Describe what is the need of this project?
The problem we are tackling is the lack of effective action regarding the leakages in the water supply network of cities.
The need for this project is to minimize the water lost in the supply network due to internal and external factors. The capital of Mexico, Mexico City loses around 40% of its water in the distribution system due to leaks, while Monterrey city loses around 33%.
The problem is solved but not in an efficient way. Users report leakages that are visible, and the water supply company fixes the leaks. This is reactive action. Ideally, preventive strategic maintenance is applied to avoid water losses and decrease response time.
The scope is to solve the problem for the city of Monterrey using Artificial Intelligence (AI) to predict leaks before they happen, to ensure preventive maintenance, and make the prediction actionable by automatically order and ensure shipment of the parts that will be needed for the fix, before the leak even happens, to ensure decrease response time. This is expected to be achieved using (1) hardware, in the form of water flowmeters or pressure sensors, and (2) real and simulated data of network variables.
The company in charge of supplying water to the whole state, Agua y Drenaje de Monterrey, has shown interested in projects related to AI and digitalization. We have previously presented ideas to the board of directors.
Describe who is affected?
Two parties are affected:
- Citizens from any city. As water is an important natural resource, losing water means less water for people.
- Companies in charge of supplying water.
Our scope is:
- Citizens from Monterrey and Mexico City.
- Mexican company in charge of supplyng water: "Agua y Drenaje de Monterrey"
What are the causes?
(from most common to least):
- Natural process of wear
- Installation defects
- Huge, complex distribution systems that are hard to monitor effectively and all around, thus possibly slowing down the detection and/or the repair of leakages.
- Illegal invasion (clandestine takings of water)
- Natural disasters such as eathquakes
What is the evidence? Also who can you interview? What can you find out? What experiment can you run?
Ing. Florentino Ayala Vazquez . Projects director, operations and sanation
Start documenting your thoughts and ideas
- Godinez Madrigal, van der Zaag & van Cauwenbergh (2018) have identified poor maintenance to leakages and lack of monitoring and law enforcement as two of the main drivers of water crises all around Mexico. Additionally, they believe adding population and economic growth to these management failures will continue to widen the gap between water supply and demand. Actions need to be taken to prevent a critical shortage of quality water resources that lead to conflict. (https://www.proc-iahs.net/376/57/2018/piahs-376-57-2018.pdf)
- "The system of getting the water from there to here [...] is also a crazy feat, in part a consequence of the fact that the city, with a legacy of struggling government, has no large-scale operation for recycling wastewater or collecting rainwater [...] Mexico City now imports as much as 40 percent of its water from remote sources — then squanders more than 40 percent of what runs through its 8,000 miles of pipes because of leaks and pilfering." (Kimmelman, 2017) (https://nyti.ms/2lqgt2O)
- Gutierrez (2019) explores water issues in Mexico: "More than a physical water scarcity problem, Mexico City has a water distribution problem. [...] Mexico City’s water challenges are multi-dimensional and include the availability and security of the water itself dependent on aquifer health; the quality of its aging distribution infrastructure that is further stressed by subsidence; and the inability of formal systems to account for and properly serve informal settlements". The author also discusses that Mexico City still has work to do to achieve the SDG 2030 to achieve access to drinking water for all, emphasizing the need for accurate data about the population actually served by the water system. One of the recommendations the author makes is to fix leaking infrastructure caused by deteriorating pipes and illegal connections, allocating more funds to repairs and maintenance. (https://penniur.upenn.edu/uploads/media/02_Gutierrez.pdf)
What is The Big Idea? What is the value proposition?
With an inexpensive machine learning solution and the data from 3 sensors in the pipeline we can detect any kind of leakage in the water system, so that the company can go check and fix the problem in a timely way. Furthermore, we can predict when a leakage will happen to make prevention less expensive than actually fixing a leakage problem.
Image retrieved from
What is the mechanism of beneficial change?
What are the key metrics?
There are some metrics that we could use to validate our solution once it is implemented:
- We will be able to measure the number of liters of water lost every second, which in turn can be translated in to cash lost.
- Also, a metric that the solution could be tested on in the future is the time taken to detect leakages.
Who is most likely to be supportive?
The state agency of Monterrey has already supported projects that allow the state to get more water for their population. In the second semester of last year, a new water dam project was accepted.
Also, this is a topic of interest for the general population of Monterrey. Monterrey has had water scarcity issues in the last years, with "Day Zero" seeming close. Just in August 2019, Nuevo León (state where Monterrey is located) scored 4.44 out of 5 in water stress (Rodríguez, 2019). More fruitful water management would better ensure water supply for the population of Monterrey.
Key foes? Who is most likely to oppose?
In our case, the state agency and the government would be the most resistant to change. Since it is a decision made by a state agency, it will also need to be funded and approved by the state government. If it is not in the interest of the current political party any project would be unfeasible without the funds.
Water infrastructure is "mature" and well established in Monterrey. Although the project in mind would not interfere much with the existing network design, we still expect some opposition from the actors previously mentioned out of skepticism.
In addition to this, Mexican culture is not one of prevention, so it is not uncommon for action to be taken once a problem has to be solved. This means that, while the water infrastructure is flawed (losing around 20% of water in leakages), its issues will probably not be addressed until they become bigger (for example, due to water becoming scarce).
Image retrieved from https://adioma.com/icons/negotiation
What is the user experience?
There are two segments of water users:
Within the segments, each segment have different experiences when a leakeage happens, in regards to their needs:
- Providers receive complaints from consumers (their users) via phone, mobile app, or other media, when a visible leakage is detected. In response, the company sends a troop to fix the leakage. Real data has shown that around 90% of the leakages are fixed on time. Providers care about fast solution. For non-visible, non-detectable leakages, we have no information to understand how the company detects and fixes them.
- Consumers report leakages to the company when they are visible. Consumers care about having a basic need: water. But they also want a fast fix, since they are likely to have their water cut.
Who has to do what. to make it happen?
In order to achieve an efficient use of water, with no water loss due to leakage:
- Water providers needs to:
- Invest in new infrastructure
- Invest in AI solutions
- Invest in labor
- Invest in IT & Cloud services
- Local government needs to:
- Provide the capital needed
- Provide the required legal permits & collaboration
- Keep reporting leakages
- Team F
- Provide the AI solution to water providers
Who are the key partners to execute? Key partners to help others evaluate your value preposition?\
Key partners to execute:
- Water supplier company
- To execute, it would be ideal to have real historic data
- A company that provides water flowmeters or pressure sensors
- To execute, we need to have real hardware, and possibly execute lab tests
Key partners to evaluate preposition:
- Open17 Staff Team
- Experts in AI
- Key staff within the water company
What are the precipitating events?
Cities worldwide are closer to day zero: the point at which there is not enough water to meet the needs of its citizens.
One recent case is that before Covid-19 the city of Monterrey only had 14 months of water left, now, after Covid-19 the city has adjusted its estimates to only 10 months left
who else is in the field?
There are some players in this field:
- Private companies
- Usually offer hardware oriented solutions, hence expensive, slow, but accurate
- We have identified many researchers working with leakage solutions. They tend to be software solutions - using AI
What's wrong? Missing? Not working?
Our solution uses open source methodologies, tools and services, that allow it to be cheaper, which is the main reason the current solutions are not used by the company, they are expensive.
Our solution is easily scalable as all the sensors needed are already used by the company. (We would have to check the connectivity of the sensors)
Physical and intellectual resources needed (besides financial resources)
Water Density, Water Flow & Water Pressure sensors. (There is a possibility the company already has the sensors that would send the needed information for our solution to work)
Informatics essentials (Server, virtual machines, etc.)
Domain knowledge on alarms and flags used in the company.
Next steps? Pilots?
Our strategy is detailed in the following steps:
- Gather real data regarding:
- Reported leakages & time needed to repair them
- Water loss (volume/time or volume)
- Water Pipe Network (Distance, pipe material, start and end points, pipe junctions, etc)
- If needed, simulate data using software
- Establish mathematical approach & equations needed to create mechanic fluids & momentum models
- Once established, translate maths into programming and fit the model to a AI model (such as Random Forest, LSTM, etc)
- Export software model into a real scenario. Prove our model in a scalated water network simulation. This will be done in a lab using real sensors and pipes.
- Connect software into a dashboard so it's user friendly
Cost structure? Financial Sustainability? Revenue streams?
Costs are segmented as follows:
- Web hosting server
- External APIs (if used)
- R&D costs
- Materials costs when lab testing (sensors, pipes, etc)
- Technical support
Once the project is concluded, we have 2 options:
- Sell the full project to "Agua y Drenaje de Monterrey", the company in charge of supplying water to the state
- Sell monthly subscription and expand the software to be used to other water companies outside Monterrey or even Mexico.
To further prove that there will be a profit, we need detailed costs data that as of today we do not have.
How might this go wrong? How might the problem evolve? What are the legal, cultural and other impediments?
There are several potential risks:
- Project ends up not being very accurate, therefore not useful
- Since water companies are regulated by local government, water companies may not invest due to limited budget and lots of local governmet approvals, which is very likely not to be approved
- Water companies could reach out to established and recognized companies such as Atos consulting, Accenture, McKinsey and they could deliver our project in a shorter time and more professional way
- Since our potential market is highly limited (only water companies), the project could go bust very easily if water companies decide to cut ties
How will i promote adoption?
Promotion needs to be specialized (direct sales) since our potential market is very limited. We could expand the project to several other potential customers such as manufacturing companies or even for residential houses, however, as of today, the scope is only limited to water companies.
¡Viva Mexico!, Ricardo & Arturo met in a water challenge in 2019, Alejandra joined the team in 2020 after the 3 of us took classes on Artificial Intelligence every Saturday with an NGO. We make a video call for this project almost every day! It also keeps us sane with this social distancing government requirement.