Merge pull request #129 from GodotMisogi/develop

Version 0.4.3: Primarily updated docs + minor fixes

.github/workflows/CompatHelper.yml

@@ -1,7 +1,7 @@
 name: CompatHelper
 on:
  schedule:
- - cron: 0 0 * * *
+ - cron: 0 0 * * 1
  workflow_dispatch:
 jobs:
  CompatHelper:

Project.toml

@@ -1,7 +1,7 @@
 name = "AeroFuse"
 uuid = "477c59f4-51f5-487f-bf1e-8db39645b227"
 authors = ["GodotMisogi <[email protected]>"]
-version = "0.4.2"
+version = "0.4.3"
 
 [deps]
 Accessors = "7d9f7c33-5ae7-4f3b-8dc6-eff91059b697"

README.md

@@ -8,43 +8,44 @@
 <img width="60%", src="https://raw.githubusercontent.com/GodotMisogi/AeroFuse.jl/main/docs/src/assets/logo.svg">
 </p>
 
-AeroFuse is meant to be a toolbox for aircraft design analyses. It currently provides convenient methods for developing studies in aerodynamics and structures, with implementations in other relevant fields such as flight dynamics and propulsion in progress.
+AeroFuse is meant to be a toolbox for aircraft design analyses. It currently provides convenient methods for developing studies in aircraft geometry, aerodynamics and structures, with implementations in other relevant fields such as flight dynamics and propulsion in progress.
 
-**Authors**: Arjit Seth, Stephane Redonnet, and Rhea P. Liem
+**Authors**: Arjit Seth and Rhea P. Liem
 
 ## Objectives
 
-The current focus is to enable tutorials in computation for undergraduates in an aerospace educational curriculum, particularly at The Hong Kong University of Science and Technology. For this purpose, the code is written in a functional style replicating the mathematics presented in textbooks as much as possible.
+The current focus is to enable tutorials in computation in an aerospace educational curriculum, particularly at The Hong Kong University of Science and Technology. An additional aim is to write code compatible with automatic differentiation libraries written in Julia for enabling multidisciplinary studies.
 
-An additional aim is to write code compatible with automatic differentiation libraries written in Julia.
+>**Disclaimer**: The implementations are work-in-progress, and hence the results may not be entirely accurate. Please exercise caution when interpreting the results until validation cases are added.
 
 ## Features
 
-**Disclaimer**: The implementations are work-in-progress, and hence the results may not be entirely accurate. Please exercise caution when interpreting the results until validation cases are added.
-
 **Functioning:**
 
-- Basic geometric tools for airfoil processing and wing design. 
+- Basic geometric tools for airfoil processing, wing (and tail), and fuselage design. 
 - Steady, inviscid, incompressible, isentropic 2D and 3D aerodynamic analyses.
  - 2D - Doublet-source panel method
- - 3D - Vortex lattice method with derivatives
+ - 3D - Vortex lattice method
 - Semi-/empirical profile drag prediction methods for wings.
 - Finite-element beam model based on Euler-Bernoulli theory for structural analysis.
+- Automatic differentiation support primarily via [`ForwardDiff.jl`](https://github.com/JuliaDiff/ForwardDiff.jl).
+- Optimization capabilities with these features.
 
 ![](plots/VortexLattice.svg)
-![](plots/LinearVortex.svg)
+<!-- ![](plots/LinearVortex.svg) -->
+![](plots/SciMLWingOptimization.svg)
 
 **In progress:**
 
-- Aeroelastic analyses coupling vortex lattice and beam element models.
-- 2D integral boundary layer solver for viscous-inviscid analyses.
-- Generic, flexible and efficient flight dynamics integrator in 2 and 3 dimensions.
-- Fuselage geometry parametrization and aerodynamic implementation.
+- Aeroelastic analyses coupling vortex lattice and beam element models. This is functioning, but a good interface is yet to be implemented.
+- Adjoints for aerodynamic, structural and aerostructural design via [`ChainRulesCore.jl`](https://github.com/JuliaDiff/ChainRulesCore.jl).
 
 ![](plots/AerostructWingTail.svg)
 
 ## Installation
 
+Please install the current stable release of [Julia](https://julialang.org/downloads/) for your operating system and execute the following commands in the REPL.
+
 ```julia
 julia> using Pkg; Pkg.add("AeroFuse")
 julia> Pkg.test("AeroFuse")
@@ -57,15 +58,10 @@ If you use AeroFuse in your research, please cite the following until any releva
 
 ```bibtex
 @software{AeroFuse,
- author = {Arjit Seth, Stephane Redonnet, Rhea P. Liem},
+ author = {Arjit Seth, Rhea P. Liem},
  title = {AeroFuse},
  url = {https://github.com/GodotMisogi/AeroFuse},
- version = {0.4.2},
- date = {2023-02-23},
+ version = {0.4.3},
+ date = {2023-03-14},
 }
-```
-
-## References
-
-1. Mark Drela. _Flight Vehicle Aerodynamics_. MIT Press, 2014.
-2. Joseph Katz and Allen Plotkin. _Low-Speed Aerodynamics, Second Edition_. Cambridge University Press, 2001.
+```

docs/src/index.md

@@ -1,42 +1,43 @@
 # AeroFuse -- Aircraft Design Platform
 
-AeroFuse is meant to be a toolbox for aircraft design analyses. It currently provides convenient methods for developing studies in aerodynamics and structures, with implementations in other relevant fields such as flight dynamics and propulsion in progress.
+AeroFuse is meant to be a toolbox for aircraft design analyses. It currently provides convenient methods for developing studies in aircraft geometry, aerodynamics and structures, with implementations in other relevant fields such as flight dynamics and propulsion in progress.
 
-**Authors**: Arjit Seth, Stephane Redonnet, and Rhea P. Liem
+**Authors**: Arjit Seth and Rhea P. Liem
 
 ## Objectives
 
-The current focus is to enable tutorials in computation for undergraduates in an aerospace educational curriculum, particularly at The Hong Kong University of Science and Technology. For this purpose, the code is written in a functional style replicating the mathematics presented in textbooks as much as possible.
+The current focus is to enable tutorials in computation in an aerospace educational curriculum, particularly at The Hong Kong University of Science and Technology. An additional aim is to write code compatible with automatic differentiation libraries written in Julia for enabling multidisciplinary studies.
 
-An additional aim is to write code compatible with automatic differentiation libraries written in Julia.
+>**Disclaimer**: The implementations are work-in-progress, and hence the results may not be entirely accurate. Please exercise caution when interpreting the results until validation cases are added.
 
 ## Features
 
-**Disclaimer**: The implementations are work-in-progress, and hence the results may not be entirely accurate. Please exercise caution when interpreting the results until validation cases are added.
-
 **Functioning:**
 
-- Basic geometric tools for airfoil processing and wing design. 
+- Basic geometric tools for airfoil processing, wing (and tail), and fuselage design. 
 - Steady, inviscid, incompressible, isentropic 2D and 3D aerodynamic analyses.
  - 2D - Doublet-source panel method
- - 3D - Vortex lattice method with derivatives
-- Empirical viscous drag prediction methods for wings.
+ - 3D - Vortex lattice method
+- Semi-/empirical profile drag prediction methods for wings.
 - Finite-element beam model based on Euler-Bernoulli theory for structural analysis.
+- Automatic differentiation support primarily via [`ForwardDiff.jl`](https://github.com/JuliaDiff/ForwardDiff.jl).
+- Optimization capabilities with these features.
 
 ![](assets/VortexLattice.svg)
 ![](assets/LinearVortex.svg)
+![](assets/SciMLWingOptimization.svg)
 
 **In progress:**
 
-- Aeroelastic analyses coupling vortex lattice and beam element models.
-- 2D integral boundary layer solver for viscous-inviscid analyses.
-- Generic, flexible and efficient flight dynamics integrator in 2 and 3 dimensions.
-- Fuselage geometry parametrization and aerodynamic implementation.
+- Aeroelastic analyses coupling vortex lattice and beam element models. This is functioning, but a good interface is yet to be implemented.
+- Adjoints for aerodynamic, structural and aerostructural design via [`ChainRulesCore.jl`](https://github.com/JuliaDiff/ChainRulesCore.jl).
 
 ![](assets/AerostructWingTail.svg)
 
 ## Installation
 
+Please install the current stable release of [Julia](https://julialang.org/downloads/) for your operating system and execute the following commands in the REPL.
+
 ```julia
 julia> using Pkg; Pkg.add("AeroFuse")
 julia> Pkg.test("AeroFuse")
@@ -49,11 +50,11 @@ If you use AeroFuse in your research, please cite the following until any releva
 
 ```bibtex
 @software{AeroFuse,
- author = {Arjit Seth, Rhea P. Liem, Stephane Redonnet},
+ author = {Arjit Seth, Rhea P. Liem},
  title = {AeroFuse},
  url = {https://github.com/GodotMisogi/AeroFuse},
- version = {0.4.2},
- date = {2023-02-23},
+ version = {0.4.3},
+ date = {2023-03-14},
 }
 ```
 

examples/aerodynamics/3d_vortex_lattice_method/vlm_aircraft.jl

@@ -87,8 +87,8 @@ ref = References(
 ##
 @time system = solve_case(
  aircraft, fs, ref;
- # print = true, # Prints the results for only the aircraft
- # print_components = true, # Prints the results for all components
+ print = true, # Prints the results for only the aircraft
+ print_components = true, # Prints the results for all components
 );
 
 ## Compute dynamics

examples/geometry/foil.jl

@@ -18,6 +18,7 @@ cst_foil = kulfan_CST(alpha_u, alpha_l, (0., 0.), (0., 0.))
 
 ## Plot
 plot(cst_foil)
+plot!(naca_foil)
 
 ## Control surface
 cst_foil_flap = control_surface(cst_foil, 

examples/geometry/wing.jl

@@ -29,7 +29,7 @@ lambda_w_c4 = sweeps(wing, 0.25) # Quarter-chord sweep angles
 lambda_w_c2 = sweeps(wing, 0.5) # Half-chord sweep angles
 
 ct = camber_thickness(wing, 60) # Camber-thickness distribution
-coords = wing_bounds(wing) # Leading and trailing edge coordinates
+coords = coordinates(wing) # Leading and trailing edge coordinates
 
 ## Plotting
 using plots

examples/optimization/sciml_wing.jl

@@ -7,10 +7,10 @@ using Ipopt
 include("wing_definition.jl")
 
 ## Initial guess
-n_vars = 30 # Number of spanwise stations
+n_vars = 32 # Number of spanwise stations
 c = 0.125 # Fixed chord
 c_w = LinRange(c, c, n_vars) # Constant distribution
-CL_tgt = 1.5 # Target lift coefficient
+CL_tgt = 1. # Target lift coefficient
 nc = length(c_w)
 
 wing_init = make_wing(c_w)
@@ -33,7 +33,7 @@ end
 ## Initial run
 sys = make_case(α0, wing_init, refs)
 init = get_forces(sys, wing_init)
-# print_coefficients(sys)
+print_coefficients(sys)
 
 # Common
 function get_res(x, w_sweep = 0.25, ref= refs)
@@ -51,17 +51,10 @@ end
 function opt_drag(x, p) 
  sys, mesh = get_res(x)
 
- # Evaluate aerodynamic coefficients
- CDi, _, _, _, _, _ = nearfield(sys)
-
- # Calculate local-dissipation/local-friction drag
- CVs = norm.(surface_velocities(sys)).wing
- CDv = profile_drag_coefficient(mesh, 0.98, CVs, sys.reference)
-
- # @info "Variables:" x
- # @info "Objective:" CDi
+ # Get forces
+ res = get_forces(sys, mesh)
 
- CDi # + CDv
+ res.CD
 end
 
 # Constraints
@@ -122,7 +115,7 @@ xopt = sol.u
 wing_opt = make_wing(xopt[2:end])
 sys_opt = make_case(xopt[1], wing_opt, refs)
 opt = get_forces(sys_opt, wing_opt)
-# print_coefficients(sys_opt)
+print_coefficients(sys_opt)
 
 # Exact solution
 y_exact = LinRange(0., 1., n_vars)
@@ -131,7 +124,7 @@ x_exact = @. √(1. - y_exact^2) * 4 / π * c
 wing_exact = make_wing(x_exact)
 sys_exact = make_case(xopt[1], wing_exact, refs)
 exact = get_forces(sys_exact, wing_exact)
-# print_coefficients(sys_exact)
+print_coefficients(sys_exact)
 
 ## Plotting
 #==========================================================================================#
@@ -158,7 +151,7 @@ plt_opt = plot(
  legend = :bottom,
  xlabel = L"x,~m",
  guidefontrotation = 90.0,
- title = LaTeXString("Planform, \$ S = $(round(Sw; digits = 4)),~C_L = $(round(CL_tgt; digits = 4)) \$"),
+ title = LaTeXString("Planform, \$ S = $(round(Sw; digits = 4)),~C_{L_{req}} = $(round(CL_tgt; digits = 4)) \$"),
  grid = false,
  # aspect_ratio = 1,
  # zlim = (-0.5, 0.5) .* span(wing_init)
@@ -215,19 +208,19 @@ plot!(
  [ -cumsum(wing_init.surface.spans)[end:-1:1]; 0; cumsum(wing_init.surface.spans) ], 
  [ wing_init.surface.chords[end:-1:2]; wing_init.surface.chords ], 
  lc = :black,
- label = LaTeXString("Initial Wing: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([init.CDi;init.CDv;init.CD]; digits = 4)) \$"),
+ label = LaTeXString("Initial Wing: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([init.CDi;init.CDv;init.CD;init.CL]; digits = 4)) \$"),
 )
 plot!(
  [ -cumsum(wing_opt.surface.spans)[end:-1:1]; 0; cumsum(wing_opt.surface.spans) ], 
  [ wing_opt.surface.chords[end:-1:2]; wing_opt.surface.chords ],
  lc = :cornflowerblue, 
- label = LaTeXString("Optimized Wing: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([opt.CDi;opt.CDv;init.CD]; digits = 4)) \$"),
+ label = LaTeXString("Optimized Wing: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([opt.CDi;opt.CDv;opt.CD;opt.CL]; digits = 4)) \$"),
 )
 plot!(
  [ -cumsum(wing_exact.surface.spans)[end:-1:1]; 0; cumsum(wing_exact.surface.spans) ], 
  [ wing_exact.surface.chords[end:-1:2]; wing_exact.surface.chords ],
  lc = :green, 
- label =LaTeXString("Inviscid Optimum: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([exact.CDi;exact.CDv;exact.CD]; digits = 6)) \$"),
+ label = LaTeXString("Inviscid Optimum: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([exact.CDi;exact.CDv;exact.CD;exact.CL]; digits = 6)) \$"),
 )
 
 plt_CL = plot(

examples/optimization/snow_wing.jl

@@ -13,10 +13,10 @@ using Ipopt # The optimizer
 includet("wing_definition.jl")
 
 ## Initial guess
-n_vars = 30 # Number of spanwise stations
+n_vars = 64 # Number of spanwise stations
 c = 0.125 # Fixed chord
 c_w = LinRange(c, c, n_vars) # Constant distribution
-CL_tgt = 1.5 # Target lift coefficient
+CL_tgt = 1 # Target lift coefficient
 nc = length(c_w)
 
 wing_init = make_wing(c_w)
@@ -51,7 +51,7 @@ function optimize_drag!(g, x, w = 0.25, ref = refs)
  res = get_forces(system, wing_mesh)
 
  # Objective
- f = res.CDi
+ f = res.CD
 
  # Constraints
  g[1] = res.CL # Lift coefficient
@@ -88,7 +88,7 @@ opts = Options(
 )
 
 ## Run optimization
-@time xopt, fopt, info = minimize(optimize_drag!, x0, ng, lx, ux, lg, ug, opts)
+@time xopt, fopt, info = minimize(optimize_drag!, x0, ng, lx, ux, lg, ug, opts);
 
 ## Substitute
 wing_opt = make_wing(xopt[2:end])
@@ -131,7 +131,7 @@ plt_opt = plot(
  legend = :bottom,
  xlabel = L"x,~m",
  guidefontrotation = 90.0,
- title = LaTeXString("Planform, \$ S = $(round(Sw; digits = 6)),~C_L = $(round(CL_tgt; digits = 6)) \$"),
+ title = LaTeXString("Planform, \$ S = $(round(Sw; digits = 4)),~C_{L_{req}} = $(round(CL_tgt; digits = 4)) \$"),
  grid = false,
  # aspect_ratio = 1,
  # zlim = (-0.5, 0.5) .* span(wing_init)
@@ -188,19 +188,19 @@ plot!(
  [ -cumsum(wing_init.surface.spans)[end:-1:1]; 0; cumsum(wing_init.surface.spans) ], 
  [ wing_init.surface.chords[end:-1:2]; wing_init.surface.chords ], 
  lc = :black,
- label = LaTeXString("Initial Wing: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([init.CDi;init.CDv;init.CD]; digits = 6)) \$"),
+ label = LaTeXString("Initial Wing: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([init.CDi;init.CDv;init.CD;init.CL]; digits = 4)) \$"),
 )
 plot!(
  [ -cumsum(wing_opt.surface.spans)[end:-1:1]; 0; cumsum(wing_opt.surface.spans) ], 
  [ wing_opt.surface.chords[end:-1:2]; wing_opt.surface.chords ],
  lc = :cornflowerblue, 
- label = LaTeXString("Optimized Wing: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([opt.CDi;opt.CDv;opt.CD]; digits = 6)) \$"),
+ label = LaTeXString("Optimized Wing: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([opt.CDi;opt.CDv;opt.CD;opt.CL]; digits = 4)) \$"),
 )
 plot!(
  [ -cumsum(wing_exact.surface.spans)[end:-1:1]; 0; cumsum(wing_exact.surface.spans) ], 
  [ wing_exact.surface.chords[end:-1:2]; wing_exact.surface.chords ],
  lc = :green, 
- label =LaTeXString("Inviscid Optimum: \$ (C_{D_i}, C_{D_v}, C_D) = $(round.([exact.CDi;exact.CDv;exact.CD]; digits = 6)) \$"),
+ label = LaTeXString("Inviscid Optimum: \$ (C_{D_i}, C_{D_v}, C_D, C_L) = $(round.([exact.CDi;exact.CDv;exact.CD;exact.CL]; digits = 6)) \$"),
 )
 
 plt_CL = plot(

examples/optimization/wing_definition.jl

@@ -3,17 +3,19 @@ function make_wing(xc, w = 0.25)
  # Planform geometry
  n_vars = length(xc)
  bs = fill(1., n_vars - 1)
+ # bs = AeroFuse.MathTools.sine_spacing(0., 1., n_vars)[end:-1:2]
+ bs = bs / sum(bs)
  wing = Wing(
  foils = fill(naca4(4,4,1,2), n_vars),
  chords = xc, # Design variables
- spans = bs / sum(bs), # Normalizing halfspan to 1
- w_sweep = w, # Quarter-chord
+ spans = bs, # Normalizing halfspan to 1
+ w_sweep = w, # Quarter-chord sweep
  symmetry = true
  )
 
  # Meshing
  wing_mesh = WingMesh(
- wing, fill(2, n_vars - 1), 6, 
+ wing, fill(2, n_vars - 1), 1, 
  span_spacing = Uniform(),
  );