Figure - uploaded by Andrea Bertolino
Content may be subject to copyright.
Comparison of different error measures values for nominal and optimized mechanisms consid- ering data used in optimization.
Source publication
In this work, we propose a novel data-driven approach for detailed kinetic mechanisms optimization. The approach is founded on a curve matching-based objective function and includes a methodology for the optimisation of pressure-dependent reactions via logarithmic interpolation (PLOG format). In order to highlight the advantages of the new formulat...
Contexts in source publication
Context 1
... remaining set of 196 reactions retaining ~10% of the total impact are excluded in further investigations. All selected reactions at CSF and CIF level are reported in Table S1 (SM). ...Context 2
... the reactions included in this work 13 out of 43 rates come from experiments together with their uncertainty factors (see table S1 in SM). These reactions were selected by applying the procedure described in Section 2.4 to each experiment inside the database ( Section 2.3 ). ...Context 3
... general, overall improvement can be appreciated in Fig. 11 . Table 1 shows the overall objective function values for the optimized models and their deviations from those of the nominal mechanism. As expected, both L1-mech and L2-mech outperform the nominal mechanism in terms of L1 and L2-norms, but the first is characterized by a lower CM index (i.e. higher 1-CM), and the second shows little improvements. ...Similar publications
The classic approach in turbomachinery design optimization considers only nominal geometries while manufacturing tolerances are evaluated in post-processing. Without any knowledge about such deviations, the optimizer chooses the solution corresponding to the highest attainable performance. However, such a shape may require tight and expensive toler...
Citations
... [20], Konnov [22], Tian [23], Okafor [24], Otomo [26], Glarborg [27], Shrestha [28], GRI-Mech3.0 [29], Dagaut [30], Duynslaegher [31], Hadi [32], Song [25], Klippenstein [33], Nakamura [34], Han [35], Stagni [36], Zhang [37], Wang [38], Gotama [39], Bertolino [40], Li [41], Lhuillier [42]). Figure 2 shows the NH 3 /air laminar flame velocity results predicted by eight chemical kinetic reaction mechanisms with low errors, at equivalence ratios ranging from 0.7 to 1.4. ...
... [20], Ko nov [22], Tian [23], Okafor [24], Otomo [26], Glarborg [27], Shrestha [28], GRI-Mech3.0 [29], Daga [30], Duynslaegher [31], Hadi [32], Song [25], Klippenstein [33], Nakamura [34], Han [35], Stag [36], Zhang [37], Wang [38], Gotama [39], Bertolino [40]). [20], Otomo [26], Glarborg [27], Nakamu [34], Han [35], Stagni [36], Zhang [37]). ...
... [20], Konnov [22], Tian [23], Okafor [24], Otomo [26], Glarborg [27], Shrestha [28], GRI-Mech3.0 [29], Dagaut [30], Duynslaegher [31], Hadi [32], Song [25], Klippenstein [33], Nakamura [34], Han [35], Stagni [36], Zhang [37], Wang [38], Gotama [39], Bertolino [40]). ...
With the increasing greenhouse effect and energy crisis, ammonia is one of the most promising alternative fuels. However, the research on the combustion characteristics of ammonia needs to be further improved. In this paper, the combustion characteristics of two kinds of ammonia and ammonia–hydrogen amino fuels (laminar flame velocity) are investigated through experimental data and kinetic mechanism analysis, and the laminar flame predictions are calculated for 20 kinds of ammonia mechanisms with different equivalence ratios, oxygen contents, and hydrogen doping ratios, after which MAPE and sensitivity analysis are used to determine the applicability of the mechanisms. The results indicate that the incorporation of hydrogen and the augmentation of oxygen concentration induce exponential and linear increases in the laminar flame speed of ammonia, respectively. The laminar flame speed of ammonia reaches its maximum at an equivalence ratio of approximately 1.1, with a value ranging from 6 to 7 cm/s. Under a hydrogen addition ratio of 0.4, the laminar flame speed of ammonia even reaches 29–30 cm/s. The Otomo and Zhang mechanisms are recommended for ammonia fuels with different equivalence ratios and oxygen contents. For different equivalence ratios and hydrogen doping ratios of ammonia–hydrogen combustion, the Gotama and Stagni mechanisms are more suitable. For the overall conditions, the Zhang mechanism is recommended in this paper to simulate the laminar flame velocity for ammonia and ammonia–hydrogen mechanisms. Based on the Glarborg mechanism, an optimized mechanism is proposed to simulate the laminar flame velocity for both fuels, which reduces to 9.55% compared to 43% for the average calculation error of the original mechanism.
... One order of magnitude of uncertainty was assumed for the rate coefficients of these PLOG reactions. Bertolino et al. [71] were the first to propose a similar approach to simultaneously optimize the parameters of all PLOG Arrhenius parameters of a reaction step, using three fitted parameters. The same procedure was applied by Wang et al. [37] in a recent article on the development of a pentane combustion mechanism using rate rule optimization. ...
Kinetics parameter optimization of the ethylene chemistry in the AramcoMech 2.0 mechanism (493 species and 2716 reactions) was carried out against a large collection of indirect (1440 data points in 153 data sets) and direct (936 data points in 58 data sets) experimental data. The indirect data collection consisted of ignition delay time measurements in shock tubes covering a temperature range of 930-2230 K and a pressure range of 0.28-63.3 atm, and laminar burning velocity measurements at preheat temperatures from 298 to 650 K, and pressures from 0.5 to 10 atm. Due to the large size of the model and the data collection, direct optimization was not feasible, therefore, we applied the recently proposed Reduction-Assisted Parameter Optimization-Based Model Development (RAPOD) procedure. First, using the Simulation Error Minimization Connectivity Method (SEM-CM), a reduced mechanism with 75 species and 612 reactions was obtained that performs similarly to the detailed mechanism regarding the indirect measurements used. This smaller model could be simulated around 50 times faster enabling efficient optimization with moderate computational effort on the large number of experimental targets. Then, influential reactions of the reduced model were identified using the novel PCALIN method, which is based on principal component analysis of the local sensitivity matrix scaled with experimental data uncertainty and parameter uncertainty. The Arrhenius parameters (ln A, n, E/R) in 18 reactions were optimized within their prior uncertainty domain against the data collection. Finally, the optimized parameters were transferred to the original AramcoMech 2.0 mechanism, whose performance was shown to improve in a similar fashion as that of the reduced model. The uncertainties of the model results were considerably reduced due to the significant reduction of the uncertainties of most of the optimized rate coefficients.
... The mechanism constructed by Bertolino et al. [11] is optimized based on LBVs, species concentrations, and IDTs from Stagni et al. [12] This optimization contains 31 species and 203 reactions, which has good agreement with the data over the whole range of experimental conditions. Gotama et al. [13] investigated the LBVs and Markstein length of NH 3 /H 2 /air flames under various conditions. ...
... Further simplification of this mechanism yields a simplified NH 3 /nC 7 H 16 mechanism with 74 species and 495 reactions. Xu et al. [27] coupled the NH 3 mechanism from Bertolino et al. [11] with the nC 7 H 16 skeletal mechanism from Chang et al. [28], developing a NH 3 /nC 7 H 16 mechanism that includes 69 species and 389 reactions. The combustion process under specified conditions can be well predicted by this mechanism while maintaining high computational efficiency. ...
For spark ignition and compression ignition ammonia engines, a typical approach to ensure stable operation involves the blending of ammonia with hydrogen and diesel, respectively. For the ammonia/hydrogen fuel, in this study a comprehensive comparison was conducted firstly for the differences among existing chemical mechanisms according to the experimental data of ignition, oxidation, and flame propagation. The result indicates that the current reaction mechanisms for ammonia/hydrogen fuel exhibit high prediction accuracy only within limited condition ranges. Subsequently, considering the completeness of combustion reaction pathways for ammonia/hydrogen fuel, a chemical mechanism of ammonia and ammonia/hydrogen fuel was developed and optimized in this study, and the comprehensive validation demonstrates the accuracy of the developed mechanism. On this basis, the ammonia mechanism was integrated with the detailed n-heptane mechanism to derive a mechanism for ammonia/diesel fuel that includes 1351 species and 6227 reactions. The good performance of this mechanism was demonstrated in terms of the experimental data of ignition and oxidation. In addition, the ignition sensitivity and reaction pathways of ammonia/hydrogen fuel were investigated based on the constructed mechanism, and the significance of C3–C7/N reactions was also analyzed for the ammonia/diesel fuel ignition process.
... Numerous attempts have been made to model Ammonia combustion, resulting in the development of several mechanisms, some of which are applicable to engine combustion scenarios. For example, a data-driven approach was implemented by Bertolino et al. [25] to derive an optimized chemical kinetic mechanism for Ammonia blends, based on analysis of 365 experimental data points. A separate experimental investigation by Stangi et al. [26] focused on obtaining a chemical kinetic mechanism for Ammonia combustion under lean conditions and within a temperature range of 500 K to 2000 K. ...
... As stated earlier, in the target combustion strategies (RCCI and dual fuel), the auto-ignition process starts from regions with high concentration of the fuel with high reactivity and then the flame propagates into the mixture containing the low reactivity fuel (or fuel blend). Therefore, IDT for Ammonia/n-heptane mixtures in low temperatures, comparison of simulations (lines with small marks) with experimental data (hollow marks) from Yu et al. [25] in the following, the LBV in the mixtures containing target low reactivity fuels, i.e. NG, Ammonia and hydrogen is investigated. ...
... Test conditions for Ammonia/n-heptane mixtures from[25] ...
div class="section abstract"> Ammonia, with its significant hydrogen content, offers a practical alternative to pure hydrogen in marine applications and is easier to store due to its higher volumetric energy density. While Ammonia's resistance to auto-ignition makes it suitable for high-compression ratio engines using pre-mixed charge, its low flame speed poses challenges. Innovative combustion strategies, such as dual-fuel and reactivity-controlled compression ignition (RCCI), leverage secondary high-reactivity fuels like diesel to enhance Ammonia combustion. To address the challenges posed by Ammonia's low flame speed, blending with hydrogen or natural gas (NG) in the low reactivity portion of the fuel mixture is an effective approach. For combustion simulation in engines, it is crucial to develop a chemical kinetics mechanism that accommodates all participating fuels: diesel, Ammonia, hydrogen, and NG. This study aims to propose a kinetics mechanism applicable for the combustion of these fuels together. The mechanism is tailored for engine conditions, including high pressures and temperatures, and diverse chemical species concentrations. To render the mechanism suitable for computationally efficient 3-D Computational Fluid Dynamics (CFD) simulations, it is reduced and contains 82 species and 636 reactions, with N-heptane serving as the surrogate for diesel fuel. The mechanism is tuned using optimization methods to match available experimental data on ignition delay time (IDT) for N-heptane. The prediction of IDT and laminar burning velocity values by the mechanism is validated with available experimental data. Additionally, 3-D CFD and quasi-dimensional multi-zone engine simulations are conducted using the new mechanism to verify engine operating parameters against available experimental data.
</div
... It underscores the significant influence of the key steps within the H 2 NO subset (R 16, R 17, R 18,R 22,R 23, and R 24). However, these reactions are still associated with considerable uncertainty [3,7,14,15,106,107], and reducing these uncertainties would be crucial for accurate characterization of the kinetic models for low temperatures. ...
... According to a recent study by Szanthoffer et al. applying a large collection of experimental data on laminar burning velocity, ignition delay time, and species concentrations [28], this mechanism is one of the best-performing mechanisms for NH 3 and NH 3 /H 2 combustion. Additionally, the mechanism is compared to four other well-performing mechanisms [29][30][31][32] at the conditions of the present study in the supplementary material S4 and is considered suitable for the present study. For CH 4 /air mixtures, the well-established CRECK high-temperature C1-C3 mechanism is applied [33]. ...
Ammonia is a promising future energy carrier because of its carbon-free nature and high volumetric energy density compared to hydrogen. However, implementing ammonia as a fuel appears challenging due to its low reactivity. This can be improved, inter alia, by cofiring with a highly reactive fuel like hydrogen. A fuel mixture of ammonia, hydrogen, and nitrogen with favorable thermochemical properties can be produced by partially cracking ammonia. To assess the combustion behavior of ammonia and partially cracked ammonia at engine conditions, this study performs experiments on an optical engine test rig. Ammonia cracking ratios of 0, 7.5, and 10%, fuel-air equivalence ratios of 0.7 to 1.2, and different turbulence conditions at variable engine speeds are investigated at a compression pressure of 7 MPa. A turbulent flame speed approach is determined from high-speed schlieren imaging in the combustion chamber. The corresponding laminar flame properties and effective Lewis number are calculated numerically and the combustion regimes are assessed. The results show that ammonia/air flames propagate significantly faster under turbulent, engine-like conditions than expected from results at laminar, ambient conditions. Additionally, the partial cracking of ammonia further improves the turbulent combustion behavior. With lean fuel/air mixtures, a cracking ratio of 10% is sufficient to achieve flame speeds close to that of methane under highly turbulent flow conditions. The observed stronger influence of turbulence on the flame speed of ammonia and partially cracked ammonia compared to methane is due to the lower effective Lewis numbers and higher Karlovitz numbers of these fuels.
... OptiSMOKE++ implements heuristic optimization strategies to iteratively calibrate kinetic parameters, to refine the agreement with target experiments (e.g. measurements in ideal reactors and 1D laminar flames [32]), as well as artificially generated data [28]. It adopts the OpenSMOKE++ libraries [33] for the ideal reactor modeling as well as the management of detailed kinetics while exploiting the DAKOTA toolkit for the optimization part. ...
... As a result, the value of each model-experiment comparison, as defined in Eq. (1), also provides information on the statistical error due to experimental uncertainty. Such a procedure has been successfully applied in previous works [32]. ...
... The experimental details are available in Ref. [49]. At the pressures of 4 and 6 atm, the improved mechanism demonstrated better predictive capability for N 2 O peak compared to other commonly used ammonia combustion mechanisms [47,[50][51][52][53][54], thereby validating the accuracy of the improved mechanism. More detailed simulations of N 2 O component evolution and experimental results can be found in Fig. S1 and Fig. S2, which further emphasize the advantages of the improved mechanism over others. ...
As a kind of substitute fuel of hydrocarbon fuel, ammonia has gained increasing attention for its potential to reduce carbon emissions. Blending with carbon-neutral methanol is expected to deal with the slow flame speed of ammonia. Nevertheless, the emission of N 2 O, a byproduct of ammonia combustion, would undermine the carbon reduction benefits due to its high global warming potential. This study employed Chemkin to construct a Chemical Reaction Network (CRN) model to analyze the reaction kinetics of ammonia combustion when blended with a small amount of methanol. It finds that the increases in the methanol blending ratio, pressure, and temperature can reduce greenhouse gas emissions effectively. Further analysis indicates that pressure and temperature have distinct effects on the emission mechanisms of N 2 O and NO. Increasing the combustion pressure weakens the chain reactions, thereby reducing the formation of both NO and N 2 O; the increase in initial temperature raises NO emissions, but also promotes the decomposition of N 2 O due to the abundance of H in the post-combustion phase. Optimizing post-combustion efficiency and controlling the formation of NO are crucial for N 2 O emission control. Thus, a staged methanol injection strategy is proposed, where early injection reduces NO emissions and late injection facilitates the decomposition of N 2 O.
... • Ammonia (NH 3 ) is a promising carbon-free fuel because it can be used in a sustainable and recyclable loop for energy production. ...
... In Fig. 2, symbols show the experimental data measured by Ronney [62], Jabbour et al. [63], Mei et al. [64], and Xiao et al. [65] via different experimental methods. Lines show the simulated results of the present and previous mechanisms (including Mei et al. [64], Stagni et al. [48], Berolino et al. [66], and Li et al [45]). The ammonia-coal mechanism captures a similar trend with both experimental and simulated data. ...
... The mass flow rate of the oxidants and fuels in Table 4 Table 3 Parameters of the C2SM model. [66], and Li et al. [45]. ...
Co-firing ammonia with coal is a promising and feasible technology for reducing coal-related carbon emissions. Pyrolysis and ignition of ammonia-coal blended fuels are the key steps for flame stability and boiler operation safety throughout the conversion process but remain unclear. In this work, an extended Euler-Lagrange framework coupled to detailed solid-phase pyrolysis kinetics and gas-phase reactions mechanism is introduced and validated against experimental results for ammonia-coal co-firing in a two-stage flat flame burner. First, the CRECK-S coal pyrolysis model was validated with TGA experiments and the CPD model at different heating rates to determine parameters for a competing two-step model for CFD simulations. Second, a detailed gas-phase mechanism with 116 species and 1513 elementary reactions was derived from the volatile and ammonia reaction mechanisms. Thirdly, the co-firing simulations were validated for the ignition delay time for various ammonia co-firing ratios. The results show that increasing the co-firing ratios from 0.0 to 1.0 results in gradually increasing ignition delay times in a low-oxygen atmosphere. Further analysis demonstrates that adding ammonia accelerates the coal particle heating rate due to the reduced coal particle number density and ammonia reaction induced gas-phase temperature increase, and thus the coal devolatilization rate is increased. The latter plays a dominant role. Hydrogen produced from ammonia pyrolysis is negligible, so ammonia predominantly participates in the ignition. Time scale analysis shows that homogeneous ignition is dominant during the ignition process. The presence of ammonia inhibits the inward oxygen diffusion and causes the reaction zone to move away from the pulverized coal particle flow. The oxygen diffusion inhibition and low reactivity of ammonia compared to volatiles lead to an increase in ignition delay time for ammonia co-firing, even though pulverized coal devolatilization is accelerated.
