Gas turbine derating is the difference a buyer must understand between reference performance and usable power at the project. Temperature, elevation, installation losses, equipment condition and plant auxiliaries can all enter that assessment, but they should not be collapsed into an unsupported percentage.
The purchasing question is specific: what can this identified package deliver at the required electrical point, under the conditions that determine whether the project can serve its load? A catalogue MW figure does not answer it.
This guide separates performance corrections from additional plant loads, works through two hypothetical net-output examples and explains what belongs in the supplier's site-performance assessment.
Key Takeaways
- Preserve the conditions and measurement boundary attached to every rating.
- Evaluate the relevant ambient temperatures, elevation and actual air-system losses.
- Account for plant auxiliaries without subtracting the same load twice.
- Test the required outage case using dependable site-net output.
Distinguish three different causes of a lower delivered MW figure#
The word derating is often used to describe every difference between a brochure and a project model. Separate the causes so the buyer can understand what might be changed and what has simply been measured differently.
First, site conditions can change the output of the turbine-generator itself. Second, condition or configuration differences can make the offered asset different from the reference. Third, additional plant loads and electrical losses can reduce the power delivered beyond the generator terminals. These effects should be identified separately even when the supplier uses an integrated performance model.
A reduction caused by an external gas compressor is not the same as a reduction in the turbine's generator-terminal capability. Both can reduce usable campus power, but they imply different equipment and operating decisions. Treating them as one generic percentage makes it harder to compare proposals or evaluate an improvement.
EIA's definition of net generation accounts for electricity consumed in operating the plant. For procurement, draw the specific delivery boundary and identify every load on the generation side of it. The word “net” should be the beginning of that conversation, not the end.
Establish the reference before applying corrections#
ISO reference conditions allow published ratings to be understood against a stated environment, but manufacturers can still use different equipment and loss boundaries. Read the footnotes before combining specifications.
Solar's Titan 250 generator-set data sheet, for example, states a nominal reference at 15°C, sea level and 60% relative humidity, with no inlet/exhaust or accessory losses. GE Vernova's TM2500 page describes net ratings that include inlet and exhaust losses while excluding balance-of-plant equipment. Those are different boundaries.
Before evaluating an offered unit, write down:
- The full configuration and rating document revision.
- Fuel composition and heating-value basis.
- Ambient temperature, elevation or pressure, and humidity assumptions.
- Included inlet, exhaust, generator and accessory losses.
- Whether the machine is new-and-clean or assessed in its present condition.
- The point where electrical output is measured.
Create a boundary ledger before applying a deduction#
Use one ledger for the reference, the offered package and the proposed plant. Mark whether each load or loss is already included in the reported output. Ask the supplier to resolve unknown entries instead of inserting a generic allowance that may duplicate an existing deduction.
| Item | Already included in quoted output? | Information needed for the site case |
|---|---|---|
| Inlet and exhaust losses | Confirm from rating notes | Proposed arrangement and applicable pressure losses |
| Package auxiliaries | Confirm the equipment list | Consumption at relevant operating points |
| External fuel compression | Confirm explicitly | Duty and power source for the site fuel conditions |
| Water or inlet-cooling systems | Confirm enhancement case | Electrical demand and availability assumptions |
| Transformer and distribution losses | Confirm delivery boundary | Electrical design and operating case |
| Shared plant services | Confirm allocation | Fixed and variable loads across dispatch cases |
GE Vernova's LM6000 reference specifications include rating qualifications that distinguish included losses from excluded balance-of-plant equipment. That is why a buyer should carry the source footnotes into the comparison sheet instead of copying only the MW row.
Keep the ledger alongside every quoted site result. If the site design later adds a system or changes a boundary, revise the delivered-power calculation explicitly. A performance number without a traceable set of assumptions becomes difficult to defend as the project evolves.
Evaluate temperature and elevation with the right curves#
Ambient conditions affect turbine performance. The EPA's combustion-turbine technical reference discusses the effect of inlet conditions and elevation on output. It supports the need for corrections, not one universal derating factor for all machines.
Request the applicable performance curves or an OEM-supported calculation for the identified unit. Ask for the site's relevant operating cases, including the temperature used to establish dependable peak-season capacity.
A cool annual average can conceal the condition that determines whether a plant serves its highest required load. Evaluate the time when demand and generation constraints coincide, not just a convenient average day.
For a used turbine, ask whether the assessment includes degradation and any required restoration work. Keep assumed recovery after cleaning or repair distinct from measured current performance.
Include the real inlet and exhaust arrangement#
An inlet filter, silencer, duct or exhaust system is part of the performance case. Identify the losses assumed in the source rating and those expected for the proposed installation.
Ask for the operating case with the relevant filter condition and the selected exhaust equipment. If aftertreatment or heat-recovery equipment is included, obtain the corresponding performance assessment rather than attaching it to an unmodified catalogue rating.
Avoid subtracting several generic allowances from one another as though the effects were always independent. Use an integrated calculation where the supplier can support it, with each assumption visible.
Evaluate inlet cooling as a complete option#
An inlet-cooling proposal should specify the expected output benefit across relevant weather conditions, its own electrical demand, water needs, maintenance and availability. Ask what performance remains if the cooling system is unavailable.
Do not compare an enhanced rating against an unenhanced alternative without showing both scopes. The purchasing comparison should state the additional capital, operating costs and dependencies required to produce the claimed output.
For systems involving water, use the gas turbine water requirements guide to identify the supply and treatment questions. Keep inlet cooling and other enhancement systems distinct from the underlying engine rating.
Convert available output into delivered power#
Suppose a hypothetical package has a 50 MW reference rating. Engineering determines that it can produce 44 MW at the generator terminals in the specified site case. Additional loads and losses outside that boundary total 2 MW per operating package. For this simplified example, assume those per-package loads remain unchanged in the outage case and that there are no additional shared plant loads.
| Calculation step | Illustrative value |
|---|---|
| Reference rating | 50 MW |
| Site-assessed generator-terminal output | 44 MW |
| Additional loads and losses outside that boundary | 2 MW |
| Power delivered at the required point | 42 MW |
The delivered result is 16% below the initial reference: ((50 − 42) ÷ 50) × 100% = 16%. That percentage belongs only to this invented example. It is not a rule for the LM2500, LM6000, SGT-800 or another turbine.
If three such units are installed, loss of one leaves 84 MW under the same assumptions. That can change a purchasing decision that looked adequate when screened as three 50 MW nameplates. Apply the actual project's contingency and transient requirements separately.
Model shared auxiliaries correctly during an outage#
The earlier 42 MW example assumed identical per-package loads and no additional shared demand. A second hypothetical case shows why that assumption matters. Suppose each operating unit produces 44 MW at its generator terminals and uses 1 MW of dedicated auxiliaries. Assume the plant also has a constant 3 MW shared load, with no other losses in this simplified calculation.
| State | Running units | Generator-terminal total | Dedicated auxiliaries | Shared load | Delivered power |
|---|---|---|---|---|---|
| Normal | 3 | 132 MW | 3 MW | 3 MW | 126 MW |
| One unit unavailable | 2 | 88 MW | 2 MW | 3 MW | 83 MW |
Allocating the shared load equally across three units would give 42 MW per unit in normal operation. Multiplying that allocated value by two after an outage would predict 84 MW. The actual result under the stated assumptions is 83 MW because the shared load remains 3 MW rather than falling to 2 MW.
The lesson is not that shared loads are always constant. Some change with dispatch, weather or plant state. The point is to model the outage state directly rather than carrying a normal-operation allocation into a different configuration.
If a 84 MW requirement had been approved using the shortcut, the one-megawatt difference would matter. Resolve it through the actual auxiliary-load schedule and performance assessment, not by rounding the result or adding unsupported optimism about the remaining machines.
Request a performance schedule with the offer#
| Required case | What the supplier should report |
|---|---|
| Reference | Configuration, conditions and rating boundary |
| Normal operation | Net delivered MW and heat rate at the expected load |
| Seasonal capacity case | Output at the specified limiting ambient conditions |
| Enhancement unavailable | Output without the relevant supporting system |
| Defined outage | Remaining plant output and associated operating assumptions |
Ask which results are estimates and which are contractual commitments. The acceptance method should define measurements, corrections, tolerances and responsibilities before the commercial terms are finalized.
This assessment complements the part-load guide: site conditions determine available performance, while the dispatch plan determines where within that capability the plant operates.
Request a matrix of conditions rather than a single corrected number#
A useful performance request defines the conditions relevant to the purchase. Ask for the required output and heat rate at the chosen seasonal capacity case, normal operation and the early demand phase. Specify which enhancement systems are available in each case.
If the project relies on cooling or another enhancement during its most demanding period, include a case where that system is unavailable. This reveals whether the dependency changes the unit count, the accepted load or the operating strategy. It also gives the owner a basis for evaluating the supporting system's maintenance and redundancy.
For a used machine, identify whether the forecast describes current condition or an assumed condition after restoration. If cleaning, repair or replacement work is expected to recover performance, state the basis and the evidence required to confirm it. Avoid treating all forecast recovery as achieved output before the work is completed and evaluated.
Ask which inputs are fixed by the project and which are estimates. A performance curve can be technically sound while relying on a fuel or inlet arrangement that the site will not actually provide. The comparison becomes useful when those interfaces are reconciled.
Agree how a site test will be interpreted#
A test occurs under the conditions present during the test. A purchase commitment may concern different reference or site-design conditions. The engineering team needs an agreed method for connecting those two cases and for reporting uncertainty.
ASME's performance-test-code overview identifies PTC 22 for gas turbines, PTC 46 for overall plants and PTC 19.1 for test uncertainty. Those references help define the scope of an appropriate procedure; this guide does not prescribe an edition, reproduce the standards or claim a particular asset complies with them.
Before the test, document the measurement boundary, instrumentation, operating sequence and correction method. Identify who reviews the data and how an inconclusive result will be handled. A seller and buyer should not first discover that they mean different things by “guaranteed output” after commissioning.
For a preliminary enquiry, the same discipline can be simplified into a performance schedule: required MW, delivery point, ambient case, fuel, included auxiliaries and enhancement status. That schedule makes the LM2500, LM6000 and SGT-800 sourcing discussion about usable power rather than an attractive reference number.
Frequently Asked Questions#
What percentage should I subtract for hot weather?#
Use the applicable curve or performance assessment. A generic percentage can produce a false sense of precision because configuration, temperature range and site systems differ.
Does “net output” include gas compression?#
Only if the stated boundary includes it. Obtain a list of included auxiliary loads rather than relying on the word “net.”
Can a turbine exceed its published reference output in cold conditions?#
Some configurations may produce more under certain conditions, but generator, controls and operating limits still apply. Use approved performance information and do not treat an extrapolated curve as an operating authorization.
Can shared auxiliary loads be divided equally between units?#
They can be allocated that way for a stated reporting purpose, but the allocation may not predict an outage case. Recalculate shared and dedicated consumption for the actual number of running units and operating conditions.
Is a site-performance estimate a guarantee?#
Not automatically. Ask the supplier to label estimates, references and contractual commitments separately. Any guarantee needs a defined measurement boundary, conditions, test procedure and commercial treatment of the result.
Source equipment against your site conditions#
Review the LM2500, LM6000 and SGT-800 dossiers. Through Get Quote → I'm looking to buy equipment, provide the location, relevant ambient conditions, elevation, fuel, electrical requirements and minimum net MW at the delivery point.