2.4.2Turbulence ModelsAs mentioned above, turbulence models arean integral part of solving fluid flow within turbulent flow regimes when usingthe RANS equations. Within ANSYS FLUENT many turbulence models are availablefor selection by the user. The main ‘industry’ turbulence models will bebriefly discussed briefly as to come to an informed decision on the mostsuitable turbulence model for this research.K-epsilonmodelsThe K-epsilonturbulence model is a two-equation model and is one of the most common modelsused as it is appropriate for flows with adverse pressure gradients andinternal flows but has been known to only perform well where the mean pressuregradients are known to be small (Cfd-online.com,2017).
This model is popular due to a relatively low requirement forcomputational power, is known for good convergence and can also applied toexternal flow around bluff bodies (COMSOL Multiphysics, 2017).K-OmegaModelsThe K-omega model is also a two-equationmodel with the second equation solving for the specific rate of kinetic energydissipation as opposed to the k-epsilon model which solves for kinetic energydissipation. It is known to have issues with convergence during the initialstages of the simulation but this can be remedied by first solving with a modellike k-epsilon (COMSOLMultiphysics, 2017). The K-omega model is more suited to internal flows thatare subject to large pressure gradients and complex geometries.Researchpublished by Bayeul-Lainé, Bois, and Issa in 2010 compared the use of both the above-mentionedturbulence models on an application much like the one investigated by thisreport and found that both gave very similar results but found that the k-omegamodel produced results that were deemed to be unsatisfactory and so thek-epsilon model was used going forward. LargeEddy Simulation (LES)Large eddy simulation is considered aneffective means of modeling turbulent flows and so will have a brief mentionwithin this part of the report. It is designed to model accurately thehigh-energy areas of turbulence known as eddies and approximate the smallerlower energy eddies. Unfortunately, LES is much more demanding computationallyso will not be considered any further within this report.
1. Methodology1.1 AimsAs discussedpreviously, the purpose of this project is to carry out CFD analysis using ANSYSFLUENT to determine how the type suction fitting of a pump intake can affectthe flow as it approaches the pump impeller. The methods undertaken in thissection of the report.As previouslydiscussed in the literature review there are many factors that can affect pumpperformance and efficiency but for this research the three that will be investigatedare vortex formation, pump pre-swirl and head loss.
The objectives of thisresearch are summarized below.· Determinewhich pump suction fitting gives the lowest average pre-swirl measured over a10-minute period as per the recommendations by Hydraulic Institute (1998)American national standard for pump intake design.· Determinewhich suction fitting generates the lowest head loss and compare withcalculated head losses.· Qualitativelydetermine the severity of vortex formation and identify if the pump suctionfitting affects the severity of vortex formation in the immediate vicinity ofthe suction intake.· Determineif optimisation techniques such as suction cones and baffles have a beneficialeffect for this application.