Context :

Aims : Despite extensive observations over the last decades , the central questions regarding the power source of the large IR luminosity of Ultra Luminous Infra Red Galaxies ( ULIRGs ) , and their evolution , are still not fully answered .
In this paper we will focus on massive star formation as a central engine and present an evolutionary model for these dust-enshrouded star formation regions .

Methods : An evolutionary model was created using existing star formation and radiative transfer codes ( STARBURST99 , RADMC and RADICAL ) as building blocks .
The results of the simulations are compared to data from two IRAS catalogs .

Results : From the simulations it is found that the dust surrounding the starburst region is made up from two components .
There is a low optical depth ( \tau = 0.1 , which corresponds to 0.1 % of the total dust mass ) , hot ( T \sim 400K ) non-grey component close to the starburst ( scale size 10pc ) and a large scale , colder grey component ( 100pc , 75K ) with a much larger column ( \tau = 10 ) .
The simulations also show that starburst galaxies can be powered by massive star formation .
The parameters for this star forming region are difficult to determine , since the IR continuum luminosity is only sensitive to the total UV input .
Therefore , there is a degeneracy between the total starburst mass and the initial mass function ( IMF ) slope .
A less massive star formation with a shallower IMF will produce the same amount of OB stars and therefore the same amount of irradiating UV flux .
Assuming the stars are formed according to a Salpeter IMF ( \Psi ( M ) \propto M ^ { -2.35 } ) , the star formation region should produce 10 ^ { 9 } M _ { \sun } of stars ( either in one instantaneous burst , or in a continuous process ) in order to produce enough IR radiation .

Conclusions : Our models confirm that massive star formation is a valid power source for ULIRGs .
In order to remove degeneracies and further determine the parameters of the physical environment also IR spectral features and molecular emissions need to be included .