Microfluid Mechanics/Modeling of Microflows

Introduction
The amount of available knowledge on a system and the complexity of the interacting physical effects are decisive role in the selection and development of a model of this system.



The multi-physical nature of the micro flows inflates the number of variables which have to be solved and/or controlled. At some stage, classical data analysis tools or models based on conservation laws can not be feasibly used to understand and/or monitor the process. Analysis tools should be selected depending on our knowledge of a system. If the system is well known, it can be modeled by conservation laws. As our knowledge of the system decreases we should select different approaches and at the low end of our knowledge only classification techniques might help. As regards to a process, there can be many sub-systems which can be modeled and analyzed by a technique appropriate for this system. Therefore, for a complex process having many subsystems, using hybrid models becomes not only feasible but also more accurate. In such a hybrid model, one can train artificial neural networks (ANN) via accurate simulations based on conservation laws and the experiments and integrate knowledge and experience based information to the hybrid by using fuzzy systems. If the hybrid is capable of reproducing the known phenomena and tendencies, this hybrid model can later be used to understand and optimize the process itself. Moreover, it can be further trained during the operation of the process.

As most of the micro flows comprise interacting multiple physical effects, mostly no analytical solution exists for them. Therefore, those flows are modeled mathematically in a form which can be numerically computed. As the knowledge on those flows are relatively high, models are based on mainly atomistic and continuum approaches both for liquid and gas flows.

Some of these methods are briefly explained in the ongoing text.