A good example of early use of computer animation is Computational Fluid Dynamics. In the early 1960s, the T-3 Group at the Los Alamos Scientific laboratory (LASL) started developing numerical techniques to solve time-dependent problems in both compressible and incompressible fluid flow, initially in two dimensions but also 3 dimensions when the computer power improved beyond their IBM 7030 (Stretch), the fastest computer available at the time.

Jacob Fromm in his 1963 paper A Method for Computing Nonsteady Incompressible, Viscous Fluid Flows wrote With the advent of the high-speed computer, it has become possible to develop methods for studying theoretically many of the nonsteady incompressible fluid flow problems which previously had been hopelessly complicated for analysis.

Harlow and Fromm in their 1965 paper Computer Experiments in Fluid Dynamics wrote The fundamental behavior of fluids has traditionally been studied in tanks and wind tunnels. The capacities of the modern computer make it possible to do subtler experiments on the computer alone

Harlow, Shannon and Welch in their 1965 paper Liquid Waves by Computer gave examples that could be simulated as the flow of water from a broken dam, the generation of water waves by an explosion, the formation of breakers on a beach, and the splash of a jet of liquid hitting a plate.

Eddie Welch in his Datamation 1966 paper Computer Simulation of Waves describes a new technique that has been developed at Los Alamos which is for the full time-dependent solution of the flow of an incompressible fluid with a free surface.

The initial method (Particle-In-Cell) was used in the paper Rise Through the Atmosphere of a Hot Bubble (1962).

Some effort was made in performing experiments to check the accuracy of the method. A 1966 paper by Mader (Theoretical and Experimental Two-Dimensional Interactions of Shocks with Density Discontinuities) is an example.

A new method called Marker-And-Cell (MAC) was used by many of the later LASL examples:

  • The area under consideration was divided into a large number of rectangular areas that could be part of the boundary, part of the fluid, part of an obstacle in the area or be the boundary between the liquid and its free surface.
  • The characteristics of the fluid and the initial position is defined
  • The changes in the fluid are calculated for a small time step for each of the cells
  • The contents of each cell are adjusted to the new values and the computation repeated
  • The computer animated film is generated by a sequence of frames at equal intervals visualising the state of the set of cells

A full description is given in THE MAC METHOD: A Computing Technique for Solving Viscous, Incompressible, Transient Fluid-Flow Problems Involving Free Surfaces (1965).

An early paper in the Science magazine entitled Liquid Waves by Computer (1965) reached a large audience and resulted in several inquiries per day by mail. Every effort was made to handle the fan mail (Fan Mail for T-3 (1965)).

Eddie Welch gave a talk at UAIDE65, Moving Picture Computer Output(1965) which impressed a different audience about the possibilities of using films computer animation in simulation.

Early on, the Group demonstrated that it provided insights over a variety of problems in the flow of viscous incompressible fluids. The computer animations can be viewed as controlled experiments to guide the development of analysis and the planning of experimental programs.

Two later papers are the Datamation paper Computer Simulation of Waves (1966) and The Splash of a Liquid Drop (1967). This used a modified version of the original MAC method.

Los Alamos continued to develop the MAC method . An example is YAQUI: An Arbitrary Lagrangian-Eularian Computer Program for Fluid Flow at All Speeds (1973).

A major problem over the years was making adjustments to get the maximum performance out of each new computer system. The CDC 7600, for example, had an extra fast store that was not that large so there was a lot of effort in ensuring that computation took place in the fast store. Only a small part of the mesh could reside in the fast memory so that shuttling the next part to be updated was a problem particularly due to the new calculation involving two of the previous three mesh rows or columns. The solution was to have multiple versions of the code rather than change the position of the variables. Also, if two variables were used in different parts of the computation, they used a single address in the store by equivalencing the address used by the two variables.

The earlier Particle-In-Cell (PIC) code was used by AWRE Aldermaston on their IBM 7030/SC4020 system, see Diagrams Drawn by Computer (1965).

ACL mounted PIC on their Atlas computer, see Numerical Solutions of Hypersonic Near-Wake Flow by the Particle-in Cell Method (1968) .