II. Theoretical & Mathematical Development
2.1 MHD Physics Review in Tokamak
2.1.7 Properties of large aspect ratio in tokamak
In tokamak, the aspect ratio, ๐ 0/๐, is assumed to be large. Introducing a small number, ๐, as an inverse large aspect ratio of tokamak, ๐ โก ๐/๐ โช 1 in tokamak.
The safety factor, ๐ is the order of 1 as ๐ = ๐๐ต๐0
๐ 0๐ต๐ ~ ๐๐ต๐0 ๐ต๐ ~ 1.
So, ๐ต๐ is the order of
๐ต๐ ~ ๐๐ต๐. (2.36)
For the equilibrium equation, โ๐ = ๐ฃร๐, each term of Eq. (2.26) is equivalent as d๐
d๐ ~ 1 ๐0{1
๐
๐
๐๐(๐๐ต๐)} ๐ต๐ (2.37)
and
d๐ d๐ ~ 1
๐0
๐๐ต๐
๐๐ ๐ต๐. (2.38)
From Eq. (2.37), ๐ is the order of
๐ ~ ๐ต๐2
๐0. (2.39)
So, poloidal beta, ๐ฝ๐, is the order of 1 as
๐ฝ๐ ~ 1 (2.40)
and toroidal beta, ๐ฝ๐ก, is the order equivalent as
๐ฝ๐ก = ๐ต๐๐2
๐ต๐๐2 ๐ฝ๐ ~ ๐2 (2.41)
and beta, ๐ฝ, is the order equivalant as
๐ต๐๐2 ๐ต๐๐2 2
(2.42)
From Eq. (2.38), ๐๐ต๐/๐๐ is the order of
๐๐ต๐
๐๐ ~ ๐0 1 ๐ต๐
d๐
d๐ ~ ๐0 1 ๐ต๐
๐ต๐2 ๐0
1 ๐ ~ 1
๐ต๐(๐2๐ต๐2)1 ๐ ~ 1
๐ต๐(๐2๐ต๐2)1
๐ ~ ๐2๐ต๐ ๐ .
From Ampereโs law, ๐ฃ = 1/๐0(โร๐) each componenets of current, ๐ฃ in Eq. (2.28) are equivalent as ๐๐ = 1
๐0
๐๐ต๐
๐๐ ~ ๐2 ๐ต๐
๐0๐ ~ ๐2 ๐ต
๐0๐, (2.43)
๐๐= 1 ๐0{1
๐
๐
๐๐(๐๐ต๐)} ~ ๐ต๐
๐0๐ ~ ๐ ๐ต๐
๐0๐ ~ ๐ ๐ต
๐0๐. (2.44)
So, ๐๐ is the order of
๐๐ ~ ๐๐๐. (2.45)
Summarizing section 2.1.7, leading order relation is shown in Table 2.2.
Table 2.2 Leading order relations in large aspect ratio tokamak.
Safety factor ๐ ~ 1
Magnetic field ๐ต๐ ~ ๐๐ต๐
Pressure ๐ ~ ๐ต๐2
๐0
Beta, ๐ฝ
๐ฝ๐ ~ 1 ๐ฝ๐ก ~ ๐2 ๐ฝ ~ ๐2
Current density
๐๐ ~ ๐2 ๐ต ๐0๐ ๐๐ ~ ๐ ๐ต
๐0๐
These properties are used to approximate the equations. The MHD equations can be solved more simply by the approximation of large aspect ratio in tokamak.
This concludes the basic physical concept review of tokamak. In this section, the principle of magnetic fields induced by currents was reviewed. Based on the magnetic fields, the concept and the role of the safety factor and beta value were also reviewed. The properties of large aspect ratio in tokamak was also reviewed. Based on this knowledge, MHD instability in the tokamak will be reviewed to study the sawtooth instability.
2.2 Review of MHD instability in tokamak
In this section, the basics of MHD instability will be reviewed.
2.2.1 Basic idea of MHD stability
Whether the equilibrium is stable or unstable depends on whether a small perturbation is growing or is damping [28]. When the perturbation is growing, the plasma is in an unstable condition whereas when the perturbation is damping, the plasma is in a stable condition. Sawtooth oscillation is the result of instability, which shows periodic behavior.
The basic idea about stability is shown in Figure 2.4.
(a) (b)
Figure 2.4 The basic idea of MHD stability. (a) Stable equilibrium, (b) unstable equilibrium.
This can be illustrated by thinking about a ball, as shown in Figure 2.4. The ball tends to move in the direction which gives the potential energy to the outer system. In Figure 2.4 (a), the ball is perturbed and potential energy increases (๐ฟ๐ > 0). As time goes by, the ball will return to its original point. The ball is determined as stable. In Figure 2.4 (b), the ball is perturbed and potential energy deceases (๐ฟ๐ <
0). As time goes by, the ball will roll away.
Stable Equilibrium
๐ฅ0 ๐ฅ0+ ๐
Unstable Equilibrium
๐ฅ0 ๐ฅ0+ ๐
2.2.2 Energy principle
Based on the perturbation introduced in section 2.2.1, the energy principle is introduced here. The energy principle is used to determine stability by the sign of change of potential energy, ๐ฟ๐. By energy conservation, the summation of the potential energy change and the kinetic energy change is zero. The change of kinetic energy from the perturbation is
๐ฟ๐พ =1
2โซ ๐ โ ๐1(๐)d๐
where ๐1 is the first order perturbation of force density. According to the law of energy conservation, the potential energy comes from the result of the plasma displacement, ๐ as
๐ฟ๐ = โ1
2โซ ๐ โ ๐1(๐)d๐. (2.46) The change of potential energy, ๐ฟ๐ is derived from ideal MHD equations and small displacement, ๐ . Using the perturbation theory, but neglecting the second order term and putting the equilibrium velocity to zero, u0(r) = 0, since the plasma is in static equilibrium. Thus
๐ฎ(๐ซ, ๐ก) = ๐ฎ1(๐ซ, ๐ก), (2.47)
๐(๐ซ, ๐ก) = ๐0(๐ซ) + ๐1(๐ซ, ๐ก), (2.48) ๐(๐ซ, ๐ก) = ๐0(๐ซ) + ๐1(๐ซ, ๐ก), (2.49) ๐(๐ซ, ๐ก) = ๐0(๐ซ) + ๐1(๐ซ, ๐ก), (2.50)
The mass conservation, separating equilibrium term and first order perturbation term, is given as
๐๐0
๐๐ก +๐๐1
๐๐ก + (๐ฎ1โ ๐)๐0= โ๐0๐ โ ๐ฎ1 The equilibrium term for mass conservation is
๐๐0
๐๐ก = 0,
For momentum conservation, the plasma is assumed to be in laminar flow, then (๐ฎ โ ๐)๐ฎ = 0 and separating equilibrium term and first order perturbation term, the momentum conservation is expressed as
๐0๐๐ฎ1
๐๐ก = ๐ฃ0ร๐0+ ๐ฃ0ร๐1+ ๐ฃ1ร๐0โ ๐๐0โ ๐๐1 The equilibrium term for momentum conservation is
๐๐0= ๐ฃ0ร๐0
and the first order perturbation term of momentum conservation is ๐0๐๐ฎ1
๐๐ก = ๐ฃ0ร๐1+ ๐ฃ1ร๐0โ ๐๐1. (2.52)
The adiabatic equation, separating equilibrium term and first order perturbation term neglecting second order perturbation term is
๐๐0
๐๐ก +๐๐1
๐๐ก + (๐ฎ1โ ๐)๐0= โ๐พ๐0๐ โ ๐ฎ1. The equilibrium term for the adiabatic equation is
๐๐0
๐๐ก = 0
and the first order perturbation term of the adiabatic equation is
๐๐1
๐๐ก = โ๐ฎ1โ ๐๐ โ ๐พ๐๐ โ ๐ฎ1. (2.53)
Combining Faradayโs law and Ohmโs law,
๐๐
๐๐ก = โ๐ร(โ๐ฎร๐) = ๐ร(๐ฎร๐).
Separating equilibrium term and first order perturbation term and neglecting second order perturbation term, the equation is
๐๐0
๐๐ก +๐๐1
๐๐ก = ๐ร(๐ฎ1ร๐0).
and the first order perturbation term of the equation is
๐๐1
๐๐ก = ๐ร(๐ฎ1ร๐0). (2.54)
The first order perturbation of ideal MHD equations are as below.
๐๐1
๐๐ก = โ๐ฎ1โ ๐๐0โ ๐0๐ โ ๐ฎ1 (2.51) ๐0๐๐ฎ1
๐๐ก = ๐ฃ0ร๐1+ ๐ฃ1ร๐0โ ๐๐1 (2.52)
๐๐1
๐๐ก = โ๐ฎ1โ ๐๐ โ ๐พ๐๐ โ ๐ฎ1 (2.53)
๐๐1
๐๐ก = ๐ร(๐ฎ1ร๐0) (2.54)
The displacement vector, ๐ is the integration of velocity perturbation as ๐(๐ซ, ๐ก) = โซ ๐ฎ1(๐ซ, ๐ก)๐๐กโฒ
๐ก 0
. (2.55)
From now on, ๐ is used instead of ๐ฎ1. Integrating Eq. (2.13), Eq. (2.15) and Eq. (2.16) using Eq. (2.55), the Eqs. (2.51) โ (2.54) are expressed as
๐1(๐ซ, ๐ก) = โ๐(๐ซ, ๐ก) โ ๐๐0โ ๐0๐ โ ๐(๐ซ, ๐ก) (2.56) ๐0๐2๐
๐๐ก2= 1 ๐0
(๐ร๐0)ร๐1+ 1 ๐0
(๐ร๐1)ร๐0โ ๐๐1 (2.57)
๐1(๐ซ, ๐ก) = โ๐(๐ซ, ๐ก) โ ๐๐0โ ๐พ๐0๐ โ ๐(๐ซ, ๐ก) (2.58) ๐1(๐ซ, ๐ก) = ๐ร{๐(๐ซ, ๐ก)ร๐0(๐ซ)} (2.59) Substituting Eq. (2.58) and Eq. (2.59) into Eq. (2.57),
๐0๐2๐
๐๐ก2= ๐ฃ0ร{๐ร(๐ร๐0)} + 1
๐0[๐ร{๐ร(๐ร๐0)}]ร๐0+ ๐(๐ โ ๐๐0+ ๐พ๐0๐ โ ๐)
Substituting Eq. (2.60) into Eq. (2.46), the change of potential energy, ๐ฟ๐, is obtained as ๐ฟ๐ = โ1
2โซ [๐ โ (๐ฃ0ร๐1) + 1
๐0๐ โ [(๐ร๐1)ร๐0] โ ๐ โ ๐(๐พ๐0๐ โ ๐ โ ๐ โ ๐๐0)] d๐. (2.61)
Using the vector relations Eq. (B.2), Eq. (B.7) and Eq. (B.10) to Eq. (2.61), ๐ฟ๐, is obtained as ๐ฟ๐ = โ1
2โซ [๐ โ {(1
๐0๐0โ ๐1) ๐} + 1
๐0๐1โ {๐ร(๐ร๐0)} + ๐ฃ0โ (๐1ร๐) + ๐ โ (๐1๐) + ๐1(๐ โ ๐)] d๐.
Using divergence theorem, ๐ฟ๐ =1
2โซ (1
๐0๐0โ ๐1+ ๐1) ๐ โ d๐ +1 2โซ {1
๐0๐ต12โ ๐ฃ0โ (๐1ร๐) โ ๐1(๐ โ ๐)} d๐. (2.62) The potential energy change, ๐ฟ๐, can be distinguished by three different regions: plasma region (0 < ๐ < ๐), ๐ฟ๐๐, plasma surface region (๐ = ๐), ๐ฟ๐๐ and vacuum region (๐ < ๐ โค ๐), ๐ฟ๐๐ฃ. At the vacuum region, there is no displacement. So, the Eq. (2.52) is divided as
๐ฟ๐ =1
2โซ {1
๐0๐ต1๐2 โ ๐ฃ0โ (๐1ร๐) โ ๐1(๐ โ ๐)} d๐
plasma
+1
2โซ (1
๐0๐0โ ๐1+ ๐1) ๐ โ d๐
surface
+ 1
2๐0โซ ๐ต1๐ฃ2 d๐
vacuum
.
(2.63)
where ๐1๐ฃ is perturbed magnetic field at vacuum and ๐ฃ1๐ฃ= 1
๐0(โร๐1๐ฃ) = 0. (2.64)
is satisfied. Each term ๐ฟ๐๐, ๐ฟ๐๐ and ๐ฟ๐๐ฃ is
๐ฟ๐๐=1
2โซ {1
๐0๐ต1๐2 โ ๐ฃ0โ (๐1ร๐) โ ๐1(๐ โ ๐)} d๐
plasma
, (2.65)
๐ฟ๐๐ =1
2โซ (1
๐0๐0โ ๐1+ ๐1) ๐ โ d๐
surface
, (2.66)
๐ฟ๐ = 1
2๐0โซ ๐ต1๐ฃ2 d๐
vacuum
. (2.67)
For steady state, the continuity equation, Eq. (2.1) is
โ โ ๐ฎ1= 0. (2.68)
Integrating Eq. (2.68) by ๐ก
โ โ ๐ = 0. (2.69)
The volume integral in tokamak is
โซ d๐ = โซ ๐d๐d๐d๐ง = โซ ๐d๐d๐๐ 0d๐ = ๐ 0โซ ๐d๐d๐d๐
Substituting Eq. (2.69) into Eq. (2.63), the surface term is removed by taking volume integral, and assuming ๐๐ ~ ๐๐โซ ๐๐ and ๐ต๐1 ~ ๐ต๐1โซ ๐ต๐1,
๐ฟ๐ =1 2โซ {1
๐0๐ต1๐2 โ ๐ฃ0โ (๐1๐ร๐)} ๐d๐d๐d๐ง + 1
2๐0โซ ๐ต1๐ฃ2 ๐d๐d๐d๐ง =1
2โซ โซ โซ {1
๐0๐ต1๐2 โ ๐ฃ0โ (๐1๐ร๐)} ๐d๐
๐ 0
d๐
2๐ 0
๐ 0d๐
2๐ 0
+ 1
2๐0โซ โซ โซ ๐ต1๐ฃ2 ๐d๐
๐ ๐
d๐
2๐ 0
๐ 0d๐
2๐ 0
= 2๐2๐ 0โซ {1
๐0๐ต1๐2 โ ๐ฃ0โ (๐1๐ร๐)} ๐d๐
๐ 0
+2๐2๐ 0
๐0 โซ ๐ต1๐ฃ2 ๐d๐
๐ ๐
= 2๐2๐ 0โซ {๐ต1๐,๐2 + ๐ต1๐,๐2
๐0 โ ๐๐โ (๐ต๐1๐๐โ ๐ต๐1๐๐)} ๐d๐
๐ 0
+2๐2๐ 0
๐0 โซ (๐ต1๐ฃ,๐2 + ๐ต1๐ฃ,๐2 )๐d๐
๐ ๐
So, the change of potential energy, ๐ฟ๐, is
๐ฟ๐ = 2๐2๐ โซ {๐ต1๐,๐2 + ๐ต1๐,๐2
๐0 โ ๐๐โ (๐ต1๐,๐๐๐โ ๐ต1๐,๐๐๐)} ๐d๐
๐ 0
+2๐2๐ 0
๐0 โซ (๐ต1๐ฃ,๐2 + ๐ต1๐ฃ,๐2 )๐d๐
๐ ๐
.
(2.70)
where the first term is plasma term and the second term is vacuum term.
The plasma term, ๐ฟ๐๐ which is the first term of Eq. (2.70), is first calculated. The displacement, ๐, analyzed by Fourier analysis as ๐ ~ ๐๐(๐๐โ๐๐), is given as
๐ = ๐0๐๐(๐๐โ๐๐)= (๐๐๐ฬ + ๐๐๐ฬ)๐๐(๐๐โ๐๐). (2.71)
Then, the relation between ๐๐ and ๐๐ is
๐๐= โ ๐ ๐
d
d๐(๐๐๐). (2.72)
Substituting Eq. (2.72) into Eq. (2.59) and using Eq. (B.18), ๐1๐= [โ๐๐๐ต๐0
๐ 0 (๐ ๐โ1
๐) ๐๐๐ฬ +๐ต๐0 ๐ 0 [๐
๐๐{(๐ ๐โ1
๐) ๐๐๐}] ๐ฬ] ๐๐(๐๐โ๐๐) where the safety factor, ๐ is
๐ = ๐๐ต๐0 ๐ 0๐ต๐.
then, the components of perturbed magnetic field at plasma region are ๐ต1๐,๐ = โ๐๐๐ต๐0
๐ 0 (๐ ๐โ1
๐) ๐๐, (2.73)
๐ต1๐,๐=๐ต๐0 ๐ 0 [๐
๐๐{(๐ ๐โ1
๐) ๐๐๐}]. (2.74)
Substituting Eq. (2.72), Eq. (2.73) and Eq. (2.74) into the first term of Eq. (2.70), the plasma term, ๐ฟ๐๐ is obtained as
๐ฟ๐๐= โซ [
2๐2๐2๐ต๐02 ๐0๐ 0
(๐ ๐ โ
1 ๐)
2
1 + ๐2๐2 ๐2๐ 02
(d๐๐ d๐)
๐ 2 0
+ {2๐2๐ต๐02 ๐0๐ 02 (๐
๐โ1 ๐)
2
+4๐2๐๐โฒ ๐ 0 (๐
๐)
2 1
1 + ( ๐๐ ๐๐ 0)2
} ๐๐2
] ๐๐๐
+2๐2๐ต๐02 ๐0๐ 0
(๐2 ๐2โ 1
๐2) 1 + ( ๐๐
๐๐ 0)2 ๐2๐๐2.
(2.75)
where ๐๐ is the radial displacement, ๐๐, at ๐ = ๐.
Subsequently, the vacuum term, ๐๐ฃ which is the second term of Eq. (2.70) is calculated. As in Eq.
(2.64), โร๐1๐ฃ= 0, and the flux function, ๐ satisfies ๐1๐ฃ = โร๐. Then
โร(โร๐) = โ(โ โ ๐) โ โ2๐ = 0 or
โ2๐ = 0 or
1 ๐
๐
๐๐(๐๐๐
๐๐) + 1 ๐2
๐2๐
๐๐2 + ๐๐
๐๐ง2= 0.
Then flux function ๐ can be expressed as ๐0๐๐(๐๐โ๐๐), and ๐ satisfies Besselโs equation as 1
๐ ๐ ๐๐(๐๐๐
๐๐) โ (๐2 ๐ 02+๐2
๐2) ๐ = 0, (2.76)
and the magnetic field components in the vacuum are expressed in terms of ๐ as ๐ต1๐ฃ,๐= โ1
๐
๐๐
๐๐ = โ๐๐
๐ ๐, (2.77)
๐ต1๐ฃ,๐ =๐๐
๐๐. (2.78)
The solution of ๐ is
๐ = ๐ถ1๐ผ๐(๐
๐ 0๐) + ๐ถ2๐พ๐(๐
๐ 0๐). (2.79)
At ๐ = ๐, ๐ต1,๐ is continuous, so
๐ต1๐ฃ,๐(๐) = ๐ต1๐,๐(๐) =๐๐
๐ ๐๐ =๐๐๐ต๐0 ๐ 0 (๐
๐โ 1 ๐๐) ๐๐ Then ๐(๐) is
๐(๐) =๐๐ต๐0 ๐ 0 (๐
๐โ 1
๐๐) ๐๐=๐๐ต๐0 ๐ 0 (๐
๐โ 1 ๐๐) ๐๐ At ๐ = ๐, ๐(๐) = 0, then the solution of ๐(๐) is
Substituting Eq. (2.77) and Eq. (2.78) into the second term of Eq. (2.70), the vacuum term is calculated as
๐ฟ๐๐ฃ=2๐2๐ 0
๐0 โซ (๐ต1๐ฃ,๐2 + ๐ต1๐ฃ,๐2 )๐d๐
๐ ๐
=2๐2๐ 0
๐0 โซ {(๐ ๐ ๐)
2
+ (d๐ d๐)
2
} ๐d๐
๐ ๐
=2๐2๐ 0
๐0 โซ (๐๐2
๐2 ๐2+ ๐d๐ d๐
d๐ d๐) d๐
๐ ๐
=2๐2๐ 0
๐0 โซ (๐๐2
๐2 ๐2) d๐
๐ ๐
+2๐2๐ 0
๐0 {[๐๐d๐ d๐]
๐ ๐
โ โซ ๐ d d๐(๐d๐
d๐) d๐
๐ ๐
}
=2๐2๐ 0
๐0 โซ {๐๐2
๐2 ๐2โ ๐ d d๐(๐d๐
d๐) d๐} d๐
๐ ๐
+2๐2๐ 0
๐0 [๐๐d๐ d๐]
๐ ๐
Then,
๐ฟ๐๐ฃโ2๐2๐ 0
๐0 โซ ๐ (๐2 ๐2 ๐ โ1
๐ d d๐(๐d๐
d๐) d๐) ๐d๐
๐ ๐
+2๐2๐ 0
๐0 [๐๐d๐ d๐]
๐ ๐
. (2.81)
Substituting Eq. (2.76) into Eq. (2.81),
๐ฟ๐๐ฃ= โ2๐2๐ 0 ๐0
๐๐(๐)๐โฒ(๐). (2.82)
Substituting Eq. (2.80) into Eq. (2.82), the vacuum term is calculated as
๐ฟ๐๐ฃ= โ2๐2๐๐ต๐02 ๐0๐ 02
๐พ๐โฒ (๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) โ ๐ผ๐โฒ (๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) โ ๐ผ๐(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0)
(๐ ๐โ 1
๐๐)
2
๐3๐๐2. (2.83)
The properties of modified Bessel function given as d
d๐ฅ{๐ผ๐(๐ฅ)} =๐
๐ฅ๐ผ๐(๐ฅ) + ๐ผ๐+1(๐ฅ), (2.84) d
d๐ฅ{๐พ๐(๐ฅ)} =๐
๐ฅ๐พ๐(๐ฅ) โ ๐พ๐+1(๐ฅ). (2.85)
Using these properties, ๐ฟ๐๐ฃ in Eq. (2.83) is calculated as ๐ฟ๐๐ฃ
= โ2๐2๐๐ต๐02 ๐0๐ 02
{๐ ๐
๐ 0
๐ ๐พ๐(๐๐
๐ 0) โ ๐พ๐+1(๐๐
๐ 0)} ๐ผ๐(๐๐ ๐ 0) โ {๐
๐ ๐ 0
๐ ๐ผ๐(๐๐
๐ 0) + ๐ผ๐+1(๐๐
๐ 0)} ๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) โ ๐ผ๐(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0)
(๐ ๐
โ 1 ๐๐)
2
๐3๐๐2 = โ2๐2๐ต๐02
๐0๐ 0 ๐ (๐ ๐โ 1
๐๐)
2
๐2๐๐2
+2๐2๐๐ต๐02 ๐0๐ 02
๐พ๐+1(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) + ๐ผ๐+1(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) โ ๐ผ๐(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0)
(๐ ๐โ 1
๐๐)
2
๐3๐๐2
=2๐2๐ต๐02 ๐0๐ 0 ๐ (๐
๐โ 1 ๐๐)
2
{ ๐๐ ๐๐ 0
๐พ๐+1(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) + ๐ผ๐+1(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0) ๐ผ๐(๐๐
๐ 0) โ ๐ผ๐(๐๐
๐ 0) ๐พ๐(๐๐ ๐ 0)
โ 1} ๐2๐๐2
Then, the vacuum term, ๐ฟ๐๐ฃ is obtained as
๐ฟ๐๐ฃ=2๐2๐ต๐02 ๐0๐ 0 ๐
{
๐๐๐พ๐+1(๐๐ ๐ 0) ๐๐ 0๐พ๐(๐๐
๐ 0) 1 +
๐ผ๐+1(๐๐ ๐ 0) ๐ผ๐(๐๐
๐ 0)
๐พ๐(๐๐ ๐ 0) ๐พ๐+1(๐๐ ๐ 0) 1 โ
๐ผ๐(๐๐ ๐ 0) ๐ผ๐(๐๐
๐ 0)
๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0)
โ 1
} (๐
๐โ 1 ๐๐)
2
๐2๐๐2.
(2.86)
Finally, summing Eq. (2.75) and Eq. (2.86), the total change of potential energy is derived as [8]
๐ฟ๐ = โซ (๐๐โฒ2+ ๐๐2)๐๐๐
๐ 0
+2๐2๐ต๐02 ๐0๐ 0
๐2๐๐2 {
๐2 ๐2โ 1
๐๐2 1 + ( ๐๐
๐๐ 0)2
+ ๐๐ฌ (๐ ๐โ 1
๐๐
)
2
}
. (2.87) where
๐ =2๐2๐2๐ต๐02 ๐0๐ 0
(๐ ๐ โ
1 ๐)
2
1 + ๐2๐2 ๐2๐ 02
=2๐2๐2๐ต๐02 ๐0๐ 0 (๐
๐โ1 ๐)
2{1 + ( ๐๐
๐๐ 0)2} โ ( ๐๐ ๐๐ 0)2 1 + ( ๐๐
๐๐ 0)2
=2๐2๐2๐ต๐02 ๐0๐ 0
(๐ ๐โ1
๐)
2
{1 โ ( ๐๐
๐๐ 0)2 1 + ( ๐๐
๐๐ 0)2
} =2๐2๐2๐ต๐02 ๐0๐ 0
(๐ ๐โ1
๐)
2
{1 โ (๐ ๐ 0
)
2
(๐ ๐)
2
}
and
๐ =4๐2๐๐โฒ ๐ 0 (๐
๐)
2 1
1 + ( ๐๐ ๐๐ 0)2
+2๐2๐ต๐02 ๐0๐ 02 (๐
๐โ1 ๐)
2
{๐2โ 1 1 + ( ๐๐
๐๐ 0)2 }
+4๐2๐ต๐2 ๐ 0 (๐
๐)
2
(๐ ๐ 0)
2
(๐2 ๐2โ 1
๐2) 1 {1 + ( ๐๐
๐๐ 0)2}
2
=4๐2๐๐โฒ ๐ 0 (๐
๐)
2{1 + ( ๐๐
๐๐ 0)2} โ ( ๐๐ ๐๐ 0)2 1 + ( ๐๐
๐๐ 0)2
+2๐2๐ต๐02 ๐0๐ 0 (๐
๐โ1 ๐)
2
{๐2โ
{1 + ( ๐๐
๐๐ 0)2} โ ( ๐๐ ๐๐ 0)2 1 + ( ๐๐
๐๐ 0)2
}
+4๐2๐ต๐02 ๐0๐ 0 (๐
๐)
2
(๐ ๐ 0)
2
(๐2 ๐2โ 1
๐2)
{1 + ( ๐๐
๐๐ 0)2} โ ( ๐๐ ๐๐ 0)2 {1 + ( ๐๐
๐๐ 0)2}
2 and
๐ฌ =
๐๐๐พ๐+1(๐๐ ๐ 0) ๐๐ 0๐พ๐(๐๐
๐ 0) 1 +
๐ผ๐+1(๐๐ ๐ 0) ๐ผ๐(๐๐
๐ 0)
๐พ๐(๐๐ ๐ 0) ๐พ๐+1(๐๐ ๐ 0) 1 โ
๐ผ๐(๐๐ ๐ 0) ๐ผ๐(๐๐
๐ 0)
๐พ๐(๐๐ ๐ 0) ๐พ๐(๐๐
๐ 0)
โ 1.
The change of potential, ๐ฟ๐ consists of plasma term, ๐ฟ๐๐, and the vacuum term, ๐ฟ๐๐ฃ, as
๐ฟ๐๐= โซ (๐๐โฒ2+ ๐๐2)๐๐๐
๐ 0
+2๐2๐ต๐02 ๐0๐ 0 ๐2๐๐2
๐2 ๐2โ 1
๐2 1 + ( ๐๐
๐๐ 0)2 ,
๐ฟ๐๐ฃ =2๐2๐ต๐02
๐0๐ 0 ๐2๐๐2๐๐ฌ (๐ ๐โ1
๐ )
2
.
Applying large aspect ratio assumption, ๐/๐ โช 1, functions ๐ and ๐ are approximated as
๐ โ2๐2๐2๐ต๐02 ๐0๐ 0 (๐
๐โ1 ๐)
2
,
๐ โ2๐2๐ต๐02 ๐0๐ 0 (๐
๐โ1 ๐)
2
(๐2โ 1) +2๐2 ๐ 0 (๐
๐)
2
{2๐๐โฒ+๐ต๐02 ๐0 (๐
๐ 0)
2
(๐ ๐โ1
๐) (3๐ ๐ +1
๐)}.
The beta values are equivalent to ๐ฝ ~ ๐ฝ๐ก ~ ๐2 and ๐ฝ๐ ~ 1. Then, the change of potential can be divided into second order and fourth order terms as
๐ฟ๐2=2๐2๐ต๐02
๐0๐ 0 โซ {(๐๐๐ ๐๐)
2
+ (๐2โ 1)๐2} (๐ ๐โ1
๐)
2
๐๐๐
๐ 0
+2๐2๐ต๐2 ๐0๐ ๐๐2๐2
{ ๐2 ๐2โ 1
๐2 ๐2 ๐ 02+๐2
๐2 +๐2
๐๐ฌ (1 ๐โ ๐
๐ )
2
} ,
๐ฟ๐4=2๐2 ๐ 0 (๐
๐)
2
โซ {2๐๐โฒ+๐ต๐02 ๐0 (๐
๐โ1 ๐) (3๐
๐ +1
๐)} ๐2๐๐๐
๐ 0
and
๐ฟ๐2 ~ 2๐2๐ 0๐ต๐02 ๐0
๐2๐ (๐ ๐ 0
)
2
,
๐ฟ๐4 ~ 2๐2๐ 0๐ต๐02
๐0 ๐2๐ (๐ ๐ 0)
4
This is the whole derivation process for change of potential energy, ๐ฟ๐, in tokamak using the
2.3 Theories of sawtooth instability
2.3.1 Internal kink, ๐/๐ = 1/1 mode
As mentioned in section 1.3, sawtooth oscillation is a type of โ๐ = 1 internal kink modeโ [9]. In the internal mode, plasma surface displacement is zero (๐๐= 0) whereas the external kink mode is nonzero (๐๐โ 0). For the external mode, stability is effected by both the integral term and the other term of Eq.
(2.87). So, the external kink mode both has ๐ = 1 mode and ๐ โฅ 2 mode. For the internal kink mode, with ๐๐= 0, Eq. (2.87) becomes
๐ฟ๐ =2๐2๐ต๐2
๐0๐ 0 โซ {(๐๐๐ ๐๐)
2
+ (๐2โ 1)๐2} (๐ ๐โ1
๐)
2
๐๐๐
๐ 0
+2๐2 ๐ 0 (๐
๐)
2
โซ {2๐๐โฒ+๐ต๐2 ๐0 (๐
๐โ1 ๐) (3๐
๐ +1
๐)} ๐2๐๐๐
๐ 0
.
(2.88)
with
๐ฟ๐2=2๐2๐ต๐2
๐0๐ 0 โซ {(๐๐๐ ๐๐)
2
+ (๐2โ 1)๐2} (๐ ๐โ1
๐)
2
๐๐๐
๐ 0
,
๐ฟ๐4=2๐2 ๐ 0 (๐
๐)
2
โซ {2๐๐โฒ+๐ต๐2 ๐0 (๐
๐โ1 ๐) (3๐
๐ +1
๐)} ๐2๐๐๐
๐ 0
.
Instability occurs when ๐ฟ๐ < 0, thus the plasma is likely to be unstable, when ๐ฟ๐2 is minimized.
Then, the poloidal number, ๐, is always ๐ = 1 for internal kink mode. So,
๐ฟ๐ =2๐2๐ต๐2
๐0๐ 0 โซ (๐๐๐ ๐๐)
2
(๐ โ1 ๐)
2
๐๐๐
๐ 0
+2๐2๐2
๐ 0 โซ {2๐๐โฒ+๐ต๐2
๐0 (๐ โ1
๐) (3๐ +1
๐)} ๐2๐๐๐
๐ 0
.
(2.89)
with
๐ฟ๐2=2๐2๐ต๐2
๐0๐ โซ (๐๐๐ ๐๐)
2
(๐ โ1 ๐)
2
๐๐๐
๐ 0
,
๐ฟ๐4=2๐2๐2
๐ 0 โซ {2๐๐โฒ+๐ต๐2
๐0 (๐ โ1
๐) (3๐ +1
๐)} ๐2๐๐๐
๐ 0
.
Finally, the change of potential energy, ๐ฟ๐, for ๐/๐ = 1/1 mode is derived as
๐ฟ๐ =2๐2๐ต๐2
๐0๐ 0 โซ (๐๐๐ ๐๐)
2
(1 โ1 ๐)
2
๐๐๐
๐ 0
+2๐2 ๐ 0
โซ {2๐๐โฒ+๐ต๐2 ๐0
(1 โ1
๐) (3 +1
๐)} ๐2๐๐๐
๐ 0
(2.90)
with
๐ฟ๐2 =2๐2๐ต๐2
๐0๐ โซ (๐๐๐ ๐๐)
2
(1 โ1 ๐)
2
๐๐๐
๐ 0
,
๐ฟ๐4=2๐2
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 + 3๐)} ๐2๐๐๐
๐ 0
.
or
๐ฟ๐ = 2๐2๐ 0โซ ๐ต๐2
๐0(1 โ ๐)2(d๐ d๐)
2
๐๐๐
๐ 0
+2๐2
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 + 3๐)} ๐2๐๐๐
๐ 0
(2.91)
with
๐ฟ๐2= 2๐2๐ 0โซ ๐ต๐2
๐0(1 โ ๐)2(๐๐ ๐๐)
2
๐๐๐
๐ 0
,
๐ฟ๐4=2๐2
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 + 3๐)} ๐2๐๐๐
๐ 0
.
Two models for sawtooth instability are introduced to minimize ๐ฟ๐2 of Eq. (2.91). Kadomtsevโs model tries to explain instability by taking d๐/d๐ โ 0 inside the ๐ = 1 surface [30], whereas Wessonโs model tries to explain instability by taking ๐ โ 1 inside the ๐ = 1 surface.
2.3.2 Kadomtsevโs model
As mentioned in section 2.3.1, Kaomtsevโs model tries to minimize ๐ฟ๐2 by d๐/d๐ โ 0 inside the ๐ = 1 surface. At the magnetic axis (๐ = 0), the displacement is ๐ = ๐0, and at the edge (๐ = ๐), ๐๐= 0. To minimize ๐ฟ๐2, d๐/d๐ = 0 except near ๐ = 1 surface (โต At ๐ = 1, d๐/d๐ cannot be zero).
Based on this picture, the radial displacement distribution, ๐(๐), is described in Figure 2.5 as
Figure 2.5 The displacement distribution, ๐(๐), based on Kadomtsevโs model.
The corresponding ๐ฟ๐ is ๐ฟ๐ =2๐2๐02
๐ 2 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 + 3๐)} ๐๐๐
๐1 0
(2.92) and ๐โฒ< 0 (โต ๐ = ๐๐๐ต๐ and ๐ is decreased as ๐ increased as shown in Figure 1.3), then it is concluded as ๐ฟ๐ < 0.
๐ ๐
๐1 ๐1โ ๐ฟ ๐0
๐ = ๐0 as ๐ < ๐1
๐ = 0 as ๐ > ๐1
๐ = 1 surface
The ๐/๐ = 1/1 internal kink mode has a radial displacement of ๐ ~ ๐๐(๐โ๐).
(a) (b)
Figure 2.6 The ๐/๐ = 1/1 internal kink within ๐ = 1 surface. (a) total view of tokamak (b) poloidal view of tokamak.
The corresponding flow pattern of Figure 2.6 (b) - โก is shown in Figure 2.7 as
Figure 2.7 The flow pattern of plasma in
โ
โก
โข
โฃ ๐ = ๐0cos(๐ โ ๐)
โ โก
โข โฃ
๐ฃ1 ๐ฟ ๐ฃ2
๐ฃ2
The velocity, ๐ฃ1, is the hot core velocity and the velocity, ๐ฃ2, is the plasma velocity which comes from the narrow layer. The thickness, ๐ฟ, is the thickness of the narrow resistive layer.
The helical magnetic field line is
๐ตโ= ๐ต๐โ (๐/๐ )๐ต๐= ๐ต๐(1 โ ๐) (2.93) and current, ๐ comes from the electric field as equivalent to
๐ ~ ๐๐ฃ1๐ตโ=๐ฃ1๐ตโ
๐ . (2.94)
From Ampereโs law,
๐~๐ตโ/๐0๐ฟ. (2.95)
Combining Eq. (2.94) and Eq. (2.95), ๐ฃ1 is expressed as ๐ฃ1= ๐
๐0๐ฟ. (2.96)
The magnetic pressure is equivalent to plasma pressure by ๐ฃ2 as ๐ตโ2
2๐0
~ 1
2๐๐ฃ22+1
2๐๐ฃ22= ๐๐ฃ22. Then, the velocity ๐ฃ2 is
๐ฃ2~ ๐ตโ
โ2๐๐0
. (2.97)
From continuity equation
๐ฃ1๐1~๐ฃ2๐ฟ, then
๐ฟ ~ ๐ฃ1๐1
๐ฃ2 =๐๐1โ2๐๐0
๐ตโ๐0๐ฟ . or
where
๐๐ =๐0
๐ ๐12, (2.99)
๐๐ด= ๐1 ๐ตโ/โ๐๐0
. (2.100)
The collapse time is given by
๐๐พ ~ ๐1
๐ฃ1. (2.101)
โ
Substituting (2.95) into (2.98), the collapse time is calculated as ๐๐พ ~ ๐ฟ2
(๐๐ด/๐๐ )1/2 ๐0
๐ ~ (๐๐ /๐๐ด)1/2(๐๐ด ๐๐ )๐0
๐ ๐12 = (๐๐ด/๐๐ )1/2๐๐ = (๐๐ด๐๐ )1/2 Finally, the collapse time in Kadomtsevโs model is
๐๐พ ~ (๐๐ด๐๐ )1/2. (2.102)
The whole process of sawtooth oscillation in Kadomtsevโs model is shown in Figure 2.9. Initially, ๐ > 1 everywhere as in Figure 2.9 (a). As the temperature at the core increases, ๐ = 1 surface appears and, instability occurs as in Figure 2.9 (b). The core starts to move by internal kink, and a new magnetic island appears on the other side, resulting in two apparent islands. One magnetic island by internal kink has ๐ < 1 whereas the magnetic island on the other side has ๐ > 1 . Then, magnetic reconnection occurs at the narrow resistive layer near ๐ = 1 surface as in Figure 2.9 (c). One magnetic island which has ๐ < 1 is annihilated by magnetic reconnection whereas the new magnetic island which has ๐ > 1 is expanded. After the old magnetic island is annihilated completely, it returns to the initial state as ๐ >
1 everywhere as in Figure 2.9 (d). Then the periodic behavior is repeated, starting from the state illustrated in Figure 2.9 (a). This is the theoretical review of Kadomtsevโs model to observe fast crash by magnetic reconnection.
Figure 2.8 The whole process of sawtooth oscillation in Kadomtsevโs model [31].
(a) ๐ > 1 everywhere
(d) ๐ > 1 everywhere (b) Internal kink
๐ < 1
New magnetic island
Core starts to move Two island systems
๐ < 1 ๐ > 1
(c) Magnetic reconnection
Magnetic reconnection
2.3.3 Wessonโs model
Wessonโs model tries to explain the inconsistencies of Kadomtsevโs model. There are a few discrepancies in Kadomtsevโs model when compared to the experimental observations. One discrepancy is the crash time. The collapse times of several types of tokamak, using Kadomtsevโ model, were calculated as in Table 2.3 with ๐ = 1 keV, ๐ = 2ร1020mโ3 , and ๐0 ~ 0.9 . The detailed calculation process is in Appendix C.2. According to Table 2.1, Kadomtsevโ collapse time is ๐๐พ โ ๐1/โ๐ต๐, and ๐๐พ ~ 2ร10โ4-3ร10โ3s. But the actual tokamaks have always ๐๐พ ~ 10โ4 s and does not depend on the tokamak geometry [32].
To explain this inconsistency, Wesson assumed a flat ๐-profile inside the ๐ = 1 surface. By taking ๐ โ 1 to Eq. (2.91),
๐ฟ๐ โ2๐2
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 + 3๐)} ๐2๐๐๐
๐ 0
(2.103) and the corresponding displacement distribution [18] is illustrated in Figure 2.10.
Kadomtsevโs collapse time calculation is not applicable for the ๐ โ 1 case, since ๐๐พ becomes infinite. So, Wesson introduced a new model [18]. In this model, there is no magnetic shear stress inside the ๐ = 1 surface, because of the magnetic shear stress, given in Eq. (2.103) [12],
๐ =๐ ๐
d๐
d๐ (2.103)
goes to zero as ๐ goes to 1. Due to the absence of magnetic shear, the magnetic surface deforms into a crescent shape, and cold plasma comes into the core as shown in Figure 2.10 [14, 33]. This flow pattern is called โhot crescent, cold bubbleโ. The magnetic reconnection takes place during the ramp phase and not in the fast collapse phase [18].
Table 2.3 Calculated Kadomtsevโs collapse time according to several tokamaks [34-39].
Tokamak
Major radius [m]
Toroidal Magnetic
field [T]
q=1 surface radius [m]
Resistive diffusion time [s]
Alfven time [s]
Kadomtsev's collapse
time [s]
DIII-D 1.67 2.2 0.1
(estimated) 2.26ร10โ1 4.91ร10โ6 1.05ร10โ3
JET 2.96 3.45 0.2 9.05ร10โ1 5.54ร10โ6 2.24ร10โ3
ADITYA 0.75 1.2 0.03
(estimated) 2.04ร10โ2 4.04ร10โ6 2.87ร10โ4
ASDEX 1.65 3.1 0.1
(estimated) 2.26ร10โ1 3.44ร10โ6 8.82ร10โ4
TCV 0.88 1.43 0.03
(estimated) 2.04ร10โ2 3.98ร10โ6 2.85ร10โ4
WEST 2.25 4.5 0.1
(estimated) 2.26ร10โ1 3.23ร10โ6 8.55ร10โ4
COMPASS 0.56 0.9 ~ 2.1 0.03
(estimated) 2.04ร10โ2 2.41ร10โ7 2.22ร10โ4
NSTX-U 0.85 1 0.1
(estimated) 2.26ร10โ1 5.49ร10โ6 1.11ร10โ3
TEXTOR 1.75 3 0.09ยฑ0.01 1.83ร10โ1 3.77ร10โ6 8.31ร10โ4
SST-1 1.1 3 0.04
(estimated) 3.62ร10โ2 2.36ร10โ7 2.93ร10โ3
EAST 1.75 3.5 0.05~0.09 1.11ร10โ1 3.23ร10โ6 5.98ร10โ4
KSTAR 1.8 3.5 0.1 2.26ร10โ1 3.32ร10โ6 8.67ร10โ4
T-15U 2.43 3.5 0.15
(estimated) 5.09ร10โ1 4.49ร10โ6 1.51ร10โ3 0.2
Figure 2.9 The displacement distribution, ๐(๐), based on Wessonโs model [18].
Figure 2.10 The flow pattern of plasma in Wessonโs model. The red region is hot region, and the blue
๐ = 1 surface
๐ ๐
๐1 ๐0
๐ = 0 as ๐ > ๐1
2.3.4 Comparison of the two models
This concludes the theoretical review of sawtooth instability. Summarizing section 2.3.2 and section 2.3.3, a comparison of Kadomtsevโs model and Wessonโs model is shown in Table 2.4.
Table 2.4 Comparison between Kadomtsevโs model and Wessonโs model.
Kadomtsevโs model Wessonโs model
Model Fast reconnection model Quasi-interchange model
Safety factor,
๐ 1 โ ๐ > ๐ 1 โ ๐ โช ๐
To diminish ๐ฟ๐2 term
d๐
d๐โ 0 except ๐ = 1 surface ๐ โ 1 inside ๐ = 1 surface Approximated
๐ฟ๐
2๐2๐02
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 +
๐ 0
3๐)} ๐๐๐
2๐2
๐ 0 โซ {2๐๐โฒโ๐ต๐2
๐0 (1 โ ๐)(1 +
๐ 0
3๐)} ๐2๐๐๐ Flow structure Hot bubble, cold crescent Hot crescent, cold bubble
Magnetic
reconnection Occurs at fast collapse phase Occurs at ramp phase
III. Experimental Methods
The ECEI system is a diagnostic system that visualizes electron temperature fluctuation in 2-D/3-D spaces of tokamak plasmas. The ECEI system operates based on the principle of ECE (Electron Cyclotron Emission). First, the physical principle of ECE will be reviewed. Then, the ECEI (Electron Cyclotron Emission Imaging) system in KSTAR will be reviewed. Finally, the experimental set up for the measurement of sawtooth instability will be described.
3.1 ECE and KSTAR ECEI System
In tokamak, it is impossible to measure the plasma electron temperature with commercial thermometers, because the plasma electron temperature is too high to be measured. Instead of a thermometer, an ECE (electron cyclotron emission) radiometer or ECE imaging system can be used to measure electron temperature by measuring radiation intensity emitted from gyrating electrons.
3.1.1 Physical principle of ECE
The electron is gyrating around a magnetic field in tokamak as shown in Figure 3.1. The cyclotron radiation is emitted from gyrating electrons.
Magnetic field line Trace of electron
In this picture, the equation of motion is ๐๐๐ฏ
๐๐ก = ๐๐ฏร๐ (3.1)
and the solution of the equation is simple harmonic motion. The angular velocity of the solution is ๐ = ๐๐ต
๐๐
. (3.2)
where ๐๐ is the mass of the electron, and ๐ต is the magnitude of magnetic field [20].
In this gyration motion, the radioactive transfer equation is given as ๐๐ผ๐
๐๐ = ๐๐โ ๐ผ๐ผ๐ฃ. (3.3)
where ๐ผ๐ is the intensity of radiation, ๐ is the radiation traveling distance, ๐๐ is the emission coefficient, and ๐ผ is the absorption coefficient [40]. The optical depth, ๐, is defined as
๐ = โซ ๐ผ๐๐
๐
. (3.4)
If the radiation travel from 0 to ๐, the solution for the equation is given as ๐ผ๐(๐) = ๐ผ๐(0)๐โ๐+๐๐(๐)
๐ผ (1 โ ๐โ๐) =๐๐(๐)
๐ผ + {๐ผ๐(0) โ๐๐(๐)
๐ผ } ๐โ๐. (3.5)
When the optical depth is thick enough, the radiation density is approximated as ๐ผ๐(๐) = ๐๐/๐ผ [41].
This means all radiation is absorbed. In this case, the plasma can be regarded as a blackbody, so it follows Rayleigh-Jeanโs law as
๐ผ๐= 2โ๐3/๐2
๐โ๐/๐๐ต๐๐โ 1. (3.6)
where โ is the Planck constant and ๐๐ is the electron temperature [42, 43]. With low frequency approximation, โ๐ โช ๐๐, Eq. (3.6) approximated as
๐ผ๐ =2โ๐2๐๐ต๐๐
๐2 . (3.7)
Finally, for thick optical depths, the electron temperature is obtained as ๐๐=2๐2๐2๐ผ๐
โ๐2๐๐ต =2๐2๐2๐2๐ผ๐
โ๐2๐ต2๐๐ต . (3.9)
This shows that the electron temperature, ๐๐, is proportional to radiation intensity, ๐ผ๐, for thick optics. In order to satisfy this characteristic, the plasma should be optically thick.
3.1.2 KSTAR ECEI System
The KSTAR ECEI system is used to visualize MHD instability in 3D by measuring relative electron temperature, ๐ฟ๐๐/โฉ๐๐โช, where ๐ฟ๐๐= ๐๐โ โฉ๐๐โช, and โฉ๐๐โช is the time averaged temperature.
In KSTAR, there are 3 detector arrays for ECEI measurement. Two of them are placed in H-port and another is placed in G-port as shown in Figure 3.2. The angle difference of G-port and H-port is 22.5ยฐ toroidally. Each of the two detector arrays in G-port is placed on HFS (High Field Side, ๐ < ๐ 0) and LFS (Low Field Side, ๐ > ๐ 0 ) as shown in Figure 3.2. The magnetic field, ๐ต can be approximated by ๐ต๐ (โต ๐ต = โ๐ต๐2+ ๐ต๐2 and ๐ต๐ โซ ๐ต๐ ), and according to Eq. (2.12), ๐ต๐ is proportional to 1/๐ . Thus, ๐ต ~ 1/๐ .
Each detector array has 24 vertical and 8 radial detection channels. In other words, each detector has 192 detection channels, and at H-port, there are 384 channels and at G-port, there are 192 channels, so in total, there are 576 channels on the ECEI system. The vertical and radial resolution is ~1.5 cm per channel and the time resolution is 0.5, 1, 2 ๐s [44].
Figure 3.2 The schematics of the ECEI system. The first system is in H-port, and the second system is in G-port [45].
3.2 Experimental Set-up
The major and minor radii of KSTAR is shown in Table 3.1. These parameters are constant in every experiment.
Table 3.1 Parameters of KSTAR [32]
Radius
Major radius ๐ 0= 1.8 m
Minor radius ๐ = 0.5 m
The intrinsic rotation experiment # 11264 is used to observe the sawtooth oscillation in 3D with the KSTAR ECEI system. The experimental set-up is shown in Table 3.2. In the next chapter, the experimental results will be shown and discussed by the results. Then, the results will be compared with the theoretical models.
Table 3.2 Experiment set-up for # 11264 Set-up
Measurement duration ๐ก = 0 ~ 10.5 s
Observed time ๐ก = 3.5 ~ 4.5 s
Time resolution โ๐ก = 2 ๐s
Mode H-mode
Average density of electron โฉ๐๐โช = 1.8ร1019mโ3 Average density of electron at the core โฉ๐๐0โช = 3ร103mโ3
Average electron temperature โฉ๐๐0โช = 1.5 ~ 2.7 keV Average electron temperature at the core โฉ๐๐0โช = 2.5 keV
Sampling frequency ๐๐ = 500 kHz
Harmonic extraordinary mode
Total plasma current ๐ผ๐= 600 kA
Total driven current ๐ผ๐ก = 20 kA
Heating power 0.765 MW
Elongation ๐ = 1.8
Safety factor at ๐ = 0.95๐ ๐95โ 5 Radial (๐ ) position /
Vertical size / Zoom factor
HFS, H-port 180 m - 190 m / 40 cm / 1.3 G-port 190 m - 200 m / 50 cm / 1.6 LFS, H-port 205 m - 215 m / 40 cm / 1.3
Vertical position 0.02 ~ 0.04 m
Toroidal magnetic field ๐ต๐ก = 3.5 T Radial spatial resolution โ๐ = 1 cm Vertical spatial resolution โ๐ง = 2 cm
Triangularity ๐ฟ = 0.8
Poloidal beta ๐ฝ๐= 0.4
Rotation speed ๐ฃ๐ = 110 km/s
IV. Results & Discussion
4.1 Experimental results
4.1.1 Observation of periodic behavior
The relative temperature fluctuation was observed in 10.5 s intervals. Within this time interval, 3.4 - 4.6 s was selected as shown in Figure 4.1. The results show the time traces of the ratio of electron temperature fluctuation to average temperature, observed through two detection channels from the ECEI system.
(a) (b)
The observed sawtooth period, theoretically calculated Kadomtsevโs collapse time, and the observed collapse time, are shown in Table 4.2.
Table 4.1 Sawtooth period, and collapse times.
Times Symbol Result
Sawtooth period (observed) ๐sawtooth 28.2 ms Kadomtsevโs collapse time
(calculated) ๐๐พ 867 ๐s
Collapse time (observed) ๐๐ 150 ๐s
According to the results, the observed collapse time is very short compared to the sawtooth period, i.e., ๐๐/๐sawtooth ~ 10โ2. That is why this phenomenon is called fast collapse.
The whole periodic behavior of at the central region (observed in high field side of H-port) and the outer region (observed in G-port) is shown in Figure 4.2. Also, the corresponding three phases are shown in Table 4.2.
(a)
Figure 4.2 1-D time trace of ๐ฟ๐๐/โฉ๐๐โช in one period (a) at the central region (b) at the outer region.
Table 4.2 The time lengths of the three phases in Figure 4.2.
Times Time interval (s) Duration Ramp phase 3.52482 ~ 3.54982 25 ms
Precursor phase 3.54982 ~ 3.55287 3.05 ms
Fast collapse phase 3.55287 ~ 3.55302 150 ๐s
The ECE images of periodic behavior are shown in Figure 4.3. The observation time is from 3.52488 s to 3.55308 s. Initially, as shown in Figure 4.3 (a), the core is relatively cold and it is stable. The temperature of the core increases over time in Figure 4.3 (a) โ Figure 4.3 (f). Then, the new magnetic island is growing in Figure 4.3 (g) โ Figure 4.3 (k). A new magnetic island was first distinctly observed as shown in Figure 4.3 (i). The old magnetic island and new magnetic island were observed alternately, as plasma was noticed to be rotating. Finally, between Figure 4.3 (g) โ Figure 4.3 (k), the crash occurred.
After the crash occurred, the heat of the core spread to the outer region. The plasma returned to its initial conditions as shown in Figure 4.3 (l). This behavior of relative electron temperature fluctuation occurs repeatedly.
The stable ramp phase is shown in Figure 4.3 (b) โ Figure 4.3 (f); the precursor phase is shown in Figure 4.3 (g) โ Figure 4.3 (j); and the fast collapse phase is shown in Figure 4.3 (k) โ Figure 4.3 (l). In the ramp phase, the core is heating. In the precursor phase, the temperature reached critical temperature, and the hot core started to move by internal kink. On the other side of the hot core plasma, a new magnetic island appeared, and a new magnetic island grew. The old magnetic island disappeared suddenly, and the electron temperature fluctuation distribution returned to initial sate, which is the first state of the stable ramp phase. This demonstrated that the experiment resembles Kadomtsevโs model more than Wessonโs model, which was described in section 2.3.2.