Global Engineering |
Enhancing Fish Feed Distribution Systems in Aquaculture Farms in Ghana

This was a project overseen by Prof. Amos Winter of MIT’s Global Engineering and Research(GEAR) Lab and in collaboration with our community partners, Tropo Farms; an aquaculture company based in Ghana. I worked in a team of 4 to design a modular device to convert off the shelf leaf blowers to affordable fish feed distribution systems

The project produced a co-authored a paper delineating the design and validation process that was submitted to the ASME 2023 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference (IDETC/CIE2023)

Key Skills : Machine Design Theory, Stakeholder engagement, analysis of developing/emerging market dynamics, user-centered product design, rapid prototyping(CAD, 3-D Printing, workshop fabrication tools), first-principle calculations, developing theoretical models to guide mechanical design

Framing the Problem

We met with our project's industry partners Innovasea and TropoFarms, both companies involved in offshore aquaculture fisheries. Innovasea specializes in advanced farm management software such as fish tracking and farm water analysis to develop solutions to various aquaculture challenges. The company sought to collaborate with Tropofarms, an aquaculture fishery in Dzorwulu, Ghana, whose main goal was to expand farming operations to double the yield while maintaining operational costs and number of employees

We analyzed the current operations at the farm to establish areas for improvement. The farm operates as clusters of offshore cages each cage 19-20m in diameter, 6m deep, with a total of 72 cages.

We were able to obtain videos of farming operations where we learned that the fish feeding process was entirely manual, with teams of two workers sailing out to the cages multiple times a day to hand toss fish feed pellets to the fish in each cage. Over the course of the day, 100+ metric tonnes of feed pellets will have been tossed into the cages.

Upon analyzing the operations, a few inefficiencies stood out. Our partners mentioned that upon harvest there was a drastically uneven distribution in the mass of fish with weight ranging from 150-1000g. The feeding method was also rather laborious for the workers and entirely depended on the reach and accuracy of their toss.
This also resulted in clustered feeding frenzies close to the cage wall where the feed was being tossed. We immediately began to toy with this possibly being a factor influencing the weight range, forming a hypothesis that bully fish, or those that come out on top during these frenzies, tended to eat more, and reached a higher weight come harvest

In addition to analyzing the farming processes, we interviewed the stakeholders for a more quantitative analysis of their key production metrics used in tracking or measuring efficiency. We learned that because feed is the highest cost area, they focused mainly on the feed conversion ratio(FCR), that is the mass of feed input, over the mass of fish output. Cost analysis calculations projected significant savings for even a slight reduction in FCR.

The stakeholders also pointed out that the FCR provided was an average, with experienced distribution teams consistently having efficient fish production with a lower FCR. The experience here ultimately affected the way in which the teams distributed the feed, either through better technique, or attuned human perception of the fish behavior influencing their decisions while feeding the fish

These observations led us to focus on improving the feed distribution method for a reduced FCR, potentially through the introduction of a mechanized feed launcher. At the same time, we recognized that the farm was a major employer in the region, and there was a lot of value in leveraging human perception and intuition during the feeding process. Therefore, our solution would need to:

• be portable, with an on board hopper to hold the feed
• implement range and flow control allowing farmers to tune distribution range
• be worker operated to maintain employment and appreciate human perception and intuition.
• be cost effective and use locally accessible materials
• be easy to use/train farmers on

Summary of Strategy Selection Process

Prototype Development Process

Having framed the problem around optimizing the feed distribution system, we defined our key performance metrics as the range of feed distribution and mass flow rate, or mass of feed ejected from the system in a given time. We tabulated our contraints, metrics and reasons for each as shown below:

Having identified a solution path and our constraints, we began the process of designing and prototyping. Our product development process centered heavily on identifying the most critical module or main component upon which the mechanism operation depended. We broke down the device components and identified the launching mechanism as the MCM for our proposed solution

We then began the process of downselecting through different possible concepts for the Launching mechanism

Concept 1 | Mechanical Contact

We explored the idea of using a high speed rotating set of blades that would hit and launch feed coming in from the attached device hopper. We obtained information about the type of pellets used, placed an order for the feed, and then used the pellet characteristics to obtain values sich as mass and volume to plug into first order calculations as an initial validation of the concept. We immediately hopped into prototyping sample rotary blades, making use of a handheld drill to provide variable rpm to the blades.

We iterated on the blade design and developed CAD models of potential enclosures for the mechanism. While this mechanical contact launching system provided a higher efficiency/energy transfer due to the direct contact with the pellets, the components required to make the mechanism viable for the launch range and feed mass flow rate required would be too expensive to locally produce and power, pose safety risks in operation, and potentially crush feed pellets causing wastage and potentially jamming

Concept 2 | Pumping out a Water-Feed Slurry

Our second concept of using water as a medium of transport was eliminated almost immediately due to the fact that breaking down the pellets into a slurry would cause the fish to not immediately recognize it as food, leading to wastage. In addition, the slurry would interfere with the water pH, deteriorating fish health. As opposed to solid pellets which are consumed much faster before the water starts to interfere with their structural integrity

Concept 3 | Air Assisted launching

The third concept explored was the use of air as a transport medium for the pellets. We analyzed commercially available pellet feed blowers designed to meet a variety of feed distribution ranges and mass flow rates, however these were too expensive, heavy or locally unavailable, barring their implementation.
We then explored the idea of using off the shelf leaf blowers for the following reasons:

• they are already portable, designed to be worn or carried by the average able-bodied adult
• range and flow control could be achieved by attaching the blower to a pivoting platform that allowed for changes in yaw and pitch. Mass flow rate would be controlled by the rate of addition of feed to the hopper, and inbuilt power/speed controls on the blower
• leaf blowers are commercially available, and after consulting the project parnters, the potential gains for this scale of farm allowed for the investment in the blowers, using a proposed price point of $500 per blower.
• the use of the blowers is relatively intuitive, and still implements a level of manual implementation, leveraging user decision making while making the feeding process far less laborious for the farmers.
• off-the-shelf leaf blowers come rated by air velocity and volumetric flow rate, which allowed us to perform quick, "back-of-the-envelope" first principle calculations to estimate our launch power needs, and purchase a couple of blowers for the experimental validation process

Our next step was to develop a first-order theoretical model to validate our air assisted power launch concept. We sought to develop a Range Model and a Mass Flow Rate Model, making executive decisions to simplify the complex turbulent, particle-laden airflow to quickly obtain an order of magnitude estimate of the range and flow rate to test concept fesibility. The extensive calculations are documented in the accompanying paper compiled by the team

The theoretical results informed our purchasing of off-the-shelf blowers for the experimentation phase, by providing key specifications such as air velocities and nozzle lengths for maximum desired output. We started out by modifying the existing nozzle attachments by drilling inlet holes for the fish feed pellets

Our initial prototype gave promising results, nearly meeting our range requirements, so we went on to experiment with modified nozzle lengths and attachments. We performed several experiments to observe the effects of these modifications as well as validate our first order model. Our focus narrowed on optimizing the attachment, which allowed us to rapid prototype several options without the need for rebuilding an entire device from scratch

Our final prototype involved developing a 3D printed blower converter that could be attached between the blower motor and the outlet nozzle, with an inlet to which a hopper containing the feed could be attached.

Experiments were conducted with the final prototype attached to a cart. The launched feed fell within a range of distances from the cart, with the maximum and minimum distanced measured and recorded. The feed weight and launch time were also measured. Further experiments would see different pellet sizes used with the current prototype to get a better intuition for the mass flow rates and continue validating the model using varied parameters.

The results met the tabulated target values, proving the viability of the converter and blowers as an improved fish feed distribution system

Future work would include:

• development of a robust mounting system for the blower that grants 2-rotational axis degrees of freedom and designed for integration on boats used in Ghana
• attachment of a robust hopper large enough to hold 20 kg of feed (equal to one whole bag of feed typically used by Tropo Farms) would also be developed so workers could refill the hopper every minute.
• increased levels of testing to determine the relationship between inlet tube size and wind speed in order to optimize the mass flow rate and range while maintaining long term blower performance
• scaling the attachment to be 3D printed on demand and tailored to dimensions of local leaf blowers
• testing the product an actual fish farm to see the drop in FCR over multiple harvest cycles