The principal objective of this study was to investigate process simulation and
optimization of an existing offshore natural gas process which needed profitability
improvements. Optimization was done using two alternative approaches: a global
approach (response surface methodology) or a local approach (successive quadratic
programming). The global approach was characterized by process performance at
selected case study points throughout the feasible operating region and made use of global
information, while the local approach was characterized by numerical iterative
computation driven by local information in the neighborhood of a single point in the
design space. A Box-Behnken design was used as a second-order response surface
design for the estimation of correlation between process simulator design variables and an
economic objective function, and the estimation model was then optimized. In the local
approach, a process simulator (ASPEN PLUS) with optimization capabilities was used.
From the investigation, three major design variables were identified that had
significant effects on the objective function of maximum product sales value. The three
variables were: the compression ratio of the Production Compressor, the heat duty of the
Gas/Gas Exchanger, and the compression ratio of the Expander. The results indicated that
profitability of the offshore natural gas process could be improved by greatly (about 60%)
increasing the production of raw condensate, even though this meant slightly lower (about
4%) sales of natural gas as a main product. The improvement found, however, was only
7% when the design variables were limited by currently installed equipment.
The global approach was found to converge more consistently because once the
global quadratic model was calculated, convergence to its unique optimum was simple.
On the other hand the local approach had non-unique termination points due to the small
effects of some design variables. The global approach provided better engineering insight
since the effects of each design variable could be easily calculated from the quadratic
model. However, the global approach was less efficient in terms of engineering manpower
because of the time required to identify the quadratic model. / Graduation date: 1995
| Identifer | oai:union.ndltd.org:ORGSU/oai:ir.library.oregonstate.edu:1957/35269 |
| Date | 23 September 1994 |
| Creators | Chaiwanakupt, Nopasit |
| Contributors | Levien, Keith L. |
| Source Sets | Oregon State University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
Page generated in 0.0015 seconds