The performance of multiphase flow processes is often determined by the distribution of phases inside the equipment. However, controllers in the field are typically implemented based on flow variables, which are simpler to measure, but indirectly connected to performance (e.g., pressure). Tomography has been used in the study of the distribution of phases of multiphase flows for decades, but only recently, the temporal resolution of the technique was sufficient for real-time reconstructions of the flow. Due to the strong connection between the performance and distribution of phases, it is expected that the introduction of tomography to the real-time control of multiphase flows will lead to substantial improvements in the system performance in relation to the current controllers in the field. This paper uses a gas–liquid inline swirl separator to analyze the possibilities and limitations of tomography-based real-time control of multiphase flow processes. Experiments were performed in the separator using a wire-mesh sensor (WMS) and a high-speed camera to show that multiphase flows have two components in their dynamics: one intrinsic to its nonlinear physics, occurring independent of external process disturbances, and one due to process disturbances (e.g., changes in the flow rates of the installation). Moreover, it is shown that the intrinsic dynamics propagate from upstream to inside the separator and can be used in predictive and feedforward control strategies. In addition to the WMS experiments, a proportional–integral feedback controller based on electrical resistance tomography (ERT) was implemented in the separator, with successful results in relation to the control of the distribution of phases and impact on the performance of the process: the capture of gas was increased from 76% to 93% of the total gas with the tomography-based controller. The results obtained with the inline swirl separator are extended in the perspective of the tomography-based control of quasi-1D multiphase flows.

Electrical resistance tomography (ERT) has been used in the literature to monitor the gas–liquid separation. However, the image reconstruction algorithms used in the studies take a considerable amount of time to generate the tomograms, which is far above the time scales of the flow inside the inline separator and, as a consequence, the technique is not fast enough to capture all the relevant dynamics of the process, vital for control applications. This article proposes a new strategy based on the physics behind the measurement and simple logics to monitor the separation with a high temporal resolution by minimizing both the amount of data and the calculations required to reconstruct one frame of the flow. To demonstrate its potential, the electronics of an ERT system are used together with a high-speed camera to measure the flow inside an inline swirl separator. For the 16-electrode system used in this study, only 12 measurements are required to reconstruct the whole flow distribution with the proposed algorithm, 10× less than the minimum number of measurements of ERT (120). In terms of computational effort, the technique was shown to be 1000× faster than solving the inverse problem non-iteratively via the Gauss–Newton approach, one of the computationally cheapest techniques available. Therefore, this novel algorithm has the potential to achieve measurement speeds in the order of 10⁴ times the ERT speed in the context of inline swirl separation, pointing to flow measurements at around 10kHz while keeping the average estimation error below 6 mm in the worst-case scenario.

Swirl-based separators are widely used in the separation of two-phase flows with phases of different densities, e.g. in the oil and gas industry. The separation is based on centrifugal forces associated with the swirl motion, that splits the original mixture into a light-phase core and a heavy-phase annulus, captured by two separate outlets equipped with control valves. Due to the complex (unsteady) dynamics of the flow and variations in the mixture that reaches the separator, real-time controllers are required to keep the separation optimal. Traditionally, the control is based on pressure measurements at the boundaries of the separator. However, pressure is only indirectly related to the separation. A more direct (and potentially more effective) control variable is the distribution of phases inside the separator, which can be monitored using soft-field tomography. This idea is explored in this work using an air-water inline-swirl separator facility equipped with an Electrical Resistance Tomography (ERT) sensor, that determines the size of the air core in real-time using a newly developed fast ERT reconstruction algorithm. The dynamics of the system was studied for different valve actions and summarized into a Transfer Function of the process, which is used to design a PI controller. The approach was successful and the control strategy implemented kept the air core near the setpoint in the presence of disturbances in the air flow rate reaching the separator.

Today's mechanical fluid separators in industry are mostly operated without any control to maintain efficient separation for varying inlet conditions. Controlling inline fluid separators, on the other hand, is challenging since the process is very fast and measurements in the multiphase stream are difficult as conventional sensors typically fail here. With recent improvement of process tomography sensors and increased processing power of smart computers, such sensors can now be potentially used in inline fluid separation. Concepts for tomography-controlled inline fluid separation were developed, comprising electrical tomography and wire-mesh sensors, fast and massive data processing and appropriate process control strategy. Solutions and ideas presented in this paper base on process models derived from theoretical investigation, numerical simulations and analysis of experimental data.

This text structures the application of Wire-Mesh sensors and Electrical Resistance Tomography in the control of an Inline Swirl Separator. It introduces a mechanistic model of the two-phase flow inside the device, which is linearized around an ideal perfect operation, and implemented in a Model Predictive Controller. The whole text is structured aiming at a future real application of the controller, briefly introducing the setup that is going to be used, the sensors and their working principles. The results obtained show a stable controller, able to regulate the process relatively fast in relation to the time resolution of the sensors. The positive response of the approach stimulates further improvements in the model developed, and the implementation of more sophisticated techniques to handle the non-linearities of the process.

Phase separation based centrifugal forces is effective, and thus widely explored by the process industry. In an inline swirl separator, a core of the light phase is formed in the center of the device and captured further downstream. Given the inlet conditions, this gas core created varies in shape and size. To predict the separation behavior and control the process in an optimal way, the gas core diameter should be measured with the minimum possible intrusiveness. Process tomography techniques such as electrical resistance tomography (ERT) allows us to measure the gas core diameter in a fast and non-intrusive way. Due to the soft-field nature and ill-posed problem in solving the inverse problem, especially in the area of low spatial resolution, the reconstructed images often overestimate the diameter of the object under consideration leading to unreliable measurements. To use ERT measurements as an input for the controller, the estimated diameters should be corrected based on secondary measurements, e.g., optical techniques such as high-speed cameras. In this context, image processing and image analysis techniques were adapted to compare the diameter calculated by an ERT system and a fast camera. In this paper, a correction method is introduced to correct the diameter obtained by ERT based on static measurements. The proposed method reduced the ERT error of dynamic measurements of the gas core size from over 300% to below 20%, making it a reliable sensing technique for controlled separation processes.