I think I understand.
It's often good to group associated, but mixed type, data items into clusters (e.g. file records), particularly to form a single array of these clusters/records. It facilitates a more compact and better documented diagram/coding.
The downside to using arrays of clusters is that you inevitably need to both index the array and then unbundle the cluster to access individual elements. Sometimes this might seem a rather onerous task and may appear to create more block diagram functions than neccessary - but it's often still the best way, as it's very important to create logical easy to read code.
I've attached an example of the sort of data structure I think you've described - and showed two alternative ways of summing all the second cluster elements (and the two alternative unbundle options). The upper method looks nicer and may run a tiny bit faster, but the lower method is more memory efficient as it does not create a second array before summing - the difference is insignificant unless you are dealing with very large arrays.
Hope this helps.
If this isn't how your code looks, you will have to attach your code example so we can see it.