An Empirical Model of the Abacus
Generally speaking, your abstract is very good. The emphasis on the human aspect of abacus use is quite appropriate, and the idea of documenting this through the EMPE is potentially exciting. I like the fact that your abstract refers to a variety of different ways in which your construal can serve an illustrative and educational purpose - this fits in well with the idea that a modelling study enables many kinds of application rather than one specific one. Given that the submission is centrally concerned with EM, it is surprising that you make no reference in the abstract to key concepts such as observable, dependency and agency - it will be important to redress this in your final submission. A key point is that what you are aspiring to is not simply implementing an abacus in EDEN - it is vital that you give full attention to what it is that is being observed and interactively experienced as an adjunct to the thought-processes of the human operator. There is an interesting parallel here with the study of data structures (cf. the heapsorting construal) - perhaps there is a sense in which the abacus can be viewed as a data structure for arithmetic.
At the moment, what you describe is rather sketchy. A major achievement for your final submission would be making the practical implications of "adding 9 and 4 on the abacus" in different ways quite clear. Though you don't explicitly refer to this, it would be good to consider ways in which you can enhance the visualisation of the abacus so that e.g. when a particular computation was being contemplated the most significant beads that would come to the mind of the expert user are highlighted. I note that you don't mention Jona Gjini's previous effort to make an abacus construal (see WEB-EM-8), but I think it would be good to make use of this if you can understand and extend it. Commenting on the issues of re-use of construals is another EM theme that you can bring in in this way. Gjini's construal also highlights the way in which the movement of beads can be most elegantly and appropriately expressed using dependency (i.e. moving one bead may simultaneously move several).
Bear in mind that fully implementing the agenda set out in your abstract may be rather too ambitious for the WEB-EM submission. You will have done enough to earn an excellent mark if you build a construal that makes it clear how you might address your full agenda in principle. Even if you don't manage to implement both kinds of abacus, you might consider writing about how your comparative study of their use might be conducted, for instance.
With reference to your title, the term "Empirical Model" is not one that I have ever been quite comfortable with. It is unfortunate that we have to use capitalisation to distinguish EM from what is otherwise understood by 'empirical modelling', and I would prefer not to extend the convention in this way. For that reason, I favour terms like "EM construal", or even "EM model", clumsy though they are.