IFP : Innovating for energy

The division's skills are embodied by four departments.

The Division's scientific and technical ambitions

• remaining internationally competitive, from the scientific and industrial standpoint, in the following fields:

• engine control in the broadest sense (for clean combustion, post-treatment, bio-fuels, hybrid vehicles), which includes the important aspect of estimating unmeasured variables
• computing and numerical performance of porous media flow computation software
   

• set up, for the Business Units, working with the "industry" research divisions:

• software development platforms
• algorithmic tools, software, and hardware for better use of experimental data (databases, experimental designs, on-line optimization, model-fitting tools for experiments)
   

Activities of the Technology, Computer Science, and Applied Mathematics division for the various Business Units

Exploration-Production Business Unit

The work done by the Technology, Computer Science, and Applied Mathematics Division in the projects of the Exploration-Production Business Unit is primarily research, design, and execution of:

• the mathematical parts of codes simulating flows in porous media, multiphase flows in the fluid stream (numerical schemes, solvers, multiphysics couplings)
• high-performance calculation, design of software architectures, data models, geometrical modeler, meshing, visualization in codes for flow simulation in porous media and for reservoir characterization
• design of software architectures in codes for fluid stream and gas treatment management

The codes mentioned involve mathematical and computing difficulties. Consider, for example, flow simulators for the storage of CO₂ in reservoirs and basins: the complexity of the fluids simulated (three phases, several components in each phase), of the discontinuities of various types (discontinuities in the type of rock, in shape), and of the couplings of complex physical phenomena (thermodynamic, geochemical, etc.) lead to difficulties in the numerical schemes and the solvers. The complex (uneven) shapes of the subsoil and the bulky data sets create difficulties in the geometrical construction (areas, volumes, meshes). The geological objects must be understood and exchanged by the various software programs of the same family even if they were created in different contexts, which aggravates the problem of data exchanges. All of these difficulties have been approached and solutions found.

The Technology, Computer Science, and Applied Mathematics Division makes an important contribution to the "industrial software" strategy of the Exploration-Production Business Unit:

• results on parallel linear solvers of internationally recognized quality (benchmark on public data set); industrial competitiveness depends to a large extent on the performance of these solvers
• lthe software architecture of the Business Unit's XXFlow line is based on the OpenFlow platform, designed and implemented by the Technology, Computer Science, and Applied Mathematics Division. This platform is based on Eclipse (www.eclipse.org)
• computer science prime contracting in the production of industrial software (the Business Unit is ISO-9001 Quality certificated for the "Development of industrial software from research results")
   

Refining-Petrochemicals Business Unit

The Technology, Computer Science, and Applied Mathematics Division contributes to the projects of the Refining-Petrochemicals Business Unit through its expertise in:

• "advanced control" (automation, identification), making it possible to improve the performance of processes installed at industrial sites
• decision aids, for better monitoring of the functioning of the units and decision making when there are malfunctions (diagnostics, fault location and identification)
• statistics and optimization to assist modern use of IFP's testing resources (e.g. "high flow" catalyst synthesis tools)
• signal and image processing to improve IFP's analysis equipment in terms of speed, ease of processing, and reproducibility
• databases for rationalized use of the experimental data of the whole Refining-Petrochemicals Business Unit.
   

Engines-Energy Business Unit

The "control" skill appears in most of the Engines-Energy Business Unit's technical themes, either explicitly or implicitly. This is because low-emissions technologies need control to work at all (and not just to work better). Bio-fuel engines, which have variable energy characteristics, need control for optimal operation. Note that by "control" we mean research and applications development work in automation, signal processing, real-time computing, and embedded electronics.
 
IFP's engine control is based on multivariable control techniques and observers (estimates of unmeasured variables) to manage transitions correctly and specifically - both sudden variations in torque demand and the slow changes resulting from the aging or fouling of the actuators and catalysts. The observers developed at IFP are preferably based on simplified 0D knowledge models, reworked to be completely real-time.
 
The contributions to the Engines-Energy Business Unit are, in methodology:

• application of control to conventional gasoline and diesel engines fitted with sophisticated turbocharging and displacement-reduction technologies
• application of control to "clean" combustion processes (HCCI, CAI)
• treatment of combustion gases
• control of hybrid vehicles
• energy management and optimization in hybrid vehicles

...and in software and hardware tools, which are instruments for doing research in control, and at the same time "Tools of the trade" for the Business Unit:

• the "Hardware in the Loop" software and hardware platform, which is the fundamental structure for all current and future tests bench equipment that integrates the models
• Acebox electronic equipment (engine control prototyping tool)
• embedded electronics for demonstrator vehicles
• IFP's real-time kernel incorporated in Morphée 2 software (engine test bench supervisor), a commercially available product of the D2T company. 

Publications

• Review articles and conference papers
>> List of review articles and conference papers for the period 2005-2007 (PDF - 180 Ko)

• Patents
>> List of patents for the period 2005-2007 (in French, PDF - 80 Ko)

• Books
Vehicle Propulsion Systems. Introduction to Modeling and Optimization, 2005 Springer, L.Guzzella; A.Sciarretta

Dissertations

The division recruits about 7 doctoral candidates a year, and was awarded the ParisTech Dissertation Prize for 2007, for a dissertation in collaboration with the Ecole des Mines of Paris (in the context of an Engines-Energy Business Unit project)

Human resources

The division's personnel include engineers (85), technicians (7), doctoral candidates (20), post-docs (2) and trainees (6).

	Oil & Gas Science and Technology - Revue de l'IFP THEMATIC DOSSIER: "Software Interoperability for Petroleum Applications" Download the articles free of charge

	Oil & Gas Science and Technology - Revue de l'IFP THEMATIC DOSSIER: "New Trends on Engine Control, Simulation and Modelling" (IFP International Conference) Download the articles free of charge

Technical overview of the division's work themes for porous media flow computation software (projects of the Exploration-Production Business Unit)

These software programs concern models of the geological storage of CO₂, basin models, reservoir simulators.

Recall that the problems dealt with in the geosciences are characterized by:
 
• the heterogeneous and multi-scale character of the geological medium: stratigraphy, faults, fractures, channels, geostatistic description of the medium
 
• the complexity of the geometry and of the mesh: representation of the well, representation of the horizons and faults, kinematics of the basins in large deformations
 
• the strong nonlinearities and the multiphysics couplings of the dynamic models:

• range of flows in porous media
  
  • from single-phase to three-phase with many components
• complex closure laws
  
  • thermodynamic and chemical equilibria
• many couplings
  
  • thermal
  • reservoir-well-surface network
  • chemical kinetics (water-rock interaction, cracking, combustion)
  • geomechanical (large deformations in basin))
     

• the growing size of the simulations on massively parallel distributed architectures
 
• large volumes of data to be displayed, to be exchanged between applications via "workflows"

The division's R&D work is aimed at increasing the power and accuracy of the discretizations, the numerical methods, and the solvers in order to adapt to this complexity, which is both physical and computational (evolution of architectures), and to the robustness constraints on industrial codes. It is also aimed at dealing with the problem posed by the data to be displayed, manipulated, and exchanged between applications that are different but operate in the same "workflow".
The solutions used are shaped essentially by the type of mathematical model, the physical phenomena studied, and the determination to use a common data model for the various applications involving transport in porous media (storage of CO₂, reservoir and basin simulations).
 
The computing design of the new-generation simulators under development emphasizes modularity and distributed processing, using the Arcane platform the CEA-DAM has been developing since 2000, the object of a joint development venture with IFP. The goal is to provide IFP's research divisions with computing and numerical tools that are more modular, suitable for distributed processing (on meshes), and allow more effective consolidation of research results.
 
It also provides a solution to the problem of data interchange between applications. The solution used today (totally compatible with the Arcane platform mentioned above) is a common geosciences data model with an exchange mechanism, globalized under the name "Openflow platform". This platform, based on Eclipse (www.eclipse.org), developed by the Technology, Computer Science, and Applied Mathematics Division, is the basis for all of the Exploration-Production Business Unit's industrial and research porous-medium flow software, today and for the next ten years.

Examples of current research themes

• Discretization of multiphase flows in porous media

• Meshes suited to the geosciences
• Finite-volumes schemes for diffusion on general meshes
   

• Modeling of faults and networks of conducting faults for basin models and the geological storage of CO₂

• Corefining algorithms
• 4D meshes for basin models in complex tectonics
• Numerical schemes
• Solving of nonlinear and linear systems
   

2D basin model in complex tectonics (mesh mobile according to the material)

3D basin model in complex tectonics (mesh mobile according to the material)

• Modeling of well and well-tank-surface network couplings

• Multiphysics couplings
  
  • reservoir-well vicinity-well-surface network
• Solving of nonlinear and linear systems
• Coupling of time scales
• Discretization in the well vicinity
  
  • Meshing, numerical schemes.

Hybrid mesh around a multi-branch well

• Compositional models of multiphase flows in porous media

• Solving of nonlinear and linear systems
  
  • Large number of variables per mesh
  • Management of calculations of thermodynamic equilibrium and of phase changes
• Solution of thermodynamic flashes
• Thermal heavy-oil recovery models in reservoir simulation
   

• Coupling with Geochemistry to model the geological storage of C02 and diagenesis in basins

• Multiphase flow and reactive transport coupling
• Transport-geochemistry coupling
• Large number of chemical species
• Coupling of time scales
• Tracking of dissolution/precipitation fronts
• Solving of nonlinear and linear systems
• Parallel algorithms
   

Example of simulation of CO₂ injection in the Utsira formation

• Parallel linear solvers

Larger simulations, more heterogeneities, more complex meshes, and the evolution of hardware architectures require a new generation of linear solvers for our systems, the solution of which accounts for the bulk of the CPU time of a simulation.

	Oil & Gas Science and Technology - Revue de l'IFP THEMATIC DOSSIER: "Software Interoperability for Petroleum Applications" Download the articles free of charge

Technical overview of the division's work themes in engine control (projects of the Engines-Energy Business Unit)

IFP's positioning in Engine Control

See: Industrial development > Engines > Engine Control

The engine control environment at IFP - skills, software, hardware

The basic skills are Automation and Signal Processing, leading in the end to embedded software when the engine is in the vehicle, or software acting on the engine on the bench, via electronic equipment, for the purpose of "prototyping" algorithms (See Figure, "Embedded control" part).

The Control tools and environments are:

• Matlab, Simulink
• AMESim, 0D modeling software developed by the Energy Applications Techniques Division.
  See also Industrial development > Engines > Engine Control (See Figure, "Engine Simulator" part)
• electronic equipment for the prototyping of control algorithms (See Figure, "Embedded control" part; ACEbox II equipment and the embedded computer are devices built by IFP)
• a software/hardware platform that can be used to develop and parameterize algorithms, taking the validation as far as possible before execution on test benches (See Figure, HiL (Hardware in the Loop) part)
• the actual system - engine on bench, or engine in vehicle - where the algorithms developed are ultimately validated.
   

Notes on the "model- based control" approach to engine control at IFP

To facilitate engine control development, 0D models, in a generic sense, are used extensively, mainly on 3 different but complementary levels:

1. control design support: 0D simulation models can help in the design stage and in the first engine control tests. They enable control engineers to design the control function and test it in a simulation with the 0D model as reference. The model then functions as a virtual engine, a simple engine control design aid. In this stage, further reducing the 0D model can be considered, to make it better suited to the control function or more useful for it. Since there are no real-world constraints in this stage, this type of 0D simulation is just an integration of differential equations, thus, as for the computing, the associated simulation platform works with a variable-time-step solver (not in real time)
    
2. formalism for control design: the 0D model, i.e., a set of differential equations, replaces the heuristic data (maps, gains, etc.). These equations describe the dynamic behavior of the subsystems to be controlled and serve as an entry point into the "state-space formalism" (another formalism in control theory is input/output). This formalism can be used to design control laws or "observers" that estimate the unmeasured variables when necessary if, of course, the systems are "controllable" or "observable", respectively. This is Model-based Control: the synthesis of control laws is based on the 0D model. This type of model is then incorporated in an embedded computer where it is coded with a discrete-state formalism, and it runs in real time.
    
3. support for the final validation of the control function and the calibration of control parameters: the 0D model is used as a reference model to calibrate, if possible automatically, the engine control parameters that act on this model in simulation. The purpose is that of gaining calibration time without affecting the testing resources. The stakes are obviously very large, given the large gain in industrial terms. The scientific interest lies in the use of optimization techniques, experimental design, and closed-loop identification techniques. The latter are rarely, if ever, used in the automobile industry.
    

Overview of R&D themes in engine control

The research and development work in progress concerns themes concerning hybrid vehicles, new combustion modes, control of the many actuators available to limit fuel consumption and the associated pollution (throttle, use of one or more turbochargers controlled by waste gate or VGT, EGR valve, use of camshaft phase shifters, etc.). Work is being done, for example, on:

• the air loop: the objective is to estimate the mass of fresh air and of burnt gases, generate an allowable trajectory, given the saturations of the actuators and the pollutant constraints, minimizing the response time with respect to the driver's demand and providing faster/more powerful control of the closed-loop dynamics,
   
• the fuel loop: the objective is to estimate the quantity injected and provide closed-loop control driven by the air loop,
   
• estimating fuel composition in Flex-Fuel (or natural gas) engines by developing a suitable, robust observer to allow on-line control of combustion,
   
• the treatment of combustion gases. The commands to be implemented will use more feedback loops and real-time models, created by simplifying physical models and refitting them on line using the available sensors
   
• control of the powertrain, where various actuators will have to be considered: ICEs or electric motors, clutches, alternators, gearboxes, etc. Each of these components has a continuous behavior and a discrete behavior. They act on the vehicle's transmission, which includes linear parts (elasticity, friction), and nonlinear elements (backlashes, tires). The control laws developed must ensure a good drivability, while enabling the hybrid operation, through management of the activations/de-activations of components and of the transitions between the various operating modes.
   
• energy management and optimization in hybrid vehicles.
   

	Oil & Gas Science and Technology - Revue de l'IFP THEMATIC DOSSIER: "New Trends on Engine Control, Simulation and Modelling" (IFP International Conference) Download the articles free of charge

Technical overview: various control functions in energy optimization involved in the ICE/electric hybrid vehicle

The problem: how to split the power between the engine and motor

The "control" function must distribute the power needed to propel the vehicle between the engine and the electric motor. This distribution must take into account contradictory requirements (e.g. minimize CO₂ emissions, maximize the life of the electrochemical batteries and the time between charges) and also many operating constraints imposed by the various technologies involved. This problem can be dealt with by optimal control techniques (off-line to estimate the best performance that can be attained by a given vehicle configuration; on-line to manage the energy flows when the vehicle is in operation). The researchers of the Technology, Computer Science, and Applied Mathematics Division master the most advanced algorithms and develop innovative techniques - up to and including their implementation and validation on board of IFP's demonstrator vehicles.
 
The goal of the research work is to extend this notion of "optimal energy management" to different control tasks that are crucial for hybrid vehicles. For example, management of the embedded electrochemical batteries is also a control task, and it requires the most accurate possible estimate of their unmeasurable internal condition - charge, general condition, temperature. Algorithms are therefore being developed to supply these estimates as efficiently as possible, given the available measurements and their uncertainties, using a physical model of the components, a model both fine enough and simple enough to be useful in real time.

"Observer" of battery charge condition, using external measurements

"Observer" of battery charge condition, using external measurements, that supplies input for the computations of the "controller"