Stephen B. Jaffe

Building useful models of complex reaction systems in petroleum refining

The identity and composition of petroleums countless molecular components is beyond the resolution of analytical techniques. Nonetheless, models of petroleum refining processes must have a molecular basis to be useful in predicting the molecular composition of products and their physical and quality properties required by industry specifications. Strategies for solving this seemingly intractable problem are presented.

Future directions in modeling the FCC process: An emphasis on product quality

Abstract A new generation FCC process model has been developed that describes fundamental cracking chemistry of FCC feedstocks using over 3000 molecular species. FCC reaction chemistry is specified using the methods of structure-oriented-lumping (SOL) with over 60 reaction rules. Application of these rules results in a complex network of over 30,000 elementary chemical reactions. Both monomolecular reactions such as cracking, isomerization, and cyclization and bimolecular reactions such as hydrogen transfer, coking, and disproportionation are included. The effects of thermal cracking and metals-catalyzed dehydrogenation are also represented. Reactivity relationships have been developed to reduce the number of parameters required to specify the kinetic model. The kinetic parameters are regressed from data spanning a wide range of FCC process conditions, feed compositions, and catalyst formulations. The riser kinetic model is coupled with equipment models for key FCC components to form an integrated FCC process model. The model accurately predicts product yields, product composition, and product quality over a wide range of FCC process conditions and is capable of predicting the complex non-linear phenomena exhibited in commercial FCC operations such as steady-state multiplicity. The model can be used to predict the quality of FCC products made from different crudes as well as to scope novel processing schemes for improving FCC product quality.

ACORN : APL to C on real numbers

a prototype APL to C compiler ( ACORN : APL to C On Real Numbers) was produced while investigating improved tools for solving numerically intensive problems on supercomputers. ACORN currently produces code which runs slower than hand-coded Cray FORTRAN, but we have identified the major performance bottlenecks, and believe we know how to remove them. Although created in a short time on a limited budget, and intended only as a proof of the feasibility of compiling APL for numerically intensive environments, ACORN has shown that straightforward compiled APL will be able to compete with hand-optimized FORTRAN in many common supercomputer applications.

Analysis of tracer experiments from commercial-scale trickle-bed reactors

Abstract Tracer experiments were conducted in commercial-scale trickle-bed reactors (catalytic hydrodesulfurization) to investigate uniformity of liquid and gas distribution in the catalyst bed. Carbon-14 tagged normal paraffins of 8, 16, and 32 carbon atoms were used as tracers to develop residence time distribution curves. Since dispersion caused by maldistribution is difficult to differentiate from spreading due to catalyst diffusion in a heterogeneous system, a model was developed to uncouple the effects. The flow model consisted of two zones with unequal mass fluxes, but otherwise ideal plug flow. The equations describing the transient tracer response were solved by numerical Laplace transform inversion and the split of mass flux to each zone was adjusted to match model output to the experimental data. The impact of the flow distribution on reactor performance was estimated by coupling a kinetic model to the flow model.

Topics for a second course in APL

After students complete the introductory APL course, after they become comfortable with the concepts and after they use it regularly in their work, then the problem of enhancing skills must be addressed. This paper presents a syllabus of three lectures which comprise part of a second course in APL. The purpose of Lecture I, MULTIDIMENSIONAL ARRAYS MADE EASY, is to help students deal with data stored in higher rank arrays. The idea of formulating such arrays frequently occurs naturally to users but they are reluctant to proceed because of the difficulty visualizing the arrangement. Six rules are presented to ease the way. Lecture II, SPECIAL USES OF THE GRADE AND SCAN, alerts students to approaches and techniques which would not typically occur to them. Interesting applications of the scan operator yield utilities which produce expansion vectors and indexing vectors based on field lengths. The grade and double grade form the core of other utilities to merge data and to affect table lookup. BUILDING BETTER USER/COMPUTER INTERFACES, Lecture III, is designed to facilitate delivering “closed” versions of “open” systems. Topics include a standard PROMPT utility, techniques in error trapping and a MENUDRIVEN MENUDRIVER.

Corporate application of APL—case studies (panel)

A melt-blended composition comprising poly(arylene sulfide) and hydrogenated conjugated diene/monovinyl aromatic block copolymer with the hydrogenated conjugated diene/monovinyl aromatic block copolymer present in an amount of up to about 5 weight percent of the total composition. A method for increasing the crystallinity of poly(arylene sulfide) by melt-blending therewith hydrogenated conjugated diene/monovinyl aromatic block copolymer in an amount up to about 5 weight percent of the total composition.

TPLAN a table driven planning system

A general broad based APL system has been developed for the storage, retrieval and analysis of financial planning or accounting type data. It is valuable for the class of problems arising from collecting, organizing and consolidating data according to set formulas or recipes. The system, called TPLAN for Table driven PLANning system, has as its foundation three principal tables: a table of line or account names; a table of formulas specifying how each account is to be calculated; and a table of data with one row for each account. TPLAN is an “Open” system that provides the user with a collection of conformable utility functions which operate on the principal tables. The package consists of 52 APL functions which contain an average of 3-4 lines of code.