TECHNOLOGY Improved control strategies correct main fractionator operating problems

Scott W. Golden
Process Consulting Services Inc.
Dallas

Heat and mass balance control of refinery main fractionators can be improved through simple process design changes.

Metering flows of internal reflux streams improves unit operability and controllability. Modifying the process system design to measure small internal reflux flow is another inexpensive way to control main fractionators.

Three case histories show how simple design changes in refinery main fractionators can solve advanced control problems, thus changing product yields and improving refinery economics.

Process control

Process control of refinery main fractionators has become more sophisticated with the arrival of faster computers, new control logic, and the application of dynamic simulation. The impact of basic process design on distillation column controllability, however, often is overlooked.

The application of some basic design principles can make advanced control easier and less error prone.

Process control of multiple-draw refinery main fractionators is a manipulation of the column heat and material balance. Main fractionator process designs improve the ability of a sophisticated process control system to manipulate the heat and mass balance. This, in turn, greatly improves product quality and yield control.

Multiple-draw columns

Distillation column performance varies with the heat and material balance, column internal performance, and with the dynamic interaction between the two. Advanced process control strategies deal with this interaction.

Process control of any refinery main fractionator is dynamically compensated manipulation of the heat and mass balance.

Sophisticated process control programs have features that can infer product composition and "feed-forward" feed composition changes, and compensate for various unit disturbances. But in all cases, it is the column mass and energy balance that is adjusted to accomplish the control objectives.

Calculation of internal reflux rates in a main column requires a complete heat and mass balance envelope around that section of the column.

It is very difficult to subtract two or more large calculated numbers to calculate an accurate small one. A small change in the large heat and material balance control variables (external flows) will yield large changes in the smaller streams (internal flows).

Process designs that simplify the system to ease controllability often are overlooked. Generally, these process design modifications require little capital.

Column operation

The two broad categories of column internal types are trayed and packed columns. Although these two types of equipment operate on different principles, they each have operating windows.

To maintain column performance, distillation column vapor-liquid traffic must be maintained within the design limits of the column internals. The variables required to make this calculation are set by the process system design.

Modifying the process design to improve vapor-liquid loading calculations is the key. Internal vapor-liquid traffic calculations can be improved dramatically by measuring the internal liquid rate at appropriate points in the column.

Mass, heat balances

Internal reflux control is the key to maintaining fractionation in a given section of the column.

Calculating column internal reflux at any product draw location is a simple matter for an advanced control program. It may be necessary to input many plant data, however, depending on the number of heat removals. Both flow meter and temperature readings are required. And, usually, oil feed composition is unknown.

In theory, a refiner knows the column's feed composition based on crude assay data and crude blend ratios, or based on reactor outlet composition from sophisticated reactor models. But actual feed composition can vary greatly from predictions made with calculations based on typical-quality field data.

In comparison, metering low-liquid-rate streams is simple and inexpensive.

The mass balance above any point in a column is important, as it relates to the operability and control of the column. In a complex oil refinery, the mass balance is subject to many disturbances. Disturbances that affect column operation impose another level of control problem that makes correct process design even more important.

The following three cases illustrate how improved control can be used to solve main fractionator problems.

Case 1: Delayed coker

Fig. 1 (72351 bytes) is a schematic of a combination crude/delayed coker main fractionator. Combination tower process control is very difficult because column crude charge and delayed coker drum effluent composition and temperature can change independently or concurrently.

The column in Case 1 has three side-draw products. These products, and their specifications, are:

• Light gas oil (LGO), ASTM D-86 95 vol % point

• Intermediate gas oil (IGO), ASTM D-86 95 vol % point

• Heavy gas oil (HGO), ASTM D-1500 color.

The design of this combination column makes it very difficult to control the heat and mass balance during drum switches. LGO is withdrawn from an active tray. In principle, the tray does not leak and all the liquid is withdrawn from the column.

The problem is that neither the operators nor the advanced control system can maintain LGO and IGO product quality through coke drum switches.

Column vapor traffic decreases significantly during drum switches, and the valve tray used for LGO product draw-off leaks. (The LGO product draw on a coker should always be a total draw [seal welded] collector tray that has been water-tested during shutdown.)

During drum switches, the vapor rate and heat to the main fractionator decrease significantly. It takes time to heat the coke drum.

The major heat input to the column is declining. Heat removal, therefore, must be decreased by a corresponding amount.

Assuming coke-drum effluent temperature and composition are constant (which is not true), and that only the vapor rate changes, the heat balance must be adjusted. To maintain a constant reflux ratio in the LGO/IGO fractionation section, the HGO pumparound must be reduced.

Additionally, the mass balance in the column must be adjusted accordingly. During drum switches, however, the vapor rate, composition, and temperature change. This complicates the heat and mass balance adjustments required to maintain product specifications.

Fig. 2 (72704 bytes) shows modifications that have greatly simplified column operation during drum switches and other disturbances. The LGO and HGO subcooled pumpdown refluxes were changed to hot reflux. The quantity of hot reflux has increased because it is no longer subcooled.

The pumparound circulation rate has increased so that pumparound heat removal equals the pumparound and pumpdown heat removal from the original design.

The system has been simplified by reducing the number of internal heat removal zones in the column from four to two.

During drum switches, the HGO pumparound duty can be reduced to maintain the LGO and HGO product quality specification. The hot LGO pumpdown can be monitored and used as the control objective during drum switches.

Case 2: Crude unit

The heaviest distillate from an atmospheric crude unit, atmospheric gas oil (AGO), usually is feed for a fluid catalytic cracking (FCC) unit, a hydrocracker, or a hydrotreater. AGO quality is set by the downstream unit contaminant limitations.

The quality of the gas oil typically is inferred from ASTM D-1500 oil color. The low oil color of a high-quality AGO product (i.e., yellow) indicates little residue entrainment. Many refiners, however, produce black AGO.

An objective of the overall process control strategy on an atmospheric crude column is minimum overflash, consistent with acceptable AGO color.

Color is something an operator can inspect without laboratory analysis. An ASTM D-1500 color of 3.0 is dark yellow; therefore, the quality of the heaviest distillate can be viewed immediately by the operator.

A glass bottle held up to a light provides a means of evaluating AGO quality and wash-section column performance. Overflash requirements of 3-5% need to be maintained to meet a 3.0 color with a trayed column.

At some minimum liquid rate, trays will lose efficiency. Low tray efficiency is a result of liquid being blown off the tray when operating below the minimum liquid rate.

Overflash rates on an 80,000 b/d crude unit need to be between 2,400 and 4,000 b/d to produce AGO with low product color. Indirect control of the overflash rate requires the overflash to be adjusted within a range of only 1,600 b/d; otherwise the color will be poor or the overflash will be high, resulting in low yields.

With indirect reflux control on this column, the AGO product rate intermittently dropped almost to zero to maintain quality.

In many units, overflash is either metered or calculated to determine the appropriate amount of reflux below the heaviest distillate draw (Fig. 3)(45764 bytes). Minimizing the overflash is a noble goal for such a wash-section control scheme.

In practice, the use of either overflash measurement systems or on-line calculation systems are unnecessarily complicated, impractical, and produce less-than-optimum economic benefit. With indirect, internal reflux control in the wash section on this column, AGO quality was difficult to control.

The following equation represents the mass balance above the flash zone:

Flash zone vapor = Crude gas + Naphtha + Kerosine + Diesel + AGO + Overflash

On this unit, about 50 vol % of the crude is vaporized in the flash zone. When all the product draw rates are set by direct mass-balance control (flow control), the overflash rate closes the mass balance above the flash zone.

The overflash rate will increase or decrease as the flash-zone vapor rate changes. To show this, the mass balance equation can be rearranged to produce:

Overflash = Flash zone vapor 2 Crude gas 2 Naphtha 2 Kerosine 2 Diesel 2 AGO

Although the calculation is performed quickly and easily, it is not necessarily correct. Crude oil pseudocomponent breakdown and enthalpy algorithms are not exact.

Overflash may represent 5% or less of the heat input of the furnace; therefore, it is a small number calculated from several large numbers. In fact, these calculations often generate a negative overflash.

Atmospheric crude columns are subject to many disturbances that change the quantity of vapor generated in the flash zone. Furnace outlet temperature, crude rate, crude composition, stripping steam rate, and preflash drum operation all affect the flash-zone vapor rate.

Overflash meters that measure the quantity of material leaving the bottom tray have been installed on many units. The principle behind such an installation is that direct measurement of the overflash will allow better control. The practical problems with metering overflash are significant:1

• First and most important, the liquid drawn from the bottom tray comprises both overflash and entrainment. The quantity of entrainment varies with the amount of flash zone vapor, and with the flash zone mechanical design. The higher the flash-zone vapor rate, the higher the entrainment rate. The metered rate is not true overflash, and the percentage of overflash in the metered stream changes.

• Second, gravity flow meters generally "paint the chart." The metered reading is very erratic, because the meter is gravity flow and because the quantity of entrainment is erratic. The instability probably is caused by liquid inventory changes in the furnace and transfer line two-phase flow regime system, which, in turn, are caused by vertical flow instabilities.

• Third, the overflash line is subject to plugging during outages because the oil in the line is very heavy and cools off. Unless the line is drained and flushed, it will plug. The tube connections in the flow meter orifice flange also are subject to plugging for the same reason.

Overflash metering is unreliable from both a process and maintenance perspective; therefore, it should be avoided.

Fig. 4(43346 bytes) shows the modified process flow scheme. The AGO product draw from the column is a total draw. The reflux to the wash section is flow-controlled. Because this is a total draw, AGO product yield is a function of the total draw to the stripper, the wash oil rate, and the quantity of material stripped.

The AGO stripper bottoms level sets the AGO product rate. Thus, in Fig. 4,(43346 bytes) the AGO product, not the overflash, closes the material balance.

The control system is simplified and improved. AGO quality is adjusted by increasing or decreasing the metered wash-oil flow rate according to the observed AGO color.

Trays were replaced with packing, with the minimum reflux (wash oil rate) set by the packing liquid-distributor design. The packing produced good-quality AGO at overflash rates as low as 0.5%.

Variation in overflash should remain between 0.5 and 1.0% to produce high-quality AGO. The lower end of acceptable wetting rates are heavily dependent on cut point and tower internal design.

Case 3: FCC unit

One control objective on an FCC main fractionator is to minimize decant oil production.

FCC main fractionator decant oil is a component of residual fuel oil. Light cycle oil (LCO), on the other hand, is a blend component for the diesel pool. Incremental LCO yield has a large price differential over fuel oil, assuming it can be used as diesel oil blendstock.

Control of the main fractionator heat and material balance is the key to minimizing decant oil production for a given reactor yield.

Typical FCC main fractionators have four or five heat removal sections. Decant oil production represents less than 5 wt % on many units. The feed to the fractionator contains a large quantity of superheat; therefore, the bottom pumparound represents a very large heat removal.

The pumparound design duties for an FCC main fractionator processing 45,000 b/d of paraffinic, atmospheric residue are:

• Heavy naphtha, 32 MMBTU/hr

• LCO, 68 MMBTU/hr

• Heavy cycle oil (HCO), 40 MMBTU/hr

• Slurry, 90 MMBTU/hr.

The LCO reflux to the LCO/HCO fractionation section is internal overflow from the collector tray, which also feeds the LCO product stripper. The column heat balance is adjusted by reducing the bottom pumparound heat removal to allow more vapor to rise up the column and increase the reflux below the LCO draw.

A common problem, and one encountered on this unit, is that the LCO reflux is reduced to essentially zero for the purpose of increasing LCO yield. The LCO end point increases, and, in an extreme situation, the LCO will go off-color.

Increasing the heat removal at or above the LCO pumparound, at the expense of reducing pumparound duty, increases the reflux to the LCO/HCO fractionation section. To a point, increasing LCO/HCO reflux increases LCO recovery from the slurry. At higher reflux rates, however, the improvement in LCO yield becomes an issue of diminishing returns and no additional product can be recovered.

An evaluation of the unit mass balance and product distillations revealed some basic operating difficulties with a system that uses internal overflow of reflux below the LCO product draw (Fig. 5).(32967 bytes)

The decant oil product distillation on most units contains 10-30 vol % of 700 F. material. A unit with 10% 700 F. product yields essentially no recoverable LCO-boiling-range material without the installation of a slurry stripper.

For example, assume that the decant oil contains 10 vol % recoverable LCO-boiling-range material. If decant oil production is 2,700 b/d, this represents 270 b/d of recoverable oil. At an LCO product rate of 16,000 b/d, a 1.7% change in LCO yield represents the total recoverable LCO-boiling-range material in the slurry. This illustration shows that maximum LCO recovery requires good heat and material balance control.

Changing reflux below the LCO draw requires adjusting one of the two lowest pumparound duties, which, together, equal 130 MMBTU/

hr. Each 5 MMBTU/hr reduction in the bottom pumparound represents 1,600 b/d of internal reflux below the LCO draw.

It is impossible to control internal reflux flow using plant data for the calculations. An operator trying to adjust the heat balance manually will find the system almost impossible to control unless the internal reflux is maintained at high values.

Fig. 5 (32967 bytes) shows modifications that allow the internal reflux below the LCO to be metered. The heat balance can be adjusted directly to meet an optimum reflux rate without having to rely on complicated heat and mass balance calculations.

Heat and material balance adjustments via direct LCO/HCO reflux control require a new line and a flow control station that meters the LCO reflux below the draw tray.

Small adjustments in heat balance have a major impact on the reflux rate below the LCO draw. Directly metering this stream is the simplest and best way to simplify control of LCO recovery.

Reference

1.Golden, S.W., Lieberman, N.P., and Martin, G.R., "Correcting design errors can prevent coking in main fractionators," OGJ, Nov. 21, 1991.

The Author