Model infeasibility

Infeasibility is attained when a solution cannot be found which meets the constraints of a specific problem. These constraints are often related to elements such as flow requirements, inventory requirements, or end to end service constraints. For example, one of the most common types of infeasibility is caused by a violation of flow-balance, in which no solution can fulfill 100% of customer demand. A model can be infeasible for a variety of reasons, and it must be examined carefully to understand the root cause(s).

Infeasibility can occur within a variety of solving methods, including:

• Network Optimization
• Greenfield Analysis

When you run a model, the design engine determines whether or not the problem is infeasible. If infeasibility is detected in the model, you receive a message specific to the solving method.

If the infeasibility is detected during Network Optimization, you can run the model through Infeasibility Diagnosis. This algorithm guides you through the process of resolving the infeasibility by providing information in the Constraint Summary output table, or in the Customer Flows, InterSite Flows, and Production Process Flows output tables.

Network Optimization

Infeasibility in Network Optimization can be attributed to network structure, constraints, or capacity limitations. By default, Network Optimization tries to achieve flow balance so that 100% of customer demand can be fulfilled. Any model elements that limit flow balance can trigger infeasibility. The infeasibility is usually linked to missing sourcing or transportation lanes, capacity limitations, inventory requirements, or production requirements.

Specific examples include:

• Sites that are excluded in the Sites table but required to bring product to customers.
• Sites with hard capacities that limit throughput and disrupt flow balance.
• Sourcing or transportation policies that are missing or have capacities that limit flow paths to customers.
• Flow constraints, inventory constraints, or production constraints that contain broken logic or prohibit flow balance:
  • Flow Constraints for site or customer sourcing policies or transportation policies that do not exist
  • Flow Constraints with Min, Max, or Fixed flow requirements that conflict with other capacities or constraints
  • Inventory Constraints for sites that do not have an Inventory Policy or conflict with other inventory requirements
  • Production Constraints for sites that conflict with other site capacities or requirements
• Count constraints that do not use groups configured as “All” or that generate conditions which are impossible to solve.
• Site constraints that do not use site groups configured as “All”, where the sites in the group do not have a site Status designated as Consider, and the site Type is not Potential Site.

Greenfield Analysis

The causes of infeasibility within Greenfield Analysis also are limited. Generally, the infeasibility is created by an inability to meet Last Mile Service Constraints or the improper setup of greenfield inputs.

Examples include:

• No valid demand points or customers exist in the model.
• Maximum sourcing distance is too constraining – the number of defined Greenfield Sites or the pool of potential locations is not sufficient to meet the Max Sourcing Distance specified in the greenfield constraints options.
• Last Mile Service Constraints are too constraining – the number of defined Greenfield Sites or the pool of potential locations is not sufficient to meet the last mile service constraints.