Manufacturing · Operations analytics & ERP
Water plant ERP — Production ERP and operations analytics for a water plant
A plant is one line, one shift and one number that matters
A bottling plant looks simple from the outside: water goes in, pallets go out. Inside, every shift spends preforms, caps, labels, film, electricity and hours, and each of them lands in the cost of a single bottle. We built the system that keeps that line honest — production booked by shift, stock counted by scan, and a cost per unit that exists on the day it is produced.
Produced
38 640 bottles
Target 42 000
- Line running 6 h 12 m
- Stopped 48 m
- Rejected 310
Consumed this shift
- Preforms 38 950 310 rejected
- Caps 38 700
- Labels 38 660
- Shrink film 104 kg
Project facts
- Project type
- Client project
- Industry
- Manufacturing
- Platforms
- Web system · Shift terminal · Management dashboards
- Languages
- Azerbaijani · Russian
Before The challenge
The plant knew how much it made. Not what it cost to make it.
Output was counted — that part is hard to lose. Everything that produced it was not: the preforms taken from the store, the film used on the pallets, the half hour the line stood still, the crate of rejects nobody booked. At the end of the month a total appeared, and nobody could take it apart.
- Shift log Output written by hand at the end of the shift
- Store notebook Material issued, sometimes recorded, sometimes not
- Maintenance notes Stoppages and reasons, kept by the mechanic
- Delivery notes What went out, filed after the truck left
- Accounting One monthly total with nothing underneath it
What it cost
-
Cost per unit was an average of everything
One number for the month, divided by everything produced, hid the shifts, lines and formats that were actually losing money.
-
Stock was found, not known
Preforms, caps, labels and film ran out on the shift that needed them, because the plant learnt it was low at the moment it stopped.
-
Downtime had no price
A stopped line was a note in a mechanic’s book. It was never converted into the bottles it did not produce and the people it still paid.
-
Rejects disappeared quietly
Material that became waste left the store like any other material, so the loss showed up only as a difference at stocktake.
Step 01 Production
The shift is the unit of truth
Everything the system knows is booked to a shift on a line: how long it ran, what it produced, what it consumed and where it stopped. Each is entered once, at the terminal beside the line, and every later number is built from those entries rather than retyped from them.
- 1 Shift opened
- 2 Materials issued
- 3 Line running
- 4 Output counted
- 5 Rejects booked
- 6 Shift closed
-
Output is counted against a target
The shift has a plan before it starts, so the difference is visible while the shift is still running rather than at the end of the week.
-
Every stop has a reason
Stoppages are classified — changeover, cleaning, breakdown, no material, no operator — because a stop without a reason can never be reduced.
-
Rejects belong to their shift
Bottles rejected before, during and after filling are booked with their cause, which is what makes material loss a number instead of a mystery.
-
Closing the shift closes the books
A shift cannot be closed with material unaccounted for, which is how the plant’s stock stays right without a monthly reconciliation.
Step 02 Stock
Preforms, caps, labels, film — counted by scan, not by memory
A bottling line eats consumables in a rhythm that is entirely predictable, right up to the moment nobody is watching. Raw material and finished pallets carry QR codes, issues and returns are scans, and the system knows which consumable will run out first and who supplies it.
What the stock knows
- What will run out first
- Consumption per thousand bottles is known per format, so cover is expressed in production days rather than in an abstract quantity on a shelf.
- Which supplier is behind it
- Every consumable carries its supplier, its lead time and its last prices, so a reorder is a decision with its own history attached.
- What is finished goods
- Pallets of bottled product are stock too: produced, palletised, labelled and reserved for a customer, each state a scan away from the last.
- What left and what came back
- Returnable pallets and crates are tracked like everything else, which is what stops them becoming an annual write-off.
Step 03 Cost per unit
What one bottle costs, on the day it is bottled
Cost per unit is the number a plant is steered by, and it is useless once it is a monthly average. The system builds it from the shift that produced the bottle: the material issued to that shift, the energy and hours it used, and the output it actually got out of them.
- Materials Preforms, caps, labels and film issued to the shift, at what they were bought for.
- Energy The line’s consumption for the hours it ran, rather than a factory bill spread evenly.
- Labour The crew on that shift, including the hours paid while the line stood still.
- Rejects Material that became waste, which raises the cost of every bottle that did come out.
- Output Good bottles produced — the only quantity the cost can honestly be divided by.
- Cost per bottle A figure per shift, per line and per format, comparable the next morning.
One shift, as the system shows it
-
38 640 bottles
produced against a target of 42 000 on a single shift.
Line L1 · sample data
-
48 minutes
of downtime, classified by reason and priced in lost output.
Line L1 · sample data
-
0.084 ₼
cost per bottle for that shift, comparable with every other shift.
Line L1 · sample data
Step 04 Control
The operator books, the approver approves, the owner sees
A plant is a small place where everyone knows everyone, which is exactly why the system is explicit about who can do what. Entry is fast and limited to what a role needs, spending goes through an approval, and every record keeps the name and the time behind it.
17 sections
-
- Shifts
- Output
- Stoppages
- Rejects
- Issues
- Returns
-
- Purchases
- Suppliers
- Acceptance
- Orders
- Dispatch
- Returns
-
- Cost per unit
- Payments
- Approvals
- Stock value
- Dashboards
What roles change
- The operator books the shift
- Output, stoppages, rejects and the material used — without prices, and without other shifts.
- The storekeeper moves material
- Acceptance, issues and returns by scan, with quantities and suppliers but not with costs.
- Procurement sees supply
- Cover in days, lead times, supplier prices and the purchase requests waiting for approval.
- Leadership sees the plant
- Cost per unit by shift, line and format, stock value, and a trail of who approved what.
Step 05 Analytics
Two shifts, the same line, a different cost
Once every shift carries its own materials, downtime and output, the interesting question stops being “what did the month cost” and becomes “why is Tuesday night more expensive than Tuesday morning”. That question has an answer in the data, and it is usually actionable.
- Preform 0.039 ₼
- Cap, label, film 0.020 ₼
- Energy 0.015 ₼
- Labour and rejects 0.010 ₼
Output against target, shift by shift
Negative shifts are the ones that missed the target. Each one opens onto its own stoppage list, which is where the reason lives.
- Cost per unit by shift and format
- The same line in the same week can produce at two different costs, and the difference is visible per shift instead of averaged away.
- Downtime by reason
- Which causes take the most minutes over a month, ranked by the output they cost rather than by how often they happen.
- Consumption per thousand bottles
- Material use normalised by output, which is what turns “we used more film” into “we used more film per pallet”.
- Procurement analytics
- Supplier prices, lead times and delivery reliability over time, so buying decisions are made against the plant’s own history.
Our role
What Atlas BIP did
-
Started at the line, not at the ledger
We followed a shift from opening to closing — what it takes from the store, what it produces, where it stops and what it throws away — and designed the entry points so that each of those is captured once, where it happens.
-
Built the ERP from scratch
Production, stock, purchasing, dispatch, approvals and reporting as one system on one data model, with roles and permissions designed in from the start.
-
Put the store and the line on scans
QR codes on raw material, finished pallets and store bays, with issues and returns booked to the shift that used them — because a plant that counts once a month is a plant that guesses for twenty-nine days.
-
Built a cost model per shift
Materials, energy, labour and rejects divided by good output, per shift and per format, which is what makes two shifts comparable and a change measurable.
-
Turned the records into the reporting
Cost per unit, downtime analysis, consumption per thousand and procurement analytics read the operational data directly, so there is one set of numbers in the building.
Engineering
How it was built
- System
-
- Web application
- Relational database
- Role-based access control
- Audit trail
- Plant floor
-
- Shift terminal
- QR labelling
- Scan-based issues
- Offline-tolerant entry
- Costing
-
- Per-shift cost model
- Material norms
- Downtime classification
- Analytics
-
- Operational reports
- Power BI dashboards
- Excel export
Questions and answers
Questions we are asked about this project
Do you connect to the machines themselves?
Where a line exposes counters, reading them is better than typing them, and we design for that. Where it does not, a terminal at the line with a two-field entry is still an enormous improvement over a notebook — and it can be replaced by a machine feed later without changing anything above it.
Is a small plant too small for this?
The opposite, usually. A small plant feels every wasted pallet, and it has nobody spare to reconstruct a month afterwards. What it needs is entry that costs seconds and numbers that are right the next morning.
How accurate is a cost per bottle really?
As accurate as the inputs allow, and we are explicit about which ones are measured and which are allocated. Materials and rejects are measured, energy and labour are allocated by the hours the line ran. That is enough to compare shifts, which is what the number is for.
Can it handle more than one line or format?
Yes — that is the point of booking everything to a shift on a line. Formats have their own norms and their own costs, and comparing them is one of the first things a plant asks the system once it trusts it.
Your project
Your system is next
Tell us how orders, money and documents move through your company today. We reply within 24 hours with questions and a concrete plan.
Shift codes, quantities and figures on this page are sample data used to show the interface. The plant is not named, and none of this is a client’s commercial information.
Back to the start