Refinery LP models

The database for these refinery LP models is a mixture of economic and technical inputs. Economic inputs include the availability and price of refinery raw materials, the variable cost of operating the individual process units, and the demand and price for refinery products. Technical inputs include refinery product specifications as well as the operating constraints, usage of equipment and utilities, product yields, and product properties for each process unit. 

An LP model for a petroleum refinery would consist of a number of modules for the different units (e.g., FCC, hydrocracker, fluid delayed coker, etc.). Each of these... 

The deterministic LP model was set upon GAM S and solved using C P LEX. Table 2.15 illustrates the computational results for the refinery Model. The planning model suggested producing 2000 t/day of gasoline, 625 t/day of naphtha, 1875 t/day of jet... 

Khogeer (2005) developed an LP model for multiple refinery coordination. He developed different scenarios to experiment with the effect of catastrophic failure and different environmental regulation changes on the refineries performance. This work was developed using commercial planning software (Aspen PIMS). In his study, there was no model representation of the refineries systems or clear simultaneous representation of optimization objective functions. Such an approach deprives the study of its generalities and limits the scope to a narrow application. Furthermore, no process integration or capacity expansions were considered. 

We demonstrate the implementation of the proposed stochastic model formulations on the refinery planning linear programming (LP) model explained in Chapter 2. The original single-objective LP model is first solved deterministically and is then reformulated with the addition of the stochastic dimension according to the four proposed formulations. The complete scenario representation of the prices, demands, and yields is provided in Table 6.2. 

The use of linear programming to optimize the flow of process streams through a petroleum refinery began in the mid-1950 s (Symonds, 1955 Manne, 1956). Now, almost twenty-five years later, it is safe to say that one half of U.S. refining capacity is represented by linear programming or LP models which are routinely optimized to schedule operations, evaluate feedstocks, and study new process configurations. 

The operation of the preprocessor is shown schematically in Figure 1. There are five basic steps o Select crude assay data o Build and report input tables o Generate crude data o Generate process data o Build and access LP data tables In the first step, the preprocessor accesses the disc file which contains all of Sun Petroleum Products Company s crude assays. The preprocessor extracts assay data for those crudes which the user has identified by card input as part of the base crude mix or to be made available to the LP model as an incremental refinery feed. The user can identify up to ten crudes any five of which can be designated as incremental. 

Phase 3. Integrate planning, scheduling, and real-time optimization by linking the optimization models to the refinery LP... 

Although the consistency of estimating crude oil relative values with gravity and sulfur-based valuations is still defended, there has been a clear tendency in the oil industry towards product-based valuations, considering the value of the products than can be obtained from a crude oil in a particular refinery. The relative value for the same crude oil may differ fi-om one refinery to another, and LP models are commonly used to maximize cmde oil value optimizing the product distribution. ... 

LP crude oil valuation is specific to a refinery and its facilities, markets and constraints. The combination of LP methodologies and refinery simulation models provides the required process and economic data to adequately evaluate crude oil feedstocks. Consistency between the quality of received crude oils and the assay data used with planning LP models is critical to guarantee the success in the optimization process. If crude oil quality suffers significant variations, a continuous assay update is required to take advantage of the benefits offered by supply-chain technologies. 

We overcome these problems by using the detailed FCC model developed in this work. We have shown that the FCC model can predict yields accurately for varying process conditions. To apply the FCC model into the refinery LP, we must first convert the large non-linear model in to a linear yield model. We can then use the coefficients from this generated linear yield model directly in the LP for the refinery. We show the process for generating the hnear yield coefficients in Figure 4.36. We have found that 4—5% is a reasonable value for CHANGE% (variable perturbation) for most of the important feed attributes in the FCC process. 

A refinery LP and linear unit model represents a set of linear correlations that predict yield given an average yield value and changes in the certain operating variables. In this section, we discuss how to apply the rigorous reforming model in the context of a hnear unit model. The key information for a linear model of a nonlinear process is the DELTA-BASE vector ... 

Transitioned the results from rigorous non-linear model to the LP model for the refinery... 

The FCC simulator program was converted to subroutine form a few years ago and incorporated into a nonlinear programming model representing a complex of process units in the Toledo refinery. It is this subroutine version which has been linked with the LP preprocessor. 

Refiners have typically solved this problem by using linear programming (LP) methods, which have been used extensively in refineries since 1950. Gary et al. [7] state that "a site-wide model of the refinery is therefore usually required to in order to properly determine refinery economics". 

---