Manufacturing companies often face challenges in maintaining machine performance to avoid operational losses. PT. XYZ is a company engaged in the flat glass production sector. The company experienced problems from water heater engine damage caused by several factors; the main problem found was the frequent occurrence of extended failures due to maintenance errors that increased the risk of damage to the engine and stopped production. Therefore, the use of the intensity function is a concept that can predict engine failure by involving historical analysis of the damage using Weibull and exponential distributions to predict the time of failure. Thus, this analysis results in two maintenance policies that are compared, namely Periodic Replacement Policy (Policy 4) and Repair Count Policy (Policy 5). The results of the study show that Policy 4, with an optimal time to replace the engine every 923 hours or 39 days, results in lower maintenance costs (Rp 11,447,000) compared to Policy 5 (Rp 15,500,000) and an estimate where engine damage will begin to occur every T=200, and it is recommended to repair and replace the Main Unit every T=800. In conclusion, updating the preventive maintenance schedule with the implementation of Policy 4 can minimize maintenance costs and the risk of machine damage, thereby increasing the company's operational efficiency.