OpenFOAM FOR DRONES: THE INTERNS OF PAANDUV

Have you ever looked up into the sky and seen an aircraft flying, or if you are lucky, possibly a drone?

We run a healthy year-round internship program at Paanduv Applications. Our next cohort of interns starts today and they will be working on boiling/condensation modeling, GPU parallelization, and CO2 synthesis modeling. Interested students and researchers with interest in OpenFOAM, solver development, software development, Artificial Intelligence, computational fluid dynamics, GUI development, hands-on prototyping, technical blogging, and state-of-art upcoming tech can send their applications to careers@paanduv.com. For general queries, contact support@paanduv.com.

The most satisfying part of our jobs at Paanduv is seeing the exceptional work of our interns, typically at cutting-edge tech. Amazing what some students can attain without a higher degree, with dedication as their main skill.

Anagh Dave

Bachelor of Technology, Mechanical Engineering

Manipal Insitute of Technology

Parth Dev Bundela

Bachelor of Technology, Mechanical Engineering

IIT (ISM) Dhanbad

This article summarizes the good work that our latest cohort of interns - Parth and Anagh - produced during their time at Paanduv. They validated the thrust coefficient of APC slow flyer 10”x7” propeller for drones using OpenFOAM CFD software with the reported experimental data and also compared the results with those reported by a commercial CFD software.

OpenFOAM performs way better than the commercial CFD software in terms of the estimation of the thrust coefficient

Backed with the confidence of superb validations for the thrust coefficient under different conditions of freestream velocity, our team extended the work to quadcopters.

The media below shows the simulation of a Quadcopter with 4 APC slow flyer 10”x7” propellers at a freestream velocity of 2.44 m/s and rotating at 3000 RPM. Simulated using advanced dynamic meshing methods in OpenFOAM and photo-realistically post-processed using advanced ray-tracing in Paraview

The rest of the article is in Anagh’s own words.

Have you ever looked up into the sky and seen an aircraft flying, or if you are lucky, possibly a drone?

Over the ages and years, air travel and aircraft have evolved from the first airplane that Orville and Wilbur flew to today's jumbo jets such as the Airbus A380 and Beluga. However, today we look up to see much smaller flying vehicles, much stealthier, ........................ you guessed it, Drones!

Drones have a rich history and have evolved immensely, from being in the form of large airplanes to tiny insects like one that you might have seen in the movie "Eye in the Sky." First-ever drones or pilotless vehicles were first developed in the First World War by the British. The development of one such vehicle called the Queen Bee inspired the name "drone”. These have secretly roamed our heavens for years far above our reach.

Today drones have immensely developed from the concepts of aerodynamics, payload mechanics, and computer science to help people worldwide. Applications of drones are focused on defense technology, delivery systems, scenic photography and videography, and agriculture. The most recent types of drones are Micro UAVs. Micro UAVs for defense. These are mainly used in military and battlefield surveillance in areas that are inaccessible to soldiers by foot or ground vehicles.

Low airspeeds with low Reynolds numbers are where these drones truly shine.
Fundamentals of this drone have come from the Biomechanics of insects such as bees, which was the inspiration for the name “drone” itself!

We know of several popular military drones manufactured by Lockheed Martin, Northrop Grumman, and others like the Predator, Darkstar, etc.

Agricultural Drones

On the other hand, Agricultural Drones are way more fundamental and essential to us as the general public in farming. They have been insanely valuable for irrigation, harvesting, and spraying pesticides. They save a considerable chunk of the farmer's time from growing crops. Hence, it helps both the farmers as well as the consumers.

But, farmers can effectively use drones only for irrigating large fields, which helps spread the water to a larger area and makes the drones cost-effective.

As these drones are expensive, more so for smaller farmers, it is paramount that we develop methods to optimize the payload, i.e., water and pesticide carrying capacity and the aerodynamics of the drones.

Computational Fluid Dynamics for improving drone designs

Computational Fluid Dynamics is a beautiful tool that one can use to analyze the impact of fluid on any arbitrarily shaped body, specifically these drones. Consequently, drone designs can be improved for better thrust coefficients. This tool primarily solves the fundamental fluid flow equations of Navier Stokes Equations using a range of numerical methods like SIMPLE, Upwind scheme, etc.

CFD is a stepwise process involving building geometry and then meshing to process the results into a meaningful conclusion. Meshing is paramount to getting accurate and precise results. As our drones are complex geometries, they result in complex 3d meshes.

Transitioning from ANSYS to OpenFOAM

Being accustomed to some commercial software such as Fluent, Hyperworks, etc, it was a challenge to transition to the more open-source OpenFOAM. Adwaith and Manoj sir have helped me understand the basics of programming in it with the “Basics of OpenFOAM” webinar being really informative and educating.
I have developed a real passion for OpenFOAM since first attending the “CFD by OpenFOAM” workshop in the month of October of 2021 and was really fascinated by how flexible the pre-processing processes like meshing and constructing geometry were. If you are interested in learning OpenFOAM, feel free to apply to our hands-on workshop here.

OpenFOAM gives you the liberty to decide the element size at any certain region of your choice in the domain thus helping to refine the mesh and get more accurate results than Fluent.

Adwaith Gupta and Manoj Joishi sir have helped and guided me through the process of understanding the challenging process behind a simulation.

Before interning at Paanduv, I tried to design and simulate an exhaust fan but I always was bamboozled by the method to actually rotating the entire fan. I was introduced to the concept of Arbitrary Meshing Interface (AMI) during this stage, but I never fully understood it.

How does one know where the fan is or what coordinate does it designate in the domain? If I find the coordinates, how do I make it rotate?

These were some of the questions I had in mind. I got my answers as soon as I joined the internship. I enjoyed learning every single bit and it's been a real education. The way we approached the problem is described below.

The Thrust Coefficient

Navier Stokes Equations are the fundamental fluid flow equations that are solved for mass and momentum conservation and fluid variables like pressure and velocities in three directions are evaluated.

The simulation was performed for an angular velocity of 3008 RPM for the propeller and with three different free stream velocities and for the validation [1] was used.

Advanced ratio, which is the ratio of the freestream fluid speed to the propeller tip speed was calculated using

where V is freestream velocity, n is the angular velocity in rps and D is the diameter of the propeller.

The coefficient of thrust (Ct) can be calculated using

Where T is the thrust force, ρ is the density of air, V is the freestream velocity and D is the diameter of the propeller. It is this parameter, the thrust coefficient that directly impacts the payload capacity of a drone.

Pre-Processing

The geometry used for simulation was a standard propeller known as APC slow flyer 10”x7”.

The simulation used AMI (Arbitrary Mesh Interface) for the dynamic meshing which rotates the propeller, kept inside the AMI.

Initial meshing was done using blockMesh which generates the base mesh in the simulation domain. Secondly, an OpenFOAM mesh optimizer called snappyHexMesh is used to import the propeller CAD file and thereby create an unstructured mesh, thus refining the mesh at desired locations.

Dynamic meshing

The meshing strategy was divided into three parts :

The preliminary process is using blockMesh to initially create a simple base Mesh.
After this, snappyHexMesh can create a more unstructured mesh to encapsulate the drone.
As drones are complex flying objects, especially in the case of our simulation, i.e., a quadcopter, it is easier to analyze one propeller and reciprocate the results to the rest of them. To rotate the propeller, we need to build an arbitrary mesh interface (AMI) that contains two parts: a slave and a master. The circular domain that encompasses the propeller and, when meshed, forms the AMI. This AMI takes care of the propeller's rotation, where the master rotates the slave and remains static by itself.

The total mesh cell count was 4,12,278.

Initial and Boundary Conditions

In the following scenario, we are posed with a problem consisting of initial boundary conditions of a range of free-stream velocities and constant turbulence intensity (0.1%).
Outlet conditions were set as inlet-outlet type with uniform pressure and velocity.

Let's talk about Results

The simulation was smooth and converged perfectly, thus verifying the results on the research paper with drag and thrust coefficients. The following results were obtained from the simulation.

The thrust coefficient decreases with an increase in velocity.

Pressure and velocity contour around propeller for J = 0.628 and the freestream velocity of 2.4 m/s

Velocity Contour at inlet velocity 10.1 m/s or J= 0.799 with K-Omega SST turbulence model.

From the above results, we can conclude that the OpenFOAM has given us results with better accuracy and the variation of thrust vs free stream velocity can be understood very well.

Page updated

Report abuse