.BondLib.Examples.Biosphere

Information

1. Amazonas rain forest biome,
2. upper and lower Savannah biomes,
3. salt-water lake biome,
4. mangrove marsh biome,
5. Southwestern desert biome, and
6. agricultural biome.

Eight people lived in this structure for two years, growing their own food and regenerating their own air. The purpose of the exercise was to test the feasibility of space colonies with current technology.

The model makes use of five large data tables containing observations of:

1. temperature,
2. humidity,
3. cloud cover,
4. solar radiation, and
5. wind velocity,

stored hourly for an entire year for the location of Oracle, AZ, i.e., the location of the Biospere 2 project.

These data items are stored on an external binary file (Bio_tables.mat) rather than in memory. This is essential, as Dymola isn't geared to dealing with large data tables in memory. Dymola essentially converts all matrices to sets of individual scalar variables in the process of compilation. This is necessary, as the most flexible automatic causality assignment depends on this feature. Yet, this is highly inefficient, when dealing with large data tables containing many thousands of data points. We must thus hide these tables from Dymola, and can do so by placing them in external data files.

The model of the hemodynamics had originally been developed by Francisco Luttmann in his Ph.D. dissertation [1]. The original Fortran code was converted to bond graphs and an early version of Dymola by Àngela Nebot during one of her postdoctoral stays in Tucson. It was converted to graphical form and a modern version of Dymola/Modelica by François Cellier. The overall model has been published in [2,3].

Experimentation:

This model uses hours as time units. If you wish to only experiment with this model, you can simulate it over 4 days by setting the final time to 96 hours. However, if you wish to reproduce the results published in the 1999 paper [1], you need to simulate the model for a full year, i.e., during 8600 hours.

Contrary to the SolarHouse example that computes internally using seconds as computing time units, accomplishing the display units of hours by means of time scaling, this model uses hours as units of time throughout.

This is a typical American model. Instead of consistently using SIunits for its computations, it uses units at will, converting between incompatible units by use of numerical conversion factors.

Unfortunately, this model is rather large. It produces far too much I/O, when all variables are being stored. Dymola by default stores each and every variable. Thus, if you wish to simulate over an extended period of time, you should turn off the automatic storage of all variables, and selectively store only those variables that you wish to look at. This will speed up the simulation dramatically. At least, you should turn off the storage of the auxiliary variables.

References:

1. Luttmann, F. (1990), A Dynamic Thermal Model of a Self-sustaining Closed Environment Life Support System, Ph.D. dissertation, Department of Nuclear and Energy Engineering, University of Arizona.
2. Nebot, A., F.E. Cellier, and F. Mugica (1999), "Simulation of Heat and Humidity Budget of Biosphere 2 without its Air Conditioning," Ecological Engineering, 13, pp. 333-356.
3. Cellier, F.E. and A. Nebot (2005), "The Modelica Bond Graph Library," Proc. 4^th Modelica Conference, Hamburg, Germany.

Contents