More than half of the operating nuclear power plants in Japan are the light water cooled reactor built in the 1970s and 1980s. After more than three decades of operation, the ageing components in these nuclear power plants starts to appear as a problem and the ageing issue is becoming one of the most important issues in the nuclear industry in Japan. The proper ageing management or maintenance is demanded to ensure the safety during operation and to avoid the unplanned outage of the plant. Ageing issues of nuclear power plants is not only appearing in Japan but also in other nuclear developed countries, such as the United States. The huge and complicated system of the nuclear power plant determines that the ageing issue has to be watched for many components. Among the components the piping consists of a large portion of the plant system and the incidents caused by piping degradation are frequently reported. Two accidents, i.e. the Surry-2 accident in the U. S. (in year 1986) and the Mihama-3 accident in Japan (in year 2004), were resulted from the rupture of degraded piping and caused fatal injuries.The observed piping degradation is regarded to be initiated or accelerated by the dynamic processes in the flow carried by the piping. The two processes studied in this thesis, liquid droplet impingement (LDI) and flow-accelerated corrosion (FAC), are among these processes.

In the nuclear power plant liquid droplet impingement (LDI) occurs in the piping carrying the high-quality two-phase flow where the steam entrains massive small droplets. When the flow is experiencing the abrupt change of velocity, the droplets do not exactly follow the change in the steam. Hence, the droplets can impact on the piping wall. The high-velocity droplet impact exerts significant mechanical loads on the piping inner surface. When the velocity is sufficiently high, damage can be resulted from a single impact. However, this is unusual in the piping system of nuclear power plants where the LDI damage is believed to be produced by massive impingement instead. Not all the impacts can produce damage. It is argued that there is virtually a threshold impact velocity only above which LDI damage is possible. The flow conditions that allow for the occurrence of LDI can usually be found in the piping from the low-pressure turbine to the condenser and the drain piping of the pre-heaters. The LDI damage is usually observed as local pitting or pinholes at the elbow, orifice or valve.

The evaluation of the mechanical load caused by LDI is an important step to understand the LDI damage mechanism and to predict the damage rate. A variation of the Moving Particle Semi-implicit (MPS) method, the MPS-AS method, is applied to simulate the single droplet impingement and to investigate the mechanical load induced by the droplet impingement. The simulation of the high-velocity impingement has been challenging because it involves both the simulation of free surface deformation and the shock generation and propagation. The MPS-AS method which inherits the capability of simulating the large deformation of free surface and holds the capability to capture the shock wave is utilized in this investigation. In the past, the evaluation of impingement pressure load has been based on the correlations. The different correlation may attempt to feature the different stage during the impingement. Beside the impingement phenomena, the numerical simulation also contributes to evaluate the major existing correlations for the impingement pressure load. The current correlations are only valid for simple impingement case. For even more complicated impingement scenario, the numerical simulation becomes the basis for the development of the new correlation. Compared with the prediction by the correlations, another advantage of the numerical simulation is that it enables the detailed study of the pressure load history which may also important to understand the damage process. The final interest of the investigation of LDI is the prediction of the damage rate caused by LDI. This investigation attempts to correlate the damage rate with the pressure load.

In the flow-accelerated corrosion, the corrosion or metal oxidation occurs on the interface between the base metal and the oxide layer. The oxidants have to diffuse through the oxide layer before they can oxide the metal. Thus the oxide layer normally protects the metal from oxidation by separating it from the oxidants. When the mass transfer between the corroded surface and the bulk flow is enhanced by the flowing stream, the concentration of metallic ion species near the surface may be reduced below its saturation value and the dissolution of the oxide layer starts. The oxide layer becomes thinner and less protective, and the corrosion rate increases. Eventually a steady state is reached where the corrosion and dissolution rates are equal and stable corrosion rates are maintained. FAC can occur in the piping carrying the single- (water) and two-phase (water/steam) flows.

In the hydraulic analysis for the flow-accelerated corrosion the turbulent mass transfer is the most important issue to be looked after. Currently, the mass transfer in the analysis of flow-accelerated corrosion is considered either with the empirical correlation or the models developed based on the empirical correlation. These models are always questionable on its physical foundation. This investigation attempts to provide the reliable and low-cost turbulence model which can be applied in the three-dimensional computation for the large-scale systems. The mass transfer in the flow-accelerated corrosion is characterized by the high Schmidt number due to which the apparent concentration variation is expected to appear in the vicinity of the wall. And the turbulence modeling at the wall has been a challenging task for a long time, especially when we need to handle the complex flows. With the aid of the DNS data, the development of the low Reynolds number models prosperous blossomed in the 1990s. Low Reynolds number models have been regarded to be advantageous in the turbulence modeling at the wall. There have been hundreds of variations of low Reynolds number model which are not possible to be tested for all. The models are selected based on the reported performance in the separated and reattaching flows, which is the important flow type leading to the mass transfer enhancement in the piping system of nuclear power plants. The low Reynolds number models demands the fine resolution near the wall, which can be a barrier preventing it from applying in the large-scale systems. Hence, the high Reynolds number models and wall functions are also surveyed and evaluated in this investigation. The evaluation is conducted in several flow geometries, the fully developed pipe flow, the backward-facing step flow and the flow through an orifice. The latter two are the typical flows with separation and reattachment. Especially, the flow through an orifice is the geometry leading to the Mihama-3 accident.

Above is the main content included in this thesis.

本研究は原子力発電プラントにおける液滴衝撃エロージョン(LDI, Liquid Droplet Impimgement)および流れ加速型腐食(FAC, Flow-Accerelated Corrosion)の流体解析に関するもので、5章より構成されている。

第1章は緒言であり研究の背景が述べられている。日本における原子力発電プラントは1970年代および1980年代に建設されたものが多く、高経年化対策が重要な課題である。適切な高経年化対策がプラントの安全性確保と計画外停止の回避のために求められている。中でも、配管減肉はしばしば報告されており、1986年の米国のサリー原子力発電所2号炉の事故および2004年の美浜原子力発電所3号炉の事故において配管破断が発生している。配管減肉のメカニズムとしてはLDIとFACが重要であり、本研究ではこれらの2種類を対象にするとしている。

第2章ではLDIの流体解析がまとめられている。LDIは曲り管やオリフィスの下流において気液二相流中の液滴が配管内面に高速で衝突することによって生じる。そこで、単一液滴が高速に壁面の衝突するシミュレーションを粒子法(MPS-AS (Moving Particle Semi-implicit for All Speed)法)によって実施した。まずシミュレーションの検証と妥当性確認のため、単一液滴が垂直に壁に衝突する解析をおこない、空間解像度に対する収束を確認するとともに、衝突速度に対する衝撃圧力の変化が従来の相関式と一致することを確認した。また、3次元計算と2次計算を比較し、衝撃圧力の最高値は次元の影響がほとんどなく、最高値の後の減少速度は3次元の方が速いことが示された。本研究では空間解像度を高めた2次元計算を主におこない、衝撃圧力の最高値を議論するものであり、次元の影響は現れないとしている。次に、液膜の影響、衝突角度の影響およびナトリウム液滴の場合についてシミュレーションを用いて評価した。液膜によって衝撃圧力が緩和されること、衝突角度の垂直成分で衝撃圧力が決まること、ナトリウムの場合でもほぼ水と同じ結果が得られることがわかった。

第3章ではLDIに関する様々なシミュレーション結果にもとづいて、衝撃圧力に対する新たな相関式を提案している。具体的にはHeymannの相関式を拡張して、衝撃圧力から潜伏期間と減肉速度を算出できるものとしている。

第4章はFACの流体解析がまとめられている。FACは配管内面の酸化皮膜からのイオンの溶出が乱流によって促進されることによって減肉が加速される現象である。そこで、低レイノルズ数型RANS(Reynolds-Averaged Navier-Stokes)モデルおよび高レイノルズ数型RANSモデルを、複数用いて流体解析をおこなった。計算体系として、円管流れ、バックステップ流れ、およびオリフィス流れの3種類を用いた。いずれも空間2次元である。計算は有限体積法にもとづくOpenFOAMコードに、独自に乱流モデルを組み込んで実施している。円管流れでは、速度分布や壁面摩擦係数はいずれの乱流モデルでもDNS(Direct Navier-Stokes)の結果とよい一致を示したが、物質伝達係数では大きな違いが現れた。これは、シュミット数が1,000と非常に大きい値であるためで、低レイノルズ数型RANSモデルでは壁面のごく近傍のわずかな乱流の違いによって結果が大きく影響されることがわかった。一方、高レイノルズ数型RANSモデルでは壁関数境界条件の中に高シュミット数の効果が適切に考慮されていれば妥当な結果が得られている。バックステップ流れおよびオリフィス流れでははく離が生じるため、これを精度良く計算することが重要であり、SSTモデルがよい結果を与えるとの知見が得られた。FACははく離が生じる場所に現れることが多いので、FACの流体解析には高レイノルズ数型RANSモデルとしてSSTモデルを用い、壁関数境界条件と組み合わせることが適切であるとの結論を得た。壁関数境界条件では壁近傍に細かいメッシュを切る必要がなく、実用性の観点からも優れている。

以上を要するに、本研究では原子力プラントの高経年化対策として重要な配管減肉の予測のため、LDIとFACに関する流体解析をおこない、LDIにおける単一液滴の衝撃に関して様々なパラメータの影響を解明し、その結果を相関式にまとめ、さらにはFACにおける乱流解析を様々に実施して高シュミット数流れによって生じるFAC特有の現象を明らかにするとともに、これを精度良く解析できる乱流モデルを提案している。これらの成果は配管減肉のメカニズムの解明と減肉速度の予測に対して大きく貢献するものである。よって本論文は博士(工学)の学位請求論文として合格と認められる。