๐Ÿ“š ๊ณต๋ถ€ ์ฃผ์ œ: ์œ ํ•œ ์š”์†Œ๋ฒ• ๊ธฐ์ดˆ

1. FEM(์œ ํ•œ ์š”์†Œ๋ฒ•) ์ด๋ž€?

FEM, ์œ ํ•œ์š”์†Œ๋ฒ•์€ ๋ณต์žกํ•œ ๊ณตํ•™ ๋ฐ ๋ฌด๋ฆฌ ๋ฌธ์ œ๋ฅผ ์ปดํ“จํ„ฐ๋ฅผ ์ด์šฉํ•ด ์ˆ˜์น˜์ ์œผ๋กœ ํ•ด๊ฒฐํ•˜๋Š” ๊ฐ•๋ ฅํ•œ ํ•ด์„ ๊ธฐ๋ฒ•์ด๋‹ค. ๋ˆˆ์— ๋ณด์ด์ง€ ์•Š๋Š” ํž˜์˜ ์ž‘์šฉ, ์—ด์˜ ์ด๋™, ์œ ์ฒด์˜ ํ๋ฆ„ ๋“ฑ ์–ด๋ ค์šด ๋ฏธ๋ถ„๋ฐฉ์ •์‹์œผ๋กœ ํ‘œํ˜„๋˜๋Š” ํ˜„์ƒ๋“ค์„ ๋ˆˆ์œผ๋กœ ํ™•์ธ ๊ฐ€๋Šฅํ•œ ํ˜•ํƒœ๋กœ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ํ•˜์—ฌ ๋ณด์—ฌ์ค€๋‹ค.

ํ•ต์‹ฌ ์›๋ฆฌ๋Š” ๋ถ„ํ• ๊ณผ ์ •๋ณต์— ์žˆ๋‹ค. ํ•ด์„ํ•˜๋ ค๋Š” ๋ณต์žกํ•œ ํ˜•์ƒ์˜ ๋Œ€์ƒ๋ฌผ์„ โ€˜์œ ํ•œ(Finite)โ€™๊ฐœ์˜ ๋‹จ์ˆœํ•œ ๊ธฐํ•˜ํ•™์  ๋ชจ์–‘(์‚ผ๊ฐํ˜•, ์‚ฌ๊ฐํ˜• ๋“ฑ)์ธ โ€˜์š”์†Œ(Element)โ€™๋กœ ์ž˜๊ฒŒ ๋‚˜๋ˆ , ๊ฐ ์š”์†Œ์˜ ๊ฑฐ๋™(์›€์ง์ž„)์„ ๋น„๊ต์  ๊ฐ„๋‹จํ•œ ์ˆ˜ํ•™์  ๊ด€๊ณ„๋กœ ๊ทผ์‚ฌํ™”ํ•˜์—ฌ ๊ณ„์‚ฐํ•˜๊ณ , ์ด๋ฅผ ์ „์ฒด์ ์œผ๋กœ ํ†ตํ•ฉํ•˜์—ฌ ์˜ˆ์ธกํ•˜๋Š” ๋ฐฉ์‹์ด๋‹ค.

  1. ๋ถ„ํ•  (Divide)
    • ์ฃผ์–ด์ง„ ๋ฌธ์ œ๋ฅผ ๋” ์ด์ƒ ๋‚˜๋ˆŒ ์ˆ˜ ์—†์„ ๋•Œ๊นŒ์ง€ ๋น„์Šทํ•œ ์œ ํ˜•์˜ ์—ฌ๋Ÿฌ ์ž‘์€ ๋ฌธ์ œ๋กœ ์ชผ๊ฐ ๋‹ค.
  2. ์ •๋ณต (Conquer)
    • ๋‚˜๋ˆ„์–ด์ง„ ์ž‘์€ ๋ฌธ์ œ๋“ค์€ ํฌ๊ธฐ๊ฐ€ ์ž‘๊ณ  ๋‹จ์ˆœํ•˜๊ธฐ ๋•Œ๋ฌธ์— ํ•ด๊ฒฐํ•˜๊ธฐ๊ฐ€ ์‰ฌ์›Œ, ๊ฐ ๋ฌธ์ œ๋ฅผ ๊ฐœ๋ณ„์ ์œผ๋กœ ํ•ด๊ฒฐํ•œ๋‹ค.
  3. ํ†ตํ•ฉ (Combine)
    • ํ•ด๊ฒฐ๋œ ์ž‘์€ ๋ฌธ์ œ๋“ค์˜ ๋‹ต์„ ๋‹ค์‹œ ์›๋ž˜์˜ ํฐ ๋ฌธ์ œ์— ๋งž๊ฒŒ ํ•ฉ์ณ, ์ตœ์ข…์ ์ธ ํ•ด๋‹ต์„ ์–ป๋Š”๋‹ค.

2. ํ•ต์‹ฌ ํ‚ค์›Œ๋“œ ์„ค๋ช…

  • ๊ทผ์‚ฌ ํ•ด : $u_h = \sum_{i=1}^{DOF}\alpha_{i}\psi_{i}$
    • $u_{h}$: FEM์œผ๋กœ ๊ณ„์‚ฐ๋œ ๊ทผ์‚ฌ ํ•ด (approximate solution). ์‹ค์ œ ํ•ด $u$๋ฅผ ์™„๋ฒฝํ•˜๊ฒŒ ์•Œ ์ˆ˜ ์—†๊ธฐ ๋•Œ๋ฌธ์—, $u$ ๋Œ€์‹  $u_{h}$๋กœ ๊ณ„์‚ฐ.

    • $\alpha_{i}$ : ํ•ด์˜ ๊ณ„์ˆ˜ (๋˜๋Š” ์ž์œ ๋„ ๊ฐ’, DOF ๊ฐ’). ์ด ๊ฐ’์€ FEM ํ•ด์„ ๊ณผ์ •์—์„œ ๊ณ„์‚ฐ๋˜๋Š” ์ˆ˜์น˜ ๊ฒฐ๊ณผ(๋ณ€์ˆ˜). ์˜ˆ: ๋ณ€์œ„, ์˜จ๋„ ๋“ฑ

    • $\psi_{i}$ : ๊ธฐ์ € ํ•จ์ˆ˜(basis function) ๋˜๋Š” ํ˜•์ƒ ํ•จ์ˆ˜(shape function). ๊ฐ ์š”์†Œ๋‚˜ ๋…ธ๋“œ์—์„œ ํ•ด๋ฅผ ๊ทผ์‚ฌํ•˜๊ธฐ ์œ„ํ•ด ์‚ฌ์šฉ๋˜๋Š” ํ•จ์ˆ˜.

    • $DOF$ : ์ „์ฒด ์‹œ์Šคํ…œ์—์„œ์˜ ์ž์œ ๋„(Degree of Freedom) ๊ฐœ์ˆ˜, FEM ์‹œ์Šคํ…œ์˜ ๋ณ€์ˆ˜ ๊ฐœ์ˆ˜๋กœ ์‚ฌ์šฉ.

    • ์‹์˜ ์˜๋ฏธ : FEM์€ ํ•ด $u(x)$๋ฅผ ์ •ํ™•ํžˆ ๊ตฌํ•  ์ˆ˜ ์—†๊ธฐ ๋•Œ๋ฌธ์—, ๊ธฐ์ € ํ•จ์ˆ˜๋“ค์˜ ์„ ํ˜• ๊ฒฐํ•ฉ์œผ๋กœ ๊ทผ์‚ฌํ•œ๋‹ค. ์ฆ‰, $u_h(x) = \alpha_1 \psi_1(x) + \alpha_2 \psi_2(x) + \cdots + \alpha_n \psi_n(x)$ ์ด์ฒ˜๋Ÿผ $\psi_{i}(x)$๋ผ๋Š” ๊ฐ„๋‹จํ•œ ํ•จ์ˆ˜๋“ค์„ ์—ฌ๋Ÿฌ ๊ฐœ ํ•ฉ์ณ์„œ, ๋ณต์žกํ•œ ํ•ด $u(x)$๋ฅผ ๊ทผ์‚ฌํ•œ๋‹ค.

Various types of finite elements

Figure 1. ์ฐจ์›๋ณ„ ์œ ํ•œ์š”์†Œ ์ข…๋ฅ˜

  • ์š”์†Œ(Element)
    • FEM์—์„œ ๋ณต์žกํ•œ ์—ฐ์†์ฒด(์˜ˆ: ๊ตฌ์กฐ๋ฌผ, ๊ธฐ๊ณ„๋ถ€ํ’ˆ, ๋ผˆ, ๋‚ ๊ฐœ ๋“ฑ)๋ฅผ ์ž‘์€ ์กฐ๊ฐ๋“ค๋กœ ๋ถ„ํ• ํ•ด์„œ ๋ถ„์„ํ•  ๋•Œ, ์ž‘์€ ์กฐ๊ฐ ํ•˜๋‚˜ํ•˜๋‚˜๋ฅผ ์š”์†Œ(Element) ๋ผ๊ณ  ๋ถ€๋ฅธ๋‹ค.
    • ๊ฐ ์ฐจ์›์—์„œ์˜ ์š”์†Œ๋Š”?
      • 1์ฐจ์› -> ์„ ๋ถ„(line segment or line element or linear element)
        • ๋‘ ๋…ธ๋“œ๋กœ ๊ตฌ์„ฑ๋œ 1์ฐจ์› ์š”์†Œ
      • 2์ฐจ์› -> ์‚ผ๊ฐํ˜•(triangular element) / ์‚ฌ๊ฐํ˜•(quadrilateral element)
        • ์„ธ ๋…ธ๋“œ๋กœ ๊ตฌ์„ฑ๋œ ํ‰๋ฉด ์š”์†Œ / ๋„ค ๋…ธ๋“œ๋กœ ๊ตฌ์„ฑ๋œ ํ‰๋ฉด ์š”์†Œ
      • 3์ฐจ์› -> ์‚ฌ๋ฉด์ฒด(tetrahedral element) / ์œก๋ฉด์ฒด(hexahedral element or brick element)
        • ๋„ค ๊ผญ์ง“์ ์˜ 3D ์š”์†Œ / ์—ฌ๋Ÿ ๊ผญ์ง“์ ๊ณผ ์—ฌ์„ฏ ๋ฉด์œผ๋กœ ๊ตฌ์„ฑ๋œ ๋ฒฝ๋Œ ํ˜•ํƒœ์˜ 3D ์š”์†Œ
  • ์ž์ฝ”๋น„์•ˆ(Jacobian)
    • ์ขŒํ‘œ๊ณ„ ๊ฐ„์˜ ๋ณ€ํ™˜์œจ์„ ๋‚˜ํƒ€๋‚ด๋Š” ๋„ํ•จ์ˆ˜ ํ–‰๋ ฌ์ด๋‹ค. ๊ฐ„๋‹จํžˆ ๋งํ•ด, ์–ด๋–ค ์ขŒํ‘œ๊ณ„(์˜ˆ: ์ฐธ์กฐ ์ขŒํ‘œ $\hat{x}$์—์„œ ๋‹ค๋ฅธ ์ขŒํ‘œ๊ณ„(์˜ˆ: ์‹ค์ œ ์ขŒํ‘œ $x$)๋กœ ๋ณ€ํ™˜ํ•  ๋•Œ, ํ•œ ์  ๊ทผ์ฒ˜์˜ ๋ณ€ํ™”๋ฅผ ์–ผ๋งˆ๋‚˜ ๋Š˜์ด๊ฑฐ๋‚˜ ์ค„์ด๋Š”์ง€๋ฅผ ๋‚˜ํƒ€๋‚ด๋Š” ๊ฐ’์ด๋‹ค.
  • ์•„ํ•€๋ณ€ํ™˜(Affine Transformation)
    • ์„ ํ˜• ๋ณ€ํ™˜(linear transformation)๊ณผ ์ด๋™(translation)์„ ํ•ฉ์นœ ์ขŒํ‘œ๊ณ„ ๋ณ€ํ™˜ ๋ฐฉ๋ฒ•์œผ๋กœ, ์ด ๋‘ ๋ณ€ํ™˜์„ ์กฐํ•ฉํ•˜๋ฉด ๋„ํ˜•์ด๋‚˜ ์ขŒํ‘œ๋ฅผ ์„ ํ˜•์ ์œผ๋กœ ์™œ๊ณกํ•˜๊ณ  ๋™์‹œ์— ์œ„์น˜๋„ ๋ฐ”๊ฟ€ ์ˆ˜ ์žˆ๋Š” ๊ต‰์žฅํžˆ ์œ ์—ฐํ•œ ๋ณ€ํ™˜ ๋ฐฉ์‹์ด๋‹ค.

3. 1์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•, 1D Finite Element Method

  • 1์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•์ด๋ž€?
    • ์œ ํ•œ ์š”์†Œ๋ฒ•(FEM)์˜ ๊ฐ€์žฅ ๊ธฐ์ดˆ์ ์ธ ํ˜•ํƒœ๋กœ, ์ฃผ๋กœ ๋ง‰๋Œ€(bar), ๋น”(beam), ์—ด์ „๋‹ฌ ๋กœ๋“œ(rod) ๋“ฑ์˜ ๊ตฌ์กฐ๋‚˜ ์‹œ์Šคํ…œ์„ ํ•ด์„ํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋œ๋‹ค.
  • ์ˆ˜์‹ ํ‘œํ˜„
    • $u_{h}(x) = \sum_{i=1}^{2}\alpha_{i}\psi_{i}(x) = \alpha_{1}\psi_{1}(x) + \alpha_{2}\psi_{2}(x)$
  • ์š”์†Œ ๊ตฌ๊ฐ„ : $x\in[0,1]$
    • ์š”์†Œ ๊ตฌ๊ฐ„์ด $[0, 1]$์ธ ์ด์œ 
      • ์ •๊ทœ ์ขŒํ‘œ๊ณ„(natural coordinate)๋ฅผ ์‚ฌ์šฉํ•˜๊ธฐ ์œ„ํ•ด
      • ์‹ค์ œ ์š”์†Œ ๊ตฌ๊ฐ„์ด [2, 5]์ผ์ง€๋ผ๋„, ์ •๊ทœ ์ขŒํ‘œ๊ณ„๋ฅผ ์‚ฌ์šฉํ•จ์œผ๋กœ์จ ๋ชจ๋“  ์š”์†Œ๋ฅผ ๋™์ผํ•˜๊ฒŒ ํ•ด์„ํ•  ์ˆ˜ ์žˆ๊ณ , ํ•œ ๋ฒˆ ์ •์˜ํ•œ ํ˜•์ƒ ํ•จ์ˆ˜ ๋ฐ ์ ๋ถ„ ๋“ฑ์˜ ๊ณ„์‚ฐ์„ ๋ชจ๋“  ์š”์†Œ์— ์žฌ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•˜๋‹ค.
      • ์‹ค์ œ ์š”์†Œ ๊ตฌ๊ฐ„(์‹ค์ œ ์ขŒํ‘œ๊ณ„, physical coordinates)์„ ์ •๊ทœํ™”๋œ ์š”์†Œ ๊ตฌ๊ฐ„ $\hat{x}\in[0,1]$์œผ๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ๋‹จ์ˆœํ•˜๊ฒŒ ํ•ด์„ํ•œ๋‹ค.
        • ์‹ค์ œ ์š”์†Œ -> ์ •๊ทœํ™”๋œ ์š”์†Œ : $x = \alpha + (b - a)\hat{x}$
        • ์ •๊ทœํ™”๋œ ์š”์†Œ -> ์‹ค์ œ ์š”์†Œ : $\hat{x} = (x - a) / (b - a)$
  • ๋…ธ๋“œ(Node) :
    • $ x = 0 $ ๊ณผ $ x = 1 $
  • ํ˜•์ƒ ํ•จ์ˆ˜(Shape function) :
    • $\psi_{1}(x) = 1 - x$
    • $\psi_{2}(x) = x$
  • ํ•ด์˜ ๊ทผ์‚ฌ :
    • $u_{h}(x) = \alpha_{1}(1-x) + \alpha_{2}(x)$
  • ํ•ด์„์  ์˜๋ฏธ :
    • $\alpha_{1}$ ๊ณผ $\alpha_{2}$ ๋Š” ๊ฐ๊ฐ ๋…ธ๋“œ์—์„œ์˜ ๋ฌผ๋ฆฌ๋Ÿ‰(์˜ˆ: ๋ณ€์œ„)
    • ํ•ด๋Š” ๋‘ ์  ์‚ฌ์ด๋ฅผ ์ง์„ ์œผ๋กœ ๋ณด๊ฐ„ํ•œ ํ˜•ํƒœ

4. 2์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•, 2D Finite Element Method

  • 2์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•์ด๋ž€?
    • 2์ฐจ์› FEM์€, 1์ฐจ์› FEM์„ ํ™•์žฅํ•œ ๊ฐœ๋…์œผ๋กœ, ํ•ด์„ํ•˜๋ ค๋Š” ์˜์—ญ(์˜ˆ: ๊ตฌ์กฐ๋ฌผ์˜ ๋‹จ๋ฉด, ํŒ, ํ‰๋ฉด ์˜์—ญ ๋“ฑ)์„ ์‚ผ๊ฐํ˜• ๋˜๋Š” ์‚ฌ๊ฐํ˜• ์š”์†Œ๋“ค๋กœ ๋ถ„ํ• ํ•˜๊ณ , ๊ฐ ์š”์†Œ ๋‚ด์—์„œ ํ•ด(๋ณ€์œ„, ์˜จ๋„ ๋“ฑ)๋ฅผ ๋ณด๊ฐ„ํ•˜์—ฌ ์ „์ฒด ํ•ด๋ฅผ ๊ทผ์‚ฌํ•œ๋‹ค.
  • ์ˆ˜์‹ ํ‘œํ˜„
    • ์„ ํ˜• ์‚ผ๊ฐํ˜• ์š”์†Œ(Linear Triangle Element)
      • $u_{h}(x, y) = \sum_{i=1}^{3}\alpha_{i}\psi_{i}(x, y) = \alpha_1 \psi_1(x, y) + \alpha_2 \psi_2(x, y) + \alpha_3 \psi_3(x, y)$
  • ์•„ํ•€ ๋ณ€ํ™˜(Affine Transformation)
    • ์™œ 2์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•์—์„œ ์•„ํ•€ ๋ณ€ํ™˜์ด ์‚ฌ์šฉ๋˜๋Š”๊ฐ€?
      • ์ผ๋ฐ˜์ ์œผ๋กœ ์œ ํ•œ ์š”์†Œ๋ฒ•์—์„œ๋Š” ์ฐธ์กฐ ์š”์†Œ(reference element)๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ •ํ˜•ํ™”๋œ ๊ณต๊ฐ„์—์„œ ๊ณ„์‚ฐ์ด ์ด๋ค„์ง„๋‹ค. ์ด๋ฏธ ๊ณ„์‚ฐ๋ผ ์žˆ๋Š” ์ฐธ์กฐ ์š”์†Œ๋ฅผ ์•„ํ•€ ๋ณ€ํ™˜์„ ์ ์šฉํ•˜์—ฌ ์‹ค์ œ ์š”์†Œ(physical element)๋กœ ๋ณ€ํ™˜ํ•œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๊ตฌํ•˜๊ณ ์žํ•˜๋Š” ์˜์—ญ์„ ์‰ฝ๊ฒŒ ๊ณ„์‚ฐํ•  ์ˆ˜ ์žˆ๋‹ค.
  • ์‚ผ๊ฐํ˜• ์š”์†Œ, Triangle Element
    • ์„ ํ˜• ์‚ผ๊ฐํ˜• ์š”์†Œ, Reference Triangle, $T_{R}$ :
      • FEM์—์„œ ๋ชจ๋“  ์‚ผ๊ฐํ˜• ์š”์†Œ๋ฅผ ๋™์ผํ•œ ๋ฐฉ์‹์œผ๋กœ ํ•ด์„ํ•˜๊ธฐ ์œ„ํ•ด ์‚ฌ์šฉํ•˜๋Š” ํ‘œ์ค€ ์‚ผ๊ฐํ˜•. ์ผ๋ฐ˜์ ์œผ๋กœ ์„ธ ๊ผญ์ง“์ ์„ ๊ฐ–๋Š” ๋‹จ์œ„ ์‚ผ๊ฐํ˜•์œผ๋กœ ์ •์˜๋จ

      • ์ด ์ •๊ทœํ™”๋œ ์‚ผ๊ฐํ˜• ์œ„์—์„œ ํ˜•์ƒ ํ•จ์ˆ˜(shape function), ์ˆ˜์น˜ ์ ๋ถ„(Gauss quadrature) ๋“ฑ์„ ์ •์˜ํ•˜๊ณ , ๋ชจ๋“  ์‹ค์ œ ์‚ผ๊ฐํ˜• ์š”์†Œ๋Š” ์ด ๊ธฐ์ค€ ์‚ผ๊ฐํ˜•์œผ๋กœ๋ถ€ํ„ฐ์˜ ์•„ํ•€ ์‚ฌ์ƒ(affine mapping)์„ ํ†ตํ•ด ๋ณ€ํ™˜๋จ.

\[T_{R} = \left\{ (\hat{x}, \hat{y}) \mid{\hat{x} \ge{0},\ \hat{y} \ge{0},\ \hat{x} + \hat{y} \le{1}} \right\}\]

* * ์‹ค์ œ ์‚ผ๊ฐํ˜• ์š”์†Œ, Physical Triangle, $T_{P}$ : * ์‹ค์ œ ํ•ด์„ ๋Œ€์ƒ์˜ ๋ฌผ๋ฆฌ ๊ณต๊ฐ„(physical domain)์— ์กด์žฌํ•˜๋Š” ์ž„์˜์˜ ์‚ผ๊ฐํ˜• ์š”์†Œ์ด๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด, 2D ๊ตฌ์กฐ๋ฌผ์˜ ๋ฉ”์‰ฌ๋ฅผ ๊ตฌ์„ฑํ•˜๋Š” ์‚ผ๊ฐํ˜•๋“ค์ด ์—ฌ๊ธฐ์— ํ•ด๋‹น๋œ๋‹ค. * ์ด ์‹ค์ œ ์‚ผ๊ฐํ˜•์€ ์„ธ ๊ผญ์ง“์ ์˜ ์ขŒํ‘œ $ (x_1, y_1), (x_2, y_2), (x_3, y_3) $๋กœ ์ •์˜๋˜๊ณ , ๊ณ„์‚ฐ์„ ์œ„ํ•ด ๊ธฐ์ค€ ์‚ผ๊ฐํ˜• $ T_R $์—์„œ ๋ฌผ๋ฆฌ ์‚ผ๊ฐํ˜• $ T_P $๋กœ์˜ ์•„ํ•€ ๋ณ€ํ™˜์ด ์ ์šฉ๋œ๋‹ค:

\[(x, y) = x_1 + (x_2 - x_1)\hat{x} + (x_3 - x_1)\hat{y}\] \[(y, y) = y_1 + (y_2 - y_1)\hat{x} + (y_3 - y_1)\hat{y}\]
    • ๐Ÿ” $T_{R}$ ์™€ $T_{P}$ ๋น„๊ต
ํ•ญ๋ชฉ Reference Triangle $ T_R $ Physical Triangle $ T_P $
์ •์˜ ์ •๊ทœํ™”๋œ ๊ธฐ์ค€ ์‚ผ๊ฐํ˜• ์‹ค์ œ ํ•ด์„ ๋Œ€์ƒ ์‚ผ๊ฐํ˜•
์ขŒํ‘œ๊ณ„ $ (\hat{x}, \hat{y}) \in [0,1] $ $ (x, y) \in \mathbb{R}^2 $
๋ชฉ์  ํ˜•์ƒ ํ•จ์ˆ˜ ์ •์˜, ์ˆ˜์น˜ ์ ๋ถ„ ํ†ต์ผ ์‹ค์ œ ๋„๋ฉ”์ธ์˜ ๋ฌผ๋ฆฌ ์ •๋ณด ํ‘œํ˜„
๋ณ€ํ™˜ ์—†์Œ $ T_R \to T_P $ ๋กœ ์–ดํŒŒ์ธ ์‚ฌ์ƒ ์ ์šฉ๋จ

5. 2์ฐจ์› FEM์˜ ์‚ผ๊ฐํ˜• ์š”์†Œ์— ์•„ํ•€ ๋ณ€ํ™˜ ์ ์šฉ

์„ธ ๊ผญ์ง“์ ์„ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์ •์˜ํ•œ๋‹ค:

๋…ธ๋“œ ์‹ค์ œ ์ขŒํ‘œ $(x_i, y_i)$ ์ฐธ์กฐ ์ขŒํ‘œ $(\hat{\xi}_i, \hat{\eta}_i)$
1 $(x_1, y_1)$ $(0, 0)$
2 $(x_2, y_2)$ $(1, 0)$
3 $(x_3, y_3)$ $(0, 1)$

์ด๋•Œ ์•„ํ•€ ํ–‰๋ ฌ $A$์™€ ์ด๋™ ๋ฒกํ„ฐ $\mathbf{b}$๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์ •์˜๋œ๋‹ค:

\[A = \begin{bmatrix} x_2 - x_1 & x_3 - x_1 \\ y_2 - y_1 & y_3 - y_1 \end{bmatrix}, \quad \mathbf{b} = \begin{bmatrix} x_1 \\ y_1 \end{bmatrix}\]

๋”ฐ๋ผ์„œ ์ „์ฒด ๋งคํ•‘์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค:

\[\begin{bmatrix} x \\ y \end{bmatrix} = \begin{bmatrix} x_2 - x_1 & x_3 - x_1 \\ y_2 - y_1 & y_3 - y_1 \end{bmatrix} \cdot \begin{bmatrix} \hat{\xi} \\ \hat{\eta} \end{bmatrix} + \begin{bmatrix} x_1 \\ y_1 \end{bmatrix}\]

6. ์—ญ๋ณ€ํ™˜ (์‹ค์ œ โ†’ ์ฐธ์กฐ)

์•„ํ•€ ๋ณ€ํ™˜์€ ์„ ํ˜•์ด๋ฏ€๋กœ ํ–‰๋ ฌ $A$๊ฐ€ ๊ฐ€์—ญ์ ์ด๋ฉด ์—ญ๋ณ€ํ™˜๋„ ๊ฐ€๋Šฅํ•˜๋‹ค:

\[\begin{bmatrix} \hat{\xi} \\ \hat{\eta} \end{bmatrix} = A^{-1} \cdot \left( \begin{bmatrix} x \\ y \end{bmatrix} - \begin{bmatrix} x_1 \\ y_1 \end{bmatrix} \right)\]

์ด ์—ญ๋ณ€ํ™˜์€ ์ ๋ถ„์  ๋˜๋Š” ๋ฌผ๋ฆฌ ์ขŒํ‘œ๋ฅผ ์ฐธ์กฐ ์ขŒํ‘œ๋กœ ๋งคํ•‘ํ•  ๋•Œ ๋งค์šฐ ์ค‘์š”ํ•˜๊ฒŒ ์‚ฌ์šฉ๋œ๋‹ค.


7. 3์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•, 3D Finite Element Method

  • 3์ฐจ์› ์œ ํ•œ ์š”์†Œ๋ฒ•(3D FEM)์ด๋ž€?

    • 3D FEM์€ ํ•ด์„ํ•˜๋ ค๋Š” ๋ฌผ๋ฆฌ์  ๊ตฌ์กฐ(์˜ˆ: ๊ณ ์ฒด, ๋ฉ์–ด๋ฆฌ ๊ตฌ์กฐ๋ฌผ, ๋ถ€ํ”ผ๋ฅผ ๊ฐ€์ง„ ์‹œ์Šคํ…œ)๋ฅผ 3์ฐจ์› ์š”์†Œ๋“ค(์˜ˆ: ์‚ฌ๋ฉด์ฒด, ์œก๋ฉด์ฒด ๋“ฑ)๋กœ ๋ถ„ํ• ํ•˜๊ณ , ๊ฐ ์š”์†Œ ๋‚ด์—์„œ ๋ฌผ๋ฆฌ๋Ÿ‰(๋ณ€์œ„, ์‘๋ ฅ ๋“ฑ)์„ ๋ณด๊ฐ„ํ•˜์—ฌ ์ „์ฒด ํ•ด๋ฅผ ๊ทผ์‚ฌํ•˜๋Š” ๋ฐฉ๋ฒ•์ด๋‹ค.
  • ๊ธฐ๋ณธ ์š”์†Œ ํ˜•ํƒœ

    • ์‚ฌ๋ฉด์ฒด(Tetrahedron), ์œก๋ฉด์ฒด(Hexahedron), ํ”„๋ฆฌ์ฆ˜(Prism) ๋“ฑ์ด ์‚ฌ์šฉ๋˜๋ฉฐ, ์ด ์ค‘ ์‚ฌ๋ฉด์ฒด(Tetrahedral element)๊ฐ€ ๊ฐ€์žฅ ๊ธฐ๋ณธ์ ์ธ 3์ฐจ์› ์š”์†Œ๋กœ ๋งŽ์ด ์“ฐ์ž„.
  • ๊ธฐ์ € ํ•จ์ˆ˜ ํ‘œํ˜„ ์˜ˆ

    • ์„ ํ˜• ์‚ฌ๋ฉด์ฒด ์š”์†Œ์—์„œ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์ด 4๊ฐœ์˜ shape function์„ ์‚ฌ์šฉํ•˜์—ฌ ํ•ด๋ฅผ ๊ทผ์‚ฌํ•œ๋‹ค:
    \[u_h(x, y, z) = \sum_{i=1}^{4} \alpha_i \, \psi_i(x, y, z)\]

8. 3์ฐจ์› FEM์˜ ์‚ฌ๋ฉด์ฒด ์š”์†Œ์— ์•„ํ•€ ๋ณ€ํ™˜ ์ ์šฉ

๊ธฐ์ค€ ์‚ฌ๋ฉด์ฒด(Reference Tetrahedron), $T_R$

  • ๋ณดํ†ต ๋‹ค์Œ์˜ ์ •๊ทœํ™”๋œ ์ ๋“ค์„ ๊ผญ์ง“์ ์œผ๋กœ ์‚ฌ์šฉ:

    • $(0, 0, 0)$
    • $(1, 0, 0)$
    • $(0, 1, 0)$
    • $(0, 0, 1)$

์‹ค์ œ ์‚ฌ๋ฉด์ฒด ์š”์†Œ(Physical Tetrahedron), $T_P$

  • ์‹ค์ œ ํ•ด์„ ๋Œ€์ƒ์— ์กด์žฌํ•˜๋Š” ์‚ฌ๋ฉด์ฒด ์š”์†Œ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์€ ๊ผญ์ง“์  ์ขŒํ‘œ๋ฅผ ๊ฐ€์ง:

    • $(x_1, y_1, z_1)$
    • $(x_2, y_2, z_2)$
    • $(x_3, y_3, z_3)$
    • $(x_4, y_4, z_4)$

์•„ํ•€ ํ–‰๋ ฌ $A$ ๋ฐ ์ด๋™ ๋ฒกํ„ฐ $\mathbf{b}$

๊ธฐ์ค€์  $\hat{\mathbf{x}} = (\hat{\xi}, \hat{\eta}, \hat{\zeta})$๋ฅผ ์‹ค์ œ ๊ณต๊ฐ„ $\mathbf{x} = (x, y, z)$๋กœ ๋งคํ•‘ํ•˜๋Š” ์‹:

\[\mathbf{x} = A \cdot \hat{\mathbf{x}} + \mathbf{b}\]

์—ฌ๊ธฐ์„œ:

  • $A \in \mathbb{R}^{3 \times 3}$: 3D ์„ ํ˜• ๋ณ€ํ™˜ ํ–‰๋ ฌ
  • $\mathbf{b} \in \mathbb{R}^{3}$: ์ด๋™ ๋ฒกํ„ฐ (๊ธฐ์ค€์  ์ด๋™)
\[A = \begin{bmatrix} x_2 - x_1 & x_3 - x_1 & x_4 - x_1 \\ y_2 - y_1 & y_3 - y_1 & y_4 - y_1 \\ z_2 - z_1 & z_3 - z_1 & z_4 - z_1 \end{bmatrix}, \quad \mathbf{b} = \begin{bmatrix} x_1 \\ y_1 \\ z_1 \end{bmatrix}\]

๋”ฐ๋ผ์„œ ์ „์ฒด ๋งคํ•‘์€:

\[\begin{bmatrix} x \\ y \\ z \end{bmatrix} = A \cdot \begin{bmatrix} \hat{\xi} \\ \hat{\eta} \\ \hat{\zeta} \end{bmatrix} + \begin{bmatrix} x_1 \\ y_1 \\ z_1 \end{bmatrix}\]

9. ์—ญ๋ณ€ํ™˜ (์‹ค์ œ โ†’ ์ฐธ์กฐ)

์•„ํ•€ ๋ณ€ํ™˜ $\mathbf{x} = A \hat{\mathbf{x}} + \mathbf{b}$๋Š” ์„ ํ˜•์ด๋ฏ€๋กœ, ํ–‰๋ ฌ $A$๊ฐ€ ๊ฐ€์—ญ์ ์ผ ๊ฒฝ์šฐ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ์—ญ๋ณ€ํ™˜์ด ๊ฐ€๋Šฅํ•˜๋‹ค:

\[\hat{\mathbf{x}} = A^{-1} \cdot \left( \mathbf{x} - \mathbf{b} \right)\]

์ฆ‰,

\[\begin{bmatrix} \hat{\xi} \\ \hat{\eta} \\ \hat{\zeta} \end{bmatrix} = A^{-1} \cdot \left( \begin{bmatrix} x \\ y \\ z \end{bmatrix} - \begin{bmatrix} x_1 \\ y_1 \\ z_1 \end{bmatrix} \right)\]

์ด ์—ญ๋ณ€ํ™˜์€ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ƒํ™ฉ์—์„œ ์‚ฌ์šฉ๋œ๋‹ค:

  • ์ ๋ถ„์ ์ด ์‹ค์ œ ์š”์†Œ ์ƒ์— ์ฃผ์–ด์กŒ์„ ๋•Œ, ๊ธฐ์ค€ ์‚ฌ๋ฉด์ฒด ์ขŒํ‘œ๋กœ ๋ณ€ํ™˜ํ•ด shape function ๋“ฑ์„ ํ‰๊ฐ€ํ•  ๋•Œ
  • ํ•ด์„ ํ•ด๋ฅผ ์‹œ๊ฐํ™”ํ•˜๊ฑฐ๋‚˜ ํ›„์ฒ˜๋ฆฌํ•  ๋•Œ, ๊ธฐ์ค€ ๊ณต๊ฐ„์—์„œ์˜ ์ •๋ณด๋ฅผ ๋ฌผ๋ฆฌ ๊ณต๊ฐ„์œผ๋กœ ๋ณ€ํ™˜ํ•˜๊ธฐ ์œ„ํ•ด

10. ๊ณผ์ œ ๋ฐ ์ถ”๊ฐ€ ์—ฐ๊ตฌ

๋ฌผ๋ก ์ž…๋‹ˆ๋‹ค! ์ง€๊ธˆ๊นŒ์ง€์˜ ๋‚ด์šฉ์„ ๋ฐ”ํƒ•์œผ๋กœ ํ•œ 10. ๊ณผ์ œ ๋ฐ ์ถ”๊ฐ€ ์—ฐ๊ตฌ ํ•ญ๋ชฉ์„ ์•„๋ž˜์— ๊ตฌ์„ฑํ•ด๋“œ๋ ธ์Šต๋‹ˆ๋‹ค. ์ด ํ•ญ๋ชฉ์€ ๊ฐœ๋… ์ •๋ฆฌ, ์‹ค์ œ ๊ณ„์‚ฐ ์ ์šฉ, ํ”„๋กœ๊ทธ๋ž˜๋ฐ ๊ตฌํ˜„, ์‹ฌํ™” ํ•™์Šต์œผ๋กœ ์ด์–ด์ง€๋Š” ๋‹จ๊ณ„์  ํ•™์Šต ๋ชฉํ‘œ์™€ ํƒ๊ตฌ ๊ณผ์ œ๋ฅผ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.


10. ๊ณผ์ œ ๋ฐ ์ถ”๊ฐ€ ์—ฐ๊ตฌ -> ChatGPT ์š”์ฒญ ๊ฒฐ๊ณผ

๊ธฐ๋ณธ ๊ฐœ๋… ๋ณต์Šต

  1. ๊ธฐ์ดˆ ์šฉ์–ด ์ •๋ฆฌ

    • ์œ ํ•œ ์š”์†Œ๋ฒ•(FEM), ์š”์†Œ(Element), ์ž์œ ๋„(DOF), ๊ธฐ์ € ํ•จ์ˆ˜(Shape Function), ์ž์ฝ”๋น„์•ˆ(Jacobian), ์•„ํ•€ ๋ณ€ํ™˜(Affine Transformation)์˜ ์ •์˜์™€ ์—ญํ• ์„ ์Šค์Šค๋กœ ์„ค๋ช…ํ•ด๋ณด๊ธฐ
    • 1D, 2D, 3D ์š”์†Œ์˜ ์ฐจ์ด์  ์š”์•ฝ ์ •๋ฆฌํ•˜๊ธฐ
  2. ์ˆ˜์‹ ์œ ๋„ ๋ฐ ์† ๊ณ„์‚ฐ

    • 1์ฐจ์› ์„ ํ˜• ์š”์†Œ์—์„œ ํ•ด $u_h(x)$์˜ ์ˆ˜์‹ ์œ ๋„ ๋ฐ ์†๊ณ„์‚ฐ์œผ๋กœ ๊ทผ์‚ฌ๊ฐ’ ๊ตฌํ•˜๊ธฐ
    • ์„ ํ˜• ์‚ผ๊ฐํ˜• ์š”์†Œ์—์„œ์˜ ์•„ํ•€ ๋ณ€ํ™˜ ํ–‰๋ ฌ $A$์™€ ์—ญ๋ณ€ํ™˜ ์ˆ˜์‹ ์œ ๋„ํ•ด ๋ณด๊ธฐ

๊ณ„์‚ฐ ๋ฐ ์ˆ˜์น˜ ์˜ˆ์ œ

  1. 1D FEM ์ˆ˜์น˜ ๊ตฌํ˜„ ๊ณผ์ œ

    • ๊ตฌ๊ฐ„ [0,1]์„ 2๊ฐœ ์š”์†Œ๋กœ ๋‚˜๋ˆˆ FEM ๋ชจ๋ธ์—์„œ ๊ฐ ์š”์†Œ์˜ ๊ฐ•์„ฑ ํ–‰๋ ฌ์„ ๊ณ„์‚ฐํ•˜๊ณ  ์ „์ฒด ์‹œ์Šคํ…œ ๋ฐฉ์ •์‹ ์กฐ๋ฆฝํ•ด๋ณด๊ธฐ
    • Dirichlet ๊ฒฝ๊ณ„ ์กฐ๊ฑด์„ ์ฃผ๊ณ  ์ž์œ ๋„ ํ•ด $\alpha_i$๋ฅผ ๊ตฌํ•˜๋Š” ์‹œ์Šคํ…œ ๊ตฌํ˜„
  2. 2D ์‚ผ๊ฐํ˜• ์š”์†Œ์˜ ์ž์ฝ”๋น„์•ˆ ๊ณ„์‚ฐ

    • ์„ธ ๊ผญ์ง“์ ์ด ์ฃผ์–ด์ง„ ์‹ค์ œ ์‚ผ๊ฐํ˜•์—์„œ ์•„ํ•€ ํ–‰๋ ฌ $A$ ๊ตฌ์„ฑ
    • ์ž์ฝ”๋น„์•ˆ $J = \det(A)$, $J^{-1}$์„ ๊ณ„์‚ฐํ•˜๊ณ  ๋„ํ•จ์ˆ˜ ๋ณ€ํ™˜์— ํ™œ์šฉ
  3. 3D ์‚ฌ๋ฉด์ฒด ์š”์†Œ์˜ ์•„ํ•€ ๋ณ€ํ™˜ ์ ์šฉ

    • ์ฃผ์–ด์ง„ ์‚ฌ๋ฉด์ฒด์˜ ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๋ฅผ ๊ฐ€์ง€๊ณ  ์•„ํ•€ ๋ณ€ํ™˜ $A$ ๋ฐ $A^{-1}$์„ ๊ณ„์‚ฐ
    • ๊ธฐ์ค€ ์ขŒํ‘œ $(\hat{\xi}, \hat{\eta}, \hat{\zeta})$์—์„œ ์‹ค์ œ ์ขŒํ‘œ $(x, y, z)$, ๋ฐ˜๋Œ€ ๋ฐฉํ–ฅ ๋ณ€ํ™˜๋„ ์ˆ˜ํ–‰

ํ”„๋กœ๊ทธ๋ž˜๋ฐ

  1. Python์œผ๋กœ FEM ๊ตฌํ˜„

    • 1D ๋˜๋Š” 2D FEM ํ•ด์„ ํ”„๋กœ๊ทธ๋žจ ๊ตฌ์„ฑ (์˜ˆ: ์—ด์ „๋‹ฌ, ๋ณ€์œ„ ํ•ด์„)
    • ์š”์†Œ ์กฐ๋ฆฝ, ํ˜•์ƒ ํ•จ์ˆ˜, ์ž์ฝ”๋น„์•ˆ, ์ ๋ถ„, ๊ฒฝ๊ณ„์กฐ๊ฑด ์ ์šฉ ๋“ฑ ์ „์ฒด ํ๋ฆ„ ํฌํ•จ
    • [์„ ํƒ] matplotlib๋ฅผ ์ด์šฉํ•œ ํ˜•์ƒ ํ•จ์ˆ˜ ์‹œ๊ฐํ™”
  2. ๊ธฐ์ € ํ•จ์ˆ˜ ์‹œ๊ฐํ™”

    • 1D: $\psi_1(x) = 1 - x, \ \psi_2(x) = x$
    • 2D: ์„ ํ˜• ์‚ผ๊ฐํ˜• ์š”์†Œ์˜ $\psi_1(\hat{x}, \hat{y}) = 1 - \hat{x} - \hat{y}$, ๋“ฑ๊ณ ์„  ๋˜๋Š” ์ƒ‰์ƒ ๋งต์œผ๋กœ ์‹œ๊ฐํ™”

๐Ÿ“š ์‹ฌํ™” ํ•™์Šต ์ฃผ์ œ

  1. ๊ณ ์ฐจ ์š”์†Œ ํƒ์ƒ‰

    • 2์ฐจ ๋˜๋Š” 3์ฐจ ๊ธฐ์ € ํ•จ์ˆ˜์˜ ์ •์˜ ๋ฐ ํ™œ์šฉ ๋ฐฉ๋ฒ• ์ •๋ฆฌ
    • 6-node ์‚ผ๊ฐํ˜• ์š”์†Œ, 10-node ์‚ฌ๋ฉด์ฒด ์š”์†Œ ๋“ฑ ๊ณ ์ฐจ ๋ณด๊ฐ„ ์š”์†Œ ๊ตฌ์กฐ ํ•™์Šต
  2. ์ˆ˜์น˜ ์ ๋ถ„(Gauss quadrature)

    • 1D, 2D, 3D์—์„œ ์‚ฌ์šฉ๋˜๋Š” Gaussian Quadrature ๊ทœ์น™ ํ•™์Šต
    • ์ •๊ทœ ์š”์†Œ ๊ธฐ์ค€์œผ๋กœ ์ˆ˜์น˜ ์ ๋ถ„ ๊ณต์‹ ์ง์ ‘ ๊ตฌํ˜„
  3. ์‘์šฉ ๋ถ„์•ผ ์กฐ์‚ฌ

    • ๊ตฌ์กฐ ํ•ด์„, ์ „์ž๊ธฐ ํ•ด์„, ์œ ์ฒด ํ๋ฆ„ ๋“ฑ FEM์ด ์‚ฌ์šฉ๋˜๋Š” ์‹ค์ œ ์‚ฐ์—… ์‚ฌ๋ก€ ์กฐ์‚ฌ
    • ์ƒ์—…์šฉ FEM ์†Œํ”„ํŠธ์›จ์–ด(์˜ˆ: ANSYS, Abaqus, COMSOL ๋“ฑ) ๊ธฐ๋Šฅ ๋น„๊ต

๋ชฉํ‘œ ์ •๋ฆฌ

๋‹จ๊ณ„ ๋ชฉํ‘œ
โ‘  ๊ฐœ๋… ์ดํ•ด FEM, ์•„ํ•€๋ณ€ํ™˜, ์ž์ฝ”๋น„์•ˆ ๋“ฑ ํ•ต์‹ฌ ์šฉ์–ด ๋ฐ ์ˆ˜์‹ ์ˆ™์ง€
โ‘ก ๊ณ„์‚ฐ ๋Šฅ๋ ฅ ์š”์†Œ ๋‹จ์œ„์˜ ์ˆ˜์‹ ๋ฐ ์ž์ฝ”๋น„์•ˆ ๊ณ„์‚ฐ ์—ญ๋Ÿ‰
โ‘ข ํ”„๋กœ๊ทธ๋ž˜๋ฐ Python/MATLAB ๊ธฐ๋ฐ˜ FEM ์ฝ”๋“œ ๊ตฌํ˜„
โ‘ฃ ํ™•์žฅ ํƒ๊ตฌ ๊ณ ์ฐจ ์š”์†Œ, ์ˆ˜์น˜ ์ ๋ถ„, ์‚ฐ์—… ์‘์šฉ ์ดํ•ด