You may see this message when you are first starting to build optimization models that include uncertainty – consider it part of the “learning experience.” You’ll be defining constraints or an objective, computed by formulas that depend on uncertain parameters: Each such formula represents an array of sample values, one for each realization of the uncertainties. For your model to be well-defined, the objective or constraint must either be summarized to a single value (such as a mean or percentile value) or transformed into a set of single-valued constraints (through an automatic transformation in the Solver Model dialog).
To correct the problem, you can (i) use the Task Pane Model tab to define the constraint as a chance constraint or the objective as an expected value or risk measure objective, or (ii) edit the formula so that its ‘top level’ value is computed by a PSI Statistics function such as PsiMean() or PsiPercentile(). It’s important to understand why you need to summarize or transform a formula that depends on uncertainties: To learn more, read the chapter “Mastering Stochastic Optimization Concept” in the Frontline Solvers User Guide.