Useful Derivatives for Numerical Propulsion System Simulation Models

The gradient of propulsion cycle closure solutions with respect to design inputs can be used to support system design and analysis capabilities using various numerical methods, such as gradient-based optimization. Towards this goal, two gradient methods have been implemented within the Numerical Propulsion System Simulation (NPSS) interpreted language: direct finite difference (DFD) and adjoints using finite difference (AFD) gradients of cycle outputs with respect to design inputs. Due to the speed of the NPSS interpreted language and reliability of its cycle solver, the DFD and AFD methods can be used to provide effective gradients to any existing NPSS model. We compare the effectiveness of DFD and AFD methods when applied to the numerical optimization of several design parameters for the N+3 engine.