Hybrid FPGA-CPU-GPU system for real-time multiphase flow monitoring using wire-mesh sensors
Benjamin Sahovic, Michael Quinten · Results in Engineering · 2025
Reliable, real-time monitoring of multiphase flows is essential for continuous industrial processes such as oil-water separation, chemical reactor control, and flow diagnostics. This paper presents a hybrid FPGA-CPU-GPU system for processing Wire-Mesh Sensor (WMS) data in real time, enabling continuous, high-resolution estimation of void fraction and gas-liquid interface velocity. The system combines high-speed signal acquisition with adaptive calibration and multiple velocity estimation methods, including histogram-based approaches and cross-correlation techniques. Calibration adjusts automatically to changing process conditions such as temperature or conductivity - without interrupting operation. The velocity estimation methods are selected based on flow structure, balancing accuracy and computational cost. Demonstrated in an inline fluid separation scenario and validated through high-speed optical tracking, the system processes tens of thousands of frames per second with minimal latency. Its modular structure makes it suitable for integration into industrial control systems where responsiveness and robustness are critical. By enabling fast and continuous interpretation of WMS data, this architecture supports real-time process control and monitoring in demanding industrial environments.