ANL/NSE-26/24 Development of SAM Code Capabilities for Safety Analysis of GCR Air-ingress Events Nuclear Science and Engineering Division About Argonne National Laboratory Argonne is a U.S. Department of Energy laboratory managed by UChicago Argonne, LLC under contract DE-AC02-06CH11357. The Laboratory’s main facility is outside Chicago, at 9700 South Cass Avenue, Argonne, Illinois 60439. For information about Argonne and its pioneering science and technology programs, see www.anl.gov. DOCUMENT AVAILABILITY Online Access: U.S. Department of Energy (DOE) reports produced after 1991 and a growing number of pre-1991 documents are available free at OSTI.GOV (http://www.osti.gov/), a service of the US Dept. of Energy’s Office of Scientific and Technical Information. Reports not in digital format may be purchased by the public from the National Technical Information Service (NTIS): U.S. Department of Commerce National Technical Information Service 5301 Shawnee Road Alexandria, VA 22312 www.ntis.gov Phone: 800-553-NTIS (6847) or 703-605-6000 Fax: 703-605-6900 Email: orders@ntis.gov Reports not in digital format are available to DOE and DOE contractors from the Office of Scientific and Technical Information (OSTI): U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 www.osti.gov Phone: 865-576-8401 Fax: 865-576-5728 Email: reports@osti.gov ANL/NSE-26/24 Development of SAM Code Capabilities for Safety Analysis of GCR Air-ingress Events by Gang Yang, Zhiee Jhia Ooi, Ling Zou, Travis Mui, Thanh Hua, Rui Hu Nuclear Science and Engineering Division, Argonne National Laboratory JUNE 30, 2026 ABSTRACT Air-ingress following a depressurized loss-of-forced-cooling (DLOFC) event is a challenging, multi- physics safety scenario for High-Temperature Gas-Cooled Reactors (HTGRs), involving coupled gas com- position transport, buoyancy-driven flow redistribution, graphite oxidation, and structural heat-up. Despite its importance — air ingress is a key scenario identified in the PIRT process for the HTGRs — existing system-level safety codes have lacked the integrated capability to simulate the complete event sequence with high confidence. This report documents the development, validation, and demonstration of three new capabilities in the SAM code to address this gap: (1) a multi-component gas mixture flow model with binary diffusion to track the helium-air composition and its effect on system density and flow; (2) a 0-D graphite oxidation model based on the Roes correlation, including oxygen consumption and exothermic heat release; and (3) an isentropic critical flow model for accurate representation of primary system depressurization through a break. These capabilities are validated against two benchmark experiments. The NSTF heavy-gas ingress experiment validates the multi-component flow model: SAM correctly reproduces the rapid buoyancy- driven flow stagnation and subsequent natural circulation recovery driven by composition-dependent den- sity changes. The NACOK graphite oxidation experiment validates the oxidation model: SAM predicts a bottom-level graphite weight loss of 25%, in close agreement with the measured 24%, and reproduces the strong axial nonuniformity and block-geometry dependence of oxidation, at a level comparable to the SPECTRA and TINTE codes. The validated capabilities are then exercised together in an integrated, reactor-scale simulation of a DLOFC air-ingress transient in a simplified HTR-PM pebble-bed reactor. In a single calculation spanning approximately 8 days, SAM reproduces the complete accident sequence: rapid depressurization, density- and diffusion-driven air ingress over ˜15 hours, onset of buoyancy-driven natural circulation, exothermic graphite oxidation with a peak fuel temperature at ˜62 hours, and eventual passive cooldown. These results demonstrate that SAM now provides the nuclear community with a preliminarily validated, modern system- level tool for HTGR air-ingress safety analysis, filling a recognized capability gap. Future extensions to broaden species tracking, improve oxidation chemistry, and refine the reactor model are discussed. i ii Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 Air-ingress Events in GCR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3 Developed capabilities for Air-ingress Modeling . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.1 Multi-component Flow Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 Critical Flow Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.3 Graphite Oxidation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.1 NSTF Heavy-gas Ingress Event Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.1.1 NSTF Facility Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.1.2 Heavy-gas Ingress Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4.1.3 SAM Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4.1.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2 NACOK Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.2.1 NACOK Facility Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.2.2 NACOK Air-ingress Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.2.3 SAM Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5 Reactor-scale Demonstration of Air-Ingress Events . . . . . . . . . . . . . . . . . . . . . . . . 41 5.1 The HTR-PM Reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 5.2 Demonstration Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 5.3 SAM Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 iii List of Figures 2.1 Air ingress scenario in HTGR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Temperature regimes of graphite oxidation. . . . . . . . . . . . . . . . . . . . . . . . . . . 6 4.1 NSTF layout and primary segments. Adapted from [39]. . . . . . . . . . . . . . . . . . . . 23 4.2 Argon enclosure in the NSTF heavy-gas ingress experiment. Adapted from [37] . . . . . . . 24 4.3 Operating modes of the NSTF heavy-gas enclosure: normal air draw (left), and argon bypass (right). Adapted from [37]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4.4 SAM model for NSTF heavy-gas ingress test. . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.5 Oxygen concentrations measured in the inlet plenum top and bottom, and average value from SAM simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.6 System mass flow rate during the NSTF heavy-gas ingress test. . . . . . . . . . . . . . . . . 28 4.7 Predicted gas temperature and mixture density at the riser duct inlet, mid-height, and outlet during the NSTF heavy-gas ingress test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.8 Overall arrangement of the NACOK test facility. Adapted from [44]. . . . . . . . . . . . . . 30 4.9 Graphite block configuration of NACOK experiment. Adapted from [44]. . . . . . . . . . . 32 4.10 SAM model of the NACOK experiment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.11 Lower boundary temperature imposed at the bottom of the graphite block stack in the SAM NACOK model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.12 Temperature evolutions at the bottom graphite level from SAM predictions and the measured data. The left column shows the graphite solid temperature and the right column the gas outlet temperature, for the Type-1 blocks (top row) and the Type-2 blocks (bottom row). Measured data from [46]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.13 Thermocouple layout in the NACOK experiment. The four thermocouples used in the present comparison are located in the bottom level: the Type-1 graphite solid (2.2) and gas outlet (2.3), and the Type-2 graphite solid (21.3) and gas outlet (22.1). Adapted from [46]. 38 4.14 Graphite weight loss of each block in the NACOK experiment: comparison of the measured data (SGL NBG-17 and UCAR PCEA grades) with the SPECTRA Roes-correlation result and SAM prediction. Blocks are labeled by level (B = bottom, M = middle, T = top) and geometry type (1, 2). Experimental and SPECTRA values are from Ref. [44]. . . . . . . . . 38 5.1 Full system layout of an HTR-PM reactor module. Adapted from [50]. . . . . . . . . . . . 42 5.2 Schematic of the simplified HTR-PM air-ingress SAM model. . . . . . . . . . . . . . . . . 44 5.3 Boundary pressure at the core inlet and outlet during the depressurization phase. The system blows down from the 7 MPa operating pressure to atmospheric from 100 to 120 s. . . . . . . 45 5.4 Mass flow rates in the two outlet pipes during the density-driven flow after depressurization . 46 5.5 Air mole fraction y1 at the two outlets and in the lower branch during the initial ingress stage. The denser air enters the lower outlet pipe (outlet-2) first, then fills the branch, and reaches the upper outlet pipe (outlet-1) last. . . . . . . . . . . . . . . . . . . . . . . . . . . 46 5.6 Air mole fraction y1 at the core outlet (bottom) and core inlet (top) during the diffusion- dominated phase. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.7 Core mass flow rates during the simplified HTR-PM air-ingress transient. . . . . . . . . . . 48 5.8 Core fuel and coolant temperature during the HTR-PM air-ingress demonstration transient. . 49 iv List of Tables 3.1 The collision integrals, ΩD based on the Lennard-Jones potential. [32]. . . . . . . . . . . . . 16 3.2 Lennerd-Jones coefficients for various gases [32]. . . . . . . . . . . . . . . . . . . . . . . . 17 4.1 Summary of the NSTF heavy-gas ingress validation . . . . . . . . . . . . . . . . . . . . . . 29 4.2 Graphite weight loss of each level (%) measured in the NACOK experiment and predicted by SPECTRA, TINTE, and SAM models. Experimental, SPECTRA, and TINTE values are from Ref. [44]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.3 Summary of the NACOK graphite oxidation validation . . . . . . . . . . . . . . . . . . . . 40 v Acronyms BDBA Beyond Design Basis Accident CFD Computational Fluid Dynamics DEGB double ended guillotine break DLOFC de-pressurized loss of forced cooling FHR Fluoride-Cooled High-Temperature Pebble-Bed Reactor HTGR High-Temperature Gas-Cooled Reactor HTR-10 10 MW High Temperature Gas-cooled Test Reactor HTR-PM High-Temperature Gas Cooled Reactor–Pebble-Bed Module HTTF High Temperature Test Facility HTTR High Temperature Engineering Test Reactor LOFC loss-of-forced-cooling MHTGR Modular High Temperature Gas-Cooled Reactor NACOK Naturzug im Core mit Korrosion NEAMS Nuclear Energy Advanced Modeling and Simulation NSTF Natural Convection Shutdown Heat Removal Test Facility PIRT Phenomena Identification and Ranking Table PLOFC pressurized loss of forced cooling RCCS reactor cavity cooling system RELAP5-3D Reactor Excursion and Leak Analysis Program - 3D RPV reactor pressure vessel SAM System Analysis Module TINTE Time-dependent Neutronics and Temperatures TRISO tristructural isotropic 1 1 Introduction High-Temperature Gas-Cooled Reactor (HTGR) is one of the candidates for the Gen-IV reactor designs and is currently under active development worldwide. There are two typical designs of HTGRs: the pebble- bed type and the prismatic type. In the pebble-bed HTGR, the reactor core is filled with fuel pebbles containing tristructural isotropic (TRISO) particles. In the prismatic type HTGR, TRISO particles are packed into pellet forms and turned into long fuel rods inserted in hexagonal graphite blocks in the core. In both HTGR designs, helium is commonly used as the primary coolant because of its chemical inertness, low neutron absorption cross section, and relatively high specific heat capacity. However, due to its low density, HTGRs are pressurized to increase the density of helium, enhancing heat removal capability. The use of pressurized helium in HTGRs presents some unique challenges, especially under accident scenarios. In an event of a de-pressurized loss of forced cooling (DLOFC) accident where the core is de-pressurized, the system pressure drops significantly such that the low density of helium may hinder the formation of natural circulation within the core. Under these conditions, heat is removed primarily by conduction and thermal radiation from the core to the decay heat removal system. In this scenario, fuel temperature may rise significantly above the nominal value during normal operating condition and risk being damaged. In addition, the evacuation of helium may cause air from the reactor cavity to enter the core. During this so-called air-ingress scenario, graphite in the pebble fuels and the reflectors in the core may undergo oxidation by reacting with the oxygen in the air. This may compromise the integrity of the graphite and release fission products from the fuel pebbles into the reactor cavity or weaken the graphite structures located under the pebble bed. The importance of the air ingress scenario is evident as it is one of the scenarios whose safety related phenomena are identified using the Phenomena Identification and Ranking Table (PIRT) process for the High-Temperature Gas Cooled Reactor–Pebble-Bed Module (HTR-PM) [1]. Air ingress can occur as a consequence of a DLOFC event, in which a loss of confinement integrity (e.g., a break in the vessel or connecting piping) is followed by a loss of coolant inventory. The process typically unfolds in four stages: 1. Depressurization – Helium coolant rapidly vents to the containment or environment until the primary system pressure equilibrates with the surroundings. 2. Density-Driven and Diffusion-Driven Gas Exchange – The higher density of air relative to helium promotes buoyancy-driven mixing and stratification, leading to gradual air replacement in the lower regions of the reactor. Depending on the core geometry and flow resistance, upward air transport into the core can be governed either by molecular diffusion or by buoyancy. This slow stage may persist for a long time before loop-wide flow is established. 3. Onset of Loop-wide Natural Circulation – As air accumulates and the density difference between the core and the return path grows, the buoyancy force eventually overcomes the core hydrostatic head and frictional losses, establishing a sustained, loop-wide natural circulation that delivers a continuous supply of oxygen to the hot core. 4. Graphite Oxidation and Core Heat-Up– Oxygen transported into the core reacts with graphite structural components and fuel element blocks. Oxidation is exothermic and can lead to material loss, weakening of structural integrity, and potential reconfiguration of the flow field. 2 The safety importance of air-ingress phenomena has motivated decades of experimental and modeling work. Key experimental campaigns have been carried out to investigate the fundamental physics. A variety of computational approaches and tools have been used or developed by many researchers to investigate the effects of air-ingress in HTGRs, ranging from system-level codes such as TINTE, SPECTRA, and GAMMA to higher fidelity Computational Fluid Dynamics (CFD) codes [2–9]. The inherently multiphysics character of the air-ingress scenario that combines thermal fluids phenomena with chemical reactions presents substantial challenges to these computational tools. Furthermore, there are also difficulties due to the difference in the time scales of the mechanisms with which air enters the core. Significant gaps remain in predicting coupled multi-physics behavior across scales, particularly for the transition from molecular diffusion–dominated ingress to buoyancy-driven convection, and for oxidation effects on flow patterns and structural integrity. As a result, the coupled simulation of an HTGR air-ingress event — from primary system depressurization through diffusion-dominated ingress, natural circulation onset, graphite oxidation, and passive cooldown — in a single system-level calculation remains a capability gap. The System Analysis Module (SAM) code [10] is under development and supported by DOE-NE’s Nuclear Energy Advanced Modeling and Simulation (NEAMS) program as a modern system-level modeling and simulation tool for advanced non-light water reactor safety analyses. The SAM code capabilities for HTGR applications have been extensively demonstrated in the simulations of various HTGR designs, such as MHTGR-350 (General Atomics) [11], a generic pebble bed gas-cooled reactor [12, 13], the HTR-PM [14], Holos Quad (Holos Gen) [15], and Horizontal-Compact HTGR (MIT/Boston Atomics) [16], under various transient conditions. Extensive SAM code validation for High-Temperature Gas-Cooled Reactor (HTGR) applications have also been performed, including the OECD/NEA High Temperature Test Facility (HTTF) thermal-hydraulics benchmark [17, 18], and the ongoing code benchmark activities using High Temperature Engineering Test Reactor (HTTR) data. To extend SAM’s applicability to air-ingress scenarios — which require tracking gas composition changes, modeling graphite oxidation, and accurately representing primary system depressurization — three new modeling capabilities have been developed: a multi-component gas mixture flow model with binary diffusion, a 0-D graphite oxidation model based on the Roes correlation, and an isentropic critical flow model for break depressurization. This report documents the SAM capabilities developed to model the air-ingress event and presents the corresponding validation and demonstration studies. The following sections first describe the underlying mechanisms and progression of air ingress in detail. The newly implemented SAM modeling capabilities for air-ingress scenarios are then introduced. Subsequently, these capabilities are validated against the Natural Convection Shutdown Heat Removal Test Facility (NSTF) heavy-gas ingress experiment and the NACOK air-ingress experiments. Finally, a demonstration analysis of an air-ingress transient using SAM is presented, followed by a summary and a discussion of future work. Together, these capabilities establish SAM as a validated, system-level tool for HTGR air-ingress safety analysis, supporting HTGR licensing efforts and filling a recognized gap in the community’s modeling toolkit. 3 2 Air-ingress Events in GCR This section provides a detailed description of the air-ingress scenario in HTGRs and a review of prior experimental and modeling studies. A loss-of-forced-cooling (LOFC) accident is considered as a critical event for a High-Temperature Gas-Cooled Reactor (HTGR). In general, there are two types of LOFC accidents, namely the pressurized loss of forced cooling (PLOFC) and the de-pressurized loss of forced cooling (DLOFC). During PLOFC, the reactor is scrammed and the helium blower is tripped, while the primary circuit of the reactor remains intact. Some possible initiating events that could lead to PLOFC include the inadvertent withdrawal of a control rod, uncontrolled acceleration of the helium blower, loss of offsite power, and loss of feed water [1]. Under PLOFC conditions, natural circulation of the pressurized helium can be established, thereby facilitating decay heat removal from the core to the reactor cavity cooling system (RCCS). The maximum fuel temperature is generally not considered a major concern during PLOFC as it is expected to remain well below the safety limit due to the enhanced heat removal from the natural circulation in the core. By contrast, the DLOFC is typically characterized by rapid de-pressurization of the core and the primary circuit followed by an emergency shutdown of the reactor. Possible initiating events include unisolable breaks of the pressure boundary, such as an rupture of instrumentation tube. Due to the low helium density in the de-pressurized system, natural circulation cannot be established, which leads to less effective redistribution and removal of decay heat than in the PLOFC. Thermal radiation and conduction become the dominant decay heat removal mechanism during DLOFC. An extreme form of DLOFC originates from the so-called double ended guillotine break (DEGB) at the horizontal coaxial duct of the core, which is considered a Beyond Design Basis Accident (BDBA) in HTGRs [19]. In this scenario, the rapid de-pressurization of the system is followed by the ingress of air into the core, initially through molecular diffusion and then buoyancy effect. Eventually, if the buoyancy force surpasses the pressure drop of the core, a stable natural circulation is established. The ingress of air into the core can lead to the graphite oxidation, which may compromise the integrity of the fuel and release fission products into the environment. At the same time, the lower graphite structures and reflectors can also undergo oxidation that can potentially challenge the structural integrity of the core. The air ingress event in HTGRs is described in detail by Oh and Kim [20] and is illustrated in Figure 2.1. Following the DEGB, the core is rapidly de-pressurized and helium is discharged from the core. After de-pressurization, air in the containment enters the reactor vessel through the bottom part of the horizontal co-axial pipe forming a stratified flow due to the difference in air and helium densities. At this stage, molecular diffusion due to the difference in air concentration inside and outside the reactor can also play a role in introducing air into the bottom plenum. However, several studies compared the time scale of stratified flow to the time scale of molecular diffusion and found that the latter had negligible effects at this stage of the transient [5, 20]. After the bottom plenum of the reactor is filled with air, the temperature difference between the air inside and outside the vessel results in a density gradient that in turn drives a counter-current flow in the bottom plenum. At this stage, the counter-current flow persists because there is always a temperature difference between the air inside and outside of the vessel. At the same time, due to heating in the lower plenum, some of the air from the counter-current flow will flow into the reactor core if the buoyancy force in the lower plenum is sufficient to overcome the hydrostatic head and frictional pressure drop in the core, thus 4 initiating the onset of natural circulation. Conversely, if the buoyancy force is insufficient to overcome the hydrostatic head and core frictional pressure drop, the flow of air from the lower plenum to the core will be dominated by molecular diffusion. In the study by Oh and Kim [20], by comparing buoyancy force to hydrostatic head, it was found that the flow of air into the prismatic core of the 600 MWth Modular High Temperature Gas-Cooled Reactor (MHTGR) was dominated by buoyancy force; whereas in the Naturzug im Core mit Korrosion (NACOK) test facility molecular diffusion was the dominant mechanism due to insufficient buoyancy force from the air in the lower plenum to overcome the hydrostatic head of the core. Liu et al. [5] performed similar calculation for the HTR-PM and found that the flow of air into the pebble bed core was dominated by molecular diffusion. These studies highlight that the design of the core plays an important role in determining the dominant mechanism for air to flow from the lower plenum into the core during an air ingress scenario. Finally, as air continues to enter the core, and the density difference between the gas mixture in the core and the riser channels eventually grows enough to provide sufficient buoyancy force to generate a stable natural circulation, a massive air-ingress is initiated. Figure 2.1: Air ingress scenario in HTGR. Once air enters the lower plenum and the core, oxidation occurs between the oxygen in the air and the graphite pebbles and structures. Oxidation is essentially a two-step process that involves the diffusion of oxidizing gas into the graphite and the chemical reactions between the gas and the graphite. Several possible chemical reactions are listed as Eqs. 2-1 to 2-4. C + O2 → CO2 (2-1) 2C + O2 → 2CO (2-2) 5 C + CO2 → 2CO (2-3) 2CO + O2 → 2CO2 (2-4) The graphite oxidation rate is strongly temperature dependent [19]. Generally, the oxidation of graphite can be divided into three temperature-dependent regimes: (1) the chemical regime for temperature below 500 ◦ C, (2) the in-pore diffusion controlled regime for temperature ranging from 500 - 900 ◦ C, and (3) the boundary layer diffusion controlled regime for temperature higher than 900 ◦ C [21], as illustrated in Fig. 2.2. Figure 2.2: Temperature regimes of graphite oxidation. In the chemical regime, the chemical reaction rate between the graphite and reaction gas is relatively low, which allows the oxidizing gas to penetrate deep into the graphite. Due to the uniform concentration of reaction gas in the graphite, the oxidation rate tends to be relatively uniform across the depth of the graphite. The oxidation rate in this regime is mainly controlled by the chemical reaction rate, and the oxidation rate is shown to increase with temperature. In the in-pore diffusion controlled regime, the chemical reaction rate is faster but the gas can still diffuse into the graphite. The reaction rate is determined by the diffusion rate of the oxidizing gas into the graphite and the chemical reaction rate. This means that deeper in the graphite the reaction rate is slower due to a lower concentration of oxidizing gas. In the boundary layer diffusion controlled regime, the chemical reaction rate of graphite and oxidizing gas is fast, which hinders the diffusion of oxidizing gas into the graphite. As a result, the reaction occurs only near the surface of the graphite with almost no reaction deep inside the graphite. Given the fast chemical reaction rate, the oxidation rate in this regime is mainly dictated by the rate of diffusion of oxidizing gas into the graphite. Factors affecting the 6 diffusion rate and chemical reaction can be categorized into two main groups, namely the internal factors of graphite materials and the environmental factors during oxidation. The former comprises factors such as the porosity of the graphite, the impurity content the graphite, and the grain size of the graphite; whereas the latter comprises factors such as the reaction temperature, reaction gas flow rate, reaction gas concentration, ambient pressure, and graphite irradiation [19]. Past experimental works carried out to study graphite oxidation primarily focused on investigating the mechanisms of graphite oxidation and the changes in graphite physicochemical properties after oxidation. An extensive review of these studies was provided by Zhang et al. [19]. Additionally, other researchers performed experiments to study the mechanisms of air ingress, including the reverse U-tube facility by Hishida and Takeda [22], the NACOK facility [23], and the h-loop by Gould et al. [24]. These studies focused on the formation of loop-wide natural circulation through molecular diffusion and/or buoyancy effects after de-pressurization. Several computational studies have also been performed to model the air ingress phenomenon in HTGRs using system-level and CFD software. One of the system-level codes used was GAMMA by Lim and No [2] to model the High Temperature Engineering Test Reactor (HTTR)-simulated test facility, and by Kim et al. [25] to model the MHTGR-600 reactor. The SPECTRA code was used by Zheng and Stempniewicz [3] to model the NACOK facility. Another widely used code was TINTE, which was used by Zheng and Stempniewicz [3] to model the NACOK facility and numerous researchers to model the HTR-PM reactor [4–7]. RELAP5-3D was used by Davis et al. [26] to model the nitrogen/air ingress in the reverse U-tube and NACOK facilities. The work by Davis et al. [26] focused solely on the flow of nitrogen/air into the facilities without the consideration of oxidation, whereas, some forms of oxidation models were included in the other studies. Kim et al. [25] considered only the C/O2 reaction in their GAMMA model. Similarly, Zheng and Stempniewicz [3] included only the C/O2 reaction in their TINTE model, though, an additional C/CO2 reaction was considered in their SPECTRA model. The models by Lim and No [2], Liu et al. [5], and Xu et al. [4, 6, 7] considered the C/O2 , C/CO2 , and CO/O2 reactions. In these studies, oxidation rate was modeled using a series of correlations that were dependent on factors such as temperature, partial pressure, gas velocity, and gas molar concentration. Furthermore, due to different concentration of oxygen and temperature, the oxidation rate and the dominance of oxidation reactions were shown to vary spatially within the core. CFD analysis was performed by Oh et al. [8] to study the patterns of the stratified flow and counter- current flow in the lower plenum after the core of the 600 MWth MHTGR reactor is de-pressurized. Liu et al. [27] performed CFD analysis to investigate the influence of temperature and Reynolds number on the graphite oxidation shape function, defined as the relation between the maximum and the average burn-off of a pebble, under different pebble stacking configurations. Ferng and Chi [9] used CFD to investigate the characteristics of graphite oxidation corrosion in the HTR-10 reactor. Franken et al. [28] constructed a CFD model of the h-loop to study the transition of air ingress from molecular diffusion to natural circulation. The above discussion highlights that air ingress in an HTGR is a complex phenomenon that remains an active area of research. The onset of natural circulation in the core is shown to be dependent on the reactor design. On top of that, the oxidation of graphite after the ingress of air into the core adds further complexity to the problem. Previous analysis tools have addressed important parts of this problem, but has certain limitations. Legacy system analysis code such as RELAP5-3D have demonstrated capabilities for molecular diffusion and natural circulation, but published applications have generally focused on selected phenomena rather than the full accident sequence with graphite oxidation and long-term response. Codes 7 such as GAMMA+, SPECTRA, and TINTE have been used to model reactor-scale air-ingress transients and include varying oxidation treatment, but their applicability is often tied to specific reactor concepts, legacy implementations, or limited extensibility to additional coupled physics and new designs. CFD studies have provided valuable insight into local stratification, countercurrent flow, and lower-plenum mixing, but are typically limited to shorter time scales, reduced domains, or simplified chemistry, and therefore are not practical as the primary tool for routine full-plant multi-day safety analysis. Therefore, integrating air-ingress capability in a modern system analysis code such as SAM is important for HTGR safety analysis and licensing. Built on the MOOSE multiphysics framework, SAM can perform whole-plant air-ingress analysis extensible to new reactor designs and can be coupled to other physics such as neutronics, structure mechanics, fuel performance, and other models as needed. It is also developed under modern software quality assurance practices with verification and validation, so that a single, well-supported tool can serve needs for safety analysis and licensing for advanced HTGR designs. The next section describes the capabilities recently developed in the SAM code to model the air-ingress event. 8 3 Developed capabilities for Air-ingress Modeling To model the full sequence of HTGR air-ingress accidents, three related capabilities were developed in SAM: a multi-component gas mixture flow model with molecular diffusion, an isentropic critical flow model for a small break discharge, and a graphite oxidation model with oxygen consumption and heat release. Collectively, these models cover the principal phases of the accident sequence introduced in Section 2: the critical flow model represents the initial choked depressurization discharge; the multi-component flow model captures the gas mixing and the density-driven and diffusion-driven ingress, and later loop-wide natural circulation; and the graphite oxidation model represents the oxygen consumption and exothermic heat release that govern the later oxidation stage. Each capability is described in turn below. 3.1 Multi-component Flow Model As discussed before, when a rupture happens on the main coolant pipes on the body of the reactor pressure vessel during a DLOFC accident in the HTGR, the air ingress scenario proceeds from diffusion to natural circulation, which leads to considerable rate of air ingress to severely oxidize the fuel and core [29]. To capture the graphite oxidation damage to the fuel and core because of potential air ingress during DLOFC, it is important to track the temporal and spatial distributions of different gas species. For this reason, a multi- component gas mixture model is needed. For simplicity, a two-component gas mixture flow model, involving component 1 and 2, is implemented in SAM. Assuming mechanical and thermal equilibrium between the two gas components, i.e., traveling at the same speed and having the same temperature. This mechanical and thermal equilibrium assumption is appropriate for the present system level application, where the slip between different gas components and temperature non-equilibrium are negligible. The two-component flow model can be described by the mass conservation equation for component 1 (or 2), and a set of mass, momentum, and energy balance equations for the gas mixture. In one-dimensional, these equations are given as follows   ∂ ρ1 ∂ ρ1 u ∂ ∂ X1 + − ρD = 0, (3-1) ∂t ∂x ∂x ∂x ∂ ρ ∂ ρu + = 0, (3-2) ∂t ∂x ∂ ρu ∂ (ρuu + p) f + + ρg + ρu|u| = 0, (3-3) ∂t ∂x 2Dh ∂ ρh ∂ ρuh ∂ + + [J1 (h1 − h2 )] − q′′′ = 0. (3-4) ∂t ∂x ∂x In the equations above, the subscript ‘1’ denotes for component 1. Symbols without subscripts denotes for the gas mixture. D is the binary diffusion coefficient, X1 = ρ1 /ρ is the mass fraction of component 1, and J1 = −ρD ∂ X1 /∂ x is the diffusion flux of component 1 following Fick’s law; the derivation of D is discussed 9 later in this section. The four nonlinear equations are solved for the four independent variables, mole fraction of component 1 (y1 ), mixture pressure (p), mixture velocity (u), and temperature (T). This equation set is closed with the equation of state of both gas components as well as the gas mixture, which are discussed as follows. Because both components contribute to the total mixture pressure, the partial pressure of each component is needed to determine its thermodynamic and transport properties. The Dalton’s Law of partial pressure is used to approximate the partial pressure of the components, such that p1 = py1 , p2 = p(1 − y1 ) (3-5) Note that y2 = 1 − y1 . The density, specific enthalpy, viscosity, and thermal conductivity of each component can then be obtained from the equation of state for each component, ρi = ρi (pi , T ), (3-6) hi = hi (pi , T ), (3-7) µi = µi (pi , T ), (3-8) ki = ki (pi , T ), (3-9) The gas mixture density ρ, by definition, is the summation of the densities of both components, ρ = ρ1 + ρ2 , (3-10) and the specific enthalpy of the gas mixture is, ρ1 h1 + ρ2 h2 ρ1 h1 + ρ2 h2 h= = , (3-11) ρ1 + ρ2 ρ The specific heat capacity of the gas mixture is, ρ1 c p,1 + ρ2 c p,2 cp = , (3-12) ρ The transport properties of a gas mixture are more complex to model. In this work, we use the method of Wilke [30] to model the viscosity, µ, of a gas mixture of two components, y1 µ1 y2 µ2 µ= + , (3-13) y1 + y2 φ12 y2 + y1 φ21 where 10 h i2 1 + (µ1 /µ2 )1/2 (M2 /M1 )1/4 φ12 = , (3-14) [8 [1 + (M1 /M2 )]]1/2 and µ2 M1 φ21 = φ12 , (3-15) µ1 M2 where M1 and M2 are the molecular weight of component 1 and 2, respectively. For the thermal conductivity of the gas mixture, the empirical relation proposed by Wassiljewa [30] is used. The Wassiljewa equation takes a form analogous to the theoretical relation for mixture viscosity. Using Mason and Saxena modification [30], and taking the numerical constant ε = 1 in the modification, the mixture thermal conductivity can be expressed as, y1 k1 y2 k2 k= + , (3-16) y1 + y2 φ12 y2 + y1 φ21 which takes a similar form as Eq. 3-13 for the viscosity of gas mixture. Note that the mass, momentum, and energy balance equations for the gas mixture (Eqs. 3-2–3-4) share the same form as the single-phase flow model already implemented in SAM; their FEM implementation and stabilization are discussed in detail in SAM’s Theory Manual [31]. The component-1 continuity equation (Eq. 3-1) and the mixture energy equation (Eq. 3-4) contain additional diffusion terms due to gradient of mass/molar fractions; their FEM implementations are discussed below. The SUPG stabilization term for Eq. 3-1 is formulated using its primitive form. Substituting ρ1 = ρX1 and applying the mixture continuity equation (Eq. 3-2), the primitive form is obtained as,   ∂ X1 ∂ X1 ∂ ∂ X1 ρ + ρu − ρD = 0. (3-17) ∂t ∂x ∂x ∂x Note that in the SUPG residual, the second-order diffusion contribution Dρ ∂ 2 X1 /∂ x2 , arising from expansion of the diffusion Laplacian, is omitted as an approximation to facilitate Jacobian computation. The weak form of Eq. 3-1 with PSPG/SUPG stabilization is given as,       ∂ ρ1 dψ ∂ X1 ∂ ψ ∂ X1 , ψ − ρ1 u, + ⟨ρ1 un̂, ψ⟩ + ρD , − ⟨ρD n̂, ψ⟩ ∂t dx ∂x ∂x ∂x   ∂ X1 ∂ X1 ∂ ρ ∂ X1 dψ + ρ + ρu −D , τSUPG = 0, (3-18) ∂t ∂x ∂x ∂x dx in which ψ is the test function and the last term is the stabilization term. The SUPG stabilization of the energy equation (Eq. 3-4) is similarly formulated using its primitive form. 11 In Eq. 3-4, the enthalpy diffusion term J1 (h1 − h2 ) arises because the two gas components have different specific enthalpies; for binary diffusion the fluxes satisfy J2 = −J1 , so the net enthalpy transport by diffusion is J1 h1 + J2 h2 = J1 (h1 − h2 ). Expanding Eq. 3-4 and applying the component continuity equations, the primitive form is obtained as, ∂T ∂T ∂T ρc p + ρc p u + J1 (c p,1 − c p,2 ) − q′′′ = 0, (3-19) ∂t ∂x ∂x where c p = (ρ1 c p,1 + ρ2 c p,2 )/ρ is the mixture specific heat capacity. The weak form of Eq. 3-4 with PSPG/SUPG stabilization is given as,       ∂ ρh dψ dψ , ψ − ρuh, + ⟨ρuhn̂, ψ⟩ − J1 (h1 − h2 ) , + ⟨J1 (h1 − h2 ) n̂, ψ⟩ ∂t dx dx   ∂T ∂T ∂T ′′′ dψ + ρc p + ρc p u + J1 (c p,1 − c p,2 ) − q , τSUPG = 0. (3-20) ∂t ∂x ∂x dx In addition to the one-dimensional components, multi-component gas mixture model is also implemented in the SAM zero-dimensional components such as PBBranch and PBVolumeBranch. PBBranch models a zero volume 0-D flow junction where multiple 1-D fluid components are connected. The governing equations of the multi-component gas mixture model without diffusion in the zero-volume PBBranch component are: n ∑ (ρ1 uAn)i = 0 (3-21) i=1 n ∑ (ρuAn)i = 0 (3-22) i=1 ∑ni=1 (ρuA)i ubranch = if (u · n)i > 0 (3-23) ρbranch Abranch ∑ni=1 (ρuAH)i Hbranch = if (u · n)i > 0 (3-24) ρbranch ubranch Abranch In which, 12 ρ1 : the density of component 1 at the connecting nodes; ρ: the mixture density at the connecting nodes; u: the mixture flow velocity at the connecting nodes; n: the flow direction at the connecting nodes; ubranch : the flow velocity at the branch; A: the flow area of the connecting components; n: the number of connecting components; H: the enthalpy at the connecting nodes; Hbranch : the mixture enthalpy at branch. Similar to the single-phase flow model, the above equations for multi-component gas mixture model are implemented in SAM using the MOOSE NodalConstraint system, along with the use of scalar variables to represent the y1 , p, u, and T of the PBBranch. The boundary conditions of the connecting pipe nodes are dependent on the branch conditions: If flow into branch, Ppipe = Pbranch + ∆Pacc + ∆Pf orm (3-25) If flow out of branch, Ppipe = Pbranch + ∆Pacc − ∆Pf orm (ρ1 u) pipe = ρ1branch u pipe (3-26) (ρu) pipe = ρbranch u pipe (ρuH) pipe = ρbranch u pipe Hbranch In which, 1 1 ρu2 branch − ρu2 pipe   ∆Pacc = 2 2 1 2  ∆Pf orm = K ρu pipe 2 Different than the PBBranch, PBVolumeBranch considers the volume effects of a junction component so that it can account for the mass and energy in-balance between inlets and outlets due to inertia. The governing equations of the mass and energy conservation of the multi-component gas mixture model for the PBVolumeBranch can be given as: n d (ρ1V ) − + ∑ (ρ1 uAn)i = 0 (3-27) dt i=1 n d (ρV ) − + ∑ (ρuAn)i = 0 (3-28) dt i=1 13 n d (ρHV ) − + ∑ (ρuAHn)i = 0 (3-29) dt i=1 In which, ρ1 : the density of component 1 at the PBVolumeBranch; ρ: the density at the PBVolumeBranch; V: the volume of the PBVolumeBranch; t: time. The diffusion model is also implemented in the PBBranch and PBVolumeBranch components by adding the following diffusion term to the LHS of Eqs. 3-21 and 3-27 for PBBranch and PBVolumeBranch, respectively: n n   X1,down − X1,up ∑ (ρ1 uAn)i − ∑ DAρ ∆x ·n =0 (3-30) i=1 i=1 i n n   d (ρ1V ) X1,down − X1,up − + ∑ (ρ1 uAn)i − ∑ DAρ · n = 0, (3-31) dt i=1 i=1 ∆x i where the subscripts up and down represent upstream and downstream according to the channel orientation. For instance, if the outlet of a channel is connected to a branch, X1,pipe is X1,up whereas X1,branch is X1,down . Conversely, if the inlet of a channel is connected to a branch, X1,branch is X1,up whereas X1,pipe is X1,down . Here, ρ is the total mixture density at the connecting pipe node, and ∆x is the length scale which is taken as half the size of the element to which the branch is connected. For PBTDJ and PBTDV, a flux boundary condition is applied for X1 , X1,down − X1,up ρ1 un − Dρ ·n (3-32) ∆x The up and down subscripts represent upstream and downstream that are defined according to the channel orientation similar to that described above for the branch components. For instance, if the outlet of a channel is connected to a PBTDJ/PBTDV, X1,pipe is X1,up whereas X1,BC is X1,down . Conversely, if the inlet of a channel is connected to a PBTDJ/PBTDV, X1,BC is X1,up whereas X1,pipe is X1,down . Here, ρ is the total mixture density at the connecting pipe node and ∆x is taken as half the size of the boundary element. According to [20], diffusion coefficient can be calculated using the formula provided by [32] as, 0.001858T 3/2 (1/MA + 1/MB )1/2 DAB = 2 Ω , (3-33) PσAB D 14 where the subscript AB represent components A and B of the gas mixture, T is temperature in [K], M is molecular weight in [g/mol], P is absolute pressure in [atm], σAB is the collision diameter (Lennard-Jones parameter) in Å, and ΩD is the collision integral. Note that the above formulation has a unit of [cm2 /s], which must be converted to [m2 /s] in SAM. According to [32], σAB can be calculated as, σA + σB σAB = , (3-34) 2 where the σ of various gases are provided in Table 3.2. The collision integral, ΩD is a dimensionless function of the temperature and the intermolecular potential for one molecule of A and one molecule of B. The values of ΩD is listed as a function of T /(εAB /κ) in Table 3.1, where κ is the Boltzmann constant, which is 1.38 × 10−16 ergs/K. εAB can be calculated as, √ εAB = εA εB , (3-35) or with κ, p εAB /κ = (εA /κ) (εB /κ). (3-36) For example, to calculate the diffusion coefficient of carbon dioxide in air at 293 K and 1 atm, using the values from Tables 3.1 and 3.2, σ in Å εA /κ in K Carbon dioxide 3.996 190 Air 3.617 97 σA + σB 3.996 + 3.617 σAB = = = 3.806, (3-37) 2 2 p √ εAB /κ = (εA /κ) (εB /κ) = 190 × 97 = 136. (3-38) T 293 = = 2.16 (3-39) εAB /κ 136 Using T /(εAB /κ) = 2.16 and Table 3.1, ΩD = 1.047. And finally, using Eq. 3-33 and the values of σAB and ΩD calculated above, 15 0.001858 × (293)3/2 (1/44 + 1/29)1/2 DAB = = 0.147cm2 /s = 1.47 × 10−5 m2 /s. (3-40) 1 × 3.8062 × 1.047 Table 3.1: The collision integrals, ΩD based on the Lennard-Jones potential. [32]. T /(ε/κ) ΩD T /(ε/κ) ΩD T /(ε/κ) ΩD 0.3 2.662 1.6 1.167 2.9 0.9576 0.35 2.476 1.65 1.153 3.0 0.949 0.4 2.318 1.7 1.14 3.1 0.9406 0.45 2.184 4.5 0.861 3.2 0.9328 0.5 2.066 4.6 0.8568 3.3 0.9256 0.55 1.966 4.7 0.853 3.4 0.9186 0.6 1.877 4.8 0.8492 3.5 0.912 0.65 1.798 4.9 0.8456 3.6 0.9058 0.7 1.729 5.0 0.8422 3.7 0.8998 0.75 1.667 6.0 0.8124 3.8 0.8942 0.8 1.612 7.0 0.7896 3.9 0.8888 0.85 1.562 8.0 0.7712 4.0 0.8836 0.9 1.517 1.75 1.128 4.1 0.8788 0.95 1.476 1.8 1.116 4.2 0.874 1.0 1.439 1.85 1.105 4.3 0.8694 1.05 1.406 1.9 1.094 4.4 0.8652 1.1 1.375 1.95 1.084 10 0.7424 1.15 1.346 2.0 1.075 20 0.664 1.2 1.32 2.1 1.057 30 0.6232 1.25 1.296 2.2 1.041 40 0.596 1.3 1.273 2.3 1.026 50 0.5756 1.35 1.253 2.4 1.012 60 0.5596 1.4 1.233 2.5 0.9996 70 0.5464 1.45 1.215 2.6 0.9878 80 0.5352 1.5 1.198 2.7 0.977 90 0.5256 1.55 1.182 2.8 0.9672 The present implementation is limited to a two-component mixture, which is sufficient for the helium–air system considered here, with air treated as a single equivalent component. To resolve the complete oxidation chemistry, the model would need to be extended to track additional components such as N2 , O2 , CO, and CO2 separately. This extension would need generalizing the mass conservation equations and mixture 16 Table 3.2: Lennerd-Jones coefficients for various gases [32]. Compound σ [Å] ε/κ [K] Acetylene 4.221 185.0 Air 3.617 97.0 Argon 3.418 124.0 Arsine 4.06 281.0 Benzene 5.27 440.0 Bromine 4.268 520.0 i-Butane 5.341 313.0 n-Butane 4.997 410.0 Carbon dioxide 3.996 190.0 Carbon disulfide 4.438 488.0 Carbon monoxide 3.59 110.0 Carbon tetrachloride 5.881 327.0 Carbonyl sulfide 4.13 335.0 Chlorine 4.115 357.0 Chloroform 5.43 327.0 Cyanogen 4.38 339.0 Cyclohexane 6.093 324.0 Ethane 4.418 230.0 Ethanol 4.455 391.0 Ethylene 4.232 205.0 Fluroine 3.653 112.0 Helium 2.576 10.22 n-Heptane 8.88 282.0 n-Hexane 5.909 413.0 Hydrogen 2.968 33.3 Hydrogen chloride 3.305 360.0 property models to an arbitrary number of components, replacing the single binary diffusion coefficient with a multi-component diffusion model, and providing thermodynamic and transport property data for each added gas components. 17 3.2 Critical Flow Model In some scenarios where there is a break in the primary loop the helium leaking into the cavity may undergo critical flow conditions (choked flow) due to the high pressure of the primary system. In this scenario the helium leaking out system will be choked and will be traveling at the speed of sound. To capture the phenomenon, a critical flow model is developed in SAM based on the isentropic ideal gas formulation [33]. It is intended for the modeling of a break in a pressurized system where fluid is evacuated to the ambient, with the break treated as a converging-diverging nozzle. The model is implemented in a PBVolumeBranch component and is enabled with a user-defined function. Once a break happens, the model first calculates the Mach number, Ma, using Eq. 3-41 which describes the ratio between the ambient pressure, Pamb and the reference pressure, P0 , γ ! γ−1 Pamb 1 = , (3-41) P0 1 + γ−1 2 Ma 2 C where γ is the specific heat ratio defined as Cvp . The reference pressure is defined as the pressure of the PBVolumeBranch. If the Mach number is greater than 0.3, which is generally accepted as the point beyond which compressibility effects start to matter, the density, ρbreak and temperature, Tbreak at the break, can then be calculated as, 1 ! γ−1 ρbreak 1 = , (3-42) ρ0 1 + γ−1 2 Ma 2 and Tbreak 1 = γ−1 . (3-43) T0 1+ Ma2 2 With that, the critical velocity at the break can be calculated as, p vcritical = Ma RTbreak γ, (3-44) where R is the molecular gas constant defined as, molar mass R= , (3-45) Universal gas constant and the critical mass flow rate can be calculated as, 18 ṁcritical = vcritical ρbreak Abreak , (3-46) where Abreak is the break area. When the Mach number is smaller than 0.3, compressibility effects are ignored. The leakage velocity, vleak , is calculated based on the pressure difference between the PBVolumeBranch and the ambient as, 1 ∆P = kloss ρv2leak , (3-47) 2 where ∆P is the pressure difference. The density, ρ is either the branch density or the ambient density depending on the direction of the flow. The loss coefficient, kloss is a user-defined value that describes the form loss as flow undergoes contraction and expansion while flowing out from the system. Similarly, the leakage mass flow rate is calculated as, ṁleak = ρvleak Abreak . (3-48) In the current implementation, the leakage mass flow rate of Eqs. 3-46 and 3-48 are added to the mass balance equation of PBVolumeBranch as a sink term. Additionally, another sink term is added to the energy balance equation as, ḣleak = ρvleak Abreak h, (3-49) where ḣleak is the rate of enthalpy leakage and h is either the enthalpy of the branch or enthalpy of the ambient based on flow direction. 19 3.3 Graphite Oxidation Model An initial capability for calculating the graphite oxidation reaction rate was completed previously to support source term evaluation modeling in FHRs [34] and is re-evaluated and discussed in this section for air-ingress modeling in HTGRs. As discussed in Section 2, there are several possible chemical reactions that can occur between the gas and the graphite, the simplest reaction being carbon dioxide production from carbon and oxygen, defined in Eq. 2-1. This reaction is calculated in SAM using the Roes correlation [35], as shown in Eq. 3-50 for graphite oxidation rate R: " #−1 16140 −1  −4 0.65 0.34 −1   2 − T  PO2 R [kg/m s] = 7500e + 8.03 × 10 v T 105 (3-50) which is a function of the local graphite surface temperature T (K), gas velocity v (m/s), and oxygen partial pressure Po2 (Pa). This reaction rate is computed as an auxiliary calculation based on local flow variables and a user-defined oxygen partial pressure and can be integrated over the solid surface to estimate the total oxidation in a graphite volume. The Roes correlation captures the combined dependence of oxidation rate on graphite temperature, local oxygen availability, and gas velocity. It represents the two principal rate-limiting processes of graphite oxidation. One is the chemical-kinetics rate that governs the reaction at lower temperatures, where oxygen is abundant at the surface and the intrinsic reactivity of the graphite limits consumption. The other is the mass-transfer-limited supply of oxygen across the boundary layer to the graphite surface; This process depends on the gas velocity and dominates at higher temperatures, where the surface reaction is fast and the overall rate is controlled by how quickly oxygen can be delivered. To fully capture the limiting effect of oxygen consumption on the progression of oxidation, the calculated rate should impose a sink term on the oxygen component of the multi-component gas mixture and introduce carbon dioxide in its place. However, this adds several complexities, including change in the diffusivity and overall enthalpy of the mixture components. In the current scope, we focus on the impact of oxidation once air ingression is well-established while still considering the role of oxygen consumption in limiting the rate of oxidation. As an engineering simplification, we model the transport of oxygen as a passive scalar concentration that advects with the mixture flow velocity with an additional sink term SR determined by the local oxidation rate R and boundary conditions informed by the multicomponent gas mixture model. ∂ co2 + ∇ · (umix co2 ) = SR (3-51) ∂t In addition, the oxidation reaction modeled by the Roes correlation (Eq. 2-1) is exothermic, producing 394 kJ/mol and in many cases accelerating the overall reaction rate. This reaction heat should be distributed according to the location of reaction (see Fig 2.2). For the present analyses on safety-limiting scenarios, we will focus on Regime 3 which occurs at elevated temperatures and predominantly at the graphite surface. In this analysis, we capture the heat of oxidation by applying it locally at the graphite surface as a coupled heat flux boundary condition defined by the local reaction rate R at 394 kJ/mol of consumed carbon. The present oxidation model includes only direct oxidation by oxygen (Eq.2-1) and does not yet include the Boudouard reaction, C + CO2 → 2CO (Eq.2-3). This reaction becomes increasingly important at higher temperatures, particularly above about 900 ◦ C, where carbon monoxide production can compete with the 20 direct oxidation reaction. Incorporation of this reaction would require extending the current two-component gas model to additional gas components, together with the corresponding transport and property models. The Boudouard reaction will be incorporated in oxidation model in the future. These three newly implemented SAM modeling capabilities for air-ingress scenarios have been verified through a series of verification cases, including comparisons against analytical solutions and benchmark problems, as documented in the previous report [36]. The present report therefore focuses on the validation of these capabilities against experimental data, presented in the following section. 21 4 Validation 4.1 NSTF Heavy-gas Ingress Event Validation The Natural Convection Shutdown Heat Removal Test Facility (NSTF) heavy-gas ingress experiment is selected as a validation case for the SAM multi-component gas flow modeling capability. This experiment measured the response of a buoyancy-driven loop when the inlet gas composition transitions from air to a denser non-air gas. Under this transient, the system response is governed not only by the imposed heat load and hydraulic resistance, but also by the spatial and temporal evolution of gas composition, the corresponding changes in mixture density, and the heat up of the gas inventory within the heated channels. The experiment therefore provides a valuable validation case of the developed SAM multi-component gas flow model for air-ingress events analysis, especially for the system level behavior associated with composition-dependent buoyancy effects. 4.1.1 NSTF Facility Description The NSTF is a large-scale integral effects experimental facility constructed at Argonne National Laboratory to study passive air-cooled RCCS behavior under prototypic natural circulation conditions [37]. The facility is designed at half-scale of the RCCS of the 350 MWth MHTGR [38] and preserves the major flow and heat transfer features of an air-cooled cavity cooling system, including the inlet downcomer, inlet plenum, heated cavity with riser channels, outlet plenum, and chimney section. The overall facility height is nearly 26 m, providing a substantial elevation difference between the inlet and outlet regions and allowing development of a measurable natural circulation driving head. The heated wall is driven by electric heaters to simulate the RPV outer surface, and the facility is instrumented with high resolution sensors that measure mass flow rate, temperature distributions, and gas composition. These features make the NSTF well suited for validation of system analysis codes for passive decay heat removal applications. During normal operation, ambient air is drawn into the facility through the inlet region and flows downward through the inlet downcomer. The gas then enters the inlet plenum, where it is distributed to twelve parallel riser ducts. Heat is supplied to the riser section through the electrically heated wall, increasing the gas temperature and reducing its density as it flows upward. The heated gas enters the outlet plenum and discharges through two chimney ducts to the atmosphere. The density difference between the relatively cool gas in the inlet downcomer and the hot gas in the riser and chimney sections produces the buoyancy head that drives the air circulation. The NSTF facility is illustrated schematically in Fig. 4.1. 4.1.2 Heavy-gas Ingress Experiment The heavy-gas ingress test was designed to examine the effects of introducing a dense, non-air gas into the inlet of an air-cooled RCCS flow path. This scenario is motivated by the potential for an external heavy-gas released from nearby chemical processing facilities that enters the passive air-cooled RCCS of HTGRs. Ingress of a heavy gas can modify the natural circulation driving head and thereby reduce the effectiveness of passive heat removal. The NSTF experiment therefore provides valuable data for validating 22 Figure 4.1: NSTF layout and primary segments. Adapted from [39]. system codes used to model multi-component gas behavior in natural circulation systems. In the test, the facility was first operated at the normal air-cooled condition. After the transient begins, the inlet gas is switched to argon. Argon was selected as the heavy gas for the experiment. Although carbon dioxide is also relevant to industrial ingress scenarios, argon provides a clear density effect while avoiding some of the additional exposure hazards associated with carbon dioxide. To perform the test, an argon enclosure with an internal volume of approximately 1,200 ft3 (roughly twice the internal gas flow volume of the NSTF loop) was installed above the inlet downcomer. The enclosure is 7 × 10 × 15 feet, as shown in Fig. 4.2. The enclosure operation modes are shown in Fig. 4.3. A 24-inch air duct was placed at the center of the enclosure to provide the normal inlet path during steady-state air operation, with a mating seal at the lower end of the duct that isolated the argon inventory from the NSTF loop. The enclosure was filled with argon before the transient, and prior to transient initiation the measured enclosure composition was greater than 98% argon with an oxygen concentration below approximately 0.4%. At the beginning of the transient, bypass vent valves were opened and the mating seal at the lower end of the air duct was broken. This transition established a flow path that allowed argon from the enclosure to be drawn into the inlet downcomer. Oxygen sensors were installed in both the argon enclosure and the inlet plenum to track the local evolution of gas composition during stagnation and recovery, while loop mass flow rate was measured as the primary integral indicator of natural circulation performance. Together, the flow rate and oxygen concentration measurements provide the principal validation targets for the SAM simulation. A companion transient was also performed with air placed in the enclosure prior to transient initiation. This air transition test confirmed that the mechanical actions associated with opening the enclosure path and breaking the mating seal did not by themselves cause the observed loss of natural circulation; the strong response observed in the argon test is therefore attributed to the gas composition change and associated density effect rather than to the actuation procedure itself. 4.1.3 SAM Model Description The SAM model used for this validation extends the previously developed single-gas NSTF model [40] to include two gas components, air and argon. The model preserves the major flow paths, heat structures, 23 (a) Diagram of the argon enclosure positioned above the NSTF inlet. (b) Top view of the argon enclosure. Figure 4.2: Argon enclosure in the NSTF heavy-gas ingress experiment. Adapted from [37] Figure 4.3: Operating modes of the NSTF heavy-gas enclosure: normal air draw (left), and argon bypass (right). Adapted from [37]. 24 control volumes, and boundary conditions of the NSTF. A schematic of the SAM model is shown in Fig. 4.4. The NSTF flow path is represented using one-dimensional pipe components for the inlet downcomer, heated riser section, and chimney, together with zero-dimensional PBVolumeBranch components for the inlet and outlet plenums. The twelve physical riser ducts are combined into a single equivalent riser duct that preserves the total flow area and the overall hydraulic response of the loop; the total electrical heat input is imposed as a uniform heat flux on the equivalent riser wall. Similarly, the two chimney ducts above the outlet plenum are modeled with a single equivalent chimney pipe, with a convective wall boundary condition that accounts for the inlet-to-outlet temperature variation along the chimney. This simplified representation is appropriate for predicting the overall natural circulation mass flow rate, although it cannot resolve individual riser channel behavior. A schematic of the SAM model is shown in Fig. 4.4. Figure 4.4: SAM model for NSTF heavy-gas ingress test. The upstream inlet region is modified to represent the heavy-gas ingress configuration. Two parallel flow paths are added upstream of the downcomer: one representing the central air duct and the other representing the argon enclosure. The argon enclosure itself is modeled as a zero-dimensional PBVolumeBranch with a specified initial 98% argon composition. During the steady-state initialization, the argon path is isolated from the rest of the loop by assigning large form loss coefficients at its inlet and outlet, while the air path remains 25 open with zero loss coefficients. At the transient initiation time, the form loss coefficients are switched so that the enclosure path becomes open and the air duct path becomes restricted, approximating the experimental transition in which the inlet flow is diverted from the central air duct to the surrounding argon enclosure. Because zero-dimensional volume components are used for both the enclosure and the inlet plenum, SAM predicts an average gas composition in these volumes rather than local sensor composition variations. To properly reproduce the ambient effects on the natural circulation loop [41], two virtual components representing the atmosphere of the NSTF building are included above the inlet of the argon enclosure and air duct. A vertical air column provides the correct hydrostatic head and ambient temperature of the air outside the NSTF building, with its inlet boundary condition set to the measured outdoor pressure and temperature. A horizontal pipe with wall heating is then used to match the measured NSTF inlet air temperature. The transient calculation is initialized from a converged steady-state air flow solution. Molecular diffusion is neglected in the base model, so the gas composition evolution is governed primarily by advection. Because air is treated as a single component in the two-component formulation, oxygen is not tracked as a separate species; the calculated oxygen concentration is instead inferred from the computed air mole fraction by assuming that oxygen accounts for 21% of air, which enables direct comparison with the oxygen sensors used in the NSTF experiment. 4.1.4 Results and Discussion Transient simulation is performed for the full NSTF heavy-gas ingress test. A null transient is first run to establish the nominal air-cooled steady-state operating conditions. The transient begins at 28.65 hr and is initialized from this converged steady state by switching the form loss coefficients of the air duct and argon enclosure paths, as described in the preceding section. The measured response and the corresponding SAM prediction are compared below in three phases that follow the evolution of the transient: rapid argon entry into the inlet region, degradation of natural circulation toward near stagnation, and heat up and recovery. In the experiment, the inlet plenum oxygen concentration decreased sharply from the normal air value immediately after the transition, dropping to approximately 11% within about 2 s and to approximately 1.8% within about 6 s. This rapid decrease indicates that the inlet plenum became argon-rich very quickly after the enclosure was opened. The SAM calculation reproduces this rapid depletion: the predicted inferred oxygen concentration in the inlet plenum, shown in Fig. 4.5, drops sharply after the argon path is opened and then rises to an intermediate value of approximately 10% as a mixture of air and argon flows from the enclosure into the inlet plenum. Because the enclosure and inlet plenum are modeled as zero-dimensional well-mixed volumes, SAM predicts an average composition rather than the local measured value. Short period oscillations are also observed in the predicted mass flow rate immediately after the transition (see Fig. 4.6). The short initial spike and oscillations in the predicted mass flow rate are considered to be primarily physical, arising from the abrupt reconfiguration of the inlet flow path when the enclosure path is suddenly opened with higher buoyancy force. The spike and oscillations dissipate within a few seconds and do not affect the predicted long-term flow degradation or recovery behavior, so they have no impact on the validation conclusions. After argon entered the downcomer and inlet plenum, the measured loop flow rate decreased rapidly and approached an essentially stagnant condition within approximately 90 s. The system mass flow rates 26 Figure 4.5: Oxygen concentrations measured in the inlet plenum top and bottom, and average value from SAM simulation. predicted by SAM and measured in the experiment are compared in Fig. 4.6. The SAM prediction follows the experimental trend closely, with the loop mass flow rate decreasing rapidly after the argon path is opened and approaching a near-stagnant condition on the same time scale as the experiment. The mechanism captured by the model is identical to that inferred from the experiment: dense argon accumulates in the inlet downcomer and inlet plenum, increasing the density of the lower loop and reducing the net buoyancy head, so that the hot riser and chimney gases alone can no longer sustain the initial air-cooled natural circulation flow rate. The agreement in Fig. 4.6 therefore indicates that the SAM multi-component gas model captures the density-driven suppression of natural circulation. During the stagnation period, the heated riser wall continued to transfer energy to the gas in the heated section. In the experiment, the flow rate began to recover approximately 16–19 min after ingress and returned toward the normal natural circulation level, while the inlet region oxygen concentration also increased as argon was depleted from the enclosure and displaced from the lower loop region. The SAM calculation reproduces this recovery in Fig. 4.6: as the gas mixture in risers is heated, its density decreases and the restored buoyancy head eventually drives the loop back toward normal natural circulation. The underlying mechanism is illustrated in Fig. 4.7, which shows the predicted gas temperature and mixture density at the riser duct inlet, mid-height, and outlet. After flow stagnation, the gas temperature in the riser rises steadily as wall heat continues to be transferred to the gas, and the corresponding mixture density decreases. This density reduction in the heated section gradually overcomes the dense argon column in the lower loop and restores the buoyancy driving force. The predicted oxygen concentration in the inlet plenum, shown in Fig. 4.5, slowly increases during the stagnation period and finally approaches the pure air level once natural circulation is re-established, matching the qualitative experimental trend. 27 Figure 4.6: System mass flow rate during the NSTF heavy-gas ingress test. Figure 4.7: Predicted gas temperature and mixture density at the riser duct inlet, mid-height, and outlet during the NSTF heavy-gas ingress test. 28 The experiment also exhibited local multi-dimensional behavior during the low flow period, as shown in the oxygen concentration measurements in Fig. 4.5. The inlet plenum sensors at the top and bottom of the plenum do not follow the same evolution: the top sensor begins to recover shortly after flow stagnation, well before any measurable loop flow returns, while the bottom sensor remains argon-rich until the system flow rate itself starts to recover. The early increase at the top sensor is attributed to local recirculation within the riser tubes, with downward flow in some risers returning a small amount of partially mixed gas back into the top of the inlet plenum [37]. The present SAM model does not resolve these effects: the zero-dimensional enclosure and plenum components assume perfect mixing, so the predicted inlet plenum oxygen concentration in Fig. 4.5 represents an average value rather than the local top or bottom signal, and the single equivalent-riser representation precludes inter-channel redistribution. As a result, the SAM curve lies between the top and bottom experimental traces and captures the overall trend rather than the spread between sensors. These simplifications affect local oxygen concentration and motivate higher fidelity modeling of the inlet plenum and riser region in future work. The comparison shows that the SAM model captures the dominant integral response of the facility, including rapid oxygen depletion, flow degradation, near stagnation, recovery of natural circulation, and the late-time increase in oxygen concentration. Overall, the NSTF heavy-gas ingress validation demonstrates that the SAM multi-component gas flow model can correctly capture the density-driven suppression and thermal recovery of natural circulation at the system level, which is precisely the mechanism controlling the air ingress into the HTGR cores. The validation also identifies the expected limitations of the present system-level model: the perfect mixing assumption in large volumes and the equivalent-riser approximation limit prediction of local stratification, local oxygen concentration, and individual riser recirculation. These limitations are acceptable for the present integral validation objective and provide useful guidance for future higher fidelity modeling of local multidimensional effects. Table 4.1 summarizes the principal validation quantities, comparing the measured facility response with the SAM prediction and a qualitative assessment of the agreement. Table 4.1: Summary of the NSTF heavy-gas ingress validation Measured value/behavior SAM prediction Assessment Inlet plenum O2 after ingress drops to ∼1.8% within ∼6 s rapid drop to ∼10% Faira (local) (well-mixed average) Time to near-flow stagnation ∼109 s ∼108 s Good Onset of flow recovery ∼18 min after ingress ∼22 min after ingress Good Return to natural circulation recovers toward nominal flow recovers toward nominal flow Good Late-time inlet-plenum O2 returns to air level approaches pure-air level Good a Because the inlet plenum is modeled as a zero-dimensional, well-mixed volume, SAM predicts an average composition rather than the local top and bottom sensor values. 4.2 NACOK Validation 4.2.1 NACOK Facility Description The NACOK (Naturzug im Core mit Korrosion, “natural convection in the core with corrosion”) facility was constructed at Jülich Research Centre in Germany to investigate phenomena associated with 29 air-ingress accidents in HTGRs [42–44]. The test facility can carry out experiments in different configurations and different graphite materials, under different temperature profiles and varying boundary conditions. A schematic overview of the facility geometry is shown in Fig. 4.8. The NACOK configuration represents a reverse U-tube flow path that reproduces the essential circulation route after rupture of the coaxial duct in the HTR primary system [42]. Ambient gas enters through a horizontal lower inlet, flows upward through a tall heated experimental channel containing ceramic or graphite internals, passes through a horizontal cross connection near the top, and then descends through a vertical return tube representing a cold-helium riser channel in the reactor side reflector. The inlet and outlet are located at the same geodetic elevation, thereby mimicking double ended guillotine break (DEGB) accident flow path. The NACOK facility can also be reconfigured into an open-chimney arrangement by disconnecting the return leg and venting the test channel directly to the atmosphere. This chimney configuration has been used for air ingress experiment carried out on October 23, 2008 [44]. The main experimental channel has a square cross section of 0.30 m × 0.30 m and is assembled from modular welded sections of approximately 800 mm height fabricated from high-temperature austenitic furnace steel, with a wall thickness of 6 mm [42]. In the reverse U-tube configuration, the maximum height difference is approximately 7.334 m, including a packed-spheres section of about 5.0 m and an upper freeboard region representing the upper plenum. In the open-chimney configuration, the total effective height can reach about 8.0 m. The return tube has an inner diameter of 0.125 m, equal to that of the horizontal inlet tube, and the upper horizontal connection between the experimental channel and return leg is approximately 1.6 m long center-to-center. Figure 4.8: Overall arrangement of the NACOK test facility. Adapted from [44]. Independent electrical furnace sections are installed around both the main experimental channel and the return tube, allowing prescribed axial wall-temperature distributions to be imposed [42]. The installed power 30 is 12.0 kW per heated section of the main channel and 3.0 kW per heated section of the return tube, for a total installed heater capacity of about 147 kW. This arrangement permits wall temperatures up to roughly 1000 ◦ C in the reverse U-tube configuration and as high as about 1200 ◦ C in the chimney configuration, while the return tube can be heated up to about 800 ◦ C. Because each section is heated independently, NACOK can reproduce a broad range of axial temperature profiles relevant to different phases of an air-ingress event, including cases with a strongly heated hot leg and a cooler return leg. This capability is essential for natural-circulation studies, since the circulation rate is determined by the integrated density difference around the loop and by the distributed hydraulic resistance of the internals. NACOK is heavily instrumented and has been used as a benchmark facility for validation of reactor system-analysis and accident-analysis codes [42, 44–46]. The available measurements include inlet air mass flow rate, distributed wall and gas temperatures, differential pressures, and gas compositions. Up to 90 Type-K sheathed thermocouples are available, with 3.0 mm diameter probes in the experimental channel and 1.5 mm probes in the return tube. Differential pressures are measured with a capacitive micro-manometer having a manufacturer-stated accuracy of approximately ±0.5 Pa, enabling characterization of low-pressure-drop natural-circulation flow and packed-bed resistance. Gas composition can be sampled at as many as 26 locations. The inlet air mass flow is measured using a thermal sensor based on a constant-temperature- difference principle, with a calibrated operating range from below 0.5 g/s to approximately 17–18 g/s and a measurement uncertainty of about ±1.4% of reading [42]. Because NACOK provides simultaneous measurements of flow rate, pressure loss, temperature, and graphite corrosion, it has been used extensively for validation of TINTE, THERMIX/REACT, SPECTRA, RELAP5-3D, ATTICA3D, and related models [44–46]. It therefore serves as an important validation for the SAM graphite oxidation capability developed in this work. 4.2.2 NACOK Air-ingress Experiment Graphite oxidation experiments were performed in NACOK by replacing some or all of the nonreactive ceramic internals with graphite specimens representative of lower reflector blocks and pebble-bed graphite material [42]. One of the graphite oxidation test was conducted within the framework of the RAPHAEL project on October 23, 2008 with open-chimney configuration [44, 47]. In this test, the return tube was disconnected and the gas discharged directly from the top of the main experimental channel. A 0.30 m high packed bed of 10 mm ceramic spheres was placed in the upper portion of the channel to provide hydraulic resistance, while three levels of graphite blocks, each 0.30 m high, were installed in the lower heated part of the channel. Two types of internal block geometries were examined: Type 1 blocks containing 14 cylindrical holes of 16 mm diameter, and Type 2 blocks containing 2 holes of 26.5 mm diameter, as shown in Fig. 4.9. Two graphite grades, SGL NBG-17 and UCAR PCEA, were tested to determine the relative importance of internal geometry and graphite material on oxidation behavior [44]. The test procedures are described as follows [44]. The entire assembly was first heated under nitrogen until thermal equilibrium was reached, with the main-channel wall maintained near 900 ◦ C. Ambient air at approximately 20 ◦ C, or 293 K, was then admitted through the lower inlet. The inlet valve was adjusted to yield a target flow of 5 g/s, with the measured value during the test being 4.32 g/s. The external heaters maintained the wall temperature during the transient. In this configuration, the dominant hydraulic resistances were the inlet valve and the upper ceramic-sphere bed, while the graphite blocks themselves contributed 31 (a) Geometric types of graphite blocks: Type 1 (14 holes, 16 mm diameter) and Type 2 (2 holes, 26.5 mm diameter). (b) Arrangement of the four graphite blocks at one axial level: two SGL NBG-17 blocks and two UCAR PCEA blocks, one of each grade paired with each hole geometry. Figure 4.9: Graphite block configuration of NACOK experiment. Adapted from [44]. only a minor portion of the total pressure loss. After the inlet valve was opened, the experiment ran for approximately 8.5 h, during which the wall, graphite, and gas temperatures together with the inlet air mass flow and outlet gas composition were measured continuously. At the end of the run, the graphite blocks were removed and weighed individually, and the relative mass loss of each block was determined by comparison with its pre-test mass. Once air entered the lower heated graphite section, oxygen was transported upward through the graphite stack and reacted at the graphite surface and within the accessible pore structure. The relevant reactions are introduced in Eq. 2-1 to Eq. 2-4. Under the conditions of this chimney experiment, the measured outlet gas composition showed that CO2 was the dominant product, indicating that complete oxidation given in Eq. 2-1 was the primary reaction. Water-gas and water-shift reactions were neglected because the inlet air was dry 32 and the steam content was low [44]. 4.2.3 SAM Model Description The SAM model of the NACOK graphite oxidation experiment is based on the chimney test configuration on October 23, 2008 described in Section 4.2.2. Because the bulk gas during the test is primarily air, and the main quantity of interest is the graphite oxidation rate, the present model employs the SAM passive scalar transport capability with air as the fluid rather than the full multi-component gas formulation described in Section 3.1. Oxygen is treated as a passive scalar advected with the bulk flow and consumed locally in the graphite block channels through a sink term provided by the graphite oxidation model. The associated heat release is imposed as a heat source in the graphite heat structures. A schematic of the SAM model is shown in Fig. 4.10. The flow path is constructed using one-dimensional and zero-dimensional SAM components. A short horizontal inlet pipe, modeled as a PBOneDFluidComponent, supplies air to the inlet plenum. The inlet pipe hydraulic diameter is 0.125 m, consistent with the experimental inlet tube. The inlet plenum is represented by a zero-dimensional PBVolumeBranch with a flow area of 0.09 m2 , corresponding to the cross section of the 0.30 m × 0.30 m experimental channel. This plenum is located immediately below the heated graphite-block stack and distributes the incoming air to the parallel graphite-block channels. The three axial graphite-block levels are modeled explicitly. As in the SPECTRA NACOK model of Ref. [44], the differences in thermal and oxidation properties between the UCAR PCEA and SGL NBG-17 graphite grades are considered to be negligible, and the two graphite types are therefore not distinguished in the SAM model. At each level, the flow region is accordingly divided into two parallel one-dimensional channels. One channel represents the Type 1 block geometry, which consists of 14 holes with a diameter of 16 mm, and the other represents the Type 2 geometry, which consists of two holes with a diameter of 26.5 mm. Adjacent block levels within the same channel are connected by PBVolumeBranch components that represent the small inter-block gas volumes. Above the third graphite-block level, a PBVolumeBranch with a flow area of 0.09 m2 merges the two parallel channel systems. The upper chimney section is represented by a 0.725 m long one-dimensional pipe, followed by the 0.30 m high ceramic sphere packed bed and a 6.075 m long exhaust pipe. The ceramic sphere bed is modeled as a pipe with the SAM built-in KTA pebble bed pressure drop closure correlation. The chimney discharges to a PBTDV boundary condition maintained at atmospheric pressure. The solid graphite block is represented using two-dimensional cylindrical PBCoupledHeatStructure components coupled to the corresponding one-dimensional graphite-block flow channels. Because each equivalent one-dimensional channel represents the combined flow area of multiple holes in the physical graphite blocks, the actual multi-hole geometry is collapsed into an equivalent cylindrical conduction problem. To preserve the effective lateral heat-transfer behavior between the graphite solid and the gas channels, an effective thermal conductivity is assigned to the graphite material. Each graphite heat structure is surrounded by a cylindrical steel wall heat structure representing the NACOK experimental channel wall. The graphite and steel wall heat structures are coupled across the annular gap through RadiationHeatTransfer. Axial radiative exchange between adjacent graphite-block levels is also represented by SurfaceCoupling objects connecting the top and bottom surfaces of successive graphite heat structures. This treatment enables the model to capture both radial heat transfer from the wall to the graphite and axial radiation between neighboring block levels. 33 Figure 4.10: SAM model of the NACOK experiment. A radiative boundary condition (CoupledRadiationHeatTransferBC) is applied on the external surface of each steel-wall heat structure, with an ambient temperature of 1100 K that represents the heated surface temperature from the experiment data [44]. To simplify the model, the heat transfer between the graphite block bottom and the support structure is not modeled explicitly. Instead, a time-dependent temperature boundary condition derived from the experimental thermocouple data is applied directly at the lower surface of the bottom graphite block, as showed in Fig. 4.11). At the loop inlet, a PBTDJ component prescribes the measured inlet conditions: an air mass flow rate of 4.32 g/s, an inlet gas temperature of 293.15 K. The coupled oxygen transport and graphite oxidation feedback is implemented through the passive 34 1000 950 900 850 800 T ( C) 750 700 650 600 550 500 0 1 2 3 4 5 6 7 8 9 10 Time (hr) Figure 4.11: Lower boundary temperature imposed at the bottom of the graphite block stack in the SAM NACOK model. scalar equation and a local oxidation rate model. The passive scalar oxygen is transported with the bulk gas using the SAM passive scalar capability. The molecular diffusivity is set to zero, so species transport is governed by advection alone as discussed in Eq. 3-51. SAM built-in function evaluates the local graphite oxidation rate using the Roes correlation (Eq. 3-50). It is selected here because previous comparisons for the NACOK test showed that the Roes correlation provided the best overall agreement with measured graphite mass loss and temperature distributions [44]. The computed oxidation rate is used to calculate a local sink in the oxygen passive scalar transport equation on each graphite block channel. This sink term is converted from the mass of carbon consumed per unit graphite surface area into the corresponding oxygen consumption per unit channel volume. In addition, the oxidation rate is converted into an heat flux on the inner surface of the graphite heat structure by using the heat of complete graphite oxidation to CO2 (Eq. 2-1) is 394 kJ/mol. The local oxidation rate is integrated over the corresponding graphite surface area and over time to compute the cumulative graphite mass loss for each block. Therefore graphite mass loss can be compared directly with the measured NACOK data. 4.2.4 Results The SAM NACOK model results are compared with the experimental data in this section. First, the temperature evolutions at the bottom graphite level are compared, where incoming air first contacts the graphite stack and the highest oxidation rates are expected. Second, the cumulative graphite mass loss of each block is compared, as it reflects the integrated effects of graphite oxidation, including oxygen transport, oxygen depletion along the flow path, the thermal response of graphite, temperature and oxygen concentration dependent oxidation, and the feedback of oxidation heat release on the local reaction rate. 35 Although the upper graphite levels were initially hotter than the bottom level graphite before the test, the bottom level was exposed to the highest oxygen concentration because air entered from below and flowed upward through the stack. As the transient progressed, graphite oxidation reaction released more heat in the lowest blocks, and after about 2 h the bottom level became the hottest region of the stack according to the experimental measurements [44]. The temperature response at the bottom level is therefore a key indicator of whether the model correctly predicts the region that dominates graphite oxidation and mass loss. Figure 4.12 presents the comparison between the SAM predictions and the measured temperature at four bottom level thermocouple locations on the bottom level, including the graphite solid temperature and gas outlet temperature for the Type 1 and Type 2 block geometries. The locations of the four bottom level thermocouples in this comparison are shown in Fig. 4.13. For the Type 1 block, the predicted graphite solid temperature at 2.2, shown in Fig. 4.12a, and the predicted gas outlet temperature at 2.3, shown in Fig. 4.12b, agree closely with the measurements throughout the transient. SAM predicts a slightly faster heat up during the first two hours, but the predicted and measured temperatures remain in close agreement during the subsequent high temperature plateau. For the Type 2 block, larger deviations are observed at 21.3 and 22.1, shown in Figs. 4.12c and 4.12d. In the simulation, the temperatures decrease initially during the first hour then increase, whereas the measured temperatures increase monotonically from the start of the transient. Consequently, SAM underpredicts the Type 2 temperatures during approximately from 1 to 2.5 h, and then slightly overpredicts the Type 2 graphite solid temperature during part of the plateau. This discrepancy is mainly related to the simplified representation of the two large Type 2 holes as a single one-dimensional channel. At later times, the predicted plateau temperatures are consistent with the measurements. In the future, the graphite will be modeled using a three-dimensional heat conduction model, which is expected to improve the accuracy of the temperature distribution predicted by SAM. The integral graphite mass loss provides the primary validation metric for the NACOK test. The total graphite mass decreased from 129.538 kg before the experiment to 118.794 kg after the experiment, corresponding to an average mass loss of approximately 8.3% [44]. The measured oxidation was strongly nonuniform in the axial direction. On average, the bottom level lost about 19% of its mass on average, whereas the middle and top levels lost roughly 4% and 2%, respectively. This distribution indicates that oxidation was governed primarily by the availability of oxygen entering from the bottom of the stack. Block geometry also affected the mass loss. At the bottom level, the Type 1 blocks oxidized more rapidly than the Type 2 blocks. The influence of graphite grade was smaller than the influence of geometry, SGL NBG-17 showed slightly greater oxidation than UCAR PCEA [44]. Table 4.2 compares the measured per block graphite weight loss with predictions from SPECTRA, TINTE, and the present SAM model. The experimental, SPECTRA, and TINTE values are reproduced from Zheng and Stempniewicz [44]. In the SAM model, each axial level is represented by two equivalent graphite block channels, one for each hole geometry. The model does not distinguish between the SGL NBG-17 and UCAR PCEA graphite grades; therefore, a single SAM value is reported for each level-geometry combination and applied to both graphite grades in that combination. The SAM predictions reproduce the principal features of the measured mass loss distribution, as shown in Fig. 4.14. The model captures the strong axial nonuniformity of oxidation, with the largest mass loss at the bottom level and substantially smaller losses in the middle and top levels. The bottom-level Type 1 prediction is 25.3%, close to the measured values of 24.3% for SGL NBG-17 and 23.8% for UCAR PCEA, and comparable to the SPECTRA result obtained with the Roes correlation, which gives 20.1% for the same 36 1000 2.2 1000 2.3 950 950 900 900 850 850 800 800 T ( ∘ C) T ( ∘ C) 750 750 700 700 650 650 600 600 550 SAM 550 SAM Experiment Experiment 500 500 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Time (hr) Time (hr) (a) 2.2 – graphite solid temperature, Type 1. (b) 2.3 – gas outlet temperature, Type 1. 1000 21.3 1000 22.1 950 950 900 900 850 850 800 800 T ( ∘ C) T ( ∘ C) 750 750 700 700 650 650 600 600 550 SAM 550 SAM Experiment Experiment 500 500 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Time (hr) Time (hr) (c) 21.3 – graphite solid temperature, Type 2. (d) 22.1 – gas outlet temperature, Type 2. Figure 4.12: Temperature evolutions at the bottom graphite level from SAM predictions and the measured data. The left column shows the graphite solid temperature and the right column the gas outlet temperature, for the Type-1 blocks (top row) and the Type-2 blocks (bottom row). Measured data from [46]. location. The model captures the principal geometry effect at the bottom level, where the Type 1 blocks oxidize more strongly than the Type 2 blocks because their larger internal surface area provides more reaction area in the oxygen-rich inlet region. Higher in the stack, the model reproduces the reversal of this ordering, with the Type 2 blocks oxidizing more than the Type 1 blocks once the local oxygen inventory becomes limiting.x SAM overpredicts the measured weight loss for the Type 2 blocks in the middle and top levels, labeled M-2 and T-2 in Fig. 4.14. Overall, the comparison shows that the SAM graphite oxidation model captures the main behavior of the NACOK experiment. The predicted bottom level temperatures agree with the measured data, and the model reproduces the experimentally observed concentration of graphite mass loss in the oxygen-rich bottom level. The remaining differences are mainly associated with the Type 2 simplified equivalent-channel treatment. The agreement is comparable to that reported for the SPECTRA and TINTE analyses of this 37 Figure 4.13: Thermocouple layout in the NACOK experiment. The four thermocouples used in the present comparison are located in the bottom level: the Type-1 graphite solid (2.2) and gas outlet (2.3), and the Type-2 graphite solid (21.3) and gas outlet (22.1). Adapted from [46]. benchmark, supporting the use of SAM for system level simulation of HTGR air-ingress transients with graphite oxidation. Future validation work will focus on more detailed representation of the graphite solid structures, with three-dimensional heat conduction and more explicit local flow channel geometry, to improve prediction of the Type 2 block temperature response and the oxidation distribution in the upper levels. 30 Experiment (SGL NBG-17) Experiment (UCAR PCEA) 25 SPECTRA (Roes) SAM 20 Weight loss [%] 15 10 5 0 B-1 B-2 M-1 M-2 T-1 T-2 Graphite block (level--type) Figure 4.14: Graphite weight loss of each block in the NACOK experiment: comparison of the measured data (SGL NBG-17 and UCAR PCEA grades) with the SPECTRA Roes-correlation result and SAM prediction. Blocks are labeled by level (B = bottom, M = middle, T = top) and geometry type (1, 2). Experimental and SPECTRA values are from Ref. [44]. 38 Table 4.2: Graphite weight loss of each level (%) measured in the NACOK experiment and predicted by SPECTRA, TINTE, and SAM models. Experimental, SPECTRA, and TINTE values are from Ref. [44]. Weight loss [%] Locationa SPECTRA Data TINTE SAM Roes CEA NKL B-1-S 24.3 20.1 15.8 22.1 25.3 B-1-U 23.8 19.5 B-2-S 11.5 12.1 6.2 17.1 12.0 B-2-U 16.5 M-1-S 3.5 4.0 9.0 3.1 3.1 M-1-U 3.4 7.5 M-2-S 4.5 4.9 13.4 4.5 7.9 M-2-U 5.0 T-1-S 0.8 0.8 0.6 0.3 0.6 T-1-U 1.5 2.8 T-2-S 2.6 2.0 1.3 0.5 4.3 T-2-U 2.8 a Location: level (B = bottom, M = middle, T = top), geometry type (1, 2), and graphite grade (S = SGL NBG-17, U = UCAR PCEA). SPECTRA, TINTE, and SAM do not distinguish between graphite grades in the values shown; each listed model value applies to both blocks of the corresponding geometry at that level. 39 Table 4.3 provides a concise qualitative summary of the NACOK validation, combining the integral mass-loss and thermal comparisons into a single assessment of the key validation quantities. Table 4.3: Summary of the NACOK graphite oxidation validation Measured value/observation SAM prediction Assessment Total graphite mass loss ∼8.3% ∼8.8% Good Bottom-level mass loss 23.8–24.3% (T1) 25.3% (T1) Good 11.5–16.5% (T2) 12.0% (T2) Middle-level mass loss 3.4–3.5% (T1) 3.1% (T1) Fair 4.5–5.0% (T2) 7.9% (T2) Top-level mass loss 0.8–1.5% (T1) 0.6% (T1) Fair 2.6–2.8% (T2) 4.3% (T2) Axial oxidation distribution bottom ≫ middle > top bottom ≫ middle > top Good Bottom-level temperatures, T1 2.2, 2.3 rise then plateau close agreement throughout Good Bottom-level temperatures, T2 21.3, 22.1 rise under-predicts at ∼1–2.5 h, Fair monotonically then agrees at plateau T1 and T2 denote the Type-1 and Type-2 block geometries; the SAM total is the mass-weighted average of the per-block predictions. 40 5 Reactor-scale Demonstration of Air-Ingress Events The air-ingress modeling capabilities developed in SAM are demonstrated together in an integrated, reactor-scale simulation. This demonstration problem is a de-pressurized loss-of-forced-cooling (DLOFC) air-ingress transient in a simplified model of the HTR-PM reactor. The objective is not to reproduce the full plant in detail, but to demonstrate that the new SAM capabilities can capture the sequence of depressurization, diffusive air ingress, buoyancy-driven flow, oxygen transport, graphite oxidation, oxidation heat release, and passive heat removal over a long transient. This section therefore summarizes the HTR-PM reactor and the air-ingress accident scenario, describes the simplified SAM model used for the demonstration, and discusses the resulting transient behavior. 5.1 The HTR-PM Reactor The High-Temperature Reactor Pebble-bed Module (HTR-PM) is a modular pebble-bed HTGR demon- stration plant built at Shidao Bay in Shandong Province, China, and is the only commercial-scale modular HTGR currently in operation [48]. Its design based on the German pebble-bed reactor program from the 1960s to 1990s and on the Chinese HTR-10 test reactor [49] during the 2000s; the overall plant layout is shown in Fig. 5.1. The HTR-PM has a core thermal power of 250 MWth . It uses helium coolant at 7 MPa, heated from a core inlet temperature of 250 ◦ C to an outlet temperature of 750 ◦ C at a primary mass flow rate of about 96 kg/s. The cylindrical pebble-bed core is 3 m in diameter with an equivalent active height of about 11 m. The reactor core contains roughly 420,000 pebbles, each 6 cm in diameter. The pebble bed core is surrounded by radial reflector, top cavity, lower, and upper reflectors. The radial reflector includes various orifices for the control rod channels, Kleine Absorber Kugel Systeme (KLAK) channels (shutdown system), and fluid riser channels, surrounded by a steel core barrel and a steel reactor pressure vessel (RPV). During normal operation, helium enters the core through the outer channel of the coaxial duct and flows upward through the riser channels in the side reflector into the top plenum. It then passes downward through the pebble bed, and enters the bottom plenum before leaving the core through the inner channel of the coaxial duct to the steam generator. After being cooled in the steam generator, the cold helium flows upward along the steam generator vessel and returns to the blower, completing the primary circuit (Fig. 5.1) [49]. HTR-PM is designed for inherent safety: the small module power, the low power density, the large graphite heat capacity, and decay heat is removed passively by conduction and thermal radiation from the core, through the reflector, core barrel, and RPV, to a reactor cavity cooling system (RCCS). The RCCS is a passive system of water-cooling panels — vertical cooling tubes welded to a cylindrical plate — mounted on the reactor-cavity wall; during accidents, heat is transferred from the RPV surface to the panels by thermal radiation and natural convection in the cavity, and the heated water rises to air–water heat exchangers in an air-cooling tower that ultimately rejects it to the atmosphere. 5.2 Demonstration Problem This demonstration problem reproduces the depressurized air-ingress accident sequence introduced in Section 2, on the simplified HTR-PM model from a steady-state operating condition (250 MW, 7 MPa helium, 41 Not to scale 11 12 1 Reactor core 2 Side reflector & carbon thermal shield 13 14 1 3 Core barrel 3 2 4 Reactor pressure vessel (RPV) 5 Steam generator 8 4 6 Steam generator vessel 9 7 Coaxial hot-gas duct 8 Water cooling panel 9 Blower 10 Fuel discharging tube 5 11 Control rod driving system 7 12 Small absorber sphere unit 13 Fuel charging tube 6 14 Top cavity 10 Figure 5.1: Full system layout of an HTR-PM reactor module. Adapted from [50]. 250/750 ◦ C). The accident is initiated by a large break of the primary boundary at t = 100 s and is followed for 7 × 105 s (about 194 h, or ≈8 days), long enough to capture the depressurization, the slow air ingress, the graphite oxidation, and the subsequent passive cool down. Event sequence. The imposed transient proceeds through the following phases: 1. Depressurization. The primary system blows down from 7 MPa to atmospheric pressure over about 20 s as the break opens. It is assumed that the reactor scrams immediately after the break and the fission power drops onto the HTR-PM decay-heat curve (to about 6% of nominal at scram). The gas surrounding the break is set to containment air at ambient temperature (≈20 ◦ C). 2. Density-driven and diffusion-driven air ingress. After depressurization, buoyancy-driven mixing and stratification leading to gradual air enters in the lower plenum. Then air advances into the core slowly by molecular diffusion. 3. Onset of natural circulation and graphite oxidation. As air accumulates in the core, the buoyancy overcomes the loop resistance and the loop-wide natural circulation begins, which thereafter continu- ously draws fresh air and oxygen up through the hot core. The oxygen delivered to the hot pebbles is consumed by graphite oxidation, which depletes the oxygen and releases heat at the pebble surface. 4. Long-term passive cool down. As the decay heat continues to decrease, heat transfer from the RPV to the RCCS eventually exceeds the heat generation, causing the core temperature to decline and leaving the reactor in a passively cooled state. This sequence captures the principal physical phenomena of an HTGR air-ingress accident (Section 2) 42 within a single simulation. Simulating this sequence provides an integrated demonstration of the SAM capabilities described in Section 3 for modeling air-ingress events. 5.3 SAM Model For the present capability demonstration, the HTR-PM reactor is represented by a simplified SAM model, shown in Fig. 5.2. The detailed primary loop and RCCS are not included. The pebble bed core is modeled as a single equivalent channel that preserves the principal HTR-PM parameters: the 3 m core diameter, the 6 cm pebble diameter, the 0.39 bed porosity, the 7 MPa helium pressure, the 250 ◦ C inlet temperature, and the 250 MW reactor power. The corresponding flow area and hydraulic diameter follow from these values. The radial heat-conduction path of the reactor is represented explicitly by concentric cylindrical solid structures: an inner graphite reflector, a riser gap, an outer graphite reflector, a core barrel, and the RPV, which reproduces the conduction path of the real reactor. The SAM model includes several simplifications relative to the detailed reactor design. First, the core is a single equivalent one-dimensional channel, so the radial power and temperature distribution within the bed is not resolved. Second, the full primary loop is not modeled. The steam generator, blower, and the remainder of the primary circuit are omitted. The core flow path is bounded only by prescribed pressure boundary conditions applied at its inlet and outlet pipes, which impose the depressurization transient and supply the ambient air. Third, to model the buoyancy-driven mixing and stratification after depressurization, the lower plenum is connected to another volume branch, outlet branch, which connected two horizontal outlet pipes with slightly different elevations. The density difference between the outside air and the light helium drives air enters the outlet branch through the lower outlet pipe and helium leaves through the upper outlet pipe. Last, the RCCS is not modeled explicitly; instead, the RPV outer surface radiates directly to a 333 K ambient. This simplified reactor representation is built from the SAM one-dimensional and zero-dimensioanl components with multi-component gas model. The core is represented with a PBCoreChannel component with sphere pebble fuel, and KTA pebble bed heat transfer and friction closures. The riser, the outlet pipes, and the inlet piping are PBOneDFluidComponents; the plena and flow junctions are zero-dimensional PBVolumeBranch components; and the loop terminates at given pressure boundary conditions with PBTDV components. Reactor decay heat is supplied through a ReactorPower component using the HTR-PM decay-heat curve. All flow components use the multi-component gas mixture equation of state for a mixture of helium and air, with a diffusivity of 7.92 × 10−4 m2 /s [20]. Oxygen is treated as a passive scalar advected with the bulk flow and consumed locally in the core through a sink term provided by the graphite oxidation model. The graphite oxidation rate is evaluated with the Roes model of Section 3.3 as a function of the local oxygen partial pressure, gas velocity, and pebble surface temperature; it supplies both the oxygen sink term and the chemical heat release. In this demonstration the oxidation heat is deposited onto the solid pebble surface as a surface heat flux. At the high pebble temperature reached during the transient, oxidation is primarily in boundary layer controller regime, according to the oxidation theory in Section 2. So the reaction occurs near the surface of the pebble. The radial solid structures are PBCoupledHeatStructure components (the inner and outer graphite reflectors, the riser gap, core barrel, and RPV). A radiative boundary condition (CoupledRadiationHeatTransferBC) is applied on the external surface of the RPV outer wall. 43 Figure 5.2: Schematic of the simplified HTR-PM air-ingress SAM model. The transient is initialized from the steady state solution. Before the break at t = 100, a null transient is performed by applying the steady state conditions. Between t = 100 and 120 s the boundary pressures are reduced from 7 MPa to atmospheric pressure to represent the depressurization, while the boundary gas temperature is reduced from 250 ◦ C to the ambient containment value of about 20 ◦ C and the fission power is dropped onto the decay heat curve. Between t = 125 and 130 s the boundary gas composition is switched from helium to air (the air mole fraction is raised from 1e-5 to 0.99). After 130 s the boundary conditions are then held constant, so that the slow diffusive ingress, the onset of natural circulation, the graphite oxidation, and the passive cool down all develop as part of the solution rather than being imposed. 5.4 Results As described before, the air ingress event begins with the depressurization process. Figure 5.3 shows the boundary pressure during the depressurization: from the 7 MPa steady state value, the pressure decreases to atmospheric over the 100–120 s window as the break happens, with the inlet and outlet boundaries blowing down together. This depressurization causes rapid loss of the helium inventory and sets the stage for air ingress. Once the system has depressurized, ambient air begins to enter through the lower outlet pipe to the 44 Figure 5.3: Boundary pressure at the core inlet and outlet during the depressurization phase. The system blows down from the 7 MPa operating pressure to atmospheric from 100 to 120 s. outlet branch. As shown in Fig. 5.4, after the depressurization, the two outlet pipes settle to opposite flows of about ±0.12 kg/s: dense air is drawn in through the lower pipe (outlet-2) while the lighter helium-rich gas is pushed out through the upper pipe (outlet-1). The resulting gas composition is tracked in Fig. 5.5: the air mole fraction y1 rises first at the lower pipe (outlet-2) just after t = 125 s, then in the outlet branch, and last at the upper pipe (outlet-1), reflecting the filling of the bottom of the loop. Within about 30 s the lower branch is essentially filled with air (y1 ≈ 0.99), establishing the air reservoir from which the subsequent air diffusion to the core proceeds. The flow in two outlet pipes does not stop after the outlet branch is filled with air. It is instead sustained by the temperature difference between the cold ambient air drawn in through the lower pipe and the hot gas in the interior of the reactor, whose density difference drives a local gravity current flow or a local natural circulation flow [20]. With the outlet branch filled but the loop still hydraulically stalled, air migrates into the core by molecular diffusion against the residual helium. Figure 5.6 shows air mole fraction y1 at the core outlet (bottom) and core inlet (top) over this diffusion-dominated phase. The bottom of the core jumps to nearly pure air (y1 ≈ 1) as the air-filled lower plenum reaches it; it then falls back to about 0.5 over the first hour, as that air diffuses upward into the helium-rich column above, and thereafter climbs again as fresh air is continuously supplied from the lower plenum and keeps diffusing in. The top of the core remains helium-rich and air mole fraction increases only slowly, the top of core air fraction climbs gradually over roughly 15 h. Once the air has progressed far enough that the buoyancy force can overcome the flow resistance, the natural circulation onsets. There are several previous analyses of the HTR-PM air-ingress scenario using the system code TINTE [4– 7]. These studies report the onset of stable, loop-wide natural circulation over a wide range: Liu et al. [5] estimated about 100–200 h from a time-scale analysis and about 120 h in their TINTE simulation, whereas 45 Figure 5.4: Mass flow rates in the two outlet pipes during the density-driven flow after depressurization Figure 5.5: Air mole fraction y1 at the two outlets and in the lower branch during the initial ingress stage. The denser air enters the lower outlet pipe (outlet-2) first, then fills the branch, and reaches the upper outlet pipe (outlet-1) last. Xu et al. reported about 73 h [7] and, depending on the bottom-reflector graphite grade, between about 43 and 73 h [6]. The present simplified SAM model predicts onset at about 15 h. The spread among these values 46 Figure 5.6: Air mole fraction y1 at the core outlet (bottom) and core inlet (top) during the diffusion-dominated phase. reflects the strong sensitivity of the onset time. Liu et al. [5] attribute the long HTR-PM onset time to the high flow resistance of the tall pebble bed and the complicated flow structure of the hot helium plenum and bottom reflector. Because these features are simplified in the present demonstration model, the earlier onset predicted here is expected. Once the air has progressed far enough that the buoyancy force can overcome the flow resistance, the loop-wide natural circulation onsets. The computed loop mass flow during the transient is shown in Fig. 5.7. During the first stage the loop is stratified and stalled: the two outlet pipes have equal and opposite flows and the core flow rate is essentially zero, so although the outlet branch fills with air within minutes (Fig. 5.5), air advances up into the core only by slow molecular diffusion (Fig. 5.6). After about 15 h the air has advanced far enough up the core that the loop-wide natural circulation onsets sharply, with the core flow rate jumping to about −0.19 kg/s within a few minutes. Thereafter the circulation continuously supplies oxygen to the core. This 0.19 kg/s circulation rate is comparable to the 0.12–0.16 kg/s reported by Liu et al. [5], indicating that the simplified SAM model captures the magnitude of the buoyancy-driven flow once it is established. After the onset of loop-wide natural circulation at about 15 h, the oxidation heat release rises sharply, peaking at approximately 0.59 MW, which is about 39% of the roughly 1.5 MW decay heat at that time. Then the oxidation heat is declined to a level of order 0.1 MW as the local oxygen supply becomes rate-limiting. Integrated over the full transient, about 2,600 kg of graphite carbon is consumed in the oxidation process, corresponding to roughly 7% of the modeled core graphite inventory, at a net consumption rate of about 14 kg/h during the natural circulation phase. In a real depressurized air-ingress accident, the consumption could be lower. That is because the present model assumes unlimited capacity of air from boundary rather than the finite air inventory of the reactor building. Compared at the same transient time of 144 h, which is the endpoint of the Xu et al. calculation, the present model has consumed about 1,940 kg of graphite, of the same 47 Figure 5.7: Core mass flow rates during the simplified HTR-PM air-ingress transient. order as the ≈2231 kg they report [7] for a TINTE analysis of the HTR-PM. The lower graphite consumption is consistent with the lower peak temperature reached in the present model. Another reason is that the present oxidation model includes only direct oxidation by oxygen and does not yet include the Boudouard reaction, C + CO2 → 2CO (Eq. 2-3). Because the transient reaches pebble temperatures at which this reaction may become important, particularly in the hotter regions once the local oxygen has been depleted, omitting the Boudouard reaction may underpredict graphite consumption. Inclusion of the CO/CO2 chemistry and its thermal feedback is planned for future work. Figure 5.8 shows the core fuel and coolant temperature. The onset of natural circulation and graphite oxidation at about 15 h produces a distinct change in slope, after which the combined decay and oxidation heat drive a gradual core heat up. The peak fuel temperature reaches about 1659 K (1386 ◦ C) at roughly 62 h (2.6 days), after which heat removal from the vessel overtakes the heat sources and the core cools down. The present value is slightly lower than the prior HTR-PM TINTE analyses, which predicted roughly 1460 ◦ C in Ref. [5]. This demonstration shows that SAM can simulate the full HTGR air-ingress sequence in a single integrated calculation. The simulation captures the expected progression from depressurization, density- driven flow at lower plenum, diffusion-dominated air ingress, transition into buoyancy-driven core natural 48 Figure 5.8: Core fuel and coolant temperature during the HTR-PM air-ingress demonstration transient. circulation, followed by oxygen transport to and consumption at the hot graphite, exothermic oxidation heat release and its thermal feedback, and eventual passive cooldown. The results reproduce the key qualitative behavior of air-ingress accidents, including an initial heatup followed by long-term cooldown over multiple days. Because the model includes several simplifying assumptions, the transient timeline may differ from that of more detailed HTR-PM analyses. This section should therefore be viewed as a capability demonstration rather than a validation study. 49 6 Conclusions This report describes the development, validation, and demonstration of new SAM capabilities for modeling air-ingress phenomena in HTGRs, a recognized multi-physics challenge for which the community has lacked a validated, integrated system-level tool. To represent the helium–air gas behavior during air ingress, a multi-component gas mixture model with diffusion was implemented in SAM. A 0-D graphite oxidation model based on the Roes correlation was developed, including oxygen consumption and oxidation heat release. A critical flow model was also implemented to improve the representation of primary system depressurization following a break. These capabilities were validated against the NSTF heavy-gas ingress experiment and the NACOK air-ingress facility. The NSTF heavy-gas ingress experiment exercised the SAM multi-component flow model in a buoyancy-driven cooling loop whose inlet composition transitions from air to a denser gas, argon; SAM captured the composition-dependent density changes and the resulting system-level flow response. The NACOK experiment provided the validation of the SAM graphite oxidation models against a measured total graphite mass loss. SAM reproduced the axial nonuniformity and the block geometry dependence of the oxidation, predicting a bottom-level weight loss of about 25% against the measured ∼24% and matching the measured bottom level temperatures. Overall, SAM reproduced the principal experimental trends and the key transient behaviors relevant to air-ingress analysis. The SAM graphite oxidation prediction is comparable to those from the SPECTRA (Roes correlation) and TINTE analyses of the same benchmark, supporting the use of SAM for system-level HTGR air-ingress simulations with coupled graphite oxidation. The developed models were then applied in an integrated, reactor-scale demonstration of a depressurized loss-of-forced-cooling (DLOFC) air-ingress transient in a simplified HTR-PM pebble-bed reactor. In a single simulation spanning approximately 8 days, SAM reproduced the expected accident progression: rapid depressurization from 7 MPa to atmospheric in 20 s, density-driven flow in the lower regions, diffusion- dominated air ingress into a stalled stratified loop over approximately 15 hours, onset of buoyancy-driven natural circulation, exothermic graphite oxidation reaching a peak fuel temperature at roughly 62 hours, and eventual passive cooldown as vessel heat rejection overtakes the decay and oxidation heat sources. These results demonstrate that the newly implemented models in SAM can reproduce the complete sequence of HTGR air-ingress events in a single integrated calculation. Future work will focus on several extensions and improvements to the capabilities developed in this work. The current gas mixture model is limited to two components and will be extended to additional components, together with further improvements in numerical robustness and continued verification and validation. The graphite oxidation model currently represents only direct oxidation by oxygen and does not yet include the C + CO2 → 2CO, reaction or its thermal feedback. Finally, the present HTR-PM analysis is intended as a capability demonstration rather than a detailed reactor analysis; future applications will incorporate a multi-channel core representation, more realistic break and flow-path geometries, and detailed modeling of the RCCS. Together, these improvements will progressively close the remaining gaps between the current capability demonstration and a full-fidelity, licensing-quality air-ingress analysis of the HTGRs. The present work represents a significant step toward that objective, establishing SAM as a validated, modern system-level tool for HTGR air-ingress safety analysis in support of the NEAMS program’s mission. 50 Acknowledgments This work was performed at Argonne National Laboratory, supported by the U.S. Department of Energy, Office of Nuclear Energy, Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, under contract DE-AC02-06CH11357. 51 References [1] F. Gou, F. Chen, and Y. Dong, “Preliminary phenomena identification and ranking tables on the subject of the High Temperature Gas-cooled Reactor-Pebble Bed Module thermal fluids and accident analysis,” Nuclear Engineering and Design, vol. 332, pp. 11–21, 2018. [2] H. S. Lim and H. C. No, “GAMMA multidimensional multicomponent mixture analysis to predict air ingress phenomena in an HTGR,” Nuclear science and engineering, vol. 152, no. 1, pp. 87–97, 2006. [3] Y. Zheng and M. M. Stempniewicz, “Investigation of NACOK air ingress experiment using different system analysis codes,” Nuclear engineering and design, vol. 251, pp. 423–432, 2012. [4] W. Xu, Y. Zheng, L. Shi, and P. Liu, “Oxidation Analyses of Massive Air Ingress Accident of HTR-PM,” Science and Technology of Nuclear Installations, vol. 2016, no. 1, p. 6419124, 2016. [5] P. Liu, Z. Chen, Y. Zheng, J. Sun, F. Chen, L. Shi, F. Li, Y. Dong, and Z. Zhang, “Study on air ingress of the 200 MWe pebble-bed modular high temperature gas-cooled reactor,” Annals of Nuclear Energy, vol. 98, pp. 120–131, 2016. [6] W. Xu, J. Sun, Y. Zheng, and L. Shi, “The influence of nuclear graphite oxidation on air ingress accident of HTR-PM,” Annals of Nuclear Energy, vol. 110, pp. 1242–1248, 2017. [7] W. Xu, H. Zhang, H. Xie, L. Shi, and J. Sun, “Numerical Study of the Air Ingress Accident of the HTR-PM and Possible Mitigation Measures,” Science and Technology of Nuclear Installations, vol. 2020, no. 1, p. 8872899, 2020. [8] C. H. Oh, H. S. Kang, and E. S. Kim, “Air-ingress analysis: Part 2—Computational fluid dynamic models,” Nuclear Engineering and Design, vol. 241, no. 1, pp. 213–225, 2011. [9] Y. Ferng and C. Chi, “CFD investigating the air ingress accident for a HTGR simulation of graphite corrosion oxidation,” Nuclear engineering and design, vol. 248, pp. 55–65, 2012. [10] R. Hu, L. Zou, D. O’Grady, T. Mui, Z. J. Ooi, G. Hu, E. Cervi, G. Yang, D. Andrs, A. Lindsay, C. Permann, R. Salko, Q. Zhou, L. Fick, A. Heald, and H. Zhao, “SAM: A Modern System Code for Advanced Non-LWR Safety Analysis,” Nuclear Technology, vol. 0, no. 0, pp. 1–20, 2024. [11] P. Vegendla, R. Hu, and L. Zou, “Multi-scale modeling of thermal-fluid phenomena related to loss of forced circulation transient in htgrs,” Tech. Rep. ANL-19/35, Argonne National Laboratory (ANL), Argonne, IL (United States), Sep 2019. [12] Z. J. Ooi, L. Zou, T. Hua, J. Fang, and R. Hu, “Modeling of a generic pebble bed high-temperature gas-cooled reactor (pb-htgr) with sam,” tech. rep., Argonne National Laboratory (ANL), Argonne, IL (United States), Sep 2022. [13] Z. J. Ooi, L. Zou, T. Hua, J. Fang, and R. Hu, “Multi-physics system-level simulations of a generic pebble bed high-temperature gas cooled reactor with coupled sam/griffin model,” Tech. Rep. ANL/NSE–23/45, Argonne National Laboratory (ANL), Argonne, IL (United States), July 2023. 52 [14] Z. J. Ooi, G. Yang, T. Mui, L. Zou, and R. Hu, “Coupled sam/griffin model of a reference pebble bed high-temperature gas cooled reactor for multi-physics simulations,” Tech. Rep. ANL/NSE–23/43 Rev.1, Argonne National Laboratory (ANL), Argonne, IL (United States), Oct 2023. [15] B. Hollrah, L. Zou, and C. Filippone, “Preliminary Thermal-hydraulics and Selected Safety Analysis for Holos-Quad Reactor Design,” Tech. Rep. ANL/NSE-22/1, Argonne National Laboratory, Mar 2022. [16] B. Hollrah, Y. Jeong, L. Zou, and R. Hu, “Primary system thermal fluids analysis model development for the compact high temperature gas reactor (hc-htgr) – final report,” tech. rep., Argonne National Laboratory (ANL), November 2024. [17] T. Hua, L. Zou, and R. Hu, “Code Benchmark of the HTTF Pressurized Conduction Cooldown Test Using SAM,” Nuclear Science and Engineering, vol. 197, no. 10, pp. 2660–2672, 2023. [18] A. Epiney, G. Strydom, R. Kile, J. Barthle, I. Gutowska, B. Nakhnikian-Weintraub, T. Hua, L. Zou, J. Fang, K. Podila, et al., “SMR safety through HTTF modeling and benchmark efforts for code validation for gas-cooled reactor applications,” Nuclear Engineering and Design, vol. 448, p. 114737, 2026. [19] X. Zhang, F. Sun, G. Xiong, W. Xu, and M. Ding, “A review of research progress of graphite oxidation in high temperature gas-cooled reactors,” Nuclear Engineering and Design, vol. 428, p. 113486, 2024. [20] C. H. Oh and E. S. Kim, “Air-ingress analysis: Part 1. Theoretical approach,” Nuclear Engineering and Design, vol. 241, no. 1, pp. 203–212, 2011. [21] H. Hinssen, W. Katcher, and R. Moormann, “Kinetics der Graphite,” Sauerstoff Reaction Juel-1875, 1983. [22] M. Hishida and T. Takeda, “Study on air ingress during an early stage of a primary-pipe rupture accident of a high-temperature gas-cooled reactor,” Nuclear Engineering and Design, vol. 126, no. 2, pp. 175–187, 1991. [23] W. Froehling, H. Hohn, M. Kugeler, Y. Sun, and Z. Zhang, “The NACOK experiment on natural convection of air through the core,” tech. rep., Forschungszentrum Juelich GmbH, 1995. [24] D. Gould, D. Franken, H. Bindra, and M. Kawaji, “Transition from molecular diffusion to natural circulation mode air-ingress in high temperature helium loop,” Annals of Nuclear Energy, vol. 107, pp. 103–109, 2017. [25] E. S. Kim, H. C. No, B. J. Kim, and C. H. Oh, “Estimation of graphite density and mechanical strength variation of VHTR during air-ingress accident,” Nuclear Engineering and Design, vol. 238, no. 4, pp. 837–847, 2008. [26] C. Davis, L. Siefken, and C. Oh, “Assessment of RELAP5-3D for Analysis of Very High Temperature Gas-Cooled Reactors,” tech. rep., Idaho National Lab.(INL), Idaho Falls, ID (United States), 2005. [27] P. Liu, Y. Zheng, W. Xu, and L. Shi, “Nonuniform oxidation on the surface of fuel element in HTR,” Science and Technology of Nuclear Installations, vol. 2016, no. 1, p. 7485602, 2016. [28] D. Franken, D. Gould, P. K. Jain, and H. Bindra, “Numerical study of air ingress transition to natural circulation in a high temperature helium loop,” Annals of Nuclear Energy, vol. 111, pp. 371–378, 2018. 53 [29] X. L. Yan, T. Takeda, T. Nishihara, K. Ohashi, K. Kunitomi, and N. Tsuji, “A study of air ingress and its prevention in htgr,” Nuclear Technology, vol. 163, no. 3, pp. 401–415, 2008. [30] B. E. Poling, J. M. Prausnitz, O. John Paul, and R. C. Reid, The properties of gases and liquids, vol. 5. Mcgraw-hill New York, 2001. [31] R. Hu, L. Zou, T. Mui, D. O’Grady, E. Cervi, Z. J. Ooi, G. Yang, T. Fei, G. Hu, G. Zhang, et al., “Sam theory manual,” tech. rep., Argonne National Laboratory (ANL), 2025. [32] J. Welty, G. L. Rorrer, and D. G. Foster, Fundamentals of momentum, heat, and mass transfer. John Wiley & Sons, 2014. [33] P. M. Gerhart, A. L. Gerhart, and J. I. Hochstein, Munson, Young and Okiishi’s fundamentals of fluid mechanics. John Wiley & Sons, 2016. [34] T. Mui, Y. Mao, T. Hua, L. Zou, R. Hu, and Q. Zhou, “SAM Code Development for Source Term Modeling in Fluoride-salt-cooled High-temperature Reactors,” Tech. Rep. ANL/NSE-23/78, Argonne National Laboratory (ANL), Argonne, IL (United States), 2023. [35] J. Roes, “Experimentelle Untersuchungen zur Grephitkorrosion und Aerosol-entstehung beim Lufteinbruch in des core eines Kugelheufenm Hochtempernturreektors,” tech. rep., Forschungszentrum Jülich GmbH, 1994. [36] Z. J. Ooi, G. Yang, L. Zou, T. Mui, T. Hua, and R. Hu, “SAM Enhancements for Air-Ingress Event Modeling in HTGRs,” Tech. Rep. ANL/NSE-25/46, Argonne National Laboratory, 2025. [37] D. D. Lisowski, C. D. Gerardi, D. J. Kilsdonk, N. C. Bremer, S. W. Lomperski, R. Hu, A. R. Kraus, M. D. Bucknor, Q. Lv, T. Lee, et al., “Final project report on rccs testing with the air-based natural convection shutdown heat removal test facility (nstf),” tech. rep., Argonne National Laboratory (ANL), 08 2016. [38] “Preliminary Safety Information Document for the Standard MHTGR,” Tech. Rep. HTGR-86-024, Vol. 1, Amendment 13, U.S. Department of Energy, 1992. [39] R. Hu, A. Kraus, M. Bucknor, Q. Lv, and D. Lisowski, “Final project report on computational modeling and analysis of air-based nstf,” tech. rep., Argonne National Lab. (ANL), Argonne, IL (United States), 06 2016. [40] B. Hollrah, M. Bucknor, D. Lisowski, Y. Hassan, R. Vaghetto, and R. Hu, “Benchmark simulation of the natural convection shutdown heat removal test facility using sam,” Nuclear Technology, vol. 206, no. 9, pp. 1337–1350, 2020. [41] R. Hu, D. D. Lisowski, M. D. Bucknor, A. R. Kraus, and Q. Lv, “Ambient and building condition effects modeling of air-cooled natural circulation systems,” Applied Thermal Engineering, vol. 137, pp. 23–31, 2018. [42] M. B. Kuhlmann, Experimente zu Gastransport und Graphitkorrosion bei Lufteinbruchstörfällen im Hochtemperaturreaktor. PhD thesis, Institut für Sicherheitsforschung und Reaktortechnik, 2002. [43] T. Schaaf, W. Fröhling, H. Hohn, and S. Struth, “The NACOK experimental facility for investigating an air ingress into the core of a high temperature reactor,” Kerntechnik, vol. 63, no. 3, pp. 107–112, 1998. 54 [44] Y. Zheng and M. M. Stempniewicz, “Investigation of nacok air ingress experiment using different system analysis codes,” Nuclear Engineering and Design, vol. 251, pp. 423–432, 2012. [45] C. B. Davis, L. J. Siefken, and C. H. Oh, “Assessment of RELAP5-3D for analysis of very high temperature gas-cooled reactors,” in 11th International Topical Meeting on Nuclear Reactor Thermal-Hydraulics (NURETH-11), (Avignon, France), 2005. [46] J. Lapins, An Advanced Three-Dimensional Simulation System for Safety Analysis of Gas Cooled Reactors. PhD thesis, Universität Stuttgart, 2016. [47] H. Nissen et al., “NACOK experiment: experimental set-up, execution of the test, experimental results,” tech. rep., 2008. [48] Z. Zhang, Z. Wu, D. Wang, Y. Xu, Y. Sun, F. Li, and Y. Dong, “Current status and technical description of Chinese 2 × 250 MWth HTR-PM demonstration plant,” Nuclear Engineering and Design, vol. 239, no. 7, pp. 1212–1219, 2009. [49] Z. J. Ooi, L. Zou, R. Hu, M. K. Jaradat, D. Reger, V. C. Leite, J. Ortensi, and C. Takasugi, “Gas-Cooled High-Temperature Pebble-Bed Reactor Reference Plant Model Updates,” Tech. Rep. ANL/NSE-25/29; INL/RPT-25-85834, Argonne National Laboratory and Idaho National Laboratory, 2025. [50] Y. Zheng, M. M. Stempniewicz, Z. Chen, and L. Shi, “Study on the dlofc and plofc accidents of the 200 mwe pebble-bed modular high temperature gas-cooled reactor with tinte and spectra codes,” Annals of Nuclear Energy, vol. 120, pp. 763–777, 2018. 55 www.anl.gov ARGONNE NATIONAL LABORATORY □ U.S. Department of Energy research facility □ Operated by the UChicago Argonne, LLC □ Midwest’s largest federally funded R&D facility □ Located in Lemont, IL, about 25 miles (40 km) □ southwest of Chicago, IL (USA) □ Conducts basic and applied research □ in dozens of fields □ Unique suite of leading-edge and rare scientific □ user facilities CONTACT Argonne National Laboratory 9700 South Cass Avenue Lemont, IL 60439 630-252-2000