Turboexpander-Compressor Technology for Ethylene Plants

AIChE T4 174869 Turboexpander-Compressor Technology for Ethylene Plants Radjen Krishnasing Senior Lead Process Engineer The Shaw Group Gabriele Mariotti Engineering Manager GE Oil & Gas Florence, Italy Kara Byrne Applications Engineer & Commercial Manager GE Oil & Gas Houston, TX, USA Radjen Krishnasing Senior lead process Engineer The Shaw Group

Prepared for Presentation at the 2010 Spring National Meeting San Antonio, TX, March 21-25, 2010 AIChE and EPC shall not be responsible for statements or opinions contained in papers or printed in its publications

Turboexpander-Compressor Technology for Ethylene Plants Radjen Krishnasing Senior lead process Engineer The Shaw Group Gabriele Mariotti Engineering Manager GE Oil & Gas Florence, Italy Kara Byrne Applications Engineer & Commercial Manager GE Oil & Gas Radjen Krishnasing Senior lead process Engineer The Shaw Group

Abstract Today’s ethylene plants incorporate Turboexpander Systems to optimize cryogenic recovery and reduce the energy demand. The molecular weight and flow rate of the residue gas depend directly on the selected upstream feedstock gas composition, conversion, and feedrates. Various recent ethylene units have generated residue gas volumetric flow ranges from approximately 100-200%. Hence, the Turboexpander system is designed and manufactured accordingly. As we are aware, the typical naphtha cracker produces a methane rich residue gas (bulk hydrogen is recovered, treated, and delivered as a high pressure co-product). On the other hand, the typical ethane or E/P cracker

produces a very high hydrogen content residue gas. Current designs and revamps require a wider range of feedstocks, and hence, a correspondingly wide range of residue gas composition and quantity. In order to meet the above demands, the Turboexpander solution must be flexible. As an overview, we will discuss the typical performance of one- and two-stage Turboexpander solutions for the expansion and recompression of the residue gas. Key mechanical design recommendations (e.g., magnetic bearings, variable nozzles, multistage control, high head wheels) will be outlined. Based on the demand from the different feedstocks and the industry requirements for feedstock flexibility, we will then discuss the technology and mechanical solutions. This presentation will also include related design improvements that have been successfully utilized in other Turboexpander applications. Part A Radjen Krishnasing

Introduction Turbo-expanders/re-compressors play a crucial role in the recovery of both ethylene and hydrogen from cracked gas in steam cracking units. A turboexpander converts energy that has been incorporated into the cracked gas, by the cracked gas compressor and by the ethylene/propylene refrigeration systems, back to refrigeration at the lowest temperature levels, to further enhance the recovery of ethylene and hydrogen. Turbo-expanders are, therefore, integrated into the cold fractionation cryogenic section of an ethylene unit. Turbo-expanders take the tail gas (mixture of hydrogen and methane) at high pressure and low temperature and drop the pressure over the expander with isentropic efficiencies of well more than 80%, producing a cryogenic stream that can be 40oC to 50oC lower than the lowest level of ethylene refrigerant. These cryogenic streams are then used for refrigeration to retrieve the last minor portion of ethylene from the tail gas that otherwise would have been lost. After providing refrigeration, the warmed up tail gas is compressed by the re-compressor to fuel gas pressure level. The driver of the re-compressor is the expander that conveys the energy liberated by the expansion through a common shaft. Effects ethylene plant feedstock A critical parameter in the integration and design of turbo-expanders is the composition of the tail gas (mixture of hydrogen and methane). Depending on the plant fresh feedstock and the potential hydrogen pre-recovery, the tail gas can be very rich in methane for one feed or very rich in hydrogen for another. Most ethylene units are designed to crack either a light feedstock, such as ethane/propane, or a heavy feedstock, such as naphtha or heavier

liquid feedstock. However, there are units with a much wider range of feedstock. Cracking a light feedstock, in particular ethane, produces a high ratio of hydrogen to methane. However, a typical ethylene complex based on ethane (or ethane/propane) needs very little hydrogen as product. The need is limited to the hydrogenation of acetylenes and small quantities of high purity hydrogen product, for use by downstream polymer units. To the contrary, an ethylene unit cracking naphtha or heavy liquid feedstock produces a lower ratio of hydrogen to methane but demands much more hydrogen co-product for the hydrogenation of unsaturated by-products that have been produced. Table 1 below demonstrates the yield patterns of different feedstock, expressed in component molar ratio with respect to ethylene. It shows a noticeable difference between ethane feed and any other feedstock: - Ethane as feed produces the highest ratio of hydrogen to ethylene, while the ratio of heavier byproducts to ethylene is the lowest. It produces very low ratio of methane leaving a tail gas high in hydrogen. -

Naphtha and gasoil as feed produce a relatively low ratio of hydrogen to ethylene, but a very high ratio of heavy byproducts to ethylene, therefore requiring very high recovery of hydrogen as product.

-

Propane as feedstock has a very interesting mid-position. It produces a tail gas that has a close resemblance to naphtha or gasoil. Propane can act as a buffer for the heavy feedstock in ethylene plants designed with a broad range of feed slate such as a unit to crack a combination of ethane and heavy feeds.

Table 1: The molar ratio of key components / ethylene in cracker effluent for typically used feedstocks. Feedstock type Ethane Propane Naphth Gasoil a Cracked Gas H2 / C2H4 1.07 0.63 0.44 0.30 Cracked Gas CH4 / C2H4 0.23 1.24 0.83 0.62 Cracked Gas (C4 & C5) / 0.02 0.07 0.22 0.24 C2H4 Cracked Gas Pygas / C2H4 0.01 0.05 0.19 0.14 Tail gas H2 / CH4 ratio 4.15 0.51 0.53 0.48 Ethylene plants cracking primarily liquid feedstock produce relatively high ratios of unsaturated C4 and heavier fractions. These fractions often require hydrogenation to either serve as recycle feed to the cracking furnaces or as finished product of the ethylene plant. A typical ethylene unit cracking liquid feed is therefore characterized by a very high recovery of hydrogen to balance this need. Recovery of hydrogen as product can be as high as 90%. Hydrogen is recovered at high pressure (3000 kPa), which means that the

recovered hydrogen can no longer be part of the tail gas that feeds the turbo-expander. The challenge in the integration and design of the turboexpander is to find the optimal balance between maximizing hydrogen recovery while maintaining a reasonable flow to the turbo expander to minimize loss of ethylene. On the contrary, an ethylene plant cracking ethane or a combination of ethane/propane is characterized by a very high ratio of hydrogen to ethylene, a low ratio of methane and an insignificant amount of C 4 and heavier fractions. As a result, the recovery of hydrogen as a product is little to none, meaning that virtually all of the tail gas is available as feed to the turbo-expander. However, as a lighter tail gas will have a richer ethylene content, maximizing the available tail gas for the turbo-expander is a critical parameter in reducing the loss of ethylene. Case study The following two cases are presented to further emphasize the design challenges when specifying and selecting a turbo-expander. The first case is for an ethylene plant where the predominant feedstock is naphtha, producing a nominal product rate of 1,000 KTA ethylene (1 million metric tons per year). This case will demonstrate that with the integration of a turbo-expander, only a single stage is needed. Hydrogen recovery is maximized while minimizing the loss of ethylene in the tail (or residue) gas. The variations in composition, and frequently the flow rate of the tail gas to the turbo-expander, are not affected if the feedstock cracked by the ethylene unit does not vary over a wide range from heavy to light naphtha. It is also not very sensitive to the cracking severity because the high hydrogen recovery results in a residue gas feeding the turbo-expander that is very rich in methane. A minimal variation of composition and flow rate to the turboexpander is then often caused by the extent of hydrogen recovery, or the overall plant capacity.

Table 2: Overview liquid (Naphtha) feedstock cracking. Key item clarifications: Am3 / min Actual cubic meters / minute dHs Isentropic enthalpy difference between inlet & outlet kmol / hr 1000 moles per hour kPA kilo pressure atmospheric Ξ (0.0145 psi / kPA) kW Hp = 0.746 kilowatts

Expander Inlet Flow rate (kmol/hr) Molecular weight Pressure (kPA) Temperature( oC) Expander outlet

Naphtha feed cracking

Higher Hydrogen recovery (less flow through turboexpander)

Lower Hydrogen recovery (more flow through turbo- expander)

2400

2176

2850

14.0 3050 -100

14.5 3050 -97

13.2 3050 -103

Flow rate (Actual 69 m3/min) Re-compressor Inlet Flow rate (kmol/hr) Mole weight Pressure (kPA) Temperature ( oC) Flow rate 3 (Am /min) Re-compressor Outlet Pressure (kPA) Expander dHs (kJ/kg) Turbo-expander RPM Expander power (kW) Expander efficiency (%)

63

83

2400

2176

2850

13.99 354 -3 214

14.52 356 -3 194

13.2 357 -4 250

604

604

604

111

105

118

28,630

27,640

30,000

970

870

1140

86

86

85

The second case (Tables 3A and 3B) is for an ethylene unit cracking light feedstock, ethane or ethane/propane. It is based on 1,500 KTA ethylene production rate (1.5 million metric tons per year). As can be seen from Table 1, that when ethane is cracked, it produces a high ratio of hydrogen and a low ratio of methane. The opposite is true if propane is cracked, resulting in a low ratio of hydrogen and a high ratio of methane. A turbo-expander designed for a hydrogen rich feed will, in general, require two single-stage expanders in series. The limitation is imposed by the recompressor section as is discussed in the second part of this paper.

Table 3A: Overview light (ethane, ethane/propane) feedstock cracking (HighPressure Machine) Key Item Clarifications: Refer to Table 2

Table 3B: Overview light (ethane, ethane/propane) feedstock cracking (Low-Pressure Machine)

100% C2 feed crackin g

50/50 C2/C3 feed cracking

HP Expander Inlet Flow rate (kmol/hr) Mole weight Pressure (kPA) Temperature( oC)

7935 4.94 2131 -114

7883 7.66 2131 -118

HP Expander outlet Flow rate (Am3/min)

LP Expander Inlet Flow rate (kmol/hr) Mole weight Pressure (kPA) Temperature( oC)

127

121

LP Expander outlet Flow rate 215 (Am3/min)

HP Compressor Inlet Flow rate (kmol/hr) Mole weight Pressure (kPA) Temperature ( oC) Flow rate (Am3/min) HP Compressor Outlet Pressure (kPA) Expander dHs (kJ/kg) Turbo-expander RPM Expander power (kW) Expander efficiency (%)

7837 4.72 630 63 580

7567 7.17 636 60 550

740 130

740 80

20,000

16,240

1360

1255

86

85

LP Compressor Inlet Flow rate (kmol/hr) Mole weight Pressure (kPA) Temperature ( oC) Flow rate (Am3/min) LP Compressor Outlet Pressure (kPA) Expander dHs (kJ/kg) Turbo-expander RPM Expander power (kW) Expander efficiency (%)

100% Ethane feed cracking

50/50 Ethane/propa ne feed cracking

7923 4.91 1165 -135

7742 7.42 1175 -134

207

7836 4.72 532 43 647

7567 7.17 541 43 612

630 124

636 82

20,000

16,340

1355

1265

89

86

Further evaluation/observations  An important turbo-expander design parameter is the isentropic enthalpy drop (dHs) across the expander. As discussed in the second part of this paper, this number is indicative of the expander or re-compressor wheel tip speed. As a general guideline, an enthalpy drop of up to 180kJ/kg is considered to set an optimal basis for the turbo-expander design. For our naphtha case, the isentropic enthalpy drop is in the order of 110kJ/kg – a number that falls in this range and does not provide unusual constraints to the design of the turbo-expander. A single-stage design is therefore very common for naphtha (or other liquid/LPG feedstock) based ethylene plants.



For our ethane cracking case, a two-stage turbo-expander/re-compressor design is used. The isentropic enthalpy drop across each expander stage is kept around 125kJ/kg. Although using a single stage expander is not impossible, the overall isentropic drop in that case would be 250 kJ/kg. In general, the constraint is not the expander side but the compressor side. As can be seen from the tables, the volumetric flow of gas flowing into the recompressor is nearly five times higher than the expander outlet flowrate. The re-compressor rotor is therefore the larger of the two wheels, becoming the limiting factor in the design.



The naphtha case demonstrates the effects of higher or lower hydrogen recovery than the design recovery of the turbo-expander. A higher recovery of hydrogen can be desired in plant operations as a way to produce more product hydrogen. This will reduce the total flow through the expander, while at the same time increasing the molecular weight. As can be seen in the second column in Table 2, the turbo-expander is still within its operable range, but it will provide less refrigeration because of the reduced flow rate through the turbo-expander. This will have to be taken into consideration when deciding on increasing recovery of hydrogen.



As the demand for raw C 4 and perhaps also raw C5 as finished co-products without hydrogenation increase, an ethylene plant cracking liquid feedstock can end up with excess hydrogen product. If there is no other output for product hydrogen, it is ultimately letdown to the fuel gas header and combusted in the cracking furnaces. Instead of letting the product hydrogen across a control valve (isenthalpic), it would be more beneficial to pass this

excess of hydrogen through the expander. The third column of Table 2 (the naphtha case) demonstrates the effects this will have. More hydrogen across the expander will result in more cryogenic duty from the turbo-expander, and as an overall effect, it will reduce the refrigeration demand from ethylene/propylene refrigeration systems. Table 2 shows that the increased flow rate combined with a reduced molecular weight will increase the RPM of the turbo-expander. How much hydrogen can be diverted to the turboexpander is a function of how much room is available in the design of the turbo-expander. A typical design comfortably will accommodate an increase such as demonstrated in the table.



The gas cracker case evaluation demonstrates the simple fact that in case a turbo-expander is designed for the tail gas of an ethylene plant cracking ethane (tail gas very rich in hydrogen), a mixed feed case of ethane and propane is less stringent to the operation of the turbo-expander. The first column of Table 3A and Table 3B are for pure ethane feedstock, while the second column of each is for a 50/50 ethane/propane case.



In these days of mega-size steam cracking units, serious challenges are presented to the sizes of major compressors and other equipment, such as separation columns. When it comes to turbo-expanders however, the sizes are far from reaching their maximum. While the naphtha case turboexpanders use a 225mm expander wheel and the gas case a 350mm wheel; these are by far not the largest sizes used in other branches of the industry for turbo-expanders. It is also interesting to note that the scale-up, which has been seen since the early use of turbo-expanders, from small ethylene units to today’s mega-size plants, hardly has affected the high (isentropic) efficiencies the industry has relied upon. This feature continues to make turbo-expanders a very important choice in maximizing the economics of ethylene plants.

Part B Gabriele Mariotti Kara Byrne

Foreward The importance of turboexpanders has increased significantly over the past few decades since the first application of a turboexpander in the oil and gas industry by the founder of Rotoflow, Dr. Judson Swearingen. Typically, turboexpanders were used to replace a Joule-Thompson (JT) valve in order to increase the overall efficiency of air separation plants. Driven by increased competition in the oil and gas market, it is increasingly common to find a turboexpander as a key component for the overall production in a hydrocarbon gas separation plant. This is especially important for designing a more efficient and competitive ethylene plant. While the turboexpander alone can easily reach isentropic efficiencies of up to 90%, when it is directly coupled to a compressor the interaction of the two machines must be taken into account. The turboexpander efficiency is limited by the compressor (and vice versa) and, therefore, cannot be optimized beyond the mechanical limitations of each machine. This paper, after a brief discussion of current technologies and the characteristics of GE Oil & Gas Turboexpanders, will focus on some typical turboexpander compressor selections showing the interaction between the selection of the turboexpander and re-compressor. Turboexpander History The turboexpander is a reaction type radial turbine originally developed to replace the Joule-Thompson (JT) valve in air separation plants. The French Engineer, George Claude, utilized the first radial turbine for air liquefaction in the early 1900s. German engineers, including Dr. Carl von Linde, further developed and improved the turbines for many other applications, such as refrigeration and jet propulsion engines. After World War II, Dr. Judson Swearingen began to develop the turboexpander for natural gas processing applications (Photo-1). He realized the overall cooling capacity of the plant and, therefore, the cost and performance, is greatly improved by replacing the JT Valve with a simple and reliable machine that expands a single-phase stream in a nearly isentropic method. The fact that the radial inflow turbine could handle two-phase flow at the discharge made the machine perfect for heavy hydrocarbon removal.

The turboexpander continues to date to develop in the natural gas industry. In the 1960s, turboexpanders were used in ethylene projects and then naturally progressed into several other markets such as liquefied natural gas, geothermal, and gas-to-liquids.

Turboexpander Applications Turboexpanders are predominantly used in refrigeration/liquefaction processes and power generation applications. The refrigeration/liquefaction process utilizes the Turboexpander for cooling fluids through nearly isentropic expansion from a higher pressure to a lower one. This is able to achieve much lower temperatures than throttling the fluid through a JT valve by isenthalpic expansion. The lower temperatures considerably increase the overall refrigeration cycle efficiency. Typical applications covered by GE Oil & Gas Turboexpanders are: Natural Gas Processing/Dew Point Control Plants, Pressure Let Down Energy Recovery, and Geothermal/Waste Heat Energy Recovery. Depending on the service required, mechanical power produced by expansion of flow in the radial turbine can be recovered or dissipated through three main machine configurations: Turboexpander-Generator Mechanical power is converted into electrical power through a reduction gear and a generator (Photo-2).

Photo-2: Turboexpander-Generator General Arrangement Turboexpander-Compressor Mechanical power drives a compressor impeller either coupled to the same shaft as the turboexpander or driven via a gearbox (Photo-3).

Photo-3: Turboexpander-Compressor General Arrangement Turboexpander-Dyno Mechanical power is dissipated through an oil brake if it is not economical to convert the excess power into another form of energy (Photo-4).

Photo-4: Turboexpander-Dyno Often it is not clear which turboexpander configuration is suitable for an ethylene plant, since the same service can be covered through either a Turboexpander-Generator or a Turboexpander-Compressor. Table-1 lists the pros and cons of both solutions.

COMPRESSORTURBOEXPANDER- GENERATOR TURBOEXPANDER-

Table-1: Comparison of Various Turboexpander Machinery Configurations PROS CONS  Very high efficiencies can be achieved. The wheel can be optimized to achieve the best aerodynamics by freely changing the RPM without other machinery constraints.  Recompressor is designed independently from the turboexpander, merging more stages into a single machine with higher efficiency.  Simpler plant layout: reduced number of piping interconnections.  Simpler machine control can easily be set up for a fully automatic control system.  A fixed speed machine can typically perform better in off-design condition when the enthalpy drop is maintained constant with process controls.



 Very robust and simple machine.  Perfect for oil free applications with the use of active magnetic bearings (AMB).  The stiff shaft design improves the operating range and the capability to withstand very high imbalances.  Labyrinth, or similar, seals and the pressurized auxiliaries system makes it very difficult for gas to escape from the machine in case of failure.  For a well-balanced machine, the turboexpander flow and re-compressor flow are linked. This reduces the size of required anti-surge systems to manage unbalances in flow between the turboexpander and compressor.











The machine has a tendency to speed up in case of electric load rejection. This limits the maximum tip speed of the wheel and tripping devices need to be redundant for safety reasons. The machine is typically more complex than a TurboexpanderCompressor due to the presence of a gearbox, generator, and other auxiliaries. Cost per unit is higher and oil free solutions are not yet economically feasible.

Efficiencies are sometimes lower than turboexpander-generator due to the balancing of the turboexpander and compressor performance and limitations. If the plant throughput (flow) is decreased while the pressure ratio is kept constant, the machine speed will reduce with a significant loss in efficiency. Units may be arranged in series, increasing the complexity and tuning of the control system.

It should be noted that dyno, pump, and blower configurations have not been included in the comparison table because they are not typically applied to medium and large sized machines that are commonly found in ethylene plants. GE Oil & Gas Product Line The GE Oil & Gas Turboexpanders product line is standardized so that most of the components are pre-designed. Parts that normally need to be customized for each project are the wheels (both turboexpander and compressor), shaft, nozzle assembly, diffuser cone, compressor follower, gear, auxiliaries and controls. The naming convention for machine standardization is the “Frame” size. The frame size is directly linked to the casing and, therefore, the overall dimension of

the machine. Each standard frame can accommodate a specific diameter range of turboexpander wheels. Frame sizes are also distinguished by the design pressure and flow rate. The design pressure sets the flange ratings. Each of the Frame Sizes are clarified further in Table-2. Table-2: GE Oil & Gas Frame Size vs. Flange Ratings & Flow FRAME #

TURBOEXPANDER RATING ACCORDING TO ANSI (PSI) 150

10 15 20 25 30 40 50 60 80 100 130 160 180

x x x x x x x x x X

300 x x x x x x x x x x x x

600 x x x x x x x x x x

900 x x x x x x x x x

OUTLET FLOW (ACMH)

1500 x x x x x

450 1000 4000 5500 9000 16000 25000 36000 45000 65000 100000 160000 200000 TURBOEXPANDER GENERATOR FRAME SIZE AVAILABLE TURBOEXPANDER COMPRESSOR FRAME SIZE AVAILABLE

Table-2 is applicable to turboexpander-compressors (EC), turboexpander-multistage compressors (ECC), and turboexpander-generators (EG) single stage or multistage integrally geared types.  Typical design limitations are as follows:  Power up to 35 MW  Wheel diameter up to 1800mm  Design temperature from –270oC to +315oC  Mechanical design in accordance with API 617 Chapter 4  Lube oil system in accordance with API 614 Chapters 1, 2, and 4  Turbine operability in accordance with IEC45 or API 612 Chapter 12 As with most turbomachinery designs, there are standard comments and exceptions to all of the industry specifications listed above. The design temperatures typically set the materials of construction for the components. For cryogenic applications the turboexpander casing is typically stainless steel, but if warm enough low temperature carbon steel can be used. The compressor casing and bearing housing are typically carbon steel due to the warmer temperatures. Other components are also affected mechanically. For example, by using a fixed nozzle instead of a variable nozzle, the design temperature limitations can exceed the values given above. While there are no size limitations for turboexpander-generators and turboexpander-compressors with traditional oil bearings, the active magnetic

bearing (AMB) units need to be checked versus the standard bearing size from AMB suppliers. GE Oil & Gas has additional experience with special “canned type” magnetic bearings that are suitable for aggressive and sour gases typically not tolerated by standard electrical devices. This design encapsulates traditional electrical components of the AMB within a metal “can” made of Inconel material that prevents any contact with process gas. This design, mainly used in natural gas applications, allows the AMB to operate without being contaminated or harmed by the aggressive gas. Photo-1 shows a machine currently installed with this technology.

Photo-3: Turboexpander-Compressor with “Canned” Active Magnetic Bearing The GE Oil & Gas product line offers a fabricated casing design, as shown in Figure-1, in addition to the traditional Rotoflow cast casing design. This recently applied technology is able to ensure the highest quality pressure-containing components while also minimizing any potential defects during the manufacturing of the unit. Moreover, the use of a fabricated casing ensures the flexibility to design for a wide range of applications, ratings, and nozzle loads. The internal parts made by castings can now be aerodynamically shaped for the best efficiency. In particular, the re-compressor discharge volute can be manufactured with a variable section scroll and a tangential nozzle to provide the best efficiency and range.

Figure-1: Turboexpander-Compressor Cross-Sectional Drawing The control of the turboexpander is primarily accomplished by means of adjustable guide vanes (nozzles). GE Oil & Gas can provide patented solutions with a traditional Rotoflow slot and pin mechanism, shown in Figure-2, which is very effective on turboexpander-compressors. Also available is a newly patented multilink mechanism, shown in Figure-3, which adjusts the guide vanes using a “progressive” opening law for precision flow control and minimal actuating forces.

Figure-2: Slot and Pin Inlet Guide Vane (Nozzle) Assembly

Precise flow regulation is useful in turboexpander-generators in order to minimize the speed fluctuations at low load and synchronize the generator to the grid without using an external control valve. The improved mechanical design of the nozzle mechanism is associated with increased aerodynamic performance design. Antifriction and anti-wear coatings on the nozzle segments minimize the losses during the first isenthalpic expansion. Nozzle segments are subjected to severe working conditions as shown in the Finite Element Analysis of Figure-3. These conditions are due to the high velocities of the gas at this location (similar to the wheel tip speed) and because of the presence of solid particles and liquid droplets passing through the turboexpander. For this reason, tungsten carbide coatings or surface induction hardening are typically applied to the nozzles to minimize erosion problems. Another key component of the turboexpander-compressor is the wheel. To ensure the reliability of the machine, the turboexpander and compressor wheels need to be carefully designed in order to avoid excessive stresses, harmful resonances, and erosion by liquid droplets. The wheel and wheel attachment has a strong influence on the rotor dynamics of the machine. As shown in Figure-4, GE Oil & Gas designs and manufactures open and closed wheel designs up to 1800 mm diameters in various materials.

In general, the most common material in ethylene plants is 7050 Aluminum. This material has a very good weight to strength ratio, which is required to reach very high tip speeds. Titanium with superior properties is not typically used when there is hydrogen in the tail gas, but is commonly used in many other turboexpander applications. Each wheel is analyzed by means of a finite element analysis (FEA) tool to assess the stress and modal behavior. The modal behavior is assessed to avoid possible resonances between the stimuli from the nozzle segments and natural modes of the wheel.

Figure-5: Finite Element Analysis of a Compressor Wheel In ethylene plants, where the compressor head requirements are very severe (Figure-5), the maximum head is determined by a compromise between the mechanical aspects (tip speed) and aero design (blade loading). GE Oil & Gas uses hirth serration (Figure-6), a splined fit, to attach the wheel to the shaft. This solution minimizes the centrifugal stresses on the wheel and, therefore, improves the maximum tip speed and head capability.

TIE ROD KEYS

Figure-6: Hirth Serration

HIRT

Turboexpander Performance Turboexpander Selection The turboexpander performance is computed as a function of a nondimensional factor called specific speed (Ns) defined as: N Q2 Ns  3/ 4 his where Q2 is volumetric flow at the discharge, his is the isentropic enthalpy drop through the turboexpander, and N is the rotating speed of the machine selected. The specific speed is the key parameter for the assessment of the efficiency of a radial turbine at the design point. The optimal range of specific speed for turboexpander design, as shown in Figure-7, is from ~1800 to ~2000.

Figure-7: Normalized Efficiency vs. Turboexpander Specific Speed The specific speed is related to the maximum enthalpy drop that one stage can handle. Typical numbers for the maximum enthalpy drop are:  Low Specific Speed (500 < Ns < 1000): 350 kJ/kg (148.2 BTU/lbm)  High Specific Speed (2000 < Ns < 2500): 180 kJ/kg (76.2 BTU/lb m) A second important parameter to consider is the u 1/Co factor. This is a nondimensional parameter where u1 is the tip speed of the wheel and C o is the spouting velocity. The spouting velocity is the fluid speed that would be achieved if the entire isentropic enthalpy drop were to be converted into speed. In other words, it is the speed that is created from putting work into the system. This is similar to converting the potential energy in a water tower into a velocity at the exit of the tower. Figure-8 further explains this idea pictorially, with H being the potential energy and w being the speed at the water tower exit. SPOUTING VELOCITY:

Co  hts ,is

Figure-8: Spouting Velocity Pictorially Represented The u1/Co factor determines the degree of reaction of the turboexpander stage and is selected during the design phase (Figure-9). The optimum u 1/Co is around 0.7, corresponding to approximately a 50% degree of reaction. In this configuration, the inlet of the turboexpander wheel is radial, improving the ability to withstand liquid at the inlet.

The u1/Co factor becomes important during the testing of a turboexpander. Current API 617 practices call for it to be one of the measured values in the machine final testing. In an ethylene plant, the gas conditions are never constant. It is important to predict the behavior of the turboexpander in off-design conditions. The turboexpander efficiency is affected by the change in two main parameters: u1/Co and Q2/N (the flow coefficient). The efficiency of the machine in off-design conditions considers the effect of variation of flow rate and u 1/Co ratio. After the calculations have been completed, formula correction factors are provided in correlation curves, based on experience (Figure-10).

Figure-10: Sample Correlation Curves for Efficiency Correction Factors

The overall plant control and machine selection should take into account the turboexpander behavior during off-design conditions. Here is a typical range for u1/Co and Q2/N turboexpander off design conditions:  % Q2/N: 30 to 140% of design case  % u1/Co: from 30 to 135% of design case Compressor Selection The compressor is used as a brake for the turboexpander. The absorbed power determines the operating speed of the turboexpander-compressor. The compressor selection is very important in ethylene applications, where very often the compressor is required to produce very high head. Recent developments in ethylene plant design also impose more importance on the re-compressor performance. The compressor is no longer seen as a “by product”, but rather an important plant component that is required to operate with good polytropic efficiency, turndown, and head rise. The compressor load influences the turboexpander efficiency. Compressors with controllable power absorption characteristics can be supplied to provide more flexibility to the turboexpander. The compressor selection is made using three main parameters: 4Q1  Flow coefficient:   D22u 2 u2  Compressor Peripheral Mach Number: Mu  a 0t h  Work Coefficient:   2 u2 where Q1 is the volumetric flow at the inlet, D 2 is the impeller diameter and u 2 is the wheel peripheral speed. The capability for a given wheel to produce power depends on both  and u2 squared and the mass flow rate that is handled by the compressor wheel. The Work Coefficient is limited by the aerodynamic design of the wheel and the peripheral speed affects the static stress on the impeller. In ethylene applications, the Mach number is normally not an issue because of the low molecular weight gas. Typical numbers for the maximum enthalpy change on the compressor wheel are as follows: 

Low flow coefficient (0.025 < <  0.100): 150 kJ/kg (63.5 BTU/lbm)

 High flow coefficient (0.180 < < 0.280): 120 kJ/kg (50.8 BTU/lbm) A well-balanced turboexpander and compressor wheel depends on the process design. The turboexpander wheel power (including mechanical losses) should be the same as the compressor absorbed power. Gexp his is  Gcomp his

It should be noted that the capability for the compressor to act as a load for the turboexpander does not depend on the polytropic efficiency. For this reason, an

optional hot bypass around the compressors can be used to artificially increase the absorbed power, also reducing the turboexpander speed. As a consequence, the efficiency of the compressor will drop because of the “internal” recirculation. Turboexpander and Compressor Interaction As seen earlier, the specific speed (Ns) is one of the main parameters to determine the efficiency of the expander. The efficiency vs. Ns curve has a flat peak portion ranging from ~1800 to ~2000 (Graph-2). Targeting a minimum value of Ns (i.e. Ns > 800), it is possible to determine the minimum rotational speed of the machine. This is important in order to stay within an acceptable efficiency range as a function of the ratio h to the expander volumetric flow at the outlet (Figure-11).

Figure-11: Minimum Rotational Speed of Turboexpander (Assuming Similar Mass Flow Rate Between Turboexpander Compressor) On the other hand, the rotational speed affects the compressor flow coefficient. The rotational speed must be limited below a given value in order to limit the compressor flow coefficient and also to increase the capability to produce head and power. This behavior is exactly the opposite of the turboexpander. The following graph (Figure-12) represents the change of compressor flow coefficient as a function of the rotational speed for two density ratios. This ratio is between the density at the expander outlet and the density at the compressor inlet. The warmer gas at lower density on the compressor side tends to increase the flow coefficient. This needs to be kept under a given value by reducing the speed, which has an impact on the expander efficiency as seen in Figure-11.

Figure-12: Compressor Rotational Speed vs. Flow Coefficient In summary, the turboexpander and the compressor selection have to be balanced. In order to do so, the turboexpander efficiency may be negatively affected. This could occur for several reasons, but the major issue that affects this “balance” is the density ratio imbalance between the turboexpander discharge and the compressor suction. Case Studies Two case studies where analyzed, to provide examples of the trends in today’s ethylene plants: a naphtha cracker producing a methane-rich residue gas and a typical ethane or ethane/propane (EP) cracker producing a light hydrogenrich residue gas were analyzed. The focus was on the turboexpander-compressor configuration since this is more complex than a turboexpander-generator in conjunction with a stand-alone re-compressor. Liquid Cracker The liquid cracker evaluation was made considering the following scenarios:  Base Case: high percentage of hydrogen recovery.  Lower Hydrogen Recovery: reduced rate of hydrogen recovery and, therefore, a larger percentage of ethylene recovery. This case reduces the C2 and C3 refrigeration to a certain extent.  Higher Hydrogen Recovery: increased rate of hydrogen recovery with decreased flow. With the margins available in cold boxes, this increased rate of hydrogen does not affect the ethylene recovery or the refrigeration. The machine selection for this service does not have any issues related to specific speed at the higher range of efficiencies. The turboexpander-compressor is at the lower end of GE Oil & Gas production capabilities, corresponding to a Frame 20 (EC201). This service can be satisfied either with oil bearings or active magnetic bearings. The selection based on compressor efficiency can be further optimized to improve the efficiency. However, based on all parameters, the initial selection fits

into a very standard unit, and both the mechanical and aerodynamic characteristics are well within proven experience. The same case study was analyzed by increasing the flow rate by 25%. Since the gas conditions remain unchanged, the machine selection resulted in a similar unit design, scaled up to the Frame 25 (EC251). Table-2 provides a summary of the machinery sizing for the Liquid Cracker case to highlight the important turboexpander factors, such as specific speed (Ns). Table-2: Liquid Cracker Turboexpander-Compressor Sizing at 100% Flow Case Description UNIT Condition RPM Ns Diameter Efficiency Wheel Power Weight Liquid Frame size

BASE H2 RECOVERY

Exp Comp Design 35,000 35,000 1,500 3,200 (mm) 200 230 (%) 84-88% 72-76% (hp) 1039 1035 (%) 15.3 EC0201

LOWER H2 RECOVERY

HIGHER H2 RECOVERY

Exp Comp Off-Design 33,800 33,800

Exp Comp Off-Design 36,630 36,630

84-88% 936 14.7

82-86% 1223 15.7

72-76% 933

71-74% 1219

GAS CRACKER Gas crackers produce a very large residue gas stream with high concentrations of hydrogen. The gas does not vary with hydrogen product demand. In fact, the demand of hydrogen product is very low. Variation occurs due to co-cracking of propane or other feedstock. This reference is based on 100% ethane cracking (the base case) with the option of 50/50 Ethane/Propane cracking. From a machinery design point-of-view, this service is considered to be more difficult due to the high enthalpy change involved. A first selection was made with a 2-stage expander compressor, a standard configuration for the 100% and 111% flows. Both units are sized into a Frame 40 (EC401) with good efficiencies and with well-referenced mechanical and aerodynamic parameters. Table-3 shows an overview of the machine performance.

Table-3: Gas Cracker Turboexpander-Compressor Sizing at 100% Flow Case Description UNIT Condition RPM Ns Diameter Efficiency Wheel Power Weight Liquid Frame size

100% Ethane BASE 100% Ethane BASE

50/50 Ethane/Propane

50/50 Ethane/Propane

Exp_HP Comp_HP Exp_LP Comp_LP Exp_HP Comp_HP Exp_LP Comp_LP Design Design Off-Design Off-Design 20,000 20,000 20,000 20,000 16,270 16,270 16,360 16,360 1,100 3,000 1,400 3,200 (mm) 325 425 350 425 83868384(%) 87% 73-77% 89% 72-76% 87% 70-74% 88% 70-74% (hp) 1629 1626 1630 1627 1509 1507 1528 1526 (%) 0.6 4.8 4.9 5.5 EC0401 EC0401

If the flow is increased by 11%, the design remains basically the same. However, the selected wheels are larger in terms of flow capability (larger flow coefficient). The flow capacity of a turboexpander can be increased by either using a wheel design with a higher flow coefficient/specific speed, or by increasing the diameter and reducing the rotational speed to keep the same peripheral speed. The second option is required to handle the different enthalpy change. With the intent of simplifying the plant layout and reducing cost, GE Oil & Gas has selected for this service a single Frame 40 (ECC401) machine, with twostage compressors directly coupled to a single expander wheel. This type of unit is referenced with oil bearings and can also be developed with AMB. Table-4: Gas Cracker Turboexpander-Multistage Compressor Sizing at 100% Flow Case Description UNIT Condition RPM Ns Diameter Efficiency Wheel Power Weight Liquid Frame size

100% Ethane BASE

Exp

(mm) (%) (hp) (%)

23,000 1,000 350 78-82% 2396 0.6

50/50 Ethane/Propane

Comp_LP Comp_HP Exp Comp_LP Comp_HP Design Off-Design 23,000 23,000 18,890 18,890 18,890 3,900 3,700 350 350 74-78% 74-78% 77-81% 71-75% 71-75% 1466 1466 27244 1386 1386 4.9 ECC401

Due to the very high enthalpy drop across the expander stage, the efficiency is highly penalized with respect to the traditional design at nearly the same specific speed. The turboexpander-compressor-compressor solution (Figure-13) could be considered as a low cost alternative solution. This arrangement would also be considered if the turboexpander enthalpy drop per stage were lower. The rotor dynamics of this arrangement needs to be analyzed carefully to ensure a robust design without harmful expander wheel-overhung modes throughout the operating range.

BRG

Expander Compressor

Figure-13: Turboexpander-Compressor-Compressor (ECC) Arrangement Conclusions This paper presents an overview of current turboexpander technology to provide information for the selection of the best machine configuration and thermodynamic design for ethylene plant applications. GE Oil & Gas has analyzed potential selections for turboexpander-compressors for large ethylene plants. The results show that there are no issues with increasing the machine capacity, due to the scalability of the unit frame sizes. However, large enthalpy drops per stage and optimization trade-offs between the expander and compressor wheels need to be carefully evaluated to find the best compromise between cost and performance.