November 2018

Columns

Digital: Improve profitability in the process industry with artificial intelligence

The digitalization of chemical processes is becoming a critical factor in ensuring the profitability of chemical plants.

Bergman, S., NAPCON

The digitalization of chemical processes is becoming a critical factor in ensuring the profitability of chemical plants. Digitalization is a wide concept: it covers real-time systems, software and algorithms that convert process and equipment-related data from various sources into valuable information and insight, enabling the online optimization of plant operations.

Many of us think that our plant’s profitability has reached its maximum level; however, by utilizing artificial intelligence (AI) and machine learning, it is possible to take profitability to the next level. As market prices of feedstock fluctuate and operating conditions vary, it is crucial to guarantee the profitability of production under any circumstances.

Accessing data.

The first challenge in digitalization is to access all relevant and available data. Traditionally, real-time optimization is based on process measurements in the process control system/distributed control system (DCS). However, other data sources are often needed to increase profitability, such as a laboratory information management system (LIMS) or data on logistics operations in an enterprise resource planning system (ERP). Modern IT communication technologies, such as OPC unified architecture (UA), enable safe and secure communication between these different systems and the digitalization platform.

The next step is to refine the data into information on unhidden phenomena that cannot be directly observed by monitoring the time series of single measurements. Models based on AI, particularly machine learning, have proven to be powerful tools for these types of tasks. The refined information based on AI models enables operation, e.g., in various transient states. Traditionally, plants are driven with caution, especially if direct and real-time data of all production conditions do not exist.

With AI and machine learning, it is possible to better understand what goes on in processes and within process equipment. This knowledge enables improved prediction and the optimization of operative actions. Production can be run at maximum levels with the maximum profit safely.

FIG. 1. AI and machine learning enable improved prediction and the optimization of operative actions.
FIG. 1. AI and machine learning enable improved prediction and the optimization of operative actions.

The key benefit of AI in the process industries is its ability to consolidate production data with other operational data, pushing a company’s operations to new levels (FIG. 1). Examples of regular incidents that can be more easily monitored and predicted with AI include corrosion and events related to deterioration and fouling. Early identification improves real-time production and maintenance optimization, as well as other operating breaks. Predictive and preventive maintenance based on AI improves profitability, as stoppages and production breaks can be planned and costly breakdowns can be avoided.

Taking action.

However, digitalization is not only about technology. It is, above all, a business and leadership decision. A key to making digitalization a profitable business case is automatization. Refined information and improved knowledge do not necessarily increase the profitability of chemical processes—improved digital insight must be put into action, automatically and in real time. Often, the easiest and most efficient way to implement this is to utilize the information in various levels of process optimization, especially in real-time optimization and advanced process control (APC). Understanding this data and analyzing it with careful insight are necessary to make correct predictions and actions.

The use of AI does not remove the need for the human workforce, nor does it remove the responsibility of human operators. This makes digitalization a leadership challenge.

The increased information that digitalization provides will change the ways operators work. They must understand new methodologies and learn to trust this new information. Therefore, digitalization projects in the chemical industry should never be left solely for ICT or automation departments. Plant engineers, shift supervisors, and panel operators and trainers must be involved in the early phases of digitalization. HP

The Author

From the Archive

Comments