Achieve Incentives For Tighter Fractionator Control

Bruce Jensen
Yokogawa Industrial Automation.
125 Huddleston Road
Peachtree City, GA. 30269

Phil Collins
Yokogawa Industrial Automation.
401 S. Dewey Ave, Suite 401
Bartlesville, Okla. 74005

Fractionator control has been published and discussed a good many times throughout the last 20 years. It is a subject so complex, detailed and industry wide that authors have written books concerning the subject.^{1, 2}

While an earlier article "Incentives For Tighter Fractionator Control" ³, provided the background for the justification of fractionator control projects and identified fractionators that make good candidates for advanced control, this discussion will concentrate on the application of fractionator control technology.

Successful application of control technology requires proper project management, control scheme flexibility, and design flexibility. A control application is not successful unless it is used by the plant operating personnel on a daily basis to help operate the tower in the most profitable manner.

We will address the subject of fractionator control in two phases. Phase one will concentrate on the project management and definition that is necessary to have a successful project at completion. Phase two will address some of the control techniques that are applicable to fractionators.

PHASE 1 - PROPERLY DEFINE THE OBJECTIVE

What seems to be the objective to control fractionation processes? Can they be the following?

1. To impress visiting V.I.P.'s with multi-colored CRT displays.
2. To keep up with the Jones' or that other refinery.
3. To reduce steam at the fractionator reboiler so that it can be later vented to the atmosphere.
4. To reduce the operator's workload.
5. To increase the operator's workload.
6. To test the I.Q. and annoyance-toleration level of process operators.
7. To upgrade low-priced product to high-priced product.
8. To allow engineers to gain experience on process units.
9. To create a topic on which to obtain patents and write papers.
10. To control each individual fractionator and process for maximum income.

There is some truth to most of the above answers, but, we hope that all can agree that answer #10 is preferred. But to achieve this, process objectives must be defined and understood.

In "The Top-Down Approach To Successful Process Control Projects" ⁴, six steps were listed for a control project:

1. Begin by defining the economic objectives.
2. Keep ownership of the project local.
3. Do not hesitate to employ assistance when needed.
4. Specify "what" instead of "how".
5. Do not ignore project staffing, as well as staffing for continued operation.
6. Never lose sight of the objective of economic performance.

DEFINE ECONOMIC OBJECTIVES

The definition of economic benefits that are measurable, significant and achievable is a major milestone. To reach this milestone, it may be necessary to employ the services of an advanced control consultant. The consultant will perform an economic feasibility study and document the results. Numerous questions must be asked and answers quantified. Such questions include:

• What are the economics?
• What are product and stream values?
• What is the throughput?
• What are the product specifications?
• What are the energy and operating costs?
• What is the performance of the fractionator with respect to meeting product specifications?
• Can better statistical control with a computer control system recover more product or save energy?
• What are the constraints that exist which limit the performance of the tower?

Economics of individual fractionators may continually change throughout the life of the plant. Product prices may determine that at one time energy savings was important, but that now recovery is paramount. The economic benefits of fractionator control include energy conservation, the shifting of less profitable components into more profitable products and increased throughput. Other benefits arise, including:

• Minimum disturbances propagated to downstream units.
• Minimum rework or recycle of off-spec products.
• More consistent product quality.

When minimization of fractionator utilities is an objective, the following guidelines are recommended:

• Operate the fractionator to produce minimum overseparation.
• Implement control to achieve composition control on all products of the fractionator.
• Be certain that the reduction in energy usage is reflected in the energy inflow to the production complex.
• Minimize energy waste from blending of overseparated products.

KEEP LOCAL OWNERSHIP

Site-wide and discipline-wide ownership is critical to the success of the project. An overly engineered and complicated control scheme to minimize impurities in the product streams, while the operator's orders are maximized throughput, is doomed from its conception. A project team whose members should include technical, operational, management and maintenance personnel should be formed. Though management may approve and provide support for a project and technical people design and implement the strategies, it must remain in the forefront of everyone's mind that operations must live with it.

EMPLOY ASSISTANCE IF NEEDED

As mentioned an advanced control contractor's services maybe required to perform the economic study. The consultant can also be a valuable member of the project team. Contractors provide new insight for old problems, an information base from past experiences, and mediation between plant disciplines. A preliminary design and specification study is a service that a contractor may provide to help identify project costs and to serve as the basis for competitive bidding.

Remember this is the project management and definition stage of the project and not the detailed design engineering phase. Proper definition of objectives and desired results in sufficient detail are paramount at this phase. Control strategies for fractionators typically control physical properties that relate to product composition, but direct control of product composition permits a more precise separation. During the definition phase keep in mind that a properly designed and implemented control system requires:

• Reliable measurement of process variables.
• Reliable indication of controlled product compositions.
• Control configuration consistent with the operating objectives and physical design of the fractionator.

SPECIFY "WHAT", NOT "HOW"

The following is a real life example of "what" instead of "how" in implementing a control design. A fractionator that separates a feed stream into a overhead stream and a bottoms product was identified for optimization. An advanced control strategy was designed and installed that included the installation of an analyzer to measure iso-butane content in the bottoms product. This strategy was based upon the measurement and control target of iso-butane in the bottoms product. At commissioning, it was discovered that the operator was required to control the tower to a RVP specification. Although attempts were made to correlate iso-butane with RVP, the control was never implemented. The analyzer provided valuable information but was not used in a closed-loop condition as designed and purchased. The "what" in this example should have been RVP instead of iso-butane specification. Although some investigation or "how to" is required to determine the desired method and inputs required to obtain the RVP measurement for control, the control techniques required for the "how to" of closed-loop control of RVP and iso- butane would be very similar if not identical.

STAFF FOR CONTINUED OPERATIONS

Advanced control applications are similar to other pieces of equipment installed in the plant where staffing is concerned. They all require upkeep, modification, support and other forms of maintenance. Unit modifications in equipment, feedstocks, etc. require adjustments in the control strategies. Support for the man-machine interfaces so that the operators can attend to the unit quickly and easily is necessary. This support will require the dedicated efforts of plant personnel following the commissioning of the application.

KEEP SIGHT OF THE OBJECTIVES

The axiom in the plant for operations has always been "Keep the units running." It is the axiom of the control engineer to "keep the units running profitably." With properly defined objectives, the project will provide economic benefits for years following commissioning.

PHASE 2- IMPLEMENT CONTROLS TO MEET THE OBJECTIVE

Once an economically justified fractionator control project with properly defined objectives has been approved, the task of designing the control strategies that will best accomplish the project objectives comes next. The control strategy must reflect the requirements of the particular fractionator as defined by the project objective. Each distillation tower is unique. Factors such as column design, operating objectives, operating environment, and instrumentation contribute to a fractionator's uniqueness. Thus, no standard or "off the shelf' control package meets the objective of every tower, but control strategies can employ similar techniques. In conjunction with control techniques, certain control philosophies are applicable. These philosophies include the isolation of internal liquid and vapor flows from external disturbances, proper use of on-line analyzers for composition control, and control of both the column's material and energy balances.

Control strategies applied to fractionators should be flexible in design. The flexibility to retain degrees of control, even if not implementable to the highest degree, is desirable. For a control strategy that consists of analyzer based control of a tower's reflux flow, it is desirable to have the ability to control the internal reflux flow in case of analyzer failure. Some of the money to be saved by implementation of internal reflux control can be realized even without the analyzer.

Control techniques include the traditional feedback and feedforward algorithms. They are prevalent in most all DCS, single and multiloop controllers. Inferential controls provides a means to control physical properties that may not be directly measurable. Other advanced techniques such as constraint networks and multivariable control are increasingly being used to great benefits.

FEEDBACK CONTROL TO ENFORCE SPECIFICATION

Feedback control involves the measurement of a variable and the use of this measurement to make a correction via a manipulated variable to maintain a specified target. The objective of feedback control for fractionators is to enforce specification targets for overhead and bottom products. Since these product specifications relate to the stream's composition the majority of the time it is advantageous to use online analyzers to provide the measurement for feedback control. Traditionally, a tray temperature has been used to infer product quality. However, variations in pressure, low sensitivity in many applications, and variations in off-key components causing errors limit the effectiveness of temperature controllers. Knowing a composition directly is certainly better than an inferred composition. In Phase 1, the emphasis was on proper definition of the control object. With the proper objective in mind, the next stumbling block for the feedback control is the proper measurement of the product composition.

The heart of the product composition chromatographic control concerns the quality and reliability of the measurement. The design of the sample system can be the deciding factor between success or failure of a fractionator control application. Analyzer sample system considerations include sample tap location, sample pressure, sample phase, sample cleanliness, and the ability to provide a timely representative sample of the desired stream. Every effort should be made to minimize cycle time of the analyzer, as process control and material balance often require conflicting analysis requirements.

FEEDFORWARD CONTROL TO MINIMIZE DISTURBANCES

While feedback control corrects a variable after it has been influenced by factor, feedforward control identifies a disturbance variable and uses its measurement to make a correction via a manipulated variable. Feed flow and reflux flow are the most commonly measured disturbance variables which are incorporated into feedforward control logic. Feed composition, feed temperature, and tower pressure are other disturbance variables sometimes used in feed forward control strategies. The strength of the feedforward technique consists of taking corrective action based upon the measured disturbance before it adversely affects the separation. In the case of feed composition, the strength of the feedforward technique is often negated by lag in the composition measurement signal. The majority of the time the composition has already adversely affect the tower separation by the time the feed composition analysis is measured by the on stream analyzer. The basis of the application of the feedforward technique involves the use of an elementary model dynamically tuned to approximate the distillation tower.

One of the most commonly used models for feedforward compensation of feed flow involves the bottoms to feed ratio (B/F). The bottoms product draw when ratioed with the feed rate is a function of the overhead and bottoms product target and at a given feed composition. The implementation of the bottoms to feed ratio typically involves the application of dynamic functions (dead time and lag) to the feed rate. The dynamically compensated feed rate is then multiplied by the desired bottoms-to-feed ratio to obtain the target bottoms rate.

Since models are only approximations of the real world, some inaccuracies exist. For this reason the bottoms-to-feed ratio target obtained by the feedforward logic is trimmed by the analysis based feedback control. In the majority of feedforward applications, feedforward techniques are applied to minimize disturbances in feedback controllers measurements. The value of feedforward control comes from addressing a variety of disturbances to a fractionator for which some form of compensation must be made in order to enforce product purity specifications. This requires that the advanced control must be able to measure, quantify the disturbance and react before the fractionator separation is adversely affected.

COMPUTED VARIABLES FOR FEEDBACK AND DISTURBANCE ISOLATION

Feedback control discussion concentrated on the control of a fractionator product specification based upon an analyzer as the measurement device. Occasionally the objective forces the feedback specification measurement to be other than what is supplied by a process chromatograph. In cases such as this product properties may be inferred from other process measurements. RVP (Reid Vapor Pressure) and stream boiling points are examples of inferred properties.

Computed process variables are also used in control strategies to help isolate the fractionator from environmental disturbances. Environmental disturbances can include ambient conditions that affect the tower's condensing system, feed enthalpy disturbances, steam system variations, changes in hot oil system reboilers and fuel gas fluctuations that influence furnace reboilers. Process variables can often be computed and included in the control logic to account for these external influences.

Internal reflux is a example of a computed variable. Reflux is a liquid stream that is pumped to the top tray of a tower after being obtained by condensing all or part of the vapor leaving the top tray of the fractionator. Internal reflux is a liquid stream that overflows the top tray internal to the fractionator and is the reflux that does the true separation. Internal reflux consists of the amount of external reflux plus the amount of vapors that are condensed internal to the column by the external reflux flow. The physical measurement of the internal reflux is not practical. However, internal reflux flow can be computed based upon several process measurements. For control purposes, the manipulated variable will be the external reflux flow to maintain a desired internal reflux flow. This relationship is expressed in the following equation:

Rx = Ri / (K1 * (1.0 + K2 * (To - Tr)))

Where:

Rx = External reflux flow rate setpoint
Ri = Desired internal reflux flow rate
To = Overhead vapor temperature
Tr = Reflux temperature
K's = Internal reflux constants

The internal reflux flow target that adjusts the rate of external reflux can be manipulated by an operator, a feedback analyzer, or feedforward compensation. The decreasing internal reflux flow results in reduced heat requirements, and, in turn, translates into reduced utility costs. Requirements for condenser capacity, reboiler capacity, and liquid and vapor flow capacity also decrease.

Heat duty is another computed variable. Most towers use steam as its reboiler heatant medium. Some towers are designed with reboilers that employ hot oil. With hot oil, other plant conditions may vary the source oil temperature or flow. Heat duty supplied by a hot oil system may be represented by the equation:

Q = Fo * c_p * (Ti - To)

Where:

Q = Heat input duty
Fo = Flow rate of oil
Ti = Temperature of hot oil entering reboiler
To = Temperature of hot oil exiting reboiler
c_p = Heat capacity of oil steam

Heat duty supplied by a steam reboiled system is represented by the equation:

Q = Fs * (ls + c_p * (Ti - To ))

Where:

Q = Heat input duty
Fs = Flow rate of steam
ls = Latent heat of vaporization at
Ti = Temperature of superheated steam entering reboiler
To = Temperature of condensed steam at its dew point
c_p = Heat capacity of steam

Of course, for saturated steam, only latent heat is exchanged.

ADVANCED TECHNIQUES FOR CONTROL

In today's competitive market, it becomes necessary to push equipment to operating limits in attempts to maximize product or to optimize the fractionation process. Methods beyond the feedforward, feedback, and computed process variable techniques are required. The techniques applied to maximization and optimization control schemes involve constraint or multivariable control concepts. The choice between constraint or multivariable depends upon factors such as: preference and familiarity, complexity of scheme, degree of optimization, hardware for application, and number of variables monitored and controlled by single strategy.

Constraint networks have been successfully applied in many applications. The application of constraints often involves the implementation of selector modules. Fractionators are frequently constrained by vapor load limits, equipment limits, condenser duties, reboiler duties. Current operating benefits are often limited by constraints. Attempts to maximize production of a product stream may drive the specification to become a constraint variable. Constraint applications compare the control objective with constraints, attempting to obtain the desired goal without violation of a constraint. Figure 1 is a control block diagram of a typical fractionator bottoms product control strategy.

Figure 1. Bottoms Product Control Block Diagram

In this example, the steam flow rate is manipulated in a feedforward manner with respect to the feed flow rate and the internal reflux flow rate. Feedback from one of three controllers biases this feedforward action. The primary feedback controller is an analyzer controller with constraint control provided by a temperature controller and a valve position controller. The temperature controller protects the fractionator from attaining a maximum temperature in the bottoms (This constraint can be altered to perform a minimum temperature constraint). The valve position controller constrains the operation whenever it is detected that either the reflux valve or the steam valve is wide open indicating a maximum heat or maximum cooling limit.

Multivariable control (MVC) can service many of the same constraint network control applications. In general, the more difficult the process is to control, the greater the expected benefits of multivariable control will be. Multivariable based control techniques deal with constraints and process lags. Highly interactive multivariable systems, where several loops are coupled, are candidates for multivariable control. Multivariable control strategies can also include economic optimum considerations. This technique requires the development of dynamic models based upon fractionator testing and data collection. Multivariable control applies the dynamic models and historical information to predict future fractionator characteristics. Predicted fractionator responses result in planned controller actions on the manipulated variables to minimize error for the dependent controlled variable, while considering constraints in the present and the future.

The control diagram shown in Figure 2. illustrates an example of an application of multivariable control.

Figure 2. Multivariable Control

In the example, two products and an impurity stream are separated using two towers. The objective is to control the composition of both products. As with all separations like this, the two composition control loops are coupled. That is, when any single control action is taken to control one composition, that action also affects the other composition. In this example the controlled variables are the two product compositions as measured by process analyzers. Disturbance variables include the feed flow rate. The steam to the first column and the temperature of the top of that column are the manipulated variables. A constraint variable is an internal flow as calculated from other tower temperatures and flows. The multivariable controller will take the appropriate steps to control both compositions, subject to the calculated constraint, by manipulating the two manipulated variables while accounting for the deadtime caused by the stripper.

CONCLUSIONS

Feedforward control techniques react to variations in a fractionator feed stream, predicts their effects, and takes corrective action before the tower is significantly affected. Feedback attempts to maintain the specification target of the tower based upon quality measurements upon exiting the tower. Frequently a combination of feedforward and feedback technique attempt to compensate for process deviations in the shortest time possible. This is accomplished by considering process dynamics (dead times and time lags), the non-linearities between separation efficiency and column loading, loop interactions, and process measurements. Control design can also address multiple controlled and manipulated variables.

Advanced control strategies in distillation controls have been integrated into plant and company-wide control systems for several years. Advanced strategies operating over and above basic regulatory systems provide the capabilities and flexibility necessary to keep the column and company profitable in times a changing objectives.

REFERENCES

1. Shinsky, F.G., "Distillation Control For Productivity and Energy Conservation, 2nd Ed.", McGraw-Hill, 1984.
2. Deshpande, P.B., "Distillation Dynamics and Control", Instrument Society of America, 1985.
3. Jensen, B.A., Collins, P.L., "Incentives For Tighter Fractionator Control", Control, Nov. 1990.
4. Christie, D.A., "The Top-Down Approach To Successful Process Control Projects", Control, Oct., 1989.
5. Stewart, W.S., Smith, D.E., and Griffin, D.E. "Distill with Composition Control", Hydrocarbon Processing, Feb. 1978.
6. Griffin D.E., "Fractionator Controls Can Save Energy", Oil & Gas Journal, Mar, 1978.
7. Buford, B.N., Bush, B.A., and Staten, H.W., "Computers Conserve Energy in NGL Fractionators", Oil & Gas Journal, Dec, 1978.
8. Bhullar, R.S., "Advanced Distillation Control", Control, May, 1990.

BIBLIOGRAPHIES

B.A Jensen, Engineer Support Manager with Yokogawa Industrial Automation, received his B.S and M.S. degrees in Chemical Engineering from Iowa State University. He has 21 years of experience applying automation solutions for the refining, petrochemical, and chemical industries, first with Applied Automation/Hartmann & Braun and then Johnson Yokogawa . He is a member of ISA and AIChE and holds 8 U.S. patents and more than 20 articles in process control.

P.L Collins is Sr. Advanced Applications Engineer with Yokogawa Industrial Automation. A graduate of Texas Tech University with a B.S. in Chemical Engineering, he began his career as a control engineer with Phillips Petroleum, and then with Applied Automation/Hartmann & Braun. Mr. Collins also has more than 20 years in the process controls arena.