Besides the system inherent functions and their combination to fulfil the overall purpose of the system, any industrial plant relies on control functions to ensure the nominal operation of specic safety- or production-relevant functions. In order to be able to model a system including its control functions, Morten Lind introduced a representation of control actions for MFM based on a logical notation of actions. This notation considers different states and expresses their connection with a temporal operator T and an operator I representing what would have happened without intervention. Thus, the schema [pTqIr] denotes that the considered system transitioned from state p to q instead of r, where r would have occurred without intervention. The underlying concept identifies a small number of elementary interventions that are adapted for MFM and shown in the table below. These control functions do not represent the details of the implementation of a specic controller but rather resemble the intended function to e.g. keep the flow in the heat exchanger steady.
Analogous to the control actions, means-end relations representing the same intentions have been introduced inMFM. These means-end relations connect the state of a function to an evaluable objective or threat. Lind discusses meaningful combinations of the means-end relations with either threats or objectives, and a control function. Limiting the combination of those ensures logical consistency. Theses combinations also reflect that the only control relation normally associated with control functions as a controller is actuate. However, neither the distinct means-end relations nor the control functions have been thoroughly interpreted with regards to the reasoning about fault cause and consequence. The influence of control relations is yet to be investigated as well. Following the notion that this concept of control serves to model the purpose or intention of the controller rather than detailing its implementation, Zhang propose to represent operational knowledge about the designed purpose of possible manual or automatic action. Thus, including control functions that are not or only partially automated can be a way of extending the knowledge represented by the MFM model and enable better operations support by incorporating more expert knowledge.