|
|||
|
Description of the group: "Norway has abundant resources of natural gas. This gas is valuable both as a source of energy and as feedstock for the production of a number of different chemicals. At present, almost all the natural gas produced in Norway is exported via transmission pipelines to continental Europe and the UK. Thus, much of the value creation from this natural resource occurs abroad. The Gas Technology Center NTNU – SINTEF has been established to coordinate research activities in Trondheim in the areas of production, processing, transportation and utilisation of natural gas. The research challenges range from obtaining better understanding of basic phenomena, through improved operation and control, to improving the basis for investment decisions. Environmental aspects are also important. Most notably, there is the issue of greenhouse gas emissions when using fossil fuels for energy production. Minimising carbon dioxide emissions from energy production, as well as handling and disposal of captured CO2, are therefore central research topics. Information and Communication Technology (ICT) is essential both in performing the research as such, and in the industrial application of the research results. In some application areas the ICT tools are applied off-line, whereas other applications involve two-way interaction with operating equipment. Thus, the ICT tools are required to work on-line (in ‘real time’). On-line and off-line use of ICT tools both typically involve extensive use of optimisation and/or simulations, and visualisation tools are necessary for presenting and interpreting results. In on-line applications the reliability of both computations and communications is essential. The communications often have to work over significant distances – for operation of a gas transmission network extending over hundreds of kilometres this should be obvious, but also within a single production facility distance and reliability of communication is important. In addition to the safety and reliability issues associated with the on-line use of ICT tools in advanced operations, ICT tools is also used to monitor and ensure the safety of operating equipment." Position paper Slide Show
Preliminary version, 17.03.2004
Future need for Information and Communication Technology in the areas of Production, Transportation and Utilisation of Natural Gas
ScopeThis memo addresses, as indicated by the heading, the needs for Information and Communication Technology (ICT) in the areas of production, transportation and utilisation of natural gas. Note that the needs for ICT in petroleum prospecting or drilling are not addressed. On the other hand, the operational challenges associated with chemical conversion of natural gas are in many aspects similar to what is encountered in other branches of the chemical processing industries. Many of the generic research needs related to gas conversion can therefore be expected to apply also to other processing industries. BackgroundNorway has abundant resources of natural gas. This gas is valuable both as a source of energy and as feedstock for the production of a number of different chemicals. The export value of natural gas is expected to surpass the export value of oil within a decade.
At present, almost all the natural gas produced in Norway is exported via transmission pipelines to continental Europe and the UKThus, much of the value creation from this natural resource occurs abroad. The Gas Technology Center NTNU – SINTEF has been established to coordinate research activities in Trondheim in the areas of production, processing, transportation and utilisation of natural gas.
The research challenges range from obtaining better understanding of basic phenomena, through improved operation and control, to improving the basis for investment decisions.
Environmental aspects are also important. Most notably, there is the issue of greenhouse gas emissions when using fossil fuels for energy production. Minimising carbon dioxide emissions from energy production, as well as handling and disposal of captured CO2, are therefore central research topics.
Information and Communication Technology (ICT) is essential both in performing the research as such, and in the industrial application of the research results. In some application areas the ICT tools are applied off-line, whereas other applications involve two-way interaction with operating equipment. Thus, the ICT tools are required to work on-line (in ‘real time’). On-line and off-line use of ICT tools both typically involve extensive use of optimisation and/or simulations, and visualisation tools are necessary for presenting and interpreting results.
In on-line applications the reliability of both computations and communications is essential. The communications often have to work over significant distances – for operation of a gas transmission network extending over hundreds of kilometres this should be obvious, but also within a single production facility distance and reliability of communication is important. In addition to the safety and reliability issues associated with the on-line use of ICT tools in advanced operations, ICT tools is also used to monitor and ensure the safety of operating equipment.
In the following, we will describe briefly a number of application areas for gas technology, with operational objectives and engineering problem areas. This will subsequently be used to motivate the need for education and research in various ICT domains.
Application Areas
Process and product design Traditionally, the process design problem has started from a requirement to produce specified amounts of well-defined products, within legislative and environmental constraints, with the objective of maximising the profit earned from the production. This is still the dominant process design scenario, and presents a number of challenging research topics. Such challenges include
Valuable tools exist for addressing each of the areas above, although completely general methodologies are not available. Integrating all aspects of the design problem into a unified methodology is also hard. Typically a number of initial designs showing economic potential have to be evaluated for safety, flexibility and operability.
Some of the tools used in process design are relatively simple to conceptualise, and is at present used systematically by engineers with a Masters degree. An example of such a tool could be the “pinch curve” used in heat integration. Other aspects of the design problem result in highly complex optimisation formulations, expressed as Mixed Integer Non-Linear Programs (MINLP’s). Few Master’s level students have the background to both formulate such problems (requiring knowledge of physics, chemistry and process technology) and to solve the resulting optimisation problems (requiring knowledge of advanced global optimisation tools). A high level of competence in quite disparate areas is required to use such programs and reliably interpret and analyse the results of computer programs used to formulate and solve complex design problems.
Process design is usually formulated using static models. Although this agrees well with the operating philosophy of most process plants (requiring stationary operations, infrequent changes in operating conditions, and a control design focusing on disturbance rejection), formulating the process design problems as static optimisation problems naturally results in a vast reduction in the degrees of freedom available in the design. It is noted above that even static design problems often result in highly complex optimisation problems. Introducing dynamic operation into the design problem will typically increase the complexity significantly.
In the future, the focus is expected to shift from ‘process design’ to a paradigm more centered on the success criteria for the product in specific applications. Thus, the required physical and/or chemical properties of the product need to be specified, without necessarily specifying what type of product should be used to achieve the specified properties. To illustrate, instead of specifying that the plant should “produce X tonnes/day of polymer Y with chain length distribution Z”, the specifications may in the future detail
This clearly adds additional complexity to the design problem, and requires connections to be made between phenomena on a microscopic (possibly down to molecular) level and product properties on a macroscopic level. Effective two-directional communication between different model scales is a central issue. These techniques must reach sufficient maturity (in terms of general applicability and reliability) in order to have realistic hope of making significant progress.
Process operationsProcess operations focuses on the operation of existing plants, and the design of the plant is considered as fixed. The time scales addressed in operations range from fractions of a second (safety, regulatory control) to weeks or even months (in production planning and supply chain management). It is natural that the system descriptions and problem formulations used will differ significantly depending on the type of operational problem and time scale in question.
On the longer time scales, static plant models (e.g., describing production capacities) and discrete decision variables (what product to produce, when to produce, possibly also where it should be produced) dominate. Although alarms and automatic shutdowns may be classified as discrete decision variables, continuous decision variables and dynamic plant models dominate the operational problems in focus at the shorter time scales.
Although significant research problems exist at all time scales, the two ends of the ‘time scale spectrum’ may be claimed to be better understood than the intermediate range of time scales (from tens of minutes to days). Many problems on this time scale require both continuous and discrete decision variables. System stability cannot be ignored on these intermediate time scales, and dynamic plant models are therefore required. Thus, many problems on this timescale are inherently complex. However, significant time is also available to solve these problems, and this is therefore a promising area for advanced ICT. This intermediate timescale also contains the interface between the ‘business operations’ (focusing on the longer timescales) and the ‘plant operations’ (focusing on the shorter timescales). These are two different ‘cultures’, and very few people have deep knowledge of both cultures (or problem areas). Research aiming to bridge the gap between the two cultures should therefore have significant potential.
Operator supportThe term ‘operator support’ is here used for techniques and applications for supporting decisions made by human operators. This includes techniques for - navigating historical data to find situations similar to the present operating scenario or operating problem, - perform ‘what if’-analysis by simulating the future effects of choices available to the operator, - provide advance warning of developing operating problems, - make equipment documentation and operational procedures available on demand, - analyse operational data to identify poorly functioning equipment or control systems, needs for maintenance, etc.
Remote operationOil fields which are remotely operated already exist. Remote operation, with processing equipment located on the sea bed, is clearly an attractive alternative for many deep sea oil and gas fields. Clearly, such remote operation places high demands on the safety and reliability of the equipment and control systems.
Remote operation is also of interest to education and research as such. NTNU has received funds from the EU to make several of its advanced laboratories in the oil and gas area available to researchers from other parts of Europe. Clearly, remote operation could save a lot of travel and subsistence expenses for researchers elsewhere in Europe who perform experiments at NTNU.
Enabling technologiesA number of information and communication technology areas are discussed briefly below, to describe needs and potentials for research in these areas. There is necessarily some overlap with the discussion of application areas above.
InstrumentationThere are many challenges for research and development of new and improved instrumentation in various parts of the natural gas value chain. Instrumentation is here understood to include both measurement and actuation devices.
Downhole (inside a production or injection well) instrumentation must be reliable in very hostile environments, due to high temperature and pressure, erosion and corrosion. Maintenance is at best difficult and costly, often impossible. There are difficulties both in supplying energy for instrument operation , and in communication with surface facilities. Of interest is both measuring instruments (providing information on temperatures, flowrates, pressures, etc.), and actuators for active well control located downhole. Actuators of interest include active gas lift valves, as well as valves for controlling production from various zones within a well.
Instrumentation for sub-sea production facilities is undergoing rapid development. Focus is also here on reliability and cost of maintenance, harsh operating conditions, limited energy supply, etc, although these challenges in general are not as extreme as for downhole installation.
There are significant research challenges also in instrumentation of many of the chemical processes using natural gas as feedstock. For instance, some polymerisation processes operate at extremely high pressures, and even measuring basic variables like temperature and pressure is difficult. More complex properties related to product quality (e.g., viscosity, chain length distribution) often require sampling and laboratory analysis even for processes operating at more benign conditions.
Many research and development needs in instrumentation are common to all branches of the chemical processing industries. These include issues such as improving self-diagnostic capabilities of the instrumentation, ease of installation, configuration and maintenance, efficient communication, etc.
CommunicationIt is noted above that in many natural gas applications the ability to communicate with the instrumentation is a major challenge. This applies in particular to instrumentation inside injection or production wells, and is difficult also for subsea production facilities. Difficulties in communicating with instrumentation is also a severe problem in subsea oil and gas transportation pipelines. Such pipelines can be several hundred kilometres in length, with measurements (and actuators) only at the inlet, outlet and booster/recompression platforms. The information available is thus very sparse, and this is an obstacle to efficient monitoring and optimisation of operations. This problem is more severe for gas transmission pipelines, which have slower dynamic response than oil pipelines.
On production platforms and in chemical conversion plants many difficulties arise from the sheer scale of the system. The number of signals to transmit to and from various instrumentation may well run into tens of thousands. Thus, installing and maintaining communication requires substantial effort. Wireless communication may hold promise in simplifying communication, and will certainly reduce the need for communication cables. Installing and maintaining wireless communication in production facilities will certainly present challenges of its own. However, more serious are the issues of information security and potential for terrorism. The typical manipulated variable in a process control loop receives a new value from the controller once every few seconds. A patient spy or saboteur listening to the communication will therefore have access to massive amounts of data that can be used for cracking data encryption codes. Conceivably, after cracking encryption codes a saboteur may attempt to take over control of parts of a plant, cause destruction to the plant and endanger both pertain personnel and people outside the plant.
ModellingPractically all advanced tools for process design, optimisation, control and monitoring require some sort of plant model. Modelling if therefore a cornerstone in all applications of process systems engineering. Different applications require very different models. Much can be said for keeping the models used as simple as possible, but several factors points towards increased use of more complicated models. These factors include requirements for greater performance and flexibility (implying that the models must be valid over larger ranges and include dynamics), applications involving effects at different scales of magnitude (from molecule to reactor), etc.
Whereas effective tools exist for deriving linear models from experiments, there is a great need for tools to aid in rigorous models (based on physics and chemistry), fitting parameters to complex models, experiment design and model validation for nonlinear models, model reduction for complex nonlinear models to increase computational speed, etc.
Many real life systems are of a distributed nature, and are most naturally modelled using partial differential equations (PDEs). Modelling tools and technology is therefore needed both for systems described by ordinary differential equations (ODEs), and for systems described by PDEs.
SimulationMathematical simulation is heavily used in advanced plant design and operations (including decisions on when and where to drill new wells, etc.). Methods for numerical integration may be said to be well developed for models of small to moderate size, although model reduction is required in some applications to obtain required computational speed. However, reservoir simulations and simulation of large processing plants both involve systems of large scale, for which efficient and accurate simulation requires effective exploitation of model structure. Large-scale system simulation is likely to remain an active research area, and requires understanding of both mathematics and the type of system in question.
VisualisationLarge-scale system simulations, as encountered in both reservoir simulations and process plant simulations, produce massive amounts of data. The ability to interpret and draw correct conclusions from such data rests critically on the ability to visualise simulation results and navigate in ‘simulated space and time’. VR techniques have already found extensive application in reservoir simulations as an aid in visualisation. VR techniques and HMI design is also important in process simulators.
Navigating, filtering, and interpreting data Large amounts of data are found not only in large scale simulations, but also in operational data for large scale systems. Although there is a great potential in using such data effectively for model fitting and validation, using historical data to monitor and diagnose present operation, etc., there are also many difficulties in doing so, including:
This area will therefore remain a source of relevant research topics for years to come.
OptimizationOptimization is heavily used both in investment decisions, supply chain management, process design, planning and control. Optimization is therefore a key enabling technology in a wide range of application areas. The types of optimisation problems applied range from LP problems (with a linear optimisation criterion, linear constraints and continuous decision variables), to MINLP problems (with non-linear optimisation criterion, nonlinear constraints, and both discrete- and continuous-valued decision variables). The ability to find fast and reliable solutions to limits the applicability of many optimisation problems, in particular in real-time applications. The effects of uncertainties in the optimisation problems are also poorly understood in many cases.
ControlControl is an area which makes active use of several of the other areas listed above, in particular modelling, simulation and optimisation. What distinguishes control from merely being a combination of other technology areas is the ability to fundamentally change system behaviour by the use of feedback. Using feedback control, an open loop unstable system can be made stable.
Understanding and recognising this fact has been central in many recent advances in areas of relevance to gas technology, including feedback stabilization of
The above examples illustrate the importance of continued applied control research.
On the theoretical side, the problem of designing a controller for a linear plant with specified inputs and outputs is well understood. On the other hand, it was realised some thirty to forty years ago that to use a single, centralised controller for a large-scale system is in general a bad idea. Nevertheless, the problem of designing control structures is still only partially understood.
Most real-life control problems are nonlinear and include constraints in controlled and manipulated variables. This is a very large research area with a number of relevant research topics.
The most widely applied control technology for problems involving constraints is what is called Model Predictive Control (MPC), which solves an optimisation problem on-line for each sample interval. Until recently, the ability to solve such optimisation problems in the time available has been a severe constraint to the range of problems for which MPC can be applied. Recent advances have dramatically reduced the need for on-line computations, at the expense of requiring heavy computations at the controller design stage. Ongoing research aims at extending these techniques to more complex nonlinear systems, addressing robustness to model uncertainty, hybrid problems (involving both continuous and discrete-valued manipulated variables), etc.
MEMS technologyThe emergence of Micro-Electro-Mechanical Systems technology presents a number of opportunities and challenges in production, processing and transportation of natural gas.
The aviation industri is experimenting with ‘active surfaces’ based on MEMS technology, which minimises aerodynamical drag on aircraft. One might envisage similar technology being applied to impeller blades in compressors or to the internal surfaces of transmission pipelines.
Furthermore, there is an abundance of problems where distributed sensing would be very helpful in process operations, in particular if the sensing elements can be made so small that they do not significantly influence process operation. Examples might be measurement of flow distribution in distillation columns or reactors, measuring temperature distribution in reactors, etc.
Members of the working group: Professor Morten Hovd, Department of Engineering Cybernetics |