Worked Example: Inputs, Checks and Adjustments for 120 Brackets

Download the complete exercise pack (SQL, field list, Python scripts, database and CSV files)

This example connects time, demand, routing, resources, materials and inventory through two operations: stamping and assembly. It produces a loadable teaching database, table files and quantity calculations, without changing your business database.

1. Business Conditions

The periods represent two production windows. When mapping real dates, also retain the planning origin, shift times and nonworking intervals.

2. Prepare the Files

From the supplied examples directory, use Python 3.10 or later. The base scripts use only the standard library. On Linux, use python3 instead of python if that is how your environment is configured.

python build_example.py --output-dir ./my-first-example
python check_example.py ./my-first-example/data.db
python calculate_candidate.py ./my-first-example
python test_examples.py

The builder creates:

An existing data.db stops the builder; choose another directory. You may also inspect the supplied sample output before rebuilding from SQL. The business records are the same.

3. Check the Inputs

check_example.py should report ok: true, 18 tables and 105 fields. There are two order lines, two operations, two adaptors, four capacity rows, two BOM rows, four readiness rows and two inventory-limit rows.

planning_result has zero rows. The manual candidate is stored separately and is never written into the input database as an actual solver result.

Use this query in a SQLite viewer to inspect how demand reaches its operations:

SELECT oi.id AS order_line, p.code AS product_code,
       oi.delivery_time, oi.number AS demand_quantity,
       r.code AS route_code, op.code AS operation_code, op.seqno
FROM order_item oi
JOIN product p ON p.id = oi.product
JOIN process_route r ON r.product = p.id
JOIN process op ON op.route = r.id
ORDER BY oi.id, r.id, op.seqno;

The query returns four rows: each of two demand lines joins to two operations. Do not sum the repeated demand values into additional demand.

4. Load the Example into DecisioWorks

With DecisioWorks and its dependencies available, run from any directory:

python /path/to/examples/load_with_decisioworks.py /path/to/my-first-example/data.db --project-root /path/to/DecisioWorks

Replace the three paths with actual locations. The script sets up the project import paths, calls load_business_data, and runs validate_batch_productivity_multiple.

The verified loaded counts are demand=2, process=2, capacity=4, bom=2, kitting=4 and inventory_limit=2, with preprocessing status ok. This confirms data-layer loading; the step does not execute a solver.

5. Recalculate the Manual Candidate

Stamping has 360 available minutes in period 1; all other resource-period combinations have 480. Cumulative consumption is 120 blanks and 240 fasteners, matching supply. Ending finished stock is 20 in each period; cumulative production is 80 and then 120.

The calculator should return aggregate_checks_passed: true. It also retains solver_executed: false and full_scheduling_feasibility_proven: false: the calculation checks aggregate relationships in this candidate, not every timing, transport, changeover or overlap condition.

ok means the inputs meet this calculator's requirements. aggregate_checks_passed covers resource, material and inventory totals. demand_issue_comparison lists order demand, assumed stock issues and their difference by period. issues_match_due_quantities compares those two quantity series only; it does not prove order allocation or on-time delivery, so demand_coverage_checked remains false. Missing records or unsupported inputs produce ok: false, diagnostic errors and a nonzero exit code.

6. Why Maintenance Changes Require Another Check

Build a separate exercise copy:

python build_example.py --output-dir ./three-hour-maintenance

Execute this SQL in the copy:

UPDATE capacity SET used=0.375
WHERE workcenter=11 AND time_unit=1;

Run validation and calculation again. Stamping availability falls to 300 minutes, still enough for its 200 processing minutes.

Now add these assumptions: maintenance occurs at shift start, all stamping finishes before assembly starts, and the operations do not overlap. Originally the chain needs 120+200+160=480 minutes. Longer maintenance raises it to 180+200+160=540 minutes. Individual resource totals passing their checks does not establish that the complete routing fits within eight hours.

Choose a business response: allow confirmed transfer-batch overlap, adjust supply and start times, revise delivery arrangements, or enable optional time. Run the appropriate scheduling capability with the selected conditions and check its actual output.

7. Move from the Exercise to Your Data

Begin with small changes that retain the example's object relationships. When changing due dates or the number of periods, follow lesson 02 to update the issue plan, inventory limits, resource calendars, supply and candidate output together. Table CSV exports are for inspection; calculations read data.db. Record the source, units, expected change and check result.

This manual calculator supports product id=1, one route, two sequential operations on separate resources, one adaptor per operation, OEE100 and no setup loss. Quantities are discrete units, both operations produce equal quantities in each period, and there is no opening work in progress. All periods must be normal production periods. You can add periods using the complete procedure; alternate routes, shared resources, substitute materials, joint processing, optional time and advanced batch rules require another calculation or model. Batch dispatch, WIP capacity, order allocation and detailed timing feasibility need separate checks through the relevant model.

Changing demand does not automatically change the stock-issue assumptions in scenario.json or the output in manual_candidate.csv. Explain the difference before revising the order, issue plan or candidate. For a business case outside the stated scope, retain field and relationship checks and use the appropriate DecisioWorks model and run configuration rather than treating this manual report as full acceptance.

Useful review materials include input tables, source mappings, rule configuration, run records, result tables and before/after explanations. Use the field reference for storage details and the semantic guide for business relationships.