A gas processing plant separates raw natural gas into methane for industrial and residential fuel and heavier hydrocarbon liquids for petrochemical feedstock. Refrigeration and a turbo-expander chill the inlet gas to cryogenic temperatures and the two products are separated in a demethanizer column. The primary product specification is the maximum volume of methane in the liquids. Colder operating temperatures will increase liquids recovery but at the risk of violating the specification. So careful regulation of heat input at the column reboiler is critical for product quality and quantity. Gas plants associated with oil production are generally operated to minimize back pressure on the oil fields. Thus the plant must tolerate significant swings in feed rate and composition while safely producing products that meet contract specifications. The data for this case study came from an Amoco Production Company plant in the Permian Basin region of Texas. This facility had a modern distributed control system (DCS) but a series of plant modifications left several instrumentation and equipment limitations that were difficult to handle with conventional control.
The Control Objectives
There are two control objectives. The first is to achieve maximum liquids recovery at product specifications. The second is to minimize the effects of inlet flow changes on column pressure to reduce alarm trips and flare emissions. Maximizing liquids recovery at spec is difficult due to the long cycle time of the bottoms product analyzer and the multiple lags in the reboiler design. Pressure control is difficult due to field operations (plunger lift pumps) that can cause a 10-20% feed rate swing in 10-20 minutes. Feed rate swings also frustrated previous attempts at dead time compensation for analyzer control. An additional internal source of pressure disturbances was the inlet and product compressors which, while automated, were not tied in to the DCS.
The QuickStudy™ Solution
QuickStudy was used to create dynamic models of the relationships between the process variables and control signals. The heat medium temperature and an upper tray temperature were used as feedforward variables to improve the temperature response. Plant inlet and outlet flowrates were used as feedforward variables to improve the pressure response. In all but one of the six blocks commissioned in this plant, the models were created off-line from historical data. In all cases, the models were created without bump tests due to the unique QuickStudy modeling technology. These models then provide the predictions used in the optimal control calculation. The sophisticated QuickStudy models allow the controller to regulate complex, non-linear processes with varying lags and deadtimes. The QuickStudy controller also has excellent response to unmeasured disturbances and load changes. When necessary, model adaptation can be invoked while controlling. Models can be saved as disk files and later restored as needed. This installation was implemented on a standard PC using Ethernet networking. A DDE data server was purchased from the DCS vendor and the existing DCS provided all physical input and output connections to the end devices. The DCS operator station was used as the QuickStudy operator station as well by configuring two new screens. This works because all QuickStudy internal parameters are accessible via OPC or DDE.
The QuickStudy controller improved both temperature and pressure control in the demethanizer. Overall project payout was one month. More specifically the improved control:
- allowed operation at a lower temperature increasing the volume of liquids recovered by 5%
- maintained methane volume consistently on spec
- allowed use of analyzer for supervisory control
- allowed use of increased product recovery vs. refrigeration cost as an economic trade-off
- reduced high pressure alarm trips by 80%
Ethane Recovery for QuickStudy and PID control plotted against column inlet temperature shows QuickStudy consistently provided better recovery with less refrigeration.
QuickStudy maintained tight control of product quality, meeting customer requirements of a maximum 1.5% by volume of Methane in the liquid ethane.