Optimization under wind direction uncertainty in FLORIS version 2 ignores wind direction heterogeneity

[pjireland]

I've been using FLORIS version 2 due to the inability to specify heterogeneous wind directions in later versions. Looking at the code, it seems that when optimizing under wind direction uncertainty, the baseline wind direction is taken as the first wind direction (self.floris.farm.wind_map.input_direction[0]), ignoring any specified wind direction heterogeneity.

In floris.tools.floris_interface.py::get_turbine_power, I see the following:

            wd_orig = self.floris.farm.wind_map.input_direction[0]

            yaw_angles = self.get_yaw_angles()

            for i_wd, delta_wd in enumerate(unc_pmfs["wd_unc"]):
                self.reinitialize_flow_field(wind_direction=wd_orig + delta_wd)

                for i_yaw, delta_yaw in enumerate(unc_pmfs["yaw_unc"]):
                    self.calculate_wake(
                        yaw_angles=list(np.array(yaw_angles) - delta_wd + delta_yaw),
                        no_wake=no_wake,
                    )
                    mean_farm_power = mean_farm_power + unc_pmfs["wd_unc_pmf"][
                        i_wd
                    ] * unc_pmfs["yaw_unc_pmf"][i_yaw] * np.array(
                        [turbine.power for turbine in self.floris.farm.turbines]
                    )

            # reinitialize with original values
            self.reinitialize_flow_field(wind_direction=wd_orig)

A possible solution is this following:

            wd_orig = self.floris.farm.wind_map.input_direction

            yaw_angles = self.get_yaw_angles()

            for i_wd, delta_wd in enumerate(unc_pmfs["wd_unc"]):
                self.reinitialize_flow_field(wind_direction=[wd + delta_wd for wd in wd_orig])

                for i_yaw, delta_yaw in enumerate(unc_pmfs["yaw_unc"]):
                    self.calculate_wake(
                        yaw_angles=list(np.array(yaw_angles) - delta_wd + delta_yaw),
                        no_wake=no_wake,
                    )
                    mean_farm_power = mean_farm_power + unc_pmfs["wd_unc_pmf"][
                        i_wd
                    ] * unc_pmfs["yaw_unc_pmf"][i_yaw] * np.array(
                        [turbine.power for turbine in self.floris.farm.turbines]
                    )

            # reinitialize with original values
            self.reinitialize_flow_field(wind_direction=wd_orig)

Is this something the team is aware of? Would there be a benefit to submitting a PR to resolve the issue, or are changes no longer being made to version 2?

[misi9170]

Hi @pjireland , sorry on the long wait time for a response here. Let me look into this a bit more and try to get back to you next week. Generally speaking, we can continue to provide updates for v2, if there is a bug and the solution is clear, which may be the case here. I'll keep you posted.

Misha

[misi9170]

Hi @pjireland , I've now had a chance to take a look through your suggestion here and don't see any obvious issues. We do still support bug fixes to v2, and this does appear to me to be a "bug" (since wind direction heterogeneity, although nominally supported, is not correctly handled in get_turbine_power()). I'd be happy to go through the process of getting your code merged and releasing a new patch release of v2 including the bug fix.

We did something similar not too long ago, where #1008 provided a bug fix to v2, after which we released v2.5.2. If you'd like to open a pull request with your bug fix pointing at the v2 branch, we can go through the process of getting it merged and then release a v2.5.3 that includes the fix.

[pjireland]

Hi @misi9170, thanks for the response, and sorry for the delay on my side. I've submitted PR #1115 against the v2 branch to address this issue. Please let me know if you have any questions for me about this. Thanks!

[misi9170]

No problem, thanks for submitting the PR! I hope to be able to review it in the next couple of days.

[misi9170]

Thanks @pjireland --- the PR looks good to me. I've added a comment to the PR that includes the change in outputs to the example as you suggested. If you are happy with that, I'll go ahead and merge the PR and create a patch release for FLORIS v2 (i.e. v2.5.3).

Misha

[misi9170]

This issue has now been corrected in v2.5.3.