In both the pre-interpolation are post-interpolation stages of the data transfer process, data reconstruction algorithms are used to create variable data on specific mesh locations (for example, nodes), given that data for this variable already exists at different mesh locations (for example, face centroids). These reconstruction algorithms are used when a source participant provides values at mesh locations different from those required by the mapper or when the mapper provides values at mesh locations different from those required by the target participant.
Conservative algorithms are used to reconstruct extensive variable data by one of the following methods:
- Creating Nodal Data from Face/Element Centroid Data
Extensive variable values (such as forces or heat rates) may be available on element (faces or cells) centroids. If the values are required on nodes, the following steps are executed:
For a given element, the contribution to the node value is calculated by dividing the value at the element by the number of nodes that define that element.
These individual element contributions are accumulated at the node for each element adjacent to this node.
Intensive variable values (such as temperatures or incremental displacement) may be available on element (face or cell) centroids. If the values are required on nodes, they are calculated by taking an area- or volume-weighted average from all connected elements.
- Creating Face/Element Data from Nodal Data
Extensive variable values (such as forces or heat rates) may be available on nodes. If the values are required on elements/faces, the following steps are executed:
For a given node, the contribution to the element value is calculated by dividing the value at the node by the number of elements adjacent to that node.
These individual nodal contributions are accumulated at each element for each node that defines that element.
Intensive variable values (such as temperatures or incremental displacements) may be available on nodes. If the values are required on elements/faces, they are calculated by taking a simple average of the nodal values on each element.