Mar-2006

Steady-state simulators are developing a dynamic personality

Process simulators have been used for years to design and model actual operation of all types of different plant processes.

John Dunlap, Crosstex Energy Services, LP
W G “Trey” Brown, Bryan Research and Engineering, Inc

Article Summary

The majority of process simulators provide a steady-state picture of plant operations and do not account for changes in inlet or ambient conditions. Steady-state simulators are very useful when first designing a plant under a certain set of conditions, or when developing a baseline for plant operation. These simulators are also much more affordable than the dynamic simulators that are available in today’s market. Unfortunately, plant operating conditions very seldom match design conditions and it is difficult for the operator to discern what effect the changing conditions have on his process without performing numerous simulations using trial and error and manual manipulation. Even then, these results are often suspect.

Crosstex Energy Services, LP and Bryan Research and Engineering, Inc undertook a project to model one of the Crosstex gas processing facilities using the ProMax simulation software. Using the program’s capabilities to rate the performance of various plant equipment, as it executes the simulation, and by utilising available parametric study features that allow numerous runs to be made consecutively, without interruption, the ProMax simulator was able to provide a series of snapshots that provided a realistic and accurate prediction of how the plant will respond to changes in conditions. While this is still a prediction of steady-state operation, the simulator has approached the dynamic threshold and only lacks the time derivative to cross over into that next dimension. This paper will show the steps that were taken to reach this point, the benefits it provided and how it might be used at other plant locations.

Introduction
Crosstex Energy Services, LP (Crosstex) is owner and operator of the Gregory gas plant facility in Southeast Texas. The plant processes third-party natural gas and delivers fractionated NGL products to pipeline and to trucks. The plant operation has changed over the years, with the inlet gas volumes declining slightly, but increasing in ethane-plus content. The plant had not been simulated under the new conditions and Crosstex was not convinced that they were operating the plant in the most efficient and productive manner. Using the ProMax simulation software, personnel from Crosstex and Bryan Research and Engineering, Inc (BR&E) undertook the task to model the entire plant facility and determine what could be done to improve operations and to find out what bottlenecks existed and where they were. This paper endeavours to show what steps were taken to simulate the plant under numerous sets of different conditions, what information was garnered from the simulation results and the impact that was realised when those simulation results were applied to actual plant operations.

Discussion
Crosstex and BR&E used a “stair-step” approach to construct a realistic simulation of the different plant processes. These steps included: 1) building the base plant model; 2) inclusion of equipment rating and sizing; 3) using the equipment ratings and sizing within the simulation itself to predict actual performance of the entire plant for a given set of conditions; and, finally, 4) inclusion of a multi-case “Scenario Tester” that allows the plant model to be run automatically any number of times under varying operating conditions, with the predicted results for each case displayed side-by-side for easy comparison. By using this defined methodology and plant design data in the construction of the plant simulation, it was hoped that a reliable and accurate predictive simulation model would be developed that would help optimise plant operation.

Building the base model
Using Plant PFDs and P&IDs, Crosstex and BR&E personnel built the base model of the Gregory plant on six different flowsheets (Figures 1, 2, 3, 4, 5 and 6) using the ProMax software package.

Each flowsheet was developed using the appropriate thermodynamic package for that particular process. The different flowsheets can interact by the use of cross-flowsheet connectors, which allow either process or energy streams to cross from one flowsheet to another. This permits the person using the simulator to view the impact a change to inlet parameters will have on a downstream system.

Rating and sizing process equipment
Once the base model was completed, actual plant operating data were collected and put into the appropriate locations within the simulation. The simulator provided final results that matched very well with actual operation including product compositions and flow, heat duties and horsepower requirements. At this point, Crosstex and BR&E believed they had a model that represented the Gregory plant well. The next step was to include information on the various equipment within the plant and rate the performance of that equipment.

Using equipment datasheets, the physical characteristics of the various heat exchangers, columns and separators were input into the rating sections of the simulator (Figures 7, 8, 9 and 10) to determine their performance and adequacy for that service and conditions. In order to confirm the simulator’s equipment ratings, the original plant design data was used in a process run. The results were remarkably accurate. The program’s predicted rating of each exchanger (including multi-pass brazed aluminum exchangers) was within +10/-5% of that predicted by the original equipment vendors.

The rating features provided information on potential areas of concern, such as calculated pressure drop through an exchanger, actual nozzle sizes versus recommended nozzle sizes and approach to flood within a tower. By having this information, Crosstex could easily determine where they were limited within the plant and determine where they might make changes. Also, by rating the exchangers and recognising that their rated performance (ie, 0% over design in the simulator) almost exactly matched actual operating performance, gave good indication that the rating program was accurate. This was very important in the next phase of building a flexible model that predicts performance under varying process conditions.

Implementing the predictive model
Now that the plant model had been built and all available equipment rated, a fully predictive plant model was developed using these ratings and incorporating them into the actual simulation run to be used to adjust parameters (such as duty or tower pressure) to meet specified (or measured) criteria, such as per cent over design in heat exchangers or pressure at the discharge of the booster compressor. This predictive capability is accomplished using a feature called “solvers”.