I hope this message finds you well. I am currently working on a research thesis where I need to incorporate stochasticity into the TIMES-CZ model owned by Charles University. I am facing challenges in configuring the stochastic variables effectively. After conducting external analyses, I have identified gas prices, EUA prices, and electricity demand as the most suitable variables.
I have successfully determined the appropriate probability distribution for these variables and made accurate predictions for their 2030 values. However, I need assistance in integrating these results into the TIMES-CZ model. If any of you have experience in this area or can provide a model with stochasticity, your guidance or support would be greatly appreciated. Thank you for your time and consideration. I look forward to any assistance or advice you can offer.
It is not clear what kind of challenges you face. But if this is the first time modelling with the stochastic mode in TIMES, you should, first of all, read carefully the documentation: Stochastic and Tradeoff Analysis
For the VEDA specification part, it is very straightforward and shouldn't have much challenge. A simple example below for the set-up specification of a two-stage stochastic program with 7 SOWs. If the probability distribution is flat, the third line can be dropped out, but this small example also demonstrates how you can define an otherwise flat distribution, but with the middle point having a user-defined value (here 0.3). Specifying the uncertain parameters is likewise quite straightforward in VEDA.
I appreciate your previous guidance and the information you shared regarding the VEDA specification for stochastic programming. I have made progress and prepared a table to address the configuration of the stochastic variables. Kindly find the table below:
I would appreciate your guidance on how to properly set the SW_LAMBDA attribute within the table. Additionally, I would like to confirm if the table accurately represents a two-stage stochastic program with nine scenarios of the world (SOWs). During my external analyses, I have assessed EUA prices and Gas prices separately. For each of them, I have identified three possible future SOWs (high prices, middle prices, and low prices), resulting in a total of nine scenarios with different probabilities. My question is whether the provided table combines these variables effectively or if I should create separate documents for each variable.
Furthermore, I would like to ensure that VEDA-BE correctly links the stochastic document with the relevant variables. Please advise on the appropriate steps to verify this linkage.
Thank you for your continuous support and clarification on these matters.
>I would appreciate your guidance on how to properly set the SW_LAMBDA attribute within the table.
Please refer to the documentation section 2.2 Alternative objective formulations. There you can see how SW_LAMBDA is applied in the formulation. It is the responsibility of the modeler to decide upon an appropriate (non-negative) value for it, when choosing to use such an alternative formulation. In section 4.3.1 User control parameters you can also see that SW_LAMBDA does not have any indexes (e.g. Stage or Sow).
>Additionally, I would like to confirm if the table accurately represents a two-stage stochastic program with nine scenarios of the world (SOWs).
That table (on the left above) does indeed include the necessary control specifications for a two-stage stochastic program with nine scenarios of the world (SOWs). However, the control specifications alone are, of course, not sufficient for any meaningful multi-stage stochastic program. You also need to define the uncertain parameters. But sure, you could make a test run with only those control specifications, and you should see it terminating with normal completion, optimal, and with OBJz equal to deterministic case.
>I have identified three possible future SOWs (high prices, middle prices, and low prices), resulting in a total of nine scenarios with different probabilities. My question is whether the provided table combines these variables effectively or if I should create separate documents for each variable.
That table (on the left above) does include the necessary control specifications for a two-stage stochastic program with nine SOWs. (The table on the right is not valid for such a program.) The modeler is responsible for giving a "meaning" to each of these SOWs, by defining the appropriate uncertain parameters for each SOW, for example, by defining high EUA prices and high Gas prices for some SOW=k.
>I would like to ensure that VEDA-BE correctly links the stochastic document with the relevant variables.
I don't know what you mean by the "stochastic document". The model results will include the SOW index for basically all model variables, and so you can immediately link all results to the SOWs.
17-07-2023, 02:34 PM (This post was last modified: 17-07-2023, 02:35 PM by SimonO.)
Dear Antti-Li,
I hope this email finds you well. I'm writing to ask for your advice once more. Your help in reviewing my two documents on stochasticity in gas prices and EUA prices would be highly appreciated. I had trouble deciding whether to set these scenarios as parametric or classical scenarios despite my best efforts to navigate the material. In addition, I experimented with three versions when studying the documents, where one symbolises the upper bound, another the lower bound, and the third a middle ground situation. If these settings are right, I'm not sure. Unfortunately, VEDA has not yet been successful in processing this data.
Once again, I extend my gratitude for your continued support and guidance. I look forward to hearing from you and eagerly await your response.
Sorry to I interfere again, but I had a look at your Excel files, and they make me conclude that instead of TIMES multi-stage stochastics, you are actually trying to use the VEDA parametric scenarios feature (Scen_Par*), i.e. a kind of sensitivity analysis. While I am only an expert on TIMES and I don't really know about the VEDA parametric scenarios any more than you do, I can point you to a working demo available for that feature, given in this thread: Paremetric Scenarios. The VEDA experts should be able to help you further with that functionality.
Even though your Excel files now appear to make it clear that you are not at all trying to formulate a stochastic program (your formulation does not quite adhere to what is described in the TIMES documentation, and the worked examples given there), just in case you, or any other Forum user, might also want to try using the stochastic feature, but have some doubts about going from the TIMES to the VEDA specification (although it's pretty straightforward), I have now added a small and quite self-contained Stochastic Demo model on the Forum Modeling library.
