William Y. Svrcek

A review of practical calculation methods for the viscosity of liquid hydrocarbons and their mixtures

This paper reviews methods for the prediction and correlation of Newtonian viscosity for pure components and mixtures of liquid hydrocarbons and petroleum fluids, which are suited for practical engineering use. The methods reviewed were chosen because they are well known and accepted or appear potentially promising. The methods are categorized as semi-theoretical or empirical and further distinguished as predictive or correlative. The applicability and average deviations for each method are discussed, with the recommended methods identified.

Phase Behaviour and Physical Property Measurements for VAPEX Solvents: Part I. Propane and Athabasca Bitumen

The saturation pressure and solubility of propane in Athabasca bitumen, as well as the liquid phase densities and viscosities, were measured for temperatures from 10 to 50°C. Equilibration proved challenging for this fluid mixture and required some experimental modifications that are discussed. Only liquid and liquid-vapour phase regions were observed at propane contents below 20 wt%. A second liquid phase appeared to have formed at higher propane contents. The saturation pressures, where only a single dense phase formed, ranged from 600 to 1,600 kPa and these were fitted with a modification to Raoults law. Viscosities less than 210 mPa.s were obtained at a propane content of 15.6 wt%. All of the viscosity data of the liquid phase were predicted from the propane and bitumen viscosities using the Lobe mixing rule.

Ultrapyrolysis of n-hexadecane in a novel micro-reactor

A novel reaction system, which provides better control compared with the tubular flow reactors of previous researchers, was employed to study the ultrapyrolysis kinetics of n-hexadecane, a gas oil model compound. Ultrapyrolysis, or ultra rapid pyrolysis, refers to thermal cracking under conditions of high temperature, very short reaction time, high heating rate and rapid product quench. Modern-day ‘millisecond’ furnaces, operating under such conditions, have demonstrated significantly higher ethylene yields due to improved selectivity. The automated micro-reaction system employed a Curie point pyrolyser to rapidly heat microgram n-hexadecane samples to high temperature. The Curie point phenomenon ensured that a reproducible and well-defined temperature was attained. A rapid direct quench system was used to stop the reactions and transfer the products to the analyser. n-Hexadecane was pyrolysed at 576–842 °C for 100–3200 ms. Peak ethylene production (28 wt%) was exhibited at ultrapyrolytic conditions of 842 °C and 500 ms. A first-order kinetic analysis performed on the pyrolysis data yielded an activation energy of 39.4 kcal mol−1.

Automatic design of neural network structures

Abstract Artificial neural networks are being extensively applied in many fields of science and engineering. Despite their wide range of applications and their flexibility, there is still no general framework or procedure through which the appropriate neural network for a specific task can be designed. The design of neural networks is still very dependent upon the designers experience. This is an obvious barrier to the wider application of neural networks. To mitigate this barrier methods have been developed to automate the design of neural networks. A new method for the auto-design of neural networks was developed, which is based on genetic algorithms (GA) and Lindenmayer Systems. The method is less computationally intensive than existing iterative design procedures, hence it can be applied to the automatic design of neural networks for complex processes. To evaluate the performance of the new design procedure, it was tested for the design of industry standard neural networks. The method was also applied to design neural networks to model the dynamics of a pH neutralization process and a CSTR reactor in which a set of nonlinear reactions takes place. The networks obtained by the new algorithm for these typical chemical processes was much simpler, yet more accurate than those designed by traditional methods.

EXTENSION OF THE SIMULTANEOUS-SOLUTION AND INSIDE-OUTSIDE ALGORITHMS TO DISTILLATION WITH CHEMICAL REACTIONS

Abstract The equations describing the steady state operation of distillation towers with chemical reactions were solved by two algorithms: (1) simultaneous solution after linearization; and (2) the inside-outside method, a tearing algorithm. The two algorithms were compared using a 13 tray ethyl acetate producing column. The inside-outside algorithm converged to the solution faster than the simultaneous-solution algorithm. The robustness of the two methods was similar, as measured by the size of the domain within which the initialized variables had to lie for convergence to take place. The temperature and concentration column profiles generated by the two methods were the same since identical thermodynamic and kinetic functions were used. The predicted column profiles matched available experimental data within 15%.

Fuzzy logic modeling of surface ozone concentrations

Abstract Due to the complex relationships and the necessity for forecasts in atmospheric studies, air pollution modeling is a task for which fuzzy logic methods are amicably suited. This research investigates the ability to predict surface ozone concentration with the use of an automated fuzzy logic method, termed modified learning from examples (MLFE). Hourly ozone concentrations during summer months in the city of Edmonton are predicted with MLFE models and the results are compared to models used by Environment Canada. The root mean square error, mean absolute error and scatter plots are used to compare the results of the MLFE, CHRONOS and CANFIS models. The newly developed model captures the trends in ozone concentrations, and based on the statistical comparisons, the MLFE consistently shows good agreement with the measured data. The MLFE model compares favourably with CHRONOS and CANFIS and is easier to implement.

Spectrophotometric Measurement of Asphaltene Concentration

Abstract Typically, when ultraviolet and visible absorbance of asphaltenes is employed to measure asphaltene concentration, linear calibrations of absorbance vs. asphaltene concentration are prepared from a sample of asphaltenes in a given solvent. This calibration is shown to be sensitive to: (a) the inorganic solids content of the asphaltenes; (b) physical–chemical differences between asphaltenes from different sources or extracted with different methods; and (c) selective adsorption of asphaltenes on liquid–liquid or solid–liquid interfaces. Calibration constants were determined at wavelengths of 288 and 800 nm for samples of Athabasca and Cold Lake asphaltenes obtained using different extraction methods, from precipitation experiments, and from adsorption experiments on water-in-hydrocarbon emulsions and on powdered metals. It was found that the inorganic solids content did not affect absorbance but the asphaltene concentrations must be corrected to a solids-free basis for accurate results. Calibration constants were found to correlate to the average associated molar masses of the asphaltenes. Therefore, any change in molar mass of asphaltenes during the course of an experiment may change the calibration constant. Partial precipitation and the selective adsorption of asphaltenes can lead to a change in the molar mass of asphaltenes left in solution. The corresponding change in the calibration constants can lead to errors of 5–25% in the estimated concentration.

A robust direct approach for calculating measurement error covariance matrix

Calculation of the measurement error covariance matrix is an essential requirement in data reconciliation methods. It is common practice to assume that the measurement errors are normal and have a known covariance matrix. A new robust method of measurement error covariance matrix calculation is presented. This approach directly treats the measured process variables but uses an M-estimator to reject the outliers and tunes the measured values for deviations from steady-state.

PLANTWIDE CONTROL STUDY OF A VINYL ACETATE MONOMER PROCESS DESIGN

ABSTRACT A simulation of a vinyl acetate monomer (VAM) process design was developed and compared with the work of Luyben and Tyreus (1998). Two incremental changes were made to the two main control substructures. Specifically, the two schemes focus on improving the liquid inventory system control and controllability of the azeotropic distillation column. The level control strategy was tested and found to produce a faster response with less oscillatory behavior. Two alternative control techniques for the azeotropic distillation column were tested, a feed-forward model predictive controller and a static feed-forward ratio controller. The model predictive controller results illustrated the large difference between the water composition analyzer sample time and the controller step size. The static feed-forward ratio controller showed excellent disturbance rejection of large feed flow variations to the azeotropic distillation column. Simulation results are presented to illustrate the effectiveness of the new control strategies.

Controllability of Heat Exchanger Networks

A heat exchanger network (HEN) is the heart of a heat-integrated plant, and classical synthesis methologies assume all design operating conditions are constant values and overlook controllability issues. The controllability measures presented in this paper are intended to assess the controllability of a HEN through a simple and comprehensive methodology consisting of five steps that allows an easy screening process to determine a more controllable HEN.