Skip to content

Space shuttle ​

  • 6 state
  • 2 control
  • 1 variable
  • Mayer cost
  • free final time
  • 6 state box
  • 2 control box
  • 1 variable box
  • 9 boundary
  • OptimalControl valid
  • OptimalControl, scalarized valid
  • JuMP valid

The Space Shuttle reentry problem is a classical benchmark in aerospace optimal control. It models the atmospheric descent of the space shuttle from high altitude to the Terminal Area Energy Management (TAEM) interface. The system includes six state variables: altitude , longitude , latitude , velocity , flight path angle , and azimuth . The control variables are the angle of attack and bank angle . The goal is to maximise the final latitude (crossrange) at TAEM while satisfying aerodynamic, gravitational, and operational constraints [Betts 2010; Bulirsch 1971; Dickmanns 1972].

Mathematical formulation ​

The problem can be stated as

The final time is free but bounded.

Qualitative behaviour ​

  • The optimal trajectory balances lift and drag to control heating and deceleration while extending crossrange.

  • The bank angle determines heading changes and crossrange capability.

  • The solution typically combines steep reentry to dissipate energy and crossrange manoeuvres to reach the target latitude.

Characteristics ​

  • Nonlinear dynamics with strongly nonlinear aerodynamic coefficients.

  • Widely used as a benchmark in optimal control and trajectory optimisation.

References ​

  • Betts, J.T. (2010). Practical Methods for Optimal Control and Estimation Using Nonlinear Programming. SIAM. Provides a comprehensive discussion of trajectory optimisation for aerospace vehicles, including space shuttle reentry problems. It details numerical methods applicable to nonlinear, constrained, free-final-time problems.

  • Bulirsch, R. (1971). Numerical solution of optimal control problems with state constraints by direct methods. Numerische Mathematik. Introduces early direct methods for optimal control problems with constraints, which are foundational for solving shuttle reentry trajectory problems with bounded states and controls.

  • Dickmanns, E.D. (1972). Numerical solution methods for nonlinear optimal control problems with state constraints. Automatica. Focuses on numerical techniques for nonlinear optimal control with state constraints, directly relevant to handling the shuttle’s aerodynamic and flight path restrictions.

  • Ascher, U.M., Mattheij, R.M.M., & Russell, R.D. (1988). Numerical Solution of Boundary Value Problems for Ordinary Differential Equations. Provides practical algorithms for solving boundary value problems, which underpin the solution of two-point boundary value problems like the shuttle reentry trajectory with fixed initial and terminal states.

  • Betts, J.T. (2001). Survey of numerical methods for trajectory optimization. Journal of Guidance, Control, and Dynamics, 24(4), 643–653. Reviews trajectory optimisation methods including direct transcription and collocation, which are commonly applied to the space shuttle reentry benchmark.

Parameters ​

ParameterDefault valueUnitMeaning
t00sinitial time
w203000.0lbweight of the shuttle
g₀32.174ft/s²gravitational acceleration at sea level
ρ₀0.002378slug/ft³air density at sea level
hᵣ23800ftscale height of the atmosphere
Rₑ20902900ftradius of the Earth
μ1.4076539e16ft³/s²gravitational parameter of the Earth
S2690ft²reference area
a₀-0.20704–lift coefficient, constant term: c_L = a₀ + a₁ α (α in degrees)
a₁0.0292441/deglift coefficient, linear term
b₀0.07854–drag coefficient, constant term: c_D = b₀ + b₁ α + b₂ α² (α in degrees)
b₁-0.00615921/degdrag coefficient, linear term
b₂0.0006214081/deg²drag coefficient, quadratic term
tf_l1500.0slower bound of the final time
tf_u2500.0supper bound of the final time
h_t02.610⁵ ftinitial altitude
ϕ_t00.0radinitial longitude
θ_t00.0radinitial latitude
v_t02.5610⁴ ft/sinitial velocity
γ_t0-0.017453292519943295radinitial flight path angle
ψ_t01.5707963267948966radinitial azimuth
α_s0.0radangle of attack of the initial guess
β_s0.0radbank angle of the initial guess
h_tf0.810⁵ ftfinal altitude, at the Terminal Area Energy Management interface
v_tf0.2510⁴ ft/sfinal velocity
γ_tf-0.08726646259971647radfinal flight path angle
h_l010⁵ ftlower bound of the altitude
ϕ_l-6.283185307179586radlower bound of the longitude
ϕ_u6.283185307179586radupper bound of the longitude
θ_l-1.5533430342749532radlower bound of the latitude
θ_u1.5533430342749532radupper bound of the latitude
v_l0.000110⁴ ft/slower bound of the velocity
γ_l-1.5533430342749532radlower bound of the flight path angle
γ_u1.5533430342749532radupper bound of the flight path angle
ψ_l-6.283185307179586radlower bound of the azimuth
ψ_u6.283185307179586radupper bound of the azimuth
α_l-1.5707963267948966radlower bound of the angle of attack
α_u1.5707963267948966radupper bound of the angle of attack
β_l-1.5533430342749532radlower bound of the bank angle
β_u0.017453292519943295radupper bound of the bank angle

Every modeling takes other values with the keyword parameters, merged with these: space_shuttle(backend; parameters=(t0=…,)).

Reference objective ​

The best objective among the solves that converged, in each modeling, on the default grid of the modeling: the modelings are independent, each has its own reference. See how the references are computed.

ModelingSchemeGridReference objectiveSolvesDistinct solutionsConfigurations agree
JuMP:trapeze500-0.595875013481yes, within 8.0e-11
OptimalControl:trapeze500-0.5958750134181yes, within 6.2e-10
OptimalControl, scalarized:trapeze500-0.595875013331yes, within 5.6e-16