Automatic CFD optimisation of biomass combustion plants. Ali Shiehnejadhesar

Transcription

1 Automatic CFD optimisation of biomass combustion plants Ali Shiehnejadhesar IEA Bioenergy Task 32 workshop Thursday 6 th June 2013

2 Contents Scope of work Methodology CFD model for biomass grate furnaces Optimisation routine Case studies performed Results Summary and conclusions Folie 2

3 Scope of work Development of a tool for the automatic performance of CFD-based parameter studies for the optimisation of biomass grate furnaces Link of automatic parameterisation and optimisation routines with an inhouse developed CFD model for biomass grate furnaces Test of the model for a 180 kw grate furnace of BE2020+ Reduction of carbon monoxide emissions and the energy demand of the secondary air fan (pressure loss) by changing the diameter and angle of the secondary air nozzles Performance of a manual parameter study Performance of an automatic parameter study and benchmarking against the manual study Investigation of the grid type influence on the overall simulation time and accuracy Application of the method developed for a 15 KW fixed bed pellet furnace Folie 3

4 Methodology CFD model In-house developed empirical packed bed combustion model for biomass grate furnaces Simulation of gas phase combustion (CFD models of Ansys/Fluent) Realizable k-ε Model: turbulence Discrete Ordinate Model: radiation Eddy Dissipation Model with global 3 step methane mechanism: interaction of turbulence and chemistry Folie 4

5 Methodology Optimisation routine I Weight function Development of a weight function to combine the two optimisation parameters (CO emissions and pressure loss over the secondary air nozzles) in a common function 1 W = A + B P YCO α A linear correlation to the weight function has been assumed regarding the pressure loss (proportional to the energy demand of the fan) A polynomial function with a strong increase at a chosen emission limit was assumed for the CO emissions Explanations: W weight function; Y CO CO mole fraction in ppmv; α constant factor; B constant factor; P pressure loss in Pa Folie 5

6 Methodology Optimisation routine II Optimisation routine Parameterisation of the geometry and definition of design points based on selected design variables Automatic performance of CFD simulations for the defined design points with the ANSYS Workbench Evaluation of the output parameters and calculation of the weight function for the design points Minimisation of the weight function to find the optimum geometric configuration Folie 6

7 Furnace investigated CFD model for the pilot-scale grate furnace of BE2020+ Symmetry plane Flue gas recirculation nozzles rimary combustion hamber Flue gas path Surface of fuel bed Furnace outlet (boiler inlet) Secondary combustion chamber Secondary air nozzles Angle of the nozzles to the horizontal plane Boiler load = 180 kw th Adiabatic flame temperature = 937 C O 2 content flue gas = 8.4 Vol% dry Total air ratio = 1.67 Primary air ratio = 0.83 Flue gas recirculation ratio = 0.3 Fuel Miscanthus Unit Value C wt.% d.b H wt.% d.b. 5.9 O wt.% d.b N wt.% d.b Moisture content wt.% w.b Ash wt.% d.b. 2.2 GCV MJ/Kg d.b NCV MJ/Kg w.b Folie 7

8 Results Automatic versus manual optimisation Lines Manual optimisation Circles Automatic optimisation K G Manual optimisation ( tetrahedral cells) Automatic optimisation (1 ( 1 million tetrahedral cells) cells) Global minimum (manual optimisation method) Global minimum (automatic optimisation method) G d min = (mm) K d min = 17 (mm) Diameter of secondary air nozzles (mm) Diameter of secondary air nozzles (mm) Calculated CO emissions and pressure loss (left) as well as weight function (right) for the automatic and manual optimisation method Folie 8

9 Results Grid dependency Circles Tetrahedral cells Crosses Polyhedral cells G H Tetrahedral cells ( 1 million cells) Polyhedral cells ( cells) Global minimum (tetrahedral cells) Global minimum (polyhedral cells) H d min = (mm) G d min = (mm) Diameter of secondary air nozzles (mm) Diameter of secondary air nozzles (mm) Effect of mesh type on the calculated CO emissions and pressure loss (left) and weight function for different mesh types (right) Folie 9

10 Results 3D Weight function Example of local minimum Global minimum d min = (mm) α min = 5.08 (deg) C Nozzle angle (deg) Nozzle diameter (mm) Weight surface from automatic optimisation and tetrahedral grid type calculated with 80 design points Explanations: tetrahedral mesh type with 1 million cells Folie 10

11 Summary and conclusions model test pilot-scale grate furnace Development and test of a new CFD-based automatic optimisation routine for a pilotscale biomass grate furnace concerning the minimisation of CO emissions and pressure loss Considerably higher impact of the nozzle diameter on CO emissions and pressure loss than the nozzle angle Good agreement between results generated by the automatic optimisation routines in comparison to the manual optimisation method Significant reduction of simulation time compared to the manual method by a factor of 8 for the tetrahedral mesh type With the polyhedral mesh type a further reduction of the calculation time could be achieved (factor of 32 compared to the manual optimisation) even though there were slight deviations from the tetrahedral grid concerning CO emissions A further improvement of the mesh concerning accuracy without a significant increase of computation time can be achieved by a local mesh refinement in the nozzle exit region Folie 11

12 Case study II CFD model of a 15 KW pellet boiler α Flue gas outlet Boiler (blue colors) Fuel feeding Pellet furnace Secondary air Grate Folie 12 Burner Fuel bed Secondary air nozzles Nominal boiler capacity : 15 kw Adiabatic flame temperature = C Primary air ratio = 0.41 Total air ratio = 1.53 Number of secondary air nozzles = 11 Fuel : Soft wood pellets (moisture content (wt.% w.b.) : 6.54)

13 Parameters investigated and target values Design parameters: Diameter of the secondary air nozzles (6 10 mm) Angle of the secondary air nozzles to burner axis (10 30 ) Target values investigated: CO concentration at the entrance to the heat exchanger Pressure drop over the secondary air supply and nozzles Minimisation of the weight function with respect to pressure drop and CO concentration Constraints: Temperature of burner (a) < 700 C Pressure drop < maximum under pressure in the furnace achievable by the flue gas fan (b) = 150 Pa b Combustion chamber a Folie 13

14 Results CO concentration and pressure drop The pressure drop increases with increasing the angle of the secondary air nozzles The pressure drop increases with decreasing the diameter of the secondary air nozzles The CO emissions decrease with decreasing the diameter of the secondary air nozzles No clear trend concerning CO burnout as a function of angle was found due to flame instabilities observed CO emission (ppmv) Pressure loss (Pascal) Pressure constraint (150 Pa) Diameter of secondary air nozzles (mm) Diameter of secondary air nozzles (mm) CO concentration [ppmv] at the entrance to the heat exchanger (left) and pressure Folie 14 drop over the secondary air supply and nozzles [Pa] (right)

15 Results Feasible design concerning nozzle angle and diameter Feasible design area Burner wall temperature ( C) P < 150 Pa Nozzle diameter (mm) Maximum temperature of the burner [ C] in dependence of the angle and diameter of the secondary air nozzles to the burner axis Folie 15

16 Results Maximum temperature of the burner wall α = 10 α = 20 α = 30 Maximum burner wall temperature [ C] in dependence of the angle of the secondary Folie 16 air nozzles at a nozzle diameter of 7.5 [mm]

17 Results Weight function The optimum nozzle configuration was found at an angle of 10 and a diameter of 7.5 mm. The pareto set (red points) shows the variants with the lowest CO emissions at a given pressure drop or with the lowest pressure drop for a given CO emission limit Weight (-) CO concentration (ppmv) 2 1: Optimum solution of weight function (7.5 mm / 10 ) 2: 8 mm / 10 3: 7.5 / 15 4: Original configuration (7.2 mm / 20 ) : Pareto front Pressure constraint (150 Pa) 1 d = 7.5 Nozzle angle (deg) α = 10 Nozzle diameter (mm) Optimisation function (left) and Pareto front (right) 3 4 Pressure loss (Pa) Explanations: Folie 17 black points: not-optimal solutions, red points: solutions on the pareto front

18 Summary and conclusions case study 15 kw pellet furnace A high exit velocity of the secondary air jets improves the CO burnout. An additional increase of the nozzle angle shortens the flame length and further reduces the CO emissions (swirling flow). However, the burner wall temperature increases with a shorter flame length. Moreover, the pressure drop increases with larger the nozzle angles. The pareto front represents the design variables which have the lowest CO emission for a chosen pressure loss and the lowest pressure loss for a chosen CO emission limit, respectively. Hence, the pareto front provides valuable information about the best design variant regarding a certain parameter (e.g. CO emission limits) for defined side constraints (e.g. fan pressure increase). Folie 18

19 Overall conclusions and outlook The new automatic CFD optimisation routine works well and shows a large potential concerning: targeted geometry optimisation considerable reduction of manpower needed Future work will focus on: implementation of a direct optimisation routine in order to reduce the number of simulated cases and the simulation time to find the optimum extension of the number of design parameters (from 2 to 4) implementation of an interface to the empirical packed bed model in order to investigate variations of operating conditions like air staging and fuel heat load Test of the routines for real-scale plants focusing on: automatic geometry optimisation automatic optimisation of plant operation (e.g. variation of air staging and simulation of CO-λ-characteristics as a basis for an optimised plant control) automatic scaling of plants Folie 19

20 Thank you for your attention Ali Shiehnejadhesar BIOENERGY GmbH Inffeldgasse 21b, 8010 Graz, Austria Tel.: +43 (0) Folie 20