Marcel P.T. Chin-A-Lien · GLIAG Intelligence B.V., Zoetermeer (The Netherlands) & Paramaribo (Suriname) · ORCID iD 0009-0007-1649-602X

1. The original study and the question that survives

The Maracaibo study of 1984 asked a question that still governs basin analysis: when did the source rock generate, how did the petroleum leave it, and which version of the basin received it. Talukdar, Gallango and Chin-A-Lien answered with an integrated method that tied Rock-Eval chemistry to restored burial, thermal history and the structural evolution of the basin (Organic Geochemistry 10:261–279). The 1986 abstract records the measured base: La Luna argillaceous limestones and calcareous shales of Cenomanian–Coniacian age with 1.5–9.6 wt% present-day TOC, a restored original TOC of 2.5–10.8 wt%, Type II kerogen with an immature HI near 700 mg HC/g TOC and H/C of 1.5, an oil window opening at TTI 10 and peaking at TTI 35, single-phase expulsion through overpressure-induced microfractures at 30–50% efficiency, and three oil families (abstract via Academia.edu).

The work was presented in three venues in 1985 before it went to print. The paper was read at the 12th International Meeting on Organic Geochemistry in Jülich and printed in Advances in Organic Geochemistry 1985, edited by D. Leythaeuser and J. Rullkötter, as volume 10 of Organic Geochemistry. A companion version was read in August 1985 at the II Simposio Bolivariano de Exploración Petrolera de las Cuencas Subandinas (EarthDoc). Two further Intevep papers at the VI Congreso Geológico Venezolano set out the crude-oil families (Gallango, Talukdar and Chin-A-Lien, 1985, Memoria III, 1661–1693) and the La Luna and Querecual source rocks (Talukdar, Gallango and Ruggiero, 1985, Memoria VI, 3606–3642), both cited by James (2000a).

The question of the right historical comparison is a matter of ledgers. Each inherited number needs its variable name, units, spatial support, interval, maturity basis and phase convention. The 290 MMbbl/km³ specific yield is the bridge between 1986 and today: Liborius-Parada (2019, printed p. 3) attributes it to Talukdar et al. (1986) as a yield at the end of the oil window, and James (2000a) reports the same value. This edition calls it the legacy specific-yield benchmark. It is not a basin total and not a recoverable volume.

1.1 How the original study was executed

I carried out the calculations by hand, using a Hewlett-Packard calculator. I drew the regional maps myself, including the formation-thickness maps used to reconstruct burial and migration. For 250 wells, I constructed the burial and maturation histories manually, drawing the graphs on paper and applying the time–temperature index, TTI. The restored thickness of Eocene strata removed during Oligocene uplift and erosion was particularly important: the present stratigraphic column alone could not reproduce maximum burial, heating or the former geometry of migration pathways.

Our calculations of generation, migration and trapping, and the resulting quantities, were discussed extensively with Dr John M. Hunt of the Woods Hole Oceanographic Institution, who served as our external project consultant for two years, 1980–1982, and with Professor Dr Parke A. Dickey. Both examined the data and calculations carefully and agreed with them. This account is my recollection as a participant in the original project, recorded for this retrospective on 2 November 2026.

The two paragraphs above are the author’s testimony. Independent records confirm the people and their standing. The Woods Hole Oceanographic Institution documents Hunt’s career as Senior Scientist (1974) and Scientist Emeritus (1984) and his textbook Petroleum Geochemistry and Geology (1979, 1996). Dickey’s Maracaibo work includes Dickey and Hunt (1972) and the Bolívar Coastal Fields crude-oil study of Bockmeulen, Barker and Dickey (1983). The project-specific consultancy dates and their review rest on testimony alone.

1.2 From hand-drawn maps to kinetic models

Every change in eroded section, palaeogeothermal gradient or stratigraphic thickness in 1984 meant redrawing and recalculating by hand, while keeping 250 well histories and the regional maps internally consistent. Within six years the same team and Lawrence Livermore National Laboratory replaced the TTI step with laboratory pyrolysis kinetics. They tested about 90 wells with an average TOC near 3.5 wt% and an expulsion onset at about 30% conversion, and confirmed the 1986 generation timing (Sweeney et al., 1990; OSTI preprint). The 1995 compositional model added pressure-driven expulsion and oil-to-gas cracking (Sweeney et al., 1995).

In my assessment the geological foundations of 1984 stand: source quality must be restored, maximum burial must include the eroded section, and migration must be read in the geometry of its own time. The AAPG Explorer (May 2026) reports several thousand feet of Upper Eocene section eroded in the Oligocene, which is the restoration the hand-drawn burial curves were built on. What has changed is resolution: kinetic and compositional models quantify what TTI could only rank. My judgement as co-author of the 1986 paper is that later models refine the framework rather than replace it, because each one has kept the source, the two burial pulses and the migration geometry and changed only the numbers attached to them.

2. The petroleum system on one page

Talukdar and Marcano (1993) defined the La Luna–Misoa(!) petroleum system over about 47,500 km². It supplies more than 98% of the 52.20 Bbbl of recoverable oil and 51.97 Tcf of gas then booked in the basin. The system developed in two phases from two separate mature to overmature pods of the same La Luna source, with Eocene and Miocene reservoirs holding 50% and 44% of recoverable oil. A second, minor system, the Orocué(.) of the south-west, covers 10,500 km², draws on Paleocene coaly Type III kerogen and holds less than 2% of reserves. The chapter in AAPG Memoir 60 set these numbers inside the Magoon and Dow (1994) framework.

Escalona and Mann (2006) tied the two pulses to tectonics. The first followed Paleogene oblique collision, when a north-eastern foredeep buried La Luna and oil moved up-dip along north–south strike-slip faults into Eocene reservoirs. The second followed Oligocene–Miocene uplift of the Sierra de Perijá and the Mérida Andes, which depressed the Maracaibo syncline and started generation in the central and southern basin, with northward migration into Miocene rocks and surface seeps along the basin margins. They give a basinal area of 50,000 km², more than 30 Bbbl produced and more than 44 Bbbl recoverable.

The Intevep–Beicip regional model of Tocco, Gallango and Parnaud (1996), with Gallango a co-author of the 1986 paper, extended the same logic over more than 100,000 km² of the Maracaibo and Barinas basins. It assigned 90% of generated hydrocarbons to the La Luna, Capacho and Navay Cretaceous source and 10% to the Oligocene Carbonera Formation and the Paleocene Orocué Group, defined three La Luna expulsion periods from the Early Eocene to the Holocene and named five petroleum systems. The 90% share matches the attribution of Sweeney (1990), who credits Talukdar et al. (1986) with the finding that La Luna generated more than 90% of the basin’s oil.

The two-pulse model has since been tested at field scale. Escobar et al. (2020) analysed 30 oils from the northern Bolívar Coastal Complex with carbon isotopes, biomarkers and diamondoids. They found a Middle Eocene charge, biodegraded in the Late Eocene to Peters–Moldowan level 6 in every sample, refreshed by a Miocene–Recent charge at about 1% Ro equivalent with degradation levels of 1 to 5. Escobar et al. (2022) modelled Urdaneta West and reported end-Eocene transformation ratios of about 35%, 98% and 100% at three wells, with early northern charge and a later, more mature southern contribution. Both studies reproduce the 1986 history in miniature.

3. Depositional architecture: from a formation name to source-rock cells

La Luna is a lithostratigraphic name, not one value of TOC, thickness or kinetics. Historical facies maps (Savian and Scherer, 1993) and TOC and Tmax maps (Scherer, 1995) record changing carbonate, shale and sand proportions and organic richness across western Venezuela. They are held by the Léxico Estratigráfico de Venezuela and were shown again at the 2000 La Luna Research Conference in Caracas and in Lorente’s 2026 review article. This edition cites them and does not reproduce them, since they belong to their authors.

Chronostratigraphy controls charge modelling because a source that begins or ends at different times in different sectors changes residence time and burial path. De Romero et al. (2003) built an integrated calcareous-microfossil and carbon-isotope framework. Zapata et al. (2003), Parra et al. (2003) and Rey et al. (2004) resolved Turonian–Campanian sections, lateral microfacies change and orbital cyclicity during OAE3. Erlich et al. (2000) traced the birth and death of the “La Luna Sea” to the Tres Esquinas phosphorites. Bralower and Lorente (2003) framed the PALAIOS special issue.

OAE terminology needs the same discipline. Machado et al. (2016) used carbonate isotopes to constrain OAE3 in La Luna, and Erba et al. (2026) showed that Coniacian–Santonian black-shale deposition was site-dependent within a supra-regional perturbation. Jenkyns (2010) gives the global frame. A model should therefore split La Luna into cells, each carrying net source thickness, age range, original potential, kinetic class and thermal history, and should hold correlation uncertainty in alternative cell assignments.

Below La Luna, Méndez Dot et al. (2015) interpreted organic-rich Machiques and Piché intervals of the Aptian Apón Formation (Cogollo Group) as effective sources linked to OAE 1a and 1b, with short migration into adjacent carbonates. That older source is real but unquantified at basin scale. None of the published La Luna–Misoa mass is reassigned to it here.

Two later data types make cell assignment more practical. Baritto’s 2009 master’s thesis at the Universidad Central de Venezuela (SABER UCV, series POST B239 2009) analysed 360 samples from two central-lake cores of La Luna and the Cogollo Group by wavelength-dispersive XRF, whole-rock XRD and Rock-Eval. It defined six chemical units per well and five element associations, including an organic association (TOC, Na₂O, Zn, Ni, Cu, Cr) and a productivity association (P₂O₅, MgO, MnO, Ba). Discriminant analysis matched formations between the wells at 76% and genetic sequences at 86%, so inorganic chemistry can carry source-cell boundaries into uncored wells. Leal et al. (2022), in Geofísica Internacional, trained a random-forest model on 960 log measurements and predicted La Luna TOC with R = 0.88 and an RMSE of 0.44 wt% against an average of about 5 wt%. Neither method replaces restored TOC0 and HI0, but together they turn a formation average into a mapped, testable property.

4. Organic facies, maturity and the original potential

The source that remains is not the source that existed. Present-day TOC and HI describe material already altered by conversion and expulsion, so a low residual HI in a mature sample can record a depleted good source rather than a poor one. The 1986 study restored original TOC to 2.5–10.8 wt% and used immature HI near 700 mg/g (Talukdar et al., 1986). The 1994 parameter set quoted by James (2000b) uses mean TOC 5.56 wt%, HI 650 mg/g, rock density 2.65 g/cm³ and 65 m net thickness. Escalona and Mann (2006) quote the same 5.6% average original TOC after Talukdar and Marcano.

The 2019 dissertation of Liborius-Parada adds interval-scale detail from four cores and one outcrop, with 153 analyses. Group averages span 3.85–9.10 wt% TOC with a Type IIS interpretation, and the work was extended in Liborius-Parada et al. (2023) and Liborius-Parada and Slatt (2016). Sulphur-rich kerogen converts at lower temperature than standard Type II, so the useful step is to assign kinetic populations by facies rather than replace one uniform model with another. Zumberge (1984) gives the Colombian La Luna geochemistry, and Peters and Cassa (1994) give the screening classes.

Maturity calibration is weakest where vitrinite is scarce, as it is in carbonate-rich La Luna. Solid-bitumen reflectance, biomarker isomerisation, Tmax and Raman methods measure different properties. Hackley et al. (2021) review the spatially resolved methods, and a model should store each measured observable and its conversion rule separately.

5. Burial, heat flow and the history of conversion

A generation model integrates over geological history, and present-day depth cannot recover that integral. The 1984 method used Lopatin’s TTI in the form calibrated by Waples (1980). The kinetic successors use parallel first-order reactions with an activation-energy distribution (Burnham and Braun, 1990; Sweeney and Burnham, 1990; Pepper and Corvi, 1995). For reaction j, the unconverted fraction falls at a rate set by A_j exp(−E_j/RT), and the total transformation is the weighted sum. The Maracaibo test of the Easy%Ro method was published by Sweeney and Burnham (1992).

The 1990 kinetic study found four conversion trajectories in its 90 wells: early completion, early generation interrupted and restarted by later burial, rapid late conversion, and continuing late generation (OSTI preprint). This is the two-pulse history expressed well by well. Generation must be incremental: for a cell with potential P, new generation in a time step is P times the increase in TR. A second heating acts only on the remaining potential, so a model that resets the source after an unconformity creates petroleum without a precursor.

Erosion is the largest source of non-uniqueness. Several pairs of eroded thickness and heat flow fit one present-day maturity value. A calibration should therefore combine corrected temperatures, more than one maturity indicator, stratigraphic restoration and, where available, thermochronology.

The size of the erosion has been disputed for 30 years. A Maraven BasinMod study by White and Mompart (1996) restored up to 1,500 ft (about 460 m) of eroded Eocene section in the lake area, against 10,000 ft (about 3,050 m) in previous studies. The 1986 restoration started from Banks (1978), a Maraven internal estimate from sonic and density logs, which Sweeney et al. (1990) then adjusted well by well against vitrinite reflectance (OSTI preprint). The White and Mompart abstract does not name the previous studies, so whether its 10,000 ft is Banks’s north-eastern maximum or a lake-wide value cannot be settled without the full Maraven report. It placed the first La Luna expulsion in the north-east in the Early rather than the Middle Eocene. It moved the oil window to the south-west and then to the north-west, left the western lake coast immature, and found gas expulsion continuing today in the Miocene trough north-west of the Andes. A sevenfold difference in eroded section changes the Eocene burial path, so the first-pulse conversion of any replay depends on which restoration is used.

Thermochronology now dates the second pulse independently. The 2009 doctoral dissertation of Mauricio A. Bermúdez at Université Joseph Fourier, Grenoble (HAL tel-00450838), supervised by Peter van der Beek and Matthias Bernet, reported 47 new apatite fission-track ages from bedrock, synorogenic and modern river sediment across the Venezuelan Andes. It found diachronous Late Miocene–Pliocene exhumation across the Boconó fault. The Sierra Nevada de Mérida exhumed at about 1.0 km/Ma between 14 and 9 Ma, slowing to 0.3 km/Ma since 6 Ma, and the Sierra La Culata at 0.8–1.1 km/Ma since 4 Ma. The dissertation gives a maximum burial of about 6 km in the Maracaibo foreland. The results were published in Tectonics (Bermúdez et al., 2010), Geology (2011) and Lithosphere (2013). Bermúdez et al. (2017) found youngest detrital ages of 5 ± 2 Ma in the Maracaibo foreland and no burial resetting in the sampled Mio-Pliocene sections, and the Maracaibo-block synthesis followed in 2019. The Andean load that drove the second pulse is therefore mostly younger than 15 Ma and peaked after about 8 Ma. This agrees with the Miocene–Recent charge at about 1% Ro that Escobar et al. (2020) found in the Bolívar Coastal oils.

6. How much hydrocarbon was generated? The quantitative core

6.1 Define the quantity before calculating it

Five quantities must stay separate: initial potential, cumulative generation, cumulative expulsion, petroleum arriving at traps, and petroleum remaining at a stated time. Recoverable resources and reserves are later engineering categories. This edition treats hydrocarbon mass as the governing quantity. For comparison only, it divides mass by 850 kg/m³ and 0.158987294928 m³/bbl, so 1 equivalent barrel equals 135.14 kg. The result is an oil-density-equivalent barrel, not an energy boe.

6.2 A dimensionally explicit source-rock equation

For source cell i, with area A in km², net thickness h in m, rock density ρ_r in kg/m³, original TOC f_C0 as a mass fraction, initial HI₀ in mg HC/g TOC and transformation ratio TR:

The factor 10⁶ converts km² to m², and HI₀/1000 converts mg/g to kg/kg. Summing over cells gives cumulative gross generation; a kinetic implementation replaces the chosen TR with the value predicted by each cell’s thermal history. The same structure underlies the volumetric method of Schmoker (1994) and the mass-balance method of Cooles et al. (1986).

6.3 The legacy benchmark and three sensitivity cases

One cubic kilometre of rock at 2,650 kg/m³ weighs 2.65 × 10¹² kg. At 135.14 kg per equivalent barrel, 290 million barrels equals 39.19 million tonnes of hydrocarbon, or 1.479 wt% of the rock. With HI₀ at 700 mg/g, the product f_C0 × TR must equal 0.0211. These are algebraic normalisations, not recovered 1986 inputs.

The three upper rows are chosen to expose sensitivity; they are not P90/P50/P10 and carry no probability. Figure 3 plots them against the benchmark, and Figure 4 shows that many pairs of richness and conversion reproduce any one yield.

6.4 The published mass balance

The 1993 abstract gives the mass balance in its authors’ own units. Generation from average TOC and consumed hydrogen totals 2.50 × 10¹⁴ kg, and known in-place hydrocarbons total 352.25 × 10¹¹ kg, so 14% accumulated in known traps (Talukdar and Marcano, 1993). The direct ratio is 14.09%. At 850 kg/m³ the two masses equal 1,849.94 and 260.66 billion equivalent barrels. The first agrees with the rounded 1.8 trillion barrels that James (2000b) attributes to the same synthesis.

6.5 New result: the reconciliation test

The published mass makes a demand that the earlier editions did not test. With the 1994 parameter set, 2.50 × 10¹⁴ kg spread over the full 47,500 km² footprint at 65 m net thickness needs a basin-average TR of 0.845. At full conversion the same mass needs at least 40,160 km² of source, which is 84.5% of the footprint. Yet the 1993 abstract places generation in two mature to overmature pods, not across the whole system. The mass and the map can only agree if the pods covered most of the footprint, if net source in the pods was thicker than 65 m, or if their richness exceeded the basin mean. At 90 m net, the required basin-average TR falls to 0.61; at 120 m, to 0.46 (Figure 5).

The same arithmetic resolves the apparent gap between the 1986 and 1994 yields. Applied uniformly, the legacy 290 MMbbl/km³ would need 98,140 km² of 65 m source to deliver the 1994 mass, twice the footprint. The mass-implied yield is 599 MMbbl/km³, 2.07 times the benchmark. One decomposition reproduces the ratio exactly: richness 5.56/3.5 = 1.59 (the 1994 mean against the 1990 kinetic-study average near 3.5 wt%), conversion 0.845/0.604 = 1.40 (basin-average total conversion against an oil-window TR that gives 290 at 3.5 wt% and HI 700), and initial potential 650/700 = 0.93. The product is 2.07.

The decomposition is not unique, but it is physically sensible. The 1986 figure reads as a liquid yield at the end of the oil window on a moderate-richness source; the 1994 figure reads as total hydrocarbons from consumed hydrogen in rich, overmature pods. Under this reading the two papers measure different stages of one system, and neither invalidates the other. The 1986 calculation sheets were Intevep working files and are not in the public record, so the practical test is a cell-by-cell model run once with each parameter set (Section 12, hypothesis H1). As co-author I read the 1986 number as a conservative liquid yield, chosen to avoid overstating charge from 250 hand-drawn burial histories.

6.6 New result: the trap ledger has grown since 1993

In 2026, Sandrea and Essenfeld report about 270 Bbbl of OOIP in 13 giant fields of the basin and more than 40 Bbo produced. The same article places 73% of Venezuela’s more than 450 Bbo of discovered conventional oil in the Western basin, which implies more than 329 Bbbl. The two figures differ by 22% inside one source, so this edition carries both as a range.

Converted to mass at 850–950 kg/m³, 270 Bbbl of oil equals 3.65–4.08 × 10¹³ kg and 329 Bbbl equals 4.44–4.96 × 10¹³ kg. Against the generated 2.50 × 10¹⁴ kg, oil alone in known fields therefore represents 14.6–19.8% of the generated mass, before any gas is counted. The 1993 figure of 14% for all hydrocarbons is at or below the bottom of that range. Either the 1993 inventory understated in-place oil, which is likely given later re-evaluation of the heavy-oil fields of the Bolívar Coastal Complex, or generation exceeded 2.50 × 10¹⁴ kg. Both are testable, and both strengthen the case that the 1984–1994 balance was conservative. [VERIFY: basin-wide OOIP by field with dates]

6.7 What production means against generation

More than 40 Bbbl produced is 14.8% of 270 Bbbl OOIP and 15.3% of the 1993 in-place mass expressed at 850 kg/m³. At 900 kg/m³, 40 Bbbl weighs 5.7 × 10¹² kg, about 2.3% of all hydrocarbons the La Luna generated in this system. Recoverable estimates have moved from 52.20 Bbbl in 1993 to more than 44 Bbbl in 2006 (Escalona and Mann, 2006) because their definitions differ. They should be compared by date and class, never subtracted from generated mass.

6.8 A carbon and petroleum ledger

At any time, generated petroleum is shared among retained, in-transit, accumulated, produced, lost and transformed compartments:

G is generated petroleum, I imported charge, R retained, T in transit, A accumulated, P produced, L lost and X transformed products such as gas from cracking or carbon from biodegradation. Gas from cracking is a transfer between classes, not a new addition. Oil that moves through two traps is counted once. The 85.91% not found in known fields in 1993 is therefore a residual of retention, dispersal, leakage, alteration and uncertainty, not a volume awaiting discovery.

7. Expulsion, retention and secondary cracking

Expulsion efficiency means nothing until its products, denominator and time window are named. The 1986 abstract gives 30–50% for oil, through overpressure-induced microfractures in a single phase (Talukdar et al., 1986). Hydrous pyrolysis on an unusually rich La Luna sample with 11 wt% TOC showed expulsion beginning at H/C near 1.25 and an ultimate efficiency near 75% (Talukdar et al., 1988, as reported in the OSTI preprint). The kinetic study assumed onset at about 30% conversion for a typical source. Richness, confinement and definition can explain the spread without any change in the basin over time; Mackenzie and Quigley (1988) and Pepper and Corvi (1995) show the same dependence on richness in their expulsion models.

If 290 MMbbl/km³ is gross generated oil, 30–50% expulsion leaves 87–145 MMbbl/km³ expelled. The remainder has not become recoverable shale oil; it may have cracked, migrated locally or stayed bound in organic pores. The AAPG Explorer (May 2026) reports continuous resources of 15.2 Bbbl and 204.8 Tcf, mostly in La Luna, while EIA (2013) gives 14.8 Bbbl and 202 Tcf for the Maracaibo–Catatumbo domain and the USGS (2017) gives a mean of 656 MMbbl and about 5.7 Tcf. These differ in footprint and recovery assumptions and are not reserves.

The 1995 compositional model links organic maturation and compaction to excess pressure and episodic fracture expulsion (Sweeney et al., 1995). Juliao et al. (2015) show in Colombian Upper Cretaceous rocks that solid bitumen both creates and fills porosity. Oil that escaped early also avoided the later heating of its source, so timing decides how much gas the second pulse delivered.

8. Migration: a changing network of pathways and barriers

Primary migration moves petroleum out of the source, secondary migration moves it through carriers and faults, and remigration redistributes what was trapped. Only conversion creates petroleum. For a flowing phase α, Darcy flux follows the gradient of phase potential, and capillary entry pressure follows interfacial tension, contact angle and throat radius:

Buoyancy drives oil up a connected system, but pressure gradients and fault juxtaposition can send it laterally or down into older carbonates of the Cogollo Group. England et al. (1987) set out the physics of secondary migration and filling. In Maracaibo, Escalona and Mann (2006) identify strike-slip, normal and inverted faults and the late Eocene–Oligocene unconformity as the main pathways. Tocco and Margarita (1999) show multiple filling and redistribution among Misoa reservoirs at Centro Lago.

Faults can transmit charge at one stage and seal the same section later, so a model needs time-dependent transmissibility for each fault segment. Restoration matters because source and trap positions at the time of charge differ from today’s. A regional two-pulse model must also predict differences at individual fields, and Escobar et al. (2022) supply that test at Urdaneta West: an early, less mature northern charge and a later, more mature southern contribution.

9. Source-to-oil correlation, mixing and alteration

The 1986 study separated marine, terrestrial and mixed oils, and the 1993 abstract correlated the main family with Type II La Luna kerogen through biomarkers and assigned terrestrial input to the Orocué system (Talukdar and Marcano, 1993). Later work refined, rather than replaced, this scheme. Escobar et al. (2020) found mixtures of a nearly terrestrial siliciclastic-sourced oil and a purely marine carbonate-sourced oil in different proportions in 30 samples. The Biomarker Guide of Peters, Walters and Moldowan sets out the methods, and Bockmeulen et al. (1983) record the crude-oil geochemistry of the Bolívar Coastal Fields.

A heavy oil today is not necessarily a heavy primary product. Biodegradation removes selected compounds and reduces surviving mass, and a late light charge into an older altered pool can blur API and maturity trends. The author’s own work on 25-norhopanes, the demethylated hopanes that record severe biodegradation, applies this to Maracaibo crude oils (Tracing Ancient Oils). Reconstructing pre-alteration mass needs the degree of alteration, its products and its duration, not a single shrinkage factor.

The molecular half of the 1986 lineage was built at Intevep alongside the mass balance, and its main author was Fernando Cassani. His 1985 University of Bristol PhD, carried out in Geoffrey Eglinton’s Organic Geochemistry Unit, dealt with the oils of the Orinoco Heavy Oil Belt (Cassani, 1985). Two Chemical Geology papers followed from it. The first used pyrolysis of asphaltenes to release bound steranes and triterpanes, which lets oils be correlated and their maturity read even after biodegradation has destroyed the free biomarkers (Cassani and Eglinton, 1986). The second ranked the alteration of the same extra-heavy crudes from steranes, triterpanes and phenanthrenes (Cassani and Eglinton, 1991; Pascal-Francis record). For Maracaibo itself, Cassani, Gallango, Talukdar, Vallejos and Ehrmann calibrated the methylphenanthrene index on La Luna extracts and oils (Cassani et al., 1988). OpenAlex counts 68, 50 and 147 citations for the three papers on 2 November 2026, so the aromatic maturity scale is the most used output of the Intevep programme after the 1986 paper.

The 1993 Cassani, Audemard and Eglinton abstract carries the method east (AAPG/SVG 1993). Steranes, triterpanes, aromatic steroids, phenanthrenes, HPLC petroporphyrins and SARA fractions show oils generated from early-mature Cretaceous source rocks in the north and severely biodegraded in shallow reservoirs. Cerro Negro, Hamaca and eastern Zuata share one source; Machete and western Zuata point to a different, less mature one. Talukdar (1991) gave the companion mass-balance view, a single Querecual–San Antonio petroleum system for the Orinoco Belt and the Maturín subbasin (OSTI 7171413), and Gallango et al. (1992) dated Querecual generation to the early Paleocene–early Miocene before thrusting, with a second phase beneath the thrust sheets (OSTI 5755211). The Querecual is the eastern twin of La Luna: more than 450 m of bathyal, upwelling-fed carbonates and laminated mudrocks on Chimana Grande, upper Albian (Rotalipora ticinensis Zone) to Santonian, with TOC highs in the upper Albian to mid-Cenomanian and the Turonian (Crespo de Cabrera, Sliter and Jarvis, 1999). Later UCV work tied Orinoco and Socororo oils to it through V/Ni and 25-norhopanes at biodegradation levels 7–8 (López et al., 2015; López and Lo Mónaco, 2017).

For the mass balance the Bristol thread does one thing the 1986 paper could not. Bound biomarkers in asphaltenes survive the alteration that strips the free fraction, so they can test whether a heavy Maracaibo oil and a light one came from the same La Luna cell before mixing and biodegradation are subtracted. This review (2 November 2026) found no public study that applies asphaltene pyrolysis systematically across the Bolívar Coastal Fields. Published Maracaibo work uses asphaltenes for V/(V+Ni) and sulfur, not for bound biomarkers (López and Lo Mónaco, 2017); any Intevep data of this kind sit in internal reports that are not public.

10. Production, reserves and undiscovered potential in two basins

10.1 What has been realised in Maracaibo

The basin has produced more than 40 Bbo from about 270 Bbbl OOIP (Sandrea and Essenfeld, 2026). They cite the Mara field at 3.9 Bbbl OOIP and 475 MMbbl produced, and remaining targets of more than 9 Bbbl of heavy oil in the Bolívar Coastal Fields, 1.6 Bbo at Centro Lago and 4.6 Bbo in West Lake. Field-level disclosure is thin. For the four PetroUrdaneta fields (La Paz, Mara Oeste, Mara Este and El Mojan), Maha Energy (2025) reports about 350 MMbbl 1P and 500 MMbbl 2P oil and about 250 MMboe of 2P gas on PDVSA reviews. Its 2025 report gives more than 500 MMbbl 2P oil, about 200 MMboe of gas, 1.4 Bbbl produced and more than 8 Bbbl oil in place (Maha, 2026), and its Q1 2026 presentation states 8.6 Bbbl OOIP with a 24% Maha interest (Maha, 2026b). These are full-field figures, not Maha’s share. Adding 1P to 2P would double count proved volumes.

10.2 What may still be discovered

An undiscovered resource comes from an assessment, never from generated mass minus production. The USGS 2025 assessment, published in March 2026, puts both provinces on one fully risked conventional basis (Schenk et al., 2026). Its 31 provinces sum to a mean of 37,584 MMbbl of oil and 281,610 Bcf of gas.

The Maracaibo mean of 787 MMbbl is about 2% of the oil already produced; the basin’s opportunity now lies in its known fields, which Sandrea and Essenfeld call silverfield production. The Guyana–Suriname mean of 7,912 MMbbl is ten times larger, and the AAPG Explorer (May 2026) records Venezuelan geologists making the same yet-to-find argument for their own basins.

10.3 One method, two kitchens

The accounting that served Maracaibo in 1984 is the one GLIAG applies to the Guyana–Suriname Basin. The author’s published SE Golden Lane model reports about 1,455 Bboe generated, 949 expelled, 171.7 trapped in place and 73.3 recoverable (SE Golden Lane). Its trapping ratio of 11.8% sits close to the Maracaibo 14.09%, but the definitions differ (energy-equivalent boe against mass) and the closeness is not independent validation. The Guyana–Suriname source architecture separates an Aptian contribution from the Albian–Cenomanian–Turonian (ACT) aggregate (A World-Class ACT; The Overlooked Aptian), the same distinction Méndez Dot et al. draw between Apón and La Luna. The methodology is published in Chin-A-Lien (2026).

Regionally, the Cenomanian–Turonian source of the Guyana–Suriname margin (Canje Formation) belongs to the same Late Cretaceous anoxic episode as La Luna (Erbacher et al., ODP Leg 207; Pindell and Kennan, 2009). The two basins differ in what came after: Maracaibo was inverted, eroded and recharged by Andean burial, whereas the Atlantic margin was buried progressively under a growing wedge (Yang and Escalona, 2011; Cedeño et al., 2021).

University theses carry much of the primary data on the Atlantic side. Almut Bunte (née Hetzel) measured 845 samples at ODP Site 1258 and 219 at Site 1260 for her 2009 Oldenburg PhD: TOC averages 17.8 wt% inside OAE2 at Site 1258 against 7.8 wt% below it, with a maximum of 30.2 wt%, and iron speciation, pyrite δ³⁴S and isorenieratane show euxinia reaching the photic zone (Bunte, 2009; Hetzel et al., 2009). She also reports the Demerara organic matter as thermally immature and well preserved, which makes the rise a reference for original potential, not for conversion. Christian März traced the Coniacian–Santonian redox history at Sites 1259 and 1261 in his 2007 Bremen PhD (März, 2007), and Yannick Mourlot’s 2018 Toulouse PhD with Total compared Demerara (TOC 5–30%, 1–9 m/Ma) with the shelf well Arapaïma-1 (TOC about 2–5%, 10–60 m/Ma) and assigned the difference to dilution and preservation (Mourlot, 2018). On the oil side, Andrés Cedeño’s 2023 Stavanger PhD split 15 onshore Suriname heavy oils into a distal-marine Group A, rich in extended tricyclic terpanes and assigned to the Canje, and a proximal Group B about 3‰ heavier in δ¹³C (Cedeño, 2023). Clyde Griffith’s 2015 MSc at the Anton de Kom University of Suriname mapped ACT maturity and charge into the ultra-deep offshore (Griffith, 2019).

The Aptian–Albian level has fewer theses but the same structure. Méndez Dot et al. place Machiques TOC of 1–5.5% at OAE1a, near 120 Ma, and the Piché member at OAE1b, near 113 Ma, in Maracaibo, while on the Guyana margin a Lower Albian OAE1 source is argued to have charged the deepwater fields alongside the Canje (S&D 2019 GTW Suriname) and Aptian and A3CT Organofacies B mudstones reach 15.8% TOC and HI 730 mg HC/g TOC (GeoExPro, 2026). The Cassani approach transfers directly: Cedeño’s two Suriname families are the Atlantic equivalent of the Cerro Negro and Machete split, and asphaltene-bound biomarkers could test whether the Tambaredjo heavy oils carry an Aptian component that the free fraction no longer shows [VERIFY].

11. Evidence statement

12. A falsifiable reconstruction programme

The next increment is a closed charge reconstruction that can fail. A replay should first reproduce the published specific yield and mass totals with their own parameter sets. If it fails, the failure should be traced to missing definitions, not hidden by tuning geology. Calibration should then use separate evidence for each process: temperatures and maturity for thermal history, oils and extracts for source, restored connectivity for migration, dated inventories for accumulation. A model tuned to an oil inventory cannot then cite that inventory as validation.

Uncertainty should travel through physically coherent scenarios, because TOC and HI covary by facies and burial covaries in space. Where probabilities cannot be defended, structured scenarios are better than borrowed P90/P50/P10 labels. The decisive tests are mass or carbon closure and out-of-sample prediction, reported by cell and time step.

12.1 A moving target: data and methods since 2009

The 1984 study used hand-contoured maps and Lopatin TTI because those were the tools of the time. Each later tool has changed one term of the source-rock equation, and none has yet been combined with the others in a public basin-scale replay. This edition is therefore an open revision: later Zenodo versions under the same concept record will add data as they become public, and earlier versions stay citable.

The next gains are likely to come from (U-Th)/He ages on the lake-area Eocene unconformity, kinetics measured on Type IIS La Luna rather than borrowed from other marine sources, and digitised legacy well and core data. Each would shrink one of the uncertainties in Section 11 [VERIFY].

13. Conclusions

The 1984 Maracaibo study asked how a changing basin turned organic matter into the oil now produced, and the answer has held for 42 years. Kinetic models confirmed its timing in 1990, the petroleum-system synthesis confirmed its source attribution in 1994, the tectonic overview confirmed its two pulses in 2006, and field studies in 2020 and 2022 found the same history in individual oils and wells. The paper is still the most cited petroleum-system study of the basin among those checked.

This continuation adds four numbers that make the framework testable. The published generated mass requires TR ≈ 0.85 over the whole system footprint, or at least 40,160 km² of fully converted 65 m source. The legacy 290 MMbbl/km³ and the 1994 mass reconcile through richness, conversion and HI factors of 1.59, 1.40 and 0.93. Oil alone in known fields now amounts to 15–20% of generated mass, against 14% for all hydrocarbons in 1993. Production of more than 40 Bbbl is about 2.3% of the mass the La Luna generated.

Generation is integrated once over each cell’s remaining potential. Expulsion moves petroleum out, migration and remigration move it again, and trapping and alteration decide what survives. A second burial or a second trap does not create a second copy of the oil. That rule carried the hand-drawn work of 1984, and it carries the GLIAG work on the Guyana–Suriname Basin.

Appendix A. Equations, units and reproducible cases

A1. Specific yield. One km³ at 2,650 kg/m³ with TOC0 0.045, HI0 600 mg/g and TR 0.65 generates 4.65075 × 10¹⁰ kg, or 344.145 million equivalent barrels at 135.1392 kg/bbl.

A2. Reconciliation. TR over the footprint = 2.50 × 10¹⁴ / (47,500 × 10⁶ × 65 × 2,650 × 0.0556 × 0.65) = 0.845. Minimum area at TR = 1 is 40,160 km². Area at 290 MMbbl/km³ and 65 m is 2.50 × 10¹⁴ / (290 × 10⁶ × 135.1392 × 0.065) = 98,140 km².

A3. Decomposition. At TOC0 3.5 wt% and HI0 700 mg/g, 290 MMbbl/km³ needs TR = 0.02113 / 0.035 = 0.604. The ratio 599.2 / 290 = 2.066 = (5.56/3.5) × (0.845/0.604) × (650/700) = 1.589 × 1.399 × 0.929.

A4. Trap ledger. 270 × 10⁹ bbl × 0.158987 m³ × 850–950 kg/m³ = 3.65–4.08 × 10¹³ kg (14.6–16.3% of generated); 328.5 × 10⁹ bbl gives 4.44–4.96 × 10¹³ kg (17.8–19.8%).

A5. Incremental generation. If TR rises from 0 to 0.40 in an early episode and from 0.40 to 0.80 later, gross generation is 0.80 P, not 1.20 P. The calculations are reproduced in the companion script GLIAG-MAR-PS-1984-2026-001_charge_ledger.py deposited with this record; it uses only standard Python.