chemicals_wrapper

chemicals_wrapper(phase, f, **kwargs)

Wrapper function for calling models in the chemicals package

Parameters:

• phase (dict) – The OpenPNM Species object for which this model should calculate values. This object should ideally have all the necessary chemical properties in its params attribute, although this is optional as discussed below in relation to kwargs.

• f (function) – The handle of the function to apply, such as chemicals.viscosity.Letsou_Stiel.

• kwargs – By default this function will use openpnm.models.phase.default_argmap to determine which names on phase correspond to each argument required by f. For instance, default_argmap['Tc'] = 'param.critical_temperature' so any functions that need Tc will receive phase['param.critical_temperature']. Some models only require the normal thermodynamic parameters, like T, Tc, and Pc, so in many cases no additional arguments are needed beyond whats in the default argument map. However, other models require values that are themselves calcualted by another model, like Cvm, or perhaps a list of constants like a0, a1, etc. It is necessary to provide these as keyword arguments, and they will be included in default_argmap as new entries or overwriting existing ones. So mu='pore.blah' will pass phase['pore.blah'] to the mu argument of f. Any arguments ending with an s are assumed to refer to the individual properties of pure components in a mixture. So mus means fetch the viscosity of each component as a list, like [1.3e-5, 1.9e-5]. This function will trim the trailing s off any argument name before checking the default_argmap so that the normal pure component values can be looked up.

Notes

This wrapper works with both pure and mixture phases, but the mixture models are very slow due to the way chemicals vectorizes code. For pure species it allows the computation of values at many different conditions in a vectorized way. The means that the conditions in each pore, such as temperature, pressure, etc can be passed and iterpreted as a list of conditions. For mixture models, however, the vectorization is done over the compositions, at a fixed condition. This means that we must do a pure-python for-loop (ie. slow) for each individual pore. As such, we have re-implemented several of the most useful mixing models offered by chemicals in OpenPNM, and include unit tests to ensure both implementations agree.