TU Darmstadt / ULB / TUbiblio

Stress and fracture prediction using geomechanical reservoir models - A case study from the Lower Magdalena Valley Basin, Colombia

Guerra, Clairet (2019)
Stress and fracture prediction using geomechanical reservoir models - A case study from the Lower Magdalena Valley Basin, Colombia.
Technische Universität Darmstadt
Dissertation, Erstveröffentlichung

Kurzbeschreibung (Abstract)

Proper characterisation of the mechanical and stress state of a hydrocarbon or geothermal reservoirs is crucial for optimal exploration and exploitation. The increasing complexity of discoveries push the boundaries of conventional methodologies, which often rely on local laboratory measurements or well-derived properties that may not represent the actual spatial distribution of relevant characteristics of the area. The early and accurate knowledge of pore pressure, mechanical properties and in-situ stress represent critical information to promote reservoir exploration, well placement and enhanced recovery techniques as well as avoidance of risky locations. Numerical modelling approaches serve as the solving tool as they can honour the heterogeneous and geometrical complexity of reservoir and associated structures. This study evaluates the potential and pertinence of geomechanical modelling techniques for stress and fracture prediction. The methodologies performed in this research are applied to a tight sand gas reservoir located in the Lower Magdalena Valley Basin in northern Colombia. The target reservoir is an Oligocene to Miocene sequence known as Porquero formation which is composed of low-permeability sandstones and shale layers. The area has been explored for more than 30 years but, due to its difficulty to be adequately characterised, has only become of economic interest in the last decade, thanks to the implementation of seismic methodologies and advanced petrophysical studies. The exploitation of the Porquero formation will require non-vertical drilling paths and hydraulic fracturing for economic production. Therefore, an accurate understanding of reservoir mechanics and stresses is of critical importance. In this study, different methodologies are proposed to study the general geomechanical behaviour of the stratigraphic column covered in the study area, as well as the inference of the natural fracturing behaviour along the reservoirs conforming the Porquero formation. Each methodology was developed around the data available, highlighting the integration between basin and petroleum systems modelling technique and geomechanical analysis as a pivotal approach to determine the mechanical state of a reservoir. The first modelling phase presents the performance of a complete geomechanical assessment using log-derived properties. This section includes the analysis of log and core data, as well as hydraulic fracturing tests leading to detailed 1D mechanical earth models (MEM) for each of the wells available in the study area. Subsequently, a 3D mechanical earth model was set up using two different property population methods that were tested and compared in terms of stresses. One method used a geostatistical approach based on well data mechanical models for property inter-/extrapolation, whereas the other, additionally used seismic inversion techniques to account for the vertical and lateral differences in mechanical rock properties. The combination of these methodologies serves as a base model of the present-day stress state of the area and is set to be the reference for potential predictive simulations. The second modelling phase introduces basin and petroleum systems modelling (BPSM) technology for geomechanical purposes. The method takes advantage of a robust model construction intended to assess hydrocarbon generation, migration and accumulation. It ensures that the temporal basin evolution and spatial properties variations are consistent and assumed under adequate reasoning. Pore pressure model was derived from porosity-dependent compaction laws that answer to subsidence and sedimentary registers of the basin. In-situ stresses were estimated through a poro-elastic approach. The definition of process-based mechanical properties, physics-driven pore pressure models and therefore enhanced resulting in-situ stresses, is presented as an alternative methodology to conventional modelling approaches, to perform a pre-drilling geomechanical assessment. The modelling results provide a spatial mechanical characterisation, pore pressure models and the complete stress tensor, not only for the reservoir but for each point in the model domain. The simulation shows that the ruling stress regime in the study area is a normal faulting regime with a governing orientation of SHmax in a WNW-ESE direction. At reservoir depth, vertical stress gradient (Sv) has a mean value of 23.29 MPa/km, and SHmax is on average 1.2*Shmin. The final stage of the modelling work includes the study of the potential of the developed geomechanical models to infer natural fracture networks. Such fracture models are crucial in enhancing field development and promoting efficient well placement to maximise hydrocarbon production. Two methods were tested. The first method corresponds to a stochastic simulation enhanced by the definition of geomechanical constraints and well-log derived fracture metrics. Paleo-stress selection is performed through a stress-inversion approach. The second model showcases the potential of forward modelling of BPSM-derived properties. The critical difference, in this case, is that the dynamic setup allows the step-wise definition of maximum stress orientation correspondent to defined geological stages. The overall analysis is made in terms of fracture orientation metrics (Dip angle and dip azimuth), fracture intensity and the potential of the models to reproduce observed data. The stochastically generated fracture network is based in a boundary element method, which is beneficial because of the computational speed. However, a concept of this method is the definition of homogeneous mechanical properties which may lead to oversimplified results when dealing with complex lithological distributions. On the other hand, the forward model approach, using the BPSM technology, displays great pertinence in gathering most of the evolutionary traits in the area. Moreover, the recognition of heterogeneous lithological distribution represents an enhancement and consequently leads to more reliable results. As a disadvantage, the construction endeavor of such models is of high complexity, as well as demanding in the necessity of the integration and cooperation of several disciplines to achieve a consistent result. Considering the mentioned generalities, the stochastic approach is advised for low-deformation regions, while the forward model approach may display its maximum capacity in areas with strong deformation. The significant contributions of this research are fully customized geomechanical reservoir models that can reproduce the reported mechanical behavior of the area but, moreover, are able to provide critical insights in the inter-well and undrilled regions of the model domain. The models are populated with actual field and laboratory data and are set to be the fundamental scenarios in any predictive simulation. The methods presented in this work are replicable and deployable in any other geographical and tectonic setting.

Typ des Eintrags: Dissertation
Erschienen: 2019
Autor(en): Guerra, Clairet
Art des Eintrags: Erstveröffentlichung
Titel: Stress and fracture prediction using geomechanical reservoir models - A case study from the Lower Magdalena Valley Basin, Colombia
Sprache: Englisch
Referenten: Henk, Prof. Dr. Andreas ; Schill, Prof. Dr. Eva
Publikationsjahr: 28 August 2019
Ort: Darmstadt
Datum der mündlichen Prüfung: 10 Juli 2019
URL / URN: https://tuprints.ulb.tu-darmstadt.de/9029
Kurzbeschreibung (Abstract):

Proper characterisation of the mechanical and stress state of a hydrocarbon or geothermal reservoirs is crucial for optimal exploration and exploitation. The increasing complexity of discoveries push the boundaries of conventional methodologies, which often rely on local laboratory measurements or well-derived properties that may not represent the actual spatial distribution of relevant characteristics of the area. The early and accurate knowledge of pore pressure, mechanical properties and in-situ stress represent critical information to promote reservoir exploration, well placement and enhanced recovery techniques as well as avoidance of risky locations. Numerical modelling approaches serve as the solving tool as they can honour the heterogeneous and geometrical complexity of reservoir and associated structures. This study evaluates the potential and pertinence of geomechanical modelling techniques for stress and fracture prediction. The methodologies performed in this research are applied to a tight sand gas reservoir located in the Lower Magdalena Valley Basin in northern Colombia. The target reservoir is an Oligocene to Miocene sequence known as Porquero formation which is composed of low-permeability sandstones and shale layers. The area has been explored for more than 30 years but, due to its difficulty to be adequately characterised, has only become of economic interest in the last decade, thanks to the implementation of seismic methodologies and advanced petrophysical studies. The exploitation of the Porquero formation will require non-vertical drilling paths and hydraulic fracturing for economic production. Therefore, an accurate understanding of reservoir mechanics and stresses is of critical importance. In this study, different methodologies are proposed to study the general geomechanical behaviour of the stratigraphic column covered in the study area, as well as the inference of the natural fracturing behaviour along the reservoirs conforming the Porquero formation. Each methodology was developed around the data available, highlighting the integration between basin and petroleum systems modelling technique and geomechanical analysis as a pivotal approach to determine the mechanical state of a reservoir. The first modelling phase presents the performance of a complete geomechanical assessment using log-derived properties. This section includes the analysis of log and core data, as well as hydraulic fracturing tests leading to detailed 1D mechanical earth models (MEM) for each of the wells available in the study area. Subsequently, a 3D mechanical earth model was set up using two different property population methods that were tested and compared in terms of stresses. One method used a geostatistical approach based on well data mechanical models for property inter-/extrapolation, whereas the other, additionally used seismic inversion techniques to account for the vertical and lateral differences in mechanical rock properties. The combination of these methodologies serves as a base model of the present-day stress state of the area and is set to be the reference for potential predictive simulations. The second modelling phase introduces basin and petroleum systems modelling (BPSM) technology for geomechanical purposes. The method takes advantage of a robust model construction intended to assess hydrocarbon generation, migration and accumulation. It ensures that the temporal basin evolution and spatial properties variations are consistent and assumed under adequate reasoning. Pore pressure model was derived from porosity-dependent compaction laws that answer to subsidence and sedimentary registers of the basin. In-situ stresses were estimated through a poro-elastic approach. The definition of process-based mechanical properties, physics-driven pore pressure models and therefore enhanced resulting in-situ stresses, is presented as an alternative methodology to conventional modelling approaches, to perform a pre-drilling geomechanical assessment. The modelling results provide a spatial mechanical characterisation, pore pressure models and the complete stress tensor, not only for the reservoir but for each point in the model domain. The simulation shows that the ruling stress regime in the study area is a normal faulting regime with a governing orientation of SHmax in a WNW-ESE direction. At reservoir depth, vertical stress gradient (Sv) has a mean value of 23.29 MPa/km, and SHmax is on average 1.2*Shmin. The final stage of the modelling work includes the study of the potential of the developed geomechanical models to infer natural fracture networks. Such fracture models are crucial in enhancing field development and promoting efficient well placement to maximise hydrocarbon production. Two methods were tested. The first method corresponds to a stochastic simulation enhanced by the definition of geomechanical constraints and well-log derived fracture metrics. Paleo-stress selection is performed through a stress-inversion approach. The second model showcases the potential of forward modelling of BPSM-derived properties. The critical difference, in this case, is that the dynamic setup allows the step-wise definition of maximum stress orientation correspondent to defined geological stages. The overall analysis is made in terms of fracture orientation metrics (Dip angle and dip azimuth), fracture intensity and the potential of the models to reproduce observed data. The stochastically generated fracture network is based in a boundary element method, which is beneficial because of the computational speed. However, a concept of this method is the definition of homogeneous mechanical properties which may lead to oversimplified results when dealing with complex lithological distributions. On the other hand, the forward model approach, using the BPSM technology, displays great pertinence in gathering most of the evolutionary traits in the area. Moreover, the recognition of heterogeneous lithological distribution represents an enhancement and consequently leads to more reliable results. As a disadvantage, the construction endeavor of such models is of high complexity, as well as demanding in the necessity of the integration and cooperation of several disciplines to achieve a consistent result. Considering the mentioned generalities, the stochastic approach is advised for low-deformation regions, while the forward model approach may display its maximum capacity in areas with strong deformation. The significant contributions of this research are fully customized geomechanical reservoir models that can reproduce the reported mechanical behavior of the area but, moreover, are able to provide critical insights in the inter-well and undrilled regions of the model domain. The models are populated with actual field and laboratory data and are set to be the fundamental scenarios in any predictive simulation. The methods presented in this work are replicable and deployable in any other geographical and tectonic setting.
Alternatives oder übersetztes Abstract:
Alternatives AbstractSprache

Die richtige Charakterisierung des mechanischen Spannungszustandes von Kohlenwasserstoff- oder geothermischen Speichern ist entscheidend, um deren Untersuchung und Abbau optimal zu gestalten. Die zunehmende Komplexität der Funde, lassen herkömmliche Methoden an ihre Grenzen stoßen, die sich oft auf lokale Labormessungen oder von Bohrlöchern abgeleitete Eigenschaften stützen, die möglicherweise nicht die tatsächliche räumliche Verteilung relevanter Merkmale des Gebietes darstellen. Die frühzeitige und genaue Kenntnis von Porendruck, mechanischen Eigenschaften und In-situ-Spannungen ist entscheidend, um die Erkundung von Lagerstätten, die Platzierung von Bohrlöchern und verbesserte Gewinnungstechniken sowie die Vermeidung von riskanten Standorten zu fördern. Numerische Modellierungsansätze dienen als Lösungsinstrument, da sie die heterogene und geometrische Komplexität von Reservoirs und zugehörigen Strukturen berücksichtigen können. Diese Studie bewertet das Potenzial und die Relevanz geomechanischer Modellierungstechniken für die Vorhersage von Spannungen und Brüchen. Die in dieser Forschungsarbeit genutzten Methoden werden auf ein dichtes Sandstein-Gas-Reservoir im unteren Magdalena-Talbecken im Norden Kolumbiens angewendet. Das Zielreservoir ist eine Oligozän-Miozän-Sequenz, die als Porquero-Formation bekannt ist und aus schwach durchlässigen Sandsteinen und Schieferschichten besteht. Das Gebiet wird seit mehr als 30 Jahren erforscht, ist aber aufgrund von Schwierigkeiten, angemessen charakterisiert zu werden, erst in den letzten zehn Jahren dank der Anwendung fortschrittlicher gesteinsmechanischer und seismischer Methoden von wirtschaftlichem Interesse geworden. Die Nutzung der Porquero-Formation erfordert nicht-vertikale Bohrpfade und hydraulisches Aufbrechen für eine wirtschaftliche Produktion. Daher ist ein genaues Verständnis der Reservoirmechanik und spannungen von entscheidender Bedeutung. Die erste Modellierungsphase stellt die Durchführung einer vollständigen geomechanischen Bewertung unter Verwendung von abgeleiteten Eigenschaften aus Bohrlochmessungen (Logs) dar. Dieser Abschnitt beinhaltet die Analyse von Log und Probendaten, sowie hydraulische Frakturierungstests, die zu detaillierten mechanischen 1D-Erdmodellen (1D Mechanical Earth Models, MEM) für jede der verfügbaren Bohrungen im Untersuchungsgebiet führen. Anschließend wurde ein mechanisches 3D-Erdmodell (3D MEM) mit zwei verschiedenen Eigenschaftspopulationsmethoden erstellt, die getestet und in Bezug auf Spannungen verglichen wurden. Eine Methode verwendete einen geostatistischen Ansatz, der auf mechanischen Modellen aus Bohrlochmessdaten für die Inter-/Extrapolation von Eigenschaften basiert, während die andere zusätzlich seismische Inversionstechniken einsetzte, um die vertikalen und lateralen Unterschiede in den mechanischen Gesteinseigenschaften zu berücksichtigen. Die Kombination dieser Methoden dient als Basismodell für den gegenwärtigen Stand des Gebietes und soll als Referenz für potenzielle prädiktive Simulationen dienen. In der zweiten Modellierungsphase wird die für geomechanische Zwecke genutzte Modellierungstechnik, BDSM (Basin and petroleum systems modelling) eingeführt. Das Verfahren nutzt die Vorteile einer robusten Modellkonstruktion zur Beurteilung der Kohlenwasserstoffbildung, migration und akkumulation. Es stellt sicher, dass die zeitliche Entwicklung des Beckens und die Veränderungen der räumlichen Eigenschaften konsistent sind und unter angemessener Argumentation angenommen werden. Das Porendruckmodell wurde aus einem porositätsabhängigen Kompaktionsgesetz abgeleitet, das auf Bodensenkungs- und Sedimentregister des Beckens reagiert. Die In-situ-Spannungen wurden durch einen poroelastischen Ansatz geschätzt. Die Definition prozessbasierter mechanischer Eigenschaften, physikalisch-basierte (physics-driven) Poren-druckmodelle und damit erhöhter resultierender in-situ Spannungen wird als alternative Methodik zu herkömmlichen Modellierungsansätzen vorgestellt, um eine geomechanische Vorbohrbeurteilung durchzuführen. Die wichtigsten Modellierungsergebnisse liefern eine räumliche mechanische Charakterisierung, Porendruckmodelle und den kompletten Spannungstensor, nicht nur für das Reservoir, sondern für jeden Punkt in der Modelldomäne. Die Simulation zeigt, dass das herrschende Spannungsregime im Untersuchungsgebiet eine normale Verwerfung mit einer vorherrschenden Ausrichtung von SHmax in einer WNW-ESE-Richtung ist. In der Speichertiefe hat der vertikale Spannungsgradient (Sv) einen Mittelwert von 23,29 MPa/km und SHmax beträgt durchschnittlich 1,2*Shmin. Die letzte Phase der Modellierungsarbeiten beinhaltet die Untersuchung des Potenzials der entwickelten geomechanischen Modelle zur Ableitung natürlicher Bruchnetze. Solche Bruchmodelle sind entscheidend für die Verbesserung der Feldentwicklung und die Förderung einer effizienten Bohrlochplatzierung zur Maximierung der Kohlenwasserstoffproduktion. Zwei Methoden wurden getestet. Die erste Methode entspricht einer stochastischen Simulation, die durch die Definition von geomechanischen Randbedingungen und Bruchmetriken aus Bohrlochmessungen (Logs) ergänzt wird. Die Auswahl der Paläospannung erfolgt über einen Ansatz der Spannungsinversion. Das zweite Modell zeigt das Potenzial der Vorwärtsmodellierung oder der von BPSM abgeleiteten Eigenschaften. Der entscheidende Unterschied besteht in diesem Fall darin, dass der dynamische Aufbau die schrittweise Definition der maximalen Spannungsorientierung entsprechend den definierten geologischen Phasen ermöglicht. Die Gesamtanalyse erfolgt in Bezug auf die Kennzahlen zur Bruchorientierung (Neigungswinkel und Azimut), die Bruchintensität und das Potenzial der Modelle zur Reproduktion der beobachteten Daten. Die wesentlichen Beiträge dieser Forschungsarbeit sind vollständig angepasste geomechanische Reservoir-Modelle, die in der Lage sind, das berichtete mechanische Verhalten des Gebietes zu reproduzieren, aber darüber hinaus auch entscheidende Einblicke in die ungebohrten Regionen und Regionen zwischen Bohrlöchern der Modelldomäne liefern können. Die Modelle sind mit aktuellen Feld und Labordaten gefüllt und stellen zu Grunde liegende Szenarien in jeder prädiktiven Simulation. Die in dieser Arbeit vorgestellten Methoden sind reproduzierbar und in jedem anderen geografischen Umfeld einsetzbar.

Deutsch
URN: urn:nbn:de:tuda-tuprints-90296
Sachgruppe der Dewey Dezimalklassifikatin (DDC): 500 Naturwissenschaften und Mathematik > 550 Geowissenschaften
Fachbereich(e)/-gebiet(e): 11 Fachbereich Material- und Geowissenschaften
11 Fachbereich Material- und Geowissenschaften > Geowissenschaften
11 Fachbereich Material- und Geowissenschaften > Geowissenschaften > Fachgebiet Ingenieurgeologie
Hinterlegungsdatum: 29 Sep 2019 19:55
Letzte Änderung: 29 Sep 2019 19:55
PPN:
Referenten: Henk, Prof. Dr. Andreas ; Schill, Prof. Dr. Eva
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: 10 Juli 2019
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen
OAI 2.0-Basis-URL: https://tubiblio.ulb.tu-darmstadt.de/cgi/oai2 TUbiblio verwendet EPrints 3.
Drucken | Impressum | Datenschutzerklärung