Carbon Capture and Storage (CCS) is a promising technology that stops the release of CO₂ from industrial processes such as electrical power generation. Accurate measurement of CO₂ flows in a CCS system where CO₂ flow is a gas, liquid, or gas-liquid two-phase mixture is essential for the fiscal purpose and potential leakage detection. This paper presents a novel method based on Coriolis mass flowmeters in conjunction with least squares support vector machine (LSSVM) models to measure gas-liquid two-phase CO₂ flow under CCS conditions. The method uses a classifier to identify the flow pattern and individual LSSVM models for the metering of CO₂ mass flowrate and prediction of gas volume fraction of CO₂, respectively. Experimental work was undertaken on a multiphase CO₂ flow test facility. Performance comparisons between the general LSSVM and flow pattern based LSSVM models are conducted. Results demonstrate that Coriolis mass flowmeters with the LSSVM model incorporating flow pattern identification algorithms perform significantly better than those using the general LSSVM model. The mass flowrate measurement of gas-liquid CO₂ is found to yield errors less than ±2% on the horizontal pipeline and ±1.5% on the vertical pipeline, respectively, over flowrates from 250 kg/h to 3200 kg/h. The error in the estimation of CO₂ gas volume fraction is within ±10% over the same range of flow rates.

碳捕集与封存（CCS）是一项有前途的技术，可阻止从工业过程（如发电）中释放CO ₂。在CCS系统中，CO ₂流量是气体，液体或气液两相混合物的准确测量CO ₂流量对于财务目的和潜在泄漏检测至关重要。本文提出了一种基于科里奥利质量流量计和最小二乘支持向量机（LSSVM）模型的新方法，用于在CCS条件下测量气液两相CO ₂流量。该方法使用分类器来识别流量模式，并使用单独的LSSVM模型来计量CO ₂质量流量并预测CO的气体体积分数₂，分别。在多相CO ₂流量测试设备上进行了实验工作。进行了一般LSSVM和基于流模式的LSSVM模型之间的性能比较。结果表明，带有流模式识别算法的LSSVM模型的科里奥利质量流量计的性能明显优于使用普通LSSVM模型的科里奥利质量流量计。发现在250 kg / h至3200 kg / h的流量下，气液CO ₂的质量流量测量在水平管道上产生的误差分别小于±2％，在垂直管道上产生的误差分别小于±1.5％。在相同流量范围内，CO ₂气体体积分数的估计误差在±10％以内。