The table below shows how the most popular transient solvers can be characterized – implicit or explicit, standard or adaptive time step methods.

During the solution, an explicit solver computes new temperature values entirely as a function of previous temperature values, which are already known from 1 or more time steps back. Implicit solvers compute new temperatures from both previous (known) and current (unknown) temperatures. This requires simultaneous solutions for all new temperatures, which can be done with iterative or direct methods, just like a steady state solution. For a given time step size, implicit methods are more accurate, and for a given accuracy requirement, implicit methods can take larger time steps. Explicit methods perform fewer computations per time step.

As with steady state solvers, the sparse matrix methods (TRSPM and ATSSPM) tend to be good for large plate-like models, or any problem that can be described as truly sparse. TRQMR and ATSQMR may prove best for large 3D-solid type models.

FWDBKL, ATSFBK, and the SNTSM family of solvers may prove best for radiation dominated problems. These solvers do not linearize radiation, handling it instead as a fourth order polynomial.

ATS type solvers should not be used with thermostats or other closed-loop thermal control simulations. These solvers can back up and re-try a time step when the error estimate is too large, which can cause problems with thermostat switching states and duty cycle calculations.

DRLXCA and ARLXCA remain important for many transient solvers, and should generally be set to smaller values than used for steady state solutions.