In optimal control models applied to farm level optimization you can choose the state variable and then select the control variable that will change the state variable. for example, if soil nitrogen is a state variable, the control variable could be the amount of urea you apply to soil, or the organic matter you apply. You could take a look at some of the papers. Eg;

1.          Babu, S. C., and R. Hassan. 1995. International Migration and Environmental Degradation - A Case Study of Mozambican Refugees and Forest Resources in Malawi. Journal of Environmental Management 43: 233-247.

2.          Babu, S. C., A. Hallam, and B. Rajasekaran.  1995.  Agroforestry Systems, Soil Fertility and Household Food Security - A Model of Component Interactions and its Implications for Multi-Disciplinary Research.  Agricultural Economics 13: 125-135.

Dear Alary

You can read these papers.

May also see the attached paper

Then the issue of access to the control variables need to be considered. 

Dear Veronique,

I'm not an expert of bio-economic model, but I recently worked on GAMEDE, a model developed in La Reunion you might certainly know, and did a bit of optimization based on Linear Programming.

From my experience, bio-economic models are most of the time constrained on an activity-based (time for milking, harvesting, etc...), and to see if the model is correct, it is useful to reach the whole-farm scale to see if the aggregation of this constraints allows you to define properly the farming system activity.

For instance, controlling the total labor, or the global gross margin, or the milk productivity (at herd scale) will help you to see respectively if constraint on labor, price/cost, and zootechnical data, are well constrained in your optimization model. 

If inconsistency are detected, then you have to track-back into the constraints of the model to understand where the model has found an "open window" to "go crazy" .

Naively, I would also recommend to have a specfic look at socio-economic constraints, cause most of the time bio-physical constraints are well-known and very accurately mathematically defined, while time allocation, cost management, are still uncertain...

Not sure i'm helping here, but for sure the topic of knowing if aggregating constraints at activity scale defines properly the constraints of your farming system is an interesting topic!

Good luck with your on-going research.

David.