Interfacial mixing of consecutive product batches in multiproduct pipeline systems has been a long-standing and economically significant issue in the petroleum industries (Li and Bidmus 2019; He et al. 2019). The U.S. liquid petroleum pipeline comprises 200,000 miles of pipe in all fifty states. These pipelines are vital to the nation’s economy and transport a wide range of materials, including gasoline, aviation fuels, kerosene, diesel fuel, and heating oil (Pharris and Kolpa 2008). In this work, we investigate one such multiproduct complex network of pipelines used to transport multiple fluids. Ideally, the production lines must be cleaned/flushed between product changeovers to avoid contamination and maintain final product integrity. The flushing operation utilizes a significant amount of finished products, which results in the contamination of the high-value products, and waste in the production operation. The process results in high economic losses to the industries and adds to a significant environmental burden. Therefore, it is crucial to understand the drawbacks of the existing flushing operation and minimize the amount of oil that gets downgraded. Hence, our research goal is to implement novel process optimization techniques and enhance the existing flushing operations in lubricant industries by formulating the operation as an optimal control problem.

The current flushing operation is based on previous operator experience to determine appropriate flush times and volumes. This, however, is not optimal, and there are potential opportunities for cost, energy, and material savings on detailed analysis and subsequent optimization of this process. In this research, we are developing models of fluid hydrodynamics in multiproduct pipelines. We will use process optimization tools to predict strategies for conducting the flushing operations with less energy consumption and material downgrade.

This research aims to develop an optimal control problem for flushing operations in the petroleum industries. The problem is formulated using various performance indices: minimum flush time, maximum product purity, and minimum oil downgrade. We use Pontryagin’s maximum principle for solving these problems. In our work, we present the formulation of these optimal control problems and analysis of the solutions. Our objectives can be categorized as follows:

• Data collection and analysis
• Model development for fluid hydrodynamics within the flushing process
• Formulation of an optimal control problem that minimizes the flushing time, maximizes the product purity, and minimizes the oil downgrade
• Solution strategy using Pontryagin’s maximum principle and
• Solution analysis and recommendations for industrial application

A 23-week lube oil flush study data was collected from our industrial partner. This data was analyzed to understand the existing flushing operations at the facility. The uncertainty observed through the data analysis gave us a strong indication that the current flushing processes can be optimized, thus minimizing the oil downgrade and overall energy consumption. Following this analysis, on-site flushing experiments were designed, and 70 product changeovers were conducted to identify the key process conditions and bottlenecks. We then developed a first-principles model of the flushing operation, which incorporated the essential parameters from the experimental flush study. The parameters include but are not limited to the viscosity of individual products, the viscosity of product blend, density, flowrate, flushing time, and downgraded/flushed oil. These parameters are incorporated into time-dependent differential equation model, and optimal control problems are formulated using various performance indices mentioned above. Pontryagin’s maximum principle is used for solving these optimal control problems (Diwekar 2008; Yenkie and Diwekar 2013). The advantage of this method over other approaches of solving optimal control problems is that it does not involve second-order differential equations or partial differential equations (Benavides and Diwekar 2013; Yenkie and Diwekar 2014).

Our research can be an excellent starting point in optimizing the existing flushing operations in the lubricant industries. It can lead to financial savings for the company and improve the environmental footprint of the process.

Keywords: Flushing, lubricants, optimal control, optimization

Acknowledgments

• S. Environmental Protection Agency’s Pollution Prevention (P2) Program
• Sustainable Design and Systems Medicine Lab
• Undergraduate Students: Joseph D’Intino, Anthony Wiley, Jacob Martin, Erik Dunn, Emily Rooney, Diana Castro, Marissa Martine, Spencer Verdoni
• Rowan University Department of Chemical Engineering

References

Pahola B. T., and Diwekar U. 2013. “Studying Various Optimal Control Problems in Biodiesel Production in a Batch Reactor under Uncertainty.” Fuel 103 (January): 585–92. https://doi.org/10.1016/j.fuel.2012.06.089.

Diwekar U. 2008. Introduction to Applied Optimization. Springer.

He G., Na Y., Kexi L. Baoying W., and Liying S. 2019. “A Novel Numerical Model for Simulating the Quantity of Tailing Oil in the Mixed Segment between Two Batches in Product Pipelines.” Research Article. Mathematical Problems in Engineering. Hindawi. August 19, 2019. https://doi.org/10.1155/2019/6892915.

Li G., and Hamid B. 2019. “Estimating Mixing Lengths in Multi-Product Pipelines.” In . OnePetro. https://onepetro.org/PSIGAM/proceedings/PSIG19/All-PSIG19/PSIG-1903/2103.

Pharris T. C., and Kolpa. R. L.2008. “Overview of the Design, Construction, and Operation of Interstate Liquid Petroleum Pipelines.” ANL/EVS/TM/08-1, 925387. Argonne National Lab. https://doi.org/10.2172/925387.

Yenkie, K. M., and Diwekar U. 2013. “Stochastic Optimal Control of Seeded Batch Crystallizer Applying the Ito Process.” Industrial & Engineering Chemistry Research 52 (1): 108–22.

Yenkie, Kirti M., and Diwekar U. 2014. “Optimal Control for Predicting Customized Drug Dosage for Superovulation Stage of in Vitro Fertilization.” Journal of Theoretical Biology 355 (August): 219–28. https://doi.org/10.1016/j.jtbi.2014.04.013.