Missing chemical information

[Jkaen]

Hi guys,

I have posted a thread on the aspen forum (

http://www.eng-tips.com/viewthread.cfm?qid=148041&page=1)

but as it seems rarely used I thought I would post here in a slightly reworded form.

Main problem is I am trying to model a reaction with various chemical streams, but dont have sufficient information on the chemicals (specifically 2-iodoethanol in this case). While I can find very basic information on a number of sites, I cannot find anywhere that offers information such as Gibbs energys.

Is anybody able to point me in the direction of a source of this more detailed information?

Thanks
Stephen

 

[sshep]

JKaen

My approach with physical properties is to make the search to see what is available (as you have done) and then build on that to the extent possible. My special favorites for general properties of non-databank pure components and mixtures are:

http://www.nist.gov/srd/online.htm

(Pure Components)

http://www.dechema.de/The_DECHEMA-lang-en.html

(Mixtures- this will cost a small amount)

Your search has hopefully got you boiling point, standard density and some other basic data. If you can't find this data readily available in the common literature and internet sources then you must generate the needed parameters for Aspen yourself. In your case structural estimation should provide the easiest route to do this because "I" is an allowable functional group in many methods.

Raw data on Gibbs Free Energy is an unusual find for any component, but would be useful only as a delta-Gibbs check in any case for Aspen-plus. DGForm uses an unusual reference condition in Aspen-Plus and the Gibbs free energy at any conditions is calculated from other major and minor properties as described in Ref Manual Vol 2, Physical Property Methods and Models, chapter 1. If you can reference a specific reaction using other databank components then the DGForm can be calculated from that- this is your best bet.

You can get pretty far with a minimal amount of data especially with respect to pure component properties. The extent to which this can be done is well illistrated in "Properties of Gases and Liquids" by Reid, Prausnitz, and Poling (or Sherwood in my day). The only real issue is to recognize when your ability to estimate with existing data is inadequate for the model purpose and when further experimental data may be needed. For example you can probably get to a fine stochiometric reaction model which adequately predicts heat of reaction and hense adequate for the energy balances needed to design heat transfer equipment etc; but that the same model is not adequate for designing a reactor.

This is what I would do if I were you:
1.) Use an activity coefficient OPSet as this is easier for building up properties from scratch. Hopefully this reaction is not in an aqueous process because those systems can be pretty complicated to build up physical properties from basic data.
2.) Make a structural estimation in Aspen-Plus using the known structure, boiling point, and any other data you have. You may need to compare a few estimation methods. Note: After estimation you are much better off to later paste your parameters back into Aspen and then turn off your estimation to insure that the integrity of the parameters is always under your control.
3.) Check your scalar properties [MW, DHForm, etc] and plot temp dependent thermo and transport properties [density, Antoines Equation, HVap (activity coef model), CPIG] to see if they are well behaved.
4.) For missing transport properties such as vapor viscosity which may not be easily estimated, point Aspen to a method (such as DIPPR vapor viscosity) which is simple to work with and then plug in parameters to either hold this constant or keep it well behaved and understandable to you.
5.) Run your model to see if you have warnings for missing properties and that the results are reasonable. Make parameter adjustments if required. For example, run your reaction as an RStoic to the side and adjust DHForm for 2-iodoethanol to get the appropriate heat of reaction for the reaction you are using.
6.) Assess the model against your intended purpose to see if it is capable of meeting you needs.

Myself (and presummably some others in this forum) have thousands of hours modeling systems where we had to generate property parameters out of some pretty thin starting data. Perhaps we could be of more help if you share some more about the process you are modeling and for what purpose.

best wishes,
sshep