rl-institut / offgridders

Models and optimizes capacity & dispatch of electricity supply systems, off-grid or connected to a (weak) central grid
GNU General Public License v3.0
21 stars 9 forks source link

How does one guarantee reliable results? #91

Open JLuij opened 4 years ago

JLuij commented 4 years ago

Hi,

When I researched some ways to verify Offgridders' results, I came to this recent study Martha co-authored: Sustainable Energy Solutions for Remote Areas in the Tropics in which Offgridders' settings were transferred to HOMER in the best possible way and it was shown that the results of both tools were roughly similar. It showed that Offgridders can deliver results that are as good as HOMER's! To me, it then seems that a reliable Offgridders result is dependent on reliable input settings. So, if I want to guarantuee reliable results, is it then simply a matter of being very precise about the input settings and case definitions? Or, is there another way I can guarantee valid results? I'd like to know other's thoughts on this.

smartie2076 commented 4 years ago

Hi @JLuij,

I am actually Martha, so I suppose I can reiterate my perspective on this. In general, Offgridders can only be so precise as the inputs are - if you use ballpoint costs for PV and battery, eg. 700/kWp and 300/kWh, then you will get optimization results that are to be treated as ballpoint results, ie. rough estimates. The better your cost data and technical data represents your specific location, the better your results will be.

As for the pre-feasibility step data is rare, there is a way to meet this bottleneck: Apply the sensitivity analysis extensively. That way you can analyse the co-dependencies of the input parameters and look into which parameters are most vital. The first parameters I would look into are the PV and storage investment costs, electricity price from the grid and the fuel costs of the generator. Analysing this sensitivity analysis will give you some certainty on the range in which the system design you chose is reliable (ie. the best system design considering input uncertainty).