Design Engineering
Showcase 2021

OceanHub Africa

Tags
CAD
Rendering
Animation
Data Science
Sustainability
Ocean Innovation

Project Details

Student
Alexandre Berkovic
Supervisor
Dr Sam Cooper
Role
Engineer in Residence
Sector
Consultancies
Links
Email
LinkedIn

OceanHub Africa is a startup accelerator based in Cape Town which focuses on impact-driven companies in Africa that take part in ocean innovation.

My primary role was to support the incubated startups in their technical development such as CAD, mechanical design, rendering, simulations, and data science. I mainly supported four start-ups (Brayfoil, GIVO, Inseco, Shark Safe Barrier), bringing ad-hoc support to another four.

I also focused on internal missions for the incubator such as organising the data segment of the Hackathon, editing a video for the 1000 Oceans Startups coalition, or developing the design of the soon-to-be FabLab workshop.

Oceanhub innovation

Demonstration of Design Engineering Thinking and Skills

CAD: Working on developing 3D models for the incubated companies quickly became one of my biggest tasks. After being initiated to CATIA, I went on to develop models and assemblies for companies such as Brayfoil, GIVO, Arena Recycling, Inseco and Shark Safe Barrier.

Simulations: With Brayfoil, I managed to leverage 3DS’s Simulia app to develop mechanical simulations as well as CFDs for their morphing aerofoil blades.

Rendering/animation: Throughout my placement, I developed projects on a very technical/engineering aspect. However, an essential aspect of an incubator is to raise funds which means that the commercial impact must also be developed. Hence, I created numerous render stills and animations for startups to use in their investment pitch decks.

Deep Learning: When working with GIVO, I quickly understood that they positioned themselves as an IoT-focused company. Hence, alongside their CTO, we decided to implement an image recognition algorithm which would help automate waste sorting.

Data science: Concerning the Hackathon, Soso Care’s challenge required a substantial amount of plastic waste data from Nigeria. However, granular data on plastic waste is nearly impossible to find for a country like Nigeria due to lack of research, the immensity of the country … Hence, I used my experience in data science to create simulations and generate data. The code development consisted in Discrete Particle Tracking which is a Lagrangian simulation that was achieved after lengthy research and close work with Delphine Lobelle.

Teamwork: Initially, I engaged in training to master Dassault’s 3DS platform alongside Laurent Marche who works in partnership with the company. I spent a lot of time on the Eduspace to learn different skills and worked closely with the 3DS team to engage with their experience of the platform especially in CAD development (Sebastien Rosel), simulations (Weronika Spiewak) and CFDs (Srikrishna Chittur).

Project management: As a part of my placement, I had to triangulate my work between the incubator, the startups and our partners. Hence, I organised bi-weekly follow-up meetings with all the startups as well as weekly meetings with the incubator to show progress and discuss potential opportunities or further developments.

Event organisation: Throughout my Hackathon mission with OHA, I have had the opportunity to present and discuss proposals to several stakeholders involved in the event’s organisation such as sponsors (AWS, Société Générale …), challenge owners (Soso Care, Coral AI, Save Local), researchers (Utrecht, UCT, MIT …) and organisers (OIA, OHA, City of Cape Town…). Working and interacting with so many stakeholders was very stimulating and allowed me to iterate over ideas in much more depth.

Role and Contributions

During my internship, I worked alongside Brayfoil, a leader in compliant structure and morphing aerofoil research and design providing unique capabilities to implement breakthrough technologies for efficiency gains and cost reductions. One of their issues was to make their technology understandable to the wider audience. Hence, alongside CEO Matthew Bray, we discussed about creating a video that would depict their technology in situ, its biomimicry inspiration, and its comparison to current products.

image 1

Alongside the animation and renders produced, I worked with Dassault System’s team to develop a setup In Simulia to perform CFDs and FEAs in which the aerofoils are being morphed through a simulated kinematic system that modifies the blade’s pitch angle depending on the incoming wind profile. The simulations were underway until Brayfoil decided to continue using Open Foam instead of 3DS.

image 2

I also collaborated with GIVO, a circular economy company that offers an end-to-end recycling solution using IoT enabled devices to collect recyclable material and process these into consumer and industrial goods. The first project which I took part in was to design GIVO’s waste processing centre which they want to dispatch 4000 of across Nigeria. Initial designs were proposed to the company’s CEO & CTO and after about 5 iterations, we were able to achieve a desirable output. I then spent time in CATIA to develop an assembly of the container and in parallel, developed a pricing sheet for all its components, contacting suppliers to outsource them. The technical drawings that I made are currently being used to build the containers, which should be finished within September.


My second project was to develop a computer vision algorithm that would identify recyclable waste and separate it into different classes to help automate the sorting phase prior to processing. Initially, the idea was to use data given by GIVO to create a pipeline that would differentiate plastic waste by colour, brand, use, material … However, due to bottlenecks, the data was not created in time and an opensource dataset with broader labels (paper, plastic, glass, cardboard, trash, metal) was used. I created a convolutional neural network in Python with FastAI, optimised the learning rate (and other metrics) to achieve a 93.3% accuracy. The code’s architecture can be used for training once their dataset is ready (although accuracies will differ).
image 4

Concerning other companies, I worked with, the contributions I made were mostly regarding mechanical design, CAD, technical drawings and renderings solely.

image 5

I was also given the task to organise the challenges for the African leg of the Ocean Hackathon. I worked with three companies to come up with the challenge definition (of which we will look at two) as well as the data.

The first challenge is that of Soso Care, a Nigerian company that aims at lowering the price of life insurance by incentivizing people to bring in plastic waste. The goal of the challenge is to determine the optimal locations for Soso’s waste management centres based on plastic influx and demographics. Demographics data was collected through our partnership with the French embassy. However, for the plastic data, I worked alongside Delphine Lobelle, who is a researcher at Utrecht university working for Ocean Parcel’s TOPIOS team, investigating how 3D ocean circulation impacts plastic transport. Together, we devised a Langrangian Simulator script that determines the location of set particles after drifting away due to oceanic currents. With this, I generated a detailed map of yearly plastic waste outfall in coastal and fluvial regions of Nigeria.

image 6

The second challenge is that of Reef Support which aims at aiding coral conservation through artificial intelligence. The goal is to use photo-quadrats form coral reefs and create an image segmentation algorithm to outline the corals, then use image recognition to determine the corals’ genus levels, species as well as defining their resiliency (bleaching percentage, maturity, branching). I managed to obtain data from the XL Caitlin Seaview Survey program and devised an algorithm to determine the feasibility of the challenge within 48h.

image 9

Summary

During my placement, I worked: GIVO to develop a technology that reduces plastic waste, Brayfoil to develop an innovative mechanism for wind turbines, Inseco to allow for the creation of a more sustainable food source, Arena Recycling to develop construction bricks made from recycled plastic waste, Shark Safe Barrier to develop a shark repelling technology that protects humans without harming the ecosystem and created three sustainability-related projects for the Hackathon. Having such diverse missions has been an incredible experience!