gPROMS® Olefins is an advanced digital operations suite for online decision support of olefins plants.
A range of applications combine current and historical plant data with high-fidelity plant models, to provide monitoring of furnace coking, real-time soft sensing, furnace section optimization and forecasting for decision support.
How it works
The gPROMS Olefins suite is implemented as a set of digital process twins that execute within the automation system. These are described below.
Pre-requisite: keeping track of the current plant state
A key prerequisite in any digital operations application is to determine the current state of plant operation. For olefins plants, this is primarily a matter of accurately determining the current state and rate of coking in each furnace.
Daily monitoring of coking
The Coking Monitor uses all available relevant plant data for the run to date to estimate the current state of coking (i.e. the thickness of the coke layer along the furnace coil) and the rate of coking. This is typically done on a daily basis.
This provides important equipment health information as well as an up-to-date model for all of the digital process twin applications listed below.
gPROMS Olefins applications
With an up-to-date plant model from the Coking Monitor, it is possible to run further applications that can add significant value to operation on a daily basis.
Measuring yields in real time
The Cracking Monitor performs real-time soft-sensing of key furnace variables such as product yield and conversion, using the up-to-date model from the Coking Monitor in conjunction with real-time plant data.
The availability of reliable yield information makes it possible to implement conversion control, resulting in yield enhancements of up to 2%.
gPROMS Olefins Cracking Monitor seen here executing within a distributed control system
Run length prediction
The Run Length Predictor determines the remaining length of run for a steam cracking furnace under different operating scenarios. It uses the up-to-date model from the Coking Monitor to simulate operation until an end-of-run criterion such as maximum tube metal temperature (TMT) or maximum critical pressure ratio is reached.
This allows operators to screen and rank operating scenarios to maximize performance, and provides maintenance with information to optimize maintenance schedules.
Furnace section optimization
The Furnace Section Optimizer determines the optimal values of key furnace section decision variables, such as severity/conversion setting and feedstock allocation. The optimization is based on the current coking state of all furnaces, and takes into account feedstock availability and costs, product demands and prices, and material recycles. It presents operators with optimized set points and an estimate of the achievable benefit over current operation.
This maximizes the economic performance of the furnace section, and allows rapid re-optimization on equipment outage or other major disturbances.
Advanced operator decision support
The Decision Support tools allow operators to predict plant performance under specified scenarios, for steady-state performance or short-term dynamic performance over typical time horizons of minutes to hours, in order to gauge the effect of their actions on KPIs.
This helps to ensure that operators take informed operational decisions to maximize economic performance, and enhances operator understanding.
Full implementation
The architecture for a full implementation of gPROMS Olefins suite is shown below. The Coking Monitor, Cracking Monitor and Run Length Predictor twins are applied to each furnace (or each furnace cell, if necessary).
The Furnace Section Optimizer aggregates information across the furnace bank to optimize the section as a whole.
Offline model construction, analysis and maintenance
For offline model construction, analysis and maintenance, PSE supplies detailed olefins furnace models as an optional gPROMS Model Library for Olefins (gML Olefins) within gPROMS Process.
With gML Olefins, Process can be used for constructing and validating high-fidelity furnace section models, and, combined with other process Process libraries, full olefins plants.
