Every workflow in the library, live
Seven tabs, one for each part of the library. Each runs a live calculation in your browser on real, public production and well records for horizontal wells in the Wattenberg field (Weld County, Colorado), from the Colorado Energy and Carbon Management Commission. Change the inputs and watch the result move. The library does the same work in Python on your own data.
Batch forecast with data cleanup
Six public horizontal wells forecast in one pass. Before fitting, each history is screened: down months are dropped, outliers are flagged, shut-ins of two months or more split the history into segments, and the fit starts at the peak of the current segment. Click a well to see what was flagged.
Public ECMC monthly oil, shown as bbl per producing day, log scale. Orange marks are flagged outliers; grey marks were not used in the fit.
| Well | Months | Flagged | Segment | qi | Di/yr | b | EUR, Mbbl |
|---|
Turn the cleanup off and watch the shut-in well's EUR move. That is why the library screens every history before it fits, and reports what it removed so an engineer can review the exceptions.
Arps decline: auto fit vs. your call
Eleven years of monthly oil from one public Niobrara well (API 05-123-41210). The dashed line is the automatic hyperbolic fit. The solid line is yours: change the initial rate, the decline or the b-factor and see what it does to recovery.
Public ECMC record. Rate in bbl/d from the peak month, log scale. Forecast to a 5 bbl/d limit or 30 years.
Your forecast here feeds tabs 03 and 05.
EUR as a range: MCMC sampling
A single best fit hides how uncertain the parameters are. A Markov Chain Monte Carlo sampler draws thousands of decline parameter sets that all fit the history, then turns each into an EUR. Here it starts with only the first 36 months, so you can check the range against the years of actual production that followed.
Same public well as tab 02. Industry convention: P90 is the low case, P10 the high case.
Slide the history from 6 months up to the full record and run it again. The range narrows as data comes in, and the held-back actuals show whether the range was honest. That is the question a reserves or A&D team needs answered before a number goes in a model.
Type curve, raw vs. lateral-normalized
Every public horizontal well in this set with three or more years of production, with completed laterals from about 3,300 to 13,800 ft. Raw, the long laterals look like better rock. Scaled to a common lateral length, the comparison is fair.
Public ECMC records, aligned on each well's peak month. Oil rate, bbl/d. Band shows P90 to P10 across wells.
From forecast to NPV, IRR and payout
This uses your forecast from tab 02. Change price, cost or well cost and trace it through the monthly cash flow. Switch to remaining life to value the same well as a producing asset from today, with no capital and no history replayed.
Forecast from the public well in tab 02. Prices and costs are illustrative inputs. Oil only, flat price, 7% production tax. Cumulative cash flow, $ million.
Well spacing and overlap from directional surveys
Ten wells drilled from one pad: seven Niobrara and three Codell laterals in the same section. The library finds each well pair's closest approach, splits it into horizontal and vertical offset, measures how far the laterals run side by side, and computes geometric overlap for a radius you choose.
| Pair | Closest, ft | Horizontal | Vertical | Side by side, ft | Overlap |
|---|
Lateral positions from the top-of-producing-zone and bottom-hole locations and TVDs in public ECMC records, drawn as straight lines. Geometric overlap only. It is not a reservoir or fracture model.
With directional survey files, the library screens every well pair in a unit or a whole field and returns pair detail, overlap summaries and gunbarrel tables you can chart in Spotfire, Power BI or a notebook.
What drives production? Regression and PCA
The same public wells, with lateral length, proppant and fluid per foot from their completion reports, TVD, and first-year oil. A regression shows how much each attribute explains, and which wells beat or miss the model. PCA shows which attributes move together.
Public ECMC records. First-year oil, Mbbl. Effects are standardized (per one standard deviation of each attribute).
| PCA loading | PC1 | PC2 |
|---|
Source: public production, permit and completion records from the Colorado Energy and Carbon Management Commission (COGIS database), Wattenberg field, accessed November 2026. Wells are shown by API number. This page runs standard decline, sampling, cash-flow, geometry and statistics math in your browser to show each workflow. It is a preview, not the library itself, and it is not a forecast or valuation of any well. Economic inputs are illustrative.
Everything inside the library
Ten Python modules. Every calculation can be inspected, tested and extended by your own engineers.
| Capability | What it does | Goes in | Comes out | Demo |
|---|---|---|---|---|
| Batch forecasting | A consistent first-pass decline workflow across a selected well set, with exceptions to review | Well IDs, production dates, oil and gas history, units, fitting settings | Rates, curves, cumulative volumes, model parameters per well | 01 |
| Production preprocessing | Drops nonpositive records, flags outliers, detects start, stop and production segments before fitting | Raw monthly production | Masks, segments, cleaned series | 01 |
| Arps decline, auto and manual | Automatic fitting, plus manual fits when engineering judgment calls for it | Production history, fit settings or chosen parameters | Forecast rates, cumulative volumes, EUR | 02 |
| MCMC sampling and EUR | Markov Chain Monte Carlo sampler with parameter bounds, so recovery is a range | Production history, bounds, proposal settings | Sampled parameters, sampled EURs, acceptance reporting | 03 |
| Type curves | Mean, median, percentile or sum type curves, with lateral-length normalization | Selected wells, lateral lengths, target length | Aggregated and fitted type curve, normalized comparison | 04 |
| Well economics | Revenue, LOE, variable costs, investment and tax into monthly cash flow | Forecasts, price deck, interests, costs, investment dates, tax | Monthly cash-flow tables, NPV, IRR, payout | 05 |
| Remaining-life economics | Values producing wells from an evaluation date without replaying their full lives | Forecast, evaluation window, costs | Go-forward cash flow and metrics | 05 |
| Well spacing and overlap | Directional-survey geometry, horizontal interval selection, well-pair screening and geometric overlap | Well IDs, latitude, longitude, TVD, survey headers, radii | Well-pair detail, distances, overlap summaries, gunbarrel tables | 06 |
| PCA and regression normalization | Explores how well attributes relate to a production or recovery target | Selected well attributes and a target | Loadings, predictions, residuals | 07 |
| Time series and discounting | Monthly alignment, shifting, arithmetic, discounting and date conversion behind the economics | Monthly series | Aligned, discounted series | 05 |
| Spotfire integration | Reads typed parameters from Spotfire document properties; Spotfire output adapters | Spotfire data functions | Tables back into Spotfire | – |
| Utilities and custom modules | Local caching, run logs, an optional SQLite helper, and well and model collections to combine and extend calculations | Your own workflows | Repeatable runs | – |
Outputs are pandas DataFrames, so results drop into Spotfire, Power BI, a notebook or an AI assistant's workflow.
How a purchase works
- 15-minute fit call. Tell us one engineering question you want to answer and what tools your team uses. We confirm your environment before you buy.
- Subscribe. An annual subscription for your team.
- Optional customer-data setup. Setup is separate from the subscription and available in prepaid 20-hour consulting blocks.
- Optional setup block. Want us to connect it to your own wells, production and price data? Add a 20-hour block.
- Every calculation open to inspect, for every module above
- Updates and support during your annual subscription
For pricing, contact our team. Most conversations start with a 15-minute call.
Contact our teamBefore you buy
Is this a black box?
No. Every calculation is open to inspect. Your engineers can read it, check it against their own methods and extend it.
Do we need Spotfire?
Not necessarily. It was built to run inside Spotfire, and its outputs are standard Python DataFrames you can use in notebooks, Power BI or with an AI assistant. We confirm your setup on the fit call, before you buy.
Is the demo on this page the library?
No. The demo runs standard math in your browser on public Colorado well records so you can see each workflow. The library does this in Python on your data, at the scale of your well set.
Does it replace ARIES, PHDwin or ComboCurve?
It is not a reserves system. It gives your engineers inspectable Python for the analysis around those systems: testing assumptions, putting ranges on EUR, comparing wells fairly, checking spacing and tracing a forecast into cash flow.
Is the spacing tool a fracture model?
No. It is survey geometry: distances, side-by-side length and geometric overlap for a radius you set. It helps you screen and compare well pairs quickly.
Who is behind it?
Blue River Analytics, Denver. Analytics for oil and gas since 2014, more than 300 clients and hundreds of Spotfire projects.