A technique to infer circumferential total pressure distortion intensity found in serpentine inlet airflow was established using wall-pressure fluctuation measurements. This sensing technique was experimentally developed for aircraft with serpentine inlets in a symmetric, level flight condition. The turbulence carried by the secondary flow field that creates the non-uniform total pressure distribution at the compressor fan-face was discovered to be an excellent indicator of the distortion intensity. A basic understanding of the secondary flow field allowed for strategic sensor placement to provide a distortion estimate with a limited number of sensors. The microphone-based distortion estimator was validated through its strong correlation with experimentally determined circumferential total pressure distortion parameter intensities (DPCP). This non-intrusive DPCP estimation technique was then used as a DPCP observer in a distortion feedback control system. Lockheed Martin developed the flow control technique used in this control system, which consisted of jet-type vortex generators that injected secondary flow to counter the natural secondary flow inherent to the serpentine inlet. A proportional-integral-derivative (PID) based control system was designed that achieved a requested 66% reduction in DPCP (from a DPCP of 0.023 down to 0.007) in less than 1 second. This control system was also tested for its ability to maintain a DPCP level of 0.007 during a quick ramp-down and ramp-up engine throttling sequence, which served as a measure of system robustness. The control system allowed only a maximum peak DPCP of 0.009 during the engine ramp-up. The successful demonstrations of this automated distortion control system showed great potential for applying this distortion sensing scheme along with Lockheed Martin's flow control technique to military aircraft with serpentine inlets. A final objective of this research was to broaden the non-intrusive sensing capabilities in the serpentine inlet. It was desired to develop a sensing technique that could identify control efforts that optimized the overall inlet aerodynamic performance with regards to both circumferential distortion intensity DPCP and average pressure recovery PR. This research was conducted with a new serpentine inlet developed by Lockheed Martin having a lower length-to-diameter ratio and two flow control inputs. A cost function based on PR and DPCP was developed to predict the optimal flow control efforts at several Mach numbers. Two wall-mounted microphone signals were developed as non-intrusive inlet performance sensors in response to the two flow control inputs. These two microphone signals then replaced the PR and DPCP metrics in the original cost function, and the new non-intrusive-based cost function yielded extremely similar optimal control efforts. / Ph. D.

Thesis (PhD)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Axial flow fan arrays form part of air-cooled steam condensers in direct drycooled power plants. This dissertation investigates the performance of axial flow fans when located at the perimeter of a fan array. The perimeter (or edge) fans may experience a reduction in air flow through the fan due to the prevalence of distorted inlet conditions upstream of the fan. The reduction in air flow leads to a reduction in the heat transfer capability of the steam condenser and a consequent reduction in the electricity output of the power plant. Due to the physical size of an air-cooled condenser, full-scale experiments are often impractical and computational fluid dynamics (CFD) is used to model its performance under various conditions. To limit the size of the CFD model the axial flow fans in the CFD analysis are represented by means of simplified methods. Three different simplified methods are presented and applied to a CFD model of a single axial flow fan, namely the pressure jump method (PJM), actuator disc method (ADM) and extended actuator disc method (EADM). The results are compared to experimental values. The comparison highlights the limitations of the models: The ADM fails to model fan performance correctly at low flow rates, while the PJM ignores the variation in fan blade properties at different locations within the fan rotor. The EADM is presented as an improvement on both the other two models. A multiple fan test facility is constructed, consisting of three 630 mm diameter fans extracting air from a common inlet chamber. The inlet chamber is constructed in such a way that one of the three fans act as the perimeter (edge) fan. The floor of the inlet chamber can be adjusted to increase or reduce the inlet flow distortion experienced by the edge fan. Six different fan configurations are tested in the position of the edge fan and an empirical method is derived by which the volumetric effectiveness of an edge fan can be predicted. The experimental results are compared to CFD results for the same facility using the three different simplified simulation methods investigated previously. Particle image velocimetry (PIV) measurements are also performed upstream of the edge fan and the velocity profiles at the inlet of the fan are compared to the profiles obtained numerically. The comparisons show that the EADM predicts the performance of the edge fan more accurately than the ADM en PJM. The effect of adding a walkway and removing the bell mouth upstream of the edge fan was investigated using the EADM. The results are used to show the location of the loss mechanisms upstream of the edge fan. The addition of a walkway moves the location of the pressure loss away from the edge fan bell mouth towards the edge of the walkway. Consequently the distortion directly upstream of the edge fan is reduced and its volumetric effectiveness increased. The effect of removing the edge fan’s bell mouth is similar to the effect of adding a walkway upstream of the edge fan. / AFRIKAANSE OPSOMMING: Aksiaalwaaiermatrikse vorm deel van lugverkoelde kondensors in direk droëverkoelde kragstasies. Hierdie verhandeling ondersoek die werkverrigting van aksiaalwaaiers wat geleë is op die rand van ‘n groot waaiermatriks. Die randwaaiers kan ’n vermindering in deurvloei ondervind as gevolg van versteurde inlaattoestande stroom-op vanaf die waaier. Die vermindering in lugvloei lei tot ’n vermindering in die warmetoordagvermoë van die stoomkondensor en ’n gepaardgaande afname in die elektrisiteitslewering van die kragstasie. As gevolg van die fisiese grootte van die lugverkoelde kondensor is volskaalse eksperimente gewoonlik onprakties en word berekeningsvloeimeganika (BVM) gebruik om die werking van die aanleg onder verskeie toestande te modelleer. Ten einde die grootte van die BVM model te beperk, word die aksiaalwaaiers in so ‘n BVM analiese voorgestel met behulp van vereenvoudigde metodes. Drie verskillende vereenvoudigde metodes word aangebied en toegepas op ‘n BVM model van ‘n enkelwaaier, naamlik die druksprongmetode, die aksieskyfmetode en die verlengde aksieskyfmetode. Die resultate word vergelyk met eksperimentele waardes. Die vergelyking benadruk die beperkings van die modelle: Die aksieskyfmetode kan nie die werking van die waaier akkuraat voorspel by lae vloeie nie en die druksprongmetode ignoreer die variasie in lemeienskappe op verskillende liggings binne-in die waaierrotor. Die verlengde aksieskyfmetode word voorgestel as ‘n verbetering op die ander twee metodes. ‘n Veelvuldige waaiertoetsfasiliteit is saamgestel, bestaande uit drie 630 mm deursnee waaiers wat lug uit ’n gemeenskaplike inlaatkamer suig. Die inlaatkamer is so saamgestel dat een van die waaiers in die fasiliteit ‘n randwaaier verteenwoordig. Die vloerhoogte van die inlaatkamer kan aangepas word om die inlaatversteuring wat deur die raandwaaier ondervind word te vermeerder of te verminder. Ses verskillende waaierkonfigurasies is getoets in die randwaaierposisie. ‘n Empiriese metode waarmee die volumetriese effektiwiteit van ‘n randwaaier voorspel kan word is afgelei. Die eksperimentele resultate word vergelyk met ooreenstemmende BVM resultate vir dieselfde fasiliteit deur gebruik te maak van die drie verskillende vereenvoudigde metodes wat vroeër ondersoek is. Partikelbeeld snelheidsmetings word ook stroom-op vanaf die randwaaier uitgevoer en die snelheidsprofiele by die inlaat van die waaier word vergelyk met profiele wat numeries bereken word. Die vergelykings wys dat die verlengde aksieskyfmetode die werkverrigting van ’n aksiaalwaaier meer akkuraat voorspel as die aksieskyf- of druksprongmetodes. Die effek van die installering van ‘n loopvlak en die verwydering van die randwaaier se inlaatmondstuk word ondersoek met behulp van BVM deur gebruik te maak van die verlengde aksieskyfmetode. Die resultate word spesifiek gebruik om die ligging van die verliesmeganismes stroom-op vanaf die randwaaier aan te dui. Die resultate wys dat die installering van ‘n loopvlak die ligging van die drukverlies wegneem vanaf die rand van die waaierinlaat na die rand van die loopvlak. Dit verminder die inlaatversteuring stroom-op vanaf die randwaaier en die volumetriese effektiwiteit word verhoog. Die verwydering van die randwaaier se inlaatmondstuk het ‘n soortgelyke effek as die installering van ‘n loopvlak stroom-op vanaf die randwaaier.

Thesis (PhD (Mechanical Engineering))--University of Stellenbosch, 1990. / Leaves printed on single pages, preliminary pages i- viii and numbered pages 1-1-R5. Includes bibliography, list of tables and figures and list of appendices. Digitized at 330 dpi black and white PDF format (OCR),using KODAK i 1220 PLUS scanner. / ENGLISH ABSTRACT: A numerical procedure for predicting the effect of inlet flow distortions on the performance of axial flow fans is proposed. The study is aimed specifically at fans of low solidity and low hub-to-tip ratios, which have attained some importance with the advent of large directly air-cooled power stations in certain arid regions of the world. The numerical model is an extension to cylindrical polar coordinates of the SIMPLEN algorithm that has been developed by the author previously. The algorithm is implemented in a computer code, FLOVAX, which solves the incompressible Navier-Stokes equations, augmented by the k-Ɛ model of turbulence, on a computational domain that is aligned with the cylindrical polar coordinate system. Several relatively simple flow problems are solved to verify the code: laminar stagnation flow, laminar flow near a rotating disk, turbulent flow near a propeller, turbulent flow through an abrupt axisymmetric expansion, and turbulent swirling flow in an annulus. Good agreement is obtained between the numerical solutions and the corresponding analytical, empirical or published experimental and numerical results. Some experimental results are also presented: measurements of shaft power, volume flow rate and static pressure rise were taken in a setup comprising an axial flow fan mounted in the wall of a wind tunnel. The wind tunnel was used to provide flow across the fan intake, thus establishing distorted inflow conditions. Detailed measurements of the velocity and static pressure distributions in the duct downstream of the fan rotor were also performed. It is clear from the results that flow across the intake of the test fan has a detrimental effect on its performance in that an increased amount of power is necessary to deliver the same flow rate as with no crossflow. In the numerical predictions, blade element theory is used to model the thrust and torque exerted by the fan blades on the air. The numerical results generally confirm the results of the experiment, although the increase in power is underestimated: an increase of only approximately half of the measured increase is predicted. Several recommendations for improvement of the numerical procedure are made. / AFRIKAANSE OPSOMMING: 'n Numeriese prosedure vir die voorspelling van die invloed van versteurings in inlaatvloei op die werking van aksiaalwaaiers word daargestel. Die studie is spesifiek gemik op waaiers van lae soliditeit en lae lempunt-tot-naafverhoudings. Waaiers van hierdie tipe het belangrikheid verwerf sedert die onlangse totstandkoming van groot lugverkoelde kragstasies in sekere droe werelddele. Die numeriese model is 'n uitbreiding na silindriese poolkoordinate van die SIMPLEN-algoritme wat voorheen deur die skrywer ontwikkel is. Die algoritme word geimplementeer in 'n rekenaarkode, FLOVAX, wat die onsamedukbare Navier-Stokes-vergelykings, aangevul deur die k-Ɛ-turbulensiemodel, oplos op 'n berekeningsgebied wat saamval met die silindriese poolkoordinaatstelsel. Verskeie relatief eenvoudige vloeiprobleme word opgelos ter verifikasie van die kode: laminere stagnasievloei, laminere vloei op 'n roterende skyf, turbulente vloei deur 'n propeller, turbulente vloei deur 'n aksiaalsimmetriese vernouing, en turbulente roterende vloei in 'n annulus. Goeie ooreenstemming tussen die numeriese oplossings en die ooreenstemmende analitiese, empiriese of gepubliseerde eksperimentele en numeriese resultate is verkry. Eksperimentele resultate word ook aangebied: metings van asdrywing, volumevloei en statiese drukstyging is geneem in 'n opstelling wat bestaan het uit 'n aksiaalwaaier wat in die wand van 'n windtonnel gemonteer is. Die windtonnel is gebruik om versteurde-invloei-toestande te genereer in die vorm van dwarsvloei oor die waaier-inlaat. Snelheids- en statiese drukverdelings in die kanaal stroom-af van die waaierrotor is ook gemeet. Dit blyk duidelik uit die resultate dat die dwarsvloei oor die inlaat van die toetswaaier 'n nadelige uitwerking het op die werking daarvan in die opsig dat meer drywing nodig is om dieselfde vloeitempo te handhaaf as wat die geval is sonder dwarsvloei. Vir die numeriese voorspellings word van lem-element-teorie gebruik gemaak om die stukrag en draaimoment wat deur die waaier op die lug uitgeoefen word, te modelleer. Die numeriese resultate bevestig oor die algemeen die eksperimentele resultate, alhoewel die drywingstoename onderskat word met sowat die helfte van die gemete toename. Verskeie aanbevelings ter verbetering van die numeriese prosedure word gemaak.

Considerable research is being carried out into the parameters that affect catalyst performance in order to meet the latest emission regulations. The conversion efficiency and the durability of automotive catalytic converters are significantly dependent on catalyst flow performance. Related investigations are commonly conducted using CFD techniques which represent an inexpensive and fast alternative to experimental methods. This thesis focuses on the flow performance of automotive catalytic converters using both experimental and computational techniques. The work describes the effects of inlet flow conditions on catalyst performance, the application of radial vanes to catalyst systems and the refinement of the CFD flow model which increases the accuracy of the predicted catalyst flow performance. the effects of inlet flow conditions on the flow maldistribution across the catalyst face and the total pressure loss through the system were assessed using a steady air flow rig. Tests were conducted over a range of Reynolds numbers typically encountered in automotive catalytic converters using a uniform and a fully-developed inlet flow condition. The results showed that the flow maldistribution significantly increases with Reynolds number notably in wide-angled diffusers. The catalyst flow performance is considerably improved when the inlet flow is uniform rather than fully-developed, the non-dimensional total pressure loss is reduced by 8% at Re=60000 and the flow maldistribution across the catalyst face is decreased by 12.5% and 15% respective Reynolds numbers of 30000 and 60000 when using a 60 degree diffuser. The total pressure loss through the system was found to be mostly associated with the monolith brick resistance. When the flow maldistribution is approximately 2, the pressure loss across the monolith brick represents 80% of the system pressure loss. The flow maldistribution across the catalyst face was improved by locating a system of radial splitters in the diffuser. The optimum flow performance was found to be a complex function of the vane design. A maximum improvement in the flow maldistrution indices M and Mi of 25% and 50% respectively was achieved at the expense of an increase in total pressure loss of 13.5% at Re = 60000. Both CFD and flow visualisation techniques were used as an aid to interpreting the flow field in the diffuser. Although a qualitative agreement was obtained using CFD, the flow maldistribution across the catalyst face was underpredected by up to 20%. The accuracy of the flow predictions was significantly improved by investigating the flow field in the monolith channels. Flow recirculation occurs in the channel entry length when the flow approaches the monolith channels at an angle which induces an additional implemented into four models of the flow through axisymmetric catalyst assemblies using various diffuser geometries and inlet flow conditions. By including the flow entrance effects in the porous media approach, the flow maldistribution was predicted within 8% instead of 15% when these effects are neglected. Further investigation of the flow in the monolith channels will be required to accurately model three-dimentional flows (racetrack catalysts) and to include various channel geometries and system flow rates.

Thesis (MEng) -- Stellenbosch University, 2014. / ENGLISH ABSTRACT: The use of air-cooled steam condensers (ACSCs) is the preferred cooling method in the chemical and power industry due to stringent environmental and water use regulations. The performance of ACSCs is however highly dependent on the influence of windy conditions. Research has shown that the presence of wind reduces the performance of ACSCs. It has been found that cross-winds (wind perpendicular to the longest side of the ACSC) cause distorted inlet flow conditions, particularly at the upstream peripheral fans near the symmetry plane of the ACSC. These fans are subjected to what is referred to as '2-D' wind conditions, which are characterised by flow separation on the upstream edge of the fan inlets. Experimental investigations into inlet flow distortion have simulated these conditions by varying the fan platform height. Low platform heights resulted in higher levels of inlet flow distortion, as also found to exist with high cross-wind speeds. This investigation determines the performance of various fan configurations (representative of configurations used in the South- African power industry) subjected to distorted inlet flow conditions through experimental and numerical investigations. The similarity between platform height and cross-wind effects is also investigated and a correlation between system volumetric effectiveness, platform height and cross-wind velocity is found. / AFRIKAANSE OPSOMMING: Die gebruik van lugverkoelde stoom kondensors (LVSK's) word verkies as 'n verkoelingsmetode in die chemiese- en kragvoorsieningsindustrie as gevolg van streng omgewings- en waterverbruiksregulasies. Die werkverrigting van LVSK's word egter grootliks beïnvloed deur die teenwoordigheid van wind. Navorsing het gewys dat die teenwoordigheid van wind die werkverrigting van LVSK's verminder. Daar was gevind dat kruiswinde (wind loodreg tot die langste sy van die LVSK) versteurde inlaat vloeitoestande veroorsaak, veral by waaiers wat aan die stroomop kant van die LVSK naby die simmetrievlak geleë is. Hierdie waaiers word blootgestel aan na wat verwys word as '2-D' windtoestande wat gekenmerk word deur vloeiwegbreking wat plaasvind by die stroomop rand van die waaierinlate. Eksperimentele ondersoeke van inlaat vloeiversteurings het hierdie toestande gesimuleer deur die waaier platformhoogte te verstel. Lae platform hoogtes het gelei tot hoër vlakke van inlaat vloeiversteuring, soortgelyk aan wat gevind word met hoë kruiswindsnelhede. Hierdie ondersoek gebruik numeriese en eksperimentele metodes om die werkverrigting van verskeie waaierkon gurasies (verteenwoordigend van kon- gurasies wat gebruik word in die Suid-Afrikaanse kragvoorsieningsindustrie) wat blootgestel word aan versteurde inlaat vloeitoestande te bepaal. Die ooreenkoms tussen platformhoogte en kruiswind e ekte word ook ondersoek en 'n korrelasie tussen die sisteem volumetriese e ektiwiteit, platformhoogte en kruiswindsnelheid word bepaal.