Level control of small intake reservoir in hydraulic system with long and complex penstock - Implemented level control at Toro 3 HPP

Transcription

1 September 013 Page 1 Level control of mall intake reervoir in hydraulic ytem with long and complex pentock - Implemented level control at Toro 3 PP Damir Dolenc, Mitja Klopčar, Jernej Mazij Litotroj Power, d.o.o. Litotrojka 50, 1000 Ljubljana, Slovenia damir.dolenc@litotrojpower.eu 1. BACKGROUND Toro 3 PP i a 48 MW power plant intalled in central Cota Rica with 1 m 3 / plant dicharge and 65 m of head. It i deigned to work in level control and ue water outflow from Toro PP. Almot 7 km long water paage in combination with mall intake reervoir make the level control difficult. Our aim i to preent the ucceful method of level control ulation of complex cacade ytem uing a flow control. In addition, baic deign iue of the group controller will be decribed.. TORO 3 PP YDRAULIC SCEME Toro 3 PP i part of the Toro cacade ytem and i intalled after Toro PP. The Toro unit have a flow capacity of 0m 3 / intalled. The intake reervoir for Toro 3, which i tail water bain for Toro at the ame time, ha a water urface area of A reervoir=753 m [1]. Intalled flow capacity at Toro 3 i 10,9 m 3 /, meaning that 1 meter of intake-reervoir level would be pent in 37 if both Toro 3 unit are in operation and Toro unit are hutdown. The urge tank water urface i 78 m. The tunnel length between intake reervoir and urge tank i 5 km, and there i km of teel pentock intalled between urge tank and powerhoue. Together, the water paage length between intake reervoir and powerhoue i cloe to 7 km. Fig. 1 ydraulic cheme of the Toro 3 PP

2 September 013 Page 3. DESIGN OF LEVEL CONTROL AT TORO 3 PP When conidering claical PID level control in the deign phae, a problem became a mall intake reervoir area for which fat ulation repone would be expected. owever for uing PID control, low repone would have to be et to obtain water ytem tability and thi type of ulation wa therefore not applicable. The idea for level ulation wa to implement a flow control loop, where Toro 3 unit would compenate the Toro dicharge []. A problem for level control in thi cae i la Flor weir which ha no dicharge meaurement intalled. Neverthele, we decided to ue the flow control at Toro 3 PP a for level ulation of the intake reervoir. The La Flor intake flow i calculated a the difference between meaured flow at Toro and turbinated flow for table level ulation. La Flor weir intake flow ha a mall change rate in time, o above aumption howed to be ufficient. So calculated La Flor intake flow wa additionally averaged for 10 min to filter-out tep change in control loop. Additional compenating flow wa implemented to make poible a level reference change and to ulate the actual reervoir level at deired level. A example - if actual level i lower a reference value, the Toro 3 flow i to be decreaed. To et the flow reference for Toro 3 unit, we combined meaured flow at Toro, calculated flow at la Flor weir and compenation flow for making the level control poible and accurate. Such flow and level control loop were initially implemented in the governor, however the repone wa poor. The control loop howed to be fairly untable, forcing Toro 3 unit to load and unload continuouly in ome cae even over full unit operating range. Since thi model wa already uccefully implemented on other Franci unit, we invetigated the difference between the project and came to concluion which later led u to improve the control loop and outcome the problem. Baic difference between the project where the ame level-flow control loop wa implemented are that at Toro 3 PP the intake reervoir i ignificantly maller and that pentock length i longer. 4. CORRECTION FACTORS FOR LEVEL CONTROL To olve the level control intability problem, ite meaurement were performed in gate opening control at firt. After tabilizing the plant operation at table reervoir level and table water preure before the unit without any ignificant ocillation, a tep change in gate opening wa performed. Meanwhile, intake reervoir level and urge tank level were monitored. The reult howed, that after the tep change in gate opening urge tank level changed practically at the ame time, however the intake reervoir level changed only after. A urge wave traveling time propagate with velocity app 950 m/ in the pentock and 100 m/ in the tunnel [1], while the level ocillation between urge tank and the intake reervoir i a function of volumetric flow between the reervoir. Slow ocillatory dumped water flow between the reervoir reult in an apparent time lag in the reervoir level change a preented on meaurement of load rejection on Fig. [3]. The ocillatory dumped wave i actually a flow of the water due to the level difference between the reervoir level and urge tank. In cae when turbine flow i changed, the level in urge tank conequently change (eg. when turbine flow decreae the urge tank level rie above the quai-tationary level for a certain flow in pentock). Changed level condition influence the flow in tunnel. If we aume that turbine flow i contant but the urge ocillation are preent in the hydraulic ytem, there i a certain dumped wave ocillatory flow preent between the urge tank and the intake reervoir. Thi additional ocillatory flow can be directed upward or downward the tunnel: tunnel t o o (1) A the tunnel ocillatory flow change with time there i immediate repone alo on the reervoir level, which in Toro 3 cae i mall and the effect on the reervoir level i ignificant.

3 September 013 Page 3 n- peed of rotation Pgen y- wicket gate poition (cloing) dt WL t urhe tank water level Fig. Surge tank and intake reervoir level repone to full load-rejection The reervoir level urface area i only about 10 time larger a urge tank water urface area, o it i expected and alo proved by meaurement, that level ocillation due to ocillatory dumped wave i preent alo in the intake reervoir. The volume of the ocillatory flow i ditributed between urge tank and reervoir. In thi way, expected level deviation in the intake reervoir i 1/10 th of the level deviation in the urge tank. Thi phenomenon i additional problem in level control a even if gate opening i contant, the level in reervoir i changing a long there i ocillatory wave preent between urge tank and reervoir Intake reervoir level ocillation correction factor A correction factor k wa introduced to compenate urge tank / intake reervoir level ocillation for level-control flow reference calculation at Toro 3. The correction factor k repreent the influence of the urge tank level t. The t_ q( T 3 deviation from the quai-tationary urge tank level which i a function of the pentock loe ) urge head deviation influence the tunnel flow and conequently the intake reervoir water level. k f ( 3) t t _ q T () k WL t t _ q( T 3) t _ q( T 3) The quai-tationary urge pentock loe between intake and urge tank ) are etablihed at a certain t _ q( T 3 pentock dicharge after urge ocillation diappear. A function wa defined by meaurement and preented in a form of a polynomial to be ued in calculation. (3) 4.. Toro flow meaurement correction factor Toro dicharge i meaured by ultraonic flow meter intalled in Toro pentock. The flow ignal itelf i filtered / averaged, o the information in the Toro 3 level control loop i not repreenting actual flow at all the time. In cae, where there are flow change on Toro, the true value of the dicharge i hown only after the flow i tabilized in the new value for a certain period of time. Meanwhile, meaured flow lag behind the true value due to the averaging of the meaured ignal. In cae of loading or unloading of the Toro unit from or to full power, the untrue value of the dicharge can be preent for few minute, making problem in the level control loop. Therefore a Toro flow meaurement correction flow corr _ filter wa introduced a a function of Toro meaured flow change rate dt and filtering time of the initial meaurement t filter : corr _ filter t dt filter (4)

4 September 013 Page 4 When the flow change rate i not equal to 0, eg. it i poitive, the flow value received from Toro ultraonic meaurement i lower a a true dicharge. By adding the correction flow to Toro flow meaurement we obtain a flow value which i then ued for level ulation Toro 3 over-pilling at the intake In cae Toro inflow i higher a Toro 3 conumption, there could be a pilling preent at the Toro 3 intake reervoir. The over-pilling dicharge a a factor of pillway width L and an overpill water level d wa etimated a: pill 0, 65 d g L (5) 4.4. Calculation of compenating flow Compenating flow define the flow compenation in a way that taking into account intake and outtake dicharge, the water level would be controlled toward the reference level. By applying the compenating flow, a level change ramping wa applied and limited to acceptable value - at Toro 3 PP maximal level ramp wa defined to R=0,m/min. The actual ulation ramp r (d) i dependant of the reervoir level deviation from the reference value d: r( d) f ( d) (6) Compenation flow i than calculated taking into account intake reervoir water level: r( d) A reervoir (7) Calculation of ulation ramp compenate the inaccuracie in Toro flow meaurement or La Flor flow calculation, thu making the loop inenitive to maller uncertaintie in meaurement or flow calculation. If the flow meaurement uncertaintie are ignificant, thi would reult in the level deviation, however the control loop would maintain table Reference flow calculation for Toro 3 level-control loop The reference flow wa initially determined a function of intake dicharge at Toro T, La Flor LFl, overpilling flow at the intake and reervoir level deviation WL. pill, compenating flow ref ) ( (8) T LFl pill Correction were introduced to compenate all above mentioned phenomena, o the final reference flow for Toro 3 unit wa defined a: T 3 ( ref corr _ filter ) k (9) A factor k v ha been ranged to compenate the reference flow however limited to ±10 % of the initially calculated flow. Flow control wa deigned a PID control where reference flow wa calculated a per equation (9) and compared to meaured unit flow on both unit. Unit flow were meaured by Winter Kennedy method [4].

5 September 013 Page 5 5. GROUP CONTROLLER Group controller control the two Toro 3 unit in a way that it can tart, top or ditribute load of flow between the unit. Depending on the intake flow form Toro PP, unit at Toro 3 are programmed to tart and top automatically. When group controller i in level ulation mode, which i a uual etting for Toro 3 PP, the controller firt calculate the reference flow a preented in chapter 0. Group controller ditribute the reference flow between the two unit according to the deigned flow ditribution cam a on Fig. 3 [3]. Prepared by: Damir Dolenc Dec / 16 / 01 Toro 3 - operating CAM 1 _unit_max _cam1_max 10 9 unit flow: 1, [m3/] 8 7 "CAM1" "CAM" _cam1_min 3 _cam_min 1 _cam_low (Toro3) = 1 + [m3/] Fig. 3 Group control flow-ditribution CAM A certain flow ditribution CAM hyterei wa deigned to protect the unit againt unneceary unit tart and top Group level control tability The group level control howed to have fat repone a well a a remarkable tability for uch a complex layout a at Toro 3 PP. Meaurement taken during everal day of Toro cacade operation howed that water level at Toro 3 intake reervoir i kept within ±0,5 m when unit are in operation. One day of plant operation in level control i preented on Fig. 4 [3]. Date of meaurement: Febrery 5th 013 Data received from SCADA Supplied by Siemene Group level control at Toro 3 PP One day of operation 100 Pgen T_u - Gen. power Toro, unit Pgen T_u - Gen. power Toro 3, unit _T3 - meaured flow at Toro 3 Level ocillation due to flow ocillation between urge tank and intake level reervoir _T - meaured flow at Toro 80 WL - intake reerv. water level 70 69, ,81 m.a..l. (-19cm) ,5 y_u ,5 ΔPgen = -6,3MW (-0,7MW/min) Pgen T_u Pgen T3_u 0 689, _T _T3 0 6:14:4 688,5 8:38:4 11:0:4 13:6:4 15:50:4 18:14:4 0:38:4 Time [h:min:] Fig. 4 Group level control of the Toro 3 unit over one day 3:0:4 Level - WL [m.a..l.] di. opening - y [%], gen power - Pgen [MW], flow - [m3/] 693,5 y_u - ditributor opening Toro 3 unit WL

6 September 013 Page 6 Load rejection at Toro PP have immediate repone with Toro 3 unit, which i achieved only by flow meaurement loop without taking in account the generating data which were otherwie available to ue. When only one unit i rejected at Toro, alo only one unit at Toro 3 goe to peed no-load waiting certain amount of time until Toro rejected unit would be connected to grid and loaded again. If thi doen t happen in minute, the redundant unit at Toro 3 top a well. 6. CONCLUSIONS Toro 3 PP wa uccefully delivered into trial operation at the end of 01 with remarkable level tability within few centimeter in table operation and le than 0,5 m deviation during Toro change in plant output. With the implemented level control Litotroj Power wa able to raie level reference for m above deign value and adjut the level only 1m below the pill elevation without any over pilling during operation of unit in level control. Increae of the operating level reference increaed the total plant efficiency / output cloe to 1 % comparing to deigned value. With all above tated characteritic Litotroj Power Unit Control Sytem howed advanced and ophiticated control capabilitie and a uch repreent an excellent reference project both for level control application a well a for Unit Control Sytem. Reference 1. J. Mazij, A. Bergant ydraulic Tranient Analyi for Toro 3 PP, Litotroj Power, Ljubljana, 011, in Spanih. D. Dolenc, Intake Level Control uing flow control, Litotroj Power, Ljubljana 011, Litotroj Power internal documentation, in Slovenian. 3. D. Dolenc, Commiioning report for Toro 3 PP, Report no. 1773, Litotroj Power, Ljubljana, 013, in Spanih 4. IEC Field acceptance tet to determine the hydraulic performance of hydraulic turbine, torage pump and pump-turbine, International Electrotechnical Commiion, 1991