1 / 24

OPERATIONAL EXPERIENCE WITH STATE ESTIMATION AT HYDRO-QUÉBEC

OPERATIONAL EXPERIENCE WITH STATE ESTIMATION AT HYDRO-QUÉBEC. S. Lefebvre, J. Prévost , J.C. Rizzi, P. Ye (IREQ) B. Lambert, H. Horisberger (TransÉnergie). Network description. Main network Generation Installed capacity of around 38 000 MW

mick
Télécharger la présentation

OPERATIONAL EXPERIENCE WITH STATE ESTIMATION AT HYDRO-QUÉBEC

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. OPERATIONAL EXPERIENCE WITH STATE ESTIMATION AT HYDRO-QUÉBEC S. Lefebvre, J. Prévost, J.C. Rizzi, P. Ye(IREQ) B. Lambert, H. Horisberger (TransÉnergie)

  2. Network description • Main network • Generation Installed capacity of around 38 000 MW Over 95% of the generation is hydro (soon 4000 MW of wind!) Asynchronous with the rest of North America • Transmission 735 & 315 kV AC systems Multi Terminal DC line 450 kV (over 1000 km) 4 back to back DC Terminals (soon 5!) • Sub transmission 230, 161, 120 & 69 kV AC systems

  3. Network Description (next …) • 735 kV Grid • Components • 11000 km of lines (charging around 33 000 MVAR) • Series capacitors: 12 000 MVAR • Switched inductors: 25 000 MVAR • Switched capacitors: 13 000 MVAR • SVC & SC: -3800 – 5800 MVAR • Characteristics • Operation constrained by stability and voltage limits(almost no thermal limit) • Generally operated well under the SIL (lines switching can even be used for voltage control) • Ramping rate becoming more and more important ( 200 MW/Min.)(may even cause voltage control difficulties) • Corona effect may suddenly become important(may reach over 2 times the thermal losses: e.g. 1000 MW)

  4. Current status of system control • Hydro Quebec EMS/SCADA control centers • One Provincial control center (EMS) and one back up center • Responsible of the bulk transmission grid (735 to 315 kV) • Main Functions: - Data acquisition - Automatic Generation Control - Economic Dispatch - Security Analysis - Exchange management - Outage Management - Voltage control • Seven regional control centers (SCADA) • Responsible of the sub transmission grid (230 kV to 69 kV) • Main Functions: - SCADA (for the full HQ’s network) - Outage Management • Operations are usually triggered by operators from the provincial center and then are executed by operators at the regional centers

  5. State estimation experience • LASER0 • In-house product • Pd and QV decoupled algorithm • Model: 735 kV network • LASER1 • Commercial product: ABB • In house simple pre-processing topology error function • LASER2 • Commercial product: SNC (formerly CAE) In house elaborate pre-processing topology error function

  6. State estimation latest development at HQ’s • Archiving system • In-house product • Main functions:- Static network model (CIM/XML) saved after each DB update - Dynamic raw input/output of SE function saved at each RTS run - Power flow case (IEEE) saved after each RTS run • Matlab SE toolbox • In-house product • Main functions:- Real time snapshot handling - Sub network extraction (by substation or voltage level) - Measurement system analysis (redundancy, identification of critical meas.) - SE algorithms (WLS, Huber, DWLS) - Cases modification & comparison - Parameter estimation - Monte Carlo simulations (evaluation of the solution sensitivity & precision)

  7. State estimation latest development at HQ’s (next …) • Reporter • In-house product • Main goal: Identification of topology and measurement errors Robust approach (no false alarm) • Operate on a continuous base (24/7) • Independent of SE solution (convergence, false rejected meas., …) • Based on a heuristic approach: a set of rules, combinatorialanalysis and iterative processing • Takes advantage of previous network & telemetry data (Hn-1, Zn-1) • Filtering reporting capability (already know bad modeling, …) • Historical reporting capability (error, start & end time, frequency, …) • Others reporting possibility (performance index degradation, …) • Web & email reporting (used by the support engineer team)

  8. SE model Near half of switches are breakers that are 100% telemetered The other half is reconfigu- ring switches and only 70% are telemetered. Thus over 750 switches are based on a manual entry Moreover not all switch are modeled. By example maintenance switches are rarely modeled

  9. SE measurements and their redundancy The 735 transmission grid model has a very good redundancy (4.1). The sub-transmission grid model has a lower redundancy (2.6) QV redundancy (3.5) is much higher than its Pd counterpart (2.0)

  10. SE problems • Topology error • Originate mainly from maintenance work • Bad series switch status (bus split/merge more diffcult to identify) • Bad shunt switch status (more diffcult to identify) • Q-V model more complex, more sensitive and less accurate than P-d model • A important quantity of reactive (accuracy can become a problem) • A lot of elements (Serie cap, reactors, SVC, …) • Weather dependant parameters • Corrona effect (from almost 0 to 2 times thermal losses) • Temperature (from -40 celcius to +40 celcius -> 30% of errror)

  11. SE model is never exact • Inequality constraint cannot be model (ex: power limit, …) • Mutual effect cannot modeled (ex: on double circuit Z11~ 5%*Z1) • Complex equipments (DC, SVC, …) generally can only be modeled as simple injection • Variable system parameters as affected by temperature and humidity are generally not considered (ex: corona loss , …) • Three-windings transformer generally modeled as two-windings • Constant LTC Transformer impedance often used • Isolation switches and/or breaker not always modeled(represented only in their normal position) • Small load not always modeled (auxiliary service) • Network modification (ex: new line) not always in sync with the model • Transmission line parameters calculation often based on typical values (height, span, sag)

  12. SE measurements is never exact • Manual entry inaccuracy (switch status, …) • Presence of time skew (ex: 25s. between provincial and the regional centers)(ex: manual entry can be delay by several minutes) • Measurement dependency (V, I, P, Q) • Presence of dead bands in the acquisition chain • Measurement bias (e.g. in CCVT) • Presence of unbalance (zero and negative sequence) • Use of phase measurements vs sequence (direct) measurements • Variable standard deviation (s = f(burden))

  13. SE solution quality Relative Performance index (%)

  14. SE usage (example 1) Topology error detection: Side effect 905 -1 VOLT 735 744.2 749.3 14.2 -0.7 0-J 735 0.0 0.0 0.0 0.0 JAC CAR LINE L7018 CP 1023.0 -189.0 40.1 -118.3 EW XFR2 T1 735 -40.1 118.3 906 -1 VOLT 735 743.6 740.9 25.0 0.4 0-J 735 0.0 0.0 0.0 0.0 CHAMO LINE L7026 CP -864.0 -24.0 -243.9 -82.0 EW MICOUA LNSX L7019 CP -822.0 -111.0 -478.4 -143.9 EW XFR2 T2 735 722.3 226.0 907 -1 VOLT 735 731.7 23.9 0-J 735 0.0 0.0 0.0 0.0 MICOUA LINE L7019 -478.5 -298.4 S-D CXC15010 478.4 298.4 Topology error Wrong manual entry

  15. SE usage (example 2) Parameter validation: Double circuit of short length, modeled as equal lengthbut in reality not exactly the same length

  16. SE usage (example 3) Accurary improvement: Flow735: smeas/ sest> 3 « Limite sud » flow evaluation: Measurements accuracy: 3s =360 MW Estimates accuracy: 3s = 120 MWCan increase the margin by 240 MW!!!

  17. Corona evaluation & minimization: SE usage (example 4) Average: 8 MW loss reduction (1%) Improved by voltage control (low & flat)

  18. SE usage (example 5) Loss evaluation & minimization: Average: 33 MW loss reduction (4%) Improved by voltage control (high & flat)

  19. Conclusion • Need for SE Technology that can handle more appropriately practical issues • Adding more measurements is not always the solution(although useful) • SE does not only provide states (X) but also a model (H)So, even if PMU may help, it will not solve all problems • Model & errors/inaccuracies cannot be avoidedSo, model should not be considered as “hard constraint” (at least for parameters like R & G, and may be even X for LTC!) • All information available should be used (inequality constraint, setpoint, previous data (Zn-1, Hn-1), quality (manual, telem., …),tag • Electrical topology (not necessary physical) error detection, identification and correction function should be de facto available • Sudden quality change (residues, rejected meas., …) should trigger a validation mechanism

  20. Conclusion • Need for SE support tools • Quality indexes evaluation (standard indexes will also be nice!) • Measurements analysis (critical meas., local redundancy, …) • Model analysis (parameter estimation, sensitivity, …) • Solution analysis (estimate accuracy, robustness in regard of meas. loss, …) • Visualization tools for analysis and debugging (ex:3D diagram showing residues, biases, rejected meas.) • Model validation tools (modification, solutions comparator, …) • Improved SE solution quality will increase its role • Transmission optimization (LM, DSA, …) • Market operation (ED, …)

  21. Questions ?

  22. Provincial / area Control Center Power plant / Transmission substation Sub transmission substation Distribution control center Distribution feeder Hydro-Québec TransÉnergieControl centers architecture Phone IEC 60870-5 ICCP Regional Control Center DNP3 7 DNP3 Proprietary MODBUS .. Phone

  23. Power flow optimization State estimation Limit service frequency control Contingency analysis Regional control centers Remote terminal units Hydro-Québec TransÉnergieProvincial control center (main information functions & information flows) Limit violations Switching advices Flow limits ATCs Controlorder Network solution Snapshot SCADA Setpoints The real-time sequence (RTS) of the network analysis tools runs every minute (500 full AC contingencies, 5 min) Telemetry

More Related