Home

Design Automation of Fluid Dynamics

DAFD is a design automation tool for microfluidic droplet generators that can suggest design parameters which will meet user-supplied desired generation rates and size.

Folder Structure

The code for DAFD is distributed in several folders based on their logical functions. We go over the various folders here.

• root - The root directory of the repository contains executable files for DAFD. DAFD_CMD.py is meant to be executed by a command line program and outputs results to the console. This program reads inputs from the file cmd_inputs.txt (as described in the README). DAFD_GUI.py produces a GUI application with which to use DAFD.
• bin - The bin folder contains secondary executable files. DAFD_Interface.py is called by DAFD_CMD.py and DAFD_GUI.py in order to delegate the processing tasks. ForwardModelTester.py is a script used for gathering data on the accuracy of our forward models.
• core_logic - The scripts here constitute the main logic of DAFD. RegimeClassifier.py predicts the regime (dripping or jetting) of an inputted chip design. Regressor.py predicts the droplet size and generation rate of an inputted chip design. ForwardModel.py uses both RegimeClassifier.py and Regressor.py to predict regime, droplet size, and generation rate. InterModel.py implements the reverse model in which a design is suggested to produce desired droplet size and generation rate.
• helper_scripts - Here, we have experimental data, the DAFD logo, and ModelHelper.py, a script which loads and normalizes the experimental data for training.
• model_data - This folder contains visualization scripts that we use for examining our models. The data visualized here can be produced from bin/ForwardModelTester.py.
• models - The subfolders forward_models and regime_models contain scripts which train or load the models for use.