Case Studies

NOx Reduction and Heat Rate Improvement in a Utility Boiler Using QuickStudy™ Adaptive, Model Predictive Process Controller

The Process

Albright Unit 1 is a single reheat coal fired, drum-type boiler producing 700,000 pounds of steam per hour with a net generating capacity of 80MW. Unit 1 has three rows of four burners, and each row is fed by one coal pulverizer. The unit was previously retrofit with low NOx burners during the early 1990’s following the passage into law of the Clean Air Act Amendments.

The Control Objectives

In coal-fired boilers, NOx emissions at a given load level are a function of excess O2 level. Typically, as excess O2 levels are reduced, NOx emission rates are reduced as well. At some point, further decreases in O2 level result in large, uncontrollable increases in CO emissions. This excess O2 level is sometimes called the “CO threshold.”

Baseline testing showed that NOx levels were fairly uniform, staying within the range of 0.65 to 0.75 pounds per million Btu acrossthe boiler operating load during the NOx season (May through September). High excess O2 levels were documented in areas at the economizer outlet during the baseline tests. With typical three mill operation, the outer two burner columns (columns 1 and 4) operated approximately 1.0% O2 above the center two columns.

Typically, the flue gases from wall-fired pulverized coal boilers are stratified, and the stratification remains essentially unchanged from the primary furnace exit through the convective pass. Because of this stratification, individual O2 readings can be related to individual burner columns.

The QuickStudy™ Solution

A three step method was used to reduce NOx emission rates. First, multiple-point O2 monitors were used as feedback for control of individual burner secondary air (SA) dampers/shrouds to perform spatial balancing of furnace excess O2 levels. Second, the windbox pressure was increased by partially closing all SA dampers/shrouds, which resulted in better distribution of secondary air between burners. This provided for an integral approach and smoothness of operating conditions to reduce the excess O2 level such that the CO threshold would not be reached.

Finally, overall excess O2 levels were reduced until the stack CO monitor showed a controllable increase in CO levels. NOx emissions are reduced by operating at lower excess O2 levels, and CO emissions are controlled at this lower O2 level as well. Spatial balancing serves to lower the CO threshold, allowing operation at lower excess oxygen levels than in a spatially unbalanced situation.

The QuickStudy model predictive controller takes into account process measurements such as CO, O2, and load, as well as the furnace combustion curve in its process model. Based on the relationships among these parameters and the dead times between control actions and their effects, it determines the magnitude and timing of optimum control actions required to achieve the predicted process variables.

Off-Line historical data provided process understanding, which was used to develop the control strategy and preliminary models. Models were then further refined while controlling the combustion process using the “Auto-Adapt” capability of QuickStudy. This resulted in better models and thus enhanced control of the combustion process.

Results

  • NOx was reduced by 15%
  • CEM heat rate was reduced by a minimum of 2%
  • CO emissions were controlled below 250 ppm
  • Average opacity levels were improved slightly
  • Furnace slag conditions were improved

These results provided low cost NOx reduction as well as reduced energy costs due to the heat rate improvement.

AlbrightResult_Image2[1]AlbrightResult_Image3[1]