FEEDBACK CONTROL FOR REDUCING FLARING PROCESS SMOKE AND NOISE

摘要

A method of reducing plant emissions includes providing a MPC model for a flaring process including one-to-one models between controlled variables (CVs) including a smoke count and/or a flare count (CV1) and a noise level (CV2), and flow of assist gas as a manipulated variable (MV) and another process gas flow as a disturbance variable (DV). The MPC model receives sensed flare-related parameters during the flaring process including a measure of CV1 (CV1*) and CV2 (CV2*). Provided CV1* is above a minimum setpoint for CV1 (CV1 setpoint) and CV2* is above a setpoint for CV2 (CV2 setpoint), the flaring process is automatically controlled using the MPC model which determines an updated flow setpoint for MV from CV1* and CV2*, the CV1 and CV2 error, and the identified one-to-one models.

主权项

1. A method of reducing emissions at a processing plant, comprising:
providing a model predictive control (MPC) model for a flaring process run at said processing plant including an assist gas comprising a plurality of identified one-to-one models between controlled variables (CVs) including (i) at least one of a smoke count and a flare count (CV1) and (ii) a noise level (CV2), and a flow of said assist gas as a manipulated variable (MV) and at least one other process gas flow as a disturbance variable (DV); said MPC model implemented by a processor receiving sensed flare-related parameters during operation of said flaring process including a measure of said CV1 (CV1*) and said CV2 (CV2*), and provided said CV 1* is above a predetermined minimum setpoint for said CV1 (CV1 setpoint) and said CV2* is above a predetermined setpoint for said CV2 (CV2 setpoint), automatically controlling said flaring process using said MPC model implemented by said processor which automatically determines an updated flow setpoint for said MV based on (i) said CV1* and said CV2*, (ii) a difference between said CV1 setpoint and said CV1* and a difference between said CV2 setpoint and said CV2*, and (iii) said identified one-to-one models.