Switchgear Submittal: What Engineers Actually Reject

I spent two days reading four public submittal records for switchgear and switchboards: the packages manufacturers sent for approval, and what the engineers wrote back. One school switchboard in Connecticut came back approved. Three switchboards for a Texas water district came back "Not Approved" with 35 numbered comments. A 4,160 volt switchgear lineup for the same district came back with two reviewers from the same firm disagreeing about whether it was approved at all, followed by the manufacturer's written replies and a resubmittal seven weeks after the first one went in. If you prepare a switchgear submittal, or price the equipment behind one, these records show what engineers actually look for.

The short version is that very few of the comments are about engineering in the sense of design. Most of them are about the specification: things it asked for that the package left out, drawings in the package that disagree with each other, and labels that do not match the engineer's drawings. Several of them change the bill of material, which means they change the price of a job that has already been sold.

That is why a quoting site cares about submittals. Everything a reviewer rejects after award was in the specification at bid time. We walked through reading a switchgear specification in how to quote switchgear; this article follows the same documents one stage later, to the point where the manufacturer's reading gets checked by the person who wrote them.

In brief

• A submittal is the package of drawings, bills of material and data sheets a manufacturer sends for approval before building. The manufacturer usually holds production until it comes back approved.
• In the records I read, the comments fall into five groups: items the specification asked for that the package left out, drawings that disagree with each other, labels that do not match the engineer's one-line diagram, coordination details, and packaging.
• Several comments change the bill of material: a fourth current transformer, a higher meter model, Ethernet communication options, a different fuse type. Those were in the specification when the job was quoted.
• Identical comments repeat across every lineup in a package. Three switchboards produced the same ten issues three times.
• Two reviewers can reach different outcomes on the same package, and a construction manager and an engineer can do the same on one cover sheet.
• The best pre-submission checklist is the requirement list you built when you quoted the job.

What a switchgear submittal is, and who reads it

After an order is placed, the manufacturer prepares a submittal: outline and elevation drawings, one-line and three-line diagrams, wiring diagrams, bills of material, data sheets for every major device, and any certificates the specification asks for. The package goes to the contractor, who checks it and certifies it, then to the engineer who wrote the specification, and often to a construction manager as well. Each of them marks it with a disposition.

The dispositions matter because they control the factory. The switchboard submittal published by Loudoun County, Virginia, includes the manufacturer's own return form on page 2. It offers five outcomes: Approved, Approved as Noted, Partial Approval, Rejected, and Revise and Re-submit. Each is tied to a release decision: release everything for manufacture, release everything and show the changes on construction drawings, release only the approved sections, or release nothing. The same form asks the customer to initial five confirmations before anything is built: lug sizes verified, top or bottom entry verified, shipping splits verified, nameplate information verified, and orientation of breakers verified.

The 4,160 volt switchgear manufacturer on the Texas project put the commercial consequence in writing on its transmittal: "If changes are submitted, the order will then be subject to revision charges and possible delays in shipment." A submittal is not paperwork around the order. It is the last point where a misread specification can be fixed cheaply, and the first point where it starts costing money.

Every reviewer's stamp in these records also carries a limit. The Connecticut engineer's stamp reads: "This review is only for general conformance with the design concept of the project and general compliance with the information given in the Contract Documents." The Texas contractor's certification says approval "does not relieve the vendor of their responsibility to comply with the contract documents." Approval does not transfer the risk of a missed requirement. The manufacturer keeps it.

The four records

The two Texas records are published as text only, so their drawings cannot be seen, and the outcome of the medium voltage resubmittal is not in the record. Everything quoted below comes from pages that can be read.

What goes inside a switchgear submittal

Three of the four packages can be read page by page or from their own table of contents, and their make-up is revealing. The part a reviewer actually checks against the specification is a small share of each one.

In the Loudoun package, the drawings of the switchboard itself are two pages out of sixty-nine. In the Connecticut package, the project drawings and bills of material add up to fewer than twenty pages out of a hundred and five, and there are more pages of chemical substance declarations for the transfer switch than there are switchboard drawings. None of this is wrong. Catalogue data, manuals and declarations are often required. But the reviewer's attention goes to the few project-specific pages, and generic pages make those pages harder to find and easier to contradict.

That is why two of the Texas comments are about packaging rather than content: remove catalogue pages for equipment not supplied at this site, and mark on each data sheet which model and options are actually being provided. The first stops the reviewer asking "Where are the LJ frame circuit breaker, PowerPact PJ circuit breaker, and Masterpact NW 2000A & 3000A circuit breakers being provided at this site?" The second stops them having to cross-check a part number in the bill of material against a brochure that lists twenty variants.

An approval, and what it still asked for

The Connecticut package is what a clean submittal looks like. Page 1 is the transmittal with three signatures on it. The contractor certifies that it "has determined and verified materials, field measurements and field criteria". The engineer ticks "Reviewed/No Exceptions Taken" from a list that also offers Rejected, Furnish as Corrected, Submit Specific Item, and Revise and Resubmit. Behind that come bills of material, the switchboard drawings (a specification sheet, front view, unit schedule and floor plan), catalogue pages for the devices, the transfer switch manufacturer's brochure, warranty and terms, and a "Comments & Clarifications" page offering a five-year warranty on the switchboard.

The construction manager's stamp on the same cover sheet reads differently: "APPROVED AS NOTED: PLEASE SUBMIT SURGE PROTECTION DEVICE FOR REVIEW & APPROVAL." The surge protection device data is actually in the package, near the back, on the page that lists it as "SPD, 120/208V, 480KA, FILTER". The request appears to be for it as a submittal of its own. The record does not say why.

Two reviewers, two outcomes, one cover sheet. For the manufacturer this is the normal case, not the exception. Each reviewer checks against their own list, and a package that satisfies the engineer can still come back with a note from the construction manager. The practical lesson is to organize the package so that each reviewer finds what they are looking for quickly, and to submit separately anything the specification treats as a separate item.

Switchboard submittal review comments: all 35, sorted

The Texas switchboard review is the most useful record of the four, because the engineer's comments are complete, numbered and specific. The package covered three 480 volt switchboards, one for each of three pump stations. It was submitted on February 28, 2012, marked urgent, and returned on March 9 with the disposition "Not Approved. Provide Additional Information. Revise and Resubmit. Provide a written response to each comment on a separate sheet of paper."

Comment 2 applies to the whole package: "The warranty shall be in accordance with section 26 24 13-11, 3.06. Currently the submittal does not include any information on the warranty." The remaining 33 comments are grouped by switchboard, and most of them repeat for each one. Removing the repeats leaves twelve distinct issues.

Comment 25 adds one site-specific change: remove the vertical neutral bus from the sections feeding the pump starters at one station, "to maintain consistency between sites", because it was shown there and not at the other station.

Three things in this table deserve attention from anyone who prices switchgear.

Four of the twelve issues change what gets built. A fourth current transformer, a higher meter model, an Ethernet communications card, and a breaker sized to the engineer's one-line. Each has a cost, and each was in the specification when the equipment was priced. If the quotation carried three CTs and the lower meter, the difference now has to be absorbed or argued about as a change, after the job was won at the original number.

The same mistakes repeat on every lineup. Three switchboards, three copies of the same ten issues. Nothing about the second and third boards was checked separately; they were built from the same template with the same gaps. One missed requirement becomes three rejected comments, and three corrections to make.

Almost none of the comments are about design judgement. The engineer is not questioning the bus rating or the breaker selection. They are checking the package against the specification and against the package itself: whether it has what was asked for, whether its drawings agree with each other, and whether its labels match theirs. That is a reading check, and it can be run before the package goes out as easily as after it comes back.

The medium voltage switchgear: two reviewers, two answers

The 4,160 volt switchgear package for the same project went in on the same day and came back on the same day, March 9, 2012. The engineering firm's system recorded a combined status that reads like a compromise: "Approved As Noted, Revise and Resubmit, Additional Information Required." Underneath it are two reviewers from the same firm. One wrote "Approved as Noted. Additional information required." The other wrote "Not Approved. Provide Additional Information. Revise and Resubmit."

Their comments overlap, and they also cover different ground. The second reviewer found:

• The meter and relay displays and indicating lights would sit at 6'-2" above the floor, against a maximum of 5'-6" including the housekeeping pad.
• The power meter's Ethernet option was not shown, and the feeder protection relay's model number, as listed in the bill of material, did not include the Ethernet Modbus TCP/IP connections the specification required.
• Auxiliary contacts from the main and tie switches, wired to terminal blocks for the owner, were required and not shown.
• A drawing titled "Power wiring interconnections & key interlocking" contained nothing about key interlocking.
• Nameplates were shown as white with black letters, against the specified black with white.

The first reviewer found different things: remove a hand-off-auto switch from the contactor controls and use open and close pushbuttons with indicating lights instead; show a ground fault current transformer on each contactor; identify where the terminal strips sit in each section; and question the fuses on the incoming switches. "Verify the 57X fuses on the main incoming switches are correct for the application. Is there a reason why 600E type fuses are not used."

If only one of these reviewers had looked at the package, half of these comments would not exist. Neither review was wrong. They were thorough in different directions, which is what happens when two people read a long package against a long specification. For the manufacturer, it means a clean response to one reviewer is not a clean package.

The reply, and what it cost

The manufacturer answered point by point on its own sheet, as the engineer had asked. Most answers are two words: "Drawing revised." Some are longer. To the fuse question: "600E type fuses to be provided." To the question about the lightning arresters: "Voltage rating for the lightning arresters is 6kV, which is embedded in the highlighted part number as per the data sheet in Tab 4." Data sheets for the key interlocks and the communications device were added.

The fuse answer is the one to notice. A reviewer asked a question, and the answer changed a component. That is a bill of material change on a sold order, arriving by way of a submittal comment. The manufacturer's transmittal of March 28 had already set out the terms: changes after the submittal would bring "revision charges and possible delays in shipment."

The dates tell the rest. First submission on February 28. Comments back on March 9 and forwarded on March 13. The manufacturer's revised documents on March 28. The resubmittal went to the engineer on April 20, 53 days after the first submission, and the record ends there, with the second review still open. Seven and a half weeks had gone by and the switchgear was not yet released for manufacture, on equipment where the factory lead time runs from the day it is released.

None of this is unusual. It is what a normal first-round rejection looks like on a municipal job. That is exactly why it belongs in the price: every rejected round costs engineering hours to answer, and calendar weeks the order does not get back.

What the comments have in common

Across the three reviewed records, the comments sort into five groups, and the groups are the same whether the equipment is a school switchboard or a pump station switchgear lineup.

1. Items the specification asked for that the package left out. Warranty information, surge protection device data, Ethernet options on meters and relays, a fourth CT, auxiliary contacts for the owner, a specified meter model, a specified fuse type. These are the expensive ones, because many of them change the bill of material.
2. Drawings that disagree with each other. Nameplate colour in the bill of material against the general notes; a meter model on the one-line against the elementary drawing; fuses on one drawing and not the other; a drawing whose title promises key interlocking and whose content has none.
3. Labels that do not match the engineer's drawings. Load names, equipment designations, a tie switch nameplate that should read "TIE SWITCH".
4. Coordination details. Lug heights from the top and bottom of the equipment, terminal strip locations, mounting heights measured from the floor including the pad.
5. Packaging. Catalogue pages for equipment not supplied at that site, a submittal not organized the way the specification asked, data sheets that do not mark which model and options are being supplied.

Groups two to five are drafting and packaging discipline, and a good internal check before release catches most of them. Group one is different. It is a reading problem, and it starts at the quote. The specification that the reviewer checks against is the same document the estimator priced from. If the quote missed the Ethernet option, the fourth CT or the fuse type, the submittal inherits the miss, and the reviewer finds it after the price is fixed.

A switchgear submittal checklist built from real rejections

Every item below comes from a comment in these records. Run it before a package goes out, and run the first half of it before the quote goes out.

1. List every device the specification names by model or option: meters, relays, communications cards, fuses, surge protection. Check each one against the bill of material and against every drawing that shows it.
2. Count the current and potential transformers on each circuit the specification describes, and check the drawings show the same count.
3. Find every mounting height rule and check it against the elevation, including the housekeeping pad.
4. Find the nameplate rules: colour, lettering, wording. Check the bill of material, the general notes and the nameplate schedule say the same thing.
5. Use the engineer's load names and equipment designations from their one-line diagram, exactly.
6. Include warranty terms, and any item the specification lists as a separate submittal, as a separate item.
7. Show lug positions and cable entry distances, and the terminal strip location in each section.
8. Remove catalogue pages for anything not supplied under this submittal, and organize the package the way the specification says: by site, by lineup or by section.
9. On every data sheet, mark which model and which options are being supplied.
10. For every lineup built from the same template, check each one separately. A missed requirement on the first board is a missed requirement on all of them.

Items one to four are also the items that decide whether the quote priced the right equipment. If they were resolved at bid time and written into the quotation's clarifications, the submittal is a formality. If they were not, the reviewer resolves them for you, after award, at your expense.

How to answer review comments

The engineer on the Texas project asked for "a written response to each comment on a separate sheet of paper", and the manufacturer provided one. A few habits from that exchange are worth copying.

• Answer every comment by number, including the ones that repeat across lineups. The reviewer will check each one.
• Answer questions with decisions. "Is there a reason why 600E type fuses are not used" was answered with "600E type fuses to be provided", not with "noted". A question left open comes back in the next round.
• Point to where the answer is. "Voltage rating ... is embedded in the highlighted part number as per the data sheet in Tab 4" saves the reviewer a search.
• Price the changes before you accept them. A comment that changes the bill of material is a commercial question as well as a technical one. If the specification required it, the cost is yours. If it did not, it is a change, and it should be raised as one before the drawings are revised.

Carry the quotation's clarifications into the submittal

Two of the packages include a page the manufacturer wrote in its own words. The Connecticut switchboard has a "Comments & Clarifications" section confirming a five-year warranty and the surge protection device. The Texas switchgear package lists "Comments and Clarifications for Section 26 13 23" and a warranty letter in its table of contents, right after the transmittal.

That page is where a quotation's clarifications should reappear. If the quote said the equipment was priced with a particular meter model, without a battery system, or with the relays excluded because the owner supplies them, the submittal should say so again, in the same words, before the reviewer reaches the drawings. A reviewer who reads an agreed exception on page 2 does not write it up as a missing item on page 40.

The reverse also holds. A clarification that was never written into the quote cannot be added in the submittal and expected to stick. If the fourth current transformer was in the specification and the quote said nothing, the reviewer will ask for it, and the specification will be on their side. The clarifications page in a submittal is only as strong as the one in the quotation that came before it.

Why the submittal starts at the quote

Mavlon reads the customer's specification and drawings when the job is quoted, cites every price-relevant requirement to its page, and puts conflicts and open items in front of the estimator. That same list of requirements is the checklist a reviewer will use against your submittal after award, which is why we think the two belong together. How long it takes to set up for your own documents depends on the packages your customers send, and we can only give a real number after looking at an actual one with you.