An optimization of redundant measurements location for thermal capacity of power unit steam boiler calculations using data reconciliation method

• Application of the generalized method data reconciliation in thermal technology.
• Proved necessity of the data reconciliation in thermal technology.
• Demonstration of the data reconciliation advantages in the thermal technology.
• Optimization of a location of measurements in the thermal systems.
• Application of the Kullback-Leibler divergence in the assessment of the measurement system.

The optimization of a location of redundant measurements under varying loads for steam boiler of a supercritical power unit using the generalized method of data reconciliation has been carried out. The method of weighted objectives has been applied as a method of optimization. This method reduce the weighted multi-criteria optimization task to task one-dimensional. Measurement values have been determined by numerical experiment and the Monte Carlo method for the designed redundant measurements system. For this purpose, a mathematical simulation model of a supercritical steam power unit with power rating of 900 MW in the Thermoflex program has been worked out. In the optimization calculations of location of redundant measurements as an objective functions minimizing the relative standard deviation of a boiler thermal capacity and maximizing the Kullback-Leibler divergence have been accepted. In the calculation the measurements were taken into account, which can be located in the water-steam system of the boiler and in the high-pressure heat recovery steam supercritical power unit. The results of calculations confirm the influence of the number of redundant measurements and places of their location in the thermal system of the boiler on the accepted criteria of optimization. Increasing the number of redundant measurements, in terms of the data reconciliation method, leads to decrease the relative standard deviation of the thermal capacity of the boiler and increase the value of Kullback-Leibler divergence, i.e.; decrease the information entropy of the measuring system.