Fixing leaks and the Internet of Things

In a previous post, I wrote about when it makes sense the fix leaks (or not) depending on local conditions. Unstated but looming large in that post are some important questions (non-exhaustive list) : 

• where are the leaks?
• how big is each leak?
• what is each leaks' financial cost?
• what is each leak's environmental cost?

I'll now discuss briefly the issue of finding leaks. Traditionally, we have looked for leaks using a combination of sound, gas, and modeling methods. Today, connected objects offer new opportunities.

Sound and gas methods basically involve surveying the water network to listen for leaks or detect leaking tracer gas. Fixed or mobile listening devices can help triangulate a leaks' location, and tracer gas will leak (or not) downstream of where it has been injected - so the search can be optimized. These methods either reveal little information about the size of the leak, or are limited in physical reach. 

Hydraulic modeling makes it possible to focus the search for leaks. However, hydraulic models need field data to be reliable. Unfortunately, it is very difficult to take an instantaneous picture of a water network unless one has a significant number of sensors that are well-calibrated, functioning, and transmitting all at once. As a result, hydraulic models, while sometimes very precise, do not provide instantaneous information about the location and volume of leaks.

Enter a number of companies (i.e.: Visenti) that have combined the power of connected sensors with Big Data to monitor water networks in real-time to identify leaks and bursts as they happen. Connected water networks can update centralized control centers like SAUR's CPO to make it possible to prioritize leaks and react accordingly. We will shortly see the proliferation of connected objects along water networks, from sensors to meters to valves, in step with the capital campaigns of water utilities. 

Water network sensors must be placed at critical points in physically extensive systems (100s to 1000s of km), often in wet, buried, or even corrosive environments. Changing batteries is expensive, and power is not always available. Sensors therefore must be highly energy efficient and resilient, and their communication protocols must be able to overcome specific challenges. Because utilities are often strapped for cash, a major challenge is establishing an appropriate price point and clearly spelling out the ROI of these investments.   

That said, the Internet of Things, and in particular the Internet of Water Networks, is clearly the way of the future for leakage management - and other aspects of water management too. It will provide that elusive instantaneous picture of what is happening on the network, and allow engineers and technicians to increase their efficiency in reducing leakage rates. 

In a future post, I'll discuss why it's not enough to deploy sensors: utilities have to learn how to use them and transform themselves correspondingly. 

Should you reduce that leakage rate?

Recent articles in the French press have reported that 20% of all drinking water is lost to leakage in France. Although it hides significant local variability, this large a number warrants attention, in particular as France misguidedly set a 15% national target in  2010 (as part of Grenelle II). 

Water is the only industrial product intended for human consumption that is delivered 24/7/365 to all homes in France (and in developing countries in general). Drinking water cannot be economically shipped over large distances, and so is most often consumed close to its production site. For a given city or area, the cost of producing and delivering water depends significantly on geography and topography, and on the quality of available water sources - it is very much a local variable.  

Drinking water production and consumption represents (on average) a small percentage of the total water consumed in any given area. Most water is used for industry (including energy) and agriculture. However, this average can be different in places where there is neither industry nor agriculture, and where drinking water outtakes represent a significant proportion of water extraction from the environment. 

Repairing leaks is a significant expense for many water utilities. Finding leaks can be tricky and repairing them typically involves digging out old pipes - and usually replacing them. Instead of applying a uniform target leakage rate, it makes sense to optimize decision-making to local conditions. We see from above that there are two main 'costs' to consider: 

• The financial cost of drinking water production and delivery is lower :  
• in a densely populated region 
• in a region with nearby, plentiful, and clean water supplies
• in a region with favorable topography (water flows downhill for free !) 
• The environmental cost of drinking water production and delivery is lower: 
• in a region where it represents a small percentage of the total water extracted from the environment. (waste water effluent doesn't count because leaked water is 'clean'.)
• in a region where it requires less treatment and/or pumping (environmental impact of shipping and using chemicals and energy).

To decide whether or not to pursue an aggressive leak reduction strategy, we can use a simple heuristic:

	Low financial cost	High financial cost
Low environmental cost	Fix only biggest leaks - tolerate relatively high leakage rate	Prioritize leaks according to financial costs
High environmental cost	Prioritize leaks according to environmental costs	Aim for low leakage rate - prioritize with financial & environmental costs

This heuristic is better than the traditional approach, which is to consider the diminishing marginal value of leak reduction (the more leaks you fix, the smaller the marginal value of each fix), because it considers the value of the environmental impact of water leak as well as their financial impacts. It is up to each regulatory body and utility to decide together on the applicable thresholds between 'low' and 'high' costs. Most importantly, it should be done at the most local level possible to ensure that specific conditions can be taken into consideration in setting the most appropriate target leakage rate.