J. Meteor. Res.  2015, Vol. 28 Issue (2): 344-357   PDF    
http://dx.doi.org/10.1007/s13351-015-0193-z
The Chinese Meteorological Society
0

Article Information

HUANG Linhong, SONG Lili, LI Gang, XIN Yu. 2015.
Variation Characteristics of Regional Synchronous Wind in Hami, Xinjiang of Northwest China
J. Meteor. Res., 28(2): 344-357
http://dx.doi.org/10.1007/s13351-015-0193-z

Article History

Received August 21, 2014;
in final form November 15, 2014
Variation Characteristics of Regional Synchronous Wind in Hami, Xinjiang of Northwest China
HUANG Linhong1, SONG Lili2 , LI Gang1, XIN Yu3    
1 School of Atmospheric Sciences, Nanjing University of Information Science & Technology, Nanjing 210044;
2 Public Meteorological Service Center, China Meteorological Administration, Beijing 100081;
3 Institute of Desert Meteorology, China Meteorological Administration, Urumqi 830002
ABSTRACT:From several towers in Hami, Xinjiang of Northwest China, built by the national wind power resources professional observation network, we selected three towers with synchronous 10-min average wind speed data for one year (May 2011-April 2012) under strict quality control. The towers are located where large-scale wind power development is projected. We analyzed the frequency and variation of extreme wind speed at low wind condition (LWC), rated wind condition (RWC), and cut-out wind condition (CWC), which may significantly impact the electric power grid configuration in large-scale wind power development. The correlation between duration and frequency of LWC/RWC/CWC is obtained. Major findings are: 1) The frequency of CWC is the lowest among all conditions, its synchronous rate at all three towers tends to be zero, and the frequency of LWC is always greater than that of RWC. 2) Among the three towers, the synchronous rate of RWC steadily increases with height, and LWC differs little between different levels. The synchronous rate of LWC concentrates in winter, while that of RWC mainly occurs in spring and summer. Diurnal variation of LWC/RWC during the entire year is significantly different. 3) During the study year, the longest durations of synchronous LWC and RWC among the three towers are up to 640 and 700 min, respectively. The duration and frequency of LWC/RWC can be quantitatively well described by a logarithmic function. Consequently, the synchronous rates of LWC and RWC over any duration in the region can be easily calculated by using the fitting function equation from observed data. These results are of value to the planning of large-scale wind power transmission and grid dispatching in this area.
Keywordsregional synchronous wind     low wind     rated wind     variation characteristics     Hami     Xinjiang    
1. Introduction

Under increasing environmental pressure, as aclean renewable energy, wind power has become one ofthe main solutions to realize low-carbon energy strategyin China. In 2010, China leapt to first place in theworld in terms of installed wind power capacity. However, many problems remain in the effective utilizationof wind power resources(Li et al., 2007), including ther and om, fluctuating, and intermittent features of windpower that have daily effects on grid stability and operationalsafety(Chi et al., 2007; Sun et al., 2007;Saifur et al., 2011). The capacity of a power systemto accept wind power depends on a number of factors, such as wind power output characteristics, power systemload characteristics, power source structure and frequency, and peak regulation capacity(Lei et al., 2002; GAQSIQ, 2011). Obviously, the negative effectsof the first two factors are mainly manifested as thefluctuation and instability of wind. Therefore, windpower forecasting has been introduced in the powerdispatching system to predict wind fluctuations and accordingly arrange power dispatching, with the purposeof harnessing wind power to the maximum possible extent within the existing grid systems and powersource structure conditions(Liu et al., 2007; Fan et al., 2008; Gao et al., 2010). The synchronous fluctuationof wind within the regional grid, especially when thesynchronous wind speed is lower than the cut-in lowwind condition(LWC), or at the cut-out wind condition(CWC)or rated wind condition(RWC), presentsgreat challenges to grid dispatching(Li et al., 2007;Zhou et al., 2009; Han, 2010).

Therefore, researchers in China and other countriesbegan studying the characteristics of regionalsynchronous wind. Leahy and Mcleogh(2013)analyzedthe synchronization variations by hour of historicalwind data from 14 meteorological stations in Irel and since 1980. They used the local wind speed profileindex to determine the anemometric height of about 10m above the ground to the height of 80 m, the usualwind turbine hub level, with the conventional windturbine cut-in wind speed as a synchronous low windincident and the rated wind speed as a synchronoushigh wind speed incident. They found that, throughoutthe country, synchronous low wind speed incidentsrarely occurred, but the frequency of synchronous lowwind speed incidents was much higher than that ofsynchronous high wind speed incidents. The authorsalso analyzed the effect of this feature on wind powergeneration, and indicated that even under favorablewind conditions, a synchronous low wind speed incidentwith a return period of 10 yr may reduce theannual power generation of a wind farm by 5%. Inhis study of variation rate vs time using wind resourcedata at 66 ground meteorological stations in the UK, Sinden(2007)found that in over 90% of the l and in thecountry, on average, synchronous low wind speed incidentsoccurred for 1 h every year. A study by Huang etal.(2014)with 5-yr meteorological assimilation datashowed that the high frequency fluctuation componentof wind power output can be substantially reduced byconnecting 5–10 wind farms in 10 states in the centralpart of the United States. Lu et al.(2013)reached asimilar conclusion in their study of coastal wind farmsin China. Han(2010)investigated the relation of windpower output variation rate with the wind power connectionrate and the system frequency regulation byanalyzing the measured wind power output data at5-min intervals in the Zhangjiakou area, and provedthat synchronous variation of regional power generationhad a direct effect on the connection rate of windpower.

The regional synchronous wind variation characteristicsare important to large-scale wind power transmissionplanning and project construction in North, Northwest, and Northeast China, because the Chinesewind power resources are concentrated in theseregions(Xue et al., 2001; Song et al., 2012). However, power dem and in China is mainly distributed inthe central and eastern regions; therefore, wind powerrequires large-scale transmission(National Energy Administration, 2012). The ±800-kV southern Hami toZhengzhou UHVDC(ultra high voltage direct current)power transmission line, launched in June 2012, was the first UHVDC power transmission project for“wind power, photovoltaic power, and thermal power”in China. In this type of projects, thermal power at acertain proportion is required to balance the fluctuationof wind power, and clearly, the synchronous characteristicsof the wind speed in the project area willbe related to the configuration ratio of wind power tothermal power. A high ratio of wind power to thermalpower will be more sensitive to the synchronization ofregional wind speed, but a wind power configurationratio as high as possible is an ideal design for suchtypes of projects. So far, few studies have been conductedto reveal the synchronous variation characteristicsof winds, especially the extreme winds, over targetedlarge-scale wind power development areas. Thepresent study tends to fill this gap.

In this study, synchronous wind measurements ofseveral wind towers in the Hami area of Xinjiang AutonomousRegion, China were analyzed to determinethe synchronous variation characteristics of extremecondition wind speed associated with wind power generationin the area represented by these towers, withthe purpose of providing valuable data for wind powergrid connection dispatching in the region and the constructionof wind power transmission projects, and exploringa technical route and analysis method that canbe applied to practical projects.2. Data sources and processing2.1 Data sources and observation setup

The basic data for this study are measured observationsfrom seven wind towers built by the nationalwind power resource professional observationnetwork in the Hami area. The time synchronizationof wind speed data collected and recorded at all towersis an important prerequisite for this study. The nationalwind power resources professional observationnetwork has set up unified data acquisition centers inall provinces, autonomous regions, and municipalities, and specified clock synchronization calibration for themeasuring instruments at all towers(based on the networkclock)concurrently with four data acquisitionsevery day(Song et al., 2014), meeting the synchronousdata observation needs of this study.

Wind towers are in regular triangle mast steelstructures with bracing wires at the height of either70 or 100 m. At the 70-m tower, the wind speedobservation levels are specified as 10, 30, 50, and 70m above the ground, while at the 100-m tower, additionalwind speed and direction sensors are located atthe 100-m level. Anemoscopes are model CAWS1000-GSW wind measuring system made by the HuayunCompany of China, and the performance indicators ofthe wind speed and direction sensors are as shown inTable 1.

Table 1. Performance indicators of the wind anemoscope
2.2 Sample selection and processing2.2.1 Sample selection

The height of most large wind turbine hubs isabout 70 m; therefore, the observations of averagewind speed every 10 min at 70-m level at all the towersfor 12 consecutive months(1 yr)are considered as theanalysis target.

For field observation, severe weather and environmentalfactors often lead to abnormal or missingdata; thus, the raw observed data are checked for quality and processed according to relevant specifications(GAQSIQ, 2002; CMA, 2007). To ensure the reliabilityof the analysis results, relatively stringent requirementsare set for the effective integrity rate ofthe original data: the time period of the synchronousobserved data of several wind towers should be 1 yr(12 consecutive months), the total valid data integrityrate of each tower should be over 95%, and this rate foreach month shall be over 85%. The valid data integrityrate is calculated by using the following equation:

where P is the integrity rate of valid data, Nall is thenumber of data that ought to be measured, Nmiss isthe missing data, and Ninvalid indicates invalid data.The data quantity in Eq.(1)is the number of averagewind speed records for every 10 min over the entireyear.

In the 3-yr dataset from the seven wind towers inHami from June 2009 to May 2012, only the observationsfrom three towers from May 2011 to April 2012met the above-mentioned valid data integrity rate conditions.The valid data integrity rate of the three windtowers for each month is shown in Table 2.

Table 2. Percentage of the valid data integrity rate from the three wind towers
2.2.2 Sample data processing

Table 2 indicates that the effective integrity rateof the selected sample data is over 99.8%. As for thesmall number of missed or aberrant data points, themethod recommended in the literature(CMA, 2007)isused for interpolation and correction in present study.2.2.3 Environment of wind towers

The three selected towers at Yiwanquan, Yemaquan, and Naomaohu are located within thescope of the planned wind fields for the integratedwind, photovoltaic, and thermal power transmission project in Hami. The towers at Yiwanquan, Yemaquan, and Naomaohu are respectively referred toas Tower 1, Tower 2, and Tower 3(Fig. 1). Thestraight-line distance between the three towers is 153–277 km(Fig. 1). The area surrounding the threetowers is roughly flat Gobi l and form, with a hillapproximately 5 km to the northeast of Tower 1(Fig. 2).

Fig. 1. Schematic of the locations of three wind towers and topography of the study domain.
Fig. 2. Photos of the underlying surface and surrounding environments of the three wind towers.
3. Parameter characteristics of wind energy resource

To underst and the wind resource conditions in thearea of the towers, the main wind energy parametersof the three towers from May 2011 to April 2012 areanalyzed and calculated, as shown in Table 3. Theresults show that there are rich wind energy resources in the area represented by the three towers, and at the70-m level, close to the wind turbine hub, the annualaverage wind speed is 6.31–7.56 m s−1, with an annualaverage wind power density of 441–569 W m−2, indicatingthat this area is good for wind power development.

Table 3. Wind energy resource parameters of the three wind towers at 70-m height

The monthly variations of average wind speed and wind power density at the 70-m level of the wind towers, as presented in Fig. 3, show slightly different variationfeatures at the three towers. At Tower 1, thewind speed and wind power density change significantlyby month, with maximum values of 9.7 m s−1 and 1390 W m−2, respectively, occurring in May;whereas minimum values of 2.9 m s−1 and 89 W m−2occur in December. At Tower 2, the magnitude of thevariation is small and exhibits a double-peak distribution.High values occur in February and Septemberwhen the wind speed and wind power are respectively8.2 m s−1 and 611 W m−2, and 8.8 m s−1 and 570 Wm−2; low values occur in January and June, and thewind speed and wind power are respectively 6.2 m s−1 and 365 W m−2, and 6.7 m s−1 and 266 W m−2. AtTower 3, the variation exhibits a single-peak distribution, with the maximum wind speed and wind powervalues in June of 8.9 m s−1 and 676 W m−2, and theminimum values in December of 2.9 m s−1 and 40 Wm−2, respectively.

Fig. 3. Variations of averaged monthly wind speed and wind power density for the three towers at 70 m. 1V st and s for the wind speed of Tower 1, 1P for the wind power of Tower 1, and so on.

Figure 4 shows azimuth distributions of the winddirection and wind energy density for the three towersat the 70-m level. The prevailing wind directionof Towers 1 and 3 is similar to the main directionof wind energy density distribution, mainly in theWNW–NNW sector, with wind direction frequency respectivelyat 35% and 58%, and wind energy densityfrequency respectively at 62% and 54%. The prevailingwind direction at Tower 2 is NE–ENE at a frequencyof 39%, and its wind energy density is mainlydistributed in the NNE–E sector at a frequency of64%. From the locations and l and forms of all the towersin Fig. 1, we can see that the prevailing winddirection and wind energy density distribution characteristicsof the three towers are closely related tothe local wind climate characteristics and surroundingl and forms, and as Tower 2 is located in a valley ofSW–NE strike, its prevailing wind direction in winter and summer is SW–NE, which differs from the winddirection distribution characteristics of Towers 1 and 3.

Fig. 4. Rose diagrams of wind direction and wind energy density of the three towers at 70 m. Wind direction is depicted for(a)Tower 1, (b)Tower 2, and (c)Tower 3, while wind energy density is depicted for(d)Tower 1, (e)Tower 2, and (f)Tower 3. Numbers in the six panels above represent the percentage of wind frequency or wind energy density.
4. Wind power parameter analysis for regional synchronous wind4.1 Definition of regional synchronous wind and determination of wind power parameter indicators

Regional synchronous wind is a natural phenomenonwhere the wind speeds at all locations(orgrid points)within the pre-designated area vary insynchronization in a specified range. This study isbased on the average wind speed data for every 10min, and the characteristic indicators of synchronouswind speed variation range are defined by the powergeneration output characteristics of wind turbine generatorsunder different conditions. Based on the effectof an increase or decrease of regional synchronous windpower on grid operation dispatching, we primarily discussthree extreme power generation wind speed conditionsthat may produce a major impact on regionalgrid operation dispatching: an LWC that produces nowind power output, an RWC that produces maximum(rated)wind power output, and a further increase inwind speed that leads to a cut-out condition(CWC).

The wind speed characteristic indicators are explainedwith the output electric power curve of a windturbine with a capacity of 1.5 MW, as shown in Fig. 5. Figure 5 shows the design power curve of a 1.5-MW wind turbine and the relation between output electricalpower(the wind turbine output) and cut-in windspeed, rated wind speed, and cut-out wind speed. Accordingto the International Electrotechnical Commissiondefinition(IEC, 2005), cut-in wind speed is thelowest wind speed at the hub height at which the windturbine starts to produce power in the case of steadywind without turbulence, rated wind speed is the minimumwind speed at the hub height at which a windturbine’s rated power is achieved in the case of steadywind without turbulence, and cut-out wind speedis the highest mean wind speed at the hub height atwhich the wind turbine system is designed to producepower in the case of steady wind without turbulence.All these three wind speeds are taken as the 10-minaverage wind speed at the wind turbine hub level. The definitions of the three extreme power generation windspeed conditions discussed in this study are summarizedas follows:

Fig. 5 Power output variation with different wind speed of a 1.5-MW wind turbine.

1)LWC: the wind speed interval wherein the windturbine generator produces no electrical power output, i.e., below the cut-in wind speed;

2)RWC: the wind speed interval wherein the windturbine generator can output electric power in fullload, i.e., equal to or greater than the rated wind speed and below the cut-out wind speed of the wind turbine;

3)CWC: equal to or greater than the cut-out windspeed of the wind turbine.

Based on the technical indicators of common largewind turbines in China and other countries today, thecut-in, rated, and cut-out wind speeds are normally 3, 12, and 25 m s−1, respectively, and will be assumed asthe three extreme characteristic wind speed indicatorsin this study(Table 4).

Table 4. Characteristic wind speed indicators in the three extreme operating conditions
4.2 Analysis of the wind power parameters in regional synchronous wind

If the 10-min synchronous average wind speedmeasurements are used as the basic dataset at all observationlevels of the three wind towers in the Hamiarea from May 2011 to April 2012, each observationlevel of each tower has 52704 wind speed sample points(the study year is a leap year, with 366 days in the entireyear, as there are 29 days in February 2012). Thefrequencies of the three conditions of LWC, RWC, and CWC are considered as the basic wind power parametersthat we use to analyze the regional synchronouswind, which is expressed as follows:

where Pi is the occurrence frequency of LWC, RWC, and CWC, respectively expressed as Plow, Prate, and Pout. Pi calculated from the data of a single tower isthe occurrence frequency of different wind speed conditionsat the tower; however, for the calculation resultsof multi-tower synchronous data, Pi is referred to asthe “synchronous rate.” Ni is the number of samples ofthe three wind conditions at a single tower or reachingthe indicator wind speed values in synchronization atseveral towers, respectively expressed as Nlow, Nrate, and Nout;N is the total number of 10-min wind speedsamples in the studied year, and there are 52704 valuesboth for a single tower and as synchronous datafor the three towers.5. Statistical analysis of various wind conditions5.1 Differences of all wind conditions at a single tower and in multi-tower synchronization

The occurrence frequency of the synchronizationof the three towers, synchronization of two towers indifferent combinations, and the three conditions ofLWC, RWC, and CWC of each single tower at levelsof 10, 30, 50, and 70 m are calculated based onEq.(1), as shown in Table 5. The results show that, at the four levels of the towers, whether as a singletower or two or three towers in synchronization, thefrequency of CWC is the lowest, occurs less than 1%for each single tower, and tends to be zero with two or three towers in synchronization. The occurrencefrequency of LWC is higher than that of RWC in allcases, similar to the results obtained by Leahy and Mcleogh(2013)for Britain. The frequency of LWC and RWC at all single towers at 70 m is 16.6%–31.6% and 13.4%–15.6%, respectively, while the occurrencefrequencies of LWC and RWC with three towers insynchronization substantially drop to 4.3% and 2.0%.The synchronous rate of two towers in different combinationsis higher than that of three towers in all cases, indicating that a lower synchronous rate correspondswith a larger regional area. There are also differencesin the occurrence frequency of LWC, RWC, and CWCin two different combinations of two towers, indicatingdifferent wind synchronization characteristics in differentareas, and such localized variability may be mainlyattributed to l and form.

Table 5. Frequency and synchronous rate of the low, rated, and cut-out wind speeds for the three towers
5.2 Variations in synchronous rate under LWC and RWC with height

Table 5 shows that, under LWC, the synchronousrate of the three towers is quite similar at low and highlevels(4.2%–4.3%). Under RWC, the synchronous rateof the three towers significantly increases with height, with the synchronous rate at 10-m level at 0.7% asminimum and at 70-m level at 2.0% as the maximum.

This variability of wind speed synchronous ratewith height under LWC and RWC is mainly relatedto differences in wind speed with height. Figure 6further shows the wind speed profiles of each of thethree towers under LWC(Fig. 6a) and RWC(Fig. 6b)conditions, and it can be seen that when the threetowers reach RWC in synchronization, the wind speedat all three towers increases with height(Fig. 6b), and agrees with the wind profile characteristics of the approximateneutral stratification produced under fairlyhigh wind speed conditions. However, when the threetowers synchronously reach LWC wind speed, the windspeed profiles of Towers 1 and 3 show negative shears, with wind speed decreasing with the increase of height.Although the wind speed profile of Tower 2 remainsa positive shear, the shear index is obviously smallerthan that under RWC, which is fairly similar to thewind profile shape of stable atmospheric stratificationunder LWC.

Fig. 6 Wind profiles of(a)LWC and (b)RWC.
5.3 Analysis of the variation characteristics of synchronous rate under LWC and RWC with time

Many studies(Fu and Li, 2008; Ju et al., 2006;Zhou et al., 2005)have recognized that the annual and daily power load variation is significant, includingthe timing of peak and valley loads. Therefore, analysis of the monthly and diurnal variations of windspeed synchronous rate of the three towers under thethree extreme conditions in a year in the Hami areahas value for grid operation departments to developdispatching plans, and to design the construction ofwind and thermal power projects in the region.

Figure 7 shows the frequencies of synchronousLWC and RWC in each month of the study year forthe three towers at the 70-m level in Hami. The frequenciesof LWC and RWC exhibit a roughly oppositepattern over the months: in November, December, and January of the subsequent year, the frequency of LWCis higher than that in other months, with 666 times asthe maximum in January(for total duration of 6660min); and in May, June, and August, the frequency ofLWC is quite low, and occurs only 7–26 times. The frequencyof RWC in various months is opposite to thatof LWC. August has the highest frequency of RWC, at310 times, and the frequency from November to Februaryin the next year is almost 0.

Fig. 7 Monthly variations of the wind frequency under different wind speed conditions for an entire year at 70 m.

By comparing Fig. 7 with the variation curve ofthe average wind speed by months at all towers in Fig. 3, it can be seen that months with frequent LWC aremainly those with low average wind speed in the area.August, with the highest frequency of RWC, is notthe month with the maximum average wind speed inthe area, while April and May, with the highest averagewind speed, respectively correspond to the second and third peaks of RWC frequency. In general, synchronousLWC normally occurs in the low wind seasonin the area, while RWC roughly corresponds to the seasonwith relatively high wind speed in the area.

Based on the variation of these two conditionsover the months, further analysis is performed to examinethe average diurnal variation over the entireyear, and in winter and summer(Fig. 8). The averagediurnal variation over the entire year of LWC(Fig. 8a)is basically high in the morning and low in the afternoon, with the highest frequency before noon and thelowest frequency at around 16:00 pm(local time). Thediurnal variation of LWC significantly differsed in winter and summer. It is similar to the annual average inwinter, only with slightly smaller amplitude, whereasin summer, it is almost zero, except for several timesaround 10:00 am. The average diurnal variation over the entire year of RWC(Fig. 8b)shows a doublepeakdistribution, with peaks at 5:00 am and 19:00pm, and valleys after noon and before dawn. In winter and summer, the diurnal variation in RWC alsosignificantly differs, and shows weak double peaks insummer, with the time periods of peaks and valleyssimilar to the yearly average, but in winter, there isalmost no RWC.

Fig. 8 Diurnal variations of the wind frequency of synchronous(a)LWC and (b)RWC in summer, winter, and the whole year at 70 m.
5.4 Characteristics of synchronization duration under LWC and RWC

Figure 9 shows an example three-day period, where the three towers synchronously met LWC and RWC wind speed at the 70-m level. Each vertical linein the figure represents the “signal” of a low wind(Fig. 9a) and rated wind(Fig. 9b)with a duration of 10min. The signals of LWC and RWC are either intermittentor continuous, with the minimum time spacingof 10 min(i.e., a record sample), and the time periodof the maximum continuous signals is the longest durationof LWC or RWC.

Fig. 9 Examples of synchronous(a)LWC and (b)RWC of the three towers at 70 m during a 3-day period in August 2012.

The duration of the extreme condition wind is oneof the important phenomena considered by grid dispatching, since the longer duration of extreme windequals greater difficulty or cost. Table 6 shows thefirst five samples of the longest duration of the low and rated wind with three towers in synchronizationat the 70-m level. It shows that in the study year, the longest duration of low wind is 640 min, and occurredfrom 5:30 am to 16:00 pm on January 7, 2012.The longest duration of rated wind is 700 min, whichoccurred from 22:00 pm on July 15 to 09:30 am on July 16, 2011.

Table 6. The first five samples of the longest duration of the low and rated wind with three towers in synchronizationat the 70-m level

Figure 10 shows the accumulated times of variousdurations of the low and rated wind with three towersin synchronization at the 70-m level. The frequenciesof LWC and RWC are significantly reduced with an increasein duration, and are mainly concentrated within1 h. However, there are still 113 LWC and 47 RWCevents that last longer than 60 min, and 10 LWC and 6 RWC events that last longer than 300 min.

Fig. 10 Distribution characteristics of the wind frequency in duration of low and rated winds of three towers in synchronization at the 70-m level.
6. Statistical fitting of duration characteristics of low wind and rated wind

According to the literature(Han, 2010; Lu et al., 2013), extreme condition incidents of synchronous lowwind without electric power output or synchronousrated wind at maximum electric power output can occurin most wind farm areas, and the frequency ofoccurrence differs in different areas. In wind powertransmission project planning or grid dispatching, it isnecessary to know the duration of various extreme conditions

under pre-determined occurrence conditions inthe areas of concern in advance, or the maximumduration of extreme conditions permissible in actualprojects or under grid operation dispatching conditions.Equation(1)is used to calculate the accumulatedfrequencies of low wind and rated wind exceeding differentgrades for different durations(Fig. 11). Theremay be a certain statistical relation between the duration and occurrence frequencies, and it can be found infurther fitting analysis that the relation between theduration of low wind and rated wind in the area and their occurrence frequencies can be better expressedin the form of a logarithmic function as follows:

Fig. 11 Distribution characteristics and the fitted curve of the wind cumulative frequency of synchronous(a)LWC and (b)RWC in duration at 70 m.

In Eq.(3), x is the duration of low wind or ratedwind(in min), y the occurrence frequency of low orrated wind exceeding a given duration, and a and bare fitting parameters. Equation(3)is used to fit thesynchronous data of the three towers in Hami, and parametersa and b are obtained(see Table 7).

Table 7. Fitting parameters and goodness of fit parameters of the function

The coefficient of the determination R2 is usedto check the goodness of fit of the regression equation(Huang, 2013)as follows:

where Yj is the measured value, the fitting value, Ythe corresponding average value, Σ(YjY)2 the totaldeflection, Σ(Y)2 the regression deflection, and Σ(Yj)2 the residual deflection(or residual error).

The fitting parameters listed in Table 7 show thatthe fitting residual variance of the equation for lowwind and rated wind is very small, the goodness offit check coefficient of determination is respectively0.9909 and 0.9823(the regression passes the significancetest with a confidence of 0.001), and the logarithmicfunction has good fitting effect(Fig. 11).

In the calculation with the regression equation fittedwith Eq.(3), the occurrence frequency of low windduration in the area represented by the three towersin Hami exceeding 60, 180, 300, 480, and 600 min(1, 3, 5, 8, and 10 h)is 3.26%, 1.59%, 0.82%, 0.29%, and 0.07%, respectively; and the occurrence frequency ofthe duration of rated wind exceeding 60, 180, 300, 480, and 600 min(1, 3, 5, 8, and 10 h)is 1.26%, 0.54%, 0.33%, 0.09%, and 0.05%, respectively. These calculatedresults are of value for the planning of large-scalewind power transmission project and grid dispatchingin the Hami area.7. Summary and discussionsThe variability of regional synchronous wind inthe Hami area of Xinjiang Autonomous Region inNorthwest China is investigated in this paper based on10-min synchronous anemometric data of three windtowers in the context of large-scale wind power developmentin the region. This study has utilized designpower curves and parameters for current mainstreamlarge wind turbines, and has been carried out by assumingLWC with no electric power output from thewind farm and RWC with wind turbines in full loadoperation, as well as considering extreme conditions, such as CWC, and the following conclusions have beenobtained:

(1)Among the three wind conditions at differentlevels of single towers and multi-tower synchronization, the frequency of CWC is the lowest at less than1% for single towers and tends to be zero in multitowersynchronization. The frequency of LWC is alwaysgreater than RWC, and the synchronous rate oftwo towers is always higher than that of three towers, showing the feature that: the larger area, thelower frequency. At all levels, the frequency of thethree-tower synchronous RWC is 0.7%–2.0% and itsteadily increases with height, and the correspondingwind profile agrees with that of neutral level stratification.In LWC, the synchronous rate of the three towersis 4.2%–4.3%, with almost no difference between low and high levels, and the corresponding wind speed profileis in negative shear shape or the profile index issignificantly reduced, which agrees with that of the neutral level stratification.

(2)The three-tower synchronous LWC and RWCvary annually and diurnally. LWC is concentratedin winter and occurs very little in summer, whereasRWC mainly occurs in spring and summer and is almostzero in winter. The average diurnal variation ofLWC in the entire year is high in the morning and lowin the afternoon, with the highest before noon, and the lowest at around 16:00 pm. The diurnal variationof LWC differs greatly in winter and summer. Theaverage diurnal variation of RWC in the entire yearexhibits a dual-peak distribution, with peaks at 5:00am and 19:00 pm and valleys after noon and beforedawn. The diurnal variation of RWC also substantiallydiffers between winter and summer.

(3)During the study year, the longest durationof LWC is 640 min and that of RWC is 700 min; thefrequencies of both conditions substantially decreasewith an increase in duration, but there are still 10LWC occurrences with a duration over 300 min and 6RWC occurrences with a duration over 300 min.

(4)The relation between the duration of LWCor RWC and their frequencies can be quantitativelywell described by a logarithmic function, and the occurrencefrequency of LWC and RWC of differentdurations in the region can be easily calculated byusing the fitting function equation.

In this study, the synchronous anemometric dataof three wind towers in the large scale wind powertransmission area in Hami, Xinjiang of NorthwestChina, are used to study synchronous variations inwind speed and wind power parameters associatedwith wind power generation in the area representedby the wind towers. Regional synchronous wind observationsare statistically analyzed to explore thetechnical and analytic methods that could be appliedto wind power project planning and wind power gridconnection. The mechanism and causes of regionalsynchronous wind are quite complicated, and involvea number of specialized disciplinary fields. Therefore, further in-depth research is required in the future.

References
Chi Yongning, Liu Yanhua, Wang Weisheng, et al., 2007: Impact of wind power connection to power system. Grid Technology, 31, 77-78. (in Chinese)
China Meteorological Administration (CMA), 2007: Technical specification for wind farm meteorolog-ical observation and data review and correction. QX/T74-2007. Standards Press of China, Beijing, 14 pp. (in Chinese)
Fan Gaofeng, Wang Weisheng, and Liu Chun, 2008: Wind power short-term prediction system based on artificial neural network. Grid Technology, 32, 72- 76. (in Chinese)
Fu Guiqin and Li Yunzong, 2008: Influence analysis of meteorological conditions on electric load. Meteo-rological Science and Technology, 36, 795-800. (in Chinese)
Gao Zonghe, Teng Xianliang, and Zhang Xiaobai, 2010: Active power dispatching and control plan for inter-connected grids adaptable to large-scale wind power connection. Power System Automation, 34, 37-38. (in Chinese)
General Administration of Quality Supervision, Inspec-tion, and Quarantine of PRC (GAQSIQ), 2002: Methodology of Wind Energy Resource Assessment for Wind Farm. GB/T18710-2002. Standards Press of China, Beijing, 19 pp. (in Chinese)
GAQSIQ, 2011: Technical Rule for Connecting Wind Farm to Power System. GB/T19963-2011. Stan-dards Press of China, Beijing. (in Chinese)
Han Xiaoqi, 2010: Influence of wind power output varia-tion on system frequency regulation. China Electric Power, 43, 26-27. (in Chinese)
Huang Jiayou, 2013: Meteorological Statistical Analysis and Forecast Method. China Meteorological Press, Beijing, 298 pp. (in Chinese)
Huang, J., X. Lu, and M. B. McElroy, 2014: The meteo-rology of wind energy in the US and its implication on the interconnection of wind farms. Renewable Energy, 62, 331-340.
IEC, 2005: IEC61400-1. Third edition 2005-08, Wind Turbine-Part 1: Design requirement. International Electrotechnical Commission (IEC), 92 pp.
Ju Ping, Chen Qian, Xiong Chuanping, et al., 2006: Load classification and comprehensive modeling based on diurnal load curve. Power System Automation, 30, 6-7. (in Chinese)
Leahy, P. G., and E. J. Mcleogh, 2013: Persistence of low wind speed conditions and implications for wind power variability. Wind Energy, 16, 575-586.
Lei Yazhou, Wang Weisheng, Yin Yonghua, et al., 2002: Wind power penetration limit calculation based on chance constrained programming. Proceeding of the CSEE, 22, 33-36. (in Chinese)
Li Zechun, Zhu Rong, He Xiaofeng, et al., 2007: Research on wind power resources evaluation technologies and methods. Acta Meteor. Sinica, 65, 708-717. (in Chinese)
Liu Yongqian, Han Shuang, and Hu Yongsheng, 2007: Research overview of short-term forecast for wind farm output. Modern Electric Power, 24, 6-11. (in Chinese)
Lu, X., M. B. McElroy, C. P. Nielsen, et al., 2013: Opti-mal integration of offshore wind power for a stead-ier, environmentally friendlier, supply of electricity in China. Energy Policy, 62, 131-138.
National Energy Administration, 2012: “Twelfth Five-Year” for the Development of Wind Power. Report No. 195, 10-13. (in Chinese)
Saifur, R., and M. Pipattanasomporn, 2011: Operating impacts and mitigation strategies with large-scale wind power penetration in the United States. Power System Automation, 35, 4-5.
Sinden, G., 2007: Characteristics of the UK wind re-source: Long-term patterns and relationship to elec-tricity demand. Energy Policy, 35, 112-127.
Song Lili, Zhou Rongwei, Yang Zhenbin, et al., 2012: Me-teorological technology application and development in wind energy resources utilization. Engineering Science, 14, 96-101, 112. (in Chinese)
Song Lili, Zhu Rong, Guo Yatian, et al., 2014: Detailed Survey and Assessment Report on Nationwide Wind Power Resources. China Meteorological Press, Bei-jing, 300 pp. (in Chinese)
Sun Yuanzhang, Wu Jun, and Li Guojie, 2007: Impact of wind power generation to power system. Grid Technology, 31, 55-57. (in Chinese)
Xue Heng, Zhu Ruizhao, Yang Zhenbin, et al., 2001: Es-timation of wind power resource reserves in China. Solar Energy Journal, 22, 167-170. (in Chinese)
Zhou Hui, Wang Wei, Qin Haichao, et al., 2005: Daily load curve prediction studies for similar days based on multi-period meteorological data judgment. Re-lay, 33, 41-44. (in Chinese)
Zhou Wei, Peng Yu, Sun Hui, et al., 2009: Dynamic economic dispatching for power systems containing wind farm. Proceeding of the CSEE, 29, 13-18. (in Chinese)