Result Averaging in Abaqus

Left: Scalars computed before averaging. Right: Scalars computed after averaging in Abaqus

Optimizing Simulation Results: Which Averaging Method to Choose 

When running simulations in Abaqus, interpreting results accurately is crucial for understanding material behavior and performance under load. This post explains the difference between two Abaqus/Viewer options: "Compute scalars before averaging" and "Compute scalars after averaging," and how each choice affects your results. 

Extrapolation: How Results Are Transferred to Nodal Values 

Abaqus calculates element variables (stress, strain, etc.) at integration points, which are located at the centroids of elements. To visualize these results in contour plots, the data must be extrapolated to the element nodes. The extrapolation method used depends on the element type. For second-order, reduced-integration elements, Abaqus applies linear extrapolation to compute the nodal values. 

After extrapolation, common nodes shared by multiple elements receive values from all contributing elements. For example, nodes between two elements will average values from both elements, as illustrated in the following diagram. 

Scalar Computation and Averaging: The Next Step in Processing Results 

Scalar Computation Scalar computation converts tensor results (e.g., stress components like S11, S22, etc.) into a single scalar value, such as the von Mises stress.

This conversion simplifies the data, allowing for easier visualization and interpretation. 

Averaging Averaging takes multiple values at a node (from different elements) and computes a single averaged result. This step smooths the data and is critical for visualizing results in the context of a continuous material or geometry. 

The Key Difference: Compute Scalars Before vs. After Averaging 

Abaqus offers two options for handling scalars and averaging: 

• Compute Scalars Before Averaging (default): Scalars are computed from the extrapolated tensors first, and then the averaged scalar value is determined. This method is useful when you want precise scalar values before averaging them across neighboring nodes. 

• Compute Scalars After Averaging: Here, tensors are averaged first, and then the scalar value is calculated from the averaged tensor. This method is more suitable when you want to smooth the results prior to scalar computation. 

Impact of Your Choice on Results 

The order of operations in scalar computation and averaging affects how the simulation results are presented, especially in regions with sharp gradients, such as near boundaries or in high-stress areas. While both options yield similar results in well-behaved meshes, the choice becomes significant when working with complex geometries or simulations that require precise gradient visualization. 

Averaging Threshold: Fine-Tuning the Smoothing of Results 

The "Averaging Threshold" setting controls how much influence neighboring elements have on each other’s results. This option is only available when "Compute scalars before averaging" is selected. The threshold determines how much discrepancy between node values will prevent averaging: 

• A threshold of 0% disables averaging entirely. 

• A threshold of 100% averages all results. 

A higher threshold will create smoother results, but may obscure sharp gradients or discontinuities in the data, which might be crucial in some analyses. 

Conclusion: When to Use Which Option 

Choosing between "Compute scalars before averaging" and "after averaging" depends on the specific needs of your simulation. For scenarios requiring high accuracy in localized areas or where sharp gradients are important, "Compute scalars after averaging" might be the better option. For more even distributions or standard cases, "Compute scalars before averaging" offers a simpler, more direct approach. 

Need Support With Result Interpretation

Understanding how to properly process and visualize simulation results in Abaqus is key to making informed engineering decisions. Whether you’re unsure about the best averaging method or want to fine-tune post-processing settings for greater accuracy, our team can help. Use our contact form or email us at sales@4realsim.com to discuss how we can support your simulation workflow.