The summary tab uses a normalized scaling called "Zero-Centered Diffs", where each respondent is put on the same scale (described further below). The individual level utilities shown in the second tab are on a raw scale that is used in the market simulator, where each respondent can have different scaling.
To convert raw utilities to zero-centered diffs, first make sure each respondent's utilities are zero-centered (average zero) within each attribute (they should already be that way, unless you have run the secondary step of calibrating the ACA utilities to the Calibration Concepts). If they aren't zero-centered, (for each respondent) subtract the mean utility within each attribute from the utility of each level of each attribute.
"Diffs" is a scaling choice such that (for each respondent) the average difference between the best level of each attribute and the worst level is equal to 100. So, (for each respondent) you need to sum the differences between best and worst levels for each attribute and then find the multiplier such that the average difference across all attributes is 100. Multiply all the zero-centered raw utilities by that multiple (a different multiple for each respondent) and now you've converted to "zero-centered diffs."