OZ-ENERGY-ANALYSIS.ORG   -   open science for the new millennium

THE STORIES   |   DATA   |   ANALYSIS   |   MODELS   |   LITERATURE   |   DISCUSSIONS

Status: round one; last active here Dec 2010

Simulated Wind Farms from Broome to Cooktown

ABSTRACT and INTRODUCTION

The Story of the Broome to Cooktown Challenge is, in short, that we construct a hypothetical network of wind farms from Broome to Cooktown (or thereabouts), and examine how the system as a whole behaves in relation to synoptic scale weather patterns. We do this using BoM wind speed data for ~40 mostly costal stations that we have selected, and by applying a Wind Power Curve model to the wind-speed data to transform it into simulated wind farms.

In previous data work on BtCC R1 data, we established 40 BoM stations. Upon obtaining and examining the (2009) data we categorised 9 of these as poor, based on the wind speed distribution and reconsideration of station siting (using Google Earth). Of the nine alternate stations that we chose, only five came back with hour level data (remainder synoptic = 3hr-ly).

We now proceed in simulating wind farms, broken into three groups:
- the initial 31 that passed inspection
- the initial 9 that were considered problematic
- the new 5 that may act as replacements

Thus far we are simply considering the first group of 31 in order to get a starting look.

DATA and METHODS

The wind speed data used here is developed in the BtCC Wind Speed Data Page. For now we are proceeding with the Round One wind speed data, as found in: BtCC_R1_BASE_hr_2009.txt.gz [332 KB] - (csv version). As at 3rd Dec, there remains some data work to do in replacing previously identified 'poor' sites with alternatives. For now we simply exclude these problematic sites from consideratio; these are:

 BoM site number:  60141  31209  31210  84143   3003   5007   9178   9741  11003
 index in files:      3.     8.     9.    28.    33.    34.    36.    39.    40.

The method (the Wind Power Curve model) for transforming BoM wind speed data into putative wind farm output is developed on the Reconcile Wind Farm output with BoM wind data page; and, in particular, is formalised in these Matlab scripts (which can be translated into other languages / forms, taking care to handle missing data). Following on from Simulating Wind Farms on the Eyre Peninsula, here we additionally include longitudinal correction for the diurnal bias phase parameter, as follows:

DPpD   = 24;     % Data Points per day 
DYst   =  1;     % Day of Year start
CF_req =  0.35;  % Capacity Factor for simulated wind farms
SC     = [0.1, 30];        % Seasonal correction parameters
DC     = [7, 13.9, 0.265]; %  Diurnal correction parameters
                           %  - the 2nd of these requires a longitude based adjustment
% With a vector of raw wind speed data 'wind', and longitude 'long'
DCp      = 13.9 - 0.0667 * (long - 137.7);
DC(2)    = DCp;
wind_cor = OzEA_wind_speed_corrections( wind,  DPpD, DYst, SC, DC );
simWF	 = OzEA_simulate_wind_farm_output( wind_cor, CF_req );

ANALYSIS and RESULTS

Using data and methods as above, the 31 simulated wind farms (1 MB .txt.gz) are generated, downsampled into daily averages and plotted:

If you click the image you get a PDF where you can zoom in and pan along for a clearer view. The black line shows the day-level average of all 31 simulated wind farms. At least on first appearences, we observe a very significant level of spatial smoothing.

REMARKS

To start, just as a reminder, and apart from all the other caveats that go with a first pass look, this is not an engineering proposal. In particular, unless and until we learn otherwise, there is not going to a be a (large) transmission capacity between the east and the west coasts any time soon. In time the analysis here will break down into looking at more realistic configurations of component parts; for now we continue with an overall level view.

Work proceeds in the comments below.

2 OzEA_ABTCCWFS0002

francis Subject: State by state view of simulated wind farms Date: 2010-12-16 (at 21:01:50)

Here is a state-by-state view of the simulated wind farm data, at the day-to-day level. Each state (WA, SA, TAS, VIC, NSW & QLD) is derived as the average of the simulated wind farms from that state (4,9,4,5,6 & 4) respectively, and these traces are treated equally in generating the average (and so the average is a little different to the one given in the head post). The figure can be viewed in more detail by linking through to the pdf and then zooming in.



With an average output of 0.35 (the imposed capacity factor) it is noteworthy that the the overall daily average sits above 0.2 most of the time. Of course, the electricity system has to match supply and demand ALL the time, and so we will shortly move to a different way of viewing this data (demand remainder profiles).

Note that before putting too much effort into analysing the spatial smoothing here, we will bring in the solar data. As this proceeds the task will turn to an optimisation problem with the aim of choosing configurations of Wind and Solar farms that given the 'best' overall smoothing.

3 OzEA_ABTCCWFS0003

francis Subject: base renewable supply Date: 2010-12-18 (at 00:04:26)

Current thinking is to look at the renewable supply in terms of what we call a demand remainder profile. Here we subtract renewable supply from overall demand to get the 'demand remainder', and then instead of plotting this as a time series (as above), we add up the fraction of the time that the demand remainder is above a given value to get:



Here are equivalent plots for TAS, VIC, NSW and QLD.

The blue curve shows the full historical demand for 2009, and the base supply, intermediate supply and peaking supply requirements are clearly seen. The red curve shows the demand profile after one half of the power overall has come from the simulated wind farms in the state, while the dashed black trace has half the power coming from the entire Broome to Cooktown network. Unsurprisingly, the greater level of spatial smoothing gives a better result.

What is especially interesting here is the way renewable power (at the 50% level) is displacing the traditional "baseload", leaving somewhat elongated intermediate and peak supply phases. Thus, we will speak of base renewable supply when dealing with high renewable penetration.

In the 'traditional' mode, the base load/supply runs constantly, with intermediate supply coming in and out on a daily cycle, and with a requirement for more occasional peaking supply. With a base renewable supply, what remains need not (by working hypothesis) impose hugely different needs on intermediate and peaking infrastructure, it may just alter the temporal behaviour.

For the time being we continue to work on the 'base', and will in time and in turn look at the intermediate and peaking parts of the problem.

9 OzEA_ABTCCWFS0009

francis Subject: Simulated wind farms for 2003 Date: 2011-03-03 (at 01:03:10)

With some differences in the BoM stations from the 2009 work above (see wind data), here are the 30 simulated wind farms (0.3 MB gziped text file) for 2003. As for 2009, a simple day-level plot, as previously, with an overall average (black trace) shows much spatial smoothing:



Here are the values of the overall day-level average.

Now these 2003 simulated wind farms are established, it's back to the solar simulations for a bit, and then on with Fifty percent Renewable Scenarios.

[Show Full Lists]

 

  TOP   |   REFRESH PAGE   |   ANALYSIS

fc - December 2010