Introduction (Extracted from Chapter 12 of Reference [3]):

 

The surface area of a heat exchanger can be calculated by Equation 1.

                                                                                                          (1)

 

Where:

A         = heat exchanger area in m² [ft²]

Q         = heat transfer rate in kW [Btu/hr]

U_o        = overall heat transfer coefficient in W/ (m^2 oC) [Btu/(hr-ft²–^oF)]

ΔT_eff    = effective temperature difference ^oC [^oF]

 

 

Figure 2 presents a temperature profile as a function of heat transfer in a chiller and a gas-gas HX. In the chiller, the process fluid is partially condensing in the tube side, and pure propane is boiling in the shell side. In the gas-gas HX, the process fluid (stream 5) is partially condensing in the tube side and the gas stream 4 is warming (no phase change) in the shell side.

 

 

Figure 2. Effective Temperature Difference Schematics [2]

 

 

The only temperatures that can be conveniently measured are at the inlet and outlet of the exchanger. The largest difference on one end of the heat exchanger will be referred to as ΔT₁, the smaller ΔT₂, which is also called the minimum approach.

 

 

Equation 2 provides a simple method for estimating the effective temperature difference (ΔT_eff) in the exchanger.

                                                                                 (2)

 

 

Where:

LMTD = log mean temperature difference

ΔT_eff    = temperature difference corrected for heat exchanger configuration

F          = TEMA MTD Correction Factor

ΔT₁      = largest ΔT (at one end of the heat exchanger)

ΔT₂      = smallest ΔT (at one end of the heat exchanger)

 

 

Equation 2 is based on several assumptions.  The primary one being that the heating and cooling curves (T vs Q) for the exchanger streams are linear.  For multi-component fluids this is true when there is little or no phase change in the exchanger.  If you apply this equation, always have a look at the cooling curves to verify this assumption is adequate the application.

 

Notices in this equation that as ΔT₂ (minimum approach) decreases, ΔT_eff decreases and the required heat transfer area increases as shown in Equation 1. This increase can be significant as ΔT_eff approaches zero. On the other hand, smaller values of ΔT₂ in the gas-gas exchanger for a refrigeration plant decreases the utility costs (power and fuel) there is more “free cold energy” transferred in the heat exchanger. For this specific process application (gas-gas exchanger) both effects should be considered in heat exchanger specifications.

 

Note that the minimum approach may occur at the “hot” end or “cold” end of the exchanger depending on the application. The minimum approach may also occur internally inside the exchanger. Typical economic minimum approaches for various heat exchangers and applications are presented in Table 1.

Table 1. Typical Heat Exchanger Approach Ranges [3]

For compact exchangers, smaller minimum approaches are usually economically justifiable. Compact heat exchangers have a greater surface area per volume than shell and tube heat exchangers. In addition, they have higher fluid velocities, which result in greater overall heat transfer coefficients.

 

 

CASE STUDY

 

Let’s consider the same case presented in Apr 2019 TOTM [1] for a rich gas with the compositions and conditions presented in Table 2 [1]. Based on the reported molecular weight and relative density for the C7+ fraction, Table 3 presents the estimated normal boiling point (NBP), critical properties and acentric factor which, are needed by the equation of state. The objective is to meet a hydrocarbon dew point specification of  -20 °C (-4 °F) at about 4000 kPa (580 psia) for the sales gas by removing heat in the “Gas/Gas” heat exchanger (HX) with a hot end TA of 5°C (9°F)  and in a propane chiller, 5 °C (-4 °F) TA, and rejecting it to the environment by a propane condenser (AC-100) at 37.8°C (100°F). Pure propane is used as the working fluid in the simulation. The pressure drops in the “Gas/Gas” HX and the propane chiller are assumed to be 34.5 kPa (5 psi).

 

 

Table 2. Rich feed gas compositions and conditions

Table 3. Estimated C7+ properties [4]

 

The feed gas is flashed in the “Inlet Separator” at 30 °C (86 °F) and 4000 kPa (580 psia) to remove any condensate. The “Inlet Separator” vapor (stream 2) is saturated with water by the “Saturate -100” to form stream “2 Wet” upstream of mixing with MEG hydrate inhibitor, stream “EG1” and the recycle stream “18A” from the deethanizer overhead vapor (located at the right-hand side of Fig. 1).

 

The estimated hydrate formation temperature (HFT) of streams 5 and 7 is 14.7 °C (58.4 °F). The hydrate inhibitor is injected at the inlet of “Gas/Gas” HX by stream “EG1” and at the inlet of the “Chiller” by stream “EG2”. Stream “5” cools to about -8 °C (17.6 °F) and stream “7” cools down to the specified temperature of -20 °C (-4 °F), which is below the HFT of 14.7 °C (58.4 °F). The injection rates of streams “EG1” and “EG2” for 80-weight % lean MEG and water solution are estimated manually or by the Adjust tool of UniSim. A design margin of 1 °C (1.8 °F) HFT below the cold temperature for streams “5” and “7” were assumed.

 

Assuming a temperature approach of 5°C (9°F) and a 6.9 kPa (1 psi) pressure drop in the propane chiller “Chiller” shell side, the pressure of saturated propane vapor leaving the chiller is 203.3 kPa (29.5 psia), and at a temperature of -25°C (-13°F).  Assuming no frictional losses in the suction line to the propane compressors “K-101” and “K-102”, the resulting suction pressure is 203.3 kPa (29.5 psia).

 

The condensing propane pressure at the specified condenser temperature of 37.8 °C (100 °F) is 1303 kPa (189 psi). The condenser “AC-100” frictional losses, plus the frictional losses in the piping from the compressor discharge to the condenser were assumed to be 34.5 kPa (5 psi); therefore, the discharge pressure of compressor “K-102” is 1338 kPa (194 psia). The compressors interstage pressure was determined by equalizing the power for “K-101” and “K-102”. The compressors adiabatic efficiency was assumed to be 75%.

 

The cold Stream 7 is flashed in the 3-phase separator “V-102” at -20 °C (-°4F) and 3931 kPa (570 psia). The vapor stream “4” from this cold separator is used to cool down the incoming warm feed gas in the “Gas/Gas” HX. The heavy liquid stream “8B” (rich MEG solution) from the cold separator is regenerated in the regeneration unit (not shown in Fig. 1) and the lean 80 weight % MEG is recycled and used in streams “EG1” and “EG2”. The cold NGL stream “8” (light liquid phase) from the cold separator, “V-102”, is combined with the plant “Inlet Separator” condensate (stream “3”) in the mixer “Mix-101” to form stream “9” at about 5 °C (41 °F) and 3945 kPa (572.2 psia). To prepare the liquid to be fed to the deethanizer, the process specification is to raise the temperature of the NGL product stream “9A” from about -4°C (25°F) and 1535 kPa (222.6 psia) to 20 °C (68 °F) and 1500 kPa (217.6 psia) in “E-102” HX. The process duty and the temperature of the NGL product stream is set by the deethanizer process requirements. The pressure drops in “E-102” HX is 35 kPa (5 psi).

 

 

Deethanizer Specifications and Performance:

 

Like the Apr 2019 TOTM [1], the deethanizer column specifications are:

► To recover 90 mole percent of propane of the feed in the bottom product

► Methane to propane mole ration equal to 5% in the bottom’s product

► Top and bottom pressures are 1450 and 1500 kPa (210.3 and 217.6 psia), respectively

► Number of theoretical stages 12 plus the condenser and reboiler (determined by the material balance and column shortcut calculations)

 

The deethanizer simulation results are summarized in Table 4.

Table 4. Summary of deethanizer key design parameters

 

 

Impact of Gas/Gas HX Hot End Temperature Approach:

The Gas/Gas HX utilizes the cold temperature of stream 4 at -20 °C (-°4F) to cool down stream 2A. The specified temperature approach (TA) sets the sales gas temperature, which is equal to stream 2A temperature minus the TA. Decreasing the hot end TA increases the heat duty of the Gas/Gas HX, which decreases the stream 5 temperature and reduces the required chiller duty. Typically, the Gas/Gas HX removes about 70% of the total required cooling duty to meet the specified sales gas dewpoint temperature. Therefore, the chiller duty decreases resulting in lower compressor power, decreased propane refrigerant circulation rate, and reduced propane refrigerant condenser duty requirements. This tip investigates the impact of the hot end TA for the Gas/Gas HX for a range of 1 to 11 °C (1.8 to 19.8 °F) while keeping the chiller TA constant at 5 °C (9 °F).

 

Figure 3 presents the impact of TA on the MEG injection rate of streams EG1 and EG2 upstream of Gas/Gas HX and the chiller, respectively. As the hot end TA increases, stream 5 temperature increases but the stream 7 temperature remains constant at the set value; therefore, the HFT depression of stream 5 (d = HFT minus Stream 5 T) decreases and the required MEG injection rate of stream EG1 decreases. However, the MEG injection rate of stream EG2 increases because the total HFT depression temperature (d = HFT minus Stream 7 T) is constant; therefore, the total MEG injection (EG1 + EG2) rate stays the same. The calculated EG1 and EG2 were summed up and presented in Figure 3.

 

 

Figure 3. Impact of the temperature approach on the MEG injection rate upstream of Gas/Gas HX (EG1) and chiller (EG2)

Increasing TA causes a decrease in the Gas/Gas-gas HX duty, which results in an increase in the chiller duty because the total cooling duty is constant. As the chiller duty decreases, the compressor power and condenser duty decreases, too. Figures 4 A and B illustrate the impact of TA on the compressor power, Gas/Gas HX, Chiller, and condenser duty in SI and FPS system of units, respectively.

 

 

Figure 4A. Impact of the temperature approach on the Gas/Gas HX, chiller and condenser duty, and compressor power

Figure 4B. Impact of the temperature approach on the Gas/Gas HX, chiller and condenser duty, and compressor power

In order to calculate the required surface area of the heat exchangers a design factor of 1.25 and the following overall heat transfer coefficients in W/m²-°C (Btu/hr-ft²-°F) were assumed: Gas/Gas HX = 283 (50), Chiller = 444 (78), and Condenser = 460 (81).

 

Figure 5 presents the impact of TA on the required surface area of the Gas/Gas HX, chiller and the refrigerant condenser. This figure indicates that the Gas/Gas HX required area is much greater than the chiller and condenser area because it has greater heat duty. Note for TA cases of 1 and 2 °C (1.8 and 3.6 °F) the required surface areas of Gas/Gas HX are greater than 6000 m² (64,590 ft²); therefore, two HXs of 50% of the required area should be used [5].

 

 

Figures 5. Impact of the temperature approach on the Gas/Gas HX, chiller and condenser surface area

 

 

The procedure suggested by Hubbard [5] was used to perform the cost analysis. The CAPEX is the sum of the HXs and compressor costs. The HXs cost was estimated based on the size and surface area for the shell and tube types HXs. It is assumed that the maximum shell diameter is 1.97 m (6 ft) and the L/D was 10. For these dimensions, the maximum surface area per shell is about 6000 m² (64,590 ft²). For TA cases of 1 and 2 ^oC (1.8 and 3.6 ^oF), the Gas/Gas HX with an area greater than 6000 m² (64,590 ft²) two HXs with 50% required area were used. The compressor CAPEX was estimated based on the required power with a design factor of 1.25. Annualized CAPEX cost was estimated based on 5 years recovery. The annual OPEX (energy cost) was estimated based on the compressor power requirement. The total cost is the sum of the annualized CAPEX and annual OPEX (energy cost). These costs are based on a design factor of 1.25. Because total MEG injection rate and deethanizer performance are independent of TA, they were excluded in the cost analysis.

 

Figure 6 presents the impact of TA on the estimated annualized CAPEX, annual OPEX (energy cost), and the total annual cost. This figure indicates that the optimum TA for the Gas/Gas HX occurs at 3 °C (5.4 °F).

 

 

Figures 6. Impact of the temperature approach on the annual OPEX, annualized CAPEX and the total annual cost

 

---

 

SUMMARY

 

This tip demonstrated the impact of temperature approach (TA) on the trade-off between CAPEX and OPEX for the mechanical refrigeration of a HCDP control plant. The heat exchanger costs are based on rough estimates; therefore, the economic evaluation presented does not represent the actual costs. The Gas/Gas HX areas are extremely large for TA cases of 1 and 2 ^oC (1.8 and 3.6 ^oF). It is unlikely that Shell and Tube exchangers of this size could even be manufactured. The Gas/Gas HX CAPEX for the TA cases of 1 and 2 ^oC (1.8 and 3.6 ^oF) assumed two exchangers of the 50 % required surface area. If a new mechanical dewpoint facility were built today, it is probably that the Gas/Gas HX and the chiller would almost certainly be a compact heat exchanger, such as a plate-fin or printed circuit.  The compact heat exchanger designs have significantly greater surface area per volume than shell and tube heat exchangers which allows them to achieve smaller economic temperature approaches for a given capital cost.

 

The minimum cost is at 3 ºC (5.4 ºF), which is where we must go to two parallel shells. This makes sense, because here we have a step change in the configuration. When temperature correction factor, F < 0.8, we usually add another shell in series, which may also give step changes. Typically, these exchangers have two or three shells in series at the final configuration.

 

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing), G5 (Practical Computer Simulation Applications in Gas Processing) and PF42 (Separation Equipment – Sizing and Selection) courses.

 

By: Mahmood Moshfeghian, Ph.D.

---

Sign up to receive the Tip of the Month!

---

 

References:

1. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2019/04/Impact-of -liquid-carryover-on-performance-of-refrigeration-system-with-a-flash-tank-economizer/, PetroSkills -John M. Campbell Tip of the Month, April 2019.

2. Campbell, J.M., “Gas Conditioning and Processing, Volume 1: The Fundamentals,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum

3. Campbell, J.M., “Gas Conditioning and Processing, Volume 2: The Equipment Modules,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, PetroSkills 2018.

4. UniSim Design R443, Build 19153, Honeywell International Inc., 2017.

5. Hubbard, R.A., Personal Communication, March 2019.

This Tip of the Month shows how a Short Cut Method (SCM), after one Performance Test Run (PTR), may be used to estimate the life of a Type 4A molecular sieve dehydrating a water-saturated feed of natural gas.

The May 2015 Tip of the Month [1] discussed the benefits of standby time.  In that Tip (which the reader is urged to revisit), a case study was presented for a 3-tower dehydration system. The system was designed to meet a three-year life; however, a PTR after one year of service predicted the total life of the molecular sieve would only be about two years. Using the available stand-by time, the molecular sieve life was extended to about 3.7 years.

The above results were calculated used the concepts outlined in Chapter 18 of Gas Conditioning and Processing, Volume 2: The Equipment Modules (9^th Edition) [2]. Due to the non-linearity of the modeling techniques, the manual calculations are tedious. The June 2016 Tip of the Month [3] showed how a computer module developed for the PetroSkills-John M. Campbell GCAP (Gas Conditioning and Processing Software) [4] could be used to perform the same calculations. Among other things, the GCAP module directly calculates the life factor, F_L, which generates more consistent results compared to visually reading the Generic Molecular Sieve Decline Curves (Figure 1).

Figure 1. A generica molecular sieve decline curves [1]

SHORT CUT METHOD (SCM):

By making a few simplifying assumptions, manual calculations can be performed on the back of an envelope. This SCM permits an operator or a facility engineer to quickly determine the life expectancy of the molecular sieves. The assumptions, which cover the majority of natural gas dehydration units in the field, include:

1. Water-saturated natural gas is the feed

2. The feed conditions remain relatively constant throughout the life of the molecular sieve system

3. Type 4A-1/8 inch (3 mm) pellets or 4×8 beads are used

4. The fresh equilibrium loading is 23 weight percent water

5. The residual loading is 4 weight percent water

6. 5 % of the bed weight is devoted to the Mass Transfer Zone (MTZ)

7. Normal life degradation following the curve shapes in Figure 1. If upsets such as liquid carryover, bed lifting, bed support failure, valve hang-ups, contaminants in the regeneration gas, flow channeling or other adverse conditions occur, the shape of the curves in Figure 1 will be quite different.

Chapter 18 of Gas Conditioning and Processing, Volume 2: The Equipment Modules (9^th Edition) [2] contains equations that permit us to calculate the total mass of the molecular sieve, the Break Through Loading (BTL, or Useful Loading), and the aged net equilibrium loading. Using the assumptions listed above together with the equations in Chapter 18 it can be shown that:

F_L = BTL/18                                                                                                    (1)

where:

F_L        = Life factor

BTL      = (100)(mass of water removed/mass of molecular sieve)                 (2)

The remainder of this Tip of the Month will compare the results of the SCM to those obtained from the rigorous manual method and the computer-generated method.

SCM vs RIGOROUS METHOD:

Figure 2 shows the Process Flow Diagram used for the Case Study [2]. Tables 1 gives the Design Basis.  Table 2 presents the Design Summary. Table 3 shows the Results of the PTR after one year of operation (note the feed flow rate and the temperature during the PTR are slightly different than the design basis).

Figure 2. Typical process flow diagram for a 3-tower adsorption dehydration system [2]

Table 1. Design basis for the case study

Table 2. Design Summary for the case study

Table 3. Results of Performance Test Run (PTR) after 12 months of operation

Additional information used in the Case Study:

• 3 tower system (2 towers on adsorption, 1 on regeneration)
• External Insulation
• Tower ID = 2.9 m (9.5 ft)
• Each tower contains 24630 kg [54300 lb_m] of Type 4A 4×8 mesh beads and is designed to last three years.
• Regeneration circuit capable of handling an extra 15% of flow
• Unit is operated on fixed time cycles
• No step-change events such as liquid carryover, poor flow distribution, etc.

Following is a recipe for using the SCM:

1. Use Equation 2 to calculate the design BTL = 10.6 wt %

• • BTL =100 (16 h)(163 kg water removed/h)/(24630 kg mol sieve) =10.6%
  • BTL =100 (16 hr)(360 lb_m water removed/hr)/(54300 lb_m mol sieve) =10.6%

2. Use Equation 1 to calculate the design F_L = 10.6/18 = 0.59

3. Locate the design F_L at 1095 cycles on Figure 3, Calculated Life Factors (F_L=0.59 & 3 years of 24-hour cycles per tower is equivalent to 1095 cycles).  This is the Design F_L Point.

4. Use Equation 2 to calculate the PTR BTL = 12.0 wt %

• • BTL =100 (20.9 h)(141 kg water removed/h)/(24630 kg mol sieve) =12%
  • BTL =100 (20.9 hr)(312 lb_m water removed/hr)/(54300 lb_m mol sieve) =12%

5. Use Equation 1 to calculate the PTR F_L = 12/18 = 0.67

6. Locate the PTR F_L at 365 cycles on Figure3, Calculated Life Factors (F_L=0.67 & one year of 24-hour cycles per tower is equivalent to 365 cycles).  This is the PTR F_L Point.

7. Because the PTR F_L Point falls on a curve lower than the Design F_L Point, we need to be concerned.  Interpolating and extrapolating the capacity decline curve from the PTR F_L Point_, we see an F_L of 0.67 (the Design F_L) will occur after a total of about 750 cycles. This is approximately one year shorter than the expected life.

Figure 3. Calculated Life Factors

Because the unit has a regeneration circuit that can handle an additional 15% of flow, the complete regeneration cycle (heating, cooling, de- and re- pressurization) can be reduced to 7.0 hours.  This allows the beds to turn around faster. Using the reduced cycle time (the complete cycle time is now 21 hours vs the original 24 hours) and the original design basis conditions with the SCM recipe:

1. Calculate End of Life BTL = 9.3 wt % (from Equation 2).

• • BTL =100 (14 h)(163 kg water removed/h)/(24630 kg mol sieve) =9.3%
  • BTL =100 (14 hr)(360 lb_m water removed/hr)/(54300 lb_m mol sieve) =9.3%

2. Calculate End of Life F_L = 9.3/18 = 0.52 (from Equation 1). This is because less water is being adsorbed per cycle.

3. Interpolating and extrapolating from the PTR F_L Point, we find the End of Life F_L of 0.52 occurs around 1400 cycles (see Figure 4).  Since 365 cycles have already occurred and going forward a reduced cycle time will be used, the molecular sieves are forecast to last a total of 3.5 years.

Figure 4. Calculated Life Factors Using Standby Time

Table 4 compares the results of the SCM to the rigorous manual calculations (May 2015 Tip of the Month) and the computer-generated calculations (June 2016 Tip of the Month).

Table 4. Comparison of Three Methods

The difference between the predicted life using standby time shown by the Computer Model and the two manual methods is primarily due to the inherent inaccuracy of trying to interpolate and extrapolate data plotted on Figures 3 and 4.  The computer model will produce the same result every time.  This cannot be said when visually reading Figures 3 and 4.   Figure 5 shows the output from the GCAP Computer Model [3]. Note that when working on one curve, the higher the calculated EOL F_L, the fewer the Number of Cycles (NOC) remaining until the beds need to be replaced.

The PTR in the above Case Study was run after 365 cycles.  The slope of the curve is fairly steep in this region and small changes in data can have a significant impact on the life predictions.  While the user can get a good indication of the state of decline of their molecular sieve unit, scheduling additional PTR’s is highly recommended. Finally, generic curves are used in these Figures. The shape of your specific molecular sieve capacity decline curve may differ from these generic curves.  Despite these caveats, the SCM offers the user a quick and easy way to assess the capacity decline of their molecular sieve unit.

Figure 5. Projected life factor (L_F = 54.3% and NOC = 1251.4) if standby time is used

SUMMARY:

We can draw the following conclusions from this case study:

• The short-cut method presented allows the user to quickly estimate the decline of their adsorbent based on only one performance test run for molecular sieve dehydrators using low-pressure regeneration. This permits the early formulation of a credible action plan. The short-cut method compares reasonably well to the rigorous manual approach and the computer-generated model which also require only one performance test run.
• Both manual methods rely on a visual interpolation and extrapolation of the generic molecular sieve decline curves. The computer-generated approach provides much more consistent results.
• All the methods presented in these Tips of the Month rely on open-art technology. The molecular sieve vendors use proprietary methods specific to their manufacturing techniques. Consequently, the results of the approaches presented in these Tips of the Month should be used to generate trends as opposed to absolute values.
• Site-specific factors will determine your unit’s decline curve.  Conducting more than one performance test is highly recommended.
• Standby time offers a large degree of operating flexibility because the decline curves tend to level off; always try to build in standby time in any new molecular sieve design.
• Adsorption capacity is a function of the number of cycles, not calendar time.
• Install a good filter coalescer or filter separator upstream of your adsorption unit to keep the contaminants out of the system.

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing) and PF-4 (Oil Production and Processing Facilities) courses.

Written by: Harvey H. Malino, P.E.

---

References

1. Malino, H.M., http://www.jmcampbell.com/tip-of-the-month/2015/05/benefits-of-standby-time-in-adsorption-dehydration-process/, PetroSkills – John M. Campbell, 2015

2. Campbell, J.M., “Gas Conditioning and Processing, Volume 2: The Equipment Modules,” 9^th Edition, 3^nd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, 2018.

3. Moshfeghian, M.  http://www.jmcampbell.com/tip-of-the-month/2016/06/projecting-the-performance-of-adsorption-dehydration-process/, PetroSkills – John M. Campbell, 2016

4. GCAP 9.2.1 Software, PetroSkills – John M Campbell “Gas Conditioning and Processing Computer Program,” Editor Moshfeghian, M., PetroSkills, Katy, Texas, 2016.

Problems in meeting sales-gas dew point specifications are not unusual in plants. A facility engineer often suspects separator carryover when troubleshooting such a plant.  Proper sizing of equipment for gas-liquid separation is essential to almost all processes. Many facility operating problems are related to improperly designed or under-sized gas-liquid separators. The following list presents items that can contribute to too much liquid (carryover) in the gas stream.

 

►The mist extractor operating K_s value is greater than the design value.

►The velocity profile through the mist extractor is poor, resulting in localized high velocities/flooding.

►The droplet sizes reaching the mist extractor are too small.

►The entrained liquid load reaching the mist extractor is too high.

►The mist extractor is damaged or plugged.

►Level control and instrumentation malfunction or failure

►Foaming

 

Continuing the December 2005, January and February 2019 [1, 2, 3] Tips of The Month (TOTM), this tip investigates the impact of the liquid carryover (LCO) on the performance of a mechanical refrigeration plant with mono-ethylene glycol (EG or MEG) injection for hydrocarbon dew point (HCDP) control. Specifically, the impact of LCO on the gas-gas heat exchanger and chiller duties, the mechanical refrigeration system, and the liquid propane recovery will be investigated and reported.

 

The details of a mechanical refrigeration plant with MEG injection and regeneration system are given in Chapters 6, 15 and 18 of the Gas Conditioning and Processing, Volumes 1 and 2 [4, 5], respectively. In addition, how to minimize the liquid carry in separation equipment are discussed in PetroSkills-John M. Campbell course titled, “PF42 – Separation Equipment – Sizing and Selection.”

 

Figure 1 presents the process flow diagrams for a typical HCDP control plant using mechanical refrigeration with MEG injection system. This figure is similar to the February 2019 TOTM [3] with the exception that the refrigeration system utilizes a flash tank economizer with two stages of compression. In this tip, all simulations were performed with UniSim Design R443 software [6] using the Peng-Robinson equation of state.

 

 

 

 

CASE STUDY:

Let’s consider the same case presented in February 2019 TOTM [3] for a rich gas with the compositions and conditions presented in Table 1 [3]. Based on the reported molecular weight and relative density for the C7+ fraction, Table 2 presents the estimated normal boiling point (NBP), critical properties and acentric factor which are needed by the equation of state. The objective is to meet a hydrocarbon dew point specification of  -20 °C (-4 °F) at about 4000 kPa (580 psia) for the sales gas by removing heat in the “Gas/Gas” heat exchanger (HX) with a hot end approach temperature of 5°C (9°F)  and in a propane chiller, 5 °C (-4 °F) approach temperature, and rejecting it to the environment by a propane condenser (AC-100) at 37.8°C (100°F). Pure propane is used as the working fluid in the simulation. The pressure drops in the “Gas/Gas” HX and the propane chiller are assumed to be 34.5 kPa (5 psi).

 

 

Table 1. Rich feed gas compositions and conditions

 

 

Table 2. Estimated C7+ properties [4]

 

 

The feed gas is flashed in the “Inlet Separator” at 30 °C (86 °F) and 4000 kPa (580 psia) to remove any condensate. The “Inlet Separator” vapor (stream 2) is saturated with water by the “Saturate -100” to form stream “2 Wet” upstream of mixing with MEG hydrate inhibitor, stream “EG1” and the recycle stream “18A” from the deethanizer overhead vapor (located at the right-hand side of Fig. 1).

 

The estimated hydrate formation temperature of streams “2 Wet” is 14.7 °C (58.4 °F). The hydrate inhibitor is injected at the inlet of “Gas/Gas” HX by stream “EG1” and at the inlet of the “Chiller” by stream “EG2”. Stream “5” cools to about -8 °C (17.6 °F) and stream “7” cools down to the specified temperature of -20 °C (-4 °F) which are below the hydrate formation temperature (HFT) of 14.7 °C (58.4 °F). The injection rates of streams “EG1” and “EG2” for 80 weight % lean MEG and water solution are estimated by the Adjust tool of UniSim. A design margin of 1 °C (1.8 °F) HFT below the cold temperature for streams “5” and “7” were assumed.

 

Assuming an approach temperature of 5°C (9°F) and a 6.9 kPa (1 psi) pressure drop in the propane chiller (“Chiller”) shell side, the pressure of saturated propane vapor leaving the chiller is 203.3 kPa (29.5 psia), and at a temperature of -25°C (-13°F).  Assuming no frictional losses in the suction line to the propane compressors “K-101” and “K-102”, the resulting suction pressure is 203.3 kPa (29.5 psia).

 

The condensing propane pressure at the specified condenser temperature of 37.8 °C (100 °F) is 1303 kPa (189 psi). The condenser “AC-100” frictional losses, plus the frictional losses in the piping from the compressor discharge to the condenser were assumed to be 34.5 kPa (5 psi); therefore, the discharge pressure of compressor “K-102” is 1338 kPa (194 psia). The compressors inter stage pressure wasdetermined by equalizing the power for “K-101” and “K-102.” The compressors adiabatic efficiency was assumed to be 75%.

 

The cold Stream 7 is flashed in the 3-phase separator “V-102” at -20 °C (-°4F) and 3931 kPa (570 psia). The vapor stream “4” from this cold separator is used to cool down the incoming warm feed gas in the “Gas/Gas” HX. The heavy liquid stream “8B” (rich MEG solution) from the cold separator is regenerated in the regeneration unit (not shown in Fig. 1) and the lean 80 weight % MEG is recycled and used in streams “EG1” and “EG2”. The cold NGL stream “8” (light liquid phase) from the cold separator, “V-102”, is combined with the plant “Inlet Separator” condensate (stream “3”) in the mixer “Mix-101” to form stream “9” at about 5 °C (41 °F) and 3945 kPa (572.2 psia). To prepare the liquid to be fed to the deethanizer, the process specification is to raise the temperature of the NGL product stream “9A” from about -4°C (25°F) and 1535 kPa (222.6 psia) to 20 °C (68 °F) and 1500 kPa (217.6 psia) in “E-102” HX. The process duty and the temperature of the NGL product stream is set by the deethanizer process requirements. The pressure drops in “E-102” HX is 35 kPa (5 psi).

 

 

DEETHANIZER SPECIFICATIONS and PERFORMANCE:

Like the February 2019 TOTM [3], the deethanizer column specifications are:

A. To recover 90 mole percent of propane of the feed in the bottom product and

B. Ethane to propane mole ratio equal to 5 % in the bottoms product

C. Top and bottom pressures are 1450 and 1500 kPa (210.3 and 217.6 psia); respectively

D. Number of theoretical stages 12 plus the condenser and reboiler (determined by the material balance and column shortcut calculations)

The deethanizer simulation results are summarized in Table 3.

 

 

Table 3. Summary of deethanizer key design parameters

 

 

IMPACT OF LIQUID HYDROCARBON CARRYOVER:

Separator “V-102” is a three-phase separator. Under ideal condition the vapor (stream 4) leaving the separator has no LCO and its dewpoint temperature is the same as the feed (stream 7) temperature.Typical range of liquid carry over is 0.013–0.27 m³ liquid/10⁶ std m³ of gas (0.1–2 gallon of liquid/MMscf) [5]. In practice due to the reasons listed in the preceding section the LCO can be even higher. In this tip, the impact of LCO was investigated for a range of 0 to 3 mole % of liquid in light liquid phase (liquid hydrocarbon phase) entrained into the gas phase. The entrained liquid consists of heavier molecules causing the dewpoint temperature of stream 4 and sales gas to go up and make it off spec. To offset the effect of LCO and bring back the sales gas dewpoint temperature to spec, the operators typically lower the chilling temperature of feed (stream 7) to separator (“V-102”). This is possible if the mechanical refrigeration system is capable of handling a higher chilling load.

 

Figure 2 presents the hydrocarbon dewpoint curves as a function of the liquid hydrocarbon carryover (CO). the cricondentherm points shift to the right as LCO increases. The bubble point curves are not presented because the LCO has negligible effect on the bubble curves. All phase envelopes are generated on the drybasis.

 

 

Figure 2. Impact liquid carryover on the sales gas hydrocarbon dewpoint temperature

 

 

Figure 3 presents the impact of LCO on the sales gas dewpoint temperature and the required cold separator feed (chilling) temperature to offset the LCO. As the LCO increases the chiller temperature should be decreased to meet the sales gas dewpoint spec of -20 °C (-4 °F). For 3 mole % LCO, the sales gas dewpoint temperature is -14.4 °C (6.1 °F). To bring back the sales gas dewpoint temperature, the process gas (stream 7) should be cooled to -28.6 °C (-19.5 °F).

 

As the chiller temperature decreases, to counter the effect of the LCO, the hydrate formation temperature depression of streams 5 and 7 increases, which requires a higher MEG injection rate. Figure 5 presents the impact of LCO on the rate of streams EG1 and EG2 upstream of Gas/Gas HX and the chiller, respectively. Note the required inhibitor injection rate for stream EG2 upstream of the chiller increases considerably with the increase in LCO.

 

If there is carryover of the hydrocarbon phase, there is also a likelihood of carryover of the glycol phase.  This can result in problems meeting the water dewpoint specification and also introduces a deleterious substance into the sales gas (MEG) which may also not be allowed in the sales gas contract.

 

 

Figure 3. Impact of liquid carryover on the sales gas dewpoint temperature (solid line) and the required cold separator feed temperature (dashed line) to offset liquid carryover

 

 

Figure 4. Impact of liquid carryover on the MEG injection rate upstream of Gas/Gas HX (EG1) and chiller (EG2)

 

Lowering the chiller temperature to counter the effect of LCO also cause an increase in the compressor power, Gas/Gas-gas HX, chiller and condenser duties. Figures 5 A and B illustrate the impact of LCO on the compressor power and Gas-Gas HX, Chiller, and condenser duty in SI and FPS system of units, respectively.

 

 

Figure 5A. Impact of liquid carryover on the compressor power, Gas-Gas HX, chiller, and condenser duty

 

 

Figure 5B. Impact of liquid carryover on the compressor power, Gas/Gas HX, chiller, and condenser duty

 

Figure 6 presents the impact of LCO on the liquid propane and sales gas recoveries. This figure indicates that as the LCO increases from 0 to 3 mole %, the liquid propane recoveries increase from about 17% to 27% on mole basis but the sales gas recovery decreases slightly from about 97 % to 96 % on mole basis. The extra propane liquid recovery is achieved by operating the chiller at a lower temperature which requires higher OPEX and CAPEX.

 

As the chiller temperature is reduced, more ethane and methane end up in the low temperature separator (LTS), V-102, liquids.  These need to get boiled out in the deethanizer, so the duty of E-102 increases, reboiler and condenser duty of the deethanizer increase, and the recompression power in K-100 alsoincreases. It may also be possible to flood the deethanizer.

 

The cold condenser on the deethanizer requires propane for cooling. These units are also designed with no condenser. The cold LTS (V-102) liquid is used as reflux, and the liquids from the inlet separator are introduced lower in the column. V-100 is not required then. E-102 is usually a feed/bottoms heat exchanger. Overhead of the deethanizer can likely go to the sales gas and may not have to be recycled.  It really should not contain anything heavier than propane, heavy key (HK).

 

Table 4 presents the impact of LCO on the key equipment incremental capacity requirement to meet the sales gas hydrocarbon dewpoint temperature by lowering the chiller temperature. Assume the system was built with a design margin factor of 1.25. Table 4 indicates that this system can handle up to one mole % LCO with higher OPEX. However, for more than one mole % LCO, the system cannot lower chiller temperature enough to meet the sales gas dewpoint temperature. As shown in Table 4, the compressor power, hydrate inhibition rate, condenser and chiller duties are the limiting factors. Under such condition it may require plant shutdown for trouble shooting to reduce the LCO.

 

Table 4. Estimate of equipment incremental capacity requirement to handle liquid carryover

 

 

SUMMARY:

The common practice to meet sales gas hydrocarbon dewpoint temperature under the condition of liquidcarry is to operate the chiller at a temperature below the sales gas hydrocarbon dewpoint spec. This is only possible if the key equipment can handle the extra load with higher OPEX. This tip demonstrated the impact of varying the LCO from 0 to 3 % on a mole basis on the process stream rates, phase behavior, the equipment sizes and the refrigeration requirement.

 

As demonstrated in this tip, it would be a good practice to size the equipment with a design margin of 1.2 to 1.3 to consider the changes in operation conditions and the liquid carryover. Most important to minimize LCO is to have a properly designed separator with good feed pipe, inlet device, mist extractor, gas gravity separation and liquid gravity separation sections.

 

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing), G5 (Practical Computer Simulation Applications in Gas Processing), and PF42 (Separation Equipment – Sizing and Selection) courses.

By: Dr. Mahmood Moshfeghian

 

---

Sign up to receive monthly Tip of the Month emails!

---

References:

1. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2005/12/impact-of-liquid-carry-over-on-sales-gas-dew-point/, PetroSkills -John M. Campbell Tip of the Month, December 2005.

2. Moshfeghian, M., https://www.petroskills.com/blog/entry/00_totm/jan19-fac-optimizing-performance-of-refrigeration-systems-with-an-external-sub-cool-economizer, PetroSkills -John M. Campbell Tip of the Month, January 2019.

3. Moshfeghian, M., https://www.petroskills.com/blog/entry/00_totm/feb19-fac-impact-of-heavy-end-on-the-performance-of-a-mechanical-refrigeration-plant-with-meg-injection, PetroSkills -John M. Campbell Tip of the Month, February 2019.

4. Campbell, J.M., “Gas Conditioning and Processing, Volume 1: The Fundamentals,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum

5. Campbell, J.M., “Gas Conditioning and Processing, Volume 2: The Equipment Modules,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, PetroSkills 2018.

6. UniSim Design R443, Build 19153, Honeywell International Inc., 2017.

Continuing the Aug 2018 Tip of the Month (TOTM) ¹ on design and operation of unconventional surface facilities, this TOTM presents process safety tips for four case – studies:

1. Direct Fired Heater Treater Burn Through Failures
2. Tank Blanket Gas / Flame Arrestors
3. Pocketing Vent / Relief Piping
4. Hot Oiling of Oil Storage Tanks to meet TVP / RVP

 

We start this tip with a quote from a colleague, James A Britch: “I never regretted buying quality.” There is a lesson in there for unconventional batteries. There is tremendous pressure to reduce capital costs, but you should be focused on life cycle cost.

If you install equipment, and then burn down the battery…you haven’t saved much.

There’s the loss of capital, and loss of revenue from shut-in production.

 

---

 

PROCESS SAFETY CASE STUDY 1: DIRECT FIRED HEATER TREATER BURN THROUGH FAILURES

 

Direct fired burners are failing due to internal flame impingement directly on the steel, and salt build up on the outside of the firetube in the process fluids. The salts build up, act as an insulator, and then steel temperature increases until a burn through occurs. Since the process side of the burner operates at a higher pressure than the natural draft burner, the process fluids enter the burner, and ignite. In many instances this has resulted in massive damage to the entire battery.

 

 

Figure 1 illustrates the location of burn through failure that occurs in the 12 o’clock portion of the burner firetube (combustion leg).

 

Solutions to this issue are to use:

1. Indirect fired heater where the process fluid flows through coils in a larger shell surrounded by a heat transfer fluid. The firetube burner is immersed in the heat transfer fluid within the same shell.
   • This solves the salt build up failure mode, but not the flame impingement failure.
   • Some operators use glycol as a heat transfer fluid. This is a cheaper alternative to expensive heat transfer fluids, but glycols degrade to acetic acid if the skin temperatures of the fire tubes exceed 350°F (177°C). These same operators rarely check the pH of their heat transfer fluid until leaks and severe corrosion are found. Heat transfer oils are a better choice.
2. Direct fired heater with a ceramic sleeve to take the higher temperature shock of direct flame impingement. Heaters normally have ceramic refractory castables and bricks to prevent direct flame impingement. As shown in Figures 2 and 3 Bartz et al. ² recommends inserting a ceramic tube to spread the heat flux of the flame and avoid burn through.

 

 

 

 

 

 

1. Best Solution: consider a separate furnace and heat exchanger using heat transfer oils. This solves both issues. The heat transfer oils are designed to operate without degradation at process temperatures required, and the metal temperatures in the heat exchanger will not cause metal failure if there are no salt deposits on the heat exchanger. This may require periodic hydro-blasting, but you will not have to rebuild the battery.
2. Think Reliability has a free excellent Root Cause Analysis tool in Excel (Figure 4)

 

 

 

 

---

 

PROCESS SAFETY CASE STUDY 2: TANK BLANKET GAS / FLAME ARRESTORS ³

 

Flame arrestors and tank blanket gas provide independent layers of protection between the ignition source – flare or thermal oxidizer – and the vapor space in the water and oil tanks. When evaluating what to use in a design consider using a layer of protection analysis or LOPA ³. This tool is discussed in the PetroSkills | John M Campbell PS4 – Process Safety Course. It provides a semi-quantitative solution to design decisions that are based of failure frequencies and not just personal preferences or gut feels. The more independent layers of protection the lower the frequency of the consequence occurring.

 

f= (IEF) x PFD₁ x PFD₂….

 

Where:

f            = Frequency of the consequence occurring for the scenario

IEF      = Frequency of the initiating event

PFD     = Probability of the failure on demand for an independent layer of protection. For example, the probability that a relief valve will not operate as intended.

 

Flame arrestors have a Probability of Failure on Demand (PFD) with a range of 1×10^-1 for arrestors without temperature indicators and an effective isolation / shutdown system, and tank blanket gas (BPCS-Basic Process Control System) has a PFD of 1×10^-1 [3]. The designer can use both or either to provide independent layers of safety protection. Flame arrestors are subject to plugging from ice, corrosion, fouling, improper or lack of maintenance. Blanket gas works well in this situation due to the narrow range of flammability of methane in air (5-15% fuel to air). Majority of stock tank incidents occur during maintenance activities with small amounts of gas and large volumes of air.

 

During the high volume of production timeframe for unconventional tank batteries the stock tanks degas and have “auto-blanketing” of the active tank vapor space. But what happens in the future when rates are very low? What happens in water tanks that are not provided with gas blanketing? This also explains why water production tanks / injection batteries experience tank fires/explosions. In general, methane has a very low solubility in water – approximately 2 SCF/STB (0.36 Sm³/STm³) of water going from 250 psi (1724 kPa) to atmospheric pressure. This small volume often results in flammable mixtures in the vapor space of the tank.

 

It is a good practice to select tank blanketing as your first line of defense to prevent internal tank corrosion and gas plant amines and TEG process corrosion/solution degradation by keeping oxygen out of the system, and internal explosions from flash back from flares and thermal oxidizers.

 

---

 

PROCESS SAFETY CASE STUDY 3: POCKETING VENT / RELIEF PIPING

 

Many unconventional tank batteries run their vent / flare / thermal oxidizer piping at grade on sleepers with zero slope, and then jump vertically into a flare knockout (see Figure 5). This pocketed piping is a liquid trap for water and heavier hydrocarbons. Once a liquid pocket forms, the tanks overpressure, and then vent locally through their pressure/vacuum reliefs and thief hatches. In the winter this pocket can freeze and block the flare (see Figure 6). This can also cause a loss of containment when a PSV activates and cannot depressure to the flare. This causes a loss of revenue, as well as an environmental and safety issue. These vapors are extremely rich and normally are much heavier than air. This creates the potential for an unconfined vapor cloud explosion or flash fire locally.

 

 

 

 

 

 

As illustrated in Figure 7, slope your vent/relief piping toward a lower elevation knock out or burn pit.

 

 

 

API Standard 521 ⁴, pressure-relieving and depressuring systems, requires that the flare piping be free draining to the flare knock out drum, and then free draining from the flare back to the flare knock out.

 

These issues causing tanks to vent heavier than air molecules (propane/butane) can lead to flash fires and unconfined vapor cloud explosions (UCVE). Heavy vapor generally finds an ignition source. Figure 8 shows how oxygen may get into the oil stock tanks. Getting oxygen into your system causes major damage to the gas plant amine systems, and TEG systems, as well as general corrosion in your facilities.

 

 

 

---

 

PROCESS SAFETY CASE STUDY 4: HOT OILING OIL STORAGE TANKS

 

As illustrated in Figure 9, some operators use hot oil trucks during winter months to heat the crude oil in the tanks to flash light ends off the crude to meet vapor pressure specifications for crude sales. The solution to this issue is not to use hot oil trucks but is to stabilize the crude or use a design using Vapor Recovery Towers (VRT) as discussed in the August 2018 Tip of the Month- Design and Operation of Unconventional Surface Facilities Issues-Stabilization ¹.

 

 

 

 

Many operators are faced with large numbers of tank batteries in the hundreds or thousands spread over a large geographic area. Many rely on unsupervised contractors to conduct hot oiling operations, oil & water loading and unloading, and other maintenance operations.

• Do you have an operator present at the location to help with hot oiling?

• Does the lease operator visit the site with the contractor to issue a Hot Work Permit and JSA?

• Do you have operating guidelines or checklists for Hot Oil Operations?

• Do they include monitoring of weather conditions? Wind Speed?
  • Shutdown for low or no wind conditions?

• Do they include stopping of other operations such as oil / water truck loading?

• Are parts of the lease blocked off to prevent other vehicles entering as ignition sources?

• Are the hot oil trucks placed up wind of the tanks?

• Are the hot oil trucks 100 ft or 50 ft from the tanks? What’s your company’s design spacing requirements for ignition sources / direct fired equipment and oil storage tanks?

• Have you conducted a HAZOP for Hot Oiling Operations? Many accidents happen during non- routine operations.

• Do you have gas detectors? Normally the contractor has a portable. Is that effective? You have RVP issues because it’s winter and cold. Where will the contractor be? In the truck… …where it’s warm.

 

The hot oil truck is a direct fired (propane) heater with propane storage, and diesel or oil storage. It is normally used to pump hot oil at high pressure down the well’s tubing to melt wax deposits. This operation is normally done for 24 hours/day during winter months to stabilize the crude. It is extremely dangerous, and many flash fires have occurred in the past few years.

• February 2, 2018 Explosion-Flaming Truck ignites Tank Battery-worker airlifted with serious burns

 

These are just some recent examples, but unfortunately there are many, many more…

• August 24, 2018 Fire Crews Battle Tank Battery Fire
• January 25, 2019 Officials identify victims of fire on FM 1788 South
• January 2, 2019 Oil tanks in West Texas catch fire

 

Heater Treater Fire: https://www.youtube.com/watch?v=XVgg0cZ7H0g

Many Tank Battery Fires are occurring in areas with new unconventional plays like the Bakken and West Texas.

 

---

 

SUMMARY AND CONCLUSIONS:

 

In this tip of the month, we have identified process safety risks with some designs, solutions to the issues, and evidence that the problem exists.

So my question for you… …after reading this tip, what action do you take to improve the safety of your designs and operations at your company? Your company and colleagues need you to take action.

• Are your designs safe?
• Are your operations safe?

Stay Safe! Let us know if you have any questions.

 

---

 

To learn more about similar cases and how to minimize operational problems, we suggest attending our G4 (Gas Conditioning and Processing), PF3 (Concept Selection and Specification of Production Facilities in Field Development Projects), PF4 (Oil Production and Processing Facilities), PF49 (Troubleshooting Oil & Gas Processing Facilities), and PS4 (Process Safety Engineering) courses.

By: James F Langer, P.E.

 

References:

1. Langer, J.F., Design and Operation of Unconventional Surface Facilities Stabilization Issues, PetroSkills-John M. Campbell Tip of the Month, January 2018.
2. “SPE 166261 Distributed-Flux Burners Improve Life of Firetubes and Process Throughput in Heater Treaters”, David Bartz, Michael Silberstein, James Gotterba; ALZETA Corporation, 2013
3. “Layer of Protection Analysis-Simplified Process Risk Analysis”, Center for Chemical Process Safety-CCPS, 2001, AIChE, Table 5.2 In-line deflagration arrestor
4. API Standard 521, Pressure-relieving and Depressuring Systems. 6^th Edition, Jan 2014

Continuing the January 2019 [1] Tip of The Month (TOTM), this tip investigates the impact the heavy end characterizations on the performance of a mechanical refrigeration plant with mono-ethylene glycol (EG or MEG) injection for hydrocarbon dew point (HCDP) control. Specifically, the impact of heavy end characterization on the gas-gas heat exchanger and chiller duties, the mechanical refrigeration system, and the liquid propane recovery will be investigated and reported. The details of a mechanical refrigeration plant with MEG injection and regeneration system are given in Chapters 6 and 15 of the Gas Conditioning and Processing, Volumes 1 and 2 [2, 3], respectively.

 

Figure 1 presents the process flow diagrams for a typical mechanical refrigeration plant with MEG injection system. In this tip, all simulations were performed with UniSim Design R443 software [4] using the Peng-Robinson equation of state.

 

 

 

Figure 1. Process flow diagrams for a mechanical refrigeration plant using a sub-cool economizer and MEG Injection system

 

 

CASE STUDY:

Let’s consider a rich gas with the compositions and conditions presented in Table 1. Based on the reported molecular weight and relative density for the C7+ fraction, Table 2 presents the estimated normal boiling point (NBP), critical properties and acentric factor which are needed by the equation of state. The objective is to meet a hydrocarbon dew point specification of  -20 °C [-4°F] at about 4000 kPa (580 psia) for the sales gas by removing heat in the “Gas/Gas” heat exchanger (HX) with a hot end approach temperature of 5°C [9°F]  and in a propane chiller and rejecting it to the environment by a propane condenser (“E-103”) at 37.8°C [100°F]. Pure propane is used as the working fluid in the simulation. The pressure drops in the “Gas/Gas” HX and the propane chiller are assumed to be 34.5 kPa (5 psi).

 

Table 1. Rich feed gas compositions and conditions

 

 

Table 2. Estimated C7+ properties [4]

 

 

The feed gas is flashed in the “Inlet Separator” at 30 °C (86 °F) and 4000 kPa (580 psia) to remove any condensate. The “Inlet Separator” vapor (stream “2”) is saturated with water by the “Saturate -100” to form stream “2 Wet” upstream of mixing with MEG hydrate inhibitor, stream “EG1” and the recycle stream “18A” from the deethanizer overhead vapor (located at the right hand side of Fig. 1). The estimated hydrate formation temperature of streams “2 Wet” is 14.7 °C (58.4 °F). The hydrate inhibitor is injected at the inlet of “Gas/Gas” HX by stream “EG1” and at the inlet of the “Chiller” by stream “EG2”. Stream “5” cools to about -8 °C (17.6 °F) and stream “7” cools down to the specified temperature of -20 °C (-4 °F) which are below the hydrate formation temperature (HFT) of 14.7 °C (58.4 °F). The injection rates of streams “EG1” and “EG2” for 80 weight % lean MEG and water solution are estimated by the Adjust tool of UniSim. A design margin of 1.1 °C (2 °F) HFT below the cold temperature for streams “5” and “7” were assumed.  Table 3 presents the estimated hydrate inhibition injection rates.

 

 

Table 3. Estimated 80 weight % lean MEG hydrate inhibition injection rates

 

 

Assuming an approach temperature of 5°C  (9°F) and a 6.9 kPa (1 psi) pressure drop in the propane chiller (“RefChiller”) shell side, the pressure of saturated propane vapor leaving the chiller is 203.3 kPa (29.5psia), and at a temperature of -25°C (-13°F).  Assuming no frictional losses in the suction line to the propane compressor “K-101”, the resulting suction pressure is 203.3 kPa (29.5 psia).

 

The condensing propane pressure at the specified condenser temperature of 37.8 °C (100 °F) is 1303 kPa (189 psi). The condenser “E-103” frictional losses, plus the frictional losses in the piping from the compressor discharge to the condenser was assumed to be 34.5 kPa (5 psi); therefore, the compressor discharge pressure is 1338 kPa (194 psia). The propane compressor adiabatic efficiency was assumed to be 75%.

 

 

External Sub-Cool Economizer:

The cold Stream 7 is flashed in the 3-phase separator “V-102” at -20 °C (-°4F) and 3931 kPa (570 psia). The vapor stream “4” from this cold separator is used to cool down the incoming warm feed gas in the “Gas/Gas” HX. The heavy liquid stream “8B” (rich MEG solution) from the cold separator is regenerated in the regeneration unit (not shown in Fig. 1) and the lean 80 weight % MEG is recycled and used in streams “EG1” and “EG2”. The cold NGL stream “8” (light liquid phase) from the cold separator, “V-102”, is combined with the plant “Inlet Separator” condensate (stream “3”) in the mixer “Mix-101” to form stream “9” at about 5 °C (41 °F) and 3945 kPa (572.2 psia). To prepare the liquid to be fed to the deethanizer, the process specification is to raise the temperature of the NGL product stream “9A” from about -4°C (25°F) and 1535 kPa (222.6 psia) to 20 °C (68 °F) and 1500 kPa (217.6 psia) in “E-102” HX. The required heat duty will be supplied from a propane refrigerant sub-cool economizer “E-104” HX. The process duty and the temperature of the NGL product stream is set by the deethanizer process requirements, thus the sub-cool economizer duty is fixed.

 

The sub-cool economizer cools the condensed propane (refrigerant stream “R4”) from 37.8°C (100 °F) at 1303 kPa (189 psia) to a cooler temperature at 1269 kPa (184 psia), depending upon the specified propane refrigerant flow rate (stream “R5”). The pressure drops in “E-102” and “E-104” HXs are 35 kPa (5 psi); respectively. The heat removed by the sub-cool economizer is fixed by the process duty required to heat the NGL process stream “9A”.

 

 

Deethanizer Specifications and Performance:

The deethanizer column specifications are:

►To recover 90 mole percent of propane of the feed in the bottom product and

►Ethane to propane mole ratio equal to 5 % in the bottoms product

►Top and bottom pressures are 1450 and 1500 kPa (210.3 and 217.6 psia); respectively

►Number of theoretical stages 12 plus the condenser and reboiler (determined by the material balance and column shortcut calculations)

The deethanizer simulation results are summarized in Table 4.

 

 

Table 4. Summary of deethanizer  key design parameters for C7+

 

 

Impact of Heavy End Characterization:

Figure 2 presents the phase envelopes for the key streams of feed (“Dry Feed”), inlet separator vapor (stream “2”) and sales gas (stream “4”). All phase envelopes are generated on the dry basis. As expected the bubble point curves are very close to each other but large deviations are observed for the dewpoint curves. Similar diagrams for the nC7 and nC8 as the heavy end  are presented in the Appendix in Figures 1A and 2A; respectively.

 

 

Figure 2. Phase diagrams for the key streams for the case of C7+ as the heavy end

 

 

Figures 3, 4, and 5 present the impact of heavy ends on the phase envelope of the key streams of feed, inlet separator vapor (stream “2”) and the sales gas (stream “4”), respectively. These figures indicate that as the heavy components are removed in the “Inlet Separator” and cold separator (“V-102”) from the process streams, the impact of heavy end characterization on the phase envelope reduces and vanishes almost completely for sales gas (stream “4”) in Figure 5.

 

 

Figure 3. The impact of heavy end on the phase envelope of the feed stream

 

 

Figure 4. The impact of heavy end on the phase envelope of the inlet separator vapor stream

 

 

Figure 5. The impact of heavy end on the phase envelope of the sales gas (Stream 4)

 

 

Table 5 presents the impact of heavy end characterization on the “Gas/Gas” HX  and “Chiller” duties. Note that the “Gas/Gas” HX duty is controlled by stream “4” composition and rate. Based on the phase envelopes in Figure 5, the sales gas composition is almost independent of heavy ends because they are removed from the sales gas but the heavy ends have more impact on the composition of streams “2”.

 

 

Table 5. Impact of heavy end on the Gas/Gas HX  and Chiller duties

 

 

Table 5 indicates that as the heavy ends become heavier,

►stream “2” flow rate decreases because there is more liquid leaving the “Inlet Separator”.

►stream “4” rate increases by about 0.27% (nC7 to C7+) because most of the C7+ has been removed.

►“Gas/Gas” HX  duty is set by stream “4” rate and fixed ΔT=25-(-20) =45 °C (81 °F) because Q = mΔ(H_Salesgas – H₄).

►“Gas/Gas” HX  duty increases slightly, less than 0.8 %, because stream “4” rate increases by about 0.27%

►stream “2A” rate decreases, “Gas/Gas” HX  duty increases, stream “5A” gets colder, chiller ΔT decreases; therefore, “Chiller” duty decreases

 

 

Assume the design-heavy end was nC8 and feed gas heavy end is C7+, not the design nC8. More liquids would leave the “Gas/Gas” HX so the chiller duty would decrease by about 38%. But the additional duty to condense the liquids in the “Gas/Gas” HX  has to come from somewhere. If the “Gas/Gas” HX has excessarea to accommodate the additional duty requirements, then there would indeed be a decrease in chiller duty.  If it does not, the duty of the chiller may actually increase.

 

If the feed gas got lighter, and the heavy end is nC7, not the design nC8, then more gas would go to the chiller (less liquids leaving the “Gas/Gas” HX) and the chiller duty would increase by about 20%.  Here, the chiller would have to have excess capacity.

 

This indicates that a change in feed gas characterization would have an effect on the ability of a refrigeration unit to make spec. For easier reference of the stream, see Figure 6.

 

 

Figure 6. Simplified schematic of the front end segment of the process flow diagram

 

 

Table 6 presents the impact of heavy end characterization on the refrigeration systems. This table indicates that the rate, compressor power, condenser and the sub-cool economizer duties decrease as the heavy end becomes heavier. Table 6 also indicates that the rate, compressor power and condenser duty for the sub-cool economizer refrigeration system are lower compared to the simple refrigeration system. Because the chiller duty decreases, the refrigeration systems become smaller; therefore, the OPEX and CPEX decrease.

 

 

Table 6. Impact of heavy end on the refrigeration systems key parameters

 

 

The heat removed by the sub-cool economizer “E-104” is used to heat stream “9A” in “E-102” HX. Location of “E-102” HX and streams “9A” and “9B” are shown in Figure 7.

 

 

Figure 7. Simplified schematic of the back end of the process flow diagram

 

 

Table 7 presents the impact of heavy ends on the rates and the molecular weights for stream “3” from the “Inlet Separator” and stream “8” from the cold separator (“V-102”) and the combined NGL stream “9”. This table indicates that as the heavy end becomes heavier, rate of stream “3” increases but the rate of stream “8” decreases. Because the rate of heavy ends entering the “Gas-Gas” HX and “Chiller” decrease, the chiller duty decreases and condensation of components decrease resulting lower streams “8” and “9” rates. Table 1A in the Appendix present components flow rates for streams “3” and “8”.

 

 

Table 7. Impact of the heavy end on streams “3” and “8” and the combined NGL stream “9” rates and molecular weight

 

 

Using “VLV-100” stream 9 pressure is reduced from 3945 kPa (572.2 psia) to 1535 kPa (222.6 psia) in stream “9A”. Table 8 presents the combined NGL streams “9A” and “9B” (see Figure 7) properties. This table indicates that as heavy end becomes heavier, the “E-102” HX duty decreases because combined NGL stream rate decreases. The required heat for this HX is supplied by the sub-cool HX (“E-104”) of the refrigeration system.

 

Table 9 presents the impact of heavy end on the plant overall material balance. This table indicates that as the heavy end becomes heavier,

►the sales gas rate increases (stream 4)

►the deethanizer feed (combined NGL stream, 9 ) rate decreases because the sales gas rate (stream 6) has increased

►the overhead vapor temperature from the deethanizer top remains almost constant because the overhead composition does not significantly change

►the overhead vapor rate from the deethanizer top decreases because the deethanizer feed rate decreased

 

 

Table 8. Impact of the heavy end on the combined NGL streams “9A” and “9B” properties

 

 

Table 9. Impact of the heavy end on liquid propane recovery

 

 

The overhead vapor of deethanizer is compressed from 1450 kPa (210.3 psia) to the feed gas inlet pressure of 4000 kPa (580 psia) by the recycle compressor (“K-100”) and cooled down to the inlet feed gas temperature of 30 °C (86 °F) in the “E-101” HX. The liquid from compressor suction scrubber is recycled and combined with deethanizer feed by the recycle pump. Table 10 presents the compressor and pump power and the “E-101” HX duty requirements. Table 10 indicates that as the heavy end becomes heavier, the recycle compressor and pump power and the cooler duty decrease because the recycle stream rates decrease.

 

Table 10. Impact of the heavy end on the recycle compressor, pump, and cooler

 

 

SUMMARY:

The feed analysis and /or heavy end characterization in natural gas play an important role in the equipment sizing and process design. Feed analysis may change when different wells of slightly different composition are brought to the production facility. This tip demonstrated the impact of heavy end characterization in the feed gas on the process streams rates, phase behavior, the equipment sizes and the refrigeration requirement by replacing, the C7+ with n-heptane (nC7) and n-octane (nC8). All other specifications and operating conditions were kept the same.

 

As demonstrated in this tip, it would be a good practice to size the equipment with a design margin of 1.2 to 1.3 to take into account the changes in feed gas heavy end composition and characterizations.

 

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing), G5 (Practical Computer Simulation Applications in Gas Processing) and G6 (Gas Treating and Sulfur Recovery) courses.

 

By: Dr. Mahmood Moshfeghian

---

To receive Tips of the Month directly to your inbox, simply sign up below!

---

 

References

1. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2019/01/optimizing-performance-of-refrigeration-system-with-an-external-sub-cool-economizer/,  PetroSkills -John M. Campbell Tip of the Month, January 2019.

2. Campbell, J.M., “Gas Conditioning and Processing, Volume 1: The Fundamentals,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum

3. Campbell, J.M., “Gas Conditioning and Processing, Volume 2: The Equipment Modules,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, PetroSkills 2018.

4. UniSim Design R443, Build 19153, Honeywell International Inc., 2017.

 

 

---

Appendix

 

Figure 1A. Phase diagrams for the key streams for the case of nC7 as the heavy end

 

 

Figure 2A. Phase diagrams for the key streams for the case of nC8 as the heavy end

 

Table 1A. Impact of heavy ends on the flow rates of streams “3” and “8”

Continuing the January 2008 [1], May 2008 [2], May 2014 [3], and December 2017 [4] Tips of The Month (TOTM), this tip demonstrates the application of an external sub-cooler to optimize the performance of a mechanical refrigeration system. Specifically, by utilizing a cold process stream we will minimize the compressor power and condenser duty. The details of three typical refrigeration systems are given in Chapter 15 of Gas Conditioning and Processing, Volume 2 [5]. They are referred to as follows:

1. A simple refrigeration system (Fig 15.1);
2. A refrigeration system employing one flash tank economizer and two stages of compression (Fig 15.7), see also the May 2014 [3] and December 2017 [4] TOTM
3. A Simple Refrigeration System with a sub-cool heat exchange economizer (Fig 15.9), see also the May 2008 [2] and May 2014 [3] TOTM.

 

Figure 1 presents the process flow diagrams for a simple system and the system with a sub-cool heat exchange economizer. As part of a hydrocarbon dew point control plant, this tip will evaluate and compare these two refrigeration systems.

 

Figure 1. Process flow diagrams for a simple refrigeration system and with a sub-cool economizer

 

 

Let’s consider cooling the process gas to -20°C [-4°F] by removing 2733 kW (9.325 MMBtu/hr) in a propane chiller and rejecting it to the environment by a propane condenser at 37.8°C [100°F]. Pure propane is used as the working fluid in the simulation. In this tip, all simulations were performed with UniSim Design software [6] using the Peng-Robinson equation of state. Assuming an approach temperature of 5°C [9°F] and a 6.9 kPa (1 psi) pressure drop in the propane chiller, the pressure of saturated propane vapor leaving the chiller is 203.3 kPa (29.5 psia), and at a temperature of -25°C     [-13°F] . Assuming no frictional losses in the suction line to the propane compressor, the resulting suction pressure is 203.3 kPa (29.5 psia).

 

The condensing propane pressure at the specified condenser temperature of 37.8 °C (100 °F) is 1303 kPa (189 psi). The condenser frictional losses, plus the frictional losses in the piping from the compressor discharge to the condenser was assumed to be 34.5 kPa (5 psi); therefore, the compressor discharge pressure is 1338 kPa (194 psia). The propane compressor adiabatic efficiency was assumed to be 75%.

 

 

 

External Sub-Cool, Economizer

 

The process streams 9A and 9B are part of a hydrocarbon dew point control plant and are shown on the top of Figure 2. This stream is the extracted NGL stream from the refrigeration plant, combined with the plant inlet condensate. The stream properties are shown in Table 1.  To prepare the liquids to be fed to the deethanizer, the process specification is to raise the temperature of the NGL product stream 9A from -3.9°C (25°F) to 20°C (68°F) in HEX E-102. The resulting duty is 713.6 kW (2.435 MMBtu/hr). This heat will be supplied from a propane refrigerant sub-cool economizer heat exchanger. The process duty and the temperature of the NGL product stream is set by the stabilization process requirements, thus the sub-cool economizer duty is fixed.

 

The sub-cool economizer cools the condensed propane (refrigerant stream R4) from 37.8°C (100 °F) at 1303 kPa (189 psia) to a cooler temperature at 1269 kPa (184 psia), depending upon the specified propane refrigerant flow rate (stream R5). The pressure drops in HEX E-102 and HEX E-104 are 35 kPa (5 psi) respectively. The heat removed by the sub-cool economizer is fixed by the process duty required to heat the NGL process stream 9A.

 

 

Figure 2. Cold process stream 9A part of the hydrocarbon dew point control plant is utilized to sub-cool refrigerant stream R5.

 

 

Table 1. Process conditions for streams 9A and 9B

Process Stream	9A	9B
Temperature, °C	-3.87	20
Pressure, kPa	1535	1500
Rate, kmol/h	666	666
Temperature, °F	25.03	68
Pressure, psia	222.6	217.5
Rate, lbmol/hr	1468.3	1468.3

 

 

 

Determination of Refrigerant Circulation Rate of Sub-Cool Economizer System

 

The refrigerant circulation rate has a considerable impact on the compressor power and consequently on the condenser duty. Figure 2 presents the variation of compressor power as a function of the refrigerant circulation rate. This figure indicates that the power is minimum at 995.5 kW (1335 hp) for a circulation rate of 690 kmol/h (1521.5 lbmol/hr), with fixed propane chiller and sub-cool exchanger duties.

 

 

Figure 3. Impact of refrigerant circulation rate on the compressor power

 

 

  

Summary

 

For the same chiller duty, chiller and condenser temperatures, adiabatic compression efficiency, and pressure drops, the results of the sub-cool exchange economizer system are compared with the results of a simple refrigeration system in Table 2. This table indicates that by utilizing an external sub-cool exchange economizer for this case study with optimized propane circulation rate, the compressor power and condenser duty are reduced by 25 % and 25.4%, respectively.

 

Table 2. Comparison of the key parameters of two refrigeration systems

Refrigerant Rate, kmol/h		Simple System, kW		Sub-Cool System, kW
Simple	Sub-Cooler	Comp	Cond	Comp	Cond	Sub-Cooler
920.5	690	1328	4061	995.5	3029	713.6
Refrigerant Rate, lbmol/hr		Simple System, MMBtu/hr (hp)		Sub-Cool System, MMBtu/hr (hp)
Simple	Sub-Cool	Comp	Cond	Comp	Cond	Sub-Cooler
2029.7	1521.5	4.532 (1781)	13.86	3.446 (1335)	10.34	2.435
Reduction, %	25.04			25.04	25.41

 

 

To learn more about similar cases and how to minimize operational problems, we suggest attending our G4 (Gas Conditioning and Processing), G5 (Practical Computer Simulation Applications in Gas Processing), and G6 (Gas Treating and Sulfur Recovery) courses.

 

By: Dr. Mahmood Moshfeghian

 

References:

1. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2008/01/refrigeration-with-flash-economizer-vs-simple-refrigeration-system/, John M. Campbell Tip of the Month, January 2008.
2. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2008/05/flash-tank-vs-hex-economizer-refrigeration-system/, John M. Campbell Tip of the Month, May 2008.
3. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2014/05/refrigeration-with-heat-exchanger-economizer-vs-simple-refrigeration-system/, PetroSkills Tip of the Month, May 2014.
4. Moshfeghian, M., http://www.jmcampbell.com/tip-of-the-month/2017/12/optimizing-performance-of-refrigeration-system-with-flash-tank-economizer/, PetroSkills Tip of the Month, December 2017.
5. Campbell, J.M., “Gas Conditioning and Processing, Volume 2: The Equipment Modules,” 9^th Edition, 3^rd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, PetroSkills 2018.
6. UniSim Design R443, Build 19153, Honeywell International Inc., 2017.

Continuing the September 2018 tip of the month (TOTM) [1], the phase behavior of ethane and water binary system was studied. Like propane–water system [2], the ethane–water system is complicated for the following two reasons:

1. Very low mutual solubility in liquid phases.

2. At lower temperatures, ice or hydrates is formed.

In this tip, we will evaluate the accuracy of water content predicted by a process simulation software against limited measured experimental data. Second, the tip studies the effect of pressure and temperature on the ethane water content in equilibrium with liquid water, ice, or hydrate phase. In addition, water content charts are presented for isobars of 14.7, 25, 50, 100, 150, and 200 psia (101.3, 172, 345, 699, 1034, 1379 kPa). For each isobar a temperature range of -60 °F to 200 °F (-51 °C to 104 °C) is covered.

 

---

 

Evaluation of the Water Content Prediction Methods

The performance of the ProMax simulation software [3], for estimating the water content of ethane in equilibrium with hydrate or liquid water was evaluated against limited GPA RR 132 experimental data [4]. A summary of water content comparisons for ethane vapor (G) or liquid (L_HC) in equilibrium with hydrate (H) and liquid water (L_W) is presented in Table 1. Three methods within ProMax were utilized. These three methods are labeled and described as follows:

►ProMax 1: Water content was estimated using a water saturator tool.

►ProMax 2: One mole of pure ethane stream was mixed with a pure water stream at the desired pressure. To determine the water content of the mixed stream, the solver tool of ProMax was used to adjust the pure water stream flow rate to match the system temperature.

►ProMax 3: Water content was estimated by performing flash calculations for a binary system of 50–50 mole % ethane–water at system pressure and temperature.

 

 

Table 1. Comparison of vapor or liquid ethane water content (PPM by mole) in equilibrium with liquid water or hydrate by ProMax against the GPA-RR 132 [4] experimental data

 

* The pressure in parenthesis are the experimental values which were adjusted to form liquid ethane. See below for detail of pressure adjustment.

 

The SRK EOS (Soave-Redlich-Kwong equation of state) [5] with its ProMax default binary interaction parameters were used. For these set of pressures and temperatures, all three methods give relatively good results. Table 1 indicates that even the ethane water contents are very low (from 2 to 733 ppm by mole), the average absolute percent deviations are relatively low and are from 11 to 18.5 %.

 

For pressures higher than 496 psia (3421 kPa), to form liquid ethane and liquid water (L_HC-L_W) phases at equilibrium, the experimental pressures reported in Table 1 were adjusted as follows:

1. For a binary system of 50–50 mole % ethane–water, at the specified temperature and 0 % vapor the pressure was calculated.

2. The calculated pressure was increased slightly, i.e. less than 2 psi (14 kPa).

3. The summary of results for the adjusted pressures is presented in Table 2.

 

 

Table 2. Pressure adjustment summary for pressure higher than 496 psia (3420.7 kPa)

 

---

Ethane Water Content Charts

The coexistence of equilibrium phases depends on the system pressure and temperature as follows:

a. Ethane vapor phase in equilibrium with liquid water

b. Ethane vapor phase in equilibrium with ice or hydrate

c. Ethane liquid phase in equilibrium with liquid water

d. Ethane liquid phase in equilibrium with hydrate

 

Figure 1 illustrates the presence of these equilibrium phases as a function of temperature for the isobar of 200 psia (1379 kPa). The propane water content was estimated by the following procedures.

a. For temperatures of 200 °F to about 47.5 °F (93 to ~ 8.6 °C), the ethane vapor is in equilibrium with liquid water phase so the water saturator tool of ProMax was used.

b. For temperatures of about 47.5 °F to -6 °F (~ 8.6 to -21.1 °C), the ethane vapor was in equilibrium with the hydrate phase, so one mole of pure ethane stream was mixed with a pure water stream atpressure of 200 psia (1379 kPa). To determine the water content of the mixed stream, the solver tool in ProMax was used to adjust the pure water stream flow rate to form hydrate at the specified hydrate formation temperature.

c. The ethane vapor phase transition to the liquid phase takes place at -6 °F (-21.1 °C).

d. For temperatures of -6 °F to -60 °F (-21.1 to -51 °C), the ethane liquid is in equilibrium with the hydrate phase, so one mole of pure ethane stream was mixed with a pure water stream at pressureof 200 psia (1379 kPa). To determine the water content of the mixed stream, the solver tool in ProMax was used to adjust the pure water stream flow rate to form hydrate at the specified hydrate formation temperature.

Figure 1. Water content of vapor and liquid Ethane as a function of temperature at 200 psia (1379 kPa)

 

 

Table 3 presents the three-phase temperatures of a ethane–water system and the saturatiom temperatures of pure ethane estimated by ProMax.

 

Table  3. Three phase temperature and saturation temperature for six isobars

 Phases: V = Vapor, I = Ice, H = Hydrate, and L_W = Liquid Water

 

Similarly, the water content charts of ethane vapor and liquid phases were prepared for the other isobars and are presented in Figures 2–4. Note in Figures 2 and 4, due to the very small values, the liquid ethane water contents for different isobars fall on the same curve.

 

Figure 2. Water content of Ethane as a function of temperature for six isobars

 

 

Figure 3. Water content of vapor ethane as a function of temperature for several isobars

 

 

Figure 4. Water content of liquid ethane as a function of temperature for several isobars

 

---

 

Conclusion

Similar to propane, estimating ethane water content requires a good understanding of the phase behavior. The process simulation programs have several tools or procedures for estimating the water content. Which one should be used to give a correct answer? It depends. In addition to the selection of a suitable equation of state, the selection of the right tool or procedure at a given set of conditions is essential. The choice of a suitable tool changes as the conditions or the equilibrium phases change.

The presented ethane water content charts can be used for facility type calculations and trouble shooting. It is a good practice to test the performance/accuracy of the selected tool against experimental data first. Obviously, for better understanding of ethane–water phase behavior and improving the thermodynamic modeling, more experimental data is needed.

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing) and G5 (Practical Computer Simulation Applications in Gas Processing) courses.

By: Dr. Mahmood Moshfeghian

---

Sign up below to receive Tips of the Month!

---

References

1. Moshfeghian, M., “Propane – Water Phase Behavior at Low to Moderate Pressures,” PetroSkills TOTM, Sep 2018.

2. Harmens, A. and E.D. Sloan, “The phase Behavior of Propane – Water System: A Review,” The Canadian J of Chem Engr, Vol 68, Feb 1998.

3. ProMax 4.0, Build 4.0.17179.0, Bryan Research and Engineering, Inc., Bryan, Texas, 2017.

4. Song, K and R. Kobayashi, “Water content of ethane, propane, and their mixtures in equilibrium with water and hydrates,” Gas Processor Association Research Report (GPA RR 132), Tulsa, Oklahoma, 1991.

5. Soave, G., Chem. Eng. Sci. Vol. 27, No. 6, p. 1197, 1972.

In facilities operations, the understanding of where the process is on a phase diagram can often help the engineer and operator to avoid extremely embarrassing design and operating mistakes. A phase diagram is a function of only mixture compositions and components. Though it is expensive and time consuming, a phase envelope can be determined experimentally by measuring a series of bubblepoints and dewpoints. Typically, a phase envelope is generated with a process simulator for specified mixture compositions. For a sensitive process, which requires an accurate phase envelope, some lab or field measured VLE (Vapor-Liquid-Equilibria) data may be required for tuning the equation of state (EOS).

Process simulation programs typically don’t have a utility or analysis tool to add isothermal and isobaric retrograde curves to a phase diagram. This tip presents two methods for how a process simulation program can be used to quantitatively determine the isothermal and isobaric retrograde curves. A phase diagram, including the retrograde curves for rich gas, is generated and presented.

Figure 1 from Volume 1 of Gas Conditioning and Processing [1] shows the qualitative phase envelope for a natural gas mixture. Reference [1] also presents the terms used to define the location of various points on the phase envelope. A selection of these terms is listed here.

 

 

 

Figure 1. Typical phase diagram for a multi-component natural gas mixture [1]

 

The blue area represents isothermal retrograde, where liquid condenses when pressure decreases. This is what happens in the path A to B to D.  The condensation begins at point B and the maximum liquid occurs at point D well below the pressure that liquids first start to form. This phenomenon occurs in most naturally occurring natural gas mixtures. We make use of this phenomenon to choose the best operating pressures for our production equipment.

The green shaded area represents isobaric retrograde vaporization. In this area, liquid vaporization occurs as temperature decreases or the amount of liquid decreases as we decrease temperature. If we come into the phase envelope from the other side, the amount of liquid increases as we increase temperature.

In facilities operations, the understanding of where the process is on a phase diagram can often help the engineer and operator to avoid extremely embarrassing design and operating mistakes [2]. A phase diagram is a function of only mixture compositions and components. Though it is expensive and time consuming a phase envelope can be determined experimentally by measuring a series of bubblepointsand dewpoints. Typically a phase envelope is generated with a process simulator for specified mixture compositions. For a sensitive process, which requires an accurate phase envelope, some lab or field measured VLE (Vapor-Liquid-Equilibria) data may be required for

Process simulation programs typically don’t have a utility or analysis tool to add isothermal and isobaric retrograde curves to a phase diagram. This tip presents two methods on how to process a simulation program can be used to determine quantitatively the isothermal and isobaric retrograde curves. To demonstrate the procedure the phase diagram including the retrograde curves for a rich gas is generated and presented.

 

 

Determination of Retrograde Curves

Table 1 presents a rich natural gas mixture composition used in this tip. The PR EOS (Peng-Robinson) equation of state [4] with its ProMax [5] default binary interaction parameters were used.

 

Table 1. Natural gas composition

 

Table 2 and Figure 2 present the calculated properties and generated phase envelope for the mixture in Table 1 by ProMax [5].

 

Table 2. ProMax (PR EOS) calculated properties for the natural gas mixture of Table 1

 

 

Note Figure 2 does not illustrate the two retrograde regions. This is true for other process simulation programs. The step-by-step procedures to calculate the isothermal retrograde condensation and isobaric retrograde vaporization curve are presented in the following section.

Figure 2. Phase diagram by ProMax (PR EOS) for the natural gas mixture of Table 1
METHOD 1: Isothermal Retrograde Condensation (isoT RG) Curve

The curve connecting points M, D, and C in the blue region of Figure 1 is the isothermal retrograde curve. This curve presents the locus of maximum liquid formation for each isotherm between the critical (Point C) and cricondentherm (Point M) temperatures. For a specified temperature and an estimate of pressure as an initial guess, the maximum liquid formation is calculated by adjusting the pressure using the Solver tool in ProMax or similar tool in other software.

1. Calculate a step change for temperature, ΔT = (Cricondentherm T- Critical T)/N, N= ~ 20

2. Select the first temperature, e.g. T₁= (Cricondentherm T) – ΔT

3. Use Solver tool to find the corresponding pressure, which maximizes the liquid formation

4. Record the specified temperature and calculated pressure corresponding to the maximum liquid formation

5. Select subsequent temperature I, e.g. T_I = T_I-1 – ΔT, and repeat steps 3 and 4 till the critical temperature is reached.

6. Plot the recorded pressures and temperatures on the phase diagram

 

 

The above procedure was executed by defining a simple process flow diagram presented in Figure 3A (isoT RG). The feed composition, flow rate (100 moles), specified temperature, and the initial estimate of pressure are defined in stream 1. The pressure drop in XCHG-100 was set equal to zero. Stream 2 temperature was set equal to stream 1 temperature. Using the Solver tool, stream 2 pressure was adjusted to maximize the liquid formation (this was done by minimizing negative of liquid formation fraction).  Table 3 presents the results for a sample calculation.

 

Figure 3. Simple process flow diagrams for isothermal retrograde (A. isoT RG) and isobaric retrograde (B. isoB RG) curve calculations

 

 

Table 3. Single point sample calculation for isoT RG curve.

Note: * Initial Estimates, ** Calculated Values

 

The calculated isothermal retrograde condensation (isoT RG) curve is illustrated on the phase diagram of Figure 4. Note as the critical point was approached better estimate of pressures were needed to achieve convergence. Therefore, a few point calculations were done manually.

 

 

Figure 4. Phase diagram with isothermal retrograde condensation (isoT RG) curve for the natural gas mixture of Table 1

 

METHOD 1: Isobaric Retrograde (isoB RG) Curve

The curve connecting points M and C in the green region of Figure 1 is the isobaric retrograde curve. This curve presents the locus of maximum liquid formation for each isobar (P) between the critical (Point C) and cricondenbar (Point N) pressures. For a specified P and an estimate of T as initial guess, the maximum liquid formation is calculated by adjusting the T.

1. Calculate a step change for P, ΔP = (Cricondenbar P- Critical P)/N, N= ~ 20

2. Select the first pressure, e.g. P₁= (Cricondentherm P) – ΔP

3. Use the Solver tool to find the temperature which maximizes the liquid formation fraction

4. Record the specified pressure and the calculated temperature

5. Select subsequent pressure for point I, e.g. P_I = P_I-1 – ΔP, and repeat steps 3 and 4 till the critical pressure is reached

6. Plot the recorded pressure and temperatures on the phase diagram

The above procedure was executed by defining a simple process flow diagram presented in Figure 3B (isoB RG). The feed composition, flow rate (100 moles), specified pressure, and the initial estimate of temperature are defined in stream 3. The pressure drop in XCHG-101 was set equal to zero, which resulted in the same pressure for streams 3 and 4. Using the Solver tool, stream 4 temperature was adjusted to maximize liquid formation (this was done by minimizing negative of liquid formation fraction). Table 4 presents a sample calculation results. The calculated isobaric retrograde vaporization (isoB RG) curve is illustrated on the phase diagram of Figure 5.

 

Table 4. Single point sample calculation for isoB RG curve.

Note: * Initial Estimates, ** Calculated Values

 

 

Figure 5. Phase diagram with isothermal (isoT RG) and isobaric (isoB RG) curve for the natural gas mixture of Table 1

 

 

METHOD 2: Isothermal Condensation and Isobaric Retrograde Vaporization Curves

In this method, the process simulation program can be used to generate a series of constant liquid (or vapor) quality curves. Figure 5 presents the phase envelope generated by the SRK (Soave-Redlich-Kwong) [6] EOS of EzThermo program [7] and the constant liquid quality curves for 30%, 25%, ….5%, and 2.5% L (liquid) curves. Each constant quality curve has its own cricondentherm and cricondenbar points. Note the isothermal retrograde (diamond symbol) curve passes through each constant quality curve cricondentherm and the isobaric retrograde (triangle symbol) curve passes through each constant quality curve cricondenbar. The properties of the critical point, cricondentherm, and cricondenbar by EzThermo are presented in Table 5. Note that these properties are slightly different from those calculated by the PR EOS of ProMax presented in Table 2 (due to different programs, EOS, and the binary interaction parameters).

 

Figure 6. Phase diagram with isothermal (isoT RG) and isobaric (isoB RG) curve for the natural gas mixture of Table 1 [7]

 

 

Table 5. EzThermo (SRK EOS) calculated properties for the natural gas mixture of Table 1

The generated phase diagram including the isothermal retrograde curve with a single run by EzThermo (SRK EOS) is presented in Figure 7. Note the retrograde curve does not approach the critical point correctly.

 

Figure 7. Phase diagram by EzThermo (SRK EOS) with iso T RG for the natural gas mixture of Table 1

 

 

Conclusions

Process simulation programs typically generate phase envelopes without the option to add the isothermal and isobaric retrograde curves.  This tip presented two methods of utilizing a process simulation program for quantitative calculations of the isothermal and isobaric retrograde curves. To demonstrate the procedure two diagrams (Figures 5 and 6) including the retrograde curves for a rich gas were presented. The proposed procedure requires several steps but can be incorporated in an excel file linked to the process simulator.

Ideally, it will be helpful if process simulation programs add these features to their phase envelope utilities/analysis tools.

To learn more about similar cases and how to minimize operational problems, we suggest attending ourG4 (Gas Conditioning and Processing), and G5 (Practical Computer Simulation Applications in Gas Processing) courses.

By: Dr. Mahmood Moshfeghian

---

Sign up to receive Tip of the Month email updates below!

---

References

1. Campbell, J.M., Gas Conditioning and Processing, Volume 1: The Basic Principles, 9^th Edition, 2^nd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, 2014.

2. Lilly, L., “Why do I care about phase diagrams?”, PetroSkills June 2005 TOTM.

3. Moshfeghian, M., ”Selecting the correct phase envelope”, PetroSkills Nov 2005 TOTM.

4. Peng, D. Y., and Robinson, D. B., Ind. Eng. Chem. Fundam., Vol. 15, p. 59, 1976.

5. ProMax 4.0, Build 4.0.17179.0, Bryan Research and Engineering, Inc., Bryan, Texas, 2017.

6. Soave, G., Chem. Eng. Sci. Vol. 27, No. 6, p. 1197, 1972.

7. EzThermo, Moshfeghian, M. and Maddox, R. N., 2008.

►Watch Mick Crabtree’s Harnessing Coriolis Webinar

 

It was not until Gaspard-Gustav de Coriolis published a paper in 1835 that the terms Coriolis Effect andCoriolis Force were actually coined [1]. Nonetheless, the effect had been described by the Italian scientist Giovanni Riccioli in 1651 who postulated that if, in the astronomical model of the solar system, the Earth rotated around the Sun (as heliocentrics believed) a cannonball fired towards the north pole should cause it to deflect to the east. Since, at that time, this was not an observable effect it was used as an argument for geocentrism, which placed the Earth at the center.

 

In recent years, the major impact of the Coriolis Effect has been in the field of meteorology – describing how it affects global wind patterns – with cyclonic rotation having an anti-clockwise direction in the northern hemisphere and clockwise direction in the southern hemisphere.

 

But the practical application of the Coriolis effect in the field of engineering has been in mass flow measurement.

 

Mass flow is a primary unit of flow measurement and is unaffected by viscosity, density, conductivity, pressure and temperature. As a result, it is inherently more accurate and meaningful for measuring material transfer. Furthermore, most chemical reactions are based largely on their mass relationship.  Consequently, by measuring the mass flow of the product it is possible to control the process more accurately.

 

Traditionally, the measurement of mass flow entailed measuring the volumetric flow rate and multiplying it by the measured density – normally achieved through the use of expensive, imprecise and potentially dangerous nuclear-based densitometry. In contrast, Coriolis-based metering measures mass flow directly.

 

 

The Coriolis force:

Consider two children, Anne and Belinda, playing on a children’s roundabout (Figure 1) which is rotating at a constant angular velocity [2][3].   Anne is situated mid-way between the axis and the outer edge of the platform while Belinda is sat at the outer edge itself. If Anne now throws a ball directly to Belinda,  Belinda will fail to receive the ball!

 

Figure 1. Anne and Belinda are playing on a children’s roundabout.  If Anne throws a ball directly to Belinda, Belinda will fail to receive the ball due to the Coriolis effect.

 

The fact is that whilst Anne and Belinda both have the same angular velocity their tangential velocities are different (Figure 2). Anne’s tangential velocity (V_A) is only half of Belinda’s tangential velocity (V_B). In fact, the peripheral speeds of each are directly proportional to the radius i.e.:

 

 

Figure 2. Belinda, at the edge of the platform, will have a peripheral speed of twice that of Anne and thus the ball’s peripheral speed needs to be accelerated from V_A to V_B .

 

Consequently, to move the ball from Anne to Belinda its tangential velocity needs to be accelerated from V_A to V_B. This acceleration is a result of what is termed the Coriolis force, named after the French scientist who first described it and is directly proportional to the product of the mass in motion, its speed and the angular velocity of rotation:

 

 

Looking at this from another point, if we could measure the Coriolis force (F_cor), knowing the tangential velocity (v) and the angular velocity we could determine the mass (m) of the ball.

 

 

How does this relate to mass measurement of fluids?

 

Consider a liquid-filled pipe sealed at both ends, rotating about an axis at an angular velocity (w).

 

The tangential velocity (v) of any individual particle of the fluid is simply the angular velocity (w) times the distance (r) from the centre of rotation (Figure 3). Thus, at distance r₁, the tangential velocity of a particle would be r1.w whilst at double the distance r₂, the tangential velocity would also double to r₂.w.

 

Figure 3. At distance r₁, the tangential velocity of a particle would be r1.w whilst at double the distance r₂, the tangential velocity would also double to r₂.w.   

 

If now, the liquid flows in a direction away from the axis (Figure 4), then as each mass particle moves, for example, from r₁ to r₂ it will be accelerated by an amount equivalent to its movement along the axis from a low to a higher tangential velocity.  This increase in velocity is in opposition to the mass inertial resistance and is felt as a force opposing the pipe’s direction of rotation – i.e. it will try to slow down the rotation of the pipe.  Conversely, if we reverse the flow direction, particles in the liquid flow moving towards the axis are forced to slow down from a high velocity to a lower velocity and the resultant Coriolis force will try to speed up the rotation of the pipe.

 

Figure 4. As the liquid flows away from the axis, each mass particle will be accelerated by an amount equivalent to its movement along the axis from a low to a higher orbital velocity.

 

Thus, if we drive the pipe at a constant torque, the Coriolis force will produce either a braking torque or an accelerating torque (dependent on the flow direction) that is directly proportional to the mass flow rate. In other words, the torque required to rotate the pipe will increase in direct proportion to the actual mass flow of the liquid.

 

 

Initial implementation:

 

The possibility of applying the Coriolis effect to measure mass flow rate was recognized many years ago and, as shown in Figure 5 [4], initial patents were registered in the early 1950’s. These were typically based on rotating systems typified by the radial-vane type meter (Figure 6).

 

Figure 5. US Coriolis patent publications per year to November 2017 (© 2018 E. Jenks, Krohne Ltd.)

 

 

Figure 6. Radial-vane type meter

Here, the flowing fluid enters the center of a centrifugal pump-type impeller driven at a constant angular velocity (w). As the fluid’s tangential velocity (r.w) increases, it reacts on the impeller with a force directly proportional to the mass flow rate. This Coriolis force is measured by the torque measuring tube.

 

 

Obvious problems include:

►Maintaining the motor speed to provide a constant angular velocity (w).

►Slip rings for power and measurement of torque

►High unrecoverable pressure loss

►Wear and tear

 

It was not until 1977 that engineer and inventor Jim Smith, founder of Micro Motion, patented the first practical system. Rather than use rotational movement, this system was based on a vibrating tube system to provide oscillatory movement.

 

Figure 7.  A pipe, formed in a loop, is vibrated around the z axis so that the straight parts of the pipe, A-B and C-D, oscillate on the arcs of a circle.

 

 

The basic principle is illustrated in Figure 7 in which a tubular pipe, carrying the liquid, is formed in a loop and vibrated around the z-axis. The straight parts of the pipe, A-B and C-D, oscillate on the arcs of a circle and without any flow will remain parallel to each other throughout each cycle.

 

If a liquid now flows through the tube in the direction shown, then the fluid particles in section A-B will move from a point having a low tangential velocity at A to a point having a high tangential velocity at B. This means that each mass particle must be accelerated in opposition to the mass inertial resistance. This opposes the pipe’s direction of rotation and produces a Coriolis force in the opposite direction. Conversely, in section C-D, the particles move in the opposite direction – from a point having a high tangential velocity at C to a point having a low tangential velocity at D.

 

The resultant effect of these Coriolis forces is to delay the oscillation in section A-B and accelerate it in section C-D. As a result, section A-B tends to lag behind the undisturbed motion whilst section C-D leads this position. Consequently, the complete loop is twisted by an amount that is directly and linearly proportional to the mass flow rate of the fluid – with the twisting moment lent to the pipe arrangement being measured by sensors.

 

Figure 8 (a) shows the oscillatory motion applied to a single tube whilst Figure 8 (b) shows the forces acting on the tube in which there is fluid flow. As a result, the complete loop is twisted by an amount that is directly and linearly proportional to the mass flow rate of the fluid (Figure 8(c)) – with the flexure of the pipe arrangement being measured by sensors.

 

 

Figure 8. (a) Oscillatory motion applied to a single tube.  (b) Forces acting on the tube in which there is fluid flow. (c)Complete loop is twisted by an amount that is directly and linearly proportional to the mass flow rate of the fluid.

 

To reduce the risk of stress fractures the oscillation amplitude is limited to between 0.1 and 1 mm which, in an optimally designed system is about 20% of the maximum permitted value.

 

The distortion caused by the Coriolis forces is about 100 times smaller (a magnitude of about 10 mm), and in order to provide a measurement resolution to meet an accuracy of, for example, ±0.1% magnitudes of the order of a few nanometres would need to be resolved. The reality is we don’t actually measure the amplitude but the phase difference (i.e. time) with flexure of the pipe arrangement measured by velocity sensors (Figure 9).

 

Figure 9. Flexure of the pipe arrangement measured by velocity sensors.

 

 

With no fluid flow (and thus no flexure) the sinusoidal outputs of the two velocity sensors would be in-phase – with no time difference (Figure 10 (a)). However, with fluid flow, the flexure of the pipe produces a phase difference directly proportional to the mass flow (Figure 10 (b)).

 

Figure 10. With no fluid flow, the sinusoidal outputs of the velocity sensors are in-phase – with no time difference (a). With fluid flow, the flexure of the pipe produces a phase difference directly proportional to the mass flow (b).

 

Measurement resolution of ± 0.1 % amounts to only a few ns. And in order to achieve an accuracy of ± 0.1 %, where the resolution needs to be typically 5 to 10 times higher, time shift difference measurements are required down to picosecond levels.

 

 

Density Measurement

 

The measurement of mass flow by the Coriolis meter is, fundamentally, independent of the density of the medium. However, the vibratory action of the oscillating tubes employed in the Coriolis meter can be harnessed to provide an independent measure of the medium density.

 

Hooke’s Law spring equation states that a mass suspended on a spring will oscillate at a resonant frequency:

                                             

 

 

This indicates that when the mass increases the natural frequency decreases and when the mass decreases the natural frequency increases.

 

How does Hooke’s Law relate to density measurement in a vibrating tube? 

 

The mass of the system (m) equals the mass of the tube (m_tube), which is fixed, plus the mass of the fluid in the tube (m_fluid) – which is variable with the process:

 

 

In turn, the fluid mass (m_fluid) is determined by the volume of the tube (V_tube), which is fixed, multiplied by the density of the fluid (ρ_fluid), which is variable.

 

 

Rearranging equations 3, 4 and 5:

 

With all other values constant, the density of the fluid (ρ_fluid) is inversely proportional to the square of the resonant frequency:

 

 

Consequently, by measuring this frequency we calculate the fluid’s density.

 

It is important to recognize that the density measurement is not based on the Coriolis effect but on the effect of the vibrating tube.

 

So, in addition to providing a direct indication of mass flow the oscillating pipe system also, independently, provides a direct indication of the density by tracking the resonant oscillation frequency. Obviously, the value of (k) in equation 6 above, hides a lot of dependencies – one of which is temperature.  Consequently, the temperature must be measured as an independent quantity and used as a compensating variable. The temperature is also available as a measured output.

 

With knowledge of both the mass flow (Q_m) and the density (r) of the fluid, it is now possible to also calculate the volumetric flow rate (Q) since:

From the foregoing, we can see the importance of vibrating the tubes at their resonant frequency. Furthermore, excitation at the resonant frequency requires less drive energy and ensures that excitation is in the primary resonant mode at all times.

 

This is accomplished through a simple feedback system from the pick-up coils (Figure 11).

 

 

 

Figure 11. Excitation at the resonant frequency is accomplished through a simple feedback system from the pick-up coils.

 

 

Moving forward

 

At the beginning of the 1980’s there was only one Coriolis meter manufacturer, but by 1990 there were more than 13 different manufacturers. And, as may be seen from Figure 5, the mid-1980s saw the start of a plethora of new designs – each one seeking to overcome the shortcomings of previous versions.

 

 

So what exactly were the major perceived shortcomings of this technology?

Not necessarily in order of importance:

►the bent tube design

►maximum pipe diameter

►entrained gas

 

 

 

Tube configurations

 

Maximising sensitivity, whilst simultaneously minimizing the effects of extraneous noise and vibration, led to a variety of different bent-tube designs. But the bent-tube arrangement, itself, produced problems.

 

In any arrangement requiring the tube to be bent the outside wall is stretched and becomes thinner whilst the inner wall becomes thicker. When the flowmeter requires two such convoluted tubes, it becomes difficult to balance them both dimensionally and dynamically.

 

Furthermore, if the fluid is abrasive, this already weakened part of the flowmeter is likely to be most severely stressed. Abrasive material can also cause erosion that will change the stiffness of the resonant elements and so cause measurement errors.

 

And for many liquids the resultant pressure drop, due to the bends, could result in flashing or even cavitation damage. Furthermore, some of the bent-tube configurations did not cater for self-draining – an important consideration in many industries including: food and beverage; pharmaceuticals; and chemical.

 

A typical dual bent-tube design, as shown in Figure 12, featured a high total cross-sectional area combined with the flexibility of two pipes. On the negative side, the flow divider introduced a high-pressure drop. In addition, the flow may not be equally divided, and the dual tube arrangement does not allow clean-in-place (CIP) to be implemented.

 

 

Figure 12. A typical dual bent-tube design.

 

 

And whilst the continuous loop configuration (Figure 13) caters for CIP, a larger cross-sectional area is required to reduce the pressure loss. This leads to increased rigidity – making it less sensitive at low rates.

 

 

Figure 13. Continuous loop configuration.

 

 

Other designs proliferated (Figure 14) [5].

 

Figure 14. A variety of dual bent-tube designs.

 

 

 

However, the next major milestone occurred with the first straight-tube design introduced in 1986 by Endress+Hauser (Figure 15).

Figure 15. Straight-tube design employing dual measuring tubes.

 

 

With no flow, flexure of the tubes takes place in the vibrational plane (Figure 16 (a) and (b)). However, in the event of fluid flow, the Coriolis forces acting on the tubes produce a distorted flexure which is detected by the sensors (Figure 16 (c) and (d)).

 

 

Figure 16.  With no flow, flexure of the tubes takes place in the vibrational plane ((a) and (b)). However, in the event of fluid flow the Coriolis forces acting on the tubes produce a distorted flexure which is detected by the sensors ((c) and (d)). (Courtesy Endress + Hauser).

 

 

But whilst this design overcame the problems associated with having a bent tube (weakening of the tubes at the bends, erosion, flashing) it’s still employed dual measuring tubes with the need to split the flow. This limitation was surmounted when, in 1994, Krohne introduced the world’s first industrial single straight tube meter.

 

However, these straight tube designs also brought in their wake several other problems. Whereas the flexibility of the bent tube arrangement easily allowed for expansion and contraction due to temperature and/or pressure variations, the rigidity of the straight tube design is less forgiving. Consequently, the use of strain gauge technology became mandatory in order to detect the slight variations in the dimensional changes of the tubes. How these measurements were incorporated to provide accurate compensation became the subject of complex algorithms.

 

Early designs were constrained to a maximum pipeline diameter of typically 25 to 50 mm (1 to 2”) which severely limited their application. However, incrementally this limit has been extended until, only late last year, 400 mm (16”) pipeline diameter straight tube system [6].

 

 

MEMS-based technologies

 

At the other end of the scale, there has been a growing demand for micro-dosing in the pharmaceutical and silicon fabrication industries as well as in the laboratory sample analyzing and processing [7]. Whilst most Coriolis meters find their application in measuring flow rates greater than 1 kg/hr these industries require dosing and a measuring rates in the order of 1 g/hr and lower.

 

These challenges have been met through the application of MEMS-based technologies (Micro Electro Mechanical Systems) in which the entire mechanical Coriolis fluidic sensing system is integrated into a single silicon microchip Figure 17.

 

 

 

 

 

Figure 17. Complete Coriolis fluidic sensing system integrated into a single silicon microchip (courtesy  Integrated Sensing Systems)

 

 

Whilst some of these designs have focussed on flow measurement at very-low flow rates (down to 10 mg/h or better) other designs have been mainly developed to measure gas density and concentration [8].

 

At first glance, the use of silicon to fabricate the actual vibrating sensing tubes might seem at odds with its inherent rigidity, as compared with metals that can bend. Nevertheless, whilst a metal tube will plastically deform over time when exposed to fatigue when a silicon structure is bent it either goes back to its original position or it breaks. Consequently, because it never deforms, it is virtually free of fatigue, hysteresis anddrift.

 

Another benefit of silicon is the resonant frequency of the vibrating tube. Conventional metal tube Coriolis mass flow meters resonate at 100 to 1500 Hz [9,10] leaving them susceptible to the spectrum of common external mechanical vibration and shock frequencies that are under 2000Hz.

 

The micro-tube resonant frequency of silicon, on the other hand, is high (typically of the order of 20 kHz) making it virtually impervious to any form of extraneous vibration.

 

A third advantage of silicon over steel is that its density is 3.4 times lower.  Thus, whilst conventional Coriolis flow sensors manufactured from steel are typically not sensitive enough to accurately measure gas density at low pressures, a significantly higher sensitivity can result from the use of silicon.

 

 

Towards the perfect flow meter

 

There have been other challenges. For example, custody flow measurement of Liquefied Natural Gas (LNG) requires the system can operate down to -165°C – a severe challenge in terms of material embrittlement. But this challenge has been met and overcome and Coriolis metering for LNG is starting to become standard practice.

 

In earlier designs, the liquid to gas ratio was confined between 4 and 6%.  Newer models, employing a number of innovative designs, maintain active measurement in all measuring conditions with gas content from 0 to 100% by volume [11].

 

So how far along the road are we towards the perfect flow meter? Although it is doubtful that there will ever be a ‘one-size-fits-all’ solution, flow metering based on the Coriolis effect provides an almost universal range of answers.

 

However, in a recent article Eric Heilveil [12] stated:  “The first barrier to Coriolis world domination can be summed up in a single word … price. A one-inch line magnetic flowmeter and transmitter, for example, can be had for $3000 or less. A comparably sized Coriolis meter can run upwards of $9000 or more.”

 

But let us rather recall the words of John Ruskin:  “There is hardly anything in the world that some man cannot make a little worse and sell a little cheaper, and the people who consider price only are this man’s lawful prey.”

 

To learn more about similar cases and how to minimize operational problems, we suggest attending ourIC-3 (Instrumentation and Controls Fundamentals for Facilities Engineers), IC-71 (PLC and SCADA Technologies) and IC-72 (Valve and Actuator Technologies) courses.

By: Michael A. Crabtree

---

Sign up to receive Tip of the Month emails below!

---

References:

[1] https://en.wikipedia.org/wiki/Coriolis_force

[2] M. A. Crabtree, ‘Mick Crabtree’s Flow Handbook – 2^nd Edition’, Crown Publications cc, 2000.

[3] M. A. Crabtree, ‘Industrial Flow Measurement’, Masters Thesis, University of

Huddersfield, 2009.

[4] E. Jukes, ‘Optimass Product Group Presentation’, Krohne Ltd. 2018

[5] Rheonik, ‘RHM 160 – 12 Coriolis Mass Flowmeter’, Page 5 of 5, v6, April 2006.

[6] ‘New large line size Coriolis mass flowmeters’, Press release, Krohne Ltd. 2017

[7] C. Huber, ‘MEMS-based Micro-Coriolis Density and Flow Measurement Technology’

Endress+Hauser Flowtec AG, Proceedings AMA Conferences, 2015.

[8] A. Rieder, (Endress+Hauser GmbH), Paul Ceglia, (Endress+Hauser Flowtec) AG, ‘New generation vibrating tube sensor for density measurement under process conditions’,

2017.

[9] N. Najafi, M. Putty, R. Smith ‘Coriolis MEMS-sensing technology for real-time fluidic measurements’, Integrated Sensing Systems, (ISS), reprinted from ‘Flow Control’  May 2017

[10]  W. Sparreboom,  ‘Miniaturization to the Extreme: Micro-Coriolis Mass Flow Sensor’, Bronkhorst High-Tech B.V. January 30, 2018..

[11] H. Zhu, A. Rieder, ‘An innovative technology for Coriolis metering under entrained gas conditions’, Endress+Hauser Flowtec AG, 2016.

[12] E. Heilveil, ‘Three Dirty Little Secrets about Coriolis Flow Meters’, Siemens Industry Inc. July 2017

The phase behavior of a light hydrocarbon compound like propane and water binary system is complicated for the following two reasons:

1. Very low mutual solubility in liquid phases

2. At lower temperatures, ice or hydrates is formed.

 

Reference [1] presents an excellent review of a propane – water system phase behavior. It gives an integral description of the phase behavior of a propane – water system from very high to very low pressures in the form of a series of consecutive isobaric temperature–composition diagrams. Special attention is given to equilibria involving hydrates.

 

Figure 1 illustrates the general behavior for a propane–water system [2]. As shown in this figure, for temperatures less than the freezing point of 32 °F (0 °C) both ice and hydrate are present above the hydrate formation (green) curve; only vapor and ice are below. Above the freezing point, hydrate is the only solid phase to the left of the hydrate forming (green) curve. Above the intersection of the vapor pressure (blue) and hydrate (green) curves, as pressure is increased the temperature will increase slowly.

 

For a pressure of 29 psia (0.2 MPa), the Hydrate+Ice region requires an overall composition of propane less than 5.56 mole %. The area marked as Hydrate+Ice also includes Vapor+Hydrate because there is actually a 3^rd dimension to the plot – composition.  For overall compositions of propane > 5.56 mole %, Vapor+Hydrate is encountered and as it cools from 50 °F (10°C) at 29 psia (0.2 MPa) to the propane saturation temperature, then it is Hydrate+Liquid propane.

 

During propane processing, transportation, and storage hydrate formation should be prevented. To calculate the hydrate formation temperature and the required amount of inhibitor (e.g. methanol) to inhibit hydrate formation, estimation of water content is required.

 

In this tip, we first will evaluate the accuracy of water content predicted by a process simulation software against limited measured experimental data. Secondly, the tip studies the effect of pressure and temperature on the propane water content in equilibrium with liquid water, ice, or hydrate phase. In addition, water content charts are presented for isobars of 14.7, 25, 50, 100, 150, and 200 psia (101.3, 172, 345, 699, 1034, 1379 kPa). For each isobar a temperature range of -60 °F to 200 °F (-51 °C to 104 °C) is covered.

 

 

Figure 1. Phase Behavior of Propane-Water System [2]

 

 

 

Evaluation of the Water Content Prediction Methods

The performance of the ProMax simulation software [3], Bukacek correlation [4], and Raoult’s law (ideal) [5] for estimating the water content of propane vapor in equilibrium with liquid water (G–L_W) was evaluated against GPA RR 132 experimental data [6]. A summary of propane vapor water content comparisons is presented in Table 1. For these set of pressures and temperatures, all three methods give good results. The SRK EOS (Soave-Redlich-Kwong equation of state) [7] with its ProMax default binary interaction parameters were used. The water content was predicted using the water saturator tool available in ProMax.

 

 

Table 1. Comparison of vapor propane water mole fractions by three methods against GPA-RR 132 [6] experimental data

 

 

Similarly, the performance of ProMax [3] for estimating the water content of liquid propane in equilibrium with liquid water or hydrate was evaluated against GPA RR 132 [6] experimental data. The SRK EOS (Soave-Redlich-Kwong equation of state) with its default binary interaction parameters was used in ProMax. A summary of liquid propane water content comparison results is presented in Table 2. Considering the very low solubility of water in liquid propane, the agreement between the predicted values and experimental data is very good.

 

For the liquid propane-liquid water equilibrium phases the ProMax water saturator tool was used to estimate the water content of the liquid propane phase. Note for the formation of liquid propane, the reported experimental pressure of Table 2 had to be increased. The adjusted pressure values are presented in parenthesis. Except for the first pressure, the adjustment of pressure is small. The 130 psia point for RR-132 must be an incorrect pressure as a liquid-liquid system will require a pressure greater than the vapor pressure of propane of ~ 144 psia [993 kPa] at the stated temperature.

 

For the liquid propane-hydrate equilibrium phases, one mole of pure propane stream was mixed with a pure water stream at the desired pressure. To determine the water content of the mixed stream, the solver tool of ProMax was used to adjust the pure water stream flow rate to form hydrate at the specified hydrate formation temperature. Table 2 indicates that even the liquid propane water contents are very low (0.00004 to 0.000388 mole fraction corresponding to 4 to 388 ppm by mole), the average absolute percent deviation is 10.2.

 

Table 2. Comparison of liquid propane water mole fractions in equilibrium with liquid water or hydrate by ProMax against the GPA-RR 132 [6] experimental data

The numbers in parenthesis are the adjusted experimental values to form liquid-liquid phases.

 

 

Propane Water Content Charts

As illustrated in Figure 1, for the binary propane–water system, the coexistence of equilibrium phases depends on the system pressure and temperature as follows:

► Propane vapor phase in equilibrium with liquid water

► Propane vapor phase in equilibrium with ice or hydrate

► Propane liquid phase in equilibrium with liquid water

► Propane liquid phase in equilibrium with hydrate

 

 

Figure 2 illustrates the presence of these equilibrium phases as a function of temperature for the isobar of 14.7 psia (101.3 kPa).

 

Figure 2. Water content of vapor and liquid propane as a function of temperature at 14.7 psia (101.3 kPa)

 

 

As illustrated in Figure 1, the propane water content was estimated by the following procedures:

► For temperatures of 200 °F to about 27.5 °F (93 to ~ -2.5 °C), the propane vapor is in equilibrium with liquid water phase so the water saturator tool of ProMax was used.

► For temperatures of about 27.5 °F to -43.3 °F (~ -2.5 to -41.8 °C), the propane vapor was in equilibrium with ice or the hydrate phase, so one mole of pure propane stream was mixed with a pure water stream at a pressure of 14.7 psia (101.3 kPa). To determine the water content of the mixed stream, the solver tool in ProMax was used to adjust the pure water stream flow rate to form hydrate at the specified hydrate formation temperature.

► The propane vapor phase transition to the liquid phase takes place -43.3 °F (-41.8 °C)

► For temperatures of -43.3 °F to -60 °F (-41.8 to -51 °C), the propane liquid is in equilibrium with the hydrate phase, so one mole of pure propane stream was mixed with a pure water stream at a pressure of 14.7 psia (101.3 kPa). To determine the water content of the mixed stream, the solver tool in ProMax was used to adjust the pure water stream flow rate to form hydrate at the specified hydrate formation temperature.

 

Table 3 presents the three phase temperatures and the saturation temperatures of propane–water system estimated by ProMax.

 

Table 3. Three phase temperature and saturation temperature for six isobars

 

Phases: V = Vapor, I = Ice, H = Hydrate, L_W = Liquid Water, and L_P = Liquid Propane

 

 

Similarly, the water content charts of propane vapor and liquid phases were prepared for the other isobars and are presented in Figures 3-5. Note in Figures 3 and 5, due to the very small values, the liquid propane water contents for different isobars fall on the same curve.

 

Figure 3. Water content of propane as a function of temperature for six isobars

 

 

Figure 4. Water content of vapor propane as a function of temperature for several isobars

 

 

Figure 5. Water content of liquid propane as a function of temperature for several isobars

 

 

CONCLUSION

 

Estimating propane water content requires a good understanding of the phase behavior. The process simulation programs have several tools or procedures for estimating the water content. Which one should be used to give a correct answer? In addition to the selection of a suitable equation of state, the selection of the right tool or procedure at a given set of conditions is essential. The choice of a suitable tool changes as the conditions or the equilibrium phases change.

 

The presented propane water content charts can be used for facility type calculations and trouble shooting. It is a good practice to test the performance/accuracy of the selected tool against experimental data first. Obviously, for better understanding of propane–water phase behavior and improving the thermodynamic modeling more experimental data are needed.

 

To learn more about similar cases and how to minimize operational problems, we suggest attending our G4 (Gas Conditioning and Processing) and G5 (Practical Computer Simulation Applications in Gas Processing) courses.

 

By: Dr. Mahmood Moshfeghian

 

---

References

1. Harmens, A. and E.D. Sloan, “The phase Behavior of Propane – Water System: A Review,” The Canadian J of Chem Engr, Vol 68, Feb 1998.

2. Campbell, J.M., Gas Conditioning and Processing, Volume 1: The Basic Principles, 9^th Edition, 2^nd Printing, Editors Hubbard, R. and Snow–McGregor, K., Campbell Petroleum Series, Norman, Oklahoma, 2014.

3. ProMax 4.0, Build 4.0.17179.0, Bryan Research and Engineering, Inc., Bryan, Texas, 2017.

4. Bukacek, R.F., “Equilibrium Moisture Content of Natural Gases” Research Bulletin IGT, Chicago, vol 8, 198-200, 1959.

5. Moshfeghian, M.,  “Ideal Water Content Correlation for Sweet Natural Gas,” PetroSkills TOTM, May 2018.

6. Song, K and R. Kobayashi, “Water content of ethane, propane, and their mixtures in equilibrium with water and hydrates,” Gas Processor Association Research Report (GPA RR 132), Tulsa, Oklahoma, 1991.

7. Soave, G., Chem. Eng. Sci. Vol. 27, No. 6, p. 1197, 1972.