Sequencing Submittals Against Lead Times

The schedule says fabrication starts in June. The PLC has a sixteen-week lead time. The spec forbids ordering before the submittal is approved, and the engineer takes three weeks to review. Nobody did that arithmetic in January, so in May the panel shop is waiting on a processor that has not been ordered because the package that covers it has not been sent. Every integrator has lived some version of this, and it is entirely a sequencing problem.

The approval-before-order chain

The chain is contractual, not customary. Division 01 says it several ways, and the spec books I hold say all of them:

• Work covered by a submittal is not included in progress payments until the submittal is approved.
• The contractor bears the cost of damages resulting from ordering material or proceeding with work before the engineer's review is complete, unless authorized in writing.
• Do not ship materials and equipment until the related shop drawings and product data have been approved.
• No material shall be delivered to the work without prior written approval.

Read together: you cannot get paid for it, you order it at your own risk, and it cannot ship or arrive until the stamp exists. On a long-lead item, the order date is therefore the approval date at the earliest, and the approval date is the submittal date plus the review turnaround plus however many cycles the package takes. Every week the submittal slips, the delivery slips a week, unless you eat the risk of ordering unapproved.

Some integrators do eat that risk on safe items, ordering a standard PLC platform the spec names before the stamp. That is a business decision, and the spec's language about bearing the cost is what you are betting against. Make it consciously, on named items, with the PM's signature, not by default because the schedule got away.

Need-by and order-by

Two dates per item, worked backward:

Need-by is the date the part must be in hand. For a panel component it is the fabrication start of the panel that contains it, not the site delivery date; the panel shop cannot wire a processor it does not have. For a field instrument it is the installation date, minus whatever receiving and inspection take.

Order-by is need-by minus the lead time. Lead time comes from the vendor's quote if the quote states it, otherwise from what that part actually took last time you bought it. Catalog lead times are optimistic; your own receiving history is not.

Then one more step back that most schedules skip: submit-by is order-by minus the review turnaround, minus the time to build the package. If review runs three weeks and the package takes a week to assemble, the submittal for a sixteen-week part needs to be in the engineer's hands twenty weeks before fabrication starts. On a job awarded in January with fabrication in June, that submittal was due the second week of January.

Some specs require you to write this ladder down. The conformed spec for one plant expansion I hold demands a schedule of submittals showing, for each submittal, the date requested from the supplier, the date submitted to the engineer, the date returned with disposition, the date revised and resubmitted, the scheduled and actual delivery dates, and the date by which the material must be at the site to avoid delaying the work. It also requires that submittals on the critical path be flagged and that submittals expected to have long review times be identified and scheduled accordingly. That is the need-by, order-by, submit-by ladder, written as a contract deliverable.

Which packages go first

Not the easy ones. The first packages out the door are the ones gating the longest lead times, in this order:

1. Anything with a lead time longer than the time to fabrication. Processors and communication modules, large drives, custom enclosures, specialty instruments. If order-by is already past or within a few weeks, that package leads.
2. Anything the spec requires approved before other work can proceed. Test procedures before testing, and on some jobs a submittal schedule or a list of proposed suppliers before any technical submittal will be reviewed. One spec I hold will not review any Division 13 shipment until the test plan submittal is approved.
3. Anything with a known-difficult review. A substitution, an "or equal" not on the list, a package to a reviewer who historically takes six weeks. Difficult reviews take extra cycles, and extra cycles are extra months.
4. Everything else, in fabrication order.

The instinct to send the easy panel package first because it is ready is the instinct that leaves the long-lead instrument package for March.

Splitting a package to release long-lead items

A panel package covers a processor with a sixteen-week lead time and forty other parts with two-week lead times. Holding the processor's approval until the whole package is ready gives the forty parts a fourteen-week head start they don't need and costs the processor fourteen weeks it can't afford.

The fix is a split: send the long-lead components as their own product data submittal first, with a cover statement that the panel shop drawings will follow under a separate number, and reference the pending drawing package. Engineers accept this routinely when the transmittal says why. Two cautions:

• Some specs forbid partial submittals ("no partial submittals will be reviewed"), which usually means an incomplete package under one number, not two complete packages under two numbers. Ask at the preconstruction meeting, and if the engineer wants the split written into the submittal schedule, write it there.
• The split must be a complete package in its own right. Product data for the long-lead items, catalog numbers marked, spec compliance shown. A datasheet dump labeled "long-lead items, drawings to follow" is an incomplete submittal with a new name.

The taxonomy of what counts as a separate submittal is in shop drawings versus product data.

The review turnaround assumption

The number to plan on is the number the engineer actually takes, not the number in the spec. Specs I hold say ten working days; the median across the submittal logs I measured was 21 calendar days, with a long tail, and the figures are in what real submittal logs say about resubmittal rates. Plan at three weeks per cycle and treat a second cycle as likely on any package you did not self-review.

Two consequences for the ladder:

• Submit-by uses the real turnaround times the expected number of cycles. A first-time package to a strict reviewer is two cycles, six weeks, not ten days.
• Chase at the median, not at the spec's limit. A package that has been out three weeks with no word is a package to call about. Waiting until day 30 to ask is a week of schedule spent being polite.

One schedule example

Lift station panel, fabrication start June 15, engineer's measured turnaround three weeks, award January 6:

Reading it: the processor's package has to leave the office in the second week of January if it might take two cycles, which on a first submittal to a new reviewer it might. That is the week after award. The enclosure and level transmitter packages are due in February and March. Everything with a short lead time can ride the panel drawing package in April. The sequence writes itself once the dates are on one page; the failure is never doing the subtraction.

Two housekeeping rules keep the table honest. Update lead times from the vendor's confirmed quote, not the catalog. And when a milestone moves, move every date that hangs off it the same day, because a need-by that quietly stayed in June after fabrication moved to May is how a schedule lies to you.

Disclosure: I build Submittal Kit, which derives each part's need-by from the schedule's milestones, subtracts a lead time taken from the quote or from your own receiving history, and flags the order-by when it comes within two weeks. The subtraction is the whole trick; a spreadsheet does it too, as long as somebody keeps it current.