R. Confair, LyondellBasell, Houston, Texas (U.S.); and M. BERRIOS-SOTO, Emerson’s Aspen Technology business, Houston, Texas (U.S.)
LyondellBasell Industries operated one of the largest refineries in the U.S.—located in Houston, Texas (U.S.), just off the Houston Ship Channel—with a rated capacity to convert 268,000 bpd of crude oil into clean fuels, lubricants, chemical intermediates, coke and other products.
Producing middle distillates, including jet fuel and ultra-low sulfur diesel (ULSD), requires optimizing operating conditions and blending streams from multiple sources, each with constraints that change over time due to factors like feed composition, equipment limitations and catalyst activity. Before implementing a dynamic optimization solution, each contributing unit's advanced process control (APC) operated independently, targeting local specifications such as sulfur levels, flash points and other quality parameters based on estimates and delayed feedback from adjacent units, rundown lines and tank quality samples/inferentials. Operators used this feedback to manually adjust APC limits, tank lineups and other targets. This approach demanded significant communication, coordination and manual effort, often leading to over/under corrections, off-spec tank completions, and decreased production of clean diesel and kerosene fuels.
Reducing product quality variability through dynamic optimization. Dynamic optimization enabled the site to reduce product quality variability by dynamically coordinating multiple units within the middle distillate envelope. This technology identified process constraints and limits sooner, automatically adjusting other units through scaled moves to achieve global economic optimization.
Utilizing this approach, engineers could enhance, rather than replace, existing APC and inferential applications, and production planning/scheduling workflows. Interfacing with these applications and enhanced workflows facilitated the implementation of global economic objectives with minute-by-minute process feedback similar to APC. This dynamic optimization approach combined non-linear models with APC dynamic models, utilizing process feedback to auto-calibrate through dynamic data reconciliation. This technology solution uses process control models and historical data to ensure the accuracy and consistency of data across the process optimization scope. This approach optimized control moves, achieving higher unit control, increasing throughput and providing tighter control on product flash and freeze points (FIG. 1).
Leveraging existing tank quality inferentials (TQIs) models, the dynamic optimizer solution efficiently corrected potentially off-spec conditions before tank completion. Adjustments considered were tank heels and controlled rundown qualities to ensure the blended tank finished closer to ULSD sulfur limits, reducing off-spec production. This dynamic process optimization solution predicted rundown targets and tank properties for five diesel tanks with five specifications each, and three jet tanks with four grades and three specifications each. A thorough evaluation of site operations resulted in the elimination of redundant inferentials. As a result, only two inferential applications were implemented—one for jet and one for diesel—reducing the total number of inferentials from 64 to 16. These prediction models enabled the dynamic optimizer to correct off-spec conditions prior to tank completion. Additionally, the application handled multiple tank lineups and lab sample inputs with no manual intervention (FIG. 2).
The dynamic optimization solution implementation leveraged existing applications interfaces in APC performance monitoring, process historian, inferential and APC, minimizing implementation costs and requiring minimal operations training (FIG. 3). This approach simplified/reduced data collection, consolidated TQIs, facilitated the addition of process variables tags and allowed APC model updates without impacting the collection of historical data.
Transformative production outcomes. The implementation of the dynamic optimization solution at LyondellBasell’s Houston refinery led to significant improvements in middle distillate production, including:
Increased ULSD production: A 2% increase in ULSD production from naphtha and gasoil.
Reduced product variability: Completed by coordinating strategies and targets across multiple units.
Enhanced control of product specifications: Achieved tighter control over flash and freeze points for kerosene products and sulfur levels in ULSD.
Operational efficiency: Significant reduction in the number of tasks performed moves made by operators and superintendents, freeing up time for other tasks.
Enhanced coordination: Key information from the dynamic optimizer solution was shared across the plant, ensuring everyone was aligned on constraints and limitations.
Low maintenance costs: Leveraged existing tools and infrastructure, reducing the need for upgrades and minimizing training requirements.
No off-spec tanks: Post-commissioning, LyondellBasell saw immediate benefits with no off-spec tanks and all tanks running closer to spec, resulting in smoother and more consistent operations.
The implementation of the dynamic optimization solution at LyondellBasell’s Houston refinery significantly enhanced middle distillate production by improving product quality, increasing ULSD output and streamlining operations. The coordinated strategies and dynamic adjustments facilitated by the dynamic optimization solution led to tighter control over product specifications and reduced operational variability. The project not only delivered immediate economic benefits but also optimized the refinery's overall performance, demonstrating the value of dynamic process control in complex industrial environments. HP
Robert Confair was formerly an Advanced Process Control Engineer at LyondellBasell’s Houston refinery. He has worked in refining and chemical plant roles as a Process Engineer, a Process Design Engineer, an Equipment Team Lead and mostly as a Process Control Engineer. Confair has implemented several APC projects across multiple units, including implementing a few global optimizers. He earned a BS degree in chemical engineering from the University of California, Los Angeles (UCLA) in 1993, and is licensed as a Professional Engineer in the state of Texas.
Melvin Berrios-Soto is a Product Marketing Manager at Emerson’s Aspen Technology business, specializing in process and control solutions. With more than 8 yr of experience in engineering and project leadership within the downstream industry, he helps customers improve production performance and reduce operational costs through the successful adoption of AspenTech control and optimization solutions. Berrios-Soto also leads the development of an active user community by organizing conferences and webinars to share best practices and gather feedback that shapes future innovations.