2025 - AVEVA World - San Francisco - Process Industries (Chemicals, MMM, Pulp/Paper)
Fueling Efficiency: Nalco's AI-Driven Solutions for Smarter Boiler Operations
This presentation showcases NALCO's digital transformation of coal-fired boilers using AVEVA System Platform integrated with AI/ML algorithms. The focus is on optimizing combustion efficiency, reducing specific coal consumption, predicting failures, and providing automated root cause analysis (RCA). Key advancements include real-time anomaly detection, historical data-driven insights, and proactive recommendations to enhance boiler performance. By addressing challenges such as data integration and unplanned outages, this approach improves efficiency, minimizes emissions, reduces downtime, and lowers fuel consumption, setting benchmarks for operational reliability and sustainability. This presentation will be delivered by CereBulb on behalf of Nalco.
Industry
Mining Metals and Minerals
Company
CEREBULB
Speaker
Divyesh Dhorajiya
Session Code
SESS-153
Transcript
Hello, everyone. Thank you, first of all, to get time to come here and look at our NALCO's presentation. It's a wonderful implementation, that initiative NALCO had taken. To introduce myself, I think Glenn did a perfect job, so I am supporting global team for business development. So I operate out of India office, but I regularly travel across the globe.
With me, during the slides, will see Santosh Kumar presenting his testimonies for the solution that we had deployed over there. Unfortunately, he couldn't be visiting San Francisco because of his other commitment, but I would like to see you and go through our presentation, surely.
So first of all, let me introduce Cerebal is a global company giving digitalization and automation solutions across the globe. Our mission is to transform data into actionable, strategic insights, so diving through AI advancements and other initiatives so that customers get value out of it, and generates large amount of AI.
Our offices are located in three different locations. We are headquartered in USA. We have one office in India, and we do have our office in Australia, but we work across the globe. I would say we have a team of one hundred and forty people growing, but they are all trained in different technology portfolios, and we cover almost seventy five plus technologies across the globe. We are SSP partners of AVEVA in all the three regions, and we do provide solutions to the companies located in other regions as well.
So this is a short brief introduction on Nalco, which they have presented.
I would like to go through Welcome to Nalco, India's pioneer in aluminum and power production.
Since nineteen eighty one, this Navratna company has been setting global benchmarks in sustainable mining, refining, and power generation.
Join us as we showcase how our rich legacy embraces modern digitization.
Nalco is more than a manufacturing giant. It's a visionary enterprise driving sustainable growth. As one of India's largest integrated aluminum producers, we manage the entire value chain from bauxite mining to aluminum smelting and power generation. With a strong export presence in forty plus countries, NALCO stands for innovation and reliability.
Beyond industry leadership, our commitment to energy efficiency and environmental stewardship makes us a trusted global name. This strong legacy fuels our bold vision for digital transformation.
Embracing Industry four point zero, NALCO is revolutionizing operations through digitalization.
By leveraging advanced analytics, AI, and real time data, we enhance efficiency, optimize fuel consumption, and enable proactive maintenance. Our digital journey strengthens Nalco's commitment to innovation and sustainability.
At Nalco's dem and Jodi side, digital transformation is in action. This pioneering project integrates AI and machine learning to optimize boiler operations through real time monitoring.
Yeah. So that was the Nalco's presentation. As beautifully highlighted in the video, they were very keen on addressing and introducing AI in their operations during their operations, and found that they heavily rely on the CPPs that power their operations. And they found that it is key for them to monitor, look after their assets on the power generation side.
So they were looking after to improve what all can be done, and how AI can be introduced in power generation operations, so that their operations become much smoother, there are less interruptions, and then they came up with a list of challenges that they would like to address during the implementation. So I will just highlight the challenges that we addressed during the use case. So first was data silos. Nalco being an operational side, and as I mentioned, it's global operations, but it has a lot many factories and plants across India, and the landscape of that is so large that they have multiple datas residing in multiple data sources, and then it's hampering them to have a common platform where they can put everything together to power up a solution or do further analysis.
Thus, they were looking for a product that can help them to combine all their data into a single source so that they can do further analysis. The second was limiting predictive insights. As mentioned earlier, all the data was residing in different platforms for all the operation guys looking for their own operations. Thus there was no clear visibility for them to where to pinpoint what was the problems, what were the issues that they can solve proactively instead of reacting to it, so that they can take corrective actions based on data driven decisions.
The third was unplanned boiler. As a result of having no monitoring system, or a monitoring system just for all operational guys to themselves, and no common platform, there were so many unplanned downtimes hampering the operations across the plant, and then that were proving to be very critical for them to pay penalties, because in India, if you consume energy, unplanned energy, from the grid, you have to pay heavy penalties. To address them, they had to first bring everything on a single platform, and then monitor, predict, and make sure that all their operations are running smoothly.
Sorry, yeah. And the fourth one was heat balance and incompetent. So even if the boiler is running at its full capacity, it needs to be having complete combustion. Losing combustion, and then incomplete combustion can make the operations use higher amount of fuel, have a lot more impact on the utilities and the other functioning areas around it that can hamper the, I would say, consumption of, let's say, oxygen that is fed to the boiler, coal consumption, so all of that needs to be refined and optimized.
That was their fourth target. Fifth one was reactive maintenance. Their maintenance team, there had limited visibility, and as it is a very critical asset, as mentioned earlier, it was important for them to resolve issue as soon as possible, but whenever they had an issue, they had limited visibility. What was the reason?
They had to go through different different types of people, different different types of operational guys communicating to them, and then coming to solve a problem. And the sixth one was lack of real time monitoring.
As in general, all the maintenance guys always waited for anomaly to occur. They had no real time visibility. What's going to happen next in the boiler? What is the current coal consumption, all of that issues, they addressed to us, and at Cerebulb, we partnered with them for transforming their boiler operations and trying to introduce the AIML models.
So while going to the plant, we figured out their latest boiler made from pail, and it was modeled VU-forty. It was their latest model with highest capacity, producing two hundred tons of steam per hour, and it was their most critical asset currently being used for the operations of the plant. So we targeted that boiler first, and then deployed it across the other boilers.
Now we have combined with them, set with them, we discussed all that we could do in a boiler, and then drilled down with including the challenges and everything, we drilled down to six objectives that I would like Santosh to speak upon.
Hi, everyone. I am Santos Kumar, AGM e n r from National Aluminum Company Limited, Nalco, Drawajuri. Nalco is situated in India, and it is the largest alumina put in Asia.
For alumina making process, we require a steam for heating purpose, and these streams are generated through wire pliers. For this purpose, we have five number polyboilers hardening at a time, and one boiler is in standby emission. So imagine a power event where every parameter is optimized in a real time. Downtime is minimized, and fuel efficiency should be at its peak value.
This is exactly what we have achieved with AI driven boiler, error prone data entry, and enabled real time AI powered logging. Our system detects deviations in critical parameters like heat balance and combustion efficiency, ensuring consistent performance. Automated route points to the underlying issues drops while real time corrective guidance helps operators make optimal adjustments. Dynamic optimization instructions and predictive analytics further enhance combustion control, maintaining stability and efficiency.
Our solution Yeah.
So that was overall six objectives that we drilled upon with them, sitting them. We did multiple workshops with all the operational guys working in the plant. He was the ENI head, so he had access to multiple systems across the operation line, and then we came down to a platform. We sat with them.
We decided upon what should be the architecture to address all the challenges that they were facing, and then we finalized the system architecture. I would just quickly run through it. So they had multiple systems, but we could introduce a single OPC server there, which was communicating through all the operational parameters coming in the system. So there were SCADA systems, there were a few other softwares, all communicating to a single OPC server, and there were many manual entries, which are very critical for them, which they used to capture, but it was not being captured by any particular system, and all that data was key because all of them were control parameters, but could have been captured and then used for analytics.
So we made sure that all of that manual entries are being kept in the system, and then being processed with the models. And in the heart of the solution, we kept AVEVA system platform. As it was more reliable, it was highly giving us quality data, we were not worried about the data part as soon as the AVEVA system platform was installed. And then in these, besides, we just installed our own AIML server with Python coding.
We deployed multiple models, looking at their operations, understanding their whole process, and then it was again communicating to system platform for visualization. So data part, capturing part, all was taken care by AVEVA system platform, and the rest modeling part was taken care by our AML server.
So if I drill down into what were the building blocks of the project, I would say that first and foremost thing that we had to focus on was collecting all the data in a single platform. But they were also having gold mine of bad data for at least past two to three years, and which had to be considered as the more good data we have available for our AVEVA model, that would be giving us more accurate results. And that's why we made sure that we collect all the past data, as well as centralize all the different systems in a single platform. The second was providing advanced analytics, using calculating multiple losses, combustion calculations, and we made sure that everything is being made or looked after as they want from the plant side.
So we took multiple inputs from the plant side. Our team had to sit there, do multiple workshops again, to just come up with all the models. Then third was understanding the deviation. So whenever there were deviations in the product sorry, in the model and in the solution, then we would go back, understand what was the reason that our model has deviations from the operations that are going on, so that we can improve upon them.
And the fourth one was taking AI driven insights. So once we were confident on the model, giving them good prescriptions, predictions, then our change management part came in. At Cerebral, we again went to plant, we sat there with them, so during this whole process, you see multiple workshops have been done with them, so that they all can feel involved in it.
It was more of their inputs that had made solution, and that could help them, so it was not being pushed to them as a solution. It was built together, and they were really happy afterwards to use that solution. Fifth was real time visualization. So earlier I mentioned all the operators had to report to someone if they want to see any deviations in the projects, or because all the datas were available in different systems to different operators.
So they had to rush to someone for getting their valuable insights or anything. But now we made sure that each guy in the process has their own dashboards, what they think is important to them, but they also have access to other data which was not there earlier. And sixth was comprehensive reporting. So as new step, as an improvement, they had to view out all the reports to the central sustainability projects and other projects so that they can justify all the savings they're doing on the fuel side.
They can justify what were the consumption against the production, and all that reporting was automated through the solution.
On the AML part, if I drill down to how we approach and modeling the AML solution, it was like collecting raw data. So it was taken care by AVEVA system platform. We just had to make sure all the systems are communicating. Our team, OT team, went there.
They checked all the data sources. Are they communicating perfectly? Is there any bad data? Is there any?
And made sure that all the systems are talking to a single system. Then on data preprocessing game. So multiple times, people have always mentioned garbage in, garbage out. So we had to make sure that all the data that we feed in, because it always involved so many manual entries that I had mentioned, and we had to make sure that all the manual entries are not being wrong.
Like, it's just a matter of pressing wrong zeros Instead of pressing zero, eight, it might be pressed eight, zero. And then whole of the operation, whole of the modeling can go wrong. So all that data preprocessing and rules around it were established. And then on, model building was done.
So I would say always, model building is not the difficult part. It is the preprocessing part and model validation. So as soon as we were Okay with model preprocessing, all the data was being correctly measured, we had prepared a model for different conditions. So we observed that India having four climates, all the climates have different operational impact on the generation side, because they store coal.
In a rainy season, their coal might be facing issues with the moisture level, and due to that, all types of different possibilities were considered, and different different AI agents and models were created in that server, so that it can give them more and more accurate results. And then model validation was done. As I told you, our guys said that all they interacted with them workshops. They'd gone through all the anomalies.
What was model predicting? What was model prescription? Are we achieving it or not? And once everything was done, we deployed that model.
We trained their guys where to look, where to go, what to do, and everything. And then reporting part, we used SSL, RS modeling reporting as the tool where they can go even download customized report. So there was a very good thing that came up was the ED sitting in their office wanted to know what was the energy consumption today against the production, because as I told you, India is changing. The grid has become more strict.
Can't just use up all the energy that you have not planned for. So all that reporting he needed to justify why they are facing so many penalties, and I would be happy to say that just by reporting, he was able to pinpoint that these operations need to be streamlined so that we reduce our penalties each day. And then lastly, it was visualization. So we wanted to make sure that whatever UI we give to them, it's built in system platform, but it's not just given to them as a Scala screens or anything.
We wanted to make sure that they have whatever insights they feel and whatever trends, graphs, on AVEVA. So we made them each dashboards based on their level. So ED has its own dashboard there, and operational guy has its own dashboard, all with rule based dashboards.
Yeah. So this is a demo of all the things going on behind and multiple dashboards that we have gone through, which a customer himself is talking about.
Nissan integrates advanced AI capabilities to the one mission life, pilot operations.
Air powered heat wireless monitoring tracks real time variations analyzing enthalpy changes and temperature refers to alert operators about energy losses in real time.
Advanced AI capabilities to the one dish knife, boiler operations.
Air powered heat wireless monitoring tracks real time variations analyzing enthalpy changes, and temperature difference to alert operators about energy losses in real time, deviating from the normal operating procedures.
Traditional manual logging has been replaced with an AI driven digital system that updates twelve properties in a real time, improving accuracy and responsiveness. Troubleshooting, which once took hours, is now accelerated through AI driven root cause analysis, which is data and process dependencies to identify failures swiftly.
Combustion optimization is continuously improved AI, which provides recommendations for air to fuel ratio, oxygen level, emission controls, ensuring complete combustion while minimizing waste. Real time dashboards offer a comprehensive overview of boiler efficiency, fuel consumption, and airflow ratios, empowering operators to make informed adjustments.
Equipment health is monitored through temperature distribution tracking in critical components like superheaters and wire drums preventing overheating and extending life span. Our systems including hands, pongs, and mutes are also monitored in real time to maintain operational stability. The impact of this AI driven approach is transformative.
Wallet efficiency is significant improvement while fuel consumption is reduced, leading to substantial cost saving.
Maintenance expenses decreases as predictive analytics prevent unexpected failures and unplanned downtime is drastically reduced. Environmental benefits include lowering results, contributing to more sustainable operations.
Pure and optimize combustion further enhance reliability and cost effectiveness. The workforce benefits from reduced manual tasks, allowing personnel to focus on a strategic decision making rather than routine monitoring.
In conclusion, our AI powered Solisar redefines boiler operations by combining real time monitoring, predictive insights, and automotive optimization.
It has also included the thermodynamics laws in which we can optimize the wireless performance.
By enhancing embracing AI, industry can achieve a smarter, more reliable, and environmentally conscious wireless performance. Thank you.
So looking at the screens, you might have seen there were multiple dashboards, but each showing different parameters. I would just also like to sum up how the solution is like, if the operator wants something as an output, they can always put the inputs to system, like saying, we want to generate this much amount of steel, and it will give them all kinds of set points that they have to put in the system to achieve that. But along with it, at the same time, the solution also gives them predictions of whatever the the set points. Like if the input in the oxygen in the coal is not at the optimal level, and the generation is not going to be achieved, they will always get the inputs from the system saying that you need to change all of them. That was then highest level use case that I would say was achieved, along with the other individual use cases that we did. And this is the success metrics of what Nalco found here, and I would like Santosh to again speak about it.
Hi.
The challenges that we are facing in our existing system doesn't have any intelligence. It controls, it gives an alarm, and it is not predicting any failure or suggesting any change so that we can optimize our system. The data are scattered in different forms, limiting real time monitoring and decision making, while restricted remote access reduces operational visibility.
Operators are mainly relying on the reactive maintenance, leading to frequent breakdowns, increased downtime, and high maintenance cost.
And also inefficient combustion and uncontrolled heat losses due to various influencing parameters.
It is resulting in excessive fuel consumption, high recycling, reduced water resistance.
So we have opted an AI and MLB system, which is optimizing through first law of thermodynamics as well as the machine learning process and gives them solution, gives them some parameters which are helping the desk engineers with proper operation of the boiler. This system integrates with AVEVA with Python based models so that it enables real time monitoring, predictive analytics, automated root cause analysis, and KPI driver decision making.
Through by implementing this solution, we have optimized enhanced water efficiency and reduced coal consumption.
Another reason benefit that we are achieving is that it help it is helping in minimizing unplanned downtime and lower maintenance cost, ensuring more reliable operations.
It is also helping to reduce the emission gases so that we can achieve sustainable industrial practices.
