Q&A: AI in the Cement Industry

After many years of talk, Articial Intelligence is coming into its own as a powerful solution to cement’s many challenges.

First published in Cement Americas - January 2022 issue

Sanjit Shewale – Digital Lead, ABB Process Industrieslinkedin

AI is a buzzword in many industries at the moment. But when you talk about it in the context of the cement industry, what do you mean and what are the technologies behind it?

AI has become more than just "a thinking machine" buzzword from computer science. At ABB we see its use accelerating and expanding across process industries, driven by 3 components:

  1. computation power and increased training speed of deep neural networks
  2. connected devices and endless data storage accessible via the Cloud
  3. algorithms getting better at finding patterns and new optimization opportunities when they have access to more data 

For the cement industry constantly looking for ways to reduce the cost of operations while maximizing the yield, improving quality and reducing emissions at the same time, AI means looking at the old problems with a fresh perspective. It also means solving challenges around accuracy and explainability of AI techniques - so that machine's recommendations can be trusted. Data cleansing, anomaly removal, analyzing the correlation of parameters, result interpretation are all key elements here.   

An artificial intelligence is like a formula that has the ability to achieve set goals in new situations, hence the formula can adapt to change rather than remaining a static algorithm. Machine Learning (ML), a subset of AI, is the principle that a machine can learn without human intervention, developing its own algorithm to improve the performance of a specific task.  Neural Networks is a set of algorithms loosely modelled on the way the human brain processes information. Deep Learning is a more sophisticated version of machine learning, used to perform more complex tasks or to produce data needed for decision-making. It uses multi-layer neural networks for a more powerful way to filter and process information. But ML is only able to solve the problems formulated for it. And not every optimization method it learns from data would make sense in real life or deliver tangible benefits.

Share this page

The cement industry faces a range of challenges in its day-to-day operations, around profitability/cost control, quality vs throughput, emissions and environmental sustainability, etc. How can cement producers use AI to help meet these multiple – and sometimes conflicting – KPIs? What advantage does AI bring to cement producers and where is there most potential for AI to benefit the cement production process?

The ability to perform advanced data analytics and smart optimization powered by AI is the gamechanger. It will allow cement producers to hit key performance indicators around sustainability, process performance, asset performance, connected workers and operational excellence. Here are some examples.

 AI can play a vital role in reaching environmental sustainability targets, and not just around reducing emissions, but also energy optimization and management. This creates immediate benefits for operating costs and margins, also enabling new business models for high-tech low-CO2 cements.

For instance, ABB's AI-based system anomaly detection app learns your plant and equipment "normal" states and uses adaptive setpoints to detect unusual patterns, anomalous behaviors. By triggering alerts, it reduces the effort to identify and rectify energy consumption deviations. No more hassle of setting manual setpoints or alarms, no more notification overload. The same way an AI app can learn from your energy usage, production schedules and other factors to deliver accurate forecasts, letting reduce peak demand charges on electricity bills.

A cement plant is constantly worried about deviating from daily SO2 emissions limits and associated hydrate consumption, juggling numerous process constraints. Due to variability in feed and fuel sources, coupled with complex dynamics, manual operators with PID control tend to remain at “safe distances” from process constraints, at the cost of plant profitability.

Today, ABB's advanced process control (APC) solution is successfully used to achieve zero violations of SO2 emissions while reducing hydrate consumption by 11%.  Operator can select either normal or aggressive optimization models, the multiple feeder points of lime hydrate get automatically adjusted.

Similar process performance solutions are addressing thermal efficiency, fuel switching, reduction of the clinker-to-cement ratio and letting the plant run more profitably, e.g.

- increasing feed by over 3 tph while reducing specific energy by 20 kcal/kg.

- delivering overall productivity increase of 4% with better and more consistent cement quality at the same time.

Traditionally, cement strength can be measured after 28 days – by then it is obviously too late to make corrections in the process. Therefore, plants usually “overdeliver” on product specs. ABB is leveraging machine learning (ML) with data-driven soft sensors to predict 28-day strength on the day of sampling, allowing for process corrections - setting new daily CaCO3 / blaine targets. More cement will be sold at correct specification, reducing additives (lime).

With analytics, AI, reinforcement learning from neural networks, APC apps can be further automated in a way that the performance and accuracy of the models are continually monitored. Analytics can re-tune the models, simulate and remodel processes, optimize additional variables - moving towards adaptive APC and autonomous operations.  This will reduce engineering time and allow the system to stay at peak performance, delivering benefits.

Leveraging AI for asset performance management is a step change in the way maintenance and reliability people can collaborate with other functions, ensuring assets are available at the time and at the performance level required by the operations - depending on changing production goals. It’s harder to predict how assets react and respond to various triggers like age or operating condition, because complex systems interact in unexpected ways and are always evolving. It’s harder to make sense of things, because degree of complexity may be “beyond the human eye”. To provide accurate target parameter predictions in near real-time and prevent failures, AI / ML models need to be continuously trained with relevant dataset – requiring deep understanding of both cement processes and asset behavior. Think of AI-enabled APM as the most cost-effective way to extend the life of the aging (and newer) assets,   to decide on the optimal timing for scheduled maintenance turnarounds (one of the biggest costs in a plant) and plan better.

Without question, AI detection systems and video analytics can be used to create a more connected workforce and improve safety.  e.g. scanning employees and equipment, identifying potential risks, such as a worker who has forgotten to wear the appropriate safety gear.

More and more cement producers on the digitalization path would like to take a more proactive approach to cyber security. ABB's analytics solution and services let continuously monitor, diagnose and resolve security issues, helping safeguard people, assets and reputation.  And because technology and cyber threats can both change unpredictably, the strategy needs to be reviewed periodically, including performing simulations under different circumstances, like a major ransomware incident. Data analytics can be used to test various what-if scenarios.

Enterprise grade AI-based solutions also have enormous potential for operational excellence, helping understand the reasons behind the differences in cement plant performance levels. Transferring knowledge or process methodology from the higher-performing to the lower-performing plants will optimize production and uncover best operating points to meet and even exceed KPIs. With less cement required in the future for modular pre-fabricated buildings, AI will play an important role in restructuring operations to retain profitability at reduced cement demand.  By analyzing how procurement was done in the past, it can also assist with better planning for supply chain management.

Cement manufacture is a complex and constantly changing process. Can a machine learning algorithm ever truly understand the dynamic conditions inside a cement plant?

First, successful ML solution implementation requires standardized data, with data management approach closely aligned to the business strategy. The best classification model requires many iterations and a rare combination of data science, very specific cement industry expertise and ingenuity. To operate assets and dynamic processes optimally, it is important not to treat them separately. The real value and true understanding will come when you start uncovering previously hidden relationships and correlations. Increasing data volume and quality will improve model performance. We are just at the beginning of this exciting journey.

Any process area like quarry, raw material / meal preparation, clinker production, cement production or dispatch and virtually every asset or asset system’s DNA can be decoded over time. Technology, capital, or people – are all valuable assets for which AI can be applied - provided that there is enough data and there are people able to teach machines. Specialists using their expertise to literally show machines how to structure their computing methods more like human thought, how to break problems into simple steps. They need to guide the machine how to solve complex tasks faster by using a ‘lesson plan’ that explains the important information, offer guidance on how to self-monitor for effectiveness.

ABB’s experts and ecosystem partners can meet the special needs of the cement plant owners, collaborating to produce business results they demand and reach decarbonization targets by leveraging ABB Ability™ Genix Industrial Analytics and AI suite.

How does AI interact with existing control systems and human operators?

Human operators tend to operate a cement plant within the limits that their control system tells them. But imagine knowing the history of your plant like you know the history of your life – with the ability to operate having a memory of exact parameters from the past. AI interacting with the control system history data can start learning from patterns, recommending optimal parameters and how to avoid a negative impact. After the AI learns the best setpoints for quality, energy, eliminating downtime, etc; the core software needs to write them to the control system. This can be done either directly by rewriting a set point (close-loop system) or indirectly by providing recommendations to a human operator (open-loop system).

AI can also contribute to higher workforce productivity – analyzing how operators interact with a control system, how quick they are to react to alarms. It can learn which priority alarms require faster actions, improve their visibility, filter and rationalize alarms. 

How are AI technologies evolving? What does the future hold for AI and its use in the cement industry?

Definitely, there is more data coming to AI, with more and more instrumentation, historians and other databases. Maybe the data from 20 years ago is gone, but more and more data will be there for AI advisory to evolve. The AI implementation effort is decreasing, more and more people will get involved, driven by the sustainability targets – since there is no way to reach those just by manual setpoint adjustments. I think, you are going to start to see the cement industry improve its carbon footprint and its reputation.

As the next step, we are going to see a lot more of unit area models, plant models able to continuously retrain themselves based on outside disturbances – because things are continuously changing on the ground. This will give confidence to cement companies, make  them more comfortable with thinking about less humans in a plant and autonomous operations. I do still think that our core solution areas like Operational Excellence, Process Performance, Asset Performance, and others are still going to be needed. These are core fundamental areas that will continue to provide value, but those type of solutions can become autonomous themselves.   To me, that’s what AI is going to bring.

Learn more

  • Contact us

    Submit your inquiry and we will contact you

    Contact us
Select region / language