In this research we propose and test two non-optimal facility layout planning models based on the AI search technique "filtered beam search". The first model is a combination of consturction type with iterative improvement type facility layout planning methods, while the second is a pure iterative improvement method. The performance of the two models (quality of solution vs. processing time) is compared against CRAFT, a widely applied method in layout planning, and both methods prove to be competitive. Solutions from the two models are compared to the solutions generated by CRAFT for problems of varied sizes. CRAFT was run 100 times for each problem, and in each time it started from a different initial layout generated randomly. The first model generated consistentiy better solutions than the best solution obtained by CRAFT, and the solutions generated by the second model were consistently among the best solutions obtained by CRAFT.