Waste heat recovery boiler

Understand the basics of packaged steam generators A custom design has many advantages over older, standard ones V Ganapathy, ABCO Industries, Abilene,

C

ustom-designed steam generators should be considered

when evaluating packaged boiler designs because standard designs have several limitations. Merits of convective superheaters over radiant designs should be understood by end-users. Evaluate operating and life -cycle costs for each boiler application. The unit with lowest life -cycle cost and good des ign should be selected, not the boiler with lowest initial cost alone. Write specifications clearly and avoid comparisons based on surface areas. Specifying or selecting boilers based on pre-engineered designs/tables is not recommended. Refractory-lined designs have poor design and performance features and serious maintenance concerns. Operating at low loads is difficult for fans and superheaters and should be reviewed carefully. Oil- and gas -fired packaged steam generators are widely used in chemical plants, refineries and cogeneration systems. They form an important part of the total steam system in any plant and are available in capacities up to 250,000 lb/hr at pressures ranging from 150 to 1,500 psig and temp eratures from saturated steam to 1,000°F. They are expected to last about 25 years, and therefore, cost-effectively generate steam. However, when making purchasing decisions, plant engineers, consultants and engineering firms spend little time on important aspects such as long-term performance and operating costs and often recommend standard, off-the shelf designs that have several limitations. Here are important design and performance aspects of packaged steam generators and recent trends in their design that engineers should be familiar with.

HYDROCARBON PROCESSING /JULY 1997

85

Standard designs have limitations. Standard boiler

designs were developed decades ago by boiler vendors to simplify the process of manufacturing and purchasing packaged boilers. Tables were developed showing major dimensions, surface areas, tube count, etc., and the consultant's job was only to select a model number for a given steam capacity from pre -engineered designs; plant layout was not difficult because the major dimensions and weight details were known. Engineering and manufacturing hours were greatly reduced for the boiler supplier, resulting in lower initial costs. However, when standard designs were developed, there were no emission regulations. Low excess air, about 5% to 10% air, was used to attain high boiler efficiencies for natural gas and fuel oils. Using flue-gas recirculation (FGR) to reduce NOx emissions was relatively unknown. Today, 15% to 20% excess air and 15% to 25% FGR rates are common for natural gas-fired boilers to attain NOx less than 30 ppm and CO less than 100 ppm. Also, the flue gas quantity flowing through the boiler is directly proportional to the amount of fuel fired: Fuel fired = boiler duty/efficiency(1) Boiler duty, or energy absorbed by steam, depends on whether there is a superheater or not. Efficiency depends on the exit gas temperature, which in turn depends on the presence of economizer. Thus, even if the boiler capacity is a nominal 100,000 lb/h as shown in Table 1, due to the effect of duty and excess air and FGR rates, a significant difference in flue gas mass flow results among the various cases. Cases 3 and 6 have no economizer, which affects the fuel input. In case 5, which has a superheater, the flue gas quantity is nearly 40%

more than that of case 1 and, there fore, gas pressure drop across the convection section is doubled. Calcula tions are based on the assumption that no efforts were made by the boiler supplier to offer a larger unit or change the tube lengths, pitches, tube rows, etc., to lower the gas pressure drop as in custom-designed units. The increase in gas pressure drop causes additional fan power consumption, which is an operating cost. Fan power consumption (kW) W x Ahw x 62.4 x 0.746 (2) (3,600 x 12 x 550 x 0.7 x 0.075) = 0.0000373 W x Ahw where W = flue gas flow, lb/h Ahw = static head or pressure drop in boiler system, in.WC

62.4 = density of water 0.746 = conversion from hp to kW 550 = conversion from ft-lb/s to hp .075 = density of air, lb/ft 3 70% efficiency was used for the fan and an additional 8 in.WC loss was assumed for the burner and duct work. Case 1 power consumption =100,385 x 18 x 0.0000373 = 68 kW and case 5 power consumption = 143,150 x 28 x 0.000373 = 150 kW. The difference is about 82 kW In reality, the difference is more if actual burner drop and duct losses are considered. Over a year, additional operating costs = 82 x 8,000 x 0.05 = $32,800 (at $0.05/kWh), which is not a small amount. The fan size is also larger. Hence, if actual operating cost is evaluated, the standard design is not costeffective, although its initial price may be attractive. This could be a costly error for the end-user, who will operate the boiler for 20 to 30 years. In Table 1, the effect of additional mass flow on boiler exit gas temperature is not considered, which makes the situation even worse. Exit gas temperature increases by 10°F to 40°F if gas flow increases in a given boiler. Note that a 40°F rise in exit gas temperature is equivalent to 1% change in boiler efficiency, or about $30,000/yr, based on a fuel cost of $3/MMBtu. A better selection would have been a model generating 20% to 40% more steam or a 130,000lb/hr to 140,000-lb/h boiler vs. the nominal 100,000-lb/h unit. This does not happen often because vendors want to be competitive and push the lowest cost option. Custom designing is the only way to arrive at optimum designs. 2

HYDROCARBON PROCESSING/ JULY 1997

Custom-designed boilers. Custom designing starts with an understanding of the boiler parameters, desired emission levels and fuels fired. The starting point is a furnace design based on discussions with a burner supplier, who reviews information on furnace dimensions, excess air and FGR rates and gives approval. The furnace, convection section, superheater and economizer are then designed. Based on actual gas flow generated, efforts are made to minimize the gas pressure drop by adjusting the boiler height, tube spacings, tube counts and even possibly using finned tubes in the convection section if the gas is clean. Thus, every boiler is designed new and not pulled up from a pre-engineered table. The result is a unit with high efficiency and low operating cost, meeting the desired emis sion levels without flame impingement concerns. Furnace performance. This is the most important component of any steam generator. Its performance affects not only the combustion process but also the heat-transfer surfaces located beyond the furnace such as superheaters, convection section and economizer. Using techniques such as low-NOx burners, staged combustion and FGR, the flame shape in the furnace will be different. Result: a possibility that the furnace dimensions of standard designs are inadequate and flame impingement may occur on the furnace walls. With custom-designed boilers, furnace dimensions are reviewed with the burner supplier along with a fuel analysis and required emission levels. The furnace dimensions are based on the burner supplier's recommendations and not vice-versa (checking if a given furnace is adequate for the project in question). Another poor practice carried over from decades ago is

using refractory in the furnace floor, front and side walls. Engineers and boiler designers who had little experience with boiling heat transfer and circulation in those days found this practice convenient to prevent overheating of floor tubes. However, with the abundance of information on boiling heat transfer and boiler circulation, backed up by the operation of several hundred units, this practice is not necessary. Using refractory also reduces the furnace effective radiant surface area and increases the area heat release rate and heat flux. An important recent development is a completely watercooled furnace (Fig. 1) with several advantages: 1. The furnace front, rear, side walls and floor are completely water-cooled and are of membrane wall construction, resulting in a leak-proof enclosure for the flame. The entire furnace expands and contracts uniformly, thus avoiding casing expansion problems. When refractory is used on front and rear walls, hot gases leaking from the furnace are always possible. Casing corrosion is also likely since a gas-tight joint is difficult to ensure between refractory-lined casings and water-cooled walls; when corrosive gases condense there is corrosion. 2. Problems associated with refractory maintenance are eliminated. There is no need for a shutdown to check the refractory or replace it. 3. Fast startup rates are difficult with refractory-lined boilers. With a completely water-cooled furnace, quick startups are possible. This is important in cogeneration projects where the packaged boiler must supply steam to the end-user as soon as the heat recovery system fails. 4. Heat-release rates, on an area basis, are lower with a water-cooled furnace (for the same furnace volume) compared to a refractory-lined unit which has less cooling surface. This also results in reduced heat flux. Note that area heat-release rate is a more significant parameter than volumetric heat-release rate, which gives an indication of residence time for combustion products and is pertinent only for difficult-to-burn fuels like solid fuels. Area heatrelease rate affects furnace heat flux and departure-fromnucleate boiling (DNB) conditions and is significant. Typical area heat-release rates vary from 100,000 to 175,000 Btu/ft2-hr for packaged boilers. It makes more sense to specify area heat-release rates rather than volumetric rates.

5. Reradiation from the refractory increases the flame's local combustion temperature, which in turn increases NOx formation. Water-cooled front walls, especially the front wall where NO, formation potential is the highest, have a beneficial effect on the flame -they cool it effectively so NOx formation is reduced. 6. Another problem with us ing refractory in the furnace is an increase in furnace exit gas temperature, which raises the radiant heat flux and causes tube failures in the radiant superheater (if present). Radiant vs. convective superheaters. Radiant superheaters are widely used in packaged steam generators (Fig 2). They are prone to frequent tube failures because of their location. Convective superheaters, located behind several screen tubes, have fewer maintenance concerns and a much longer life due to their lower tube wall temperatures; but their size and cost are higher due to a lower log-mean temperature difference. The following points on radiant vs. convective designs should be understood by potential endusers, who can influence design specifications and evaluation. 1. Radiant superheaters are located at the furnace exit or turning section as shown in Fig 2. The furnace exit gas temperature is a difficult parameter to estimate. Variations in excess air, FGR rates and flame shape also add to the difficulty. The furnace exit gas temperature could be off by 100°F to 200°F from predicted values. The turning section is also subject to turbulence and nonuniformity in gas temperature profiles, which also hinders an accurate superheater performance evaluation. Thus, radiant superheater tube wall temperatures could be underestimated significantly, leading to tube overheating and failures. 2. Several boilers operate at partial loads of less than 50% for significant time periods. The radiant superheater, by its nature, absorbs more enthalpy at partial loads compared to convective designs. Also, at partial loads, steam flow distribution inside the superheater tubes is less uniform and often questionable. If at 100% load, the superheater pressure drop is 30 psi, then at 25% load the pressure drop is barely 2 psi-this may not ensure good steam flow distribution through all the tubes. Gas -side mixing will also be poor due to low gas velocities. So there is a double negative of higher radiant energy and poor steam and gas flow distribution, which is likely to cause overheating of a few tubes in the radiant superheater. The convective superheater, conversely, is located HYDROCARBON PROCESSING/ JULY 1997

3

Table 2. Comparison of surface areas Boiler 1 90,500

68,700

2

148,900

116,500

Heat release rate, Btu/ft -hr Furnace length, ft Furnace width, ft Furnace height, ft Furnace exit gas, °F Boiler exit gas, °F Economizer exit, °F 2 Furnace proj area, f t (duty) 2

Boiler surface, f t (duty) 2 Economizer surface, f t (duty) Geometry Tubes/Rows No. deep

22 6

Overall heat transfer coefficient

Developing boiler specifications. Consultants and AE firms responsible for purchasing steam generators should develop good and clear specifications. Highlight the following aspects. Steam parameters such as flow, pressure, degree of superheat and feed water temperature should be stated along with feed water quality and steam purity desired. This enables the designer to select proper drum internals. Amount of blowdown to be used while determining the boiler duty can also be estimated if feed water quality is specified. This is important since it affects boiler duty and the amount of fuel fired. If feed water is used for interstage attemperation, the water should be demineralized with preferably zero solids. Otherwise, solids from the spray water will carry over into the steam and deposit in the superheaters or steam turbine. If demineralized water is not available, then the boiler designer can engineer a condensate spray system, which essentially condenses the desired quan tity of steam using feed water and uses it for desuperheating (Fig 3). Often, specifications do not state if steam for deaera tion is from the boiler or another source. This is important if the deaerator is supplied by the boiler vendor. The total steam must be increased by 10% to 15%, depending on the temperature and amount of condensate 83

HYDROCARBON PROCESSING /JULY 1997

29 6

10

10

2,364 683 315 802 (36.6)

2,255 611 315 1,026 (40.4)

3,972 (53.7) 8,384 (10.5) evap/econ

4,760 (52.1) 8,550 (8.3) evap/econ

11/15 66/14

10/15 87/10

Length, ft 9.5/11 Eco fins/in, ht, thick, serr 3x0.75x0.05x0.157 Transverse pitch, in. 4.0/4.0

beyond several screen tubes. Thus, better gas -side mixing is likely. The heat flux and gas temperature entering it are lower, resulting in a less hostile environment at all loads. So, their performance can be predicted more accu rately than radiant designs. The screen section can be larger when the required steam temperature is less, thus ensuring low tube-wall temperatures. Radiant superheaters, however, are always located at the highest gas temperature zone irrespective of whether the degree of superheat is 20°F or 400°F. 3. Multistage superheaters with interstage desuperheating can be used with convective designs to ensure that steam temperatures are not exceeded and the tube wall temperatures are predictable and under control. End-users are better off with convective superheater designs-their size and cost may be more but their life is longer, with fewer maintenance concerns.

Boiler 2

3

Heat release rate, Btu/ft -hr

18.0/7.35

9.5/10 5x0.75x0.05x0.157 4.375/4.0 17.0/6.25

Parameters: 100,000 Ib/h, 300 psig steam, 230°F feed water, 2% blowdown, nat sat gas fuel, 10% excess air: Boilerduty = 100.8 ural efficiency (HHV) = MMBtu/h, 84.3%, furnace back pressure = 7.0 in.

steam for deaeration is supplied from the boiler. Ignoring this could result in a smaller boiler. Some consultants think that if a boiler is designed for 700 psig, then it can operate at any lower pressure, even 100 psig. This is not so. Due to the significant difference in specific steam volume, the velocity of steam at lower pressure will be high, about 300 to 400 ft/s in the superheater or in the steam lines and could be a serious operating problem. The steam drum internals will also not operate well at lower pressures and carryover of water into steam is likely. A possible option is to reduce the steam capacity at lower pressures or design the boiler for the vary ing pressures. But state this point up front in the specifications. If superheaters are present, the steam temperature control range should be specified. Typically, 50% to 100% load range is feasible. However, some consultants not familiar with flow distribution problems at low loads, suggest a load range of 10% to 100%. This is not meaningful since it is very difficult to predict superheater performance when gas/steam flow maldistribution problems are likely. Fuels and emissions. Fuel analysis and emissio ns to be met should be stated clearly. Gaseous fuels containing hydrogen have a higher combustion temperature, which increases NO, formation. Low-Btu fuels result in large amounts of flue gas to be handled by the boiler. This affects fan power consumption and efficiency due to a higher exit gas temperature. The burner supplier must also ensure that emission levels can be met with the fuels in question and suggest appropriate excess air and FGR rates. If both natural gas and fuel oils are used, the specifications should state this. Furnace exit gas temperature is higher for natural gas compared to oil. If convective superheaters are used, the desired steam temperatures have to be attained on oil firing. Desuperheating could be done on gas firing to control it. Presence of nitrogen in fuels also affects NOx formation. The burner supplier must be aware of this if NOx guarantees are made.

One reason for using an economizer and not an air heater as the heat recovery system is the impact on NO, by the higher combustion temperature with hot air. Also the gas- and air-side pressure drops are higher in an air heater, which is an operating penalty. Fan operation. A small margin should be used for flow and head while sizing fans. This is because, unlike utility boilers where multiple fans are used, a single fan is used in packaged boilers. If a large margin is used, the fan is not likely to operate well below 30% to 40% (Fig. 4) unless a variable speed drive is used. At low loads, even with fully closed damper positions, the leakage air flow could be enough to blow out the flame. These aspects must be discussed with the burner and fan supplier and with the end-user to check if low-load operation for long duration is really likely and necessary. Surface areas can be misleading. A common problem even among experienced engineers is the comparison of surface areas of different designs. Some consultants even specify required surface areas. This is a poor practice and should be avoided. Surface area is defined as: S = Q/(U •T) (3) where Q= duty or energy absorbed by the surface, Btu/hr U= overall heat transfer coefficient, Btu/ft2-hr-°F AT= log-mean temperature difference, °F. U depends on gas velocity, temperature, tube pitch and arrangement. Also, when extended surfaces are used, the variations in S could be 100% to 300%.1, 1,2 Using a large fin density decreases U while a lower fin density increases U. Therefore, unless one knows how to compute U for bare and finned tubes, comparing surface areas can be misleading and should be avoided. Rules of thumb for surface areas should also be avoided as they can lead to improper conclusions. Table 2 shows the design of two boilers for the same duty, efficiency and gas pressure drop with different surface areas. The reason for variations in S is that the amount of energy absorbed in the furnace, convection and economizer sections are different. Also, using different fin configurations in the economizer distorts the picture. LITERATURE CITED Ganapathy, V, "Design and evaluate finned tube bundles," Hydrocarbon Processing, September 1996. Ganapathy, V., "Evaluate extended surfaces carefully," Hydrocarbon Processing, October 1990.

The author V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications and packaged water tube steam generators. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler technology from Madras. University. Mr. Ganapathy is the author of over 175 articles on boilers, heat transfer and steam plant systems and has written five books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization, Basic Programs for Steam Plant Engineers (book and diskette), and Waste Heat Boiler Deskbook, copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Design, Vols. 25 and 26, Marcel Dekker, New York.

Understand boiler performance characteristics Use these suggestions when buying, designing or optimizing steam generators V Ganapathy, ABCO Industries, Abilene,

A

n understanding of the major differences in per-

formance characteristics of steam generators is essential to better use and integrate them into plant steam systems. Process and cogeneration plants widely use gas- or oil-fired packaged steam generators (Fig. 1) and gas-turbine-exhaust, heat-recovery steam generators (HRSGs) (Fig. 2) to meet steam demands. The most important differences are efficiency versus load characteristics, gas/steam temperature profiles and partial load behavior. Also, steaming in the economizer is a concern at low steam flows for HRSGs, but not for packaged boilers.

Packaged boilers. Completely shop-assembled pack-

aged boilers (Fig. 1) are used to generate steam up to 200,000 lb/h, 1,000 psig and 850°F. Slight deviations in these parameters are feasible on a case-by-case basis and depend on permitted boiler shipping dimensions. Typically, these steam generators fire natural gas and distillate oils with burners located in the furnace's front wall. A superheater and economizer may be used if superheated steam is required and a higher efficiency level is sought. Two major configurations are available for packaged boilers: the D-type and O-type. In the D-type (Fig. 3a), combustion products of flue gases leaving the furnace make a 180 degree turn and flow over the convection tubes that may contain a superheater. Gases leaving the convection section then transfer energy to an economizer, which preheats the feed water. Generally, air heaters are not used in packaged boilers due to cost considerations, larger gas/air pressure drops and increased NOx formation due to higher flame temperatures. In the 0-type boiler (Fig. 3b), the burner is mounted on the front wall. Combustion products travel to the furnace's end, make a 180 degree turn and flow towards the front via the two convection banks. These transfer energy to the convection tubes. In another option (Fig. 3c), the

gases travel straight without making the turn. In this case, the boiler will be longer because its length includes

H Y D R O C A R B O N P R O C E S S I N G / A U G U S T 1 9 9 4 131

formation. But this method has less impact on conversion of fuel- the convection pass and furnace. The bottom drum in the 0-type may be replaced by two smaller drums resulting in an A-type configuration (Fig. 3d). Generally, packaged boilers operate with a pressurized furnace design. A forced-draft fan sends combustion air through the burner. It then sends resulting flue gases all the way to the stack. Furnace pressures as high as 30 to 40 in H 20 are common. Natural circulation principles are used to circulate a steam-water mixture through the riser tubes in all of these units. How emissions impact design of packaged boilers. Generally, NOX levels of 30 to 80 ppm and CO

levels of 150 to 300 ppmv can be attained by using gas recirculation, staged fuel or air combustion, low-NO x burners, steam injection and excess-air control. Some regions require less than 9 ppmv NOx and over an 85% reduction in CO. This can be attained only with a selective catalytic reduction system (SCR). These are very expensive, on the order of 20% to 35% of boiler cost. Boiler design is impacted by emissions. Gas recirculation increases gas pressure drop through the boiler and also affects gas/steam temperature profiles. Typically, 5% to 15% gas recirculation is used. High excess air, around 15%, may also be required to control NOx and CO. If gas recirculation is used, a separate recirculation fan can transport cool flue gases from the boiler's rear to the flame region. Alternatively, a forced-draft fan may induce flue gases. Addition of cool flue gases at the burner region reduces flame temperature and, thus, limits NO x bound nitrogen to NOx, as with liq uid fuels.

Furnace dimensions should be discussed with the burner supplier so that modeling of flame characteristics can be done for emissions and burnout. The partition wall that separates the furnace and convection section must prevent leakage of flue gases because gas pressure can be 10 to 30 in H 20 higher in the furnace. Leakage can result in higher CO emissions due to incomplete combustion. Completely water-cooled furnaces. A completely watercooled membrane wall design that has the front and rear walls cooled in addition to the sides (Fig. 1) offers the maximum cooling surface for a given volume. This results in lower heat release rates and, therefore, a lower heat flux. The cool front-wall design also helps to minimize NO x formation. This is because most NOx forms in a zone close to the start of the flame where a cooler front wall helps. A refractory-lined front wall or floor radiates energy back to the flame which increases NOx formation. A few decades ago, refractory-lined floors and front/rear walls were common. But a completely water-cooled design results in lower heat release rates, lower emissions and fewer refractory maintenance problems. So plant engineers and consultants generally prefer a boiler with little or no refractory. Superheater design. Superheaters in packaged boilers are preferably the convective, drainable type. They are located at an appropriate place in the convection section depending on required steam temperature and load range over which the steam temperature should be maintained. If the superheat requirement is small, around 20°F to 50°F, the superheater may be located between the evaporator and economizer. An interstage desuperheater can control the steam temperature if its actual value exceeds the desired value. Packaged boiler performance characteristics. Major boiler performance characteristics of interest to the plant engineer are: • Efficiency • Steam/gas temperature profiles • Emissions • Efficiency Bo iler efficiency depends mainly on excess air and exit gas temperature. Fig. 4 shows major boiler heat losses and the effect from exit gas temperature and excess air on efficiency. Either the lower heating value (LHV) or higher heating value (HHV) should be used when specifying boiler efficiency. The relation between the two is: efficiency (LHV basis) x LHV =efficiency (HHV basis) x HHV (1) The variation of several parameters with load or duty for a typical gas-fired packaged boiler is in Fig. 4. The gas flow decreases at lower loads, and with the same surface area, a larger decrease in gas temperature occurs. A 40°F decrease in stack gas temperature is equivalent to about a 1% improvement in efficiency. However, efficiency does not vary significantly with load as radiation losses increase in proportion to load. If the radiation loss is 0.5% at 100% Load, it would be 2% at 25% load, thus compensating for a lower exit gas temperature. Also, a higher excess air level

32

HYDROCARBON PROCESSING /AUGUST 1994

may be required at lower loads for proper combustion, thus decreasing efficiency. At a load between 25% and 100%, efficiency peaks (Fig. 4) due to exit gas loss and radiation loss. Gas/steam temperature profiles. The gas temperature throughout the boiler decreases as load decreases, starting from the furnace outlet. As a result, the convective superheater also absorbs less energy at lower loads, resulting in a lower steam temperature. If a constant steam temperature is desired from 60% to 100% load, the d esign approach is to ensure the desired steam temperature at 60% load. At higher loads, steam temperature is higher and can be controlled using interstage spray. Feed water temperature leaving the economizer decreases at lower loads. The approach point (d ifference between saturation steam temperature and water temperature leaving economizer) increases with load. This is because the ratio of gas flow to steam flow is maintained near unity in packaged boilers. With a given surface area in the economizer and a lower inlet gas temperature, the energy transferred is lower. Conversely, the approach point decreases at lower loads in gas turbine HRSGs, leading to steaming conditions in the economizer. This is due to a higher gas flow to steam flow ratio in gas turbine units, where gas mass flow does not decrease with load as in packaged boilers. Hence, the economizer transfers a large amount of energy, though the steam flow is low, resulting in lower approach points at lower loads.

Heat-recovery steam generators. Large HRSGs are generally the convective type (Fig. 2). But they can resemble large utility boilers with radiant furnaces if the firing temperature is above 1,600°F to 1,700°F. The duct burner is located ahead of the HRSG, which con sists of a superheater, evaporator and

economizer. Additional modules may be required in multi-pressure units. If a constant steam temperature is required at all loads, the design philosophy is to make sure that the steam temperature is achieved at unfired conditions. In the fired condition, although steam generation is larger, steam temperature is also higher due to a higher inlet gas temperature. Desuperheating may be restored to control the steam temperature. Simulation methods can be used to predict HRSG performance at different conditions.3

134

Table 1. Gas-turbine temperature Amb. temp, °F

exhaust conditions vs. ambient Gas flow, Ib/h 10

60 100

Table 2. HRSG performance (unfired) temperature Ambient temp, °F Gas flow, Ib/h Gas temp to superheater, Gas temp to evaporator, Gas temp to economizer, Gas temp leaving economizer Steam flow, Ib/h Steam temp, °F Feed water temp, °F Leaving economizer, °F Approach point, °F Note: steam pressure = 650 psig; = 10.2, N2 = 73.6, 02 =12.9 HRSG

Gas turbine HRSGs, unlike packaged steam generators, have fewer options for controlling emissions because exhaust gases are generated at the turbine. Modifications in gas-turbine combustors and steam/water injection have resulted in a low-NO, exhaust, around 40 ppmv. Gas turbine HRSGs refer to NOx and CO at 15% oxygen dry volume whereas packaged steam generators use 3% oxygen as the basis. 1,2 If a NOx level down to 9 ppmv is desired, an SCR is presently the only option. But combustors are now being developed by some large gas-turbine manufacturers to achieve less than 10 ppmv NOx . SCRs may be located at suitable gas temperature zones to maximize emission reductions by separating the evaporator or superheater modules. Basic differences between HRGs and packaged boilers. Effect of ambient temperature. In packaged boilers, the required combustion air is the same for a given fuel at any ambient temperature if excess air is main

Gas temp, °F

588,600 545,600 474,300

vs ambient 10 588,600 °F 900 °F 853 °F 516 °F 388 74,000 632 250 492 7 3% blow down. Exhaust analysis, vol %: consists of superheater, evaporator and

60 545,600 979 920 518 374 80,700 647 250 482 17 economizer

tained. Hence, flue-gas mass flow at a given load does not vary with ambient conditions. It is important to select a forced-draft fan to handle the desired combustion air mass flow at the lowest density case, which results in the largest volume of air. Control methods, such as inlet vane modulation, adjust the combustion air flow to maintain desired excess air or air/fuel ratio. Conversely, a gas turbine is a volume machine. The exhaust gas flow and temperature characteristics vary with ambient conditions in single-shaft machines (Table 1). This has some effect on HRSG performance, particularly on approach point, exit gas temperature and steam generation. Table 2 shows the performance of a HRSG for different ambient conditions based on data from Table 1. Note that the exit gas temperature is higher and approach point is smaller at low ambient conditions.

Efficiency vs. load. Fig. 5 shows the characteristics of a gas turbine HRSG vs. load (steam generation) at a given ambient temperature. At the lowest load, exit gas temperature is the highest and as the supplementary firing increases, more steam is generated at a higher efficiency. The ASME efficiency is defined in power test code 4.4 for HRSGs.4 The reason for the difference between HRSGs and packaged boilers is: Gas turbine exhaust typically contains 15 vol% oxygen and the mass flow through the HRSG varies only slightly with load. At higher steam demand conditions, this excess oxygen is used without adding air by raising the exhaust gas temperature to generate more steam. In effect, the excess air is reduced at higher loads. Also, due to the smaller ratio of gas/steam flows at higher loads, the exit gas temperature decreases with increased load. All these factors, coupled with lower radiation losses, result in significant improvement in efficiency with load. Steaming in economizer. This is a concern at lower loads in HRSGs because the gas to steam ratio is very high and increases at lower loads. It is nearly unity in packaged boilers at all loads. Hence, a large increase in water temperature occurs in the economizer at lower loads in HRSGs, thus reducing the approach point. Steam temperature. If the superheater is designed for a particular temperature in the unfired mode, it will increase at higher loads due to higher firing temperatures. This trend applies to HRSGs and packaged boilers. Generally, convective superheater designs are used in larte HRSGs (Fig. 2).

The author V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications and packaged water tube steam generators. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler technology from Madras University. Mr. Ganapathy is the author of over 175 articles on boilers, heat transfer and steam plant systems and has written fi ve books: Applied Heat Transfer, Stream Plant Calculations Manual, Nomograms for Steam Generation and Utilization, Basic Programs for Steam Plant EE nn gg ii nn ee ee rr ss (book and diskette), and Waste Heat Boiler DD ee ss kk bb oo oo kk ,, copies of which are available from him. He also has contributed several chapters to the EE nn cc yy cc ll oo pp ee dd ii aa of Chemical Processing and Design, Vols. 25 and 26, Marcel Dekker, New York.

Emissions. As mentioned above, the only option for emission control in HRSGs is using an SCR, while in packaged boilers several options are available. LITERATURE CITED Ganapathy, V, Waste Heat Boiler Deskbook, Fairmont Press, Atlanta, 1991. 2

Ganapathy, V, "Converting ppm to Lb/MMBtus: an easy method," P o w e r

Engineering, April 1992.

s Ganapathy, V, "Simplify HRSG performance evaluation," Hydrocarbon Processing, March 1990. 4 ASME Power test code PTC 4.4, "Gas turbine heat recovery steam generators," 1981.

BOILERS

What you should know about boilers and performance Specifying a boiler is straightforward if you know what to look for in a boiler

V. Ganapathy, ABCO Industries, Abilene, Texas Most industrial plants use oil-

and gas fired steam generators. Whenever a plant needs a steam generator or boiler, it hires a consultant or an A/E firm to develop specifications, evaluate bids from ven dors, and purchase the boiler. Often, plant engineers and operators that must live with the boiler for several decades are never consulted. First cost considerations prevail in many cases and the end

some recent trends in their design. You then can ask the right questions of consultants or A/E firms and be involved in the evaluation process before making a purchasing decision.

Trends in boiler designs

Packaged steam generators are available in capacities to 250,000 lb/h of steam at pressures to 1,500 psi and steam temperatures to 1,000 degr F Emissions of

Avoid air heaters as heat recovery equipment because of their impact on combustion temperatures and NOx levels. user or the plant is left with a boiler that has either too many maintenance problems or high operating costs. Before this happens to you, consider a few important aspects of boiler performance and

NOx are typically less than 30 ppmv for natural gas and less than 100 ppmv for CO. Low-NO X burners with fuel/air staging and flue gas recirculation achieve these results. A few places such

as California require emissions of NOx and CO below 9 ppmv and selective cat alytic reduction systems achieve those levels. The high flame temperatures of fuels of high hydrogen content yield higher NOx emissions. The temptation A/E firms and con sultants face is selecting or accepting standard models or pre -engineered de signs to capitalize on their low initial cost. However, these designs are not suitable for every situation. These designs must make serious compromises in performance -efficiency, fuel costs, boiler gas pressure drop, fan operating costs assuming that desired emission levels can be met. The design of packaged steam generators has seen several improvements. For example, water-cooled furnaces have advantages over refractory lined casings so prevalent decades ago. Radiant or semi-radiant superheaters were the norm in early designs. These exhib it operating problems such as tube over-

Table 1: Effect of excess air, FGR on flue gas

October 1997

Plant Services 131

BOILERS

Why standard designs and models are unsuitable

heating and frequent tube failures. The trend is to use shielded convectioe superheaters with interstage desuperheating to achieve longer life and lower tube wall temperatures. Avoid air heaters as heat recovery equipment because of their impact on combustion temperatures and NO x levels. Instead, opt for economizers. Also, the pressure drops for gas/air are higher with air heaters. That leads to higher fan operating costs. Back end corrosion is more of a concern in air heaters com pared to economizers. Manipulating tube pitch, length, and size or using extended surfaces in the convection section significa ntly reduces gas pressure drop through boiler while optimizing energy transfer. To summarize, look at custom -d esigned boilers when evaluating options for packaged steam generators. Look for reduced maintenance concerns, lower operating costs, and meeting the emission levels. Don't focus exclusively on first cost.

Why standard designs and models are unsuitable A 100,0001b/h model delivers a nominal 100,000 lb/h steam. In the early days, several boiler firms developed standard models or designs for various capacities. Unfortunately, one reason these stan dards are still being used today is that they save a lot of engineering effort and

drafting time for the boiler firms. It is also convenient for the consultants as they can select a model with known overall dimensions, weights, and other measures to let them proceed with layout and engineering. However, what the consultant does not consider is the fact that boiler duty, fuel heat input, air flow, and flue gas quantity, and efficiency vary with steam parameters. Excess air and flue gas recirculation rates vary by fuel and the emission levels of NOx and CO to be attained. These, in turn, affect the total flue gas quantity flowing through the boiler. Also, a superheater and economizer affect the duty.

Decades ago when concerns over emissions were minimal, excess air was typically 5 to 10 percent with natural gas and flue gas recirculation was unknown. The objective then was to maximize efficiency alone. As shown in Table 1, the flue gas quantity flowing through the boiler varies significantly depending upon duty and efficiency, even at constant boiler ca pacity. This results in high gas pressure drop and fan power consumption. The table assumes that the standard boiler is not modified to reduce gas pressure drop. It assumes constant tube size, pitch, and dimensions. With a given boil er

Lacking a proper evaluation, a standard boiler furnace is likely to have flame impingement problems. Furnace geometry is a function of emission levels. The burner supplier must evaluate flame shape using modeling techniques and the furnace dimensions. Lacking a proper evaluation, a stan dard boiler furnace is likely to have flame impingement problems. Often, 10 to 20 percent excess air and 10 to 20 percent flue gas recirculation rates meet emis sion levels of less than 0.05 lb NOx per million BTU input.

size, the exit gas temperature increas es with increased flu e gas quantity. This leads to lower efficiency and more fuel consumption and, in turn, higher flue gas flow. Thus, the boiler performance is completely different from what the stan dard boiler was supposed to do. For example, in Case 5 of Table 1, the additional fan power consumption is 70 kW over Case 1. With electricity cost at 5 Continued on page 134 132 Plant Services October 1997

Continued from page 132

cents/kWh, the annual incremental cost is $28,000. Hence, simply by overlooking the gas pressure drop, the consultant does a disservice to the plant or end-user. These calculatio ns do not consider the effect of lower efficiency. With a lower efficiency resulting from higher exit gas

the amount of fuel wasted in heat-up procedure. Fast boiler startup and heating rates are not a concern with water-cooled designs. The entire furnace expands and contracts uniformly as a unit. This elim inates relative expansion problems

In several boiler designs, floor refractory added to the problems of failures of radiant superheater tubes located at the furnace exit. temperature, the fuel input increases that occur at the interfaces between along with the flue gas quantity and gas watercooled walls and refractory pressure drop. The design of a custom casings. Fast heat up rates are designed boiler considers the duty and important in cogeneration projects the excess air, and flue gas recirculation in which the packaged steam rates on a case-b y-case basis. A good generator must deliver steam to the evaluation considers the operating costs customer within minutes of shutfor fuel and electricity. down of the gas turbines and Also the furnace dimensions of HRSGS. Fast startup rates also stan dard units may not suitable for the conserve substan tial fuel over the fuels available and current emission long-term. levels. Flame shape is a function of For a given volume, the waterexcess air and flue gas recirculation cooled unit has lower furnace area rates. The furnace width, height and heat release rates and heat fluxes length may need to be increased to because the front and rear wateravoid flame impingement. As a result, a cooled walls provide additional standard model may have a much effective surface area. Also, some higher furnace exit gas tem perature that boiler suppliers still use refractory leads to radiant superheater tube on the furnace floor that decreases failures, high boiler exit gas the effectiveness of the cooling temperatures, and lower efficiency. surfaces. With a standard or prepackaged design, the customer gets a poorly compromised op tion though the price and delivery terms may be attractive. The additional cost of a custom - The cooling of the flame and the reradesigned boiler may have a payback of diation from the refractory walls under a year. impacts NOx formation. In the case of completely water-cooled furnaces, the Water-cooled f urnaces cooler en velope for the flame reduces One of the recent improvements in pack the NOx levels. In the case of boilers aged boiler design is the water-cooled with a refractory-lined front wall, the furnace that offers several advantages reradiation from the refractory over refractory lined furnaces. The furincreases the combustion temperature, nace front, rear as well as side walls and thus adding to the NOx formation. A floor are of membrane wall construction. significant amount of NOx forms near This results in a leak-proof enclosure for the flame front and a cool envelope in the products of combustion. It eliminates the form of watercooled front wall gas leakage and corrosion of casing as helps. seen in boilers with refractory lined Some boiler vendors even use refracfront/rear walls. tory on the floor, a practice taken from The water-cooled design eliminates decades ago when engineers were not problems associated with refractory familiar with heat fluxes and boiler maintenan ce. Plant engineers and circulation. They had to resort to this personnel are familiar with the time it practice takes to replace boiler refractory and

to prevent floor tubes from overheating. However there are hundreds of boilers without floor refractory in operation. The increased area available for cooling the flue g ases reduces the furnace exit gas temperature. In several boiler designs, floor refractory added to the prob lems of failures of radiant superheater tubes located at the furnace exit. Floor refractory also adds to the flue gas recirculation levels and increases the furnace heat release rate and heat flux and in creases furnace exit gas temperature. The welded membrane partition wall between the furnace and convection section also avoids bypassing of the hot flue gases from the furnace to the convection section. Earlier designs used tangent tube construction for this partition. That promoted the formation of CO as it provided inadequate residence time for the combustion products. In creases in CO content not only decrease boiler efficiency but also add to emission levels. Radiant and convective superheaters: a comparison Radiant superheaters were the norm in the earlier designs of packaged boilers. They are located at the furnace exit di rectly facing the flame or at the turn to the convection section. Cus tom-d esigned boilers locate the superheater

Predicting the convective superheater performance is easier and more accurate. beyond a screen section and the size of the screen section can be varied depending upon the degree of superheat. The radiant superheater is located at the furnace exit. Variations in excess air, flue gas recirculation rates, and flame patterns make the exit gas temperature difficult to predict. It could vary by 100 to 200° F from predicted values. The boiler supplier could be under-estimating the tube wall temperature by 50 to a few hundred degrees thus end angering the life of the tubes. At 2,300 to 2,500° F the radiant energy is intense and signifi cant. The region is also subject to signif icant turbulence and non-uniformity in gas temperature profiles due to the turn involved. Predicting the tube wall tem 134 Plant Services October 1997

BOILERS Tubes with high fin densities have lower heat transfer coefficients and vice versa. perature is difficult. The only advantage is that the superheater is lower in cost because it is smaller. It requires less surface area because of a higher logmean tem perature difference. The convective superheater, on the other hand, is located at a zone of low gas temperature. The superheater is shielded by several screen rows that ensure not only the cooling of the flue gas but also aid in better mixing and uniformity in gas temperature profiles. Predicting the con vective superheater performance is easier and more accurate. The screen section can be designed such that the gas temperature at the superheater inlet could vary from 1,000 to 1,900° F depending on steam temperature. The advantage is longer life derived from lower and predictable tube wall temperatures . Several boilers operate at part loads for significant periods of time. The radiant superheater operates at higher radiant fluxes at lower loads compared to convective designs. Also, at partial loads, the steam flow distribution through superheater tubes as well as the gas flow across it will not be uniform. It is diffi cult to ensure uniform flow through the tubes because of the low steam side pres sure drops involved. If, say, at 100 percent load the superheater pressure drop was 50 psi, then at 25 percent load it is hardly 3 psi. Low pressure drop causes flow maldistribution, keeping some tubes from minimum steam flow required for . cooling and thus overheating them. When the radiant flux in the radi ant superheater is more, the cooling effect is reduced -a double negative. In convective designs, the gas temperatures are lower at lower loads, the heat fluxes are lower and lower steam flows will not cause increases in tube wall temperatures compared to radiant des igns It is possible to have interstage desu perheating with convective designs, while most radiant designs have a single stage, that can result in higher steam temperatures and tube wall

temperatures at lower loads. Interstage desuperheating ensures that the steam temperatures do not exceed the desired final steam temperature and that temperatures Surface area can tube be wall misleading Now that computers and spread sheet programs are easily available, a common problem among even experienced boiler engineers is comparing different designs based on surface areas. I strongly recom mend against this practice. Surface area (S) is given by the equation Q/(U x ∆T) where Q is the energy absorbed by the surface (Btu/h), U is th e overall heat transfer coefficient (BTU/ft' h° F), and ∆T is the log-mean temperature difference (° F). Unless you are familiar with computing heat transfer coefficients, do not compare on the basis of surface area. Also ∆T could vary depending on the ga s temperature in the zone where the surface is located. Use of finned tubes also distorts the picture significantly. With finned tubes, the surface areas can be 100 to 200 percent higher while transferring the same duty. This is due to poor choice of fin configuration. Tubes with high fin den sities have lower heat transfer coefficients and vice versa. In a packaged steam generator it is possible to transfer duty among radiant section, convection, and economizer in different ways. This results in different surface areas as shown in Table 2. Note that the boiler duty, efficiency, and gas pressure drop are the same for both options. The economizer uses a higher fin density in boiler 2 thus requiring more surface area while trans ferring lesser duty. Unles s you know how to compute U and can develop the gas/steam temperature profiles, com paring S values is meaningless. If gas pressure drops were different, the variations in surface areas would have been more compelling.

October 1997

Plant Services 135

HEAT TRANSFER

Boiler circulation calculations Steam

Steam generator studies can be complex. Use these guidelines to perform them effectively V Ganapathy, ABCO Industries, Abilene, Texas

N

Fig. 1. Schematic of a water tube boiler.

atural circulation water tube and fire tube boilers

(Figs. 1 and 2) are widely used in the chemi cal process industry. These are preferred to forced circulation boilers (Fig. 3) where a circulation pump ensures flow of a steam/water mixture through the tubes. In addition to being an operating expense, a pump failure can have serious consequences in such systems. The motive force driving the steam/water mixture through the tubes (water tube boilers) or over tubes (fire tube boilers) in natural-circulation systems is the difference in density between cooler water in the downcomer circuits and the steam/water mixture in the riser tubes. This flow must be adequate to cool the tubes and prevent overheating. This article explains how circulation ratio or the ratio of steam/water mixture to steam flow may be evaluated. Circulation ratio (CR) by itself does not give a complete picture of the circulation system. Natural-circulation boiling circuits are in successful operation with CRs ranging from 4 to 8 at high steam pressures (1,500 to 2,100 psig) in large utility and industrial boilers. In waste-heat boiler systems, CR may range from 15 to 50 at low steam pres sures (1,000 to 200 psig). CR must be used in conjunction with heat flux, steam pressure, tube size, orientation, roughness of tubes, water quality, etc., to understand the bo iling process and its reliability. Tube failures occur due to conditions known as departure from nucleate boiling (DNB) when the actual heat flux in the boiling circuit exceeds a critical value known as critical heat flux-a function of the variables mentio ned above. When this occurs, the rate of bubble formation is so high compared to the rate at which they are carried away by the mixture that the tube is not cooled properly, resulting in overheating and failure.

deaerator. This mixes with the steam/ water mixture inside the drum. Downcomers carry the resultant cool water to the bottom of the evaporator tubes while external risers carry the water/steam mixture to the steam drum. The heat transfer tubes also act as risers generating steam. The quantity of mixture flowing through the system is determined by calculating the CR. This is a trial-and -error procedure and is quite involved when there are multiple paths for downcomers, risers and evaporator circuits. Each boiling circuit has its own CR depending on the steam generated and system resistance. One can split up any evaporator into various parallel paths, each with its own steam generation and CR. Splitting up is done based on judgment and experience. A particular circuit may be examined in detail if the process engineer feels that it offers more resistance to circulation or if it is exposed to high heat flux conditions. Several low heat flux circuits can be clubbed into one circuit to reduce computing time. Hence, an average CR for the entire system does not give the complete picture. Circulation ratio. CR is defined as the ratio of the mass of steam/water mixture to steam generation. The mass of the mixture flowing in the system is determined by balancing the thermal head available with various system losses, including: • Friction and other losses in the downcomer piping, including bends Two -phase friction, acceleration and gravity losses in the heated riser tubes Continued

Circulation process. Fig. 1 shows a typical water-tube, natural-circulation waste -heat boiler with an external steam drum and external downcomers and riser pipes. Feed water enters the drum from an economizer or HYDROCARBON PROCESSING/ JANUARY 1998

101

• Friction and other losses in the external riser piping • Gravity loss in the riser piping • Losses in drum internals. COMPUTING THE VARIOUS LOSSES Total thermal head. The total thermal head available

in psi = H/vl/144 where H is the thermal head, ft (Fig. 1) vl is the specific volume of 3 water, ft /lb Downcomer losses. Let the average CR for the system = CR and the total steam generation = WS lb/hr. The total mixture flowing through the system = W S x CR Let the effective length (including bends) of the downcomer piping in ft = L e The frictional pressure drop, psi = 3.36 X 10-s x f 2 5 Le vi (W d) /di (Here, it is assumed that the average flow in each downcomer pipe is W d ). d i is the inner diameter of the downcomer pipe in inches. f is the friction factor. If there are several parallel paths or series -parallel paths, then the flow and pressure drop in each path is determined using electrical analogy. This calculation may require a computer. In addition to the frictional drop, the inlet (0.5 x velocity head) and exit losses (1 X velocity head) must be computed. Sometimes the pipe inner diameter is larger than the inner diameter of the nozzle at the ends, in which case the higher velocity at the nozzles must be used to compute the inlet/exit losses. Velocity V in ft /s = 0.05 Wd 2 v l/die and velocity head, psi = V /2 g vl/144. Heated riser losses. The boiling height must first be

determined. This is the vertical distance the mixture travels before the boiling process begins. It can be shown by calculation that the mixture's enthalpy entering the evaporator section is usually less than that of saturated liquid. The following is the energy balance around the steam

Fig. 3. A forced-circulation system showing multiple streams to reduce pressure drop.

Steam drum, as in Fig. 1: Wmh+W f hf =Wmh m+W s h„ Wm = mixtureflowing through the system, lb/hr =Ws x CR hv, hm, hf, and h are the enthalpies of saturated steam mixture leaving the drum, feed water entering the drum and mixture leaving the drum, Btu/lb.

h=(hv/CR)+(1-1/CR)hl where hv, h j = enthalpies of saturated vapor and liquid, Btu/lb. From the above, hm is solved for. The boiling height or the distance the mixture travels before boiling starts, Hb, is determined from: Hb = He X W S X CR X (hl - hm)/Qs where He = height of evaporator tubes, ft (For conservative calculations, Hb may be assumed to be zero.) There are basically three losses in boiling evaporator tubes: Friction loss. -1 0 2 ∆pf= 4 X 10 vl X f L X Gi X r3/di where Gi= mixture mass velocity inside tubes, 2 lb/ft hr f= fanning friction factor L= effective length, ft di= tube inner diameter, in. r3= Thom's multiplication factor for twophase friction loss (Fig. 4a). Gravity loss in tubes. ∆Pg = 0.00695 (He - Hb) r4/v1 Thom's multiplication factor for gravity loss, r4 is shown in Fig. 4c.

2

HYDROCARBON PROCESSING /JANUARY 1998

Fig. 4. Thom's two-phase multiplication factors for: a) friction loss, b) acceleration loss, and c) gravity loss.

Acceleration loss. ∆pQ=1.664x10-11x vlXGi2 xr2 where r2, Thom's multiplication factor, is shown in Fig. 4b. External riser losses. These are similar to the downcomer losses except that the specific volume is that of the mixture and not saturated liquid. Mixture specific volume v„,, ft 3/lb, is computed as: vm = vs/CR + (1- 1/CR) v l Riser gravity loss. ∆p, = (H - He)1vm/144 where vm is the specific volume of the mixture. Losses in drum internals. These usually consist of losses in the baffles and cyclones if used and range from 0.2 to 1 psi. Total losses are calculated and balanced against the thermal head available. If they balance, the CR assumed is correct, otherwise, the iteration is repeated by assuming another CR until the losses balance with the head available. When there are several boiling circuits, one can split up the total steam flow based on steam generation in each circuit until the losses balance. A simple manual procedure is to compute losses in the circuits as a function of flow and see where it intersects the available head line, Fig. 5. Since the available head and pressure drops in the riser and downcomer system are same for all the evapo

HYDROCARBON PROCESSING /JANUARY 1998

3

Table 1: Boiler data for circulation study Duty, Steam Gas Rows Surface, Fins/in x height x thickness MMBtu/hr flow, temp, 2 1-4 691 bare 11.5 11,500 1,650 5-7 2,967 2.5 x 0.75 x 0.075 27.1 27,000 1,500 8-20 20,216 4.5 x .75 x .05 39.0 39,000 1,130 Riser 3. 8 in., 8-ft long, 2 bends Downcomers: 2. 6 in., 24-ft long, s: 8 in., 12-ft long, 4 bends 6 in., 26-ft2long, 8 in., 8-ft long, 2 bends 3 Steam pressure = 645 psia. Total head =18 ft. Drum internal loss Evaporator tube ID =1.738 in length=11 ft.

rator circuits, this graphical method is sometimes used, although it is tedious. If the downcomer or external riser piping consists of several parallel or parallel-series paths, the electrical analogy is used to determine flow and pres sure drop in each circuit. A computer program best handles this problem. EXAMPLE CALCULATION

Fig. 1 is a waste-heat boiler operating under the following conditions: Gas flow = 200,000 lb/hr Gas inlet temperature = 1,650°F (vol% C02 =7, H2O = 18, N2 = 69, 02 = 6) Steam pressure = 500 psia Feed water temperature = 230°F The total steam generated is 63,5001b/hr. The first four rows are bare, followed by six finned tubes and then 10 more with higher fin density. Details of downcomers, riser pipes and other pertinent information are in Table 1. Determine the system's circulation ratio and the flow in each pipe circuit. Note that the boiler design calculations must be done before circulation studies can be taken up. Also, one must have a good feel for the downcomer and riser pipe sizes and their layout. Often piping layout is changed at the last moment to accommodate other equipment in the plant without reevaluating circulation. A computer program was developed to perform this analysis. The evaporator 4 HYDROCARBON PROCESSING / JANUARY 1998

Heat flux, Btu/ft 2-hr 20,500 83,000 58,000 bends bends

streams or paths for evaluating circulation, even though the program can analyze more circuits. Results are shown in Table 2. Analysis of results . Boiler heat -

transfer calculations have to be done before a circulation study can be undertaken. The maximum steam 0.3 psi. generation case is usually evaluated. The heat -transfer calculations give the steam generation, flux tube and gas temperatures in each section. heat In water boilers, the heat flux inside the tubes is computed, while in fire tube boilers the heat flux outside the tubes is important. For water tube boilers, heat flux, q = Uo x ( t g - ts) x At /Ai ,Where Uo = overall outside heat -transfer 2 coefficient, Btu/ft -hr-°F t g, i s are gas and steam temperatures, °F At , Ai are the tube outside and inside surface areas, 2 ft /ft. This ratio is for bare tubes, while for finned tubes it could be high, say 5 to 12. Hence, one must be careful while analyzing finned tube bundles, as the heat flux can be very high inside the tubes. In fire tube boilers, q = Uo x ( t g - ts) Based on preliminary analyses, the CR in each circuit and overall basis seems to be reasonable. The maximum heat flux at the inlet to each section is in Table 1. Correlations are available in the literature for allowable heat flux as a function of pressure, quality and tube size, etc. These are mostly based on experimental data conducted in laboratories and are often used for guidance only. The actual permissible heat fluxes are much lower and are based on industry experience and could be 10% to 30% of the values given in correlations in handbooks. Vertical tubes can handle much higher heat fluxes than horizontal tubes, up to 40% to 50% more. Limits of 2 120,000 to 175,000 Btu/ft -hr inside tubes are permitted at pressures ranging from 1,000 to 2,000 psig, while in fire 2 tube boilers the limit is around 100,000 Btu/ft -hr. The higher the steam pressure, the lower the allowable heat flux. Similarly, the higher the steam quality (lower CR), the lower the allowable heat flux. As the CR increases, the quality decreases and higher fluxes are permissible. With higher flow, the tube periphery is wetted better and is considered safer. Another approach that is widely used is the comparison between allowable steam quality and actual steam quality. Fig. 6 shows a radiant boiler furnace, where the steam quality, x, (x=1/CR) is plotted against the height. Based on heat flux distribution along the height, the allowable quality is calculated using a correlation similar to that shown below. The allowable and actual steam qualities should be apart in order to avoid DNB problems. This type of analysis is similar to that using allowable and actual heat fluxes. The McBeth correlation shown below shows the relation among the variables involved in boiling inside vertical 1 tubes: 6 -0.1 6 0.51 q, = 0.00633 x 10 x hfg di (Gi/10 ) x (1-x) 2 where q, =critical heat flux, Btu/ft -hr h f g= latent heat of steam, Btu/lb

Fig. 6. Actual vs. allowable quality and heat flux variation with furnace height.

x= steam quality, fraction (x = 1/CR) 2 Gi Gi= mass velocity lb/ft -hr di = tube inner diameter, in. For example, the critical heat flux at a steam pressure of 1,000 psi (latent heat = 650 Btu/lb), di = 2 1.5 in., Gi = 600,000 lb/ft -hr and x = 0.2 is: 6 -0.1 51 qa = 0.00633 x 10 x 650 x 1.5 x 0.6° (1-0.2) 6 2 = 2.43 x 10 Btu/ft -hr. As discussed earlier, the above equation may be used to study the effect of various variables involved and not for determining critical heat flux. Actual allowable critical heat fluxes are much lower on the order of 10% to 30% of the above value. Fire tube boilers. A similar procedure may be adopted for fire tube boilers, Fig. 2. The frictional losses in the evaporator section are usually small. The heat flux at the tube sheet inlet is high and must be considered. CR ranges from 15 to 30 due to the low steam pressures compared to water tube boilers. Generally, there is only one evaporator circuit. The correlation for allowable heat 1 flux by Motsinki is: 0.35 0.9 qc = 803 Pc x (P s /Pe) x (1 -Ps /Pc ) Where PS and Pc are steam pressure and critical pres sure of steam, psia. At 400 psia, qc = 803 x 3,208 x 0.35 0 9 6 2 (400/3,208) x (1- 400 / 3,208) . = 1.1 x 10 Btu/ft -hr. As mentioned earlier, the actual allowable flux would be 10% to 30% of this value. Forced-circulation boilers. In forced-circulation systems, the losses are determined as indicated above. However, the available head is generally too small, so a circulating pump is added (Fig. 3) to ensure desired CR. The CR may be selected unlike in a natural-circulation system, where it is arrived at through an iterative process. If the evaporator circuits are of different lengths then orifices may also be added inside tubes to ensure flow stability. CR could range from 2 to 6 in such systems to reduce operating costs. Pump reliability is a must. In gas turbine HRSGs that use horizontal tubes, the pressure drop inside tubes is

quite high compared to vertical tubes used in naturalcirculation boilers. To reduce the pressure drop, multiple streams could be considered as shown or the pump may be eliminated by locating the drum sufficiently high, resulting in a naturalcirculation system.Final thoughts. Circulation studies are complex and preferably done using a computer. The analysis of results requires experience and is generally based on feedback from operation of similar boilers in the field. Specifying a minimum CR for a boiler is not the right approach since CR varies with different circuits. One has to review the heat fluxes and steam quality at various points in the system to see if there could be problems. Some evaporator circuits could be more critical than others and require careful analysis. For example, Fig. 7 shows the front wall of a packaged water tube boiler with completely watercooled furnace design. This wall has basically two parallel flow systems between the bottom mud drum and the steam drum, namely the tubes that connect the bot tom header to the top header and the header itself, which has an L-shape. Flow calculations were done and orifices were used to ensure proper flow distribution in all the heated circuits. LITERATURE CITED Ganapathy, V, Steam plant calculations manual, 2nd edition, Marcel Dekker, New York, 1994. 2 Thom, J. R. S., "Prediction of pressure drop during forced circulation boiling of water," International Journal of Heat Transfer No. 7, 1964. 3. Roshenow, W, and J. P. Hartnett, Handbook of heat transfer, McGraw-Hill, 1972. 1

The author:V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications and packaged water tube steam generators. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler technology from Madras University. Mr. Ganapathy is the author of over 175 articles on boilers, heat transfer and steam plant systems and has written five books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization, Basic Programs for Steam Plant Engineers (book and diskette), and Waste Heat Boiler Deskbook, copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Designs, Vols. 25 and 26, Marcel Dekker, New York.

MAINTENANCE: HEAT TRANSFER

Cold end corrosion: causes and cures Calculating dew points of various acid gases and options for reducing cold end corrosion of heat recovery exchangers are presented V Ganapathy , ABCO Industries, Abilene, Texas WHENEVER FOSSIL FUELS containing sulfur are fired in heaters or boilers, sulfur dioxide, and to a small extent sulfur trioxide, are formed in addition to C02 and water vapor. The S03 combines with water vapor in the flue g as to form sulfuric acid and condenses on heat transfer surfaces, which could lead to corrosion and destruction of the surfaces. This condensation occurs on surfaces that are at or below the dew point of the acid gas. Also, when cooled below the water vapor dew point, C02 can combine with water vapor to form carbonic acid, which though weak, can attack mild steel. While thermal efficiency of the equipment is increased with reduction in exit gas temperature (or enthalpy), lower temperatures than the acid gas dew point are not advisable for metallic surfaces in contact with the gas. In municipal solid waste fired plants, in addition to sulfuric acid, one has to deal with hydrochloric and hydro bromic acid formation. This article deals with methods for solving cold, or back end corrosion as it is called, with the most commonly used heat recovery equipment, namely economizers or water heaters. These are used to preheat feed water entering the system (Fig. 1) and operate at low metal temperatures, thereby increasing their susceptibility to corrosion by sulfuric, hydrochloric, hydrobromic and carbonic acid. Estimating the dew point of these acid gases is the starting point in understanding the problem of back end corrosion. Appendix 1 gives the dew points of the various acid gases as a function of their partial pressures in the flue gas.' Fig. 2 gives the dew point for sulfuric acid.2 C0 2 + = 87%, H2 0 = 12%, N2 = 73%, S02 = 0.02%, HCL = 0.015%, 02 = 6 % , HBR = 0.01%, all by volume. To compute the sulfuric acid dew point, one should know the amount of S03 in the flue gases. The formation of S0 3 is primarily derived from two sources. 1. Reaction of S02 with atomic oxygen in the flame zone. It depends on the excess air used and the sulfur content. 2. Catalytic o xidation of S02 with the oxides of vanadium and iron, which are formed from the vanadium in the fuel oil. It is widely agreed that 1 to 5 % of S02 converts to S03 . Hence the % volume in our case would be 4 ppm, assuming a 2 % conversion. Using these numbers and after proper conversion and

substitution in the equations in Appendix 1, we have: dew point of sulfuric acid = 267° F, dew point of hydrochloric acid = 128° F, dew point of hydrobromic acid = 134° F and dew point of water vapor = 121°F. Hence, it is apparent the limiting dew point is that due to sulfuric acid and any heat transfer surface should be above this temperature (267°F) if condensation is to be avoided. There is a misconception even among experienced engineers that the gas temperature dictates the metal temperature of surfaces such as economizers. It is not so. To explain this, an example will be worked to show the metal tempera ture of an economizer with two different gas temperatures . Appendix 2 shows this calculation. Continued Hydrocarbon Processing,

January 1989 57

It can be seen that the water side coefficient is so high that the tube wall temperature runs very close to the water temperature in spite of a large difference in the gas temperatures. Thus, the tube wall temperature will be close to the water temperature and the water temperature fixes the wall temperature and hence, the dew point. Some engineers think that by increasing the flue gas temperature the economizer corrosion can be solved; not so. It should be noted also that the maximum corrosion rate occurs at a temperature much below the dew point (Fig. 3). Methods of dealing with cold end corrosion. Basically there are two approaches used by engineers to combat the problem of cold end corrosion: A. Avoid it by using protective measures such as main taining a high cold end temperature so that condensation of any vapor does not occur. B. Permit condensation of acid vapor or both acid and water vapor, thereby increasing the duty of the heat transfer surface, and use corrosion resistant materials such as glass, teflon, etc. Methods of avoiding cold end corrosion: 1. Maintain a reasonably high feed water inlet tempera ture. If the computed dew point is say 250°F, a feed water of 250°F should keep the minimum tube wall temperature above the dew point. With finned heat transfer surfaces, the wall temperature will be slightly higher than with bare tubes. The simplest way would be to operate the deaerator at a slightly higher pressure, if the feed water enters the economizer from a deaerator (Fig.1). 58 Hydrocarbon Processing, January 1989

At 5 psig the saturation is 228°F and at 10 psig it is 240°F 2. In case the deaerator pressure cannot be raised, a heat exchanger may be used ahead of the economizer (Fig. 4) to increase the feed water temperature. It may be steam or wa ter heated. 3. Fig. 5 shows a method for using an exchanger to pre heat the water. The same amount of water from the economizer exit preheats the incoming water. By controlling the flow of the hotter water, one can adjust the water te mperature to the economizer so that a balance between corrosion criterion and efficiency of operation can be maintained. 4. Hot water from either the economizer exit or the steam drum (Fig. 6), can be recirculated and mixed with the incoming water. The economizer has to handle a higher flow, but the exchanger is eliminated and a pump is added. Note that some engineers have the misconception that bypassing a portion of the economizer (Fig. 7) would solve the problem; not so. While bypassing, the heat transfer surface reduces the duty on the economizer and increases the exit gas temperature; it does not help to increase the wall temperature of the tubes, which is the most important variable. A higher exit gas temperature probably helps the downstream ductwork and equipment, but not the economizer. One benefit, however, from bypassing is

The author

APPENDIX 1-Dew points of acid gases' HCI, HBr, HN03 and S0 2 correlations were derived from vaporliquid equilibrium data .4 The H2 SO4 correlation is from reference 5. Hydrobromic acid: 1,000/TDP = 3.5639 - 0.1350 In (PH20 ) 0.0398 1 n (PHBr) + 0.00235 In (PH20 ) l n (PHBr)

V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and MSc(eng) inMr boiler tech nology from MadrasanUniversity. Ganapathy is the author of

over 150 articles on boilers, heat transfer and steam plant systems and has written four books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization and Basic Programs for Steam Plant Engineers (book and diskette), copies of which are available from him. He also has contributed several chapters to the that steaming possibilities in the economizer are minimized. Permitting condensation on surfaces. By using proper materials one can protect the heating surfaces from corrosion attack, if condensation is likely. This concept has now been extended to recovering the sensible and latent heat from the flue gases, thereby increasing the thermal efficiency of the system by several percentage points in what are called condensing heat exchangers. If flue gases contain say 10% by volume water vapor, by condensing even half of it, approxi mately 30 Btu/lb of flue gas can be recovered. This is nearly equivalent to a 120°F drop in gas temperature if sensible heat alone is transferred. A large amount of sensible and latent heat in the flue gas can be recovered if the gas is cooled below the water dew point. This implies that sulfuric acid, if present in the gas stream, will condense on the heat transfer surfaces as its dew point is much higher than that of water vapor. Borosilicate glass and teflon coated tubes have been widely used as heat transfer surfaces for this service. Glass is suitable for low pressures and temperatures (less than 450°F and 20 to 100 psig). However, presence of fluorides and alkalis is harmful to the glass tubes. One manufacturer of condensing heat exchangers uses teflon coated tubes. A thin film (about 0.015 in.) is extruded onto carbon or alloy steel tubes, and the surface is resistant to corrosion of sulfuric acid. Finned tubes cannot be used as teflon cannot be extruded onto these surfaces. Hence, these exchangers will be larger than those with extended surfaces, however, the higher heat transfer rates with condensation process improves the overall heat transfer coefficients and partly compensates for the lower surface area per linear foot of bare tubes. The high initial investment associated with condensing heat exchangers has to be carefully reviewed along with the energy recovered, fuel costs, etc. If the fuel cost is not high, then the payback period for this type of equipment may be long. Materials such as cast iron and stainless steels probably have better corrosion resistance than carbon steel, but still they are not Conclusion The article outlined the importance of the dew point of acid gas and methods for dealing with the problem of condensation on heating surfaces such as economi zers. Similar methods could be used for air heaters. The basic difference lies in the fact that the back end temperature is a function of both the gas and air temperatures. Steam air heating or air bypassing have been used to combat the problem of corrosion. Replaceable matrices and corrosion resistant materials such as enamels have been used at the cold end of regenrative air beaters.

Hydrochloric acid: 1,000/T DP = 3.7368 - 0.1591 In (PH2 0) – 0.0326 In (PHCI) + 0.00269 In (PH2o) In (PHCI) Nitric acid: 1,000/TDP = 3.6614 - 0.1446 In (P H ) Sulfurous acid: 1,000/T DP = 3.9526 - 0.1863 In (P H30) + 0.000867 In (PS0 2) - 0.000913 in (PH20) In (PS02) Sulfuric acid: 1,000/T DP = 2.276 - 0.0294 In (P H20) – 0.0858 In (PH3 SO4 ) + 0.0062 In (PH20) In (PH2 SO4 ) Where: T DP is dew point temperature (K) and P is partial pressure (mmHg). Compared with published data, the predicted dew points are within about 6K of actual values except for H2 SO4 which is within about 9K. REFERENCES 3

Pierce, R. R., "Estimating acid dewpoints in stack gases," Chem. Eng., Apt. 11, 1973. 4 Perry, R. H., and Chilton C. H., ed., "Chemical Engineers' Handbook," 5th ed., McGraw-Hill, New York, 1973. 5 Verhoff, F H., and Banchero, J. T, "Predicting Dew Points of Flue Gases," Chem. Eng. Prog., August, 1974.

APPENDIX 2-Determining tube wall temperatures of economizers The average wall temperature of a bare tube economizer is given by the simple equation: t w = 0.5[t; + t g - U(tg - t;) (1/h a - 1/h,)] Where:

h; = heat transfer coefficient inside tubes, Btu/ ft 2 h °F It. = heat transfer coefficient outside tubes, Btu/ ft 2 h °F t; = temperature of water inside tubes, °F tg = temperature of gas outside tubes, °F t w = average tube wall temperature, °F U = overall heat transfer coefficient, Btu/ft 2h °F 11U = 1 /h i + 11h., neglecting fouling and metal resistance, which are much smaller.

Typically h; = 1,000, h a = 15 and hence U = 4.77 Case 1: Determine t w when t g = 750°F and t i =50°F tw = 0.5 [250 + 750 - 14.77 (750 - 250) (0.066 - 0.001)] = 260°F Case 2: t g = 350°F, t; = 250°F t w = 0.5 [250 + 350 - 14.77 (350 - 250) (0.066 - 0.001)] = 252°F Thus, for a variation of 400°F gas temperature, the tube wall temperature hardly changes by 8°F Thus, the water temperature fixes the tube wall temperature.

LITERATURE CITED 1 2

Kiang, Yen-Hsiung, "Predicting dewpoints of acid gases," Chemical Engineering, Feb. 9, 1981, p. 127. Ganapathy, V., "Nomograms for steam generation and utilization," Fairmont Press, 1986, p. 15.

Hydrocarbon Processing, January 1989

59

Fouling-the silent heat transfer thief Better boiler water chemistry can improve overall heat duty and efficiency by minimizing scale and sludge buildup V Ganapathy, ABCO Industries, Inc. Abilene, Texas

A

boiler's primary function is to achieve and to maintain

maximum heat duty with the least operating costs and downtime. Scale and sludge are silent heat transfer thieves who slowly steal heat duty by reducing the overall heat transfer coefficient. The effects of scale and sludge are more pronounced in finned tube boilers. Tube side fouling on finned tubes generates higher tube wall temperatures. Ultimately, high heat fluxes result in tube failures. Implementing a quality feedwater program for boilers pays off in improved exchanger efficiency, reduced operating costs and reduced downtime maintenance. Clean is for better. Boilers or heat recovery steam generators perform efficiently under clean conditions. Their performance is significantly affected by fouling

Fig. 1. Bare tube HRSG for incineration heat recovery.

either on the tube or gas side whether it is a fire tube or water tube exchanger. In addition to reduced duty, steam side cleanliness impacts the tube wall temperature leading to its overheating and failure in the long run. Good water chemistry is an easy, efficient way to reduce the effects of steam-side fouling on boiler performance and tube wall temperature. Water tube waste heat recovery boilers shown in Figs. 1 and 2 will be used as examples. The concept applies to fired water tube or fire tube boilers and heat recovery steam generators also. Water tube boilers. Typical water tube waste heat boilers (Fig. 1) are used in applications such as heat recovery from municipal waste incinerator exhaust or effluents from fluidized bed cat crackers. Bare tubes minimize fouling from particulates or ash in flue gases. Finned tube heat recovery boilers (Fig. 2) need clean gas streams such as exhaust from gas turbines or fume incinerators to perform well. The bare tube boiler usually operates at low heat flux inside the tubes, in the range of 10 to 30,000 Btu/ft 2/hr, while the finned tube waste heat boiler could operate under heat flux of 50 to 150,000 Btu/ft 2hr. It is extremely important that the proper water chemistry be maintained in finned tube exchangers. A small increase in steam side fouling factor on finned tubing can increase the tube wall temperature significantly compared to the bare tube boiler. A few calculations will demonstrate the different fouling effects between the tube types.

HYDROCARBON PROCESSING / OCTOBER 1992

49

S = Q/U∆T (1) where the duty is found by Eq. 2: Design calculations. The surface area for the boiler is determined from Eq. 1

Table 1. Suggested water quality limits*

Q = WCp (T l - T 2)h1= WS∆H (2) If U is computed based on tube inner diameter, then the tube inner surface area should be used for S. Similarly, if U is based on tube outer diameter, S should be computed using the tube outer diameter. Remember that

Boiler type: industrial watertube, high duty, primary fue l fired, drum type Makeup water percentage: up to 100% of feedwater Conditions: includes superheater, turbine drives or process restriction on steam purity Saturated steam purity target Drum operating MPa 0-2.07 2.08-3.10 3.11-4.14 4.15-5.17 5.18-6.21 6.22-6.89 6.90-10.34 10.35-13.79 Pressure (psig) (0-300) Feedwater Dissolved oxygen (mg/1 02) measured before oxygen scavenger addition <0.04

(301-450)

(451-600)

<0.04

<!0.007

(601-750)

<0.007

(751-900)

(901-1,000)

< 0.007

<0.007

!

(1,001-1,500) (1,501-2,000)

<0.007

<0.0.007

Total iron (mg/1 Fe)

<0.100

<0.05

<0.03

<0.025

<0.020

<0.02

<0.01

<0.01

Total copper (mg/1 Cu)

<0.05

<0.025

<0.02

<0.02

<0.015

<0.015

<0.01

<0.01

total hardness (mg/1 CaCO3)

<0.300

<0.300

<0.200

<0.200

<0.100

<0.05

<0.05

7.5-10.0

7.5-10.0

7.5-10.0

7.5-10.0

7.5-10.0

8.5-9.5

9.0-9.6

pH range @ 25°C Chemicals for preboiler system protection

-Not detectable 9.0-9.6

Use only volatile alkaline materials

Nonvolatile TOC (mg/1 °C)

<1

<1

<0.5

<0.5

<0.5

As low as possible, <0.2

Oily matter (mg/1) water Silica (mg/1 S102) Total alkalinity (mg/1 CaCo3) Free hydroxide alkalinity (mg/1 CaC03) Specific conductance (pmho/cm) @ 25°C

<1

<1

<0.5

<0.5

<0.5

As low as possible, <0.2Boiler

<150

<90

<40

<30

<20

<8

<350

<300

<250

<200

<150

<100

without neutralization

<3500

Not specified

<3000

<2500

'Adapted from ASME 1979 consensus. See ASME 1979 for a complete discussion. HYDROCARBON PROCESSING / OCTOBER 1992

50

<2

<1

-Not Specified-

Not detectable

<2000

<1500

<1000

<150

<100

the product U x S is the same whether U and S are based on Table 2. Watertube boilers recommended tube ID or OD. U based on tube outer diameter is given by the water limits and associated steam purity following equation for bare tube boilers: At steady state full load operation 1/U o = 1/ho + (1/hi)(d./di) +ffi (d./di) +ffo Drum Range pressure, total dissolved + (d/24Km)ln (do/di) psig solids' boiler (3) and by the following equation for the finned tube boilers: water, ppm (max) 1/U= (AT/hiAi) +ffi (AT/Ai) +ffo + (ATd/Aw24Km) In 0-300 700-3,500 (d/di) + 1/η ho 301-450 600-3,000 (4) The boiling heat transfer coefficient hi inside the tubes 451-600 500-2,500 2 200-1,000 will be very high on the order of 2,000 to 3,000 Btu/ft /hr °F. 601-750 751-900 150-750 An error of 10% to 20% in its value will not affect Uo. 901-1,000 125-625 1,001 -1,800 1,801 -2,350

drum type boilers ABMA-1982 Range total alkalinity boiler water, Ppm

2

Suspended solids boiler water, ppm (max)

140-700 120-600 100-500 40-200 30-150 25-125

15 10 8 3 2 1

100 50

N/A

3

1

Correlations for computing ho. The gas-side coefficient for 2,351 -2,600 25 N/A bare and finned tube boilers consists of two components, 2,601 -2,900 15 N/A namely hc- the convective heat transfer coefficient and h, Once through boilers 1,400 and the nonluminous heat transfer coefficient. The hc is obtained above 0.05 N/A N/A from the Grimsons equation for bare tubes and from ESCOA 1 reflect the TDS in the feed water. Higher values are 1-3 Actual values within the range correlations for finned tube boilers. Tube-side fouling solids, lower values are for low solids in the feed water. factor ff i affects duty and tube wall temperatures and the 2 are directly proportional to the actual value of TDS Actual values within the range impact is more significant in a finned tube boiler compared to water. Higher values are for the high solids, lower values are for low water. a bare tube design. 3 Scale formation and fouling factor ffo. Typically, with treated feedwater and boiler water that is maintained according to ASME or ABMA guidelines, a fouling factor of 0.0005 to 0.001 ft 2/hrF/Btu could be used. If the water chemistry is not properly maintained, then sludge and scale can accumulate on the inside of tubes hindering heat transfer. ASME and ABMA guidelines for water chemistry are listed in Tables 1 and 2. Scale is a relatively hard and adherent deposit, while sludge is softer and can be easily dislodged. The buildup of scale is most severe in high heat flux areas. Scale buildup is associated with compounds whose solubilities decrease with increasing temperatures. Conversely, sludges are precipitated directly from the boiler water when their solubilities are exceeded. Scale and sludge increase the resistance to heat transfer and decrease U. Most importantly, sludge and scale raise the tube wall temperature. Fouling factor could be approximated by dividing the scale layer thickness by its conductivity: ff i = thickness of scale/conductivity Heat flux. Using the electrical analogy as an example, one can show that the heat transfer across tubes is analogous to flow of current in an electrical circuit. Current is analogous to heat flux, while voltage drop is analogous to temperature difference and resistance and fouling factor are analogous. To compute the temperature drop across the fouling layer, multiply the heat flux by the fouling factor: ∆Tf= q x ffi (6) Heat flux q computed on inner tube diameter basis is calculated by Eq. 7:

4

Dictated by boiler treatment.

silica.

These values are exclusive of

Table 3. Thermal conductivities of scale materials Material Analcite Calcium phosphate Calcium sulfate Magnesium phosphate Magnetic iron oxide Silicate scale (porous) Boiler steel Firebrick Insulating brick Table 4. Results of calculations Case 1. Gas temp in 2. Exit temp 3. Duty 4. Steam flow 5. (5) ffi The thermal conductivity of some c 6. Heat flux 7. Wall temp 8. Fin temp 9. A t/A; 10. Fins 11. Tubes/row 12. No. deep 13. Length 14. Surf. area 15. Gas Ap Units: Temp., °F Flow, Ib/hr Duty, MMBtu/hr

Thermal conductivity, B

1

2

3

1,000 520 19.65 19,390 .001 9,314 437

1,000 545 18.65 18,400 .01 8,162 516

1.13 bare

1.13 bare

20 60 8 5,024 3.0

20 60 8 5,024 3.1

1,000 520 19.65 19,39 .001 35,36 490 730 5.6 (2x.75 x.05 x.157 20 16 8 6,642 1.80

ff, ft 2/hr°F/Btu heat flux, Btu/ft 2/hr Surf. area, ft 2 Gas Ap, in., water column Length, ft

HYDROCARBON PROCESSING/ OCTOBER 1992

51

q = Uo x (At/Ai) x (T - t) (7) At, Ai refer to external and internal surface areas. T, t refer to gas and boiling water temperatures. Note that At/A i will be very large for finned tubes compared to bare

tubes. Hence, the temperature drop due to the same fouling factor can be large for boilers with finned tubes. Example. A water tube boiler for a fume incineration system is required to cool 150,000 lb/hr of clean flue gases from 1,000°F to 520°F. Gas analysis is: vol% C02 7 H2O 12 N2 75 02 6 Gas pressure is atmospheric. Steam pressure is 285 psig and feedwater is at 230°F. Carbon steel tubes of size 2 in. x 1.77 in. are used. Assume that the gas side fouling factor = 0.001 ft2/hr°FBtu; metal thermal conductivity = 25 Btu/ft/hr°F; steam side coefficient = 2,000 Btu/ft 2/hr°F; and heat loss = 2%. Design the boiler using a steam side fouling factor ffi of 0.001 ft 2/hr°FBtu (0.025 in. thick calcium phosphat e scale) and check the performance with a ffi = 0.0 1(0.006 in. thick silicate scale). Study three options: bare tubes, finned tubes with 2 x 0.75 x 0.05 x 0.157 (2 fins/in., 0.75 in. high, 0.05 in. thick with 0.157 in. serration) and finned tubes with the geometry 5 x 0.75 x 0.05 x 0.157 and transverse and longitudinal pitch = 4 in. Solution. Calculations were performed using the methodology discussed in references 1-3. The results are listed in Table 4. Several important aspects may be noted: 1. Tube wall and fin tip temperatures increase significantly when ffi increases, though Uo and heat flux are lower with increased fouling factor. The product of heat flux and ffi determines the temperature drop across

Boilers or heat recovery steam generators perform efficiently under clean conditions.

the fouling layer which increases the tube wall and fin tip temperature. 2. Duty decreases with increase in ffi. The decrement is large as a percentage with finned tubes compared to bare tubes. With 5 fins/in. design, the duty is much lower compared to 2 fins/in. design and significantly lower compared to bare tubes. We generate only 14,4001b/hr steam with 5 fin/in. compared to 18,4001b/hr with bare tube boiler for the same fouling factor, though the basic design is for the same steam generation with the same fouling factor of 0.001. 3. More surface area does not mean more duty. Increased fin density requires a larger surface area to transfer the same duty with a lower Ua . Also, the tube wall and fin tip temperatures are higher with increases in fin surface area for the same fouling factor. 4. Compared to bare tube design, a finned tube boiler is more compact, weighs less and has lower gas pressure drop for the same duty. However, one has to be careful about the falloff in performance and possible overheating of tubes with increase in tube-side fouling or scale formation.

Water chemistry is very important in boilers. Increases in the steamside fouling factor due to formation of thick layer of scale or sludge can result in reduced duty and higher tube wall temperatures. The problem is exacerbated when the heat flux across the tubes increases due to use of extended surfaces. The larger the fin surface area (obtained by using high fin density), the higher the temperature at the tube wall and fin tip. Even in fired bare tube boilers, the furnace tubes should have proper cooling. Otherwise, high heat flux and scale formation can produce overheating at the tube wall and result in tube failure. Hence, one must be very watchful and monitor tube side fouling which is affected by the feedwater quality and proper boiler water maintenance. Even if tube failures may not be the immediate concern, the decrement in energy transferred may be substantial and prompt a review of current water treatment practices. NOMENCLATURE inner tube area, ft 2/ft total external area, ft 2/ft area of average wall, ft 2/ft gas specific heat, Btu/lboF do tube outer diameter, in. di tube inner diameter, in. ff fouling factors inside tubes, ft 2/hr°F/Btu ffo fouling factors outside tubes, ft 2/hroF/Btu hi heat transfer coefficients inside tubes, Btu/ft 2/hr°F ho heat transfer coefficients outside tubes, Btu/ft2/hroF ∆HS enthalpy absorbed by steam, Btu/lb Kn, thermal conductivity of tube, Btu/ft/hr°F q heat flux, Btu/ft 2/hr Q duty, MMBtu/hr S surface area, ft 2 ∆T log mean temperature difference, °F T,t gas and steam temperatures, °F U overall heat transfer coefficient, Btu/ft2/hr°F W gas flow, lb/hr W s steam flow, lb/hr η effectiveness of fin, fraction Subscripts f temperature drop across fouling layer 1 entering 2 leaving Ai At Aw Cp

LITERATURE CITED 1 Ganapathy, V, Applied Heat Ransfer, Pennwell Books, 1982. Ganapathy, V, Waste Heat Boiler Deskbook, Fairmont Press, 1991. ue Ganapathy, V, "Evaluating extended surfaces carefully," Hydrocarbon Processing, Vol

68, October 1990, p. 65.

The author V Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications, and packaged water tube steam generators. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler technology from Madras University. Mr. Ganapathy is the author of over 175 articles on boilers, heat transfer and steam plant systems and has written five books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization, Basic Programs for Steam Plant Engineers (book and diskette), and Waste Heat Boiler Deskbook, copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Design, Vols. 25 & 26, Marcel Dekker, New York.

HEAT RECOVERY,

Simulation Aids Cogeneration System

C ogeneration systems using

Simulating theperformance of heat-recovery steam generators provides valuable information for system design, as well as for the operation of an existing system.

gas turbines and heat -recovery steam generators (HRSGs) are widely used in chemical pro cess industries (CPI) plants. Because these plants are quite expensive and the HRSG is an important part of the system, it is prudent to analyze the heat -recovery system or simulate its performance well in advance of finaliz ing plant specifications. Simulation is a method of predicting the performance of the HRSG under different operating modes and gas and steam conditions without physically des igning the equipment. Such a study will provide the engineer with valuable information about the HRSG and its performance capabilities. The simulation results could influence the choice of steam system parameters and the selection of the steam or gas turbine. In addition, one may also obtain information about the performance of the HRSG and the cogeneration system. This article explains what HRSG simulation is and the basic methodology. Its applications are then illustrated through several examples. Wha t is HRSG simulation?

heat -transfer coefficient U and then the surface area S from the equation S=Q/U∆T

The simulation process does not require the computation of U, which requires information on tube size, pitch, geometry, fin configuration, length, and so on. In simulation analysis, the term US = Q/ ∆T is computed for each surface and used in the prediction of off-design performance. The HRSG's mechanical details are not important at this stage. Thus, engineers can use the simulation methodology to obtain valuable informa tion about the thermal performance of the HRSG before developing its specifications or committing to a particular design or supplier. Two limitations apply to the simula tion technique discussed here: • The HRSG should be of the convec tive type. Nearly 85%-95% of the HRSGs for gas turbines fall into this category. • The gas and the steam sides should be clean. Gas turbine exhaust, even with oil firing, may be considered clean compared to municipal solid waste or similar applications. Why simulate?

V. Ganapathy, ABCO Industries

It is not necessary to design an HRSG in terms of surface area, tube size, fin configuration, and so on in order to evaluate its performance under different gas and steam conditions. HRSG designers arrive at the geome try of the HRSG by computing the overall

One might wonder why simulation is necessary when the HRSG supplier can provide information on the thermal performance of the HRSG. The supplier may not have the time to

CHEMICAL ENGINEERING PROGRESS OCTOBER 1993 • 27

early stages of a study or development of a cogeneration project, particularly if the steam system has multiple pressure levels and is complex. The engineer who is more familiar with the plant require ments can perform the simula tion and conduct several "What if" studies to arrive at firmer data on steam parame ters and pressure levels and to optimize the gas/steam temperature profiles based on the gas turbines being considered. Such a study may also eliminate certain gas or steam turbines from consideration due to incompatibilities between actual and required steam parameters. The HRSG designer also benefits by having fewer options to work on. And, since the plant owner will know what to expect in terms of steam-side performance, less time is spent on evaluating or second guessing the HRSG design. Some specific applications of HRSG simulation are as follows: • The exhaust gas flow rate, temperature, and gas analysis of a gas turbine vary with ambient conditions, load, fuel used, and NOx control requirements. Using simulation analysis, one can see how a given HRSG behaves when gas-side or steam-side parameters such as steam pressure, feed water temperature, or steam temperature change. • Energy recovery can be optimized by placing appropriate heat recovery surfaces one behind the other or by relocating heat -transfer surfaces, particularly in multiple pressure HRSGs, such that the gas and steam temperature profiles are closely matched. One may also con sider adding secondary recovery surfaces, such as condensate heaters, deaerators, or fluid heaters, to maxi mize energy recovery. Simulation helps in analyzing gas/steam temperature profiles in order to optimize energy recovery. • Plant engineers may use simula tion to predict the performance of 28

•

their existing units and study the effects of variations in steam and gas parameters. Such a study can also pre dict fouling of heat -transfer surfaces. If the actual duty or energy transferred is less than what is simulated under clean conditions, then one can infer that fouling has occurred. The simulation methodology

The entire theory and procedure for performing simulation analysis for a complex HRSG is outlined in (1). Here we will briefly describe the procedure for simulating the design and off

OCTOBER 1993 •CHEMICAL ENGINEERING PROGRESS

design performance of a boiler evaporator, such as the unit in Figure 1. 1. Given the gas flow Wg, inlet gas temperature T91 and steam pressure, let us assume that the evaporator has been designed to cool the gases to a temperature Tg2 and thus transfer a duty Q . The relationship among the variables (neglect ing heat losses for the sake of simplicity) can be described by the following equation, which we call the design case: Q = (US)d ∆T = Wg Cp (Tg1 -Tg2 ) = W s ∆H

(1)

The steam flow W s is obtained by computing the enthalpy added to steam, ∆H , and solving Eq. 1. We now have all of the process informa tion about the evaporator except the surface area or the geometry. One of the objectives of the simu lation process is to see how the above evaporator performs under different conditions of gas flow, temperature, and steam conditions. This is outlined below. 2. From Eq. 1, compute the term (US)d = Q / ∆T (2) Under the new conditions of gas flow and temperature, the heat-transfer coefficient will be different. This is accounted for by correcting the term (US)d for the new conditions as follows: (US)p = (US)d (Wgp/Wg)0.65F(t) (3) where Wgp is the gas flow under the new performance conditions and F(t) corrects for the different gas temperature in the off-design mode [1]. The gas -side heat transfer gov erns the overall heat -transfer coeffi cient; hence, the design heat-transfer coefficient may be corrected for the off-design case by using the fac tor (Wgp/Wg)0.65 as shown in Eq. 2. [A convective HRSG has been assumed

in which the gas-side heat-transfer coefficient and hence the overall heattransfer coefficient is proportional to the gas flow raised to the power of 0.65. [2]. 4. The new duty is Qp = (US)p ∆T= WgpCp (Tglp - Tg2p )(4) Equation 4 can easily be solved by assuming the exit gas temperature Tg2p computing ∆T , and then calculating Qp . One or two iterations may be required. Steam flow is obtained by simply dividing Qp by AH. Thus, we can arrive at the new gas exit temperature, steam generation, and heat transfer duty without computing the surface area S per se. In other words, we have simulated the off-design performance of a boiler. The complete simulation proce dure is quite involved and requires several trial-and-error calculations and iterations, particularly for multipressure, complex HRSGs with auxiliary firing. In such cases, commercial software developed for this purpose (such as (3)) may be used. [For information on general simulation packages and heat-transfer software, see CEP's Software Directory, the next issue of which is scheduled to be pub lished in December. – Editor or contact the author.]

Examples Example 1. An engineer is developing specifications for a process plant and is considering three gas turbines (designated GT1, GT2, and GT3). The major gas/steam data are shown in Table l. The HRSG operates in unfired and fired modes. Simulate the design with 15°F pinch and approach points; obtain the firing temperature and fuel consumption required to generate 40,000 lb/h of saturated steam in the off-design mode using natural gas containing 95% methane and 5% ethane. The HRSG consists of an evapora tor and economizer. "Pinch point" in this case refers to the difference between the temperature of the gas leaving the evaporator and the steam CHEMICAL ENGINEERING PROGRESS OCTOBER 1993

• 29

saturation temperature; "approach point" refers to the difference between the saturation tempera ture and feed water temperature entering the evaporator. (Reference 1 provides for more information on the selection of these values and their influence on HRSG size.) Using a simulation program, the design and off-design performances for the three machines were evaluated. The results are summarized in Table 2. It can be seen that GT2 generates more steam than GT1 in the unfired mode (27,300 lb/h vs. 25,450 lb/h). GT3 requires the least fuel (11.76 MM Btu/h) to generate 40,000 lb/h steam. The HRSG for GT3 will be larg er and, hence, slightly more expensive than the HRSG for GT2. If fuel costs are not high, then GT2 may be a better choice if the other aspects, such as electrical power from the turbines and the cost of the turbines, are favorable. The purpose of simulation is only to see how a HRSG would perform and what range of steam-side performance one can expect. Example 2. While develop ing specifications for the HRSG, engineers often assume that incorporating large surface areas can lower the exit gas tempera ture to any desired level. This may not always be thermodynamically feasible, as this example illustrates. Consider the case where 150,000 lb/h of turbine exhaust gases at 900°F generate superheated steam at 450 psig and 650°F with feed water at 240°F. Is a 300°F exit gas temperature feasible for this single pressure system?

One need not physically design an HRSG in order to obtain information about its performance.

Through simulation, the gas and steam temperatures were obtained using a 20°F pinch and a 10°F approach, as shown in Figure 2. It can be seen that the exit gas temperature is 372°F, nowhere near 300°F. Even

with a pinch and an approach of 0 (which, of course, is physically not possible as it would require infinite surface area), the exit gas temperature can be lowered to only 342°F. To lower the temperature to 300°F, sec ondary

3O •

OCTOBER 1993 •

CHEMICAL ENGINEERING PROGRESS

heat recovery or a multiple pressure HRSG is required. Engineers often compute steam generation based on a certain exit gas temperature without checking to see if it is even thermodynamically feasible. Simulation can help them in ana lyzing the gas and steam temperature profiles and make sensible demands on HRSG performance, thus saving valuable time when interacting with HRSG suppliers. Example 3. It was mentioned earlier that temperature profiles could be maximized by relocating heat -transfer surfaces. This example shows how this is done for a multiple -pressure HRSG.

steam temperature profiles obtained after performing the simulation. In summary A dual-pressure gas turbine HRSG is to be simulated using the data shown in Table 3. The high-pressure (HP) steam flow required is 31,500 lb/h and the low-pressure (LP) steam is to be maximized. Figure 3a shows an HRSG configuration consisting of a HP superheater, evaporator, and economizer followed by a LP evaporator and economizer. This design will get the job done, but it does not maximize the energy recovery.

Figure 3b shows another configuration, which generates the same HP steam but more LP steam. This is achieved by relocating the economizer, thus making the heat sink behind the LP evaporator larger. The exit gas temperature is now lower. Note that this system has four modules, while the system in Figure 3a has five modules. Table 3 shows the data and the results. Figure 4 shows the gas and

HRSG simulation is a valuable tool and has several applications. One need not physically design an HRSG in order to obtain information about its perfor mance. Such analysis can he lp consul tants, plant engineers, and HRSG design ers to learn more about the HRSG and the steam system and, hence, arrive at optimum HRSG configurations. ISO

CHEMICAL ENGINEERING PROGRESS

OCTOBER 1993 • 31

ENVIRONMENT/HEAT TRANSFER

Recover heat from waste incineration Improved waste-heat boiler design criteria enhance thermal energy recovery, reduce unit size and increase heat transfer efficiency V Ganapathy, ABCO Industries, Abilene, Texas

Using these guidelines, engineers can address critical

design problems associated with burning processwaste streams and select cost-effective waste-heat boilers. Incinerating contaminant streams is a win-win situation: 1) Complete destruction of pollutant(s) is attained and 2) Valuable thermal energy is recovered as steam and returned to process, thus conserving energy. However, recovering thermal energy from incinerated flue-gas streams contains some caveats. This treatment method creates a large high-temperature flue gas from which valuable thermal energy is recovered as saturated or superheated steam. Unfortunately, because a processwaste stream is used as feed, this stream will have variations in contaminant and component concentrations which influence the load on the boiler. Also, burning contaminants may create acid gases which will accelerate corrosion problems for the boiler at elevated temperatures. The following guidelines and checklist clarify the do's

and don'ts when designing waste-heat boilers. Packaged boilers are not designed for waste-incineration purposes. Therefore, plant engineers must supply boiler designers important operating parameters such as gas stream analysis, steam parameters, gas flowrates, etc. Developing a better understanding of the waste stream can reduce boiler size and cost, and improve its operability. Profit from waste. Incineration is a widely accepted technique to dispose of various types of chemical and refining solid, liquid and gaseous wastes. Due to stricter federal and state regulations and the need for cleaner environment, incineration has been under much criticism. During incineration, large amounts of flue gases are generated at temperatures ranging from 1,400°F to 2,400°F depending upon the wastes incinerated, destruction levels desired and equipment used. Energy from these gas streams can be recovered as saturated or superheated steam, which can be used for process or power generation. Gas stream nature affects boiler design. The starting point when evaluating a suitable boiler type for incineration-heat recovery is the nature of the gas stream, whether it is clean, dirty, dusty, corrosive, etc. If fumes, VOCs or gaseous pollutants are incinerated which is often done in refineries, the gas stream will be usually clean; thus, finned tubes can be used for heattransfer surfaces such as

HYDROCARBON PROCESSING /SEPTEMBER 1995

1

superheater, evaporator and economizer to make the boiler compact (see Fig. 1). In this boiler, the superheater is located behind a screen section, which shields it from hot gases and cavity radiation and which also helps to minimize the fluctuations in steam temperatures due to load. Gas velocities are limited only by gas pressure drop consideration and associated operating costs. In the boiler, gas velocities can be high, exceeding 80 to 100 ft/s; however, erosion is not a concern. If a hazardous waste liquid or solid waste is incinerated, the composition of the waste-gas stream, ash concentration and its analysis must be known. Table 1 lists the melting point of a few eutectics containing salts of sodium and potassium which have a low melting point. These salts can form a molten deposit on boiler tubes if the local temperature is high enough. These salts can be cooled, forming a solid mass on the tubes. This mass deposit blocks the passage for gases and also attacks the protective oxide layer on the boiler tubes, thus causing corrosion. A large water-cooled radiant furnace may be required to cool gases below their ash-slagging temperatures, followed by a convection section and an economizer, all of bare tube construction (see Fig. 2). The resulting boiler will be huge and costly. Retractable soot blowers will be required to clean accumulated slag to keep the boiler s urfaces reasonably clean. Wide -tube spacings may be necessary at the boiler front end to minimize plugging and blockage of tubes and ash hoppers may be needed. Operating parameters. Gas analysis is also important when determining steam parameters. If the gas stream contains a large amount of hydrogen chloride, high-temperature corrosion may occur. This situation is accelerated at high tube-wall temperatures exceeding 800°F to 900°F (see Fig. 3). It is preferred to avoid superheaters in such cases or if they are used, then operate them at low tube -wall temperatures. To ensure lower tube -wall temperatures, select a lower steam temperature such as 650°F to 700°F instead of 850°F to 900°F. Also, the superheater can be located in a cool gas temperature zone (see Fig. 4), buried deep into the boiler. A screen section shields the superheater from hot gases and also minimizes wild swings in tube-wall temperatures due to varying loads. The superheater can also be designed as a parallel-flow unit instead of a counter-flow unit, which has a higher tube -wall temperature due to the combination

2

HYDROCARBON PROCESSING / SEPTEMBER 1995

of higher gas and steam temperatures. The log-mean temperature difference with a parallel-flow design will be lower, resulting in larger surface area requirements. However, this is a consideration in contrast to handling problems associated with high temperature superheater design. If a high steam temperature is a must, then consider the combination of a waste-heat boiler generating saturated steam and a separately fired superheater, operating on clean fuel. Fire-tube or water-tube boilers. The choice between fire tube and water-tube boilers is often based on cost aspects and steam parameters. Fire -tube boilers can withstand high gas pressures, are cost-effective up to 70 to 100,000 lb/h gas capacity and are ideal for saturated steam generation. When steam pressure exceeds 600 to 700 psig, the tube thickness must withstand external pressure from steam increases and the weight. Result: Cost of the boiler increases rapidly. A supe rheater if used with a fire -tube boiler must be

located either at the front or at the exit of the evaporator, which affects the superheater's performance and metallurgy. When gas inlet temperatures are high, exceeding 1,600°F to 1,900°F, do not locate the superheater at the gas inlet zone due to the higher tube wall temperatures and the large fluctuations of steam temperature with load. Locating it behind the evaporator would result in a very large superheater because of the lower gas temperature. Also, a high steam temperature may not be achievable. Water-tube boilers are very flexible in design, and

3

HYDROCARBON PROCESSING !SEPTEMBER 1995

are suitable for large gas and steam flows and high steam pressures. They can be made compact by using extended surfaces which can transfer energy at much lower pinch points (difference between saturation steam temperature and gas temperature exiting evaporator) compared to firetube boilers. Superheater location is very flexible, often in between evaporator modules and hence can result in better designs with lower tube-wall temperatures. Water holdup is also less compared to fire-tube boilers, resulting in quicker transient performance. Due to compact designs, they often have lower gas pressure drop, thus reducing operating costs. With gas temperatures exceeding 1,800°F to 2,000°F, an elevated fire -tube boiler (Fig. 5) is preferred because the tube sheet at the front can be completely cooled by a circulating steam-water mixture. Often refractory is also used at the tube sheet with ferrules to protect the tube sheet and minimize the temperature drop across it. An elevated-drum design enables using better steam drum internals, hence improving steam purity. A single shell-fire tube boiler (Fig. 6) is inexpensive and may be used when gas temperature is less than 1,600°F. Typically, these boilers operate at heat fluxes of

Table 2. Dew points of acid gases' Hydrobromic acid: 1,000/TDP = 3.5639-0.1350 In (P H20)0.0398 In (P HBr) + 0.00235 In (P H20) In (P HBr ) Hydrochloric acid: 1,000/TDP = 3.7368-0.1591 In (P H20 ) - 0.0326 In (P HCI) + 0.00269 In (P H20 ) In (P HCI) Nitric acid: 1,000/TDP = 3.6614-0.1446 In (P H20) 0.0827 In (PHN03) + 0.00756 In (P H20) In (PHN03) Sulfurous acid: 1,000/TDP = 3.9526-0.1863 In (PH20)+ .000867 In (PSO 2) - 0.000913 In (PH20)In (PSO 2) Sulfuric acid: 1,000/TDP = 2.276-0.0294 In (PH20) 0.0858 In (PH3S04) + 0.0062 In (PH20) In (PH2s04) Where: TDP is dew point temperature oK and P is partial pressure, mmHg. Compared with published data, the predicted dew points are within about 6°K of actual values except for H 2SO4 which is within about °K.

Table 3. Engineering data or checklist to design waste-heat boiler2 (or aspects to be considered while developing specifications) 1. Application: Describe process; give flow diagram for gas/steam; source of deaeration steam; distribution of process and superheated HP, IP and LP steam. 2. Space limitations: Describe or provide drawings. Is a site visit required? 3. Gas stream parameters: a. Gas flow, pph: (at different loads/ambient conditions). b. Inlet gas temperature, °F: (associated with gas flow) c. Analysis, o/a volume: C02, H 2O, N 2, 0 2, S0 2, HCL, S03, H2S, CL 2; etc., corresponding to each gas flow condition d. Gas pressure, psig e. Suggested gas pressure drop: (at a given gas flow and inlet condition) or the maximum value. f. Nature of gas: dirty/clean; particulate concentration in grains/scf; ash analysis to indicate slagging tendencies. 4. Duty or suggested steam generation/temperature profile or exit gas temperature: Part load conditions; a/o of time in each operating mode; steam temperature control if any; provide flow diagram showing HP/IP/LP steam, makeup water, condensate returns at various loads. 5. Auxiliary fuel data: Fuel analysis, augmenting air for burners if any. 6. Emission data: NOx and CO at boiler inlet and outlet; Contribution by burner (from burner vendor); Emission control equipment suggested. 7. Feedwater or makeup water analysis: a/o condensate returns if any. Is demineralized water available for spray temperature control? Injection water TDS should be very low. 8. Cost of fuel, electricity and steam: In addition, the % of time in each operating point in order to optimize life cycle cost. 9. Steam purity requirements: Drum holdup time criteria if any.Quick startup or load change requirements. 10. Other special requirements, if any.

perature. This result differs from a packaged boiler, where the steam temperature falls with load. Influential design factors. Since several variables influ ence

less than 25,000 Btu/ft 2 h. Heat fluxes up to 100,000 Btu/ft 2 h can be tolerated by steam at low to medium pressures. 1 If the gas stream does not contain hydrogen and large amount of water vapor, which is the case of reformed gases in hydrogen plants, the heat flux is not a concern.2 Boiler performance. As with any boiler, lowering the exit gas temperature increases energy recovery. But if the gas stream contains acid vapors such as hydrochloric acid, hydrobromic acid, sulfuric acid, etc., one must evaluate dew points so that the lowest tube-wall temperature at the economizer is close to or above the stream's dewpoint. Table 2 shows dew point correlations for several common acid vapors. Since the tube-side heat transfer coefficient in the economizer is very high compared to the gas-side coefficient, the feedwater temperature will determine the tubewall temperature and not the gas temperature. Incinerators typically operate at high gas temperatures with the mass flow changing due to load. Hence boiler performance is different from a conventional packaged steam generator. Fig. 7 shows the effect of mass flow and gas temperature on the performance of the waste heat boiler. It appears that the superheater steam and tubewall temperatures increase at lower loads from a combination of lower steam generation and high gas inlet tem 56

HYDROCARBON PROCESSING/ SEPTEMBER 1995

boiler type and its design features, plant engineers involved in design or operation of an incineration facility should provide proper information to boiler designers. Table 3 lists important parameters that should be furnished to boiler vendors. Of particular importance are the gas analysis, ash analysis (if present), steam parameters and gas flow in mass units. LITERATURE CITED Ganapathy, V, Steam plant calculations manual, Second Edition, Marcel Dekker, New York, 1992. z Ganapathy, V., Waste heat boiler deskbook. Fairmont Press, Atlanta, 1991.

The author V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications and packaged water tube steam generators. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler technology from Madras University. Mr. Ganapathy is the author of over 175 articles on boilers, heat transfer and steam plant systems and has written five books: Applied Heat Transfer, Stream Plant Calculations Manual, Nomograms for Steam Generation and Utilization, Basic Programs for Steam Plant Engineers (book and diskette), and Waste Heat Boiler Deskbook, copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Design, Vols. 25 and 26, Marcel Dekker, New York.

Optimize energy efficiency of HRSG With a better understanding of temperature profiles, plant engineers can increase steam production and minimize losses V. Ganapathy, ABCO Industries, Abilene, Texas

When optimizing the efficiency of heat-recovery steam

Arbitrary selecting the exit-gas temperature or pinch and approach points to estimate steam generation can be faulty and cause "temperature cross." For example, assume that the gas flow, inlet-gas temperature, desired steam temperature and feedwater temperature in the design case are known. Assuming a modest pressure drop in the superheater,then the drum pressure and saturated steam temperature can be derived steam tables. The pinch and approach points are set in the design mode. (For the off design case pinch/approach points and steam flow, an evaluation of the HRSG performance must be done using iterative procedures where: 1,2

generators (HRSGS,) fully understanding the temperature profiles of these units is crucial. Gas turbine HRSGS are unique and pose peculiar problems as compared to conventional gas/oil-fired boilers when calculating efficiency or steam output. For conventional oil/gas -fired steam generators or boilers, one can do heat balances, efficiency calculations and fuel estimates by assuming an exit-gas temperature of 300°F to 340°F from the steam generator and assuming that the feedwater is at 220°F to 250°F irrespective of steam pressure. However, these assumptions are often not Pinch point (PP) = gas temperature thermodynamically valid with leaving evaporator-saturation HRSGS. Energy recovered in temperature--(t g3 -ts)Approach point a HRSG or the exit-gas (AP) =saturation temperature-water temperature from a HRSG is a temperature leaving economizer(ts function of several variables tw2).By assuming a nominal pinch including: point (10°F to 30°F in unfired mode) • Gas inlet temperature to and a similar value for the approach the HRSG Fig. 1. Pinch and approach points in a simple HRSG. point, the amount of steam generated • Steam pressure • Steam temperature in the HRSG may be calculated as: • How the heat recovery surfaces are arranged tg3 = ts + PP tw2 = ts -AP • Number of steam pressure levels • Pinch and approach points. The energy absorbed by superheater and evaporator, Q 1 2 , The reasons are: is given by: • Low gas-inlet temperature to HRSGS (900°F-1,100°F in Q12 = Wg x Cpg x (tg1-tg3)xhl = Ws X (hs2 - hw2) unfired mode compared to the adiabatic combustion tem( where: Wg = gas flow, lb/h perature of around 3,300°F in oil/gas-fired steam generators) Cpg = gas specific heat at the average gas • Large ratio of exhaust gas flow to steam generation in temperature, Btu/lb F unfired mode (about 6 vs. 1.1 in conventional steam Tg1,tg2,tg3,tg4= gas temperature at various locations as generators) • How gas-to-steam flow ratios change with shown, °F steam generation. W S = steam flow, lb/h In conventional steam generators, the ratio of gas -to-steam flow does not change with steam generation. While in gas hs2, hw2 = enthalpy of steam at superheater exit turbine HRSGS, the exhaust gas flow remains nearly and water at economizer exit, Btu/lb°F hl= heat loss factor, constant irre spective of steam generation, which affects the typically 0.99 to 0.995. From Eq. 1, Q12 and then the steam gas/steam temperature profiles significantly. Several flow, Ws, may be calculated. In Eq. 1, blow down was methods can be used to understand HRSG temperature neglected. profiles and offer ways to improve operating efficiency or If we consider the energy absorbed in the energy recovery. superheater,Q1 Pinch and approach points determine HRSG gas/steam temperature profiles. In a typical HRSG, the gas and steam Q1=WgxCpgx(tg1-tg2)= Ws X (hs2 - hv) (2) temperature profiles are dictated by the design values for pinch and approach points. Fig 1 shows the gas/steam temperature profiles in a simple HRSG consisting of a HYDROCARBON PROCESSING / DECEMBER 2001 41 superheater, evaporator and economizer.

where hv = saturated steam enthalpy, Btu/lb Since the right hand side of Eq. 2 is known, Q1 is known, then the only unknown is the gas temperature leaving the superheater, tg2,which may be solved for. Since Q12 is known from Eq. 1, then Q2 = energy absorbed in the evaporator-Q12-Q1. Now from the economizer energy balance, the energy absorbed Q3 is given by: Q3 = Wg x Cpg x(tg3-tg4) hl = Ws(hw2-hw1) (3) where hw1 = enthalpy of feed water in, Btu/lb. The unknown is tg4,which can be solved for. Thus, all of the gas/steam temperatures and duty in each section can be determined along with the steam generation. The exit gas temperature, tg4 or steam flow cannot be arbitrarily assumed in a HRSG. An analysis similar to that presented should be performed. Arbitrarily selecting tg4 or pinch and approach points can lead to "temperature cross" situations an undesirable condition. Never arbitrarily assume pinch and approach points. The

overall energy transferred is: Q13 = Wg x Cpg x (tg1-tg4) x hl = W S x (hs2 - hw1)

(4)

From Eqs. 1 and 4, we have (tg1-tg3)/(tg1-tg4) = (hs2 - hw2/(hs2 - hw1) = K

(5)

This equation neglects small variations in gas specific heats. K is a function of steam/water properties and is nearly a constant for given steam/water conditions. For steam generation to occur, two conditions must be met: tg3>ts and tg4>tw1 If the pinch or the approach point is arbitrarily selected, then there is a high probability that tg4 can be lower than tw1 or tg3 is lower than ts. The lowest limit for tg3 is ts and for tg4 is tw1. Substituting these conditions into Eq. 5 and calling the tg1 as tg1c, we have: (tglc-ts)/(tglc-twl)=K or tglc=(ts-Ktwl)/(l-K)

(6) For gas-

inlet temperatures greater than tglc, the feedwater

42

HYDROCARBON PROCESSING / DECEMBER 2001

temperature limits the temperature profile. For values below tglc, the pinch point governs the temperature profile. Now let us illustrate these issues using a few examples. Example 1. Assume the steam pressure in a HRSG = 585 psig; steam temperature = 700°F; feedwater temperature = 250°F. Let the approach temperature = 20°F, ts. = 488° F, tw2 = 468°F. From steam tables, hs2 = 1351.8, hw1 = 219.5 and hw2 = 450.7 Btu/lb. From Eq. 5, K = (1351.8 -450.7)/(1351.8 - 219.5) = 0.796 From Eq. 6, tglc = (488 -.796 x 250)/(1 -.796) = 1,416°F. Below 1,416°F, pinch point determines the profile and above this value, the feedwater temperature sets the profile. If a pinch point is arbitrarily selected, let us assume that gas-inlet temperature is 1,600°F and the pinch point is a 30°F: Tg3=ts+30=488+30=518°F From Eq. 5, then: (1600 - 518)/(1600 – tg4) = 0.796 or tg4 = 240°F, which is below the 250°F feedwater temperature; thus, not a feasible profile. However, if we used a lower tw2 or higher K, it may work out. Assume tw2 = 400°F and K = 0.862, or tg4 = 345°F, which is feasible. Example 2. Assume the exit-gas temperature = 290°F. From Eq. 5, then: (1600 – tg3)/(1600 - 290) = 0.796 or tg3 = 557° F or pinch is (557 - 488) = 69°F, which is feasible. Example 3. Assume tg1 = 900°F and pinch point = 20°F, then: (900 - 508)/(900 – tg4) = 0.796 or tg4 = 408°F, which is feasible. Example 4. Assume an entry temperature of 900°F and an exit gas temperature = 300°F. Is it feasible? (900 – tg3)/(900 - 300) = 0.796 or tg3 = 422°F, which is below ts and thus, not feasible. Arbitrarily fixing the exit gas temperature in a single-pressure HRSG to determine the possible steam generation is not a good idea. Table 1 suggests typical pinch and approach points to be used in waste-heat boilers. A detailed evaluation should be done to confirm these assumptions of pinch and approach points. Also, in off-design modes, the pinch and approach points must be determined through complex iterative calculations. Also, it is a good idea to select pinch and approach

points for the HRSG evaporator in unfired mode since firing introduces uncertainty and possible temperature cross situations. EFFECT OF STEAM PRESSURE AND TEMPERATURE Steam pressure and temperature impacts the exit gas temperature, and hence, the steam generation. Table 2 shows the effects of steam pressure and temperature on exit-gas temperature in an unfired gas turbine HRSG. Some observations are: The higher the steam pressure, the higher the exit-gas temperature. The saturation temperature is greater at higher pressures; thus, the gas temperature, tg3, leaving the evaporator is higher with less steam generation as compared to lower steam pressure case. Steam generation is proportional to (tg1-tg3 ). With less steam generated, the water flow through the economizer is lower and the heat recovery potential is also reduced. From Eq. 3, tg3 is fixed by assuming a pinch point. Q3, the economizer duty, is smaller, but tg4 is higher as shown in Eq. 3. Higher the steam temperature (for the same pressure), the higher the exit gas temperature. Less steam is generated in the evaporator as shown by Eq. 1. If the enthalpy absorbed by steam is higher (as with superheated steam), less is generated, which results in higher tg4 . This is the reason for the lower exit gas temperature with saturated steam in Table 2. With less steam and less water flowing through the economizer, Q3 is smaller, which causes less heat recovery and higher exit-gas temperatures. Exit-gas temperature cannot be arbitrarily assumed or determined. If someone had assumed a 300°F exit gas for the last case of 600 psig, 750°F steam, and determined the steam generation, the error would be very significant, about 17%. Also, it is not thermodynamically feasible. The author has seen inexperienced engineers still do this. HRSG simulation methods (as discussed previously for the design case) must be used to determine gas/steam profiles and steam generation. 1,2 Energy-recovery calculations for HRSGs should be done with the methods previously described since arbitrary assumptions can lead to errors. Improving energy recovery by lowering pinch and approach points. With these procedures to evaluate HRSG temperature profiles, let us investigate how to improve a typical HRSG.

Several methods are available. One method to optimize energy recovered or lower exit-gas temperature is reducing the design pinch and approach points when thermodynamically feasible. Lower pinch and approach point imply that the log-mean temperature difference in the evaporator and economizer would be reduced and hence more heating surfaces would be required; thus adding to the total cost of the HRSG. A techno-economic evaluation may be done based on operating hours to see if the lower pinch/approach point design is worth it. Table 3 lists operating conditions of a HRSG for a 4500-Kw gas turbine generating 150-psig saturated steam. It was designed with a large pinch/approach point in the unfired mode (Case A) and with smaller pinch and approach points in Case B. For these two design basis, the performance is also verified in the fired mode, when 30,000-lb/h of steam is required. It is obvious that in Case A, the firing temperature will be higher; thus more fuel is consumed. Also due to the larger surface areas of evaporator and economizer in Case B, the gas pressure drop across the HRSG will also be higher than in Case A. In this example, assume: Fuel cost = $5/MM Btu/ h Cost of steam = $4/1,000 lb Electricity = 7c/kwh. Also assume an additional 4 in. WC in the HRSG, which equals 1.1 % decrease in gas turbine power output of nominal 4500 Kw. Assume 8,000 hours of operation annually with 50% of the time in unfired and fired modes. The following simplistic evaluation may be done. Design B has an edge over design A in terms of operating costs: Higher steam generation in unfired mode: (22,985 22,107) x 4 x 4,000/1,000 = $14,048 • Higher fuel consumption in fired mode: (8.1 - 6.9) x 5 x 4,000 = $ 24,000 • Higher gas per drop of 1.2 in WC: 1.2 x 4,500 x 0.07 x 8,000 x 1.1/(4 x 100) =-$8,316 The net benefit of design B over A = $(14,048 + 24,000 8,316)=$29,732/yr.

HYDROCARBON PROCESSING/ DECEMBER 2001 3

If the additional cost of Design B over A is about $30,000, then the payback is about one year. One may perform such an evaluation to see if the lower pinch/approach points are worth the expense. If the HRSG will operate continuously for years, then the author suggests investigating the lowpinch-point designs. Such conditions may justify the additional investment for the long-term. The cost of the HRSG is also not directly proportional to surface areas; an additional 10% to 15% may be included in the total HRSG cost for instrumentation, controls, duct work, burner, etc., which can be a substantial increase to the total cost. Improving HRSG efficiency through auxiliary firing . One of the simplest methods to improve the efficiency of a HRSG is to increase the steam output through supplementary firing. Unlike conventional steam generators where the ratio of gas-to-steam flow remains nearly constant at all loads, the ratio varies for a HRSG as shown in Table 4. Auxiliary firing without additional air is feasible in HRSGs since the typical exhaust contains 13 to 15 vol% oxygen. The relation between oxygen content in exhaust gas and fuel consumption is shown in Eq. 7:1 • 2 Q = 58.4 x 10-6W gO

(7)

where: Q = fuel input (oil/natural gas) in MMBtu/h on LHV basis (lower heating value) Wg = exhaust gas flow, lb/h O = vol% oxygen consumed. Example: If the exhaust gas is at 900°F and must be raised to 2,000°F, then the fuel input required = Wg x 0.31 x (2000 900) x 10-6 MMBtu/h. Equating with Eq. 7,we have: O = 5.83 vol%. If we started with 15 vol% oxygen, we would end up with about 9 vol%; thus, there is excess oxygen in the exhaust gases. Gas turbine HRSGs, which are free oxygen at 14 to 15 vol%, can easily fire up to about 3,000°E The HRSG design would require water-cooled membrane walls beyond 1,700°E Fig. 2 shows a HRSG firing to 2,300°E It consists of an 0-type boiler with water-cooled furnace section followed by a screen section, superheaters, convection bank and economizer. Fired HRSGS are efficient because: • When additional fuel is fired, air is not generally added. This reduces the exit gas losses and improves the HRSG efficiency. More steam is generated with less heat losses from exhaust gases. • With a single-pressure HRSG, the steam flow is higher with firing; the economizer acts as a larger heat sink and lowers the exit-gas temperature as compared to unfired case (see Table 3).

4

HYDROCARBON PROCESSING / DECEMBER 2001

The additional HRSG duty between unfired and fired case is 7 MM Btu/h in case B. We only used 6.9 MM Btu/ h (LHV) as fuel input. Thus, it is more than 100% efficient. Plant engineers should always think of increasing steam generation in cogeneration plants by auxiliary firing in HRSGs. The efficiency is much higher than in steam generators, which typically have an efficiency of 93% (LHV basis) for natural-gas firing. Thus, fuel utilization is improved by 6% to 7% when compared to regular steam generators. • In conventional plant steam generators when excess air is increased, the losses increase and efficiency is reduced. In HRSGs, auxiliary firing cuts excess air requirements. The oxygen levels in the exhaust gas are lowered and the efficiency is improved. Improve efficiency by adding secondary surfaces. By adding secondary surfaces such as condensate heater or deaerator coil or a feedwater exchanger, as shown in Fig 3 below, more energy can be recovered from the exhaust gases. These options reduce the amount of steam required for deaeration by preheating the makeup or generating low-pressure (LP) steam. Corrosion may occur in the economizer or condensate heater when fuel oils are fired and should be investigated. Optimize efficiency through temperature profiles. With multiple pressure HRSGs, it is possible to relocate the heating surfaces and lower the exit-gas temperature and generate more steam. Fig 5 shows two options for generating high-pressure (HP) steam and LP steam in a HRSG. System conditions are: Gas flow = 500,000 lb/h Exhaust-gas temperature = 900°F Gas analysis = vol%-CO2 = 3, H2 O = 7, N2 = 75, OZ = 15 HP steam = 31,500 lb/h at 600 psig, 600°F LP steam: maximum in unfired mode at 150-psig sat Feedwater = 230°F Assume heat loss = 1%. Using HRSG simulation methods, two options were studied as shown in Fig. 4. In Case A, the HP stage is followed by the LP evaporator with a common economizer feeding both the evaporators. In Case B, the HP stage is followed by the LP stage each with own individual economizer. In Case A, the common economizer has a larger water flow (equal to the sum of the steam flows in HP and LP stages.) Result: A much larger heat sink and the gas temperature is 306°F and is much lower than Case B, which is 342°F. More steam is generated in case 6. The LP steam is 43,450 lb/h in Case A vs. 39,000 lb/h in Case B while the HP steam is 31,500 lb/h in both cases. Rearranging the heating surfaces or relocating the surfaces can improve the gas/steam temperature profiles, particularly with multiple pressure steam generation. HRSG simulation-program

developed by the author-can be used to study complex multiple pressure unfired or fired HRSGs.1,2 Overview. HRSG efficiency can be improved by several methods.

One can use any or the entire presented methods on a HRSG project. Plant engineers must understand the significance of HRSG temperature profiles. One must also look at the cost implications and the period of operation to justify any additional expense. With HRSGs being used widely in cogeneration and combined cycle plants, a consideration for all of the presented methods will assist when optimizing the total plant efficiency. LITERATURE CITED t Ganapathy, V, Steam Plant Ca lculations Manual. Second Fd., Marcel Dekker, New York, 1993. z Ganapathy, V, Waste Heat Boiler Deskbook, Fairtnunt Press. .Atlanta, 1991.

V.Ganapathy is a heat transfer specialist at ABCO Industries, Abilene, Texas, a subsidiary of Peerless Manufacturing, Dallas. He has a bachelors degree in mechanical engineering from LI.T, Madras, India, and a masters degree from Madras University. At ABCO, Mr. Ganapathy is responsible for steam generator, HRSG and waste heat boiler process and thermal engineering functions, and has 30 years of experience in this field. He has authored over 250 articles on boilers and related subjects, written four books and contributed several chapters to the Handbook of Engineering Calculations and the Encyclopedia of Chemical Proc essing and Design. He can be reached via e-mail: [email protected].

Specify Packaged Steam Generators Properly Packaged Packaged steam generators are no longer purchased

"off the shelf." Here's how to evaluate the options.

V. Ganapathy, ABCO Industries

G2

• SEPTEMBER 1993 •

steam

generators

firing

natural gas and fuel oils are widely used in chemical process industries (CPI) plants, petroleum refineries, and cogeneration systems to gener ate steam for process use or for power. These units are generally shop assembled (Figure 1) and can generate saturated or superheated steam at capacities up to 200,000 lb/h, pressures ranging from 100 to 1,200 psig, and steam temperatures from saturated condit ions to 900°F. (The 200,000 lb/h capacity is set by ship ping constraints on generator size. When the required capacity exceeds this, the steam generator has to be field-erected.) Packaged steam generators were gen erally considered "off-the-shelf" items until a few years ago. That is, manufac turers standardized on several sizes (for example, 20,000 to 120,000 lb/h in steps of 20,000 lb/h) and selected the unit that was closest to the needs of the customer. If the manufacturer had standard 80,000and 100,000-lb/h units, and if one customer wanted an 89,000-lb/h boiler while another wanted a 105,000-lb/h boiler, both customers would get the same 100,000lb/h boiler. In the past, manufacturers and customers did not concern themselves much with the implicat ions of variations in excess air, gas recirculation, staged combustion, and so on. However, these factors influence emissions of nitrogen oxides (NO) and carbon monoxide and affect the boiler size and performance - and they are becoming more and more imp ortant in boiler selection and design.

CHEMICAL ENGINEERING PROGRESS

• Figure 1. Large packaged steam generator.

(The implications of emissions on boiler design and the need for customized design will be discussed later in the article.) Of course, a standard "off-the-shelf" design is less expensive than a "customdesigned" unit, which considers each requirement on a caseby -case basis. This article explains what packaged steam generators are, discusses various aspects of design, and outlines what information the user needs to provide to the supplier when specifying packaged steam generators.

Generator configurations Two main types of packaged steam generators are the D-type and the Otype, shown in Figure 2. In the D-type, the burner is mounted in the front wall. The products of combus tion, on leaving the furnace, make a 180 deg. turn and pass through the convection bank of tubes, which may contain a superheater if superheated steam is gener ated.

The economizer is generally located outside the boiler. In the O-type, the burner is (as in the Dtype) mounted in the front wall. The products of combustion leave the furnace and either come back toward the front (denoted O-type-1 in Figure 2) or move straight on (denoted 0-type-2). The bottom drum in the O-type may be replaced by two smaller drums and another possible configuration results namely the A-type. The choice of boiler type varies among manufacturers. Natural gas and oil are the main fuels used in these types of boilers, although solid fuels have been fired with equipment modifications. This article deals only with clean fuels such as gas and oil. Natural circulation moves the steamwater mixture through the evaporator tubes. The rear convection tubes, where the gas temperature is low (within 100°F to 250°F above the saturation temperature), act as downcomer tubes, while the rest of the tubes in the radiant and convection section act as risers, carrying the steam-water mixture to the steam drum. Depending on how the steam is utilized, cyclones and chevron steam separators (Figure 3) are used to achieve the desired steam purity, which is in the range of 10 ppb to 1 ppm. If a superheater is required, it is generally preferred that it be drainable. It can be located in the furnace (radiant type) or in the convection zone (convective type). Radiant superheaters should be designed with greater care, as they are subject to flame radiation and frequent tube failures. Convective superheaters (Figure 4) are the choice for steam temperatures less than 750°F ; a semiradiant design may be used for steam temperatures up to 950°F. Steam temperatures are generally controlled using an interstage desuperheater or a spray attemperator, which injects water into steam to control its temperature. If a spray attemperator is used for steam temperature control, the feed water should preferably be demineralized and have a very low solids content. Otherwise, the superheated steam could be injected with solids, which could be deposited inside the superheater tubes, leading to corrosion

• Figure 2. Boiler configurations.

N Figure 3. Arrangement of drum internals.

and overheating, or deposited in the steam turbine, leading to turbine failure. If the degree of superheat is very low, say 20-50°F, the superheater may also be located between the evaporator and economizer. An economizer is often used to recover energy from the flue gases once they leave the evaporator.

CHEMICAL ENGINEERING PROGRESS SEPTEMBER 1993

63

Extended surfaces are utilized to make the economizer compact. The feed water entering the economizer should be deaerated to mimimize tube side corrosion and pitting. Feed water temperatures above 240°F are preferred when firing fuels containing sulfur to avoid dew-point corrosion. In gas-fired units, the typical exit gas temperature from the economizer ranges from 300°F to 350°F at 100% load. As the load decreases, the exit gas tempera ture decreases. Air heaters are rarely used for heat recovery, because they are less compact and because they also increase the flame temperature, which results in higher NOx emis sions. In addition, the gas and air pressure drops are higher and, hence, result in higher operating costs. Air heaters may be required when firing "difficult" fuels, such as coal and solid waste, however these are not considered in this article. 3

• SEPTEMBER 1993 •

Boiler efficiency depends on the excess air and exit gas temperature, as illustrated in Table 1. Efficiency is based on either the lower or the higher heating value (LHV and HHV) and is denoted E, or Eh , respectively. The relationship between the two efficiencies is simple: E, x LHV = Eh x HHV Simple calculations are also available to relate excess air and efficiency based on data from flue gas analysis (1,2). In large boilers (above about 100,000-lb/h capacity), the efficiency is of great significance, because it impacts operational costs. For example, if the difference in efficiency between two 100,000-lb/h boilers is just 1%, about 1.3 million Btu/h of additional fuel is consumed by the less-efficient unit. This translates into $26,000/yr (based on fuel costs of $2.5/million Btu and 8,000 h/yr operating time).

CHEMICAL ENGINEERING PROGRESS

Packaged steam generators are usually of the pressurized furnace type, in which a forced-draft fan forces the combustion air and flue gases to the stack via the burner, furnace, convection tubes, and economizer. Large field-erected boilers are usually of balanced-draft construction, in which a forced-draft fan handles the air and an induced-draft fan moves the flue gases. Balanced-draft designs are not economical for packaged boilers. Several decades ago, boiler furnaces were constructed of tubes set against a refractory/tile enclosure, as shown in Figure 5a. This design is no longer used because of problems associated with leakage of gases to the casing, corrosion, and the massive weight associated with the refractory. The tangent tube design (Figure 5b) is an improved construction, but it still lacks the integrity and popularity of the membrane wall design (Figure 5c),

which acts as a gas-tight enclosure for the flue gases and also minimizes problems associated with thermal expansion and movement of the various parts of the furnace. In the membrane wall design, the entire furnace operates at a uniform temperature, so differential expansion concerns are absent. On lowpressure units, 2-in. O.D. tubes at a 4-in. pitch are generally used for the furnace construction, while in highpressure units (those exceeding 800 psig), a lesser pitch is used to minimize the fin tip temperature. Calculations have to be performed by the boiler designer to ensure that the fin tip temperature does not exceed the value suggested for the fin material under consideration. The choice of tube size and pitch varies among boiler manufacturers. Recent trends In order to meet emission criteria, parameters such as furnace geometry, excess air, and heat release rates must be commensurate with the recommendations of the burner supplier. In addition, depending on the method adopted for NOx control, the furnace and boiler design will be impacted. Some of the widely used methods of NO x control in packaged boilers are described below. Flue gas recirculation. In flue gas recirculation (FGR), a certain amount of flue gas is drawn from the boiler exit and introduced into the flame region to reduce the flame temperature. Recirculating flue gas may be induced into the forced-draft fan suction, at the furnace close to the flame via a separate recirculation fan, or by the forced-draft fan itself. NOx formation is basically of two types - thermal NOx, which is a result of the combustion process and the resulting combustion temperature, and fuel NO.,, which is the conversion of the fuel-bound nitrogen to NO x . A fuel that is rich in hydrogen, for example, can lead to a higher flame temperature and hence

increased thermal NO x. FGR reduces the combustion temperature and the availability of oxygen in the flame, which result in lower thermal NOx . However, FGR has less impact on fuel NOx . The amount of flue gas recirculation may range from 5% to 20%, depending on the extent of NO x reduction required. This results in a larger boiler or higher gas pressure drop through the boiler for the same duty, compared to a boiler without FGR, and therefore higher operating costs. An additional gas pressure drop of 1 in. w.c. in the boiler is nearly equivalent to 4 kW more of fan power consumption in a 100,000lb/h boiler; at 7¢/kwh, this is equivalent to $2,240/yr. Staged combustion. Staged air burners utilize off-stochiometric combustion under controlled conditions. Air is introduced into the furnace in two stages - the primary (60-80% of the air), and the secondary. The short age of oxygen generates high partial pressures of carbon monoxide and hydrogen. These reducing gases limit the NOx formation and also reduce any NOx that is formed to molecular nitrogen. The secondary air is introduced at a different location to complete the combustion after some energy has been transferred by the flame, thereby reducing thermal NO x formation. Furnace geometry must ensure that proper mixing occurs so that carbon monoxide and soot formation are minimized. Staging of fuel accomplishes a similar objective. A portion of the fuel and all of the air are mixed in the primary combustion zone; rapid combustion is achieved in a highexcessair atmosphere, resulting in a lower flame temperature. Additional fuel is introduced downstream at a region where the presence of flue gases results in a lower flame temperature, leading to thermal NO, reductions of as much as 50 %. Up to 30-60% reduction in NOx can be attained by the above methods alone or in combination.

Use of catalysts. When NOx lev els of 6-9 ppmv (dry) are specified, as in some California regions, selective catalytic reduction (SCR) is used,

CHEMICAL ENGINEERING PROGRESS

SEPTEMBER 1993

•4

although it is very expensive. In this method, ammonia is injected into the flue gas upstream of a catalyst reactor and the reactants pass through the catalyst blocks, which are of some proprietary precious metal construction. Depending on the catalyst, a gas temperature range of 550°F to 750°F is recommended at the SCR to ensure that the reactions for NOx reduction are favorable and complete. The use of SCR affects the boiler design in a few ways. In order to accommodate the SCR, which operates efficiently only within a narrow band of gas temperatures, methods such as gas bypassing via dampers must be used to ensure the proper gas temperature range to the SCR at all boiler loads. This gas bypassing may be done externally or internally to the boiler. The SCR also adds to the gas pressure drop through the system by 2-4 in. w.c. A convenient location for the SCR is between the evaporator and economizer.

problems and maintenance con cerns associated with refractory are also absent, because the entire unit expands and contracts uniformly. Another factor that leads to increased CO formation is the leakage of the products of combustion from the furnace to the convection section. This can occur if the partition walls are not of the membrane wall design. Tangent tube construction for the partition can lead to leak age of the combustion products because the differential pressure between the furnace and convection exit can be in the range of 16-30 in. w.c. in pressurized furnace designs. A membrane wall construction elimi nates the leakage concerns.

HRR may be specified on a lower or higher heating value basis. Typical HRR values range from 60,000 to 180,000 Btu/ft 2 •h depending on boiler capacity (smaller capacity units gener ally have the lower heat release rates). The heat release rate affects the furnace absorption and the furnace exit gas temperature. Energy absorbed in the furnace may be computed by evaluating the net energy released in the furnace and subtract ing the energy at the furnace exit, which varies with the furnace exit gas temperature. It may also be not iced that the furnace exit gas temperature is higher for natural gas compared to oil.

Furnace design and emissions All of the above methods of NOx control have significant impact on boiler design, particularly on the furnace and convection sections. Gas recirculation also affects the gas temperature profile through the boiler, which in turn affects the steam tem perature if a superheater is used. Furnace construction has a signifi cant impact on emissions. A completely water-cooled design (with front, rear, and side walls cooled) ensures adequete cooling of the flame as soon as it is formed. The absence of refractory on the front wall results in a benign environment for the flame, because the refractory can cause reradiation to the flame. The completely watercooled membrane wall furnace (Figure 6) also results in a lower heat release rate per unit of area for the same furnace volume, since the front and rear walls are effectively cooled. Thermal expan sion 5 • SEPTEMBER 1993 •

Furnace heat release rates A distinction must be made between the heat release rate and the heat flux. The heat release rate (HRR) is the amount of energy released by the fuel and air divided by the effective projected area of the furnace. Effective projected area is the flat projected area of the furnace walls (tubes, fins, and the membranes between the tubes) including the opening to the convection section. One should be careful when comparing bids, because some manufacturers use the circumferential area of the tubes plus the flat projected area of the fins, thereby resulting in a larger surface area.

CHEMICAL ENGINEERING PROGRESS

Typically the furnace absorbs 35% to 50% of the total energy. The average heat flux is the amount of energy absorbed in the furnace divided by the effective project ed area. This is an important variable, particularly in high-pressure boilers. Allowable heat fluxes range from 130,000 to 200,000 Btu/ft 2 •h, depending on steam pressure, tube size and orientation, mass velocity inside the tubes, and circulation ratio. If the maximum local heat flux exceeds the allowable heat flux, a condition called "departure from nucleate boiling" (DNB) results, leading to overheating and tube failures (1,2). This is quite rare, as pack aged steam generators

operate at a low heat flux and steam pressures are low compared to large utility boilers. Customizing designs. Packaged steam generators have come a long way since they were introduced a century ago. For instance: Furnace dimensions are set in consultation with burner suppliers, whose modeling of NO, and CO emissions may not favor a given geometry for different emissions. Borrowing the concepts from gas turbine heat-recovery steam generators (HRSGs), packaged steam generators have been made compact by utilizing extended surfaces in the convection section. Lowtemperature superheaters have been designed with finned tubes. The use of extended surfaces reduces gas pressure drop, thus reducing operating costs. The location of the superheater may also be optimized to match the steam temperature requirements, particularly in units where a large portion of the saturated steam is extracted for process use and the balance is superheated. Operating costs could be reduced by designing the convection section with appropriate tube spacings rather than a standard value. Dimensions such as boiler width, depth, or height may be changed to fit a given space or shipping limitation. Finally, the completely watercooled membrane wall design results in less maintenance, as well as the other advantages discussed earlier regarding NO, formation and heat release rates. Specifying steam generators In order to properly specify a suit able packaged steam generator, the engineer should provide, as a minimum, the following information: 1. Steam parameters, such as flow rate, pressure, and temperature, as well as the feed water temperature. If a portion of the saturated steam is taken off

for deaeration or process use, this should be stated, as only the balance flows through the superheater. Specifications should state whether the net output from the boiler should include the deaeration steam or not. 2. Steam temperature control, if required. It is very difficult to control steam temperatures over the entire range from 10% to 100% load, which is sometimes specified by engineers not familiar with boiler design. One can imagine that even if a superheater is designed for, say, 75-100 psi pressure drop, at 10% load the flow through the tubes cannot be properly distributed at 1/100th the pressure drop, which would be less than 1 psi. The same concern occurs outside the tubes, where the gas velocity is only 1/10th and heat transfer will not be in the forced convection regime. A more realistic spread for control could be 60% to 100% load. Also, it does not make sense to control the steam temperature when the degree of superheat is only 30°F to 50°F. A more sensible way of specifying the design is to allow the steam temperatures to float. Many engineers do not even ask the question, "What will happen if the steam temperature drops by, say, 40°F at lower loads?" One has to understand the implications on downstream equipment before calling out unrealistic specifications on the steam temperature control range. 3. Emission requirements, if any. These should be stated early in the design process and not after the boiler is ordered or built, because, as discussed earlier, the impact on boiler size or performance is significant. 4. Feed water analysis. Such information helps determine the blowdown requirements and whether the feed water can be used as spray for a steam attemperator. Any solids present in the water can be deposited within the superheater tubes, causing problems for the superheater. The final steam will also have a low purity. 5. Fuel analysis - that is, the composition of the fuel gas, such as methane, ethane, propane, hydrogen, and so on.

These data help in analyzing the boiler performance and combustion process and the impact on NOx and CO emissions. The presence of higher hydrocarbons increases the combustion temperature and hence NO,, emissions. The presence of nitrogen in fuel oil also contributes to high er NOx. A low-Btu gas, for example, generates a larger amount of flue gas for the same duty than a high-Btu gas and, therefore, results in higher pres sure drop through the system. 6. Space restrictions, if any. The boiler configuration may be changed to accommodate a horizontal or vertical gas flow economizer if space is a concern. Furnace dimensions and tube pitch may also be modified to fit the boiler within the given space. To receive a free copy of this article, send in the Reader Inquiry card in this issue with No. 149 circled.

CHEMICAL ENGINEERING PROGRESS • SEPTEMBER 1993 • 6

FOR HEAT TRANSFER by gases flowing through tubes at high pressure, the effect of pressure on specific heat, vis cosity and thermal conductivity cannot be neglected. Since this situation exists for many commercial heat exchangers, graphs are presented to id entify the pressure effect.

The basic equation for convective heat transfer is the same as for Part 1 of this series: Nu = 0.023 Re 0 . 8 Pr 0 . 4

(1)

Equation 1 is expanded to give the following: h = 2.44 (W0.8/d1.8)k 0.6 Cp0.4 /µ0.4

To get heat transfer coefficients

(2)

Where Cp = fluid specific heat, Btu/ (lb) °F d = tube inside diameter, inches Nu,Re,Pr are Nusselt ,Reynolds and Prandtl numbers h = heat transfer coeff., Btu/ (hr) (sq ft) (°F) k = thermal conductivity, Btu/ (hr) (sq ft) (°F/ft) W = mass flo w rate per tube, lb/hr = G (area) µ = fluid viscosity, lb/(ft) (hr) = 2.42 (centipoise).

Fluid properties shown in Equation 2 are combined into a single term as follows: C = k 0.6 Cp 0.4/µ0.4

(3)

Then Equation 2 becomes h = 2.44 (W0.8/d1.8) (C)

Pressure effect on gases

V. Ganapathy, Bharat Heavy Electricals Ltd., Tiruchirapalli, India

About the author is a senior development engineer with Bharat Heavy Electricals Ltd., Tiruchirapalla, India. He holds a B. Tech. degree from Indian Institute of Technology Madras and a M. Sc. (Engg) degree in boiler technology from Madras University. His work includes studies pertaining to optimization of heat exchangers, waste heat boilers and related equipment. V. GANAPATHY

Circle 180 on Reader Service Card

(4

Pressure effects on the various thermodynamic and transport properties were determined with a computer program using Gambill's method for variations of specific heat and using the Stiel and Thodos method for thermal conductivity and viscosity.1 , 2 The program can be used for individual gases as well as for mixtures of gases at various pressure and temperature. The data from the program for many individual gases were checked with available data 3 , 4 and variations were found to be insig nificant. A subroutine was developed to calculate the value of C as defined by Equation 3. The results for some common gases are shown in the accompanying figures. The effect of pressure on the value of C is more predominant for gases at lower temperatures. At higher temperatures, the pressure effects usually can be neglected. Heat transfer coefficients for these gases are found by substitution of C from these figures into Equation 4 or by entering the C scale on one of the nomographs pre sented in Part 1 (October 1977). LITERATURE CITED

' Reed, Robert C. and Sherwood,. Thomas K., The

Properties liquids, 1st Ed., McGraw-Hill, 1958, Chapt. 7 and 8.

of

gases and

z Holland, F. A., et al, Heat transfer, Am. Elsevier Pub. Co., 1970, pp. 324 340. s "Thermodynamic and transport properties of gases, liquids and solids," papers presented at the symposium on thermal properties, ASME, McGrawHill. Standards of Tubular Exchanger Manufacturers Assoc. (TEMA).

30

TO GET HEAT TRANSFER COEFFICIENTS

304

November 1977

HYDROCARBON PROCESSING

HYDROCARBON PROCESSING

November 1977 305

TO GET HEAT TRANSFER COEFFICIENTS

306

C i r c l e 1 8 1 o n R e a d e r S e r v i c e C a r d -*

To get heat transfer coefficients

HEAT TRANSFER coefficients for fluids flowing over tube bundles are calculated with a slightly different equation than the one used in Parts 1 and 2. For a staggered arrangement of tubes, the equation is as follows:

NU = 0.33Re 0.6 Pr0.33

(1)

Expanding Equation 1 to include the terms making up the dimensionless numbers, defining the tube diameter in inches and rearranging gives the following: h = 0.9(G0.6 /d 0.4 ) (k0.67Cp 0.33/µ0.27)

(2)

where Cp= fluid specific heat, Btu/ (lb) (°F) d = tube outside diameter, inches G = mass velocity, lb/ (hr) (sq ft) h = heat transfer coeff., Btu/ (hr) (sq ft) (F) k =thermal conductivity, Btu/ (hr) (sq ft) (°F/ft)

Pressure effect on crossflow

V. Ganapathy, Bharat Heavy Electricals Ltd., Tiruchirapalli, India

µ = fluid viscosity, lb/ (f t) (h r) = 2.42 (centipoise) . Fluid properties shown in Equation 2 are combined into a single term as follows: F =(k0.67Cp 0.33/µ0.27) (3) where the value for F for some common gases can be obtained directly from the accompanying figures. Take note that the powers for the physical properties in F are different from those used for C in Parts 1 and 2. The figures for F are the results of a computer program in which pressure effect on specific heat is determined using Gambill's method and the effects on thermal conductivity and viscosity are by the method of Stiel and Thodos. The figures cover a wide range of pressures; namely, 1 to 250 atmospheres. Hence they can be used for the more common cases at atmospheric pressure as well as for the higher pressures encountered in waste heat recovery applications. For in-line arrangement of tubes, the value obtained for heat transfer coefficient by the equations herein should be multiplied by 0.8. For a baffled heat exchanger, an additional multiplier of 0.6 should be used. BIBLIOGRAPHY ' McAdams, William H., Heat Transmission, 3rd Ed., McGraw-Hill, 1954. 2 Rohsenow, Warren and Hartnett, J. P., Handbook of Heat Transfer, McGrawHill 1972. s Kern, Donald Process Heat Transfer McGraw-Hill, 1972. ' Perry, Robert. and Chilton, Cecil H., Chemical Engineers' Handbook, 5th Ed., McGraw-Hill, 1973. s "Steam, its generation and use," Babcock & Wilcox, USA. ' Reed

Robert C. and Sherwood, T homas K., The Properties of Gases and

Liquids, 2nd Ed., McGraw-Hill, 1966. 7

Lord, R. C., et al, "Design of heat exchangers," Chemical Engineering, Vol. 77, No. 2, Jan. 26, 1970, pp. 96-118. s Ganapathy V., Quick estimation of gas heat-transfer coefficients," Chem-

TO GET HEAT TRANSFER COEFFICIENTS

106

December 1977 HYDROCARBON PROCESSING

HYDROCARBON PROCESSING

December 1977 107

TO GET HEAT TRANSFER COEFFICIENTS

108

Circle 150 on Reader Service Card -

HEAT TRANSFER

Simplify heat recovery steam generator evaluation Insights, equations and examples illustrate a simpler method for predicting heat recovery steam generator V Ganapathy, ABCO Industries, Abilene, Texas HEAT RECOVERY STEAM generators (HRSGs ) are widely used in process and power plants, refineries and in several cogeneration/combined cycle systems (Fig. 1). They are usually designed for a set of gas and steam conditions but often operate under different parameters due to plant constraints, steam demand, different ambient conditions (which affect the gas flow and exhaust gas temperature in a gas turbine plant), etc. As a result, the gas and steam temperature profiles in the HRSG, steam production and the steam temperature differ from the design conditions, affecting the entire plant performance and economics. Also, consultants and process engineers who are involved in evaluating the performance of the steam system as a whole, often would like to simulate the performance of an HRSG under different gas flows, inlet gas temperature and analysis, steam pressure and feed water temperature to optimize the entire steam system and select proper auxiliaries such as steam turbines, condensers, deaerators, etc. HRSG suppliers can provide this information, but if a simpler approach is made available to every engineer involved with HRSGs, it would be a powerful tool for plant engineers, consultants and cogeneration system engineers, who would only like to simulate HRSG performance. Usually they do not want to get involved in the thermal and mechanical design of HRSGs, which is best done by the supplier of the HRSG. This article describes a simplified approach to predicting the performance of HRSGs. Soft ware is available that can save valuable time and drudgery, as performance evaluation of HRSGs is a tedious process. 1,2 Advantages of this approach. Configuration of the HRSG or its geometrical details such as the width, height, tube size, pitch or fin density, number of rows, etc., need not be known. Based on known or assumed pinch and approach points (see Fig. 2 for definitions), a "design" is simulated, the gas/steam temperature profiles are determined and the steam flow is obtained. Then, using the procedure described below, the "performance" at any other gas or steam parameters can be obtained using quick converging iterative logic. The procedure may be used for both unfired and fired HRSGs. Heat transfer coefficients or surface area need not be computed per se. The product of U (overall heat transfer coefficient) and S (surface area) is computed for each surface

based on the duty and log-mean temperature difference, corrected for gas flow, analysis and temperature, and used in the performance calculations. Since US can be obtained for any HRSG surface for a particular set of design or operating conditions, this information is in everybody's domain. Importance of pinch and approach points. To obtain the performance of an HRSG with a different set of gas/ steam conditions, one should have either t he design condi Hydrocarbon Processing, March 1990

TABLE 1 -Data for "design" and "performance" calculations 1 . 2 3. . 4.

Case no. Gas flow, pph Exhaust temp., °F % vol C02 H2 O N2 02 5. Steam press., psig 6. Steam temp., °F 7. Feed water temp., °F 8. Blow down, % 9. Process steam, pph 10. Heat loss + margin, 11. SH press.%drop, psi 12. Pinch point, °F 13. Approach point, °F 14. Steam flow, pph 15. Ambient temp., °F

Desig 1

150,00 0 900 3 7 75 15 450 650 240 2 1 7 20 10 ? 80

Perf

Perf

Perf

2 3 4 165,000 165,00 165,000 840 840 840 0 3 3 3 7 7 7 75 75 75 15 15 15 450 450 300 ? ? 650 240 240 240 2 2 2 2,500 1 1 1 ? ? ? ? ? ? ? ? ? ? 26,000 26,000 50 50 50

Natural gas used: % vol C1= 96, C2 =2,C3=2 C, Note that steam is required at a controlled temperature of 650°F in case 4. In cases 2 and 3 it is uncontrolled. Also, in case 4, 2,500 pph of saturated steam is taken off the drum and the balance of 26,000 pph is to be superheated to 650°F. The steam exit pressure is 300 psig in case 4. It will be seen later that cases 3 and 4 are fired and cases 1 and 2 are unfired.

TABLE 2a-Gas properties-unfired case (% vol C0 2 = 3, H20 = 7,N2=75,O 2=15 µ Temp., °F Cp 900 650 400

0.2736 0.2658 0.2584

0.083 0.0724 0.0612

k

0.0304 0.0261 0.0218

TABLE 2b -Gas properties-fired case (% vol Temp., 1,050 700 350

CO2=3.45,H2O=7.87,N2=74.65,O2=14.01 N = 74.65, k µ Cp 0.2800 0.0887 0.0330 0.2689 0.0743 0.0267 0.2583 0.0586 0.0208

Units: Cp-Btu/Ib°F, µ - Ib/ft h, k-Btu/fth°F (Interpolate for gas properties at intermediate temperatures) tions or the HRSG performance under a set of gas/steam parameters. HRSGs for gas turbine exhaust are usually designed in unfired conditions and the performance evaluated at other unfired or fired conditions. The reason for this is that two of the important variables which affect the gas and steam temperature profiles, namely the pinch and approach points, cannot be arbitrarily selected in the fired mode. The follow ing problems can result if this is done: 1. A realistic temperature profile may not result. That is, a temperature cross can occur in the economizer, with the gas exit temperature being lower than the incoming feed water temperature. ; 2. An unrealistic boiler size can result due to very small pinch or approach points especially if that is the basis for selecting the temperature profile in the fired mode. Pinch and approach points up to 10°F can be achieved with practical boiler configurations in unfired modes. Unless one has a great deal of experience in designing HRSGs, pinch and approach points should be selected in unfired modes and the HRSG performance evaluated in fired modes. 3. Steaming in the economizer is likely if one selects an 78 Hydrocarbon Processing, March 1990

HRSG temperature profile in the fired mode and operates the boiler in unfired mode.',' The performance of' the HRSG has to be checked in the unfired coldest ambient conditions, when the gas flow is the highest and the gas inlet temperature to the HRSG is the lowest to make sure the economizer does not steam. If it does, the original temperature profile selected has to be revised, resulting in a waste of time and effort. 4. The amount of spray water used for steam temperature control cannot be simulated a priori in the fired mode. This is particularly important in HRSGs where the steam temperature has to be controlled over a wide load. The steam temperature will fall off as the gas inlet temperature reduces and as a result, if the steam temperature is chosen (say 700°F) in the fired mode, it will be lower in the unfired mode. Several performance checks would have to be made to ensure that the steam temperature is being achieved over the desired load range. On the other hand, if the steam temperature is selected in the unfired mode, it will increase with an increase in gas inlet temperature and hence, can be controlled with spray water or other means. Hence, it is prudent to arrive at a design temperature profile in gas turbine HRSGs based on cold ambient, unfired conditions and then check the HRSG performance in other unfired or fired conditions, even if the HRSG operates in the fired mode most of the time. Pinch and approach points lie in the range of 10 to 40°F in unfired conditions for clean applications such as gas turbine exhaust, where extended surfaces can be used. Higher numbers may be selected if the steam generation can be lower. In the case of HRSGs for applications such as incineration exhaust or chemical plants where the gas inlet temperature could be in the range of 1,400 to 1,800°F, and in HRSGs where extended surfaces cannot be used due to dirty gas, a higher pinch point in the range of 100 to 250°F should be used.

Design temperature profile and calculations. A superheater and economizer are assumed to be in counterflow arrangement, which is the widely used configuration.

Example 1. A gas turbine HRSG is to be designed for the parameters shown in Table 1. Determine the gas/steam profiles and the steam flow. Let the gas pressure drop = 6.0 in. WC. The drum pressure = 450 + 7 = 457 psig. The saturation temperature is 460°E Gas temperature leaving the evaporator = 460 + 20 = 480°E Compute the gas properties for the given analysis.' The data are shown in Table 2. Using an instantaneous specific heat of 0.267 for the range 900 to 480°F, and a heat loss factor of 0.99, the duty in the superheater and evaporator is: Q1 + Q2 = 150,000 (0.267) (0.99) (900 - 480) = 16.65 x 106 Btu/h = Wsd[(1,330.8 - 431.2) + 0.02(442.3 - 431.2)] = 899.8 Wsd. Where 1,330.8 = enthalpy of superheated steam at 450 psig, 650°F, 442.3 = enthalpy of saturated water at drum pressure, 431.2 = enthalpy of water entering the evaporator at 450°E 0.02 is the blow down factor. From the above, Wsd = 18,510 pph. Superheater duty, Q1 = 18,510(1,330.8 - 1,204.4) = 2.34 x 106 Btu/h, where 1,204.4 is the enthalpy of 'saturated steam. Gas temperature drop in the superheater = 2.34 x 106/(150,000) (0.273) (0.99) = 58°E Hence, gas temperature to evaporator = 900 - 58 = 842°F. Q2 = Evaporator duty = 16.65 - 2.34 = 14.31 x 106 Btu/h. Economizer duty = 18,510 (1.02) (431.2 - 209.6) = 4.19 x 106 Btu/h, where 209.6 is the enthalpy of feed water at 240°F.

Gas temperature drop in the economizer = 4.19 x 106 / (150,000) (0.99) (0.26) = 109°F. The gas specific heat at the average gas temperature in the economizer, obtained from Table 2 by interpolation is 0.26. Hence, the exit gas temperature = 480 - 109 = 371°F. The temperature profile is shown in Fig. 2. Using a similar approach, the temperature profiles for any other pinch or approach points can be obtained. To proceed with the performance calculations for case 2 shown in Table 1, a few parameters should first be computed, as discussed in Appendix 2. These parameters help relate the heat transfer coefficients in the "design" mode to those in "performance." For the superheater: K1 = Q1/(∆T1 ) (Wg0.65 ) (Fg) where ∆T1 = log-mean temperature difference = [(842 - 460) (900 - 650)]/ln[(842 - 460)/(900 - 650)] = 311°F, Fg = Cp 033 k0.67 /µ0.32 = 0.135, using a Cp = 0.273, k = 0.029, and µ = 0.0826. Hence K1 = 2,340,000/150,0000.65/311/ 0.135 = 24.10. Similarly for the evaporator K2 = 387.6 and K3 = 218.4 for the economizer. K1, K2 and K3 will be used to compute (US)p the product of U and S in the performance modes as discussed in Appendix 2. Performance calcbulations. Let us see how the unit per-

forms when the conditions are as shown in case 2, Table 1. The gas flow is 165,000 pph at 840°E The gas analysis, feed water temperature and steam pressure remain the same as earlier. The performance of the HRSG is arrived at through an iterative process described in Appendix 1, using the equations discussed in Appendix 2. Trial 1. As a first approximation, assume that the steam flow is proportional to the gas flow and temperature drop. Ws = 18,510(165,000/150,000) (840 - 371)/(900 - 371) = 18,050 pph. Superheater performance. Let ts 2 , the steam exit temperature = 640° F. Then, from steam tables, the enthalpy = 1,325 Btu/lb. The assumed duty = 18,050 (1,325 - 1,204.4) = 2.177 x 106 Btu/h. Gas temperature drop = 2,177,000/(165,000) (0.99) (0.271) = 49°F. Hence, gas temperature leaving the superheater = 840 - 49 = 791°F. Compute the transferred duty, Q1t, using Eq. 2 in Appendix 2. Fg = 0.135, Wg = 165,000, K, = 24.1, Wsd = 18,510, Ws = 18,050. Hence (US)p = 165,0000.65(0.135) (24.1) (18,050/18,510)°'5 = 7,974. ∆T = log-mean tempera turedifference = [(840 - 640) - (791 - 460)]/ln[(840 - 640)/ (791 - 460)] = 260°F. Hence, Q1t = 7,974(260) = 2,074,000 Btu/h. This is close to the assumed value. If it were not, we would have to assume another steam temperature and repeat the steps. Let us continue.Evaporator performance. Compute Fg at the average gas temperature in the evaporator. Fg = 0.129, K2 = 387.6. Then, (US)p = 165,0000.65(0.129) (387.6) = 123,123. Using Eq.8,[(791460)/(Tg3 -460)]= e(123,123/165,000/0.99/0.266) = 17,00. Hence Tg3 = 480°F; Q2 = 165,000(0.99) (0.266) (791 - 480) = 13.522 x 106 Btu/h. Note that the gas properties have to be interpolated for the values at the average gas temperature in the section. Economizer performance. Let the water temperature leaving the economizer be 450° F. hW2 = 431.2 from steam tables. Assumed duty Q3a = 1.02(18,050) (431.2 - 209.6) = 4.08 x 106 Btu/h. The gas temperature drop = 4,080,000/ 165,000/0.99/0.26 = 96°F, exit gas temperature = 480 - 96 = 384°F. Fg = 0.130, K3 = 218.4. Hence (US)p = 218.4

(165,000065 ) (0.120) = 64,535. Transferred duty = Q3t = 64,535(72.7) = 4.69 x 106 Btu/h, where 72.7 is the log mean temperature difference. Since the transferred duty is more than the assumed, let us repeat the calculations with say tw2 = 457° F.Q3a = 18,050( (1.02) (439 209.6) = 423,000 Btu/h. The exit gas temperature = 381. ∆T= 65. Then, Q3t = 64,535(65) = 4,190,00( Btu/h. Since this is closer to Q3t let us continue. The total transferred duty = Q1t + Q2t + Q3t = 2.07 + 13.52 + 4.19= 19.78 MMBtu/h. The corrected steam flow, Ws = 19.78x 106 /[1,325 - 209.6 + 0.02(442 - 209.6)] = 17,660 pph. per Eq. 14, Appendix 2. Since this is not close to the assumec value of 18,050 pph, another trial is warranted. Try Ws = 17,770 pph. Trial 2. Let the revised steam flow = 17,700 pph. Follow a similar procedure as before. Superheater performance. Let ts 2 = 640° F . Q1a = 17,700( (1,325 - 1,204.4) = 2.134 MMBtu/h. Gas temperature drop = 2,134,000/(165,000) (0.99) (0.271) = 48°F. Tg2 = 840 - 48 = 7920 F. ∆T = 260°F. Fg = 0.135. K, = 24.1. Then, (US)p = 165,000°65 (0.135) (24.1)(17,700/18,510)0.15 =7,957. Q1t = 7,957(260) = 2.07 MMBtu/h. Since Q1 , is less than Q1a, try a lower steam temperature, say 635° F. Then Q1a = 17,700(1,322 - 1,204.4) = 2.081 MMBtu/h. Gas temperature drop = 47°F. Tg2 = 840 - 47 = 793°F. ∆T = 264°F. Hence, Q1t = 7,957(264) = 2.1 MMBtu/h. This is close enough. Continue. Evaporator performance. Solve for Tg3 as before. [(793 460)/(Tg3 - 460)] = 17.00; hence Tg3 = 480°F. Q2 = 165,000 (0.99) (0.266) (793 - 480) = 13.6 MMBtu/h. (The factor 17 computed from Trial 1 is unchanged.) Economizer performance. Let tw2 = 455; hW2 = 436.8; Q3a = 17,700(1.02) (436.8 - 209.6) = 4.1 MMBtu/h. Gas temperature drop = 96°F. Tg4 = 480 - 96 = 384°F. ∆T = 68°F. Using the same (US)p as before, Q3t = 64,535(68) = 4.36 MMBtu/h. Since the variation between Q3a and Q3t is large, try tw2 =458° F. Then, Q3a = 4.14 MMBtu/h. Tg4 = 383°F. ∆T = 64.6°F. Hence, Q3a = 64.6(64,535) = 4.16 MMBtu/h. This is quite close. The total transferred duty = 2.1 + 13.6 + 4.16 = 19.86 MMBtu/h. The corrected steam flow, Ws = 19.86x106 /[(1,322 - 209.6) + 0.02(442 - 209.6)] = 17,770 pph.

Hydrocarbon Processing, March 1990 3

lier one. However, additional steps are necessary to iterate for the firing temperature, as discussed in Appendix 2. The method of computing the fuel input, firing temperature and gas analysis is discussed elsewhere. 6 Let us only check the final results which are shown in Fig. 4. Superheater performance. Table 2b shows the gas properties for the gas analysis after combustion. From the printout, Fig. 4, it is seen that the HRSG gas inlet temperature is 1,034°F and the burner fuel input is 9.29 MMBtu/h (LHV basis). Wg = 165,430; Ws = 26,000; ts 2 = 677°F. Fg at the average gas temperature is 0.142. The saturation temperature is 462° F, at the corrected drum pressure of 463 psig. Q3a = 26,000(1,346.0 - 1,204.3) = 3.69 MMBtu/h. Gas temperature drop= 3,690,000/(165,430) (0.99) (0.278) = 81°F. Exit gas temperature, Tg2 = 1,034 - 81 = 953°F; ∆T = 420°F. K1 = 24.1. (US)p = 165,430°65 (0.142) (24.1) (26,000/18,510)°15 = 8,840. Then, Q3t = 420(8,840)=3.71 MMBtu/h. Evaporator performance. Fg = 0.135; K2 = 387.6; hence (US)p = 165,430°65 (0.135) (387.6) = 129,437. Using Eq. 8, [(953 - 462)/(Tg3 - 463)] = e(129,437/165,430/0.99/0.27) = 18.67. Hence, Tg3 = 489°F. Q2 = 165,430(0.99) (0.27) (955 - 489) = 20.52 MMBtu/h.

Since this is close to the assumed value of 17,700, let us stop here. The final temperature profile is shown in Fig. 3. The gas pressure drop, using Eq. 15, Appendix 2 = 6(165,000/ 150,000)2 [0.5(840 + 383) + 460]/[0.5(900 + 371) + 460)] = 7.1 in. WC. Let us check the performance for case 3 shown in Table 1, where it is desired to make 26,000 pph of steam. The steam temperature is uncon trolled. It is obvious that with the same inlet gas conditions as in the earlier case, we need additional fuel input to the HRSG to generate 26,000 pph. The procedure is similar to the ear 4 Hydrocarbon Processing, March 1990 Performance check-fired case.

Economizer performance. Tw2 = 435; hw2 = 414.45; Q3a = 26,000(1.02) (414.45 - 209.6) = 5.43 MMBtu/h. Gas temperature drop = 128°F; Tg4 = 489 - 128 = 361°F. ∆T = 83°F. K3 = 218.4; Fg = 0.120; (US)p = 165,4300.65 (218.4) (0.120)=65,000. Hence Q3t = 83(65,000) = 5.4 MMBtu/h. Total energy transferred = 3.71 + 20.52 + 5.4 = 29.63 MMBtu/h. Ws=29.63 x 106 /[(1,346.7 - 209.6) + 0.02(442.6 - 209.6)] = 25,970 pph. The gas pressure drop could be corrected as before. This gives an idea of the complexity of performance calculations if fuel firing is involved. Several iterations of performance calculations would be required before the correct firing temperature is arrived at. Also, if the steam temperature has to be controlled, the superheater has to be split up into two stages with a spray desuperheater in between. The method of computing the spray water for steam temperature control is discussed elsewhere.' In such an HRSG, more iterations are involved before the spray water flow and the final temperature profiles are arrived at. Without a computer it would be extremely tedious and time consuming. Fig. 5 shows the results of case 4 where steam temperature control and fuel firing are involved. Note that gas inlet temperature is 1,067° F. The spray quantity has been arrived at based n a split in the ratio of 6:4 in design U times S values between the first and second stages of the superheater. This ratio is built into the program. Slight changes in the temperature profile and spray quantity can result due to a different split in the surfaces between the two stages of the superheater while actually building the HRSG. Also, note the higher steam pressure drop in the superheater due to the lower steam pressure. The economizer flow includes the 2,500 pph saturated steam taken off the drum. A note of caution on U, S and U times S values. Note that US values could be computed for each surface from its Q and ∆T data. For instance in the "design" case, for the superheater, US = 234,000/311 = 7,524. These would naturally change depending upon the gas flow, analysis and temperature profile. Hence, these values should be interpreted with caution.After arriving at the US values,some enginners

Appendix 1: Performance calculation procedure The procedure is discussed for a single pressure HRSG. Fig. 6 shows the various configurations of HRSGs consid ered. The first case is quite involved. The methodology for this case will be discussed. The gas flow, gas inlet temperature and analysis, steam pressure and feed water temperature are assumed to be known. The design calculations, which are the basis of establishing an initial design, are as sumed to be done and the results available, along with K 1, Kz, K 3 factors. 1. Assume the steam flow. A good estimate is obtained by using a ratio of the "performance" to "design" gas flows and temperature drop. 2. Solve the superheater performance. This is an iterative process. See Appendix 2, Eqs. 1 to 5. If the transferred and assumed duty are not equal, repeat with another steam temperature or else continue. 3. Solve the evaporator performance. Obtain the duty and exit gas temperature using Eqs. 6 to 9. 4. Solve economizer performance using Eqs. 10 to 13. This is again an iterative procedure. Calculate the total transferred duty. 5. The steam flow is then corrected based on the total transferred duty and enthalpy rise, Eq. 14. If this is close to the assumed steam flow in step 1, continue or else repeat steps 1 to 5.

6. If the final steam temperature is greater than that desired, the steam flow is corrected for the desired steam temperature. 7. If the desired steam flow is zero (unfired mode) or less than the corrected flow, proceed to step 11. 8. If the desired steam flow is larger than the corrected flow, calculate the fuel input required to raise the gas temperature to the required level to achieve the desired steam flow. This again involves several iterations, and for each firing temperature, all the steps from 1 to 8 have to be repeated until they match. 9. If the final steam temperature is higher than desired, calculate the interstage spray quantity based on a split superheater. 10. Another round of fine tuning is done to check the temperature profiles and steam flow. 11. It can be easily seen that a lot of iterative calculations are involved. For each round, the gas and steam prop erties have to be computed based on the gas analysis and temperature. If there is steaming in the economizer, the economizer is split up into two stages, a small evaporator and an economizer and calculations are done to evaluate the extent of steaming. It is obvious that without a com puter, the calculations can be overwhelming, particularly if there are several alternate performance conditions, and steam is generated at several pressure levels.

Appendix 2: Equations used in performance calculations Superheater performance. Assuming that the steam flow = Ws, from energy balance we have: Q1a = Ws(hs 2 - hs,) = Wg(Cp) (hf) (Tg1 - Tg 2) (1) where ts2 = exit steam temperature and hs 2, the enthalpy. Compute the exit gas temperature, Tg2, from the above. The transferred duty is then: Q 1 t = (US) p ∆T (2 a) ∆T = log-mean temperature difference ∆T = [(Tgl - ts 2 ) - ( T g2 - ts l )]/ln[(Tg, - ts2)/(Tg2-ts1)1

assuming counter flow configuration, which is widely (Tg2 -(US tSi)])p is the product of S and U in performance ( b) used. mode and is obtained from the (US) value in the design case by adjusting as follows for the gas properties and flow. (US) p = Wgo.65 FgK 1 (WS/Wsd)o.15 (3) K, is obtained from Q1, ∆T, Wg and Fg values in design case: 0.65 ) (Fg)) (4) K 1 = Q1/(∆T(Wg 0.32

k /(µ ) (5) If for the assumed steam temperature Q1a and Q1, do not come close (say within 0.5%), another iteration is warranted. All of the above steps are repeated until Q 1a and Q1, match. Evaporator performance. From energy balance,Q2 = Wg(Cp) (hf) (Tg2 - Tg3) = (US) p ∆T (6) where ∆T= [(Tg2 - ts) - ( T g3 - ts)]/ln[(Tg2 - ts)/ (Tg3 - ts)] = (Tg2 – Tg3)/ln[(Tg2 - ts)I(Tg3 - ts)] (7) Fg

= (Cp

o.33 o.67

From Eqs. 6 and 7 after simplification, we have: [(Tg2 - ts)/(Tg3 - ts)] = e[(US)p/(Wg)(cp)(hlf)] (8) try to split up the U and S values and compare alternate designs based on S values alone. This can lead to very misleading conclusions and the author strongly recommends against it, particularly if extended surfaces are used. With finned tubes, the gas side heat transfer coefficient and fin efficiency are affected by variables such as fin den sity, height, thickness and fin or tube material .6, ',s By using

where:

(US)p = Wgo.65FgK 2

(9)

K2 is computed as in Eq. 4 from the design conditions. Fg is computed for the performance conditions. Tg3 is solved from Eq. 8 without iteration. Q 2, the duty, can be obtained from Eq. 6. Economizer performance. Assume tw2, the water exit temperature. Then, Q3a = Ws(hw2 – hw1)(1+bd)=W gCp(Tg3 -T g4 )hf (10) Obtain Tg4 and then the ∆T, assuming counter flow conditions ∆T = [(Tg4 – tw 1) - (Tg3 - tw2)]/ln

[(Tg4 - tw I )/(Tg3 - tw2)] Transferred duty: transferred duty Q3t = (U S)p ∆T where 0.65 (US) p = Wg

.Ks

(11) (12) (13)

K 3 is obtained as in Eq. 4 from design conditions. If Q3a

and Q3 , are close, continue or else the iteration continues from Eqs. 10 to 1 3 with a different tw2. The steam flow is then corrected as follows: W S , =(Qlt+ Q 2 t+ Q 3t )/[(hs 2 - hw,) + bd(hf – hw1)] (14) If Ws, is not close to the assumed flow, Ws, the calculations are repeated starting with the superheater. The gas pres sure drop is corrected for performance conditions: ∆P = (∆P)d (Wgl Wgd) 2 [(Tavg + 460)1(Tavgd + 460)] (15) tubes with high fin density, say six, one could show more surface in the HRSG, but due to the lower U associated with it, it does not mean that the energy transferred is more com pared to a design which has a lower fin density, say two to four, and hence, lower S. Lower fin density should be used whenever possible to increase U and minimize gas pressure drop and fin and tube wall temperatures. This is more im Hydrocarbon Processing, March 1990

5

HRSGs fall under the category discussed in Fig. 6 , and hence the methodology discussed can be applied to a wide variety of HRSGs used in the industry. While the method of predicting performance using U values based on actual tube geometry, fin configuration, etc., gives accurate results, this methodology has been checked against several designs and operating results. For the pur poses of engineering analysis, trend projections, evaluation of alternate designs and for studying the effect of different gas/steam parameters on performance, this approach is very effective and hence a powerful tool. Considering the complexity of the calculations and iterative nature of the procedure, particularly if multipressure HRSGs are involved, a program has been developed by the author for HRSG design and performance evaluation. For more information on the software and its availability, contact the author at PO. Box 673, Abilene, Texas 7 9 6 0 4 , U S A . NOMENCLATURE bd-Blow down fraction; if blow down = 2%, then bd = 0.02 Cp-Gas specific heat, Btu/1b °F Fg-A factor accounting for gas properties, de fined in Eq. 5 hlf--Heat loss factor; if heat loss = 2%, then h l f = 0.98 hs2, hs1-Enthalpy of superheated steam and inlet steam, Btu/lb hw t, hw 2 -Enthalpy of water at eco inlet and exit, Btu/ lb k-Gas thermal con ductivity, Btu/ft h°F

portant in surfaces with low tube side heat transfer coefficients such as superheaters. One could show that S can be 100 to 200% more by using six fins/in. compared to two, but due to the higher U, the duty can be the same or even more. The author has performed studies on optimization of finned tubes', 7 and advises engineers against comparing and select ing HRSGs simply because the surface area, S, is more com pared to another design which uses lower fin density. Unless the engineer is familiar with all aspects of heat transfer with extended surfaces and the impact of each variable on U, comparisons of S alone can be misleading and should be avoided.

K1, K 2, K3-Factors obtained from design conditions, Eq. 4 Q1, Q2, Q3 -Energy absorbed in superheater, evaporator and economizer, Btu/h; subscript a = as sumed and t = transferred Tgl, T g2 , T g3 , T g4-Gas temperature distribution, S-Surface area, sq ft ∆T-Log-mean temperature difference, °F Tavg, Tavgd-Average gas temperature in HRSG in per formance and design modes Tw 1 , tw2-Water temperature at inlet and exit of economizer, F ts l, ts2 -Saturated and superheated steam temperature, °F U-Overall heat transfer coefficient, Btu/sq ft hF (US) p -Product of U and S in performance mode Wg, Wgd-Gas flow in performance and design modes, pph Ws, Wsd-Steam flow in performance and design mode

(∆P)d,p -Gas pressure drop in design and performance, in. WC µ-Gas viscosity, lb/ft h

Note: "Design" case is the basis used to arrive at initial temperature profiles, steam flow, and the design. "Performance" case predicts the performance of the HRSG so designed at different gas or steam parameters.

Limitations and software. The approach discussed has a limitation. It cannot be used in HRSGs which have a radiant section. However, the author is of the view that 80 to 9 0 % of

The author

V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute of Technology, Madras, India, and an MSc(eng) in boiler tech nology from Madras University. Mr. Ganapathy is the author of over 150 articles on boilers, heat transfer and steam plant systems and has written four books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization and Basic Programs for Steam Plant Engineers (book and diskette), copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Design, Vols. 25 & 26, Marcel Dekker, New York.

6

Hydrocarbon Processing, March 1990

LITERATURE CITED

1.Ganapathy, V., Applied heat transfer, Pennwell Books, Tulsa, 1982. 2 Ganapathy, V., "HRSG features and applications," Heating, piping and air -condi tioning, Jan. '89. s Ganapathy, V., et al., "Heat recovery boilers for process and cogeneration applications," Seventh Industrial Energy Technology Conference, Houston, May '85. 4 Ganapathy, V., "HRSG temperature profiles guide energy recovery," Power, Sept. '88. s Ganapathy, V., "HRSGs for gas turbine applications," Hydrocarbon Processing, An g. '87. b Ganapathy, V., "Charts simplify spiral finned tube calculations," Chemical Engi neering, April 25, 1977, p. 117. 7 Ganapathy, V., "Charts help evaluate finned tube alternatives," Oil and Gas ,Journal, Dec. 3, 1979, p. 74. e Ganapathy, V., Nomograms for steam generation and utilization, Von Nostrand Reinhold, 1988, p. 77. 9 Ganapathy, V., "Program computes fuel input, combustion

temperature," Power Engineering, July '86. '0 Ganapathy, V., "Determine spray water to desuperheat steam," Healing, piping and air-conditioning, Dec. '87.

Understand Steam Generator Performance The key performance variables are excess air, fuel type, exit gas temperature, load, and emissions.

S

everal variables affect plant engineers plan their opera tion better. This article discusses the effects of such variables as excess air, fuel type, exit gas temperature, load, and emissions on generator design and operation. It also discusses some of the potential benefits of customized steam generators over standard, prepackaged designs, which often compromise on overall performance. The focus of the article is limited to gaseous and oil fuels.

BOILER EFFICIENCY V. Ganapathy, ABCO Industries

42 •

DECEMBER 1994

Fuel composition

the

• Figure 1. Large packaged steam generator.

The single most important variable from a performance standpoint is the steam generator efficiency. This is particularly true for base-load steam generators that will operate most of the time (unlike a standby boiler, which operates for only a few hours per year). During the design stage, the consultant or the end user specifies a certain efficien cy. Efficiency is primarily affected by the fuel composition, unburned carbon losses, excess air, exit gas temperature, and the type of fuel (Table 1). CHEMICAL ENGINEERING PROGRESS

The fuel composition is important, as it affects the flue gas composition, which in turn affects the various heat losses. While variations may not be significant between typical natural gases, differences between a low-Btu and a high-Btu gas do matter.

increase the fuel moisture loss. Similarly, the percentages of hydrogen and carbon in oil fuels affect the fuel moisture loss and hence the efficiency, as shown in Table 1.

Unburned carbon losses The various boiler heat losses are evaluated at the design stage using the American Society of Mechanical Engineers' Power Test Code heat-loss method, ASME PTC 4.1. One of the losses impacted by the com bustion process is the unburned carbon

loss. Carbon in the fuel is converted to carbon monoxide instead of carbon dioxide, which results in lower carbon utilization. With gaseous fuels this may be insignificant. However, with fuel oils, the amount of CO formed can be very high - on the order of several hundred ppm. The type of burner used, the amount of excess air, turbulence in the combustion zone, and the type of furnace construction (that is, whether it is a membrane wall or tangent tube design) influence this loss. Leakage of combustion products from the furnace to the convection section via a tangent tube partition wall contributes a great deal to CO formation, because the combustion products do not have the residence time in the furnace to complete combustion before entering the convection section. A highexcessair operation may be required to minimize this loss; however, this decreases the efficiency due to increased heat losses, as shown in Table 1. The loss L (in Btu/lb) due to CO formation is given by: L = 10,160 C [CO/(CO + CO2)] where CO and CO 2 are the volume percentages in the flue gas and C is the weight fraction of carbon in the fuel.

Excess air Excess air affects efficiency significantly, as indicated in Table l. The choice of how much excess air to use depends on the type of fuel used and the desired levels of NOx and CO emissions, as well as the degree of flue gas recirculation (FGR). Burner suppliers often recommend the amount of excess air after reviewing the emission levels to be guaranteed, the fuel analysis, and the furnace dimensions. A high excess air (on the order of 10-15%) is often suggested even for natural gas fuels. This is because FGR is used to limit NO, which in turn affects the burnout of CO; higher excess air helps to complete combustion. Figure 2 shows the typical relationship between excess air and emissions. CHEMICAL ENGINEERING PROGRESS • DECEMBER 1994 •

43

For each fuel there is range of excess air that achieves the desired CO and NOx levels. Higher molecular weight hydrocarbons have a higher flame temperature, which produces higher NOx, which in turn requires a higher degree of FGR to limit it, which in turn may require higher excess air, depending on the CO levels to be guaranteed. The experience of burner suppliers with units burning similar fuels and having similar emissions often determines this parameter. Exit gas temperature

The lower the exit gas temperature, the higher the boiler efficiency. A rule of thumb is that every 40°F difference is equivalent to a 1 % change in efficiency. However, if the temperature of the feed water entering the economizer is higher, then the stack temperature will also be higher. Otherwise, a very large economizer may be required to maintain the same exit gas temperature compared to a low feed water temperature case. One factor influencing the exit gas temperature is the sulfuric acid dew point. When fuels containing sulfur are fired, SO2 is formed, and some of it (1-5%) is converted to SO3, particularly if vanadium is present in the oil ash, which acts as a catalyst. When the acid vapor gets cooled by the feed water, condensation may occur if the temperature of the tube surface is at or below the acid dew point. Dew point is a function of the partial pressure of the acid vapor and water vapor in the flue gases. Table 2 shows a few correlations for acid vapors (1, 4, 5). There is a misconception, even among experienced engineers, that condensation in the economizer can be avoided by maintaining the exit gas temperature high enough. When an economizer is used as the heat recovery equipment, the cold end temperature is mainly a function of the feed water temperature entering the boiler and the gas temperature has little effect on it. This is due to the high tube-side heat-transfer coefficient. Table 3 shows a simple calculation where the exit gas temperature varies by as much as

44 • DECEMBER 1994 • CHEMICAL ENGINEERING PROGRESS

400°F, while the economizer tube wall temperature varies by only a few degrees (/, 2, 6). However, it is not necessary to have the feed water temperature at or above the acid dew point to minimize corrosion problems. While this is a sure way to prevent acid condensation, research has shown that corrosion is significant at 50-100°F below the acid dew point, as illustrated in Figure 3 (/, 6). Hence, one need not specify a high feed water temperature, which results in a lower efficiency. If the acid dew point is, say, 275°F, a 230-250°F feed water temperature is a good compromise. An exit gas temperature of 300 -320°F can be achieved with moderate costs. Air heaters are often avoided as backend heat -recovery equipment. This is because they contribute to high er combustion temperature and hence NOx, which calls for higher FGR rates, which in turn results in a higher gas pressure drop through the boiler or a larger boiler or both. Also, the gas pressure drop through an air heater is much higher, by 2 -3 in. w.c., which is a continuous loss of energy. When an air heater is used, it is usu ally one of two types - the tubular or recuperative, or the regenerative or rotary air heater. Rotary air heaters have the additional problem of leakage from the air to the gas side, which affects the fan size and air heater performan ce. An advantage of air heaters, though, is that they may be more compact than tubular air heaters. Air heaters are primarily recommended when difficult fuels, such as solid fuels and low-Btu gaseous fuels, are fired. Flue gas quantity Using higher excess air and FGR rates for the same boiler duty increases the flue gas quantity to be handled by the boiler. This naturally results in a larger boiler or, if dimensions are limited because of shipping restrictions (which is often the case), a higher gas pressu re drop is incurred in the convection section and economizer. Table 4 shows the flue gas quantity produced with different excess air and FGR factors.

This is one of the reasons why it is often beneficial to go with a custom designed steam generator - tube spacings, tube height, and the number of tube sections can be varied to minimize gas pressure drop. Unfortunately, packaged boilers are often treated as predesigned or off -theshelf items. Some consultants even suggest model numbers while developing specifications. This practice should be avoided. Otherwise you could be purchasing a design that was developed several decades ago, when emis sions were not a concern and FGR was not considered while sizing the convection section and economizer. These steam generators were designed with skimpy furnaces and low excess air factors and without any FGR. Hence, even if the steam parameters may be the same, the flue gas flow through the unit

can be nearly 30 -40% more, resulting in higher gas pressure drops, higher exit gas temperatures, and therefore lower efficiency and much higher operating costs. Engineers should understand these aspects and opt for custom designed units, which can incorporate several design features to minimize operating costs and improve efficiency. As an example, if 130,000 lb/h of flue gases have to be forced through an additional 3 in. w.c. in the boiler because a standard rather than custom design is used, the additional fan power consumption is 14.5 kW, based on a 70°F air temperature and a fan efficiency of 70%. Assuming that the boiler operates for 8,000 h/yr and that electricity costs 1 ¢/kWh, the additional cost is $11,600/yr. Capitalizing this cost over the lifetime of the boiler CHEMICAL ENGINEERING PROGRESS • DECEMBER 1994 • 45

ENERGY TRANSFER/CONVERSION

shows that when one considers both the operating and the initial costs, it pays to select a custom designed unit, since the additional capital costs for a custom-designed system is generally only about $20,000 to $30,000, and in some cases is virtually nil.

at lower loads, they absorb more energy due to radiation and thus have higher steam temperatures. The spraying of feed water into the steam for steam temperature control can increase the solids content of the

L o w- l o a d o p e r a t i o n There are also a few issues of concern at low loads, particularly with superheater and fan operation. The number of streams (the area through which steam flows in a super

SUPERHEATER DESIGN It is very important to know where the steam generated by the boiler is used. Steam turbines require a high steam purity, which calls for good drum internals such as chevrons and cyclones. If saturated steam is generated, simple mist eliminators may be adequate. Another aspect to be determined is the range of load over which the steam temperature has to be maintained. A wide load range calls for a large and, therefore, expensive superheater. Consultants must discuss with clients and turbine suppliers before specifying this requirement. While 70-100% load range of superheat temperature control is common, some designers unknowingly specify steam temperature control from 40% to 100% load, which complicates the superheater design. The superheater is the equipment most significantly affected by parameters such as excess air, flue gas recirculation, and furnace sizing. A larger furnace results in a lower exit gas temperature, which increases the superheater size due to the lower log mean temperature difference available. A high FGR rate increases the size of the superheater and the gas pressure drop. Typically, the steam temperature is maintained at 70% to 100% load. With convective type superheaters (Figure 4), this means that the steam temperature will be higher at higher loads. Interstage attemperation should be incorporated to control the steam temperature at higher loads. Radiant superheaters, which are located in the furnace zone or exposed to direct flame radiation, generally operate at higher tube wall tempera tures. Thus, unless these units are very carefully designed, failures are more likely. Radiant superheaters behave differently than convective superheaters -

46 -

• Figure 4. Superheater for steam generator. steam. Thus, the feed water should have the same amount of solids as the final steam - in the ppb range. Demineralized water is preferred for such applications. If demineralized water is not available, saturated steam may be condensed in a heat exchanger and sprayed into steam, as shown in Figure 5a and 5b.

DECEMBER 1994 CHEMICAL ENGINEERING PROGRESS

heater) has to be chosen so that a good distribution occurs even when the boiler operates at the lowest load. Some consultants specify turndown of 1:8 or even 1:10. From a practical point of view, a turndown this high is not recommended. The real problem is in the ability of the superheater and fan to handle such

low-load conditions. If the steam pres sure drop is, say, 50 psi at 100% load, it will be 3 psi at 25% load. At lower loads, it is difficult to ensure that the flow distribution will be uniform through all the tubes. One has to be also concerned about reverse flows, which can result in overheating of some tubes and possibly failure. The author recommends a load range of 50% to 100%, not 10% to 100%, since performance is difficult to predict at low gas and steam velocities. Another problem with low-load operation is the fan performance. If the fan is selected with high margins on flow and head, then at low loads, the operating point may fall below the capacity of the fan even at the lowest vane opening position (Figure 6). This may cause problems with fan operation, such as vibrations, instability, and noise. This is likely in packaged boilers, which typically have one forced draft fan. If two fans each having half the capacity are used instead, then a higher turndown is feasible. Also, at low flows, the fan operating point can drift into the unstable operating regimes of the fan curve. Using a high margin for the fan flow and head should be avoided. The author suggests 10% on flow and 20% on head.

LOAD VS. PERFORMANCE Figure 7 shows the performance of a boiler at different loads. The efficiency peaks at a certain load and then drops off. This is expected, as the nature of the two important losses, namely radiation and flue gas heat losses , differ. At higher loads, the radiation loss will be lower and the heat losses due to the flue gases will be higher; the opposite is true at lower loads. The combination of these losses results in a peak efficiency, at some load between 0% and 100%. The exit gas temperature drops off with load. An economizer acts as a heat sink, which limits the gas temperature so that gas is not cooled to dew point levels.

The approach point, or the feed water temperature leaving the economizer, decreases when load decreases (unlike in a gas turbine heat -recovery steam generator, or HRSG). Hence, steaming is not a problem at low loads. CUSTOMIZED DESIGNS As mentioned earlier, adopting a standard design developed decades ago to present -day operating conditions with high excess air and FGR

rates results in a compromise on performance and operating costs. Customized designs can overcome these concerns. Following are some of the aspects reviewed in customized designs. 1. Furnace design, which considers the excess air and F GR rates and flame dimensions for each fuel, after discussions with the burner supplier so that the flame is contained entirely within the furnace.

CHEMICAL ENGINEERING PROGRESS • DECEMBER 1994

• 47

2. Convection section design, which can have longer tubes, wider tube spacing, and more tube sections to reduce gas velocity and hence pres sure drop to acceptable levels.

3. The possible use of extended surfaces in the convection section if clean fuels are fired. Extended surfaces can result in compact designs, lower gas pressure drop, and lower

exit gas temperatures from the convection section (1). 4. Superheater location and whether it is in the appropriate gas temperature zone in the convection section to operate safely over a wide load range. If the steam temperature is low enough, the superheater could even be located downstream of the convection section and ahead of the economizer. 5. Horizontal or vertical economizers to match the layout requirements. To receive a free copy of this article, send in the Reader Inquiry card in this issue with the No. 180 circled.

48 •

DECEMBER 1994 PROGRESS

CHEMICAL ENGINEERING

Superheaters: design and performance Understand these factors to improve operation

V Ganapathy, ABCO Industries, Abilene, Texas

Steam

superheaters are widely used in steam generators and heat-recovery steam generators (HRSGs). Their purpose is to raise steam temperature from saturation conditions to the desired final temperature, which can be as high as 1,000°F in some cases. When used in steam turbines, superheated steam decreases steam heat rate of the turbine and thus improves the turbine and overall plant power output and efficiency. Also, steam conditions at the steam turbine exit will have little or no moisture, depending on the pressure ratio; moisture in the last few stages of a steam turbine can damage the turbine blades. This article outlines some of the design considerations and performance aspects of superheaters, which should be of interest to plant engineers.

Superheaters in packaged steam generators and HRSGs- general features. Packaged steam

generators generate up to 300,000 lb/h steam, while a few gas turbine HRSGs generate even more depending on the gas turbine size. Steam pressure in cogeneration and combined cycle plants typically ranges from 150 to 1,500 psig and temperature from saturation to 1,000°F Seamless alloy steel tubes are used in superheater construction. Tube sizes vary from 1.25 to 2.5 in. Commonly used materials are shown in Table 1. Allowable stress values depend on actual tube wall temperatures. Tube thickness is determined based on this using formulae discussed in the ASME Code, Sections 1 and 8. Different designs are available for superheaters depending on gas/steam parameters and space availability. The inverted loop design (Fig. 1) is widely used in packaged boilers, while the vertical finned tube design is common in HRSGs. The horizontal tube design with vertical headers is used in both. Bare tubes are generally used in packaged steam generators, where gas temperatures are high (typically 1,500-2,200°F) and tube wall temperature is a concern. However, in gas turbine HRSGs, finned superheaters are used. Gas inlet temperature is generally low, on the order of 900-1,400'F, which requires a large surface area. Use of finned tubes makes their design compact. Superheaters can be of convective or radiant design or a combination of these in packaged boilers. Final steam temperature may or may not be controlled. In unfired and supplementary fired HRSGs, the superheaters are or convective design only.

Steam velocity inside superheater tubes ranges from 50 to 140 fps depending on steam pressure, allowable pressure drop and turndown in load. Typical pressure drop in industrial applications ranges from 10 to 70 psi depending on size, pressure and load turndown conditions. In utility boilers, where multiple stage superheaters are used, pressure drop will be much higher, say 150-200 psi. If the superheater has to operate over a wide load range, a higher steam pressure drop at full load ensures reasonable flow at lower loads.

Convective and radiant superheaters in packaged boilers. Fig. 2 shows typical location of superheaters in a packaged boiler. Superheaters are basixally

HYDROCARBON

PROCESSING

/JULY

2001 4

cally classified as radiant or convective, depending on their location. The convective superheater is shielded away from the furnace and the flame, while the radiant design is located at the furnace exit, facing the flame and direct radiation from the furnace. While the radiant design requires less heating surface (due to the higher LMTD and direct radiant energy contribution), it has several drawbacks compared to the convective design. • A convective superheater is located behind a screen section, which helps to cool gases from the furnace and also ensures that a uniform gas mixture enters the superheater. This permits the designer to predict superheater performance with much higher reliability and accuracy. The furnace is a difficult section to evaluate due to the complexity of the combustion process. Adding to the difficulty is use of varying excess air and flue gas recirculation rates (used for NOX control) for different fuels at different loads, which in turn affects flame temperature and its temperature distribution along the flame. Only models based on experience of similar units could be considered reasonable since no simple mathematical formulae can accurately predict the furnace energy balance. Hence actual furnace exit gas temperature can easily be off from predicted values by 50-150°F, which affects the radiant superheater performance significantly. If act ual gas temperature is higher than predicted, we have tube overheating problems and if it is lower, we may not obtain the desired steam temperature. • The radiant superheater is located in a region where the flue gases make a turn and, hence, the gas flow distribution pattern over the tubes is difficult to predict at various loads. • The radiant superheater receives direct radiation from the furnace, which in turn depends on exit gas temperature. Again, if we do not predict exit gas temperature accurately, radiant heat flux will be off from estimates,

which affects steam and tube wall temperatures. Margin of error on gas temperature estimation is much higher at furnace exit compared to the convective superheater inlet, which is located behind a screen section. • Steam temperature characteristics are different between convective and radiant designs. The radiant design absorbs more energy at lower loads, while the convective design absorbs more at higher loads (Fig 3). This is due to the increase in convective heat transfer rates. In large field-erected industrial or utility boilers, the superheater is in multistages. The combination of radiant and convective designs helps ensure a uniform steam temperature over a wide load range. However, when only a single-stage superheater is usedas in packaged boilers-the radiant design is subject to higher steam temperatures at low loads, when flow distribution on both gas and steam sides is poor. If at 100% load the steam pressure drop is, say, 30 psi, at 25% load it will be hardly 2 psi, which cannot ensure good steam flow distribution in various elements. The same is true of flow distribution on the gas side at low loads due to the low gas velocity. Thus, this fact, along with lack of steam temperature control methods, can result in overheating and possible tube failure. Creep analysis using Larson-Miller parameter methods may be done to estimate life at various tube wall temperatures and remaining superheater life, based on number of hours of operation at each load. The convective design, on the other hand, is located in a much cooler gas temperature zone-1,600-1,800°F, compared to 2,200-2,400°F for the radiant design. Hence, it runs much cooler and tube wall temperatures are also more accurately predictable. At lower loads, gas temperature to the superheater will be lower as well as the heat transfer rate and tube wall temperature. Thus, convective superheater life is longer than the radiant design. • Steam turbines usually require a constant steam temperature. Lower steam temperatures affect the heat rate; however, this occurs at a lower load with convective designs and, hence, loss in output is not significant. Oversizing convective superheaters may also be done to ensure that desired steam temperature is achieved over a wider load range if necessary-say from about 30% to 100%. • With convective designs, it is possible to have twostage designs with interstage attemperation. With radiant designs in packaged boilers, single stages are generally used, which causes concerns with steam temperature fluctuations and tube overheating. 42

HYDROCARBON PROCESSING ' JULY 2001

Steam temperature control methods in superheaters. Generally, steam temperature is maintained

constant from about 60% to 100% load. Interstage attemperation or spray water injection (Fig. 4) is done to achieve the desired final steam temperature. Water injected should be demineralized since solids contained in feed water can get carried into the superheater and turbine and selective deposition can occur. Salt deposits in the superheater can result in tube overheating. Turbine blade deposition is a big concern with turbine maintenance engineers since it reduces power output, restricts flow passages, causes corrosion and can damage the blades. Hence, high steam purity on the order of 20-50 ppb is generally desired in high steam temperature applications. Good steam drum internals using a combination of baffles and Chevron separators can achieve the desired steam purity. In case demineralized water is not available for spray, some of the steam may be condensed using a heat exchanger as shown in Fig. 4, and the condensate is sprayed into the desuperheater. Steam flow through the exchanger and superheater should be balanced in the parallel paths either by using flow restrictions, control valves in each parallel path or by raising the exchanger level to provide additional head for control. Feed water from the economizer cools and condenses steam used for desuperheating (Fig. 4a). In Fig. 4b, the feed water is directly injected into the steam between the stages. Desuperheating beyond the superheater is not recommended since moisture can be carried to the steam turbine along with the steam if downstream mixing is not good. Also, this method permits steam temperature in the superheater to increase beyond the desired final steam temperature and, hence, the premium on materials used for superheater construction will be high. There are several other methods used for steam temperature control such as varying excess air, tilting burners, recirculating flue gases, etc., but in packaged boilers and HRSGs, interstage attemperation is generally used. The basic difference in superheater design used in steam generators and HRSGs is that in HRSGs, as mentioned earlier, finned tubes may be used to make the design compact. The large duty and large gas-to-steam flow ratio coupled with the low LMTD necessitates this. However, while selecting finned tubes, a low fin density should be used conside ring the low steam side heat transfer coefficient inside the tubes. The heat transfer coefficient due to superheated steam flow is small, on the order of 150300 Btu/ft2h°F, depending on steam flow, pressure, temperature and tube size. A large fin area would only increase heat flux inside the tubes, tube wall temperature and possibly gas pressure drop as discussed in an earlier article.' Note that the gas side heat transfer coefficient is lower with higher fin density or surface area. Hence, it is misleading to evaluate finned superheater designs based on surface areas.' In large gas turbines, steam after expanding from the steam turbine is again reheated in the HRSG to generate additional power. Design considerations

Superheaters in HRSGs.

of the low-pressure superheater, also known as a reheater, are similar to those discussed previously for superheaters. There are two basic types of calculations with any heat transfer surface. One is the design calculation, in which the objective is to arrive at the surface area, tube layout, steam and gas-side pressure drops, and preliminary material selection. The off-design performance calculation tells us how the same surface will perform at other gas or steam conditions. There is only one design calculation, but performance could be checked at different loads or offdesign conditions. A computer program is generally used for these purposes since the calculations are quite involved and iterative. Only after these two types of calculations are completed can we say that the engineering process is over. Here's the energy balance equation for a superheater. Total energy absorbed by the superheater is:

Sizing procedure.

Qs = Ws(hs2 - hsl) = Qc + Qn + Qr (1)

Relating this to the gas temperature drop:

(Qs - Qr) = Qn + Qc = US∆T = Wg(hgl - hg2)

(2)

External radiation, Qr, is generally absent in convective type designs. Also, if a screen section is used, Qr gets absorbed in about four to six rows of screen tubes depending on tube spacing. The fraction of energy, F, absorbed in each row is given by: F=

3.14(d / 2St) - (d / St) [sin - 1(d / St) +

√{(St /d)2 –1} - (St 1d)]

(3)

If S/d = 4, then F = 0.361. The first row absorbs HYDROCARBON PROCESSING /JULY 2001

3

The procedure for determining these are well documented. 2.3 Gas side heat transfer coefficient, hc, for finned and bare tubes may be obtained from published charts or equations 1.2.3 as also the procedure for determining hn. A simplified approach to estimating hi is:

h i = 2.33w

0.361 of the external radiation. The second row absorbs: (1- 0.361)0.361= 0.23 and so on. If (S / d) were smaller, fewer rows would be required to absorb the direct radiation. Qr is estimated from gas emissivity and furnace exit gas temperature. Overall heat transfer coefficient is obtained from: 1 / U = ( A t / A i) / h i + ffi ( A t / A i ) + ffo + ( A t / A w ) d / 24Km ln(d/ di) + 1 /ηh o

(4)

for finned tubes. For bare tubes, the same equation is used, however, A t / A i = d / d i and fin effectiveness η=1 =1

Fouling factor, ffo, is typically 0.001 ft 2hF/Btu in clean gas applications, while ffi ranges from 0.0005 to 0.001. The gas side heat transfer coefficient, ho =

hn + hc.

0.8 c

/ di 1.8

(5)

(

where factor C is given in Table 2. Once the duty Qs are known and resultant gas and steam temperatures at the superheater inlet and exit, then the LMTD may be estimated. Knowing the various gas and steam side heat transfer coefficients, fouling factors and tubes sizes, overall heat transfer coefficient, U, is obtained. Surface area, S, is determined as shown previously from Eq. 2. Then the tubes are laid out and the gas/steam side pressure drops are evaluated. Several factors such as tube size, number of streams carrying the steam flow, tube spacing, gas mass velocity, etc., are selected based on experience. Gas side pressure drop may be found from equations discussed in citations 2 and 3. Tube side pressure drop is give by:

∆P = 3.36 x 10-6 fw2 Le v /di 5

(6)

If superheaters are of such a design (say the inverted loop design in Fig. 2) that tube length in various elements is different, a flow balance calculation has to be performed to determine steam flow in each element. The tube with the lowest flow is likely to have the highest tube wall temperature. If there are, say, four elements, the following equations help evaluate flow in each. Pressure drop across the headers is the same across each element and from the previous equation for pressure drop: w1 2 RI = w22 R2 = w3 2 R3 = w4 2 R4 = M(constant) R = resistance of each element = fLe/di5

Thus, we first determine resistance of each element with different tube lengths, R1, R2, R3, R4. Then we can solve for flow through each element: w 1 + w 2 + w 3 + w 4 = known as the total steam flow is known or √MR, + √ MR2 +√ MR3 +√`MR4 = total flow or M can be obtained from the above since all resistances are known. Then wl, w2, w3 and w4 can be obtained. The tube wall temperature calculations are done. In simple terms, heat flux is first estimated: Q = U ( t g - t s ) ] Temperature drop across the steam film is:

heat flux / hi.

Temperature drop across the fouling layer inside is: heat flux x ffi. Then drop across the tube wall is determined by multiplying tube wall resistance, (d/24K)ln(d/di), by heat flux. All these are added to the steam temperature to arrive at the tube wall temperature. This calculation may be done at different tube locations using local tg, ts, hi and U values. Off-design performance. To arrive at the offdesign performance, one can resort to the NTU method. 2,3 In this calculation, surface area is known and gas flow, steam flow and their inlet temperatures are known. It is desired to predict the duty and exit gas and steam temperatures. This method is discussed in various textbooks 2,3 and will not be explained here. Based on actual gas/steam flow conditions, an estimation of U is done. Using the NTU method, superheater duty V Ganapathy is a heat transfer specialist at ABCO Industries, Abilene, Texas, a subsidiary of Peerless Manufacturing, Dallas. He has a bachelors degree in mechanical engineering from I.I.T., Madras, India, and a masters degree from Madras University. At ABCO, Mr. Ganapathy is responsible for steam generator, HRSG and waste heat boiler process and thermal engineering functions, and has 30 years of experience in this field. He has authored over 250 arti cles on boilers and related subjects, written four books and contributed several chapters to the Handbook of Engineering Calculations and the Encyclopedia of Chemical Processing and Design. He can be reached via e-mail. [email protected].

can be found. Tube wall temperatures at various locations are again evaluated. Based on both design and off-design conditions, a final material selection is made. One may revise the design if off-design performance is not up to expectations. NOMENCLATURE Af, Ai, At = area of fins, inside tube area and total tube surface area per unit length, ft 2/ft C = constant for determining tube side coefficient d, di = tube outer and inner diameters, in. F = fraction of direct radiation absorbed f = friction factor inside tubes ffo, ffi = fouling factors outside and inside tubes, fth°F/Btu hg l, hg 2 = gas enthalpy at inlet and exit of superheater, Btu/lb hc, hn, ho = convective, nonluminous and outside heat transfer coefficients, Btu/ft 2hoF hsl, hs2 = steam enthalpy at superheater inlet and exit, Btu/lb K= tube thermal conductivity, Btu/fth°F Le = tube effective length, ft M = a constant Qc, Qn, Qr, Qs = energy due to convection, nonluminous heat transfer, direct radiation and that absorbed by steam, Btu/h S = surface area, ft 2 Tg, is = local gas and steam temperatures, °F Wg, WS = gas and steam flow, lb/h w = steam flow per tube, lb/h ∆T = log-mean temperature difference, °F ∆P = pressure drop inside tubes, p si v = steam specific volume, ft 3/lb η = fin effectiveness 1

LITERATURE CITED Ganapathy, V, "Evaluate extended surface exchangers carefully,"

Hydrocarbon Processing, October 1990.

Ganapathy, V, Steam. Plant Calculations Manual, Marcel Dekker, New York, 1994. s Ganapathy, V, Waste Heat Boiler Deskbook, Fairmont Press, Atlanta, 1991. 2

How Important is Surface Area .? "

Fire-tube It's important, but it should not be the only criterion you use to size and specify boilers and safety valves. Consider the factors outlined here as well.

V. Ganapathy, ABCO Industries

and water-tube boilers are widely used in the chemical process industries (CPI), for example, to recover energy from flue gas streams and to generate steam in gas- or oil-fired packaged steam generators. One of the main criteria that engineers use to specify or eval uate boilers is surface area. Packaged firetube boilers, for instance, are often specified as requiring 5 ft= per boiler horsepower (one boiler horsepower is equivalent to 34,500 Btu/h of output). However, surface area is a misleading variable because heat transfer depends on other factors as well, including gas velocity, the size of the tubes, the tube pitch and arrangement, the configuration of tube fins, fouling factors, and others. For the same duty or energy transferred, one can develop different designs with significant differences in surface areas, and the various designs can have widely different costs. This article outlines how to size and specify boilers other than by simply stating surface area. In addition, it discusses the selection of safety valves, which is still done based on surface area, and describes a more practical approach. FIRE-TUBE BOILERS

In fire-tube boilers (Figure 1), flue gas flows inside the tubes while the steam is generated outside the tubes. Depending on the cleanliness of the gas, tube sizes can vary from 1.5 to 3.5 in. O.D. If slagging is a concern, as in municipal solid waste incineration applications, the boiler should be of a multipass design, where the first pass is a pipe with a diameter ranging from 30 to 48 in. and subsequent passes consist of smaller diameter tubes. Packaged oil- or gas-fired boilers have a similar configuration. The OCTOBER 1992

CHEMICAL ENGINEERING PROGRESS

specified gas velocities can vary, depending on the allowable gas pressure drop. Both of these factors - tube size and gas velocity - influence the heat -transfer coefficients and, hence, the surface area. Sizing procedure

The procedure for sizing a fire-tube boiler is as follows. The required surface area, S, is calculated from: ( 1a)

S = Q/(UAT)

If U is based on the tube outer diameter, then the surface area is also based on the tube outer diameter; likewise, if U is based on the tube inner diameter, then the surface area should be based on the tube inner diameter. This can also be expressed as Uo So = Ui Si , where So = πdo NL/12 and Si = = πdi NL/12. Thus, Eq. 1 a can be rewritten as either Si = Q/(Ui∆T)

(lb)

or So = Q/(Uo∆T)

(1c)

The energy transferred, Q, is: Q = Wg Cp,(T1-T2)h l=Ws Hs

(2)

The term h, represents the heat loss factor and is equal to one minus the losses due to radiation and convection from the boiler surfaces. A 2% loss, or h l = 0.98, is typical. The log mean temperature difference, ∆T, is determined by: ∆T= (T 1-t s)-(T2 -ts)/ln[(T 1-t s)(T 2 –t s)

(3)

The overall heat-transfer coefficient, Uo, is given by:

The tube-side heat-transfer coefficient, hi, is the sum of the convective heat-transfer coefficient, h,., and the nonluminous heattransfer coefficient, h n . The value of hn depends on the partial pressures of the tri atomic gases in the flue gas (e.g., CO2 , H2 O) and is usually small - on the order of 5% of h c in fire -tube boilers. Thus, many designers are conservative and neglect hn. (In watertube boilers, however, h n is very significant and cannot be neglected.) Further details on calculating hn can be found in (1). The value of h i is obtained from the DittusBoelter equation: Nu = 0.023Re0.8 Pr0.4 (5) where Nu = h c di /12k, Re = 15.2w/( µd i), and Pr = µCp/k. Substituting these expressions into

Eq. 5 and simplifying yields: h,. = 2.44w 0. 8F/di 1.8 (6) 0.4 0.6 where F = (Cp /µ) k The inside and outside fouling factors are denoted by ffi and ffo , respectively. For

CHEMICAL ENGINEERING PROGRESS

OCTOBER 1992

* 71

clean gases and boiler water, they can be assumed to be 0.001 ft2•h•°F/Btu. For gas streams that can cause fouling, ff can be much higher - on the order of 0.05 ft 2•h•°F/Btu. Tables of fouling factors are available in several published sources, such as (2) and (3). The boiling heat-transfer coeffi cient, ho is very high - on the order of 2,000 Btu/ft2h°F. Thus, even a 20% variation in its value will not impact U, because the tube-side coefficient, hi, which is typically on the order of 10-20 Btu/ft 2h°F, governs U. The last term in Eq. 4 is the resis tance of the tube wall to heat transfer. The thermal conductivity of the tube material, K, is about 20 -25 Btu/ ft•h•°F for carbon steel, the typical material used for boilers. To size the boiler, the mass flow per tube, ranging from 120 to 200 lb/h for a 2-in. tube, and the gas velocity, typically ranging ranging from 60 to 170 ft/s, are assumed and the tube count is calculated. The relationship between mass flow and velocity is:

72

OCTOBER 1992

V = 0.05wv/d i2 (7) While it may seem easier to assume a number of tubes than to assume a mass flow rate and gas velocity, in practice, because these calculations are done by computer the terms are essen tially conceived in parallel. Based on the temperature and properties of the gas, h c and then U are determined. Then Eq. 1 is used to calculate S, which is in turn used to determine the tube length, L. The gas pressure drop is then calculated based on geometry (1): ∆Pg = (93x10-6)fL e vw2/d i 5 (8) If the computed pressure drop is higher than that allowed by the specification, another mass flow rate per tube is assumed and the procedure is repeated. Example 1 Consider a fire-tube waste heat boiler required to cool 100,000 Ib/h of flue gas from 1,300°F to 474°F. The gas is at atmospheric pressure and consists of (by volume) 12% CO2 12% H 2O, 70% N2 and 6% O2. Feed

CHEMICAL ENGINEERING PROGRESS

water temperature is 220°F and steam pressure is 150 psig. Fouling factors of ffi = 0.002 ft2h°F/Btu for the gas and ff0 = 0.001 ft2h°F/Btu for the steam, heat losses of 2%, and an outside heat-transfer coefficient of h = 2,000 Btu/ft2hF are assumed. Tube sizes of 1.75 x 1.521, 2 x 1.773, and 2.5 x 2.238 (outer x inner diameter) will be considered. What are the effects on surface area requirements of tube size and gas velocity (which can range from 90 to 170 ft/s)? For simplicity, most of the calculation details are omitted. The results of the calculations for the various tube sizes and velocities are summarized in Table 1. For the same amount of' energy transferred, one can see significant variations in the surface area - by as much as 50%. As the gas velocity increases, U increases, which brings down the surface area,

and the gas pressure drop increases. Also, as the tube size increas es, U decreases for the same velocity. This, along with the fewer larger tubes, results in longer tube lengths. The main point to be noted is that for the same duty, the surface area can vary depending on the tube size and gas velocity. These conclusions also apply to packaged fire-tube boilers firing oil or gas. A rule of thumb that, unfortunately, is still being used by specifying engineers is 5 ft2 of surface per boiler horsepower. One can, by using a higher gas velocity or smaller tube size, develop a boiler design that will work fine with up to 10% to 20% less surface area. However, through lack of knowledge of heat-transfer design, several good designs are being overlooked by potential buyers, consultants, and end users. Boiler cost generally increases with an increase in surface area. However, it does not rise proportion ately because other items, such as boiler trim, controls, casing, insulation, and so on, account for a consid erable part of the total cost and these may not increase proportionately. Labor costs are significant and may not be proportional to surface area. Each case, therefore, must be reviewed independently. WATER-TUBE BOILERS

extended surfaces, often called finned tubes, may be used. The use of finned tubes makes the design very compact. Other advantages include lower weight and lower gas pressure drop. If the gas stream is dirty, as in municipal solid waste incineration systems, only bare tubes should be used. Design procedure

As with fire-tube boilers, the heat transfer duty, Q, is calculated by:

Q = Wg Cp(T1-T2)hl, = W s∆H s. = U S ∆T

(9)

U refers to the overall heat-transfer coefficient, and it is usually based on the outside surface area of the tube. Uo is given by:

where A, and Ai, which refer to the total external and internal surface area per foot of tube, are used instead of do and di. (In the case of bare tubes, At/Ai = d o / d i ) . Fin effectiveness is represented by n, which equals 1 in the case of bare tubes. In wat er-tube boilers, h o is the gas -side heattransfer coefficient, which is the sum of hc and h n ; h, is the tube-side boiling heat transfer coefficient, which is in the range of 2,000 -3,000 BtU/ft2hF. Bare tubes. The procedure for computing h o for bare tubes is as follows. Grimson's correlation for convective heat transfer is used for tubes in either an in -line or staggered arrangement (depicted in Figure 3):

In water-tube boilers (Figure 2), if the gas stream is clean (such as with gas turbine exhaust gases), tubes with

The gas properties are evaluated at

CHEMICAL ENGINEERING PROGRESS

0 OCTOBER 1992 73

the gas film temperature, and the coefficients B and N are obtained from Table 2. (The ratio of transverse pitch to outside diameter (S t/d o) and of lon gitudinal pitch to outside diameter (Sl/d o ) are computed. (S t /d o) is read across the top of the table, and Sll d o down the side, under "Staggered" or "In-line," as appropriate. The values of B and N are then read from the chart.) Gas mass velocity, G, is calculated by:

triatomic gases present, and the beam length, L b . b is given by:

where "surface" refers to the total external surface area touched by the gas. Hottels' charts (2) are used to determine the gas emissivity, s, from the above data. One can then calculate h from:

The gas pressure drop, ∆Pg , is then obtained from:

where f is the friction factor. For an inline arrangement, f is:

(18a) The nonluminous heat-transfer coefficient, h„, could be significant depending on the tube pitch, the partial pressures of water vapor and other OCTOBER 1992

Once ho is computed (here, too, ho = hc + hn), Eq. 10 can be used to calculate Ua. S is then obtained from:

CHEMICAL ENGINEERING PROGRESS

and for a staggered arrangement, f is:

Finned tubes. The correlations for heat transfer in finned tubes are more complex, so only the chart technique for computing hc will be discussed here, since the objective of this article is only to show the effect of a few variables on surface area and not the complete design procedure for finned tubes. [The interested reader can find further details in (1).[ Figure 4 can be used to estimate h,., the convective heattransfer coefficient, for in -line arranged tubes having a 2-in. O.D. and a 4-in. square pitch. The heat-transfer coefficient, fin effectiveness (11), and gas pressure drop are shown for 10 tube rows based on a gas turbine exhaust at an average temperature of 600°F. The real value of the chart, however, is not in estimating hc. Rather, it illustrates the effect of fin configuration (that is, fin density, n, and fin height, h) on hc . From this figure, one can see that: • as fin density increases, h decreases; • the higher the fin density, the higher the gas pressure drop will be, even after adjusting for the effect of the different number of rows required; and • fin effectiveness decreases with fin height. Hence, the simple conclusion that can be drawn from Figure 4 is that a higher fin density (or surface area per unit length) results in a lower hc and a lower Uo, which in turn means that more surface area is required. Let us now look at two specific examples to see how different fin configurations, and how the difference between bare and finned tubes, can significantly affect surface area. This example also illustrates the advantages of using extended surfaces, particularly in clean gas applications. A boiler evaporator needs to be designed for a gas turbine

exhaust. Gas data: Flow = 150,000 lb/h. Iinlet gas temperature = 1,000°F. Exit gas temperature = 382°F. Feed water temperature = 240°F. Inside and outside fouling factors = 0.001 ft2h°F/Btu. ho = 2,000 Btu/ft2•h•°F. The geometry is as follows: 18 tubes/row, tube O.D. = 2 in., I.D. = 1.77 in., length = 10 ft, in-line arrangement with transverse and longitudinal (square) pitch = 4 in., and material of construction is carbon steel. How do designs using bare tubes and serrated finned tubes (fin density = 5 fins/in., height = 0.75 in., thickness = 0.05 in.) compare? Again, the calculation details are omitted. [The complete procedure can be found in (1).] Here we will discuss the most

The con vective and nonluminous heat-transfer coefficients, hc and hn are computed for bare tubes using the procedure described above, and Uo is evaluated from Eq. 10. The outside tube area, So is then computed and is used, along with the assumed tube length (L) and number of tubes wide (Nw), to determine Nd, the required number of rows of tubes deep. Then Eq. 17 is used to calculate ∆P g. For finned tubes, the gas mass velocity is calculated and is used to obtain h . from Figure 4. The fin effectiveness, η and the gas pressure drop for 10 rows are also read from the chart. At and Aw can be obtained from standard reference charts or can be calculated based on the fin geometry (1). Beam length is calculated and used to determine h„. Then Eq. 10 is employed to compute U(), which is used to determine Nd and ∆P g. Table 3 summarizes the results and compares the bare-tube and finned tube cases. The advantages of using finned tubes are clear: The finned tube boiler is more compact (it has only 20 rows deep, vs. 122 rows of bare tubes), has a lower gas pressure drop (3.2 vs. 4.5 in the bare tube boiler), and weighs less (26,000 lb vs. 48,000 lb). On the other hand, the sur

CHEMICAL ENGINEERING PROGRESS * OCTOBER 1992

* 75

LITERATURE CITED 1. (:antipathy, V., "Waste Heat Boiler Deskbook," Fairmont Press, Atlanta (1991) (also available from the author). 2. Ganapathy, V, "Applied Heat Transfer," Pennwell Books, Tulsa, OK (1982). 3. Tabular Exchanger Manufacturers Association, "Standards of the Tubular Exchanger Manufacturers Association (TEMA), 7th ed., New York, NY (1988). 4. American Society of Mechanical Engineers, Sec. I, Rules for Construction of Power Boilers, 1989.

face area is much higher - nearly twice the area of the bare-tube boiler - because of the lower overall heattransfer coefficient.

Example 3 The choice between bare and finned tubes is not the only factor that affects surface area. One can also see significant variations in surface area for the same duty with finned tubes of different configurations. Consider a finned-tube superheater being designed for the following condi tions: Gas flow = 200,000 lb/h. Gas inlet temperature = 1,200°F. Gas analysis (vol. %) = 7% CO2 ,12% H2O, 75% N 2 and 6% O2. Steam flow= 100,000 Ib/h. Steam inlet temperature = 491°F (saturated) at 600 psig. Fouling factors for the gas and steam= 0.001 ft2-•h•°F/Btu. The tube configuration is: tube O.D. = 2 in., I.D. = 1.738, 22 tubes/row, length = 10 ft, inline arrangement with square pitch = 4 in., countercurrent flow, and 22 streams (100,000 lb/h of steam flows through 22 tubes). The duty is between 14 and 18 million Btu/h. What happens when the number of rows deep is varied, the fin density is varied from 2 to 5 fins/ in., and the fin height is varied from 0.5 to 0.75 in. (fin thickness is constant at 0.075 in.)? The procedure is very similar to that used in Example 2. In this case, though, the tube-side heat-transfer coefficient, hi must be computed using Eq. 6. The results are presented in Table 4. Though cases l and 2 transfer the same energy, the surface areas are significantly different, varying by nearly 100%. The reason is that the high fin density coupled with a

Btu/ft2h°F compared with 2,000 BtU/ft2h°F for boiling water) results in a much lower U . Hence, more surface area is required for the same duty. Similar results are obtained for cases 3 and 4, which have the same duty. Comparing cases 2 and 3, we see that case 3 transfers more energy with less surface area. This is due to a better fin configuration [as explained fully in (l)]. Thus, the bottom line is that simply relying on surface area for specifying a boiler is simplistic and can lead to wrong decisions by eliminating designs that can transfer the same duty but with a lower surface area.

SAFETY VALVE SIZING AND ASME CODE Section 1 of the American Society of Mechanical Engineers' Code (4) on boilers recommends that safety valves should be sized based on the surface area used in the boiler. This is not a prudent way of sizing safety relief valves. As we have just seen, one can transfer the same duty with variations in surface area of 50 to 100%. It is more practical to use the steaming capacity as the criterion for relief valve sizing. For example, in gas turbine exhaust applications, one can generate anywhere from 20% to

smaller hi (on the order of 200-250 76

OCTOBER 1992

0

CHEMICAL ENGINEERING PROGRESS

100% of the steam simply by varying the fuel input to the burner. To illus trate, a heat recovery boiler for a typical 3-5 MW gas turbine can generate about 20,000 lb/h of steam in the unfired mode but can be designed to generate up to 100,000 Ib/h of steam with the same surface area simply by firing auxiliary fuel. That is, the same boiler with th e same surface area can generate a maximum of only 30,000 lb/h at one site because the demand is only 30,000 Ib/h or 90,000 Ib/h at another location because that site needs 90,000 lb/h of steam. Now, according to ASME code, the safety valve sizing remains unaffected, as the surface area is the same!

Avoid heat transfer equipment vibration Plant heat transfer modifications and additions can change equipment dynamics. Use these design checks to predict and reduce vibration and noise problems V Ganapathy, ABCO Industries, Inc., Abilene, Texas TUBE BUNDLES in heat exchangers, boilers, superheaters and heaters are often subject to vibration and noise problems. Vibration can lead to tube thinning and wear, resulting in tube failures. Excessive noise can be a problem to plant operating personnel. Large gas pressure drop across the equipment is also a side effect, which results in large operating costs. 1 ,2 With the design checks presented here, one can predict during design if problems associated with noise and vibration are likely to occur. Vibration causes. Vibration and noise problems are

caused when air or flue gases flow over tube bundles, which may be arranged inline or staggered (Fig. 1). Vortices are formed and shed beyond the wake of the tubes, resulting in harmonically varying forces on the tubes perpendicular to the flow direction. It is a self-excited vibration. If the frequency of vibration of the Von-Karman vortices, as they are called, coincide with the natural frequency of vibration of the tube bank, resonance occurs which leads to tube vibration. Another phenomenon that occurs with vortex shedding is acoustic vibration, leading to noise and high gas pressure drop. The duct or the bundle enclosure vibrates when the acoustic oscillation frequency coincides with the vortex shedding frequency.' The acoustic oscillation is normal to both the direction of gas flow and tube length. Design methods to check vibration and noise. The

first step in the analysis for possible vibration or noise is the estimation of the vortex shedding frequency, fe Vortex shed

ding is prevalent in the Reynolds number range of 300 to 100,000, which is the operating range of many boilers, heaters and exchangers. The vortex shedding frequency may be estimated once the Strouhal number, S, is known which is given by the expression: S=fe d/(12V) (1) Here d is the tube outer diameter, V is the average gas velocity and S is a function of tube geometry. Figs. 2 to 5 give typical values of S. The natural frequency of vibration of the tubes is then determined. For a uniform beam supported at each end, fn is given by the expression 8 : fn =C(EI/MeL4 ) /2π (2) C is a constant depending on end conditions and is given in Table 1. The tube length in feet is l and Me is the total weight of the tube, which includes the contribution of the fluid weight inside and outside the tubes. For carbon steel tubes the above equation may be simplified and written as 8 : fn = 90C[(d 4 - d,4 )/Me ]0.5 /l2 (3) The next step is estimation of acoustic frequency, fa fa =Vs/λ (4) Vs is the sonic velocity of the gas and λ is the wave length. λ = 2w/n where w is the width of the duct in feet and n is the mode of vibration.For air or flue gases, Vs is approximately 49√T where T is the gas temperature in degrees R. For a cylindrical duct f, = NV s/D (5) ( N is a constant - 0.5681 for mode 1, 0.9722 for mode 2 and 1.337 for mode 3. 0.5

Checks and analysis for vibration and noise. To

analyze for possible vibration or noise in the tube bundles caused by flow of gases across tube banks, the following calculations are performed: 1. Calculate fn for different modes and load conditions. Compute fe. If fn and fe are within 20% of each other, vibra tion is almost certain to occur. 2. Estimate fa at different loads. Compare fa with fe. If Hydrocarbon Processing, June

Large gas pressure drop across the equipment is also a side effect o f noise and vibration, which results in large divide the gas column into smaller channels or ducts and thereby increase the acoustic frequency, moving it away from the vortex shedding frequency. If the gas temperature is high, the materials for baffles must be chosen with care. Acoustic vibrations usually lie in the range of 40 to 100 Hz.

Example problem. A tubular air heater 11.7 ft wide, 17.5

ft deep and 10 ft long is used in a plant. Carbon steel tubes of 2 in. OD and 0.08 in. thick are arranged inline with a transverse and longitudinal pitch of 3.5 in. The bundle is 40 tubes wide and 60 tubes deep. Air flows over the tubes, while flue gas flows inside. Air flow is 300,000 lb/h at an average temperature of 260°F. The tubes are fixed at each end in tube sheets. Analyze the bundle for possible noise and vibration problems.

Solution: Estimate fe . For st/d = s1/d = 3.5/2 = 1.75,

TABLE 1-Values of

C Mode of vibration

End support conditions Both ends clamped One clamped, one hinged Both hinged

1 2 22.37 61.67 15.42 49.97 9.87 39.48

TABLE 2-Summary of results n 1 fn fe fa (no baffles) fa 1 baffle) fa (2 baffles)

33.1 54 56.1 112.2 168.3

3 120.9 104.2 88.8

2

3

91 54 112.2 224

179 54 168.3 336.6

336

504

they are within 20% of each other, excessive noise is likely. 4,5,6The first mode of vibration is the most critical one as the amplitude of vibrations is large.

Eliminating noise and vibration problems. By

changing the tube span, tube pitch, or end conditions, the natural frequency may be altered keeping„ and fe apart to avoid vibration problems. Gas velocity can also be changed so that fe is altered. This may be done by changing the tube length and number of tubes wide. Primary correction devices for noise are baffles. 5,7 Baffles

62

Hydrocarbon Processing, June

from Fig. 3, S = 0.3. From Fig. 5, we see that S = 0.31. Calculate the air velocity, V. Air density = 0.081(492)/ (460 + 260) = 0.055 lb/ft.' V = 300,000(12)/[366(0.055)40 (3.5 - 2)10] = 30 ft/s. Hence fe = 12SV/d = 12(30)0.30/ 2 = 54 Hz. Estimate fn using Eq. 3. l = 10, d = 2, di = 1.84, M = 1.67 lb/ft = Me (neglecting weight of air/gas). For the first three modes, C1 = 22.37, C2 = 61.67 and C3 = 120.9, from Table 1. Then, fn1 = 33.1,fn2 = 91 and fn3 = 179 Hz, using Eq. 3. Let us compute the acoustic frequencies, fa . Sonic velocity, Vs = 49(460 + 260)°-' = 1,315 ft/s. Width, w = 11.7 ft and X = 2(11.7)/n, fa1 = Vs/λ = 56.1, fa2 = 112.2, f a3 = 168 Hz. The summary of results is shown in Table 2, which also shows the fa data with one and two baffles (w being 11.7/ 2 = 5.85 ft and 11.7/3 = 3.9 ft). Note that fa and fe are very close to each other in the very first mode. Hence, acoustic vibration leading to noise is likely. If one baffle is used, fa and fe are kept well apart in all the modes. Also, fa and fn are well apart in all modes, and tube vibrations are unlikely.

Conclusion. The above calculations show how one can

check a tube bundle design for possible vibration or noise problem. A simple approach was discussed. For elaborate analysis, one would use the methods discussed in literature.-`' However, noise and vibration problems are better predicted based on field operating experience of similar sized units. Performing the above calculations and modifying a design to keep the forcing frequencies well apart may not avoid noise/vibrations in all cases, as vibration and noise phenomenon are inexplicable at times. Damping effect of finned tubes, presence of' ash in flue

gases, manufacturing tolerances used and effect of end connections are variables that cannot be quantified. Hence, field experience coupled with analysis would be the ideal way to deal with the problem of noise and vibration.

NOMENCLATURE C Constant used in Eq. 3 d

E fa fa fn

Tube outer diameter, in. Tube inner diameter, in. Youngs modulus of elasticity, psi Acoustic frequency, hertz Vortex shedding frequency, hertz Natural frequency of vibration of tubes, hertz

L M n S sl st T V V

Tube length, ft Total we ight of tube per foot, lb Mode of vibration Strouhal number Longitudinal pitch, in. Transverse pitch, in. Gas temperature, R Gas velocity, ft/s Sonic velocity, ft/s

dj

I

w λ

Moment of inertia of tube

Width of duct, ft Wave length, ft

LITERATURE CITED Chen, Y. N., "Flow induced vibration and noise in tube bank heat exchangers due to Von Karman Streets," Trans ASME, Jour. of Engg for Industry, Vol 1, 1968, pp. 134-146 2 Rogers, J. D., et al., "Vibration prevention in boiler banks of industrial boilers," American Power Conference, 1977 s Fitzhugh,.J. S., "Flow induced vibration in heat exchangers," Symposium on vibration problems in industry, UK, April 1973 } Rogers, J. D., and Peterson, C. A. "Predicting sonic vibration in cross flow heat exchangers-experience of model testing," ASME 1977 WA/DE 28 Barrington, E. A., "Acoustic vibrations in tubular exchangers," Chemical Engineer ing Process, Vol 69, No 7, July 1973 t' Putnam, A. A., "Flow induced noise in heat exchangers," Trans ASME, Jour. of Engg for Power, Oct. 1959, p. 417 Deane, W. J., and Cohan, L. J., "Baffle plates cure boiler vibration," Power, Feb. 66, P. 82 t¢ Ganapathy, V., "Applied Heat Transfer," Pennwell Books, Tulsa, Okla. 82, pp. 650-658 "Symposium on Flow Induced Vibrations, Vol 3, Vibration in heat exchangers, ASME, 1984, pp. 87-101

The author V. Ganapathy is a heat transfer specialist with ABCO Industries Inc., Abilene, Texas. He is engaged in the engineering of heat recovery boilers for process, incineration and cogeneration applications. He also develops software for engineering of heat recovery systems and components. He holds a B Tech degree in mechanical engineering from Indian Institute ofMr. Technology, India, and an nology from Madras University. GanapathyMadras, is the author of over 125 articles on boilers, heat transfer and steam plant systems and has written four books: Applied Heat Transfer, Steam Plant Calculations Manual, Nomograms for Steam Generation and Utilization and Basic Programs for Steam Plant Engineers (book and diskette), copies of which are available from him. He also has contributed several chapters to the Encyclopedia of Chemical Processing and Design, Vol. 25, Marcel Dekker New York.

Hydrocarbon Processing, June 1987

63