Monte Carlo Simulation

[mbr1818]

Dear VEDA Forum,

I am Mike a 3rd year phd student learning TIMES to use with EPA's models when they become available.

I have two questions. The first question is whether it is possible to use VEDA in such a way that approximates a monte carlo simulation?

What I would like to learn to do is solve a model and record the objective function in different scenarios. Specifically I would like to run the model with varying taxes on different emissions. For instance I would like to implement a CO2 tax of $1, $5, $10, $15, $20, etc. and essential build a CO2 abatement supply curve from a model.

Second question is whether there is an easier way to derive abatement supply curves for specific pollutants than running the model with varying levels of a cap or tax and recording the objective function value?

Thank you!

Mike

[Antti-L]

As far as I know, VEDA does not provide tools for approximating a Monte Carlo simulation.

As I am sure you recall, TIMES does have stochastic and tradeoff analysis features, which include  a simple sensitivity analysis facility, and can be used under VEDA. Specifically, one could easily run a model with varying costs imposed on different emissions, by using the S_DAM_COST parameter for defining a number of different emission price trajectories, for instance, a CO2 tax of $1, $5, $10, $15, $20, etc. By running the model under the sensitivity analysis facility, you would get the model results for all these prices from a single run, indexed by the SOW index given to each CO2 tax scenario. Similarly, you could define a number of different emission caps, by using the S_UC_RHSxxx parameters, run the model, and obtain the results for each cap. The sensitivity analysis feature solves the model in a loop for each of the sensitivity analysis scenarios defined (SOW=1,…,N).  With the GAMS BRATIO option you can also control whether to use a "hot start" between successive solves or not.

However, as you already must be aware of this feature, it is seems that what you need is indeed something more sophisticated.

[mbr1818]

Thanks so much for the great info Antti-L! This is great I will do some research on this method and try out! Specifically I think the loop option for multiple caps could work well.