STAR Global Conference 2017 Berlin Mar 6-8, 2017 Coupled CFD-STH analysis of liquid metal flows: STAR-CCM+ - RELAP5 Marti Jeltsov, Kaspar Kööp, Pavel Kudinov Division of Nuclear Power Safety KTH Royal Institute of Technology Stockholm, Sweden 1
Outline Physical phenomena of interest In LFRs In TALL-3D experiment Coupling methodology Code qualification methodology Results Standalone Star-CCM+ Coupled Star-CCM+ - Relap5 Conclusions 2
Thermal-hydraulic phenomena in LFRs The use of numerical modeling and simulation (M&S) in the design and safety analysis of nuclear reactor systems is increasing. Commonly, 0D/1D System- Thermal Hydraulics (STH) codes have been used in the design, safety analysis and licensing of LWRs. Pool-type LFRs are characterized by multidimensional flows with large fraction of natural circulation (sensitive to small changes in the system). Numerical tools used for LFR analysis need to be validated for these specific phenomena. Single phase phenomena Free shear Wall shear Thermal mixing Thermal stratification Jet impingement of a surface Thermal inertia of structures Natural circulation Multi-phase phenomena Bubble/particle transport Boiling Sloshing Melting/solidification Model type Turbulence model Turbulence model, wall functions Momentum convection, bulk turbulence Thermal diffusion, buoyancy models Turbulence model (production) Conjugate heat transfer Integrally all model Model type Interfacial momentum transfer (drag) and buoyancy Irrelevant ALE, Eulerian+interphase tracking (e.g. VoF, Level Set) MYRRHA ALFRED ELFR 3
TALL-3D loop design The facility loop is comprised of 3 sections: main heater leg (left/red) 3D test section leg (center/green) heat exchanger leg (right/blue) Essential components Main heater Straight run 3D Test section Heat exchanger (secondary loop is not shown) Secondary loop utilizes Dowtherm HT fluid as coolant and uses a fan as a secondary heat exchanger Electric Permanent Magnet (EPM) pump 3 Ball valves (one per each section) for fine tuning of flow rate 6 expansion compensators Oxygen sensor Expansion tank Instrumentation includes Thermocouples ~330 TCs (154 in the test section) Differential pressure measurement system 5 DP groups (each contains one pressure transducer and several control valves), 11 measurement points total 15 differential pressures around the loop can be measured Flow meters: Two Carioles flow meters on the primary loop One transit time ultrasonic flow meter on the secondary loop Dimensions: Total height 6980 mm; loop height 5800 mm, width 1480 mm Pipe ID 27.86 OD 33.4 4
TALL-3D test section Designed to pronounce transient 3D effects in the system The dimensions are selected such that at high flow rate (forced convection) the jet mixes the pool and at low flow rate thermal stratification develops (natural circulation) 5
TALL-3D test section: Flow regimes Full mixing at m ሶ > 1.0 kg/s jet penetrates the stratified layer and mixes the pool Partial mixing at mሶ 0.7 kg/s jet penetrates the stratified layer in the middle of the pool, but has not enough momentum to penetrate the buoyant wall jet at the heated wall Thermal stratification at m ሶ < 0.3 kg/s jet is too weak to break the stratified layer 6
TH phenomena in TALL-3D Phenomenon Description Regime Experiments Thermal mixing and stratification in the test section are governed by the jet dynamics. A necessary condition for CFD validation, Re, Ri N01, F01, T01 Free jet flow then, is that the codes accurately simulate free jets in a pool-like geometry in steady state conditions, both in forced and in natural circulation. Measurements: Temperature measurements on the inner circular plate are the validation data for this phenomenon. Jet impingement on a surface It is a key factor to produce the recirculation patterns inside the test section. The design simulations show that the jet will not reach the disk in natural circulation conditions therefore this validation can only be performed against steady state, forced circulation experiments. Measurements: Minimum mass flow rate for the jet to reach the disk measured by a Coriolis FM is used in validation metric. Re F01, T01 Jet induced circulation in the pool Thermal stratification Mixing Thermal inertia of the structure Thermal conduction through the plate Turbulence Circulation in the pool is produced by the diversion of the inlet jet due to the circular inner plate. Validation data can be therefore obtained in forced or fast natural circulation experiments. Measurements: As the circulation drives mixing, validation data for this phenomenon is the temperature field in the test section. The codes capability to capture transient development of stratification after switching on the heater can be assessed against temperature data obtained in forced (at small flow rates to prevent jet reaching the plate to enhance mixing) or natural circulation thermal transients in the loop. Measurements: Inner pool thermocouples. The codes capability to capture thermal mixing can be assessed against temperature data obtained in steady state forced circulation experiments. Measurements: Inner pool thermocouples. Heat transfer between LBE and heater/ambient is dependent on the thermal conductivity and capacity of the wall materials. Thermal inertia is important in transients where the temperature changes on either side of the walls, e.g. heater is switched on/off. Measurements: Temperature at the outer and inner side of the wall is measured in order to validate thermal conductivity model. Fluid temperatures below and above the circular plate can be different resulting in heat transfer through the plate. The circular plate is made of conductive material and the temperature gradient through that material is therefore a subject to validation. Measurements: Thermocouples at the bottom and top of the plate. Turbulence modeling is regarded as the one of the major contributors to uncertainty in the CFD M&S. TALL-3D provides integral data for turbulence studies. Turbulence produced in the wall boundary and jet shear layers is responsible for the flow energy loss in the test section. Measurements: Pressure loss over the test section is considered in validation metric. Re Gr, Re Re Re F01, T01 F01, N01, T01 F01, T01 T01 F01, N01, T01 F01, N01, T01 NB! In initial or final steady states belonging to transient tests, Reynolds number is between 3,600 and 58,000. Lower and higher Reynolds values occur during flow oscillations and in the dedicated test for different steady states. 7
HX mass flow rate MH power TS power HX mass flowrate MH power TS power Oil inlet temperature Tests performed in THINS project Initial steady state Final steady state # Name Notes kg/sec kw kw kg/sec kw kw C 1 C01 DPs measurements 2 C02 TCs offset 3 T01.08 4.1 2.6 4.8 0.6 2.6 4.8 65 Forced to natural circulation transient 4 T01.09 4.3 2.6 4.8 0.6 2.6 4.8 61 Forced to natural circulation transient 5 T01.10 3.3 3.2 4.0 0.6 3.2 4.0 85 Forced to natural circulation transient 6 T02.03 4.3 6.3 0 0.5 2.8 4.0 95 Forced to natural circulation transient 7 T02.04 4.2 2.1 0 0.6 2.1 5.2 145 Forced to natural circulation transient 8 T02.06 4.2 1.7 0 0.5 1.7 5.2 145 Forced to natural circulation transient 9 T03.01 0.5 2.3 0 0.5 2.3 4.8 140 Natural to natural circulation transient 10 T06.01 0.6 2.6 4.8 4.3 2.6 4.8 61 Natural to forced circulation transient 11 T09.01 0.6 3.2 4.0 0.4 0 4.0 85 Natural to natural circulation transient 12 T11.02 0.5 2.8 4.0 0.5 1.8 4.9 95 Natural to natural circulation transient 8
CFD+STH validation CFD validation STH validation Phenomena validation matrix Physical Phenomena FC 3DH Off Steady states FC 3DH On NC 3DH Off NC 3DH On Transients T01 T02 T03 T06 T09 T11 1. Drag in forced circulation x x x x x 2. Drag in natural circulation x x x x x x x x 3. Transient drag x x x x x x 4. Heat transfer in forced circulation 5. Heat transfer in natural circulation 6. Stability of natural circulation 7. Thermal inertia of the loop sections transient 8. Heat losses as a function of the system temperature steady state 9. Heat losses as a function of the system temperature transients x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x 1. Free jet flow x x x x x x x x x x 2. Jet impingement on a surface x x x x x 3.Jet-induced recirculation x x x x x 4.Stratification development x x x x x x 5. Mixing x x 6. Buyoant wall jet flow x x x x x x 1. Transient response of the CFD test section 2. Transient response of the loop 3. Nonlinear feedbacks and instabilities x x x x x x x x x x x x x x x x x x 9
Code coupling The aim of the code coupling is to achieve required accuracy with affordable computational effort. CFD domain Codes that are coupled: CFD: Star-CCM+ STH: RELAP5/Mod3.3-LBE Coupling the transition and transformation of information between 1D and 3D codes. Controlled by a Java code. 10
Code coupling aspects What information is exchanged? Amount and type of coupling variables (values, profiles, sources, sinks etc.) Information that one code lacks must be provided by another Where is the information exchanged? Definition of the spatial division of system between codes Options: domain decomposition and domain overlapping Coupling boundary location must be selected such that data losses are minimized when transformed during transition (e.g. 1D 3D in fully developed flow region) When is the information exchanged? Frequency of the data exchange (i.e. coupling time-step) The rate of change of coupling variables must be captured Options: constant and automatically variable time-stepping How is the information exchanged? The exchange of information can be uni- or bi-directional Unidirectional approach can be used to zooming while bidirectional is required to analyze behavior of systems featuring feedbacks between components 11
Coupling in space Domain overlapping vs decomposition - Domain overlapping chosen to avoid STH solution stability issues 12
Exchanged variables During a coupling time-step: Inlet b.c. are provided to CFD STH temperature field and pressure drop are corrected. Virtual heater Qሶ ሶ Q calculation T outlet Works in both flow directions K loss K loss calculation Flow direction STH CFD T inlet mሶ 13
Coupling in time Start coupled calculation Execute STH for 1 coupling time step Execute CFD for 1 coupling time step Convergence? NO Iterate with STH until converged YES coupling time step NO End of simulation? YES coupled simulation End coupled calculation 14
Coupling algorithm STH and CFD solutions have to be converged at the beginning of coupled calculations Either a common steady state, or continuation of paused coupled calculation. Coupling starts with STH calculation Mass flow rate and inlet temperature are provided to CFD p n-1 n n+1 CFD calculates the same time-step using b.c. from STH CFD STH m, ሶ T in allowed ΔT out < ΔT out ΔT out > ΔT allowed out Q virtual The difference in solutions is evaluated against convergence criterion Δt STH < Δt CFD Based on the difference, correction terms are calculated Δt coupling = Δt CFD time STH calculates new solution until convergence criterion is met (or number of allowed STH iterations is reached) 15
Architecture Controls time marching (Δt C ), extrapolates and averages b.c., modifies CFD and STH models, executes codes, logs data. T inlet CFD T outlet ሶ Q T inlet mሶ ΔP K loss mሶ Coupling code features: Written in Java (a macro nested in Star-CCM+) OS agnostic (Windows/Linux, given that respective executables are available) Automatic creation of STH/CFD inputs Automatic switch between and calculation of steady state/transient models Logging of variables (only desired part of solutions are stored) LBE properties from Sobolev (2011) (since this is implemented in Relap5) 16
V&V process Goal: To develop sufficient evidences for robust decision making Successful V&V process is necessarily: Systematic and complete All uncertainty sources addressed Iterative Obtaining new data affects code inputs and models Converging: to a decision. User-independent 17
Base input model description 2D axisymmetric model to increase the computational efficiency (accuracy vs speed) Model geometry was defined according to the manufacturing specifications, CAD drawings and on-site measurements 2 nd order implicit solver for unsteady calculations Models for fluid Coupled Energy solver Coupled Flow (2 nd order upwind implicit integration) solver Gradients (2 nd order) Gravity Turbulence Reynolds-Averaged Navier-Stokes k ε Realizable Two Layer All y + Wall treatment No-slip 1 fluid region (LBE) LBE fluid Stainless steel Heater Insulation Models for solids Constant density Coupled Solid Energy solver Gradients (2 nd order) 6 solid regions (steel walls, heater, 4 insulation regions) Heat losses Ambient temperature: from experiment Convection Contact resistance between the heater and pool walls Total resistance coefficient accoiunting for radiation and thermal conductivity effects. 18
Definition of SRQs Phenomena in TALL-3D experiment: Thermal stratification, Mixing, Laminar pipe flow, Turbulent pipe flow, Thermal inertia. Heat losses, Free jet dynamics. Proposed list of SRQs for CFD valiation: Steady state Outlet temperature Pressure drop Maximum LBE temperature IPT gradient (e.g. IPT8-IPT2) Transient Maximum temperature at the outlet Timing of outlet temperature peak Decay ratio (in coupled simulations) 19
Verification Mesh size Divide and conquer separate mesh study for different parts Fluid (convection, turbulence scales) Solids (diffusion scales) Wall mesh (turbulence model requirements) Grid Convergence Index (GCI) (Roache, 1994) ε GCI = F s r p 1 Time-step size Time-step is selected according to following criteria: Numerical criterion (e.g. CFL condition) Stability Accuracy Physical criterion Characteristics of unsteady phenomena (e.g. boundary conditions, flow field) Time-step convergence Converegence of finite-difference approximation of the numerical problem 0.02 0.015 GCI 0.01 10-3 0 coarser mesh <- Refinement step -> denser mesh 20
Sensitivity Analysis SA is a useful to identify the most important UIPs Morris MOAT method by coupling with Dakota Hundreds of CFD calculations Iterative Process 1. Define initial ranges Conservative, and Based on limited knowledge at the time. 2. Perform first SA Screening of influential/non-influential parameters is obtained 3. Revise ranges based on new evidence (e.g. additional experiments) Use calibration to reduce conservative ranges, and Some ranges may be enlarged based on new knowledge available. 4. Perform second SA 5. Compare first and second SA results 6. Iterate until ranges converge 7. Screening Discard/fix non-influential input parameters SA matrix helps to find UIPs with the largest effect (i.e. most influential) in terms of number of affected SRQs Helps to focus on only important input parameters! UIPs that affect a particular SRQ the most (normal Morris plot type info) Tells which UIPs can be calibrated using that SRQ! SRQs that are mostly affected by a particular UIP Tells which SRQ should be used to calibrate this input parameter! 21
Calibration Natural convection SS Integral Outlet temperature must be predicted correctly to predict SS heat losses. x3 Local Thermal conductivities in 6 insulation regions were calibrated to model the spatial distribution of heat losses In-pool TCs were used as SRQs x2 x4 x4 x5 x2 22
LOSS OF FLOW TRANSIENT (T0206) 23
Standalone CFD Mass flow rate 24
Standalone CFD Temperature 25
Standalone CFD Animation 26
Standalone CFD CIP temperature 27
Coupled CFD-STH Mass flow rate Coupled simulation predicts the mass flow rate dynamics better than standalone STH Timing of flow reversal duration and recovery is improved Oscillatory behavior is captured Residual deviation can be reduced by further calibration using experimental evidence 28
Coupled CFD-STH Temperature Coupled simulation improves the pool inlet and outlet temperature prediction Timing of peaks is improved Oscillatory behavior is captured STH sees no temperature oscillations due to under-resolving the complex 3D flow field 29
Conclusions Multi-scale flows are important in liquid metal cooled nuclear reactors STH codes are inadequate to resolve 3D flow phenomena and CFD codes are computationally demanding for large systems Coupling of CFD and STH improves the computational efficiency - desired accuracy at affordable computational costs Coupling introduces new physical models Translation of data between 1D and 3D codes Validation of coupled codes cannot be reduced to validation of standalone codes TALL-3D facility is designed and instrumented for validation of standalone and coupled thermal-hydraulic codes Pre-test analysis demonstrated the desired transition between thermal mixing and stratification in the 3D test section Coupled simulations reveal features that would not be captured with standalone STH code 30
Thank you! 31