Adaptive control of stochastic batch lumber kilns

Abstract Lumber drying is a stochastic batch process. Thousands of dimension lumber are stacked and charged into a kiln where circulating air with controlled and varying temperature and relative humidity is used as the drying medium. The objective is to maximize overall lumber quality within reasonable drying time. Due to varying individual lumber properties and other product and process conditions, boards show a distribution of drying related degrade; and furthermore, consecutive batches do not necessarily behave the same way. The control problem is to continuously adapt a drying schedule for the remaining time to maximize the performance of each individual batch. Final product value can be inferred through a stochastic process model which in turn can be updated for accuracy by an observed process variable. Actual product value can only be assessed at the end by direct measurements.