Profit Maximization Techniques for Operating Chemical Plants. Sandip K. Lahiri

Чтение книги онлайн.

СКАЧАТЬ

      2.3.1 Attaining a New Level of Functional Excellence

      Data analytics and AI‐based interpretation is helping efficiency improvement of all core business processes of the chemical industry, including manufacturing, marketing and sales, and R&D. Data‐based decision making, called digital in short, provides the means to unlock a new level of productivity enhancement (Klei et al., 2017).

      2.3.1.1 Manufacturing

Manufacturing operations consume most of the production costs and digital technology can bring the highest impact in this area. This is true for all segments of the chemical industry, from petroleum refinery to petrochemicals to pesticides to specialty chemicals. The global management firm Mckinsey estimate the potential for a three‐ to five‐percentage point improvement in return on sales from employing digital in production operations (Klei et al., 2017). With the advent of data historians two decades ago, an enormous amount of production‐related data has been collected by all of the major chemical industries. However, due to the absence of proper data analytics software, most of these data remains unutilized. AI‐based algorithms can extract knowledge from these data and utilize that knowledge to achieve higher efficiency and throughput, lower energy consumption, and more effective maintenance. For many companies, these are low hanging fruits and benefit can be achieved immediately using existing IT (information technology) and process control systems.

      The contribution to profits can be substantial. Examples are many in all leading chemical industries around the world. A major petrochemical company applied advanced AI‐based data analytics to a billion data points that it collected from its naphtha cracker manufacturing plant. With the help of an AI‐based stochastic optimization algorithm, this plant optimizes different process parameters that lead to an increase in the ethylene production by 5% without making any capital investments and generated cost savings by reducing energy consumption by 15%. A leading refinery company takes another approach at one of its main plants: it used AI‐based advanced analytics to model its production process and make a virtual plant, and then used the model to provide detailed, real‐time guidance to DCS (distributed control system) panel operators on how to adjust process parameters to optimize performance. Once it was implemented, profit from this plant increased by over 25% and yields increased by seven percentage points, thus saving on raw materials, while energy consumption fell by 26% (Holger Hürtgen, 2018).

      Besides this AI‐driven analytics‐based opportunity, there are other digital‐enabled advances that have started creating profit in the manufacturing operations area. Examples include IoT‐based steam trap monitoring, IoT‐based wireless vibration and temperature monitoring of critical pumps and other single‐line rotating equipment, the use of digital sensors to monitor vent gas composition, etc. These advances help to reduce maintenance costs and improve process reliability and safety performance. At the same time, deploying a holistic automated and centralized data analytics and plant performance management system should enable the plant engineers to monitor the plant better and take proper corrective and preventive actions faster.

      2.3.1.2 Supply Chain

      Digital technology also can bring enormous value to the entire supply chain, including inbound and outbound logistics and warehousing. From past historical data, an intelligent algorithm can significantly improve accuracy of forecasting, which helps to optimize the entire sales and operations planning process (Klei et al., 2017). Digital technology can be used to leverage better scheduling of batch production, shorter lead times, and lower safety stocks with a higher level of flexibility. A digital enabled holistic system can be built to develop integrated “no touch” ordering and scheduling systems.

      2.3.1.3 Sales and Marketing

      Data analytics and AI‐based digital technology can be used for intelligent decision making in sales and marketing. Mckinsey estimate‐ that digital‐enabled initiatives in marketing and sales could improve the industry's average return on sales (ROS) by two to four percentage points.

Digital initiatives in marketing and sales include developing intelligent pricing systems, generating growth opportunities from data, and using algorithms to predict churn at the individual‐customer level and then suggesting countermeasures to the sales force. The impact of these initiatives can be significant. A large polymer company used advanced analytics to reset prices for hundreds of thousands of product‐customer combinations in three core countries, based on individual risk and willingness to pay. By developing an AI‐based intelligent algorithm, the company was able to achieve price increases of 3 to 5%, compared to 1% increases in previous years. In some other petrochemical companies, the company's manufacturing unit is connected with the sale and marketing unit by an optimization algorithm and the company's production plant process parameters, product split in a multi‐product plant, and capacity are adjusted by the demand scenario coming from sales and marketing forecasting.

      2.3.1.4 Research and Development

      Due to plastic pollution, pollution from cars and from various carcinogenic chemicals, the usage patterns and demands of various chemicals across the globe is changing very fast. This poses a challenge to chemical industries who makes those products.

      One of the ways a research and development department of chemical plants can respond to this challenge is by creating higher‐value‐added, higher‐margin products at a faster pace, in particular in specialty chemicals and crop‐protection chemicals (Klei et al., 2017). Through intelligent algorithms, chemical companies will be able to use high‐throughput optimization to develop and adjust molecules that offer more value. They will also be able to deploy advanced analytics and machine learning to simulate experiments, to use digital predictive power to systematically optimize formulations for performance and costs, and to data‐mine information available from past successful and failed experiments. Not least, they will be able to identify the best possible resource allocation to enhance the performance of R&D teams and the innovation pipeline. Many of these practices are already established in the pharmaceutical industry but were largely unaffordable for chemical companies. With the emergence of inexpensive computing power on a massive scale, this is likely to change.

2.4 Using Advanced Analytics to Boost Productivity and Profitability in Chemical Manufacturing

      As of now, it is quite clear that digital will have a significant impact on many areas of the chemical industry, with the gains in manufacturing performance potentially among the largest companies (Holger Hürtgen, 2018). Chemical companies have already created the infrastructure to collect and store enormous amounts of process data from hundreds of thousands of sensors, but very few have succeeded so far to take advantage of this data gold mine of potential intelligence. With the availability of cheaper computational power, IoT‐based cheap sensors, and intelligent advanced analytics tools, all chemical companies can now use those data to make more profit, extract knowledge from those data, and using machine‐learning and visualization platforms to uncover ways to optimize plant operations (Wang, 1999).

      AI‐based machine learning tools can be used to develop insights into what happens in a chemical plant's complex manufacturing operations; this can help chemical companies solve previously impenetrable problems and reveal those that they never knew existed, such as hidden bottlenecks or unprofitable production lines.

There are three major areas where applications of advanced analytics tools can give an enormous profit increase, namely predictive maintenance; yield, energy, and throughput analytics; and value‐maximization modeling, as shown in Figure 2.3 (Wang, 1999).

СКАЧАТЬ