Edge Computing: Automatic fine-tuning of predictive models in water plants

• Platform as a Service (PaaS) - Edge Computing Platforms

• Oil & Gas
• Utilities

• Maintenance

• Edge Computing & Edge Intelligence

• Software Design & Engineering Services

ACCIONA spent significant time and resources manually testing 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 Due to the time it took to obtain these results, they were often outdated and unreliable.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 and regulatory penalties.

By implementing real-time optimized Machine Learning control algorithms at each of its desalination plants, ACCIONA was able to minimize the use of reactive chemicals, eliminate associated regulatory penalties, and provide an efficient edge infrastructure to implement new applications for predictive maintenance, energy efficiency, sensing, optimization or reinforcement learning.

ACCIONA

ACCIONA is a global group that develops and manages sustainable solutions for infrastructure, including water, concessions, construction, services, and renewable energy. The company plays a leading role in the water treatment sector, participating in designing, constructing, and operating various water treatment plants. These include drinking water treatment plants, water purification plants, wastewater treatment plants, reverse osmosis desalination plants, and tertiary treatment plants for water reuse. ACCIONA places special emphasis on providing services to cities.

ACCIONA aimed to achieve the following objectives:

1. Precise control of the variables to be monitored: The algorithm had the ability to predict the values of chemical elements in the water, which could only be detected by laboratory analysis. By continuously monitoring correlated parameters, ACCIONA aimed to achieve more precise control of these variables.
2. Cost reduction by optimizing reagent dosing control loops ACCIONA aimed to minimize the costs of heavy chemical usage, which would result in savings on penalty payments.
3. Scale the solution globally to all its plants.

In order to achieve these objectives, ACCIONA needed a tool that would:

1. Efficiently ingest data from different sensors, machines, and devices.
2. Securely store and manage all their data.
3. Easily deploy their AI models.
4. Automatically adjust their algorithms with environment variables.
5. Remote and centralized monitoring and maintenance of the complete lifecycle of applications and devices. 

Barbara's Industrial Edge AI Platform was instrumental in deploying, monitoring, and orchestrating algorithms in real time, while complying with the highest cybersecurity standard (IEC62443-4-2 safety level 1 standard).

• [Cost Reduction - Operation]
  
  1. Achieved 250.000 USD savings per plant only in the first year. 
  2. Expected to save more than $70 million in the first 3 years across its 89 desalination plants.

   

  
   

• [Process Optimization - Real Time Alerts]
  
  1. Remotely predict and monitor in real-time the water component values in all their facilities centrally.
  2. Overcome fragmentation of equipment and heterogeneity of plant environments.
• [Financial Growth - Market Opportunities]
  
  1. Incorporate new applications and models for predictive maintenance,  energy efficiency, sensing, and reinforcement learning to optimize other plant operations.
  2. Accelerate go-to-market of new models by 800%, thanks to Barbara's remote and automatic management of distributed application deployment.