TOWARDS THE OPTIMAL OPERATION OF AN ORGANIC RANKINE CYCLE UNIT BY - PowerPoint PPT Presentation

TOWARDS THE OPTIMAL OPERATION OF AN ORGANIC RANKINE CYCLE UNIT BY MEANS OF MODEL PREDICTIVE CONTROL Andres Hernandez*, Adriano Desideri, Clara Ionescu, Sylvain Quoilin, Vincent Lemort and Robin De Keyser Electrical energy, Systems and Automation Thermodynamics laboratory Ghent University, Belgium University of Liège, Belgium 1

Outline 1. Motivation 2. Organic Rankine Cycle 3. Optimal conditions 4. Control strategies a. PI, SW-PI and MPC. 5. Results 6. Conclusion 2

Energy Efficiency in Industry In the In the product Energy product Energy Low-temperature Waste heat Low-temperature Recovery Waste heat system Energy Losses Losses recovered Challenge: System dynamics change due to the fluctuating nature of the waste heat Process Optimal Control Motivation Results Conclusions 3 description conditions strategies

Outline 1. Motivation 2. Organic Rankine Cycle 3. Optimal conditions 4. Control strategies a. PI, SW-PI and MPC. 5. Results 6. Conclusion 4

Organic Rankine Cycle Organic Rankine Cycle Operate in safe conditions Waste heat Maximize the output power Superheating ∆𝑈 𝑡ℎ = 𝑈 𝑓𝑦𝑞 , 𝑗𝑜 − 𝑈 𝑡𝑏𝑢 , 𝑓𝑤 Evaporating temperature 𝑈 𝑡𝑏𝑢 , 𝑓𝑤 = 𝑔 ( 𝑞 𝑡𝑏𝑢 , 𝑓𝑤 ) Process Optimal Control Motivation Results Conclusions 5 description conditions strategies

Test-rig Organic Rankine Cycle 1. Heating loop: Electrical Boiler (therminoll66) 2. Cooling loop: glycol-water system 3. Regenerative cycle 4. Working fluid: SES36 5. Expander capacity 11Kwe (single screw) (Ghent University - Campus Kortrijk) Process Optimal Control Motivation Results Conclusions 6 description conditions strategies

Simulation environment Dymola/Modelica: Matlab: System dynamics and fluid Model Predictive Control ( MPC ) properties and optimization algorithm. FMI toolbox Control algorithm is then compiled as a ‘ dll ’ and linked to the current Labview interface. Process Optimal Control 7 Motivation Results Conclusions description conditions strategies

Model validation Adriano Desideri et.al. COMPARISON OF MOVING BOUNDARY AND FINITE-VOLUME HEAT EXCHANGERS MODELS IN THE MODELICA LANGUAGE, ASME ORC 2015, Brussels, Belgium Process Optimal Control Motivation Results Conclusions 8 description conditions strategies

Control variables Disturbances ṁ hf T hf ∆T sh N pp Inputs Outputs T sat,ev Process Optimal Control Motivation Results Conclusions 9 description conditions strategies

Outline 1. Motivation 2. Organic Rankine Cycle 3. Optimal conditions 4. Control strategies a. PI, SW-PI and MPC. 5. Results 6. Conclusion 10

The concept of an optimized control Optimized control: Basic Control: Main optimization parameter: Evaporating temperature Process Optimal Control Motivation Results Conclusions 11 description conditions strategies

Relationship between power and controlled variables a) Outpower as function of evaporating temperature b) Outpower as function of superheating 4500 4500 T hf =125 °C T hf =125 °C 2000 2000 4000 4000 T hf =110 °C T hf =110 °C T hf =100 °C T hf =100 °C 1900 1900 3500 3500 T hf =90 °C T hf =90 °C Pump Speed - N pp [rpm] Pump Speed - N pp [rpm] 3000 3000 1800 1800 2500 2500 W exp [W] W exp [W] 1700 1700 2000 2000 1600 1600 1500 1500 1500 1500 1000 1000 500 1400 500 1400 0 0 70 80 90 100 110 120 0 10 20 30 40 50  T sh [°C] T sat,ev [°C] 𝑛 ℎ𝑔 = 1.0 𝑙𝑕/𝑡 Process Optimal Control Motivation Results Conclusions 12 description conditions strategies

Optimal evaporating temperature (Static Map) 𝑈 𝑡𝑏𝑢,𝑝𝑞𝑢 = −290.915 + 183.33 ∗ 𝑚𝑝𝑕 10 𝑈 ℎ𝑔 + 10.636 ∗ 𝑛 ℎ𝑔 𝑛 ℎ𝑔 = {1.5 ; 1.0 ; 0.5} 𝑙𝑕/𝑡 Valid in the range between 0.5 < m hf < 1.5 kg/s and 90 < T hf < 125 °C, for P sat,cd =1.4 bar Process Optimal Control Motivation Results Conclusions 13 description conditions strategies

Extremum Seeking (ES) algorithm 1) Modulation phase The adaptation signal shifts the sine wave towards the gradient direction 2) Obtaining the gradient direction 3) Computing the adaptation law Process Optimal Control Motivation Results Conclusions 14 description conditions strategies

Outline 1. Motivation 2. Organic Rankine Cycle 3. Optimal conditions 4. Control strategies a. PI, SW-PI and MPC. 5. Results 6. Conclusion 15

Control architecture Controllers: 1. PI 2. Switching PI 3. Model Predictive Control Process Optimal Control Motivation Results Conclusions 16 description conditions strategies

Control Design Switching PI strategy EPSAC - MPC 𝑂 2 𝑠 𝑢 + 𝑙 𝑢 − 𝑧 ( 𝑢 + 𝑙 | 𝑢 ) 2 v = 𝑙 = 𝑂 1 NO min U If DT sh > DT sh,min Subject to: 𝑧 + = 𝑔 𝑧 , 𝑣 YES 𝑉 𝑛𝑗𝑜 < 𝑉 < 𝑉 𝑛𝑏𝑦 PID T sat PID DT sh 𝑍 𝑛𝑗𝑜 < 𝑍 < 𝑍 𝑛𝑏𝑦 SP = T sat,opt SP = DT sh,min ∆𝑉 𝑛𝑗𝑜 < 𝑉 𝑢 − 𝑉 𝑢 − 1 < ∆𝑉 𝑛𝑏𝑦 Ui T sat = Ui DT sh Ui DT sh = Ui T sat N 1 =1, N 2 =15, N u =1 End U min = 1320 rpm U max = 2100 rpm ∆U = 100 rpm/s Process Optimal Control Motivation Results Conclusions 17 description conditions strategies

System Identification Identification experiment: Saturation temperature Identification experiment: Superheating Data; measured 2.5 Data; measured sys; fit: 88.71% 6 sys; fit: 71.24% 2 4 1.5 1 2 Temperature [°C] Superheating [°C] 0.5 0 0 -2 -0.5 -4 -1 -6 -1.5 -2 -8 -2.5 -10 200 400 600 800 1000 1200 1400 1600 1800 2000 200 400 600 800 1000 1200 1400 1600 1800 2000 Time [s] Sampling time T s = 1s Process Optimal Control Motivation Results Conclusions 18 description conditions strategies

Outline 1. Motivation 2. Organic Rankine Cycle 3. Optimal conditions 4. Control strategies a. PI, SW-PI and MPC. 5. Results 6. Conclusion 19

Heat Source conditions Heat source conditions Heat sink conditions 130 25 125 20 T hf (°C) T cf (°C) 120 15 115 110 10 100 200 300 400 100 200 300 400 1.5 4 3.5 M hf (kg/s) M cf (kg/s) 1 3 2.5 0.5 2 100 200 300 400 100 200 300 400 Time (s) Time (s) Process Optimal Control Motivation Results Conclusions 20 description conditions strategies

Control performance for MPC strategy Control performance for MPC strategy 20  T sh (°C) 10 0 100 150 200 250 300 350 400 450 110 T sat,ev (°C) ES 100 Map 90 100 150 200 250 300 350 400 450 N pp (rpm) 1800 1600 100 150 200 250 300 350 400 450 5 W exp (kW) 4 3 100 150 200 250 300 350 400 450 Time (s) Process Optimal Control Motivation Results Conclusions 21 description conditions strategies

Control performance for PI strategy Control performance for PI strategy 20  T sh (°C) 10 0 100 150 200 250 300 350 400 450 110 T sat,ev (°C) ES 100 Map 90 100 150 200 250 300 350 400 450 N pp (rpm) 1800 1600 100 150 200 250 300 350 400 450 5 W exp (kW) 4 3 100 150 200 250 300 350 400 450 Time (s) Process Optimal Control Motivation Results Conclusions 22 description conditions strategies

Control performance for SW-PI strategy Control performance for Switching PI strategy 20  T sh (°C) 10 0 100 150 200 250 300 350 400 450 110 T sat,ev (°C) ES 100 Map 90 100 150 200 250 300 350 400 450 N pp (rpm) 1800 1600 100 150 200 250 300 350 400 450 5 W exp (kW) 4 3 100 150 200 250 300 350 400 450 Time (s) Process Optimal Control Motivation Results Conclusions 23 description conditions strategies

Performance comparison Controller Net Energy Advantages Profit produced Switching PID 2.79 kWh Safe operation 100 % MPC 3.37 kWh Safe and optimal 120 % operation Process Optimal Control Motivation Results Conclusions 24 description conditions strategies

Experimental results using correlation as optimizer Basic PI Basic MPC Switching PI Optimal MPC 40 40  T sh [°C]  T sh [°C] 40 40  T sh [°C]  T sh [°C] 20 20 20 20 0 0 0 0 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 T sat,ev [°C] T sat,ev [°C] T sat,ev [°C] T sat,ev [°C] 95 95 95 95 90 90 90 90 85 85 85 85 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 2200 2200 2200 2200 N pp [rpm] N pp [rpm] N pp [rpm] N pp [rpm] 2000 2000 2000 2000 1800 1800 1800 1800 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 Time [s] Time [s] Time [s] Time [s] Unsafe 115% 100% 121% Process Optimal Control Motivation Results Conclusions 25 description conditions strategies

Conclusions and perspectives 1. There exists an optimal evaporating temperature which maximizes the output power given some heat source conditions 2. Small amount of superheating is necessary to guarantee stable operation of the ORC unit. 3. MPC is able to produce the highest net output power since it operates closer to boundary conditions, due to a better constraint handling. 4. Explore the influence of the expander´s speed on the system. Process Optimal Control Motivation Results Conclusions 26 description conditions strategies