Data-Driven Virtual Viscosity Meter Augments Inline, Laboratory Measurements

Abstract

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 211393, “Introducing Data-Driven Virtual Viscosity Measurements,” by Volodmir Karpan, Shell, and Samya Al Farsi and Hanaa Al Sulaimani, SPE, Petroleum Development Oman, et al. The paper has not been peer reviewed. _ For polymer-based chemical flooding projects, controlling the viscosity of the injected polymer solution is critical because polymer cost is one of the more significant elements in project economics. The polymer viscosity is measured routinely in the laboratory using fluid samples taken manually at different sampling points. In the case of large-scale projects, however, such viscosity monitoring becomes time-consuming and requires a dedicated field staff. The complete paper introduces a data-driven virtual viscosity meter (VVM) as a tool to augment inline and laboratory viscosity measurements. Introduction In field conditions, proven ways to monitor polymer viscosity of the injected solution include laboratory measurements using injector wellhead samples and inline measurements using a viscometer. In the former, injected chemical solution is sampled regularly to test its viscosity in the laboratory. This is a time- and labor-intensive method with safety and environmental risks during sampling. The latter method is a way to monitor viscosity using inline viscosity-measurement devices. Measuring the viscosity inline reduces the risk of polymer degradation significantly and provides more viscosity data. An inline viscometer uses pressure-drop and flow-rate measurements for the fluid flowing in a bypass line. Pressure drop is measured in the insulated coiled tube, while the flow rate is measured using a Coriolis-type flowmeter. During operation, the polymer solution temperature and pressure drop at a fixed flow rate are measured continuously. The system then calculates the viscosity of the polymer solution passing through the viscometer using the curve extrapolated from the calibration points. Such a viscometer measures a wide range of polymer viscosity and could be used for both high-viscosity and diluted solutions at low and high pressures. Use of inline viscosity measurement, while advantageous in many ways, comes with significant costs. An inline viscometer usually is installed on the foundation in a cabinet to protect it from harsh weather conditions (Fig. 1). Additionally, the injection line upstream of the wellhead must be equipped with a bypass line and drainage system. Maintenance activities could affect the uptime of the inline viscometer. This could become critical under field conditions, where problems with water treatment could translate into plugging of the viscometer filters, leading to erroneous measurements and eventual downtime for cleanup. Such problems grow with the number of devices in operation. Therefore, using a few viscometers is often considered a compromise that comes at the cost of less-effective viscosity monitoring. The authors address a solution to this problem wherein a machine-learning (ML) tool is proposed to augment viscosity measurements by calculating the viscosity using field data for the periods when no viscosity measurements are available.