Optimization under uncertainty for integrated models

A method, comprising:

• inputting realizations of a model of a reservoir that comprises at least one well into a computing system comprising a processor, wherein the realizations represent an uncertainty in a multidimensional space;
• wherein the processor is configured to reduce run-time overhead by:
• selecting a portion of the realizations in a reduced dimensional space to maintain a same amount of the uncertainty,
• wherein the portion of realizations is less than a number of the realizations, and
• wherein the same amount of the uncertainty is maintained by using multidimensional scaling of the realizations to reduce dimensionality of the realizations,
• generating a parameter value by performing a preliminary optimization of an objective function that is based on the portion of the realizations,
• running, based on the parameter value, the model to generate a field operations plan;
• and
• adjusting operation of one or more pieces of wellbore equipment based on the field operations plan generated from the model to maximize production of hydrocarbons from one or more reservoirs.