Numerical simulation of welding process in large structures

Zavarivanje, kao jedan od osnovnih postupaka spajanja materijala koji je i danas u širokoj upotrebi zahvaljujući svojoj pouzdanosti, jednostavnosti i niskim troškovima, predmet je mnogih istraživanja. Cilj je zadržati cjelovitost i produžiti radni vijek zavarenih konstrukcija, ali i smanjili troškove korekcija konstrukcije deformirane uslijed zavarivanja. Tijekom procesa zavarivanja dolazi do velikog lokalnog unosa topline, što uz brzo hlađenje dovodi do pojave zaostalih naprezanja i plastičnih deformacija u području zavara i njegovoj okolini. Zaostala naprezanja, uz pojavu dodatnih naprezanja uslijed eksploatacije, često uzrokuju lomove konstrukcija, dok plastične deformacije uvode dimenzijska odstupanja koja mogu predstavljati problem pri montaži konstrukcije. Za većinu realnih konstrukcija ne postoji analitičko rješenje za izračun zaostalih naprezanja odnosno deformacija uslijed zavarivanja, a eksperimentalna ispitivanja često su zahtjevna, skupa, dugotrajna i samim time neisplativa. Zaostala naprezanja i deformacije stoga se računaju primjenom različitih metoda numeričkih simulacija. Budući da je zavarivanje izrazito nelinearan proces, numeričke simulacije i uz određena pojednostavljenja, koja neznatno smanjuju točnost, još uvijek predstavljaju računalno vrlo zahtjevne postupke koji kod velikih i složenih zavarenih konstrukcija zahtijevaju veliko vrijeme računanja. Dosadašnje metode također uz veliko vrijeme računanja zahtijevaju i znatno korisničko vrijeme pripreme simulacije pri čemu nerijetko može doći do grešaka zbog kojih se simulacija prekida. U ovom radu prikazana je novo razvijeni, računalno učinkovitiji numerički postupak simulacije procesa zavarivanja kod kojeg je računalno vrijeme trajanja značajno smanjeno. Numeričke analize provedene su u komercijalnom paketu ABAQUS, uz korištenje novog modula Abaqus Welding Interface (AWI). Bitna razlika između dosadašnjih klasičnih metoda i nove metode na kojoj se temelji AWI je u modeliranju unosa topline zavarivanjem. Za razliku od dosadašnjih klasičnih metoda kod kojih se toplina unosi modeliranjem toplinskog toka, u novo razvijenoj metodi unos topline definira se preko Dirichletovog rubnog uvjeta temperature zavara. I kod jedne i kod druge metode unos dodatnog materijala zavara modelira se primjenom tehnike rađanja elemenata. Cilj rada je verificirati i validirati navedenu metodu kroz nekoliko različitih problema: problem sučeono zavarenih ploča, problem kutno zavarenih ploča u obliku T spoja, te konačno na problemu velikog panela čime bi se ispitala mogućnost metode na problemima realne veličine. Rad se sastoji od 7 poglavlja. U prvom poglavlju dan je kratki uvod i osvrt na dosadašnje metode kroz dostupnu literaturu. U drugom poglavlju prikazane su teorijske osnove procesa zavarivanja s osnovnim jednadžbama prijenosa topline i čvrstoće materijala koje čine matematički model procesa zavarivanja, dok su u trećem poglavlju dani temelji numeričkog modela metodom konačnih elemenata. U četvrtom poglavlju opisane su dosadašnje klasične metode simuliranja procesa zavarivanja. Također, u istom poglavlju detaljno opisana je nova AWI metoda i njezina primjena sa svojim prednostima, ali i nedostacima koje su premošćene u daljnjem dijelu rada. U petom poglavlju prikazan je utjecaj odabira temperature zavara na pomake i raspodjelu zaostalih naprezanja sučeono zavarenih ploča u usporedbi s klasičnom metodom s rađanjem elemenata. Također, ispitana je varijacija metode čija je mehanička analiza izvedena bez rađanja elemenata, te je zaključeno da su rezultati zadovoljavajuće točnosti uz dodatno ubrzanje vremena trajanja. U šestom poglavlju metoda je primijenjena na modeliranje kutnog zavarivanja ploča u obliku T spoja kao uvod u modeliranje zavarivanja velikog panela. U okviru ovog poglavlja pronađena je optimalna zadana temperatura zavara, za koju se smatra da ovisi o parametrima zavarivanja, debljini zavarenih ploča, širini zavara, materijalnim svojstvima te gustoći mreže konačnih elemenata. Optimalna zadana temperatura zavara dobivena je iz usporedbe rezultata s rezultatima klasične metode rađanja čime je metoda još jednom verificirana, te usporedbom s eksperimentalnim mjerenjima iz literature čime je metoda validirana. Također, parametarskim modeliranjem od nekoliko modela s različitim udjelom 3D i ljuskastih elemenata odabran optimalni kombinirani model, odnosno potrebna širina 3D zone. Zaključeno je da su rezultati zadovoljavajuće točnosti kako u usporedbi s eksperimentom tako i s klasičnom metodom rađanja elemenata uz znatno smanjenje računalnog vremena trajanja simulacije što omogućuje simulacije velikih problema kakav je problem zavarivanja panela. Kako je model T spoja referentan modelu velikog panela prema debljinama zavarenih ploča, parametrima zavarivanja, materijalnim svojstvima i širini zavara, sve zaključke ovog poglavlja (iznos zadane temperature zavara i širina 3D zone kombiniranog modela) moguće je direktno primijeniti u zadnjem, 7. poglavlju na problemu zavarivanja velikog panela. Model panela, kao i model T spoja, preuzet je iz literature u kojoj su navedena i rješenja eksperimentalnih mjerenja u usporedbi s u literaturi korištenom inherent strain metodom. Unatoč veličini i složenosti numeričkog modela, simulacija je završena u konačnom i zadovoljavajućem vremenu čime je potvrđena njezina učinkovitost i mogućnost primjene u industrijskim problemima simulacija velikih zavarenih konstrukcija. Također, dobiveni rezultati su se pokazali zadovoljavajuće točnosti uz realnu ovisnost o načinu zavarivanja za razliku od u literaturi prikazane inherent strain metode, što je također bitno naglasiti.Welding is the subject of many studies, as it is one of the primary procedures of material joining which is still widely used due to its reliability, simplicity and low cost. The goal is to maintain the integrity and extend the service life span of welded structures, while concurrently reducing the cost of the corrections of structures deformed due to the welding. During the welding process a great amount of heat is locally generated which, together with a rapid cooling, leads to residual stresses and plastic deformation in the area in and around of the weld. Residual stresses, along with the appearance of additional stresses due to exploitation, often cause construction fractures, while plastic deformations introduce dimensional deviations which can cause problems during the structure assembly. For most real structures there is no analytical solution for residual stress and strain field distribution, while experimental tests of the welding process are often very challenging, expensive and time-consuming, hence unprofitable. Therefore, residual stresses and deformations are calculated using different methods of numerical simulation. Since welding process is highly nonlinear, numerical simulations together with some simplifications, which slightly reduce accuracy, still represent a very computer-intensive procedures which in the problems of large and geometrically-complex structures require unacceptably high computation time. Existing methods along with the high computation time require considerable user time for preparing the simulation, during which errors may occur and cause the simulation to abort. This paper aims to present the newly developed, computer-efficient numerical simulation method of welding processes in which the computation time is significantly reduced. The numerical analyzes were performed using commercial software package ABAQUS, with the use of new plugin called Abaqus Welding Interface (AWI). The fundamental difference between the previous methods and this method is heat generation modelling. Unlike the previous methods where the heat generation is modelled via heat flux input, in the newly developed method heat input is defined via the Dirichlet boundary condition of weld temperature. In both previous and the new one the entry of additional weld material is modeled by element birth and death technique. The aim is to verify and validate the specified method through several different models: the butt-welded plates, fillet-welded plates in the form of T joint, and finally, a large panel with longitudinal stiffeners which would examine the potential of method on the real size problems. The paper consists of seven chapters. The first chapter is a brief introduction and overview of the current methods described in the literature. In the second chapter, the theoretical groundwork of the welding process is given along with the basic equations of heat transfer and strength of materials that make up the mathematical model of the process of welding. The foundations of the numerical model based on the finite element method are given in the third chapter. The fourth chapter describes the current classical methods of the welding process simulation. In the same chapter the new method is described in detail along with its application, and its advantages and disadvantages which are overcome in the further part of the paper. The fifth chapter analyses the effect of different temperatures of the weld set on the distribution of displacements and residual stresses in butt welded plates in comparison with the classical element birth and death method. Also, the variation of the method, whose mechanical analysis was carried out without the element birth and death technique, is examined and it was concluded that the results are satisfactory in both accuracy and further acceleration of analysis. In the sixth chapter the method has been applied to the modeling of the fillet-welded plates in the form of a T joint as an introduction to the modeling of the welding process in the large panel with longitudinal stiffeners. Within the framework of that chapter, the optimum temperature of the weld set is found. It is considered to be dependent on the welding parameters, the thickness of the welded plates, the fillet weld leg width, material properties and the density of the finite element mesh. The optimum temperature of the weld set was obtained from the comparison of the results with the classical element birth and death method, and from the comparison with the experimental measurements obtained from the literature. By doing that, the method was validated and verified once again. Furthermore, the optimal combined model is chosen by parametric modelling of few models with different contribution of 3D and shell elements. It is concluded that the results obtained were satisfactory in accuracy both in comparison with the experimental data and the classical element birth and death method with significant reduction of the computer simulation duration which allows for the simulation of large problems such as the welding process in the large panel with longitudinal stiffeners. As the panel model is analogous to the model of T joint in the aspect of welded plates thickness, welding parameters, material properties and the fillet weld leg width, all the conclusions of this chapter (the optimum temperature of the weld set and the width of the 3D zone on the combined models) can be directly applied on panel model in the chapter 7. Panel model, as well as the T joint model, was taken from the literature along with the experimental measurement data which were compared with the results of the inherent strain method used in the same literature. Despite the size and complexity of the numerical model, simulation was completed within reasonable timeframe which confirmed methods effectiveness and applicability to the industrial problems of large welded structures