Mesoscale Simulation and Machine Learning of Asphaltene Aggregation Phase Behavior and Molecular Assembly Landscapes

Abstract

Asphaltenes constitute the heaviest fraction of the aromatic group in crude oil. Aggregation and precipitation of asphaltenes during petroleum processing costs the petroleum industry billions of dollars each year due to downtime and production inefficiencies. Asphaltene aggregation proceeds via a hierarchical self-assembly process that is well-described by the Yen-Mullins model. Nevertheless, the microscopic details of the emergent cluster morphologies and their relative stability under different processing conditions remain poorly understood. We perform coarse-grained molecular dynamics simulations of a prototypical asphaltene molecule to establish a phase diagram mapping the self-assembled morphologies as a function of temperature, pressure, and n-heptane:toluene solvent ratio informing how to control asphaltene aggregation by regulating external processing conditions. We then combine our simulations with graph matching and nonlinear manifold learning to determine low-dimensional free energy surfaces governing asphaltene self-assembly. In doing so, we introduce a variant of diffusion maps designed to handle data sets with large local density variations, and report the first application of many-body diffusion maps to molecular self-assembly to recover a pseudo-1D free energy landscape. Increasing pressure only weakly affects the landscape, serving only to destabilize the largest aggregates. Increasing temperature and toluene solvent fraction stabilizes small cluster sizes and loose bonding arrangements. Although the underlying molecular mechanisms differ, the strikingly similar effect of these variables on the free energy landscape suggests that toluene acts upon asphaltene self-assembly as an effective temperature.

Asphaltene aggregation studied by molecular dynamics simulations: role of the molecular architecture and solvents on the supramolecular or colloidal behavior

“…This fact, also observed in the CN∕N(%) curves, indicates that these molecules are slightly soluble in n-heptane and with the increased concentration of this solvent, they interact more with the solvent and less with each other. For the other molecules, within the concentration limits studied, adding n-heptane to the toluene solution has no effect on the nanoaggregation formation and stabilization, in line with other works (Rogel 1995;Sedghi et al 2013;Wang et al 2017a;Headen et al 2017) In this way, helped by its miscibility with toluene, n-heptane can interact selectively with the asphaltene molecules in function of their lateral chain length. This fact indicates that during fractionation of asphaltene molecules following the "precipitation in n-heptane" procedure, it is possible that the asphaltenes are unwillingly selected in function of their lateral chain lengths (Fan and Buckley 2002;Kharrat et al 2007;Sabbah et al 2011;Langevin and Argillier 2016;Qiao et al 2017).…”

Tetranucleosome Interactions Drive Chromatin Folding

“… 24 We employ the density-adaptive variant of diffusion maps, which we find to be particularly useful for handling the large inhomogeneities in sampling densities observed in our chromatin simulations. 44 We provide a brief summary of the approach below, but direct the reader to prior publications for mathematical and algorithmic details. 24 , 41 − 43 …”

“…A Gaussian kernel is applied to d ij to construct a threshold pairwise distance matrix A , where ϵ is the kernel bandwidth and defines the local neighborhood of each point and α is a parameter that globally rescales pairwise distances to smooth out large density fluctuations between densely and sparsely sampled regions of configurational state space. 44 Matrix A is then row-normalized to form the transition matrix, where D is a diagonal matrix with elements, The transition matrix, M , is then diagonalized to calculate its eigenvectors ψ i and eigenvalues λ i . By the Markov property, the top eigenvalue–eigenvector pair (ψ 0 = 1⃗, λ 0 = 1) is trivial, corresponding to the steady-state distribution of a random walk.…”

Aggregation Behavior of Model Asphaltenes Revealed from Large-Scale Coarse-Grained Molecular Simulations

“…One of the models was further studied in Ref. 75 to understand the effect of temperature, pressure, and solvent composition on the averaged cluster size. The behavior of asphaltenes at oil-water interfaces was studied by Ervik et al 76 using multiscale techniques.…”