An axial flow fan design methodology is developed to design large diameter, low pressure rise, rotor-only fans for large air-cooled heat exchangers. The procedure aims to design highly efficient axial flow fans that perform well when subjected to off design conditions commonly encountered in air-cooled heat exchangers. The procedure makes use of several optimisation steps in order to achieve this. These steps include optimising the hub-tip ratio, vortex distribution, blading and aerofoil camber distributions in order to attain maximum total-to-static efficiency at the design point. In order to validate the design procedure a 24 ft, 8 bladed axial flow fan is designed to the specifications required for an air-cooled heat exchanger for a concentrated solar power (CSP) plant. The designed fan is numerically evaluated using both a modified version of the actuator disk model and a three dimensional periodic fan blade model. The results of these CFD simulations are used to evaluate the design procedure by comparing the fan performance characteristic data to the design specification and values calculated by the design code. The flow field directly down stream of the fan is also analysed in order to evaluate how closely the numerically predicted flow field matches the designed flow field, as well as determine whether the assumptions made in the design procedure are reasonable. The fan is found to meet the required pressure rise, however the fan total-to-static efficiency is found to be lower than estimated during the design process. The actuator disk model is found to under estimate the power consumption of the fan, however the actuator disk model does provide a reasonable estimate of the exit flow conditions as well as the total-to-static pressure characteristic of the fan.

An axial flow fan developed in the previous study is tested in order to characterize its performance. The M-fan, a 7.3152 m diameter rotor only axial flow fan was designed to perform well under the challenging operating conditions encountered in air-cooled heat exchangers. Preliminary computational fluid dynamics (CFD) results obtained using an actuator disk model (ADM) as well as a periodic three dimensional model indicate that the fan meets the specified performance targets, with an expected total-to-static efficiency of 59.4% and a total-to-static pressure rise of 114.7 Pa at the operating point. Experimental tests are performed on the M-fan in order to determine its performance across a full range of flow rates. A range of fan configurations are tested in order to ascertain the effect of tip clearance, blade angle, and hub configuration on fan performance. Due to the lack of a suitable facility for testing a large diameter fan, a scaled 1.542 m diameter model is tested on the ISO 5801 type A fan test facility at Stellenbosch University. A Reynolds-averaged Navier–Stokes CFD model representing the M-fan in the test facility is also developed in order to provide additional insight into the flow field in the vicinity of the fan blades. The results of the CFD model will be validated using the experimental data obtained. Both the CFD results and the experimental data obtained are compared to the initial CFD results for the full scale fan, as obtained in the previous study, by means of fan scaling laws. Experimental data indicate that the M-fan does not meet the pressure requirement set out in the initial study at the design blade setting angle of 34 deg. Under these conditions, the M-fan attains a total-to-static pressure rise of 102.5 Pa and a total-to-static efficiency of 56.4%, running with a tip gap of 2 mm. Increasing the blade angle is shown to be a potential remedy, improving the total-to-static pressure rise and efficiency obtained at the operating point. The M-fan is also shown to be highly sensitive to increasing tip gap, with larger tip gaps substantially reducing fan performance. The losses due to tip gap are also shown to be overestimated by the CFD simulations. Both experimental and numerically obtained results indicate lower fan total-to-static efficiencies than obtained in the initial CFD study. Results indicate that the M-fan is suited to its intended application, however, it should be operated with a smaller tip gap than initially recommended and a larger blade setting angle. Hub configuration is also shown to have an influence on fan performance, potentially improving performance at low flow rates.

In this paper the procedure and results of the aeromechanical optimization of a mixed-flow compressor impeller to be used in a 600 N micro gas turbine (MGT) are presented. Today’s unmanned aerial vehicles (UAVs) demand high thrust-to-weight ratios and low engine frontal area. This may be achieved using mixed-flow compressors. The initial impeller design was obtained using a 1-D turbomachinery layout tool. A multi-point optimization of the impeller aerodynamic performance was completed, followed by a mechanical optimization to reduce mechanical stresses in the impeller. A coupled aero-mechanical optimization was concluded with the purpose of increasing the choke limit and reducing stresses while conserving aero-performance. Subsequently, a modal analysis of the rotor was conducted to determine its vibrational characteristics. The optimization process was set up and controlled in an integrated environment that includes a 3-D Navier-Stokes flow solver and a 3-D finite element (FE) structural solver. The optimization process is based on the use of a database, an artificial neural network (ANN), a user-defined objective function and a genetic algorithm (GA). The overall optimization process achieved an increase in pressure ratio (total-to-total) of 30.6% compared to the initial design while the efficiency (isentropic total-to-total) was increased by 5% at design conditions. A decrease in the surge margin was experienced, but the final surge margin was still acceptable (12%). The choke limit was increased meaningfully. This was achieved while also decreasing the peak von Mises stress from far above the material yield strength to 30% below the yield limit.

Large axial flow fans are used in forced draft air cooled heat exchangers (ACHEs). Previous studies have shown that adverse operating conditions cause certain sectors of the fan, or the fan as a whole to operate at very low flow rates, thereby reducing the cooling effectiveness of the ACHE. The present study is directed towards the experimental and numerical analyses of the flow in the vicinity of an axial flow fan during low flow rates. This is done to obtain the global flow structure up and downstream of the fan. A near-free-vortex fan, designed for specific application in ACHEs, is used for the investigation. Experimental fan testing was conducted in a British Standard 848, type A fan test facility, to obtain the fan characteristic. Both steady-state and time-dependent numerical simulations were performed, depending on the operating condition of the fan, using the Realizable k-ε turbulence model. Good agreement is found between the numerically and experimentally obtained fan characteristic data. Using data from the numerical simulations, the time and circumferentially averaged flow field is presented. At the design flow rate the downstream fan jet mainly moves in the axial and tangential direction, as expected for a free-vortex design criteria, with a small amount of radial flow that can be observed. As the flow rate through the fan is decreased, it is evident that the down-stream fan jet gradually shifts more diagonally outwards, and the region where reverse flow occur between the fan jet and the fan rotational axis increases. At very low flow rates the flow close to the tip reverses through the fan, producing a small recirculation zone as well as swirl at certain locations upstream of the fan.

The design and validation of a Compact Crossover Diffuser (CCD) to replace the size-limited radial diffuser and axial de-swirl cascade of an existing Micro Gas Turbine (MGT) is discussed. A CCD strives to combine the performance of a channel diffuser with the operating range and efficiency of a vaneless diffuser. The development of a one-dimensional Mean-Line Code (MLC) is presented, which aids the designer in preliminary design and performance evaluation of the CCD. Design graphs indicating the performance effects of changing the primary design variables are developed and shown. The MLC is numerically validated using Computational Fluid Dynamics (CFD). Good agreement is seen between the MLC and CFD results, predicting the design point PRss(2-4) to within 1.4 %. A CFD optimized CCD was manufactured and tested. Agreement between the CFD and experimental results for PRts(0-4) is within 7.58 % at 106 kRPM. A numerically predicted increase in PRts(0-4) from 3.31, to 3.53, to 3.83 is seen for the vaneless-, MLC optimized-, and CFD optimized-design respectively. An experimental increase of 82.3 % in engine thrust and 80.0 % in total-to-static pressure recovery across the compressor stage was measured when retrofitting the BMT120KS with a new impeller and CCD.