Optimal wind farm operation

Abstract

The present application is concerned with a flexible way of operating a wind farm with a plurality of degrading wind turbine components. According to the invention, maintenance scheduling and power production in the wind farm are handled concurrently in a single optimization step. Instead of a serial approach first scheduling maintenance activities and subsequently adapting the power production and/or wind turbine operation the two aspects are optimized together. The wind farm operation takes maintenance aspects into account by adapting life index or health status based on modeled mechanical and electrical stress. Accordingly, the wind farm owner may decide when and how much energy to produce accepting which level of stress to the turbine equipment. The proposed optimization of wind farm operation includes all aspects of transmission network operator settings, the topology of wind farms and the underlying collector grid, the short and long term wind conditions forecasts, the conditions of the turbines, the estimated remaining operational time under different usage patterns and times, as well as aspects of the electricity market.

Claims

1. A method of operating a wind farm including plurality of wind turbines with a plurality of turbine components comprising a first and further turbine components, said plurality of turbine components subject to degradation, comprising: predicting, for a turbine component of a first wind turbine and for each of a sequence of time intervals (t.sub.1 . . . t.sub.N1) into the future, based on a sequence of first turbine control input values (u.sub.1(t.sub.1) . . . u1(t.sub.N1)) including a component maintenance action at a maintenance interval tM1, a component life index L(t) of the first turbine component, determining a sequence of optimum turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N1)) including a) an optimum maintenance interval t.sub.Ml* for the first turbine component and b) optimum maintenance intervals t.sub.Mj* for each of the further turbine components that optimize an objective function J(u) depending on the component life index L of the first turbine component and on component life indices L.sub.j of the further turbine components, and operating the first wind turbine according to at least one optimum turbine control input value u.sub.1*(t.sub.1), wherein the method further comprises predicting the component life index of the component based on a sequence of second turbine control input values (u.sub.2(t.sub.1) . . . u.sub.2(t.sub.N2)) of a second wind turbine electrically connected to a same branch of a collector grid of the wind farm as the first wind turbine, by evaluating voltage phase and amplitude difference between the first turbine and the second turbine, and determining optimum first and second turbine control input values u.sub.1*(t), u.sub.2*(t).

2. The method of claim 1, comprising predicting the component life index of the component based on the sequence of second turbine control input values (u.sub.2(t.sub.1) . . . u.sub.2(t.sub.N2)) of the second wind turbine located upstream of the first wind turbine by evaluating predicted wind turbulences at the first turbine and caused by operational behavior of the second wind turbine, and determining optimum first and second turbine control input values u.sub.1*(t), u.sub.2*(t).

3. The method of claim 2, comprising providing for each of the sequence of time intervals (t.sub.1 . . . t.sub.Nj) into the future a wind forecast, and calculating a sequence of electrical power output values (p.sub.j(t.sub.1) . . . p.sub.j(t.sub.Nj)) of each of the plurality of turbines of the wind farm based on the wind forecast and based on a respective sequence of turbine control input values (u.sub.j(t.sub.1) . . . u.sub.j(t.sub.Nj)), providing for each of the sequence of time intervals (t.sub.1 . . . t.sub.Nj) into the future an electrical power demand or power generation forecast P(t) for the wind farm, and providing an objective function J penalizing a deviation of a sum of the calculated electrical power outputs p.sub.j(t) of the plurality of turbines of the wind farm from the power demand or power generation forecast P(t).

4. The method of claim 2, comprising determining the sequence of optimum turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N1)) at least once per hour.

5. The method of claim 2, wherein the wind turbines of the wind farm are electrically connected to a collector grid, comprising providing an objective function including a term penalizing power flow imbalance in the collector grid.

6. The method of claim 2, wherein the objective function includes a term indicative of earnings from electrical power generated by the first turbine over a time span including the component maintenance time t.sub.M.

7. The method of claim 1, comprising providing for each of the sequence of time intervals (t.sub.1 . . . t.sub.Nj) into the future a wind forecast, and calculating a sequence of electrical power output values (p.sub.j(t.sub.1) . . . p.sub.j(t.sub.Nj)) of each of the plurality of turbines of the wind farm based on the wind forecast and based on a respective sequence of turbine control input values (u.sub.j(t.sub.1) . . . u.sub.j(t.sub.Nj)), providing for each of the sequence of time intervals (t.sub.1 . . . t.sub.Nj) into the future an electrical power demand or power generation forecast P(t) for the wind farm, and providing an objective function J penalizing a deviation of a sum of the calculated electrical power outputs p.sub.j(t) of the plurality of turbines of the wind farm from the power demand or power generation forecast P(t).

8. The method of claim 7, comprising determining the sequence of optimum turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N1)) at least once per hour.

9. The method of claim 7, wherein the wind turbines of the wind farm are electrically connected to a collector grid, comprising providing an objective function including a term penalizing power flow imbalance in the collector grid.

10. The method of claim 7, wherein the objective function includes a term indicative of earnings from electrical power generated by the first turbine over a time span including the component maintenance time t.sub.M.

11. The method of claim 1, comprising determining the sequence of optimum turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N1)) at least once per hour.

12. The method of claim 11, wherein the wind turbines of the wind farm are electrically connected to a collector grid, comprising providing an objective function including a term penalizing power flow imbalance in the collector grid.

13. The method of claim 11, wherein the objective function includes a term indicative of earnings from electrical power generated by the first turbine over a time span including the component maintenance time t.sub.M.

14. The method of claim 1, wherein the wind turbines of the wind farm are electrically connected to a collector grid, comprising providing an objective function including a term penalizing power flow imbalance in the collector grid.

15. The method of claim 14, wherein the objective function includes a term indicative of earnings from electrical power generated by the first turbine over a time span including the component maintenance time t.sub.M.

16. The method of claim 1, wherein the objective function includes a term indicative of earnings from electrical power generated by the first turbine over a time span including the component maintenance time t.sub.M.

17. A wind farm management system for operating a wind farm including a plurality of wind turbines with a plurality of turbine components comprising a first and further turbine components, said plurality of turbine components subject to degradation, comprising: a processor; memory in electronic communication with the processor; and instruction stored in the memory, the instructions being executable by the processor to: predict a component life index L(t) of a turbine component of a first wind turbine, for each of a sequence of time intervals (t.sub.1 . . .t.sub.N) into the future, based on a model of the first wind turbine and based on a sequence of first turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N)) including a component maintenance action at a maintenance interval t.sub.M, determine a sequence of optimum turbine control input values (u.sub.1*(t.sub.1) . . . u.sub.1*(t.sub.N)) including a) an optimum maintenance interval t.sub.M* for the first turbine component and b) optimum maintenance intervals t.sub.Mj* for each of the further turbine components that optimize an objective function J(u) depending on the component life index L of the first turbine component and on component life indices L.sub.j of the further turbine components, and operate the first wind turbine according to at least one optimum turbine control input value u.sub.1*(t.sub.1), predict the component life index of the component based on a sequence of second turbine control input values (u.sub.2(t.sub.1) . . . u.sub.2(t.sub.N2)) of a second wind turbine electrically connected to a same branch of a collector grid of the wind farm as the first wind turbine, by evaluating voltage phase and amplitude difference between the first turbine and the second turbine, and determine optimum first and second turbine control input values u.sub.1*(t), u.sub.2*(t).

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The subject matter of the invention will be explained in more detail in the following text with reference to preferred exemplary embodiments which are illustrated in the attached drawings, in which:

(2) FIG. 1 schematically shows a wind farm layout,

(3) FIG. 2 depicts turbine control input values and corresponding life index evolution, and

(4) FIG. 3 depicts various aspects of a wind farm control concept.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

(5) FIG. 1 depicts a wind farm layout with three parallel rows 11 each comprising seven wind turbines connected to a branch 12 of a collector grid. Exemplary first and second wind turbines are denoted 111 and 112. All collector grid branches converge at a Medium Voltage MV bus bar 13 of a main substation. A transformer 14 connects the MV bus bar to a Point of Common Coupling PCC and further to a transmission grid 15 under control of a grid or network operator.

(6) FIG. 2 is a graph showing, at distinct time steps beginning with t.sub.0 and ending at t.sub.N, turbine control input values u(t) (small circles) as well as an interpolated component life index evolution L(t) or lifetime trajectory. The latter is, in this representation, indicative of a consumed lifetime or inverse health condition of the turbine component. Starting from a base index value at present time to the index increases steadily in relation to the values of the turbine control input. The life index is re-set to a lower value at a maintenance time step t.sub.M.

(7) FIG. 3 depicts various aspects of a wind farm control concept. The wind farm has a certain wind farm layout specifying an electrical collector grid topology and a mutual geographical arrangement of the individual wind turbines. The latter allows to account for wind turbulences, or wake effects, between different wind turbines in the wind farm management system. Three wind turbine instances are depicted, each provided with a Wind Turbine Control WTC module and an optional Condition Monitoring System CMS including sensors and diagnostics tools to determine a life index or health condition of the turbine.

(8) The wind farm management system receives measurement data from the wind farm including the data from the CMS and the current wind conditions. Additionally, the wind farm management system receives the current network operator dispatch points specifying the operation mode of the wind farm and the associated operating points on a farm level.

(9) The wind farm management system receives short, mid and long term wind forecasts for wind power prediction as well as short, mid and long term energy price forecasts. Both information can be provided by external services or by the wind farm operator. Likewise, demand or load forecasts, possibly in the form of repeating seasonal patterns, may be provided to the system. Here, seasonal patterns can be envisioned that specify periods where energy demand is high and periods were it is likely to have low demands from the grid perspective. Furthermore, external tools or services are used to predict the long term network operator dispatch points, wherein energy demands may also be modelled within such long term forecast of the network operator dispatch points.

(10) The wind farm management system includes a wind turbine condition module for determining a life index indicative of a health condition, or lifetime consumption, or aggregated wear, of each wind turbine, or even of individual components thereof. Typical components considered include blades, converter, generator, and transformer of the turbine. An updated life index indicative of a present component status may be obtained from the CMS and/or determined from corresponding sensor data.

(11) Alternatively, a life index may be determined or predicted based on recorded or forecast hours of operation of the turbine components and/or corresponding operating conditions or turbine control inputs. Turbine operation gives rise to a mechanical stress for the components of wind turbine to which adds an electrical stress mainly caused by the currents carried by electrical components of the turbine. At least two ways of determining, linking, a life index to the operating hours and/or turbine control inputs may be considered: a) The operating hours and/or turbine control inputs may be mapped to, or compared with, an expected life index provided by the component manufacturer and based on design operating parameters. b) The operating hours and/or turbine control inputs may be fed into models of the turbine components to identify the deterioration rate and hence the life index of the component. The models may include empirical models of the turbine components that are trained on observed deterioration data of turbines of an at least similar kind than the turbine under consideration. Turbine model training, or model calibration, also includes suitable evaluation of observed premature turbine failures.

(12) Determining a life index or forecasting a lifetime trajectory of a turbine component may include one of crack growth or creep models of thick-walled components such as turbine blades subject to mechanical stress, fatigue estimation in the blades related to start ups and shut downs, turbine converter power electronics failure estimation, e.g. of power semiconductor switching elements such as IGBTs, related to high load such as speed, torque, power operation and/or large number of switching events, wear estimation, e.g. of coal brushes, slip ring, motor bearings, gearbox, related to number of rotations of the turbine.

(13) The wind farm management system includes an inter-turbine interaction module for evaluating aerodynamic and electrical or electro/mechanical interactions between individual wind turbines of a wind farm. In a simplified picture, if an upstream turbine creates wakes, a downstream turbine experiences more turbulent conditions and corresponding mechanical vibrations. The inter-turbine interaction may be quantified as a function of the operational points the individual turbines. Accordingly, by suitably balancing the operational points of neighboring turbines, the mechanical load can in some cases be transferred or migrated from one turbine to another.

(14) The wind farm management system ultimately may include a production and maintenance planning module for determining production and maintenance schedules and for issuing corresponding turbine control inputs including set-points for active or reactive power to the individual wind turbine control modules. The maintenance planning module is aware of possible maintenance intervals and constraints, i.e. the number of turbines may be serviced in a given maintenance interval.

(15) The inclusion, in an objective function, of a life index allows to influence an availability of a wind turbine component through controlled ageing or lifetime engineering, by prediction of component degradation based on turbine control inputs. Specifically, a method of operating a wind farm including a plurality of wind turbines with turbine components subject to degradation may comprise the steps of

(16) a) determining a trajectory of candidate turbine control input values u(t.sub.i) at N future time steps or intervals t.sub.1 . . . t.sub.N, including values of active/reactive power generated P, Q, and/or set points for pitch angles, yaw angles, generator/rotor speed, and including a maintenance time t.sub.M,

(17) b) determining, by means of wind turbine Model Predictive Control MPC, from the trajectory of turbine control input values u(t) a simulated lifetime trajectory L(t) of a first turbine component,

(18) c) computing an objective function J(u) comprising, inter alea, the simulated lifetime trajectory L(t),

(19) d) iteratively repeating steps a) through c) with an optimisation module varying the trajectory of turbine control input values u(t) until an optimised value of the objective function J[u] is obtained for a trajectory of optimum turbine control input values u*(t) including an optimum maintenance time t.sub.M*,

(20) e) applying to the turbine at least a first optimum turbine control input value u*(t.sub.1) of the trajectory of optimum turbine control input values.

(21) The method may also comprise, wherein an operational behaviour of a second wind turbine is predicted from a trajectory of multi-turbine input values, determining the simulated lifetime trajectory for the component by predicting and evaluating wind turbulence or wake effects from the operational behaviour of the second wind turbine on the first turbine.

(22) The features of the method of operating a wind farm and the wind farm controller as described herein may be performed by way of hardware components, firmware, and/or a computing device having processing means programmed by appropriate software. For instance, the wind farm controller can include any known general purpose processor or integrated circuit such as a central processing unit (CPU), microprocessor, field programmable gate array (FPGA), Application Specific Integrated Circuit (ASIC), or other suitable programmable processing or computing device or circuit as desired. The processor can be programmed or configured to include and perform features of the exemplary embodiments of the present disclosure such as a method of operating a wind farm. The features can be performed through program or software code encoded or recorded on the processor, or stored in a non-volatile memory accessible to the processor, such as Read-Only Memory (ROM), erasable programmable read-only memory (EPROM), or other suitable memory or circuit as desired. In another exemplary embodiment, the program or software code can be provided in a computer program product having a non-transitory computer readable recording medium such as a hard disk drive, optical disk drive, solid state drive, or other suitable memory device or circuit as desired, the program or software code being transferable or downloadable to the processor for execution when the non-transitory computer readable medium is placed in communicable contact with the processor.

(23) While the invention has been described in detail in the drawings and foregoing description, such description is to be considered illustrative or exemplary and not restrictive. Variations to the disclosed embodiments can be understood and effected by those skilled in the art and practising the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word comprising does not exclude other elements or steps, and the indefinite article a or an does not exclude a plurality. The mere fact that certain elements or steps are recited in distinct claims does not indicate that a combination of these elements or steps cannot be used to advantage, specifically, in addition to the actual claim dependency, any further meaningful claim combination shall be considered disclosed.