Final Design

Shown above is a fully constructed sensor array. Here, one Arduino controls 6 CO2 sensors.

MODULAR DESIGN

Our final design's primary configuration consists of a single Arduino Mega 2560 microcontroller connected to six SCD30 CO2 sensors, each connected to an Adafruit TCA9548A I2C Multiplexer. Various rooms have various shapes, so our final design keeps modularity in mind. By adding more arrays, our modular sensor systems are able to cover separate areas of a room simultaneously. By utilizing long jumper cables, each array is able to cover a 6.6 foot radial space. With this open design, we are able to measure and collect data on critical areas in a room no matter their location.

Shown above is a configuration utilizing 2 Arduinos. This design allows for CO2 sensors to be placed at vital locations around a spacious room. The inherent modularity of this configuration would allow for a total of 15 CO2 sensors if additional sensors were to be purchased.

Shown above is a configuration utilizing one Arduino. This configuration allows for higher density sensing for more higher quality measurement.

REVIEW OF DESIGN COMPONENTS

2 CO2 Sources

1 Arduino Mega 2560 Microcontroller

1 Adafruit TCA9548A Multiplexer

6 SCD30 CO2 sensors

2 SPS30 Particulate Matter Sensors

CO2 Molecule

KEY DESIGN DECISION:
THE SURROGATE

To mimic the flow of human breath, CO2 was used as a surrogate. Affordable, measurable, and easily obtained from our sponsor, our sensor array detects CO2 emanating from a source. CO2 has a history of being used in airflow experiments and is detectible with relatively cheap sensors. Even though our final product is incapable of accurately measuring the flow of aerosols inside of a room, CO2 concentrations inside of living spaces heavily impact humans.

Although everyone exhales CO2, it is still dangerous! While the normal background concentration of CO2 is about 250 to 400 PPM, humans feel adverse affects starting at concentrations of 1000 PPM. To ensure the safety of our team members, a buddy system was used whenever we conducted experiments with gas.

KEY DESIGN DECISION:
THE CO2 SENSOR

After testing multiple different CO2 sensors, our team decided on utilizing six SCD30 sensors from Sensirion. With accuracy, range, sampling rate and cost taking part while deciding which sensor to use in our array, Sensirion's SCD30s were the cheapest on the market which tailored to the needs of our project. Sampling every two seconds with an accuracy of 30 PPM over a range of 0 to 40000 PPM, the SCD30 provided fast and reliable results.

SCD30 Sensor

Arduino Mega 2560 Microcontroller

KEY DESIGN DECISION:
THE MICROCONTROLLER

To operate all of our CO2 sensors at the same time while keeping the array portable and cheap, an Arduino Mega 2560 microcontroller was utilized. With its large number of varied communication ports, it was perfect for connecting to all of our SCD30 sensors. In choosing a microcontroller, our team attempted to find one that was not only cheap, reducing the cost of future sensor arrays, but also large, easily able to hold as many SCD30 sensors as needed. Even though our design utilized 6 SCD30 sensors in total, our microcontroller had space for more! Our code can also be used on multiple separate microcontrollers, allowing for smaller arrays to be spread out across a larger room.

KEY DESIGN DECISION:
PARTICULATE MATTER SENSOR

Along with measuring CO2, our sponsor requested that the sensor array also measure particulate matter inside of a room. Particulate matter comes in different sizes, infiltrating human lungs and causing many serious health effects, including heart and lung disease. Examples of such particles include dust, soot, and various airborne metals.

To measure these particles, our team utilized the Sensirion SPS30. Boasting accurate and fast sampling, the SPS30 provides mass concentration measurements of PM10, PM4, PM2.5 and PM1 particles. Knowing concentrations of these particles in living conditions alludes to the safety of living in such environments.

SPS30 Sensor

Adafruit TCA9548A Multiplexer

KEY DESIGN DECISION:

THE MULTIPLEXER

With our array using multiple of the same sensor, our microcontroller needs a way of telling one sensor apart from another. Although each sensor is individually different, every sensor communicates with the same factory-programmed address. Being that multiple sensors try to communicate with the same address, the microcontroller has a hard time telling each sensor apart.

This is where the Adafruit TCA9548A comes in. This Multiplexer distributes unique addresses to each individual sensor, so the microcontroller can tell each sensor apart. If the microcontroller were the brains of the operation, and the sensors were the hands and feet, the multiplexer acts as the body, ensuring the overall task can be completed. If desired, the Adafruit TCA9548A can connect to up to 15 sensors, allowing large amounts of data to be collected simultaneously.

SIMULATION & EXPERIMENTATION

To verify the accuracy of our sensor array, we utilized CONTAM's 3D computational fluid dynamics functionality to recreate rooms and experiments that we performed in the real world. By accounting for the largest factors of air circulation in a room, such as air conditioning, open windows, and people, we are able to model scenarios, and predict where high concentrations of CO2 gas would be. These concentrations can be located all throughout a room, and the modularity in our design allows us to monitor these locations.

Shown above is a CONTAM test modeling the flow of CO2 inside of our test room.

A CONTAM simulation

KEY DESIGN DECISION: SIMULATION SOFTWARE

While performing experiments, it was necessary to validate our results. After research and sponsor recommendation, the multizone indoor air quality and ventilation analysis computer program CONTAMW paired with its sister computational fluid dynamics addon CONTAM CFD0 were utilized to produce three dimensional simulations of gas flow inside of a room. CONTAM's main attraction is its ability to simulate transient conditions. Doors, windows, and CO2 sources can be open and closed at any moment in time, allowing us to recreate conditions observed during our experimentation. CONTAM, being a simulation software, only responds to whatever it is given. Reality has many irregularities, such as infiltration, that aren't easily measured and that CONTAM can't easily account for. Oftentimes these discrepancies illuminate key factors that guide future testing.

Above is the gas canister valve used to regulate the flow of CO2 inside of a room. Set to .5 LPM, it mimics the breath of 2 humans.

To capture the flow of CO2 given by CONTAM and verify the simulation, CO2 sensors were placed in various locations along the desk to capture the flow of the CO2 cloud.

THE RADIAL EXPERIMENT

Of every experiment done for the project, below is the most archetypal one performed. Wanting to quantify the radial spread of our CO2 canisters, we placed our CO2 sensors radially around it. With the exfiltration minimized as much as possible, closing all doors and windows, it was observed that CO2 seemed to drift to the right. From real world observation, we theorized that this rightward spread was due to the exfiltration caused by a fireplace, and that our CO2 cloud was sensitive to drafts.

Image of experimental setup. CO2 sensors are placed radially from our CO2 canister.

Graph of data obtained from our CO2 sensors. Sensors to the right of the experiment generally picked up more CO2 than their counterparts.

Side view of CONTAM simulation. Wire-frame boxes simulate obstructions in the room, such as desks, beds and couches. The simulated source emits gasses radially.

Top-down view 5 minutes into the simulation

The same experimentation was then simulated in CONTAM. After simulation, it was observed that the CO2 cloud produced concentrations to the rightmost side of the room. This was due to the geometry of the room, with simulated obstructions pushing the cloud to the right.

Chart of PPM concentrations obtained from simulation and experiment. While magnitudes vary, trends of concentration remain the same.

After comparing our simulated and experimental results, many properties about CO2 concentration flow were obtained. Below is a non-exhaustive list of conclusions obtained from our data.

Whereas for different reasons, both the simulated and real cloud of CO2 tended to produce larger concentrations rightward relative to the initial source. Both factors, an open fire place and room geometry, played a role in this clouds shape, and later experiments attempted to reduce exfiltration to strictly observe the effects of room geometry.
CONTAM's CO2 source radially emits gas throughout the simulated room, while our physical CO2 canister emits CO2 in one direction. This discrepancy was observed when calculations were compared to physical results. Where trends tended to be the same, magnitudes between the experiment and simulation heavily differed. Designs were made to produce a radial CO2 nozzle for our physical experiment.
Shown in the graph of data obtained from our sensors, large sporadic spikes of CO2 concentration can be seen over time. These spikes were not observed in the CONTAM simulation. This discrepancy led us to experiment on the CO2 tank itself, eventually coming to the realization that the CO2 canister didn't emit gas at a constant rate.

Page updated

Report abuse