Oct-2024

Evaluating tray efficiency impacts on column design

Steps to avoid undersizing or oversizing of columns along with the selection of property packages essential for the model to generate reliable tray traffic data.

Tek Sutikno
Fluor Enterprises

Article Summary

Commercial simulation programs are commonly used for generating tray traffic data for distillation column design specifications. The model developed to simulate a column can generate tray traffic data, such as flow rates and physical properties of vapour-to-trays and liquid from trays, along with the number of stages needed to meet the product specifications and the heat and mass balance.

The column diameter(s) and length are primarily specified or assessed based on the simulated tray traffic or loading data, which are also utilised to check hydraulic parameters, such as jet flood, downcomer flood, weir loading, and others resulting from the selected tray design and dimensions. To properly specify the distillation column design parameters, tray traffic data from the simulation model need to realistically reflect the flow rates, phase equilibrium, and properties of the fluids in the column.

Two simulation options, ideal (theoretical) stages and the actual number of trays, are generally available in commercial programs for column design simulation. In some, if not most instances, simulation models based on ideal stages are developed, and the simulated tray traffic data are directly utilised for specifying tray dimensions and the column diameter. The actual number of trays required to attain the simulated ideal stages is determined using a known or reference tray efficiency applicable to the particular fluid system. Different efficiencies can be specified for a column with multiple sections. The column height can be specified mainly based on the actual number of trays and tray spacing.

The other simulating option is specifying the actual number of trays and tray efficiency for generating the tray traffic data. The generated tray traffic data in this option, with reliable efficiency estimates, can be expected to be the same as those generated by the similar model using ideal or theoretical stages or tray efficiency of 100%. If the ideal stages (100% efficiency) and the actual stages with assigned efficiency provide significantly different tray traffic data, the specifications for tray and column will likely be underdesigned or overdesigned. This is especially the case when the traffic data selected for design are less or more than the real operation data.

Three-column simulation models are presented to evaluate the potential impacts of tray efficiency on column and tray design specifications.

Property packages and tray efficiency
For designing new columns, especially for first-of-its-kind types where design references and field data are not available, the designer will typically develop the simulation model to determine the number of ideal stages required to meet the product specification/separation requirements. One or more property packages available in the simulation program and applicable to the fluid system may be selected and used in the model.

The simulated theoretical stages will likely be unreliable or erroneous if the property package selected does not realistically represent the properties and vapour-liquid equilibrium of the fluid compositions. Property package usage guidelines are typically provided in commercial simulation programs and should be reviewed.

When the fluid compositions are not explicitly described in the guidelines, but the components are all included in the property package available for selection, the ideal stages simulated from the selected package may be best checked against ideal stages estimated using the short-cut approximation methods, the Fenske method as an example,1 especially for first-of-its-kind type design.

Tray efficiency is necessary to specify the number of actual trays from the simulated ideal stage model. For first-of-its-kind columns, without ‘goby’ or field data, tray efficiency inputs will need to be estimated and could involve uncertainty. In this case, the O’Connell correlation² may be considered for estimating the overall or sectional tray efficiencies.

The overall or sectional tray efficiency (OTE), defined as the ratio of the number of ideal stages to the actual number of trays, is commonly used in column simulation models using the actual column’s number of trays.

This overall efficiency option is available in most commercially available programs. However, some simulation programs like ProII Version 10.2 do not seem to directly provide the equivalent OTE feature but offers Murphree and other point efficiency options. Definitions of these options are described in the ProII reference manual,³ which also alerts inconsistent results from using the point efficiency options.

Typical tray efficiency data are available for commonly known columns and are mainly derived from relevant field operating data. Simulation models for these columns can be based on theoretical stages or the actual number of trays. For revamping existing columns, discussed in the next section, field operating data should be reviewed to finalise the selected property package and verify the estimated tray efficiencies as necessary.

Crude column
Crude columns are common in oil refining, and the applicable property packages or tray efficiencies are relatively well-known. An example of crude column simulation is presented to evaluate the tray loading data from using different property packages with or without tray efficiencies.

The first example model uses ProII (Version 10.2) to simulate an existing 41-tray crude column using 29 ideal stages (30 including the condenser). Three different property packages – Grayson-Streed (GS), Soave-Redlich-Kwong (SRK), and Peng-Robinson (PR) – are individually used in the separate versions of the model. These packages are commonly used for simulating crude columns, and the default option is used on each package. Input specifications, top and bottom pressures, and pump-around duties are the same in the column simulation model using the three packages.

Figure 1 shows the actual volumetric flow rates of the vapour to the ideal stages simulated separately using three different packages. As shown, the calculated vapour loading rates in ft3/sec from all packages are in good agreement, with less than 1% maximum deviation. Not shown in Figure 1, other calculated results, such as the condenser duties, deviate as much as 6%, with GS resulting in the lowest duty. For this example, the typical 10% exchanger design margin on flow rate and duty appears necessary to cover the duty variation from different property packages.

Liquid flow rates from ideal stages calculated with SRK and PR have good agreements with much less than 1% absolute differences, except for a 2.5% difference for the liquid flow rate from the ideal stage above the feed stage. However, relative to SRK or PR, GS results in about 20% maximum difference for liquid stage-to-stage flow rates. This difference becomes more than 10 times higher for liquid flow from the stage above the feed stage. Changing the VLE option in the GS package from GS VLE to SRK VLE reduces the difference to about 7% maximum. This shows that the property estimating options available within a package could result in large differences in tray loading data and need to be properly selected.