Build a cost-per-tonne model for a made-up iron-ore mine
Whether a restarted mine can make money depends on what it costs to dig, haul, process and ship each tonne, yet students rarely see how those costs stack up.
A spreadsheet model with every cost input labelled as made up, a cost-per-tonne result split by stage, and a one-page note on which input changes the result most.
This is a financial model project. Build a small model of costs, revenue or cash and test what changes the answer.
The approach: Write every assumption in its own labelled cell → Build the calculation from the assumptions → Change one assumption at a time → Say which assumption the answer depends on most.
Think first
Whether a restarted iron-ore mine in Goa can make money depends on what each tonne costs from the pit to the ship. You are building a small, honest model with made-up inputs to see how those costs stack up and which one matters most.
- Which cost stages will you include, from extraction to closure provision, and which will you leave out on purpose?
- Which costs vary with each tonne, and which are fixed whatever the output?
- How will you treat payments to the state, which depend on rules you should read but must not interpret as legal advice?
- Which single input do you expect to matter most, before you test it?
A workbook with an inputs sheet where every made-up figure is labelled, a calculation sheet split by stage, a cost per tonne result, a one-at-a-time sensitivity table, and a note naming the input the result depends on most.
Treating the made-up model as a real estimate for any actual mine. The point is to learn the structure and the drivers, and the note must say so plainly.
Gather
- The central mineral rules or state policy, to list which payments to the state exist (names only)
- A public description of the ore transport chain in Goa, so your stages match reality
- A list of your own made-up inputs, each labelled as practice figures
AI must not present any cost as a real industry figure. Every input is your own made-up number in a labelled cell, and the note says the model is for practice.
Build with AI
- 3.1 Challenge the cost stages firstCLAUDE CHAT
Why this tool: Claude is good at spotting a missing cost, such as closure provision or rehandling at a jetty, before you build.
Write your list of stages and ask Claude what is missing. Add only stages that fit the Goa transport chain you read about.
- 3.2 Set up the workbook from your inputsCLAUDE WORKING ON YOUR FILES (FORMERLY COWORK)
Why this tool: Claude working on your files can build the sheets, formulas and labels from your list in one go, so you can spend time checking rather than typing.
Put a file with your stage list and made-up inputs in a folder. Ask Claude to create an inputs sheet, a calculation sheet and a result, with every input referenced from its own cell.
Before you let it act: Before it builds, confirm the folder holds only your stage list and made-up inputs. After, click into five formulas and check each points to a labelled input cell, not a typed number.
- 3.3 Add the sensitivity tableGOOGLE SHEETS
Why this tool: A simple table of one change at a time shows clearly which input drives the result.
Copy the workbook into Sheets or keep it in Excel. For each input, raise it by a fixed step you choose, record the new cost per tonne, then reset it before testing the next.
- 3.4 Explain the driver in plain wordsCLAUDE CHAT
Why this tool: Claude can check whether your explanation of the main driver follows from the table.
Paste the sensitivity table and your draft note. Ask Claude to point out any claim the table does not support.
Check the AI's work
- Recompute the cost per tonne for one stage by hand and confirm it matches
- No formula contains a typed number; every input is a labelled cell
- Every input carries the label 'made-up practice figure'
- Each sensitivity test changed only one input and reset it afterwards
- No number typed inside a formula: every input is a labelled cell
- Recompute the headline result by hand
AI may quietly hard-code a number inside a formula or add a cost stage with a figure it assumed. Check every formula and delete any input you did not choose yourself.
Finish and show
- An inputs sheet with labelled made-up figures
- A calculation by stage and a cost per tonne result
- A one-at-a-time sensitivity table
- A note on the main driver, stating it is a practice model
Show the inputs sheet, the stage breakdown as a simple bar chart, and the sensitivity table, with a one-paragraph note underneath.
Explain which input surprised you as a driver and what that would mean for a real mine's managers, without claiming real numbers.
Which tool fits
- Claude chat: Claude is good at spotting a missing cost, such as closure provision or rehandling at a jetty, before you build.Free plan works; Pro gives longer conversations.
- Claude working on your files (formerly Cowork): Claude working on your files can build the sheets, formulas and labels from your list in one go, so you can spend time checking rather than typing.Paid plans. Rolled into ordinary chat from 16 Sep 2026; check what your plan shows.
- Google Sheets (free alternative): A simple table of one change at a time shows clearly which input drives the result.Free with a Google account.
- Claude for Excel: formulas, checks and explaining the modelPaid plans; needs Excel.
Add a second transport route, direct by road to the port, and compare the cost per tonne of both routes under the same inputs.
AI helps you think, check and polish. The work, and the judgement, stay yours.
This is a free sample. Full Access opens every project in full and lets you submit your finished work.
See Full Access →