IEAGHG Modelling and Risk Management Combined Network

Publication Overview

IEAGHG’s combined Modelling and Risk Management Network, hosted by the EERC, took place in Grand Forks, North Dakota between 18th and 22nd June 2018. These meetings bring together leading experts from research and industry to discuss the latest work and developments, with over 30 speakers and 71 attendees representing 8 countries. The theme for the meeting was ‘How advances in modelling and risk management improve pressure management, capacity estimation, leakage detection and the prediction of induced seismicity’. Sessions included project updates, the application of oil and gas production experience, modelling capacity, unconventional reservoir risk assessments and active pressure management. The third day focused on conformance and regulation with a keynote presentation by Lynn Helms from the North Dakota Industrial Commission on Class VI well regulations.

Publication Summary

Key findings from the meeting included further refinement for a global capacity estimation methodology needs to be agreed. SPE SRMS provides a solid foundation although further refinement is necessary. Geological heterogeneity still remains very difficult to simulate. The development of models based on a diversity of analogues would improve model predictability. Capillary pressure simulations can help to explain the pattern of CO₂ plume migration better than porosity/permeability simulations alone. Fluid events can now be simulated at pore-scale.

Residual Oil Zones (ROZs) potentially have significant CO₂ storage capacity but field-specific studies are needed to improve predictions. CO₂ has been tested as a fracture/production fluid to extract oil in the Bakken Formation but conformance is a major challenge and further optimisation is required.

Evidence from the Norwegian sector shows ~40% of injector wells have integrity problems. Producer wells are less prone to integrity issues. A Neural-Genetic Algorithm could be a useful method for screening wells in depleted O&G fields as storage candidates. Statistical methods have been developed to rationalize the number of variables that affect wellbore integrity. The technique has been field tested on data from fields with background data on well integrity.

Good progress has been made with CarbonSAFE projects notably the initial stages on the storage potential of the Atlantic Shelf and across Michigan. Probabilistic Risk Maps have been developed which could aid project design.

A comparative study of known leaking and sealing faults in CO₂ traps on the Colorado Plateau has shown that where CO₂ pressure exceeds the fracture gradient then seepage occurs. The research has shown that these criteria apply equally to CO₂ traps and conventional oil and gas traps. Knowledge of reservoir stress changes and fracture closure pressures adds uncertainty to predicting fault and fracture-related leakage. Sophisticated numerical modelling tools are available for predicting fluid flow at faults but building models is difficult, material properties are not well known and few have been thoroughly calibrated.

Micro-seismicity detected at Quest is at very low levels and has no correlation with injection rates or pressures. Large events are extremely unlikely and none of the events represent a risk to containment.

Decatur induced seismicity events range from -2.13 to 1.17 and 95% are 0 or less. ~85% of events are in Precambrian basement.

Pipeline risk assessment needs to include transient multiphase flow models for CO₂. Fluid and thermodynamics within pipelines are tightly coupled. Large-scale experiments (with many effects) are difficult to use for model development. When designing CO₂-transport pipelines against running-ductile fracture, we need to take into account that CO₂ can exert large forces on the pipe flanks due to the phase change occurring during decompression.

Active pressure management at basin scale is likely to become an essential tool for carbon storage operations. A single study has shown central injection and pressure management using brine extraction, concentrated in the centre of a basin, may be easier to manage than onsite injections dispersed across the basin. A single study has shown that approximately the same volume of brine needs to be produced as the volume of CO₂ being injected. Consequently large volumes of brine would need to be treated or re-injected in other formations.

Smart reduction in model complexity can gain valuable efficiency without sacrificing accuracy. Integration of geological modelling and simple dynamic pressure simulation are necessary for understanding where to invest in detailed characterization and simulation.

Recommendations from the meeting included the development and application of more appropriate constitutive models such as anisotropic cap-plasticity models with creep; more research on the relationship between pressure increase and the effects elsewhere on fractures; and further modelling on hydrate formation in or near wellbores if water is present in CO₂.