CHEMISTRY AS A COMPASS: MULTIVARIATE EXPLORATION OF WELL CONNECTIVITY IN RESERVOIR SYSTEMS

- 328727
Resumo expandido (Oral)
Favoritar este trabalho
Como citar esse trabalho?
Resumo

Introduction

Assessing reservoir connectivity is a fundamental challenge in reservoir management, directly impacting decisions on well placement, production strategy, and resource estimation (Asemani et al., 2021). Traditional approaches to evaluate compartmentalization often provide limited resolution or require long-term operational data. In contrast, geochemical fingerprinting offers a powerful complementary approach by using compositional differences among fluids to infer reservoir communication.

In this context, multivariate statistical tools can provide objective and data-driven means to interpret complex chemical information. Among them, principal component analysis (PCA) stands out as a robust exploratory technique capable of summarizing high-dimensional geochemical data and highlighting underlying patterns. By projecting samples into a reduced-dimensional space, PCA can reveal groupings or trends that suggest whether fluids from different wells are compositionally similar or distinct information that may reflect physical communication or isolation between reservoir compartments.

The objective of this study was to demonstrate the use of PCA as a decision-support tool for evaluating connectivity between wells in a shared reservoir. Using high-resolution compositional data obtained from ultra-high resolution mass spectrometry, we explored whether differences in molecular composition could be used to infer communication or isolation between sampled wells. This approach provides an interpretable and reproducible framework that supports reservoir characterization efforts and assists in the integration of geochemical and engineering perspectives.

 

Experimental

Nine crude oil samples were collected from three different wells belonging from the same reservoir. These wells had previously established connectivity status based on conventional analysis, operational and geological data, serving as a reference for validating the proposed chemometric approach. The samples were selected to represent both connected and compartmentalized regions within the reservoir.

Each sample was analyzed using two ionization modes: electrospray ionization in negative mode (ESI (–)) and atmospheric pressure photoionization in positive mode (APPI (+)). Mass spectrometric analyses were carried out using a SolariX 7T 2xR Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR MS, Bruker Daltonics, Bremen, Germany). The mass spectra were acquired over an m/z range of 140 to 1200 Da, with 300 scans accumulated per sample to ensure analytical consistency.

Raw data were recalibrated using DataAnalysis® 5.0 SR1, and molecular formula assignments were performed using Composer® 1.5.3 (Sierra Analytics, USA). The resulting composition tables, containing monoisotopic abundances and molecular formulas, were exported to MATLAB R2024a (The MathWorks, Natick, USA). Separate data matrices were created for ESI (–) and APPI (+), and each was submitted to PCA. The analysis aimed to explore patterns in the molecular composition of the oils, with a focus on identifying whether the samples clustered according to their known connectivity relationships.

Following the successful validation with the reference wells, the methodology was applied to a second set of samples (11 samples) from three additional wells (D, E, and F), for which reservoir connectivity was still under evaluation. The goal was to test whether PCA could reveal grouping patterns that support or reject a proposed connection among these new wells based on geochemical similarity.

 

Results and Discussion

PCA revealed distinct clustering patterns that were consistent with the known connectivity between wells. In both ESI (–) and APPI (+) ionization modes, wells A and B were clearly separated from well C along PC2, indicating consistent geochemical distinction and supporting the known compartmentalization of well C. This separation was robust across ionization techniques and confirms the ability of PCA to reveal connectivity-related compositional patterns (Figure 1).

Figure 1. PCA score plots of crude oil samples from wells A, B and C analyzed by ESI (–) (left) and APPI (+) (right) FT-ICR MS.

An additional pattern emerged in the APPI (+) data. Wells A and B, although connected based on geological reference methods, appeared more separated along PC1, unlike in the ESI (–) data, where they clustered closely. This divergence is likely due to the nature of the ionization mechanisms. APPI (+) preferentially ionizes less polar, more apolar compounds, which may vary between connected wells due to local geochemical processes or migration effects, while ESI (–) focuses on more polar, heteroatom-rich species that often reflect more conservative features of oil composition.

Importantly, the separation between A and B in APPI (+) should not be interpreted as lack of connectivity, but rather as a difference in the molecular composition of apolar fractions. This highlights the importance of combining multiple ionization techniques to capture a fuller molecular profile and to avoid misinterpretation of chemometric results. The consistent separation of well C from A and B across both modes strengthens the case for PCA as a valid tool to assess reservoir connectivity.

Following the validation using wells A, B, and C, the approach was extended to evaluate a second group of wells (D, E, and F), for which reservoir connectivity was hypothesized based on geological interpretation. As shown in Figure 2, the PCA score plots obtained from both ESI (–) and APPI (+) analyses show that the three wells are not separated along PC2, which is the principal axis used in this methodology to assess connectivity.

In both ionization modes, samples from wells D, E, and F remain distributed along PC1 but maintain close proximity along PC2, suggesting that they share a geochemically similar profile in the dimensions most associated with compartmentalization. This supports the hypothesis that these wells are part of a connected reservoir system, despite some variability captured along PC1, likely reflecting subtle compositional gradients unrelated to physical disconnection.

Figure 2. PCA score plots of crude oil samples from wells D, E, and F analyzed by ESI (–) (left) and APPI (+) (right) FT-ICR MS.

These findings reinforce the utility of PCA for assessing reservoir connectivity: in the first case, the method correctly distinguished between connected and disconnected wells, while in the second case, it confirmed the geochemical similarity among wells believed to be connected. The complementary nature of the ESI (–) and APPI (+) data further strengthens the interpretation, as different molecular fractions contribute to the consistent clustering. PCA thus proves to be a valuable analytical and interpretative tool for reservoir geochemistry, offering rapid insight into fluid continuity across production wells.

 

Conclusions

This study demonstrates the effectiveness of PCA combined with FT-ICR MS as a reliable tool for assessing reservoir connectivity. In the validation set, PCA successfully distinguished between geochemically connected and disconnected wells, with complementary ionization modes (ESI (–) and APPI (+)) offering a broader compositional view. Notably, the separation direction in PC2 was inverted between the two modes, reflecting differences in ionization behavior highlighting the importance of using orthogonal analytical techniques. In the second set of wells under evaluation, PCA results from both ionization modes consistently showed tight clustering among all three wells, supporting the hypothesis of full geochemical connectivity. The agreement across ionization modes strengthens confidence in this chemometric approach as a decision-support tool for reservoir characterization. By revealing compositional similarity and compartmentalization patterns at the molecular level, PCA applied to petroleomic data offers a fast, interpretable, and non-invasive methodology to complement traditional reservoir engineering techniques and inform development strategies.

 

Acknowledgements

The authors thank Petrobras for financial support.

 

Reference

Asemani, M., Rabbani, A.R., & Sarafdokht, H. (2021). Implementation of an integrated geochemical approach using polar and nonpolar components of crude oil for reservoir-continuity assessment: Verification with reservoir-engineering evidences. SPE J., 26(5), 3237–3254.

 

Compartilhe suas ideias ou dúvidas com os autores!

Sabia que o maior estímulo no desenvolvimento científico e cultural é a curiosidade? Deixe seus questionamentos ou sugestões para o autor!

Faça login para interagir

Tem uma dúvida ou sugestão? Compartilhe seu feedback com os autores!

Instituições
  • 1 Universidade Federal de Goiás
  • 2 CENPES - PETROBRAS
Eixo Temático
  • ST-04 - Geoquímica do Petróleo e Novas Tecnologias para Remediação de Impactos Ambientais
Palavras-chave
Reservoir geochemistry
PCA
FT-ICR MS
Multivariate analysis
Compartmentalization