Many industrial facilities consist of multiple units working in parallel to reach a production target or to maximize production revenues. These units deteriorate due to stress that is often dependent on the production speed, and thus maintenance needs to be performed to ensure that the system remains functional. Maintenance generally requires a considerable amount of time and maintenance capacity is often limited. Therefore, during maintenance of a specific unit, a failure of another unit may lead to additional downtime until maintenance capacity becomes available again. In this study we examine the value of condition-based production in such a system, and jointly optimize production and maintenance decisions. The problem is formulated as a Markov decision process, and the results are compared with a benchmark policy without an adjustable production speed. A numerical analysis shows that adjustable production speeds are used to actively desynchronize the deterioration levels of units for systems with a production target. For systems that aim to maximize total production output, considerable profit increases are obtained by lowering the production speed (i) at high deterioration levels to reduce the risk of failure, and (ii) when the maintenance station is busy to reduce the risk that multiple units require maintenance simultaneously. A sensitivity analysis is used to illustrate under what circumstances joint condition-based production and maintenance decisions are most beneficial.