Jones D.S.J., Pujado P.R. Handbook of Petroleum Processing

Подождите немного. Документ загружается.

1327

1328

1329

Appendix D

Conversion factors

1330

A DICTIONARY OF TERMS AND EXPRESSIONS 1331

To Convert From To

Multiply By

Length

Feet

Inch

Statute Mile

Meter m

Millimeter m m

Kilometer

Area

Square inches

Square feet

Acres

Square millimeters

Square centimeters

Square meters

hectare

645.2

6.452

0.000645

0.0924

0.4047

Volume

Cubic inches

Cubic feet

Fluid ounces

Gallons (US)

Cubic millimeter

Cubic centimeter

Cubic meter

Milliliter

Liter

16387

16.387

0.00001639

0.0281

29.57

3.785

Mass

Pounds

Ton ( short)

Ton (long))

Kilogram kg

Metric ton tonne

0.4536

0.9072

1.016

Pressure

Pounds per square inch

Kilograms per sq meter

Bar

Pascal

Kilopascal

Pascal

Kilopascal

6895

6.895

9.807

100

Force

Pounds force

Kilogram force

Newton

4.448

9.807

Work

British thermal unit

Foot pound

Calorie

Joule

1055

1.356

4.186

Power

Btu / hour

Btu / sec

Horsepower

Watt

Kilowatt

0.293

1055

0.746

Flow Rate

Cubic feet per minute

Gallons (US) / minute

Barrels (US) oil

Cubic meters / minute

Liter / minute

Gallons Oil (US)

0.0283

3.785

Specific Energy. Latent Heat

BTU / Pound Joule / Kilogram 2326

Specific Heat, Specific

Entropy

BTU / Pound – deg F

Joule / Kilogram - Kelvin 4184

Miscellaneous

(US) Barrels per day

(US) Gallons

(US) Gallons / hour

Imperial gallons

1.75

0.8326

0.304

25.4

1.609

Appendix E

An example of an exercise using linear programming

1332

Linear programming aids decisions

on reﬁnery conﬁgurations

D. S. J. Jones

Fluor (England) Ltd.

and

J. N. Fisher

Bonner & Moore Associates, Inc., Houston, Texas

Mathematical modeling using linear programming can solve many problems associ-

ated with reﬁnery operation and planning. The technique, compared with the cost of

an error of judgment, represents only an insigniﬁcant ﬁnancial outlay.

It is becoming more and more evident that there is a deﬁnite economic incentive to

studying problems associated with reﬁnery planning by mathematical modeling using

linear programming.

Ever-increasing investment costs associated with the tightening of product speciﬁca-

tions, changing crude slates, alterations in the energy pattern in marketing countries,

and expanding petrochemical requirements, all make an error in decision judgment of

reﬁnery processing increasingly costly. Whereever this basic risk can be reduced by

modern mathematical techniques, the potential saving in capital or investment could

make the ﬁnancial outlay on such a study insigniﬁcant.

What follows is a description of a typical reﬁnery simulation study but this is only one

of an increasing number of problems that can be solved by mathematical modeling

using linear programming.

Wider impetus. The application of linear programming techniques using comput-

ers has long been used in the oil reﬁning industry for the development of planning

and operating policies. With the introduction of modern high-speed, large-capacity

computers, this technique grew considerably within the industry.

Brieﬂy, linear programming is the developing of linear sub models which mathemat-

ically describe many of the various operations within the industry, such as reﬁnery

processing, crude and product ﬂows, marketing demands. etc. These linear sub models

are inter-related to build up a complete mathematical model of the speciﬁc operation.

By the use of the computer, the equations within the model can be solved to optimize,

on a selective basis, the operation under study.

1333

1334 A DICTIONARY OF TERMS AND EXPRESSIONS

The growth of mathematical models using this technique provides management with a

means of making an increasing number of decisions which do have a calculable basis.

Thus, in many cases, the need for decisions based only on individual experience or

‘feel’, with its obvious inherent dangers, is being eliminated.

One such study illustrates a relatively simple reﬁnery problem application of mathe-

matical modeling and linear programming techniques.

Deﬁnition of problem

The client, wished to build a new reﬁnery. It had already executed a marketing survey

in the area and could specify quantity and quality, together with prices, of the products

which would meet its market requirements. Its management now had to decide the

economic optimum reﬁnery conﬁguration that would satisfy its crude and product

slate. At this stage only one type of crude was intended for the reﬁnery, and to

some extent this simpliﬁed the problem. However, to satisfy other considerations,

management required the solutions to the following premises:

The reﬁnery conﬁgurations, which would satisfy a MINIMUM investment, when

producing a high volume of gasoline with and without a low sulfur content limita-

tion of the fuel oil. All other products were to meet quality and quantity require-

ments.

The reﬁnery conﬁgurations which would give the MAXIMUM return on investment

to satisfy a ﬁxed crude throughput with no quantity restriction on the product slate,

and then to satisfy a limited restriction on the product slate with no limit to the

crude throughput.

Such a problem lends itself readily to linear programming and thus a reﬁnery simu-

lation model was developed to solve these two premises.

Process consideration

The ﬁrst step in constructing the model was to establish as many processing units as

could conceivably contribute to the solution of the problem. For instance, with such

large requirements of gasoline there would obviously be required a cracking unit of

some kind. Thus, the model included a reformer, ﬂuid cracker, hydrocracker, coker

and visbreaker. Some combination of these processes must satisfy the premises of the

problem.

Similarly, the lower sulfur content of the fuel oil would probably require some form

of residue treating. Thus two severity desulfurizers for both short and long residue

respectively were included together with a process for hydrocracking these residues.

A DICTIONARY OF TERMS AND EXPRESSIONS 1335

The many process units now included were then deﬁned in terms of feed streams,

product yields and quality, and operating costs, all based as a percentage on the feed

streams. This part of the study was the ﬁrst important step which required the expertise

of specialists. This data forms the basis for the rational solution to the problem, and

therefore it was necessary to be accurate and to augment prediction and theory with

realism and technical experience.

For instance, in arriving at the yields from the crude and vacuum units the effect of

fractionation on the product yield was considered. Realistic ASTM distillation gaps

were used that could be met by a commercial distillation unit.

In the ﬂuid catalytic cracking units a more sophisticated approach was needed to

correlate the yields from the many feedstocks which would be independent of thermo-

dynamic considerations. Here a base case feed yield data (in this case a straight run

waxy distillate) at a conversion of 75% using zeolite catalyst was used.

Yields from all other feedstocks (including those which had been hydrotreated) were

related by ﬁrst principle kenetic and thermodynamic considerations to the base case.

A short and simple computer program was used for this purpose, and it was also

possible to simulate the effect of changing the quantity of zeolite catalyst by this

means. The results of these computerized calculations were checked against existing

plant data before being incorporated into the study.

In other processes such as hydrocracking, hydrotreating, visbreaking, etc., care was

taken that only proven yield data or correlations were used.

Catalytic reforming yield data was obtained from a correlation which related yield

to severity for a basic naphthene and aromatic content of the feed-stock. A whole

range of severity operations from 95 to 105 O.N. (Research) clear was encompassed

in the study. Spot checks of the predicted yield by this method against actual yield

from an operating unit showed that the method was viable and acceptably accu-

rate.

Basic economic data

Having developed the physical yield structure of the ‘model’, the next step was to

complete the basic data by providing investment and maintenance costs.

There is, of course, a considerable wealth of plant cost data available to a contractor

from the projects he has completed over the years. However, there is always the need

to analyze these costs, and to review them in terms of up to date material and labor

cost changes.

1336 A DICTIONARY OF TERMS AND EXPRESSIONS

For this study, a large amount of cost data was statistically analyzed for each type of

plant. From this analysis, a base cost and an empirical expotential factor was developed

in order to relate a total investment cost to capacity in as realistic a way as possible.

This realtionship can be expressed mathematically in a non-linear form.

C = C

(T/T

)

where

C = Investment cost

= Base cost at a base throughput T

T = New throughput

K = An empirical constant

The inclusion of a non-linear form for investment costs in a linear program required

special consideration, and we shall see later how this was utilized.

In the models, many of the units considered were licensed processes, for which a

royalty would be paid. A value in terms of a paid up royalty in dollars per barrel of

throughput was included in the investment. Where chemicals and catalysts were used,

the ﬁrst inventory of these was also included as part of investment.

Chemicals and loss of catalyst was considered as an operating cost based on usage as

were utilities. Labor, a ﬁxed operating cost, wasincluded with the return on investment.

Maintenance cost was ﬁxed as a percentage of the total maintenance cost.

Model Development

At the same time as the process and cost data were being generated, the basic form

that the model would ﬁnally take was also being developed. This consisted of deﬁning

the various optional routes of each stream within the simulated reﬁnery model.

The optional routing of the streams was carefully selected. This selection had to satisfy

at least one of two requirements. Firstly, would such a routing actually contribute to

satisfying the product slate and the premises of the problem? Secondly, would such a

routing be feasible under actual operating conditions?

Just as a reﬁnery is described by the units of which it is comprised, so also was

the reﬁnery linear model described. Here, each processing unit was considered a

submodel in itself and these submodels were deﬁned by their process and economic

data.

These data were arranged in tabular from which were easily accessed and listed in

recognizable terminology. An example of a submodel tabulation as used in this study