A T sector SEMINAR Modeling Simulation and Control of

A&T sector SEMINAR Modeling, Simulation and Control of CERN cryogenic systems Benjamin BRADU CERN, EN-ICE CNRS, Laboratoire des Signaux et Systèmes Thursday 25 th March 2010

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 2

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 3

Motivations Develop a dynamic simulator for CERN cryogenic systems Model Large-Scale helium refrigerators n CERN cryogenic systems: large scale complex systems Similar to large industrial systems (Petroleum refineries, food industry , etc. ) ü LHC cryogenics : 42 000 I/O & 5 000 control loops ü n Non-linearity of helium properties (wide operation ranges) Temperature : 1. 9 K to 300 K ü Pressure : 14 mbar to 20 bar ü n Unique systems Built to be operated at nominal conditions ü Few information about transients and out of predefined operation points ü n Dynamic simulation is a good tool for : Train operators safely and in degraded conditions ü Test new control strategies without disturbing real operation ü Validate control and supervision systems in simulation : « Virtual Commissioning » ü 4

Multidisciplinary Approach Compute physical values In Space and Time Fluid Mechanics Thermodynamic Choose a set of PDEs Computer Science Parametric Identification Numerical implementation Mathematics Control Theory Validation with Experimental data Control Theory No New Control command developments / optimizations No Successful simulations Engineering Test on real plant 5

State of art Dynamic simulators of large-scale cryogenic systems Team Simulated Process Modeling Control Modeling Optimization R. Maekawa NIFS Japan Commercial liquefier + LHD Refrigerator 10 k. W @ 4. 5 K Physics DAE* (cryo lib) Partial High Pressure control with feed-forward action I. Butkevitch Kapitza insti. Russia Commercial liquefier for university education Mathematical Heuristic No No H. Quack UT Dresden Germany Commercial liquefier Physics DAE* (cryo lib) No No Physics DAE* (standard lib) No Pulse management for future tokamaks C. Deschildre 800 W refrigerator CEA/AL/Gipsa + France Commercial liquefier *Differential Algebraic Equations 6

PROCOS : Process & Control Simulator n CERN processes : ü ü ü n Commercial liquefier Medium and very large helium refrigerators @ 4. 5 K Refrigeration units @ 1. 8 K with cold-compressors Modeling tasks ü Use of equations from physics n ü Identification techniques n n Differential Algebraic Equation (DAE) with Ecosim. Pro Matlab Constraint : Simulation of control systems ü Use of existing control programs n ü PLC Schneider & Siemens (Programmable Logic Controller ) Use of existing CERN supervision system n PVSS 7

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 8

CERN control architecture for cryogénics Supervision controles (PVSS) Data Server SCADA (PVSS) Supervision Layer Réseau technique Ethernet / MOD ou S 7 Control Layer PLC Unity Schneider (PLC simulator) I/O (OPC) bus) I/O (Field Simatic Win. LC PLC (PLC Siemens simulator) I/O (OPC) bus) I/O (Field Process model (Ecosim. Pro) Field Layer OPC client C++ application C++ class Ecosim. Pro model Real process Ecosim. Pro algorithm Cryogenic Process Simulator (CPS) 9

Refrigerator operation Com pres Wa ter coo ler Water cooler sor C 1 1 T 1 Mechanical Work 2 C 1 1 bar HX 2 HX 3 re ne HX 2 rbi Low pre HX ssu 4 rb ine Tu HX 1 HX HX 3 Hig hp HX 4 Tu 15 bar 300 K HX 1 HX 5 T 2 HX 4 Mechanical Work HX 5 JT valve Expansion valve Joule-Thomson HX 5 5 K Heat Extraction 1 300 K HX res 2 sur e Temperature - Log (T) Electrical power 1 bar / 4. 5 K Q = m * Lv Thermal load Entropy - S Lv ~ 19 g/J for helium Thermal load 10

A component library for cryogenics Interface components Inputs/Outputs Analog/Digital Helium Thermodynamic properties (HEPAK tables) Storage components - pipes - Phase separators Calcul : P, hout Helium port P, h, m Ecosim. Pro Library Helium cryogenics Hydraulic components - Valves - Turbines -Screw compressors -Hydrodynamic compressors Calcul : m, hout PLC components Logical blocks Controllers (PID) Material Properties Stainless steel 304 L Aluminum 6061 T 6 Copper Thermal components -Thermal loads -Heat Exchangers Calcul : hout, Pout, min 11

Interconnections between components Rule 1 Hydraulic component Between 2 storage components Rule 2 Thermal component in series or between a storage component and an hydraulic component 12

Variational approach n Some parameters are difficult to find in large cryogenic systems Geometrical parameters (length, pipe shape, HX parameters, etc. ) ü Thermal parameters (insulation) ü “Secret” parameters (turbines) ü n Generally, manufacturers guarantee a nominal operation point : ü Static calculations performed by manufacturers n If X depends on an unknown constant K 1 : n Known design point : n Ratio between both: n Non-linearity of ‘f(P, T)’ are kept and unknown constants are removed

Example in Ecosim. Pro Example of a model in Ecosim. Pro Linde 18 k. W cold-box for the LHC 14

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 15

Component model validation n Individual validation Constant boundary conditions ü Transient checking ü Operation point checking ü n Validation of a reduced set of components ü Ex : Valve + pipe + turbine 16

Multi-component model validation n CMS cryoplant ü ü ü n n n Cooldown CMS magnet (225 tons) Air Liquide – 1. 5 k. W @ 4. 5 K 3310 Algebraic equations 244 Differential equations Simulation speed during cool-down : x 15 Validation of a complete system Simulation architecture validation Operator training tool 17

Virtual Commissioning n CERN central helium liquefier (B 165) ü ü ü n Provide liquid helium for small CERN experiments Commercial Linde TCF 50 – 70 Lhe / hour 2060 Algebraic equations 170 Differential equations Simulation speed during cool-down : x 20 Test and improve PLC code and supervision (collaboration with TE-CRG-CE) ü ü ü Bad calibration of sensors PLC-coding errors Sequence errors (timers, threshold, etc. ) New turbine starting sequence PI tuning 18

Large-Scale system simulation: LHC refrigerator n 4. 5 K LHC refrigerator Linde 18 k. W @ 4. 5 K ü 4600 Algebraic equations ü 400 Differential Equations ü Simulation speed: x 3 ü n n n Operator training tool High Pressure control optimization (IMC) New control strategies to reduce operation costs (floating pressure) 19

LHC refrigerator simulations After model validation : Optimization in simulation (real refrigerator in operation non available) Cool-down cold-box alone Stable state reached with LHC (16 k. W @ 4. 5 K) 20

High Pressure control optimization (IMC) -IMC : « Internal Model Control » -Model Synthesis -Model uncertainties evaluations -Synthesis of the controller Q using a robust tunning èGuarantee stability for the worst case èAdapt model in real-time (according to compression state) èTake into account saturation of valves ( « anti-windup » ) Simulation results: Better set-point tracking Better disturbance rejection Better robustness 21

Floating pressure System n High Pressure (HP) influences refrigeration power ü n Floating Pressure system: Adaptation of HP and MP to loads applied to the refrigerator ü ü ü n Load adjusted with an electrical heater in the phase separator Compressor flow rate decreases : Electrical consumption decreases First tests in 1994 on 12 k. W @ 4. 5 K refrigerators (LEP) Manual or semi-automated management Automatic control system to fluctuate HP and MP ü Objective : Stabilize the electrical heater at a desired value (ex : 1 k. W) 22

Floating Pressure approach n Contrôle direct du niveau en pilotant la Haute Pression : Régulation Cascade niveau Heater Level set-point +10 90% Heater 1 + 2 Level set-point 80% 60% Heater 1 = 1 k. W Level set-point -20 t Level set-point + - Floating Pressure Controller HP set-point Refrigerator Level set-point -20% Level set-point + 10% + Level controller 1 - + - Level controller 2 t liquefaction Sat=1 k. W Phase Separator Liquid level Heater 1 Heater 2 (secu) 23

Non-Fragile PI tuning Floating Pressure controller (FPC) Non-fragile Level controllers (LC 240/LC 241) Non-fragile 24

Simulation results High Set-Point FPC start FPC set-point Load decrease Load Increase Low Set-Point Activation of 2 nd heater No Saturation on 1 st heater Heater Set-point = 1 k. W 25

Variable-Frequency drive for compressors With ideal VFD 26

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 27

Large cryogenic systems at 1. 8 K n Cool-down dysphasic helium from 4. 5 K to 1. 8 K ü Pumping on helium baths from 1 bar to 16 mbar n Large cryogenic systems (pumping flow rate > 60 g/s for LHC) ü Cold Compressors (helium at 4 K, compression ratio ~ 60) ü Pumping on long cryogenic lines (3. 3 km for the LHC) n 1. 8 K refrigeration unit for the LHC 2800 Algebraic equations ü 210 Differential equations ü Simulation speed during pumping : x 80 ü 28

Cold-Compressor : characteristic identification n Cold compressor= hydraulic component Strict pressure field to respect: m = f(ratio pressure, speed) “Move” theoretical pressure field from measurements to fit real data 29

LHC cryogenic distribution line n n QRL : Transport cold helium along LHC Return Pumping Line (Line B) ü n Low pressure cold helium : 3 K /16 mbar Development of a dynamic model ü ü ü Low pressure and temperature helium flow Convection heat transfers Interconnections every 107 m n Euler equations: n Simplification for QRL (PDE 1 D) : with: 30

Jacobian calculation Perfect GAZ Equation of state: Pressure: Sound speed: Jacobian: 9% at 5 K Eigen values: 2% at 10 K Gaseous low pressure helium Equation of state: Pressure: Sound Speed: Jacobian: Eigen values : 31

Discretization n Finite element method with an upwind scheme Discretization according to the propagation direction ü Mass and energy along the flow direction (transport) ü Momentum on the inverse direction of the flow (pumping) ü n Dirichlet boundary conditions: Input density ρ(0, t) ü Input energy E(0, t) ü Output momentum M(L, t) ü n Implicit temporal discretization (backward Euler) : 32

Interconnections and heat transfers n Interconnections every 107 m ü n Source term is augmented (mass, momentum and energy) Heat transfers: ü Constant term (conduction, radiation) : 1, 92 W/m 3 on line B ü Variable term(convection) : 33

Simulation of the final pumping n Pumping between 100 mbar and 16 mbar 34

Simulation after a « quench » n « Quench » : Resistive transition between the superconductor state and the resistive state: ü n Release a large amount of energy (heat) Simulation of the “heat wave” induced by a quench in the return pumping line ü Comparison with a quench in the LHC sector 5 -6 during hardware commissioning (May 2008) Simulation with Δx=107 m Simulation with Δx=10, 7 m 35

Contents n Introduction ü n Dynamic simulator for cryogenic systems ü ü n CMS cryoplant Central Helium Liquefier LHC refrigerator Simulation of LHC 1. 8 K refrigeration units ü ü n Simulation and control architecture Cryogenic Modeling Simulations of cryogenic systems operating at 4. 5 K ü ü ü n Motivations and state of art Cold compressors Cryogenic Distribution Line Conclusion & Perspectives 36

Main Contributions n A dynamic simulator for CERN cryogenic plants « Cryo Simulation Lab » available at CERN for TE-CRG (building 36) ü « Virtual commissioning » in collaboration with TE-CRG-CE (B 163 & B 165) ü 37

Main Contributions n Control Improvements Optimization of the High Pressure control on LHC refrigerators (IMC) ü Development of a floating pressure control to reduce operational costs on LHC cryoplants ü n QRL model ü Dynamic model of low pressure helium flow in long pumping lines 38

Perspectives n Export simulator to other large cryogenic plants Helium refrigerator @ 2 K for the XFEL project at DESY (in discussion) ü Dynamic behavior during pulsed heat loads for tokamaks (ITER ? ) ü n IMC and floating pressure control test on real LHC refrigerators Possible windows in may/june 2010 ? ü Speed variator study (in discussion) ü n Extension of the « cryo » library to other industrial processes Water cooling (already done for STP 18) ü Electronic cooling for detectors (C 02, C 6 F 14 …) ü LHC ventilation systems ü 39

Thank you for your attention Questions ?

Cryo Simulation Lab