KM and the environment: Water management uses analytics, big data and collaboration to handle complexity.
Weather and retailing
When IBM purchased The Weather Company in 2015, its role in IBM’s portfolio was unclear to many observers. By June 2016, IBM had announced a “hyperlocal” forecast product for predicting weather within a mile of a specified location. Aimed at retailers who wanted to better predict customer needs and behaviors that might be weather-related, it was followed in April 2017 by a product developed by The Weather Company in collaboration with LiveRamp to use real-time weather data to drive personalized digital marketing programs.
Retailers have long been using weather forecasts to make decisions about what products to stock during various weather events. Many stock extra batteries, flashlights and bottled water when flooding and power interruptions are expected. Walmart, for example, has emergency operations managers who prepare for weather-related events both in terms of their merchandise and facility management.
The use of highly granularized environmental big data for marketing campaigns is relatively new but is likely to take hold as another method of personalization. In addition, it may also be used to prove or disprove retailers’ beliefs in the extent to which weather affects their sales, which has been an area of dispute. In dramatic situations such as flooding, the impact can be clear, but the effect of more subtle fluctuations is not well understood at present.
Water utilities use KM for rapid response, knowledge retention
Water utilities are facing a complex mix of challenges, including downward pressure on budgets, an aging infrastructure and a workforce that is entering retirement age, taking with it a large corpus of institutional knowledge. Utilities are dealing with those issues with a variety of strategies, including automating processes that were previously handled manually and using various techniques for knowledge capture.
In Cincinnati, the municipal utility is meeting those challenges through increased use of real-time analytics and optimization. The department evaluates the weather forecast four hours in advance, examines radar data in the area and if a heavy rainfall is expected, adjusts the collection system to make room for the influx. “The sensor and meteorological data can be voluminous, requiring sophisticated analytics and models to interpret it,” says Raja Kadiyala, VP, senior technology fellow, global practice director for intelligent systems at CH2M, a professional services firm that specializes in complex infrastructure and natural resource problems.
A company specialty is the use of real-time analytics and optimization to alleviate water management issues. “Historically the information required for decisions might have been in someone’s head,” Kadiyala says, “but now real-time algorithms are used that consider such factors as how much rain is expected and what the impact on the watershed will be.”
Hydraulic models are available from a number of sources including the U.S. Environmental Protection Agency, which provides open source versions. Commercial models are also available. “Most large utilities have a hydraulic model,” Kadiyala explains. “The model, in conjunction with optimization and analytics, can be used to maximize water quality, minimize energy consumption and respond to wet weather events. The combination of these technical components can do a much better job of optimizing across multiple constraints than even experienced humans because of the ability to ingest all the data and analyze across a much larger set of scenarios.”
Operation of the drinking water system, in particular water quality, entails similar monitoring. Online sensors can be used to keep track of water conditions, and the utilities have response plans to deal with the anomalous water quality.
Focusing efforts to reduce impact
Because it is not possible to have a sensor at all locations, the human factor also enters in as a data source. An example of that was the ability of a large U.S. city to track and remediate algae that had made it into the distribution system. By monitoring water quality-related customer calls coming into its 311 system, the city could discover where the algae were and perform localized flushing to reduce the impact.
“The city used real-time geospatial clustering of the calls coming in to follow the progress of the bloom, which was causing a metallic taste in the water,” Kadiyala says. “That information allowed the water department to know where to focus flushing efforts to minimize impact.”
Water is a resource that is often taken for granted. “Generally, people don’t think too much about their water until there is an issue,” Kadiyala adds. “But when there is, they want it fixed as soon as possible.” The combination of physical data and input from customers provides a more comprehensive view that supports rapid interpretation and response.
Like many organizations, water utilities have a workforce that is retiring at record rates. That segment of the workforce also has extensive institutional knowledge, some of which was acquired before electronic documentation or automation was the norm. Utilities, worried about knowledge loss, are using some well-recognized techniques for capturing and converting tacit knowledge into explicit (and therefore shareable) knowledge.
“In many utility systems, retiring workers are interviewed to extract their decades of knowledge on operating the system,” Kadiyala says. “These interviews are captured on video and then indexed according to several parameters to make access to the knowledge more accessible. More recently, utilities are running the interviews through speech recognition software to make them more searchable.”
Water management is a global issue and will require a combination of approaches to ensure steady progress. Scientific and technological advances, along with improved collaboration and decision support, will help determine the path forward.