15th European Conference on Turbomachinery Fluid dynamics & Thermodynamics
Authors
Abstract
The development of pumps for applications in wastewater systems is a complex and extensive task. In this setting, a pump needs to convey a multitude of different contaminants while maintaining functionality and, ideally, efficiency. Studies have shown that the main culprit for clogging in wastewater pumps are wipes made from long, fibrous materials. Therefore, a major insight into a pumps suitability for wastewater application is its susceptibility to clogging with these materials. At Technische Universität Berlin, Chair of Fluid System Dynamics, we operate a test stand dedicated to analyzing the behavior of pumps under artificial wastewater conditions and evaluate their susceptibility to clogging. The artificial wastewater is composed of varying degrees of fibrous wipes in water, called contamination classes. During multiple long-time functional performance tests with varying degrees of contamination and variations in the operation point, some 50 different pump setups have been tested. For each of these tests, the main output is the recording of the main operational parameters, i.e., electrical power, head, flow, and efficiency. Further, the weight of the contaminants being introduced to the test medium as well as the weight of contaminants being stuck in the pump at the end of the test are recorded. In this paper, these extensive tests are leveraged to draw conclusions regarding the behavior of different pump setups in wastewater operation. While previous studies have proven already that, e.g., ball passage and impeller type do not unequivocably influence a pumps suitability for wastewater application, this extensive evaluation goes beyond anecdotal evidence to show there actually are some common trends to be observed. For this investigation, correlations of the different meta-parameters, e.g., impeller diameter, operation point or ball passage, as well as time-resolved peculiarities are calculated via Spearman rank-order to showcase commonalities. In particular, four different characteristic patterns are leveraged to break down time-resolved behavior, namely, periods of high fluctuation, positive and negative monotonicity, as well as jumps in the signals. These patterns are quantified and subsequently correlated with the different meta-parameters of the pump setups. In an additional analysis, the different signal traces of tests are computationally compared with the technique of dynamic time warping. This can be leveraged to analyze (dis-)similarities in the full-length records of signal traces. The preliminary results show that at least over all test runs, including the multiple operation points and contamination classes, there are no explicit clusters to be observed when investigating the entire length of the signal traces. This supports the assumption that all pumps, be it with similar dimensions or impeller types, have individual characteristic behavior in wastewater applications, signifying the need for further disaggregation of time-resolved patterns. Therefore, in the full version of this paper, there will be a structured and elaborate analysis of the aforementioned characteristic patterns with quantifications of their correlations to several meta-parameters. The assumption to investigate is, that even if pumps behave very differently over the entire hour of testing, there are common time-related patterns, which do correlate with the meta-parameters of the pumps to some degree.
ETC2023-237
Paper is available for download