Computational fluid dynamics methods in model- and ship-scale

Pekka Kanninen

Computational fluid dynamics methods in model- and ship-scale

Julkaisun otsikon käännös: Virtaussimulointimenetelmiä malli- ja laivamittakaavassa

Pekka Kanninen

Energia- ja konetekniikan laitos

Tutkimustuotos: Doctoral Thesis › Collection of Articles

Abstrakti

Laivamittakaavassa tehdyn virtaussimuloinnin luotettavuus on tärkeä aihe laivojen energiatehokkuuden ennustettavuuden kannalta. Laivan virtausvastuksen simulointi laivamittakaavassa vältytään skaalausvirheeltä, mikä on mallikoetekniikan tarkkuutta rajoittava tekijä. Todenmukainen laivan hydrodynamiikan simulointi edellyttää kaksifaasivirtausratkaisua, jossa veden vapaa pinta mallinnetaan. Laskenta-alueessa vapaa pinta muodostaa epäjatkuvuuden useisiin fysikaalisiin suureisiin, mikä edellyttää kehittyneitä numeerisia menetelmiä luotettavan laskentatuloksen saavuttamiseksi. Laskennallisessa virtausmekaniikassa yleisesti käytetty kontrollitilavuusmenetelmän soveltaminen epäjatkuvien kenttien simulointiin aiheuttaa numeerisia virheitä. Ensimmäinen keskeinen tutkimusaihe tässä väitöskirjassa haamufluidimenetelmän (ghost fluid method) soveltaminen vapaan pinnan diskretointiin. Menetelmä ei aiheuta numeerista virhettä kenttäratkaisuun vapaanpinnan välittömässä läheisyydessä, jota esiintyy perinteisissä menetelmissä. Haamufluidimenetelmää on tutkittu aiemmin kirjallisuudessa, mutta laajempaa kokemusta menetelmän soveltuvuudesta laivasimulaatioihin on niukasti. Tässä työssä menetelmä toteutetaan ja testataan käyttämällä avoimen lähdekoodin OpenFOAM ohjelmistoa. Menetelmästä esitetään myös korjattu versio jonka havaitaan parantavan aaltokuviota. Toinen keskeinen tutkimusaihe tässä työssä on ratkaistun turbulenssin hyödyntäminen laivasimulaatioissa. Mallimittakaavatulosten skaalaustarkkuus laivamittakaavaan liittyy virhelähteitä, jotka riippuvat Reynoldsin luvusta. Eräs virhelähde liittyy aallon murtumiseen. Ratkaisemalla turbulenssi suoraan, mallinnetun turbulenssin sijaan, laskentamallin kyky ennustaa murtuvaa aaltoa paranee. Murtuva peräaalto voi aiheuttaa ns. kastuneen laivan peräpeilin ja lisävastuksen. Ilmiö on haastava mallintaa aikakeskiarvotetussa laskennassa. Peräaaltoilmiöitä tutkittiin upotetulla tasolevyllä ja todellisella laivan geometrialla. Ratkaisemalla turbulenssi aallon murtuminen voidaan ennustaa paremmin kuin käyttämällä aikakeskiarvotettua ratkaisua. Laivamittakaavassa turbulenssin suora ratkaisu on kuitenkin laskennallisesti raskasta. Tässä työssä käytettiin päävirtauksessa ratkaistun turbulenssin laskemiseen alimallia laivan peräosassa. Alimallin sisäänvirtausreunaehdot otettiin aikakeskiarvotetusta mallista. Työn kolmas tutkimusaihe liittyi laivan rungon pinnankarheuden mallinnusmenetelmiin. Pinnankarheuden vaikutusta peräaallon muotoon tutkittiin laivamittakaavassa aikakeskiarvotetulla ja ajansuhteen tarkalla menetelmällä. Numeeristen tulosten perusteella pinnankarheus lisää rajakerrospaksuutta laivan perässä, vaikuttaa peräaallon muotoon ja irtoamiseen laivan peräpeilistä. Työssä havaitaan, että kasvava pinnakarheus lisää kastuneen peräpeilin todennäköisyyttä. Lisäksi huomattiin, että kastunut peräpeili aiheuttaa vastuskomponentin mikä ilmenee, kun nopeuden muutos on riittävän suuri peräpeilin nurkassa.

Julkaisun otsikon käännös	Virtaussimulointimenetelmiä malli- ja laivamittakaavassa
Alkuperäiskieli	Englanti
Pätevyys	Tohtorintutkinto
Myöntävä instituutio	Aalto-yliopisto
Valvoja/neuvonantaja	Vuorinen, Ville, Vastuuprofessori Vuorinen, Ville, Ohjaaja
Kustantaja	Aalto University
Painoksen ISBN	978-952-64-1074-6
Sähköinen ISBN	978-952-64-1075-3
Tila	Julkaistu - 2022
OKM-julkaisutyyppi	G5 Artikkeliväitöskirja

Tutkimusalat

laivamittakaava
haamufluidimenetelmä

YK:n kestävän kehityksen tavoitteet

Tämä tuotos edistää seuraavia kestävän kehityksen tavoitteita:

Pääsy asiakirjaan

http://urn.fi/URN:ISBN:978-952-64-1075-3

Siteeraa tätä

@phdthesis{560e64159950454b9dad472278c46c86,

title = "Computational fluid dynamics methods in model- and ship-scale",

abstract = "The reliability of a full-scale flow simulation is an important topic in terms of predictability of the energy efficiency in ships. Using full-scale resistance simulations, it is possible to avoid a scaling effects, which limit accuracy of model-scale predictions. An advanced ship hydrodynamic simulation requires a two-phase flow solution where the free surface is taken into account. In the computational domain, the free surface causes a discontinuity to several physical quantities, which requires advanced numerical methods. The finite volume method (FVM) is commonly used in fluid mechanics to discretize the governing equations. However, the standard FVM causes a numerical error at the discontinuity location. The first research topic in this thesis is the application of the ghost fluid method (GFM) for the free surface discretization. The method does not cause unphysical numerical oscillation in the free surface. Respectively, traditional discretization methods cause so-called spurious velocity. The GFM is already presented in the literature, but more ship simulations testing is needed. In this work, the method is implemented and evaluated using the open source OpenFOAM software. A corrected version of the method is also presented, which is found to improve the wave pattern. Another key research topic in this work is the utilization of resolved turbulence in the ship simulations. The extrapolation of model-scale results to the ship-scale depends on the Reynolds number. One error source relates to the wave breaking. By resolving turbulence directly, instead of modelling it, the capability of the computational model to predict the breaking wave increases. A breaking stern wave causes a so-called wetted transom and additional resistance. This phenomenon is typically challenging to resolve with a time-averaged simulation. The stern wave phenomenon is investigated with an immersed flat plate and real ship geometry. By using resolved turbulence simulation, the wave breaking can be predicted better compared to the time-averaged solution. However, resolving turbulence in ship simulations is computationally demanding. In this work, a submodel approach was utilized where the turbulence is resolved only in the ship aft part. The inflow boundary conditions to the submodel were taken from the time-averaged results. The third research topic of this thesis is related to the ship hull surface roughness modeling methods. The surface roughness effect to stern wave shape was investigated by using time-averaged and time-accurate methods. Based on the numerical results the surface roughness increases the boundary layer thickness in the aft ship, affects the stern wave shape and the wave separation from the transom. Here, it was noticed that increasing the surface roughness increases the probability of a wet transom. Additionally, it was noted that the wetted transom causes a resistance component which appears when the speed change is large enough in the transom corner.",

keywords = "Ship-scale, CFD, Ghost Fluid Method, laivamittakaava, haamufluidimenetelm{\"a}, Ship-scale, CFD, Ghost Fluid Method",

author = "Pekka Kanninen",

year = "2022",

language = "English",

isbn = "978-952-64-1074-6",

series = "Aalto University publication series DOCTORAL THESES",

publisher = "Aalto University",

number = "195",

address = "Finland",

school = "Aalto University",

}

TY - BOOK

T1 - Computational fluid dynamics methods in model- and ship-scale

AU - Kanninen, Pekka

PY - 2022

Y1 - 2022

N2 - The reliability of a full-scale flow simulation is an important topic in terms of predictability of the energy efficiency in ships. Using full-scale resistance simulations, it is possible to avoid a scaling effects, which limit accuracy of model-scale predictions. An advanced ship hydrodynamic simulation requires a two-phase flow solution where the free surface is taken into account. In the computational domain, the free surface causes a discontinuity to several physical quantities, which requires advanced numerical methods. The finite volume method (FVM) is commonly used in fluid mechanics to discretize the governing equations. However, the standard FVM causes a numerical error at the discontinuity location. The first research topic in this thesis is the application of the ghost fluid method (GFM) for the free surface discretization. The method does not cause unphysical numerical oscillation in the free surface. Respectively, traditional discretization methods cause so-called spurious velocity. The GFM is already presented in the literature, but more ship simulations testing is needed. In this work, the method is implemented and evaluated using the open source OpenFOAM software. A corrected version of the method is also presented, which is found to improve the wave pattern. Another key research topic in this work is the utilization of resolved turbulence in the ship simulations. The extrapolation of model-scale results to the ship-scale depends on the Reynolds number. One error source relates to the wave breaking. By resolving turbulence directly, instead of modelling it, the capability of the computational model to predict the breaking wave increases. A breaking stern wave causes a so-called wetted transom and additional resistance. This phenomenon is typically challenging to resolve with a time-averaged simulation. The stern wave phenomenon is investigated with an immersed flat plate and real ship geometry. By using resolved turbulence simulation, the wave breaking can be predicted better compared to the time-averaged solution. However, resolving turbulence in ship simulations is computationally demanding. In this work, a submodel approach was utilized where the turbulence is resolved only in the ship aft part. The inflow boundary conditions to the submodel were taken from the time-averaged results. The third research topic of this thesis is related to the ship hull surface roughness modeling methods. The surface roughness effect to stern wave shape was investigated by using time-averaged and time-accurate methods. Based on the numerical results the surface roughness increases the boundary layer thickness in the aft ship, affects the stern wave shape and the wave separation from the transom. Here, it was noticed that increasing the surface roughness increases the probability of a wet transom. Additionally, it was noted that the wetted transom causes a resistance component which appears when the speed change is large enough in the transom corner.

AB - The reliability of a full-scale flow simulation is an important topic in terms of predictability of the energy efficiency in ships. Using full-scale resistance simulations, it is possible to avoid a scaling effects, which limit accuracy of model-scale predictions. An advanced ship hydrodynamic simulation requires a two-phase flow solution where the free surface is taken into account. In the computational domain, the free surface causes a discontinuity to several physical quantities, which requires advanced numerical methods. The finite volume method (FVM) is commonly used in fluid mechanics to discretize the governing equations. However, the standard FVM causes a numerical error at the discontinuity location. The first research topic in this thesis is the application of the ghost fluid method (GFM) for the free surface discretization. The method does not cause unphysical numerical oscillation in the free surface. Respectively, traditional discretization methods cause so-called spurious velocity. The GFM is already presented in the literature, but more ship simulations testing is needed. In this work, the method is implemented and evaluated using the open source OpenFOAM software. A corrected version of the method is also presented, which is found to improve the wave pattern. Another key research topic in this work is the utilization of resolved turbulence in the ship simulations. The extrapolation of model-scale results to the ship-scale depends on the Reynolds number. One error source relates to the wave breaking. By resolving turbulence directly, instead of modelling it, the capability of the computational model to predict the breaking wave increases. A breaking stern wave causes a so-called wetted transom and additional resistance. This phenomenon is typically challenging to resolve with a time-averaged simulation. The stern wave phenomenon is investigated with an immersed flat plate and real ship geometry. By using resolved turbulence simulation, the wave breaking can be predicted better compared to the time-averaged solution. However, resolving turbulence in ship simulations is computationally demanding. In this work, a submodel approach was utilized where the turbulence is resolved only in the ship aft part. The inflow boundary conditions to the submodel were taken from the time-averaged results. The third research topic of this thesis is related to the ship hull surface roughness modeling methods. The surface roughness effect to stern wave shape was investigated by using time-averaged and time-accurate methods. Based on the numerical results the surface roughness increases the boundary layer thickness in the aft ship, affects the stern wave shape and the wave separation from the transom. Here, it was noticed that increasing the surface roughness increases the probability of a wet transom. Additionally, it was noted that the wetted transom causes a resistance component which appears when the speed change is large enough in the transom corner.

KW - Ship-scale

KW - CFD

KW - Ghost Fluid Method

KW - laivamittakaava

KW - haamufluidimenetelmä

KW - Ship-scale

KW - CFD

KW - Ghost Fluid Method

M3 - Doctoral Thesis

SN - 978-952-64-1074-6

T3 - Aalto University publication series DOCTORAL THESES

PB - Aalto University

ER -

Computational fluid dynamics methods in model- and ship-scale

Abstrakti

Tutkimusalat

YK:n kestävän kehityksen tavoitteet

Pääsy asiakirjaan

Sormenjälki

Siteeraa tätä