Introducing modelling-to-generate alternatives (MGA) approach into VEDA-TIMES

[Xin Wang]

In a recently published journal paper "Modelling to generate alternatives: A technique to explore uncertainty in energy-environment-economy models", the MGA approach is introduced into TIMES model by set a slack for the optimal system cost. Did anyone have experiences in doing so? It seems that we have to modify the source GAMS codes related to the objective functions. Glad to hear any suggestions on how to conduct this research. 
Thank you all~

Steps listed in the journal paper
1. The model is solved in standard formulation and a least cost energy system transition pathway obtained.
2. The total system cost of this pathway, scaled up by a small amount or slack (usually >1%), is entered into the model as a new constraint. In principle, the scope of possible formulations for the new objective function is large and does not necessarily have to be related to the maximization of difference across the model solutions. It could, for instance, maximise the amount of primary energy from wind or minimise the utilisation of certain end-use technologies, with both energy systems being only marginally more expensive than the optimal run. As our focus in this study is finding energy systems that are as diverse as possible and yet still nearly cost optimal, here we use an objective function formulation that searches for a set of transition pathways that are very nearly least cost but also maximally different from one another in terms of the fuel mix of their cumulative primary energy consumption
3. A new objective function is formulated with the specific aim of exploring the near optimal region defined by the constraint in step 2. This reformulation of the model is also subject to all constraints from the standard formulation in step 1.

[Antti-L]

I would suggest to contact the corresponding author of the recent paper, James Price at UCL. Indeed, the authors do mention in the paper that they have built their own implementation of MGA into the GAMS source code of TIMES using a mixed integer formulation. If you ask politely and it does not require much effort for them, they might well be willing to share their implementation with you, or even with the ETSAP community at large.

There are many different algorithmic formulations of MGA.  One was implemented into the ETSAP MARKAL code about 15 years ago by Dan Loughlin and Gary Goldstein, see e.g. the following link: Dan Loughlin.pdf

The MARKAL implementation was used for several studies by Dan Loughlin at US EPA. It was also suggested for TIMES at some point, but as the approach has several challenges and issues (briefly mentioned in the link above), it was deemed to have a very low priority for ETSAP, and has not been implemented into the common TIMES code. But perhaps there might now be some renewed interest in it.

[Xin Wang]

I would suggest to contact the corresponding author of the recent paper, James Price at UCL. Indeed, the authors do mention in the paper that they have built their own implementation of MGA into the GAMS source code of TIMES using a mixed integer formulation. If you ask politely and it does not require much effort for them, they might well be willing to share their implementation with you, or even with the ETSAP community at large.

There are many different algorithmic formulations of MGA.  One was implemented into the ETSAP MARKAL code about 15 years ago by Dan Loughlin and Gary Goldstein, see e.g. the following link: Dan Loughlin.pdf

The MARKAL implementation was used for several studies by Dan Loughlin at US EPA. It was also suggested for TIMES at some point, but as the approach has several challenges and issues (briefly mentioned in the link above), it was deemed to have a very low priority for ETSAP, and has not been implemented into the common TIMES code. But perhaps there might now be some renewed interest in it.

Thank you, Antti-L!
I will try to get more details about introducing MGA into VEDA-TIMES from the corresponding authors. And the document you share is also quite helpful for me to get a clear understanding of MGA. Have a nice day!