Abstract
Ph.D.
The use of cutting edge computer simulation tools such as Computational Fluid Dynamics (CFD) enables engineers to design, analyse and predict how effective and efficient a new design concept will be long before the plant or system is actually built. Although CFD software codes have progressed in vast leaps and bounds over the past ten to fifteen years, their application can still be found to be limited when complex systems have to be simulated. In such cases it is often possible for engineers to adapt or modify the software codes in order to cater for a specific need.
This research is based on such a problem where the mere application of a standard CFD code was not sufficient to simulate the complexities and interactions which are found when analysing the performance and effectiveness of an Air Cooled Condenser (ACC). In the case of a mechanical draught ACC the multitude of fans which are employed to feed the cooling system with sufficient cooling air, require special treatment when attempting to simulate them using a standard CFD code. Most codes cater for detailed simulation of a rotating fan, however such simulation techniques require numerical meshes in excess of 40 000 computational cells per fan.
In the case of an ACC such as the one at Matimba power station in South Africa, 288 fans would have to be simulated which would require a numerical mesh in excess of 11 million
computational cells. Although this size of numerical problem could possibly be solved on some of the worlds fastest computers, it would not provide a practical engineering solution to the problem.
The ACC at Matimba power station is the largest of its kind in the world. However, poor
availability resulting from high sensitivity to changing ambient conditions such as high wind
speeds, high ambient temperatures and changing wind directions prompted an urgent need for detailed simulation of the entire ACC system. As will be shown in the literature study, some attempts have been made to simulate air cooled condensers or parts thereof, however two main factors constantly limited the accuracy and usability of CFD codes for this application, (i) the interactive simulation of the fans with prevailing ambient conditions and (ii) the interaction between the performance of the ACC and the response of the turbine, which is thermodynamically coupled to the ACC via the steam pressure and temperature in the steam ducts.
The abovementioned factors have in most cases restricted simulations to steady-state solutions and have also required tremendous computational and human effort resulting from the complexities surrounding the treatment of the high number of fans which need to be simulated. This research study submits a unique new integrated approach to simulation of a complex system such as an ACC, including its multitude of fans together with the complex interaction between the entire ACC system and the changing ambient conditions. A fan simulation sub-model was developed and tested and good agreement was achieved with the fan design data. Further sub models were developed in order to simulate the interaction between the ACC and the turbine generator. A test case was simulated and final results were compared to on site measurements achieving good agreement with physical test data and unit operating data.