1. Yu-Wei Wu Yi-Yu Shi Sudip Roy Tsung-Yi Ho3
Obstacle-Avoiding Wind Turbine Placement for Power-Loss and Wake-Effect Optimization
Yu-Wei Wu
Yi-Yu Shi
Sudip Roy
Tsung-Yi Ho3
ASP-DAC’15

2. Outline Introduction Preliminaries Proposed Method Experimental result
Conclusions

3. Introduction
The main difficulty in designing a wind farm is the wind turbine placement problem that would directly influence the produced power.
The most considered factor in the wind turbine placement problem is to reduce the wake effect.

4. Introduction
When the wind flows through the wind turbines, the wake effect is caused due to the fact that each wind turbine produces abnormal turbulence behind it and effects downstream indirectly.
Hence, minimizing the wake effect can lead to maximizing the power generation.

5. Introduction
In the wind turbine placement problem the blocks are in irregular shape and they can overlap without violation of the center of the wind turbine into the obstacle.
Along with the wake effect, we additionally consider wirelength and obstacle-avoiding at the same time.

7. Preliminaries
Let X = {x1, , xn} and Y = {y1, , yn} be the sets of x and y-coordinates of n wind turbines with m obstacles in the plane.
We use the penalty function to incorporate the constraint and wirelength cost as

8. Wirelength Model
As the half-perimeter wirelength(HPWL) is not differentiable, we use the log-sum-exp wirelength model [10] in our wirelength computation given by Eq.(2).

9. Bell-shaped Function
In order to smooth the non-differentiable and non-smoothing density, we use the bell-shaped function [11] as follows

10. Constraints
Constraint 1: Wind turbines may be damaged by turbulence when they are too close to each other.
Constraint 2: An obstacle means that the location is not suitable to set up the wind turbine.

11. Power Output
We use the Park wake model [4] to evaluate the wake effects.
The expected value of power generation (η) of the whole wind farm is given by

12. Proposed Method Region Separation Based on Wake Effect
Obstacle-Avoidance Method
Gradually Approximation Method
Wind Turbine Placement

13. Region Separation Based on Wake Effect

14. We consider the wake effect block as an irregular shape in the plane
We consider the wake effect block as an irregular shape in the plane. We adjust the size of wake effect block (wv) for computing the wake effect density in a bin as follows

15. 12

16. Obstacle-Avoidance Method
After obtaining the next location of a wind turbine by separateRegion(),the center of the wind turbine may move on to the obstacle Ob.
In the next round, the wind turbine may escape from the obstacle Ob and again it hides.

17. Gradually Approximation Method
we diffuse the wind turbines through gradually increasing the wake effect distances using a procedure increasingWakeEffect(). Fig.7 shows how the wake effect distances gradually increase.

18. Wind Turbine Placement

19. Experimental result
We implement the proposed method using C++ on a 64-bit Linux machine with 3.4-GHz processor and 8GB RAM.

22. Conclusion
In this paper, we have proposed an algorithm to solve the wind turbine placement problem with the help of the analytical method and based on the gradually approximation method.
In the best of our knowledge, for the first time, we have considered wake effect, wirelength and obstacle-avoiding all together at the same time.