(57c) Furnace Flame Instability Detection Using Pressure Measurement with Advanced Diagnostics

Furnaces and fired heaters are utilized throughout refineries. A common problem in furnaces is instability in the burner flames. If the instability is not corrected, it can lead to a flame-out, a dangerous condition in which unburned fuel is left in the fire box. If a flame-out occurs, the unit must be shut down immediately, to prevent the unburned fuel from igniting. Thus, it is very important for plant operators to be able to take early corrective action to return burner flames to a stable condition. Various technologies, including flame sensors, combustion analyzers, and image processing have been used for detection of unstable flames. All, of these traditional technologies have shortcomings, such as being expensive, difficult to install and maintain, or prone to false alarms.

One or more pressure transmitters are typically installed on the fire-box to measure the draft pressure of the furnace. When a flame goes unstable, it tends to flicker, resulting in an increase in the variation, or process noise, of the pressure measurement signal. In traditional pressure transmitters, this process noise is filtered out of the 4-20 mA signal, and so it is never seen in the distributed control system.

However, pressure transmitters with advanced diagnostics can be used to gain insight into the stability in furnace flames. Using Statistical Process Monitoring (SPM) technology, the transmitter measures the draft pressure at a fast sampling rate, and then statistically calculates the standard deviation. The increase in process noise due to flame instability causes the standard deviation to increase. The standard deviation is then transmitted to the control system as a HART digital parameter. By monitoring the standard deviation, a furnace operator can get an indication that burner flames are going unstable, and take early corrective action to avoid the flame-out.

Recent furnace testing further substantiates the use of SPM technology for detecting unstable burner flames. Various tests were conducted, running the furnace with different fuel types and different heat rates, and mounting the pressure transmitters at various locations on the fire box. In every case, flame instability was detected in the SPM data, as an increase in the standard deviation. By detecting flame instability early, a plant can avoid a flame-out, and prevent an expensive process shutdown.