Added ship resistance from headboxes and transverse tunnels in full scale CFD

Abstract

Eit fullskala forsyningsskip med to baugtunnelar og headboksar er simulert i stille vatn for å estimera tilleggsmotstanden frå desse modifikasjonane. Holtrop sin formel for baugtunnelar med koeffisientar 0.005 og 0.009 for fremre og bakre tunnel svarar godt til målt motstand. Likninga ITTC føreslår for appendiksmotstand passar headboksmålingane best med ein formfaktor lik 2.9, men den venta samanhengen med Reynoldstalet kan ikkje påvisast. Summen av gjorde funn tyder at ein like godt kan simulera i modellskala for å spå motstanden frå tunnelar og headboksar.

Simuleringane løyser Reynoldsmidla Navier-Stokes likningar (RANSE). Likningane er lukka med turbulensmodellen k-omega SST. Fri overflate er modellert med vekta volumdelar (VOF). Skipsrørsler er funne med ein modell som nyttar låst jamvekt. Logaritmisk grensesjiktstilnærming er brukt for å finna væskefarten nær skrogoverflata.

Totalmotstand viser jamt god gridkonvergens, men konvergens er ikkje tydeleg for alle undersøkte verdiar. Samanlikna med Gridkonvergensindeksmetoda (GCI), gjer kurvetilpassing med minste kvadrat meir konservative og truverdige estimat for den numeriske utryggleika. For fem vurderte snøggleikar er snittet av numerisk utryggleik for totalmotstand sju prosent.

A full scale offshore supply vessel with two bow tunnels and headboxes is simulated in calm water to estimate the added resistance from these hull modifications. Holtrop's formula for bow thrusters with drag coefficients of respectively 0.005 and 0.009 is shown to approximate the added resistance from the foremost and aftmost tunnel well. Least squares fitting with ITTC's recommended formula for appendix resistance gives a form factor of 2.9 for the headboxes, but the expected Reynolds number dependence of their resistance can not be shown. By consideration of all findings, there is a clear argument that model scale simulations may instead be used for the prediction of tunnel and headbox resistance.

Simulations apply the Reynolds-averaged Navier-Stokes equations. They are closed with the k-omega SST turbulence model. The free surface is resolved with a Volume of Fluid approach, and an equilibrium model is applied for ship motions. Near wall velocities are approximated by wall functions.

While total resistance generally shows good grid convergence, convergence is not clear for all investigated values. When compared to the Grid-convergence index method, a least squares fitting approach gives more appropriately conservative estimates for the discretization uncertainty. For five considered velocities, the mean numerical uncertainty of resistance is estimated to seven percent.