Ashwin Salvi

Optimal Energy Management for a Hybrid Vehicle Using Neuro-Dynamic Programming to Consider Transient Engine Operation

This paper proposes a self-learning approach to develop optimal power management with multiple objectives, e.g. to minimize fuel consumption and transient engine-out NOx and particulate matter emission for a series hydraulic hybrid vehicle. Addressing multiple objectives is particularly relevant in the case of a diesel powered hydraulic hybrid since it has been shown that managing engine transients can significantly reduce real-world emissions. The problem is formulated as an infinite time horizon stochastic sequential decision making/markovian problem. The problem is computationally intractable by conventional Dynamic programming due to large number of states and complex modeling issues. Therefore, the paper proposes an online self-learning neural controller based on the fundamental principles of Neuro-Dynamic Programming (NDP) and reinforcement learning. The controller learns from its interactions with the environment and improves its performance over time. The controller tries to minimize multiple objectives and continues to evolve until a global solution is achieved. The control law is a stationary full state feedback based on 5 states and can be directly implemented. The controller performance is then evaluated in the Engine-in-the-Loop (EIL) facility.Copyright

Real-Time Transient Soot and NOx Virtual Sensors for Diesel Engine Using Neuro-Fuzzy Model Tree and Orthogonal Least Squares

Diesel engine combustion and emission formation is highly nonlinear and thus creates a challenge related to engine diagnostics and engine control with emission feedback. This paper presents a novel methodology to address the challenge and develop virtual sensing models for engine exhaust emission. These models are capable of predicting transient emissions accurately and are computationally efficient for control and optimization studies. The emission models developed in this paper belong to the family of hierarchical models, namely “neuro-fuzzy model tree”. The approach is based on divide-and-conquer strategy i.e. to divide a complex problem into multiple simpler subproblems, which can then be identified using simpler class of models. Advanced experimental setup incorporating a medium duty diesel engine is used to generate training data. Fast emission analyzers for soot and NOX provide instantaneous engine-out emissions. Finally, the Engine-In-theLoop is used to validate the models for predicting transient particulate mass and NOX.

Hydraulic Hybrid Powertrain-In-the-Loop Integration for Analyzing Real-World Fuel Economy and Emissions Improvements

The paper describes the approach, addresses integration challenges and discusses capabilities of the Hybrid Powertrain-in-the-Loop (H-PIL) facility for the series/hydrostatic hydraulic hybrid system. We describe the simulation of the open-loop and closed-loop hydraulic hybrid systems in H-PIL and its use for concurrent engineering and development of advanced supervisory strategies. The configuration of the hydraulic–hybrid system and details of the hydraulic circuit developed for the H-PIL integration are presented. Next, software and hardware interfaces between the real components and virtual systems are developed, and special attention is given to linking component-level controllers and system-level supervisory control. The H-PIL setup allows imposing realistic dynamic loads on hydraulic pump/motors and accumulator based on vehicle driving schedule. Application of fast analyzers allows characterization of the impact of dynamic interactions in the propulsion system on engine-out emissions. Therefore, the H-PIL facility allows optimization of the hybrid system for both high-efficiency and low emissions. The impetus is provided by previous work showing that more than half of the soot emissions from a conventional diesel powertrain over the urban driving schedule can be attributed to transients. The setup includes a 6.4L V-8 International diesel engine, highly dynamic dynamometer, Radial piston pump/motors supplied by Bosch-Rexroth and dSPACE real-time environment with in-house developed simulation of the virtual vehicle.

An Iterative Learning Control Approach to Improving Fidelity in Internet-Distributed Hardware-in-the-Loop Simulation

Abstract : One of the main challenges of co-simulating hardware-in-the-loop systems in real-time over the Internet is the fidelity of the simulation. The dynamics of the Internet may significantly distort the dynamics of the network-integrated system. This paper presents the development of an iterative learning control based approach to improve fidelity of such networked system integration. Towards this end, a new metric for characterizing fidelity is proposed first, which, unlike some existing metrics, does not require knowledge about the reference dynamics (i.e., dynamics that would be observed, if the system was physically connected). Next, using this metric, the problem of improving fidelity is formulated as an iterative learning control problem. Finally, the proposed approach is illustrated on a purely simulation-based case study. The conclusion is that the proposed approach holds significant potential for achieving high fidelity levels.

Reducing Soot Emissions in a Diesel Series Hybrid Electric Vehicle Using a Power Rate Constraint Map

This paper considers a diesel series hybrid electric vehicle (SHEV) and proposes the utilization of an engine-generator power rate constraint map to reduce soot emissions without a significant compromise in fuel economy. Specifically, model predictive control (MPC) is used to split the vehicle power demand between the engine-generator unit and the battery. To achieve a reduction in soot, the engine-generator power rate is constrained. Unlike existing strategies, the power rate limit is not a fixed value but varies, depending on the power level, resulting in a map. This constraint map is designed by formulating the soot emission reduction problem as an optimization problem, which is solved through a three-step offline discrete optimization process. The optimization relies on a quasi-static soot emissions map that captures the trends, even during transients, but underestimates the magnitudes. Therefore, to evaluate the performance of the MPC-based power management with the power rate constraint map, experiments are conducted through an engine-in-the-loop simulation framework. Experimental results show that compared with a constant power rate constraint, soot emissions can be reduced by 44.5% while compromising fuel economy by only 0.3% through the proposed approach. As a tradeoff, the ampere-hour (Ah) processed in the battery, which is a variable that has been shown to correlate with battery capacity loss, increases by 5.5%.

Effect of Volatiles on Soot Based Deposit Layers

The implementation of exhaust gas recirculation (EGR) coolers has recently been a widespread methodology for engine in-cylinder NOX reduction. A common problem with the use of EGR coolers is the tendency for a deposit, or fouling layer to form through thermophoresis. These deposit layers consist of soot and volatiles and reduce the effectiveness of heat exchangers at decreasing exhaust gas outlet temperatures, subsequently increasing engine out NOX emission.This paper presents results from a novel visualization rig that allows for the development of a deposit layer while providing optical and infrared access. A 24-hour, 379 micron thick deposit layer was developed and characterized with an optical microscope, an infrared camera, and a thermogravimetric analyzer. The in-situ thermal conductivity of the deposit layer was calculated to be 0.047 W/mK. Volatiles from the layer were then evaporated off and the layer reanalyzed. Results suggest that volatile bake-out can significantly alter the thermo-physical properties of the deposit layer and hypotheses are presented as to how.Copyright

Virtual sensors for transient diesel soot and NO x emissions: Neuro-fuzzy model tree with automatic relevance determination

The paper describes development of virtual sensors for transient diesel particulate and NOX emissions. The emission models developed in this paper belong to the family of hierarchical models, namely “neuro-fuzzy model tree”. The modeling techniques are motivated by the idea of divide and conquer the input-output space. The complex problem is divided into multiple simpler subproblems, which are then identified using simpler class of models. A specially designed multi-pseudo random perturbation signal and experimental tests are proposed to generate training data. The diesel engine is tested using integrated hardware and software tools for automated testing with high speed data recording. The engine out transient NOX and soot emission is recorded using fast emission analyzers. The data is then used to construct neuro-fuzzy model with Gaussian validity functions and local neural networks. An automatic relevance determination (ARD) derived from Bayes framework is derived and applied for choosing appropriate model inputs and reducing the model complexity. Finally, the model is validated with testing data recorded during Engine-in-the-Loop (EIL) testing of engine coupled to virtual hybrid powertrain. It is shown that the prediction accuracy of the proposed models, both qualitatively and quantitatively, are very good with low computational cost.

Engine-in-the-Loop Validation of a Frequency Domain Power Distribution Strategy for Series Hybrid Powertrains

Abstract This paper presents an engine-in-the-loop validation of a power management strategy for hybrid electric powertrains that splits the power demanded by driver between the engine and battery depending on the frequency content. In particular, a series hybrid electric powertrain is considered where the engine is a real 6.4L diesel engine and the rest of the vehicle, including the battery, generator, motors, vehicle dynamics, and driver, is modeled in computer. A networked engine-in-the-loop experiment is considered where the engine and generator constitute one site, and the rest of the system constitutes another site. This networked setup is used to compare the abovementioned power management strategy to a thermostatic strategy as the baseline. For the specific drive cycle considered, the proposed strategy yields about 12% increase in fuel economy, a performance that exceeds the previously reported purely model-based simulation results. In addition, an improvement in battery life can also be expected.

In-situ thermophysical properties of an evolving carbon nanoparticle based deposit layer utilizing a novel infrared and optical methodology

The use of exhaust gas recirculation (EGR) in internal combustion engines has significant impacts on engine combustion and emissions. EGR can be used to reduce in-cylinder NOx production, reduce fuel consumption, and enable advanced forms of combustion. To maximize the benefits of EGR, the exhaust gases are often cooled with liquid to gas heat exchangers. However, the build up of a fouling deposit layer from exhaust particulates and volatiles result in the decrease of heat exchanger efficiency, and increase the outlet temperature of the exhaust gases, and decrease the advantages of EGR.This paper presents experimental data from a novel in-situ measurement technique in a visualization rig during the development of a 378 micron thick deposit layer. Measurements were performed every 6 hours for up to 24 hours. Results show a non-linear increase in deposit thickness with an increase in layer surface area as deposition continued. Deposit surface temperature and temperature difference across the thickness of the layer was shown to increase with deposit thickness while heat transfer decreased. The provided measurements combine to produce deposit thermal conductivity.A thorough uncertainty analysis of the in-situ technique is presented and suggests higher measurement accuracy at thicker deposit layers and with larger temperature differences across the layer. The interface and wall temperature measurements are identified as the strongest contributors to the measurement uncertainty. Due to instrument uncertainty, the influence of deposit thickness and temperature could not be determined. At an average deposit thickness of 378 microns and at a temperature of 100°C, the deposit thermal conductivity was determined to be 0.044 ± 0.0062 W/mK at a 90% confidence interval based on instrument accuracy.Copyright

Visual Study of In-Situ EGR Cooler Fouling Layer Evolution

Exhaust gas recirculation (EGR) is a major technology to reduce NOx from diesel engines for future emission standards. The implementation of EGR coolers has been a common methodology to provide engine in-cylinder NOx reduction. However, EGR cooler fouling is a common problem. The particulate matter in exhaust tends to form a deposit layer on the wall of the heat exchangers. This effect leads to a reduction of thermal effectiveness of the heat exchangers resulting in insufficient EGR cooling and subsequently higher engine NOx emission.This paper utilized a unique test rig offering visible and infrared optical access to the deposit layer in a simulated diesel EGR cooler to study the evolution of the layer from fresh to heavy deposit. A 460μm thick deposit layer was built during a 37 hour exposure. Time lapse videos were taken provide visual in-situ evidence for the investigation of the layer thickness development and morphology change during the deposition. The layer growth tended to stabilize from about 22 hours of deposition. The shear force exerted by the gas flow moves surface particles of 20μm in diameter or larger. This could contribute to the stabilization phenomenon.Copyright