Optimisation

Openwind has the ability to optimise turbine positions with respect to the amount of energy produced by the entire workspace. It does this using a fairly simple stochastic optimisation process that involves randomly perturbing the positions of the turbines and then assessing whether the changes are beneficial—and rejecting those that are not.

Figure 1: Optimiser Options

Figure 1 shows the Optimiser Options, which are explained below:

• Maximum attempts to place each turbine (before forcing) - This is quite simply the number of attempts using only random numbers to finding a legal layout before changing tack.

• Maximum attempts to force a legal layout - If random numbers fail to create a legal layout—as often happens in very constrained sites—Openwind attempts to pack turbines into the available space at 60-degree intervals.

• Number of successive fruitless iterations before stopping - If this many iterations pass without any improvement in net energy yield, the optimiser will stop.

• Maximum number of iterations - This is simply another stopping criterion.

• Attempt to dynamically add a turbine - If a user checks this box, the optimiser will attempt to add turbines to each site that is currently accepting new turbines (see Site Layers ) as long as the array losses for each site are below the specified threshold and the capacity factor is above the specified threshold and any set cap has not been met.

• Only while incremental cost of energy is falling - this option must be used in conjunction with the optimiser for cost of energy (see Optimiser for the Cost of Energy (OCOE) ) but can be used to optimise the capacity of a project just beyond the point at which adding a turbine no longer improves the financial viability of the project. At this point, economies of scale have been exhausted and although each previous turbine contributed to a drop in cost of energy, the most recent one did not and therefore no more can be added.

• Auto-save layouts every X iterations as Y (deprecated) - This acts as a safety in case the software crashes or in case the computer turns itself off or loses power for any reason. This functionality is now redundant with the advent of automatic backups and has been marked for removal in a future release.

• Turn on large array optimiser (Enterprise and Basic versions only) - This is a modification aimed at addressing the weaknesses of the optimiser for very large and/or constrained layouts. It scans the site for legal positions using the grid resolution specified below.

• Attempt to auto-place turbines when layout not legal (Enterprise and Basic versions only) - this option scans the site areas in a grid (resolution set below) and then orders the legal grid locations by either

  • Windiest - this orders the legal grid locations from highest mean wind speed to lowest. Turbines are then placed by working down this list, respecting turbine separation distances.

  • Least environmental impact* - this orders the legal grid locations starting with the locations that are furthest from an environmental sensor with a noise limit. Turbines are then placed in order such that they respect minimum separation distances

• Perturbation – this button activates the

• Delete turbines it was not possible to place - allows the user to over-size the turbine layout to find out the maximum number which can be placed using the gridded locations

• The above options require gridding. Grid resolution in meters is - this is the grid resolution used by the options immediately above. As with most gridded applications, smaller resolutions are better but require more processing and memory.

*We have found this method to be as much as 30% better than human turbine placement whereas with previous methods, humans could easily beat Openwind by 20-30% as measured by the total number of turbines that can be placed in noise-restricted site.

Figure 1: Optimiser Peturbation Options

This dialog is a little arcane and only applies to optimisation for cost of energy optimisation and even then only when we turn on the large array optimiser. Within the perturbation algorithm, we want to limit the size of the allowed jumps within the one dimensional array of legal turbine positions that have been sorted by mean wind speed. The assumption here is that the cost of energy might go down along with the mean wind speed but that this is only true up to a point and that if we limit the decrease in mean wind speed then we will be more likely to find a good move for each turbine and the optimisation will be more efficient.

• Minimum decrease - this is the allowed fractional decrease in the mean wind speed for large jumps regardless of how well the turbine is doing.

• Maximum decrease – this is the maximum allowed fractional decrease in mean wind speed for any large jump. The size of the allowed decrease in mean wind speed decreases with increasing turbine performance (lower cost of energy or higher IRR).

• Don’t jump if better than average – when checked, turbines whose performance (low cost of energy or high IRR) is better than average do not make large jumps but instead concentrate on fine tuning around their current location.

• Max tries 1 – the maximum number of attempts to find a legal large jump

• Max tries 2 – the maximum number of attempts to find any kind of move. It should be larger than “max tries 1” as when we allow large jumps, the number of attempts to find a small jump is limited by the difference between max tries 1 and max tries 2.