Critical Limitations of Wind Turbine Power Curve Warranties

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Critical Limitations of Wind Turbine Power Curve Warranties

A. Albers
Deutsche WindGuard Consulting GmbH, Oldenburger Straße 65, D-26316 Varel, Germany
E-mail: a.albers@windguard.de, Tel: (++49) (0)4451/9515-15, Fax: (++49) (0)4451/9515-29

Summary ence from activities as technical adviser in


contract negotiations with a variety of turbine
Wind turbine power curve warranties of suppliers. Another technical basis of the
many turbine suppliers became subject to observed effects are several power curves
increasing limitations over recent years. This tests as performed for verifying power curve
more and more questions the economic warranties, as well as experience as official
value of power curve warranties. Some of expert in court cases.
the technical risks linked to the limitations
remain hidden for non-experts, e.g. for tur- 2 Limitation on Verification Period, Test
bine purchasers and sales people. Of par- Turbines and Announcement of Tests
ticular importance are limitations on the va-
lidity of the warranted power curve to certain A power curve warranty is expected to pro-
environmental conditions and turbine condi- vide a guarantee that the power output of all
tions, which often mean that the warranted contractual wind turbines as function of wind
power curve is not valid for the average speed is kept at an agreed level for the war-
conditions expected at a certain wind farm ranty period. Thus, a verification of the war-
site, leading to too optimistic yield expecta- ranty should be possible at each turbine,
tions. Turbine suppliers should warrant site where a power curve test is possible, and
specific power curves rather than generic within the entire warranty period. Also a
power curves to overcome this situation. repetition of the power curve test should be
The expected revision of the power curve possible for the owner after some time for
testing standard IEC 61400-12-1 provides the case that doubts on the power curve
guidance on how power curves can be bet- arise. However, this is not the case in the
ter adjusted to site specific conditions and standard warranties as presented by many
how uncertainties of power curves due to turbine suppliers. The test period is often
the impact of environmental conditions can limited to a short period after commissioning
be treated. Other critical factors of power of e.g. one year, and sometimes the verifica-
curve warranties are mainly related to limita- tion test is allowed only once. Thus, the
tions on the allowed period for verifying the warranty gets useless for the owner in the
fulfilment of the warranty, to limitations on case that the power performance decreases
the selection of test turbines, to the right of with time. Furthermore, the possible test
the turbine supplier to optimise the turbines turbines are often contractually fixed. By
before the verification, to limitations on the this, verifications at other turbines, where
testing procedure, to an often found artificial doubts about the power curve may arise, are
increase of the uncertainty of verification not possible for the owner.
tests and thus to an effective (and hidden) Critical in this respect is also the fact that
lowering of the warranty level, to improper many standard warranties provide only a
compensation rules and to unreasonable warranty on the mean power curve of all
high limitations of the liabilities of power turbines of a wind farm as tested on a set of
curve warranties. exemplary machines. No compensation for a
significant lack of the power curve of outlier
1 Introduction machines is provided in this case. The war-
ranties should rather provide a power curve
A trend has been observed over the past warranty for single turbines in addition to the
few years that warranties on wind turbine warranty for the mean of the turbines, while
power curves became subject to increasing the warranty level may be reduced for the
limitations. Often, neither turbine purchas- single turbines compared to the level related
ers, nor sales people of turbine suppliers are to the mean power curve of all turbines. The
aware of the risks and consequences linked lowering of the warranty level for single tur-
to such limitations. The reported findings bines can be oriented on the higher uncer-
and recommendations are based on experi- tainty of power curve tests for single turbines
compared to the uncertainty of a mean cases, advanced nacelle anemometry ac-
power curve as measured at a set of tur- cording to the draft standard CD IEC 61400-
bines. Instead, most of today’s standard 12-2 [2] is a proper power curve verification
warranties include a lowering of the warranty procedure. Also the application of lidars on
level for the mean power curve as tested at nacelles for power curve testing is expected
a set of turbines on the basis of the uncer- to become a good alternative power curve
tainty of a single power curve test. verification procedure in the near future [3].
Another limitation of many standard warran- Especially offshore, where the application of
ties is the obligation of the owner to inform met masts is extremely expensive, power
the turbine supplier about power curve tests. curve verifications may be feasible also on
Sometimes, the turbine supplier has explic- the basis of wind measurements with a
itly the right to optimise the power curve scanning lidar mounted on the turbine sup-
before the power curve verification starts. port structure at the tower foot. This proce-
This regulation makes it nearly impossible dure is very close to the draft revision of the
for the owner to proof a breach of the power standard IEC 61400-12-1 [4].
curve warranty, which was present before All of the alternative power curve verification
such an optimisation. The right to optimise procedures are still linked to higher uncer-
the power curve before the verification of the tainties than the application of IEC 61400-
warranty seems justified only if the condition 12-1. This should be no reason for the tur-
of the turbine is not in the responsibility of bine supplier to reject these methods, as the
the turbine supplier. Otherwise, e.g. in case warranty level is normally reduced by the
of the presence of full service contracts be- uncertainty of the power curve verification.
tween the owner and the turbine supplier, On the other hand, the owner can consider
the warranted power curve should always be the alternative verification procedures as
reached without optimisation by the turbine trade-off of the provided warranty level and
supplier prior to a power curve test, and the the effort for power curve verifications.
owner should have the right to perform blind It is pointed out that the author has success-
tests. fully implement alternative verification pro-
cedures in power curve warranties.
3 Limitation on General Verification
Procedure 4 Limitation on Special Test Conditions

Most power curve warranties require verifi- A clear trend has been observed over the
cation tests being performed on the basis of past few years that power curve verification
the standard IEC 61400-12-1 [1]. This procedures as defined in standard warran-
seems being justified as IEC 61400-12-1 is ties require more and more special test con-
so far the only international standard on ditions, often far beyond the requirements
power curve tests and as the general proce- defined in IEC 61400-12-1 and sometimes in
dure outlined in this standard is well ac- contradiction to some requirements defined
cepted. in IEC 61400-12-1.
However, the standard IEC 61400-12-1 has Many standard warranties require additional
practical shortcomings: Tests according to data filtering on atmospheric conditions, like
this standard are expensive as met masts on turbulence intensity, wind shear, vertical
reaching hub height are required. In com- flow inclination, air density, air temperature
plex terrain, the method is applicable only if and wind direction. This seems justified from
a site calibration with additional masts has the viewpoint of the turbine supplier, who
been performed prior to the installation of may intend to warrant the power curve only
the turbines. Furthermore, only turbines at under well defined environmental conditions
the boarder of wind farms can be tested. (warranty of generic power curve). However,
These shortcomings are a true burden for the consequence for the turbine owner can
turbine owners to perform verifications of the be that the power curve tested under such
warranted power curve at all, and some- special atmospheric conditions may not be
times power curve tests are just impossible. representative for the wind farm site any-
Thus, alternative procedures for the verifica- more and may thus be of limited value for
tion of power curves should be seriously economic considerations (e.g. wind re-
considered for power curve warranties, at source assessments). Sometimes, the addi-
least for cases, where an application of the tional data filtering leads to very idealised
standard IEC 61400-12-1 is difficult. In many measured power curves, which are just too
optimistic for economic considerations, i.e.
the warranty level is indirectly lowered by Other special testing conditions defined in
the filtering. In other cases, the definition of some standard power curve warranties are
special test conditions seems being quite related to certain conditions of the tested
arbitrary, and the intention of the turbine turbines. The justification of such turbine
supplier to well define the testing conditions conditions must be checked case by case: It
is clearly not met. Figure 1 illustrates the is mostly plausible to exclude special turbine
effect of special data filters as defined in conditions from power curve tests, for which
power curve warranties on the measurement the turbine supplier is not responsible, e.g.
results in terms of the annual energy pro- situations with improper grid conditions or
duction (AEP). The influence on the AEP noise reduced operation. Other special con-
varies approximately from -3 % to +3 %, but ditions, like e.g. automatic load reduction,
in the majority of the cases, the AEP is im- may be considered as the turbine condition
proved by the special data filtering, i.e. the representative for the wind farm site and
verified power curve tends being overopti- should thus not be excluded from power
mistic for the wind farm site. curve tests. A few standard warranties re-
In order to overcome the conflict of warrant- quire even the cut-in hysteresis to be ex-
ing a power curve for well defined conditions cluded from power curve verifications. This
(intention of turbine supplier) and getting a is completely unjustified, as the cut-in hys-
warranted power curve representative for teresis can be well measured and forms a
the wind farm site (intention of owner), war- normal, although turbulence dependent, part
ranted power curves should be defined site of the power curve. Excluding the cut-in
specific rather than generic. The new draft hysteresis leads to a systematic improve-
revision of the standard IEC 61400-12-1 [4] ment of measured power curves in favour of
provides methodologies to normalise power turbine suppliers. Many standard warranties
curve data in terms of the turbulence inten- require the application of database B ac-
sity and wind shear to pre-defined site spe- cording to IEC 61400-12-1 instead of data-
cific conditions [5]. In terms of the air den- base A (main result according to current
sity, such data normalisation is common version of IEC 61400-12-1). Database A
practice since a long time. Potential short- includes the high wind cut-off hysteresis,
comings of the normalisation procedures are while database B excludes the cut-off hys-
covered by additional uncertainties, which teresis from power curve tests. The use of
again reduce the warranty level. This should database B is supported by the fact that
be acceptable for the turbine supplier and often not much data is present in the wind
for the owner. Another advantage of data speed range where the high wind hysteresis
normalisation about data filtering is that the takes place, what can lead to quite arbitrary
extension of the measurement period, as measurement results. On the other hand,
sometimes caused by data filtering, is database B does not cover potential prob-
avoided. lems of a turbine to keep rated power at high
wind speeds and can lead to overoptimistic
4 AEP calculations. Furthermore, the arbitrari-
AEPspecial filter - AEPIEC filter [%]

3 ness of database A is expressed by a higher


2 statistical uncertainty of the measurement,
1 what reduces automatically the warranty
0 level. Thus database A should be accept-
-1 able for the turbine supplier and for the
-2
owner.
-3
0 5 10 15 20 25 30
Some standard power curve warranties in-
test number [-]
clude special conditions on the terrain at the
test site, which by far exceed the require-
Figure 1: Influence of special data filtering ments defined in the standard IEC 61400-
as defined in power curve warranties on the 12-1. Most of these requirements are rather
measurement results in terms of the annual arbitrary, what is expressed by the fact how
energy production (AEP). Shown is the AEP different these additional requirements are
of the power curve under special data filter- among the turbine suppliers and how often
ing minus the AEP of the power curve fil- such requirements are changed by certain
tered according to IEC 61400-12-1 in per- turbine suppliers. Sometimes, the additional
cent for 26 test cases. requirements make a power curve verifica-
tion impossible, what invalidates the whole uncertainty of the power curve verification.
power curve warranty. Additional require- However, most standard warranties include
ments on the terrain conditions are justified a reduction of the warranty level by one
only at extreme sites. standard uncertainty of the power curve
Some standard power curve warranties verification. By assuming a Gaussian distri-
have additional requirements on the posi- bution for the uncertainty assessment, this
tioning of met masts beyond the require- regulation implicitly means that the true
ments defined in IEC 61400-12-1, like posi- power curve of the test turbine does not
tioning the mast upwind of the turbine in the reach the warranted power curve by a prob-
main wind direction or in short distances to ability of 84 % as condition for a breach of
the turbine. Such additional requirements the power curve warranty. The choice of the
normally intend to improve the correlation of probability of 84 % is rather arbitrary. Refer-
wind speeds measured at the met mast and ence [6] shows examples where clear short-
present at the test turbine. However, in comings of power curves were present al-
complex terrain positioning of the mast up- ready in the case of power curve tests
wind of the test turbine can be an improper where the true power curve does not reach
choice leading to systematic errors of the the warranted power curve by a probability
site calibration. Furthermore, lowering the of only 70 %. The probability of 70 % is
distance between mast and turbine is some- equal to the case that the deviation of the
times not practicable. Thus, the feasibility measured and warranted power curve is
and appropriateness on additional require- 0.52 times the standard uncertainty of the
ments on mast positions must be judged for test. Thus, the warranty level provided by
each wind farm site separately. power curve warranties should be reduced
Certain standard power curve warranties only by 0.52 times the standard uncertainty
contain additional requirements on the de- rather than by the full standard uncertainty.
sign of met masts, exceeding the require- Some standard power curve verifications
ments defined in IEC 61400-12-1. Most of include minimum uncertainties to be as-
these requirements are intended to improve sumed for the assessment of the verified
the measurement accuracy, can be easily power curve. Some of these assumptions
fulfilled and do not change the measured are very conservative and do lack any tech-
power curve in a certain direction. However, nical reasoning. Such conservative uncer-
there are more critical requirements, like e.g. tainty assumptions lead to an improper,
requesting the masts not exceeding hub systematic and hidden lowering of the war-
height, leading to small systematic im- ranty level. The uncertainty assumptions
provements of the measured power curves should rather be as realistic as possible.
in the case that masts reaching exactly hub The standard IEC 61400-12-1 suggests a
height cannot be realised. methodology how to accumulate different
uncertainties of power curve measurements
5 Critical Aspects of Criteria of Fulfil- to an uncertainty in terms of the annual en-
ment of Power Curve Warranties ergy production. The exact formulation is
given by formula E.4 of the standard IEC
There are still power curve warranties exist- 61400-12-1. IEC 61400-12-1 includes fur-
ing, where the treatment of the uncertainties ther an approximation of formula E.4 given
of power curve verification tests is not de- by formula E.5. Formula E.5 systematically
fined. Experience as court adviser shows, increases the uncertainty in terms of the
that courts normally count the uncertainty of AEP over formula E.4 by about 10 %. De-
the verification procedure in favour of the spite the fact that formula E.4 represents the
turbine supplier if the treatment of the uncer- exact formulation, some standard power
tainty is not contractually agreed otherwise. curve warranties require formula E.5 to be
As the uncertainties of power curve verifica- applied for the calculation of the uncertainty
tions are considerable, the treatment of the in terms of the AEP. This is considered as
uncertainties should always be defined in being extremely unfair and reduces system-
the warranty. atically the warranty level (clearly hidden for
Most standard power curve warranties in- non-experts).
clude a reduction of the warranty level by Most standard power curve warranties re-
the uncertainty of the power curve verifica- quire the measured power curves being
tion. This principle is justified as the turbine extrapolated to the cut-off wind speed for the
supplier cannot be hold responsible for the comparison with the warranted power curve
in terms of the AEP. This procedure is fair power curve should rather be evaluated as
only if the cut-off wind speed is realistic, follows:
especially if the measured power curve is ● The single uncertainty components
required being evaluated according to data- should be cumulated between the single
base B (high wind cut-off hysteresis not power curve measurements in a physi-
included in measurement). Most cut-off wind cally meaningful way in terms of the AEP
speeds given by turbine suppliers do not by considering for each component the
refer to 10-minute periods, i.e. most turbines correlation among the measurements.
effectively cut-off at wind speeds referring to ● Based on this cumulating of uncertain-
10-minute periods as relevant for wind re- ties, a weighted AEP should be calcu-
source assessments below the cut-off wind lated such from the single measurements
speeds given by turbine suppliers. In addi- that a minimum uncertainty results for the
tion, most warranted power curves do not weighted AEP, rather than just calculat-
represent the high wind cut-off hysteresis of ing the arithmetic average of the meas-
the turbine but are rather extended to the ured power curves in terms of the AEP.
cut-off wind speed relevant for an averaging This weighted AEP and the associated
period below 10 minutes (re-cut-in not rep- uncertainty should be relevant for the
resented at all). Extending the measured evaluation of the criterion of fulfilment of
power curve to this less representative cut- the warranty.
off wind speed leads to a reduction of the It is pointed out that both of the above
deviation of the measured and warranted measures are linked to a (small) reduction of
power curve. This can be critical at high the uncertainty of the mean AEP, i.e. the
wind sites. Another problem of this practice warranty level is slightly increased com-
is that the uncertainty calculation according pared to the less sophisticated procedure of
to IEC 6140-12-1 refers only to the non- most standard warranties.
extended measured power curve, i.e. the As is explained in chapter 2, an appropriate
uncertainty assumed for the reduction of the power curve warranty should cover also
warranty level or for the evaluation of the significant shortcomings of power curves of
criteria of fulfilment of the warranty does not single turbines in addition to a warranty on
fully match the compared AEP’s. A more the mean of the turbines. In terms of the
appropriate procedure is to perform the AEP criterion of fulfilment of the warranty of the
calculations for the measured power curve power curves of single turbines, it makes
and the warranted power curve only up to sense to account the uncertainty of each
the wind speed covered by the measured single power curve test for each individual
power curve. By this, the AEP calculations machine. As this uncertainty is normally
are treated equally for both power curves, higher than the above described uncertainty
no speculation of the true power curve in the of the mean power curve, the warranty level
wind speed range not covered by the power for individual turbines is automatically re-
curve has to be made, and the calculated duced.
uncertainty matches to the AEP-calculation.
Another critical aspect of most standard 6 Critical Aspects of Compensation
power curve warranties is that the criterion Rules
of fulfilment is defined for the mean of the
measured power curves (normally for the Some standard power curve warranties pro-
mean AEP) by taking the mean uncertainty vide compensation in case of underperfor-
of the power curve tests into account, i.e. mance only for the warranty period. This is
the power curve warranty is fulfilled if the very critical for turbine owners if the war-
mean of the measured power curves plus ranty period is only in the range of 2 to 5
the mean standard uncertainty of the power years, because then the warranty covers
curve tests exceeds the warranted power only underperformance for a small fraction
curve in terms of the AEP. In fact, the mean of the expected lifetime of the turbines.
of the measured power curves is linked to a Thus, compensation should be provided by
slightly lower uncertainty than the mean the power curve warranty for the entire pe-
uncertainty of single power curve measure- riod where the underperformance is present,
ments, because some uncertainty compo- even if this period exceeds the warranty
nents are independent between the single period. However, it is considered being fair
measurements (e.g. uncertainty of site ef- that the fulfilment of the power curve war-
fects). Thus, the criterion on the mean
ranty can be verified by the owner only els, the warranty cannot be expected to
within the warranty period. compensate true significant underperfor-
Another shortcoming of most standard mance in terms of the power curve. The
power curve warranties is that compensation liability should cover at least 20 % of the
for underperformance is paid only after un- turbine purchase price.
successful trials of optimisation of the power
curve by the turbine supplier. The key issue warranted power level warranted availability level
true power level true availability level
of this regulation is that the optimisation of 1.2
the power curve and the re-measurement to 1.0
not covered
covered by power curve warranty

power level [-]


proof the optimisation can take very long. 0.8

avail. warranty
Furthermore, quite often problems with the

covered by
0.6
acceptance and interpretation of measure- 0.4

ment results by the turbine owner and by the 0.2


0.0
turbine supplier are observed. Thus, some- 0.0 0.2 0.4 0.6 0.8 1.0 1.2
times compensation gets due only after time [-]
years of measurements and discussions
(sometimes never), while the turbine owner Figure 2: illustration of problem of scaling
is suffering from underperformance of the compensation for underperformance in
wind farm. In order to overcome this, com- terms of power curve and availability propor-
pensation should be payable by the turbine tional to the true energy yield
supplier as soon as the owner proofs under-
performance according to the procedures
defined in the warranty, i.e. the compensa- 7 Conclusions
tion period should start with the start of the
relevant measurements of the owner, inde- ● Many regulations offered these days by
pendent from possible re-measurements or turbine suppliers in standard power curve
optimisation by the turbine supplier. Com- warranties are deemed being unbal-
pensation should be payable until the tur- anced.
bine supplier proofs by the procedures de- ● Many power curve warranties are hardly
fined in the warranty that the warranted capable to significantly reduce the risk of
power curve is met. turbine owners linked to power curve un-
A further shortcoming of many standard derperformance. The associated risks
power curve warranties is that the compen- should be considered in economic con-
sation for underperformance is kept propor- siderations, e.g. in uncertainties of wind
tional to the true energy yield in reference resource assessments and when com-
periods. If then the warranted power curve paring prices of wind turbines.
and the warranted availability level are not ● Wind turbine suppliers should rather
reached at the same time, the compensation warrant a site specific power curve,
for the underperformance of power curves is which is adjusted to the conditions ex-
lowered by a too low availability (see illustra- pected in a certain wind farm, rather than
tion in Figure 2). Thus, the compensation for generic power curves, which are valid
underperformance in terms of the power only under idealised conditions. This
curve should be kept proportional to the true would contribute to more realistic yield
energy yield scaled by the ratio of the war- expectations of wind farms.
ranted and true availability if both warranties
are not met at the same time. Alternatively, 8 References
the compensation for underperformance in
terms of the availability should be kept pro- [1] IEC 61400-12-1, Wind turbines - Part
portional to the true energy yield scaled by 12-1: Power performance measure-
the ratio of the warranted and true power ments of electricity producing wind tur-
curve. Another, less sophisticated, method bines, 2005
is to compensate underperformance in [2] IEC 61400-12-2 Committee Draft, Wind
terms of the power curve as a lump sum turbines - Part 12-2: Power performance
proportional to the expected energy yield. measurements of electricity producing
Another often observed critical aspect of wind turbines based on nacelle anem-
power curve warranties is a very low maxi- ometry, 2008
mum liability, e.g. of only 5 % of the turbine [3] Wagner R., Pedersen TF., Courtney M.,
purchase price. Under such low liability lev- Gottschall J., Antoniou I., Møller R.,
Pedersen SM., Le N., Mouritzen AS.,
Velociter T., Bardon M.; Power perform-
ance measured using a nacelle-based
lidar, proceedings of EWEA 2011, Brus-
sels
[4] IEC 61400-12-1, Edition 2 Committee
Draft, Wind turbines - Part 12-1: Power
performance measurements of electricity
producing wind turbines, September
2011
[5] Albers A.; Turbulence and Shear Nor-
malisation of Wind Turbine Power Curve
Data, proceedings of EWEC 2010, War-
saw
[6] Albers A.; Relative and Integral Wind
turbine Power Performance Evaluation,
proceedings of EWEC 2004, London

You might also like