sailing Foudra MRP
Reference manual chevron_rightplanning/index

Planning

MRP II planning, both ways. This page answers the question every other page has been building towards: given what we've promised, what we keep in stock, and how things are built, what should we make or buy, how many, and by when? And then the question MRP II adds: with the hours we actually have and the deliveries actually coming, when will it really be ready?

The run plans backwards from when things are needed and proposes what to make and buy. The schedule works forwards from now, through stock, purchase orders' expected dates, routing times and each workstation's working hours, and predicts when every order and every demand will really be ready, and why when that is late.

What the run does

Run planning (planner) recalculates for this company. The nightly job does the same for every company, before the tasks job, so the page and its tasks are fresh each morning. A run:

  1. Starts from demand: every open demand order, plus every stocked part's reorder point (minimum stock), plus the components that open work orders (drafts included) still need to build what they will build.
  2. Walks each demanded part's bill of materials top down, multiplying quantities through. Phantoms (unstocked sub-assemblies) are expanded into their own components.
  3. Nets each part: what is wanted, plus its reorder point, minus what is on hand, minus what is inbound on purchase orders (drafts included), minus what open work orders will still deliver (drafts included).
  4. Anything still short becomes a proposal: a make if the part has a routing, otherwise a buy. Makes are rounded up to the routing's batch size. A buy goes to a supplier chosen by fixed rules, in order: of those that can deliver by the date it is needed, the least total spend for the quantity (rounded up to that supplier's minimum order and pack size, at its price break, in your currency); then the preferred one; then the quickest. If none can deliver in time, the soonest. The quantity is rounded up through the chosen supplier's minimum order and pack size.
  5. Dates: needed by is the earliest date that wants the part. Release by is that less the lead time (the part's for a make, the supplier part's for a buy). A component is needed by the day its parent is released. A reorder point alone carries no date.

One proposal per date. Requirements for the same part on different dates stay apart: three boats due in three months become three proposals for each part, each with its own needed-by and release-by, so October's parts can be started without November's tying up the printer. Stock and open orders are handed to the earliest date first, and any excess from lot sizing covers the next. A reorder point is an undated requirement, satisfied last.

Priority. Each proposal carries the priority a work order firmed from it will inherit: the most urgent demand it serves, directly or through what it goes into (a customer order 2, a forecast 3, or whatever a planner set, 1 to put it ahead of everything), and 4 for a reorder-point top-up. The schedule books work in that order, then by needed-by date, and the printer takes released orders the same way, so a spare-stock run never elbows a customer's order aside.

Every run replaces the previous proposals; anything you firmed or cancelled stays.

The table

Column Meaning
Do make or buy
Part Indented under the proposal that needs it
Qty After lot sizing
Needed by / Release by See above; blank when only a reorder point drives it
Predicted For a make: when the schedule says it will really be finished, red and marked late when that is after its needed-by date. Hover for why
Source "routing" for a make; for a buy, the supplier chosen (step 4) with why underneath, in a list to choose another of the part's suppliers before firming; no supplier when the part has none (add one on the part's sourcing page)
Why The demand orders, parent proposals and reorder point behind it

Firming

Tick proposals and press Firm selected (planner). Each ticked make becomes a draft work order for the quantity, due on the needed-by date, at the chosen location for both components and output. Ticked buys are grouped into one draft purchase order per supplier, with the supplier's part number and last price on each line, expecting delivery on the needed-by date or, when the supplier's lead time from today makes that impossible, the soonest they can deliver. Nothing is released, approved or sent from here; the work order and purchase order pages do that as usual.

A firmed proposal moves to the lower table with a link to what it became. The next run counts those drafts as supply, so it will not propose the same thing twice.

The × on a row cancels a proposal until the next run, for "I know, and no".

Late demand

Above the proposals, when there is any: each open demand the schedule predicts will miss its due date, with when it will be ready and why, following the chain to where the delay starts:

Cycle Hub Galway, order 4471: 8 × Trail 29 Comp, due 19 Oct, ready Tue 24 Nov. Waiting for WO-00013, predicted 24 Nov; behind that, the rear wheels are waiting for tyres on a proposed buy from Maxxis, not ordered yet, 45 days' lead.

That is the point to act: order the tyres today, move the work, raise a priority, or tell the customer. The schedule never moves a date itself.

The schedule

Predictions come from one forward pass, recomputed every time a page asks, so they are never stale:

The reasons name what held each thing back: the component waited for and where it is coming from, a group of workstations that was full (Welding busy until Thu 8 Oct 13:00), something not in stock, on order or planned at all, and steps whose routing has no time, which count as instant. The same prediction shows on each work order and in the demand list.

It is a single greedy pass, not an optimiser: good enough to see trouble weeks ahead, and simple enough that every date can be explained.

Capacity

Below the proposals, one chart per workstation group with work on it: the schedule's bookings in that group, open orders' remaining steps in blue and proposals' in grey, anything finishing after its due date in red and listed above the chart: WO-00023 · Print finishes 3 Dec, due 30 Nov. The header shows each group's stations and their hours a week together. That is the answer to "with one printer, will the second boat be ready?"

A group with no active station, or none with working hours, can't be scheduled and says so. Timings come from the routing, so set setup and run minutes on every step.

Tasks

A proposal whose release date has arrived and that nobody has firmed raises a planning.release task on the dashboard and the alert bar. A demand the schedule predicts will be late raises a planning.late task with its reasons: urgent for priority 1 and 2, high otherwise. Both are refreshed nightly, or with Refresh on the dashboard.