A PhD position is available on methods/tools for “control architecture” design, with deadline May 23, 2021. For details, see PhD Research Fellow in Process, Energy and Automation Engineering (205482) | University of South-Eastern Norway (jobbnorge.no). Applications must be made via web form in the above link.

The idea of the project is that for a given dynamic model, the zero dynamics between chosen inputs and outputs will (to some degree) limit how fast it is possible to make the closed loop system. By changing inputs and/or outputs, the zero dynamics will change. With more inputs (actuators) and outputs (sensors), the limitation of the zero dynamics may be reduced. When using optimal control, other types of zero dynamics than the transmission zero dynamics may become relevant.

With *control architecture*, it is understood the choice/design of inputs and outputs.

So – how does one find a balance between choices of inputs (actuators) and outputs (sensors), and the cost of purchasing, installing, and maintaining/operating the instrumentation of the system? Key questions relate to:

- How to quantify how fast a system can be made, depending on choice of inputs and outputs (location, number, etc.)
- How to quantify the cost of instrumentation?
- How to find the optimal solution, possibly for uncertain, large scale systems.
- Initially, methods and tools should be studied for linear dynamic systems, with a possibility to extend this to more advanced models depending on time and potential.

The problem is particularly relevant in cases where it is costly to change the control architecture once a system has been built. In this case, the application is oil/gas production, but the ideas are relevant for many other types of systems.

The PhD position will be part of a larger project. Some models in this larger project will be based on MATLAB tools, but it is of particular interest to see how the ideas can start from a model described in, e.g., the Julia ModelingToolkit, and extract simplified models from there. Optimization to find the best control architecture will possibly involve mixed integer programming.

A suitable background for a candidate is within control engineering or numerical mathematics, but candidates from other quantitative engineering disciplines are also welcome.