Embedded Files
Post date: Apr 9, 2011 8:26:25 AM
The title may look strange but for sure Ethiopian wind data is unique. First because it is recorded only five times a day (few missing values) second, there is no night time data third long term standard measurements are absent. Therefore if you are designing and simulating an energy system consisting of wind turbines, because of these errors in the wind data your model is invalid. The good news is this can be tackled , thanks to the brilliant method developed by Dr Getachew Bekele, [1]. Here is the detail step by step explanation of the method. The method is even more explained in [2] and [3].
The wind data
Only for few location in Ethiopia have a well recorded data at the national metro-logy services agency (NMSA).Recent data (2000 onwards) for major towns are recorded at 6:00, 9:00, 12:00, 15:00, and 18:00, at a height of 10 meters. At the previous times, the data are at 2m (non standard) and taken for 3-7 times a day. Besides lack of hourly data, the data have a few recordings missing here and there.
The pre requisitesThe basic steps involve using the software HOMER with monthly average data and wind coefficients. HOMER--Hybrid Optimization Model for Electric Renewable is a free software used for optimization of hybrid energy systems. The monthly average wind speed should be calculated from the daily average values. The best way to fill the missing values is to use the average wind speed of the previous and the next hourly data. This is like the data for 9:00 is the average of the 6:00 and the 12:00. This is actually very rough approximation but keep in mind that if the data has too many missing values, it should not be used in the first place. The wind characteristics coefficients required are Weibull K, diurnal pattern strength, peak hour and auto correlation. How do we generate 24hr data?The day time average wind speeds are known for all of the locations. Night time (19:00-05:00) data are not recorded. Hence to compensate them, HOMER is fed with monthly average values of the day time only data. The software then interprets this to be 24hr average monthly data. Weibul K, auto-correlation, diurnal pattern strength and peak hour are fed in addition to the monthly average wind data to generate 24hr data. From this data, daytime (06:00-18:00) data are filtered out to calculate new monthly average values.
Then the measured values are divided by the these values to determine the scaling factor. The parameters are varied until best scaling factor and minimum mean square error is obtained. The scaled down monthly average values are fed into HOMER to generate 24hr data and this is then assumed to compensate the missing night time data.The scaling factor is determined by averaging the scaling factor for each of the 12 months. This alone, however, doesn’t tell how well the data is generated. It is possible that a scaling factor of greater than one is possible for one or some of the months when the measure value is higher than the synthesized value. This increases the average scaling factor and the mean square error. Therefore, simultaneous use of these two parameters leads to accurately synthesized data. [1] Getachew et al. 2009, STUDY INTO THE POTENTIAL AND FEASIBILITY OF A STAND ALONE SOLAR WINDHYBRID ELECTRIC ENERGY SUPPLY SYSTEM, PhD thesis, KTH
[2] Moges A, 2010, MODELING AND ANALYSIS OF WIND HYDROGEN ENERGY SYSTEM IN ETHIOPIA, Msc Thesis, KTH
Page updated
Google Sites
Report abuse