(84au) Developing Efficient and Sustainable Packaging Processes in the Downstream Operations of the Oil and Gas Sector

Oil and gas is the major industry in the energy market and it involves various upstream (exploration, production), midstream (transporting, storing), and downstream (refining, packaging, selling the finished products) activities (Nelson 1969). Each of these activities involves high risks and therefore significant efforts are made by industries to reduce the impact of their processes on the environment. Companies are exploring systematic operational procedures to manage and improve the greenness of these activities. In this work, we study one such downstream operation from a manufacturing and packaging plant of refined petroleum lubricants. We use a Machine Learning (ML) approach for improving the operational performance and sustainability of their packaging activities.

The performance index of a manufacturing company is evaluated by the company’s capability to attain the desired production goals and effectively utilize its available resources. Thus, excellent performance can be achieved by developing and implementing an appropriate manufacturing strategy. The lubricant industry produces over 2000 unique lube oil compositions in a single production year. The manufactured products within a certain limit will have comparable physical properties, similar specifications, and applications. Therefore, these manufacturing plants are categorized as multiproduct plants(Zlokarnik 2001). Multiproduct plants use the same process for the sequential production of different products belonging to the same product family, e.g. engine oil family with products for car engines for different brands.

The multiproduct plants are batch operated to sequentially produce the desired products as per the market demands. Therefore, a special necessity of these plants is to avoid cross-contamination between changeovers and efficiently clean the production vessels, ancillary equipment (valves, fittings, filters) and the associated pipeline system. The straight pipeline sections at the plant are cleaned using polymeric pigs. However, pigging cannot be used for variable diameter pipelines and ancillary equipment having complex design (Stewart 2016). Therefore, a technical solution to this problem is flushing the residual oil in the pipes with new oil that is to be packaged next. This results in the formation of a mixed oil that does not meet the specifications of any of the two batches and is regarded as low value mixed oil. The existing flushing operations lack a standardized procedure which result in generation of large volumes of mixed oil leading to economic losses exceeding millions of dollars annually. Moreover, this leads to poor resource management and with a large energy and environmental footprint. With the growing awareness of sustainability and policies related to pollution prevention and mitigation, it has become a necessity to not only make good quality products but also make their manufacturing and packaging processes environment friendly.

To this end, we present a machine learning approach to strategically optimize the cleaning operations of the multiproduct plants. We considered a case study of the lubricant industry where we have 12 months of data consisting of over 900 features and 300 data points. Out of the total data points, 70% were used for training, 15% for validation, and the remainder for testing. Through a multi regression machine learning algorithm (Ray 2019), we develop an approach for systematically optimizing the flushing times and the associated product volume for a particular changeover. Furthermore, we also analyze the associated environmental impacts. Since the environmental impacts are associated with the mass and energy flows of a product, the respective first principles conservation equations for the system add valuable information. Next, we collect process-related sustainability metrics data via LCA (Life Cycle Analysis) tools such as GREET (Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies) model by Argonne National Lab (Wang 2007), SPIonWeb (Shahzad et al. 2014), and GREENSCOPE (Ruiz-Mercado, Gonzalez, and Smith 2014) to predict the CO₂ emissions, Global Warming Potential and Sustainable Process Index (SPI) (Papadokonstantakis et al. 2016; Klemes et al. 2014; Goedkoop and Spriensma 2001). Through this work, we present an approach to making the operations of the multiproduct plants economically profitable and environment friendly. Figure 1. shows a schematic of our overall approach.

References:

Goedkoop, Mark, andSpriensma. R. 2001. “The Eco-Indicator 99: A Damage Oriented Method for Life Cycle Impact Assessment,” January.

Klemes, Jiri J., Petar Sabev Varbanov, Peng Yen Liew, Peng Yen Liew, Petar Sabev Varbanov, and Peng Yen Liew. 2014. 24th European Symposium on Computer Aided Process Engineering: Part a and B. Oxford, NETHERLANDS, THE: Elsevier. http://ebookcentral.proquest.com/lib/rowan/detail.action?docID=1718627.

Nelson, W. L. 1969. Petroleum Refinery Engineering. 4th Edition. McGraw-Hill Book Co.

Papadokonstantakis, Stavros, Paraskevi Karka, Yasunori Kikuchi, and Antonis Kokossis. 2016. “Challenges for Model-Based Life Cycle Inventories and Impact Assessment in Early to Basic Process Design Stages.” In , 295–326. https://doi.org/10.1016/B978-0-12-802032-6.00013-X.

Ray, Susmita. 2019. “A Quick Review of Machine Learning Algorithms.” In 2019 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COMITCon), 35–39. https://doi.org/10.1109/COMITCon.2019.8862451.

Ruiz-Mercado, Gerardo J., Michael A. Gonzalez, and Raymond L. Smith. 2014. “Expanding GREENSCOPE beyond the Gate: A Green Chemistry and Life Cycle Perspective.” Clean Technologies and Environmental Policy 16 (4): 703–17. https://doi.org/10.1007/s10098-012-0533-y.

Shahzad, Khurram, René Kollmann, Stephan Maier, and Michael Narodoslawsky. 2014. “SPIonWEB – Ecological Process Evaluation with the Sustainable Process Index (SPI).” In Computer Aided Chemical Engineering, edited by Jiří Jaromír Klemeš, Petar Sabev Varbanov, and Peng Yen Liew, 33:487–92. 24 European Symposium on Computer Aided Process Engineering. Elsevier. https://doi.org/10.1016/B978-0-444-63456-6.50082-X.

Stewart, Maurice. 2016. “Pipeline Pigging.” In Surface Production Operations, 897–960. Elsevier. https://doi.org/10.1016/B978-1-85617-808-2.00013-4.

Wang, M. 2007. “The Greenhouse Gases, Regulated Emissions and Energy Use in Transportation (GREET) Model.”

Zlokarnik, Marko. 2001. Stirring: Theory and Practice. Wiley. https://doi.org/10.1002/9783527612703.