Skip to content

DEPRO Calculator

Calculate Daily Equivalent Production Offline (DEPRO) for an oil/gas well using the canonical AgentCraft formula and run a Monte Carlo uncertainty band.

Type: Script Domain: Drilling Vertical: Upstream Version: v3.2.1 Last tested: 2026-06-03 Cost per run: Free (Python stdlib only)

:material-github: Source on GitHub · Drive canon PDF · CHANGELOG

Trigger prompts

Run depro_calc.py on inputs.yaml for Well A-12.

Calculate DEPRO for the attached well events with 1000 MC runs and output JSON.

What's the DEPRO for Q1 2026? Use the events I pasted above.

What it produces

A text report (default) or JSON (with --json) showing:

  • Design rate, productive hours, downtime hours, availability %
  • Actual production (bbl over the window)
  • DEPRO (bopd) — the headline number
  • Monte Carlo P10/P50/P90 + mean ± stdev (if --mc-runs > 0)

Sample output — Well A-12, Q1 2026

============================================================
  DEPRO Report — Well A-12
============================================================
  Design rate:              4,200 bopd
  Analysis window:        2,160.0 hours
  Productive hours:      2,038.00
  Downtime hours:          122.00
  Availability:             94.35 %
  Actual production:    8,559,600 bbl
  Downtime events:              4
------------------------------------------------------------
  DEPRO:                 3,962.78 bopd
============================================================

  Monte Carlo (1000 runs):
    Mean:     3,962.70 bopd
    Stdev:        4.20 bopd
    P10:      3,957.08 bopd
    P50:      3,962.70 bopd
    P90:      3,968.27 bopd
    Range:  [3,952.82, 3,972.60]
============================================================

Inputs

  • --input <path> — YAML or JSON well events file (YAML shape: name, design_rate_bopd, analysis_window_hours, events: [...])
  • --mc-runs N — Monte Carlo runs (0 = skip, 1000 = recommended)
  • --seed N — MC random seed for reproducibility (default: 42)
  • --json — output machine-readable JSON
  • --output <path> — write XLSX (requires openpyxl; falls back to JSON)

Example input

name: "Well A-12"
design_rate_bopd: 4200
analysis_window_hours: 2160

events:
  - start: "2026-01-15T08:00:00"
    end:   "2026-01-17T14:00:00"
    reason: "ESP failure - pulled and replaced"
    error_hours: 2.0
  - start: "2026-02-21T11:00:00"
    end:   "2026-02-23T19:00:00"
    reason: "flowline leak - hot tap repair"
    error_hours: 3.0

Source files

File Path Version
Source PDF drive/00-canon/drilling/upstream/depro-calc/00-depro-formula-v3.2.1-2026-06-03.pdf v3.2.1
Canonical script canon/drilling/upstream/depro-calc/depro_calc.py v3.2.1
Example input canon/drilling/upstream/depro-calc/inputs.example.yaml v3.2.1
CHANGELOG canon/drilling/upstream/depro-calc/CHANGELOG.md v3.2.1
Public skill skills/skill-depro-calc/SKILL.md v3.2.1

Install

# No install — the script is stdlib only
git clone https://github.com/ah-alhawar/agentcraft
python3 agentcraft/canon/drilling/upstream/depro-calc/depro_calc.py \
  --input agentcraft/canon/drilling/upstream/depro-calc/inputs.example.yaml

Version + last tested

  • v3.2.1 — released 2026-06-03
  • Last tested: 2026-06-03 against the Well A-12 Q1 2026 events log. Result: 3,962.78 bopd DEPRO, 94.35% availability, MC P10/P90 = [3,957 / 3,968] bopd.

Known limitations

  • Naive datetime parsing (assumes UTC, no DST handling)
  • Single-well per run (multi-well aggregation is a future skill)
  • No timezone-aware input — convert all timestamps to UTC before analysis

← Drilling · Library · Home