(37e) Simultaneous OPTIMIZATION of Feedstock Composition and Operating Conditions for A SYSTEM of Crude Oil Distillation Units

Keywords: Crude Distillation Unit System, Optimization, Energy Integration, Metamodels, NLP, GAMS, PRO/II, Latin Hipercube Sample -LHS

Crude Oil Distillation Units (CDU) have relatively small operational windows due to the limitations set by restrictions like hydraulic capacities to pump crude oil and draw-off products, products specifications, hydraulic and separation capacities of the towers and thermal integration with crude oil preheat exchangers train. Techniques like rigorous modeling, linear programming and unit specific empirical correlations have been reported to optimize CDU [1,2,3]. In all these papers, CDU operation optimization deals only with the atmospheric tower and doesn't take into account the thermal integration with preheat exchangers train. Despite this simplification accelerates model development and the model is easy to use, it can produce misleading conclusions and even operational conditions not allowed in the industrial unit [4].

The objective of this paper is to present the development of an optimization model for a CDU system which belonging to a Colombian refinery of ECOPETROL S.A. involving the typical restrictions (flow according to pipeline capacity, pumps, distillation columns, etc) and a restriction that has not been included in bibliographic reports for this type of models: the energy integration of streams from Atmospheric Distillation Towers (ADT) and Vacuum Distillation Towers (VDT) with the heat exchanger networks for crude pre-heating.

The optimization model was NLP, maximizing the system profit. This model was implemented in GAMSide 22.2 using the CONOPT solver and it found new operational points with better economic results than those obtained with the normal operation in the real plants. The model calculated simultaneously Optimum Feedstock Composition and all Operating Conditions of 3 ADTs, besides the yields and properties of atmospheric products, additionally to temperatures and duties of 27 Crude Oil exchangers.

The optimization involved crude oils blending, distillation processes, final products blending and plant restrictions like feedstock availability, market demand, product qualities, equipment integrity and energy integration. Crude oils blending considered five extra-heavy crudes (API gravity<13); crude pre-heating trains or HENs were represented with mass and energy balances and design equation for each heat exchanger; Jet-1A and Diesel were the final products prepared. Non-linear semi-rigorous models (Metamodels) of each atmospheric distillation tower were constructed. The methodology [13,14] used to develop the Metamodels included: 1) Selection of input variables, 2) Design of Experiments (Latin Hypercube Sample -LHS), 3) Generation of data using PRO/II models, 4) Parameterization of Metamodels using multivariate stepwise regression in Matlab, 5) Validation and evaluation of their significance.

The optimal profit found by the NLP model was over 20% respect to initial estimate, using low time execution for determining the solution in GAMS (between 26 and 62 seconds).

The extra-heavy crude oils in the CDUs optimum charges required more severe operating conditions for extracting the most valuable components by distillation, requiring more duty in the furnaces to increase the inlet temperature to the towers, a larger amount of stripping steam flows along the bottom and a lower top pressure of the towers.

Metamodels (second degree polynomial) showed an excellent performance to calculate flow rates, properties characterization and temperatures of atmospheric streams with average error below 4%, 3% and 1% respectively

The existence of the heat exchangers in the CDUs system NLP model moved the optimum to different operating zones compared to those found if the towers were used alone; the new restriction in the problem ensured safety operation within the design capabilities of the industrial plants.