ABSTRACT
Air jets for active flow control have proved effective in postponing the onset of stall phenomenon in axial compressors. In this paper, we use a combination of machine learning and genetic algorithm to explore the optimal parameters of air jets to control rotating stall in the axial compressor CME2. Three control parameters are investigated: the absolute injection angle, the number of injector pairs and the injection velocity. Given an experimental dataset, the influence of the air jet parameters on the surge margin improvement and power balance is modeled using two shallow neural networks. Parameters of the air jets are then optimized using a genetic algorithm for three rotational velocities, i.e., [Formula: see text]. First, surge margin improvement and power balance are being maximized independently. Then, a bi-objective optimization problem is posed to explore the trade-off between the two competing objectives. Based on the Pareto front, results suggest that a globally optimal set of parameters is obtained for a velocity ratio (defined as the ratio of the injection velocity to the rotor tip speed) ranging from 1.1 to 1.6 and an injection angle attack varying from [Formula: see text]. These outcomes point out a potential generalization of the control strategy applicable to other compressors.
ABSTRACT
Improved turbulence modeling remains a major open problem in mathematical physics. Turbulence is notoriously challenging, in part due to its multiscale nature and the fact that large-scale coherent structures cannot be disentangled from small-scale fluctuations. This closure problem is emblematic of a greater challenge in complex systems, where coarse-graining and statistical mechanics descriptions break down. This work demonstrates an alternative data-driven modeling approach to learn nonlinear models of the coherent structures, approximating turbulent fluctuations as state-dependent stochastic forcing. We demonstrate this approach on a high-Reynolds number turbulent wake experiment, showing that our model reproduces empirical power spectra and probability distributions. The model is interpretable, providing insights into the physical mechanisms underlying the symmetry-breaking behavior in the wake. This work suggests a path toward low-dimensional models of globally unstable turbulent flows from experimental measurements, with broad implications for other multiscale systems.
ABSTRACT
BACKGROUND: Postoperative pancreatic fistula (POPF) is the most serious complication following pancreaticoduodenectomy (PD). Identifying patients at high or low risk of developing POPF is important in perioperative management. This study aimed to determine a predictive risk score for POPF following PD, and compare it to preexisting scores. METHODS: All patients who underwent open PD from 2012 to 2017 in two high-volume centers were included. The training dataset was used for the development of the POPF predictive risk score (using the 2016 ISGPS definition), while the testing dataset was used for external validation. The proposed score was compared to the fistula risk score (FRS), the NSQIP-modified FRS (mFRS), and the alternative FRS (aFRS). RESULTS: Overall, 448 and 213 patients were included in the training and testing datasets, respectively. A probabilistic predictive risk score was developed using four independent POPF risk factors (increasing age, no preoperative radiation therapy, soft pancreatic stump, and decreasing main pancreatic duct diameter). The discriminative capacities of the new score, FRS, mFRS, and aFRS were similar (AUC ranging from 0.73 to 0.79 in the training cohort and from 0.73 to 0.76 in the testing cohort). However, the new score identified more specifically patients at low risk of POPF compared with other scores, in both cohorts, with a 6% false-negative rate. CONCLUSIONS: Preoperative radiation therapy is an independent protective factor of POPF following PD. It should be included in the risk score of POPF to identify more precisely patients at low risk for this complication.
