1
00:00:10,490 --> 00:00:14,106
Hello, in this six-minute
video we are going to see how

2
00:00:14,106 --> 00:00:18,052
the mesh preset of Heliciel's
CFD bench changes the result,

3
00:00:18,052 --> 00:00:21,734
on one and the same wind
turbine calculated five times.

4
00:00:22,866 --> 00:00:31,657
Here is a wind turbine 4 metres in
diameter, modelled in Heliciel;

5
00:00:31,657 --> 00:00:41,646
it has 3 blades and turns at 166.8 rpm
in a wind of 5.5 metres per second.

6
00:00:41,696 --> 00:00:47,939
By the blade element method, the
designer calculates 231 newtons

7
00:00:47,939 --> 00:00:53,596
of thrust on the blades and
35.9 newton-metres of torque.

8
00:00:53,921 --> 00:00:58,367
Let's see what the CFD bench finds,
depending on how fine the mesh is.

9
00:01:00,442 --> 00:01:05,444
In the CFD bench, the only decision
to make is the mesh preset.

10
00:01:05,494 --> 00:01:09,366
Each level cuts the space more
finely around the blades,

11
00:01:09,366 --> 00:01:15,105
and Heliciel announces for each one the
number of cells and the expected duration.

12
00:01:15,155 --> 00:01:24,824
We have calculated the same wind
turbine with five of these levels:

13
00:01:24,824 --> 00:01:28,431
Levels 1, 2, 3, 5 and 7.

14
00:01:30,548 --> 00:01:36,324
At the first level, Level 1, about
21,000 cells fill the domain,

15
00:01:36,324 --> 00:01:39,662
and the mesh is built in one minute.

16
00:01:39,712 --> 00:01:44,684
Around the blades, the cells remain large.

17
00:01:44,734 --> 00:01:48,062
On the blade, the pressure
is only a sketch,

18
00:01:48,062 --> 00:01:53,052
and the thrust does not reach half
of the designer's: 99 newtons.

19
00:01:54,860 --> 00:02:00,234
The cut already shows the wake behind
the rotor, but in coarse blocks.

20
00:02:00,284 --> 00:02:03,991
As for the torque, it even
comes out with the wrong sign:

21
00:02:03,991 --> 00:02:06,202
at this level, it cannot be used.

22
00:02:08,517 --> 00:02:14,400
At the second level, about 47,000 cells
and two minutes of meshing:

23
00:02:14,400 --> 00:02:17,649
the mesh tightens around the blades.

24
00:02:17,699 --> 00:02:22,905
Across the axis, the cut shows the
disc of slowed wind behind the rotor.

25
00:02:22,955 --> 00:02:28,791
The thrust rises to 192 newtons,
17 percent below the designer.

26
00:02:31,921 --> 00:02:39,693
At the third level, about 84,000 cells
and four and a half minutes of meshing.

27
00:02:39,743 --> 00:02:44,427
On the face turned to the wind, the
blade is now clearly in overpressure;

28
00:02:45,725 --> 00:02:52,605
on its back, the air is sucked,
above all towards the blade tips.

29
00:02:52,655 --> 00:02:56,289
A cylindrical cut follows
the wind at a given radius:

30
00:02:56,289 --> 00:02:59,032
it slows down as it crosses the blades.

31
00:02:59,082 --> 00:03:04,950
The thrust reaches 214 newtons,
7 percent below the designer.

32
00:03:07,121 --> 00:03:13,903
At the fifth level, about 235,000 cells
and seven minutes of meshing.

33
00:03:13,953 --> 00:03:16,034
On the face turned to the wind,

34
00:03:16,034 --> 00:03:20,733
the overpressure is now detailed
right up to the edges of the blades;

35
00:03:20,783 --> 00:03:24,875
on the back, the low pressure
gathers towards the blade tips,

36
00:03:24,875 --> 00:03:27,357
where the relative speed is highest.

37
00:03:28,399 --> 00:03:32,582
The flow lines slow down as they cross
the rotor and spread out behind it:

38
00:03:32,582 --> 00:03:36,313
it is the signature of a machine
that takes energy from the wind.

39
00:03:36,363 --> 00:03:42,439
The thrust joins the designer
within 4 percent: 221 newtons.

40
00:03:42,489 --> 00:03:44,013
In the vertical plane,

41
00:03:44,013 --> 00:03:48,380
the axial velocity cut and the
flow lines tell the same story:

42
00:03:48,380 --> 00:03:52,469
a wide wake of slowed wind
stretches far behind the rotor.

43
00:03:52,519 --> 00:03:58,355
The iso-surfaces wrap the zones of
low pressure around the blades.

44
00:04:00,432 --> 00:04:03,756
The second phase makes
the rotor really turn,

45
00:04:03,756 --> 00:04:08,188
and it confirms the thrust of
the first phase: 225 newtons.

46
00:04:10,317 --> 00:04:14,151
Let's finally go up to the seventh level,

47
00:04:14,151 --> 00:04:19,013
on the same wind turbine,
set this time to 180 rpm.

48
00:04:19,063 --> 00:04:23,952
More than 15 million cells:
the cuts become much sharper,

49
00:04:23,952 --> 00:04:28,497
and the wake can be read
down to its finest details.

50
00:04:28,547 --> 00:04:33,973
The price matches: more than seven hours
of meshing on a sixteen-core processor,

51
00:04:33,973 --> 00:04:36,957
and an hour and a half
for the first phase.

52
00:04:37,007 --> 00:04:43,723
It is an overnight calculation, which we
keep for the final check of a project.

53
00:04:45,949 --> 00:04:50,802
Let's put the figures side by
side. The thrust converges fast:

54
00:04:50,802 --> 00:04:56,985
from 57 percent of gap at the first level,
it falls to 4 percent at the fifth.

55
00:04:57,035 --> 00:05:02,171
The torque, however, stays far from
the designer even at the fifth level,

56
00:05:02,171 --> 00:05:06,463
and the bench flags it with its
“torque not reliable” alert:

57
00:05:06,463 --> 00:05:10,579
it needs an even finer mesh,
able to describe the boundary

58
00:05:10,579 --> 00:05:14,695
layer that carries the suction
on the back of the blades.

59
00:05:17,111 --> 00:05:21,751
With Heliciel, we can start with a
quick mesh to rough out a project,

60
00:05:21,751 --> 00:05:23,903
then go up a level to conclude:

61
00:05:23,903 --> 00:05:26,897
a result that no longer
moves from one level

62
00:05:26,897 --> 00:05:29,619
to the next is a result we can rely on.