Systems, methods and computer program products leveraging digital twin modeling and cognitive computing to predict lubrication replacement for a physical asset. Predictions of lubrication replacement consider one or more various parameters such as operating conditions, usage parameters, the surrounding environment, overall health and state of repair of the physical asset, lubricant properties and historically collected data from the physical asset (or similarly comparable assets). Timing for optimal lubrication replacement is identified using the collected data of the physical asset, along with historical data, to simulate changes in a state of lubricants and lubricated parts within a physical asset using digital twin modeling to make predictions how one or more actions upon the physical asset impact the health, stability and/or longevity of the lubricant's lifespan. Based on the simulation results, recommended action(s) suitable for increasing and optimizing the overall life of the lubrication are provided and/or implemented.