Transformer Loss Capitalization: The 30-Year Cost of a Cheap Bid

Two transformers arrive on a bid tabulation. Same kVA rating, same voltage class, same standard, same delivery week, both compliant with the technical specification. One is cheaper. On a price-based evaluation that is the end of the analysis, and the cheaper unit is awarded. It will then sit in a substation for thirty years, energised continuously, converting a fraction of everything that passes through it into heat that somebody pays for every hour of every one of those years.

The difference in that heat between two nominally identical bids is routinely larger than the difference in their prices. It is also entirely visible at tender, printed on the bidder's own guarantee sheet, and ignored by the evaluation method being used to compare them.

Two numbers that behave completely differently

A transformer has two losses and they do not share a schedule.

No-load loss, sometimes called core or iron loss, is present whenever the unit is energised. It does not care whether the transformer is serving full load or nothing at all. On a continuously energised asset that is roughly 8,760 hours a year, every year, for the life of the installation. It is set by the core: the grade of grain-oriented electrical steel, the stacking and joint design, or the decision to use amorphous metal instead.

Load loss, the winding or copper loss, varies with the square of the load ratio. At half load it is a quarter of its rated value. It is set principally by conductor cross-section and winding geometry.

The commercially important consequence is that a manufacturer can trade these against each other and against price. More core steel, or a better grade of it, lowers no-load loss and raises cost. More conductor lowers load loss and raises cost and weight. Every bid on the table is a point somebody chose on that trade-off surface. The price alone does not tell you where they chose to sit, and if the enquiry gave them no reason to sit anywhere else, they chose the cheapest point available.

The formula, and why it has to be published

Loss capitalisation converts those two guaranteed loss figures into present value so they can be added to the price and compared on one line:

Evaluated cost = quoted price + (A × guaranteed no-load loss) + (B × guaranteed load loss)

A is the present value of a unit of no-load loss across the evaluation period. B is the same for load loss. A is normally the larger of the two, because no-load loss runs continuously while load loss is scaled by a loading factor that on most installations sits well below unity and then gets squared. The method is long established and documented in IEEE C57.120, the loss evaluation guide for distribution and power transformers and reactors, which most utilities and many large industrial buyers already work from.

The part that gets missed is not the arithmetic. It is that the factors have to appear in the enquiry documents. A buyer who computes A and B privately once bids are in has made the evaluation fairer without changing the outcome much, because the bidders designed and priced against the only criterion they were given, which was first cost. Publishing the factors changes what is offered, not merely what is selected. Manufacturers hold multiple designs for the same rating. Tell them losses are being valued and at what rate, and a different design comes back. That is the entire mechanism, and it operates before bid submission or not at all.

Where this goes wrong in practice

Nobody publishes factors, because nobody in the room pays the energy bill. Utilities generally do this properly. Industrial plants, developers and EPC contractors building for handover frequently do not. The structural reason is that the party selecting the transformer is often not the party that will operate it. An EPC working to a fixed lump sum has a direct incentive to buy the cheapest compliant unit and no exposure whatsoever to what it costs to run. Unless the owner imposes the evaluation basis into the subcontract, the decision is made by someone with the wrong incentive, and it is made once, permanently.

Factors get borrowed instead of derived. A and B come from the owner's own numbers: cost of energy, cost of capital, expected loading profile, and the evaluation horizon. Copying a value from a previous project or from a utility in another market produces a calculation that is arithmetically tidy and economically meaningless. Set A too low and the evaluation quietly collapses back into a price comparison wearing a formula.

Guarantees are written without penalties. This is the failure that surprises people. Under IEC 60076-1, guaranteed losses carry tolerances: total losses may exceed the guarantee by up to 10 percent, and an individual component loss by up to 15 percent provided the total tolerance is not breached. A bidder can therefore win an award on a guaranteed no-load loss figure, come in materially above it at factory test, and remain fully compliant with the standard. If the contract does not attach a loss penalty priced at the same A and B rates used in the evaluation, the guaranteed number was never a commitment. It was a bidding instrument. The penalty is what converts it into a number the manufacturer has to mean.

Regulatory minimums get mistaken for specifications. In the EU, Regulation 548/2014 has set minimum efficiency requirements since 2015, with the stricter Tier 2 applying from 1 July 2021 and amendments introduced by Regulation 2019/1783. Equivalent floors exist in other markets. These establish what may lawfully be sold. They do not establish what is economically correct for a specific loading profile and a specific cost of energy. A buyer who specifies compliance and stops there has delegated a thirty-year economic decision to a legal minimum.

How we structure a transformer enquiry

At D1R7K0N the loss basis goes into the enquiry package, not into the evaluation spreadsheet afterwards. Four elements travel with the technical specification.

The A and B factors are stated explicitly, derived from the owner's energy cost, discount rate and evaluation period rather than inherited from a previous package. The expected loading profile is stated alongside them, because B is meaningless without it and bidders will otherwise assume rated load, which almost nothing actually runs at.

Guaranteed no-load loss and guaranteed load loss are required as separate declared line items on the bid form, not as a combined total, since a combined figure conceals exactly the trade-off the evaluation exists to expose. A loss penalty is written into the contract at the same rates used in the evaluation, so that the cost of missing a guarantee matches the credit the bidder received for offering it. And the test basis is fixed in advance: measurement method, applicable tolerance, correction to reference temperature, and whether the owner witnesses the test or accepts the certificate.

One question to ask internally

Find the last transformer your organisation bought and ask what A and B factors were used to evaluate it. If the answer is a number, the mechanism is working. If the answer is that nobody recalls, or that the evaluation compared prices against a compliant specification, then the unit was bought on first cost, the energy consequence started the day it was energised, and it will continue accruing quietly against an operating budget that has no visibility of the procurement decision that caused it.

None of this requires new technology, a longer lead time or a larger capital budget. It requires two numbers to be calculated before the enquiry goes out rather than after the bids come back. The transformer market will price losses accurately for any buyer who asks. Very few ask.