Simulator Development of 1000MW Class Ultra Super Critical Coal-Fired Power Plant with advanced control algorithm

Transcription

1 Simulator Development of 1000MW Class Ultra Super Critical Coal-Fired Power Plant with advanced control algorithm Ki-Yong Oh Energy & Mechatronics Group, Strategy Technology Laboratory, KEPRI (Korea Electric Power Research Institute) Science Town, Daejeon , Korea and Geon-Pyo Lim and HoYol Kim I&C Group, Plant Generation Laboratory, KEPRI (Korea Electric Power Research Institute) Science Town, Daejeon , Korea ABSTRACT Even though efficiency of coal-fired power plant is proportional to operating temperature, increase of operating temperature is limited by a technological level of each power plant components. It is an alternative plan to increase operating pressure up to ultra super critical (USC) point for efficiency enhancement. It is difficult to control the system in that pressure within safety guideline that many unexpected phenomena are happen because that region is highly nonlinear region. Also it is another reason for hard to control that power plant is a combination of many complicated components and controllers. In this paper, Advanced Process Control algorithm (APC), ARX and Fuzzifier, is introduced. Then control logic applied Unit step Response Optimizer, which is combination of ARX and Fuzzifier, is designed. Its performance is tested and analyzed with design guide line. Keywords: robust control, advanced process control algorithm (APC), ARX, Fuzzifier, fuzzy logic, ultra super critical (USC), coal-fired power plant, Simulator 1. INSTRODUCTION As global oil and coal price are increased drastically, efficiency enhancement of thermal power plant is more important for cost reduction in power industrial field. Also it is a hot issue to enforce Kyoto Protocol to the United Nations Framework Convention on Climate Change 1) in electric industrial field. So pollutants like as NO X, SO X, CO 2 and dust should be decreased. The amount of Pollutants decreased is proportional to the amount of fuel consumption decreased. So efficiency of coal-fired power plant is important because the amount of fuel consumed is heavily related with efficiency of that. Even though efficiency of coal-fired power plant is proportional to operating temperature, operating temperature is limited by a technological level of each power plant components as durability, thermal resistance of plant components. It is an alternative plan to increase operating pressure up to ultra super critical point for efficiency enhancement instead of increasing operating temperature. Despite many advantages as high efficiency, low cost and eco-friendly, it is difficult to operation in that condition because system is hard to control in that region. In this paper, robust controller is proposed to control the USC power plant. Also 1000MW class USC power plant specification and total procedure of simulator development is described sequentially. Then control logic design and gain tuning are described. Its performance is simulated and analyzed with design guide line MW Class USC Power Plant 1000MW Class USC Power Plant is designed that generation power is 1100MWe in valve wide open (VWO) state and 1024 MWe in maximum guaranteed rate (MGR) state. Main steam pressure 265kg/cm 2, main steam temperature 610 and Reheat steam temperature 621. These specifications are chosen to USC power plant considering the present condition of material thermal resistance, economical efficiency, a technological level of power plant manufacturer. Its main/reheat steam temperature is very high in comparison with existing super critical power plant because main/reheat steam temperature of existing super critical power plant are 565~593 /565~593. So Efficiency of USC power plant is anticipated about 45%. It will result in lower annual fuel consumption levels and pollutants. So USC power plant could be high efficiency, low cost and eco-friendly coalfired power plant. More information is displayed in table 1. Overall Plant Gross electrical output 1100 MWe (VWO) 1024 MWe (MGR) Main steam pressure 265 (kg/cm 2 ) Main steam temperature 610 ( ) Reheat steam temperature 621 ( ) Steam Generator Main steam flow 2942 (ton/h) Main steam pressure 274 (kg/cm 2 ) Main steam temperature 613 ( )

2 Feedwater temperature ( ) Temperature control HP 3 spray attemperators HP Superheater stages 4 superheater stages RH steam pressure 49.9kg/cm 2 g at BMCR (kg/cm 2 ) RH steam temperature 624 ( ) Damper control RH control 1spray for emergency RH steam bypass RH stages 2 reheater stages Flue gas temperature downstream of air heater 143 ( ) Steam Turbine Type Tandem Compound Speed 3600 (rpm) HP Turbine Single flow turbine IP Turbine Double flow turbine LP Turbine Two double flow turbine sections HP steam conditions 265 / 610 (kg/cm 2 / ) IP steam conditions 46.7 / 621 (kg/cm 2 / ) LSB Length 1143 (mm) Condenser pressure 1.5 (inhga) Generator Type 3 phase, Synchronous Capacity 1222 (MVA) Cooling medium Rotor/Hydrogen, Stator/Water Power factor (cosφ) 0.9 Emission Control Nitrogen oxide (ppmdv, O 2, 6%) Less than 40 with SCR System Sulfur dioxide (ppmdv, O 2, 6%) Less than 40 with FGD System Dust (mg/sm 3, O 2, 6%) Less than 40 with EP & FGD System <Table 1> Design Specification of 1000MW Class USC Power Plant Temperature change limit of USC power plant is designed with table 2. This guideline is applied to existing 500MWe super critical power plant in Dangjin. Even though large temperature fluctuation would be happen because load change rate is large, USC power plant is designed with same guideline of existing power plant. So advanced process control algorithm which is consisted of Auto Regressive with exogenous input (ARX) and Fuzzifier should be applied to USC power plant for reliability and stability. State Main steam Reheat steam Temp. Sec. Temp. Sec. Constant load ±3 - ±3-30~40%NR ± %~ ±8 900 ± Step change 50%~ ± ± GOV free ±8 - ±8 - <Table 2> Guideline of Temperature Fluctuation 2) 3. Control Logic Design Control Logic Design and Tuning: Control logic design is based on PID controller. But it is difficult to stabilize the main/reheat temperature and other components with the guideline because dynamic load change range is larger than that of existing power plant. The first work is control logic design of all part and each part gain tuning. Control logic is designed based on existing power plant. But APC is added to total control logic for enhancement of the reliability. Gain tuning is executed based on Ziegler-Nichols method 3). But it can t be guaranteed that Optimal Tuning is executed because each part is tuned in steady state condition while another part is kept ideal value in passive mode. So robust controller should be add to control logic to optimize the system for stability and efficiency. Unit Step Response Optimizer (URO) Mode: There are three modes in existing power plant. One is boiler following mode, another is turbine following mode and the other is coordinate mode. URO mode is newly added in USC power plant. A drawing of URO control logic is shown in figure 1. URO mode consists of many ARX, PID controller and Fuzzifier. It is classified turbine control part and boiler control part. Turbine side controls Megawatt. PID controller of this side acts as a trim that ensures the megawatts are controlled to set point. Boiler side controls throttle pressure. This side classified feed forward section (A) and feedback section (B). Section (A) consists of two ARX, which are called slow kicker and fast kicker, and Fuzzifier. Fuzzifier combines the fast and slow kicker signal to create the control signal based on megawatt error. Three ARX of section (B) are transfer function of boiler dynamics via Megawatt, throttle pressure and Superheater Temperature. PID controller of this side is a trim to ensure throttle pressure is controlled to set point. <Figure 1> Drawing of URO Control Logic ARX algorithm: Two kind of ARX algorithm are designed in USC power plant. One is slow and fast kicker in figure 2. ARXs of section A in figure 1 are this algorithm designed from the load index to the boiler. Each response function is

3 designed for slow and fast response. These kicker s signals are combined by Fuzzifier to optimize the load change. There are no rules to select these kickers. In this research, Generating megawatts are tested to check the best kicker using the rule of trial and error. <Figure 2> Slow and Fast Kicker The other is constructed based on the transfer function of system dynamics. That is used to control transient response in load change. At first, each system dynamics is measured with unit step input in time domain. Then system dynamics are modeled based on measured unit step response of system. Each model is designed with 2~4 order transfer function because there is no physical meaning in high order model. (a) Turbine Master Demand vs. Megawatts TMMW s = s + 1 e 2s 10s s + 1 (b) Turbine Master Demand vs. Throttle Pressure TMTP s = 2 30s + 1 e 2s s s e 2s TMST s = 100s s s + 1 Modeled transfer functions with turbine master demand are shown in (1). TMMW Megawatts via turbine master, TMTP Throttle Pressure via turbine master, TMST Superheater Temperature via turbine master. Figure 3 shows dynamic responses of system and constructed model correspond with dynamic response. It is used with PID controller to change the set point. Error between set point and state value is decreased with this ARX algorithm because each set point value is calculated by expected system dynamics. So these are one kind of feed forward 4) elements to predict the set point. These transfer function are changed to discrete time transfer function 5). These transfer function is shown in (2). Sampling time is set 1second considering simulator capability. These are implemented to simulator as discrete time transfer function. System has high reliability as well as anti-windup 6) ability with this algorithm. General adaptive controllers 7) or nonlinear robust controllers is designed to change control gain or other parameter. But this algorithm change set point to reduce the error. TMMW z = z z 2 z z z 2 TMTP z = z z 2 z z z z z 2 z 2 TMST z = 10 7 z z z 2 (c) Turbine Master Demand vs. Superheater Temperature <Figure3> System Dynamics Modeling in Turbine Master ARX algorithms for Boiler and feedwater are designed with same method. These are shown in (3). BMMW s = BMTP s = 14e 30s 50s s + 1 (2500s + 1) 2.5e 30s 50s s + 1 (2500s + 1) 9.5e 30s BMST s = 100s s + 1 (1200s + 1) 3.5(1200s + 1)e 10s FWMW s = 100s s + 1 (450s + 1) 0.3(4500s + 1)e 10s FWTP s = 400s s e 10s FWMW s = 10s s + 1 (250s + 1) (3) Fuzzifier: Fuzzifier is designed with general fuzzy logic 8). It combines two ARX algorithms which are called slow kicker and fast kicker based on the megawatt error. Slow kicker has small overshoot while it has fast response speed.

4 BOILER PART SIGNAL OF APC Fast kicker has large overshoot while it has slow response speed. Fuzzifier combines response signals of two kickers or selects that of one kicker. Figure 4 is fuzzy theory in this Fuzzifier. Fuzzifier selects response of slow kicker in S.K. region, while it selects response of fast kicker in F.K. region. Also it combines responses of two kickers as equation (4) in L.K. region. So it is important to select proper pslow/pfast points for stability and performance enhancement. Designed Fuzzifier with two kickers is shown in figure 5. Pslow, pfast, slow kicker and fast kicker are chosen to 30, 150, 150s+1 300s+1 100s+1 in fine tuning by trial and error. 50s+1, OIS workstation 2 SET Model Development workstation Program Development workstation <Table 3> Simulator Hardware Construction DCS controller acquires process values of virtual power plant from power plant model server. Then DCS controller transfers operation and control values to model server after it executes control logic with process values in internal flash memory. Model Development workstation contains virtual USC power plant model. It calculates process values of virtual power plant with model and communicates with DCS controllers. EWS Developer supports DCS controllers, engineering works and operation display with DCS network. Process Control Logic Builder can develop control logic, upload control logic to DCS controller and download control logic from DCS controller. OIS workstation can operates virtual power plan like real power plant. It contains GUI. Details of GUI in OIS are described in below. <Figure 4> Fuzzifier L. K. = Diff pfast pfast pslow S. K. + Diff pslow pfast pslow SIMULATION RESULTS OF 1000MW USC CONTROL LOGIC URO BOILED MASTER BOILER MASTER ARX BOILER MASTER TARGER <Figure 5> Boiler Part of URO F. K. (4) <Figure 6> Schematic diagram of Simulators Graphic User Interface (GUI): Operators monitor and work with GUI in power plant. So Graphic display has a various function as alarm, event handling, log, mode change and etc. GUI with this function is added to simulator to realize the real power plant. All of GUI are designed for safety and easy operation based on existing power plant. So operators excise the power plant operation as well as control logic verification. Main display is shown in figure Simulator Development Simulator hardware set is shown in table 3. It consists of 6 DCS controllers, DCS DI O AI O and 7 workstations. Each component is linked with network. Schematic diagram is shown in figure 6. Hardware DCS Controller Unit DCS DI, DO, AI, AO Module Process Control Logic Builder workstation Instructor Station workstation EWS Developer workstation Quantity 6 SET <Figure 7> GUI of Unit Master Control

5 TEMPERATURE( O C) MEGAWATTS All of important data is linked in this main display. Other needed information is linked with other GUI. Also mode change is possible in this display. Other displays are designed like this. So operators can do all of work with these GUI whenever they need. Simulation Results: Simulations are executed with USC control logic. Results of load change are shown in Figure 8. Generator MW overshoot is 8.4MW in 800MW and 4.33MW in 1000MW. Maximum/minimum main temperatures are / Maximum/minimum reheater temperatures are / Differences from standard are /-3.48 in main steam and /-3.18 in reheat steam. Other components are moved with guide line. So reliability and stability of UCS power plant is verified in this simulation results. EMAMI-NAEINI, FEEDBACK CONTROL OF Dynamic Systems 4 th edition, Prentice Hall, 2002, pp 221. [4] Bernard Friedland, CONTROL SYSTEM DESIGN (An Introduction to State-Space Methods, Dover, 2005, pp238. [5] Katsuhiko Ogata, Discrete-Time Control Systems (Second Edition), Prentice Hall, 1995, pp [6] Pyung H. Chang and Suk H. Park, The Development of Anti-windup Scheme and Stick-Slip Compensator for Time Delay Control, Proceedings of American Control Conference, 1998, pp [7] JEAN-JACQUES E. SLOTINE, WEIPING LI, APPLIED NONLIEAR CONTROL, Prentice Hall, 1991, pp [8] F.A. Alturki, A. Abdennour, Design and simplification of adaptive Neuro-Fuzzy inference controllers for power plant, Electrical Power and Energy Systems, 1999, pp ACKNOWLEDGMENTS SIMULATION RESULTS OF 1000MW USC CONTROL LOGIC This work has been financially supported by R from Power R&D Program of ETEP (Electric power Technology Evaluation & Planning center). URO SIGNAL 800 GENERATOR MW FIANL TARGET MAIN STEAM REHEATER STEAM <Figure 8> Simulation Results 5. CONCLUSION In this paper, control logic design for ultra super critical coal-fired power plant is introduced. Advanced process control algorithm is proposed. These logics are applied to USC control logic for reliability and performance enhancement. Total control logics with APC algorithm are verified in simulator. Although load change rate is larger than existing super critical power plant, all of components are operating with a permissible range. 6. REFERENCES [1] Bjart Holtsmark, Cathrine Hagem, Emission Trading under the Kyoto Protocol, Center for International Climate and Environmental Research (CICERO), [2] H.Y. Kim, Control Logic Design of 1000MW Class Ultra Super Critical Power Plant (5 th Technical Report), KEPRI, 2007, pp27. [3] GENE F. FRANKLIN, J. DAVID POWELL, ABBAS