Juan Ramón Trapero Arenas

Universidad de Castilla-La Mancha

Forecasting in techno-economic analysis.

Forecasting in techno-economic analysis.

Yesterday, I received by e-mail the publication details of my last collaboration with my chemical engineers colleagues. The article is entitled “Is methanol synthesis from co-gasification of olive pomace and petcoke economically feasible?” and during 50 days the access will be free (until July 27th.).

This work deals with a problem very related to a circular economy application, where residuals of the olive industry can be used to generate methanol. It shows the techno-economic analysis of the process involved. However, there is a part referring to forecasting that I consider interesting and somehow overlooked. Essentially, I’ve read many papers that deals with techno-economic analysis and, in the best cases, a Monte Carlo simulation is run to estimate potential risks of the investment. However, and this is the key part, only few (if any) have tried to forecast important variables that are subject to uncertainty. In my experience, most of the works have assumed those variables to be iid (independent, identically distributed), that in christian words it means that the mean and variance are assumed to be constant and, therefore, no forecasting is needed. Although as you can imagine that assumption is frequently not fulfilled.

In this article, we cope with the methanol price forecasting problem, which is not iid by far. That problem is a very difficult one, because the forecasting horizon is also quite a big one. Here, we used a damped trend exponential smoothing and, not only, we provide a point forecast for the methanol price but, we also include a density forecast (probabilistic forecast) that fed the Monte Carlo simulation and that allows us to estimate the investment risks by means of metrics as the Value at Risk.

I think that, currently, many forecasting models that offer a probabilistic forecast are readily available by many free software and, thus, techno-economic analysis should (must) incorporate forecasts of key variables subject to uncertainty, overall, when past information of those variables are available. In my view, that exercise will help considerably the decision makers to optimize their investment strategies by getting a better idea of the future investment risks.

Leave a Reply

Your email address will not be published. Required fields are marked *