NOx emission estimation in gas turbines via interpretable neural network observer with adjustable intermediate layer considering ambient and boundary conditions

Measurement ◽  
2021 ◽  
pp. 110429
Author(s):  
Dawen Huang ◽  
Shanhua Tang ◽  
Dengji Zhou ◽  
Jiarui Hao
Author(s):  
Weiqun Geng ◽  
Douglas Pennell ◽  
Stefano Bernero ◽  
Peter Flohr

Jets in cross flow are one of the fundamental issues for mixing studies. As a first step in this paper, a generic geometry of a jet in cross flow was simulated to validate the CFD (Computational Fluid Dynamics) tool. Instead of resolving the whole injection system, the effective cross-sectional area of the injection hole was modeled as an inlet surface directly. This significantly improved the agreement between the CFD and experimental results. In a second step, the calculated mixing in an ALSTOM EV burner is shown for varying injection hole patterns and momentum flux ratios of the jet. Evaluation of the mixing quality was facilitated by defining unmixedness as a global non-dimensional parameter. A comparison of ten cases was made at the burner exit and on the flame front. Measures increasing jet penetration improved the mixing. In the water tunnel the fuel mass fraction within the burner and in the combustor was measured across five axial planes using LIF (Laser Induced Fluorescence). The promising hole patterns chosen from the CFD computations also showed a better mixing in the water tunnel than the other. Distribution of fuel mass fraction and unmixedness were compared between the CFD and LIF results. A good agreement was achieved. In a final step the best configuration in terms of mixing was checked with combustion. In an atmospheric test rig measured NOx emissions confirmed the CFD prediction as well. The most promising case has about 40% less NOx emission than the base case.


Author(s):  
Toshiaki Sakurazawa ◽  
Takeo Oda ◽  
Satoshi Takami ◽  
Atsushi Okuto ◽  
Yasuhiro Kinoshita

This paper describes the development of the Dry Low Emission (DLE) combustor for L30A gas turbine. Kawasaki Heavy Industries, LTD (KHI) has been producing relatively small-size gas turbines (25kW to 30MW class). L30A gas turbine, which has a rated output of 30MW, achieved the thermal efficiency of more than 40%. Most continuous operation models use DLE combustion systems to reduce the harmful emissions and to meet the emission regulation or self-imposed restrictions. KHI’s DLE combustors consist of three burners, a diffusion pilot burner, a lean premix main burner, and supplemental burners. KHI’s proven DLE technologies are also adapted to the L30A combustor design. The development of L30 combustor is divided in four main steps. In the first step, Computational Fluid Dynamics (CFD) analyses were carried out to optimize the detail configuration of the combustor. In a second step, an experimental evaluation using single-can-combustor was conducted in-house intermediate-pressure test facility to evaluate the performances such as ignition, emissions, liner wall temperature, exhaust temperature distribution, and satisfactory results were obtained. In the third step, actual pressure and temperature rig tests were carried out at the Institute for Power Plant Technology, Steam and Gas Turbines (IKDG) of Aachen University, achieving NOx emission value of less than 15ppm (O2=15%). Finally, the L30A commercial validation engine was tested in an in-house test facility, NOx emission is achieved less than 15ppm (O2=15%) between 50% and 100% load operation point. L30A field validation engine have been operated from September 2012 at a chemical industries in Japan.


Author(s):  
Dawen Huang ◽  
Shanhua Tang ◽  
Dengji Zhou

Abstract Gas turbines, an important energy conversion equipment, produce Nitrogen Oxides (NOx) emissions, endangering human health and forming air pollution. With the increasingly stringent NOx emission standards, it is more significant to ascertain NOx emission characteristics to reduce pollutant emissions. Establishing an emission prediction model is an effective way for real-time and accurate monitoring of the NOx discharge amount. Based on the multi-layer perceptron neural networks, an interpretable emission prediction model with a monitorable middle layer is designed to monitor NOx emission by taking the ambient parameters and boundary parameters as the network inputs. The outlet temperature of the compressor is selected as the monitorable measuring parameters of the middle layer. The emission prediction model is trained by historical operation data under different working conditions. According to the errors between the predicted values and measured values of the middle layer and output layer, the weights of the emission prediction model are optimized by the back-propagation algorithm, and the optimal NOx emission prediction model is established for gas turbines under the various working conditions. Furthermore, the mechanism of predicting NOx emission value is explained based on known parameter influence laws between the input layer, middle layer and output layer, which helps to reveal the main measurement parameters affecting NOx emission value, adjust the model parameters and obtain more accurate prediction results. Compared with the traditional emission monitoring methods, the emission prediction model has higher accuracy and faster calculation efficiency and can obtain believable NOx emission prediction results for various operating conditions of gas turbines.


2007 ◽  
Vol 40 (20) ◽  
pp. 166-171
Author(s):  
Alejandro Garcia ◽  
Alex Poznyak ◽  
Isaac Chairez ◽  
Tatyana Poznyak

Author(s):  
T. Shudo ◽  
K. Omori ◽  
O. Hiyama

Hydrogen is expected as a clean and renewable alternative to the conventional hydrocarbon fuels. Because the only possible pollutants from the hydrogen combustion are nitrogen oxides (NOx), it is crucial to reduce the NOx emission in the hydrogen utilization. The rich-lean combustion is well known as a technique to reduce the emission of the Zel’dovich NO from the continuous combustion burners for such as gas turbines and boilers. Because the Zel’dovich NO occupies a large part of the total NOx emission, the rich-lean combustion is quite effective to reduce the NOx emission. However, the NOx reduction effect of the rich-lean combustion has not yet been proven for the hydrogen fuel, while the effect has been demonstrated for the hydrocarbon fuels. On the other hand, the prompt NO is emitted from the hydrocarbon combustion especially under the fuel-rich conditions. Though the amount of the prompt NO is quite small for premixed or diffusion combustion, it could be a relatively significant part in the total NO emission from the rich-lean combustion due to the decreased Zel’dovich NO. The authors estimate that hydrogen is more suitable for the rich-lean combustion compared with hydrocarbons, because hydrogen does not emit the prompt NO even under the fuel-rich conditions which necessarily exist in the rich-lean combustion. This research proposes the rich-lean combustion as a method to reduce the NOx emission from hydrogen combustion and experimentally analyzes the characteristics using a coaxial rich-lean burner with varying the mixture conditions.


Sign in / Sign up

Export Citation Format

Share Document