Harnessing hybrid digital twinning for decision-support in smart infrastructures

POSITION PAPER
Ha nessing hyb id digi al winning o decision-suppo in
sma in as uc u es
Huangbin Liang
1
, Bea iz Moya
2,3
, Eugene Seah
4
, Ashley Ng Kwok Weng
5
, Dominique Bailla gea
2
,
Jonas Joe in
1
, Xiaozheng Zhang
6
, F ancisco Chines a
2,3
and Eleni Cha zi
1,7
1
Singapo e-ETH Cen e, Singapo e, Singapo e
2
CNRS@CREATE, Singapo e, Singapo e
3
PIMM Labo a o y, ENSAM Ins i u e o Technology, Pa is, F ance
4
Meinha d G oup, Singapo e, Singapo e
5
CETIM-Ma co , Singapo e, Singapo e
6
TÜV SÜD, Singapo e, Singapo e
7
Depa men o Ci il En i onmen al and Geoma ic Enginee ing, ETH Zu ich, Zu ich, Swi ze land
Co esponding au ho : Bea iz Moya; Email: [email p o ec ed]
Recei ed: 05 Augus 2024; Re ised: 01 Ap il 2025; Accep ed: 01 May 2025
Keywo ds: decision-making; hyb id digi al win; esilience suppo ; sma in as uc u es; in as uc u al managemen
Abs ac
Digi al Twinning (DT) has become a main ins umen o Indus y 4.0 and he digi al ans o ma ion o manu ac u ing
and indus ial p ocesses. In his s a emen pape , we elabo a e on he po en ial o DTas a aluable ool in suppo o he
managemen o in elligen in as uc u es h oughou all s ages o hei li e cycle. We highligh he associa ed needs,
oppo uni ies, and challenges and discuss he needs om bo h he esea ch and applied pe spec i es. We elucida e he
ans o ma i e impac o digi al win applica ions o s a egic decision-making, discussing i s po en ial o si ua ion
awa eness, as well as enhancemen o sys em esilience, wi h a pa icula ocus on applica ions ha necessi a e
e icien , and o en eal- ime, o nea eal- ime, diagnos ic and p ognos ic p ocesses. In doing so, we elabo a e on he
sepa a e classes o DT, anging om simple images o a sys em, all he way o in e ac i e eplicas ha a e con inually
upda ed o e lec a moni o ed sys em a hand. We oo ou app oach in he adop ion o hyb id modeling as a seminal
ool o acili a ing winning applica ions. Hyb id modeling e e s o he syne gis ic use o da a wi h models ha ca y
enginee ing o empi ical in ui ion on he sys em beha io . We pos ula e ha mode n in as uc u es can be iewed as
cybe -physical sys ems comp ising, on he one hand, an a ay o he e ogeneous da a o di e si ied g anula i y and, on
he o he , a model (analy ical, nume ical, o o he ) ha ca ies in o ma ion on he sys em beha io . We he e o e
p opose hyb id digi al wins (HDT) as he main enable o sma and esilien in as uc u es.
Impac S a emen
We ad oca e o he adop ion o Hyb id Digi al Twinning (HDT) as a main enable o ans o ming s a egic
decision-making and enhancing sys em esilience wi hin he domain o in as uc u e. In cla i ying he modus
ope andi o DT echnologies, his pape highligh s he s eng hs and po en ial o digi al win echnologies and
aspi es o lay he ounda ions o he de elopmen o nex -gene a ion digi al wins o sma in as uc u es. This
s udy summa izes he insigh s gained om a ound- able discussion on Decision Suppo o In as uc u al Asse
© The Au ho (s), 2025. Published by Camb idge Uni e si y P ess. This is an Open Access a icle, dis ibu ed unde he e ms o he C ea i e Commons
A ibu ion licence (h p://c ea i ecommons.o g/licenses/by/4.0), which pe mi s un es ic ed e-use, dis ibu ion and ep oduc ion, p o ided he
o iginal a icle is p ope ly ci ed.
H.L. and B.M. con ibu ed equally o his wo k.
Da a-Cen ic Enginee ing (2025), 6: e43
doi:10.1017/dce.2025.10015
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Managemen , which was held as a join ini ia i e o he Fu u e Resilien Sys ems (FRS) p og am a he
Singapo e-ETH Cen e and he DESCARTES in e disciplina y excellence p og am a CNRS@CREATE.
1. In oduc ion
Enginee ing in as uc u es o m he backbones o ou socie y. Unde he manda e o Indus y 4.0, he
digi al e olu ion has b ough abou a pa adigm shi in how we design, p oduce, and in e ac wi h
physical asse s (Oz emel and Gu se , 2020). While digi iza ion has been b oadly adop ed in he con ex o
manu ac u ing and p oduc ion echnologies and he handling o indus ial asse s, i emains unde u ilized
in la ge-scale buil en i onmen s, such as in as uc u es. Building In o ma ion Modelings (BIMs)
domina e he ield, p ima ily se ing as s a ic images o he design and cons uc ion phases (Sacks
e al., 2020). Howe e , also on his scale, he concep o Digi al Twinning (DT) has he po en ial no only
o deli e in o ma ion on he s a e o he sys em “as is,”bu also o in o m decision suppo amewo ks
u he . These amewo ks ope a e h oughou he s uc u al li e cycle, namely om he s age o
manu ac u ing/cons uc ion, o he s age o ope a ion unde s anda d as well as ex eme loads and
haza ds, and inally o he decommissioning phase. To add alue, DT ep esen a ions should enable a
closed-loop exchange be ween digi al and physical asse s. This in ol es ex ac ing in o ma ion ga ne ed
om ope a ing physical sys ems (e.g., by means o moni o ing) and dis illing his in o ma ion ia he use
o digi al ep esen a ion. Finally, his analysis would be exploi ed o ac on he physical asse o p o ec
c i ical in as uc u e and gua an ee i s esilience (A gy oudis e al., 2022).
In as uc u e esilience is used he e as he main c i e ion based on which s a egic decision-making
can be made. I can be de ined as he abili y o an icipa e, p epa e o , and adap o en i onmen al changes,
as well as cope wi h, espond o, and eco e apidly om ex eme dis up ions (Cimella o e al., 2016).
Nume ous s udies in ecen yea s ha e ocused on in as uc u e esilience unde ad e se en i onmen al
impac s and exposu e o ex eme e en s. These s udies pu o h amewo ks o quan i ying and
enhancing esilience ac oss scales, om componen s and indi idual asse s, o in e connec ed ne wo ks
(Ouyang e al., 2012; Cimella o e al., 2016; Dha and Khi an, 2017; Koliou e al., 2020; Blagoje iće al.,
2023; Liang e al., 2023). This analysis is ypically conduc ed in he p e-inciden phase using simula ed
scena ios wi h s ochas ic de e io a ion/ agili y and es o a ion models, wi hou accoun ing o in o ma-
ion ha is ga he ed om he ac ual sys em o e ime. Ou p ima y ocus lies on decision-making in he
con ex o du ing-inciden and pos -inciden phases, which usually equi e as (some imes e en eal-
ime) decision-making. The p emise o such an in es iga ion assumes he a ailabili y o da a om
in as uc u al asse s and sys ems. This is nowadays jus i ied by he g owing a ailabili y o in o ma ion,
which includes no jus digi ized logs wi h inspec ion in o ma ion on s uc u al sys ems, bu also he
inc easing use o sensing echnologies o moni o hese sys ems, on bo h a pe iodic (e.g., Non Des uc i e
E alua ion) and con inuous (e.g., S uc u al Heal h Moni o ing) e alua ion (Kama io is e al., 2024).
Cu en ly, he e is no in eg a ed amewo k o quan i ying and enhancing in as uc u e esilience
based on he usion o such da a wi hin DT echniques. Hence, he objec i e o his pape is o
•cla i y he cu en landscape in e ms o a ailable DT ep esen a ions,
•de ine Hyb id Digi al Twins (HDTs) as a class o DTs ha is pa icula ly sui ed o in as uc u al
asse s, when iewed unde he p ism o cybe -physical sys ems,
•illus a e he po en ial applica ion o HDTs in suppo o decision-making o pe o man and
esilien in as uc u es,
•and inally, highligh he associa ed challenges and oppo uni ies in his espec .
2. Mo i a ion o in eg a ing HDTs in in as uc u al managemen
DT e e s o he de elopmen he c ea ion o i ual ep esen a ions o physical asse s ha in eg a e
senso da a, sys em simula ions, and analy ics. This in eg a ion p o ides decision-make s wi h
e43-2 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
unp eceden ed, o en eal- ime, insigh s in o he condi ion and beha io o physical asse s, which e e
o any physical objec , sys em, o in as uc u e holding economic alue o an o ganiza ion(e.g.,
building, b idge, wind ene gy s uc u es). Unlike adi ional pe iodical and eac i e decision-making
me hods, he in eg a ion o DT in oduces p edic i e analy ics, which a e in o med based on eal- ime
and his o ical da a collec ed om he physical asse in ope a ion. This o ecas ing po en ial suppo s
p oac i e managemen o ope a ions, main enance, and esilience agains isks and haza ds. A signi i-
can limi a ion o pu ely da a-d i en Digi al Twin (DT) models is hei lack o gene alizabili y and
in e p e abili y. This issue o en a ises om he insu iciency o ep esen a i e da a, which can lead o
o e i ing and poo pe o mance in unseen scena ios. Addi ionally, he absence o physical knowledge
in eg a ion in hese models can hinde he abili y o in e p e model p edic ions and accu a ely cap u e
in ica e dynamics needed o accu a e p edic ions, limi ing hei e ec i eness in decision suppo o
in as uc u e managemen .
Aiming o g ea e accu acy and e ec i e decision-making, we use he e m hyb id digi al winning
(HDT) o e e o an ad anced o m o digi al winning ha explici ly inco po a es a physics-based
model o he sys em (which can be nume ical, analy ical, o empi ical) wi hin a p ocess ha u he
eeds om da a. The in eg a ion o physics-based models undamen ally dis inguishes HDTs om
pu ely da a-d i en DTs by enhancing p edic i e capabili ies and ensu ing physically g ounded
p edic ions. HDTs uniquely enable he gene aliza ion o p edic ions beyond he posi ions o senso
obse a ions, acili a ing i ual sensing o sys em esponses in c i ical, unmoni o ed loca ions
(Papa heou e al., 2023; Ve o i e al., 2023). This gene aliza ion capabili y is essen ial o managing
asse s unde ex eme and changing condi ions, whe e eliance solely on a ailable senso da a would be
insu icien . By g ounding p edic ions in physical p inciples, HDTs enhance in e p e abili y, ensu ing
ha p edic ed ou comes can be alida ed and us ed, which is pa icula ly c ucial o c i ical decision-
making p ocesses. Consequen ly, HDTs empowe decision-make s o espond swi ly o dis up ions
and adap o dynamic condi ions by p o iding insigh s in o bo h moni o ed and unmoni o ed pa s o
he sys em, suppo ing a p oac i e and anspa en decision-making p ocess. This anspa ency,
essen ial o accoun abili y and us , is i al in sec o s whe e decisions ha e signi ican economic
and sa e y impac s.
This in eg a ed app oach allows decision-make s o de elop a scalable amewo k, which is adap able
ac oss dimensions such as asse size and deg ees o eedom, in e dependency, and h oughou he li e
cycle o asse s. Adap abili y e e s o he capaci y o he amewo k o emain e ec i e whe he applied o
a single asse o when scaled up o encompass an en i e ne wo k o asse s, and i s po en ial o be
consis en ly implemen ed h oughou all phases, om design and cons uc ion o ope a ion and end-o -
li e managemen . Du ing he manu ac u ing and cons uc ion phase, HDT echnology acili a es he
in eg a ion o eal- ime da a and model-based p edic i e analy ics, allowing o he op imiza ion o design
p ocesses and he embedding o esilience measu es ailo ed o an icipa ed ope a ional challenges. As
asse s ansi ion in o he ope a ional phase, HDT echnology con inuously lea ns ope a ional s a egies in
eal ime o e ec i ely manage eme ging isks and minimize down ime, especially unde ex eme
condi ions. Finally, in he decommissioning phase, HDT echnology p o ides a da a- ich basis o
execu ing cos -e ec i e s a egies by le e aging he comp ehensi e his o ical da a accumula ed o e
he ope a ional li e ime o he asse . An HDT embodies a closed-loop, dynamic, possibly eal- ime, da a-
d i en app oach o asse managemen ha no only accoun s o complex in e dependencies and cu bs
assessmen unce ain y bu also ope a es based on he cu en s a e o he sys em a he han i s ini ial
deploymen condi ions.
Despi e he clea ad an ages o he use o DTs and, in pa icula , HDTs in he con ex o
in as uc u e managemen and esilience, hei adop ion has been slow in p ac ice. This eluc ance
o en s ems om he di e se in e p e a ions and lack o cla i y su ounding he de ini ion and
applicabili y o a DT/HDT, as well as he ela i e lack o s anda ds and p o ocols o o mally aming
he use o such ools. This s a emen pape aims o cla i y he de ini ion and po en ial use o HDTs
wi hin he domain o sma in as uc u es, explo ing he need o enhance hei u ili y and maximize
hei up ake.
Da a-Cen ic Enginee ing e43-3
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
3. Hyb id digi al wins—HDTs
The implemen a ion o digi al wins p esen s i s own se o challenges. Da a in eg a ion, modeling
complexi y, anspa ency, communica ion among agen s, and e hical conce ns ela ing o au oma ed
decision-making a e signi ican challenges ha mus be add essed o ensu e an ac ionable applica ion.
Howe e , he i s s ep is o p opose a amewo k o c oss-disciplina y unde s anding ha se s he
ounda ion o any u u e de elopmen .
3.1. De ini ion and in e p e a ion o digi al wins
The concep o digi al win (DT) inds i s oo s in NASA’s Apollo XIII p ojec , whe e digi al simula o s
and a physical eplica we e connec ed o he eal spaceship o ecei e in o ma ion om i o upda e i s
ope a ing condi ion and p opose mission ules based on i s s a e, especially in c i ical condi ions (Sha o
e al., 2010). As epo ed in his documen , his was he case wi h he explosion o he oxygen anks ha
damaged he engine du ing he mission, a si ua ion in which he simula o s helped o e alua e damage and
solu ions o pe o m in o med c isis managemen .
Wi h he su ge o Indus y 4.o, DTs became a go- o e m in se e al ields; howe e , he de ini ion o he
e m may s ill appea blu ed and unclea (W igh and Da idson, 2020). Ce ain sou ces ((Alam and
Saddik, 2017; Hughes, 2018; Pla enius-Moh e al., 2020) o name a ew) de ine DTs as models,
simula o s, eplicas o exis ing phenomena, i.e., digi al eplicas o eal asse s. Al hough pa ially co ec ,
his de ini ion lacks an essen ial elemen , namely he in e ac ion wi h he physical asse . Mo e ecen
amewo ks wi hin he enginee ing con ex desc ibe a DTas a p ocess ha de ines a closed loop be ween
he physical en i y and he digi al eplica (AIAA, 2020; McClellan e al., 2022). This equi es a digi al
wo k low o in o ma ion, pa ame ized models, diagnos ic and p ognos ic algo i hms, and con ol ools,
o en agg ega ed in a isualiza ion laye , which gene a es alue o he use and acili a es decisions.
The o igin o he DTconcep may be aced back o a p esen a ion by Michael G ie es a he Uni e si y
o Michigan in 2002, which aimed o es ablish he so-called P oduc Li ecycle Managemen (PLM)
amewo k (G ie es, 2002). Howe e , he i s known de ini ion o he DT is conside ed o be he one
published by NASA in (Sha o e al., 2010). In his de ini ion, a DT is claimed o be an in eg a ed
mul iphysics, mul iscale, p obabilis ic simula ion ha uses he bes a ailable physical models, senso
upda es, lee his o y, e c., o mi o he li e o i s lying win o ecommending changes in mission p o ile
o inc ease bo h he li e span and he p obabili y o mission success, al eady signi ying he key aspec o
wo-way in e ac ion be ween he physical and digi al coun e pa .
Following his spi i , simila desc ip ions ha e been assigned o DTs (Glaessgen and S a gel, 2012;
Saddik, 2018; Xu e al., 2019; Liu e al., 2021; Kene and Bo man, 2022). The ecen AIAA posi ion
pape (AIAA, 2020) de ines a digi al win as:
A se o i ual in o ma ion cons uc s ha mimics he s uc u e, con ex and beha iou o an
indi idual/unique physical asse , o a g oup o physical asse s, is dynamically upda ed wi h da a
om i s physical win h oughou i s li e cycle and in o ms decisions ha ealise alue.
We disce n h ee main cha ac e is ics o a DT in he a ious de ini ions o e ed:
•A physical asse om which in o ma ion is ex ac ed, implying he p esence o a moni o ing sys em.
•A digi al ( i ual) ep esen a ion o he physical elemen , ep esen ed by a model ha cap u es he
beha io o he physical coun e pa . He e we dis inguish ou le els o desc ip ion: componen ,
asse , sys em, and p ocess.
•A one- o wo-way in o ma ion low p ocess, depending on he applica ion, ha links he digi al and
physical coun e pa s o ensu e con inuous acking o he beha io o he physical asse . This is used
o upda e he s a us o he digi al eplica, o e ing aluable augmen ed in o ma ion on he s a e o he
sys em, and allows o ac ing on i wi h imp o ed con idence ma gins. A one-way p ocess is also
called a “digi al shadow”(Be gs e al., 2021).
e43-4 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Dynamic Da a-D i en Applica ion Sys ems (DDDAS) (Blasch e al., 2013), is p oposed as a amewo k
o he dynamic upda e o simula o s (models) wi h da a ob ained om senso ne wo ks and moni o ing
de ices. Al hough his amewo k ocuses on he aspec o upda ing a digi al mi o (essen ially) o he
ope a ing physical sys em, he pu pose o DTs ex ends beyond compu a ional modeling and upda ing o
include pe o mance and condi ion assessmen , analysis, and op imiza ion o physical asse s h oughou
hei li e cycle.
In he li e cycle o in as uc u e sys ems, we can dis inguish i e main phases: design, cons uc ion,
ope a ion, main enance, and decommissioning. Each o hese phases can be coupled wi h digi al wins,
accompanying he e olu ion o he sys em and enhancing i s managemen and op imiza ion h oughou i s
li e cycle. Following (G ie es and Vicke s, 2017), in his wo k, we de ine DTs using a classi ica ion in
h ee essen ial ca ego ies (classes), acco ding o he pu pose se ed by he win h oughou he li e cycle:
• he Digi al Twin P o o ype (DTP)
• he Digi al Twin Ins ance (DTI)
• he Digi al Twin Agg ega e (DTA)
The i s DT class we e e o he e is he Digi al Twin P o o ype (DTP), which e lec s a i ual
ep esen a ion o a physical objec , encompassing he essen ial in o ma ion se s needed o cha ac e ize
and ab ica e a physical coun e pa ( o ins ance, equi emen s, 3D models, lis s o ma e ials,
p ocesses, se ices, and disposal p ocedu es). This class is ypically used du ing he design phase
and is closely associa ed wi h he ea u es and goals o Building In o ma ion Modeling (BIM)
(De ini ion, 2014).
In he wo k o G ie es and Vicke s (2017), a Digi al Twin Ins ance (DTI) is desc ibed as a speci ic
physical asse o which a digi al coun e pa emains linked h oughou he li e o ha physical p oduc .
He e, we adop he in e p e a ion o McClellan e al. (2022) in ela ion o he no ion o an ins ance and
de ine a DTI as he DTo an indi idual ins ance o he p oduc , once i is manu ac u ed and equipped wi h
senso s ha gene a e da a. This implies ha he DTI embodies he no ion o in o ma ion low be ween he
physical and digi al coun e pa s.
The Digi al Twin Agg ega e (DTA) (G ie es and Vicke s, 2017; McClellan e al., 2022) is desc ibed as
he agg ega ion and analysis o da a om nume ous DTIs, allowing o e iew and possible in e en ion
ega ding a se o asse s. Essen ially, i desc ibes a compu ing cons uc ha allows o ga he and analyze
da a om a ious DTIs o gain insigh s wi h espec o a b oade ange o physical p oduc s o p ocesses.
A DTA can agg ega e ins ances anging om di e en DTIs o componen s comp ising an assembly, o
mul iple ins ances om simila sys ems ha ha e agg ega ed a collec ed beha io . In he la e , DTA
ela es o he concep o lea ning om lee s o popula ions (Wo den e al., 2020), e lec ing a mo e
massi e collec ion o da a, which can enhance p edic i e and p ognos ic capabili ies a he sys em le el.
Each class o DTs will equi e di e en le els o dep h, abs ac ion, and en ichmen o p ope ly
accompany he o iginal win h oughou a ious phases o he asse ’s li e cycle. Figu e 1 has now been
e ised o illus a e hese DT classes, delinea ing he sys ema ic applica ion o DTPs, DTIs, and DTAs
ac oss a ious s ages o physical asse s. The igu e employs he example use case o wind u bine
ope a ions: DTPs aid in he design phase by simula ing and e ining u bine s uc u es. Mul iple DTIs
ep esen eal- ime ope a ional uni s equipped wi h senso s, acili a ing ongoing moni o ing and imme-
dia e adjus men s. The DTA syn hesizes insigh s om indi idual DTIs o guide sys em-wide pe o mance
assessmen s and p edic i e main enance s a egies, enhancing o e all ope a ional e iciency and he
longe i y o he asse s.
Based on he desc ip ion o each DT and he needs speci ic o each phase o he li e cycle, a ying
le els o de ail, abs ac ion, and enhancemen will be necessa y o e ec i ely accompany he o iginal
win. This e olu iona y spi i o DTs is e lec ed in Figu e 2. In hese de ini ions, in o ma ion low is
assumed o be a ailable h oughou he asse ’s li e. Models ha do no con inuously ollow a physical
asse a e me ely snapsho s, no ue DTs. In enginee ing, Real-Time Digi al Twins (RTDTs) a e digi al
ep esen a ions upda ed online, in eal o nea eal- ime, as da a become a ailable.
Da a-Cen ic Enginee ing e43-5
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.

DTs a e powe ed by he use o simula o s/models ha p o ide ep esen a ions o complex sys ems,
p ocesses, o phenomena o in e es . Cu en ly, BIM (Building In o ma ion Modeling) ep esen a ions
seem o p e ail in e ms o adop ion in p ac ice, despi e hem la gely comp ising geome ic ep esen a-
ions and me ada a eposi o ies o buil objec s. This obse a ion is p ima ily e idenced by insigh s
ga he ed om indus y ound ables, whe e expe ienced p ac i ione s emphasized he obus ness and
in eg a ion capabili ies o BIM in he cons uc ion and enginee ing sec o s. Whe eas BIMs, as mainly
adop ed oday, a e close o wha one would de ine as “as-designed geome ic models,”e ec i e DTs
equi e mo e compu a ional capabili ies. Such mo e e icien models can be ob ained ia he use o
s uc u al ( ini e elemen ) models and well-es ablished o mula ions such as luid mechanics, ansien
dynamics, and deg ada ion models. To make such models ac ionable wi hin a winning amewo k, i is
necessa y o deli e eliable, ye educed-o de ep esen a ions ha can inco po a e physics in a way ha
is manageable o he p ocess a hand. Reduced O de Models (ROMs) signi ican ly con ibu e by
o e ing swi emula ions o a moni o ed sys em wi h manageable compu a ional expenses (F angos
e al., 2010; Chines a e al., 2011; Amsallem e al., 2012; Fa ha e al., 2018; Kap eyn e al., 2020; Vlachas
e al., 2021; Aga hos e al., 2022,2024; Id issi e al., 2022). ROMs a e ma hema ical ep esen a ions o
complex sys ems ha aim o p o ide simpli ied bu accu a e p edic ions o sys em beha io . When
inco po a ing physics p inciples, such ROMs a e o en e e ed o as in usi e (Chines a and Cue o,
2014). Al hough he e a e nonin usi e, ha is, pu ely da a-d i en echniques ha employ da a om
simula ions o expe imen s o bypass physics (Ibáñez e al., 2018; He nandez e al., 2021), he imposi ion
o physics biases is o en desi able o ensu e in e p e abili y (Vlachas e al., 2012; Bacsa e al., 2023; Liu
e al., 2025).
Figu e 1. Li e cycle in eg a ion o digi al win echnologies o physical asse s, using he example o wind
a m managemen . DTPs aid in he design and decommissioning phases by simula ing and op imizing
u bine s uc u es and he decommissioning p ocess, while mul iple DTIs ep esen eal- ime ope a ional
uni s equipped wi h senso s, acili a ing ongoing moni o ing and immedia e adjus men s. The DTA
syn hesizes insigh s om indi idual DTIs o guide sys em-wide pe o mance assessmen s and p edic i e
main enance s a egies, enhancing o e all ope a ional e iciency and longe i y o he asse s. DTI and
DTA can e ol e on a empo al scale depending on he equency o he collec ing da a, whe e Real-Time
Digi al Twins (RTDTs) a e speci ic DTs ha a e upda ed in a mo e equen , eal- ime manne .
e43-6 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Accompanying he eal asse along i s use ul li e equi es he capaci y o adap a ion and e-enginee ing
along i s di e en phases, wi h lexible con igu a ions ha may ha e o espond o p e iously unseen
condi ions. In his ega d, McClellan e al. (McClellan e al., 2022) also highligh he ole o cu en
de elopmen s such as a i icial in elligence (AI), machine lea ning (ML), deep lea ning (DL), and da a
analy ics o co ec ly ill he gap be ween he simula ion model, usually de ined by known physics, and he
eal beha io pe cei ed as a manne o ex end he capabili ies o he o iginal ROMs ha ep oduce he
physics o he eal asse .
AI-in o med ROMs s ongly depend on da a quali y and a ailabili y. To o e come his limi a ion, new
echniques d i en by physical knowledge may ind pa e ns and econs uc missing in o ma ion. This
in ol es emb acing he sma da a egime, which in ol es he igh in o ma ion, a he igh momen , and
igh place. ML and DL can syne gis ically be combined wi h hyb id models, enhancing hei explain-
abili y and p edic i e po en ial (Mon áns e al., 2019; Champaney e al., 2022; Kene , 2024). Such an
ins ance has eme ged in physics-enhanced o physics-in o med modeling, which capi alizes on he usion
o physics p inciples, da a, and ML, wi h his mixing assigning di e en weigh s o he mixed compo-
nen s, as explained in (Haywood-Alexande e al., 2023). Physics-in o med digi al wins (PIDT) a e hose
digi al win ep esen a ions ha inco po a e domain-speci ic knowledge o physics p inciples and laws,
o e ing in e p e able models ha e ec i ely cap u e he sys em’s inhe en dynamics (Kap eyn and
Figu e 2. Landscape o he DT pa adigm. The HDT includes hyb id modeling o en ich simula ions wi h
aspec s o physics and machine lea ning (ML) o accu a ely mimic he beha io o eal sys ems. Such a
cons uc o e s highe in e p e abili y. Finally, cogni i e digi al win (CDT) would combine p e ious
echnologies wi h scene unde s anding and au onomous decision-making. As a esul , he DT p og es-
si ely inc eases in complexi y and oppo uni ies.
Da a-Cen ic Enginee ing e43-7
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Willcox, 2020; Liu e al., 2025). While PIDTs equi e mo e de elopmen e o , hey p o ide anspa ency
and ideli y, making hem well-sui ed o applica ions whe e unde s anding and ce i iabili y a e essen ial.
The choice be ween hese app oaches depends on he speci ic equi emen s o he p oblem a hand, which
balance p edic i e powe wi h in e p e abili y and eliabili y. Some e sa ile examples a e hose ha
employ known desc ip ions o he sys em, such as pa ial di e en ial equa ions, o algo i hms ounded in
known physical laws (Ta sis e al., 2022; Vlachas e al., 2022; Haywood-Alexande and Cha zi, 2023;
Zhang and Zhao, 2023; Yang e al., 2024), such as hose o he modynamics (He nandez e al., 2022;
Cue o and Chines a, 2023), and p ese a ion o physical quan i ies (Ki chdoe e and O iz, 2016; Bacsa
e al., 2023).
Unde his p emise, we e e o hyb id digi al wins (HDTs) as win cons uc s ha c ea e a mo e
comp ehensi e and accu a e ep esen a ion o a sys em o p ocess. He e, accu acy e lec s he abili y o
he digi al win o emain aligned wi h eal-wo ld beha io , including in p e iously unseen con ex s o
esponse o e ol ing loads and en i onmen s. As p o ided in Figu e 2, HDTs in eg a e mul iple modeling
pa adigms—combining physics-based (whi e-box) models ha o e anspa en insigh s in o unde lying
physical mechanisms wi h da a-d i en ML (black-box) app oaches ha enhance p edic i e accu acy. The
esul ing g ey-box models use in e p e abili y wi h adap abili y, enabling a iche and mo e obus digi al
ep esen a ion o physical asse s o sys ems. Speci ically, HDTs may inco po a e physics knowledge as a
ha d cons ain (physics-guided o physics-encoded) by di ec ly embedding di e en ial equa ions wi hin
he neu al ne wo k a chi ec u e, ensu ing ha p edic ions adhe e o known physical laws. Al e na i ely,
HDTs can ea physics knowledge as a so cons ain (physics-in o med) by adding he esidual o
physics-based models o he loss unc ion o guide he lea ning p ocess o o e ine he ou pu s o ML
algo i hms (Chines a e al., 2020; Haywood-Alexande e al., 2023). This in eg a ion enhances bo h he
explainabili y and anspa ency o he wins’ou pu s, while imp o ing hei capaci y o adap o a ying
loads and en i onmen s (Wagg e al., 2025). Fu he mo e, hyb id modeling allows in e p e able diag-
nos ics and gene aliza ion o hei p edic i e abili y o he win, while main aining compu a ional
e iciency. Pu ely physics-based models, while s ong in in e p e abili y, ypically lack p ac ical e i-
ciency due o slowe compu a ional speeds equi ed o p ecise simula ions. HDTs hus p esen a
compelling ad an age by combining he s eng hs o bo h physics-based models and da a-d i en
app oaches o deli e mo e eliable p edic ions and enable eal- ime moni o ing and decision suppo
ac oss a wide ange o applica ions (Wagg e al., 2020).
In his pa adigm, he e is an incipien subclass o DTs ha is expec ed o lead he nex de elopmen s in
he domain: he cogni i e digi al win (CDT) (Abbu u e al., 2020; Unal e al., 2022). Cogni ion e e s o
he se o abili ies ha encompass sensing, hinking, and easoning (Bundy e al., 2023). Al hough
esea ch applica ions ha mimic cogni ion a e s ill limi ed ( he mos common use case being la ge
language models), he app op ia e design o algo i hms can lead o he in eg a ion o some o hese
abili ies. The eme ging concep o cogni i e, o sma , digi al wins (CDT) e e s o sys ems ha
can in e ac wi h bo h physical and i ual en i onmen s o au onomously make sma e decisions
based on con ex (Abbu u e al., 2020; Zheng e al., 2022). Al hough bo h HDTs and CDTs use ML o
en ich hemsel es, HDTs end o use da a and ML o ill in gaps in he knowledge o he sys em. In
con as , CDTs use da a o complex in e p e a ion—also called pe cep ion (Moya e al., 2023)—
easoning (au onomously making decisions abou hei pe o mance), au oma ic calib a ion o imp o ed
decision-making (A cie i e al., 2021), and in e ac ion wi h he use . Al hough one o he ou comes can be
he en ichmen o HDTs, we expec CDTs o mo e comp ehensi ely cap u e he ela ionship be ween da a
and physics models. The expe in he loop complemen s he cogni i e and in e ope abili y equi emen s
o CDTs (Niloo a e al., 2023). The inco po a ion o he human cogni i e dimension wi hin he digi al
win pa adigm le e ages he expe ise and expe ien ial knowledge, se ing as a c ucial acili a o in
unde s anding he unde lying a ionale o decisions and hei app op ia eness wi hin a speci ic con ex .
Consequen ly, he expe -in- he-loop pa adigm unde sco es he signi icance o model explainabili y, a
salien ea u e du ing a ious in e ac ion phases wi hin a Cogni i e Digi al Twin (CDT).
When ex ending p edic ion/es ima ion a he sys em le el, DTs may equi e he inco po a ion o
ep esen a ions and simula ions o in e connec ed sys ems o componen s (Heussen e al., 2011; Ouyang,
e43-8 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
2014; Schluse e al., 2018; Liang and Xie, 2021). Such ep esen a ions a e de ined as Sys em-Le el
Models. Fo example, ene gy sys em ne wo k models (Heussen e al., 2011; Ouyang e al., 2017) p o ide a
de ailed unde s anding o how ene gy lows h ough a ious componen s, helping o op imize ene gy
consump ion and iden i y po en ial ine iciencies.
AR (augmen ed eali y), VR ( i ual eali y), and DT echnology connec he physical and digi al
wo lds (Badías e al., 2019; Moya e al., 2022; Ve o i e al., 2023), enhancing use in e aces o imp o e
unde s anding, collabo a ion, and decision-making in a ious ields (Michalik e al., 2022).Speci ically,
AR allows use s o o e lay digi al in o ma ion on o he eal wo ld, enhancing he abili y o unde s and
complex sys ems and p ocesses in si u. Howe e , VR c ea es a comple ely imme si e simula ion
en i onmen ha is ideal o aining scena ios, sa e y d ills, and isualiza ion o scena ios ha a e ei he
dange ous o imp ac ical o eplica e in he eal wo ld. Toge he , AR and VR enhance DTs by imp o ing
isualiza ion, in e ac ion, and simula ion capabili ies, allowing s akeholde s o analyze po en ial ou -
comes in a con olled i ual se ing, acili a ing mo e in o med decision-making. This p oac i e app oach
ans o ms indus y p ac ices in o ecas ing, oubleshoo ing, and op imizing ope a ions, u he es ab-
lishing digi al wins as essen ial in digi al ans o ma ion.
Vi ual en i onmen s o en use i ual sensing o simula e he beha io o senso s ha exis in he eal
wo ld.Al hough emo e sensing acili a es he c ea ion o accu a e DTs o in as uc u e sys ems (Do a shan
e al.,2018; Phillips and Na asimhan, 2019; Bado e al., 2022;Kaa inene al.,2022), he ea es illscena ios
whe e i is imp ac ical, expensi e, o insu icien , such as he case o assessing he load and p edic ion o he
pe o mance o DTs o wind u bine blades(Ve o ie al., 2022). These i ual senso s gene a e da a wi hin a
i ual en i onmen , which can hen be used o simula e ealis ic scena ios, es algo i hms o senso da a
p ocessing and analysis, and pe o m dynamic adap a ion wi hin i ual en i onmen s.
3.2. Role o In e ne o Things, eal- ime da a analy ics
The In e ne o Things (IoT) in ol es senso selec ion, deploymen , acquisi ion, and connec i i y. IoT
ep esen s no only he deployed sensing ne wo k, bu also he pu pose o connec ing and ans e ing
in o ma ion. Mos o he in o ma ion comes in he o m o ime se ies o image-based ep esen a ions,
collec ed ia app op ia e comp ession schemes. IoT egimes o en in ol e mul iple and he e ogeneous o
mul imodal da a sou ces. Hence, DTs mus be designed o lexibly ackle di e si ied ypes o da a inpu ,
which is usually ackled ia he aspec o usion. E en hough some measu emen s (s ains, p essu e,
empe a u e) can be di ec ly co ela ed o quan i ies o in e es , his is no ue o o he sou ces, which
deli e indi ec in o ma ion (such as ib a ion-based ones). Physically in used hyb id modeling is
equi ed o ex ac physical insigh s om di e se and indi ec da a.
In his con ex , we e isi he p e iously in oduced concep o RTDTs, which is based on eal- ime
pe o mance, e lec ing a g owing desi e o he indus y. I is impo an o p ope ly de ine wha eal ime
implies in p ac ice and o conside he app op ia e ime scale o assess he pe o mance o he sys em and
he equi ed da a low a e. We de ine an RTDTas a digi al win ha e ol es synch onously o i s physical
coun e pa , measu ing and p ocessing he changes ha occu in he physical coun e pa and co es-
pondingly upda ing he i ual eplica, and possibly implemen ing eedback (in he o m o ac ions) o he
physical asse , in an online ashion (Zippe and Died ich, 2019). Howe e , achie ing pe ec ly synch on-
ous, ha d eal- ime esponse wi h minimal delays and high sampling a es can be ine icien , equi ing
excessi e esou ces and in as uc u e, and inc easing isks o o e head and la ency. Thus, “ eal- ime”
pe o mance in a DT a ies depending on i s pu pose, anging om immedia e o pe iodic upda es,
in luenced by da a collec ion a es and iming o ela ed ac ions o decisions.”
3.3. The sma da a pa adigm
The da a collec ion p ocess can pose challenges ha equi e a comp ehensi e amewo k o in elligen
da a collec ion, p ocessing, and use. Table 1 summa izes p ima y sou ces o da a used in he cons uc ion
o DTs. Sys em loads and esponse da a a e c i ical because hey p o ide eal- ime eedback on
in as uc u e pe o mance and condi ion, o ming he basis o ope a ional digi al wins; ex e nal
Da a-Cen ic Enginee ing e43-9
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Open da a exchange. Challenges in open da a exchange include ambiguous da a owne ship, da a
p i acy conce ns (Wang e al., 2023), da a quali y and consis ency a iabili y, leading o po en ial dispu es
and limi ing he a ailabili y o ele an da a o digi al win sys ems.
Secu i y and us wo hiness o algo i hms/da a. Da a may be co up ed, ampe ed wi h, o manipu-
la ed; algo i hms used in DTs may exhibi bias and may no unde go ho ough alida ion p ocesses; he
explainabili y o AI models is o en limi ed. All hese can lead o inaccu a e ep esen a ions and lawed
decision-making ou comes (Ame i ad e al., 2023).
S anda diza ion and ce i ica ion o DT. Cu en digi al win s anda ds, including he IFC and ISO
se ies (ISO.ISO/TR 24464-2020; ISO.ISO 23247-2021; ISO.ISO 19650-1:2018; ISO.ISO 37100-2016;
ISO.ISO/IEC AWI 30173; ISO.ISO/IEC AWI 30172), IEEE se ies (IEEE.IEEE SA-P2806.1; IEEE.IEEE
SA-P3144), IEC se ies (IEC.IEC 61850-2024; IEC.IEC 62832-2020) and ITU se ies (ITU.ITU-TY.3090;
In e ope abili y amewo k o digi al win sys ems in sma ci ies and communi ies), encoun e limi a ions
hinde ing hei widesp ead adop ion and e ec i eness. One no able challenge is he lack o comp ehen-
si e co e age ac oss indus ies and applica ion domains, leading o in e ope abili y issues. Addi ionally,
he apid e olu ion o digi al win echnologies ou paces s anda d de elopmen , esul ing in ou da ed
guidance o eme ging use cases. Achie ing consensus among s akeholde s and alloca ing esou ces o
compliance also pose signi ican challenges, especially o smalle o ganiza ions o hose wi h legacy
sys ems (Bice skis e al., 2017; Hong and Huang, 2017; Ki chen e al., 2017; Bu ns e al., 2019).
Dealing wi h alse posi i es/ esponsibili y o he decision. In a legal con ex , he a ibu ion o
esponsibili y becomes a c ucial aspec , as s akeholde s may ques ion accoun abili y o any ad e se
e ec s esul ing om alse posi i es o e oneous decisions. This challenge is exace ba ed by he e ol ing
na u e o digi al win echnologies, making i essen ial o na iga e legal amewo ks ha may no ha e
caugh up wi h he apid ad ancemen s.
Human elemen / e hics o alle ia e dange s om au oma ion. Balancing he ad an ages o au oma ion
wi h e hical conside a ions, such as ai ness, accoun abili y, and anspa ency, is essen ial o p e en
dange s s emming om unchecked au oma ion, and a obus amewo k is needed o he in eg a ion o
human expe ise and e hical guidelines in o au oma ed decision-making in DTs o mi iga e isks and build
us . In addi ion, aining use s o unde s and and wo k wi h he win is c ucial o he app op ia e
in e p e a ion and use o i s in o ma ion.
Add essing hese challenges equi es collabo a i e e o s om s akeholde s ac oss indus ies, in ol -
ing policymake s, s anda ds o ganiza ions, echnology p o ide s, and end-use s, o de elop amewo ks,
s anda ds, and bes p ac ices ha p omo e he esponsible and e ec i e use o DTs o decision-making in
a apidly e ol ing echnological landscape.
5.3. Oppo uni ies
Recen pe spec i e pape s ha e highligh ed he limi a ions o cu en digi al win ools in u ban planning,
pa icula ly ega ding hei ocus on sho - e m goals e sus he long- e m ocus o ci y planning policies
(Ba y, 2024; Be encou , 2024). They no e issues such as s a ici y, limi ed agg ega ion capaci y, and a
p ima y ocus on isualiza ion. Emphasizing he need o imp o emen , hey ad oca e o modeling
mul ile el and mul idomain as well as mul i-spa io empo al scale ne wo ks be e o cap u e in e ac ions
and he dynamic na u e o u ban en i onmen s acing a ious s esso s. Fu he mo e, hese pape s
unde sco e he impo ance o obus e i ica ion, alida ion, and unce ain y quan i ica ion me hods o
enhance he eliabili y and accu acy o digi al win models. In addi ion, au ho s in (Mohammadi and
Taylo , 2021) discuss he impo ance o u ilizing Sma Ci y DT o disas e decision-making in ci ies
acing a ious s esso s. They emphasize he in eg a ion o as and slow modes in decision-making
p ocesses and highligh he need o cap u ing, p edic ing, and adap ing o u ban dynamics a a ying
paces o e ec i ely manage disas e - ela ed mo ali y and economic losses.
The ongoing s anda diza ion o DTs p esen s nume ous oppo uni ies o indus ies and s akeholde s.
S anda dized amewo ks and p o ocols acili a e seamless in e ope abili y and in eg a ion, os e ing
collabo a ion and inno a ion while educing implemen a ion cos s and isks h ough clea guidelines and
e43-16 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.

bes p ac ices. In addi ion, s anda dized da a o ma s and communica ion p o ocols enhance da a quali y,
consis ency, and secu i y, building us and con idence.
Finally, he demand o open pla o ms ha in eg a e exis ing echnologies is g owing in he as -
changing ech landscape. (Robles e al., 2023). These pla o ms a e designed o acili a e he in eg a ion o
a ious da a sou ces, senso s, de ices, and applica ions wi hin a sma ci y en i onmen . Pla o ms like
iTwinJS (Inco po a ed Ben ley Sys ems) and Open wins (Robles e al., 2023) exempli y he pi o al ole o
openness in os e ing collabo a ion, inno a ion, and in e ope abili y wi hin he digi al ealm. Ano he
example is he Digi al Twin Pla o m (DTCC Pla o m), de eloped a he Digi al Twin Ci ies Cen e, ha
inco po a es a DTCC builde (Logg e al., 2023) (Somana h e al., 2023), model and simula ion, and
isualiza ion. An example o he implemen a ion o he p ojec is ha o he ci y o Go henbu g
(Gonzalez-Cace es e al., 2024).
The s udy o au oma ion may esul in he eplacemen o human labo in a posi i e sense. Al hough
human expe ise is pi o al in he digi al win cycle, he p oposed new echnology can in e ene o
au oma e as decision-making in c ucial scena ios and imp o e he e iciency, sa e y, and well-being o
po en ial human use s.
DTs mus be buil o empowe he human, no he machine. The exploi a ion o AR, VR, o i ual
spaces (me a e se) as acili a o s can democ a ize access o in o ma ion and insigh s, enabling a b oade
audience, including s akeholde s wi h a ying le els o echnical expe ise, o in e ac wi h and unde -
s and complex sys ems and da a. This os e s c oss- unc ional collabo a ion, accele a es decision-making
p ocesses, and imp o es he o e all e ec i eness o digi al win ini ia i es.
6. Conclusion
This s a emen pape aims o se he ounda ions o he de elopmen o nex -gene a ion DTs and hei
applica ion o sma in as uc u es. We ha e iden i ied challenges in he da a acquisi ion and simula ion
ha could be add essed h ough he so-called sma pa adigms. The sma use o da a enhances da a
collec ion and p ocessing e iciency by selec ing wha , when, whe e, and a wha scale o a oid p oblems
de i ed om big da a. This, combined wi h analy ics en iched wi h physics, imp o es he in e p e a ion
and quali y o he esul s. Addi ionally, hyb id modeling p o ides an e ec i e s a egy o in eg a ing
di e se modeling me hodologies, including physics-based and da a-d i en app oaches, he eby imp o -
ing he p ecision, adap abili y, and e ec i eness in simula ing complex eal-wo ld sys ems.
Ou analysis highligh s he need o uni y languages o imp o e communica ion among pla o ms and
s akeholde s handling a ious ypes o da a. Fu he mo e, we ad oca e o explo ing he in eg a ion o
elemen s and agen s wi hin he digi al win amewo k o ully accoun o ope a ional in e ac ions and
connec ions a di e en le els. Las ly, we ecommend u he in es iga ion in o he de elopmen o he
sma digi al win amewo k o acili a e au oma ion and in elligen decision-making p ocesses ha
would enhance eac ion o unp edic able, and possibly c ucial, new scena ios.
We ad oca e o a pa adigm shi om adi ional decision-making p ac ices in in as uc u e man-
agemen owa ds mo e p oac i e, da a-d i en app oaches. We p opose de eloping digi al win–enabled
decision-making amewo ks h oughou he p ojec ’s li e cycle and discuss ad anced applica ions
including au onomous managemen , p edic i e main enance, adap i e beha io , and esilience enhance-
men . Fu he mo e, we ou line he u u e ou look o augmen ing such digi al win–enabled decision-
making amewo ks by applying expe -guided pa adigms, o ming sys em-le el pe spec i es, and
conside ing unexpec ed ex eme e en s, o make mo e in o med and comp ehensi e decisions in suppo
o in as uc u e esilience.
Acknowledgmen s. This posi ion pape has been de eloped as pa o a ound able session on he heme o Digi al Twinning and
Decision Suppo o Asse Managemen . The ound able was held in he con ex o join collabo a ion be ween he Fu u e Resilien
Sys ems (FRS) o he Singapo e-ETH Cen e and he DESCARTES in e disciplina y p og am o excellence by CNRS@CREATE.
All in ol ed sec o s akeholde s, including TÜV SÜD, ARUP, MEINHARDT, CETIM-Ma co , NAVAL G oup, Minis y o
Na ional De elopmen (MND), Land T anspo Au ho i y (LTA), and GOVTECH, a e acknowledged o hei pa icipa ion and
ac i e eedback.
Da a-Cen ic Enginee ing e43-17
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Au ho con ibu ion. Concep ualiza ion: H.L; B.M; F.C; E.C. Me hodology: H.L; B.M; F.C; E.C. P ojec adminis a ion: F.C; E.
C; D.B; J.J. Da a cu a ion: H.L; B.M; F.C; E.C. Resou ces: E.S; A.W; X.Z; F.C; E.C. Da a isualiza ion: H.L; B.M; E.C. W i ing
o iginal d a : H.L; B.M. Supe ision: F.C; E.C. W i ing – e iew/edi ing: H.L; B.M; E.S; A.W; D.B; J.J; X.Z; F.C; E.C. All au ho s
app o ed he inal submi ed d a .
Compe ing in e es s. None.
Da a a ailabili y s a emen . In his manusc ip , no da a we e p oduced o used o pu sue he esea ch s a ed.
Funding s a emen . The esea ch was conduc ed a he Singapo e-ETH Cen e, which was es ablished collabo a i ely be ween
ETH Zu ich and he Na ional Resea ch Founda ion Singapo e, and CNRS@CREATE h ough he DESCARTES p og am; bo h
esea ch p og ams suppo ed by he Na ional Resea ch Founda ion, P ime Minis e ’s O ice, Singapo e unde i s Campus o
Resea ch Excellence and Technological En e p ise (CREATE) p og amme. E. Cha zi would also like o acknowledge he suppo o
he InBlanc p ojec , i led “INdus ialisa ion o Building Li ecycle da a Accumula ion, Nume acy and Capi alisa ion,” unded unde
he Ho izon Eu ope p og amme wi h he G an Ag eemen ID 101147225. B. Moya acknowledges suppo om he F ench
go e nmen , managed by he Na ional Resea ch Agency (ANR), unde he CPJ ITTI.
E hical s anda ds. The esea ch mee s all e hical guidelines, including adhe ence o he legal equi emen s o he s udy coun y.
Re e ences
Abbu u S,Be e AJ,Jacoby M,Roman D,S ojano ic L and S ojano ic N (2020) Cogni i e digi al wins o he p ocess
indus y. In P oceedings o he he Twel h In e na ional Con e ence on Ad anced Cogni i e Technologies and Applica ions
(COGNITIVE 2020), Nice, F ance, pp. 25–29.
Abbu u S,Be e AJ,Jacoby M,Roman D,S ojano ićL and S ojano ic N. (2020) Cogni win –hyb id and cogni i e digi al
wins o he p ocess indus y. In 2020 IEEE In e na ional Con e ence on Enginee ing, Technology and Inno a ion (ICE/ITMC),
pp. 1–8. h ps://api.seman icschola .o g/Co pusID:221846414.
Aga hos K,Ta sis KE,Vlachas K and Cha zi E (2022) Pa ame ic educed o de models o ou pu -only ib a ion-based c ack
de ec ion in shell s uc u es. Mechanical Sys ems and Signal P ocessing 162, 108051.
Aga hos K,Vlachas K,Ga land A and Cha zi E (2024) Accele a ing s uc u al dynamics simula ions wi h localised phenomena
h ough ma ix comp ession and p ojec ion-based model o de educ ion. In e na ional Jou nal o Nume ical Me hods in
Enginee ing 125, e7445.
AIAA (2020) AIAA Digi al Enginee ing In eg a ion Commi ee E al. Digi al Twin: De ini ion & Value—An AIAA and AIA Posi ion
Pape . Res on, VA: AIAA.
Alam KM and Saddik AE (2017) C2ps: A digi al win a chi ec u e e e ence model o he cloud-based cybe -physical sys ems.
IEEE Access 5, 2050–2062.
Ame i ad B,Ca aneo M,Kene RS and Luciano E (2023) Ad e sa ial a i icial in elligence in insu ance: F om an example o
some po en ial emedies. Risks 11(1), 20.
Amsallem D,Zah MJ and Fa ha C (2012) Nonlinea model o de educ ion based on local educed-o de bases. In e na ional
Jou nal o Nume ical Me hods in Enginee ing 92(10), 891–916.
A cie i G,Wöl le D and Cha zi E (2021) Which model o us : Assessing he in luence o models on he pe o mance o
ein o cemen lea ning algo i hms o con inuous con ol asks. a Xi p ep in a Xi :2110.13079.
A cie i G,Hoelzl C,Schwe y O,S aub D,Papakons an inou KG and Cha zi E (2023) B idging POMDPs and Bayesian
decision making o obus main enance planning unde model unce ain y: An applica ion o ailway sys ems. Reliabili y
Enginee ing & Sys em Sa e y 239, 109496.
A gy oudis SA,Mi oulis SA,Cha zi E,Bake JW,B ilakis I,Gkoumas K,Vousdoukas M,Hynes W,Ca luccio S,Keou O,
F angopol DM and Linko I (2022) Digi al echnologies can enhance clima e esilience o c i ical in as uc u e. Clima e Risk
Managemen 35, 100387.
Bacsa K,Lai Z,Liu W,Todd M and Cha zi E (2023) Symplec ic encode s o physics-cons ained a ia ional dynamics in e ence.
Scien i ic Repo s 13(1), 2643.
Badías A,Cu i S,González D,Al a o I,Chines a F and Cue o E (2019) An augmen ed eali y pla o m o in e ac i e
ae odynamic design and analysis. In e na ional Jou nal o Nume ical Me hods in Enginee ing 120(1), 125–138.
Bado MF,Tonelli D,Poli F,Zon a D and Casas JR (2022) Digi al win o ci il enginee ing sys ems: An explo a o y e iew o
dis ibu ed sensing upda ing. Senso s 22(9), 3168.
Ba iah L,Sa i H and Debbah M (2023) Digi al win-empowe ed communica ions: A new on ie o wi eless ne wo ks. IEEE
Communica ions Magazine 61(12), 24–36.
Bassey KE,Opoku-Boa eng J,An wi BO and N iakoh A (2024) Economic impac o digi al wins on enewable ene gy
in es men s. Enginee ing Science & Technology Jou nal 5(7), 2232–2247.
Ba y M (2024) Digi al wins in ci y planning. Na u e Compu a ional Science,4(3), 192–199. h ps://doi.o g/10.1038/s43588-024-
00606-7.
e43-18 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Be gs T,Gie lings S,Aue bach T,Klink A,Sch akneppe D and Augspu ge T (2021) The concep o digi al win and digi al
shadow in manu ac u ing. P ocedia CIRP 101,81–84.
Be mejo-Ba banoj C,Moya B,Badías A,Chines a F and Cue o E (2024) The modynamics-in o med supe - esolu ion o sca ce
empo al dynamics da a. a Xi p ep in a Xi :2402.17506.
Be encou LMA (2024) Recen achie emen s and concep ual challenges o u ban digi al wins. Na u e Compu a ional Science
4(3), 150–153. h ps://doi.o g/10.1038/s43588-024-00604-9.
Bice skis J,Bice ska Z and Ka ni is G (2017) Execu able da a quali y models. P ocedia Compu e Science 104, 138–145.
Bigoni C,Zhang Z and Hes ha en JS (2020) Sys ema ic senso placemen o s uc u al anomaly de ec ion in he absence o
damaged s a es. Compu e Me hods in Applied Mechanics and Enginee ing 371, 113315.
Blagoje ićN and S ojadino ićB(2022) A demand-supply amewo k o e alua ing he e ec o esou ce and se ice cons ain s
on communi y disas e esilience. Resilien Ci ies and S uc u es 1(1), 13–32.
Blagoje ićN,He i F,Henken J,Didie M and S ojadino ićB(2023) Quan i ying disas e esilience o a communi y wi h
in e dependen ci il in as uc u e sys ems. S uc u e and In as uc u e Enginee ing 19(12), 1696–1710.
Blasch E,See ha aman G and Reinha d K (2013) Dynamic da a d i en applica ions sys em concep o in o ma ion usion.
P ocedia Compu e Science 18, 1999–2007.
Bogoe ska S,Spi idonakos M,Cha zi E,Dumo a-Jo anoska E and Hö e R (2017) A da a-d i en diagnos ic amewo k o
wind u bine s uc u es: A holis ic app oach. Senso s 17(4), 720.
Bo olini R,Rod igues R,Ala i H,Vecchia LFD and Fo cada N (2022) Digi al wins’applica ions o building ene gy
e iciency: A e iew. Ene gies 15(19), 7002.
B uneau M,Chang SE,Eguchi RT,Lee GC,DO’Rou ke T,Reinho n AM,Shinozuka M,Tie ney K,Wallace WA and Von
Win e eld D (2003) A amewo k o quan i a i ely assess and enhance he seismic esilience o communi ies. Ea hquake
Spec a 19(4), 733–752.
Budia djo A and Miglio i D (2021) Digi al Twin Sys em In e ope abili y F amewo k. Technical epo , Tech. ep. Digi al Twin
Conso ium, Eas Lansing, Michigan.
Bundy A,Cha e N and Muggle on S (2023) In oduc ion o ‘cogni i e a i icial in elligence’.Philosophical T ansac ions o he
Royal Socie y A 381.2251: 20220051.
Bu ns T,Cosg o e J and Doyle F (2019) A e iew o in e ope abili y s anda ds o indus y 4.0. P ocedia Manu ac u ing 38,
646–653.
Cap a i G,Cas elli G,Mon uo i M,Cama delli M and Mal ezzi R (2022) Digi al win o u ban planning in he g een deal e a:
A s a e o he a and u u e pe spec i es. Sus ainabili y 14(10), 6263.
Chabane S,El-Haouzi HB and Thomas P (2022) Towa d a sel -adap i e digi al win based ac i e lea ning me hod: An
applica ion o he lumbe indus y. IFAC-Pape sOnLine 55(2), 378–383.
Champaney V,Amo es VJ,Ga ois S,I as o za-Vale a L,Ghna ios C,Mon áns FJ,Cue o E and Chines a F (2022) Modeling
sys ems om pa ial obse a ions. F on ie s in Ma e ials 9, 970970.
Champaney V,Chines a F and Cue o E (2022) Enginee ing empowe ed by physics-based and da a-d i en hyb id models: A
me hodological o e iew. In e na ional Jou nal o Ma e ial Fo ming 15(3), 31.
Chauhan N (2020) Digi al wins: De ails o implemen a ion. ASHRAE Jou nal 62(10), 20–24.
Chines a F and Cue o E (2014) PGD-Based Modeling o Ma e ials, S uc u es and P ocesses. Swi ze land: Sp inge In e na ional
Publishing.
Chines a F,Lade eze P and Cue o E (2011) A sho e iew on model o de educ ion based on p ope gene alized decomposi ion.
A chi es o Compu a ional Me hods in Enginee ing 18(4), 395–404.
Chines a F,Cue o E,Abisse -Cha anne E,Du al JL and Khaldi FE (2020) Vi ual, digi al and hyb id wins: A new pa adigm in
da a-based enginee ing and enginee ed da a. A chi es o Compu a ional Me hods in Enginee ing 27, 105–134.
Cimella o GP,Renschle C,Reinho n AM and A end L (2016) Peoples: A amewo k o e alua ing esilience. Jou nal o
S uc u al Enginee ing 142(10), 04016063.
Cue o E and Chines a F (2023) The modynamics o lea ning physical phenomena. A chi es o Compu a ional Me hods in
Enginee ing 30(8), 4653–4666.
Dab owski JJ,Pagendam DE,Hil on J,Sande son C,MacKinlay D,Hus on C,Bol A and Kuhne P (2023) Bayesian
physics in o med neu al ne wo ks o da a assimila ion and spa io- empo al modelling o wild i es. Spa ial S a is ics 55, 100746.
De ini ion BIM (2014) F equen ly asked ques ions abou he na ional BIM s anda d-Uni ed S a es-na ional BIM s anda d-Uni ed
S a es. Na ionalbims anda d.o g. A chi ed om he o iginal on 16 Oc obe 2014.
Dembski F,Wössne U,Le zgus M,Rudda M and Yamu C (2020) U ban digi al wins o sma ci ies and ci izens: The case
s udy o He enbe g, Ge many. Sus ainabili y 12(6), 2307.
Deng T,Zhang K and Shen Z-JM (2021) A sys ema ic e iew o a digi al win ci y: A new pa e n o u ban go e nance owa d
sma ci ies. Jou nal o Managemen Science and Enginee ing 6(2), 125–134.
Dha TK and Khi an L (2017) A mul i-scale and mul i-dimensional amewo k o enhancing he esilience o u ban o m o
clima e change. U ban Clima e 19,72–91.
Di Lo enzo D,Champaney V,Ma zin JY,Fa ha C and Chines a F (2023) Physics in o med and da a-based augmen ed lea ning
in s uc u al heal h diagnosis. Compu e Me hods in Applied Mechanics and Enginee ing 414, 116186.
Didie M,B occa do M,Esposi o S and S ojadino ic B (2018) A composi ional demand/supply amewo k o quan i y he
esilience o ci il in as uc u e sys ems ( e-codes). Sus ainable and Resilien In as uc u e 3(2), 86–102.
Da a-Cen ic Enginee ing e43-19
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Digi al Twin Ci ies Cen e.h ps://gi hub.com/d cc-pla o m. [Online].
Do a shan S,Thomas RJ and Magui e M (2018) Fa igue c ack de ec ion using unmanned ae ial sys ems in ac u e c i ical
inspec ion o s eel b idges. Jou nal o B idge Enginee ing 23(10), 04018078.
Enoguanbho EC,Gollnow F,Walke BB,Nielsen JO and Lakes T (2021) Key challenges o land use planning and i s
en i onmen al assessmen s in he Abuja ci y- egion, Nige ia. Land 10(5), 443.
Fang Y-P and Sansa ini G (2019) Op imum pos -dis up ion es o a ion unde unce ain y o enhancing c i ical in as uc u e
esilience. Reliabili y Enginee ing & Sys em Sa e y 185,1–11.
Fa ahmand H,Xu Y and Mos a a i A (2023) A spa ial– empo al g aph deep lea ning model o u ban lood nowcas ing
le e aging he e ogeneous communi y ea u es. Scien i ic Repo s 13(1), 6768.
Fa ha C,Bos A,A e y P and Soize C (2018) Modeling and quan i ica ion o model- o m unce ain ies in eigen alue
compu a ions using a s ochas ic educed model. AIAA Jou nal 56(3), 1198–1210.
F ancis R and Beke a B (2014) A me ic and amewo ks o esilience analysis o enginee ed and in as uc u e sys ems.
Reliabili y Enginee ing & Sys em Sa e y 121,90–103.
F angos M,Ma zouk Yand Willcox K (2010) Su oga e and educed-o de modeling: A compa ison o app oaches o la ge-scale
s a is ical in e se p oblems. In La ge-Scale In e se P oblems and Quan i ica ion o Unce ain y. Wiley Online Lib a y,
pp. 123–149.
Galdelli A,D’Impe io M,Ma chello G,Mancini A,Scaccia M,Sasso M,F on oni E and Cannella F (2022) A no el emo e
isual inspec ion sys em o b idge p edic i e main enance. Remo e Sensing 14(9), 2248.
Gi elman LD,Kozhe niko MVand Kaplin DD (2020) Asse managemen in g id companies using in eg a ed diagnos ic de ices.
Ene gy Resou ces and Policies o Sus ainabili y,211.
Glaessgen E and S a gel D (2012) The digi al win pa adigm o u u e nasa and us ai o ce ehicles. In 53 d AIAA/ASME/ASCE/
AHS/ASC S uc u es, S uc u al Dynamics and Ma e ials Con e ence 20 h AIAA/ASME/AHS Adap i e S uc u es Con e ence
14 h AIAA, pp. 1818. IAAA.
Gobeawan L,Lin ES,Tandon A,Yee ATK,Khoo VHS,Teo SN,Yi S,Lim CW,Wong ST,Wise DJ,Cheng P,Liew SC,Huang
X,Li QH,Teo LS,Feke e GS and Po o MT (2018) Modeling ees o i ual Singapo e: F om da a acquisi ion o Ci yGML
models. The In e na ional A chi es o he Pho og amme y, Remo e Sensing and Spa ial In o ma ion Sciences 42,55–62.
González Chá ez CA,Bä ing M,F an zén M,Annepa a A,Gopalak ishnan D and Johansson B. 2022. Achie ing
sus ainable manu ac u ing by embedding sus ainabili y KPIs in digi al wins. In 2022 Win e Simula ion Con e ence (WSC),
pp. 1683–1694. IEEE.
Gonzalez-Cace es A,Hunge F,Fo ssén J,Somana h S,Ma k A,Nase en in V,Bohlin J,Logg A,Wäs be g B,Komisa czyk
D,Edel ik F and Hollbe g A (2024) Towa ds digi al winning o mul i-domain simula ion wo k lows in u ban design: A case
s udy in Go henbu g. Jou nal o Building Pe o mance Simula ion,1–22.
G ie es M (2002) Concep ual ideal o PLM. In P esen a ion o he P oduc Li ecycle Managemen (PLM) Cen e , Uni e si y o
Michigan.
G ie es M and Vicke s J (2017) Digi al win: Mi iga ing unp edic able, undesi able eme gen beha io in complex sys ems. In J.
Kahlen, S. Flume el and A. Al es (eds), T ansdisciplina y Pe spec i es on Complex Sys ems: New Findings and App oaches,
Cham: Sp inge , pp. 85–113.
Hämäläinen M (2021) U ban de elopmen wi h dynamic digi al wins in Helsinki ci y. IET Sma Ci ies 3(4), 201–210.
Haywood-Alexande M and Cha zi E (2023) Physics-in o med neu al ne wo ks o one-s ep-ahead p edic ion o dynamical
sys ems. In 14 h In e na ional Wo kshop on S uc u al Heal h Moni o ing (IWSHM 2023). Lancas e , PA: DES ech Publica ions,
pp. 2253–2262.
Haywood-Alexande M,Liu W,Bacsa K,Lai Z and Cha zi E (2023) Discussing he spec a o physics-enhanced machine
lea ning ia a su ey on s uc u al mechanics applica ions. a Xi p ep in a Xi :2310.20425.
He nandez Q,Badias A,Chines a F and Cue o E (2022) The modynamics-in o med g aph neu al ne wo ks. In IEEE
T ansac ions on A i icial In elligence. IEEE.
He nandez Q,Badias A,Gonzalez D,Chines a F and Cue o E (2021) Deep lea ning o he modynamics-awa e educed-o de
models om da a. Compu e Me hods in Applied Mechanics and Enginee ing 379, 113763.
Heussen K,Koch S,Ulbig A and Ande sson G (2011) Uni ied sys em-le el modeling o in e mi en enewable ene gy sou ces
and ene gy s o age o powe sys em ope a ion. IEEE Sys ems Jou nal 6(1), 140–151.
Hong J-H and Huang M-L (2017) Enabling sma da a selec ion based on da a comple eness measu es: A quali y-awa e app oach.
In e na ional Jou nal o Geog aphical In o ma ion Science 31(6), 1178–1197.
Hughes A (2018) Fo ging he Digi al Twin in Disc e e Manu ac u ing, a Vision o Uni y in he Vi ual and Real Wo lds. LNS
Resea ch e-book
Ia a e F (2014) A jou ney om big da a o sma da a. In Digi al En e p ise Design & Managemen : P oceedings o he Second
In e na ional Con e ence on Digi al En e p ise Design and Managemen DED&M 2014, pp. 25–33. Sp inge .
Ibáñez R,Abisse -Cha anne E,Amma A,González D,Cue o E,Hue a A,Du al JL and Chines a F (2018) A mul idi-
mensional da a-d i en spa se iden i ica ion echnique: The spa se p ope gene alized decomposi ion. Complexi y 2018(1),
5608286.
Id issi MEF,P aud F,Champaney V,Chines a F and Me aghni F (2022) Mul ipa ame ic modeling o composi e ma e ials
based on non-in usi e pgd in o med by mul iscale analyses: Applica ion o eal- ime s i ness p edic ion o wo en composi es.
Composi e S uc u es 302, 116228.
e43-20 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
IEC.IEC 61850-2024. Communica ion p o ocols o in elligen elec onic de ices a elec ical subs a ions. h ps://webs o e.iec.ch/
publica ion/6028. [Online].
IEC.IEC 62832-2020. Indus ial-p ocess measu emen , con ol and au oma ion-Digi al ac o y amewo k. h ps://webs o e.iec.ch/
publica ion/65858. [Online].
IEEE.IEEE SA-P2806.1. S anda d o Connec i i y Requi emen s o Digi al Rep esen a ion o Physical Objec s in Fac o y
En i onmen s. h ps://s anda ds. ieee.o g/ieee/2806.1/10370/. [Online].
IEEE.IEEE SA-P3144. S anda d o Digi al Twin Ma u i y Model and Assessmen Me hodology in Indus y. h ps://h ps://
s anda ds.ieee.o g/ieee/3144/10837/. [Online].
Igna ius M,Wong NH,Ma in M and Chen S (2019) Vi ual Singapo e in eg a ion wi h ene gy simula ion and canopy modelling
o clima e assessmen . In IOP Con e ence Se ies: Ea h and En i onmen al Science, olume 294, pp. 012018. IOP Publishing.
Inco po a ed Ben ley Sys ems. iTwin.js. h ps://www.i winjs.o g. [Online].
In e ope abili y amewo k o digi al win sys ems in sma ci ies and communi ies.h ps://www.i u.in /ITU-T/wo kp og/wp_
i em.aspx? [Online].
ISO.ISO 19650-1:2018. O ganiza ion and digi iza ion o in o ma ion abou buildings and ci il enginee ing wo ks, including
building in o ma ion modelling (BIM) - In o ma ion managemen using building in o ma ion modelling - Pa 1: Concep s and
p inciples. h ps://www.bsig oup.com. [Online].
ISO.ISO 23247-2021. Au oma ion sys em and in eg a ion-Digi al win amewo k o manu ac u ing. h ps://www.iso.o g/s and-
a d/75066.h ml. [Online].
ISO.ISO 37100-2016. Sus ainable ci ies and communi ies - Vocabula y. .h ps://s anda ds.i eh.ai/ca alog/s anda ds/sis /0d35 35d-
85e7-467e-b8ed-984 a9a66590/iso-37100-2016. [Online].
ISO.ISO/IEC AWI 30172. Digi al Twin-Use cases. h ps://www.iso.o g/s anda d/81578.h ml. [Online].
ISO.ISO/IEC AWI 30173. Digi al win-Concep s and e minology. h ps://www.iso.o g/s anda d/81442.h ml. [Online].
ISO.ISO/TR 24464-2020. Au oma ion sys ems and in eg a ion- –Indus ial da a-Visualiza ion elemen s o digi al wins. h ps://
www.iso.o g/s anda d/78836.h ml. [Online].
ITU.ITU-TY.3090. Digi al win ne wo k-Requi emen s and a chi ec u e. h ps://www.i u.in / ec/T-REC-Y.3090-202202-I/en.
[Online].
Kaa inen E,Dunphy K and Sadhu A (2022) Lida -based s uc u al heal h moni o ing: Applica ions in ci il in as uc u e
sys ems. Senso s 22(12), 4610.
Kama io is A,Cha zi E and S aub D (2022) Value o in o ma ion om ib a ion-based s uc u al heal h moni o ing ex ac ed ia
bayesian model upda ing. Mechanical Sys ems and Signal P ocessing 166, 108465.
Kama io is A,Cha zi E,S aub D,De ilis N,Goebel K,Hughes AJ,Lombae G,Papadimi iou C,Papakons an inou KG,
Pozzi M,Todd M and Wo den K (2024) Moni o ing-suppo ed alue gene a ion o managing s uc u es and in as uc u e
sys ems. a Xi p ep in a Xi :2402.00021.
Kap eyn MG and Willcox KE (2020) F om physics-based models o p edic i e digi al wins ia in e p e able machine lea ning.
a Xi p ep in a Xi :2004.11356.
Kap eyn MG,Kneze ic DJ and Willcox K (2020) Towa d p edic i e digi al wins ia componen -based educed-o de models and
in e p e able machine lea ning. In AIAA Sci ech 2020 Fo um, pp. 0418.
Kene RS (2024) Enginee ing, emula o s, digi al wins, and pe o mance enginee ing. Elec onics 13(10), 1829.
Kene RS and Bo man J (2022) The digi al win in indus y 4.0: Awide-angle pe spec i e. Quali y and Reliabili y Enginee ing
In e na ional 38(3), 1357–1366.
Khamesi AR,Shin E and Sil es i S (2020) Machine lea ning in he wild: The case o use -cen e ed lea ning in cybe physical
sys ems. In 2020 In e na ional Con e ence on COMmunica ion Sys ems & NETwo kS (COMSNETS), pp. 275–281. IEEE.
Ki chdoe e T and O iz M (2016) Da a-d i en compu a ional mechanics. Compu e Me hods in Applied Mechanics and
Enginee ing 304,81–101.
Ki chen I,Schü z D,Folme J and Vogel-Heuse B (2017) Me ics o he e alua ion o da a quali y o signal da a in indus ial
p ocesses. In 2017 IEEE 15 h In e na ional Con e ence on Indus ial In o ma ics (INDIN), pp. 819–826. IEEE.
Kla R,A idsson N and Angelakis V (2023) Digi al wins’ma u i y: The need o in e ope abili y. IEEE Sys ems Jou nal 18(1),
713–724.
Koliou M, an de Lind JW,McAllis e TP,Ellingwood BR,Dilla d M and Cu le H (2020) S a e o he esea ch in communi y
esilience: P og ess and challenges. Sus ainable and Resilien In as uc u e 5(3), 131–151.
Kuo Y-H,Pila i F,Qu T and Huang GQ (2021) Digi al win-enabled sma indus ial sys ems: Recen de elopmen s and u u e
pe spec i es. In e na ional Jou nal o Compu e In eg a ed Manu ac u ing 34(7–8), 685–689.
Labaka L,He nan es J and Sa iegi JM (2016) A holis ic amewo k o building c i ical in as uc u e esilience. Technological
Fo ecas ing and Social Change 103,21–33.
Lenk A,Bono den L,Hellmanns A,Roedde N and Jaehnichen S (2015) Towa ds a axonomy o s anda ds in sma da a. In 2015
IEEE In e na ional Con e ence on Big Da a (Big Da a), pp. 1749–1754, . h p://doi.o g/10.1109/BigDa a.2015.7363946.
Li S and Wu T (2022) Deep ein o cemen lea ning-based decision suppo sys em o anspo a ion in as uc u e managemen
unde hu icane e en s. S uc u al Sa e y 99, 102254.
Liang H and Xie Q (2021) Sys em ulne abili y analysis simula ion model o subs a ion subjec ed o ea hquakes. IEEE
T ansac ions on Powe Deli e y 37(4), 2684–2692.
Da a-Cen ic Enginee ing e43-21
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.

Liang G,Liu G,Zhao J,Liu Y,Gu J,Sun G and Dong Z (2020) Supe esolu ion pe cep ion o imp o ing da a comple eness in
sma g id s a e es ima ion. Enginee ing 6(7), 789–800.
Liang H,Blagoje ićN,Xie Q and S ojadino ićB(2023) Seismic esilience assessmen and imp o emen amewo k o elec ical
subs a ions. Ea hquake Enginee ing & S uc u al Dynamics 52(4), 1040–1058.
Liu M,Fang S,Dong H and Xu C (2021) Re iew o digi al win abou concep s, echnologies, and indus ial applica ions. Jou nal
o Manu ac u ing Sys ems 58, 346–361.
Liu W,Lai Z,Bacsa K and Cha zi E (2022) Physics-guided deep ma ko models o lea ning nonlinea dynamical sys ems wi h
unce ain y. Mechanical Sys ems and Signal P ocessing 178, 109276.
Liu W,Lai Z,S ou a CD,Bacsa K and Cha zi E (2025) Model-based unknown inpu es ima ion ia pa ially obse able ma ko
decision p ocesses. Mechanical Sys ems and Signal P ocessing 225, 112233. h ps://doi.o g/10.1016/j.ymssp.2024.112233.
Logg A,Nase en in V and Wäs be g D (2023) DTCC builde : A mesh gene a o o au oma ic, e icien , and obus mesh
gene a ion o la ge-scale ci y modeling and simula ion. Jou nal o Open Sou ce So wa e 8(86), 4928.
Makhoul N,Roohi M, an de Lind JW,Sousa H,San os LO,A gy oudis S,Ba bosa A,De as B,Ga doni P,Lee JS,e al.
(2024) Seismic esilience o in e dependen buil en i onmen o in eg a ing s uc u al heal h moni o ing and eme ging
echnologies in decision-making. S uc u al Enginee ing In e na ional 34(1), 19–33.
Ma yko skiy Y,Cla k T,Day J,Wiens M,Hende son C,Quick J,Abdallah I,Semp e i a AM,Calbimon e J-P,Cha zi E,
e al. (2024) Knowledge enginee ing o wind ene gy. Wind Ene gy Science 9(4), 883–917.
Mashaly M (2021) Connec ing he wins: A e iew on digi al win echnology & i s ne wo king equi emen s. P ocedia Compu e
Science 184, 299–305.
McClellan A,Lo enze i J,Pa one M and Fa ha C (2022) A physics-based digi al win o model p edic i e con ol o
au onomous unmanned ae ial ehicle landing. Philosophical T ansac ions o he Royal Socie y A 380(2229), 20210204.
Medina FG and He nandez VM (2025) P oduc digi al wins: An umb ella e iew and esea ch agenda o unde s anding hei
alue. Compu e s in Indus y 164, 104181.
Mema zadeh M and Pozzi M (2016) Value o in o ma ion in sequen ial decision making: Componen inspec ion, pe manen
moni o ing and sys em-le el scheduling. Reliabili y Enginee ing & Sys em Sa e y 154, 137–151.
Michalik D,Kohl P and Kumme A (2022) Sma ci ies and inno a ions: Add essing use accep ance wi h i ual eali y and
digi al win ci y. IET Sma Ci ies 4(4), 292–307.
Mille C,Hil on J,Sulli an A and P akash M (2015) Spa k–a bush i e sp ead p edic ion ool. In En i onmen al So wa e
Sys ems. In as uc u es, Se ices and Applica ions: 11 h IFIP WG 5.11 In e na ional Symposium, ISESS 2015, Melbou ne, VIC,
Aus alia, Ma ch 25–27, 2015. P oceedings 11, pp. 262–271. Sp inge .
Mohammadi N and Taylo JE (2021) Thinking as and slow in disas e decision-making wi h sma ci y digi al wins. Na u e
Compu a ional Science 1(12), 771–773.
Mon áns FJ,Chines a F,Gómez-Bomba elli R and Ku z JN (2019) Da a-d i en modeling and lea ning in science and
enginee ing. Comp es Rendus Mécanique 347(11), 845–855.
Moya B,Badías A,Al a o I,Chines a F and Cue o E (2022) Digi al wins ha lea n and co ec hemsel es. In e na ional Jou nal
o Nume ical Me hods in Enginee ing 123(13), 3034–3044.
Moya B,Badias A,Gonzalez D,Chines a F and Cue o E (2022) Physics pe cep ion in sloshing scenes wi h gua an eed
he modynamic consis ency. IEEE T ansac ions on Pa e n Analysis and Machine In elligence 45(2), 2136–2150.
Moya B,Badías A,González D,Chines a F and Cue o E (2023) A he modynamics-in o med ac i e lea ning app oach o
pe cep ion and easoning abou luids. Compu a ional Mechanics 72(3), 577–591.
Niloo a P,Laza o a-Molna S,Omi aomu F,Xu H and Li X (2023) A gene al amewo k o human-in- he-loop cogni i e
digi al wins. In 2023 Win e Simula ion Con e ence (WSC), pp. 3202–3213. IEEE.
Noch a T,Wan L,Schooling JM and Pa likad AK (2021) A socio- echnical pe spec i e on u ban analy ics: The case o ci y-scale
digi al wins. Jou nal o U ban Technology 28(1–2), 263–287.
Oli o i D,D eye S,Lebek B and B ei ne MH (2019) C ea ing he ounda ion o digi al wins in he manu ac u ing indus y: An
in eg a ed ins alled base managemen sys em. In o ma ion Sys ems and e-Business Managemen 17,89–116.
Ouyang M (2014) Re iew on modeling and simula ion o in e dependen c i ical in as uc u e sys ems. Reliabili y Enginee ing &
Sys em Sa e y 121,43–60.
Ouyang M,Dueñas-Oso io L and Min X (2012) A h ee-s age esilience analysis amewo k o u ban in as uc u e sys ems.
S uc u al Sa e y 36,23–31.
Ouyang M,Xu M,Zhang C and Huang S (2017) Mi iga ing elec ic powe sys em ulne abili y o wo s -case spa ially localized
a acks. Reliabili y Enginee ing & Sys em Sa e y 165, 144–154.
Oz emel E and Gu se S (2020) Li e a u e e iew o indus y 4.0 and ela ed echnologies. Jou nal o In elligen Manu ac u ing
31, 127–182.
Papacha alampopoulos A,Giannoulis C,S a opoulos P and Mou zis D (2020) A digi al win o au oma ed oo -cause sea ch
o p oduc ion ala ms based on KPIs agg ega ed om IoT. Applied Sciences 10(7), 2377.
Papa heou E,Ta sis KE,Ba u RS,Aga hos K,Haywood-Alexande M,Cha zi E,De ilis N and Wo den K (2023) Vi ual
sensing o shm: A compa ison be ween kalman il e s and gaussian p ocesses. In P oceedings o ISMA2022 Including
USD2022, pp. 3792–3803.
Pa k KT,Son YH and Noh SD (2021) The a chi ec u al amewo k o a cybe physical logis ics sys em o digi al- win-based
supply chain con ol. In e na ional Jou nal o P oduc ion Resea ch 59(19), 5721–5742.
e43-22 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Phillips S and Na asimhan S (2019) Au oma ing da a collec ion o obo ic b idge inspec ions. Jou nal o B idge Enginee ing
24(8), 04019075.
Pla enius-Moh M,Malaku i S,G üne S,Schmi J and Goldschmid T (2020) File-and API-based in e ope abili y o digi al
wins by model ans o ma ion: An IIoT case s udy using asse adminis a ion shell. Fu u e Gene a ion Compu e Sys ems 113,
94–105.
Poulin C and Kane MB (2021) In as uc u e esilience cu es: Pe o mance measu es and summa y me ics. Reliabili y
Enginee ing & Sys em Sa e y 216, 107926.
Psa omma is F and May G (2023) A s anda dized app oach o measu ing he pe o mance and lexibili y o digi al wins.
In e na ional Jou nal o P oduc ion Resea ch 61(20), 6923–6938.
Rasheed F,Yau K-LA and Low Y-C (2020) Deep ein o cemen lea ning o a ic signal con ol unde dis u bances: A case s udy
on Sunway Ci y, Malaysia. Fu u e Gene a ion Compu e Sys ems 109, 431–445.
Rehak D,Seno sky P and Sli ko a S (2018) Resilience o c i ical in as uc u e elemen s and i s main ac o s. Sys ems 6(2), 21.
Robles J,Ma ín C and Díaz M (2023) Open wins: An open-sou ce amewo k o he de elopmen o nex -gen composi ional
digi al wins. Compu e s in Indus y 152, 104007.
Sacks R,Gi olami M and B ilakis I (2020) Building in o ma ion modelling, a i icial in elligence and cons uc ion ech.
De elopmen s in he Buil En i onmen 4, 100011.
Saddik AE (2018) Digi al wins: The con e gence o mul imedia echnologies. IEEE Mul imedia 25(2), 87–92.
Sai ullah M,And io is C and Papakons an inou KG (2023) The ole o alue o in o ma ion in mul i-agen deep ein o cemen
lea ning o op imal decision-making unde unce ain y. In 14 h In e na ional Con e ence on Applica ions o S a is ics and
P obabili y in Ci il Enginee ing 2023. Dublin, I eland.
Schlech ingen M,San os IF and Achiche S (2013) Wind u bine condi ion moni o ing based on SCADA da a using no mal
beha io models. Pa 1: Sys em desc ip ion. Applied So Compu ing 13(1), 259–270.
Schluse M,P iggemeye M,A o L and Rossmann J (2018) Expe imen able digi al wins—S eamlining simula ion-based
sys ems enginee ing o indus y 4.0. IEEE T ansac ions on Indus ial In o ma ics 14(4), 1722–1731.
Schöbi R and Cha zi EN (2016) Main enance planning using con inuous-s a e pa ially obse able Ma ko decision p ocesses and
non-linea ac ion models. S uc u e and In as uc u e Enginee ing 12(8), 977–994.
Sch o e G and Hü zele C (2020) The digi al win o he ci y o Zu ich o u ban planning. PFG–Jou nal o Pho og amme y,
Remo e Sensing and Geoin o ma ion Science 88(1), 99–112.
Se les B (2009) Ac i e Lea ning Li e a u e Su ey. Uni e si y o Wisconsin-Madison.
Sha o M,Con oy M,Doyle R,Glaessgen E,Kemp C,LeMoigne J and Wang L (2010) D a modeling, simula ion, in o ma ion
echnology & p ocessing oadmap. Technology A ea 11,1–32.
Somana h S,Nase en in V,Ele he iou O,Sjölie D,Wäs be g BS and Logg A (2023) On p ocedu al u ban digi al win
gene a ion and isualiza ion o la ge scale da a. a Xi p ep in a Xi :2305.02242.
Tajnsek V,Pihle J and Rose M (2011) Ad anced logis ical sys ems o he main enance o o e head dis ibu ion lines h ough
dcc wi h he use o lase moni o ing. IEEE T ansac ions on Powe Deli e y 26(3), 1337–1343.
Ta sis KE,Aga hos K,Cha zi EN and De imanis VK (2022) A hie a chical ou pu -only bayesian app oach o online ib a ion-
based c ack de ec ion using pa ame ic educed-o de models. Mechanical Sys ems and Signal P ocessing 167, 108558.
To A,Liu M,Bin Muhammad Hai ul MH,Da is JG,Lee JSA,Hesse H and Nguyen HD (2021) D one-based Ai and 3D
econs uc ion o digi al win augmen a ion. In In e na ional Con e ence on Human-Compu e In e ac ion, pp. 511–529.
Sp inge .
Unal P,Albay ak O,Jomâa M and Be e AJ (2022) Da a-d i en a i icial in elligence and p edic i e analy ics o he
main enance o indus ial machine y wi h hyb id and cogni i e digi al wins. In Technologies and Applica ions o Big Da a
Value, pp. 299–319. Sp inge .
U mene a J,Izquie do J and Le u iondo U (2023) A me hodology o pe o mance assessmen a sys em le el—Iden i ica ion o
ope a ing egimes and anomaly de ec ion in wind u bines. Renewable Ene gy 205, 281–292.
Ve o i S,DiLo enzo E,Pee e s B and Cha zi E (2022) Vi ual sensing o wind u bine blade ull ield esponse es ima ion in
ope a ional modal analysis. In Model Valida ion and Unce ain y Quan i ica ion, Volume 3: P oceedings o he 39 h IMAC, A
Con e ence and Exposi ion on S uc u al Dynamics 2021, pp. 49–52. Sp inge .
Ve o i S,Gomes G,Di Lo enzo E,Pee e s B and Cha zi E (2023) In luence o he inpu model o i ual sensing o wind u bine
blades. P oceedings o ISMA2022 Including USD2022, 4537–4550.
Ve o i S,Di Lo enzo E,Pee e s B and Cha zi E (2024) Assessmen o al e na i e co a iance unc ions o join inpu -s a e
es ima ion ia Gaussian p ocess la en o ce models in s uc u al dynamics. Mechanical Sys ems and Signal P ocessing 213,
111303.
Vlachas K,Ta sis K,Aga hos K,B ink AR,Quinn D and Cha zi E (2022) On he coupling o educed o de modeling wi h
subs uc u ing o s uc u al sys ems wi h componen nonlinea i ies. In Dynamic Subs uc u es, Volume 4: P oceedings o he 39 h
IMAC, a Con e ence and Exposi ion on S uc u al Dynamics 2021, pp. 35–43. Sp inge .
Vlachas K,Naje a-Flo es D,Ma inez C,B ink AR and Cha zi E (2012) A physics-based educed o de model wi h machine
lea ning-boos ed hype - educ ion. In Topics in Modal Analysis & Pa ame e Iden i ica ion, Volume 8: P oceedings o he 40 h
IMAC, A Con e ence and Exposi ion on S uc u al Dynamics 2022, pp. 131–139. Sp inge .
Vlachas K,Ta sis K,Aga hos K,B ink AR and Cha zi E (2021) A local basis app oxima ion app oach o nonlinea pa ame ic
model o de educ ion. Jou nal o Sound and Vib a ion 502, 116055.
Da a-Cen ic Enginee ing e43-23
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.
Wagg DJ,Wo den K,Ba ho pe RJ and Ga dne P (2020) Digi al wins: S a e-o - he-a and u u e di ec ions o Modeling and
simula ion in enginee ing dynamics applica ions. ASCE-ASME Jou nal o Risk and Unce ain y in Enginee ing Sys ems, Pa B:
Mechanical Enginee ing,6(3), 030901, 2332–9017. h ps://doi.o g/10.1115/1.4046739.
Wagg DJ,Kei h Wo den and Ga dne P (2020) Digi al wins: S a e-o - he-a and u u e di ec ions o modeling and simula ion
in enginee ing dynamics applica ions. ASCE-ASME Jou nal o Risk and Unce ain y in Enginee ing Sys ems, Pa B:
Mechanical Enginee ing 6(3), 030901.
Wagg DJ,Bu C,Shephe d J,Con i ZX,Enze M and Niede e S (2025) The philosophical ounda ions o digi al winning.
Da a-Cen ic Enginee ing 6, e12.
Wang Y,Su Z,Guo S,Dai M,Luan TH and Liu Y (2023) A su ey on digi al wins: a chi ec u e, enabling echnologies, secu i y
and p i acy, and u u e p ospec s. IEEE In e ne o Things Jou nal 10(17), 14965–14987.
Whi e G,Zink A,Codecá L and Cla ke S (2021) A digi al win sma ci y o ci izen eedback. Ci ies 110, 103064.
Wo den K,Bull LA,Ga dne P,Gosliga J,Roge s TJ,C oss EJ,Papa heou E,Lin W and De ilis N (2020) A b ie
in oduc ion o ecen de elopmen s in popula ion-based s uc u al heal h moni o ing. F on ie s in Buil En i onmen 6, 146.
W igh L and Da idson S (2020) How o ell he di e ence be ween a model and a digi al win. Ad anced Modeling and Simula ion
in Enginee ing Sciences 7(1), 1–13.
Xia K,Sacco C,Ki kpa ick M,Saidy C,Nguyen L,Ki caliali A and Ha ik R (2021) A digi al win o ain deep ein o cemen
lea ning agen o sma manu ac u ing plan s: En i onmen , in e aces and in elligence. Jou nal o Manu ac u ing Sys ems 58,
210–230.
Xu Y,Sun Y,Liu X and Zheng Y (2019) A digi al- win-assis ed aul diagnosis using deep ans e lea ning. IEEE Access 7,
19990–19999.
Yang J,Langley RS and And ade L (2022) Digi al wins o design in he p esence o unce ain ies. Mechanical Sys ems and
Signal P ocessing 179, 109338.
Yang S,Kim H,Hong Y,Yee K,Maulik R and Kang N (2024) Da a-d i en physics-in o med neu al ne wo ks: A digi al win
pe spec i e. a Xi p ep in a Xi :2401.08667.
Yu W,Dillon T,Mos a a F,Rahayu W and Liu Y (2019) A global manu ac u ing big da a ecosys em o aul de ec ion in
p edic i e main enance. IEEE T ansac ions on Indus ial In o ma ics 16(1), 183–192.
Zhang J and Zhao X (2023) Digi al win o wind a ms ia physics-in o med deep lea ning. Ene gy Con e sion and Managemen
293, 117507.
Zhang W-H,Qin J,Lu D-G,Thöns S and Fabe MH (2022) Voi-in o med decision-making o shm sys em a angemen .
S uc u al Heal h Moni o ing 21(1), 37–58.
Zhao J,Feng H,Chen Q and de So o BG (2022) De eloping a concep ual amewo k o he applica ion o digi al win
echnologies o e amp building ope a ion and main enance p ocesses. Jou nal o Building Enginee ing 49, 104028.
Zheng X,Lu J and Ki i sis D (2022) The eme gence o cogni i e digi al win: Vision, challenges and oppo uni ies. In e na ional
Jou nal o P oduc ion Resea ch 60(24), 7610–7632.
Zhong C,Cheng S,Kasoa M and A cucci R (2023) Reduced-o de digi al win and la en da a assimila ion o global wild i e
p edic ion. Na u al Haza ds and Ea h Sys em Sciences 23(5), 1755–1768.
Zippe H and Died ich C (2019) Synch oniza ion o indus ial plan and digi al win. In 2019 24 h IEEE In e na ional Con e ence
on Eme ging Technologies and Fac o y Au oma ion (ETFA), pp. 1678–1681. IEEE.
Ci e his a icle: Liang H, Moya B, Seah E, Ng Kwok Weng A, Bailla gea D, Joe in J, Zhang X, Chines a F and Cha zi E (2025).
Ha nessing hyb id digi al winning o decision-suppo in sma in as uc u es. Da a-Cen ic Enginee ing, 6, e43. doi:10.1017/
dce.2025.10015
e43-24 Huangbin Liang e al.
Downloaded om h ps://www.camb idge.o g/co e. 09 Sep 2025 a 08:21:10, subjec o he Camb idge Co e e ms o use.