Heat Transfer

What’s the Price

of Steam?
Realistic steam prices are necessary
Robin Smith for the efficient operation of a
Centre for Process Integration,
Univ. of Manchester
chemical production site, both to allocate
Petar Varbanov realistic costs to the different businesses
Technical University of Berlin on the site and to provide a true economic
incentive for energy conservation.

T
he basic problem with the question “What is the price of and supplementary firing are two independent degrees of freedom.
steam?” is that it is largely a meaningless question. Unless Steam can also be generated from waste heat within a process.
steam is purchased from a third party according to some However, such in-process steam generation will be assumed here
contract price, steam does not have a cost. Purchased fuel has a to be fixed according to the operation of the process.
cost. Purchased power has a cost. Power exported from the site Multiple steam turbines. Generally, steam turbines have dif-
and sold has a value. Water and chemicals purchased for steam ferent efficiencies, depending on their size, design, age and
generation have costs. Labor and maintenance also have costs. maintenance. For a given turbine, the efficiency varies with
However, steam does not have a direct cost. It is simply an load. Hence, if there are two or more steam paths through the
intermediary between the primary costs (e.g., fuel) and the end utility system via different steam turbines connecting two steam
users. Yet somehow we need to distribute to the businesses on headers, this introduces additional degrees of freedom for inter-
the site the primary costs that result from their use of steam. nal flow distribution.
Letdown stations. Steam can be transferred between headers
Modeling and optimization of utility systems via letdown stations rather than steam turbines. Usually, large let-
To analyze a utility system, it is first necessary to develop a down flows indicate a missed opportunity for power generation.
simulation model (1), which can be done using commercially However, in some instances, letdown station flows can be
available software. The simulation model should allow part-load exploited to bypass constraints in the steam turbine flows at one
performance of the steam system components. It should provide a level in order to exploit the letdown flow at a lower level for
simulation of the complete material and energy balance around the power generation. Also, if the letdown station involves de-super-
steam system, and be capable of predicting the fuel, power genera- heating by injection of boiler feedwater (BFW), the temperature
tion, water requirements, etc. for any condition of the steam sys- at the exit of de-superheating is an additional degree of freedom.
tem. The model must take into account the operating constraints Condensing turbines. Condensing steam turbines provide
around the system, for example with respect to steam flows from utility systems with additional degrees of freedom, generating
steam generation devices and steam flows through steam turbines. extra power, but rejecting heat to atmosphere.
Once such a simulation model has been developed, it can Vents. As with condensing steam turbines, venting steam
be subjected to optimization. The important degrees of free- from low-pressure headers also provides additional degrees of
dom in utility systems are: freedom to increase power generation. While this might seem a
Multiple steam generation devices. Each steam generation waste of steam, if there is a significant price differential between
device within the utility system can use a different fuel or a differ- the price of power and heat, it can be economic. Again, heat is
ent combination of fuels, and usually has its own efficiency that rejected to the atmosphere.
varies with the steam load. The firing in the gas turbine combustor The optimization model for existing utility systems can be
CEP www.cepmagazine.org July 2005 29
Heat Transfer

used to make continuous and dis- Coal Fuel Oil

crete decisions. Discrete decisions Natural
Current: 200.0 Current: 99.8 Gas
relate to the operational status Min: 50.0 Min: 40.0 GT
(on/off) of the devices. For exam- Boiler 1 Max: 250.0 Boiler 2 Max: 180.0 18 MW
ple, it might be possible to switch HRSG
between a steam turbine or an elec- Current: 25.6
Current: 95.9
tric motor on a particular drive. Min: 60.0 VHP (101 bar abs.)
Max: 165.0
The energy balances of the sys- Current: 209.5
20.0
Min: 115.0 Current: 35.5 T6
tem elements include nonlinear Current: 44.0 T4 Max: 335.0 0.0
Min: 15.0 Min: 15.0
terms that result in a nonlinear Max: 75.0 Max: 65.0
HP (41 bar abs.)
optimization, with the potential to Current: 51.9 Current: 65.0
54.5
T5 Min: 30.0 5.0
bring all of the associated prob- Min: 30.0
Max: 90.0 Max: 70.0
lems of local optima (1). Fortu- MP (15 bar abs.)
nately, this difficulty can be over- Current: 20.0 Current: 109.0 55.9
2.0 Min: 20.0 Min: 60.0
come by fixing both the tempera- Max: 20.0 T7 Max: 150.0 1.0
ture and the pressure of the steam LP (3 bar abs.)
LP Vent Current: 65.0
mains during the optimization to Min: 40.0 CW 0.073 128.0
produce a linear optimization Max: 75.0 bar abs.

model, which is straightforward to

solve (1). This is followed by a ■ Figure 1. A typical site utility system for a CPI facility.
rigorous simulation after each opti-
mization step. The linear optimization is repeated, followed fuel, but could include other costs as well. Having calculated the
again by rigorous simulation, and so on, until convergence is cost of the highest-pressure steam, the model will also provide the
achieved. This procedure usually requires no more than four or amount of power generated by expansion of the highest-pressure
five iterations to reach convergence (1). steam to the next-highest pressure. The value of this power can
then be subtracted from the cost of generating the highest-pres-
Using the optimization model sure steam in the utility boilers to give the cost of the next-highest
to set steam prices for fixed loads pressure steam. The calculation is repeated for the next lower
Consider first the case where the steam heat loads for the pressure level, and so on. Thus, the cost of each steam level is the
processes on a site are fixed and the objective is to calculate the cost of the next highest level minus the value of the power gener-
cost of the steam for the allocation of costs to the processes and ated from the expansion from the next highest level. This model
businesses on the site. Establishing the true steam costs for an incorporates the performance data and constraints associated with
existing steam system requires a model that reflects, as much as an actual system, rather than an idealized model.
possible, the performance of the existing equipment and the In extreme cases, this approach to costing steam might lead
constraints in the existing equipment and steam network. The to steam at low pressure having a negative cost. This can happen
model should also include the existing steam heating demands if fuel is relatively cheap and power is relatively expensive. This
for the various processes. is not a fundamental problem, as it reflects the true economics.
This model should first be optimized, as previously described. The site utility system is benefiting from the availability of a heat
The cost to generate the steam in the utility boilers can be calcu- sink for the low-pressure steam. However, it should be empha-
lated from the optimized model. This will mainly be the cost of sized that if the low-pressure steam has a negative cost, this does

Table 1. Fuel data for Example 1. Table 2. Site configuration data for Example 1.

Fuel 1 Fuel 2 Fuel 3 Ambient Temperature 25°C

Coal Fuel Oil Natural Gas BFW Temperature 110°C
Cooling Water Temperature 25°C
Normal Site Power Demand 68.0 MW
Heating
Minimum Power Import 0 MW
Value, kJ/kg 28,000 40,000 52,000
Maximum Power Import 50.0 MW
Price $0.065/kg $0.12/kg $0.22/kg Minimum Power Export 0 MW
$65.0/t $120.0/t $220.0/t Maximum Power Export 50.0 MW
$0.0084/kWh $0.0108/kWh $0.0152/kWh Power Price $0.05/kWh
Cooling Water Price $0.005/kWh

30 www.cepmagazine.org July 2005 CEP

 isentropic efficiency model pre-
Coal Fuel Oil
Natural dicts costs that are too low for
Current: 250.0 Current: 55.4 Gas lower-pressure steam levels. (Of
Min: 50.0 Min: 40.0 GT
Boiler 1 Max: 250.0 Boiler 2 Max: 180.0 18 MW course, the results for the isentrop-
HRSG ic efficiency model depend on the
Current: 28.0
assumption for the isentropic effi-
Current: 105.0
Min: 60.0 VHP (101 bar abs.)
ciency.) It is also clear that there
Max: 165.0 Current: 208.4 are significant differences
Min: 115.0 20.0
Current: 15.0 T4 Max: 335.0
Current: 59.5 T6 0.0 between the costs calculated for
Min: 15.0 Min: 15.0
Max: 75.0 Max: 65.0
HP (41 bar abs.) the base case and the optimized
Current: 40.9 case. Finally, the cost of the low-
Current: 90.0
Min: 30.0 54.5
Min: 30.0 T5
Max: 70.0
0.0 est-pressure steam is negative
Max: 90.0
MP (15 bar abs.) when based either on the isentrop-
Current: 20.0 Current: 108.0 55.9 ic efficiency model or the simula-
0.0 Min: 20.0 Min: 60.0
Max: 20.0 T7 Max: 150.0 0.0 tion model. The most appropriate
LP (3 bar abs.) costs are those from the right-
LP Vent Current: 75.0
0.073 128.0 hand column of Table 3.
Min: 40.0 CW
Max: 75.0 bar abs.

Steam prices
for changing loads
■ Figure 2. Typical site utility system after optimization.
Consider now the case where
not necessarily mean that it is economic to increase low-pressure the steam price needs to be determined for a change in process
steam consumption significantly. As will be seen later, the cost of heat load, for example for the retrofit of a heat exchanger net-
steam is likely to change as its consumption changes. work for increased heat recovery. Alternatively, a project might
It should also be noted that variation in electricity costs will involve an increase in heat demand as a result of commission-
change the optimization and therefore the steam costs. For this ing a new plant or expansion of an existing one.
situation, an average can be taken according to the relative dura- The starting point is a model for the steam system, again
tion of the costs. optimizing for the existing steam heating loads. It might be sus-
pected that the steam cost calculated from the model for the
Example 1: Steam prices for fixed steam loads existing steam loads can be used to provide steam costs for an
Consider now the steam prices for the existing utility system in increase or decrease for a given steam main. Unfortunately, this
Figure 1 when the loads on the various steam headers are fixed. is not the case. The optimum settings for the steam system
The fuels in use are summarized in Table 1 and the site configura- change once the loads for the steam mains change. Constraints
tion data are given in Table 2. There is a demand for very-high- on the existing equipment will also be encountered, and all of
pressure (VHP) steam from various processes on the site to drive this needs to be accounted for.
steam turbines on fixed drives with condensed exhaust. There are Consider again Figure 2 showing the existing site utility sys-
also demands for high-pressure (HP), medium-pressure (MP) and tem (2). Suppose that it is possible to reduce the HP steam
low-pressure (LP) steam for process heating. In addition, steam is demand. This could be done, for example, by improving the
expanded through various steam turbines. heat recovery within the processes that use HP steam.
Figure 2 shows the same utility system after optimization. But what is such a steam saving actually worth? The saving
The steam cost can now be calculated for the base case and the in steam from the HP mains means that it now does not need to
optimized case. Table 3 presents the costs of the
various steam levels based on the fuel value of the Table 3. Steam prices for fixed process steam loads for Example 1.
enthalpy difference between the steam mains, an
Price Based on Price Based on
isentropic efficiency steam turbine model (assuming
Price Based Ideal Steam Turbines Actual Steam Turbines
a typical 85% isentropic efficiency and 95% Steam on Enthalpy Base Base
machine efficiency) and the full simulation model. Level Difference Case Optimized Case Optimized
Table 3 shows that the cost model based on the
VHP $8.32/t $8.32/t $7.94/t $8.32/t $7.94/t
fuel value of the enthalpy difference between steam HP $7.78/t $5.17/t $4.78/t $5.78/t $5.32/t
mains significantly over-predicts the prices of MP $7.26/t $2.21/t $1.83/t $3.34/t $2.61/t
lower-pressure steam levels. On the other hand, the LP $6.54/t –$1.47/t –$1.86/t $0.17/t –$0.55/t

CEP www.cepmagazine.org July 2005 31

Heat Transfer

be expanded through steam turbines from the VHP level. This, to an increase in the HP generation. This leads to a surplus of
in turn, means that there is a surplus of steam at the VHP level. HP steam, which in turn leads to a surplus of VHP steam. Then
The first obvious action to take is to reduce the steam gener- the same arguments apply as to what is the most efficient way
ation in the utility boilers and accept a saving in fuel costs as a to exploit the surplus of VHP steam.
result. Unfortunately, the saving in fuel costs would also be The only way to reconcile the true cost implications of a
accompanied by an increase in power costs. This results from reduction in steam demand created by an energy reduction proj-
the reduction in the flow of steam through the steam turbines, ect is to use the optimization techniques described earlier. An
and additional power would have to be imported (or export optimization model of the existing utility system must first be
power decreased) to compensate. It is therefore not so straight- set up. Starting with the steam load on the main with the most
forward to determine exactly what the cost benefits associated expensive steam, this is gradually reduced and the utility system
with a steam saving would be. re-optimized at each setting of the steam load. The steam load
Another way to deal with the surplus of steam at the VHP can only be reduced to the point where the flowrate constraints
level created by a steam saving would be to pass the heat are not violated.
through an alternative path, say to the condensing turbine. This The concept of steam marginal cost is used as an indicator in
would allow additional power to be generated, with a resulting the analysis. It is defined as the change in utility system operat-
cost benefit. In a complex utility system, the heat can flow ing cost for unit change in steam demand for a given steam
through the utility system through many paths. The flow main (change in steam main balance) (3):
through different paths will have different cost implications.
In assessing the true cost benefits associated with a steam MPSTEAM = ∆Cost/∆mSTEAM
saving, the steam and power balance for the site utility system
must be considered, together with the costs of fuel and power where MPSTEAM = marginal price of steam, ∆Cost = change in
(or power credit if power is exported). cost and ∆mSTEAM = change in steam flowrate.
If steam is being generated, rather than used, then the same It is important to emphasize that the change in the operating
basic arguments apply. For example, suppose HP steam was cost is taken between the optimum operation before the steam
being generated by a process into the HP main in Figure 2. The demand change and the optimum operation after the steam
project to improve the heat recovery in this process might lead demand change for the current step.

Approaches to Steam Pricing: What Not to Do

A common approach to pricing steam starts with first cal- efficiency model (3). Knowing the amount of power that
culating the cost for the generation of high-pressure can be extracted allows the value of the power to be esti-
steam in the utility boilers or heat-recovery steam genera- mated, and this can be subtracted from the price of the
tors (HSRGs) associated with the gas turbines. This is high-pressure utility steam to obtain the price of the
usually dominated by the cost of fuel. Other costs include lower-pressure steam. This seems to be more logical for
those for auxiliary steam and power required for steam utility systems with power generation in steam turbines.
generation, water and treatment chemicals, labor, and so Unfortunately, even this approach has a number of
on. The cost of the fuel could be used as a first approxi- shortcomings. First, it assumes a single steam turbine in
mation of the cost of the high-pressure steam generation. each expansion. There will usually be a number of tur-
One approach to setting the price of lower-pressure bines between each two mains.
steam is based on the enthalpy difference between the Second, the efficiency of power generation is assumed
high- and low-pressure steam. The fuel value of the differ- to be fixed. In practice, the efficiency of steam turbines
ence in enthalpy between two levels is calculated and depends on the load. The load for some turbines can be
subtracted from the price of the high-pressure steam. If a varied, and for these turbines, the efficiency will vary
mix of fuels is being used, then a mean fuel cost weighted according to the steam loads in various parts of the sys-
according to the calorific value of the fuels can be used. tem. Other turbines, perhaps on direct drives, might have
While this sounds reasonable, it is completely inappro- their loads (and hence their efficiencies) fixed.
priate for all but the smallest utility systems, because it Third, it does not account for the existing equipment
does not recognize the potential for the high-pressure for power generation, either from the point of view of its
steam to generate power by expansion in steam turbines. performance, or the constraints within the utility system.
Estimating the potential for power generation from steam Some turbines are larger than others and are likely to be
turbines by expansion between two different steam pres- correspondingly more efficient. All turbines are con-
sures is not difficult. Starting from the highest-pressure strained to operate between minimum and maximum
steam cost, if the pressure of the next-lowest steam level steam flowrates. Complex turbines involving extraction
is fixed, then the amount of power that can be generated or induction have constraints on the extraction or induc-
by expansion can be estimated based on an isentropic tion flowrates.

32 www.cepmagazine.org July 2005 CEP

 Steam demand for a given main can increase as a result of caused when, after optimization, further saving in steam along
an increase in production rate or decrease in order to improve some path in the utility system is restricted by a constraint, typi-
the energy efficiency of the site. The same approach can be cally a lower bound on a flowrate.
used to deal with any context. An important point to emphasize is that when the process
The first step is to optimize the operation of the utility system steam load changes, the cost of a given level of steam can
for the initial steam demands, as in Figures 1 and 2 (3). Next, the change, depending on the size of the change in steam load. The
steam main with the highest steam marginal cost for the current cost of steam in this situation not only depends on the costs of
steam demand is identified. The potential for decreasing the fuel and power, but also on the utility configuration, equipment
demand of this main is then determined by gradually decreasing performance and the constraints within the utility system.
the demand for that steam. At each step in the reduction, the Another important point to emphasize is that Figure 3 does
whole utility system is re-optimized. If a constraint on decreased not represent the feasibility of steam saving within the process-
usage (or increased generation) is reached, or the marginal cost es. The steam savings, as represented in Figure 3, might not be
for the steam main changes significantly, the stepwise decrease in possible as far as the processes are concerned. Thus, the proce-
the steam demand (or increase in steam generation) is terminated. dure should be interpreted as a kind of “what if?” analysis, i.e.,
The procedure is then repeated for the steam main with the high- what would be the economic consequences of eventual steam
est steam marginal cost for the new conditions until no further savings in processes?
decrease in steam usage (or increased generation) is possible. Figure 3 demonstrates that it is, in general, incorrect to attrib-
ute a single economic value for steam at a given level. Its value at
Example 2: Steam prices for changing steam loads a given level depends on how much is being consumed, as well
Consider again the utility system in Figures 1 and 2, but now as fuel cost, power cost, and so on. These marginal prices, togeth-
the steam loads on the various steam headers are varied. The er with the limits on steam saving, provide a strategy for energy
current process steam requirements will be examined for poten- conservation on the site without modification to the utility system.
tial benefits of saving steam. Even for a relatively simple utility The step changes in the marginal price also provide indications
system such as this, there are many complex interactions to be for potential modifications to the utility system. CEP
explored in order to determine the true economic value of steam
saved at any one of the pressure levels.
Figure 3 is a plot of steam marginal price versus the poten-
Literature Cited
tial steam savings. It features six segments: three for HP steam
savings and three MP segments. Saving LP steam has no value. 1. Varbanov, P.S., et al., “Modeling and Optimization of
Utility Systems,” Trans. IChemE, 82A, pp. 561–578 (2004).
Within each segment, there are slight marginal price variations, 2. Varbanov, P.S., et al., “Top-Level Analysis of Site Utility
usually forming a stable trend, resulting from the nonlinearity of Systems,” Trans. IChemE, 82A, pp. 784–795 (2004).
3. Smith R., “Chemical Process Design and Integration,”
the steam turbine performance. The steps in the curve are Wiley, Hoboken, NJ (2005).

ROBIN SMITH is director of the Centre for Process Integration in the School of
8.000 HP Steam MP Steam Chemical Engineering and Analytical Science at the Univ. of Manchester (P.O.
Box 88, Manchester, M60 1QD, U.K.; Phone: 44-161-200-4382; Fax:
7.000 44-161-236-7439; E-mail: robin.smith@umist.ac.uk). He has extensive
industrial experience with Rohm & Haas and ICI, and as a consultant on
6.000 process integration projects. He has published widely in the field of process
Marginal Price, $/t

integration, and is the author of “Chemical Process Design and Integration”

5.000 (Wiley). He holds bachelor’s, master’s and doctoral degrees in chemical
engineering from the Univ. of Bradford in the U.K. He is a Fellow of the Royal
4.000 Academy of Engineering, a Fellow of the Institution of Chemical Engineers in
the U.K., a member of AIChE, and a Chartered Engineer. In 1992, he was
3.000 awarded the IChemE’s Hanson Medal for his work on waste minimization.

2.000 PETAR VARBANOV is a researcher in the Dept. of Process Dynamics and

Operation, Technical University of Berlin (Strasse des 17, Juni 135, KWT9,
1.000 D-10623 Berlin, Germany; E-mail: varbanov@dynamik.fb10.TU-Berlin.de). He
has worked in the field of energy efficiency, specializing in heat integration,
0.000 at IChE — Bulgarian Academy of Sciences, and his professional interests
0 10 20 30 40 50 60 70 include process modeling and optimization. He graduated from the Univ. of
Steam Savings, t/h Chemical Technology and Metallurgy in Sofia, Bulgaria, with an MSc in
chemical engineering, and studied for a PhD, in the field of optimization and
■ Figure 3. Marginal steam price versus potential steam savings synthesis of process utility systems, at the Univ. of Manchester.
when process steam loads change.

CEP www.cepmagazine.org July 2005 33