Introduction

Although I have decided to limit as much as possible posts about papers and reports published from the 1984 European Wind Energy conference and onwards, I will make an exception for this report.

As I wrote earlier, I was lucky enough to receive a bunch of documents from Tom Zambrano, who worked for AeroVironment. It includes a gem, that is a report by Peter Lissaman on the wind turbine park siting tool developed for the DOE in the late 80’s, see below:

• (Lissaman, Foster, Hibbs 1987): Operational Model for the Design of Optimum Wind Farm Arrays

The aim of the project was to develop a tool which allowed for “rapid estimates of energy production for a general array in a nonuniform wind field on complex terrain“.

This report consists only in the phase 1 of the project, it seems that the final report is located at the DOE archives, I am awaiting for good news from them. At that stage the software was not finished yet, but included beta versions of orographic- and turbulence models, as well as interactive layout work, see below:

What strikes at first, is the statement that there were no tools available for performing production estimates with a personal computer, and indeed: WAsP from Risø was released in 1987, and I could not find examples of it being used before 1988 (see for instance this Master Thesis).

The AeroVironment program is called AVFARM, its basic modules are showed in the diagram below:

I could not find very detailed information about the WAsP graphical interface in 1987, apart of the Master Thesis from Botha (1988), but AVFARM must have looked prettier I guess :P, see below:

Please note that the flow model had not yet been implemented, at the time the report was written. It is quite a key component of the software, and it is worth focusing on the approach taken here by AeroVironment.

They envisaged to use the model from Taylor, MS3DJH, a model which is similar with the model used in WAsP, see mor details in the European Wind Atlas. The reports states that they would seek consultancy from Mr. Taylor in the phase II of the project for the implementating the model. The report makes reference to the Askervein validation, but does not mention any publication from Risø, for instance the Danish Wind Atlas or the EWEC October 1986 paper on the WAsP orography model by Troen and De Baas, but that is not surprising since the AV report is dated from January 1987, so very much at the same time. I am still awaiting from the DOE archives, whether the final report is available, it may help understand how much was realised.

Highlights

Four features of AVFARM, integrated to the software, sounded very interesting in that beta-version (but computing speed was a clear limitation):

1) An integrated wake model, combined with the orography model. WAsP, at least from its 1993 user manual, required the user to run a seperate program (PARK). The limitation for AVFARM was the computational time (15 seconds for one direction and one speed, without terrain effect calculation), but many code improvements are mentioned that may have succeeded in reducing it.

2) The consideration of turbulence: the report mentions first the effect of turbulence on the turbine power curve, see below. The analysis is based on what I understand to be BEM results from the AV code called “PROP” (see a later DOE version here).

Surprisingly, the report does not refer to previous works on turbulence and blades, for instance as early as 1978 by Jensen and Larsen in Amsterdam, or this 1984 report from Risø on unsteady loads, but expresses concerns about the lack of knowledge and consideration for turbulence, from the developers side. See below:

The great idea in this report is then the need to assess the site conditions, also from the turbine structual perspective. This comes quite early, at least as concern from siting engineers like AeroVironment advising large commercial projects. Many turbines failed, in the early days of industrical wind energy, due to fatigue loads primarily. The first 61400-1 standards came out in 1999, and WAsP engineering (site assessment tool) the following year, if I remember well. It is unclear to me what design standards were used then, in California in the late 80’s. Overall a good suggestion from AeroVironment, it makes this report very well comparable with the work carried out today by consultants, definitively.

3) Calculations of uncertainties: the whole Section 3.5 describes in great detail a way to assess uncertainty on the yield estimate. It includes the main uncertainty contributors (measurements, long-term correction, etc).

4) Layout optimisation: it was planned during the phase 2, to develop a layout optimisation algorithm, and set of tools to help to user place the turbines.

Thoughts

I don’t know what the final version AVFARM ended up looking like. If it included the orography model, the wake calculation as well as the other features presented in this report, this would have made it very suitable for performing production estimate, in a similar fashion to WindPRO for instance. With one limitation though: it did not consider roughness changes, possibly because that wasnt the main driver behind the flow pattern in California, where most of the AV projects were located (it seems).

And it really tells something about the difference in code development compared with Europe, and especially Denmark at the time. As highlighed previously, the Danish Wind Power program caught up quite late with the American and Swedish ones, but very quickly set up three very successful programs at the National Laboratory (Risø) in the early and mid 80’s:

1. A 3-bladed demo turbine program in two phases: first refurbishing the Gedser turbine and then building the two NIBE machines. Those included both atmospheric, power and structural measurements.
2. At the same time, develop computer codes that could model the unsteady behavior of the turbine and thereby for a more robust design, as well optimisation of the blades and controllers.
3. Building up a wind resource assessment toolkit, focusing on the European wind turbine marker where both roughness changes and orography (simple terrain) were the norm.

All this work was carried out in public research labs (DK, NL, SE), while it seems that in the USA the work was mostly carried out by a mix of engineering consultants and aeronautics/space national labs. As I remember NREL was founded in the early 90’s only. The approach developed by AeroVironment regarding siting was very project based it seems, while the work at Risø was part of a more generic framework (the European Wind Atlas). Let us not forget that support for wind power in California stopped in 1986, after that the marker mainly became European (Germany took off at this stage), such a hectic commerical market was not the most favourable environment for developing tools for the engineering community outside of the research labs.

Also, it is worth noting that WAsP would have been of no practical use in the case of the Californian sites in complex terrain, with bimodal wind distributions (see examples here) and all sorts of features WAsP is not designed to handle. The approach from AV was clearly the best one for those sites (I would still not use a flow model there today, without having a solid measurement dataset..).

So.. this 1987 report from AV clearly calls for a set of tools which was difficult to make at the time, because of the computational power limitations. But it really speaks to me more than any Risø report (sorry guys :P), I think this was really great consultancy work. They must have enjoyed it. Too bad that the US market, both from a regulatory and from a technical perspective was a bit too immature at that time.

I am really curious to see if I can get hold of the final version, that’s for sure !

Comments and suggestions are welcome,

Rémi