The Iteration options menu provides control over the iteration processes towards a solution for a steady-sate or quasi-steady state (transient) operating point:

•Accuracy

This is the accuracy required for the solution of the set of virtual non-linear equations to be solved in GSP (it is the maximum tolerated value for each error variable). This accuracy can be considered to be roughly equal to (or of same order) the other calculated performance parameters (default 0.001).

•Relative state perturbations for jacobian calculations

The relative perturbation of the state variables used to determine partial derivatives for the Jacobian matrix (default 0.001). It is a means to control the relative range across which the model is locally linearized. For stable iteration calculations the absolute value of this variable must be equal or smaller than the accuracy.

•State correction step factor

This is a factor to increase or decrease the stepsize GSP uses to iterate towards the solution using the inverse Jacobian of the Newton-Raphson method (default 1).

It is a means to avoid too large steps which cause the solution to become unstable due to non-linearities.

•Relative state variable correction limit

This parameter limits the maximum relative correction of a state variable during iteration (default 0.05). It is a means to avoid too large steps which cause the solution to become unstable due to non-linearities.

•Convergence test vector

This factor represents the factor GSP uses to determine whether there is conversion progress (default 0.9). The default is 0.9, meaning that GSP requires at least 10% less of the total square of the error variables per iteration step before the inverse Jacobian is updated.

•Max. nr. of Jacobian calculations

This is the maximum number of times the Jacobian is recalculated per operating point calculation, i.e. per time step for transients (default 50). It is a means to stop the iteration process if there is little chance of finding a solution. When the default of 50 is encountered it is often useless to extend te limit. Instead, improving the model or control system consistensy, trying to take smaller steps towards/between new succeeding operating points or adaption of the other iteration parameters is often more successful.

•Max. nr. of Boyden updates

This is the maximum number of times the Jacobian is updated (using the Broyden method) per operating point calculation, i.e. per time step for transients (default 0). It is a means to fine-tune optimum iteration speed by taking the best trade-off between recalculating the Jacobian completely which consumes much computation time and accepting an approximate Jacobian using Broyden with little computation time.

•Iteration step limit factor

This factor limits the total number of iteration steps by the number of model state variables and this factor (default 50). It is a means to stop the iteration process if there is little chance of finding a solution (see Max. nr. of Jacobian calculations).