Microalgae Bioreactor Control by Model Predictive Techniques

Tuesday, October 18, 2011: 4:15 PM
101 J (Minneapolis Convention Center)
Javad Abdollahi, Chemical and Materials Engineering, University of Alberta, Edmonton, AB, Canada and Stevan Dubljevic, Chemical & Materials Engineering, University of Alberta, Edmonton, AB, Canada

Bio-fuel is a sustainable replacement for fossil fuels because of less pollution and its renewablity, however production costs makes it not competeble with traditional fuel sources [1]. Today, industrial biodiesel production is based on oleaginous crops. Cultivation of these crops needs vast land and high amount of water. By the ndings from [2], the total farmland area of the earth is needed to produce biodiesel to support current power plant and transportation fuel consumption. Also, it is expected that the world will need almost 60% more energy in 2030 if current policies are held [3]. Bio-fuel production from phototrophic and heterotrophic microalgae is a novel method of oil production that is going to replace bio-fuel production from crops, see [1]. In this method there is no need for vast land, high amounts of water and in the case of heterotrophic microalgae there is no need for sunlight, see [1]. Furthermore, increasing oil production rate can ensure the economic feasibility of biodiesel production. In this work we explore model predictive techniques to maximize oil production rate. 

We built this work on an existing heterotrophic microalgae fed-batch model [4] which describes the algae growth and oil production. The fed-batch microalgae reactor is fed by two nutrients, glucose as carbon source and glycine as nitrogen source. The experimental reactor setup is realized such that the temperature and pH are regulated at constant values by a conventional controller and the reactor is monitored by Raman spectroscopy laser which measures certain species and substrate concentrations. In addition from the model [4], growth, nitrogen uptake and oil production rates are highly nonlinear and the maximum value of rate functions are time varying [4].

In this work moving horizon estimator (MHE) and model predictive control (MPC) are used to estimate the states and parameters of microalgae bioreactor and consequently optimize and control the oil production rate. The highly nonlinear rate functions, time varying unknown parameters, unavailable full state measurements and uncertainties in modeling make the observer and controller design challenging. Unknown state/parameters estimation is realized by the moving horizon estimator which is necessary to be applied to microalgae bioreactor system since it is suitable for constrained state estimation [5, 6]. In particular concentrations in the bioreactor admit positive values and also the estimated parameters should be maintained in the reported ranges, see [7]. The control objective is to maximize the oil production rate, since feasibility studies showed a potential of about 20 times more production compared to the traditional cultivation of phototrophic microalgae [4]. Finally, the moving horizon estimator and model predictive control is implemented in order to solve the state/parameter estimation and control problem.


[1] Yusuf Chisti, Biodiesel from microalgae, Biotechnology Advances 25 (2007) 294-306 

[2] Martin I. Hoert, et al, Advanced Technology Paths to Global Climate Stability: Energy for a Greenhouse Planet, Science, vol. 298 1 November (2002), 981-987

[3] Vishwanath Patil, et al, Towards Sustainable Production of Biofuels from Microalgae, International journal of molecular Science (2008), 9, 1188-1195

[4] H. De la Hoz Siegler, A. Ben-Zvi, R.E. Burrell, W.C. McCarey, The dynamics of heterotrophic algal cultures, Bioresource Technology 102 (2011) 5764-5774

[5] C.V.Rao, J.B. Rawlings, D. Q. Mayne, Constrained State Estimation for Nonlinear Discrete-Time Systems: Stability and Moving Horizon Approximations, IEEE Transaction on Automatic control, Vol. 48, No. 2, February 2003

[6] A. Alessandri, M. Baglietto, G. Battistelli, Moving-horizon state estimation for nonlinear discrete-time systems:new stability results and approximation schemes, Automatica 44 (2008) 1753-1765

[7] K. Surisetty, H. De la Hoz Siegler, W. C. McCarey, A. Ben-Zvi, Model reparametrization and output prediction for a bioreactor system, Chemical Engineering Science 65 (2010) 4535-4547

Extended Abstract: File Not Uploaded
See more of this Session: Emerging Control Applications
See more of this Group/Topical: Computing and Systems Technology Division