I believe this will do what you want. Give it a try. It scans along work getting the cumulative sum from that index onwards. Then it finds the first index where that cumulative sum exceeds 2. When that happens, it records the sum for that chunk (typically in the range 2-3) and the original index where that chunk ends.
% Create sample data.
Work = rand(2400,1);
% Define the limit of sums that the chunks need to surpass.
sumLimit = 2;
% Preallocate more memory than needed for W
sumValues = zeros(size(Work));
indexLocations = zeros(size(Work));
chunk = 1; % Index for what chunk we're working on.
index = 1;
while index < length(Work) && chunk < length(Work)
% Find the sums from this index onwards to the end of the Work vector.
cdf = cumsum(Work(index:end));
% Find the first time the sum exceeds the limit.
limitLocation = find(cdf > sumLimit, 1, 'first');
% Remember, limitLocation is relative to Index, not to element 1.
% If no remaining sum is more than sumLimit, just break out of the loop.
if isempty(limitLocation)
break;
end
% Save this sum value.
sumValues(chunk) = cdf(limitLocation);
% Save this index value. The index value relative to index 1, not relative to limitLocation.
indexLocations(chunk) = limitLocation + index - 1;
% Move pointer to the next element.
index = indexLocations(chunk) + 1;
% Increment what chunk number we're working on.
chunk = chunk + 1;
end
% Trim trailing zeros from sumValues and indexLocations.
sumValues(sumValues == 0) = []
indexLocations(indexLocations == 0) = []
