10/6/2024
-
4
min
Optimizing Chemical Usage in Water Plants Through Machine Learning at the Edge.
By implementing real-time optimized Machine Learning algorithms at each of its desalination plants, ACCIONA a global infrastructure operator, managed to minimize the use of reactive chemicals, eliminate associated regulatory penalties, and leverage an edge infrastructure to implement new predictive applications for water quality control. In this article, we explore the intricacies of the project.
Machine Learning in Distributed Water Plants
The volume of data handled in the water management infrastructures is expanding daily. To navigate the complexities and sheer scale of this data, infrastructure managers are increasingly turning to Edge infrastructures like Barbara, that enables real-time processing of edge data without risking the integrity of the operational technology (OT) network.
ACCIONA a global infrastructure operator, spent significant time and resources examining water samples in a laboratory to determine chemical concentrations. Due to the time it took to obtain these results, they were often outdated and unreliable. This resulted in additional costs related to chemical supply as well as possible regulatory penalties.
The Role of Machine Learning in Reagent Dosing Control
Reagent dosing control loops are crucial for targeting variable chemical compounds in water. Machine Learning algorithms offer a groundbreaking solution by enabling automatic fine - tuning of chemical dosage bearing in mind the variables of the environment of each facility plant.
What was Acciona looking for?
1. Virtual Chemical Predictions: The objective was to utilise algorithms capable of forecasting chemical compositions in water, a task typically achievable only through laboratory analysis. By consistently monitoring associated parameters, ACCIONA aimed to attain enhanced control over these crucial variables.
2. Cost Reduction: By optimizing the control loops for reagent dosing, the goal was to reduce the significant expenses associated with heavy chemical usage, and also achieving substantial savings on penalty payments.
3. Global Scalability: The company aimed to extend the solution across its 89 plants worldwide, ensuring consistent efficiency and operational excellence.
4. Efficient Data Capture from Diverse Sources: Efficient data ingestion from a variety of sensors, machines, and devices was crucial. Alongside, ensuring the secure storage and management of all data was paramount.
5. AI Model Deployment and Fine-Tuning: The strategy included seamless deployment of AI models, coupled with the capability to automatically refine these models based on environmental variables, ensuring adaptability and resilience.
6. Remote Monitoring and Maintenance: Centralized, remote monitoring and maintenance of the entire lifecycle of applications and devices was essential for streamlined operations and enhanced performance.
Barbara´s Edge AI Platform
With the explosion of connected devices and the need for real-time data processing, it is no longer practical to send all data to a centralized data centre. To achieve ACCIONA´s goals an Edge Infrastructure was needed. This was crucial for the following reasons:
1. Cutting down costs linked to data capture and transmission to the Cloud, particularly in situations requiring ongoing data collection.
2. Complying with stringent security and data privacy standards was critical to ACCIONA.
3. Accessing to real-time data processing capabilities.
Barbara's proposed solution included the Barbara OS, the software to deploy, orchestrate and maintain AI models very close to the data source, at the water plants and Barbara Panel, (The Edge Management and Orchestration tool), that allows the remote management, configurations and updates, granting complete governance over the lifecycle of edge nodes.
More precisely the solution involved:
- Connectivity and data processing with alarms in case of incidents.
- Running, updating and configuring “dockerized” applications remotely and securely.
- Lifecycle management of edge nodes / edge gateways (upgrades, reconfigurations, etc).
- A graphic interface to view measurements in real-time.
Main Challenges of the Project
The biggest challenge involved the fine-tuning of distributed models to accommodate new environmental variables. The llater, was essential for achieving optimal performance from a single model across various plants, each with its own distinct conditions such as temperature, operational state, and positioning.
Another major hurdle? Ensuring robust and secure connectivity, especially when working with the OPC-UA protocol and its complex authentication requirements.
Barbara´s technological solution enabled a seamless integration of Barbara Panel ( for the remote management of applications and edge nodes in the field) with Acciona´s water plant servers safeguarding data privacy and device security through encrypted storage and transmission.
Expected impact of Edge AI in numbers
250.000 USD per plant just in the first year.
An 800% increase in the speed of introducing new applications to any facility.
The deployments of new predictive model such as water turbidity model.
Want to learn more? Download Acciona´s story here
4
min
By implementing real-time optimized Machine Learning algorithms at each of its desalination plants, ACCIONA a global infrastructure operator, managed to minimize the use of reactive chemicals, eliminate associated regulatory penalties, and leverage an edge infrastructure to implement new predictive applications for water quality control. In this article, we explore the intricacies of the project.
Water Management
5
min
To operate more efficiently, water management operators face the challenge to monitor in a unified and comprehensive way all components that make up the OT and industrial automation architectures. The use of Edge Nodes allows connectivity with different equipment in each installation regardless of the various data reporting equipment. Discover how utility companies can access sophisticated intelligence to support water operators using Edge Computing.
Water Management
min
Although much of the water infrastructures in place are dozens of years old, in recent times their processes have changed radically thanks to a large extent, to digital technology. Edge computing is one of those technologies that will change the shape of the water industry due to the speed and reliability it provides when modernizing this type of infrastructure.
Water Management
