Virtual sensors for microfluidic systems control – A short review

Here it is presented  a short review on the article “Virtual Sensor Development for Continuous Microfluidic Processes“, written by María García-Camprubí, Cristina Bengoechea-Cuadrado  and Salvador Izquierdo  and published in the journal “IEEE Transactions on Industrial Informatics” (2020) [1]. The experiments performed in this review were realised by Remigijus Vasiliauskas, Marie Curie Postdoctoral fellow at Elveflow.

Abstract – Virtual sensor

In continuous microfluidic processes, non-invasive physical sensing is not always feasible and virtual sensors arise as great tools to support the development of enhanced control systems. This is the case, for instance, of the shape and position tracking of the flow interfaces in a flow-focusing microdevice.

In this work, the authors present a methodology for building an accurate virtual sensor, based on Computer-Aided Engineering (CAE) simulations. To demonstrate the methodology, a case study is presented, where the virtual sensor aims to define the flow pattern within a flow-focusing microfluidic chip.

METHODOLOGY – Virtual sensor

The methodology to build the virtual sensor consists in two major stages:

1. Detailed Computational Fluid Dynamics (CFD) simulations, performed in the open-source platform OpenFOAM [2].
2. Model order reduction (MOR) techniques applied upon the CFD results to generate the real-time response models standing for the virtual sensors. To this end, TWINKLE, a Digital-Twin-Building Kernel for Real-Time Computer-Aided Engineering [3], was employed.

THE MICROFLUIDIC SYSTEM

The case study to demonstrate the methodology is shown in Figure 2.

It consists of:

• A PDMS flow focusing chip
• A pressure-driven flow controller (Elveflow OB1)
• 3x flow sensors (Elveflow MFS2)
• 4x fluid reservoirs
• + required tubing to connect the elements

Figure 2 – Microfluidic system

Key findings – Virtual sensor

Materials & methods

• PDMS flow-focusing chip [4]

• Pressures & flow rates per video clip [4]

*Video 1.82 – pulse change of the pressure, from lower to higher and back (3s).  Pressure profile: _|ˉ|_
Video 1.83 – gradual change of the pressure (2s), from lower to higher and back. Pressure profile: /\/\
Video 1.84 – gradual (just slower than 1.84 (3s)) change of the pressure, from lower to higher and back. Pressure profile: /\/\

Results & discussion

The outcome is a real-time model (virtual sensor) able to predict the shape and location of the multiphase fluid interfaces from the volumetric flow rate measured at the system inlets. Results are in good agreement with in-house experimental data, as it is shown in Figure 3 and in video 1.

Figure 3. Microfluidic chip flow pattern corresponding to different operating conditions: (left column) experimental image, (central column) CFD results, and (right column) ROM prediction (points).

Video 1: Movie from microfluidic chip flow patterns corresponding to different operating conditions of pressures and flow rates controlled by the OB1 Mk3 pressure-driven flow controller. [4] Work performed by Remigijus Vasiliauskas, Marie Curie Postdoctoral fellow at Elveflow.

The methodology presented in this work allows the generation of real-time predictive models with the accuracy of very time-consuming full CFD simulations.

For the particular case under study, the resulting reduced order models can predict the position of fluid interfaces in real time, so they can be used to support Model Predictive Control (MPC) methods for improved microfluidic pressure controllers.

Find out more about virtual sensors…

To learn more about these exciting results, please check the original paper: “Virtual Sensor Development for Continuous Microfluidic Processes“, written by María García-Camprubí  , Cristina Bengoechea-Cuadrado  and Salvador Izquierdo  and published in the journal “IEEE Transactions on Industrial Informatics” (2020).

This work proposes a methodology for building an accurate virtual sensor, based on Computer-Aided Engineering (CAE) simulations. The combination of the fine control of the flows performed by the pressure-driven flow controller OB1 Mk3 Elveflow and the accuracy of the virtual sensor developed in this study ensures optimum performance for your experiment.

References