Organic Rankine cycle working fluid considerations for waste heat to power applications.
Lawrence Berkeley National Laboratory (LBNL) predicts there is 100 GW of waste heat to electric power potential in the United States (Baily and Worrell 2005). This figure represents over 10% of the current installed capacity of electric generators in the country. This potential presents an opportunity to produce low cost, virtually zero emissions, local generation to assist in meeting power quality needs and pending clean energy regulations such as Renewable Portfolio Standards.
In many cases industrial process heat is discharged to the atmosphere still containing 60+% of the heat from the combustion process. Unless there is an opportunity for use of this heat at the industrial site, it is almost certainly wasted. However, conversion of the waste heat to electricity provides energy that can be used either at the site, or it can be economically transported over long distances to another customer. As long ago as 1979, studies investigated technology capable of transforming waste heat into electricity (General Electric 1979). With increases in fuel and electricity prices, the economic case for harvesting this wasted energy has become even more compelling.
Industrial heat sources vary widely with respect to size, exhaust temperature, primary fuel source, duty cycle, and contaminant content. Depending on the relative influence of these factors, different technologies may be more appropriate than others for converting the heat to power. Rankine cycle power generation is a well known technology and, with steam as a working fluid, is the basis for the vast majority of power generation worldwide. It remains the premier choice for power generation for waste heat to electric power conversion in many high temperature applications. Because of the thermal stability of steam it can be used in cases where the source temperatures are very high without fear of thermal decomposition. However, because of turbine size, vacuum conditions in the condenser, and the need to avoid condensation in the turbine, steam is most appropriate for the largest sources of high temperature waste heat. Also, because of design considerations with small molecular weight working fluids, steam turbines generally have lower efficiency than organic working fluid turbines for sizes below several megawatts (Table 1 after Abbin and Leuenberger 1974). For waste heat sources below 1000[degrees]F (538[degrees]C) or smaller than 1 MW output capacity, organic working fluids are more likely to provide better overall economics. Exceptions include industrial facilities that require process steam or existing power plants where steam is already being used as a working fluid. Even in cases where very high temperature sources are present, environmental conditions such as access to water, permitting requirements, and maintenance costs may influence the ultimate decision regarding whether or not to consider steam as a working fluid.
Table 1. Comparison of Turbine Isentropic Efficiency vs. Size and Working Fluid molecular Weight Turbine Isentropic Efficiency % Power Level Steam High Molecular Weight > 10 MW 70-80 75-80 1-5 MW 50-70 75-80 200-500 MW 30-50 78-80 10-100 MW 25-50 60-75
The objective of the study described in this paper was to evaluate the impact of working fluid decomposition on the opportunity for waste heat to power using industrial waste heat and organic Rankine cycle (ORC) based generation within the United States.
To estimate the amount of heat that may be available for conversion to electric power, the U.S. Environmental Protection Agency National Emissions Inventory Criteria Air Pollutants database (EPA CAP) was analyzed to determine both the number and size of waste heat sources as a function of temperature. The EPA CAP database was chosen for this analysis because it contains exhaust gas temperature, flow rate, industry classifications, and process classifications. This makes it a useful tool for estimating the power that could be generated by recovering the waste heat. The 2002 inventory was used in the study as it was the most recent complete inventory available at the time the analysis was performed. (U.S. EPA 2007)
In addition to being several years old, this source of information has other limitations. First, the database is limited to sources that are reported to the U.S. EPA by individual states. The states in turn gather their data from various local government agencies. As such, the data are subject to variations in quality as well as omissions and redundancies. Another limitation is that duty cycle is not disclosed in the database. Even with these limitations, the authors believe this to be the best available information on industrial waste heat sources.
Using the EPA CAP database as an input, recoverable power (the amount of electricity that can be generated from waste heat) was calculated for each individual source using available waste heat and modifications to the Carnot efficiency under assumptions described below. These estimates were then combined to produce an overall estimate of recoverable power as a function of exhaust temperature.
To reduce material and size related costs of an exhaust gas to working fluid (or thermal oil) heat exchanger, exhaust exit gas temperature was limited to 225[degrees]F (107[degrees]C) for purposes of the analysis. This constraint both reduces the required size of the heat exchanger and prevents condensation of water on the heat exchanger. Available heat for each source was then calculated based on the temperature difference between the EPA reported exhaust gas temperature and a heat capacity assumed to be 0.018 Btu/[ft.sup.3] * [degrees]F (1207 J/[m.sup.3] * [degrees]C).
WORKING FLUID CONSIDERATIONS
Two important factors determining the economic viability for ORC based waste heat to power applications are the installation and maintenance cost of the equipment and the amount of heat that can be successfully recovered in the form of electrical power. The choice of working fluid has significant implications for both of these factors. The Rankine cycle uses the phase change of a fluid from liquid to gas to replicate the Carnot cycle as closely as possible. The ultimate efficiency achieved is therefore limited by the high and low temperatures of the working fluid as well as by non-isentropic compression, heating, and expansion and by parasitic losses from pumps, fans, and other auxiliary equipment. The T-s diagram of several working fluids (Figure 1) illustrates that none of them are ideal. Also, if the working fluid cannot be used at the maximum temperatures available from the heat source due to decomposition of the fluid, efficiency will be reduced. In the case of fluids such as water and ammonia with negative slope of the saturated vapor line, efficiency will be reduced by the requirement for superheat to avoid condensation in the turbine. In cases where the working fluid is operated at temperatures above its critical point, efficiency will be reduced, compared to a fluid which could operate at that temperature without supercritical operation as it more closely replicates the trilateral cycle than the Carnot cycle and reduces overall cycle efficiency (Crook 1994). Choice of working fluid therefore will determine how closely the cycle matches the Carnot ideal as well as the maximum allowable temperature of the cycle. One of the advantages of organic working fluids is that many of these compounds have saturated vapor lines with near vertical or slightly positive slope. This allows expansion through a turbine without the need for superheat. This property is tied to their larger number of atoms/molecule and the larger number of possible configurations that this allows. Note that water, ammonia, and propane, (3, 4, and 11 atoms per molecule respectively) have negative saturation slopes, while all of the other organic molecules (atoms per molecule > 13) shown in Figure 1 have positive slopes. It is also generally true that the enthalpy loss will be lower for a higher molecular weight fluid than a lower molecular weight fluid at a given boiler temperature (Marciniak et al. 1981). This allows for smaller and fewer turbine stages with a higher molecular weight fluid. Adding turbine stages for lower power applications tends to decrease turbine efficiency (Abbin and Leuenberger 1974).
[FIGURE 1 OMITTED]
Higher molecular weight organic molecules generally have higher critical points than lower molecular weight ones. High critical point allows higher temperature sources to be used without superheat and therefore at greater efficiency. However, continuing to increase molecular weight causes the slope of the saturated vapor line to become more positive (Figure 2). A positive slope implies that the fluid will be expelled from the turbine as a superheated vapor, increasing the exit temperature of the turbine and reducing cycle efficiency. Some of this excess heat can be recaptured using a regenerative heat exchanger. This increases efficiency but at the expense of additional capital equipment and a reduction in the maximum achievable efficiency.
[FIGURE 2 OMITTED]
The effects of increasing molecular weight on the slope of the saturated vapor line can also be partially mitigated by changing the molecular structure to limit the number of available molecular orientations. Ring structures do not allow the twisting that occurs in straight chain or branched alkanes, they therefore have fewer degrees of freedom and more vertical saturated vapor lines. As an example of this behavior, both heptane and toluene shown in Figure 2 have seven carbons, but the ring structure of toluene shifts the critical point up while simultaneously making the saturated vapor line closer to vertical. Effectively its saturated vapor line behavior is more like that of a lower molecular weight fluid but it retains the high critical point of a high molecular weight fluid. This makes toluene a potentially more efficient working fluid than pentane when higher temperature sources are available.
Based solely on their T-s behavior it would appear that molecules such as R-245fa and pentane for lower temperature sources and toluene for higher temperature sources would be nearly ideal working fluids, and in fact these are common fluids used in ORC equipment. All have nearly vertical saturated vapor lines and the critical point can be as high as 600[degrees]F (316[degrees]C). However, thermal and chemical stability is another issue that needs to be considered, and it is generally in conflict with choices that would be made solely based on the thermodynamic properties of the fluids themselves. This is an area with significant uncertainty for ORC technology. All organic materials undergo degradation and cracking processes as temperature increases. Degradation is accelerated by the presence of air in the system. For this reason, the maximum working fluid temperatures for organic Rankine cycle systems must be limited to avoid temperatures which would cause decomposition. In addition, care must be taken to ensure that the system is free from atmospheric contamination.
In theory, smaller molecular weight organic molecules should be more resistant to thermal cracking. However, very little decomposition data is available in the academic literature in the temperature range of interest for waste heat recovery. Research measurements were conducted in the early 1900's at substantially higher temperatures with the intent of determining the mechanism of decomposition and measuring decomposition rates (Marek and McCluer 1931; Paul and Marek 1934; Morgan et al. 1935; Frey and Hepp 1933; Pease 1928). More recent measurements were taken in the temperature range of interest but were not performed over a range of temperatures (Andersen and Bruno 2005). The results of these studies have been re-plotted in Figure 3. Reaction rate constants as a function of temperature were determined using the Arrhenius equation:
[FIGURE 3 OMITTED]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)
k = reaction rate constant in [s.sup.-1]
A = pre-exponential constant
[E.sub.a] = activation energy
R = gas constant
T = temperature in absolute units
The rate of decomposition may then be determined by:
r = [-d(C)/dt] = k(C) (2)
r = decomposition rate
C = concentration of the working fluid
t = time
The amount of working fluid remaining is then given by:
C = [C.sub.0][e.sup.-kt] (3)
where [C.sub.0] is the initial concentration of the working fluid.
The early temperature dependent studies cited here have concluded that the decomposition mechanism is a simple first-order decomposition reaction for all molecules studied. While there may be other, faster, processes that dominate decomposition at lower temperatures first order decomposition should provide a lower bound of the decomposition rate. Extrapolating the results into a lower temperature regime allows an estimate of decomposition rate at the temperatures of interest for ORC's.
There are several trends of interest from Figure 3. First, smaller molecules decompose more slowly than larger ones. This is expected assuming entropy is the driving force for first order thermal decomposition. Branched molecules decompose more rapidly than straight chains, and ring structures, particularly benzene, are more stable than linear chains. This is also expected because of the added stability introduced by carbon's resonant structures. Finally, when similar molecules were tested at lower temperatures, their stability was significantly worse than predicted by high temperature data extrapolation. This may indicate that some mechanism other than thermally activated first order decomposition becomes dominant at lower temperatures. One possibility is that the surface of the reaction vessel catalyzes the decomposition at low temperatures.
In Figure 3, k is the reaction rate constant, the dashed vertical line near 1000[degrees]F is the low temperature bound of the early temperature dependent studies, and the horizontal dashed lines represent the threshold for the rate constant where 1% of the fluid decays in the time period indicated. These can be used as a guide to estimate the maximum allowable working fluid temperature for a given fluid decomposition tolerance. However, the lack of data in the temperature range of interest for technologically important working fluids is a concern that needs to be addressed by the research community.
An ORC working fluid is circulated at temperatures between ambient and the maximum temperature of the system. So if, for example, one percent of the fluid decomposing is tolerable and the maximum temperature of the system is such that one percent is expected to decompose per year at that temperature, satisfactory operation for ten or more years may be possible because the fluid is not exposed to the maximum temperature for a large fraction of the cycle. To determine more accurate measures, these temperature dependent rate constants could be combined with modeling that takes into account the temperature profile of the fluid with time. This also highlights the importance of avoiding hot spots in a heat exchanger that is near the maximum allowable temperature.
In addition to the data presented in Figure 3, there is some variation in reports regarding acceptable working temperature for some fluids. For example, Andersen and Bruno (2005) concluded that toluene has an unacceptable decomposition rate at 600[degrees]F, but Marciniak et al. (1981) reported that it can be used to 750[degrees]F (399[degrees]C), though they do so without reference or supporting data. Cole found toluene to be a stable working fluid to 677[degrees]F (358[degrees]C) and was expected to be stable at least to 750[degrees]F (399[degrees]C) provided that oxygen was excluded from the system (Cole et al. 1987). Researchers in that study suggested that years of operation should be possible between fluid changes. Baton (2000) reported that in one facility operating with toluene, working fluid decomposition products of toluene were found after several thousand hours of operation at 700[degrees]F (371[degrees]C) hot side temperature. But another facility operating at 750[degrees]F (399[degrees]C) hot side temperature had not shown any signs of decomposition. Differences between laboratory measurements of decomposition and field observations might be explained by differences in measurement methodology. Specifically, the field workers are looking for visibly obvious signs of decomposition such as black chunks or residue, while laboratory workers are using instruments to measure concentrations of decomposition products quantitatively. If this is the case, it may indicate that some of the decomposition products are largely benign.
Figure 4 highlights the conflicting needs of high critical point and high resistance to thermal decomposition for the n-alkanes. For example, if 1%/year degradation is the maximum allowable for a heat source at 500[degrees]F (260[degrees]C) a fluid would have to be chosen which is working above its critical point. The data displayed for decomposition is meant to illustrate the tradeoff that exists between stability and critical point, the data used is based on the early high temperature work which may significantly overestimate stability. For example, Figure 3 shows more recent low temperature data for n-pentane which shows decomposition of pentane faster than 1%/30 days at 600[degrees]F (316[degrees]C) while extrapolating higher temperature data into this range suggests it would be closer to 1%/year.
[FIGURE 4 OMITTED]
Fluorinated refrigerants such as R-245fa (1,1,1,3,3-pentafluoropropane) are also actively used as ORC fluids. Angelino and Invernizzi (2003) reported that this compound is stable for at least 50 hours at 572[degrees]F (300[degrees]C) but at 626[degrees]F (330[degrees]C) decomposition is rapid (Angelino and Invernizzi 2003). Their results are in contrast with a representative of the manufacturer of R-245fa, who suggests that working fluid temperatures much above 300[degrees]F (149[degrees]C) should be avoided due to observations of fluorine formation attributed to decomposition (Zyhowski 2008). Further, manufacturers of ORC equipment using R-245fa generally limit the maximum working fluid temperature to 300[degrees]F. While the C-F bond strength is high, the contributions of entropy driving the decomposition is also larger for the refrigerants than it is for alkanes of the same chain length due to the larger number of atom types. This is what drives the auto ignition temperature of the HCFC's to be lower than alkanes of the same chain length (The Engineering Toolbox 2008; BOC Gases 2008). Therefore, a lower decomposition temperature should be expected for these compounds than their non-halogenated relatives. An explanation for the difference in opinion regarding operating temperature may simply be the length of time over which the experiments were conducted and the sensitivity limits available to measure decomposition. Because the decomposition process is a thermally activated one, moving from 572 to 626[degrees]F (300 to 330[degrees]C) could increase the decomposition rate from 1%/day to 1%/ hour (which was the stated maximum sensitivity for Angelino and Invernizzi 2003). A change this dramatic is consistent with the activation energies for alkane thermal decomposition. The greater bond strength of the C-F bond compared to the C-H or C-C bond would likely make the activation energy larger and the change in decomposition rate even more abrupt.
Thermal decomposition of ORC working fluids is normally avoided by using an intermediate heat transfer fluid to separate the working fluid from the high temperature exhaust streams. This has the disadvantage of reducing the attainable conversion efficiency by reducing the maximum temperature of the cycle. It also increases capital cost by requiring an additional heat exchanger. Even when the maximum temperature is limited by use of a secondary heat exchanger and thermal oil, decomposition of pentane has been noted in some facilities where low winter operating temperatures cause vacuum conditions in part of the system (Sweetser and Leslie 2007). Vacuum conditions can result in air infiltration into the system, particularly if it was designed for positive pressure operation. Entrained air will significantly lower decomposition temperature.
In addition to the concern of loss of the working fluid itself to decomposition, there is the concern of potential safety hazards and equipment damage depending on the nature of decomposition products. In the case of alkanes as working fluids, products of decomposition will likely be smaller alkane chains, hydrogen gas, and eventually carbon (Marek and McCluer 1931; Paul and Marek 1934; Morgan et al. 1935; Frey and Hepp 1933; Pease 1928). These products are not particularly concerning unless the working fluid is diluted to the point that system performance is affected or carbon is deposited as a solid. Solid carbon can reduce the efficiency of heat exchangers as well as eventually clogging them. Also, the turbine could be damaged by carbon particulates. Fluorinated hydrocarbons, on the other hand, may form HF or [F.sub.2] during decomposition in addition to possibly forming carbon and hydrogen. There is a significantly greater safety concern in this case. This is an interesting point considering that safety, by virtue of not having a flash point, is frequently a reason given for using these refrigerants as working fluids as opposed to other hydrocarbons. Any air or water leaks into the system provide for the possibility of still other reaction productions such as carbon dioxide, hydrofluoric acid, or carbon monoxide.
The current state of understanding of working fluid decomposition is a barrier to adoption of ORC based waste heat to power applications. Advances in the understanding of both the rates of decomposition and the reaction products of the decomposition could facilitate more widespread use. Even if decomposition of working fluids cannot be eliminated, understanding it may make it possible to produce decomposition products that are largely benign and to predict when maintenance is necessary. The effect of decomposition of organic working fluids needs to be taken into account in the calculation of recoverable power as it will affect the efficiency of the conversion of the waste heat into electric power.
Carnot efficiency was calculated using a cold source temperature of 120[degrees]F (49[degrees]C) and a heat source temperature either equal to the exhaust gas temperature or limited at some lower temperature to avoid working fluid decomposition. It is estimated that efficiency of approximately half of the Carnot ideal is achievable in most waste heat to power applications due to loss in heat exchange, non-isentropic heating, pumping, and expansion and in the generation equipment and gearbox. This estimate appears to be reasonable based on a recent technology assessment performed by the authors that evaluated the current state of ORC technology. This assessment found that under the assumptions described above the overall waste heat to electric power efficiency was roughly half of the calculated Carnot efficiency. The results of the calculated efficiency compared to measured and quoted values from various suppliers are shown in Figure 5 (Leslie 2009). Clearly the actual efficiency will be influenced by a wide variety of factors in addition to working fluid minimum and maximum temperature including the thermodynamic properties of the fluid, overall system design and optimization, and ambient temperature conditions. However, for the purpose of the present analysis the assumptions given here should be sufficient to estimate the overall opportunity and the impacts of working fluid temperature limitations.
[FIGURE 5 OMITTED]
A histogram of all sources over 300[degrees]F (149[degrees]C) is shown in Figure 6. Half of all potential sources are at 450[degrees]F (232[degrees]C) or below. This result may be somewhat misleading, however, as it considers only the number of sources and not their power production capability. Looking specifically at the temperature range important for ORC technology, between 300 and 1000[degrees]F (149 and 538[degrees]C), the EPA CAP database was analyzed first under the assumptions above assuming that the working fluid is capable of being used near the exhaust temperature without decomposition problems.
[FIGURE 6 OMITTED]
Under these assumptions total recoverable power potential is calculated to be 63 GW with 44 GW coming from sources under 1000[degrees]F (538[degrees]C). Results of this analysis are presented in Figure 7 as a function of waste heat source temperature. The total recoverable power available in each temperature regime remains roughly constant over the range of 300 to 1000[degrees]F (149 to 538[degrees]C) as a result of a much larger number of low temperature sources but poorer efficiency in recovering it. In the range of interest for ORC technology, 80% of recoverable power generation is in the temperature range of 500 to 1000[degrees]F (260 to 538[degrees]C) and 50% is in the range of 750 to 1000[degrees]F (399 to 538[degrees]C). Also, the average recoverable power per source does not get over 1 MW until exhaust gas temperatures rise above 600[degrees]F (316[degrees]C). Therefore, while most of the sources of industrial waste heat are at relatively low temperatures, most of the opportunity for waste heat to power applications is actually at intermediate temperature, 500 to 1000[degrees]F (260 to 538[degrees]C). These temperatures are high enough that working fluid decomposition must be considered when designing a system.
[FIGURE 7 OMITTED]
This analysis was repeated limiting maximum working temperature to 450[degrees]F based on the earlier discussion of working fluid decomposition. The resulting decrease in efficiency caused by this limitation reduces the total recoverable electric power from sources under 1000[degrees]F to 32 GW from 44 GW and increases capital costs, in the form of an additional heat exchanger for all sources over 450[degrees]F. The results of this analysis as a function of temperature are shown in Figure 8.
[FIGURE 8 OMITTED]
Including sources over 1000[degrees]F (538[degrees]C) assuming steam as a working fluid, with maximum temperatures equal to the source temperature, the total recoverable power opportunity is then calculated to be 51 GW. Even considering the high temperature limitations of the organic fluids, the overall opportunity for waste heat to power is still greater for ORC, 32 GW, than for steam, 19 GW. Figure 9 shows the same analysis as Figure 8 but includes sources 1000[degrees]F and hotter.
[FIGURE 9 OMITTED]
In the case of source temperatures near 1000[degrees]F (538[degrees]C), limiting working fluid temperature to 450[degrees]F brings the conversion efficiency to only 30% of the Carnot ideal. In this temperature range it may become advantageous to use a transcritical [CO.sub.2] cycle instead of an ORC cycle, eliminating the issue of thermal decomposition completely. Another possibility for improvement is use of smaller chain alkanes such as propane or butane. Even though their critical temperatures are relatively low, the ability to use them at higher maximum temperature may provide better overall efficiency and may also provide better economics if the use of a thermal transfer fluid and additional heat exchanger can be avoided.
It is important to note that the estimates of recoverable power presented here almost certainly substantially overestimate the opportunity, as sources with hostile exhaust compositions have not been eliminated from the analysis and duty cycle is not contained in the EPA source database. However, this study illustrates the impact that working fluid decomposition has on the total opportunity for waste heat to power applications and the relative opportunity for ORC based recovery compared to steam Rankine cycle recovery.
The overall waste heat to power opportunity from industrial sources in the U.S. based on analysis of the EPA National Emissions Inventory CAP database is estimated to be 51 GW. This result does not consider limitations that would certainly exist of exhaust gas suitability for heat exchange, difficulty in connecting some remote processes to the power grid, or low duty cycle of some sources included in the CAP database that would tend to decrease the opportunity. It also does not include opportunities that may exist from sources not included in the CAP database either accidentally or intentionally, such as non-stationary sources, that would tend to increase the opportunity.
Consideration of working fluid decomposition in the opportunity analysis reduces the potential for ORC energy recovery by 12 GW, 27%. However, 63% of the estimated recovery opportunity is still from sources below 1000[degrees]F (538[degrees]C) for which ORC is the most appropriate recovery technology.
The authors gratefully acknowledge the National Rural Electric Cooperative Association Cooperative Research Network for supporting this work.
Abbin, J.P. and W.R. Leuenberger. 1974. Program CYCLE: A rankine cycle analysis routine. Report SAND 74-0099 (revised), Sandia National Laboratories, USA.
Andersen, W.C. and T.J. Bruno. 2005. Rapid screening of fluids for chemical stability in organic rankine cycle applications. Industrial and Engineering Chemistry 44:5560-5566.
Angelino, G. and C. Invernizzi. 2003. Experimental investigation on the thermal stability of some new zero ODP refrigerants. International Journal of Refrigeration 26:51-58.
Bailey, O. and E. Worrell. 2005, Clean energy technologies: A preliminary inventory of the potential for electricity generation. Report LBNL-57451 Ernest Orlando Lawrence Berkeley National Laboratory, USA.
Baton, B. 2000. Organic rankine cycle engines for solar power. Solar 2000, June 18, Madison WI.
BOC Gasses, Safety Data Sheet: 1,1,1,3,3-Pentafluoropropane, HFC-245fa. http://www1.boc.com/uk/sds/special/R245fa.pdf
Cole, R.L., J.C. Demirgian, and J.W. Allen. 1987. Predicting toluene degradation in organic rankine-cycle engines. Proceedings of the Intersociety Energy Conversion Engineering Conference, Philadelphia, PA, pp. 1402-1407.
Crook, A.W. 1994. Profiting form Low-Grade Heat: Thermodynamic Cycles for Low-Temperature Heat Sources, Institute of Electronics Engineers.
Frey, F.E. and H.J. Hepp. 1933. Thermal decomposition of simple paraffins. Industrial and Engineering Chemistry, 25:441-449.
General Electric. 1979. Pipeline bottoming-cycle study: operational reliability and maintainability assessment. Report GESP-813. General Electric Company, Cincinnati, OH.
Leslie, N.P. 2009. Waste heat to power technology and market assessment. NET 2009, January 22, Tucson AZ.
Marciniak, T.J., J.L. Krazinski, J.C. Bratis, H.M. Bushby, and E.H. Buyco. 1981. Comparison of rankine-cycle power systems: effects of seven working fluids, Report ANL/ CNSV-TM--87, Argonne National Laboratory, USA.
Marek, L.F. and W.B. McCluer. 1931. Velocity constants for the thermal dissociation of ethane and propane. Industrial and Engineering Chemistry 23:878-891.
Morgan, J.J. and J.C. Mucay. 1935. Thermal decomposition of n-pentane. Industrial and Engineering Chemistry, 27: 1082-86.
Paul, R.E. and L.F. Marek. 1934. Primary thermal dissociation. Industrial and Engineering Chemistry, 26:454-57.
Pease, R.N. 1928. The thermal dissociation of ethane, propane, normal butane, and isobutane. Preliminary study. Journal of the American Chemical Society, 50:1779-85.
Sweetser, R. and N. Leslie. 2007. Final report for the basin electric project at northern border pipeline company's compressor station #7, ND. Report ORNL/TM-2007/158, Oak Ridge National Laboratory, USA.
The Engineering Toolbox. 2008. Fuels and chemicals--Auto ignition temperatures. http://www.engineeringtoolbox. com/fuels-ignition-temperatures-d_171.html.
U.S. EPA. 2007. 2002 national emissions inventory data, http://www.epa.gov/ttn/chief/net/2002inventory.html.
Zyhowski, G. 2008. Zyhowski with Honeywell, telephone conversation with the author, April 2008.
David Schroeder is assistant professor in the Department of Engineering Technology, Northern Illinois University, Dekalb, IL. Neil Leslie is R&D manager in End Use Solutions at the Gas Technology Institute, Des Plaines, IL.
David J. Schroeder, PhD
Neil Leslie, PE