Timothy A. Morrison
Setpoint Inc.
Houston
Diane Laflamme
Petromont Ltd.
Varennes, Quebec
Petromont Ltd. implemented advanced controls on a two-shell depropanizer, with the shells operating at significantly different pressures. The depropanizer project was part of a larger computer control project for its Olefins II ethylene plant in Varennes, Quebec.
Improving the control of the depropanizer tower was particularly important because the tower was difficult to operate and it limited plant throughput. The new advanced controls stabilized composition control and significantly reduced column loading.
These control improvements have reduced operating costs 12% and increased capacity by 9%.
PETROMONT'S COMPLEX
The Petromont Ltd. ethylene plant was constructed in 1963. Union Carbide Corp. and SGF, a provincial government corporation, are joint owners of Petromont.
The current plant capacity is 220,000 metric tons/year of ethylene.
Projects are planned to increase capacity to 265,000 metric tons/year by June 1991.
Petromont studied the feasibility of computer control and decided the best initial applications were controls on the furnaces, refrigeration systems, demethanizer, cold box, deethanizer, C2 splitter, and plantwide optimization.
The Phase 1 project was started in October 1985 and completed in July 1987. During Phase 1, Petromont also expanded its control room, switched to distributed control, and installed computer hardware.
In February 1988 the Phase 2 project was initiated. In Phase 2, advanced controls were installed on the depropanizer, debutanizer, acetylene recovery unit, and acetylene converter. Commissioning was completed in May 1989.
The Phase 3 project concerned acetylene conversion optimization. This phase was completed in November 1989,
The advanced control strategies discussed here, apply to the depropanizer in the Phase 2 project.
PROCESS OPERATION
The depropanizer column (Fig.1) performs a single distillation, but it is built in two stacked shells with the bottom shell operating at much lower pressure. The intent of the design is to reduce the polymer fouling frequently seen in this service by reducing the stripping shell operating temperature.
Unfortunately, the complexity of the design makes it difficult to operate the system properly. A large percentage of the rectifier bottom liquid flashes when it enters the stripper.
This flashed vapor is condensed along with stripper top vapor and pumped back to the rectifier sump. Thus, this portion of the column internal liquid flow does not continue down the column.
The stripper reflux line was added by Petromont, but the flow was normally not adjusted in response to process changes.
The stripper pressure is limited by the cooling water temperature, causing pressure control problems. The operators relieved this problem by operating at a higher rectifier control temperature.
Consequently, the column operated with an inordinate share of the load on the rectifier resulting in difficult control of the bottoms composition.
CONTROL OBJECTIVES
AND STRATEGIES
Because the depropanizer was a limit to plant capacity, the primary goal of the advanced controls was to maximize the capacity of the column.
This is achieved by meeting other basic control objectives:
- Stabilize column operation.
- Control both product compositions to their optimum setpoints. This eliminates excess reflux and boilup.
- Balance the internal flows between shells to make effective use of all the trays. This makes the separation easier.
- Reduce operating pressure. This increases relative volatility and reduces polymer formation.
The control strategies on the depropanizer include:
- Composition controls
- Hard target
- Soft target
- Flow balancing controls
- Rectifier internal vapor control
- Stripper internal liquid control
- Pressure minimization
Fig. 2 shows the depropanizer control strategies.
COMPOSITION CONTROL
The hard-target composition is always controlled to its setpoint, whether or not the column is constrained. The soft-target composition is controlled to its setpoint unless a constraint is encountered.
The hard-target composition is the butane product specification. There are specifications for maximum methyl acetylene content as well as total C3S.
The most limiting of these two compositions, typically the methyl acetylene, is selected for control. A separate deadtime-compensated, proportional-integral-derivative (PID) algorithm is provided for each composition.
The maximum of the outputs is selected as the target for the reboiler steam flow controller.
The steam flow is also adjusted by moves from a feed-forward controller to compensate for changes in condensate stripper bottoms and by an interactive decoupler based on changes in the calculated stripper top internal liquid.
The original column controls included a stripping temperature controller setting the steam flow. This was never reliable and the steam flow was set manually.
It would be desirable to use a composition model based on temperature for advanced control. However, during step testing to identify the model, no consistent correlation could be found.
It was also found that there was virtually no response of bottom composition to changes in reboiler steam flow unless the rectifier vapor module was in operation to link the two shells. Without the rectifier module,any change in stripper vapor was essentially recycled to the stripper via the rectifier sump level control.
The soft-target composition is the C4 content of the distillate. The value for the composition target is either set by the operator or by an energy optimizer.
The soft-target composition controller also uses a deadtime-compensated, PID algorithm. The composition controller sets the setpoint of an internal liquid/feed ratio controller. The output of the ratio controller is a target internal liquid flow. This target is checked in a constraint controller and modified if necessary to avoid violating column constraints. The constraints considered are:
- Maximum pressure controller output
- Maximum rectifier pressure drop
- Maximum rectifier reboiler steam controller output
- Maximum approach to rectifier flooding
- Maximum reflux flow controller output
- Maximum stripper reboiler steam controller output
Note that all the constraints except of the last are related to the rectifier.
SHELL BALANCING
The column is constructed in two shells but performs a single separation. To obtain maximum separation efficiency, the internal flow profiles should be continuous throughout the column as in a single shell column.
In other words, the internal flows at the bottom of the upper shell should be equal to the corresponding flows in the top of the lower shell (Fig. 3). Together, the rectifier internal vapor and stripper internal liquid control cascades perform this balancing function.
These can also be considered to be optimizing controls because separation efficiency is maximized when the internal flows in the two shells are in balance. Secondarily, the internal vapor constraint controller shifts loads between shells as required to relieve constraints.
The soft-target constraint controller takes action only when the column is so constrained that load shifting cannot relieve it.
In this way, the soft-target composition can be maintained up to a higher column loading.
The optimum rectifier bottom vapor is equal to the condensed liquid flow from the stripper accumulator to the rectifier because, when the stripper is properly refluxed, this flow is equal to the top internal vapor. Due to the swings in this flow from the accumulator level controller, using this value directly as the target rectifier vapor causes instability.
This is primarily due to the small size of the accumulator which requires fast tuning of the level controller. The solution was to set the rectifier vapor target equal to the stripper bottoms vapor plus a bias.
A target bias is calculated to make the rectifier vapor equal to the average stripper distillate flow. This target bias, then, represents the deviation of actual column operation from constant molal overflow.
The bias is filtered toward this target value, thus continuously moving the rectifier vapor flow toward the optimum.
The rectifier vapor target is input to a constraint controller. The purpose of the constraint controller is to shift load between shells to relieve constraints before the soft-target composition is affected.
Constraints in this controller, duplicated in the soft target, have lower limits to ensure that this controller acts first. If any constraint action is taken, of course, the shell internal flows will no longer be balanced.
Also, during such constraint action, the filtering of the bias is suspended. The constraints considered are:
- Maximum stripper pressure
- Maximum stripper pressure drop
- Maximum stripper reboiler steam controller output
- Maximum approach to stripper flooding
- Minimum rectifier temperature
- Maximum rectifier level controller output
- Maximum rectifier pressure controller output
- Maximum rectifier pressure drop
- Maximum rectifier reboiler steam controller output
- Maximum approach to rectifier flooding
- Maximum rectifier reflux controller output
- Maximum rectifier temperature
The rectifier temperature maximum and minimum constraints are provided primarily for operator acceptance because this temperature is a controlled variable when the advanced controls are not in use. The approach to flooding used in the constraint controllers is calculated using standard correlations, but is adjusted by a fouling factor to account for polymer buildup on the trays over time.
This fouling factor is calculated from a measured pressure drop, and a pressure drop is calculated for clean trays. The operator has the option to use a manually entered pressure drop instead of the measured value, and the fouling factor adjustment may be turned off if desired.
The constraint controller output sets the setpoint for the rectifier internal vapor controller.
The calculation of the internal vapor includes the effect of the subcooled stripper distillate.
Because the stripper is run at roughly half the pressure of the rectifier, it is subcooled relative to rectifier bottoms liquid, and the sensible heat effect is a significant portion of the rectifier reboiler duty.
The optimum stripping shell top liquid flow (excluding the feed from the condensate stripper) is equal to the rectifier bottom flow. Approximately 24% of the rectifier bottoms flashes to vapor at stripper operating pressure.
If the stripper reflux is held constant, as was the case before computer control, this flashed vapor essentially returns to the rectifier via the stripper distillate. The stripper internal liquid control cascade ensures that changes in rectifier bottom internal liquid flow are reflected in the stripper top internal liquid.
This is accomplished by adjusting the stripper reflux to hold a target internal liquid.
The target for the stripper internal liquid controller is a calculated rectifier bottoms flow.
The rectifier bottoms flow is calculated by material balance around the sump. A bias is included in the target calculation to allow bumpless transfer.
This bias is filtered to zero, as long as no rectifier vapor constraints are active, to bring the rectifier bottom and stripper top internal liquid flows into balance. Heat and material balances around the stripper top section are used to calculate the current stripper internal liquid minus the stripper feed. This is the controlled value for the internal liquid controller.
PRESSURE MINIMIZATION
As pressure decreases, the relative volatility increases making separation easier. An added advantage of reducing the pressure in the rectifier is the possibility of reducing the operating temperature. This reduces the rate of polymer fouling on the lower rectifier trays.
The pressure minimization strategy slowly lowers the pressure controller setpoint while checking for constraints. The goal is to run at one of the constraint limits.
The constraints considered in the rectifier pressure minimizer are:
- Maximum accumulator level controller output
- Minimum condenser subcooling-
- Maximum pressure con-troller output
The pressure controller output limit is set lower than the equivalent limit in the rectifier vapor constraint controller, which is in turn lower than the one in the reflux constraint controller. The subcooling constraint is typically the most limiting.
The constraints considered in the stripper pressure minimizer are:
- Maximum pressure controller output
- Maximum bottoms flow controller output
- Maximum accumulator level controller output
The major column limitation before the control project was that the stripper pressure control valve was usually full open. This was thought to be due to insufficient condenser capacity.
However, during commissioning it was found that, often, the reflux vapor pressure exceeded the pressure controller setpoint. The pressure controller minimum setpoint clamp was set above the old operating setpoint to help reduce the problem.
The stripping shell top temperature did not increase, because the average C3 content increased. This also allows the rectifier to run cooler.
RESULTS
The advanced computer control strategies have improved product quality, increased column capacity and improved energy efficiency. The composition controls maintain operations within product specifications, which was very difficult with manual control.
Balancing load between the two shells has resulted in a capacity increase of 9%. Energy consumption has been reduced 12%.
The rectifier operating pressure has been reduced about 10% and is now typically at the bottom of the instrument range. The rectifier control temperature has been reduced 8 C. on average. This is expected to reduce the polymer fouling on the rectifier trays.
ACKNOWLEDGEMENT
The authors would like to thank James Jones, now with Amoco, and Gerald Lebrun and Dinh Doan of Petromont for their contributions during the execution of this project and the preparation of this paper.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.
