- b b?QrM BM 6B;@
m`2 jUV M/ U#VX Ai Bb rQ`i? TQBMiBM; Qmi i?i i?2 3@
MQ/2b ?2t?2/`H 2H2K2Mi +M #2 i`Mb7Q`K2/ BMiQ
key TQbbB#H2 +QM};m`iBQMb- #mi 6B;m`2 jUV QMHv
b?Qrb Rd mMB[m2 +b2bc i?2 `2KBMBM; 26017 = 243
+b2b +M #2 Q#iBM2/ #v `QiiBM; M/ KB``Q`BM;
i?2b2 Rd +b2bX KQM; i?2b2 Rd mMB[m2 +b2b-
Cs = 65 Bb i?2 bT2+BH +b2- r?2`2 Bi +QMiBMb irQ
TQbbB#H2 K`+?BM; +m#2b- ?B;?HB;?i2/ BM 6B;m`2 jUVX
aBKBH` iQ i?2 T`2pBQmb k. 2tKTH2- +QMM2+i2/ M/
/Bb+QMM2+i2/ bQHB/ /QKBMb 7Q` i?2 T`2b2Mi j. +b2
`2 b2H2+i2/ r?2M ˆ
Se(i)Siso M/ ˆ
Se(i)>S
iso-
`2bT2+iBp2HvX
h?2 2MiB`2 bKQQi?BM; T`Q+2bb 7Q` j. bi`m+im`H
iQTQHQ;B2b Bb bBKBH` iQ i?2 rQ`F~Qr /2p2HQT2/ 7Q`
k. +b2bX 6B`bi- SnBb +H+mHi2/ 7Q` HH MQ/2b mbBM;
1[miBQM eX a2+QM/- Siso Bb bT2+B}2/ iQ bbB;M HH
MQ/2b rBi? St =y Q` R #b2/ QM Sn- b /2b+`B#2/ BM
3X h?B`/- HH 2H2K2Mib `2 `2b?T2/ 7`QK i?2 HQQFmT
i#H2 ++Q`/BM; iQ St mbBM; 1[miBQM NX 6Qm`i?- i?2
b?T2 Q7 i`Mb7Q`K2/ 2H2K2Mib `2 7m`i?2` KQ/B}2/
mbBM; 1[miBQM RyX .Bz2`2Mi iQ k. bKQQi?BM;- M
//BiBQMH bi2T Bb `2[mB`2/ BM j. +b2bX
9
Linear
interpolation
Replace
(a) (b) (c)
A
B
t
A
B
Boundary
condition
Design
domain
Loading
condition
Structural
topology
Smoothed
structural
topology
Internal points External points
Boundary points
0
0Cs
1
32
1Cs 2Cs 3Cs 4Cs
6Cs 7Cs 8Cs
9Cs 10Cs 11Cs 12Cs
13Cs 14Cs 15Cs
Modified Marching Squares
S
:
Solid:
V
:
Void:
*
Boundary:
t
new
5Cs
ˆ( )
e iso
S
i Sd
ˆ( )
e iso
S
i S!
ˆ( )
e iso
S
i Sd
ˆ( )
e iso
S
i S!
6B;m`2 k, aKQQi?BM; i?2 bi`m+im`H iQTQHQ;v Q7 k. +MiBH2p2`X UV AMBiBH b2imT Q7 i?2 QTiBKBxiBQM T`Q#H2K- r?2`2 i?2 /2bB;M
/QKBM Bb /Bb+`2iBx2/ rBi? [m/`M;mH` 2H2K2MibX U#V h?2 HQQFmT i#H2 Bb #mBHi mbBM; i?2 KQ/B}2/ J`+?BM; a[m`2b H;Q`Bi?K
i?i ?b R3 TQbbB#H2 +QM};m`iBQMbX U+V h?2 i`Mb7Q`KiBQM `2bmHi Bb Q#iBM2/ i?`Qm;? `2b?TBM; HH Q`B;BMH 2H2K2MibX
b b?QrM BM 6B;m`2 9- i?Bb bim/v `2KQp2b HH BM@
i2`MH 7+2b Q7 j. bi`m+im`H iQTQHQ;B2bX Ai b?QmH/
#2 ?B;?HB;?i2/ i?i HH i`Mb7Q`K2/ 2H2K2Mib `2
+HQb2/ bm`7+2 K2b?2bX h?2v 7Q`K bKQQi?2/ +QM@
};m`iBQM i?i BM+Hm/2b +QBM+B/2Mi 7+2b- b b?QrM
BM 6B;m`2 9UVX h?2b2 +QBM+B/2Mi 7+2b `2 B/2MiB@
}2/ b BMi2`MH 7+2bX _2KQpBM; HH BMi2`MH 7+2b
;Bp2b j. bm`7+2 K2b?2b- b b?QrM BM 6B;m`2 9U#V-
r?B+? +M #2 +QMbB/2`2/ b *.@+QKTiB#H2 KM@
B7QH/ K2b?2bX
kX8X am`7+2 `2K2b?BM;
h?2 T`Q/m+ib Q7 i?2 #Qp2 k. M/ j. bKQQi?@
BM; T`Q+2bb2b `2 bm`7+2 K2b?2b (98)X b b?QrM BM
6B;m`2 8- i?2 [mHBiv Q7 i?2b2 bm`7+2b +M #2 BK@
T`Qp2/ i?`Qm;? dzbm`7+2 `2K2b?BM;ǴX Ai b?QmH/ #2
MQiB+2/ i?i KMv `2K2b?BM; i2+?MB[m2b `2 /2p2H@
QT2/ #b2/ QM KMB7QH/ *. KQ/2Hb HBF2 bm`7+2
K2b?2b (9e)X
h?Bb bim/v mb2b i?2 BM+`2K2MiH `2K2b?BM; H;Q@
`Bi?K /2b+`B#2/ BM (98- 9d- 93)X h?Bb H;Q`Bi?K +M
#2 7QmM/ BM KMv *. bQ7ir`2- bm+? b _?BMQ dX
h?2 `2K2b?BM; H;Q`Bi?K `2T2i2/Hv KQ/B}2b i?2
K2b? 2/;2b M/ p2`iB+2b #b2/ QM HQ+H QT2`iBQMb
mMiBH HH 2/;2 H2M;i?b `2 +HQb2 iQ i?2 T`2b+`B#2/ i`@
;2i pHm2c i?2 HQ+H QT2`iBQMb BM+Hm/2 bTHBiiBM; HQM;
2/;2b- +QHHTbBM; b?Q`i 2/;2b- ~BTTBM; K2b? 2/;2b-
M/ `2HQ+iBM; p2`iB+2bX b `2bmHi Ub22 6B;m`2 8V-
i?2 2H2K2Mib QM i?2 bm`7+2 +M #2 `2@/Bbi`B#mi2/
mMB7Q`KHv mbBM; i`BM;mH` 2H2K2Mib iQr`/ i?2 i`@
;2i 2/;2 H2M;i?X Ai +M HbQ #2 b22M i?i i?2 ;2Q@
K2i`B+ 72im`2b- bm+? b +Q`M2`b M/ b?`T 2/;2b-
`2 T`2b2`p2/ /m`BM; i?2 `2K2b?BM; T`Q+2bbX
kXeX S`2b2`pBM; `2 Q` pQHmK2
h?2 bKQQi?2/ +QM};m`iBQMb Q#iBM2/ 7`QK i?2
#Qp2 K2i?Q/b +M #2 mM/2`biQQ/ b BbQ@b2MbBiBpBiv
KQ/2Hb- b i?2v `2 /2i2`KBM2/ #b2/ QM i?2 Siso
8
(b)
(a)
01
5
4
3
2
6
7
Modified Marching Cubes
1
0
2
3
Modified Marching Tetrahedra
58Cs 67Cs 74Cs
105Cs
51Cs
50Cs
5Cs 3Cs
1Cs
113Cs 165Cs 177Cs 178Cs
255Cs
0Cs
1Cs
2Cs 3Cs 4Cs 5Cs 6Cs 7Cs
8Cs 9Cs
10Cs
11Cs
12Cs 13Cs 14Cs 15Cs
ˆ( )
e iso
S
i S!
0Cs
65Cs
ˆ( )
e iso
S
i Sd
6B;m`2 j, JQ/B}2/ j. HQQFmT i#H2bX h?2 `2/ MmK#2`b `2T`2b2Mi i?2 BM/B+2b Q7 i?2 2H2K2MiH MQ/2bX UV Rd mMB[m2 +b2b BM i?2
KQ/B}2/ J`+?BM; *m#2b K2i?Q/ Q7 i`Mb7Q`K2/ ?2t?2/`H 2H2K2MiX U#V HH Re TQbbB#H2 +b2b BM i?2 KQ/B}2/ Jh K2i?Q/
Q7 i`Mb7Q`K2/ i2i`?2/`H 2H2K2MiX
pHm2X JQ`2Qp2`- +?M;BM; Siso +M /B`2+iHv BM~m@
2M+2 i?2 }MH `2 M/ pQHmK2 Q7 k. M/ j. BbQ@
b2MbBiBpBiv KQ/2Hb- `2bT2+iBp2HvX AM Q`/2` iQ K22i
i`;2i `2 Q` pQHmK2- i?Bb bim/v i`2ib Siso b
p`B#H2 M/ BMi`Q/m+2b i?2 #Bb2+iBQM K2i?Q/ iQ
KQ/B7v Siso miQKiB+HHvX
Siso =Sup +Slo
2URRV
r?2`2 Slo =0M/ Sup =2×Sth `2 i?2 BMBiBH
HQr2` M/ mTT2` #QmM/- `2bT2+iBp2HvX h?2 Sth Bb i?2
i?`2b?QH/ b2MbBiBpBiv MmK#2` K2MiBQM2/ BM a2+iBQM
kXRX h?2 #QmM/b `2 mT/i2/ mbBM;
Slo =Siso, if V > V
Sup =Siso, if V VURkV
r?2`2 VBb i?2 +m``2Mi pQHmK2f`2 Q7 i?2 BbQ@
b2MbBiBpBiv KQ/2H- M/ VBb i?2 i`;2i pQHmK2f`2X
e
(a) (b)
Faces to be
removed
6B;m`2 9, _2KQpBM; +QBM+B/2Mi 7+2b Q7 j. 2H2K2MibX UV HH +QBM+B/2Mi 7+2b `2 B/2MiB}2/ b BMi2`MH 7+2b- r?B+? `2 `2[mB`2/
iQ #2 `2KQp2/X U#V _2KQpBM; HH BMi2`MH 7+2b ;Bp2b j. +HQb2/ bm`7+2 K2b?X
Remeshing
6B;m`2 8, AKT`Qp2 i?2 bm`7+2 [mHBiv mbBM; i?2 BM+`2K2MiH
`2K2b?BM; H;Q`Bi?K bQ i?i i?2 2H2K2Mib QM i?2 bm`7+2 +M
#2 `2@/Bbi`B#mi2/ mMB7Q`KHv #b2/ QM i`;2i 2/;2 H2M;i?
M/ T`2b2`p2 i?2 ;2QK2i`B+ 72im`2bX
h?2 #B@b2+iBQMH Bi2`iBp2 T`Q+2bb Bb `2T2i2/ mM@
iBH |VV|
VD
Vtol URjV
r?2`2 VDBb i?2 iQiH pQHmK2f`2 Q7 i?2 /2bB;M /Q@
KBM- Vtol =0.1% Bb i?2 HHQr#H2 iQH2`M+2X h?Bb
2[miBQM bBKTHv `2T`2b2Mib i?i i?2 }MH bi`m+im`H
iQTQHQ;v rBHH TQbb2bb i?2 iQiH pQHmK2f`2 +HQb2
iQ i?2 i`;2i pHm2 rBi? M ++2Ti#H2 /Bz2`2M+2X
h?2 2MiB`2 bKQQi?BM; T`Q+2bb i?i +QMbB/2`b T`2@
b2`pBM; `2 Q` pQHmK2 T`Q+2bb Bb bmKK`Bx2/ BM
i?2 ~Qr+?`i b?QrM BM 6B;m`2 eX
jX LmK2`B+H MHvbBb
MmK2`B+H MHvbBb K2i?Q/ Bb /2p2HQT2/ iQ
2tKBM2 i?2 +T#BHBiB2b Q7 i?2 #Qp2 bKQQi?BM;
K2i?Q/bX JQ`2 bT2+B}+HHv- i?2 Q#iBM2/ "1aP
bi`m+im`H iQTQHQ;B2b `2 +QKT`2/ rBi? i?2B` +Q`@
`2bTQM/BM; bKQQi?2/ `2bmHib iQ Q#b2`p2 Mv p`B@
iBQMb BM b?T2 M/ bi`m+im`H T2`7Q`KM+2X
h?Bb bim/v mb2b i?2 +QKK2`+BH }MBi2 2H2K2Mi
MHvbBb U61V bQ7ir`2- #[mb- M/ Svi?QM
b+`BTi iQ T2`7Q`K bi`m+im`H MHvbBb M/ "1aP
iQTQHQ;v QTiBKBxiBQM- `2bT2+iBp2Hv- rBi? /2iBHb
7mHHv /2b+`B#2/ BM (je)X PM+2 i?2 QTiBKH bi`m+im`H
iQTQHQ;B2b `2 Q#iBM2/- *O +Q/2 Bb mb2/ iQ 2t2@
+mi2 i?2 #Qp2 H;Q`Bi?Kb 7Q` ;2M2`iBM; bKQQi?2/
KQ/2Hb BM i?2 _?BMQ d *. bQ7ir`2X h?2 Q`B;B@
MH M/ bKQQi?2/ QTiBKH bi`m+im`H iQTQHQ;B2b `2
i?2M BKTQ`i2/ BMiQ #[mb iQ T2`7Q`K 61 bi`m+@
im`H MHvbBb iQ +QKT`2 i?2B` biBzM2bb T2`7Q`@
KM+2X AM Q`/2` iQ ?p2 7B` +QKT`BbQM- i?2B`
61 K2b?2b `2 b2i iQ ?p2 i?2 bK2 K2b? ivT2 M/
i`;2i 2/;2 H2M;i?X aT2+B}+HHv- k. M/ j. +b2b
`2 MHvx2/ mbBM; i`BM;mH` M/ i2i`?2/`H 2H2@
K2Mib- `2bT2+iBp2Hv- rBi? i?2 i`;2i 2/;2 H2M;i? b2i
iQ #2 i?2 bK2 pHm2 b mb2/ BM i?2 `2K2b?BM; T`Q@
+2bbX LQi2 i?i 2[mBpH2Mi iQTQHQ;B2b `2 /Bb+`2iBx2/
mbBM; i?2 bK2 K2b?BM; H;Q`Bi?K 2K#2//2/ BM i?2
+QKK2`+BH #[mb bQ7ir`2 iQ +`2i2 61 KQ/2Hbc
K2b?BM; H;Q`Bi?Kb Q7 61 KQ/2Hb `2 MQi BM+Hm/2/
BM i?2 b+QT2 Q7 i?Bb bim/vX
h?2 Ki2`BH mb2/ BM i?Bb bim/v Bb bbmK2/ iQ #2
BbQi`QTB+ M/ HBM2` 2HbiB+- rBi? uQmM;Ƕb KQ/mHmb
Q7 E=1JS M/ SQBbbQMǶb `iBQ Q7 v=0.3X lM@
H2bb Qi?2`rBb2 bii2/- "1aP iQTQHQ;v QTiBKBxiBQM
T`K2i2`b `2, ert = 2%-p=3- M/ rmin =3
KKX
h?2 j. bKQQi?BM; T`Q+2bb Bb }`bi i2bi2/ mbBM;
+MiBH2p2` 2tKTH2- b b?QrM BM 6B;m`2 dX qBi?
`272`2M+2 iQ 6B;m`2 dUV- i?2 /2bB;M /QKBM Q7 i?2
d
Topology
optimization
Topology
optimization
Topology
optimization
Surface
Remeshing
Iso-sensitivity
model
Iso-sensitivity
model
Nodal
sensitivity field
Nodal
sensitivity field
No
Yes
No
Final design (smoothed)Final design (smoothed)Final design (smoothed) Yes
V
ˆ
e
S
up iso
SS
up iso
SS
lo iso
SS
lo iso
SS
*
tol
D
VV V
V
*
tol
D
VV V
V
2
up lo
iso
SS
S
2
up lo
iso
SS
S
*
VV
*
VV
iso
S
Surface
mesh
6B;m`2 e, h?2 ~Qr+?`i Q7 i?2 2MiB`2 bKQQi?BM; T`Q+2bbX h?2 T`Q+2bb Q7 T`2b2`pBM; `2 Q` pQHmK2 rBHH 2Mbm`2 i?i i?2 bKQQi?2/
`2bmHi K22ib i`;2i `2 Q` pQHmK2X
+MiBH2p2` Bb ey KK ×Ry KK ×9y KKX F=1
L TQBMi HQ/ Bb TTHB2/ i i?2 +2Mi2` Q7 i?2 7`22
2M/X }t2/ #QmM/`v +QM/BiBQM Bb bbB;M2/ #2?BM/
i?2 r?QH2 +MiBH2p2`X h?2 Q#D2+iBp2 pQHmK2 7+iBQM
Q7 "1aP Bb b2i b R8WX aKQQi?2/ +QM};m`iBQMb
`2 Q#iBM2/ 7`QK bi`m+im`H iQTQHQ;B2b rBi? /Bz2`@
2Mi K2b? bBx2b- b b?QrM BM 6B;m`2b dUV@U#V M/
U+V@U/VX 6Q` Q`B;BMH "1aP bi`m+im`H iQTQHQ;B2b-
bQHB/ +m#B+ 2H2K2Mib i?i ?p2 M 2/;2 H2M;i? Q7
R KK M/ yX8 KK `2 mb2/- b b?QrM BM 6B;m`2b
dUV M/ U+V- `2bT2+iBp2HvX 6Q` bKQQi?2/ bi`m+im`H
iQTQHQ;B2b- Q`B;BMH 2H2K2Mib `2 `2b?T2/ mbBM; i?2
KQ/B}2/ J* HQQFmT i#H2 M/ i?2M i?2 bb2K#HB2b
`2 `2K2b?2/ iQ bm`7+2 K2b? rBi? i`BM;mH` 2H2@
K2Mib i?i ?p2 M p2`;2 2/;2 H2M;i? Q7 yX8 KK-
b b?QrM BM 6B;m`2b dU#V M/ U/VX HH bi`m+im`H
iQTQHQ;B2b `2 /Bb+`2iBx2/ mbBM; bQHB/ i2i`?2/`H 2H@
2K2Mib rBi? M 2/;2 H2M;i? Q7 yX8 KK BM 61- bQ
i?i i?2 +QKTHBM+2 pHm2b- C- `2 +QKT`#H2X
AM 6B;m`2 d- Bi +M #2 b22M i?i i?2 xB;@x; #QmM/@
`B2b Q7 Q`B;BMH bi`m+im`H iQTQHQ;B2b ?p2 #22M bm+@
+2bb7mHHv bKQQi?2/X Ai +M HbQ #2 b22M i?i i?2
;2QK2i`B+H 72im`2b ?p2 #22M T`2b2`p2/- r?2`2
i?2 b?T2b Q7 i?2 Q`B;BMH M/ bKQQi?2/ bi`m+@
im`H iQTQHQ;B2b `2 HKQbi B/2MiB+HX JQ`2Qp2`- i?2
bi`m+im`H pQHmK2b `2 +?M;2/ rBi?BM M HHQr#H2
iQH2`M+2X hQ;2i?2`- Bi +M #2 +QM+Hm/2/ i?i i?2
T`QTQb2/ bKQQi?BM; K2i?Q/ +M ;2M2`i2 2[mBp@
H2Mi bi`m+im`H iQTQHQ;B2b iQ i?2 Q`B;BMH iQTQHQ;v
QTiBKBxiBQM `2bmHibX
6m`i?2`KQ`2- i?2 +QKTHBM+2 Q7 6B;m`2b dUV@U/V
`2 8Xjje LKK- 9X388 LKK- 8Xyey LKK- M/ 9XNdj
LKK- `2bT2+iBp2HvX Ai Bb b22M i?i i?2 bKQQi?2/
bi`m+im`2b `2 biBz2`- +Q``2bTQM/BM; iQ HQr2` +QK@
THBM+2X JQ`2 bT2+B}+HHv- i?2 +QKTHBM+2 Q7 6B;@
m`2b dUV M/ U+V `2 `2/m+2/ #v NXykW M/ RXdjW
7i2` bKQQi?BM;- `2bT2+iBp2HvX h?2 2M?M+2/ bi`m+@
im`H T2`7Q`KM+2 Kv #2 ii`B#mi2/ iQ i?2 bi`2bb
#2BM; KQ`2 mMB7Q`KHv /Bbi`B#mi2/ BM i?2 bKQQi?2/
KQ/2Hb- i?mb `2bmHiBM; BM HQr2` +QKTHBM+2- /2KQM@
bi`i2/ KQ`2 +H2`Hv BM 6B;m`2b dUV@U#VX >Qr2p2`-
i?2 BKT`Qp2K2Mi Bb H2bb MQiB+2#H2 mbBM; }M2`
K2b?- b i?2 `2bQHmiBQM Q7 i?2 Q`B;BMH bi`m+im`H
iQTQHQ;v Bb H`2/v ?B;?- K2MBM; i?i i?2 b?T2
Bb bBKBH` iQ Bib +Q``2bTQM/BM; bKQQi?2/ 7Q`K- b
b?QrM BM 6B;m`2b dU+V@U/VX L2p2`i?2H2bb- i?Bb 2tK@
TH2 biBHH /2KQMbi`i2b i?i M 2M?M+2/ bi`m+im`H
T2`7Q`KM+2 +M #2 Q#iBM2/ i?`Qm;? bKQQi?BM;X
9X .Bb+mbbBQM
9XRX _2;mH` M/ B``2;mH` k. K2b?2b
h?Bb bm#b2+iBQM 2tKBM2b i?2 bi`m+im`H T2`7Q`@
KM+2 Q7 bKQQi?2/ iQTQHQ;B2b Q#iBM2/ 7`QK `2;m@
H` M/ B``2;mH` k. [m/`M;mH` K2b?2b (9N)- b
b?QrM BM 6B;m`2b 3UV M/ U#V- `2bT2+iBp2HvX
6B;m`2 3UV b?Qrb k9y KK ×9y KK J"" #2K
BMBiBHHv K2b?2/ rBi? R KK b[m`2 2H2K2MibX
F=1L TQBMi HQ/ Bb TTHB2/ i i?2 KB/TQBMi
Q7 i?2 iQT 2/;2X 6Q` iQTQHQ;v QTiBKBxiBQM- i?2
Q#D2+iBp2 pQHmK2 7+iBQM Bb b2i b 8yW- M/ rmin
Bb +?M;2/ iQ k KKX .m`BM; i?2 bKQQi?BM; T`Q@
+2bb- i?2 KQ/B}2/ Ja HQQFmT i#H2 Bb mb2/ iQ Q#iBM
i`Mb7Q`K2/ [m/`M;mH` 2H2K2Mib- M/ i?2M i?2v
3
S, Mises
Max
Min
(a) (b)
(c)
C = 5.336 Nmm
F
60
10
40
Smoothed
(d)
C = 5.060 Nmm
C = 4.855 Nmm
C = 4.973 Nmm
Original
6B;m`2 d, 61 `2bmHib Q7 Q`B;BMH M/ bKQQi?2/ bi`m+im`H iQTQHQ;B2bX UV@U#V b?Qrb i?i +m#B+ 2H2K2Mib i?i ?p2 M 2/;2 H2M;i?
Q7 R KK `2 bKQQi?2/ BMiQ bm`7+2 K2b? rBi? i`BM;mH` 2H2K2Mib i?i ?p2 M p2`;2 2/;2 H2M;i? Q7 yX8 KKX U+V@U/V b?Qrb
i?i +m#B+ 2H2K2Mib i?i ?p2 M 2/;2 H2M;i? Q7 yX8 KK `2 bKQQi?2/ BMiQ bm`7+2 K2b? rBi? i`BM;mH` 2H2K2Mib i?i ?p2
M p2`;2 2/;2 H2M;i? Q7 yX8 KKX UV M/ U+V `2 Q`B;BMH bi`m+im`H iQTQHQ;B2bc U#V M/ U/V `2 bKQQi?2/ bi`m+im`H iQTQHQ;B2bX
AM 61 UV@U/V `2 K2b?2/ mbBM; bQHB/ i2i`?2/`H 2H2K2Mib rBi? M p2`;2 2/;2 H2M;i? Q7 yX8 KKX
`2 `2K2b?2/ mbBM; i`BM;mH` 2H2K2Mib rBi? M p@
2`;2 2/;2 H2M;i? Q7 R KKX 6Q` 61 bi`m+im`H MHv@
bBb- i?2 Q`B;BMH M/ bKQQi?2/ bi`m+im`H iQTQHQ;B2b
`2 #Qi? /Bb+`2iBx2/ mbBM; i`BM;mH` 2H2K2Mib rBi?
M p2`;2 2/;2 H2M;i? Q7 RKKX
AM 6B;m`2 3UV- Bi +M #2 b22M i?i i?2 J"" #2K
?b #22M bm++2bb7mHHv bKQQi?2/ mbBM; i?2 T`QTQb2/
k. bKQQi?BM; K2i?Q/- r?B+? `2/m+2b i?2 Q`B;BMH
+QKTHBM+2 #v jX9eW U7`QK N9X9e8 LKK iQ 3NXNN3
LKKVX h?2 bK2 2tKTH2 ?b #22M bKQQi?2/ T`2@
pBQmbHv mbBM; i?2 6hPS K2i?Q/ (kj) rBi? }MH
+QKTHBM+2 Q7 3NXe3 LKK- r?B+? Bb bBKBH` iQ i?2
T`2b2Mi bKQQi?2/ `2bmHi BM b?T2 M/ +QKTHBM+2X
>Qr2p2`- i?2 6hPS K2i?Q/ `2[mB`2b KQ`2 i?M
jyy Bi2`iBQMb iQ Q#iBM i?2 }MH bKQQi?2/ `2bmHi
rBi? bKQQi?2/ bi`m+im`H iQTQHQ;B2b ;2M2`i2/ BM
2+? Bi2`iBQMX AM +QMi`bi- i?2 T`QTQb2/ bKQQi?BM;
K2i?Q/ Bb Km+? KQ`2 2{+B2Mi- r?B+? +M #2 //2/
iQ i?2 Hbi bi2T Q7 i?2 "1aP T`Q+2bbX h?Bb 2tK@
TH2 QMHv mb2b dy Bi2`iBQMb iQ Q#iBM i?2 bi`m+im`H
iQTQHQ;v- M/ i?2 iQiH iBK2 bT2Mi QM i?2 bKQQi?@
BM; T`Q+2bb Bb QMHv yX9j b2+QM/b mbBM; M Q`/BM`v
HTiQT rBi? M AMi2H Bd@Nd8y> kXey:>x T`Q+2bbQ`X
h?2 ddd KK ×dRj KK +?B` 2tKTH2 b?QrM
BM 6B;m`2 3U#V Bb BMBiBHHv K2b?2/ rBi? B``2;mH`
[m/`M;mH` 2H2K2Mib BM Q`/2` iQ +`2i2 bKQQi?
#QmM/`B2b- `2bmHiBM; BM HH 2H2K2Mib ?pBM; /Bz2`@
2Mi b?T2b rBi? M p2`;2/ 2/;2 H2M;i? Q7 9Xk8
KKX h?2 +?B` Bb HQ/2/ i Bib MQM@/2bB;M /QKBMb
rBi? `2HiBp2 7Q`+2 /2MbBiv Q7 F:F=3:2(8y)-
r?2`2 F=yXyj L M/ F=yXyk LX h?2 Q#D2+iBp2
pQHmK2 7+iBQM Bb b2i b jjW iQ Q#iBM i?2 "1aP
bi`m+im`H iQTQHQ;vX AM i2`Kb Q7 bKQQi?BM; T`Q+2bb-
i?2 KQ/B}2/ Ja HQQFmT i#H2 BM 6B;m`2 kU#V Bb mb2/-
N
r?2`2 2+? bm#@};m`2 Kv #2 `2b?T2/ /2T2M/BM;
QM i?2 b?T2 Q7 i?2 b2H2+i2/ B``2;mH` 2H2K2Mi mb@
BM; 1[miBQM RyX 6BMHHv- i?2 i`Mb7Q`K2/ 2H2K2Mib
`2 `2K2b?2/ mbBM; i`BM;mH` 2H2K2Mib i?i ?p2
M p2`;2 2/;2 H2M;i? Q7 jX8 KKX 6Q` 61 bi`m+@
im`H MHvbBb- i?2 Q`B;BMH M/ bKQQi?2/ bi`m+im`H
iQTQHQ;B2b `2 #Qi? /Bb+`2iBx2/ mbBM; i`BM;mH` 2H@
2K2Mib rBi? M p2`;2 2/;2 H2M;i? Q7 jX8 KKX
AM 6B;m`2 3U#V- Bi Bb b22M i?i i?2 +?B` ?b #22M
bm++2bb7mHHv bKQQi?2/- K2MBM; i?i i?2 T`QTQb2/
bKQQi?BM; K2i?Q/ Bb +T#H2 Q7 /2HBM; rBi? B`@
`2;mH` K2b?2bX aBKBH` iQ i?2 #Qp2 2tKTH2b
rBi? `2;mH` K2b?2b- i?2 +QKTHBM+2 HbQ `2/m+2b
7i2` bKQQi?BM;- r?2`2 kXNjW BKT`Qp2K2Mi Bb
+?B2p2/c CBb `2/m+2/ 7`QK 33XdkN LKK iQ 3eXRjj
LKKX Ai Bb rQ`i? TQBMiBM; Qmi i?i +QKTH2t
;2QK2i`B2b `2 ivTB+HHv /Bb+`2iBx2/ rBi? B``2;mH`
K2b?2b iQ T2`7Q`K iQTQHQ;v QTiBKBxiBQMX lbBM;
i?2 T`QTQb2/ K2i?Q/- QTiBKH bi`m+im`H iQTQHQ@
;B2b Q7 bm+? +QKTH2t ;2QK2i`B2b +M #2 +QMp2MB2MiHv
bKQQi?2/- r?B+? K2Mb i?i i?2 T`QTQb2/ K2i?Q/
Bb ?B;?Hv #2M2}+BH 7Q` +QKTH2t ;2QK2i`B2b BMBiBHHv
/Bb+`2iBx2/ rBi? B``2;mH` K2b?2bX
9XkX j. T`+iB+H TTHB+iBQM
ddR KK ×9yy KK ×dRj KK j. +?B` 2t@
KTH2 Bb ;Bp2M ?2`2 iQ /2KQMbi`i2 i?2 TQi2MiBH
T`+iB+H TTHB+iBQMb Q7 i?2 T`QTQb2/ bKQQi?BM;
K2i?Q/- b b?QrM BM 6B;m`2 NX 6B;m`2 NUV b?Qrb
i?2 b2imT Q7 i?2 QTiBKBxiBQM T`Q#H2K- 7`QK r?B+?
3RW Q7 i?2 Ki2`BH Bb `2KQp2/ i?`Qm;? iQTQHQ;v
QTiBKBxiBQMX Ai Bb bbmK2/ i?i i?2 +?B` +``B2b M
3y F; /mHi- ?2M+2 i?2 MQM@/2bB;M /QKBM Bb HQ/2/
rBi? F=0.0003 LfKK2M/ F=0.0002 LfKK2
QM i?2 b2i T`i M/ i?2 #+F@`2bi T`i- `2bT2+iBp2Hv
(8y)X 6Qm` ky KK ×ky KK `2;BQMb QM i?2 #QiiQK
bm`7+2 `2 bbB;M2/ iQ ?p2 }t2/ #QmM/`v +QM/B@
iBQMX rmin Bb +?M;2/ iQ jy KK iQ Q#iBM i?2 "1aP
bi`m+im`H iQTQHQ;vX h?2 KQ/B}2/ Jh HQQFmT i#H2
Bb mb2/ 7Q` bKQQi?BM;X h?2 Q`B;BMH M/ bKQQi?2/
+?B`b `2 #Qi? K2b?2/ rBi? bQHB/ i2i`?2/` i?i
?p2 M 2/;2 H2M;i? Q7 Ry KKX
6B;m`2b NU#V@U+V b?Qr i?2 Q`B;BMH M/ bKQQi?2/
bi`m+im`H iQTQHQ;B2b- `2bT2+iBp2HvX Ai Bb b22M i?i
i?2 bKQQi?2/ `2bmHi BKT`Qp2b MQi QMHv i?2 pBbmH
2z2+i #mi HbQ i?2 bi`m+im`H T2`7Q`KM+2- r?2`2
i?2 +QKTHBM+2 Q7 i?2 Q`B;BMH `2bmHi Bb `2/m+2/ #v
eX8kW U7`QK eRXkjR LKK iQ 8dXkj8 LKKVX 6QH@
HQrBM; i?2 /Bb+mbbBQM Q7 a2+iBQM 9XR- +QKTH2t j.
;2QK2i`B2b `2 ivTB+HHv /Bb+`2iBx2/ rBi? B``2;mH`
i2i`?2/`H K2b?2bX 7i2` T2`7Q`KBM; iQTQHQ;v QT@
iBKBxiBQM QM bm+? +QKTH2t j. ;2QK2i`B2b- i?2 T`Q@
TQb2/ bKQQi?BM; K2i?Q/ +M i?2M #2 2KTHQv2/ iQ
+`2i2 KMB7QH/ *. KQ/2Hb i?i TQbb2bb bKQQi?
#QmM/`B2b- b b?QrM BM 6B;m`2b NU+V@U/VX h?2`2@
7Q`2- Bi Bb +QM}`K2/ i?i i?2 T`QTQb2/ bKQQi?BM;
K2i?Q/ +M #2 2z2+iBp2Hv /QTi2/ 7Q` j. TTHB+@
iBQMb rBi?Qmi i?2 HBKBiiBQM Q7 i?2 K2b? 2H2K2Mi
ivT2X
8X *QM+HmbBQM
AM i?Bb rQ`F- M2r TQbi@T`Q+2bbBM; K2i?Q/ Bb
T`QTQb2/- r?B+? mb2b T`2@#mBHi HQQFmT i#H2b iQ
bKQQi? i?2 xB;@x; #QmM/`B2b Q#iBM2/ 7`QK iQTQH@
Q;v QTiBKBxiBQMX Ai Bb b?QrM i?i i?2 KQ/B}2/
J`+?BM; a[m`2b UJaV- J`+?BM; *m#2b UJ*V-
M/ J`+?BM; h2i`?2/` UJhV T`QpB/2 2z2+iBp2
HQQFmT i#H2b 7Q` bi`m+im`2b BMBiBHHv /Bb+`2iBx2/
rBi? [m/`M;mH` Uk.V- ?2t?2/`H Uj.V- M/
i2i`?2/`H Uj.V 2H2K2Mib- `2bT2+iBp2HvX b2@
`B2b Q7 2tKTH2b `2 i2bi2/ iQ b?Qr i?i, URV i?2
M2r K2i?Q/ +M +`2i2 bKQQi? bi`m+im`H iQTQHQ@
;B2b i?i T`2b2`p2 i?2 pQHmK2 M/ ;2QK2i`B+ 72@
im`2b 7`QK i?2 Q`B;BMH QTiBKH iQTQHQ;B2b- M/ UkV
i?2 +QKTHBM+2 +M #2 `2/m+2/ 7i2` bKQQi?BM;
iQ +?B2p2 biBz2` bi`m+im`2bX JQ`2Qp2`- i?2 M2r
K2i?Q/ Bb b?QrM iQ #2 ?B;?Hv 2{+B2Mi- b Bi Bb T2`@
7Q`K2/ b TQbi@T`Q+2bbBM; bi2T 7i2` i?2 iQTQHQ;v
QTiBKBxiBQM Bb +QKTH2i2/X h?Bb bim/v HbQ b?Qrb
i?i i?2 T`QTQb2/ K2i?Q/ bmTTQ`ib p`BQmb K2b?
2H2K2Mi ivT2bX 6BMHHv- +QKTH2t j. 2tKTH2 Bb
mb2/ iQ /2KQMbi`i2 i?i i?2 M2r K2i?Q/ +M #2
`2/BHv TTHB2/ iQ rB/2 `M;2 Q7 T`+iB+H TTHB@
+iBQMbX
*_2/Bh mi?Q`b?BT +QMi`B#miBQM bii2K2Mi
w?B GB, *QM+2TimHBxiBQM- J2i?Q/QHQ;v- aQ7i@
r`2- .i *m`iBQM- oHB/iBQM- oBbmHBxiBQM-
q`BiBM; @ P`B;BMH .`7iX hBM;@l2B G22, q`BiBM;
@ P`B;BMH .`7i- oBbmHBxiBQM- q`BiBM; @ _2pB2r
1/BiBM;X umM uQ, *QM+2TimHBxiBQM- J2i?Q/@
QHQ;vX uB JBM sB2, *QM+2TimHBxiBQM- J2i?Q/@
QHQ;v- amT2`pBbBQM- q`BiBM; @ _2pB2r 1/BiBM;-
6mM/BM; +[mBbBiBQMX
.2+H`iBQM Q7 +QKT2iBM; BMi2`2bi
h?2 mi?Q`b /2+H`2 i?i i?2v ?p2 MQ FMQrM
+QKT2iBM; }MM+BH BMi2`2bib Q` T2`bQMH `2HiBQM@
Ry
F’
F
(a) (b)
F
Non-design domain Design domain
Structural topology
Smoothed structural topology
C = 94.465 Nmm
C = 89.998 Nmm
C = 88.729 Nmm
C = 86.133 Nmm
Problem definition
Regular Irregular
6B;m`2 3, aKQQi?BM; k. bi`m+im`H iQTQHQ;B2b rBi? `2;mH` M/ B``2;mH` K2b?2bX am#@};m`2b 7`QK i?2 iQT `2, QTiBKBxiBQM
T`Q#H2K- Q`B;BMH bi`m+im`H iQTQHQ;v- M/ bKQQi?2/ bi`m+im`H iQTQHQ;vX UV J"" #2K 2tKTH2 BMBiBHHv K2b?2/ rBi? b[m`2
2H2K2MibX U#V *?B` 2tKTH2 BMBiBHHv K2b?2/ rBi? B``2;mH` [m/`M;mH` 2H2K2MibX
F’
F
C = 61.231 Nmm C = 57.235 Nmm
Non-design domain
Design domain
Symmetry
Fixed support
(a) (b) (c)
Structural
topology
Smoothed
structural
topology
Tetrahedral
meshes
(d)
6B;m`2 N, SQi2MiBH T`+iB+H TTHB+iBQMb Q7 i?2 T`QTQb2/ bKQQi?BM; K2i?Q/ /2KQMbi`i2/ rBi? j. +?B` 2tKTH2X UV
AMBiBH b2imT Q7 i?2 QTiBKBxiBQM T`Q#H2KX U#V P`B;BMH bi`m+im`H iQTQHQ;v Q#iBM2/ 7`QK iQTQHQ;v QTiBKBxiBQMX U+V aKQQi?2/
bi`m+im`H iQTQHQ;vX U/V JMm7+im`2/ j. KQ/2HX
RR
b?BTb i?i +QmH/ ?p2 TT2`2/ iQ BM~m2M+2 i?2
rQ`F `2TQ`i2/ BM i?Bb TT2`X
+FMQrH2/;2K2Mib
h?Bb rQ`F rb bmTTQ`i2/ #v i?2 mbi`HBM _2@
b2`+? *QmM+BH (;`Mi MmK#2` 6GRNyRyyyR9)X h?2
mi?Q`b rQmH/ HBF2 iQ i?MF umHBM sBQM;- umMx?2M
>2- CK2b EB`#v- ZBM; w?Qm- ZBM uM;- CBKBM;
J M/ JBM;?Q "B 7Q` i?2B` ?2HT7mH /pB+2 QM p`@
BQmb i2+?MB+H Bbbm2b 2tKBM2/ BM i?Bb TT2`X
_272`2M+2b
(R) "2M/bǠ2 JS- aB;KmM/ PX hQTQHQ;v QTiBKBxiBQM, h?2@
Q`v- K2i?Q/b- M/ TTHB+iBQMbX ki? 2/X aT`BM;2`c kyy9X
(k) sBQM; uG- uQ a- w?Q wG- sB2 uJX M2r TT`Q+? iQ
2HBKBMiBM; 2M+HQb2/ pQB/b BM iQTQHQ;v QTiBKBxiBQM 7Q`
//BiBp2 KMm7+im`BM;X //Bi JMm7 kykycjk,RyRyyeX
?iiTb,ff/QBXQ`;fRyXRyRefDX//KXkyRNXRyRyyeX
(j) "B J- h`M S- sB2 uJX hQTQHQ;v QTiBKBxiBQM Q7 j.
+QMiBMmmK bi`m+im`2b mM/2` ;2QK2i`B+ b2H7@bmTTQ`iBM;
+QMbi`BMiX //Bi JMm7 kykycje,RyR9kkX ?iiTb,ff/QBX
Q`;fRyXRyRefDX//KXkykyXRyR9kkX
(9) sm "- >M u- w?Q G- sB2 uJX hQTQHQ;B+H QTiB@
KBxiBQM Q7 +QMiBMmmK bi`m+im`2b 7Q` //BiBp2 KMm@
7+im`BM; +QMbB/2`BM; i?BM 72im`2 M/ bmTTQ`i bi`m+@
im`2 +QMbi`BMibX 1M; PTiBKBx kykyc8j,kRkk@9jX ?iiTb,
ff/QBXQ`;fRyXRy3yfyjy8kR8sXkykyXR39NRdyX
(8) ;2 L- KB` P- *Hmb2M - >/` G- JB2` .- aǠM@
/2`;`/ X /pM+2/ iQTQHQ;v QTiBKBxiBQM K2i?Q/b
7Q` +QM+2TimH `+?Bi2+im`H /2bB;MX AM, S`Q+22/BM;b Q7
i?2 /pM+2b BM `+?Bi2+im`H :2QK2i`v kyR9X *?K,
aT`BM;2`c kyR8X TX R8N@dNX
(e) sB2 uJ- wmQ w>- >mM; s- hM; Cq- w?Q "- 62HB+2iiB
SX `+?Bi2+im`2 M/ m`#M /2bB;M i?`Qm;? 2pQHmiBQM`v
bi`m+im`H QTiBKBbiBQM H;Q`Bi?KbX AM, S`Q+22/BM;b Q7
i?2 AMi2`MiBQMH avKTQbBmK QM H;Q`Bi?KB+ .2bB;M 7Q`
`+?Bi2+im`2 M/ l`#M .2bB;M- hQFvQc kyRRX
(d) *mB *- P?KQ`B >- abFB JX *QKTmiiBQMH KQ`T?Q@
;2M2bBb Q7 j. bi`m+im`2b #v 2ti2M/2/ 1aP K2i?Q/X C
AMi bbQ+ a?2HH aTiBH ai`m+i kyyjc99URV,8R@eRX ?iiTb,
ff/QBXQ`;fRyXRyRefDX2M;bi`m+iXkyR9XydXyjkX
(3) "?QQb?M o- 6m+?b J- "?QQb?M aX j.@T`BMiBM;- iQTQH@
Q;v QTiBKBxiBQM M/ biiBbiB+H H2`MBM;, +b2 bim/vX
AM, avKTQbBmK QM aBKmHiBQM 7Q` `+?Bi2+im`2 M/ l`@
#M .2bB;M- hQ`QMiQc kyRdX TX R@3X
(N) J C- GB w- w?Q wG- sB2 uJX *`2iBM; MQp2H 7m`@
MBim`2 i?`Qm;? iQTQHQ;v QTiBKBxiBQM M/ /pM+2/
KMm7+im`BM;X _TB/ S`QiQivT C kykRckdUNV,Rd9N@83X
?iiTb,ff/QBXQ`;fRyXRRy3f_SC@yj@ kykR@yy9dX
(Ry) w?Q wG- w?Qm aq- 62M; sZ- sB2 uJX PM i?2 BM@
i2`MH `+?Bi2+im`2 Q7 2K2`;2Mi THMibX C J2+? S?vb
aQHB/b kyR3cRRN,kk9@jNX ?iiTb,ff/QBXQ`;fRyXRyRefDX
DKTbXkyR3XyeXyR9X
(RR) J C- w?Q wG- GBM a- sB2 uJX hQTQHQ;v Q7 H27 p2BMb,
2tT2`BK2MiH Q#b2`piBQM M/ +QKTmiiBQMH KQ`T?Q@
;2M2bBbX C J2+? "2? "BQK2/ kykRcRkj,Ry9d33X ?iiTb,
ff/QBXQ`;fRyXRyRefDXDK##KXkykRXRy9d33X
(Rk) "2M/bǠ2 JS- EBFm+?B LX :2M2`iBM; QTiBKH iQTQHQ;B2b
BM bi`m+im`H /2bB;M mbBM; ?QKQ;2MBxiBQM K2i?Q/X
*QKTmi J2i? TTH J2+? 1M;`; RNN3cdRUkV,RNd@kk9X
?iiTb,ff/QBXQ`;fRyXRyRefyy98@d3k8U33VNyy3e@ kX
(Rj) >mM; s- sB2 uJX M2r HQQF i 1aP M/
"1aP QTiBKBxiBQM K2i?Q/bX ai`m+i JmHiB/Bb+BT
PTiBK kyy3cj8,3N@NkX ?iiTb,ff/QBXQ`;fRyXRyydf
byyR83@yyd@ yR9y@9X
(R9) aB;KmM/ PX NN HBM2 iQTQHQ;v QTiBKBxiBQM +Q/2 r`Bi@
i2M BM JiH#X ai`m+i JmHiB/Bb+BT PTiBK kyyRckR,Rky@dX
?iiTb,ff/QBXQ`;fRyXRyydfbyyR83yy8yRdeX
(R8) aB;KmM/ P- Jmi2 EX hQTQHQ;v QTiBKBxiBQM T@
T`Q+?2bX ai`m+i JmHiB/Bb+BT PTiBK kyRjc93,RyjR@88X
?iiTb,ff/QBXQ`;fRyXRyydfbyyR83@yRj@ yNd3@eX
(Re) >mM; s- sB2 uJX 1pQHmiBQM`v iQTQHQ;v QTiBKBxiBQM
Q7 +QMiBMmmK bi`m+im`2b, J2i?Q/b M/ TTHB+iBQMbX
CQ?M qBH2v aQMbc kyRyX
(Rd) >mM; s- sB2 uJX "B@/B`2+iBQMH 2pQHmiBQM`v iQTQH@
Q;v QTiBKBxiBQM Q7 +QMiBMmmK bi`m+im`2b rBi? QM2
Q` KmHiBTH2 Ki2`BHbX *QKTmi J2+? kyyNc9j,jNj@9yRX
?iiTb,ff/QBXQ`;fRyXRyydfbyy9ee@yy3@ yjRk@yX
(R3) "2M/bǠ2 JS- aB;KmM/ PX Ji2`BH BMi2`TQHiBQM
b+?2K2b BM iQTQHQ;v QTiBKBxiBQMX `+? TTH
J2+? RNNNceN,ej8@89X ?iiTb,ff/QBXQ`;fRyXRyydf
byy9RNyy8yk93X
(RN) 6m u6- _QH72 "- qM; u- >mM; s- :?#`B2
EX a1J.Ph, aKQQi?@2/;2/ Ki2`BH /Bbi`B#m@
iBQM 7Q` QTiBKBxBM; iQTQHQ;v H;Q`Bi?KX /p 1M;
aQ7ir kykycR8y,RykNkRX ?iiTb,ff/QBXQ`;fRyXRyRefDX
Rk
/p2M;bQ7iXkykyXRykNkRX
(ky) am#2/B a*- o2`K *a- am`2b? EX `2pB2r Q7 K2i?@
Q/b 7Q` i?2 ;2QK2i`B+ TQbi@T`Q+2bbBM; Q7 iQTQHQ;v QTiB@
KBx2/ KQ/2HbX C *QKTmi AM7 a+B 1M; kykyckyUeV,yey3yRX
?iiTb,ff/QBXQ`;fRyXRRR8fRX9y9d9kNX
(kR) w?mM; w- sB2 uJ- w?Qm aqX `2+iBQM /BzmbBQM@
#b2/ H2p2H b2i K2i?Q/ mbBM; #Q/v@}ii2/ K2b? 7Q` bi`m+@
im`H iQTQHQ;v QTiBKBxiBQMX *QKTmi J2i? TTH J2+?
1M;`; kykRcj3R,RRj3kNX ?iiTb,ff/QBXQ`;fRyXRyRefDX
+KXkykRXRRj3kNX
(kk) 6m u6- _QH72 "- *?Bm GLa- qM; u- >mM; s-
:?#`B2 EX aKQQi? iQTQHQ;B+H /2bB;M Q7 j. +QM@
iBMmmK bi`m+im`2b mbBM; 2H2K2MiH pQHmK2 7`+iBQMbX
*QKTmi ai`m+i kykyckjR,RyekRjX ?iiTb,ff/QBXQ`;fRyX
RyRefDX+QKTbi`m+XkykyXRyekRjX
(kj) >mM; sX aKQQi? iQTQHQ;B+H /2bB;M Q7 bi`m+im`2b mb@
BM; i?2 ~QiBM; T`QD2+iBQMX 1M; ai`m+i kykycky3,RRyjjyX
?iiTb,ff/QBXQ`;fRyXRyRefDX2M;bi`m+iXkykyXRRyjjyX
(k9) >mM; sX PM bKQQi? Q` yfR /2bB;Mb Q7 i?2 }t2/@
K2b? 2H2K2Mi@#b2/ iQTQHQ;v QTiBKBxiBQMX /p 1M;
aQ7ir kykRcR8R,RykN9kX ?iiTb,ff/QBXQ`;fRyXRyRefDX
/p2M;bQ7iXkykyXRykN9kX
(k8) . .- sB G- GB :- >mM; sX 1pQHmiBQM`v
iQTQHQ;v QTiBKBxiBQM Q7 +QMiBMmmK bi`m+im`2b rBi?
bKQQi? #QmM/`v `2T`2b2MiiBQMX ai`m+i JmHiB/Bb@
+BT PTiBK kyR3c8d,kR9j@8NX ?iiTb,ff/QBXQ`;fRyXRyydf
byyR83@yRd@ R39e@eX
(ke) J`iőM2x@6`miQb C- >2``2`Q@Sû`2x .X :Sl +@
+2H2`iBQM 7Q` 2pQHmiBQM`v iQTQHQ;v QTiBKBx@
iBQM Q7 +QMiBMmmK bi`m+im`2b mbBM; BbQbm`@
7+2bX *QKTmi ai`m+i kyRdcR3k,RRN@jeX ?iiTb,
ff/QBXQ`;fRyXRyRefDX+QKTbi`m+XkyReXRyXyR3X
(kd) EQ;m+?B - EBFm+?B LX bm`7+2 `2+QMbi`m+@
iBQM H;Q`Bi?K 7Q` iQTQHQ;v QTiBKBxiBQMX 1M;
*QKTmi kyyeckk,R@RyX ?iiTb,ff/QBXQ`;fRyXRyydf
byyjee@yye@ yykj@yX
(k3) J`bM G- .mii .X *QMbi`m+iBQM Q7 bm`7+2 KQ/2H
M/ Hv2`2/ KMm7+im`BM; /i 7`QK j. ?QKQ;2MBx@
iBQM QmiTmiX C J2+? .2bB;M RNNecRR3UjV,9Rk@R3X ?iiTb,
ff/QBXQ`;fRyXRRR8fRXk3keNyRX
(kN) uQ2Hv uJ- KB` P- >MMB2H AX hQTQHQ;v M/ b?T2
QTiBKBxiBQM rBi? 2tTHB+Bi ;2QK2i`B+ +QMbi`BMib mbBM;
bTHBM2@#b2/ `2T`2b2MiiBQM M/ }t2/ ;`B/X S`Q+2@
/B JMm7 kyR3ckR,R3N@NeX ?iiTb,ff/QBXQ`;fRyXRyRef
DXT`QK7;XkyR3XykXRRyX
(jy) CBM; *- qM; >- AMx o*- .2HHBM;2` 6- _Bbi 6- qHH@
M2` C- SQiiKMM >X lbBM; BbQK2i`B2b 7Q` +QKTmiiBQMH
/2bB;M M/ 7#`B+iBQMX *J h`Mb :`T? kykRc9yU9V,R@
RkX ?iiTb,ff/QBXQ`;fRyXRR98fj98yekeXj98N3jNX
(jR) CB w- GBm G- qM; :X ;HQ#H HTH+BM bKQQi?BM;
TT`Q+? rBi? 72im`2 T`2b2`piBQMX AM, LBMi? AMi2`M@
iBQMH *QM72`2M+2 QM *QKTmi2` B/2/ .2bB;M M/ *QK@
Tmi2` :`T?B+b- >QM; EQM;, A111c kyy8X TX keN@d9X
(jk) 6B2H/ .X GTH+BM bKQQi?BM; M/ .2HmMv i`BM;m@
HiBQMbX *QKKmM TTH LmK2` J2i? RNN3c9UeV,dyN@RkX
?iiTb,ff/QBXQ`;fRyXRyykf+MKXRejyy9yeyjX
(jj) "++B;HB - *2`miB - GBp2`MB X am`7+2 bKQQi?BM;
7Q` iQTQHQ;B+H QTiBKBx2/ j. KQ/2HbX ai`m+i JmHiB/Bb@
+BT PTiBK kykRce9,j98j@dkX ?iiTb,ff/QBXQ`;fRyXRyydf
byyR83@ykR@ yjykd@eX
(j9) GB s- ZBM *- q2B S- am *X #QmM/`v /2MbBiv 2pQ@
HmiBQM`v iQTQHQ;v QTiBKBxiBQM Q7 +QMiBMmmK bi`m+@
im`2b rBi? bKQQi? #QmM/`B2bX AMi C LmK2` J2i? 1M;
kykkcRkj,R83@dNX ?iiTb,ff/QBXQ`;fRyXRyykfMK2Xe38RX
(j8) G`b2M a- C2Mb2M *:X *QMp2`iBM; iQTQHQ;v QTiBKBxiBQM
`2bmHib BMiQ T`K2i`B+ *. KQ/2HbX *QKTmi B/2/
.2b TTH kyyNceUjV,9yd@R3X ?iiTb,ff/QBXQ`;fRyXjdkkf
+/TbXkyyNX9yd@9R3X
(je) wmQ w>- sB2 uJX bBKTH2 M/ +QKT+i Svi?QM
+Q/2 7Q` +QKTH2t j. iQTQHQ;v QTiBKBxiBQMX /p
1M; aQ7ir kyR8c38,R@RRX ?iiTb,ff/QBXQ`;fRyXRyRefDX
/p2M;bQ7iXkyR8XykXyyeX
(jd) aB;KmM/ P- S2i2`bbQM CX LmK2`B+H BMbi#BHBiB2b BM
iQTQHQ;v QTiBKBxiBQM, bm`p2v QM T`Q+2/m`2b /2HBM;
rBi? +?2+F2`#Q`/b- K2b?@/2T2M/2M+B2b M/ HQ+H KBM@
BKX ai`m+i PTiBK RNN3cRe,e3@d8X ?iiTb,ff/QBXQ`;f
RyXRyydf"6yRkR9yykX
(j3) "Qm`/BM "X 6BHi2`b BM iQTQHQ;v QTiBKBxiBQMX AMi C Lm@
K2` J2i? 1M; kyyRc8yUNV,kR9j@83X ?iiTb,ff/QBXQ`;f
RyXRyykfMK2XRReX
(jN) .2`pB2mt - h?QKbb2i 6X }MBi2 2H2K2Mi K2i?Q/
7Q` i?2 bBKmHiBQM Q7 _vH2B;?@hvHQ` BMbi#BHBivX AM,
TT`QtBKiBQM J2i?Q/b 7Q` LpB2`@aiQF2b S`Q#H2Kb-
"2`HBM, aT`BM;2`c RN3yX TX R98@83X
(9y) JTH2 *X :2QK2i`B+ /2bB;M M/ bT+2 THMMBM; mbBM;
i?2 K`+?BM; b[m`2b M/ K`+?BM; +m#2 H;Q`Bi?KbX
Rj
AM, S`Q+22/BM;b Q7 i?2 A111 AMi2`MiBQMH *QM72`2M+2
QM :2QK2i`B+ JQ/2HBM; M/ :`T?B+b- GQM/QM, A111c
kyyjX TX Ny@8X
(9R) :QM; a- L2rKM haX +Q`M2` 72im`2 b2MbBiBp2 K`+?@
BM; b[m`2bX AM, kyRj S`Q+22/BM;b Q7 A111 aQmi?2bi@
+QM- C+FbQMpBHH2, A111c kyRjX TX R@eX
(9k) G+?m/ CPX hQTQHQ;B+HHv /2}M2/ BbQ@bm`7+2bX AM,
JB;m2i- aX- JQMiMp2`i- X- l#û/- aX U2/bV .Bb+`2i2
:2QK2i`v 7Q` *QKTmi2` AK;2`v- "2`HBM, aT`BM;2`c
RNNeX TX Rj@8X
(9j) GQ`2Mb2M q1- *HBM2 >1X J`+?BM; +m#2b, ?B;? `2bQ@
HmiBQM j. bm`7+2 +QMbi`m+iBQM H;Q`Bi?KX *J aA:@
:_S> *QKTmi :`T? RN3dckRU9V,RejĜNX ?iiTb,ff
/QBXQ`;fRyXRR98fjd9yRXjd9kkX
(99) .QB - EQB/2 X M 2{+B2Mi K2i?Q/ Q7 i`BM;mHiBM;
2[mB@pHm2/ bm`7+2b #v mbBM; i2i`?2/`H +2HHbX A1A*1
h`Mb AM7 avbi RNNRcd9URV,kR9@k9X
(98) "Qib+? J- EQ##2Hi G- SmHv J- HHB2x S- Gûpv "X SQHv@
;QM K2b? T`Q+2bbBM;X *_* T`2bbc kyRyX
(9e) am`x?bFv o- :QibKM *X 1tTHB+Bi bm`7+2 `2K2b?BM;X
AM, S`Q+22/BM;b Q7 i?2 kyyj 1m`Q;`T?B+bf*J aA:@
:_S> avKTQbBmK QM :2QK2i`v S`Q+2bbBM;- kyyjX TX
ky@jyX
(9d) EQ##2Hi G- "Qib+? JX `2K2b?BM; TT`Q+? iQ KmH@
iB`2bQHmiBQM KQ/2HBM;X S`Q+22/BM;b Q7 i?2 kyy9 1m`Q@
;`T?B+bf*J aA::_S> bvKTQbBmK QM :2QK2i`v
T`Q+2bbBM;c kyy9X TX R38ĜNkX
(93) EQ##2Hi G- "`2mi?2` h- a2B/2H >SX JmHiB`2bQHmiBQM
b?T2 /27Q`KiBQMb 7Q` K2b?2b rBi? /vMKB+ p2`i2t
+QMM2+iBpBivX *QKTmi :`T? 6Q`mK kyyycRNUjV,k9N@eyX
?iiTb,ff/QBXQ`;fRyXRRRRfR9ed@3e8NXyy9RdX
(9N) *?HQ2 X J2b?BM; BM 61, bi`m+im`2/ pb mM@
bi`m+im`2/ K2b?2b- ?iiTb,ffQMb+H2X+QKf#HQ;f
K2b?BM;@BM@ 72@bi`m+im`2/@pb@ mMbi`m+im`2/@K2b?2bfc
kyky (++2bb2/ Rj P+iQ#2` kykR)X
(8y) w?m >X JQ/2HBM; Q7 T`2bbm`2 /Bbi`B#miBQM Q7 ?mKM
#Q/v HQ/ QM M Q{+2 +?B` b2iX .2T`iK2Mi Q7 J2@
+?MB+H 1M;BM22`BM;- "H2FBM;2 AMbiBimi2 Q7 h2+?MQHQ;v-
Jbi2` h?2bBb- E`HbF`QM- ar2/2M- kyRjX
R9
... Smooth boundaries and irregular surfaces of the optimized designs through an additional postprocessing stage to generate smooth layouts for design validation. This can be done using the Grasshopper plugin Ameba or other smoothing techniques [48]. ...
Article
Full-text available
In this study, a topology optimization technique is developed based on the bi-directional evolutionary structural optimization (BESO) method for the creation of nonreciprocal complaint mechanisms (NCMs). The internal contact surface model is proposed as a simple and innovative approach to making complaint mechanism systems nonreciprocal. The design problem is formulated as maximizing the flexibility of NCMs with a desired level of nonreciprocity subject to a volume constraint. Based on the BESO method, a novel type of NCMs is developed with potential applications in various engineering fields. The topology optimization of a nonreciprocal inverter mechanism is studied, and the effectiveness of the proposed method is verified through experiments. The numerical and experimental results indicate that topologically optimized designs of NCMs and their asymmetric deformation can be significantly controlled by the degree of nonreciprocity. The findings from this study can be used as a basis for designing a wide range of nonreciprocal structural systems.
... The authors previously developed the smoothing algorithm utilized in this study based on a pre-built lookup table [48]. Notably, the smoothed model can be exported to external CAD software. ...
Article
Full-text available
Innovative load-bearing structures often emerge from a fine balance between creative forms and engineering principles. While preference-based topology optimization methods have advanced structural design by considering designers’ geometric preferences, they struggle with visualizing and editing complex 3D details crucial for diverse design options. To overcome this bottleneck, here we propose the improved bidirectional evolutionary structural optimization considering subjective preferences (ISP-BESO) method. This method introduces a similarity constraint that enables precise control over subjective preferences in optimized structures. Then, a design exploration strategy is proposed by integrating virtual reality (VR) with topology optimization for the interactive creation of desirable 3D structures. The strategy employs VR sculpting to offer immersive visualization and real-time feedback, guiding material redistribution during optimization. This workflow can iteratively produce innovative and efficient structures. Adjusting target similarity in ISP-BESO steers designs toward performance-driven or preference-driven outcomes. A museum design example demonstrates the practical potential of this strategy.
... The authors previously developed the smoothing algorithm utilized in this study based on a pre-built lookup table [26]. Notably, the smoothed model can be exported to external CAD software. ...
Conference Paper
Full-text available
Innovative load-bearing structures often emerge from a fine balance between creative forms and engineering principles. Recent studies have advanced the structural design process by developing preference-based topology optimization methods. Such methods optimize material distribution for structural performance while integrating designers' preferences for certain geometric features. However, these methods face challenges in the subsequent exploration phases, including observing and editing complex 3D structural details vital for achieving diverse and satisfactory design options. This paper proposes a novel design exploration strategy by integrating virtual reality (VR) with topology optimization to create desirable 3D structures interactively and iteratively. Our strategy uses VR sculpting to offer immersive visualization, intuitive design exploration, and real-time feedback, with the sculpted models guiding and influencing material redistribution in topology optimization. This sculpting-optimization workflow can be repeated in multiple cycles, creating various innovative and efficient structures. Our parametric study demonstrates that adjusting the workflow parameters can control the formation of final structures toward performance-driven or preference-driven designs. We present several computational design examples that demonstrate practical applications of the proposed strategy and highlight its potential in solving real-world problems.
Article
The Solid Isotropic Microstructure with Penalization (SIMP) model, a material interpolation model in the variable density method, has been widely applied in structural topology optimization. When combining the Sigmund sensitivity filtering method with the Optimality Criterion (OC) method for solving the SIMP model, numerical instabilities and relatively low optimization performance are often observed. To address these challenges, an improved scheme based on the variable density method is proposed. Firstly, an improved material interpolation model is proposed to overcome the deficiencies of the SIMP model in terms of penalty efficiency and convergence speed. Secondly, an improved sensitivity filtering method based on the Gaussian weight function is proposed, which not only successfully suppresses checkerboards and mesh-dependence but also contributes to a superior filtering effect. Furthermore, a new gray-scale suppression operator is designed to develop a function that suppresses gray-scale elements. This function is subsequently integrated into the iterative formula of the OC method, forming the improved OC method, which is capable of completely eliminating gray-scale elements. Finally, the feasibility and effectiveness of the improved scheme are verified through several typical numerical examples.
Article
Converting topology-optimization results into parametric models is crucial for manufacturing lightweight, high-stiffness products. However, currently available technologies cannot effectively automate and perform this conversion, especially for three-axis CNC machining. To bridge this gap, this study proposes an automatic approach for generating parametric models from topology-optimization results. First, integrating the machining characteristics of three-axis CNC machining, surface voxel accessibility analysis and different voxel clustering are carried out to determine the optimal machining directions and removable geometry of the (common) raw material model. Then, a parametric sketch contours generation method is presented for the removable geometry. This also provides the essential preparation for generating a parametric model by adding subtractive features to the raw material model. Particularly, a classified layered projection method is developed to ensure the final parametric model preserves the shape of the topology-optimization result as much as possible. This method can project and fit the removable geometry into quadratic curve sketch contours in a layer-by-layer scheme. Based on the sketch contour of each layer, the corresponding subtractive feature can be generated and added to the raw material model to remove the corresponding removable geometry. Performing this subtractive process layer by layer can generate the final parametric model of the topology-optimization result. Herein, certain constraints are also implemented during the subtractive process to ensure that the final model can preserve the stiffness of the topology-optimization result. Finally, the automatic conversion experiments on two complex and representative topology-optimization results show an average reduction of 19.3% in maximum displacement (i.e., compliance) and an average increase of 45.2% in mass as well as 28.4% increase in volume when averaging the changes across both generated parametric models compared with their original topology-optimization results. The methodological comparisons also show that the presented approach has special benefits, including the ability to convert topology-optimization results automatically and effectively into parametric models, while maintaining stiffness and being processable by three-axis CNC machining.
Book
Full-text available
Smooth Topological Design of Continuum Structures focuses on the use of a newly-proposed topology algorithm for structural optimization called Smooth-Edged Material Distribution for Optimizing Topology (SEMDOT). The book presents the basic theory of SEMDOT and explains connections between this method and the corresponding optimizers. The method is used to address the long-standing jagged edge problem facing classical topology optimization algorithms and is presented as a perfect tool for combining additive manufacturing and topology optimization, which are increasingly coupled together to produce new part designs. A range of representative case studies are also included to illustrate applications. This serves as a textbook and reference for graduate and senior undergraduate students in the area, as well as engineers in the structural optimization field. Full modifiable MATLAB codes for each chapter are available online.
Article
Full-text available
The study explores possibilities on how to approach cross-field methods, such as the design of mechanical systems via finite element modeling, with the contribution of reinforcement learning as a machine learning technique for guidance in design space. The application of the epsilon-greedy algorithm for optimizing parametric finite element model is illustrated by simulations through practical examples, namely the design of a cantilever beam and a JetVest. The results obtained clearly show that this approach can be beneficial in the field of rapid prototyping.
Article
Full-text available
Purpose Furniture plays a significant role in daily life. Advanced computational and manufacturing technologies provide new opportunities to create novel, high-performance and customized furniture. This paper aims to enhance furniture design and production by developing a new workflow in which computer graphics, topology optimization and advanced manufacturing are integrated to achieve innovative outcomes. Design/methodology/approach Workflow development is conducted by exploring state-of-the-art computational and manufacturing technologies to improve furniture design and production. Structural design and fabrication using the workflow are implemented. Findings An efficient transdisciplinary workflow is developed, in which computer graphics, topology optimization and advanced manufacturing are combined. The workflow consists of the initial design, the optimization of the initial design, the postprocessing of the optimized results and the manufacturing and surface treatment of the physical prototypes. Novel chairs and tables, including flat pack designs, are produced using this workflow. The design and fabrication processes are simple, efficient and low-cost. Both additive manufacturing and subtractive manufacturing are used. Practical implications The research outcomes are directly applicable to the creation of novel furniture, as well as many other structures and devices. Originality/value A new workflow is developed by taking advantage of the latest topology optimization methods and advanced manufacturing techniques for furniture design and fabrication. Several pieces of innovative furniture are designed and fabricated as examples of the presented workflow.
Article
Full-text available
The unique, hierarchical patterns of leaf veins have attracted extensive attention in recent years. However, it remains unclear how biological and mechanical factors influence the topology of leaf veins. In this paper, we investigate the optimization mechanisms of leaf veins through a combination of experimental measurements and numerical simulations. The topological details of three types of representative plant leaves are measured. The experimental results show that the vein patterns are insensitive to leaf shapes and curvature. The numbers of secondary veins are independent of the length of the main vein, and the total length of veins increases linearly with the leaf perimeter. By integrating biomechanical mechanisms into the topology optimization process, a transdisciplinary computational method is developed to optimize leaf structures. The numerical results show that improving the efficiency of nutrient transport plays a critical role in the morphogenesis of leaf veins. Contrary to the popular belief in the literature, this study shows that the structural performance is not a key factor in determining the venation patterns. The findings provide a deep understanding of the optimization mechanism of leaf veins, which is useful for the design of high-performance shell structures.
Article
Full-text available
The topology optimization methodology is widely applied in industrial engineering to design lightweight and efficient components. Despite that, many techniques based on structural optimization return a digital model that is far from being directly manufactured, mainly because of surface noise given by spikes and peaks on the component. For this reason, mesh post-processing is needed. Surface smoothing is one of the numerical procedures that can be applied to a triangulated mesh file to return a more appealing geometry. In literature, there are many smoothing algorithms available, but especially those based on the modification of vertex position suffer from high mesh shrinkage and loss of important geometry features like holes and surface planarity. For these reasons, an improved vertex-based algorithm based on Vollmer’s surface smoothing has been developed and introduced in this work along with two case studies included to evaluate its performances compared with existent algorithms. The innovative approach herein developed contains some sub-routines to mitigate the issues of common algorithms, and confirms to be efficient and useful in a real-life industrial context. Thanks to the developed functions able to recognize the geometry feature to be frozen during the smoothing process, the user’s intervention is not required to guide the procedure to get proper results.
Article
Full-text available
The level set method can express smooth boundaries in structural topology optimization with the level set function's zero-level contour. However, most applications still use rectangular/hexahedral mesh in finite element analysis, which results in zigzag interfaces between the void and solid phases. We propose a reaction diffusion-based level set method using the adaptive triangular/tetrahedral mesh for structural topology optimization in this work. Besides genuinely expressing smooth boundaries, such a body-fitted mesh can increase finite element analysis accuracy. Unlike the traditional upwind algorithm, the proposed method breaks through the constraint of Courant-Friedrichs-Lewy stability condition with an updating scheme based on finite element analysis. Numerical examples for minimum mean compliance and maximum output displacement at specified positions, in both 2D and 3D, converge within dozens of iterations and present elegant structures.
Article
Full-text available
Element-based topology optimization algorithms capable of generating smooth boundaries have drawn serious attention given the significance of accurate boundary information in engineering applications. The basic framework of a new element-based continuum algorithm is proposed in this paper. This algorithm is based on a smooth-edged material distribution strategy that uses solid/void grid points assigned to each element. Named Smooth-Edged Material Distribution for Optimizing Topology (SEMDOT), the algorithm uses elemental volume fractions which depend on the densities of grid points in the Finite Element Analysis (FEA) model rather than elemental densities. Several numerical examples are studied to demonstrate the application and effectiveness of SEMDOT. In these examples, SEMDOT proved to be capable of obtaining optimized topologies with smooth and clear boundaries showing better or comparable performance compared to other topology optimization methods. Through these examples, first, the advantages of using the Heaviside smooth function are discussed in comparison to the Heaviside step function. Then, the benefits of introducing multiple filtering steps in this algorithm are shown. Finally, comparisons are conducted to exhibit the differences between SEMDOT and some well-established element-based algorithms. The validation of the sensitivity analysis method adopted in SEMDOT is conducted using a typical compliant mechanism design case. In addition, this paper provides the Matlab code of SEMDOT for educational and academic purposes.
Article
Full-text available
The potential of additive manufacturing (AM) in fabricating structurally efficient yet geometrically complicated designs generated from topology optimization is widely recognised. However, various constraints presented in the additive manufacturing process are not directly considered in conventional topology optimization frameworks, potentially impairing the manufacturability of the generated topology. A most noteworthy example is the presence of overhangs, which are downward-facing regions that are not self-supporting. In such areas, traditional AM processes utilise sacrificial supports to act as scaffoldings, which result in additional time, effort and the corresponding cost. Here we present a novel method that can effectively address overhanging features in the bi-directional evolutionary structural optimization framework, creating self-supporting yet structurally efficient designs. By using a simple expression, the overhang problem is formulated as a layer-wise relationship, and elements whose modification does not create overhang are selected as candidates for the element updating scheme. To validate the effectiveness of the algorithm, numerical examples under different design conditions are presented. A comparison of the self-supporting designs against their non-self-supporting counterparts obtained by conventional topology optimization demonstrates their competitiveness in structural performance. Subsequently, physical 3D prints of the examples further prove the effectiveness of the proposed method in creating self-supporting designs.
Article
Based on the variable density method, this paper proposes a boundary density evolutionary topology optimization method. The method uses a material interpolation model without penalization. Combined with the density grading filtering method, an optimal topology with only 0/1 cells can be obtained. Compared with the solid isotropic microstructures with penalization method (SIMP), no penalty factor is required in the material interpolation model; compared with the evolutionary structural optimization method (ESO), intermediate‐density elements are allowed in the optimization process, but the concept of gradually removing the low‐utilization materials near the boundary in the ESO method is retained. After the optimal result is obtained, the structural boundary element is processed by the level set of nodal strain energy, and the optimization result with smooth boundaries similar to the level set method (LSM) can be obtained. The proposed method has the superiority of the variable density method, and it also combines the advantages of the evolutionary method and the level set method, so which is named as boundary density evolution (BDE) method. The four static and one dynamic optimization examples illustrate the stability and efficiency of the proposed method.
Article
The traditional element-based topology optimization based on material penalization typically aims at a 0/1 design. Our numerical experiments reveal that the compliance of a smooth design is overestimated when material properties of boundary intermediate elements under the fixed-mesh finite element analysis are interpolated with a material penalization model. This paper proposes a floating projection topology optimization (FPTO) method for seeking a smooth design using the ersatz material model or a 0/1 design using a material penalization model. The proposed floating projection constraint combining with the upper and lower bounds heuristically simulates 0/1 constraints of design variables in the original discrete optimization problem. Numerical examples demonstrate the capability of the proposed element-based topology optimization approach in obtaining 0/1 or smooth designs for 2D and 3D compliance minimization problems. The proposed topology optimization approach can be easily implemented under the framework of the fixed-mesh finite element analysis and provides an alternative way to form explicit topologies of structures, especially when the ersatz material model is adopted.
Article
This article proposes a design method to minimize the compliance or maximize the fundamental natural frequency of continuum structures under thin feature and support structure constraints in additive manufacturing processes. The objective functions are respectively to minimize the compliance of continuum structure under static loads and maximize the fundamental natural frequency of vibrating continuum structure. The topology optimization is performed by the bi-directional evolutionary structural optimization (BESO) method. The sensitivity expressions for minimizing the compliance or maximizing the fundamental natural frequency are derived in detail. Different thin feature constraints and the support structure constraints in different printing directions are investigated in this article. Several 2D and 3D numerical examples involving compliance minimization and eigenfrequency maximization show that the proposed method is effective in achieving convergent solid-void optimized solutions for a variety of optimization problems of continuum structures and the designs resulting from the new topology optimization algorithm are additive manufacturing friendly.
Article
Topology optimization (TO) has rapidly evolved from an academic exercise into an exciting discipline with numerous industrial applications. Various TO algorithms have been established, and several commercial TO software packages are now available. However, a major challenge in TO is the post-processing of the optimized models for downstream applications. Typically, optimal topologies generated by TO are faceted (triangulated) models, extracted from an underlying finite element mesh. These triangulated models are dense, poor quality, and lack feature/parametric control. This poses serious challenges to downstream applications such as prototyping/testing, design validation, and design exploration. One strategy to address this issue is to directly impose downstream requirements as constraints in the TO algorithm. However, this not only restricts the design space, it may even lead to TO failure. Separation of post-processing from TO is more robust and flexible. The objective of this paper is to provide a critical review of various post-processing methods and categorize them based both on targeted applications and underlying strategies. The paper concludes with unresolved challenges and future work.