Home

Welcome to the GenuFlow Project Team Homepage!

This project was sponsored by 

Jose Doval, Dan Kuehler, and Kirt Winter.

Overview

    GenuFlow is a San Diego startup committed to saving water in irrigation systems of all sizes by creating an inexpensive, non-invasive device to detect leaks. Detecting leaks quickly after they form allows them to be fixed before they waste costly volumes of water. This sponsored project consisted of researching flow characteristics of PVC sprinkler systems, developing methodology for sensor-based leak detection, and constructing a testbed for experimentation and demonstration. Code was developed to perform statistical analysis on the acoustic energy data, and successfully identified ruptures. The code created, data obtained, and testbed built will be used to further develop GenuFlow's leak detection method and product.

What Problem are We Solving?

       As humanity marches further into the future, engineers need to help protect the Earth's natural resources for the generations that follow to use and enjoy. However, the issue of water scarcity is rising as a major problem that needs to be addressed. Despite the fact that approximately 70% of the Earth's surface is covered by water, less than 1% of that water is available for human use. While one leak may not seem to pose much of a risk, collectively multiple leaks have caused losses of over 1 trillion gallons of fresh water just through domestic leaks in the United States annually. Industrial and agricultural sectors also consume vast amounts of water resources each year, so one could imagine the losses leaks inflict on the entirety of the United States' economical and ecological resources.

Image credits: The Illustrated Atlas of Wildlife and WaterSense EPA

Methodology

       To tackle this major source of preventable water loss, we strived to develop a methodology to detect leaks by analyzing pipe flow data. We conducted experiments in the backyard of a project member's yard, in his front yard, and on the testbed that we created for Genuflow. In the backyard, the Teensy 3.2 microcontroller was used to collect acoustic energy data, which was processed by a computer connected via USB. The data was obtained by an acoustic sensor that was contained in a plastic box and attached to the pipe downstream from the solenoid valve.  The same electrical setup was used in the frontyard and in the testbed.  

    Eventually for the testbed trials, a Raspberry Pi setup was used with a pressure sensor and flow meter, in addition to the acoustic sensor.  The acoustic sensor, pressure sensor, and flow meter were attached sequentially, just downstream from the solenoid valve.  The microcontroller allowed us to increase the sampling rate and collect data wirelessly.  In all our trials, leaks were simulated by replacing a sprinkler with a rise and ball valve, and turning the valve at variable amounts to release varying amounts of water.  A rupture was simulated by replacing a sprinkler with a riser.

One of our sponsors, Dan Kuehler adjusts a sprinkler head in our backyard setup.  There is one sprinkler at the corner of the "L" shape and one sprinkler at the end of the "L" shape.

A microcontroller and processing board are connected to an acoustic sensor attached downstream from the solenoid valve on the riser.

This is our preliminary testbed.  Blue tarp protects the wooden surface until we apply polyurethane later to water-proof it.  The acoustic sensor is attached downstream from the in-line solenoid valve.

Final Product and Results

This is our final testbed. Inside are PVC pipes on which sensor tests were conducted.  

    The project consisted of two components: advancing and developing new elements for a sensor based method of leak detection in small irrigation systems, and building a testbed equipped with pipes, valves, and sprinklers.  It consists of a plywood table with drainage holes, and transparent housing made of vinyl sides and a polycarbonate top.  The housing is framed by 80/20 T-slot aluminum bars. Inside the testbed is 100 ft of 3/4 in. PVC piping and a solenoid valve on which to conduct tests with sensors.  One coat of polyurethane was applied to the testbed surface to water-proof the wood. Overall, the table surface is 4 ft. x 8 ft. and the housing is 2 ft. high.  

CAD model of testbed.

CAD model of piping configuration.  Pipe unions allow interchangeability of pipe sections, and dozens of possible configurations can be created by switching ball valves.

Several conclusions were made from data obtained in all our different trials. 

Backyard Trench Results

• • Leaks detected by using multiple averages to identify changes in acoustic response

  • The presence or absence of dirt did not influence characteristic changes

  • Two Acoustic Signatures

• 1. Initial Transients: from opening of the valve until the response settles

  2. Steady-State: consistent, stable signature

Testbed Results

• • Characteristic response for different setups determined by taking multiple runs and averaging the data

• Ruptures detected in systems using the full 30 m  (100 ft) or more with zones utilizing 8-10 sprinklers

This data was collected from the backyard sprinkler system, which contained one sprinkler and one rupture.

Above is a flow diagram representing the prototype leak detection technique.  This method was created based off the patterns we observed in the data.  

Conclusion

    As water security becomes a growing concern in times of climate change, it has become more important to manage how efficiently water is used.  The startup company GenuFlow strives to invent a non intrusive and inexpensive sensor-based device to detect leaks in small and large scale irrigation systems.  They assigned a project that had two components: advancing and developing new elements for a sensor based method of leak detection, and building a testbed for testing and demonstration.  An acoustic sensor was used to detect energy of normal and ruptured sprinkler systems in the backyard and front yard of a home, and on the piping of the testbed we created. Different trials were completed by shutting on and off a solenoid valve, and collecting data while the water flows through pipes. Leaks could be detected by comparing trials using statistical methods. The GenuFlow sponsors will refer to the our prototype leak detection method, our data obtained, and the testbed built to further develop their leak detection method and future product.

Thank you to our GenuFlow sponsors, Jose Doval, Dan Kuehler, and Kirt Winter.

Special thanks to our professors Dr. Jerry Tustaniwskyj and Dr. Jack Silberman, and our teaching assistant Pedro Franco who gave thoughtful perspectives throughout the project.

Additional thanks to Steve Roberts, Chris Cassidy, Tom Chalfant, Ian Richardson, and Gregory Specht for their invaluable insights and consultation.

News and Updates

The final draft of our team's poster for the 2017 UCSD has been completed! You can find our poster in pdf format in the Reports general page under the Current Poster folder, as well as the Docs and Presentations subpage by clicking on their names. Please shoot an email our way if you encounter any difficulties and we would be happy to assist!

- June 11, 2017

The final draft of our team's report has been posted in the Reports general page. The report still requires additional tidying of labels, listing, and updating pictures which are being addressed and may be updated over the course of the weekend. Any additional edits outside those criteria will be recorded in a separate errata text alongside updated final drafts.

- June 9, 2017

The first ratification of our project manifesto's can be found here. A more refined version will be posted weekly, stayed tuned!

-April 14, 2017