OCR Text |
Show Optimizing Boilers for. NOx Control and Fuel Savings: Three Utility Examples Using Sequential Process Optimization Introduction Peter D. Patterson Vice President Power~ Service Ultramax Corporation Cincinnati, Ohio 45246 In recent years, awareness has grown throughout industry of the need to utilize process modeling and optimization tools in design, development and production. While useful tools have been around for many years, they are not widely applied, especially in a production environment. These tools include empirical approaches such as various statistical design-of-experiment methods, evolutionary operations and the Taguchi Method. Spurred by Japanese competition, American companies began expanding the use of these methods in the early eighties to achieve notable quality improvements in the automotive and electronics industries. However, effective applications were mostly limited to design and development functions. In production, these methods have severe limitations that have inhibited their use. They introduce too much risk, do not deal with variables which are not being controlled, are project oriented and require substantial expertise that is often lacking. They tend to be time-consuming, ineffective and expensive to apply in production. One well-known and popular tool, statistical process control (SPC), offers a way to achieve greater quality consistency in production processes. However, SPC does not help with the fundamental needs to model processes, understand relationships and optimize settings of control parameters that contribute most to better process performance. Special Needs for Optimization of Production Systems Optimization of production systems calls for unique requirements that exceed the capabilities of these early methods. Such systems also resist attempts to model them with first-principles theories of energy balance, heat transfer and fluid dynamics. Expert systems are good diagnostic tools but are based on old knowledge, are difficult to customize and contribute little to optimization. It has become apparent that new and better tools must be created and applied in innovative ways in order to meet the needs of today's demanding and competitive environment. Empirical methods using data taken directly from the production process offer a particularly encouraging avenue to develop new knowledge. Some of the special characteristics of production systems include: • They are throughput driven and cannot be interrupted or upset by testing procedures to collect data. • They are often influenced by variations in raw materials, fuels, environmental conditions and equipment components. • They must satisfy multiple objectives relating to efficiency, quality, production rate. yield. fuel usage. emissions. cost and economic performance. • Valuable improvements can be achieved with existing equipment by optimization of operating parameter settings (tuning). • They are typically difficult. if not impossible. to model using first-principles techniques. 1 |