翻译部分
英文原文
THE EFFECT OF MALDISTRIBUTION ON SEPARATION IN PACKED DISTILLATION COLUMNS
M. J. Lockett and J. F. Billingham
Praxair, Inc., P. O. Box 44, Tonawanda, NY 14151, USA ABSTRACT
For a packed bed in a distillation column, fmax is the maximum fractional liquid
maldistribution that can be tolerated in a parallel column model whilst still being able to achieve the design separation. It was previously shown that fmax can be easily calculated from a conventional column simulation output and it is a measure of the sensitivity of a packed bed to maldistribution.
In this paper, fmax is applied to examples taken from air separation and ethylbenzenestyrene distillation. Using air separation plant data, it is shown that design separation shortfalls can be correlated against fmax. When fmax <0.05, it is extremely difficult to achieve the design separation. A case study is given where fmax was increased from a very low value by splitting the bed, thereby achieving the design separation. Application of fmax to ethylbenzene-styrene distillation leads to the conclusion that two and four packed beds should be used in the rectifying and stripping sections, respectively. By varying the number of stages in the beds in the stripping section to equalize the sensitivity to maldistribution, it is shown that it may be possible to use only three beds in future designs. INTRODUCTION
Liquid or vapor maldistribution in packed distillation columns reduces the separation that is attained. Over the years, many workers have studied the problem in an attempt to predict the extent of maldistribution that occurs and to estimate its effect on separation. In a recent paper [1], the present authors introduced the concept of fmax to characterize the sensitivity of a packed bed to maldistribution. To understand the significance of fmax, consider the parallel column model
shown in Figure 6.1 that represents liquid maldistribution in a packed bed. One side of the packed bed, represented by the left column in Figure 6.1, has a liquid flow rate of (1+f)L and the other (1-f)L. Hence, f is a measure of the fractional liquid maldistribution. Because of the different operating line slopes in the two columns, the overall separation obtained from the two column system is less than is obtained if the liquid is equally distributed. A typical calculated result is shown in Figure 6.2 where the effective number of stages from the combined two-column system decreases as the maldistribution fraction f increases. Also shown on Figure 6.2 is the limiting case labeled fmax. It represents the maximum maldistribution that could possibly be allowed while still being able to achieve the required separation. For a fixed arbitrary value of the maldistribution fraction f (say 0.04), whether or not a given bed is sensitive to maldistribution depends only on the value of fmax for the bed. Thus, in Figure 6.2, a bed having 10 theoretical stages in each parallel column corresponds to the line that cuts the ordinate at N=10 when f=0. For this bed, fmax >0.1 from Figure 6.2 and if f=0.04, the combined bed actually delivers about 9.5 effective theoretical stages, so that the effect of maldistribution is negligible. On the other hand, a bed having 40 theoretical stages in each parallel column (for which fmax=0.038) provides only 24 effective stages if f=0.04. Thus, the designer need only determine the value of fmax for the desired separation that is required to characterize the sensitivity of the overall bed to maldistribution.
Figure 6.1 parallel column model Figure 6.2 Result from the parallel column model
?=1.538,y0?0.5, yn=0.95 xN?1?0.95
It was shown in reference [1] how fmax varies with the number of stages,relative volatility,
terminal concentrations and molar liquid to vapor ratio. The following equation was derived for a binary system from which fmax can easily be calculated.
The concentrations involved are shown on Figure 6.1, where in addition,
And
Note that it is not necessary to construct a parallel column model to determine fmax from equation (1) and the concentrations required are usually readily available from the output of a conventional column simulation program.
In the present paper, we apply the concept of fmax to two important industrial separations that are carried out by distillation using packed columns namely, air separation and ethylbenzene-styrene separation.
AIR SEPARATION
A typical three-column system used in air separation is shown in Figure6.3. Saturated air is fed to the bottom of the high pressure lower column 1.Condensation of nitrogen at the top of the lower column is used in a reboiler-condenser to generate oxygen vapor for the low pressure upper column 2. Nitrogen and oxygen products are taken from the top and bottom of the upper column, respectively. A side stream is taken from part way up the upper column that contains 5-15% molar argon. This stream is fed to the argon column 3 from which an argon stream is removed as an overhead product. Condensation at the top of the argon column is provided by boiling a liquid stream taken from the bottom of the high pressure lower column。
Figure 6.3 Three column system for air separation
The internals of the three distillation columns involved can consist of trays, packing (typically structured) or a combination of both. It is evident from Figure6.3 that a number of distinct column sections are involved, particularly in the upper column, because of the existence of multiple feeds and draws. In addition, when packing is the internal of choice, a given column section is often subdivided into two or more packed beds using liquid collectors and redistributors between the beds. In the argon column, when very pure argon is produced by distillation directly (oxygen <5 ppm), about 180 theoretical stages are needed. It is then usual to have upwards of 10 beds of packing in the argon column stacked one above the other.
Structured packing has been used in Praxair’s air separation columns for more than 15 years and there are many hundreds of packed beds in operation. The great majority of these beds operate at or better than design. However, there have been a comparatively small number of beds which, on start-up, gave a separation that was initially less than design. In most cases the discrepancy was minor and no action needed to be taken, but in a few others remedial action was necessary.
A survey was undertaken of those beds for which the separation performance was initially less than design. At the same time, the value of fmax was calculated for those beds using equation (1) in conjunction with an appropriate column simulation. The results are shown in Figure 6.4 as a plot of the measured theoretical stages as a percent of design versus fmax.
Figure 6.4 Air separation plant data for packed beds where a separation shortfall was observed
The data shown on Figure 6.4 can be arbitrarily divided into four bands. When fmax>0.20, the beds are insensitive to maldistribution. The small number of apparent theoretical stage shortfalls can be readily explained by composition sampling uncertainties between beds and by rounding the number of theoretical stages in a bed to an integer.
When 0.10 When 0.05 < fmax < 0.10, packed beds are quite sensitive to maldistribution and although the design separation can usually be obtained, the potential exists for a significant shortfall in the theoretical stages obtained unless the quality of distribution is very good. When fmax <0.05, the theoretical stage shortfall can be very substantial and it can be difficult to achieve the design separation even with very uniform distribution.The following describes a case history of a plant that, on start-up, did not perform satisfactorily because of poor performance of the packed bed labeled A in Figure 6.3. The concentration of nitrogen in the liquid entering this bed was 44% molar. The mixture was assumed to be a pseudo-binary, where oxygen and argon were lumped together as one component. The design concentration for nitrogen in the liquid leaving the bed was 17 ppm molar. The latter concentration is important because all the nitrogen that enters the argon column 3 in the feed from the upper column 2 ends