**3. Nuclear data process**

The underlying nuclear data is the basis of core calculation and also the fundamental elements in the proposed software structure. The reactor physics data is divided into four categories by A. Trkov, namely, (1) basic nuclear data, (2) evaluated nuclear data files, (3) processed nuclear data, and (4) averaged nuclear data [17]. They are the similar data in the thermal-hydraulics analysis, which correspond to the physical property information of the fluid, such as the light water property data in the core calculation [18]. These different types of data involve representation, storage, and computer algorithms. For example, the cross-sectional library of deterministic reactor physics calculation needs group-wise data, such as few-group

homogenized cross section in diffusion calculation, with the affection of multiple state variables that include temperature, density, and so on [19]. The class of used nuclear data can be directly computed in the core calculation or be generated in advance. The pre-generated nuclear data (database) is commonly used in a two-step method of core calculation, one is multidimensional interpolation table, the other is parameterized library, and they have different effects on computation and parallel algorithm [20].

### **3.1 Multidimensional interpolation table**

The table is the indirect format of cross-sectional parameters, with the grids inside the value range for each state variable. The core calculation uses interpolation algorithm to obtain the nuclear data.

Assume there is only one state variable *ρ* (such as coolant density), and its value range is ½ � *a*, *b* , then the grid of a single state variable is divided as in Eq. (1):

$$\{\rho\_i\}\_{i=1}^{N+1}, \rho\_i \in [a, b] \tag{1}$$

where *ρ<sup>i</sup>* is not necessarily uniform distribution. The corresponding parameters at each discrete grid can be obtained by the lattice program as follows in Eq. (2):

$$\left\{\sum\_{i}\right\}\_{i=1}^{N+1}, \sum\_{i} \coloneqq \sum(\rho\_i) \tag{2}$$

In this way, the continuous parameter values can be calculated by polynomial interpolation method in the range of state variable. The coefficients of interpolation polynomials are stored in the table:

$$\sum^{\tilde{\nu}}(\rho) = P(\rho) = \sum\_{i=0}^{N+1} a\_i \rho^{i-1} \tag{3}$$

#### **3.2 Parameterized library**

The library method is the complex function model of nuclear data that is related to various state variables. To build an accurate model, many factors need to be considered, and the range of each state variable is also defined separately:

$$
\sigma = f(\rho, T, P, \dots) \tag{4}
$$

Generally, the tabulation method uses space for time, and the method is relatively simple, but the storage mode of nuclear data has a greater impact on the parallel algorithm of the core calculation. The method consumes more computing time, but the data storage is smaller, and it is easy to improve and modify the model so that there is less impact on the parallel algorithm.

The software implementation of the nuclear data corresponds to the model layer and model framework layer mentioned in Section 2. **Figure 1** shows two abstract forms of the nuclear data. The left side that represents the processing of each kind of nuclear data is independent, and the right side explains the design of consistent data attributes, and formats are independent underneath and dependent on top in the fixed software structure so that each type of nuclear data only needs format conversion and minor programming. This understanding of the underlying nuclear database focuses on software reuse and performance, which can fully reuse various

*Parallel Algorithm Analysis in Reactor Core Calculation DOI: http://dx.doi.org/10.5772/intechopen.92759*

**Figure 1.**

*The association between underlying nuclear data and software structure.*

known nuclear data that have fine readability and correctness. This association also enables independent design and programming of core calculation at all levels, which makes the following parallel algorithm analysis match it.
