Reaping The Rewards Of Foresight

Cement has been used in construction for over two thousand years. The production process has evolved and become more efficient, increasing production capacities to respond to new market demands. While demand is forecasted to keep growing in the coming years, cement companies must ensure operations are efficient, safe, and profitable, while simultaneously driving changes to improve sustainability KPIs (key performance indicators).

With a process that has been improved across two millennia, only so much can be optimised without radical changes. Automation has also evolved considerably in the past decade, so digitalisation is the main area to invest in and drive fast business gains. Today, the so-called ‘Digital Transformation’ affects a production company’s engineering and operational areas. The management of data in a digital environment is critical to enable models and analytics that can improve efficiency and ensure safety. Organisations invest billions in operational assets that produce a massive amount of data; however, if asset information is poorly managed, it can cost them several percentage points of revenues.

The first step in a digital transformation roadmap is to build the data infrastructure. An operational data management system must be in place, and it is critical to enable the effective implementation of other advanced applications. Once historical and real-time data are aggregated, contextualised, and visible throughout the organisation, it is time to evaluate how to leverage this data to accomplish its business goals.

One of the advanced methods with a shorter implementation and payback period is artificial intelligence (AI) and machine learning (ML), which can be applied to critical equipment as part of a predictive and prescriptive maintenance strategy. When historical conditions are combined with real-time asset behaviour and infused with AI and ML, the operations and maintenance teams can detect anomalies in asset behaviour and predict failures weeks or months in advance. The constant monitoring of the equipment operating conditions allows for the identification of issues that could lead to an unplanned shutdown, equipment damage, financial losses, and even safety incidents. Based on early warning notifications of abnormal behaviour, in-depth diagnostics, and prescriptive guidance, workers can remediate the failure to avoid those undesired outcomes.

Predictive maintenance

A digital transformation is a journey often constrained by limited resources. Predictive analytics deliver a measurable and fast impact and return on investment (ROI), but moving to a predictive maintenance strategy will not happen overnight. Companies must lay the right data foundations and establish change management strategies to ensure optimal outcomes. Industrial production companies have traditionally used reactive (run to failure) and preventive (planned based on time, calendar, or usage statistics) maintenance strategies.

Many years ago, many companies moved to condition-based maintenance, which uses rules-based logic fed by sensor data to determine the best time to perform a maintenance procedure. However, this can only improve asset performance up to a certain point. According to ARC research, “80% of all assets fail randomly despite having rigorous reliability and maintenance programmes in place.” This means that, even if a company is willing to invest more in maintenance to prevent failures, it is almost impossible to know which asset is more likely to fail by only tracking sensor data and applying pre-defined rules. Preventive maintenance will work only for about 20% of asset failures.

The answer to being able to predict abnormal behaviour in assets and processes is to adopt AI and analytics models capable of using historical and current operational data to identify potential issues well ahead of time and forecast the time to failure. The anticipation of the prediction is vital to schedule maintenance at the most convenient time, with the best team and resources in place, avoiding unplanned shutdowns and optimising procedures to reduce maintenance costs.

Decades ago, machine learning solutions required significant efforts to be implemented. Specialised resources like data scientists were required, pushing too much dependency on vendors, and the model required long periods of historical data to be built. Technology has since evolved, and data scientists are no longer needed to deploy predictive analytics models when using the proper tools and solutions. Further, the models can be ‘trained’ using shorter periods of historical data than before. Due to such improvements in the implementation process and a quicker payback time, the adoption of predictive analytics applications has been increasing across the process industries, including cement.

Enjoyed what you've read so far? Read the full article and the rest of the June issue of World Cement by registering today for free!

Read the article online at: https://www.worldcement.com/special-reports/19062023/reaping-the-rewards-of-foresight/