Data collection

We first collected the available irradiation-embrittlement data for Chinese domestic and international RPV steels (e.g., the US, France, Germany, Japan, Korea, etc.). The data included the neutron fluence (unit: n/cm², E > 1 MeV), transition-temperature shift, ΔRT_NDT, and chemical compositions of base and weld metals. There are 53 and 144 groups of valid Chinese and international RPV steel embrittlement data, respectively.

The main chemical compositions (Cu, Ni, and P) of the collected international RPV materials are summarized in Fig. 1. The Cu contents of these international RPV steels are mainly in the range of 0.02–0.20 wt.%. Most RPV materials have a relatively high Ni content (0.05–0.09 wt.%) and low P content (0.005–0.015 wt.%). In addition to the chemical composition, the neutron fluence directly affects the transition-temperature shift. These embrittlement data have similar fast neutron (E> 1 MeV) fluence ranges, which are 1.8×10¹⁷–6.4×10¹⁹ n/cm² for the collected Chinese RPV data and 3.2×10¹⁸–7.4×10¹⁹ n/cm² for the international RPV surveillance data.

Fig. 1Full-screen View

(Color online) Ranges of the Cu, Ni, and P content of the collected international RPV materials. The Y-axis represents the amount of data corresponding to a certain range of Cu/Ni/P content listed in the X-axis

Data screening

To develop a prediction model suitable for Chinese domestic RPV steels and those with similar chemical compositions, we filtered the collected international RPV embrittlement data, using the chemical compositions (mainly Cu, Ni, P, and Mn) of Chinese domestic RPV steels as the criteria. Consequently, 285 groups of international embrittlement data for RPV base metals were selected as the database to fit our prediction model. The RPV material types in these data were mainly 16MND5, SA508-3, 20MnMoNi55 forging, and SA533-B1 plate. Considering the design standards of Chinese RPV materials, the data for the 20MnMoNi55 and SA533-B1 steels were excluded.

Finally, 126 groups of embrittlement data for international RPV base metals were retained and divided into training (99) and test (27) sets. Similarly, we screened out 46 groups of data for international RPV weld metals as the training set and 44 groups as the test set. The training dataset was used for model fitting and the test dataset was used to assess the fit quality of the developed model.

In addition to international RPV embrittlement data, we screened out 69 and 53 groups of ΔRT_NDT data for Chinese domestic RPV base and weld metals, respectively. The average contents of the main chemical compositions (Cu, Ni, Mn, P, Si, and C) of the screened international and Chinese RPV steels are listed in Tables 1 and 2. It can be seen that the chemical compositions of the screened international RPV steels are very similar to those of Chinese RPV data. The only difference is that Chinese RPV steels have a much lower Cu content. Nevertheless, it is reliable to employ international RPV embrittlement data to fit the new prediction model for Chinese domestic RPV steels.

Data selection

The testing method and definition of neutron fluence and ΔRT_NDT differ for different nuclear power plants. The neutron fluence used as a parameter in the ETC is usually the average of the measured neutron-fluence values. Some Chinese domestic power plants amend the neutron fluences of different base-metal or weld-metal samples, which have different locations in the irradiation surveillance capsule. The amended neutron fluence values are then used to calculate the predictive value of ΔRT_NDT. Generally, the base- and weld-metal samples are located on the positive and negative sides of the irradiation surveillance capsule, respectively. The amended neutron-fluence values are typically no greater than 20%.

To evaluate the effects of neutron fluence values on the formula of the developed prediction model, we modified the neutron fluence. In other words, the neutron fluence of the base metal was multiplied by a coefficient of 1.2 and the neutron fluence of the weld metal was multiplied by a coefficient of 0.8. The ΔRT_NDT values were refitted using the modified neutron fluences as input parameters. The difference in the calculated ΔRT_NDT values was less than 0.23 °C when using the modified neutron fluences as variables in the model. Therefore, considering the measurement uncertainty, the amendment of the different measured neutron fluences can be disregarded in the fitting process of the prediction model.

The ΔRT_NDT values are obtained from Charpy impact experiments, which correspond to the temperature change at certain Charpy impact energies or lateral expansions. The determination of the ΔRT_NDT values is different for different standard regulations. In the Chinese RPV surveillance program, one way to determine ΔRT_NDT is to choose the larger value between the temperature shifts corresponding to the impact energy of 56 J (ΔT_56J) and the lateral expansion of 0.89 mm (ΔT_0.89mm). Another method is to consider the transition-temperature shift at the Charpy impact energy of 41 J (ΔT_41J) as ΔRT_NDT, following the ASTM E185 standard regulation.

Some foreign nuclear power plants have also adopted larger TTS values between ΔT_68J and ΔT_0.9mm as ΔRT_NDT. Based on the experimental data, we collected 121 groups of irradiation surveillance data from Chinese pressurized water reactors (PWRs) and calculated the corresponding TTS values, i.e., ΔT_41J, ΔT_56J, and ΔT_{0.89 mm}, based on the experimental data. According to the results, the ΔT_56J and ΔT_41J values have a linear relationship, as follows: $\text{Δ}{T}_{56\text{J}}=0.98\times \text{Δ}{T}_{41\text{J}}.$ (1)

The residual between ΔT_41J and ΔT_56J shows a normal distribution between -7 °C and 9 °C and the average value lies in the range of -0.6–6 °C. This indicates that the ΔT_41J and ΔT_56J values can be treated as the same.

We also examined the average difference between ΔT_41J and ΔT_56J (ΔT_41J−ΔT_56J) as a function of the neutron fluence, which varies by only 1 °C, and the correlation coefficient between (ΔT_41J−ΔT_56J) and the neutron fluence is 0.046, which is far smaller than 1. This suggests that ΔT_41J and ΔT_56J have almost the same relationship with the neutron fluence.

Similarly, we fitted the relationship between ΔT_41J and ΔT_{0.89 mm}, which is also linearly dependent as shown in the following equation: $T_{0.89\text{mm}}=0.904\times T_{41\text{J}}+\mathrm{7.305.}$ (2)

The mean value of ΔT_{0.89 mm} is slightly higher than that of ΔT_41J. If ΔRT_NDT is calculated as the maximum value between ΔT_41J and ΔT_0.89mm, it is only 2.93 °C higher than the ΔT_41J value. According to the above discussion, there were no significant differences in the three values of ΔT_41J, ΔT_56J, and ΔT_0.89mm. Therefore, we employed the measured ΔT_41J as ΔRT_NDT to fit the embrittlement prediction model.

Model development

The transition-temperature shifts were demonstrated to be related to features such as the neutron fluence, neutron fluence rate, temperature, and RPV chemical composition. These parameters contribute differently to the transition-temperature shifts.

To obtain an appropriate weight for each parameter in the embrittlement prediction model, we first calculated the correlation coefficients of these parameters using the measured ΔRT_NDT data. The correlation coefficient is calculated based on the covariance and the standard deviation between the variables, which is defined as ${\rho }_{x,y}=\frac{\text{cov}\left(x,y\right)}{{\sigma }_x{\sigma }_y}$, where cov(x, y) is the covariance between variable x (e.g., neutron fluence) and variable y (e.g., ΔRT_NDT). σ_x and σ_y are the calculated standard deviations of variables x and y, respectively.

The calculated correlation coefficients between the different parameters (neutron fluence, neutron fluence rate, temperature, and chemical composition) and the measured ΔRT_NDT are shown in Fig. 2. It can be observed that the fast neutron fluence has the largest correlation coefficient with the measured ΔRT_NDT, which is greater than 0.5 for both the RPV base and weld metals. This suggests that the fast neutron fluence will have the greatest effect on the transition-temperature shift compared to the other parameters.

Fig. 2Full-screen View

(Color online) Correlation coefficients between different parameters with the measured ΔRT_NDT data. The blue and magenta bars represent the cases of RPV base and weld metals, respectively

In addition to the fast neutron fluence, the Cu content is a critical factor for ΔRT_NDT, with a correlation coefficient larger than 0.3. The other chemical compositions, such as Mn, Si, and P, had a weaker correlation with ΔRT_NDT. The neutron fluence rate and temperature also have little influence on the transition-temperature shift according to the analysis of the correlation coefficients. Furthermore, these parameters are not independent of each other. The correlation coefficients between the fast neutron fluence and Cu content were 0.332 and 0.413 for the base and weld metals, respectively, but were still smaller than 0.5. This indicates that overfitting should be avoided when these correlated parameters are incorporated into a single fitting equation.

Because the fast neutron fluence was found to be the only parameter that had a strong correlation (correlation coefficient > 0.5) with ΔRT_NDT, we first fitted the measured ΔRT_NDT data using the fast neutron fluence as the main variable, and then included the influence of different chemical compositions of RPV steels.

3.1.1.

Fluence dependence

According to current regulatory models and the characteristics of our collected surveillance embrittlement data, the ΔRT_NDT values usually show a power-law relationship between the fast neutron fluence (f) and energies larger than 1 MeV. Thus, we employed four power functions to fit the ΔRT_NDT data to the neutron fluence; they are defined as fluence-dependent models. The equations for these four functions are as follows: $\text{Δ}R{T}_{\text{NDT}}=a\times f^n$ (3) $\text{Δ}R{T}_{\text{NDT}}=a\times f+b$ (4) $\text{Δ}R{T}_{\text{NDT}}=a\times f^{\left(b+c\times \text{log}f\right)},$ (5) $\text{Δ}R{T}_{\text{NDT}}=a\times {\text{log}}_{b}f+c.$ (6)

Coefficients a, b, and c and power exponent n are all fitted parameters.

Figure 3 shows the predicted trends of ΔRT_NDT varying with the fast neutron fluence using the above four equations. The corresponding measured ΔRT_NDT data for Chinese and international RPV steels are shown as scatters in Fig. 3 for comparison. Overall, the predicted embrittlement-trend curves (ETCs) cover the measured ΔRT_NDT scatters, indicating the reasonable fitting accuracy of these functions. It can be found that the ETCs given by Eq. (6) form the lower bounds for both the base and weld metals, which means that the ΔRT_NDT may be underestimated.

Fig. 3Full-screen View

(Color online) Embrittlement trend curves predicted by four different functions for (a) the base metal and (b) the weld metal. The measured ΔRT_NDT data for Chinese and international RPV steels are denoted by blue circles and red triangles, respectively

The model of Eq. (4) yields the highest overall predicted ΔRT_NDT values, indicating an overestimation tendency. The ETCs predicted using Eqs. (3, 4) and (5) are similar and fall into the intermediate region of the four ETCs. These two functions performed better in fitting the measured ΔRT_NDT values than Eqs. (4, 5) and (6) did.

We calculated the standard deviation (S_d) obtained from the training ΔRT_NDT datasets using the four functions, which were calculated by $\sqrt{s^2}=\sqrt{\frac{{\displaystyle {\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}{n-1}}$, where $\overline{x}$ is the mean value of ΔRT_NDT and n is the total amount of data. The calculated S_d values are listed in Table 3. Only a small difference appeared between the S_d values of these four functions, which was within 0.5 °C. Notwithstanding, Eqs. (3, 4) and (5) yielded the same S_d values for both the base and weld metals, which were also the lowest among the four models (S_d = 10.73 °C for the base metal and S_d = 11.08 °C for the weld metal).

Table 3Full-screen View

Standard deviations (S_d) obtained by the four established fluence-dependent functions

Models	Sd (℃)
	Base metal	Weld metal
$a\times f^n$(Eq.3)	10.76	11.13
$a\times f+b$(Eq.4)	10.79	11.34
$a\times f^{\left(b+c\times \text{log}f\right)}$(Eq.5)	10.73	11.13
$a\times {\text{log}}_{b}f+c$(Eq.6)	11.22	11.69

Show more

In the above discussion, the four models were separately fitted for the RPV base and weld metals. In other words, the fitting coefficients of one function are different for the base and weld metals. Therefore, two independent functional relationships were obtained between the RPV base and weld metals. Current embrittlement prediction models, such as the US. RG1.99 (Rev. 2) and French FIS/FIM, adopt the powerful form of Eq. (3) to describe the relationship between the transition-temperature shift and neutron fluence. The fitting results also suggest that Eqs. (3, 4) and (5) yield the lowest standard deviation. In addition, Eq. (3) is simpler than Eq. (5). Therefore, considering the convenience of employing the model to predict ETC in a surveillance program, we chose Eq. (3) to fit the collected ΔRT_NDT data and derived the final fluence-dependent embrittlement prediction models for RPV base and weld metals, i.e., $\{\begin{array}{c}\text{Δ}R{T}_{\text{NDT}}=15.311\times {\left(f\times {10}^{-19}\right)}^{0.525}\text{for base metal} \\ \text{Δ}R{T}_{\text{NDT}}=16.637\times {\left(f\times {10}^{-19}\right)}^{0.512}\text{for weld metal}\end{array}$ (7)

3.1.2.

Chemical-composition dependence

On the basis of the established fluence-dependent embrittlement-prediction model (Eq. 7), the influences of RPV chemical compositions were further included in the model. To obtain appropriate weighting coefficients for each chemical composition, the correlation coefficients between the chemical composition and residual of ΔRT_NDT calculated using Eq. (7) were analyzed, as shown in Fig. 4(a). The chemical compositions that have a relatively larger relevance to the ΔRT_NDT residual are mainly Cu, Ni, the combined CuNi term, P, Cr, and C. This indicates that these elements may have greater impacts on the fitting results of ΔRT_NDT, based on the fluence-dependent model.

Fig. 4Full-screen View

(Color online) (a) Correlation coefficients between different chemical compositions and the predicted ΔRT_NDT residual. (b) and (c) show the correlation coefficients between different chemical compositions for the RPV base and weld metals, respectively

The Cr content has the largest correlation coefficient with the ΔRT_NDT residual, which is -0.234 for the base metal. The Cu content remains an important factor in the transition-temperature shift of the RPV base metal, and the correlation coefficients between Cu and CuNi and the ΔRT_NDT residual are approximately 0.171 and 0.160, respectively. For the weld metal, the Ni content has the largest correlation coefficient with the ΔRT_NDT residual (–0.275), while the P and C content also have small effects on the ΔRT_NDT to a certain extent.

Moreover, the relevance of different chemical compositions was analyzed to avoid overfitting. For example, as shown by the correlation matrices in Figs. 5(b) and (c), the chemical compositions strongly interrelate with each other. In particular, the correlation coefficients between P and Cu, P and CuNi, and Mn and Mo are all greater than 0.5. This implies that if one chemical composition is implanted into the model as a parameter, other elements with large correlation coefficients (> 0.5) with this element should be avoided.

Fig. 5Full-screen View

(Color online) Distribution of scatterplots of the ΔRT_NDT predicted by the fluence-and-chemical-dependent model vs.the measured ΔRT_NDT for (a) the RPV base metal and (b) the RPV weld metal

We note that the calculated correlation coefficients between different chemical compositions and the ΔRT_NDT residual were all lower than 0.3. This means that the RPV chemical compositions may not have a significant effect on the fitting ΔRT_NDT values of the fluence-dependent model. On the other hand, most of the collected datasets contained information on Cu, P, Ni, and Mn, but lacked other elements. Therefore, we chose Cu, P, Ni, Mn, and C as the key parameters to conduct the new ΔRT_NDT fitting procedure by incorporating both the neutron fluence and different chemical compositions (Cu, P, Ni, Mn, C, and CuNi²) into one function. Moreover, we constructed different forms of equations with different combinations of RPV chemical compositions, based on the fluence-dependent model.

We found that the changes in the predicted ΔRT_NDT values were all less than 1 °C using these equations, regardless of the combination forms of the chemical compositions. Among these functions, Eq. (8) yielded the lowest S_d values for the predicted ΔRT_NDT, which were 10.23 °C for the base metal and 10.48 °C for the weld metal. $\{\begin{array}{c}\text{Δ}R{T}_{\text{NDT}}=\left(a+b\times \text{Cu}+c\times \text{P}+d\times \text{Ni}+e\times \text{Mn}\right)\times f^n\text{ for base metal}\\ \text{Δ}R{T}_{\text{NDT}}=\left(a+b\times \text{Ni}+c\times \text{P}+d\times \text{C}+e\times \text{Cu}+g\times \text{Mn}\right)\times f^n\text{ for weld metal}\end{array}$ (8)

It should be noted that the models considering the effects of both the neutron fluence and RPV chemical compositions on the transition-temperature shift have lower standard deviations than the fluence-dependent models (Eq. 7); however, the differences are less than ~0.8 °C. The results suggest that the fluence-dependent model may be sufficiently accurate to predict the ETC of RPV steels.

By contrast, the fluence-and-chemical-dependent model uses RPV chemical compositions as the main parameters; thus, the predicted ΔRT_NDT will be disturbed by the RPV material types. Therefore, the fluence-and-chemical-dependent model is limited to certain types of RPV steels that have chemical compositions similar to those in our collected datasets.

Model assessment

According to the above discussions, we have developed two models, i.e., the fluence-dependent models and the fluence-and-chemical-dependent models (Eq. 7 and Eq. 8, respectively). These models have the same functional forms, but different fitting coefficients for the RPV base and weld metals.

Taking the fluence-dependent models (Eq. 7) as an example, the fitting coefficients are a= 15.311 and n= 0.525 for the base metal and a= 16.637 and n= 0.512 for the weld metal. It can be seen that the fitting coefficients for the RPV base and weld metals are very similar to each other, especially the power exponent n of the neutron fluence. Thus, we attempted to use a unified equation to fit the measured ΔRT_NDT data for RPV steels without discriminating base and weld metals. The developed unified fluence-and-chemical-dependent prediction model is given as follows: $\{\begin{array}{c}\text{Δ}R{T}_{\text{NDT}}=A\times B\times \left(0.029+1.236\text{Cu}+16.932\text{P}-0.555\text{Ni}+0.230\text{Mn}\right)\times {\left(f\times {10}^{-19}\right)}^{0.504}\\ A=1.067\left(\text{base metal}\right), A=1.010\left(\text{weld metal}\right)\\ B=-0.855\left(T-292℃\right)+29.238\end{array}$ (9)

where Cu, P, Ni, and Mn are the mass fractions of the elements. Coefficient A is 1.067 for the base metal and 1.010 for the weld metal, which are almost the same.Coefficient B reflects the effect of the temperature on the transition-temperature shift, where T represents the inlet temperature of the RPV.

The fluence-dependent model can also have a uniform form for both RPV base and weld metals, which is $\{\begin{array}{c}\text{Δ}R{T}_{\text{NDT}}=A\times 15.919\times {\left(f\times {10}^{-19}\right)}^{0.521}\\ A=0.967\left(\text{base metal}\right), A=1.037\left(\text{weld metal}\right)\#\end{array}$ (10) where coefficient A is 0.967 for the base metal and 1.037 for the weld metal. The power exponent of the neutron fluence is 0.521 for both the base and weld metals. The S_d values of the predicted ΔRT_NDT by the unified fluence-dependent model (Eq. 10) are 10.73 °C for the base metal and 10.48 °C for the weld metal. The S_d values given by the unified fluence-and-chemical-dependent model (Eq. 9) are 10.23 °C for the base metal and 11.08 °C for the weld metal, which are very close to those derived by Eq. 10.

The advantage of the above two models is that they have an explicit scope of application, namely, they can be applied to predict the transition-temperature shift of RPV steels irradiated in neutron-fluence ranges of 1.0×10¹⁷–1.11×10²⁰ n/cm² and temperature ranges of 280–296 °C. The fluence-and-chemical-dependent model can be applicable to RPV steels with the main chemical compositions (Cu, P, Ni, and Mn) listed in Table 4.

Next, we assessed the fitting quality of these two models by plotting the measured ΔRT_NDT as a function of the predicted ΔRT_NDT, as shown in Figs. 5 and 6 for the fluence-and-chemical-dependent and fluence-dependent models, respectively. The measured ΔRT_NDT values were derived from the screened training dataset, which included Chinese and international RPV embrittlement data. The predicted ΔRT_NDT values were calculated using Eqs. (9) and (10), respectively. The differences between the predicted and measured Chinese ΔRT_NDT are indicated by the triangles and multiple signs in Figs. 5 and 6, respectively. The differences between the predicted and measured international ΔRT_NDT are indicated by the diamonds and circles in Figs. 5 and 6, respectively.

Fig. 6Full-screen View

(Color online) Distribution of scatterplots of the ΔRT_NDT predicted by the fluence-dependent model vs. the measured ΔRT_NDT for (a) the RPV base metal and (b) the RPV weld metal

It can be seen that most of the data points fall in the vicinity of the 45° line, which indicates that the predicted ΔRT_NDT equals the measured ΔRT_NDT. In other words, the predicted ΔRT_NDT showed a good overall consistency with the measured ΔRT_NDT data. Moreover, the majority of data were distributed within the one-fold standard-deviation boundary, the corresponding confidence of which was 68.3%, and almost all the data are within the 95.4% confidence interval.

The deviation-band analyses suggest that our newly developed model has a high accuracy in predicting the irradiation-embrittlement trend for Chinese and similar international RPV steels. Furthermore, the deviation bands of the fluence-and-chemical-dependent and fluence-dependent models are very similar to each other, demonstrating again that the RPV chemical composition has little influence on predicting the transition-temperature shifts.

To evaluate the performance of our developed model compared to current embrittlement-prediction models, such as RG1.99 Rev. 2 and RSE-M2000 and 2010, we employed three foreign models to calculate the transition-temperature shifts based on our collected ΔRT_NDT dataset. The standard deviations obtained using our new models and the other three empirical models are listed in Table 5.

The new fluence-and-chemical-dependent and fluence-dependent models produced the lowest S_d values for both the RPV base and weld metals. This suggests that the newly developed models have higher prediction accuracy than the current US and French prediction models. More importantly, the two new models were suitable for evaluating the embrittlement trends of Chinese domestic RPV steels. We believe that the newly developed fluence-dependent model may have more general application because it avoids the influence of different RPV chemical compositions on the transition-temperature shifts.

Model application

In this section, we employed the established fluence-dependent model to predict the embrittlement trend and the mechanical-property degradation for a Chinese 300-MW nuclear power plant. The embrittlement trend curve estimated by the fluence-dependent model is shown by the red line in Fig. 7; the surveillance data are indicated by black squares for comparison. The available surveillance data fall within the one-fold standard deviation boundary of the calculated ETC. In particular, in the low-flux neutron fluence range (< 1.0×10¹⁹ n/cm²), the estimated ETC is consistent with the measured ΔRT_NDT data. Therefore, the newly developed fluence-dependent model can be used to predict the irradiation-embrittlement trend for Chinese RPV steels with reasonable reliability.

Fig. 7Full-screen View

Embrittlement-trend curve (red line) predicted by the new fluence-dependent model for the Chinese 300-MW nuclear power plant. The black squares represent the surveillance ΔRT_NDT data

In a practical surveillance program, the irradiation-induced degradation of the mechanical properties of RPVs during operation is usually evaluated using two temperature indicators: the adjusted reference temperature (ART) and the pressure–thermal–shock reference temperature (RT_PTS). The ART is defined as the sum of the initial RT_NDT before irradiation, the irradiation-induced transition-temperature shift ΔRT_NDT, and the standard deviations, which are calculated as $2\sqrt{S_{\text{d}0}^2+S_{\text{d}\text{Δ}}^2}$. Here, $S_{\text{d}0}$ and $S_{\text{d}\text{Δ}}$ represent the standard deviations of the measured initial RT_NDT and the shifted ΔRT_NDT, respectively.

The ART values increase with increasing neutron fluence, indicating that the RPV is at risk of a brittle fracture when its service temperature exceeds the ART. Therefore, according to standard regulation NRC-RG1.99 (Rev. 2), the ART at the one-quarter wall thickness of the RPV cylinder should not be higher than 93 °C to avoid a sudden failure. On the other hand, the pressure–thermal–shock temperature will also cause the RPV to fracture; thus, the RT_PTS should not exceed 132 °C to avoid a sudden RPV failure.

Assuming that the lifetime of RPVs in Chinese nuclear power plants could be extended to 60 years, we calculated the ART at the one-quarter wall thickness of the RPV cylinder and the RT_PTS using the newly developed fluence-dependent model to evaluate the embrittlement level of RPVs. The calculation results are listed in Table 6. The calculated ARTand RT_PTS of RPV steels in the Chinese 300-MW nuclear power plant are 53 °C and 62 °C, respectively.

In the standard regulation NRC-RG1.99 (Rev. 2), the service temperature limits of RPVs at the end of 60 years of operation are ART = 93 °C and RT_PTS = 132 °C. We found that the predicted ART and RT_PTS values using the new fluence-dependent model were 40 °C and 70 °C lower than the temperature limits, respectively. This means that there are still service temperature margins of 40 °C (ART) and 70 °C (RT_PTS) for the safe operation of RPVs prior to a brittle fracture failure. Based on the prediction results, the Chinese 300-MW nuclear power plant has no tendency to fracture within 60 years of operation. Therefore, the design lifetime of the Chinese 300-MW reactor can be extended from 30 to 60 years.