The research, led by Hua Zhang and Cejun Cao with colleagues including sustainable operations expert Sachin Kumar Mangla, builds a mathematical model of a two-firm shale gas supply chain: an upstream enterprise that extracts and treats the gas, and a downstream enterprise that transmits and distributes it to consumers. The central question is deceptively simple. When should each firm invest in production technology innovation, and should it behave farsighted, weighing how today’s decisions shape the technology stock of tomorrow, or myopic, chasing immediate profit and ignoring the future? The answer, the authors show, depends critically on one contractual number: the share of the upstream firm’s innovation costs that the downstream firm agrees to carry.

What makes the model distinctive is its engineering grounding. Rather than treating research and development as an abstract black box, the authors map innovation effort onto the actual stages of shale gas production. Upstream effort covers well planning, drilling and completion, hydraulic fracturing, flowback-fluid recovery, gas gathering, separation, dehydration, compression, and process monitoring. Downstream effort spans transmission, distribution, quality assurance, and demand-side coordination. These stages are linked by material, energy, and information flows, and the study represents their combined capability as a technology stock that grows with both firms’ innovation efforts and decays over time as equipment ages, staff turn over, and routines become obsolete.

The model also separates two distinct forces that push production costs down. The first is the cost-learning effect, a phenomenon first documented in 1936 when engineer T. P. Wright observed that the direct labour cost of aircraft production fell by roughly 20 percent with every doubling of cumulative output. In shale gas, where the dominant production cost is labour rather than materials, learning arises from standardised procedures, crew experience, and reductions in non-productive time across repeated drilling and fracturing campaigns. The second force is the direct cost-reduction effect of technological innovation itself: equipment upgrades, real-time sensing, predictive maintenance, and integrated scheduling that cut unit costs immediately. Previous studies typically captured one effect or the other; this work captures both simultaneously, alongside endogenous wholesale and retail pricing.

Formally, the study constructs a differential game, a framework for analysing decisions that unfold continuously over time, with the upstream firm acting as the Stackelberg leader. This reflects the real structure of China’s shale gas sector, where Sinopec, PetroChina, and CNOOC together account for roughly 82 percent of domestic production, making the concentrated upstream a natural first mover that sets the wholesale price and its own innovation effort before the downstream firm responds. The authors solve four behavioural combinations: both firms farsighted, upstream farsighted with downstream myopic, upstream myopic with downstream farsighted, and both myopic. Using Hamilton-Jacobi-Bellman equations, they derive equilibrium prices, innovation efforts, technology trajectories, and long-run profits for each case.

The headline finding is strikingly asymmetric. No matter what the downstream firm does, the upstream enterprise always benefits from farsighted behaviour. Because technology stock depreciates without sustained investment, and because the upstream firm holds long-lived sunk assets in wells and processing infrastructure, it has every incentive to keep investing in drilling, fracturing, treatment, and process improvement. The downstream firm, by contrast, is a swing player. When the cost-sharing ratio it offers is low, it prefers to behave myopically, contributing nothing to innovation itself and merely subsidising a fraction of upstream costs to hedge risk. Once the ratio crosses a threshold, roughly 0.4 in the benchmark simulations, the downstream firm switches to farsighted behaviour, launching its own innovation effort in demand-information sharing, quality feedback, and distribution optimisation. At that point both firms can achieve a Pareto improvement, meaning neither loses and at least one gains, and total supply chain profit peaks when both are farsighted at a cost-sharing ratio near 0.58.

The temporal dynamics add a twist that complicates any simple prescription. When the cost-sharing ratio is low, the equilibrium shifts over time: in the earliest phase both firms behave myopically, then the upstream turns farsighted while the downstream lags, and only after more than two decades of simulated time does the fully farsighted combination emerge. Remarkably, in that earliest short-run window, the myopic-myopic combination can itself deliver a Pareto improvement, suggesting that short-term gains are possible even without long-term thinking. But the authors caution against reading this as licence for complacency. Myopic behaviour may be individually rational when commitments are cheap, yet it leaves cross-stage process integration and long-term clean-production investment chronically underfunded, precisely the capabilities needed to unlock deep and ultra-deep shale resources.

The study also examines what happens when the cost-sharing ratio is not fixed in advance but negotiated between the firms using Nash bargaining, with the upstream firm holding stronger bargaining power. The result is sobering for system performance: when the ratio is endogenous, the downstream firm tends to settle into myopic behaviour, and a Pareto improvement may become unattainable. In other words, leaving the innovation subsidy to haggling can undermine the very cooperation it is meant to foster. The authors interpret the cost-sharing threshold not as a mere contractual parameter but as a minimum level of governance commitment required to sustain cross-stage coordination, and they suggest that when bilateral negotiation fails, energy regulators could step in with common data standards, third-party verification, and targeted incentives for joint clean-process innovation projects.

Real-world evidence from the Fuling field illustrates the farsighted upstream strategy the model describes. By November 2024 the field had cumulatively produced more than 70 billion cubic metres of shale gas, sustaining annual output above 8 billion cubic metres for several consecutive years. Its operator developed six core technology systems, issued 176 technical standards, secured 404 national patents, and applied slim-hole drilling to 208 wells, saving nearly 500 million yuan. Well-factory construction, electric fracturing, and closed-loop produced-water management achieved complete wastewater reuse while cutting land use and energy consumption. Downstream, PipeChina’s Southwest Pipeline network had transmitted more than 37 billion cubic metres by December 2024, connecting fields, storage, and markets in a way the model identifies as essential: production-side innovation realises its full value only when transmission, storage, and distribution are coordinated.

Sensitivity analyses reinforce the practical message. Stronger cost learning and more effective innovation both raise profits for both firms, but they work through different channels. Cost learning lowers variable costs enough that the upstream firm can cut wholesale prices and stimulate demand, benefiting everyone. Stronger innovation effects, however, spur the upstream firm to invest so heavily that it may raise wholesale prices to cover the expense. The two levers, operational learning and technology upgrading, are complementary, and the study argues that managers should institutionalise both: long-term innovation portfolios on the production side, and genuine information-sharing and cost-sharing commitments on the distribution side rather than treating subsidies as passive financial transfers. As China’s dual-carbon deadline approaches, the mathematics suggests that the future of its shale gas revolution depends less on any single breakthrough than on whether firms across the supply chain can commit, credibly and jointly, to thinking beyond the next quarter.

Subject of Research: Behavioural game-theoretic modelling of vertical cooperation, cost learning, and innovation in shale gas production supply chains

Article Title: A behavioural choice perspective on shale gas production technological innovation considering cost learning and innovation effects

Article References: Zhang, H., Cao, C., Liu, Y., & Mangla, S. K. (2026). A behavioural choice perspective on shale gas production technological innovation considering cost learning and innovation effects. Cleaner Engineering and Technology, 34, Article 101326. https://doi.org/10.1016/j.clet.2026.101326

Image Credits: AI Generated

DOI: 10.1016/j.clet.2026.101326

Keywords: shale gas, supply chain, differential game, cost learning, technological innovation, carbon neutrality, Fuling field, cost sharing, farsighted behaviour, myopic behaviour, natural gas, China energy transition

The central challenge is what geologists call a dual-medium reservoir. In these formations, gas lives in two very different storage systems at once: the tiny pores between sand grains, and the sparse network of natural fractures that cuts across them. Fractures act like high-speed highways for gas flow, while the matrix pores act like vast parking lots. Knowing which system dominates in any given interval of rock is essential for deciding where to drill and how to stimulate a well. But conventional classification methods rely on discrete analytical samples, mostly from core pieces pulled from a handful of wells. Most wells are never cored at all, so point-based analysis simply cannot capture how fractured and matrix-dominated zones alternate continuously down the borehole.

The research team’s first contribution is a revised classification framework, which they describe as a three-major and six-minor scheme, built on pore structure. It separates the Xujiahe sandstones into nine reservoir types spanning matrix-controlled, fractured, and hybrid matrix-fractured varieties. A key analytical device is the fracture identification factor, a quantity derived from well logs that distinguishes true fractured reservoirs from matrix-fractured ones. Because well logging curves are recorded continuously as instruments travel up the borehole, using them as classifier inputs means every centimeter of rock in every well can be typed, not just the rare intervals where core was recovered.

But continuous logging data brought a familiar artificial intelligence problem to the surface: class imbalance. Some reservoir types, particularly the highest-quality fractured varieties, are genuinely rare in the subsurface, so their examples are heavily outnumbered in any training dataset. Machine learning classifiers trained on such lopsided data tend to simply ignore the minority classes, which is precisely the wrong behavior when those rare classes are the exploration targets. To fix this, the team applied ADASYN, the adaptive synthetic sampling technique. ADASYN works by generating synthetic minority-class examples, concentrating its effort near the decision boundaries where classification is hardest. Rather than flooding the dataset with generic duplicates, it teaches the model to pay attention to the ambiguous frontier zones between reservoir types, which the authors found markedly strengthened recognition of the scarce but valuable classes.

With rebalanced data in hand, the researchers then assembled what they call a multi-ensemble learning architecture. Instead of trusting a single algorithm, they trained four different base classifiers: random forest, support vector machine, multilayer perceptron, and XGBoost. Each brings a distinct inductive bias. Random forests average many decision trees trained on random subsets of the data, support vector machines carve maximal-margin boundaries through feature space, neural networks learn flexible nonlinear mappings, and XGBoost builds boosted trees that iteratively correct earlier errors. The team optimized each base learner and then combined their votes using voting theory, testing both hard voting, where each classifier casts a single categorical vote, and soft voting, where the classifiers’ predicted class probabilities are averaged before a decision is made.

The results were striking. The best configuration, dubbed ADASYN-voting-soft, or AVS, achieved a mean accuracy of 90.25 percent across the reservoir classes. That figure represents an improvement of 12.22 to 16.84 percentage points over standalone machine learning methods. Just as important, the soft-voting ensemble proved more stable and more accurate than hard voting, especially for reservoirs whose well log signatures are ambiguous. This makes intuitive sense: when a rock sample sits near the boundary between two classes, individual classifiers may disagree, and hard voting can flip-flop from one measurement to the next. Averaging probabilities dampens that noise, producing smoother, more geologically plausible classifications along the borehole.

The AVS method also addresses a subtle risk in machine learning practice: the exclusion of small, high-quality sample sets. In geoscience, curated expert-labeled examples are expensive and scarce. A single classifier can effectively discard such precious minority data by underfitting them, but the ensemble approach compensates for the uncertainty of any individual model. Combined with ADASYN’s synthetic augmentation, the framework ensures that even thinly represented reservoir types exert real influence on the final prediction. The interpretability emphasis of the study matters here too, because geologists need to understand which logging features drive a classification before they commit drilling capital on the model’s advice.

Applied across the Xujiahe Formation’s second member, the workflow revealed the spatial distribution of reservoir types in the Xinchang area of the western Sichuan Depression, mapping where matrix-dominated zones give way to fracture-enhanced sweet spots. That spatial picture is more than an academic exercise. Accurate identification of high-quality reservoir intervals directly guides well placement, completion design, and production optimization in a basin that Chinese researchers have described as a super gas basin. Fracture-developed tight sandstone intervals hold disproportionately large shares of recoverable gas, so a tool that reliably flags them in uncored wells translates into fewer dry or underperforming wells.

The study also fits into a broader trend sweeping the geosciences: the migration of ensemble learning and imbalance-aware sampling from computer science into subsurface characterization. The authors’ bibliography traces a lineage from classic oversampling methods like SMOTE and Borderline-SMOTE through ADASYN’s adaptive variant, alongside a growing body of work applying random forests, boosting, and stacking to lithofacies identification, fracture detection, and fluid typing in basins from the Ordos to the Tarim. What distinguishes this contribution is the combination of a geologically meaningful classification scheme, an imbalance-correction step tuned to decision boundaries, and a probability-averaged ensemble, all applied to the notoriously tricky dual-medium problem.

Funded by China’s National Science and Technology Major Project and the National Natural Science Foundation of China, the work offers a template that could travel well beyond the Sichuan Basin. Any fractured tight reservoir, whether sandstone, carbonate, or volcanic, poses the same pairing of continuous well log data with rare, high-value target classes. As exploration pushes into ever deeper and more heterogeneous plays worldwide, tools like the AVS model suggest that the winning approach will not be any single clever algorithm, but disciplined engineering: classify the rocks in a way geologists understand, fix the data imbalance before it quietly biases the model, and let a well-calibrated team of classifiers vote where no individual model should be trusted alone.

Subject of Research: Machine learning classification of dual-medium tight sandstone gas reservoirs in the Sichuan Basin

Article Title: Interpretable Prediction of Dual-Medium Tight Sandstone Reservoir Types of the Sichuan Basin: A Multi-ensemble Learning Approach for Imbalanced Data Optimization

Article References: Zhang, J., Yue, D., Wang, W., Qu, L., Liu, Y., Li, W., Jin, W., & Fu, J. (2026). Interpretable Prediction of Dual-Medium Tight Sandstone Reservoir Types of the Sichuan Basin: A Multi-ensemble Learning Approach for Imbalanced Data Optimization. Natural Resources Research. https://doi.org/10.1007/s11053-026-10761-5

Image Credits: AI Generated

DOI: 10.1007/s11053-026-10761-5

Keywords: tight sandstone, Sichuan Basin, reservoir classification, machine learning, ensemble learning, ADASYN, imbalanced data, well logging, natural fractures, Xujiahe Formation, natural gas, dual-medium reservoir