Steam Path Audit: Finding Lost Turbine Efficiency
Author
Hisham Abdalla
Date Published

Disclaimer: Research and analysis by the engineering team. The worked example uses hypothetical plant data. Sources referenced below.
A steam path audit finds where a steam turbine has lost efficiency and puts a power figure on each loss. It runs in two stages. With the machine running, an enthalpy drop test gives the isentropic efficiency of each section. With the casing open, measurements of seal clearances, surface roughness, erosion and deposits turn that figure into kilowatts by cause, and the list of causes becomes a repair list in order of value.
Strictly, the name belongs to the open-casing survey, but the survey is worth far more with a running test to read it against. The running test tells you whether there is a loss and in which section. The open-casing audit tells you why, and what putting it right is worth. This article works one efficiency calculation from steam tables, names the test codes, and shows how a loss tally sets repair priorities. It gives no study price, because we cannot source one; it lists what to ask a provider for instead.
Stage One: What a Running Test Tells You
A running test needs no outage. You measure steam pressure and temperature at the inlet and exhaust of a section, calculate its efficiency, and compare it with the baseline from when the machine was new or just overhauled. The result locates a loss. It does not name the cause, because leakage, deposits, erosion and a damaged nozzle can all produce the same fall in efficiency.
Section efficiency by enthalpy drop. Isentropic efficiency is the actual enthalpy drop across a section divided by the drop an ideal, loss-free expansion would give between the same inlet state and the same exhaust pressure. EPRI's 2012 summary of monitoring, inspection and repair calls the test straightforward in superheated sections, given accurate inlet and exhaust pressure and temperature. A section that exhausts wet steam needs a moisture measurement as well, because temperature no longer fixes the enthalpy.
First stage pressure against flow. EPRI offers a simple screen: plot first stage pressure against flow and watch for drift. In our reading, a higher pressure at the same flow means the path downstream of the tapping has narrowed, usually from deposits, and a lower one means the area has opened up through erosion or damage. Our steam turbine diaphragm repair article works the arithmetic for a nozzle row.
Hold the conditions constant. Test at the same valve position every time. We recommend the governing valves wide open, so that throttling loss does not move between tests. Take the baseline when the machine is at its best, after an overhaul with new seals.
Instruments. You need calibrated pressure and temperature measurement at each section inlet and exhaust, a flow figure from a calibrated meter or a heat balance, and electrical output at the generator terminals. Instrument error matters more than most plants expect, as the example below shows.
What goes wrong. Unsteady load, because a section efficiency taken while load is moving means little. A wet exhaust, where an enthalpy worked out from temperature is invalid. A flow figure from a meter nobody has calibrated. Thermowells with poor immersion or no insulation, which read the pipe wall and not the steam. Wait for steady conditions, log long enough to average out noise, and photograph the installation so the next test repeats it.
Standards. ASME PTC 6 is the full code for accurate testing of steam turbines; the edition is PTC 6-2004, reaffirmed in 2014 and now on stabilised maintenance. ASME PTC 6S covers routine tests through the machine's life. Its edition is 1988, reaffirmed in 2019. Each of its procedures carries an estimated repeatability value, and ASME says to use it when judging whether a change from earlier tests is significant. On the IEC side, IEC 60953-1:1990 sets high accuracy rules for large condensing turbines, IEC 60953-0:2022 covers a wide range of accuracy for turbines of any type and size and replaces the old Part 2, and IEC 60953-3:2022 covers retrofitted turbines. Check which edition your contract or guarantee cites.
How often. Run the test after every overhaul to set the baseline, then at a fixed interval, and after any event that could change the steam path: a trip, a rub, a chemistry upset or a load excursion. The PTC 6S repeatability value tells you how large a change you can believe.

The two stages of a steam path audit and the question each one answers. Source: EPRI, Steam Turbine Monitoring, Inspection and Repair, 2012; MIMAH engineering practice.
A Worked Isentropic Efficiency Example
Steam properties below come from IAPWS-IF97, the industrial standard formulation, calculated with the open-source Python package iapws (version 1.5.5). They agree within 0.2 kJ/kg with the NIST Chemistry WebBook, which uses the IAPWS-95 formulation, so you can check them yourself. If you do, expect today's efficiency below to come out at 81.1 percent and not 81.0: that is the formulation, not the plant.
Take a hypothetical back-pressure turbine. Steam enters at 4.0 MPa absolute and 450 °C and leaves at 0.8 MPa absolute.
Inlet state. Enthalpy h1 = 3331.0 kJ/kg and entropy s1 = 6.938 kJ/kg·K.
Ideal exhaust. An isentropic expansion to 0.8 MPa holds the entropy at 6.938 kJ/kg·K. That state has h2s = 2898.4 kJ/kg at 226.1 °C. Saturation temperature at 0.8 MPa is 170.4 °C, so the steam is still superheated and the temperature method works.
Ideal enthalpy drop. 3331.0 − 2898.4 = 432.6 kJ/kg.
Baseline test, after overhaul. The measured exhaust temperature is 258 °C, so h2 = 2967.7 kJ/kg. The actual drop is 3331.0 − 2967.7 = 363.3 kJ/kg. Efficiency is 363.3 / 432.6 = 84.0 percent.
Test today. At the same inlet and exhaust pressure, the exhaust temperature reads 264 °C, so h2 = 2980.6 kJ/kg. The actual drop is 3331.0 − 2980.6 = 350.4 kJ/kg. Efficiency is 350.4 / 432.6 = 81.0 percent.
Efficiency has fallen by 3.0 points, which showed up as a 6 K rise in exhaust temperature. With a hypothetical steam flow of 20 kg/s, the baseline power from the steam is 20 × 363.3 = 7,266 kW and today's is 20 × 350.4 = 7,008 kW. That is a loss of 258 kW before mechanical and generator losses.
On a back-pressure machine the steam flow is set by the process, so the lost power has to be bought or generated elsewhere. The energy has not vanished, though. The exhaust now carries 2980.6 − 2967.7 = 12.9 kJ/kg more, and 20 × 12.9 = 258 kW of extra heat goes to the process. If the process can use that heat, the real cost is the power price less the value of the heat.
Now see how fragile the number is. One kelvin of error in the exhaust temperature changes h2 by about 2.1 kJ/kg, which moves the efficiency by 2.1 / 432.6, about 0.5 points. An inlet pressure reading that is 1 percent high, 4.04 MPa for 4.00, moves it by about 0.5 points as well. A thermocouple reading 6 K high would show the same 3 point loss on a healthy machine. Calibrations get recorded and tests get repeated the same way each time for exactly this reason.

A worked isentropic efficiency example with hypothetical plant data. Steam properties from IAPWS-IF97, checked against the NIST Chemistry WebBook. Source: IAPWS-IF97; NIST Chemistry WebBook.
Stage Two: What the Open Casing Adds
When the casing is open, the audit measures what the running test could only infer. The usual list is seal and packing clearances, surface roughness of nozzles and blades, trailing edge thickness and throat openings, erosion and damage on blades, deposits, and leakage paths at joints.
Clearances come first for a reason. The 2013 symposium paper by Siemens and Saudi Aramco engineers puts steam leakage past the blade shrouds and roots, and between the shaft and the diaphragms, at 76.5 to 81.0 percent of turbine performance deterioration. Its source is a 1999 Demag Delaval document on retractable packing, drawing on research from the former USSR and the USA, so treat the share as indicative and not as a rule for your machine. The direction is plain physics: steam that leaks does no work.
Measure as found, before cleaning, and again as left after repair. The difference is the recovered kilowatts, and without the as-found numbers you cannot prove what the repair achieved.
Each measurement is then compared with design, stage by stage, and the difference is converted to a section efficiency loss. A 2005 conference paper by a US waste-to-energy plant operator describes the result. A steam path audit with computer modelling of the blade path found a 2.5 MW shortfall against design. Work done at the regular overhaul then gained almost 1 MW, and the audit guided which seal packing to replace or repair. That is about 40 percent of the shortfall on one plant, so discount recoverable kilowatts before you promise them to anyone.
From Kilowatts Lost to Repair Priorities
Return to the 258 kW. Suppose the open-casing audit attributes it as follows: shaft packing leakage 110 kW, tip and root seal leakage 70 kW, nozzle trailing edge erosion 48 kW, and surface roughness and deposits 30 kW. These add to 110 + 70 + 48 + 30 = 258 kW.
Now put a value on each kilowatt. Suppose the plant runs 8,000 hours a year, replacement electricity costs $0.10 per kWh, and the extra exhaust heat has no value to the process, all labelled inputs. One kilowatt recovered is worth 8,000 × $0.10 = $800 a year. The shaft packing leakage is then worth 110 × $800 = $88,000 a year, the tip and root leakage $56,000, the nozzle erosion $38,400 and the roughness and deposits $24,000, which total $206,400.
The payback rule follows. Divide any repair quote by the yearly value of what it recovers. A packing renewal quoted below $88,000 pays back within a year if it recovers all 110 kW, and every $100,000 of cost pays back in 100,000 / 206,400 × 12, about 5.8 months, if the whole 258 kW comes back. Those are arithmetic on invented inputs, not forecasts. Real repairs recover part of a loss.
Rank by value recovered per unit of cost, then adjust for two things the kilowatts miss. Cracks, rubs and thrust damage go to the top whatever their power value, because they threaten the machine. And an item that needs the casing opened anyway costs less than one that would extend the outage. Our steam turbine overhaul cost article covers pricing and contract models for that work.

A hypothetical loss tally from an open casing audit, valued at 8,000 hours a year and $0.10 per kWh with no value placed on the extra exhaust heat. Source: MIMAH engineering practice; invented inputs.
What a Running Condition Assessment Costs, and What to Ask For
We will not quote a price. It depends on the number of sections tested, who supplies and calibrates the instruments, the days on site and whether open-casing measurements are included, and we found no published figure we would trust.
Ask the provider for these items as separate lines: the sections and test points covered, the instrument list with calibration status, the test code and edition used, an uncertainty statement for each efficiency, the baseline you will be compared against, and a report that gives kilowatts by cause. Then compare the quote with the yearly value of the loss it is looking for. In the example that is $206,400 on invented inputs. If your first test finds no loss, you have bought a baseline, and you will need one anyway.
Where to Start
Take a running test at the next steady load and keep it as your baseline. Start trending first stage pressure against flow now, since the instruments are usually there already. Put the open-casing measurements into the scope of your next outage, using the steam turbine inspection checklist for what to record, and read steam turbine blade damage for what the damage patterns mean. If the audit points at a diaphragm, steam turbine diaphragm repair covers the options. A wider plant view, including condensers and steam traps, is in our power plant energy audit article.
MIMAH's industrial engineering team does root cause analysis, performance diagnostics and technical audits on turbines and rotating equipment, and is independent of any turbine manufacturer. If you want a baseline test or a loss tally for a machine you suspect has drifted, get in touch.
