Main-Tie-Main Switchgear: What the Spec Asks You to Price

If you have quoted a double-ended lineup, you know how little the one-line diagram says about it. Two mains, one tie, three breaker symbols and a note that the tie is normally open. Main-tie-main switchgear looks like the simplest drawing in the package. It is also the part of a low voltage quote where two manufacturers reading the same page most often price different things.

I spent two days with the documents that say what the drawing does not: the written standards of six universities that publish their rules for double-ended substations, a relay maker's application note for a three relay transfer scheme, and a key interlock maker's guide. The six owners agree on the three breakers. They disagree on almost everything those breakers are told to do.

This article is for the applications engineer or estimator who has to price that disagreement. It covers what a main-tie-main is, the five decisions the one-line does not show, what each of the six owners requires, where the cost hides in closed transition, key interlocks and the transfer scheme, and seven questions to settle before the number goes out. It is the companion to our article on UL 891 vs UL 1558, which covers the enclosure and breakers the scheme sits in.

In brief

• A main-tie-main has two sources, two buses and a tie breaker between the buses that is normally open. Either source can carry both buses when the other is lost.
• The drawing shows three breakers. The price is in what tells them when to move: the transfer scheme, the interlocks, the control power and the testing.
• Six owners give six different answers. Yale wants automatic retransfer with closed transition and names the controller. Pennsylvania allows closed transition only by hand, under a synchronism check, and names a different controller.
• Closed transition ties the two sources together for a moment. A relay maker's own application note warns that the fault current available during that moment is doubled and can exceed what the breakers are rated to interrupt.
• Key interlocks and closed transition pull in opposite directions. Keys make it impossible to close all three breakers. Closed transition requires it, briefly.
• One owner requires a witnessed factory test of every automatic transfer scheme. That is days of someone's time, and it belongs in the quote.

What a main-tie-main is, and what double-ended switchgear means

What is a main-tie-main? It is two of everything, joined in the middle. Two incoming sources, usually two transformers, each feed their own bus through a main breaker. A third breaker, the tie, joins the two buses and is normally open, so in normal running each source carries half the building. If one source is lost, its main opens, the tie closes, and the remaining source carries both buses.

The same arrangement goes by other names. Double-ended switchgear and double-ended substation mean the lineup has a source at each end. Engineers who design them also call the arrangement secondary selective, because the choice between sources is made on the low voltage side.

Three things follow from that picture, and all three reach the price.

Each source has to carry everything. The University of Michigan's design guideline puts numbers on it. A single-ended substation is sized so the transformer's self-cooled rating is about 150 percent of the projected peak demand. In a double-ended substation, each transformer's self-cooled rating is about 115 percent of the peak demand of the whole substation, both ends together. The University of Pennsylvania's standard reaches the same place by a different route: with one end out, the surviving transformer carries the whole diversified load at 70 percent or less of its fan cooled rating. For the manufacturer the consequence is simple. Both mains, the tie and both buses are sized for the whole load, not for half of it.

The tie is a full size breaker. Michigan requires the tie breaker to be identical to the mains and able to change places with either one. Northwestern University requires mains and ties on a frame of at least 3,200 amps. Nobody should price the tie as a feeder.

It is bought for buildings that cannot go dark. Dartmouth College's guidelines say double-ended substations are preferred for critical facilities only, and give examples: data centers, chiller and heating plants, research buildings. That tells you who is on the other side of the quote. It is someone who is paying for continuity and will read the transfer scheme closely.

Main-tie-main switchgear: five decisions the one-line does not show

A one-line diagram for main-tie-main switchgear shows what is connected to what. It does not show what the breakers do when a source fails, when it comes back, or when an electrician wants to take a transformer out for maintenance on a Tuesday. Those are five separate decisions, and the specification may make all of them, some of them, or none.

The rest of this article takes those five in turn, after a look at how differently six owners answer them.

What six owners require

These six universities publish their electrical standards, and each one has something specific to say about the tie. The table compresses each document into one row. The paragraph numbers are each document's own.

Look at the first two rows together. Yale and Pennsylvania are both campuses with their own medium voltage distribution, and they have reached opposite conclusions about the same question. Yale wants the lineup to go back to normal by itself, without an outage, ten seconds after the source has been stable. Pennsylvania does not let the equipment parallel two sources unless a person has put it in manual and a relay has confirmed the sources are in step.

Each also names its controller, and they name different kinds of device from different makers: a programmable controller at Yale, protective relays and an automation controller at Pennsylvania. If your standard scheme is built around a third make, both documents ask you to rebuild it around theirs.

Dartmouth is the opposite of prescriptive. It asks the manufacturer to present the transfer controls and discuss them during design. That is an invitation, and also a cost: someone on your side prepares that presentation before there is an order.

Open or closed transition, and the fault current that comes with it

An open transition is break before make. The main on the failed side opens, and only then does the tie close. For a moment the affected bus is dead. A closed transition is make before break. The tie closes while both mains are still closed, and then one main opens. Nothing goes dark, but for a moment the two sources are joined through the lineup.

Two facts help in reading a specification on this point.

First, a transfer caused by a failed source is open by nature. The source is gone, so there is nothing to parallel with. Closed transition applies to the planned moves: transferring a healthy bus so a transformer can be serviced, and the return to normal after a failed source comes back. That is why Yale's list asks for closed transition on the retransfer, and why the relay maker's application note I read, General Electric's GET-8558, lists automatic transfer as open transition and automatic retransfer as an optional closed transition.

Second, the open transfer is not instant, and the reason is motors. Motors on a bus that has just lost its source keep turning and keep generating voltage for a short time. Closing the tie onto that voltage while it is out of step with the new source is hard on the motors and on the gear. The scheme in the GE note is called a residual bus scheme for that reason: it waits for the leftover voltage to decay, to a quarter of normal by default, before it closes the tie. If a specification says residual transfer, that waiting logic and the voltage sensing behind it are part of what you are quoting.

Now the part that is easy to miss. The same application note describes what happens while all three breakers are closed. Two transformers are paralleled, and if a fault occurs on a feeder in that moment, the short-circuit current is doubled. The note says plainly that the increased current may exceed the interrupting rating of the breakers, and that the paralleling time should therefore be set as short as possible.

So a closed transition requirement asks three things of a quote, and a one-line shows none of them:

• A synchronism check. Pennsylvania requires every closed transition to be supervised by the sync check function of its named relay. Yale asks for a sync check relay if it is required by the manufacturer, which hands the decision to you.
• Logic to end the parallel. Pennsylvania calls it trip select: after the parallel, the scheme automatically trips the breaker the operator chose. Something has to time that and act on it.
• An answer on the fault rating. Either the lineup and its feeder breakers are rated for the fault current of both sources together, or the time in parallel is limited and somebody states the limit. None of the six owner documents says which. That is a question to send before pricing, because the two answers are different lineups.

Michigan requires closed transition whenever an energized bus is transferred or retransferred. Dartmouth wants it where feasible. Pennsylvania allows it only in manual. Three owners, three positions, and a default scheme that satisfies one of them will be an exception at the other two.

Main-tie-main Kirk key interlocks: two keys for three breakers

A key interlock is a lock on a breaker that traps a key. The interlock maker's own guide describes the principle with a larger example, four mains and three ties, and the rule is the same at any size. Each main breaker holds its key while the breaker is closed. The key comes out only when that main has been opened and locked open. The tie is normally locked open and holds no key of its own. To close the tie, an operator must bring it a key, and the only way to get one is to lock a main open first.

Apply that to one tie and two mains and you have two keys for three locks. Any two of the three breakers can be closed. All three never can. The guide makes the point that this is stronger than a padlock and a tag, because the unsafe operation is not forbidden, it is impossible: the key that would allow it is somewhere else.

Four of the six owners write key interlocks into their standards, each with a condition attached:

• Michigan wants a key interlock on double-ended substations that are designed without automatic transfer controls, so the tie cannot be operated unless one main is open.
• Houston wants Kirk key interlocks between main, tie and main on a manually operated throw-over scheme, as shown on the one-line.
• Pennsylvania allows manual-only schemes only with permission, and then requires key interlocks or another feature that cannot be defeated.
• Northwestern asks for key interlocks as shown on the drawings, mountings and hardware where future interlocks are indicated, and, in its list of maintenance materials, two spare sets of keys that the university's chief electrician receives and signs for.

That last requirement is worth a second look. A key interlock works because there is exactly one key. A spare key in a desk drawer defeats it. Northwestern's answer is custody: the spares exist, and one named person signs for them. If you quote that section, the spare keys are a line item and the handover is a step in your closeout.

The other thing to notice is the conflict built into the hardware. Key interlocks make it impossible to close all three breakers. Closed transition requires all three to be closed for a moment. A specification that asks for both has to say how. Houston's master text shows the tension on the page: the sentence that requires key interlocks is followed by bracketed optional text about a main-tie-tie-main that allows momentary paralleling. The bracket is there for a project editor to accept or delete. If the project section you receive kept both, ask which one governs before you price the locks.

Main-tie-main transfer scheme: what is inside the word automatic

A specification can spend one word on this: automatic. The documents show how much is inside it. The table of contents of the GE application note reads like a list of things someone has to design, wire, set and test: a transfer mode selector switch, synchronism check, lockout functions, breaker operations in the test position, live source seeking, and what to do when both sources are lost at once. Its scheme uses three relays, one for each breaker.

Putting the six owner documents beside that note gives a fair bill of material for a main-tie-main transfer scheme. Not every job needs every line. Every line appears in at least one of the documents.

Three lines in that table deserve a comment.

Control power. A transfer scheme has to operate at the exact moment a source has failed. If the control power came from that source, the scheme is dead when it is needed. That is why Northwestern asks for two control power transformers in separate compartments with relays to switch between them, and why Yale and Dartmouth want an uninterruptible supply. It is a small part of the price and the first thing that fails a factory test.

Settings and responsibility. Pennsylvania supplies standard wiring diagrams and settings files for its named devices, which sounds like free engineering. Paragraph 5.0.H takes most of it back: using the university's material does not relieve the design firm or the manufacturer of responsibility for a properly operating system, and both are expected to be well versed in programming those devices. The files save typing. The hours to understand, adapt and prove the scheme are still yours.

The witness test. Pennsylvania requires a witnessed factory test for every automatic scheme. The GE note adds a caution of its own for that test: the breakers should never be closed with their mechanical close buttons during a factory acceptance test, because that bypasses the relay logic the test is meant to prove. A witnessed test means a test procedure, a simulated loss of each source, the owner's people in your plant, and time on the floor. We wrote about how widely those requirements vary between owners in the guide on how to quote switchgear.

Main-tie-main vs main-tie-tie-main

Some lineups have two tie breakers in series between the buses. Two of the six owners mention the arrangement. Michigan tells its designers to discuss it with the university's primary system engineer project by project, and to modify the specification if it is approved. Houston's master text carries it as an option and gives the reason in a bracket: with two ties, a tie can be tested without a total outage.

That reason is physical. A single tie breaker sits in a cell that is connected to both buses. To work safely on that cell, both buses have to be dead, which means the whole lineup is out. With two ties, each one can be isolated from one side by the other, so the lineup can stay half alive while a tie is serviced.

For the quote, main-tie-main vs main-tie-tie-main is the difference of a fourth breaker, and under Michigan's rule that breaker is identical to the mains. It also adds a fourth cell, the bus between the two ties, another set of interlocks, and a transfer scheme that has to know about four breakers instead of three. If the one-line shows one tie and the text mentions two, or the other way round, that is a count question, and the count comes from the drawing.

What the tie does to the ratings and the layout

Beyond the scheme, the same documents attach a handful of physical requirements to the main and tie sections. They are easy to price once you know to look.

• Barriers. Northwestern's switchgear section asks for the barriers between main and tie breaker compartments to be extended to the rear of the section, and for the line bus to be isolated from the load bus at each main and tie.
• Ground fault sensing. The same section lists a four-wire, double-ended substation as its own arrangement for ground fault protection, separate from an ordinary four-wire system. It needs its own sensor arrangement, so check that your standard scheme covers it.
• Surge protection on both sides. Pennsylvania asks for surge protective devices on both sides of the tie in a double-ended lineup. That is two devices where a single-ended lineup has one.
• Infrared windows. Houston asks for infrared scanning windows in the rear covers of the main sections, including the main-tie-main sections.
• Remote operation. Dartmouth wants the transfer controls to let electricians operate the lineup remotely. Houston wants remote racking and operation of the breakers. A remote operator station is hardware and wiring.
• Trip settings. Michigan requires the instantaneous trip on mains and ties to be defeatable, and the main and tie trip units to include a way of reducing arc flash energy during maintenance.

The enclosure these sit in matters too. A scheme with relays, a controller and compartmented mains and ties usually means switchgear, not a switchboard, and the companion article linked at the top explains how to tell which one a specification is describing.

Seven questions to settle before you price a main-tie-main lineup

These come straight from the gaps between the six documents. If the project section answers all seven, price it. If it does not, each unanswered one is a clarification in the quote or a question to the engineer.

1. Is the transfer manual, automatic, or both with a selector? And is the baseline what Dartmouth calls enhanced manual, or a full automatic scheme?
2. Is any transfer closed transition? In which direction, and in automatic or only in manual?
3. During a closed transition, what is the fault rating? Is the lineup rated for both sources together, or is the time in parallel limited, and who states the limit?
4. What keeps the sources apart? Keys, electrical interlocks, or both. If keys and closed transition both appear, which governs?
5. Is the controller named? Who supplies the settings, who programs it, and who is responsible when it does not work?
6. Where does control power come from when one source is dead?
7. Who witnesses the test, and where? And does another scheme upstream set the timing of yours?

Each of those is a row in a compliance matrix. Our specification compliance matrix has a row for transfer schemes in its switchgear starter set, and a blank sheet for the rest. The answers you write there become the clarifications page, which is where a buyer comparing two prices finds out that the two manufacturers priced different schemes.

A real case shows what happens when they are not settled. One of the public submittal records we read for an earlier article was a 4,160 volt main-tie-main lineup for a Texas water district. Two reviewers from the same engineering firm reviewed the same package. One approved it as noted and the other rejected it. That record is in what engineers actually reject.

Why a quoting company wrote this

Mavlon builds software that reads a customer's specification and drawings when the request for quote arrives. It lists every requirement that changes the price, ties each one to its page, and flags the ones that disagree, such as a section that requires key interlocks and closed transition in the same paragraph. Your estimator decides what to do about each flag, and the quote is drafted in your own pricing rules. A main-tie-main section is a good test of any reading tool, because the expensive sentences are short and scattered. If your team quotes double-ended lineups from specifications, we would be glad to read one with you.

Sources

• University of Pennsylvania, Design Standards 26 10 00, Power Distribution System and Equipment, revision July 2019, sections 4.0, 5.0 and 9.0.
• Yale University, Design Standards, Subdivision 16300, Transmission and Distribution, revision 3, April 2009, secondary unit substations.
• University of Michigan, Design Guideline 26 11 00, Unit Substations, February 2023.
• Dartmouth College, Design and Construction Guidelines 26 24 13, January 2023, paragraph 1.2.E.
• University of Houston, Master Specification 26 24 14, Low Voltage Switchgear, August 2020, paragraphs 2.3.H and 2.3.I.
• Northwestern University, Section 26 2300, Low-Voltage Switchgear, issued 29 March 2017.
• GE, Application Note GET-8558, Residual Bus Main-Tie-Main Automatic Transfer Scheme Using Three GE 850 Relays, revision 00, 18 October 2013.
• Kirk Key Interlock Company, Examples of Key Interlock Systems, white paper.

Found a mistake? Write to atishay@mavlon.co and we will fix it.