The managerial skills involved in planning and coordinating the activities required in a production process are critical in any business, including agriculture. Poor design may lead to poor performance, e.g., bad timing of activities and unbalanced utilisation of resources. Consequently, production processes need to be carefully analysed in order to evaluate their potential performance, identify vulnerability or flaws, and contribute to the design of improved systems. This paper presents a software platform that makes it possible to model and simulate agricultural production processes, including the management processes responsible for the dynamic scheduling of activities. The successful completion of any agricultural production process requires solving successive resourceconstrainedscheduling problems, which are defined as the assignment of a limited set of resources to a collection of relevant activities, with the intent of satisfying the farmer's constraints and preferences. The most important resources include labour, equipments (e.g., tractors) and energy.The dynamic nature of the scheduling process is made necessary by the changing environment in which themanagement problem takes place (e.g., weather, epidemic hazard, resource failures, etc.). Due to the unpredictability of changes, scheduling must be reactive in order to produce activities that are coherent with intentions and current circumstances. However, production management is not purely reactive. Farmers rely on activity plans that include a type of proactive anticipatory behaviour. Proper articulation between anticipation and reaction highly conditions production performance. The simulation framework presented here aims at enabling the study of production management by virtual experimentation.In this framework, dynamic scheduling complies with organisational constraints specified in the form of a flexible plan that makes explicit the temporal and procedural dependencies between the activities devised toward the overall objective. The preservation of the flexibility of this anticipatory specification is obtained via the ability to incorporate alternative execution paths, optional activities and the underspecification of both the objects to be processed by the activities and the resources to be used in these activities. Moreover, the plan may be adjusted in response to particular events. The simulation of the application of such a plan requires a continuous interpretation process that aims at generating and executing the flow of activities that obeys the constraints and preferences coming with the plan. This involves the iteration of four steps:1. Monitoring the intended activities that are eligible for execution according to the plan;2. Allocating resources to the largest possible subsets of these activities;3. Choosing the subset of allocated activities to execute; and4. Executing the activities concurrently.This paper only addresses Steps 2 and 3, and outlines the modelling framework developed to simulate the application of management strategies that are explicitly and formally represented. The first originality of this work lies in the characterisation of the different types of resources (e.g., discrete state or capacitated, atomic or aggregated) and the various kinds of constraints that restrict their use (e.g., state or capacity-related availability, simultaneous use of a resource in different activities and with other resources). The second one concerns the algorithm of dynamic resource allocation. It efficiently processes the various constraintsapplying to the resources and activities, and assigns resources to the set of all currently pertinent activities, ifpossible or to a subset of them otherwise.