Overview
This project was born from a need for fresh, bio-diverse food and a stubborn resistance to chores such as grocery shopping. This aquaponics system is automated to provide a stable, closed-loop environment for vegetables and fish, which in turn provide lunch for 2-3 people daily.
Automating this aquaponics system was a long process of experimentation and design iteration. This is a collaboration with Daniel Theobald, CEO of Vecna, and over time we have gradually automated more processes and improved their reliability. Now the system can be left for several days at a time without any decline in functionality, and only a couple minor tasks must be done manually every few days.
My ultimate goal is to couple this existing system with robots that harvest and prepare vegetables and fish. Someday I intend to order lunch by calling the aquaponics system from a custom application on my phone.
Automation
The effects of variables such as water temperature and pH, as well as the quantity of dissolved oxygen, nitrates, ammonia, and other chemicals are well understood (Rakocy et. al., 2004). Using these data as guidance, we warmed the tank to 27 degrees Celsius using four 200W aquarium heaters. The temperature is maintained using a thermometer sensor that triggers these heaters to turn on when the temperature drops below our minimum threshold of 26.4 degrees Celsius. The aquaponics system is located indoors in an area that does not rise above 29 degrees Celsius, so we do not have a mechanism for cooling the tanks. The pH of the water is regulated similarly; when the pH sensor measures a pH below our minimum threshold of 6.5, it triggers an aqualifter pump to draw pH UP into the water.
Optimizing E&M Spectrum for Plant Growth
To maximize energy efficiency, we use LEDs that emit only the frequencies required for plant growth. This enables us to reduce the total light emitted, and thus the overall power usage of the system.
To determine the combination of frequencies that is optimal for this system, we set up a series of 8 identical trays, each containing the same initial quantity of a plant called duckweed. We selected duckweed because it is a high-protein plant that would provide a healthy diet for the fish. It also reproduces quickly, enabling us to evaluate the effects of various light frequencies on growth in a matter of weeks.
We compared the growth of the duckweed by jostling it into a single layer, and then measuring the surface area it covered. We found that the best growth was under blue light, followed closely by red light, which agrees with previous research. Although this has been thoroughly researched, we wished to quantify the effects for our own system before investing in LEDs.
A second reason we are interested in optimizing the LED spectrum is to guide the growing cycle of the plants. Blue light maximizes plant growth, while red light combined with blue light induces flowering. For most plants, we aim to maximize foliage growth and are therefore increasing the number of blue LEDs.
The spectrum of this lamp is approximately 90% red and 10% blue. In this separate experiment, we are characterizing the effect of this heavily red balance of light on arugula plants. The leaves are thicker and have a far stronger flavor than the arugula grown under broad spectrum lighting.
As expected (but still interesting to see), green plants absorb red light and thus appear black if illuminated by red light.
The frequency at which the fish are fed and the content of their food impacts the nutrients they provide to the system. As fish metabolize protein, they produce more nitrogen and phosphorous; thus a higher protein food will result in a higher concentration of these nutrients. Our first priority is to provide a healthy environment and diet for the fish, and we were not particularly concerned with the speed of their growth. We therefore selected a fish food with a balanced diet including protein and vitamins and minerals required by the fish. An automatic feeder that is triggered by the aquarium controller releases fish food into the tank three times per day.We also supplement this commercial food with duckweed, a highly nutritive plant that contains more protein than soybeans. We grow the duckweed on the surface of water, separated from the fish by a black cloth bag. Duckweed also removes nitrates from the water, and is used to modulate the amount of nitrogen in the water when the plants in the grow beds are harvested.
If you're squimish, you may want to skip this part. I describe a humane method for harvesting fish.
Harvesting the fish is currently a cumbersome task. They must be caught using a net, descaled, killed, and cleaned before cooking them. I am working on an extension to the system that catches, kills, and descales a fish. The system first lures a fish into a trap. To select only the larger fish, the trap has escape routes for fish that are small enough to fit through an opening. The system detects whether a fish is in the trap using a simple turbine, which measures water flow. Once a fish is inside, the fish is swept into another compartment that is well isolated from the tank. I researched the methods used to kill fish in aquatic farms, but I have decided not to use them because I prefer a quicker and more humane method. An electric current is applied through the water. Electrofishing is a method commonly used to count fish in natural lakes. This system follows the same basic principle, but with an increased voltage.
Although aspects of this system are fully functional, there are many opportunities for improvements. I eventually plan to integrate this system into my kitchen and partner it with a cooking robot. I've begun working to design the next steps toward that goal, and am developing separate methods for automatically harvesting the plants and the fish.
To investigate the feasibility of scaling the system for use in urban areas, I am measuring the edible content and monetary cost per unit area, among other metrics, for several types of plants and fungi. Mushrooms score well for each of these scalablility metrics; they are nutrient rich, grow relatively densely compared with plants, and inexpensive. I researched mushroom varieties and selected 3 that have antiviral, antibacterial, or anti-inflammatory properties:
-
Turkey Tail mushrooms have antiviral and antibacterial properties (Stamets, 2005).
-
Reishi mushrooms have strong anti-inflammatory properties antioxidant properties that have been shown to inhibit the growth of leukemia cells (Stavinoha, 1990).
-
Shiitake mushrooms also contain antiviral compounds and broad antibacterial properties (Hirasawa et al. 1999).
I wanted to set up a prototype of the mushroom system very quickly, so I purchased a kit for each type of mushroom, which includes a substrate that has been infused with spores. Because different mushroom species prefer different substrates, I keep them separated from each other. I also keep them separated from the plants growing in the aquaponics system to avoid unintentionally spreading fungi. In the first version of this mushroom system, each of the three mushroom species is placed into a separate stackable plastic bin.
Modified System For Growing Mushrooms
I configured a timer to spray the mushrooms four times per day for 1 minute. Determining the frequency and duration to water the mushrooms took some trial and error, since the mycelium (a part of the mushroom) forms a water resistant coating over the substrate, sloughing off the water instead of absorbing it. The mushroom fruitbodies prefer to be damp, thus more frequent watering was desirable. A few weeks later, had a burgoning crop of mushrooms.
A few months later, the mushrooms were thriving. These are the Turkey Tail variety.
Automated System For Harvesting Fish
Automated System For Harvesting Plants
After some experimentation, we have determined that some plants can survive even if all of their leaves are clipped. Therefore I am working on a system that would use computer vision to identify an area of the garden where plants meet a minimum requirement for foliage. I am working on controlling a mechanical arm with a mounted camera to use computer vision techniques to determine whether a plant meets a minimum height threshold. If so, the mechanical arm would position a pair of scissors at the base of the plant, snip, and then place the leaves into a box. I have a few ideas on how to automatically wash the greens, but I have not begun working on that step yet.
Automated Aquaponics System
Page Contents
(1) Fish live in a 1360 liter tank that serves as the base of the system. Urea from the fish provides nutrients such as nitrogen to the plants.
(2) A pump sitting inside the fish tank draws water into a PVC tube and deposits it in the top of the 2 grow beds.
(4) The water level rises 8cm to the base of the plants, entirely saturating the roots. The grow beds are slightly tilted toward the front of the system so that water can drain, leaving no standing water. It is important to eliminate standing water in the grow beds; otherwise there is an increased risk of root rot and pests.
(5) Water drains through a single hole near the front, and then passes through a second cloth filter before entering the lower grow bed.
(6) Identical to the upper grow bed, the water level rises in the lower one and then drains through a hole at the front of the system, returning to the fish tank.
WATER FLOW
(3) Water from the fish tank passes through a cloth filter (not shown) before flowing into the upper grow bed.
System Design
Many of the design choices for an aquaponics system are interdependent, and balancing them is a question of which variables are intended to be optimized. Designing the fish tank to be compatible with the grow beds is a good example of this. Nutrients for the plants come from the fish, and in turn the plants remove compounds from the water that become toxic to the fish in sufficient quantities. The containers for both must be adequate for their health and growth. Seeking to maximize the vegetation of the system, we selected a grow bed 1.36 wide by 2.6 meters long (recognizing that we must be able to comfortably reach the center of the grow bed, the width must not exceed twice the length of an arm, or approximately 1.5 meters). Because the plants require a large surface area to grow, we can make the most efficient use of space by stacking the grow beds on top of each other, leaving sufficient room for plants to grow between each layer. Leaving the plants growing space of 40-45 cm, and accounting for the space required by the lighting and the grow bed, the current system is limited by the ceiling height to just 2 grow beds, however a system requiring the same footprint could have a deeper fish tank and more grow beds stacked on top. Using the surface area of the grow beds as a measure of the quantity of vegetation, we can next estimate the quantity of fish required to balance the system, and finally determine the dimensions of a suitable fish tank.
To estimate the quantity of fish needed, we began with a ratio reported by Dr. James Rakocy and his team at the University of Virgin Islands. They began collecting data on aquaponics systems more than 30 years ago and have published numerous useful studies on their findings (Rakocy et al. 2004). His feeding rate ratio was the first to predict the quantity of fish feed per area growing space: (60-100 grams) / m^2. To determine the dimensions of the fish tank required for our system, we must calculate the volume of water it must hold. Estimating 1kg of food per 50kg of fish and 40-52 liters of water per kg of fish we calculate the volume of water needed per m^2:
(0.06 - 0.1 grams of fish feed)/m^2 * (50kg fish)/(1kg feed) * (1kg fish)/(40-52 liters of water) = 120 - 260 liters of water/m^2
Each grow bed has a surface area of 1.36*2.6 = 3.5 m^2. There are 2 grow beds, giving us a total of 7 square meters. Thus the volume of the tank must be between 840 - 1820 liters.
This is a broad range, and once we are within the healthful range of the system we can optimize for space. The width of the tank was selected to match the width of the grow beds, making it possible to neatly stack them. The length of the tank is 32cm longer than the grow beds to provide space to feed the fish and to access the water for cleaning and other monitoring. With the lights above the upper grow bed already at ceiling height, this leaves 1 meter of clearance under the lower grow bed. Leaving sufficient space to draw oxygen into the water, we selected 60cm for the depth of the tank. The resulting tank comfortably holds 1360 liters.
Rakocy, James E., et al. "Update on tilapia and vegetable production in the UVI aquaponic system." New Dimensions on Farmed Tilapia: Proceedings of the Sixth International Symposium on Tilapia in Aquaculture, Held September. 2004.
Stamets, Paul. Mycelium running: how mushrooms can help save the world. Random House LLC, 2005.
Stamets, Paul. "Novel antimicrobials from mushrooms." Herbal Gram 54 (2002): 28-33.
Hirasawa, Masatomo, et al. "Three kinds of antibacterial substances from Lentinus edodes (Berk.) Sing.(Shiitake, an edible mushroom)." International Journal of Antimicrobial Agents 11.2 (1999): 151-157.
Future Work
References
There are numerous events that must occur on schedule each day to keep an aquaponics system in good health. Lights increase in brightness over a 30 minute 'sunrise' period and similarly dim at night; the fish are fed three meals per day; plants are watered four times per day; and adjustments are made to keep the water at the right temperature and pH. In the first prototype of this system, all tasks were performed manually, and over time we have adopted automation techniques for almost all maintenance.
We initially automated the system basic digital timers intended for automatically watering lawns or other hydro systems. While it is certainly possible to keep the system going on these rudimentary devices, we found it to be too much effort. In instances of power outages or other disruptions, these timers failed. After cleaning up after one too many floods, we decided to purchase an aquarium controller by Neptune Systems. This controller comes with a 32-bit microprocessor that servers as the base of the system. The first set of temperature, pH, and conductivity sensors we tried required re-calibration tests at least once per month, which we found to be too frequent, and we replaced them with the sensors in the Neptune controller package.
One of the most important features of this controller is the integrated router that provides internet access to data and controls. Using this web interface we configured separate controls for each light, pump, feeder, and sensor (left). Each component is plugged into a designated outlet in the base controller (right).
Sensors for pH, conductivity, and temperature are held against the side of the tank with magnets (below). Analyzing the sensor data enables us to better assess the impact of changes to the system, such as upgrading a pump or adding fresh water. As we collect more data, we adjust the frequency at which the fish are fed, fresh water is added to the tank, and plants are watered.
The pump responsible for circulating water from the fish tank to the grow beds is an AquaForce model made by Aquascape. It has a sufficiently high flow rate of 850 GPH at a height of 1.6 meters, and filters solids from the water. As depicted in the overview, the water is pumped to the upper grow bed through PVC pipe, and flows through a finer cloth filter before entering the grow bed.
Automation
Optimizing Health of Fish and Vegetation
These stackable bins provide a convenient and space efficient environment for growing mushrooms. Similar to the aquaponics system, I drilled a single hole in the bottom of each bin to drain excess water, then covered the hole with a wire mesh to avoid clogging. The bottom container is raised on a set of bricks, which gives enough clearance for a hose that runs from the hole in the bin to a floor drain. I keep the mushroom environment dark and humid by loosely covering it with large black trash bags.
The mushrooms are automatically watered using a micro drip irrigation system. There are 2-3 sprayers in each bin.