Excel Solver Not Converging? Fix the Model

Errors & fixes · Updated November 18, 2026

If Excel Solver is not converging, first check that the selected solving method matches your model, then verify the starting values, formulas, and constraints. A model can also stop because it is infeasible, poorly scaled, or difficult for Solver to search; the status message and a few targeted tests usually reveal which problem you have.

What “not converging” means

Solver searches for values in selected worksheet cells that optimize a target cell while meeting the constraints you specify. Convergence means the search has reached a point where its method’s stopping conditions are satisfied. It does not automatically mean that the result is the best possible answer, or even that the model represents the real-world problem correctly.

People use “not converging” to describe several different outcomes: Solver may run for a long time, stop without a solution, report that it cannot find a feasible solution, or return a result that is technically allowed but looks wrong. These cases need different responses. Read the Solver Results message rather than treating every unsuccessful run as the same failure.

A useful distinction is between a feasible solution and an optimal one. A feasible solution satisfies all stated constraints. An optimal solution is the best feasible value Solver found for the chosen objective. For many nonlinear models, Solver may find a local optimum: a result that cannot be improved by nearby changes, even though a better solution exists elsewhere.

Check the Solver method first

In Excel for Microsoft 365 on Windows, open the Data tab and select Solver. In the Solver Parameters dialog, review the method in the Select a Solving Method box. The three methods suit different kinds of models:

  • Simplex LP is for linear models: the objective and constraints are built from linear relationships among the changing cells. A product such as quantity multiplied by a fixed unit profit is linear; multiplying two changing cells together is not.
  • GRG Nonlinear is for smooth nonlinear models, where small changes in the variables generally produce gradual changes in the formulas. It is often a sensible choice for models involving smooth curves or nonlinear rates.
  • Evolutionary is designed for models with discontinuities, jumps, or relationships that are not smooth enough for GRG. It can be useful with logical choices and some complex formulas, but may take longer and does not guarantee the global best result.

Choosing Simplex LP for a model containing nonlinear formulas can make Solver’s assumptions inappropriate. Conversely, using Evolutionary for a straightforward linear model may waste time. If your objective and every constraint are linear, select Simplex LP and, in the Solver Options, consider selecting Assume Linear Model when that option is available. Do not claim a model is linear merely because its formulas look simple: check for changing-cell multiplication, variable exponents, ratios, or nonlinear functions.

Verify the objective, variables, and formulas

Before changing advanced options, inspect the basic setup. The objective must be a single cell containing a formula that depends, directly or indirectly, on the changing cells. The changing cells must be the actual decision variables, and each constraint must compare the intended calculated result with its limit. A blank objective, a hard-coded value, or an omitted decision cell can leave Solver with nothing useful to optimize.

Check the worksheet for formula errors, text where numbers are expected, accidental references to the wrong row, and formulas that do not update when a changing cell changes. A quick test is to alter one decision cell manually and see whether the objective and relevant constraint calculations respond as expected. Then undo the manual edit before running Solver.

Also confirm that your constraints express the intended rules. For example, a resource limit of 100 units usually means the calculated resource use must be less than or equal to 100, not greater than or equal to 100. Review whether variables should be nonnegative, bounded, integer, or binary. Solver does not infer these requirements from labels such as “number of trucks” or “production quantity”; add the appropriate restrictions explicitly.

Test whether the constraints are feasible

A model is infeasible when no combination of changing-cell values can satisfy all its constraints at once. This is different from Solver merely needing more time. Conflicting minimums and maximums, an incorrect inequality, or a missing allowance for unavoidable demand can make the feasible region empty.

For a practical check, temporarily remove or relax constraints in a copy of the model, then add them back in small groups. If Solver can find a solution without one particular constraint but not with it, inspect that constraint and the formulas feeding it. You can also calculate the smallest and largest possible values of key quantities separately. If a requirement demands at least 80 units but the model’s maximum possible output is 70, no solving method can make the original model feasible.

Do not permanently loosen a real business requirement just to get a successful status. The test is for diagnosis. If the requirements genuinely conflict, the model needs a decision about which rule may change, or a revised representation of the problem—not a different Solver button.

Improve starting values and scaling

GRG Nonlinear is sensitive to where its search begins. If the variables start far from a useful region, Solver may follow an unproductive path or converge to a poor local solution. Give changing cells plausible starting values that satisfy simple bounds and, if possible, come close to meeting the main constraints. Run Solver again from several reasonable starting points and compare results. Different outcomes can be evidence that the model has multiple local optima.

Scale can matter too. A model that combines values around 0.00001 with values around 10,000,000 may be numerically difficult. Where appropriate, express quantities in more consistent units—for example, thousands of dollars instead of dollars—or use Solver’s Use Automatic Scaling option. Keep the underlying meaning and constraints consistent when changing units, and check the final result in the original units.

In Solver Options, changing the convergence setting can affect when a nonlinear search stops. A stricter setting may encourage further work, but can increase run time and will not repair incorrect formulas, conflicting constraints, or a badly chosen method. Similarly, raising time or iteration limits may help a complex search finish, but it is not proof that the model is sound. Make one change at a time so you can tell whether it helped.

Handle integers, binary choices, and discontinuities

Integer and binary restrictions make a model harder because Solver cannot freely adjust those variables through every fractional value. Use integer constraints for quantities that must be whole numbers and binary constraints for yes-or-no decisions. Avoid adding these restrictions to cells that do not need them: a model with fewer discrete decisions is generally easier to search.

Some worksheet formulas create abrupt changes. For example, a cost that switches from one rate to another at a threshold is not smooth at the switch. A formula using IF, lookup-based price bands, or a rounded decision may therefore be unsuitable for a smooth GRG search. Depending on the model, Evolutionary may be more appropriate; a genuinely linear model with integer restrictions may instead be handled with Simplex LP and the integer constraints.

Do not use rounding in a formula simply to make a fractional answer look like a whole number. Rounding the displayed result is different from constraining the decision variable to be an integer. If the real requirement is a whole number, represent that requirement with Solver’s integer constraint and verify that the returned values satisfy it.

A troubleshooting sequence that saves time

Work through a small, controlled sequence rather than changing many Solver settings at once:

  1. Save a copy of the workbook and note the current Solver settings and status message.
  2. Check that the target cell is a formula, the changing cells are correct, and all relevant formulas recalculate from those cells.
  3. Test the constraints individually for sign, reference, units, bounds, and possible conflicts.
  4. Select Simplex LP for a linear model, GRG Nonlinear for a smooth nonlinear model, or Evolutionary when discontinuities make the other methods unsuitable.
  5. Set plausible starting values and use consistent units. For GRG, try more than one reasonable starting point if local optima are possible.
  6. Run Solver and read the full result message. If it stops early, adjust an appropriate time or iteration limit, or a convergence option, only after checking the model itself.
  7. Recalculate and independently verify the reported solution: check each constraint, the objective value, and any integer or binary requirements.

Keep a record of which change produces a different outcome. A result that changes dramatically when starting values change deserves closer scrutiny, especially in a nonlinear model. A result that violates an apparent business rule may indicate a missing constraint even if Solver reports success.

Solver status messages and how to interpret them

A message that Solver found a solution and the constraints and optimality conditions are satisfied is encouraging, but it does not validate the model’s assumptions or prove a global optimum for a nonlinear problem. Check whether the result makes sense and whether the method’s assumptions fit the formulas.

A message indicating that Solver could not find a feasible solution points toward conflicting constraints, impossible bounds, or a setup error. A message that Solver converged to the current solution can mean it has reached a stopping point near its current search path; it is not the same as proof that no better solution exists anywhere. If the run stops at a limit, increasing that limit may help only when the model is otherwise correctly configured.

For important work, retain the Solver result and supporting assumptions, and test the solution against the original requirements. Solver optimizes the model you give it—not an unstated goal or real-world rule that was left out of the worksheet.

Excel version and platform notes

These steps refer to the Solver add-in in desktop Excel for Microsoft 365 on Windows. If Solver is missing from the Data tab, enable it through File > Options > Add-ins. At the bottom, set Manage to Excel Add-ins, select Go, check Solver Add-in, and select OK.

In Excel for Mac, Solver is also an add-in, but the route to manage add-ins differs: use Tools > Excel Add-ins and select Solver if it is available. Excel for the web does not provide the same desktop Solver add-in workflow; open the workbook in the desktop application to use that add-in. Menu wording can vary slightly by release, so consult the options shown in your installed version.

Frequently Asked Questions

Why does Excel Solver keep running without finishing?
It may be exploring a difficult nonlinear or integer search, or the model may be poorly scaled or incorrectly configured. Check the method, starting values, formulas, and constraints first; raise a time or iteration limit only if the setup is sound and more search is warranted.

Does Solver converging mean I found the best answer?
Not necessarily. A converged nonlinear search may have found a local optimum rather than the global best. Try multiple reasonable starting points, review the model’s structure, and verify that the result meets every real requirement.

Which Solver method should I use?
Use Simplex LP for a linear model, GRG Nonlinear for smooth nonlinear relationships, and Evolutionary when the model has discontinuities or other features that make smooth search unsuitable. If unsure, inspect the formulas and test a simplified version of the model.

Why does Solver say there is no feasible solution?
At least one constraint may conflict with another, a bound may rule out every possible answer, or a formula or inequality may be set up incorrectly. Test constraints in groups and calculate whether the required limits are achievable before relaxing anything permanently.

Can changing the starting values fix Solver?
Yes, especially with GRG Nonlinear, because the starting point can affect the search path and the local solution it reaches. Try plausible alternative values, but treat widely different results as a reason to investigate multiple optima or model problems.

Related

Try it: the Formula Explainer: paste any formula and see each function and reference in plain English.

More in Errors & fixes

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.