National Technical University of Athens (NTUA)
Development of a low-cost device and software for measuring surface wetting properties
Abstract
dc:descriptionWetting phenomena are fundamental and ubiquitous in nature and technology. Especially, in industrial manufacturing and processing, controlling and monitoring wetting properties such as interfacial tension or contact angle of a liquid drop/film on a solid is not only a matter of scientific interest — it is a fundamental requirement for product quality, process efficiency, and operational safety. In most cases, surface tension and contact angle are measured with optical methods. These methods include instrumentation for capturing high resolution back-lighted images of sessile or pendant drops and software that implements the ADSA (Axisymmetric Drop Shape Analysis) method. However, most commercial devices are expensive, bulky and designed to work robustly only under experimental, controlled conditions. Thus, an in situ and real-time quality monitoring of an industrial or real-time process is rarely performed with these devices. Micro devices based on Smartphones, microcontrollers or single board computers can successfully reduce the cost, complexity and size of commercial goniometers/tensiometers. However, finding alternatives only to the traditional instrumentation is not sufficient for dealing with the challenges of real-time measurements in industrial environments. Standard ADSA method is prone to errors or even failure caused by low quality images, non-ideal illumination or dynamic conditions. This implies the need to also develop alternative methods for measuring wetting properties from images of pendant drops or sessile droplets. In this Thesis we develop a fully integrated low-cost, robust system based on Smartphones, for optical measurement of wetting properties in real-time, industrial, engineering and academic applications. The frame of the device is based only on 3D printed parts and easily available, low-cost metal parts. To deal with the challenges associated with measurement conditions, we develop DropUI, an open-source software that implements novel machine learning methods and standard ADSA. Deep learning models are trained with both experimental and numerical generated data and demonstrate great generalization even when are used to measure wetting properties in very diverse conditions. Alongside the low-cost system, we also develop computational models for simulating complex dynamic wetting phenomena with the intention of better understanding the impact of wettability on macroscopic properties and generate numerical data that can be used for training DL models. The utilization of a novel computational approach which treats the entire droplet surface (both the liquid/vapor and liquid/solid interfaces) in a unified context enables realistic simulation of complex dynamic wetting phenomena from droplet evaporation on heterogeneous surfaces to droplet detachment from micro-structured surfaces. First, a continuum level, sharp-interface, ALE (Arbitrary Lagrangian- Eulerian) numerical model is developed that can efficiently deal with the problem of the moving three-phase contact line (TPL) during evaporation. It is found that, depending on the surface roughness and the material wettability, the evaporation dynamics proceeds following the Constant Contact Angle (CCA), Constant Contact Radius (CCR) or a mixed mode (stick-slip behavior). Numerical predictions concerning droplets evaporating on rough, hydrophobic substrates, are experimentally validated showing exceptional agreement. The stick-slip motion is found to appear during evaporation and is particularly affected by local roughness features of the surface. Using the same modeling approach, we performed simulations of droplet detachment and jumping dynamics over flat and micro-structured surfaces. Preliminary calculations reveal a considerable connection between substrate micro-topography and energy efficiency of the process. The latter results could be extended to the optimal design of micro-structured solid surfaces for electrowetting-induced droplet removal in ambient conditions. The next section is dedicated to Deep Learning Assisted Measurement Of Wetting Properties. Pre-trained Convolutional Neural Networks (CNNs) were fine-tuned on a dataset of 91,000 augmented pendant drop images acquired with a commercial camera. Testing was conducted on images from a commercial tensiometer under varied conditions, including suboptimal lighting and optical aberrations. Smartphone-acquired images and independent user-submitted data from DropUI were also evaluated to assess robustness and generalization. The models estimated surface tension with less than 13% error even when ADSA failed due to lighting conditions or optical aberrations. Inference on a GPU was at least an order of magnitude faster than ADSA running on a CPU. Extending the above idea to contact angle measurement, we presented a novel data augmentation approach implementing CycleGANs (Cycle-Consistent Generative Adversarial Networks) to learn the style of experimental sessile drop images and map it to binary images, that are produced either by i) the original experimental images or, ii) by numerical models. Pre-trained CNN models for contact angle regression, fine-tuned with hybrid data, achieved significant generalization performance over a wide contact angle range and for sessile drop images acquired with different cameras, outperforming other results listed in the literature. With the computational framework already available, we present the implementation of DL models and standard ADSA in our Open-Source web application software. Next, we designed and manufactured the device frame. This was implemented using 3D-printed parts interconnected with low-cost, easily accessible mechanical components. The overall modular design shows several important advantages such as portability, stiffness, customizability and easy assembly. In the end of this Chapter, detailed instructions on how to assembly the device and setup the experiments are provided. In the final Chapter, we independently validated our fully-developed system against a commercial optical tensiometer/goniometer and an open-source ADSA package, for both surface tension and contact angle measurement. The comparison of the results confirmed the ability to measure wetting properties from drop images acquired with cheap smartphone cameras under non-optimal lighting conditions, thanks to the integration of both standard ADSA and DL methods.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Sourais, Alexandros-Georgios
- Σουραής, Αλέξανδρος-Γεώργιος
Subjects
dc:subject × 21- Τενσιομετρία Κρεμάμενης Σταγόνας
- Γωνιομετρία Επικαθήμενης Σταγόνας
- Αξονοσυμμετρική Ανάλυση Σχήματος Σταγόνας
- Μοντελοποίηση Δυναμικών Φαινομένων Διαβροχής
- Βαθιά μάθηση
- Τριδιάστατη Εκτύπωση
- Έξυπνο Κινητό Τηλέφωνο
- Διεπαφή χρήστη
- Pendant Drop Tensiometry
- Sessile Drop Goniometry
- ADSA
- Modeling of Dynamic Wetting Phenomena
- Deep learning
- 3d printing
- Smartphone
- User interface
- Επιστήμη Χημικού Μηχανικού
- Επιστήμες Μηχανικού και Τεχνολογία
- Χημική μηχανική
- Chemical Engineering
- Engineering and Technology
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/62788