M. Teresa Sanchezllanes, Ana Lilia Perez G., Martha P. Martinez G., Enrique Aguilar-Rodriguez, Rodolfo del Rosal D.
Instituto Mexicano del Petroleo
Mexico City
Simple process analysis tools can be used to determine the best separation scheme for multi-component, wide-boiling hydrocarbon mixtures such as those found in refineries and petrochemical plants.
The selection of a separation sequence for fractionating hydrocarbon mixtures will determine the economics of the separation.
Separation economics depend mainly on investment for major equipment (columns and heat exchangers) and energy consumption. This relationship, together with the fact that, in most cases, many alternative schemes will be proposed, make it essential to find an optimum scheme that minimizes overall costs.
Practical solutions are found by applying heuristics-exploratory problem - solving techniques that eliminate alternatives without applying rigorous mathematical procedures.
These techniques have been applied to a case study. In the case study, a hydrocarbon mixture will be transported through a pipeline to a fractionation plant, where it will be separated into commercial products for distribution. The fractionation will consist of a simple train of distillation columns, the sequence of which will be defined by applying heuristic rules and determining the required thermal duties for each column.
The facility must separate ethane, propane and mixed butanes, natural gasoline (light straight-run, or LSR, gasoline), and condensate (heavy naphtha). The ethane will be delivered to an ethylene plant as a gaseous stream, the propane and butanes will be stored in cryogenic tanks, and the gasoline and heavy naphtha also will be stored.
HEURISTIC THEORY
For many years, design engineers have faced the problem of determining the most economical distillation sequence for separating multi-component mixtures. Many methods have been developed but most consider only the use of simple, sharp separators. In a sharp separator, each component entering the feed stream leaves in only one of the product streams.
During the last two decades, a large number of "synthesis" approaches have been proposed to identify optimal distillation sequences among a large number of alternative sequences. These approaches can be characterized as:
- Heuristic methodologies, which are based on engineering experience, or
- Algorithmic approaches, which employ optimization tools via mathematical programming.
It is possible, in a distillation sequence, for the condenser of one column to provide some or all of the heating energy required in the reboiler of a second column operating at a lower temperature than the first. If this type of heat integration between columns is allowed, the separation-sequence determination becomes much more difficult.
The number, n, of top and bottom products from n - 1 columns can be separated in several ways. In a direct method, the most volatile components are removed one by one as overhead in successive columns, the heaviest one being the bottom product of the last column.
The number of possible ways of separating components, however, grows dramatically with the number of products, increasing from 2 arrangements for three products to more than 100 for seven products (Table 1).
The Heaven equation, shown in Table 1, specifies the number of possible sequences, SR, for R-1 columns that separate R products.1 Given the fact that S1 = 1, values of SR can be determined.
It is a rather difficult task to examine all possible sequences, so it is recommended that the two or three most adequate schemes be selected by applying heuristic rules."
The most useful of these are:
Heuristic Rule A: Separations in which the relative volatility of the key components is close to 1.0 should be performed in the absence of non-key components, which occupy vapor and liquid space. The most difficult separations, consequently, should be reserved for last in the sequence.
Heuristic Rule B: Sequences that remove mixture components one by one in column overheads should be favored. Adding non-key components to the overhead increases vapor flow, which results in higher reboiler and condenser duties.
Heuristic Rule C: Sequences that yield a more-nearly equimolal split of the feed between the distillate and bottoms product should be favored.
With an equimolal split, the operating and equilibrium lines are better balanced, resulting in more-symmetric staging, greater reversibility, and lower energy requirements.
Heuristic Rule D: Separations involving the most constricted recovery of products should be reserved for last in the sequence.
FEEDSTOCKS, PRODUCTS
Table 2 presents some characteristics of the fractionation-plant feedstock. The plant will operate with a service factor equivalent to 330 days/year and a capacity of 55,308 b/d at 60 F. ambient temperature.
Product specifications are shown in Table 3.
SEQUENCE SELECTION
Given the multi-component feed stream of known conditions (composition, flowrate, temperature, and pressure) and the specified product streams, the "problem" is to find the most economical distillation sequence for achieving the required separations.
The basic assumptions that will be made to solve this problem are:
- Each distillation column represents a "simple" separation.
- Each distillation column performs "sharp" separations.
- Heat integration among the columns is not permitted. (This is assumed only for the sequence selection stage, because heat integration makes the problem much more difficult.)
Once all possible sequences have been defined, as shown in Fig. 1, the next step is to eliminate inappropriate sequences by applying heuristic rules.
The true boiling point values shown in Table 2 imply that ordinary distillation is the separation method of choice. The values also indicate that pressure will be higher in the first separator and lower in the second.
The pressure in each column depends on both the product specifications and the cooling medium in the condenser, which is a function of the nature and composition of the overhead mixture. For columns with an ethane-rich mixture as overhead, it is necessary to use a refrigerant for condensation.
The other condensers will operate using air at 104 F. because cooling-water availability is restricted at the plant location. Refrigerant cost, however, is an important factor to be considered in the sequence-selection process.
It is also important to take into account that sometimes these rules conflict because several sequences favor one rule over another. For every case study, the rules will influence different factors.
For example, the use of Heuristic Rule C leads to one sequence (Sequence No. 7 in Fig. 1). This sequence, however, involves two columns with ethane as overhead. This configuration requires the use of refrigerant in both, which increases operating costs.
For the case study, all the sequences that involve more than one column with ethane overhead must be eliminated. It follows that Sequence Nos. 6, 7, 8, 9, 10, 11, 12, 13, and 14 should be excluded. This reduces the number of alternatives from 14 to 5.
Applying Heuristic Rule A eliminates Sequence Nos. 1, 4, and 5, leaving only Sequences 2 and 3 as the most adequate to achieve the desired fractionation. Both sequences are depicted in Fig. 2.
ALTERNATIVE 1
The first processing alternative, shown in Fig. 2a, involves a separation train using the following sequence of splatters: [SEE FORMULA]
In this sequence, the feed stream enters the deethanizer (de-C2), where ethane is removed. The bottoms are sent to the de-C4 spotter, the overhead of which is a mixture of propane and butane. The de-C4 bottoms are the heavy fraction.
The overhead product is sent to the de-C3 splitter to obtain propane and butanes. The heavy fraction is separated in a de-C5 splitter to produce LSR gasoline and heavy naphtha.
ALTERNATIVE 2
The second processing alterative, shown in Fig. 2b, involves a [SEE FORMULA] sequence. This arrangement is the direct, conventional route for the fractionation.
In this sequence, the feed stream enters the deethanizer, where ethane is removed. The heavier fraction is then sent to the depropanizer (de-C3) to obtain propane, then to the debutanizer to produce butanes. The stream is then sent to the depentanizer (de-C5 splitter) to produce LSR as overhead and heavy naphtha as the bottom product.
PROCESS SIMULATION
Both sequences were simulated using Instituto Mexicano del Petroleo's general process simulator (Simproc). Simproc can produce mass and energy balances for refining and chemical processes in steady state, and calculate the thermophysical properties of all the streams in the process.
Once the feed stream has been characterized, a state equation must be selected for predicting liquid-vapor equilibria. Simproc recommended use of the Soave equation for the case study.4 The main parameters for each splitter were determined using shortcut distillation methods. The operating pressure of each column initially was obtained by using typical values for similar separations.
Using this information, a rigorous mass and energy balance was carried out for each column by the tridiagonal matrix technique (performed by the simulation tool).
At this point, the approach needs to be more accurately, "tuned" according to product specifications.
Several parameters, including reflux ratio (RR), operating pressure (P), feed tray (FT), and tray number (TN), thus were fitted (Table 4).
As can be observed in Fig. 2, the first splitter in both sequences has the same parameters. It is in the remainder of the train that the important differences exist.
At first glance, Alternative 1 (Fig. 2a) reveals an important disadvantage-its pressure values-because it is necessary to compress the de-C4 overhead to send this stream to the de-C3 splitter.
To confirm the selection of Alternative 2, the quantity QAT, representing the utility consumption of a single distillation task, was obtained (Table 5). The quantity Q can be represented by either reboiler or condenser duty for a given separation task. The quantity AT represents the temperature range over which heat is degraded to effect a given separation.
The temperature drop across a column is the difference between the reboiler and condenser temperatures. For both sequences, the quantities [SEE FORMULA]-the sum of the [SEE FORMULA] for the individual separations were determined.
The analysis results, shown in Table 5, indicate that Alternative 2 is the most efficient sequence for the case study.
REFERENCES
- King, C. J., Separation Processes, McGraw Hill Book Co., 1980.
- Henley, E. J., Equilibrium-Stage Separation Operations in Chemical Engineering, John Wiley & Sons Inc., 1981.
- Bojnowski, J. H., and Hanks, D. L., "Low-energy., separation processes," Chem. Eng., 86 (10), 1979, P. 67.
- Soave, G., "Equilibrium Constants from a Modified Redlich-Kwong Equation of State," Chem. Eng. Sci., Vol. 27, 1972, P. 1197.
Copyright 1993 Oil & Gas Journal. All Rights Reserved.
