Real Time Simulation of Power Plants Torsten Dreher 1 System Simulation Group Friedrich-Alexander-University Erlangen-Nuremberg Siemens Simulation Center, Erlangen December 14, 2008 1 torsten.dreher@informatik.uni-erlangen.de
Outline Introduction to Thermal Power Plants Usage of Simulation in Thermal Power Plants Problems and Solutions (Examples) Conclusions
Outline Introduction to Thermal Power Plants Usage of Simulation in Thermal Power Plants Problems and Solutions (Examples) Conclusions
Introduction to Power Plants Usage of Simulation Problems and Solutions Thermal Power Plants Heat energy sources can be fossile, nuclear, biomass, waste, geothermal, solar etc. LEHRSTUHL FU R INFORMATIK 10 (SYSTEMSIMULATION) Conclusions
Scheme of Thermal Power Plant
Scheme of the Thermodynamic Cycle
The Thermodynamic Cycle
The Combined Cycle Power Plant (CCPP) ˆ efficiency up to 60% ˆ by gas or integrated gasification combined cycle (IGCC) ˆ also as singleshaft
Power Plant Automation
Outline Introduction to Thermal Power Plants Usage of Simulation in Thermal Power Plants Problems and Solutions (Examples) Conclusions
Usage of Simulation in Thermal Power Plants for ˆ the design of all major parts ˆ error detection ˆ error reconstruction ˆ design changes ˆ automation purposes
Introduction to Power Plants Usage of Simulation Problems and Solutions Conclusions Simulation for Finding the Design, e.g. Evaporator LEHRSTUHL FU R INFORMATIK 10 (SYSTEMSIMULATION)
One-dimensional mass-, momentum- and energybalance leads to system of partial differential equations (PDEs): ρ t = 1 A ṁ z ṁ t = 1 (ṁ2 A z ρ h t = ṁ A ρ Flow in the Evaporator ) ( p A ( 1 ρ p z h z + 1 ρ τ U A z + ρ g sin(ϕ) + τ U ) A ) + U q mf A ρ + 1 ρ p t
Spatial Discretization of the Evaporator Discretization leads to differential algebraic equations (DAEs): ρ 1 t = 1 A z (ṁ 1 ṁ in )
Solving the System Solve the DAE-system with an implicit method (Euler, Runge-Kutta etc.) Resulting matrix is sparse
More Simulation A story... ˆ a module detects an error ˆ an other one finds its reason ˆ the source of fault can be removed
Error Detection I Power output: Detected discrepancy between simulated value (blue) and metered value (red)
Error Detection II Correlates with loss of steam: Simulated (blue) and metered (red)
Finding the Physical Error
Reproducing the Source of Fault Analysis of instabilities in the evaporator Without deflector plates With deflector plates
Taking Measures in Designchanges
Simulation for the Automation System Simulation replaces physical process and several levels of automation; same automation used as in plant
For the Entire Lifecycle
Requirements Simulate ˆ the whole plant ˆ dynamically ˆ in real-time ˆ on a single PC ˆ with given accuracy ˆ in an easy to parameterize environment
Outline Introduction to Thermal Power Plants Usage of Simulation in Thermal Power Plants Problems and Solutions (Examples) Conclusions
Water-Steam-Properties IAPWS-IF97 implementation ˆ by International Association for the Properties of Water and Steam ˆ standard for industrial use ˆ too slow ˆ no derivatives ˆ numerical problems
Water-Steam-Properties: (Bi-)Cubic Spline Interpolation Problems with spline interpolation (1-D example) S 0 (x), x [x 0, x 1 ] S S(x) = 1 (x), x [x 1, x 2 ] S n 1 (x), x [x n 1, x n ] S j (x) =a j + b j (x x j )+ c j (x x j ) 2 + d j (x x j ) 3 j =0,..., n 1 match nodes, be twice continuous differentiable in nodes
Water-Steam-Properties: Numeric Problems Choice of the appropriate equation system A PDE-System with density, enthalphy and massflow as states requires the property p = p(ρ, h) A PDE-System with pressure, enthalphy and massflow as states requires the property ρ = ρ(p, h)
Water-Steam-Properties: Density-Enthalpy Pane p = p(ρ, h)
Water-Steam-Properties: Pressure-Enthalpy Pane ρ = ρ(p, h)
Increasing Performance Parallelization on Many-Core-Processors ˆ as known: not the clockspeed but number of cores will increase ˆ reason: power consumption increases exponentially with the clock frequency ˆ more low-frequency cores on a single chip ˆ Moore s Law still valid ˆ from quad-core to many-core processors ˆ gap between the slow memory access and fast CPU execution ˆ data locality and data communication ˆ parallel solver
Parameterizing the Simulation for the Automation System Problems ˆ not all physical parameters are known ˆ simple model due to real time requirement ˆ sensitivities of parameters are unknown Leads to an optimization problem
Dynamic Behaviour Transient operating conditions ˆ starting process lasts approximately 8 hours ˆ at several points the manual intervention of the operator ˆ big number of analogue values
Optimization Problem Minimize the difference between metered and simulated curves ˆ for any transition ˆ do not exceed 10% difference at any point ˆ meet static behaviour even more accurate
Approaches Problem: Time consuming evaluation of objective function ˆ parameterize single components (e.g. evaporator) first ˆ parametreize static behaviour first Problem: Huge number of parameters ˆ look only at sensitive parameters ˆ use heuristic optimization to get close to global minimum, then proceed with a local method Problem: Unlimited number of transient operating conditions ˆ use a model that is based on physics ˆ limit the parameters to physically reasonable values
Outline Introduction to Thermal Power Plants Usage of Simulation in Thermal Power Plants Problems and Solutions (Examples) Conclusions
Conclusions We have learned about simulation applications in power plants ˆ the design of all major parts ˆ error detection ˆ error reconstruction ˆ design changes ˆ automation purposes
Conclusions Take home message ˆ simulation is a wide field in industry ˆ not only for design ˆ several approaches only possible with latest or future computer technology ˆ there are several research fields with still unsolved ploblems Go, solve them!
Conclusions Take home message ˆ simulation is a wide field in industry ˆ not only for design ˆ several approaches only possible with latest or future computer technology ˆ there are several research fields with still unsolved ploblems Go, solve them!
Acknowledgements ˆ Prof. U. Ruede ˆ Siemens Simulation Center ˆ Siemens E F ES EN 72 ˆ Siemens E F ES EN 11 ˆ Siemens CT PP 2 Thank you for your attention!