Why Plants Lose Output Over Time — and How Testing Finds It

A plant rarely loses output in one dramatic failure. It slips: a few megawatts here, a point of efficiency there, until someone looks at the load curve and asks when the plant got smaller. It leaked away, a little at a time, through fouling, wear and auxiliary load — most of it recoverable, or at least explainable, if someone had measured it properly along the way.

Plants get plenty of attention when they trip. Slow decline gets shrugged off as age, blamed on the weather, or buried in a monthly report comparing this month's megawatts with last month's, as if the two months were the same plant in the same weather.

Fouling steals the easy megawatts

Start with the cheapest loss. A gas turbine compressor ingests a huge volume of air, and whatever is in that air ends up on the blades. Dust, salts and oily deposits roughen the surface and narrow the passages, so the machine moves less air and burns fuel less efficiently for the same output. Wash it properly and a chunk of the loss comes straight back. Leave it, and people start treating a fouled machine as a normal one.

The same happens on the steam side. Deposits on HRSG tubes act as insulation between hot gas and water, so heat that should have raised steam goes up the stack instead. Annual HRSG inspections to API 510, the kind I carried out at combined cycle plants, are partly about pressure-part integrity — creep, corrosion, the findings that decide whether a tube stays in service. They also show, tube by tube, where heat transfer is being strangled.

Fouling is reversible, which makes it the first suspect whenever output drifts down without an obvious cause, and the first thing to rule out before anyone talks about overhauls.

Wear is slower, and harder to see

Wear does not wash off. Hot-section parts in a gas turbine survive on coatings, cooling air and tight clearances. Over thousands of hours, coatings thin and clearances open up, and the machine that once made its rated output at a given firing temperature needs more fuel — or more temperature — to make the same power. At Korangi, Level-3 maintenance on the LM6000 fleet meant module replacement, hot-section exchange and HPC blade replacement: the point at which wear stopped being a trend line and became a maintenance decision.

Steam turbines wear the same way, more quietly. Level B and Level C overhauls open the machine to different depths, and what you find inside — erosion, deposits, clearances drifted from design values — explains losses no control-room trend could pin down. Across machine types, wear shows up first as heat rate, the fuel burned per unit of electricity, and only later as lost megawatts. Watch output alone and you catch wear late; watch heat rate and you catch it early.

Parasitic load creeps up quietly

A third loss gets almost no attention, because it never appears on a nameplate. Gross output is what the generators make; net output is what leaves the station after the plant has fed itself — pumps, fans, compressors, cooling systems, all the auxiliaries that keep the cycle running. When I ran parasitic-loss and thermal-efficiency improvement programmes across the Omni plants, this gap was where a surprising amount of output was hiding.

Auxiliary load creeps up for boring reasons: a pump left running that used to cycle off, a fan damper stuck half open, a cooling system working harder because its own heat exchangers are fouled, a leak nobody chased because the make-up pump kept up. None of these trips the plant. Together they widen the gap between gross and net, and the station's commercial output falls while its generators look perfectly healthy. Trend gross output alone and you will never see it.

The control room can mislead you

All of this is tangled up with weather. A gas turbine makes less power on a hot day because hot air is thinner — less mass through the machine, less output, and nobody has done anything wrong. Fuel quality shifts the picture too, as does the load the grid actually asked for. Comparing raw megawatts from January with raw megawatts from June proves nothing. It is the most common false alarm in plant operations, and the most common excuse for a real loss.

That is why a proper performance test matters. It does not compare raw numbers. It corrects them.

A proper test separates weather from wear

A performance test run to ASME PTC 46 for the whole plant, or PTC 22 for a gas turbine on its own, measures output and heat input under controlled, steady conditions, then corrects the results back to an agreed reference: fixed ambient conditions, a stated fuel, a defined test boundary. What comes out is corrected output and corrected heat rate — numbers comparable with the plant's earlier tests, whatever the weather was on either day.

At the 900 MW BQPS-III combined cycle plant, running on RLNG and HSD, the test followed PTC 46 and PTC 22 and the report went to NEPRA. At the 1,263 MW PTPL RLNG plant, where I was team lead, the report went to CPPA-G. In both cases the point was the same: a corrected result every party could check, rather than a raw number every party could argue about.

A good test also splits the loss. Whole-plant output tells you the plant is down, not where. Breaking the result into gas turbine, steam turbine, HRSG and auxiliaries turns a disappointing test into a work list: wash here, inspect there, overhaul this, leave that alone. Without that split, the usual response is the most expensive one — overhaul everything, or overhaul the wrong thing first.

Testing fixes nothing by itself. A test is a diagnosis. Its value is that it stops a plant spending overhaul money on a fouling problem, or washing a machine whose real problem is worn blades.

The Takeaway

Keep one corrected trend and defend it. Pick a reference condition, record corrected output and corrected heat rate on the same basis after every major job, and watch heat rate as closely as megawatts — it moves first. When the trend bends, split the loss before you spend: fouling, wear, auxiliaries, in that order of cheapness. Wash and clean first, inspect second, overhaul last. And before approving major maintenance on performance grounds, test first, so the money goes to the section that actually lost the megawatts.

Further Reading

What a Performance Test Really Proves — ASME PTC 46 & PTC 22 in Plain Words — What a test actually measures, who watches, and what the final report settles.

Heat Rate: The Number That Decides a Plant's Tariff — Why a few points of heat rate move millions, and how a tariff petition argues it.

Preventive, Predictive, Condition-Based — Choosing the Right Mix — How to decide what each asset actually gets, so washing, inspection and overhaul land in the right order.

Performance testing — pillar guide — The full guide to performance testing, where corrected results sit alongside the test codes and reporting.

ASME PTC 46-2015 (R2025) — Overall Plant Performance — The standards listing for the whole-plant test code referenced in this post: corrected power, corrected heat rate and corrected heat input.