Back to Insights

What 10,000 Permian Wells Tell Us About Completion Design: Data-Driven Insights for Frac Engineers

Dr. Mehrdad Shirangi | | Published by Groundwork Analytics LLC
Editorial disclosure

This article reflects the independent analysis and professional opinion of the author, informed by publicly available production and completion data, published research, and professional experience. No vendor reviewed or influenced this content prior to publication. Data ranges cited are derived from public sources including the Texas Railroad Commission (RRC), FracFocus, the EIA, and peer-reviewed SPE papers. Individual well economics will vary.

The Permian Basin has been the proving ground for unconventional completion design for over a decade. More than 10,000 horizontal wells have been completed across the Delaware and Midland sub-basins since 2018, and the public data trail they leave behind -- production records at the Texas RRC, chemical disclosures on FracFocus, and well attributes aggregated by commercial data providers -- tells a remarkably clear story about what works and what does not.

Yet most completions teams still design wells using what amounts to a copy-paste approach: take last year's best well, replicate the design across the next pad, adjust for lateral length, and move on. The institutional knowledge embedded in that process is real and valuable. But it also leaves money on the table -- sometimes seven figures per well -- because it treats geology as a constant when it is anything but.

This article examines what the data actually shows about completion parameter-production correlations, where diminishing returns set in, how the Delaware and Midland sub-basins differ in ways that matter for design, what extended laterals are teaching us about completion scaling, and where machine learning methods can outperform the empirical rules that most teams rely on today.

The Public Data Landscape

Before diving into the analysis, it is worth understanding what data is actually available and what its limitations are.

Texas Railroad Commission (RRC)

The Texas RRC publishes monthly production data for every well in the state, including oil, gas, and water volumes. For Permian Basin horizontal wells, this data provides the production outcomes that any completion analysis ultimately needs to explain. The RRC also publishes completion reports (Form W-2) that include perforated interval, total depth, and basic well information.

The limitation is that RRC production data lags by 2-6 months and is subject to revision. Early-month data frequently understates actual production. Normalizing for this requires either using only mature data (6+ months old) or applying state-specific correction factors.

FracFocus

FracFocus (fracfocus.org) is the national hydraulic fracturing chemical disclosure registry. For each well, operators report the total volume of water and proppant pumped, along with a breakdown of chemical additives. Since 2012, FracFocus has accumulated disclosures for tens of thousands of Permian Basin wells.

What FracFocus gives you: total proppant mass, total fluid volume, and the ability to calculate proppant loading (lbs/ft) and fluid intensity (bbl/ft) when combined with lateral length data. What it does not give you: stage count, cluster spacing, perforation design, pump rate, treating pressure, or any measure of how the total proppant and fluid were distributed along the lateral. Those parameters require either commercial databases (Enverus, TGS/WellDatabase, Novi Labs) or operator disclosures.

Enverus, Novi Labs, TGS

Commercial data providers like Enverus (DrillingInfo), Novi Labs, and TGS aggregate completion data from multiple sources -- FracFocus, state agencies, investor presentations, and proprietary datasets -- into normalized databases that include stage count, cluster spacing, proppant type, fluid system, and other design parameters alongside production data.

These platforms are where the most detailed completion analytics happen. Novi Labs, for example, publishes research using their Permian dataset that includes spacing, vintage, and landing zone normalization -- corrections that are critical for isolating the effect of completion design from the effect of geology and well interference.

What You Can Actually Analyze

By combining these sources, a completions team can build a dataset that includes, for each well: lateral length, proppant loading (lbs/ft), fluid intensity (bbl/ft), stage count, cluster spacing, landing zone, operator, vintage year, spacing unit, and 6/12/24-month cumulative production. For the Permian Basin specifically, there are enough wells to do statistically meaningful analysis at the formation level -- something that is not possible in most other basins.

The Six Parameters That Matter

Not all completion parameters are created equal. Some have strong, well-documented correlations with production; others are more nuanced. Here is what the data shows, ranked by the strength and clarity of their relationship to well performance.

1. Lateral Length

Lateral length is the single strongest predictor of total well production, and this is not surprising -- longer laterals contact more reservoir. The more relevant question is whether production scales linearly with lateral length, and the answer is: approximately, but not perfectly.

In the Midland Basin, average lateral lengths have climbed from roughly 7,500 feet in 2016 to over 10,000 feet in 2024, with the share of wells exceeding 11,000 feet growing from less than 1% in 2014 to approximately 25% by 2024. Delaware Basin laterals have historically been shorter due to surface constraints and lease geometries, averaging around 8,700 feet, but the gap has narrowed.

On a per-foot basis, production tends to be relatively flat up to about 10,000 feet of lateral length. Beyond that, some studies show a modest decline in per-foot productivity -- on the order of 5-10% for laterals in the 10,000-15,000 foot range. The reasons are partly mechanical (heel-to-toe flow distribution challenges, increasing friction pressure) and partly geological (longer laterals are more likely to traverse different rock quality along their length).

2. Proppant Loading (lbs/ft)

Proppant loading -- measured in pounds of proppant per lateral foot -- is the most debated completion parameter in the Permian Basin and the one where the data tells the clearest story about diminishing returns.

The industry trajectory has been steadily upward. In 2015, the average Permian well used approximately 1,200 lbs/ft. By 2018, that figure had risen to 2,250 lbs/ft. Current basin-wide averages sit around 1,800-2,200 lbs/ft, with some operators pushing 2,500-3,000 lbs/ft and a few experimental wells exceeding 5,000 lbs/ft.

What the data shows: Production increases with proppant loading, but the relationship is clearly non-linear. The steepest gains come between 1,000 and 1,800 lbs/ft. Between 1,800 and 2,500 lbs/ft, incremental gains are real but smaller. Above 2,500 lbs/ft, the data gets noisy, and the incremental production per additional pound of proppant drops significantly.

As one reserve evaluation firm CEO noted in reviewing the data: "3,000 pounds in my eyes isn't any better than 2,500 pounds." This observation aligns with what we see in multi-variable regression and machine learning models -- above approximately 2,500 lbs/ft, the marginal production gain from additional proppant rarely justifies the additional cost, which includes not just the sand itself but the incremental water, pumping horsepower, and time on location.

The Delaware Basin Wolfcamp sub-play, with an average proppant loading near 2,000 lbs/ft, has been estimated to have room for up to a 35% increase in proppant intensity before approaching its economic sweet spot -- suggesting that many Delaware operators are still under-loading relative to the technical optimum.

3. Fluid Volume (bbl/ft)

Fluid intensity -- barrels of fluid pumped per lateral foot -- correlates with production, but the relationship is less clean than proppant loading because fluid serves multiple functions: it creates fracture volume, transports proppant, and (in slickwater designs) creates complexity through natural fracture activation.

Typical Permian fluid intensities range from 40 to 80 bbl/ft. The trend has been toward higher fluid volumes, particularly in the Delaware Basin where natural fracture density and higher stress differentials favor slickwater systems over crosslinked gel.

The data suggests diminishing returns on fluid volume begin around 55-65 bbl/ft, though this is more formation-dependent than proppant loading. Bone Spring completions, which tend to be in more naturally fractured rock, often show a stronger fluid-production correlation than Wolfcamp completions, where fracture geometry is more controlled by the in-situ stress field.

4. Cluster Spacing

Cluster spacing -- the distance between perforation clusters within a stage -- has tightened significantly over the past decade. In 2015, 60-75 foot cluster spacing was common. By 2020, most operators had moved to 20-40 foot spacing, and by 2024-2025, aggressive designs use 15-25 foot spacing with limited-entry perforating to ensure uniform cluster efficiency.

The logic is straightforward: tighter cluster spacing creates more fracture initiation points, increasing the stimulated reservoir volume and drainage density. The data supports this -- tighter cluster spacing correlates with higher production, particularly in the first 12 months.

But cluster spacing interacts strongly with stage length and perforation design. Simply adding more clusters without increasing pump rate or using limited-entry techniques often results in uneven cluster efficiency, where a few dominant clusters take most of the fluid and proppant while others barely contribute. Fiber optic diagnostics (DAS/DTS) have made this problem visible, and the data from instrumented wells shows that cluster efficiency -- the percentage of clusters that meaningfully contribute to production -- ranges from 40% to 85% depending on the design.

5. Stage Count and Stage Length

Stage count is largely a function of lateral length and stage length. The trend has been toward shorter stages (150-250 feet, down from 300-400 feet a decade ago), which means more stages per well. A 10,000-foot lateral with 200-foot stages has 50 stages; with 300-foot stages, it has 33.

More stages generally correlate with better production, but the effect is partially confounded with cluster spacing -- shorter stages with the same number of clusters per stage effectively tighten the overall cluster spacing. Isolating the independent effect of stage length from cluster spacing requires careful multivariate analysis.

What the data suggests is that stage lengths below 200 feet in the Permian show minimal incremental benefit over 200-250 foot stages, while stage lengths above 300 feet consistently underperform relative to shorter-stage designs. The operational cost of more stages (more plug-and-perf cycles, more wireline runs) creates a natural economic floor on stage length.

6. Proppant Type

The Permian Basin has been a major driver of the shift from northern white sand and resin-coated proppant to locally sourced in-basin sand. In-basin sand mines in West Texas now supply the majority of proppant used in the Permian, significantly reducing logistics costs.

The production data comparing in-basin sand to northern white sand is surprisingly ambiguous. While laboratory crush strength tests show northern white sand outperforming in-basin sand at higher closure stresses, field production data does not consistently show a statistically significant difference in well performance -- likely because the closure stresses in many Permian formations are within the performance envelope of in-basin sand, and because the cost savings allow operators to pump more proppant per foot.

Some operators are innovating further. Chevron has deployed a proprietary lightweight proppant derived from refinery coke that reportedly improves oil recovery rates by approximately 15% compared to conventional sand. These specialty proppants are niche but point toward a future where proppant selection is optimized per-zone rather than standardized across a development.

Diminishing Returns: The Proppant Loading Sweet Spot

The concept of a "sweet spot" in proppant loading deserves its own discussion because it is the single most actionable finding from large-scale completion data analysis.

The physics are intuitive. At very low proppant loadings (below 1,000 lbs/ft), fractures close without adequate proppant packs, conductivity is poor, and production suffers. As proppant loading increases, fracture conductivity improves, the effective drainage area grows, and production increases. But at some point, the formation's ability to deliver hydrocarbons to the fracture network becomes the bottleneck -- not fracture conductivity. Beyond that point, additional proppant does not meaningfully increase production because the reservoir is already well-connected to the fracture system.

The data from thousands of Permian wells places this inflection point in the 2,000-2,500 lbs/ft range for most Wolfcamp and Bone Spring targets. Specifically:

  • Below 1,500 lbs/ft: Production is clearly sub-optimal. Wells in this range consistently underperform peers.
  • 1,500-2,000 lbs/ft: Strong production response to increasing proppant. This is where the steepest part of the curve sits.
  • 2,000-2,500 lbs/ft: Still positive returns, but the slope is flattening. The incremental cost of proppant and associated fluid/pumping is approaching the incremental production value.
  • 2,500-3,000 lbs/ft: Marginal returns. Some formations and operators show small gains; others show no statistically significant improvement over 2,500 lbs/ft.
  • Above 3,000 lbs/ft: The data is sparse and noisy, but the available evidence does not support consistent production improvement. Operational risks (screenouts, higher treating pressures) also increase.

The economic optimum depends on commodity prices, proppant costs, and individual well economics. At $70/bbl WTI, the economic sweet spot for most Permian operators sits around 2,000-2,200 lbs/ft. At $80+/bbl, pushing to 2,500 lbs/ft becomes more defensible. At $60/bbl, pulling back to 1,800 lbs/ft may be optimal.

This is exactly the kind of analysis that a completions team should be running against their own data -- not basin averages, but operator-specific, formation-specific curves that account for their actual costs and well performance.

Basin-Specific Insights: Delaware vs. Midland

The Permian Basin is not one basin -- it is two geologically distinct sub-basins (Delaware and Midland) separated by the Central Basin Platform, each containing multiple target formations with different rock properties, stress regimes, and optimal completion designs. Treating them as interchangeable is a common and costly mistake.

Geological Differences That Drive Completion Design

The Delaware Basin is deeper (16,000 feet to the top of the Strawn, compared to 9,500 feet in the Midland Basin), which means higher temperatures, higher pressures, and higher closure stresses. The Wolfcamp formation in the Delaware is thousands of feet thicker than in the Midland, with different organic richness profiles and mechanical properties across the A, B, and C benches.

The Midland Basin has shallower targets, lower drilling costs, and a longer history of horizontal development. The Spraberry formation (Midland Basin) and the Bone Spring formation (Delaware Basin) are time-correlative but have different compositions, requiring different completion approaches.

Wolfcamp A: Delaware vs. Midland

In the Midland Basin, the Wolfcamp A is the most prolific target, with predictable mechanical properties and well-established completion designs. Average proppant loading sits around 2,000-2,200 lbs/ft, with relatively uniform cluster spacing designs across operators.

In the Delaware Basin, the Wolfcamp A is deeper, hotter, and more mechanically heterogeneous. The formation has higher clay content in some areas, which affects fracture containment and proppant embedment. Delaware Wolfcamp A completions tend to use higher fluid volumes (55-70 bbl/ft vs. 45-60 bbl/ft in the Midland) and have shown a stronger response to increasing proppant loading -- suggesting that many Delaware operators have not yet reached the diminishing returns threshold that Midland operators hit several years ago.

Wolfcamp B

The Wolfcamp B is generally more carbonate-rich and more brittle than the Wolfcamp A, which makes it more amenable to complex fracture network creation but also more prone to fracture height growth into adjacent zones. Completion designs in the Wolfcamp B tend to use slightly lower fluid volumes and rely more on limited-entry techniques to control fracture height.

Production data shows the Wolfcamp B generally underperforms the Wolfcamp A on a per-foot basis, but the gap has narrowed as operators have optimized landing points and completion designs. The key insight from the data is that Wolfcamp B performance is more sensitive to landing zone precision than the Wolfcamp A -- a 50-foot difference in landing depth can mean a 20-30% difference in production.

Bone Spring

The Bone Spring (primarily a Delaware Basin target) presents distinct completion challenges. The formation is shallower than the Wolfcamp, with lower closure stress and more natural fracture complexity. The 2nd and 3rd Bone Spring intervals have been the primary horizontal targets.

Recent data (2023-2024) shows concerning trends in Bone Spring productivity. Wells drilled between 2023 and 2024 showed a 12% decline in productivity per lateral foot compared to those drilled between 2021 and 2022, with breakeven prices rising by 16%. This decline is partly attributed to spacing issues (parent-child interference) and partly to inventory quality -- operators are moving into less prospective acreage as core inventory is developed.

Completion designs that perform well in the Bone Spring tend to be higher-fluid, moderate-proppant (1,800-2,200 lbs/ft, 55-70 bbl/ft), reflecting the formation's natural fracture complexity and lower closure stress. The data suggests that proppant loading above 2,200 lbs/ft in the Bone Spring shows minimal incremental benefit -- a lower threshold than the Wolfcamp, consistent with the lower closure stress environment.

Extended Laterals: What 3- and 4-Mile Wells Are Teaching Us

The trend toward longer laterals is one of the most significant shifts in Permian Basin development strategy. Three-mile laterals (approximately 15,000 feet) became common in 2023, and operators are now testing 4-mile (21,000+ foot) laterals. The economics are compelling on paper: a single 3-mile lateral replaces two 1.5-mile laterals, eliminating one wellbore's worth of drilling cost, surface equipment, and pad infrastructure.

But the completion data from these extended laterals reveals challenges that are not yet fully solved.

The Per-Foot Productivity Question

Comparing 2-mile and 3-mile laterals in comparable rock with similar completion designs, some analyses show a 10-20% degradation in productivity per lateral foot for the longer wells. This degradation comes from multiple sources:

Flow distribution: In a 3-mile lateral, the pressure differential between the heel and toe during production is significant. Heel stages produce at higher drawdown than toe stages, leading to faster depletion near the heel and under-drainage at the toe. This effect increases with lateral length.

Completion uniformity: Achieving uniform stimulation across 60-80 stages in a 3-mile lateral is mechanically harder than across 30-40 stages in a 1.5-mile well. Plug integrity, ball seat reliability, and wireline operations all become more challenging at greater distances from surface.

Geological variability: A 3-mile lateral traverses three times as much geology as a 1-mile lateral. The probability of encountering unfavorable rock -- high-water zones, depleted parent intervals, faulted rock -- increases proportionally.

Completion Scaling: What Does Not Scale Linearly

The data shows that simply scaling a proven 2-mile completion design to 3 miles does not produce 1.5x the production. Operators who are succeeding with extended laterals are making specific design modifications:

  • Variable completion intensity along the lateral: Using higher proppant loading and tighter cluster spacing in the toe section to compensate for the production disadvantage of being farther from the heel.
  • Inflow control: Some operators are experimenting with inflow control devices (ICDs) to balance production along the lateral and reduce the heel-toe flow imbalance.
  • Real-time completion adjustments: Using fiber optic diagnostics to monitor cluster efficiency during pumping and adjusting parameters (rate, fluid type, proppant concentration) in real time.

Operational Innovations

Extended laterals have driven adoption of simulfrac and trimulfrac completion techniques, where two or three wells on the same pad are fractured simultaneously. Chevron reports that 50-60% of its 2025 Permian wells will be completed using triple frac, up from roughly 20% the prior year. Matador Resources has demonstrated a 25% reduction in completion time and $1.1 million in savings per well using trimulfrac operations compared to conventional zipper frac methods.

These techniques increase pumping utilization (more pumping hours per day, less idle time), which partially offsets the higher per-well cost of more stages in longer laterals. Average capital cost savings with trimulfrac are approximately $125,000/well compared to simulfrac and $525,000/well compared to zipperfrac, assuming equivalent lateral length.

Where Machine Learning Outperforms Empirical Rules

The conventional approach to completion optimization in the Permian is empirical: plot proppant loading vs. production for your offset wells, eyeball the trend, pick a design, and pump it. This approach works reasonably well in homogeneous, well-understood acreage. It fails when geology varies, when spacing is uneven, or when multiple parameters interact in non-linear ways.

Machine learning methods can outperform empirical rules in several specific ways.

Multi-Variable Interaction Effects

Empirical analysis typically examines one parameter at a time: proppant loading vs. production, fluid volume vs. production, cluster spacing vs. production. But these parameters interact. The optimal proppant loading depends on cluster spacing, fluid volume, rock properties, and well spacing -- simultaneously. A machine learning model (gradient-boosted trees, neural networks, or even well-designed multivariate regression) can capture these interaction effects, identifying that, for example, 2,200 lbs/ft is optimal at 25-foot cluster spacing in the Wolfcamp A but 2,500 lbs/ft is optimal at 35-foot spacing.

Controlling for Confounders

The biggest challenge in completion data analysis is confounding variables. Operators drill their best acreage first, so earlier-vintage wells tend to be in better rock. Operators with higher proppant loadings may also have better acreage positions. Well spacing changes over time. Without controlling for these confounders, correlations between completion parameters and production can be misleading.

Machine learning models can incorporate geological features (log-derived porosity, GR thickness, distance to nearest offset producer, spacing unit area) as control variables, isolating the effect of completion design from the effect of rock quality and spacing. This is where ML analysis produces fundamentally different (and more actionable) conclusions than simple cross-plots.

EUR Prediction

For production forecasting, traditional decline curve analysis (DCA) using Arps or stretched exponential models requires 12-24 months of production history before producing reliable estimated ultimate recovery (EUR) estimates. Machine learning models trained on completion parameters, geological features, and early production data (30-90 days) can produce EUR estimates that, while less precise than mature DCA, provide directional guidance months earlier -- enabling faster feedback loops for completion optimization.

Research has shown that the stretched exponential production decline (SEPD) model combined with Arps provides the most accurate EUR prediction for Permian shale oil wells with more than 2 years of production history. ML models complement this by providing earlier estimates when production data is sparse.

What ML Cannot Do

Machine learning is not magic. Models trained on Permian completion data have real limitations:

  • They require data quality. Garbage in, garbage out. If lateral lengths are wrong, if production is mis-allocated between wells on the same pad, if vintages are not correctly assigned, the models will learn noise.
  • They do not explain physics. A gradient-boosted tree can tell you that production increases with proppant loading up to a point and then flattens. It cannot tell you why -- whether it is fracture conductivity saturation, limited reservoir contact, or proppant embedment. Physics-based understanding is still required to extrapolate beyond the training data.
  • They are sensitive to spatial bias. Permian data is spatially concentrated in core development areas. Models trained on Midland Basin data may not transfer to the Delaware Basin without retraining.
  • They cannot replace subsurface understanding. The best ML models for completion optimization incorporate geological features as inputs. Teams that treat ML as a black box without integrating subsurface knowledge will produce models that overfit to historical data and fail to generalize.

Actionable Recommendations for Completions Teams

Based on the patterns in the data, here are concrete steps that completions teams can take to improve outcomes.

1. Build Your Own Completion Database

Stop relying on basin averages and vendor-provided "benchmarking." Build an internal database that links your completion designs to your production outcomes, normalized for lateral length, vintage, spacing, and landing zone. The Texas RRC provides your production data for free. FracFocus provides your completion volumes. Combine them with your internal AFE data for stage count, cluster spacing, and proppant type.

If you operate 50+ wells in a given formation, you have enough data to do statistically meaningful analysis. If you operate fewer, partner with a data analytics provider (Novi Labs, Enverus, or an independent consultant) to benchmark against the basin while protecting your proprietary data.

2. Test the Proppant Loading Curve -- With Your Data

Do not assume that the basin-average proppant loading sweet spot applies to your acreage. Plot your own data. If you see a clear flattening of the production response above a certain threshold, you have an opportunity to either reduce costs (if you are above the threshold) or increase production (if you are below it).

Design deliberate experiments: on your next multi-well pad, vary proppant loading across wells (e.g., 1,800, 2,200, and 2,600 lbs/ft) while holding other parameters as constant as possible. This controlled experimentation generates far more insight than analyzing historical data where everything changed simultaneously.

3. Account for Delaware vs. Midland Differences

If you operate in both sub-basins, do not use the same completion design. The data consistently shows that the Delaware Basin -- particularly the Wolfcamp A -- responds more strongly to higher proppant loading and higher fluid volumes than the Midland Basin equivalent. The Bone Spring responds differently than the Wolfcamp. Design by formation and by basin.

4. Re-Evaluate Extended Lateral Completion Designs

If you are drilling 3-mile laterals, do not simply scale up your 2-mile design. Analyze your production data on a per-foot, per-stage basis along the lateral. If heel stages are outproducing toe stages by more than 20%, consider variable intensity completions (higher loading at the toe) or inflow control technologies.

5. Invest in Fiber Optic Diagnostics -- Selectively

Distributed acoustic sensing (DAS) and distributed temperature sensing (DTS) provide direct measurements of cluster efficiency and fluid distribution that no amount of production data analysis can replace. You do not need fiber on every well. But instrumenting 2-3 wells per year in each formation you develop will calibrate your understanding of what your completion designs are actually doing downhole, and whether your clusters are actually contributing uniformly.

6. Start With Structured Analytics Before ML

Machine learning is powerful but not necessary for every operator. Start with structured multivariate analysis: multiple regression with completion parameters and geological controls. If your team can build a regression model that explains 60-70% of the variance in your well performance with 5-7 variables, you already have actionable insight. Move to ML methods when you have enough data (200+ wells) and when the interaction effects between variables are too complex for linear models to capture.

7. Use Public Data to Benchmark Competitors

Your competitors' completion designs are public information. FracFocus tells you how much proppant and fluid they pump. The RRC tells you how their wells produce. You can reconstruct their completion intensity (lbs/ft, bbl/ft) and compare it to their production outcomes. If a competitor is consistently outperforming you in the same formation, their public data can tell you whether the difference is in completion design or acreage quality.

The Bottom Line

The Permian Basin's public data record is one of the most valuable -- and underutilized -- resources available to completions engineers. The patterns in the data are clear: proppant loading has a sweet spot around 2,000-2,500 lbs/ft for most formations; the Delaware and Midland sub-basins require different designs; extended laterals demand more sophisticated completion approaches; and machine learning can extract insights that empirical analysis misses.

But the data also shows something more fundamental: the operators who are winning in the Permian are not the ones with the most aggressive completion designs. They are the ones who treat completion design as an optimization problem -- measuring, analyzing, experimenting, and iterating with discipline. In an era where "industrialized stability" is replacing growth-at-all-costs, completion efficiency is one of the last remaining levers for creating real economic value.

The wells have already been drilled. The data already exists. The question is whether your team is using it.

Dr. Mehrdad Shirangi is the founder of Groundwork Analytics and holds a PhD from Stanford University in Energy Systems Optimization, with research focused on data-driven decision-making under uncertainty in reservoir and completions engineering. He has been building AI solutions for the energy industry since 2018. Connect on X/Twitter and LinkedIn, or reach out at info@petropt.com.

Related Articles

Have questions about this topic? Get in touch.

Back to Insights