Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90235

PERFORMANCE DEGRADATION MONITORING OF CENTRIFUGAL COMPRESSORS USING DEVIATION ANALYSIS

Mohd Shahrizal Jasmani PETRONAS Carigali Sdn. Bhd. Kerteh, Terengganu, Malaysia Thomas Van Hardeveld Strategic Maintenance Solutions Inc. Calgary, Alberta, Canada

Mohd Faizal Bin Mohamed PETRONAS Carigali Sdn. Bhd. Kerteh, Terengganu, Malaysia

ABSTRACT Performance degradation monitoring of centrifugal compressor provides a means for the operators predict the behavior of their machines. Understanding the key principles in performance evaluation is essential for operators to benefit from this approach. In this paper, common performance degradation mechanisms found in centrifugal compressors for the oil and gas industry are outlined and related to their associated performance characteristics. Various analysis and evaluation techniques and approaches are elaborated with relevant requirements and assumptions for practical site application. A case study is also presented to demonstrate the application of performance degradation monitoring in a real-life operating environment. The benefits and limitations of the approach are also discussed. When combined with other condition monitoring approaches, this method provides a powerful tool to analyze and monitor centrifugal compressor performance which will then lead to useful recommendations for maintenance and operational interventions. INTRODUCTION Centrifugal compressors are used widely in the oil and gas industry, particularly in upstream production, hydrocarbon refineries, petrochemical processing and pipeline transmission. This particular type of equipment is suitable for low to high pressure ratios with the advantage of an ability to handle high flow rates as compared to positive displacement machines. As with any rotating machinery, centrifugal compressors are susceptible to performance degradation during their operation.

The challenge is for operators to identify and quantify these degradations as the performance impact of an underperforming compressor is usually significant to the operation of the facility and ultimately to production capacity and even the profit of the operating company. Traditionally, operators rely on condition monitoring information such as vibration and temperature data and lube oil analysis to assess the condition of a gas compressor. While effectively providing some indication of the health of a machine, one of the main drawbacks of this approach is that they are not directly indicative of performance losses and these symptoms usually occur after major degradation has taken place. This leaves operators little time to prepare for maintenance intervention prior to unplanned equipment breakdown. Performance analysis on the other hand is able to provide an early indication of degradation and can potentially minimize downtime required for rectification efforts. It provides strong justification to perform corrective actions and allow ample preparation and planning prior to repair and overhaul activities. If possible, process interventions can also be made to slow down the declining effects. When combined with an existing monitoring approach, the introduction of performance analysis and monitoring provides a holistic evaluation of any machinery. Centrifugal compressors function based on the principles of aerodynamics and thermodynamics, and hence a clear understanding of how these elements relate to compressor performance is important for operators. The Original Equipment Manufacturer (OEM) is very conversant with the design of their products and they usually furnish selective design parameters and performance curves to their customers. This available information can be used by the operators as the baseline to evaluate compressor performance in service.

Copyright 2012 by ASME

Fundamental knowledge on degradation mechanism, evaluation techniques and calculation approaches will assist operators to realize the benefits of performance assessment in their operation. The use of standard computerized tools such as spreadsheets greatly facilitates the evaluation process. NOMENCLATURE H HP k LP MW N n P Q R s T Head Efficiency High Pressure Ratio of Specific Heats Low Pressure Molecular Weight Rotational Speed Polytropic Coefficient Pressure Flow Rate Density Universal Gas Constant Entropy Temperature Mass flow rate Compressibility factor

DEGRADATION MECHANISMS When the performance of a centrifugal compressor in service is affected, the effects of degradation will at some point be evident to the operating staff from basic field measurement readings. For example, a decrease in head capability will affect the discharge pressure of the compressor and jeopardize the downstream processes. A reduction in compressor efficiency will increase energy consumption and affect the overall operating cost. Restrictions in gas flow due to fouling or blockage will decrease the flow throughput and will be detrimental to operational performance especially in gas export or volumetric gas transfer. A reduction in gas flow will drive the operating point closer to surge, which becomes intolerable if the maximum power of the driver has also been reached. The major mechanisms for performance degradation are mechanical damage, fouling or blockage, erosion and corrosion and increased internal recirculation. Mechanical damage may be caused by external upstream devices such as filters or cracking of internal gas path components. Sudden increases in vibration levels will quickly be evident and likely cause a shutdown so this type of degradation is not a good candidate for medium and longer term performance monitoring. A more applicable and common degradation mechanism is fouling or blockage. Operating a centrifugal compressor in hydrocarbon services is potentially damaging to compressor performance due to a multitude of gas component mixtures and the presence of contaminants such as wax, sand and free liquids in the natural gas, particularly close to the wellhead. Fouling is also encountered in petrochemical applications such as

refrigeration and coal gasification [1]. Fouling restricts the flow of the gas path and distorts the gas inlet and exit angles at the impeller vanes causing excessive turbulence from flow separation both in the impeller and the diffuser. These conditions reduce the head produced, the compressor flow capacity and compression efficiency, resulting in higher gas power requirements and an increase in gas discharge temperature. Higher levels of vibration due to imbalance usually result. Erosion can be caused by impingement of free liquids or solids onto the compressor rotating or stationary parts. Changes in the dimensions will disturb the flow vectors of the impellers and lower the head produced and compression efficiency. Severe erosion can cause rotor imbalance due to metal loss and create high radial vibrations, which could damage the bearings or erode the internal seals. Corrosion can be caused by presence of water in the natural gas or other corrosive agents such as mercury, hydrogen sulfide or carbon dioxide. If the contaminants enter a closed-loop dry gas seal system, liquid or solid carry over would damage the seal faces, stationary elastomer or pusher springs and would eventually cause the seal assembly to seize. A somewhat different degradation mechanism is internal recirculation or leakage caused by increased labyrinth seal clearances. It has been reported that higher internal leakages can result in flow distortions of up to 15-20% which may increase the internal flow into the choke region and reducing the net flow in the process by the same amount [2]. The effect is especially severe for multi-stage compressors where performance losses are multiplied by number of impellers, interstage seals and balance pistons. A possible cause of higher internal leakage is one or more surge events that could excite severe vibration and damage the compressor internals. The labyrinth or honeycomb seals at the balance piston or disk could be eroded, which alters the rotor axial thrust. This condition would also cause high leakage rate at the dry gas seal due to pressure imbalance at the discharge end. EVALUATION REQUIREMENTS Prior to any performance analysis of a compressor being conducted, adequate measurements need to be readily available at site. A comprehensive measurement setup is required for shop testing such as outlined in the ASME Performance and Test Code PTC-10 [3] but a more practical guideline for onsite testing has been produced by the GMRC [4]. Compressors installed at site are usually equipped with minimal measurements without redundancies, as shown in Figure 1, as follows: a. Suction Pressure, Ps b. Discharge Pressure, Pd c. Suction Temperature, Ts d. Discharge Temperature, Td e. Inlet Flow Rate, Qa f. Rotational Speed, N1

Copyright 2012 by ASME

Figure 1: Typical Measurement Arrangement of a Centrifugal Compressor at Operating Site The accuracy and precision of these measurements are critical toward the accuracy of the thermodynamic calculations. Hence, it is very important that the field measurement devices to be inspected and calibrated prior to any performance analysis. However, even if the measurements are well calibrated, it is imperative that measurement errors need to be considered in the evaluation as defined by the device manufacturers. Table 1 shows typical acceptable uncertainties for site performance testing measurement [4]. Measurement Pressure Temperature Flow Torque Gas Composition Achievable Uncertainty 0.3 2.0% full scale 0.3 4.0 C 0.5 2.0% of value 0.5 1.5% of value 0.2 3.0% of value

Ideal and Actual Compression Processes An ideal compression process usually involves no entropy change and is termed isentropic compression, i.e. without any heat loss (adiabatic) and a reversible process. In actual fact, most compression takes place with a positive entropy change as per the Second Law of Thermodynamics. This produces a higher temperature compared to the isentropic process to achieve the same discharge pressure and is referred to as polytropic. Isentropic or adiabatic compression is satisfactory for lower pressure ratios but polytropic will be more accurate for pressure ratios higher than about 2. This phenomenon can be best visualized using a temperature versus entropy (T-s) diagram as illustrated in Figure 2.

Figure 2: T-s Diagram for Ideal and Actual Compression Processes Compressor Performance Curve The compressor performance curve is typically expressed in isentropic head versus actual flow with constant rotational speed, commonly expressed in rotation-per-minute (RPM) with efficiency lines at constant gas molecular weight, as shown in Figure 3. Note that some OEMs use polytropic head, pressure ratio or discharge pressure with varying gas molecular weight to cater for changes in process conditions. In this paper, isentropic head is mostly used. Compressor Head Compressor head can be defined as the measure of energy being imparted per amount of gas being compressed, usually expressed in kilojoule per kilogram of gas (kJ/kg). The isentropic head is usually related to changes in pressure ratio or gas properties as shown in Equation 1. Note that polytropic head can be calculated using the same equation by substituting the ratio of specific heats, k with polytropic coefficient, n.

Table 1: Typical Measurement Uncertainties [4] On top of the accurate measurements, the operator also needs to have basic information on the compressor such as design data, compressor drawings, operating philosophy, auxiliary diagrams and most importantly, the compressor performance or wheel curve at design conditions from the OEM. This curve will serve as the baseline comparison as discussed in the following sections. Accurate gas composition data is also important to determine gas properties such as molecular weight (MW), compressibility factor (z) and ratio of specific heats (k). PERFORMANCE FUNDAMENTALS A dynamic compressor increases the gas energy by converting the mechanical rotational energy into elevated pressure and temperature. The derivations of fundamental thermodynamic equations are discussed in detail in a number of literature sources [5]. In this paper, only relevant concepts and performance parameters such as head, flow and efficiency are discussed.

H isen

P R k = z ave Ts d Ps MW k 1

k 1 k

(1)

Copyright 2012 by ASME

parameter can be expressed in isentropic or polytropic forms. For isentropic efficiency, the relationship is shown in Equation 4.

isen

Pd k 1 P s 100% = Td 1 T s

k 1

(4)

Figure 3: Typical Centrifugal Compressor Performance Curve Note that all thermodynamic calculations in this paper are using SI units, hence absolute pressures are in Pascal (Pa) and temperatures are in degrees Kelvin (K). The natural gas properties can be calculated from the gas composition by using an equation of state such as those by Benedict-Webb-Rubin (BWR) and Redlich-Kwong (RK). The final selection of the equation of state will depend on the actual gas properties and process conditions and the most critical thing is to maintain consistency between all phases of the testing and analysis. [4] Standard and Actual Flow In operation, the inlet gas flow into the compressors is usually measured and converted to standard flow as reference with Standard Temperature, Tstd of 15 Celsius (288.15 K) and Standard Pressure, Pstd of 1 atm (101.325 kPa). For compressor performance evaluation, this standard flow must be converted into actual inlet conditions i.e. at particular inlet pressures and temperatures as shown in Equation 2.

Gas Power Gas power is the amount of energy required to compress a volume of gas to a specific discharge pressure and temperature expressed in Watts. It is a function of compressor head, actual inlet gas mass flow rate and efficiency as shown in Equation 5. This parameter is another performance indicator to determine if the compression process is consuming more power than it should.

Powergas =

mact H isen

(5)

isen

P Tact zact Qact = Qstd std Pact Tstd zstd

(2)

COMPRESSOR PERFORMANCE DEVIATION The basis for monitoring compressor performance is recognizing that the compressor characteristic curve as shown by the speed lines shifts downward and to the left as degradation occurs. In other words, to achieve a certain head and flow a higher speed is required. At the same time, there is probably also a decrease in the efficiency. From the previous discussion on degradation mechanisms, it was seen that the relative amounts of head and flow decrease is dependent on the nature of the degradation so this cannot be easily predicted. However, if the speed lines are fairly flat, a reasonable approximation is to monitor the vertical drop in head and ignore the change in flow for trending. Since the purpose is to detect changes in performance, even if this approach overstates the situation somewhat, the ultimate objective is still served. Therefore the primary measure being proposed here is the deviation in head at the same flow as will be shown in the different methods now described below. Method #1: Manual Plotting Although it is best to automate the calculation and trending of head deviation, there are instances where all one wants to do is to plot one point and check the deviation from the compressor curve. An example is shown in Figure 4. First the calculated head (using Equation 1) and flow are plotted on the compressor curve. For this point, the speed according to the compressor curve is estimated by interpolation, in this case 19,700 rpm. The actual operating speed however is 20,200 rpm. This means that without degradation, the compressor should be able to achieve a head of 126 kJ/kg at the same flow. The deviation in head can then be calculated as -6.34%.

The gas mass flow rate can be calculated from the volumetric flow rate, Qact and gas density, in at inlet conditions as in Equation 3.

mact = Qact in

(3)

Compressor Efficiency Efficiency is the ratio of useful work versus the total work supplied to a system. Since for a compressor, a part of the energy supplied results in an increase in temperature instead of the desired pressure increase, compressor efficiency is the ratio of useful pressure work versus the total work. It is a good indicator of overall compressor performance where a lower efficiency indicates less effectiveness of the compression process and additional energy losses such as fouling, recirculation and seal leakage. Similar to compressor head, this

Copyright 2012 by ASME

Figure 5: Compressor Head Deviation Calculation

Ref. N1 N2 Q1 Q2 Description Current Rotational Speed Reference Rotational Speed Current Actual Flow Corrected Actual Flow Current Operating Head Corrected Head Expected Head Head Deviation Definition Measured rotational speed of the degraded compressor at the point of evaluation Reference speed closest to the available constant speed line on the reference curve Measured inlet flow rate into the compressor Corrected flow rate (using Fan Law) of the degraded compressor at corresponding N2 Calculated head based on inlet and outlet conditions and gas properties Corrected compressor head (using Fan Law) of the degraded compressor at corresponding N2 The head that should be achieved at N2 and Q2 on the reference curve The percentage of difference between H2 and H3 Unit RPM RPM m /hr m /hr
3 3

Figure 4 Head Deviation by Manual Plotting Using Method #1 Method #2: Similarity Condition Matching For ease of calculation and trending, it is easier to automate this process, possibly by using a spreadsheet, as illustrated in Figure 5 with nomenclature in Table 2. This concept is described in [3,4] as the similarity condition approach via parameter matching. This method starts by curve fitting the speed lines to describe the baseline compressor curve. The head at the actual operating point is then corrected to the nearest speed line using the well-known Affinity or Fan Laws [6] which state that the head varies as the square of the speed. The corrected head can then be compared to the curve fit of the design curve at that speed line. Depending on the shape of the speed lines, a second order curve fit may be adequate or it may be necessary to use a third order polynomial to obtain a proper fit for the higher flow end of the curve.

H1 H2

kJ/kg kJ/kg

H3 H

kJ/kg %

Table 2: Nomenclature for Compressor Head Deviation Evaluation Note that in order to minimize inaccuracy of Fan Law estimation, the reference speed, N2 on the reference curve shall be selected as close as possible to current operating speed, N1. For example, if the operating rotating speed is 10,235 RPM and the available lines on the reference curve are 10,000 and 11,000 RPM, the former shall be selected as N2. Thus, the head deviation can then be determined from Equation 6.

H =

H2 H3 100% H3

(6)

Copyright 2012 by ASME

After understanding the fundamentals of compressor performance and the significance of using the Fan Laws, the following are the steps and assumptions to conduct head deviation monitoring with reference to Figure 5: 1. Acquire a snapshot of the operating parameters such as Ps, Pd, Ts, Td, Qs and N1 from field measurement devices 2. Determine relevant gas parameters such as z, k, and MW from gas composition analysis 3. Calculate H1 from Equation 1. This indicates the compressor head at current operating conditions (Point 1) 4. Calculate operating actual flow, Q1 from Qact relationship in Equation 2 5. Choose reference speed, N2 nearest to N1 on the reference compressor performance map 6. Calculate Q2 and H2 from the Fan Law relationship in Equation 6 and 7 respectively to determine corrected operating flow and head at reference speed, N2 (Point 2). 7. Now both the speed and flow are normalized for both degraded and design conditions, head comparison can be made. Draw a vertical line on Q2 in the reference map and read the head value on the Y-axis at N2 intersection. This value is the expected operating point (Point 3) per compressor design with expected head, H3. 8. Finally, calculate the head deviation, H from Equation 8. Using the data from the example above and the nomenclature in Table 2, the corrected head will be
! N2$ ! 20,000 $ H 2 = H1# & = 118 # & = 115.67 kJ/kg " N1 % " 20,200 %
2 2

actual operating point is at lower speeds which may make this a bit more inaccurate, particularly if it is a high pressure ratio compressor where the Fan Law correlation will not be as applicable. Method #3: Performance Software Some software is available to trend degradation of which the following is an example. A product of Windrock Inc., MHealth [7] has a number of features that facilitate the trending of degradation. The first one is the ability to calculate head, actual flow, compressor efficiency (isentropic or polytropic) and gas-related parameters from a gas composition. It further models the compressor characteristic curve using a second order curve fit for both head and efficiency as

H = h1 N + h2 NQ + h3Q 2 Q Q E = e1 + e2 + e3 N N
where
2

(10)

h1 , h2 , h3 are constants for the head equation e1 , e2 , e3 are constants for the efficiency equation H is head in kJ/kg Q is actual flow in m3 /s N is compressor speed in rpm E is efficiency in %
Once the constants are derived by curve fitting either actual test data or points from the manufacturer curve, these two equations are then used to calculate and trend the deviation in head and efficiency. Limitations A few words of caution are appropriate in order to obtain useful results. 1. Using a tested performance curve should give the best comparison but most manufacturer curves will be quite adequate. 2. If the compressor is operating in choke, the compressor curve will probably not be very accurate so the deviation data will be less reliable. 3. The accuracy and applicability of the input pressures, temperatures and flows will of course affect the calculated results. For example, the location of discharge instrumentation should not be too far downstream. 4. If the gas composition changes, regular samples will have to be taken and analyzed. 5. The gas composition, especially the molecular weight, should be close to that used for the compressor curve.

(7)

and the corresponding inlet flow is

! N2$ ! 20,000 $ 3 Q2 = Q1# & = 550 # & = 544.55 m /hr 20,200 % N1 % " "

(8)

From the curve fit, the head H3 at inlet flow Q2 will be 122.44 kJ/kg so that the deviation in head will be

H =

115.67 122.44 = 5.53% 122.44

(9)

This deviation value is a bit lower than estimated by the manual method. This is due to the fact that the similarity condition method also takes into account the decrease in flow which then decreases the difference in head. As a variation of this method, there are cases where the manufacturer supplies only the 100% speed line. The only thing this will affect is potentially the size of the correction if the

Copyright 2012 by ASME

6. 7. 8. 9.

ASME PTC 10 states the acceptable deviations in test parameters for similarity conditions [3] as shown in Table 3. It is better if the compressor is not recycling although, if the flow measurement is inside the recycle loop, the calculated data should still be valid. Repeatability is actually more important than absolute accuracy and this can be verified by data variation on the trend. It is important to start trending at the beginning with a clean compressor because then the agreement between the curve and the measured point can be verified. Figure 6: LP Compressor Head Deviation Trending (2008 to 2010) Using Method #2

Table 3: ASME acceptable deviations in test parameters for similarity conditions [3] CASE STUDY This case study is based on an offshore oil producing platform located at South China Sea, to the East of Peninsular Malaysia which is equipped with a gas compression train for well gas lifting purposes and is operated by PETRONAS Carigali Sdn. Bhd. The train consists of two compression stage casings, namely LP and HP compressors; both are the C1607 model manufactured by Solar Turbines Inc. The compressors are coupled in tandem with a speed-increaser gearbox and driven by a Solar Centaur 40 dual-shaft industrial gas turbine, as illustrated in Figure 5.

Figure 7: HP Compressor Head Deviation Trending (2008 to 2010) Using Method #2 Note that there are durations where no data is available for performance analysis as the compressor is out of service due to maintenance activities or operational constraints. Using the M-Health software, the trend of the head deviation for both LP and HP compressors as shown in Figure 8 gives the same result as the spreadsheet method above including data from 2008.

Figure 5: Gas Compressor Train Layout Performance Analysis Since 2006, these compressors have suffered from severe fouling due to wet feed gas and have required to be overhauled every six months. Performance analysis using head deviation trending was done for both LP and HP compressors from January 2009 to December 2010. The results are shown in Figure 6 for the LP compressor and Figure 7 for the HP compressor using method #2.

Figure 8: Deviation in Head for LP and HP Compressors (2008-2010) Using Method #3

Copyright 2012 by ASME

Fouling should also result in a drop in efficiency relative to the compressor curve but as Figure 9 illustrates, this does not seem to occur consistently. This is perhaps not surprising since compressor temperatures are much more difficult to measure accurately with a major effect on the calculation of efficiency.

(a) Figure 9: Deviation in Efficiency for LP and HP Compressors (2008-2010) Using Method #3 Finally, it is interesting to see that for the LP compressor the flow also decreases with fouling with the speed remaining more or less constant. Other than in 2009, for the other two years this does seem to be the case.

(b) Figure 11: (a) Impeller Fouling (b) Damaged Labyrinth Seals An improvement project was also introduced to inject a corrosion inhibitor and anti-fouling chemical into the compressor system. Injection points were located between the cooler and suction scrubber for both the LP and HP compressors and injection rates were optimized at 2-3 L/mmscf based on water analysis results. By May 2010, the installation was completed and the newly overhauled compressors were put back online. Instantaneously, the head deviation was reduced for both LP and HP compressors and the effectiveness of the anti-fouling chemical is observed by monitoring the head deviation gradients. As shown in Figure 6, the slope for LP compressor significantly reduced before and after chemical injection was introduced. However, for HP compressor there is only slight improvement as shown in Figure 7. The project team was notified and it was recommended to fine tune the chemical injection rate. It was also interesting to note the effect of fouling on compressor vibration levels. Figure 12 shows the vibration

Figure 10: Changes in Head and Actual Flow for LP Compressor (2008-2010) Using Method #3 Discussion Based on the trending, it was decided to remove both compressors for a full overhaul at an onshore workshop in October 2009 as the compressors could not deliver the required discharge pressure for gas lift purposes. This decision was supported with the rapid decreasing head deviation trending, which reached -9.75% for LP and -13.02% for HP compressors as shown in Figure 6 and 7. During overhaul, the compressors internals were found to be heavily fouled with sludge as shown in Figure 11(a) and labyrinth seals were found damaged (see Figure 11(b).

Copyright 2012 by ASME

levels for the HP compressor from 2008 until 2010 and verifies the positive impact of the anti-fouling chemical injection.

As a conclusion, understanding and utilization of knowledge of the compressor dynamics plus consistent performance monitoring is recommended as a good practice for prudent operation and maintenance of centrifugal gas compressors, particularly where degradation is known to be a problem. ACKNOWLEDGEMENT The authors would like to convey our gratitude to PETRONAS Carigali Sdn. Bhd and Windrock Inc. for the permission to utilize the proprietary data, information and applications used in this publication. REFERENCES [1] De Maria, R.L. and Gresh, M.T, The Role of Online Aerodynamic Performance Analysis, Proceedings of the 35th Turbomachinery Symposium, 2006, pp. 55-61. [2] Akhtar, M.S. Determining the real performance of centrifugal compressors operating in oil & gas production facilities, Ninth European Fluid Machinery Congress, Institution of Mechanical Engineers (IMechE), May 2006. [3] ASME Performance Test Code on Compressors and Exhausters PTC 10-1997, ISBN: 0791824500. [4] Brun, K. and Nored, M.G., Guideline for Field Testing of Gas Turbine and Centrifugal Compressor Performance, Gas Machinery Research Council, Southwest Research Institute, Release 2.0 August 2006. [5], Mohitpour, M., Botros, K.K. and Van Hardeveld, T., 2008, Pipeline Pumping and Compression Systems: A Practical Approach, ASME Press, ISBN10: 0791802786. [6] McChane, R., 1992. The Use of Fan Laws to Curve Fit Centrifugal Compressor Operating Data, Pipeline Simulation Interest Group, Corpus Christi, Texas. [7] www.windrock.com/online-files/Mhealth.html, accessed Oct 25, 2011.

Figure 12: Trend of Vibration Levels for the HP Compressor CONCLUSION As demonstrated in the case study, head deviation monitoring is useful to identify decreasing trends of centrifugal compressor performance. It was demonstrated that the method can also be used for monitoring of process improvement initiatives. Recommendations can be made with high confidence on maintenance interventions such as repair and overhaul, or proposed changes to the operating parameters such as optimizing the chemical dosing rate. When used in combination with other monitoring data plus computerized tools, the method can be a powerful analysis for compressor performance. However, there are a few limitations that need to be considered and observed. Measurement accuracy and validity are a major challenge as they highly influence trend behavior. Suitable locations of measurement instrumentation also need to be assessed, as well as variation of suction conditions and gas properties.

Copyright 2012 by ASME