Flow investigation in a centrifugal pump

• Main Author: Ang, Aaron Guo Hao
• Other Authors: Chan Weng Kong
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Mechanical engineering
• Online Access: http://hdl.handle.net/10356/64479
• Institution: Nanyang Technological University

Centrifugal blood pumps are becoming the popular choice for VADs because of their simple design and low costs. Nevertheless, the presence of flow stagnation or excessive leakage flow would induce thrombosis and hemolysis. Since these risks are flow related, ANSYS Workbench v15.0 was employed for the CFD simulations to investigate flow characteristics in the pump. CFD simulations of the pump were performed for four gap widths (0.2mm, 0.3mm, 0.4mm and 0.5mm) at operating flow rates of two to six liters per minute for impeller speed of 2000rpm. It was found that pump head, P increases as operating flow rate, L and gap width decrease. Additionally, total leakage flow rate, QL increases along with differential pressure in the gap, ΔP and the gap width but decreases as the operating flow rate, Q escalates. These trends are congruent with available experimental results. The only difference is that the experiments obtained QL values from experimental trials of a pump with a dummy. Generally, the maximum QL discrepancy between the CFD simulations and experiments does not exceed 15.7% except for the 0.5mm gap pump, making it difficult to ascertain the correlation of the CFD and experimental results. Consequently, CFD simulations of the pump with a dummy for the similar four gap widths at dummy speed of 2000rpm were carried out. The maximum difference in QLvalues between simulations of the pump and the pump with a dummy is 3.4% for the 0.2mm and 0.3mm gap widths and 15.9% for the 0.4mm and 0.5mm gap widths. Thus, using the pump with a dummy to predict QL values is relatively reliable. Comparing the pump with a dummy between the CFD simulations and the experiments, CFD simulations consistently under predict the ΔP values for a given QL due to its inability to accurately predict the losses. However, it over predicted the ΔP values for the 0.5mm gap pump with a dummy. This abnormally is related back to and accounts for the disparities in QL values between CFD simulations and the experiments, justifying the suspicion that there are systematic errors when the experiment for the 0.5mm gap pump with a dummy was conducted. Concluding, the CFD software predicts the leakage flow rates quite accurately.