Mohammad Sayyafzadeh

Dr Mohammad Sayyafzadeh

Lecturer

Australian School of Petroleum & Energy Resources

Faculty of Engineering, Computer and Mathematical Sciences

I joined Australian School of Petroleum as a lecturer in 2013. I have a B.Sc. in Chemical engineering from Tehran Polytechnic, an M.Sc. in Reservoir Engineering from Tehran Polytechnic and a Ph.D. in Petroleum Engineering from the University of Adelaide. My main field of research is Applied and Computational Mathematics targeting Reservoir Engineering problems, such as history matching, field development optimisation and uncertainty quantification, and my teaching interest includes mathematical and numerical modelling of fluid flow in reservoir rocks, inverse modelling, optimisation and geostatistics.

My main field of research is Applied and Computational Mathematics targeting Reservoir & Production Engineering problems. That includes 

  • Computer-assisted algorithms for history matching 
    History matching is a nonlinear and computationally intensive inverse problem in which it is sought to tune (calibrate) the coefficients, initial and/or boundary conditions of nonlinear PDEs corresponding to multi-phase flow in porous media, based on observed data.
  • Techniques for uncertainty quantification
    The propagation of uncertainty in the parameters of interest (e.g., reservoir performance forecasts) can be obtained by drawing samples from a nonlinear and high-dimensional posterior probability density function (in a Bayesian framework).
  • Robust optimisation algorithms for field development planning and reservoir flooding improvement under geological uncertainties
    These are numerical optimisation exercises with a highly nonlinear and uncertain (noisy) objective (fitness) function. Each function evaluation needs multiple execution of computationally expensive reservoir simulation.  
  • Surrogate (proxy) modelling techniques for reducing computational costs
    Reservoir production optimisation problems are computationally intensive, due to the nature of PDEs solved numerically for production forecasting. The computational costs associated with optimisation and calibration problems can be reduced by applying properly an approximation functions in the workflow.  
  • Linear and nonlinear Algebraic transformation methods for dimensionality reduction 
    A dimensionality reduction technique can improve the calibration problems, given the fact that the data are correlated to some extent and finding a basis that spans the search space will improve the process.
  • Modelling of unconventional resources (such as CBM and tight-sand reservoirs), ECBM, CO2 sequestration and uncharacteristic phenomena in conventional reservoirs
    In order to simulate reservoir performance in some cases, e.g., unconventional plays, sometimes, it is required to tweak the existing tools or solve numerically a new set of governing PDEs that replicates the phenomenon.  
  • Data Analytics (DA) for fast decision making
    DA is gaining popularity in Oil & Gas industry in the recent years, due to the massive information gathered everyday with less time to analyse. DA can be an alternative to the physical-based simulators for making fast decisions.
  • Full-parameterised history matching by stochastic wavelet bases for highly heterogeneous reservoirs (Lead Investigator), SANTOS Ltd, 79k. (2016-2018)
  • Enhanced gas recovery using improved flow-back in fracture treatment in tight gas reservoirs, Cooper Basin, (Co-Investigator), client: SANTOS Ltd, 80k. (2014-2017)

My teaching interest includes mathematical and numerical modelling of fluid flow in reservoir rocks, inverse problem theory, uncertainty quantification, numerical optimisation and geostatistics. I currently teach the following courses: 

1. Reservoir Simulation: The course gives the theoretical basis and practical fundamentals for mathematical modelling and numerical simulation of fluid flow in petroleum reservoirs. The governing laws and equations required for the modelling of single-phase and multi-phase flow in porous media, such as mass conservation, Darcy, equation of state, rock compressibility, capillary pressure and relative permeability, are reviewed. By combining these laws and equations, the corresponding partial differential equations are derived. The numerical methods for solving the governing partial differential equations using finite difference methods are presented.

2. Reservoir Characterisation and Modelling: The course has three main components. 1) Data sources, quality and analysis, including spatial analysis. 2) Generating 3D models of reservoir properties - classical gridding and mapping, kriging as a data-driven (variogram) form of classical mapping (estimation) and a means of data integration. Simulation techniques are introduced as a means of assessing uncertainty resulting from heterogeneity. 3) Scaling of grids and property models for the purpose of reservoir simulation is the final topic.

3. Advanced Topics in Numerical Reservoir Simulation: This course reviews the governing PDEs of multi-phase flow in porous media derived with a black-oil phase-behaviour approach, and presents the derivation of the PDEs with a compositional phase-behaviour approach (using both 2-parameter and 3-parameter equation of state). A commonly-used numerical method (finite volume method) for solving the governing PDEs is discussed, and space discretisation (27-point and 7-point) using quadrilateral grids, nonorthogonal (corner-points) and orthogonal (block-centred), is reviewed. An overview of Newton-Raphson linearisation methods in fully-implicit, IMPES and AIM scheme, is given. Iterative linear solvers for sparse matrixes are reviewed, and a few techniques for paralleling and tuning the solvers are discussed. The course, in addition to the fundamentals, covers several practical and special topics in reservoir simulation, such as, placement of deviated and multilateral wells, group controls for constraint handling, local grid refinement and coarsening, miscible and immiscible gas flooding, gas condensate, regionalisation (PVT, equilibrium and SCAL), dual porosity model for naturally fractured rocks, adsorption models, aquifer models, rock compaction/swelling and history matching. 

4. Uncertainty Modelling: The course gives the theoretical basis and practical fundamentals for uncertainty quantification and modelling (forward and backward), inverse problems and numerical optimisation. It outlines the types and sources of uncertainty, and the importance of uncertainty modelling in decision-making processes. The forward propagation of the uncertainty in the parameters of interest using different techniques, such as Monte Carlo simulation and experimental design methods, is discussed, and techniques used for drawing samples (unconditioned or directly conditioned) from multivariate distributions are reviewed. A particular attention is paid to inverse modelling (in linear and nonlinear problems) with a Bayesian approach. Popular calibration algorithms, gradient-based (steepest descent and quasi-Newton) and derivative-free used for approximating/estimating Maximum a Posteriori (MAP) and Maximum Likelihood (ML) are discussed. Gradient computation/approximation techniques in high-dimensional problems are also reviewed. The fundamentals of Markov chain Monte Carlo (MCMC) are discussed, and different techniques used for the approximation (sampling) of posterior probability density function, such as Metropolis–Hastings algorithm, data assimilation (ensemble Kalman filter) and reduced-order-model-assisted and surrogate (metamodel)-assisted algorithms, are presented and discussed. This course also reviews the algorithms and techniques used to optimise noisy single and multi-objective functions (with and without constraints), such as might be found field development and production optimisation under geological uncertainty problems.

5. Introduction to Petroleum Engineering (Production Engineering Part): The aim of the course is to provide students with a broad overview of introduction to petroleum engineering in order that advanced courses in subsequent years can be understood within a broader petroleum engineering context. 

 

    Expand

  • Current Higher Degree by Research Supervision (University of Adelaide)
    Date Role Research Topic Program Degree Type Student Load Student Name
    2019 Co-Supervisor Reservoir characterization using stochastic wavelet basis Doctor of Philosophy Doctorate Part Time Mr Roozbeh Koochak
    2018 Co-Supervisor Optimisation of Well Placement, Trajectory and Control Under Uncertainty Doctor of Philosophy Doctorate Full Time Mr Yazan Arouri
    2018 Co-Supervisor Innovative Techniques for Reservoir Surveillance and Interpretation in the Mature Windalia Waterflood Doctor of Philosophy Doctorate Part Time Mr Daniel O'Reilly
    2016 Co-Supervisor CO2 as an Agent for Enhanced Oil Recover: A Reservoir and Geomechanical Analysis Doctor of Philosophy Doctorate Part Time Mr Abbas Movassagh
    2014 Co-Supervisor The Potentiality of Enhanced Oil Recovery by Alkaline, Sunfactant and Polymen Flooding Based on Applied Reservoir Simulation Doctor of Philosophy Doctorate Part Time Mrs Sume Sarkar
  • Position: Lecturer
  • Phone: 83138023
  • Email: mohammad.sayyafzadeh@adelaide.edu.au
  • Campus: North Terrace
  • Building: Santos Petroleum Engineering, floor 3
  • Room: 3 02a
  • Org Unit: Australian School of Petroleum & Energy Resources
Connect With Me