Sérgio G. Oliveira

Real Time Fouling Diagnosis and Heat Exchanger Performance

The issue of fouling in preheat trains of crude oil distillation units in Petrobrass refineries is a major concern—especially now, as heavier Brazilian crudes with higher asphaltene content are being refined. As the efficiency of the preheat train plays an important role in the energy consumption of a distillation unit, its performance must be tracked as precisely as possible in order to identify operational problems. This work describes an online heat exchanger performance evaluation system based on rigorous simulation of the equipment in order to predict both the operational and clean overall heat transfer coefficient. A real-time comparison between these two values indicates the actual performance of the heat exchanger and of the preheat train. The use of a rigorous process simulator (Petrox from Petrobras) together with a rigorous calculation of the global heat transfer coefficient (using the program Xist from HTRI) allows one to consider aspects that are not usually taken into account in this kind of evaluation. These aspects include crude vaporization after the desalters and variations of crude and products composition with the distillation unit run. The system is being implemented at the biggest Petrobras refinery (360,000 bpd) in a 25 heat exchanger preheat train.

Parameter Estimation of Fouling Models in Crude Preheat Trains

Several fouling mitigation techniques depend on the capacity of predicting fouling rates. Therefore, the identification of accurate fouling rate models is an important task. Crude fouling rates are usually evaluated through empirical or semiempirical models. In both alternatives, there are parameters that must be determined through laboratory or process data. In this context, the article presents an analysis of the parameter estimation problem involving fouling rate models. A proposed procedure for addressing this problem is described through the development of a computational routine called HEATMODEL. An important aspect of this study is focused on the obstacles associated to the search for the optimal set of parameters of the Ebert and Panchal models and its variants. This optimization problem may present some particularities that complicate the utilization of traditional algorithms. In the article, the performance of a conventional optimization algorithm (Simplex) is compared with a more modern numerical technique (a hybrid genetic algorithm) using real data from a Brazilian refinery. The results indicated that, due to the complexity of the parameter estimation problem, the Simplex method may be trapped in poor local optima, thus indicating the importance of the utilization of global optimization techniques for this problem.

Fouling Management in Crude Oil Preheat Trains Through Stream Split Optimization

Abstract In general, the first main step of petroleum refining consists in the distillation of the crude oil stream. In order to provide adequate fractionation, the crude stream must be fed in the atmospheric distillation column at about 380°C. Aiming to reduce energy consumption, heat from hot streams of side products and pumparounds is transferred to the crude stream in a heat integration scheme, called crude preheat train. The final heating of the crude stream is executed in a furnace. However, during the operation of the preheat train; the thermal effectiveness of the heat exchangers diminishes due to fouling and, as a consequence, fuel costs increases. The large volumes of crude oil processed and the scenario of crescent energy prices justify the importance of this problem for the oil companies. Seeking to provide a solution to reduce the impact of this problem, this paper presents the exploration of stream splitting in crude preheat trains composed by several parallel branches. In this case, each branch may present different fouling levels, which allows the exploration of different distributions of the stream flow rates along the system, through a proper optimization algorithm. This optimization algorithm searches the set of stream splitters related to the maximization of the final temperature of the crude preheat train. A mathematical model of the preheat train works coupled to the optimization method. An important focus of this paper is to explore the introduction of constraints in order to guarantee feasible operating solutions, i.e., the optimum solution must attend different operational aspects related to bounds on fluid flow velocities and heat exchanger capacities. The performance of the proposed approach is illustrated through a typical example of a petroleum refinery.

Thermohydraulic Simulation of Heat Exchanger Networks Subjected to Fouling

Abstract Fouling is a complex problem which affects thermal equipment. The deposits over the heat exchanger surface diminish the thermal effectiveness and increase flow resistances. The consequences of fouling are related to important economic penalties, such as, more expensive equipment with larger thermal surfaces, higher energy consumption costs (heat/power), shutdown costs for exchanger cleaning, fluid treatment costs, etc. Despite the considerable research efforts about this subject, the engineering practice is still based on traditional approaches, e.g., fouling factors. Several recent papers have been focused on the schedule optimization of heat exchanger cleaning during a certain time horizon. In this case, optimization algorithms demand the availability of a heat exchanger network model in order to predict the network behavior. Aiming to this important problem, this paper presents the modeling and simulation of heat exchanger networks subjected to fouling. A fundamental aspect of the proposed simulation scheme involves the capacity to represent the network behavior considering heat transfer and fluid flow aspects simultaneously. For a given set of fouling models, it is possible to predict the temperature and flow rate along a heat exchanger network during a certain time span in the future. The model is represented by a set of matrix equations, where the network structure is parameterized using graph theory concepts. The potentiality of the proposed simulation scheme is illustrated through a typical example of an industrial heat exchanger network.