Gas, Gas Turbines, Reciprocating Engines

A Cheaper HRSG with Advanced Gas Turbines

Issue 3 and Volume 121.

When and How It Can Make Sense

By S. Can Gülen, Ilya Yarinovsky, and Dave Ugolini, Bechtel Infrastructure & Power Inc.

Present state-of-the-art in gas turbine combined cycle (GTCC) design is three-pressure, reheat (3P-RHT) steam bottoming cycle with steam generation at three different pressure levels. The goal is to maximize total steam generation and steam turbine generator power output for a given gas turbine exhaust energy and, thus, to maximize combined cycle efficiency. There are three “knobs” available to the designer to dial in that maximum – all dictated by fundamental thermodynamic considerations:

  • Steam mass flow rate
  • Steam availability (exergy)
  • Heat rejection temperature (steam condenser pressure)

All three knobs have a significant impact on bottoming cycle equipment size (footprint and weight) and cost via following mechanisms:

  • HRSG (Heat Recovery Steam Generator) heat transfer surface area
  • Condenser and cooling tower heat transfer surface area
  • Pipe, tube and steam turbine/valve casing/shell materials (high grade stainless and/or alloy steels)
  • Steam turbine exhaust annulus area (last stage bucket (LSB) length)

Furthermore, constant pressure-temperature boiling characteristic of the cycle working fluid, H2O, necessitates steam generation at multiple pressure levels to minimize the heat transfer irreversibility in the HRSG. Presently, three pressure levels (high, intermediate and low, HP, IP and LP, respectively) is the industry standard. State-of-the-art design parameters are summarized in Table 1, which has two columns: moderate and aggressive. Admittedly, these qualitative monikers are somewhat arbitrary (e.g., why 125 barg for the “moderate” HP steam pressure and not, say, 115 barg?). Nevertheless, there is no purely physics-based, clear-cut delineation that one can use as a yardstick. This difficulty can be traced back to the fact that there is no fixed product family classification for the bottoming cycles analogous to the “class hierarchy” for heavy-duty industrial gas turbines (i.e., the “topping cycle” of the combined cycle). Thus, some amount of fuzziness in labeling bottoming steam cycle designs is unavoidable.

The term “maximizing” used in reference to the ST power output has two connotations: “make as much as possible” and “make the best use of [what?]”. The [what?] in question is, of course, capital cost of the resulting system. Otherwise, simply building equipment as large as possible using the most exotic materials with no regard to cost, footprint and ease of construction would lead to higher and higher performances (up to a certain limit, of course, set by the second law of thermodynamics). This, in fact, is pretty much the approach taken by the OEMs in advertising “world record” combined cycle efficiency ratings as well as achieving such “world record” performances in showcase power plants with highly advantageous site characteristics (e.g., proximity to a year-round available cooling water source such as a river or ocean). Unfortunately, this is not exactly a widely reproducible and/or sensible business approach to the problem at hand.

In fact, each bottoming cycle is a tailor-made system specific to the particular project strongly dependent on owner/developer’s financial criteria, site conditions and prevailing (or projected) economic climate. There are some discontinuities (or break-points) introduced mainly by steam turbine OEMs’ product line portfolios (essentially casing/shell configuration and LSB size) and, to some extent, other equipment vendors’ in-house design practices (e.g., HRSG “box” sizes, cooling tower cell/fan sizes, etc.) but, otherwise, this is a fairly continuous design spectrum.

In this article, it is postulated that, in the existing power generation climate and economic environment, aggressive bottoming cycle designs are not warranted. Justification for this postulate is provided via a deterministic approach (i.e., LCOE calculations) by (i) considering a two-pressure reheat (2P-RHT) design and (ii) evaluating the feasibility of advanced steam cycle parameters (steam pressure and temperature). In lieu of a rigorous probabilistic approach (well beyond the scope of the paper), a sensitivity analysis is done to show that the conclusions are pretty robust to reasonable fluctuation in key parameters. The article is a condensed version of the full paper presented in PGI 2016 conference in Orlando, FL, in December 2016.


There are four important parameters in the LCOE equation and the “tug-of-war” between them constitutes the key to optimization (i.e., LCOE minimization):

  • The tug-of-war between specific capital cost, k in $/kW, and (i) plant load factor, λ, and (ii) annual operating hours, H
  • The tug-of-war between fuel price, f, and thermal efficiency, h0

Budgetary Price Data -1

In particular, investing a lot of capital into a power plant (i.e., high k) in order to buy as much efficiency as possible (i.e., high h0) can only be justified if

  • the expected/projected electric energy generation (kWh or MWh) is commensurately large, i.e.,

– High plant load factor (i.e., more kW or MW) and/or

– High annual operating hours (i.e., high capacity factor)

  • the fuel price f is high

Each parameter is looked at separately below.

What amount of extra capital investment into the bottoming cycle is justified by the improved combined cycle efficiency via increased steam turbine generator (STG) output? This is a fundamental question, whose answer is dictated by the basic principles of gas turbine (GT) combined cycle power plant thermodynamics and economics. This can be easily verified by data extracted from the budgetary price numbers listed in Gas Turbine World 2014-15 Handbook for simple and combined cycle GT power plants (Figure 1). For a large heavy-duty gas turbine generator (GTG) of a 300+ MWe size, say, each kilowatt from the bottoming cycle costs more than six times that from the topping cycle (see Figure 1). Note that budgetary prices reflect a “bare bones” EPC turnkey scope assuming “overnight construction”. Transportation, project-specific options, indirect costs such as contingencies, owner’s costs and interest during construction are not included. These items can typically add 30-40 percent to the budgetary price.

EIA Capacity Factor Data for Natural Gas-Fired Combined Cycles – 2

U.S. DOE Natural Gas Price Forecasts and Actual Prices – 3

In order to put the situation depicted in Figure 1 into a plant-level quantitative perspective, assume a 500 MWe GTCC power plant at 60 percent net efficiency (5,687 Btu/kWh). Additional 5,000 kW bottoming cycle output costs about $7.5 million in budgetary price and “buys” 0.6 percentage points of efficiency or 57 Btu/kWh heat rate. In other words, assuming $675/kW for the CC budgetary price per GTW 2014-15 Handbook, each one Btu/kWh reduction in net heat rate comes at a cost of $75,000. Is this a good trade-off? In order to answer this question, we need to point out several key factors based on available industry data.

Annual operating hours are typically expressed in terms of the capacity factor. U.S. Energy Information Administration, Electric Power Monthly, Table 6.7a provides monthly capacity factors for 16 different fossil and non-fossil fuel and technology combinations. The data for natural gas-fired combined cycle power plants is summarized in Figure 2. Prior to 2010, these plants were run at a very low capacity factor (CF) but the situation changed quite dramatically in recent years. Clearly, shale gas “boom” and ensuing low natural gas prices played a significant role in this. Even so, it is hard to envision that the annual average CF for natural gas-fired CC plants will be much higher than 55-60 percent in the foreseeable future (especially with increasing renewable resource penetration). Translation from CF to annual hours, H, is subject to uncertainty since the annual average load factor is not known and there is significant HRSG supplementary firing and GT inlet conditioning to boost output, especially during summer. A wide variation from plant to plant is to be expected (more on this later). For example, a daily-cycled CC power plant with weekend shutdown and two weeks of scheduled maintenance, will run only 50x5x16 = 4,000 hours per year, which corresponds to a CF of 0.75×4,000/8,760 = 35 percent (load factor of 0.75, no supplementary firing or GT inlet conditioning). However, capacity factors in Figure 2 are significantly higher, which is an indication of significantly higher load factor (e.g., more hours at full load), power augmentation (via supplementary firing and/or GT inlet conditioning) or a combination thereof.

Comparison of Gaseous Fuel Prices in U.S., Europe and Japan – 4

GTCC Efficiency Evolution 1985-2015 – 5

Long term natural gas (NG) price forecasts are difficult to make as illustrated by the chart in Figure 3, which superimposes outlooks by the U.S. DOE (consistently predicting increasing scarcity and rising prices) and The National Petroleum Council (NPC), with the latter comprising pessimistic (reactive path) and optimistic (balanced future) scenarios.

Except for the 2003-2008 period, when prices spiked above historical levels due to a tight market caused by several factors, i.e., weak supply and growth in demand for peaking power in particular, long term expectation of annual ~2 percent growth in NG prices pretty much held (going back to the Carter administration era and the Alaska Natural Gas Transportation System (ANGTS) project). Right after that peak price period, development of new sources of shale gas, driven by hydraulic fracturing and horizontal drilling technologies, has more than compensated for the decline in conventional supply, and has led to major increases in reserves of US natural gas. The so-called shale gas boom, although not a guarantee by any means, is expected to prevent non-seasonal, years-long price spikes and exorbitant long-term growth rates in the USA. The situation is somewhat different in Europe and Japan (see Figure 4).

What about the thermal efficiency? Historical GTCC efficiencies (rating, i.e., “advertising”, numbers as well as selected “field-clocked” values) are depicted in Figure 5. Also included in the same graph are the average efficiency of top twenty (in terms of heat rate) gas fired GTCC plants in the U.S. in 2004-2015 – including duct-fired units – which squeaked past 55 percent (LHV basis) only in the last couple years. (This was probably driven by the commissioning of the more advanced FA/H class units and increasing load factor – it is difficult to glean from the data, which includes only generation and fuel consumption numbers.) As illustrated by the min-max range, a select few registered as high as 57 percent whereas most plants (remember: these are among the twenty best in terms of performance – imagine the rest!) were clocked at only about 53 percent!

Coming back to the question posed at the beginning (i.e., additional 5 MW bottoming cycle output at $7.5 million extra cost – a good trade-off or not?), using the LCOE formula and assumptions, the answer is “it depends”. For a GTCC plant with cyclic duty, the value of the proposed improvement of extra 5 MW output is about $6 million for a fuel price of $4/MMBtu (HHV) or about $50,000 per each one Btu/kWh reduction in net heat rate. The fuel price to make it worthwhile at $7.5 million cost adder is about $6.50/MMBtu (HHV). Alternatively, at $4 fuel, the plant should run around 5,800 hours per year at base load duty (load factor of 0.9) to justify the $7.5 million cost adder.


Today’s state-of-the-art in the bottoming cycle technology makes incremental improvements very costly to the point that, at least in the USA, at prevailing natural gas prices, even a third pressure level in the HRSG becomes a “luxury”. (This is explained in more detail in the PGI 2016 conference paper.) In order to take a closer look at this premise, a detailed performance-cost trade-off analysis is undertaken.

Three HRSG OEMs are provided heat and mass balance data roughly corresponding to three variants with (i) same natural gas -fired GT exhaust gas conditions (J class, ~1,500 lb/sec and 1,175°F) at ISO base load and (ii) same HP throttle conditions (nominal 1,800 psig and 1,050°F):

  1. Base (Conventional) Case: 3P-RHT with normal HRSG pinch deltas
  2. “Cheap” Case A: 3P-RHT with large HP evaporator pinch delta
  3. “Cheap” Case B: 2P-RHT with normal HRSG pinch deltas

HRSG equipment price differential of the “cheap” designs from the OEMs are summarized in Table 2. In addition, quantities and man-hour savings resulting from the elimination of the IP section of a similar HRSG unit by one of the three OEMs (in terms of size, configuration and steam production), which was erected at a recent U.S. combined cycle project. Deleted commodities and associated labor included large-bore pipe, valves, supports, and welds, small-bore piping, IP steam drum, pressure relief valves (flanged), IP relief valve silencers and their support steel and platforms, instruments, HRSG hydro testing, chemical cleaning and pipe installation and IP two-row wide box deletion. The resulting saving was equivalent to slightly above 10,000 man-hours.

It is quite clear that nearly $1 million saving in equipment price is achievable via a cheaper HRSG. This could be obtained via either the removal of the IP section or a cheaper HP section (i.e., less HP steam production). Including the savings in erection materials and labor, the former is the preferable route with around $1.7 million total saving per HRSG (i.e., averages for Case B, ~$1 million in price plus ~$750K for construction in Table 2).

Per OEM feedback, HRSG price delta between 2,400 and 1,800 psig cycle is around $500K based on tube, drum and pipe thicknesses with consideration for valve classification.

Two cases are evaluated in order to estimate capital investment savings in a 1x1x1 single-shaft GTCC similar to that proposed for an actual CC project. The base case is set as follows:

  • J Class gas turbine (natural gas-fired, with inlet evaporative cooler)
  • Design ambient conditions 90°F, 40 percent relative humidity
  • 3P-RHT unfired steam cycle: 2,415 psia HP throttle with 1,050°F for HP and hot reheat steam admission
  • Air-cooled condenser at 3.5 inches of mercury
  • HRSG evaporator pinch deltas 15 degrees F

The second “cheap” case is based on a 2P-RHT steam cycle with 1,815 psia cycle and 25 degrees F HP evaporator pinch delta. LP admission pressure is the same as in the base 3P-RHT case. In this case, GT fuel gas performance heating (to 410°F, same as in the base case) utilizes hot feed water from a dedicated economizer section. The GT is fired 8 degrees F higher to maintain the same GTCC net output as the base case. (The implicit assumption here is that the GT in question is one of the latest H/J class machines with the most advanced technology – superalloys, coatings, cooling schemes, etc. – allowing the OEM limited “wiggle room” about the nominal TIT of 1,600°C. For a GT quoted by the OEM at its extreme capability, of course, this is not a feasible option.) Cycle performances are calculated using Thermoflow’s GT PRO software. Total overnight cost is calculated using the PEACE add-in with calibration per above. The results are summarized in the Tables 3-5 below.

Gas turbine and generator price is from PEACE with some adjustment per GTW 2014-15 Handbook budgetary price data. Steam turbine price is also from PEACE with calibration based on in-house data. HRSG equipment price difference in Table 3 can be broken down as follows:

  • $500K for 2,400 to 1,800 psig steam cycle
  • $525K for HP evaporator pinch increase by 10 degrees
  • $1 million for IP section elimination

The “Mechanical” cost bucket in Table 4 includes on-site transportation, rigging, equipment erection assembly plus piping (materials plus labor). The difference of about $2.4 million between the “Base” and “Cheap” versions can be broken down as follows:

  • $750K for IP section elimination (see Table 2)
  • $1 million for 2,400 to 1,800 psig steam cycle
  • $600K for HP evaporator pinch increase by 10 degrees

The latter two are estimated by the PEACE program and mainly driven by smaller and lighter HRSG.

They have not been verified by detailed construction material take-off and labor estimates.

The “Civil” cost bucket in Table 4 includes site work, excavation and backfill, and concrete foundations (including rebar). The difference of about $2.4 million between the “Base” and “Cheap” versions can be broken down as follows:

  • $900K for 2,400 to 1,800 psig steam cycle
  • $400K for HP evaporator pinch increase by 10 degrees
  • $900K for IP section elimination

All three are estimated by the PEACE program and mainly driven by reinforced concrete foundation material and labor for the smaller and lighter HRSG.

They have not been verified by detailed construction material take-off and labor estimates.

Clearly, even at $5 natural gas, which is on the expensive side for the U.S. market in the foreseeable future, investing into the bottoming cycle for a few Btus of heat rate clearly does not pay off. At $5 fuel, for LCOE parity between base and “cheap” cycles

  • TOC saving of ~$2.6 million is sufficient for cyclic operation whereas
  • TOC saving of ~$4.1 million is required for baseload operation

At ~$11 million TOC saving, for LCOE parity between base and “cheap” cycles

  • Fuel price must exceed $30/MMBtu for cyclic operation whereas
  • Fuel price must be nearly ~$16/MMBtu for baseload operation

It is amply clear that, unless natural gas prices are exorbitantly high and/or the power plant in question is intended to assume a truly baseload duty, there is no case to be made for an expensive bottoming cycle. (Note that even if PEACE “Mechanical” and “Civil” estimates are off by 50 percent, the TOC saving is $8.7 million and contains enough margin to support this conclusion.)

One may justifiably object to the comparison in Table 5 by pointing out the 8 degrees F higher firing temperature for the “cheap” case.

Here’s the rationale: The reason for the higher firing temperature is to equalize the net output of the two cases.

Otherwise, the cheaper 2P-RHT design would have 3.7 MWe lower output (with about the same cost delta and slightly higher LCOE, i.e., $96.92 and $67.87 per MWh for cyclic and baseload duties, respectively, but still lower than those for the more expensive 3P-RHT variant).

The chain of thought goes as follows:

  • By going from the cheaper bottoming cycle to the expensive one, extra 3.7 MWe output is “bought” by paying $11 million.
  • This is equivalent to 45 Btu/kWh better heat rate – at exact same fuel consumption!
  • The question to ask is this: Which one is cheaper?

– Buying 3.7 MWe output for extra $11 million, or

– Buying 3.7 MWe output by extra fuel consumption

The answer, via LCOE analysis, turns out to be the latter. (Note that heat rate improvement more than compensates for marginally higher fuel burning and the heat rate delta improves to 37 Btu/kWh.)

Another reasonable objection would be “what about the expensive bottoming cycle and the 8 degrees F higher firing temperature?.” To answer that, consider the performance and LCOE comparison of four possible design permutations in Table 6.

On a truly “apples-to-apples” basis, obviously, the performance delta between the “expensive” and “cheap” bottoming cycles is about 3.7 MWe and 45 Btu/kWh of heat rate. At $5 fuel, the LCOE comparison favors the latter.

On a “gross margin” basis (the difference between the market price of energy and the variable generation costs), it is true that the “Base” variant has a slight advantage.

However, it is not significant enough to severely impact the eventual place of the particular GTCC configuration selection (from owner/developer perspective) in the “merit/economic dispatch order” in a large ISO. Thus, the difference between the “levelized revenue requirement” quantified by the LCOE and the forecasted gross margin is the determinant in the selection of one variant over the other. In this case, the “Cheap” variant with the smaller difference should be the preferred configuration.

In general, as long as marginal improvements within a small band, i.e., ±1 percent or less on net output and heat rate, erring on the side of less capex is probably a good bet.

Beyond that, however, commercial considerations, which cannot be encapsulated in a simple metric like LCOE, can take precedence.

In addition, the uncertainty aspect may become more critical to the extent that erring on the side of better performance (a more robust number than fuel prices over the next twenty years) may be the more prudent course of action.


Using fundamental thermodynamic and economic arguments, it is shown that performance improvement via larger HRSG is an uneconomic choice – unless justified by high fuel price and/or plant capacity factor (i.e., a base-loaded unit).

Conceptual analysis predictions are verified by OEM supplied prices and detailed construction estimates.

The output delta is marginal enough that it can be achieved via a small increase in gas turbine firing temperature.

The capital cost saving is sufficiently large to more than compensate for the increase in heat rate so that, under most operating scenarios, the life-cycle LCOE favors the “cheaper” bottoming cycle – albeit with caveats enumerated in the preceding paragraphs.