Crude planning and scheduling system decide when, where, and the amount of crude oil to blend to satisfy the market demand and for using the resources efficiently. The refinery depends on the receipts and the type of crude they are about to get in the forthcoming months or within a period. Depending on the type, quantity, quality, and received time of the crude oil, the refinery follows a scheduling system. It uses the planning and scheduling expert system of a third-party source to execute it.
This planning and scheduling system mostly consists of mixed-integer programming. It also considers the crude oil in the operating plant that includes tank quality control schedule, crude tank inventory, and its constitutions, operational and physical constraints. Evaluating a mixed-integer plus various other considerations makes this an extremely complicated process. Specifically, considering receipt crude, its schedule, and while feeding the same to the tanks in the refinery and to the crude processing unit. This complex evaluation sometimes gives correct results; sometimes it does not provide reliable results.
Functions of crude blending
To solve this problem, refineries split it into small linear programming problems. For instance, first, the crude planning and scheduling module is done and then they evaluate “how to optimally store crude oil in the tanks.” Now the tanks one through four need a mixed crude oil of varying compositions. This helps the process to be quite simple linear programming. Since crude is blended on the sulfur and density; unlike the fuel blending.
However, we also require some crude tank monitoring modules with regard to the quantity inventory and composition. From there it is directed to the process units, after being blended, or line blended. There might arise a few complications in this section, which will be discussed shortly. These complications arise in terms of what is feed to the crude unit. And, linear programming uses only the density parameter as the controlled quality. Since only density decides the output of the crude unit.
Types of crude blending
The receipt crude can arrive at the refinery through pipelines, other than the ship. Have experience of working in a refinery which consists of about 11 to 12 pipelines of crude coming in varying types and from sources. The first stage is the storage of crude and the second stage involves feeding the tanks for blending. Depending on the type of crude, they must be stored in different tanks.
Linear programming is used by the refinery to direct the pipeline to obtain optimize results by directing crude type to the tank. For instance, say a specific tank 966 can use the crude from different upstream crude tanks to supply input to the crude unit. Assume this refinery includes five different crude units that can process various crude types. And hence based on the processing ability and the requirement of the tank, the various crude types are blended in-line (parallel) before being fed to the unit with custom-blended crude. Here we can find different types of linear programming. Additionally, if we include the scheduling of pipelines as well then it changes into mixed-integer programming. Therefore, there are only two approaches to this problem. Either the entire problem can be split into simple problems, or it can be treated as one giant problem. It is wiser to prefer the simple approach as that is a better approach.
This is a direct method of transferring crude from the ship to the crude storage tanks. This is done by observing the quantity, composition, and inventory of recipient crude tanks. The information about the crude quality, type, and quantity is already provided to the crude planning and scheduling module. This method helps achieve the optimum routing of the crude from the ship direct to the crude blending tanks.
This is another direct blending method that uses a simple linear programming algorithm. In this blending method, crude is transferred from tanks to units where it is subjected simple linear programming algorithm. This algorithm determines the crude type, quality, and quantity of crude to be fed to the crude unit. The crude qualities that are controllable are density and Sulphur, based on which online analyzers are fixed to supervise the blended qualities of the crude. Usually, online analyzers are placed after the blend header section. It is important to consider that multi-crude blending is needed since each individual crude unit can only process a crude type. Therefore, we can conclude that the crude blending optimizes the feed crude to satisfy the crude specifications for a unit.
Multi-headers crude blending
We have discussed the multi-headers crude blending which uses several blend headers to blend various crudes. Depending on the crude quality this is then fed differently to various crude units. We must keep in mind that all these mentioned applications are operating at the same time. It is not possible to do the crude blending for a unit at a time. In such a case, you require a single large LP program that can control all five headers at a time. Previously we discussed that each crude tank has various crude types that are mixed in best proportions. This is then given to the custom blended crude to each crude unit. This implies that the huge LP is having control of the feed to each of the five units at the same time. Since it takes the supply from various tanks and blending is done in-line.
Integration of crude blending
We are going to discuss the integrated scheme of feedforward and feedback to the product blending and to the crude. The modules for crude scheduling and planning are on the left-hand side. The crude composition, scheduling, and blend control are all supervised using different modules. After being monitored and controlled by the modules, it is directed to the crude unit where the output or the production of the side streams is again optimized.
Obtained side streams are directed to the storage tanks. Here there are two main sections that mix the fuels from the component tanks they are, “Optimization system and fuels blending control.” The final output product is sent to the product storage tank. With regard to the feedback, there is a crude unit model that optimizes the product output based on the crude type. Based on this model, we can fix targets for the crude side stream quality and estimate set points for rates. This in turn fixes a target for crude oil and then moves forward to the feedback.
Thus, this is a large mixed-integer nonlinear programming that binds the crude scheduling completely to the product blending, and it also provides some amount back to the crude receipt. This process continues like a cyclic process and it seems like a complicated problem. But in my experience, this was successful in a few of the refineries that I have worked. There are some refineries, where it can be split into sub-systems and sub-modules to get the same output. Consider some of the fuel blending modules that are typical examples of linear programming or nonlinear programming in mathematical programming.
It is a known fact that about 80% to 85% of crude oil is made into fuel oil, gasoline, and diesel. This process will need a lot of automation modules such as tank gauging system, online control, offline control, lab analysis, blend header analyzers, tank farm, additive control, DCS, and optimizer, after which it goes to the shipment. Before the final shipment, it must send for dispatch, to the tanks, tankers, and then to the pipelines, etc. This is a place where nonlinear programming and linear programming are done.
- Al-Qahtani, K. and Elkamel, A (2008). Multisite facility network integration design and coordination: An application to the refining industry, Computers & Chemical Engineering 32(10):2189-2202 DOI:10.1016/j. compchemeng. 2007.10.017, https://www.researchgate.net/publication/222195639_Multisite_facility_network_integration_design_and_coordination_An_application_to_the_refining_industry/citations, accessed on December 28, 2019.
- Méndez, C.A.,& Grossmann, I.E (2004). OPTIMIZATION TECHNIQUES FOR BLENDING AND SCHEDULING OF OIL-REFINERY OPERATIONS, https://www.semanticscholar.org/paper/OPTIMIZATION-TECHNIQUES-FOR-BLENDING-AND-SCHEDULING-M%C3%A9ndez-Grossmann/82fd8662b73bce6b4e33f65c9cabac40639e5db0, accessed on December 28, 2019.