Abstract
Pixel-based multiple-point statistical (MPS) modelling is an appealing geostatistical modelling technique as it easily honours well data and allows use of geologically-derived training images to reproduce the desired heterogeneity. A variety of different training image types are often proposed for use in MPS modelling, including object-based, surface-based and process-based models. The purpose of the training image is to provide a description of the geological heterogeneities including sand geometries, stacking patterns, facies distributions, depositional architecture and connectivity. It is, however, well known that pixel-based MPS modelling has difficulty reproducing facies connectivity, and this study investigates the performance of a widely-available industrial SNESIM algorithm at reproducing the connectivity in a geometrically-representative, idealized deep-water reservoir sequence, using different gridding strategies and training images. The findings indicate that irrespective of the sand connectivity represented in the training image, the MPS models have a percolation threshold that is the same as the well-established 27% percolation threshold of random object-based models. A more successful approach for generating poorly connected pixel-based MPS models at high net:gross ratios has been identified. In this workflow, a geometrical transformation is applied to the training image prior to modelling, and the inverse transformation is applied to the resultant MPS model. The transformation is controlled by a compression factor which defines how non-random the geological system is, in terms of its connectivity.
You have full access to this open access chapter, Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
Many deep-water lobe depositional sequences are characterized by laterally extensive, but finite, sand beds interbedded with continuous low permeability shales. Non-random geological processes such as compensational stacking result in these systems often having poor sand connectivity at high net:gross ratios (NTG). It is challenging to reproduce this low connectivity in reservoir models, but it is important to do so if the models are to be used to predict reservoir performance.
The advantage of using a pixel-based method to model these sequences is that it reproduces geological patterns while honouring conditioning well data. The connectivity of a number of training image (TI) inputs including object- and surface-based models (OBM and SBM respectively) are investigated in this study and compared to the connectivity of output MPS models built using the SNESIM algorithm [1]. The findings from these models, along with the connectivity characteristics of OBM models, aided in the development of workflow for generating poorly connected MPS models using a simple object-based TI followed by a grid transformation using the compression algorithm.
2 Pixel-Based MPS Modelling with Common Training Images
The dimension of the model is 45 km × 45 km and 20 m thick. The target system to be reproduced contains sand beds which are about 4.5 km wide and 0.5 m at their thickest, NTG equal to about 40% and amalgamation ratio (AR) equal to about 10%. This AR implies that 10% of the total area of sand bed bases should be connected to an underlying sand bed. AR cannot be measured in an MPS model, so the connectivity measure used to compare models is the proportion of total sand that is connected to a vertical well at the centre of the model, which is about 5% in this case (Fig. 1).
An unconditioned OBM is easily generated and works by placing user defined objects within the model volume until the target NTG is reached. A downside of OBM modelling is that are no under-defined constraints on the sand connectivity, and it is clear in Fig. 1a that the object-based TI does not have the target low connectivity since about 95% of the sand beds are connected to the well, either directly or via their connections with different beds. Therefore, although the MPS model has reproduced the connectivity of the TI, it does not have the low connectivity of the target system. The second TI is a SBM generated using a number of depositional rules. These rules control the placement of sand elements and shale interelements in a stratigraphic order, and the probability that an element is capped by an interelement controls the degree of sand amalgamation. An important feature of the TI that impacts the MPS modelling is that the TI contains an irregular grid structure since each sand or shale element is contained within an individual layer which is one grid cell thick. Hence, although the TI represents the target system (Fig. 1b), the MPS algorithm produces very different model characteristics consisting of thin sand and shale layers resulting in complete sand connectivity. The final TI is a SBM constructed in a regular grid (Fig. 1c). This TI is a simple model generated with the same object-dimensions and 90% probability of shale drapes and therefore has low connectivity representative of the target model. However, again the MPS algorithm is unable to reproduce the low connectivity represented in the TI, and all of the sand is connected.
These results indicate that a simple workflow of constructing an accurate TI and applying this directly in MPS modelling is inappropriate for geological systems with non-random connectivity, such as the deep-water deposits considered. The only MPS model which reproduced the connectivity of the TI was the object-based case. The connectivity of MPS models generated by both OBM and SBM models are investigated in much greater detail [2] and it was found that these models follow the well-known percolation threshold seen in OBM whereby models with sand proportion greater than about 27% are well connected.
3 Pixel-Based Modelling with Low Connectivity
There are many modelling methods which allow the user to have control over the output connectivity including surface- or rule-based modelling, process-based and object-based methods. One such method, compression-based modelling, was developed originally in conjunction with OBM [3] and is based on the observation that the NTG and AR are equal in these models if all sand elements are of equal thickness. The compression method is a two-step process where modelling is undertaken with an initial low NTG (equal to the target AR) and cells containing sand and shale are expanded and compressed using the compression algorithm to reach the target NTG but preserved the low connectivity.
Combining the compression method with the MPS workflow requires that a low NTG OBM model is used as the TI input and, for the case considered here, it must equal the target low AR of 10% (Fig. 2a). The resulting MPS model reproduces the low connectivity of the TI (Fig. 2b). Then the compression algorithm is applied to the MPS model to rescale its NTG to the target value of 40% (Fig. 2c).
The properties of the training image, such as the thickness of the objects, are defined by the compression algorithm. This means that the decompression algorithm [4] is applied to the conditioning data prior to modelling. This is the inverse of the compression algorithm and ensures that the conditioning well data will be honoured when the compression algorithm is applied to the final model (Fig. 2d).
4 Summary
Several commonly used TIs have been investigated to examine both the suitability the TI to honour low connectivity at high NTG, and the ability of the MPS algorithm to reproduce the connectivity represented in the TI (Table 1). Although the object-based TI does not contain low connectivity at NTG values greater than 27%, MPS models based on these TIs reproduce accurately their connectivity. In contrast, the two surface-based TIs have the target low connectivity, but this connectivity is not reproduced in the MPS models. A detailed investigation indicates that MPS models follow the same relationship between NTG and connectivity as object-based models when built using surface- or object-based TIs [2]. This equivalence provides the basis for a new workflow in which MPS models are built using OBM TIs, and then rescaled to the target NTG using the compression algorithm. The conditioning well data is incorporated by using the inverse transformation prior to the MPS modelling step. The final MPS model honours the well data and contains independently user defined NTG and connectivity.
References
Strebelle, S.: Conditional simulation of complex geological structures using multiple-point statistics. Math. Geol. 34, 1–21 (2002)
Walsh, D.A., Manzocchi, T.: Connectivity in pixel-based facies models. Math. Geosci. 53, 415–435 (2021)
Manzocchi, T., Zhang, L., Haughton, P.D.W., Pontén. A.: Hierarchical parameterization and compression-based object modelling of high net:gross but poorly amalgamated deep-water lobe deposits. Pet. Geosci. 26, 545–567 (2020)
Walsh, D.A., Manzocchi, T.: A method for generating geomodels conditioned to well data with high net:gross ratios but low connectivity. Marine Pet. Geol. 129 (2021)
Acknowledgements
This presentation has emanated from research supported in part by a research grant from Science Foundation Ireland (SFI) under Grant Number 13/RC/2092_P2 and is co-funded by PIPCO RSG and its member companies. Schlumberger are thanked for the provision of an academic Petrel license.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Copyright information
© 2023 The Author(s)
About this paper
Cite this paper
Walsh, D.A., López-Cabrera, J., Manzocchi, T. (2023). The Suitability of Different Training Images for Producing Low Connectivity, High Net:Gross Pixel-Based MPS Models. In: Avalos Sotomayor, S.A., Ortiz, J.M., Srivastava, R.M. (eds) Geostatistics Toronto 2021. GEOSTATS 2021. Springer Proceedings in Earth and Environmental Sciences. Springer, Cham. https://doi.org/10.1007/978-3-031-19845-8_10
Download citation
DOI: https://doi.org/10.1007/978-3-031-19845-8_10
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-19844-1
Online ISBN: 978-3-031-19845-8
eBook Packages: Earth and Environmental ScienceEarth and Environmental Science (R0)