Dynamic Simulator-Based Soft Sensing at a Hydrocarbon Purification Plant
A section of naphtha (lighter than petrol hydrocarbon fluid) sweetening plant demonstrates dynamic soft sensing efficacy. The process involves purifying naphtha of sulfur naturally present in the mixture by decomposing the complex sulfur compounds present in the mix in a hydrogen-rich environment into H2S (hydrogen sulfide), followed by stripping off H2S to generate sweetened naphtha.
Our first step was to generate a high fidelity dynamic simulation model of the process plant on a commercial dynamic simulator platform validated against the actual process plant data. Next, the model was tuned to match existing plant operating conditions at a steady state. The model was then linked with process instrumentation and distributed control-system feed using a compatible software interface. The dynamic simulator then began predicting all the process parameters that were not physically measured in the plant and the equipment operating efficiency indicators.
The above figure illustrates how some of the soft-sensed process parameters and estimated equipment performance indicators are computed by the simulation model. The critical parameters soft sensed by the model, and their relevance in the plant operations, include:
● Dissolved oxygen concentration in naphtha liquid exit oxygen stripper: A higher value than 5 ppm indicates a fouling threat for the downstream P-heater exchanger train.
● Oxygen stripper overheads gas flow rate: A higher percentage means more gas being fed to the fuel gas system, leading to the possibility of flaring of excess gas that is not utilized as a fuel in the process.
● Oxygen content in flue gas exit naphtha vaporizer: A higher value indicates more flow of combustion air in the furnace, which will reduce the thermal efficiency of the process.
● Hydrogen sulfide concentration in treated effluent exit foul water stripper: A higher value indicates increased pollutant concentration in effluent water discharged from the plant.
Similarly, the equipment performance indicators estimated by the simulator that cannot be directly measured in the plant are:
● Exchanger fouling in P-heater train: A higher value indicates reduced exchanger heat transfer efficiency, leading to higher fuel consumption in the naphtha vaporizer and extra cooling load on the air cooler fans.
● Naphtha vaporizer firebox thermal efficiency: The heat pickup ratio by the feed stream against the total heat released in the firebox. A lower number will indicate heat loss and increased fuel consumption.
● Pump power efficiency: A lower value will indicate increased power consumption in the system, leading to low energy efficiency in the process.
● Reactor catalyst activity: This indicates the effectiveness of the catalyst in the reactor. A low activity would require increasing reactor temperature, leading to increased fuel consumption and loss of motivation working life.
● Reduced off-gas feed flow to the oxygen stripper
● Auto-tuning the simulation model to ensure soft-sensed parameter fidelity