Laser powder bed fusion is a metal 3D printing process where a laser melts thin layers of metal powder along carefully planned paths to build up a part. Simulating the heat generated during this process is important for predicting defects and optimizing print quality, but it is computationally expensive because the laser moves along thousands of tiny scan lines and the thermal effects of each one must be tracked across the entire part and throughout the build.
A promising approach to this problem combines two mathematical tools: analytical solutions, which use exact formulas to describe how heat spreads from each laser scan line, and finite element methods, which numerically solve for temperature across the part on a mesh. The trouble is that as more scan lines are completed, the number of analytical contributions that must be evaluated keeps growing, making the simulation slower and slower over time. The authors address this by introducing a "localized" version of the method that cuts off analytical contributions once they are far enough away in space or old enough in time to have a negligible direct effect. Crucially, they also derive correction terms that feed the missing thermal information back into the finite element part of the simulation, so that heat that has already spread through the part is not simply forgotten.
Testing the method against both a full finite element simulation and the older hybrid approach shows that the localized formulation stays highly accurate while delivering substantial speed gains. In a representative case involving 500 scan lines, the time spent on analytical calculations dropped by a factor of roughly 25, and the overall simulation ran about four times faster, with the temperature field differing from the reference solution by less than 0.14 percent. The correction terms proved essential: without them, cutting off old contributions introduced much larger errors, confirming that the bookkeeping step is what makes the speedup practical for real engineering simulations.