Imported QA track1027 items

DISKOS-QA

This page exposes the imported DISKOS-QA track inside the FormationEval suite. It is kept separate from the evaluated MCQ benchmark, quiz mode and leaderboard. Provenance and licensing notes are shown per item and in the repository level notices.

March 2026 suite update

FormationEval now also includes the imported DISKOS-QA track. The public leaderboard and quiz still reflect the evaluated 505 question MCQ v0.1 track. A full rerun on the expanded suite is pending because this is a self funded one person project and QA evaluation requires materially more token spend.

If you want to collaborate, support reruns or discuss related research and engineering work, contact almaz.ermilov@gmail.com.

Showing 20 of 1027 QA items

Page 1 of 52

The evaluated public leaderboard still covers 505 MCQs and 72 models.

Rock and CoreformationSANDNES FMEntailed

What percentage of GECO Klinkenberg-corrected horizontal permeability measurements on 9/2-1 Sandnes Formation cores exceed 100 mD?

Reference answer

Less than 10%.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

10 . 2 POROSITY AND PERMEABILITY He-porosity and Klinkenberg-corrected horizontal permeability as measured by GECO on inch core plugs taken at approx. intervals are illustrated in histograms fig. and plotted against depth in fig. 4. Within the cored interval, mean values of porosity and permeability of the Sandnes s...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: SED_REPORT_1.PDF, 09-02-04-S_PB-706-1203.PDF, PETROPHYSICAL_1992-09-30_REPORT_1.PDF, GCH_1996-02-28_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_53dd17e25a52
Test fluid and pressureformationSOGNEFJORD FMEntailed

What are the implications of a 1.14 g/cm3 Draupne gradient for hydrostatic depth estimation and overpressure trend?

Reference answer

A 1.14 g/cm3 gradient indicates elevated pore pressure (overpressure) relative to normal hydrostatic, so hydrostatic depth is effectively shallower and an increasing overpressure trend exists toward the top reservoir.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

5.2 FMT and test summary 5.2.1 FMT results Prior to setting 5/8 casing, the formation multi tester (FMT) was run in hole to define reservoir pressure and pressure gradients. The pore pressure is equivalent to g/cc at the top reservoir (1561 RKB) decreasing to g/cc at the gas/water contact at approximately 1570 RKB....

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: COMPLETION_REPORT_2.PDF, 35-11-07_PB-706-0829.PDF, DRILLING_PROGRAMME_1.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_269d38564f8d
Test fluid and pressureformationNOT FMNot entailed

How do D-exponent, resistivity, and gas compare as pore-pressure indicators versus depth in well 2/7-30?

Reference answer

D‑exponent steadily decreases with depth indicating increasing pore pressure, while resistivity proved a poor pore‑pressure indicator. Gas partially supports the pore‑pressure trend (rose from 0.2% TO to 1.2% TO at 5700 ft, ~1.0% TO to 9440 ft, then fell to 0.04%–0.08% TO).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

Mud filtrate Cl concentration mg/kg Potassium ratio v/v decimal Quality Flag Medium Residual gas saturation v/v decimal Residual oil saturation v/v decimal Resistivity tool Flag LWD SXO determination method in gas Fixed SGR SXO determination method in oil Fixed SOR Shale Porosity v/v decimal Source of Mud Filtrate D...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: COMPLETION_REPORT_1.PDF, COMPLETION_REPORT.PDF, PETROPHYSICAL_REPORT_1.PDF, MUD_LOG_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_e8b8cd87c4f1
VSP - Well seismicwell2/7-29Entailed

What are the Ricker zero‑phase center frequencies (Hz) used for the 2/7-29 GEOGRAMs, listed in ascending order?

Reference answer

20, 25, 30, and 40 Hz.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

S 5.5 Synthetic Seismogram Table The Synthetic Seismogram Table summarizes all relevant data used to generate the GEOGRAM. Two way travel time (TWT), depth, velocity, density, magnitude of computed reflection coefficients for primaries and multiples together with attenuation coefficients are listed for all computed...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: VSP_PROC_1994-01-28_REPORT_1.PDF, COMPLETION_WORKOVER_1993-07-31_PROGRAMME_1.PDF, COMPLETION_GEOLOGICAL_1994-03-31_REPORT_1.PDF, WELL_TEST_1994-01-06_PROGRAMME_1.PDF
Question format: QA
formationeval_v0.2_diskos_vsp_well_seismic_527fac97bdbc
GeologyfieldTROLL BRENT BEntailed

For the 31/2-N-11 AH sonic, if Δtdrift is positive downhole, should we apply calibration, smoothing, and a -35.5 m bulk shift?

Reference answer

Yes; apply calibration and smoothing to correct positive Δtdrift and apply the -35.5 m bulk shift.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

Commercial In Confidence 1 Introduction On the 20th August 2005, Vertical Incidence Vertical Seismic Profile (VIVSP) was carried out for Norsk Hydro on the deviated well 31/2-N-11-H in the Troll field, offshore Norway. The objective of the survey was to obtain an accurate time depth profile and VSP image to tie to s...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: VELOCITY_REPORT.PDF, VSP_REPORT1.PDF, VELOCITY_REPORT_1.PDF, WL_FPR-GR-AUX_2A_3A_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_geology_bf0efc636065
VSP - Well seismicwell31/2-Y-14Entailed

For well 31/2‑Y‑14, list open‑hole cement placements with depth intervals, volumes, and plug/slurry IDs.

Reference answer

Plug#1: 1894–1700 m, 15.78 m3 SG cement; Plug#2: 1699–1523 m, 15.17 m3 SG cement; Open‑hole plug/slurry: 1590–1500 m, 8.24 m3 SG cement slurry.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

12 1/4” PILOT HOLE SECTION Introduction This section was drilled in one rock bit run with AutoTrak and MWD Triple Combo. F.I.T. /LOT. An F.I.T. was performed giving SG EQMW at casing shoe 1548m (1427.0m TVD) Lithology Claystone at top of the section and mainly sandstone with thin calcareous stringers after entering...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: MWD_REPORT.PDF, MUD_LOG_REPORT.PDF, COMPLETION_REPORT.PDF
Question format: QA
formationeval_v0.2_diskos_vsp_well_seismic_570cf4dfb9ed
Test fluid and pressurewell31/2-P-23Entailed

For well 31/2‑P‑23, upper DHSV control line at the tubing hanger: what were the test pressure (bar) and test volume (mL)?

Reference answer

600 bar and 100 mL.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

Page : 16 Tubular Data Casing specification Makeup Ratings Typical design factors Type Size Weight Grade Connections ID, drift Capasity Closed end displace- ment Torque (min/opt/max) Collapse Burst Tension Test Pressure (differential) Collapse Burst Tension Triax Inches lb/ft mm m3/m m3/m ft-lbs Bar Bar kN Bar Condu...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: DRILLING_PROGRAMME.PDF, COMPLETION_REPORT.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_cdb7734f5ca7
Rock and CoreformationSNADD FMEntailed

How do heavy-mineral-rich Snadd core densities influence porosity and water saturation (Sw) when using the S/S model and Archie?

Reference answer

Higher grain (matrix) densities from heavy‑mineral‑rich Snadd cores reduce density‑derived porosity and, when applied in the S/S model with Archie, produce higher calculated water saturation (Sw).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

8 PETROPHYSICAL EVALUATION SUMMARY The petrophysical evaluation of the well shows the primary and secondary objectives have low porosity, poor permeability and are wet. RFT's in the objectives recovered mud filtrate with some solution gas. The only hydrocarbons found are in gas sand at 2385 in the Snadd Formation. T...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: 7321-07-01_PB-706-0619.PDF, GCH_RAW_REPORT_2.PDF, GCH_1988-05-31_REPORT_1.PDF, 7728-02-01_PB-706-0549.PDF, DISCOVERY_EVALUATION_2001-08-15_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_3013776e254d
VSP - Well seismicwell7120/9-1Entailed

From GEINIDAS seismic lines and survey tie to well 7120/9-1 log, what are depth accuracy and static shift estimates?

Reference answer

Depth accuracy: up to ~4%; static shift: roughly 10 ms (one-way travel time).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

PRCGNOSIS Well 7120/9-1 was drilled to test sandstone reservoirs of Middle to Lower Jurassic age. The structure was an east - west trending anticline plunging gently to the east. To the north and west the structure terminated against downfaulted shales. The structure was wholely within the Hammerfest Basin. Explorat...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: MUD_LOG_1982-09-30_REPORT_1.PDF, SITE_SURVEY_REPORT_2.PDF, STRAT_REPORT_1.PDF, 7120-09-01_PB-706-0318.PDF, CORE_DESC_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_vsp_well_seismic_557ef7ba8867
Rock and CoreformationFANGST GPEntailed

For well 6507/8-E-1 AHT2, summarize porosity, water saturation (Sw), and KLOGH assumptions used in petrophysical interpretation?

Reference answer

Porosity was derived from bulk density (RHOB) with the Lange Formation log average 17.3% and Sw was calculated from the Archie equation with Lange average 71%; KLOGH (horizontal permeability) was derived from porosity–permeability relations specified in the “Full Field Update – Petrophysics” document.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

3.4 Processing Processing software: Geolog version Interpretation model: Statoil_Heidrun_model_2010_05_05 The general interpretation model is based on water saturation (SW) from the Archie-equation, porosity from bulk density (RHOB) and horizontal permeability (KLOGH) derived from the porosity-permeability relations...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: PETROPHYSICAL_REPORT_1.PDF, PETROPHYSICAL_1992-02-28_REPORT_1.PDF, GEOLOGICAL_2002-06-30_REPORT_1.PDF, CORING_REPORT_1.PDF, DRILLING_1990-07-10_PROGRAMME_1.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_bafb082adcdf
Test fluid and pressurewell2/8-F-18Entailed

For well 2/8‑F‑18 Valhall Ora at 3125 MD, what are the fluid‑fraction intervals, predicted phase, and pumped volume?

Reference answer

Fluid-fraction intervals: 634–637.5 min (oil) and 889.5–891.5 min (water); predicted phase: mixture of oil and water; pumped volume: 124.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

• Medium • 10 Low (5 Valid, Supercharged & Unstabilized) • Ungraded (6 Tight, No Seal, and Lost Seal) Gradient Interpretation Summary No meaningful fluid gradient can be concluded from the formation pressure survey: A mud gradient corresponding to density of g/cm3 was calculated from the hydrostatic pressure reading...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: FM_PRESS_RAW_RUN1D-1G_EWL_2019-12-06_REPORT_1.PDF, DRILLING_1995-01-23_REPORT_1.PDF, MWD_1999-01-25_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_230c93667aab
Rock and Corewell29/6-1Not entailed

How do the lithostratigraphic positions and formation tops of Brent, Heather, Balder, Statfjord, and Shetland GP compare in well 29/6-1?

Reference answer

From shallow to deep the succession is Balder Fm above Shetland GP, then Heather Fm, then Brent Fm (top ≈ 3800 m), and then Statfjord Fm (top ≈ 4340 m).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

-152- Page The first twelve metres of the Brent sand were drilled before coring began at 4219 after depth adjustment on strapping out of the hole. A small bed of dark grey siltstone occurred between 4211 in to 4213 in which had no cut or show. Detailed descriptions of the seven cores cut in the Brent are included at...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: 29-06-01_PB-706-0596.PDF, PETROPHYSICAL_1982-08-06_REPORT_1.PDF, DRILLING_REPORT_1.PDF, DRILLING_PROGRAMME_1.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_d470312cd47e
DrillingfieldTAMBAREntailed

How does the operability of the Deepsea Express cement head manual override compare to the hydraulic system on well 1/3-9?

Reference answer

The manual override was difficult to operate after the hydraulic system failed, indicating inferior operability compared with the hydraulic system.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

The shaker room on the Maersk Jutlander filled with steam from hot mud, making it difficult to see especially with safety glasses. This was partly caused by the crew leaving the doors open and disturbing the airflow pattern, of the HVAC system. A good proportion of the Drill-Reps time was spent on logistic issues, r...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: 01-03-09-S-JU8-AND-01-03-09-S-T2_PB-706-0588.PDF, DRILLING_2001-09-18_PROGRAMME_1.PDF, DRILLING_2001-05-31_PROGRAMME_1.PDF, DRILLING_2001-09-04_PROGRAMME_1.PDF
Question format: QA
formationeval_v0.2_diskos_drilling_8f0e2489d7f4
GeologyfieldATLAEntailed

What are the predicted thin-sand horizon depths and uncertainties (m; ms TWT) in Mariner Formation at well PWL 25/5-7?

Reference answer

199 m ± 3 m (245 ms TWT) and 227 m ± 4 m (274 ms TWT).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

Quaternary The Quaternary geology throughout the survey area is expected to comprise four formations. The stratigraphy prognosis at the PWL is presented in Table 3.4. Table 3.4: Unit name Formation name Depth in metres (MSL) Depth in ms (TWT) Lithology Unit Undifferentiated Holocene and late Pleistocene 119m ±1m– 13...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: SITE_SURVEY_REPORT_1.PDF, VSP_REPORT_2.PDF
Question format: QA
formationeval_v0.2_diskos_geology_8e69363f565b
Test fluid and pressurewell34/7-16Entailed

What were the TDS (mg/l) and produced volume when well 34/7-16 R reached clean formation water?

Reference answer

Approximately 36,500 mg/l TDS and 300 Sm3 produced.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

- E9 - NET CONFINING PRESSURE MEASUREMENTS The samples, 100% saturated with simulated formation water, were mounted in core holders specially designed for high hydrostatic pressures. Prior to the analyses the net confining pressure was set to 20 bar in order to displace water from the sample surface and to avoid lea...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: CORE_SCAL_1991-12-31_REPORT_1.PDF, DRILLING_1990-03-31_PROGRAMME_1.PDF, 34-07-16-AND-16-R_PB-706-0445.PDF, 34-07-C-03-H_PB-706-1991.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_3056363e5961
Test fluid and pressureformationLYSING FMNot entailed

Will a 1.75 sg mud and a 1.86 sg FIT (Cl 140 mg/L) overbalance Lysing Formation (1.61 sg) without causing fracturing?

Reference answer

Yes — both 1.75 sg and 1.86 sg exceed the Lysing Formation pore pressure of 1.61 sg; 1.86 sg is slightly below the observed breakdown at 1.88 sg, so fracturing is unlikely.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

6406 - follow the trends established in the original well but the effects of the fluid migration as result of crossflow from 4596m to 3970m in the original well were unknown. Based on data from the kick the pore pressure gradient at 4596m was estimated to be sg and the possibility of fractured and/or charged fonnati...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: MUD_LOG_1997-03-25_REPORT_1.PDF, COMPLETION_REPORT.PDF, DISCOVERY_EVALUATION_2000-11-02_REPORT_1.PDF, DISCOVERY_EVALUATION_2000-12-31_REPORT_1.PDF, COMPLETION_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_767d7e91532a
Rock and CorefieldVEST EKOFISKEntailed

Does a gas composition of 82–85% methane with low N2 indicate acceptable gas quality, compositional uniformity, and methane as the primary constituent?

Reference answer

Yes — 82–85% methane with low N2 indicates high-quality, compositionally uniform gas with methane as the primary constituent.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

Petroksen Reservoir Engineering *AMPLE DEPTH PERMEA•ILITY POROSITY PER CENT PORK SAMPLE DESCRIPTION HUMMER PEET MILLIDARCYS PER CENT OR. TOTAL WATER AND REMARKS Core No.3 Continued • 10,340 Ls, wh chalky, gr, hd, re patches? soln cays. 41 AA re patches, fine soln cays. 42 AA sl coarser, weak styl develop. 43 AA re p...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: 02-04-5X_PB-707-1405.PDF, 02-04-5X_PB-707-1408.PDF, 02-04-D-15-A_PB-706-2131.PDF, 02-04-D-03_PB-706-1801.PDF, 02-04-5X_PB-707-1406.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_32e310c86f2b
Rock and Corewell6506/11-3Entailed

For well 6506/11-3, within the standard porosity workflow, which techniques were used to obtain grain volume and bulk volume?

Reference answer

Grain volume was measured by Boyle’s law porosimeter using helium; bulk volume was obtained by immersion in mercury.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

D. SAMPLE CLEANING The plugs were placed in refluxing soxhlet extractors. Tolueno was used to remove any residual hydrocarbons, and methanol was used to remove the water anc any residual salts. Chemical and visual checks were made to ensure all the pore contaminants hid been removed. The clean samples were dried in...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: CORE_CONV_1992-10-31_REPORT_1.PDF, GCH_1993-02-02_REPORT_1.PDF, MUD_LOG_1992-09-10_REPORT_1.PDF, CORE_PETROGRAPHY_1995-01-18_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_rock_and_core_61519c487321
PetrophysicsformationRANNOCH FMEntailed

What MDT pressure points were recorded and how many valid measurements were obtained for well 34/10-A-46?

Reference answer

44 MDT pressure points were recorded, of which 41 were valid measurements.

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

19 of 88 5 Table : Geological formation tops, 34/10-A-46 T2 Formation /group Actual MD RIO Actual mTVD RICH Prognosed MD RICB Prognosed mTVD RICH Deviation TVD Hordaland gp early Eocene 040,2 356 350 979 Drake formation 287 315 925 -16 Broom formation 563,5 984,2 Rannoch fm, member 574,2 985,9 Rannoch fm, member 631...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: COMPLETION_WORKOVER_REPORT_1.PDF, COMPLETION_REPORT.PDF, DRILLING_PROGRAMME_1.PDF, WELL_TEST1-3_REPORT_2.PDF, DISCOVERY_EVALUATION_REPORT_1.PDF
Question format: QA
formationeval_v0.2_diskos_petrophysics_55f33f37094a
Test fluid and pressurewell35/3-1Entailed

What are the primary purposes and expected diagnostic indicators when choosing clean-out cement plus 3 m new hole versus staged pump/pressure buildup?

Reference answer

Clean-out cement + 3 m new hole: exposes and isolates the formation by removing cement for a valid test; diagnostics are successful 3 m penetration, hole condition and absence of cement at the shoe. Staged pump/pressure buildup: determines leak‑off/fracture and safe working pressure; diagnostics are pressure‑vs‑volume build‑up, bleed‑off, surface holding pressure and sudden volume increases (mud take).

Shown as provided by the upstream DISKOS-QA benchmark.

Context preview5 snippets

FORMATION PRESSURE TEST After drilling out casing strings the formation below the shoe will be subjected to pressure leak off test. 1. Depending on the type of formation and estimated future drilling conditions this test will be run either to predetermined pressure or to actual bleedoff. If predetermined maximum pre...

Full supporting snippets load on demand to keep the page lighter.

Benchmark provenance

DISKOS-QA benchmark

License note: NLOD 2.0 (as stated in upstream README)

Open source link

Underlying corpus provenance

Large Oil and Gas industry text dataset from Norwegian, UK and Dutch public oil and gas documents

License note: CC BY 4.0

Open source link
Document: 35-03-01_PB-707-715.PDF, DRILLING_1976-05-31_PROGRAMME_1.PDF, 35-03-01_PB-707-728.PDF, 35-03-01_PB-706-0453.PDF
Question format: QA
formationeval_v0.2_diskos_test_fluid_and_pressure_2c58047c2f3d
...

About the maintainer

Former petrophysicist, now full time software engineer. Current work and research focus on LLM transparency, control and security in high hazard industries.

If you want to collaborate, support reruns or discuss related research and engineering work, contact almaz.ermilov@gmail.com