Evaluation and optimization of PolyCor - a single-use Coriolis flowmeter

Sammanfattning: In the pharmaceutical industry it has become common to use single-use components in the production line to save time and money. Therefore, a team at General Electric started to develop a single-use Coriolis flowmeter, PolyCor M13, targeted at liquid chromatography systems. A Coriolis flowmeter in this embodiment is simply a tube put into vibration at its natural frequency. When there is a flow through the vibrating tube the Coriolis force arises, causing a phase shift of the pickup signals on each side of the actuator. This phase shift is linearly proportional to the mass flow through the tube.  The approach of PolyCor M13 is a separate oscillator and flow path. The oscillator is a metal skeleton holding the flow path, a silicone tube. The initial requirement flow range was 0.16-7.8 kg/min with an error less than 2%. Initial tests indicated that the prototype fulfilled these requirements at ambient temperature. This thesis is a further evaluation of the performance of M13. The main goal was to establish the pressure and temperature dependence and find an model to compensate for these. Investigation if M13 can manage flows up to 10 kg/min was also a part of this scope.  Control measurements showed that flows less than 2 kg/min could give large errors, over 5%. Flows higher than 2 kg/min up to 10 kg/min had errors less than 5%. The error was determined using a reference mass flow. By increasing the pressure in the system, from 0 to 4 bar, the error increased substantially. For the two lowest flows tested, 0.5 and 1 kg/min, the computed mass flows gave errors over 100%. The error for 2 kg/min was around 70% and the error successively decreased for higher flows and at 6-10 kg/min the error was around 20%. A compensation model was established by fitting a polynomial to the data. The best compensation model managed to reduce the error on new test data to 5-10% for flows between 2-10 kg/min. For smaller flows the error was still high but less than 100%.  Significant deviation from the temperature at which the proportionality was determined caused large errors. Errors for low flows, 0.5 and 1 kg/min, gave errors mainly up to around 50% but some errors were over 100%. For higher flows, 2-10 kg/min the error is up to 30% with some occasional errors up to 60%. The procedure to establish a compensation model for temperature was similar to that for the pressure compensation model. The best model for temperature compensation managed to reduce the error to 5-10% for flows between 2-10 kg/min. For lower flows the error was still high but slightly better, some error was still over 100%.  In conclusion, M13 is not as reliable as initial tests showed. The error limit is exceeded, especially for flows less than 2 kg/min. A more suitable range would be 2-10 kg/min. The pressure and temperature effects have an enormous impact but can be compensated to some extent. Since the accuracy of M13 is not perfect, errors less than 5% can be difficult to obtain in the current state of M13.