Takashi Misaka
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Takashi Misaka has a total of 101 co-authors. Here are the top co-authors by number of shared publications:
Shigeru Obayashi
Tohoku University, Sendai, Japan
15 shared publications- Shigeru Obayashi
Institute of Fluid Science, Tohoku University, Sendai, Japan
7 shared publications - Frank Holzäpfel
DLR, German Aerospace Center, 82234 Oberpfaffenhofen, Germany
6 shared publications - Mitsuo Yoshimura
Tohoku University
4 shared publications - Shigeru Obayashi
Institute of Fluid Science, Tohoku University
4 shared publications Yoshiyuki Furukawa
4 shared publications
Latest Publications
- Open Access
- 21 January 2026
- Published by Elsevier in Computers & Fluids
- Vol. 308
Flow separation prediction in a simplified notchback car model by assimilating global luminescent oil film measurements
- 21 January 2026
- Published by Elsevier in Computers & Fluids
- Vol. 308
Open AccessFlow separation prediction in a simplified notchback car model by assimilating global luminescent oil film measurements
- Open Access
- 12 December 2025
- Published by Elsevier in Journal of Computational Physics
- Vol. 548
Data assimilation with physics-informed neural network surrogates constructed against prediction uncertainty
- 12 December 2025
- Published by Elsevier in Journal of Computational Physics
- Vol. 548
Open AccessData assimilation with physics-informed neural network surrogates constructed against prediction uncertainty
- 6 September 2023
- Published by Elsevier in Journal of Computational Physics
- Vol. 493
Space-time adaptive model order reduction utilizing local low-dimensionality of flow field
- 6 September 2023
- Published by Elsevier in Journal of Computational Physics
- Vol. 493
Space-time adaptive model order reduction utilizing local low-dimensionality of flow field
- 1 August 2022
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 61 (16) , 5392-5410
A proposal for improving production efficiency of existing machining line through a hybrid monitoring and optimisation process
In a machining line with a Lean manufacturing environment, such as one used for mass production of automotive components, tool life is determined by a tool counter that is set on the safe side to avoid producing a failed product, so the cutting tools are typically underutilised. On the other hand, cutting parameters optimisation, which has a significant impact on tool life, is not feasible due to the rule that prohibits changing the cutting parameters during mass production to maintain the quality standard. To address both issues, this paper proposes a hybrid monitoring and optimisation process that can be carried out concurrently without interfering with production activities. The monitoring process involves two methods, namely threshold limit-based monitoring and machine learning-based monitoring, for ensuring product quality during the optimisation process. Meanwhile, Bayesian optimisation (BO) is used for the optimisation process due to its capability for autonomous search for the optimum condition of an unknown and expensive objective function. The case study demonstrated BO’s ability to find the optimal condition with a small number of samples, while the machine learning method used can monitor tool wear and surface roughness with average errors of 5.5% and 7.7%, respectively.
- 1 August 2022
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 61 (16) , 5392-5410
A proposal for improving production efficiency of existing machining line through a hybrid monitoring and optimisation process
In a machining line with a Lean manufacturing environment, such as one used for mass production of automotive components, tool life is determined by a tool counter that is set on the safe side to avoid producing a failed product, so the cutting tools are typically underutilised. On the other hand, cutting parameters optimisation, which has a significant impact on tool life, is not feasible due to the rule that prohibits changing the cutting parameters during mass production to maintain the quality standard. To address both issues, this paper proposes a hybrid monitoring and optimisation process that can be carried out concurrently without interfering with production activities. The monitoring process involves two methods, namely threshold limit-based monitoring and machine learning-based monitoring, for ensuring product quality during the optimisation process. Meanwhile, Bayesian optimisation (BO) is used for the optimisation process due to its capability for autonomous search for the optimum condition of an unknown and expensive objective function. The case study demonstrated BO’s ability to find the optimal condition with a small number of samples, while the machine learning method used can monitor tool wear and surface roughness with average errors of 5.5% and 7.7%, respectively.
- Open Access
- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 125-137
Augmentation of Cross-Sectional Spray Measurements with Discrete Droplet Model Using Ensemble Kalman Filter
Spray flows containing droplets and particles are used in various industrial fields. In this study, we investigate an efficient and reliable way to predict the spray flow of droplets by combining the discrete droplet model (DDM) with ensemble data assimilation for application to such industrial problems. The aim is to augment cross-sectional measurements such as particle image velocimetry (PIV) with fast DDM simulations of droplets. In this paper, we focus on the numerical experiment of data assimilation, which is also known as twin experiments, and discuss how such cross-sectional measurements and DDM can be integrated by the ensemble Kalman filter. The results showed that the position and velocity of the droplet and the spray nozzle's state were estimated by assimilating the time-averaged velocity measurements on the cross-section using a carefully prepared ensemble of droplets. Furthermore, the droplet size distribution was estimated indirectly through DDM.
- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 125-137
Open AccessAugmentation of Cross-Sectional Spray Measurements with Discrete Droplet Model Using Ensemble Kalman Filter
Spray flows containing droplets and particles are used in various industrial fields. In this study, we investigate an efficient and reliable way to predict the spray flow of droplets by combining the discrete droplet model (DDM) with ensemble data assimilation for application to such industrial problems. The aim is to augment cross-sectional measurements such as particle image velocimetry (PIV) with fast DDM simulations of droplets. In this paper, we focus on the numerical experiment of data assimilation, which is also known as twin experiments, and discuss how such cross-sectional measurements and DDM can be integrated by the ensemble Kalman filter. The results showed that the position and velocity of the droplet and the spray nozzle's state were estimated by assimilating the time-averaged velocity measurements on the cross-section using a carefully prepared ensemble of droplets. Furthermore, the droplet size distribution was estimated indirectly through DDM.
- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 112-124
Optimal Design for Suppressing Time Fluctuation Part of Two-Dimensional Jet in Crossflow
This paper proposes a shape optimisation method for suppressing a time fluctuation part of the two-dimensional jet-in-cross flow. Previously, the author has formulated a shape optimisation problem for suppressing transient fluids using Snapshot proper orthogonal decomposition (POD). This study extends it to a two-dimensional jet-in-cross flow. For this purpose, the eigenvalues of Snapshot POD are integrated to a cost function, where the constraint functions are the nonstationary Navier–Stokes equation and the eigenvalue equation of Snapshot POD. An objective functional is described using Lagrange multipliers and the finite element method as the sum of the cost and constraint functions. To verify the proposed optimisation strategy, using the gradient method for domain deformation, all triangles over a mesh are deformed, which causes the cost function to decrease. Finally, the eigenvalues of Snapshot POD in the initial and optimal domains are compared.- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 112-124
Optimal Design for Suppressing Time Fluctuation Part of Two-Dimensional Jet in Crossflow
This paper proposes a shape optimisation method for suppressing a time fluctuation part of the two-dimensional jet-in-cross flow. Previously, the author has formulated a shape optimisation problem for suppressing transient fluids using Snapshot proper orthogonal decomposition (POD). This study extends it to a two-dimensional jet-in-cross flow. For this purpose, the eigenvalues of Snapshot POD are integrated to a cost function, where the constraint functions are the nonstationary Navier–Stokes equation and the eigenvalue equation of Snapshot POD. An objective functional is described using Lagrange multipliers and the finite element method as the sum of the cost and constraint functions. To verify the proposed optimisation strategy, using the gradient method for domain deformation, all triangles over a mesh are deformed, which causes the cost function to decrease. Finally, the eigenvalues of Snapshot POD in the initial and optimal domains are compared.- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 152-166
Bayesian Uncertainty Reduction of Generalised k-ω Turbulence Model for Prediction of Film-Cooling Effectiveness
In this study, Reynolds-averaged Navier-Stokes (RANS) simulations were carried out to predict the film cooling effectiveness of an inclined round jet in crossflow using turbulence model parameters optimised based on measurement data. The posterior distributions of the generalised (GEKO) turbulence model parameters were estimated using a computationally efficient surrogate model with the Markov chain Monte Carlo method, which provides a framework for probabilistic parameter estimation based on measurement data. The results show that using the maximum a posterior parameters for a blowing ratio of 0.5 gives better predictions than using the default parameters of the GEKO model. The estimated parameters were then applied to flows with a higher blowing ratio and different hole geometry to evaluate the generalisation performance. In both cases, the results were improved by properly predicting the spread of the cooling flow.- 7 February 2022
- Published by Taylor & Francis in International Journal of Computational Fluid Dynamics
- Vol. 36 (2) , 152-166
Bayesian Uncertainty Reduction of Generalised k-ω Turbulence Model for Prediction of Film-Cooling Effectiveness
In this study, Reynolds-averaged Navier-Stokes (RANS) simulations were carried out to predict the film cooling effectiveness of an inclined round jet in crossflow using turbulence model parameters optimised based on measurement data. The posterior distributions of the generalised (GEKO) turbulence model parameters were estimated using a computationally efficient surrogate model with the Markov chain Monte Carlo method, which provides a framework for probabilistic parameter estimation based on measurement data. The results show that using the maximum a posterior parameters for a blowing ratio of 0.5 gives better predictions than using the default parameters of the GEKO model. The estimated parameters were then applied to flows with a higher blowing ratio and different hole geometry to evaluate the generalisation performance. In both cases, the results were improved by properly predicting the spread of the cooling flow.- 28 May 2021
- Published by ASME International in Journal of Fluids Engineering
- Vol. 143 (10)
Estimation of Vortex-Induced Vibration Based on Observed Wakes Using Computational Fluid Dynamics-Trained Deep Neural Network
The vibration of a circular cylinder due to fluid forces is of interest in various engineering fields. In this study, we investigate an approach to estimate the fluid forces acting on a circular cylinder in a flow field based on experimental flow visualizations using a deep neural network (DNN). Specifically, the wake patterns and fluid forces are correlated in a computational fluid dynamics (CFD) simulation, and the forces in the experiment are estimated by comparing experimental and computational wake patterns using a DNN. The approach is tested via dye-ink visualization around a circular cylinder at a Reynolds number of 560, referring to Seyed-Aghazadeh et al. (2015, “An Experimental Investigation of Vortex-Induced Vibration of a Rotating Circular Cylinder in the Crossflow Direction,” Phys. Fluids, 27(6), p. 067101). First, the CFD simulation of a circular cylinder with forced vibration in the crossflow direction is conducted with various vibration frequencies. Subsequently, the visualized wake images of the resulting flow fields and corresponding fluid forces are used as training data for the DNN. In the estimation, the images from the experiment are detected by the CFD-trained DNN. Thus, we can recall the correlated fluid forces using CFD simulation. The average drag coefficient and peak value of the lift coefficient estimated using streaming experimental images, have standard deviations of 2.1–13.7% and 6.6–18.6%, respectively, depending on the number of training images. The root-mean-square value of the lift coefficient obtained from the estimation is 0.82, which is comparable to the experimental value of 0.8, under the same flow and oscillation conditions.
- 28 May 2021
- Published by ASME International in Journal of Fluids Engineering
- Vol. 143 (10)
Estimation of Vortex-Induced Vibration Based on Observed Wakes Using Computational Fluid Dynamics-Trained Deep Neural Network
The vibration of a circular cylinder due to fluid forces is of interest in various engineering fields. In this study, we investigate an approach to estimate the fluid forces acting on a circular cylinder in a flow field based on experimental flow visualizations using a deep neural network (DNN). Specifically, the wake patterns and fluid forces are correlated in a computational fluid dynamics (CFD) simulation, and the forces in the experiment are estimated by comparing experimental and computational wake patterns using a DNN. The approach is tested via dye-ink visualization around a circular cylinder at a Reynolds number of 560, referring to Seyed-Aghazadeh et al. (2015, “An Experimental Investigation of Vortex-Induced Vibration of a Rotating Circular Cylinder in the Crossflow Direction,” Phys. Fluids, 27(6), p. 067101). First, the CFD simulation of a circular cylinder with forced vibration in the crossflow direction is conducted with various vibration frequencies. Subsequently, the visualized wake images of the resulting flow fields and corresponding fluid forces are used as training data for the DNN. In the estimation, the images from the experiment are detected by the CFD-trained DNN. Thus, we can recall the correlated fluid forces using CFD simulation. The average drag coefficient and peak value of the lift coefficient estimated using streaming experimental images, have standard deviations of 2.1–13.7% and 6.6–18.6%, respectively, depending on the number of training images. The root-mean-square value of the lift coefficient obtained from the estimation is 0.82, which is comparable to the experimental value of 0.8, under the same flow and oscillation conditions.
- Open Access
- 1 January 2021
- Published by Japan Society of Mechanical Engineers in Transactions of the JSME (in Japanese)
- Vol. 87 (898) , 21-00010
Thermal state estimation based on Assisted Ensemble Kalman Filter
Thermal analysis of spacecraft is one of the most important processes to ensure the safety of spacecraft systems. However, the thermal mathematical model has uncertainty such as thermal contact conductance or thermal optical properties. These uncertain parameters in the model are non-negligible for long-term missions because these parameters can change during operation on orbit. Despite the uncertainties, the spacecraft system has only a few onboard temperature sensors compared to large and complex systems. In this study, an advanced thermal analysis method based on data assimilation is proposed to estimate the thermal state of a complex system with limited temperature data. Firstly, this paper describes a new state estimation algorithm called Assisted Ensemble Kalman Filter, which is an advanced state estimation algorithm based on Ensemble Kalman Filter (EnKF). Here, an external estimation algorithm by calculating the heat balance equation was applied to the conventional method to improve the estimation performance of the EnKF. Secondly, we propose a new parameter that indicates observability based on heat flux and temperature sensitivity, and the influence of temperature sensor location on estimation performance was discussed. These proposed approaches were applied to a simple thermal mathematical model, and numerical experiments have confirmed their availability.
- 1 January 2021
- Published by Japan Society of Mechanical Engineers in Transactions of the JSME (in Japanese)
- Vol. 87 (898) , 21-00010
Open AccessThermal state estimation based on Assisted Ensemble Kalman Filter
Thermal analysis of spacecraft is one of the most important processes to ensure the safety of spacecraft systems. However, the thermal mathematical model has uncertainty such as thermal contact conductance or thermal optical properties. These uncertain parameters in the model are non-negligible for long-term missions because these parameters can change during operation on orbit. Despite the uncertainties, the spacecraft system has only a few onboard temperature sensors compared to large and complex systems. In this study, an advanced thermal analysis method based on data assimilation is proposed to estimate the thermal state of a complex system with limited temperature data. Firstly, this paper describes a new state estimation algorithm called Assisted Ensemble Kalman Filter, which is an advanced state estimation algorithm based on Ensemble Kalman Filter (EnKF). Here, an external estimation algorithm by calculating the heat balance equation was applied to the conventional method to improve the estimation performance of the EnKF. Secondly, we propose a new parameter that indicates observability based on heat flux and temperature sensitivity, and the influence of temperature sensor location on estimation performance was discussed. These proposed approaches were applied to a simple thermal mathematical model, and numerical experiments have confirmed their availability.
- 8 June 2020
- Published by American Institute of Aeronautics and Astronautics (AIAA)
Data Assimilation for Clear Air Turbulence by Upstream LIDAR Observation
We numerically investigated possibility of a data assimilation methodology for reproduction of Clear Air Turbulence (CAT) observed by LIDAR. The three-dimensional vortices induced by the shear layer instability were predicted by combining pseudo Lidar observation and the 6th-order accuracy numerical simulation based on the Euler equations by using the data assimilation. The assimilation performance was verified in identical-twin experiments; the true state and the observation data were generated artificially. We considered three methods for pseudo observation to verify the sensitivity of observed variables to prediction accuracy: observing only horizontal velocity, horizontal and vertical velocity components on the flight path, and wind speeds projected on the LIDAR’s line of sight. The results of the identical-twin experiments indicated that observing u and vertical wind components reduced the forecast error, and the data assimilation was able to restore the information of these two essential wind components from a wind speed data observed by LIDAR when its line of sight was inclined in the z-direction.
- 8 June 2020
- Published by American Institute of Aeronautics and Astronautics (AIAA)
Data Assimilation for Clear Air Turbulence by Upstream LIDAR Observation
We numerically investigated possibility of a data assimilation methodology for reproduction of Clear Air Turbulence (CAT) observed by LIDAR. The three-dimensional vortices induced by the shear layer instability were predicted by combining pseudo Lidar observation and the 6th-order accuracy numerical simulation based on the Euler equations by using the data assimilation. The assimilation performance was verified in identical-twin experiments; the true state and the observation data were generated artificially. We considered three methods for pseudo observation to verify the sensitivity of observed variables to prediction accuracy: observing only horizontal velocity, horizontal and vertical velocity components on the flight path, and wind speeds projected on the LIDAR’s line of sight. The results of the identical-twin experiments indicated that observing u and vertical wind components reduced the forecast error, and the data assimilation was able to restore the information of these two essential wind components from a wind speed data observed by LIDAR when its line of sight was inclined in the z-direction.