Scaling of axial fan noise

Simple item page
dc.contributor.advisorVan der Spuy, S. J.en_ZA
dc.contributor.advisorReuter, H. C. R.en_ZA
dc.contributor.authorDu Plessis, Clariceen_ZA
dc.contributor.otherStellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering.en_ZA
dc.date.accessioned2022-03-01T12:44:59Zen_ZA
dc.date.accessioned2022-04-29T09:41:46Zen_ZA
dc.date.available2022-03-01T12:44:59Zen_ZA
dc.date.available2022-04-29T09:41:46Zen_ZA
dc.date.issued2022-04en_ZA
dc.descriptionThesis (PhD)--Stellenbosch University, 2022.en_ZA
dc.description.abstractENGLISH SUMMARY: Axial flow fans play an important part in industrial cooling applications, such as air-cooled condensers (ACCs). Reliable methods of predicting the noise generated by a full-scale axial fan from a scale model at the design stage would result in less uncertainty and financial risk to cooling system suppliers. Axial fan noise is separated into its main mechanisms, namely trailing edge (TE), tip vortex formation and turbulence ingestion noise. TE noise of a 1.542 m diameter M-fan is predicted by dividing the fan blade into segments using strip theory. Large Eddy Simulation (LES) is performed for NACA 0012, LS 0409 and LS 0413 airfoils for a range of chord lengths, free stream velocities and angles of attack. Far field noise is predicted by implementing the Ffowcs-Williams Hawkings analogy with unsteady pressure fluctuations on the airfoil surface as source term. A method is developed with which the noise computed for an airfoil with a small modelled spanwise length can be corrected for an increased spanwise length if the main noise mechanism is not turbulence. Existing airfoil noise scaling methods are investigated, and a modified method is proposed. Free stream velocity, angle of attack and TE boundary layer displacement thickness are averaged over each fan blade segment and the LS 0409 and LS 0413 self-noise are scaled to represent each segment of the M-fan, which is logarithmically summed to represent the full fan. Simulations of the 4- and 8-bladed M-fans are performed using a steady state Reynolds Averaged Navier Stokes (RANS) turbulence model at various volume flow rates and tip clearances. From these simulations, the tip vortex formation noise is computed from a semi-empirical model. The tip vortex causes increased turbulence in the flow, resulting in turbulence ingestion noise generated at the blade leading edge. Assuming isotropic turbulence, turbulence ingestion noise can be calculated using an analytical model, validated on airfoils. For each fan blade segment, the inputs to this model are the relative tangential velocity, turbulent intensity, kinetic energy and dissipation rate at the relative velocity streamline reaching the airfoil leading edge. It is shown that the inputs from a scale model can be modified and applied to predict the turbulence ingestion noise for the same fan for a different diameter. Experimental measurements of the 1.542 m diameter M-fan noise are conducted in an ISO 5801, Type A (free inlet, free outlet) test facility. Corrections are made for the effect of reflections on the measured SPL spectrum. The predicted fan noise agrees well with the measured spectrum (within 0.06 dB). The noise for an M-fan with a 3.084 m diameter is also predicted. Finally, it is recommended by the frequency spectrum of the total noise be scaled with a factor proportional to the ratio of Dfan,1/Dfan,2 (Dfan,1 is the original fan diameter, Dfan,2 is the fan diameter to which the SPL is scaled). For a doubling in fan diameter, the total SPL in this case is found to scale well with the ratio of diameters to the power of 1.67.en_ZA
dc.description.abstractAFRIKAANS OPSOMMING: Aksiaalvloeiwaaiers speel ‘n belangrike rol in industriële verkoelingstoepassings, soos lugverkoelde kondensors. Betroubare maniere om geraas wat deur ‘n volskaal aksiaalwaaier vrygestel word te voorspel vanaf ‘n skaalmodel tydens die ontwerpstadium sal minder onsekerheid en finansiële risiko bied vir verskaffers van verkoelingstelsels. Aksiaalwaaiergeraas word opgebreek in sy hoofmeganismes, naamlik agterrand-, lempuntkolkvorming- en turbulensie-inname geraas. Agterrand geraas van ‘n 1.542 m diameter M-waaier is voorspel deur die waaierlem in segmente op te breek met strookteorie. LES is uitgevoer vir NACA 0012, LS 0409 en LS 0413 lemme vir ‘n reeks van koordlengtes, vrystroom snelhede en aanvalshoeke. Ver-veld geraas is voorspel deur die Ffowcs-Williams Hawkings analogie toe te pas, met ongestadigde drukfluktuasies op die lemoppervlak as bron. ‘n Metode is ontwikkel waarmee geraas wat vir ‘n lem met ‘n klein spanwyse lengte bereken is gekorrigeer kan word vir ‘n toename in spanwyse lengte, as die hoofbron van geraas nie turbulensie is nie. Bestaande lem geraasskaleringsmetodes is ondersoek en ‘n gewysigde metode is voorgestel. Gemiddeldes van die vrystroom snelheid, aanvalshoek en grenslaag verplasingsdikte by die agterrand is bereken oor elke waaierlem segment. Die LS 0409 en LS 0413 eie-geraas is geskaleer om elke segment van die M-waaier voor te stel, wat dan logaritmies gesommeer is om die hele waaier voor te stel. Simulasies van die 4- en 8-lem M-waaiers is uitgevoer met ‘n gestadigde RANS turbulensiemodel vir ‘n reeks volumevloeitempo’s en lempuntgapings. Vanaf hierdie simulasies is die lempuntkolkvorming geraas bereken met behulp van ‘n semi-empiriese model. Die lempuntkolk veroorsaak ‘n verhoogde vlak van turbulensie in die vloei, wat turbulensie inname geraas by die lem leirand veroorsaak. Deur isotropiese turbulensie te aanvaar kan die turbulensie inname geraas bereken word met ‘n analitiese model wat as akkuraat bewys is vir lemprofiele. Vir elke waaierlem segment is die relatiewe tangensiale snelheid, turbulente intensiteit, kinetiese energie en dissipasie tempo van die stroomlyn wat die lem leirand bereik, gebruik as insette tot die model. Dit is bewys dat die insette vanaf ‘n skaalmodel aangepas kan word om die turbulensie-inname geraas te voorspel vir dieselfde waaier met ‘n ander deursnee. Eksperimentele metings van die 1.542 m diameter M-waaier geraas is voltooi in ‘n ISO 5801, Tipe A (vrye inlaat, vrye uitlaat) toetsfasiliteit. Die metings is gekorrigeer vir die effek van refleksies op die gemete klankspektrum. Die voorspelde waaiergeraas stem goed oorreen met die gemete spektrum (binne 0.6 dB). Daar word aanbeveel dat die frekwensie van die totale geraas geskaleer word met ‘n faktor proporsioneel tot die verhouding Dwaaier,1/Dwaaier,2 (Dwaaier,1 is die oorspronklike waaier deursnee, Dwaaier,2 is die waaier deursnee waarna die SPL geskaleer word). Vir ‘n verdubbeling in waaier deursnee skaleer die SPL in hierdie geval goed met die verhouding van die waaier diameters tot die mag 1.67.en_ZA
dc.description.versionDoctoralen_ZA
dc.format.extent202 pages : illustrationsen_ZA
dc.identifier.urihttp://hdl.handle.net/10019.1/124929en_ZA
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch Universityen_ZA
dc.rights.holderStellenbosch Universityen_ZA
dc.subjectAeroacousticsen_ZA
dc.subjectAerodynamic noiseen_ZA
dc.subjectCooling systemsen_ZA
dc.subjectAxial flowen_ZA
dc.subjectUCTDen_ZA
dc.titleScaling of axial fan noiseen_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
duplessis_scaling_2022.pdf
Size:
7.45 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Masters Degrees (Mechanical and Mechatronic Engineering)