Nuve Recipe

Overview

Nuve Recipe provides recipe (formula × routing × process parameters) master data management and production batch / lot traceability for process industries (food, chemicals, activated carbon, cosmetics, detergents and so on). It is an independent recipe layer that sits between Oracle ERP (material/finance master), Asprova APS (planning) and shop-floor execution systems (MES): the ERP feeds the formula, Nuve Recipe owns the routing, the two are merged here, and the planning/execution systems take that combined recipe as input.

Key Concepts

• Recipe — the combination of a product's formula (BOM) and routing components; versioned and passed through an approval flow.
• Routing & Operation — production steps, precedence relationships (one operation cannot start before another finishes) and resource/work centre assignment.
• Process parameter — the temperature/pressure/time/ratio standard of an operation: target value plus min/max limits. In a process industry the recipe is not just the formula but this regime (for activated carbon: carbonisation at 550 °C, activation at 900 °C ± 25, steam ratio 1.2 kg/kg). If parameters marked critical (CPP) go out of spec, the batch cannot be completed.
• Variant & output fraction — the static model of the product fractions that come out of one raw input. The fraction set is not fixed; it is defined per industry: mesh for activated carbon (4x8, 8x30, 12x40, 30x60, PAC, FINES), solid/flake for food.
• Calibre distribution — the dynamic, batch/vessel-level data entered by planning showing what percentage came from each variant; combined with the variant catalogue it produces a weighted production projection.
• Version & approval — the recipe's Draft → Submitted for Approval → Approved → Published state transitions and the validation checks (missing resource, circular precedence, duplicate variant/fraction, inverted parameter limit, mass balance).
• Content lock — recipe content (formula, routing, operation, process parameter, variant) can only be changed in the Draft state. It freezes once submitted for approval, so that the standard used for batches already produced against that version is not retroactively altered. Commercial fields such as list price and sales unit remain editable at all times.
• Revision — to change a published recipe you open a New Revision: the formula tree, routing/operations (including the precedence chain), process parameters and variant fractions are copied verbatim into a new draft version. When the new version is published the previous ones are automatically retired — a product code has exactly one published version at a time.
• Change trail — which field was changed by whom, when, and from what to what. Version & Approval only shows state transitions; this tab shows content changes, and both the recipe and the batch have their own trail.
• Concurrent edit protection — every record carries a version counter. If two people keep the same recipe/batch open and save, the second one gets a "someone else changed this" warning; nobody's work is silently overwritten.
• Trash — deletion is a two-step process: the recipe/batch first goes to the trash (and can be restored), permanent deletion is a separate step. A recipe that has batches, and a batch whose output lot has been consumed in another batch, can never be deleted.
• Resource catalogue — a shared work centre/machine list, independent of any recipe.
• Production batch — a single execution of one recipe version. The recipe is "what should happen", the batch is "what happened". When a batch is opened the recipe skeleton is copied (input items, operation steps, parameters to be measured and the spec limits in force at that moment) — so even if the recipe is revised later, the standard the batch was produced against stays intact.
• Lot & genealogy — the lot number of every input and output item. Semi-finished items are linked to their source batch to build the chain; you can then go backwards from a shipment lot to every raw material lot, and forwards from a raw material lot to the affected shipment lots (recall impact analysis).
• Yield — accepted output / input that turned into product (%). Utility inputs such as steam, natural gas and wash water are recorded as consumption, but the "Counts toward yield" box is left unticked so that they do not inflate the denominator. This flag is set once on the recipe's formula line and carried from there to the batch.
• Unit of measure (UoM) and dimension — every unit belongs to a dimension (mass, volume, count, length, time, energy) and has a conversion factor to that dimension's base unit (1 TON = 1000 KG, 1 BB = 1 EA). Quantities are normalised to the base unit when saved; yield and mass balance are calculated in the mass dimension only. That way 320 SM³ of natural gas and 1,000 kg of char are not added into the same total; other dimensions are reported on a separate line as "Consumption outside yield".
• Product quality specification — the properties the finished product must meet (iodine number, hardness, ash, moisture) and their limits. The difference from a process parameter: that one says "how do we run the furnace", this one says "what must come out". If no fraction is selected, the property applies to every output fraction (wildcard row).
• Lot quality status — for an output lot: In spec (all measurements present and within limits), Out of spec (a critical property is outside its limit), Not measured (at least one measurement missing). Conformity is never declared with a missing measurement.
• Certificate of Analysis (CoA) — the conformity document issued to the customer for an output lot. It can only be produced for lots that are In spec and dispositioned Accept; a lot dispositioned scrap/rework, or with a missing measurement, cannot be certified. Properties with the "On CoA" flag turned off (internal control values) do not appear in the document.

Typical Workflow

Defining the recipe

1. Open Nuve Recipe from the left-hand menu; pick an existing recipe from the list or create a New Recipe. You can also load a ready-made example data set with the industry selector plus Load Demo (tuna, activated carbon, chocolate, detergent, cosmetics, dairy).
2. On the Formula (BOM) tab, edit the product's multi-level tree (e.g. big bag → bulk product → activation output).
3. On the Routing & Operations tab, define the steps, precedence relationships and resource assignments. Press the thermometer button on an operation row to enter that step's process parameters (code, name, unit, target, min, max, critical).
4. On the Calibre Variants tab, enter the variants and output fractions. The columns are not fixed: with Add Fraction you define the codes that fit your industry (e.g. 8x30, PAC, FINES); the system calculates total output and yield for each row and flags it in red if the output exceeds the raw input.
5. On the Quality Specification tab, enter the properties the product must meet (property code, name, unit, target, min, max, test method). If you leave the fraction column blank, the property applies to all fractions. If a property marked critical is out of spec, that lot cannot be accepted; the On CoA flag determines whether it appears in the document.
6. On the Calibre Distribution tab, enter the batch-level percentage distribution; the system automatically calculates the weighted output projection per fraction.
7. On the Version & Approval tab, run the validation, take the recipe through the approval flow and publish it.

   Content is locked from the moment you submit for approval. If you spot a mistake, move the recipe back to Draft (from the Version & Approval tab), fix it and submit again.

8. Use the Asprova Bridge tab to preview the recipe for transfer to the planning system (JSON export).

Units are managed from Resource Catalogue → Unit Catalogue. The default set (KG, TON, G, M³, SM³, L, EA, BB, CASE, PALLET, COIL, M, MIN, HR, KWH) is installed automatically on first use; when adding your own unit you supply its dimension and conversion factor to the base unit. The base unit of a dimension, and any unit used on batch lines, cannot be deleted.

Changing a published recipe

A published recipe cannot be edited directly. To make a change:

1. Press New Revision in the recipe header; give a version number if you want (leaving it blank derives one automatically, e.g. 1.0 → 2.0).
2. Make the change in the draft version that opens — the content arrives copied verbatim.
3. Take the new version through the approval flow and publish it. The previous version automatically moves to Retired, and this is written into the version history of both versions.
4. The batches of the old version stay exactly as they are; the standard they were produced against does not change.

The Change Trail tab shows, for every save, which field was changed from what to what, by whom and (if entered) why.

Opening a production batch and traceability

9. Switch to the Production Batches tab at the top, choose a published recipe with New Batch, and enter the batch number and planned quantity. Input items, operation steps, parameters to measure and lot quality measurement rows are created automatically from the recipe.

   A batch can only be opened from a published recipe; starting production against a draft or pending-approval recipe counts as non-standard production.

   Input quantities are calculated by exploding the formula tree: the sub-tree of a component that produces an intermediate is walked with a required / produced factor, and the scrap allowance increases consumption. When you plan 1 big bag (500 kg of bulk product), the system applies the 500/220 factor and asks for 2,455 kg of char — you do not have to type the quantities. If the unit you plan in is not in the same dimension as the root product's unit, no scaling is applied and a warning is shown when the batch is created.

10. On the Input Lots tab, enter the raw material lots consumed (lot number, supplier, supplier lot, actual quantity). If an item is a semi-finished product, pick the batch that produced it from the Source Batch column — the genealogy chain is built through that link. For utility inputs such as steam/gas/water, leave the Counts toward yield box unticked.
11. On the Operation Record tab, enter the start/end time of each step; open the measurement rows with the thermometer button and record the measured values. Out-of-spec measurements are flagged in red immediately.
12. On the Output Lots tab, enter the fractions produced, their lot numbers and their disposition (Accept / Rework / Scrap). The accepted total determines the batch's actual quantity and yield. The Quality column on each lot row shows the lot's status; open the row and enter the laboratory measurements. Target/min/max values arrive as a snapshot from the recipe and cannot be changed — even if the recipe is revised later, the limits the lot was assessed against stay intact. The Download CoA button on an in-spec lot row produces the certificate of analysis as a PDF.
13. On the Validation tab, review the errors and warnings; the deviation table lists out-of-spec measurements and duration deviations above 10% side by side with the recipe target.
14. Move the batch through Released → In Production → Completed → Closed. If something goes wrong, the batch can be put On Hold rather than cancelled.

    If validation reports an error (time integrity, mass balance, an out-of-spec measurement on a critical parameter), the batch cannot be completed — it is blocked so the record is not frozen with incorrect data.

15. Open the genealogy from the Traceability tab: the Backward direction descends from a shipment lot to raw material/supplier lots, the Forward direction finds which batches a lot was consumed in. You can also enter by searching a lot number, independently of any batch.
16. The Batch Record (PDF) button in the batch header produces the complete record for an audit: input lots, operation times, process measurements, output lots, yield and validation findings in a single document.

Feature Reference

• CRUD for recipes, routings, operations, process parameters, resources, variants/output fractions and changeover/setup times
• Calibre distribution entry and weighted production projection per fraction
• Version/approval state machine, validation checks and audit trail
• Resource catalogue: shared work centre/machine list
• Changeover times: product-to-product transition time and cleaning/cooling record
• Production batch: skeleton derivation from the recipe (explode), input/output lot recording, operation actuals, process measurements, state machine (Planned → Released → In Production → Completed → Closed; plus On Hold/Cancelled)
• Yield & mass balance: accepted output / input that turned into product; utility inputs (steam, gas, water) can be excluded from the denominator per item
• Batch validation: time integrity (end ≥ start), mass balance, out-of-spec measurement (error on a critical parameter, warning otherwise), unmeasured critical parameter, missing lot number, unclosed operation, yield anomaly. A batch cannot be completed while errors remain.
• Deviation report: out-of-spec measurements and duration deviations above 10%, side by side with the recipe target
• Lot genealogy: two-way tracing (backward to raw material lots, forward to affected shipment lots), lot number search, cycle and depth protection
• JSON export preview for Asprova APS (formula + routing + operation + variant)
• Master data protection: content lock (a non-draft recipe freezes), revision copying, single-published-version rule, field-level change trail, concurrent edit protection (version counter), soft delete + trash + restore, product code × version uniqueness
• 5 staging tables for the Asprova/Cedriks integration bridge (actuals, shift, calendar/capacity, stock, plan output)
• Sales order sub-module: customer management, order/line recording, Turkish Central Bank rate snapshot, order state machine and price-quantity comparison by customer/product
• Unit (UoM) catalogue: dimension + conversion to base unit; quantities are stored normalised, yield and mass balance are calculated in the mass dimension only, other dimensions are reported separately
• Multi-level explosion: the sub-tree of a component producing an intermediate is resolved with a required / produced factor, the scrap allowance increases consumption, and intermediates are not double-counted as inputs
• Product quality specification and lot quality: fraction-specific plus wildcard property rows, lot measurements whose limits are snapshotted onto the batch, In spec / Out of spec / Not measured lot status; an out-of-spec measurement on a critical property blocks an accepted lot in validation
• Certificate of Analysis (CoA) and batch record: per-lot CoA and full batch record produced as PDF; the CoA shows only properties flagged "On CoA" and lot-numbered inputs that turned into product
• 77 API endpoints (/api/nuve-recipe/...), 29 tables; authorisation is restricted to the nuve_recipe:manage role

Industry Adaptation

The module is not tied to any one industry; concepts specific to process manufacturing are defined as data:

Need	Where it is defined
Temperature/pressure/time regime	Routing & Operations → process parameters
Separation into product fractions (mesh, calibre, fat content…)	Calibre Variants → output fractions
Product quality specification (iodine number, hardness, ash, moisture, viscosity…)	Quality Specification → property rows
Industry-specific units of measure (big bag, SM³, case, coil…)	Resource Catalogue → Unit Catalogue
Utility consumption (steam, gas, water)	Input Lots → item with "Counts toward yield" unticked
Cooling & cleaning between furnace/line changeovers	Changeover Times
Intermediate production of semi-finished goods	Recipe-less batch + source batch link

The Activated Carbon / Chemicals demo set shows this whole structure working end to end: granular activated carbon from coconut shell (carbonisation → steam activation → acid wash → drying → mesh screening → packaging), 7 operations plus 15 process parameters, an alternative chemical activation route, 6 mesh fractions, a 13-row product quality specification and 4 production batches with a 3-level lot chain. The batches are planned as 1 big bag: input quantities are not typed by hand, they are calculated by exploding the formula tree (1 BB → 500 kg bulk → 863 kg activation output → 2,455 kg char).

Constraints / Known Limits

• Formula (BOM) lines are currently entered by hand; read-only synchronisation from Oracle ERP does not exist yet.
• No real live connection has been established with Asprova or Cedriks (MES) — the staging tables are ready at schema level but there is no data flow. Batch actuals are also entered by hand; there is no automatic feed from MES.
• The closed loop (actual yield/duration automatically updating the recipe standard) has not been designed yet — but now that actual yield and measurement data is collected, its precondition is in place.
• There is no separate laboratory/quality module: process measurements live inside the operation record and product measurements inside the output lot. A per-lot certificate of analysis (CoA) is produced, but sample management, test plans/schedules, instrument calibration tracking and retest flows are out of scope.
• The quality specification is one-dimensional: target plus min/max per property. Customer-specific specifications (different limits for the same product for a different customer) and shelf-life/stability curves are not modelled.
• There is no link between batch output lots and sales orders/shipments; which lot went to which customer is not yet recorded (forward genealogy therefore stops at the factory gate).