Optimal Control of Waste Heat Recovery Systems Applying Nonlinear Model Predictive Control (NMPC)

Transcription

1 Optimal Control of Waste Heat Recovery Systems Applying Nonlinear Model Predictive Control (NMPC) Philipp Petr, Christian Schröder, Prof. Dr.-Ing. Jürgen Köhler, Dr. Manuel Gräber ASME ORC rd Seminar on ORC Systems, Brussels, October 14 th 2015

2 Waste Heat Recovery System in a Long Distance Bus Total vehicle model (thermal, longitudinal dynamics) ASME ORC Philipp Petr Slide 2

3 Waste Heat Recovery System in a Long Distance Bus Modelled ORC Concept (design stage) Working fluid: Ethanol Evaporator type: Fin-and-Tube Expander type: Effiency Based Condenser type: Tube-and-Tube ASME ORC Philipp Petr Slide 3

4 Waste Heat Recovery System in a Long Distance Bus Control concept Expander inlet pressure controlled by expander speed Expander inlet enthalpy controlled by pump speed ASME ORC Philipp Petr Slide 4

5 Why Do We Need Advanced Control Strategies? 1. Transient heat source temperature and mass flow rates ASME ORC Philipp Petr Slide 5

6 Why Do We Need Advanced Control Strategies? 1. Transient heat source temperature and mass flow rates 2. Interactions between different subsystems ASME ORC Philipp Petr Slide 6

7 Why Do We Need Advanced Control Strategies? 1. Transient heat source temperature and mass flow rates 2. Interactions between different subsystems 3. Predicted states offer futher potential for energy recovery ASME ORC Philipp Petr Slide 7

8 Why Do We Need Advanced Control Strategies? 1. Transient heat source temperature and mass flow rates 2. Interactions between different subsystems 3. Predicted states offer futher potential for energy recovery 4. ORCs shows a high grade of nonlinear behavior in transient operation Linear approaches not feasible in all operating conditions Nonlinear approaches are beneficial, but complex Nonlinear Model Predictive Control (NMPC) is one method to take this challenge NMPC is a repetetive solving of an optimal control problem for finite prediction horizons ASME ORC Philipp Petr Slide 8

9 Brief Overview on Presented Research Development of a transient mathematical long-distance bus model with a waste heat recovery system Development of a software tool chain for NMPC Development of a differentiable High-Speed Model of the ORC for NMPC Virtual test drive in the European Transient Cycle to test the concept ASME ORC Philipp Petr Slide 9

10 Block Diagram of the Nonlinear Model Predictive Control NMPC High-Speed- Model Controlled System Optimization Control Variables u State Variables x Sophisticated ORC-Model ASME ORC Philipp Petr Slide 10

11 Block Diagram of the Nonlinear Model Predictive Control NMPC Nonlinear fast system model High-Speed- Model Controlled System Optimization Control Variables u State Variables x ASME ORC Philipp Petr Slide 11

12 Block Diagram of the Nonlinear Model Predictive Control Target function Constraining conditions Exhaust gas enthalpy flow rate NMPC High-Speed- Model Optimization Control Variables u Computation of the optimal control variable trajectory Controlled System State Variables x ASME ORC Philipp Petr Slide 12

13 Software Tool Chain Target function Constraining conditions Exhaust gas enthalpy flow rate NMPC DYMOLA TILMedia FMI Suite ORC High-Speed Model Controlled System DYMOLA TILMedia TISC Optimizer TISC Sophisticated ORC-Model ASME ORC Philipp Petr Slide 13

14 Computation Time Differential and algebraic states Cycle Time (ETC) ComputationTime (Intel Core 3.40GHz) min min 5 s High-Speed- Model Sophisticated ORC-Model Sophisticated ORC-Model High-Speed- Modell ASME ORC Philipp Petr Slide 14

15 Benchmarking NMPC in Partial Load Conditions System is shut down in partial load conditions due to low mass flow rates Constant Set Point Steady State Optimized Set Points Linear control approach. Gain scheduled controller parameter developed with AMIGO approach Prediction Horizon: 4s (real-time capable) Nonlinear Model Predictive Control ASME ORC Philipp Petr Slide 15

16 Results of the Virtual Test Drive (Urban Section of the European Transient Cycle) Expander inlet pressure ASME ORC Philipp Petr Slide 16

17 Results of the Virtual Test Drive (Urban Section of the European Transient Cycle) Expander inlet enthalpy ASME ORC Philipp Petr Slide 17

18 Results of the Virtual Test Drive (Urban Section of the European Transient Cycle) Expander power ASME ORC Philipp Petr Slide 18

19 Results of the Virtual Test Drive (Urban Section of the European Transient Cycle) Pump work ASME ORC Philipp Petr Slide 19

20 Results of the Virtual Test Drive (Urban Section of the European Transient Cycle) Higher net power output due to (optimized) ORC part load operation 7% + 8 % + 15 % ASME ORC Philipp Petr Slide 20

21 Conclusion and Outlook Implementation of advanced control strategies are necessary for small ORC systems operating under transient boundary conditions Development of a software tool chain to realize a prototype NMPC Development of an ORC High-Speed Model Virtual Test Drive of a long distance bus proved the potential of NMPC in the part load section of the European Transient Cycle (ETC) Outlook Improvement of the High-Speed Model regarding computational time and accuracy Implementation of physically motivated expander models Proof of concept by means of an ORC test rig ASME ORC Philipp Petr Slide 21

22 Optimal Control of Waste Heat Recovery Systems Applying Nonlinear Model Predictive Control (NMPC) Philipp Petr, Christian Schröder, Prof. Dr.-Ing. Jürgen Köhler, Dr. Manuel Gräber ASME ORC rd Seminar on ORC Systems, Brussels, October 14 th 2015

23 Contact Information Philipp Petr Mail. Tel. +49 (0) Dr.-Ing. Wilhelm Tegethoff Mail. Tel. +49 (0) Technische Universität Braunschweig Institut für Thermodynamik Hans-Sommer-Str Braunschweig Germany TLK-Thermo GmbH Hans-Sommer-Str Braunschweig Germany ASME ORC Philipp Petr Slide 23