Multidisciplinary Design Optimization of Renewable Energy Systems

Because of the concerns about the negative environmental impacts of fossil fuels, wind energy generation has been increased at a faster pace. Despite the fact that wind energy has evolved considerably in recent years, its cost in general is higher compared to fossil fuels. Therefore, many advancements are required to design reliable and economic wind turbines to compete with fossil fuels and mitigate the global warming. A promising technique that can contribute to the economical design of wind turbines is multidisciplinary design optimization. Prior experience with design optimization of wind turbines with all disciplines, components and constraints involved to minimize the cost of energy is not well-established.

Our research addresses some shortcomings in this field by developing an integrated multidisciplinary design optimization method to simultaneously design different components and disciplines at the same time to reduce the cost of energy significantly. The design made by our approach is superior to the design made by optimizing each component or discipline independently, and our research is the first work that developed and employed such a capability. Our results published in the prestigious journals in the field of renewable energy proves the effectiveness of the developed methodology, and it has resulted in a significant reduction in the cost of energy.

We used this technique to design the world's largest wind turbine that is publicly available to worldwide researchers, and the computational codes and its corresponding design data can be found from below links:

Wind turbine model and data; Research paper describing the approach

Physics-Based Aerodynamic Shape Optimization

Currently, most wind turbine blades are designed using low-fidelity models such as blade-element-momentum theory, due to lower computational cost, ease of implementation, and fast convergence time to a feasible solution. However, such models do not capture compressibility, viscosity, and three-dimensional effects present in wind turbine operation, and this results in a suboptimal design. These important design considerations can only be predicted accurately using computational fluid dynamics techniques.

In this work, we overcome the challenges of CFD-based wind turbine optimization for the first time by developing an efficient and robust aerodynamic shape optimization technique. We used an arbitrary Lagrangian–Eulerian Reynolds-averaged Navier-Stokes model for the CFD analysis, and a discrete adjoint method to efficiently compute the gradients of the function of interest. This novel approach captures the complex flow phenomena present in wind turbine operation, and it results in a more optimized design, and ultimately more cost savings.

This is a collaborative work with researchers from University of Michigan, Ann Arbor, Delft University of Technology, and Arkansas Tech University.

Uncertainty Quantification of Engineering Structures

Monte Carlo simulation is the traditional technique to approximate the probability of outcomes by running multiple random trials. In this implementation, computational cost increases significantly with the addition of new sources of input uncertainty. We developed a new uncertainty quantification methodology based on C-type Gram-Charlier series expansion and dimensional reduction technique to reduce significantly the computational cost of high dimensional uncertainties.

Our results show that the proposed approach can accurately estimate the probability distribution for both the Gaussian and non-Gaussian input distributions. The results demonstrate comparable accuracy with those from direct Monte Carlo simulation, while the computational cost is several hundred times lower.

This is a collaborative work with researchers from University of Texas at Dallas, Dalian Jiaotong University, and Arkansas Tech University.

Optimal Controls

Many energy devices like wind, wave and gas turbines often operate in suboptimal conditions, which is referred to as performance degradation. This is because of factors such as device dirt and dust accumulation, erosion, and corrosion of surfaces. Performance degradation reduces the energy production, and it increases the cost of energy. Almost all the controllers used today for energy devices are model-based. A model-based controller cannot cope with performance degradation, because it uses a model of the ideal state of the device, and it neglects these effects of unknown nature.

In this study, we focused on increasing the energy production from wind turbines to reduce the cost of generated electricity per kilo-Watt-hour using a model-free control algorithm. We used the existing software and hardware of the wind turbine to search online for the optimal condition of the wind turbine after such effects happen. Our results show that when performance degradation happens 7% energy lost is experienced that translates into 7% increase of the cost of energy. Our algorithm can reduce up to 1% of the energy loss. This in turn reduces the cost of energy by 1% when performance degradation happens.

This is a collaborative work with researchers from University of Texas at Dallas, and Arkansas Tech University.

Wind Farm Power and Loads Optimization using Yaw Controls

Clustering wind turbines as a wind farm to share the infrastructure is an effective strategy to reduce the cost of energy. However, this results in aerodynamic wake interaction among wind turbines. Yawing the upstream wind turbines can mitigate the losses in wind farm power output. Yaw-misalignment also affects the loads, as partial wake overlap can increase fatigue of downstream turbines.

In this research, we study multi-objective optimization of wind farm wakes using yaw-misalignment to increase power production, and reduce loads due to partial wake overlap. This is achieved using a computational framework consisting of an aerodynamic model for wind farm wake, blade-element-momentum model to compute the power and the loads, and a gradient-based optimizer. Our preliminary results show that yaw-misalignment is capable of increasing the power production of the wind farm, while reducing the loading due to partial wake overlap.

This is a collaborative work with researchers from Delft University of Technology, University of Texas at Dallas, and Arkansas Tech University.