RESEARCH PROJECTS
  1. Economic Model Predictive Control
    The performance of a process (like a chemical, a petrochemical or an oilsand process) includes many aspects such as profitability, efficiency, variability, capability and sustainability. These objectives traditionally are handled in a two-layer (or multi-layer) hierarchical architecture. In the upper-layer called real-time optimization (RTO), optimizations are solved based on steady-state process models to calculate the best process operating set-points. The optimal set-points are then used by the lower-layer process control systems which drive the process to and maintain the process at these optimal set-points. This hierarchical approach has been widely used and has shown great versatility. However, as the development of process industries, products have been greatly diversified and the time scales on which a process reacts to economic demands has shrunk. In many problems, the dynamic performance is crucial and the RTO-control separation is either inefficient or inappropriate. Efficiency and flexibility are requirements that a control system needs to offer now.

    Economic model predictive control (EMPC) removes the separation between optimization and control and addresses both in one single layer in the framework of model predictive control. EMPC optimizes a general economic cost function which in general is not quadratic. It has been shown that EMPC may lead to time-varying process operation instead of steady-state operation. In our work, we focus on the development EMPC algorithms that ensure closed-loop stability and give guaranteed improved economic performance over a given controller or steady-state operations.

    2. Back to top

  2. Distributed Predictive Control and State Estimation
    Large-scale, highly-integrated complex processes are increasingly used in modern process industries in order to get increased profits while meeting safety and environmental regulations. Model predictive control (MPC) has been widely used in the control of economically important units in these large-scale processes. Typically, centralized MPC is not preferred in plant-wide control of large-scale processes due to organizational complexity, fault tolerance as well as computational issues. For plant-wide control of large-scale processes, the decentralized MPC is one of the solutions. However, it is well known that the coupling between units in many processes such as integrated process networks with large recycles can not be neglected and that the plant-wide performance under decentralized control is limited because the interactions between subsystems in general are not considered.

    The above considerations motivate our research in distributed MPC (DMPC) in which subsystem MPCs communicate with each other to exchange information to coordinate their actions. In our work, we focus on the development of rigorous yet practical DMPC algorithms that give ensured plant-wide closed-loop stability and plant-wide control performance. Our objective is to develop DMPC algorithms that require a minimum level of modifications to the existing decentralized MPC.

    Distributed state estimation is an integral part of distributed process control. In order to maintain the structural flexibility of DMPC, distributed or decentralized state estimation systems should be used instead of centralized ones. In our work, we focus on the development of distributed moving horizon state estimation (DMHE) schemes and distributed Kalman filters (DKF). Our objective is to develop robust and adaptive distributed state estimation algorithms that give provable convergence.

    3. Back to top

  3. Optimal Operation of Irrigation Systems
    The basic need of the human species for survival is food and water, where food production-agriculture in itself requires water. As a result of these basic needs along with all other uses of water, water usage at global level has grown to a rate more than twice the rate of population increase in the 20th century. If this explosive population growth continues by the year 2050, 60 percent more food will be needed to satisfy the demand of more than 9 billion people worldwide, which will necessitate the need for even more water. This implies that there will be more water stress on the available global fresh water resources. Clearly then if policies related to water conservation and management are not carried out with, 'water scarcity' will become a global 'environ-political' issue in the near future.

    Agriculture is the sector where water scarcity has greatest relevance due to the fact that 70 percent of global fresh water is used for agriculture. In the past the problem of water scarcity was only an issue of arid or semiarid land. Even in countries that have enough water resources, water usage in the agricultural sector continues to increase because of emphasis on more local agricultural production as well as health related regulations. According to the statistics from the 'Program on Water Governance' (PoWG), Canada, each year the agriculture sectors consumes 3,036 million cubic meters (MCM) that is 71% of total withdrawn water, in contrast to this the manufacturing industry consumes 552 MCM (13%), thermal power 508 MCM (12%), municipal 119 MCM (3%) and mining only 46 MCM (1%). It is clear then even incremental improvements in agriculture water usage via efficient irrigation may result in substantial saving of fresh water. According to Alberta Irrigation, in the province of Alberta even a saving of as small as 1% irrigated water would result in savings of about 23 million cubic meters of water each year.

    From a systems engineering perspective, the current water usage (at least in Alberta) in the irrigation network is considered to operate under open loop conditions. If water were to be only supplied to farm crops as necessitated from feedback signals comprising of measured soil water content, evapotranspiration rate, nitrogen content and other on-line sensors then the irrigation network could be defined as operating under closed-loop conditions. Preliminary studies have shown that closed-loop operation can lead to a significant saving of water. The ultimate objective of this research is to develop an automated irrigation system which is able to determine precisely the amount of water required for irrigation taking into account different types of measured data such as soil water content, precipitation, temperature as well as satellite imagery information.

    4. Back to top