SPE-183555-MS

Reverse Engineering of a Five Stage Centrifugal Compressor Rotor

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Stephan Schwab, TST Turbo Service & Trading GmbH – EU; Dominic Gilardone, TST Turbo Service & Trading
GmbH – ME; Anis Haj Ayed, B&B-AGEMA GmbH

Copyright 2016, Society of Petroleum Engineers

This paper was prepared for presentation at the Abu Dhabi International Petroleum Exhibition & Conference held in Abu Dhabi, UAE, 7-10 November 2016.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
This paper reports about the reverse-engineering of a five stage centrifugal compressor rotor by a Non-
OEM turbo machinery service provider. In order to reach the aim of increasing the lifetime of the rotor, new
enhanced material is used for some compressor rotor parts. The paper describes the method to manufacture a
new rotor based on a used spare rotor. Furthermore, the results of the numerical calculations – stress analysis,
rotor dynamics and flow analysis – are discussed to show how the required performance is achieved and
a safe operation is given.
In the first step, the geometry of the complete rotor is captured with a highly precise and portable 3D-
Scan device. The scan was performed on the assembled rotor and for all single parts after disassembly of the
rotor. Since the scan is performed on a used rotor, a numerical flow analysis and stress analysis are executed
to ensure that deviations or washouts of the used geometry have no effects on the performance and the safe
operation of the rotor. In addition to that, a rotor dynamic calculation is executed for the rotor with original
material as well as for the rotor with the new material. The comparison between these two configurations
provides an indication for a safe operation.
The stress analysis shows a high stress level at the fillet radius to the cover plate of the first impeller
stages. To decrease the stress level in these areas, the geometry is adjusted. The numerical flow analysis
evaluates the flow field in the complete compressor. It is therefore also required to capture the geometry of
the diaphragms. The results show that the performance of the compressor is assured. The rotor dynamics of
the rotor with the new material show only insignificant differences to the rotor with the original material.
After the numerical analysis the manufacturing, assembly and balancing of the rotor will be shortly described
in the current paper.
The successful reverse engineering – which was executed without manufacturing drawings and only
based on an old/used sample – of a centrifugal compressor including the use of enhanced material by a
mid-size Non-OEM company is of substantial benefit to the state of knowledge in the petroleum industry.
Commonly practiced in the European market, especially for industrial steam turbines, reverse engineering
of turbo machinery equipment – without compromising on quality or performance – is to be considered a
major cost saving opportunity for the petroleum industry.
2 SPE-183555-MS

Capture of the Geometry

The geometry was captured three-dimensional with a laser scanner. Besides the rotor also the flow channel
of the diaphragms was captured. In the first step the rotor is scanned without cleaning, after this the rotor
was cleaned and the scan was carried out again. With these two steps all necessary global dimensions are
captured. After scanning of the assembled rotor, the rotor is being disassembled to scan all the single parts.
Besides the capturing of the rotor geometry, the flow channel of the diaphragms was scanned, too. The
capturing of the diaphragms flow channel is necessary to carry out the numerical flow calculation to ensure

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

the necessary performance of the compressor is reached. After the scanning of the geometry, the raw-data
is preprocessed and the geometry is checked for any defects in the scanned raw data. It is also checked if
the entire geometry was scanned satisfactory. The result of this step is a data which is usable to generate
technical drawings with a CAD program and also a data format to carry out the numerical calculations.
Figure 1 shows the old used rotor on the left hand side and the bundle with mounted labyrinths on the right
hand side, which serves as a basis for the scan. The CAD data from the scanning process is depicted in
Figure 2.

Figure 1—Old/used Rotor (left) and Bundle (right) for Scan as the Basis

Figure 2—Rotor mounted in the Bundle (CAD data)

SPE-183555-MS 3

Flow Analysis – Approach

To ensure reaching the required performance, a flow calculation has to be carried out. The following
procedure was done for the calculations. The first step was to calculate the first two stages, which have a
3D-geometry, with the boundary condition of a static inlet pressure and a constant mass flow rate. A result
of this calculation is the outlet pressure of the second stage. The next step is to calculate the last three stages
(2D-geometry). The calculated static outlet pressure from the first step is used as the static inlet pressure.
With this mentioned procedure a good convergence and with that a short calculation time is given. In the

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

last step the complete impellers were calculated. One result of this calculation is the outlet pressure, which
can be compared to the specification of the original machine. If the pressure value is not acceptable, the
impellers can be re-designed as long as the required outlet pressure is reached. After the criteria are reached
the next steps of calculation can be started. The complete calculation area is shown in Figure 3.

Figure 3—Model for Flow Calculations

4 SPE-183555-MS

A gas analysis was executed to determine a representative gas for performing the numerical flow
calculations. A calculation was executed to check the accuracy of the representative gas. In this calculation
the efficiency was given and the result was the outlet temperature, the polytropic head and the power. The
deviations between the calculated values to the representative gas and the values based on the data sheet are
listed in Table 1. The maximal deviations are at 1%, so that the gas can be used for the calculation.

Table 1—Evaluation of the Representative Gas

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Based on Data Sheet Calculated with Absolute Deviation Relative
representative Gas Deviation [%]

Outlet 405.4 406.1 0.7 0.17

Temperature [K]

Polytropic 150.54 152.17 1.63 1.08

Head [kJ/kg]

Power [kW] 10.261 10.304 43 0.42

Stress Analysis and Rotor Dynamics – Approach

Besides ensuring the performance, the safe operation of the compressor rotor is the most important subject
for the engineering phase. For that reason stress analysis and rotor dynamic analysis were executed. The
reason of the necessity of these calculations is the new material of the impellers, the sleeves and the balance
drum, which has a higher density compared to the original material, leading to an increased rotor weight.
Due to that, higher centrifugal forces are expected at the impellers, which need to be checked and evaluated.
Furthermore, the critical speed can be affected by the increased weights.
To ensure that the stress level is uncritical two different extreme cases are considered. The first case will
be called "hot" here below and is defined as follows:

• Complete rotor shaft has the lowest operating temperature

• Each complete impeller has the temperature according to the outlet temperature of each impeller

• Rotation speed is the trip speed

The second case ("cold") is defined as follows:

• Surrounding temperature for the shaft (20°C)

• Impeller temperature (20°C)

• No rotation

The case "hot" simulates the highest load for the impellers and, therefore, the highest stress levels at the
impellers can be expected for this case. The assumption of the temperature levels are also extreme and give
a hint whether the shrinkage is too tight. That means the case "cold" gives a hint about the maximum stress
level at the hub. This condition is apparent if the rotor is out of operation.

Results and Evaluation – Flow Analysis

Figure 4 shows exemplary the evaluation of the skewness angle for the computational grid. The grid was
generated and evaluated for the complete flow path except for the geometry of the labyrinth seals.
SPE-183555-MS 5

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 4—Evaluation of the Skewness Angle of the Grid

The influence of the leakage flow was considered with a 1D-model for the leakage flow path "high-to-
low" pressure. The "high-to-low" leakage flow path is shown exemplary for impeller no. 2 in Figure 5. The
model does not consider the flow path and labyrinth seal geometrically, but surfaces are defined as inlet
and outlet faces at the location shown with a green line in Figure 5. The clearances of the labyrinths are
considered and also the numbers of the labyrinths. Both are important for the determination of the amount of
leakage mass flow. The calculated leakage mass flow is between 0.4% (impeller no. 1) and 0.8% (impeller
no. 5) related to the main mass flow.
The determination of the labyrinth clearance is performed based on the provided new labyrinths
(measured with 3D-scanner and caliper) and the measurements on the rotor parts. Furthermore, labyrinth
clearances from other similar compressors were studied and also an empirical formula was used, which
provides an approximative approach. In addition to that, also the bearing clearances are considered.
6 SPE-183555-MS

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 5—Definition of Leakage Flow Path "high-to-low" Pressure and "high-to-high" Pressure

The second leakage flow path ("high-to-high" pressure or rather "sleeve sealing") was not considered
within the simulation as the leakage mass flows are negligible due to the small pressure difference between
the inlet and the outlet.
Figure 6 and Figure 7 depict the complete calculated flow field. The calculated pressure ratio is 2.706
and hence very close to the specification of 2.7. That leads to the fact that the compressor is overperforming
sligthly. Possible reasons for that are that the calculation might underestimate surface roughness effects,
leakage-main-flow-interaction etc. It is also obvious that the diaphragms only deflected the flow. The
distribution of the Ma-number shows that the Ma-number level is lower than 0.6 in the entire flow field, see
Figure 7. Altogether, the performance of the compressor is given with the applied geometry. No adjustments
in the geometry are necessary.

Figure 6—Aerodynamic Results – Absolute Pressure (static) Distribution

SPE-183555-MS 7

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 7—3D-Visualization of Ma-Number Distribution in the Flow Field

Results and Evaluation – Stress Analysis and Rotor Dynamics

The stress analysis shows maximum local stress much below yield strength for impeller 3, 4 and 5. For
impeller 2 the maximum local stress is at 98% and for impeller 1 at 110% of the yield strength, related to the
lowest value of the yield strength range. The maximum local stress level for impeller 1 is shown in Figure
8. The highest local stress level at impeller 1 is obviously due to its heaviest cover plate.
8 SPE-183555-MS

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 8—Maximum local Stress at Impeller 1, case "hot"

The highest stress level appears at the fillet radius to the cover plate. For that reason this area will be
adjusted with larger fillet radius to reduce the local stress. In Figure 9 the results for all impellers and the
different calculated cases are depicted. Besides the maximum local stress also the average stress is shown.
It is obvious that with the adjustment in the geometry of impeller 1 and 2 the maximum local stress can
be decreased significantly and is lower as for the last three impellers. Furthermore, the average stress level
for both cases ("hot" and "cold") are below 70% of the yield strength. The stress levels ("maximum" and
"average") for the "cold" case are smaller compared to the "hot" case, see Figure 9 below. The reason for
this is that no rotation applies for this case and for that no influence of the centrifugal forces is existent.
Besides the evaluation of the stress, also the amount of shrinkage was evaluated. For that the contact area
to the shaft was considered with the criteria, that there is full contact between rotor and shaft for all cases.
SPE-183555-MS 9

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 9—Comparison of the maximum local stress for each impeller and all calculation cases
(top) and comparison of the average stress level for all impeller and all calculation cases (bottom)

Several rotor parts will be affected by a material change from old/original material to the new material.
This material change especially affects the impellers, the sleeves and the balance drum. Since the density
of the new material is higher compared to the original material used for these parts, a certain change of the
rotor critical speeds is expected. In order to quantify the change in rotor critical speeds, a modal analysis
of the rotor has been performed for two cases:

• Reference case: all parts have been assigned their originally used materials

• New rotor: the material change of several parts to the new material has been considered

The modal analysis is simplified using rigid rotor constrains (infinite bearing stiffness). This
simplification leads to higher critical speeds compared to the real condition where the bearing stiffness
applies. However, the relative change of the rotor critical speeds due to mass differences is insensitive to
this simplification if the simplification is applied to both – the reference case and the new rotor – in the
10 SPE-183555-MS

same way. The calculated relative change in critical speed is used to evaluate the new position of the rotor
in the Campbell diagram and assess its safety margins towards rotation harmonics. This approach is proven
and commonly applied to evaluate the rotor dynamic safety of large steam turbine rotors after retrofitting
using new and modified rotor blading.
The results of the rotor dynamic analysis with rigid bearings show a small deviation between the rotor
with the original material and the rotor with the new material. The critical speed of the rotor with the new
material is a little bit lower due to the higher density, see Table 2. The little difference of the critical speed

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

between the different rotors based on the relative comparision leads to the statement that a safe operation
is given with regards to rotor dynamics.

Table 2—Comparison of calculated Critical Speed between Rotor with

the original Material and the Rotor with the new Material – Rigid Bearings

Calculated Critical Speed (rpm)

Mode Original Material New Material Difference [%]

1 6304 6302 0.038

2 12201 11989 1.741

3 18258 18247 0.059

Based on the calculated relative changes of the critical speeds from the reference rotor to the rotor with
new material, the expected critical speeds of the new rotor have been calculated, see Table 3. The Campbell
diagram (see Figure 10) has been drawn considering a lower speed limit of 5350 rpm and an upper speed
limit of 9732 rpm (trip speed). The smallest safety margin is found between the 1st critical speed of the old
rotor (4600 rpm) and the 1st rotation harmonic at the lowest speed limit (5350 rpm). The safety margin is
approx. 750 rpm. The application of the new material leads to a slightly lower critical speed (4598 rpm)
and thus a larger safety margin (752 rpm). All other calculated safety margins are also larger. Thus, it can
be concluded again that the material change affecting several rotor parts has no significant influence on the
dynamic behavior of the rotor.
SPE-183555-MS 11

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Figure 10—Campbell Diagram – Comparision between Rotor with original (VIRGO 38) and new (INCONEL 718) Material

Table 3—Comparison of calculated Critical Speed between Rotor with the old and the new
Material – Adjusted to the Specification "Centrifugal Compressor Data Sheet – BCL 505"

Calculated Critical Speed (rpm)

Mode Original Material New Material Difference [%]

1 4600 4598 0.038

2 11700 11496 1.741

3 14800 14791 0.059

Procurement of Material and Manufacturing

The procurement of the new material and the manufacturing was done in accordance to [1], Furthermore,
amendments from the client were also taken in to consideration for the execution, see [2] and [3], That means
in the case of procurement of the raw material lot material testing was executed. For that for each piece a test
pieces, which has exectly the same heat treatment was cutted to carry out the testing. The material testing
was executed by first cutting a test piece of every raw material piece and then executing the following test:

• Tensile test (tensile strength and yield strength)

• Microstructural analysis

• Hardness test

• Notch impact test etc.

12 SPE-183555-MS

All test results was evaluated and fits to the requirements described in [1] and [2].

Overspeed Test, Assembly and High Speed Balancing

After the manufacturing of the impeller an overspeed test was executed. Before the overspeed test was
executed the geometry are measured with CMM measurement system. Subsequently a low speed balancing
of each single impeller was done before the overspeed test was executed. After the overspeed test the same
CMM measurement was carried out to observe if big deformation are existent. After these steps the complete

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

rotor was assembled. The assembly was carried out with axial and radial run-out measurements. After this
the complete rotor was high speed balanced, see Figure 11. Before the delivery of the rotor, the rotor was
mounted in the provided bundle for a comprehensive check of the labyrinth clearances.

Figure 11—Rotor at the High Speed Balancing Facility

Conclusion
In the current project a re-enginnering of a five stage centrifugal compressor was done. Based on an old
and used rotor the original geometry was captured. A detailed flow analysis was carried out to ensure the
required performance. Due to the retrofit of the compressor rotor with different material for the impellers,
sleeves etc. the rotor weight increases. To ensure a safe operation of the rotor and the complete compressor,
a stress analysis and rotor dynamic calculations were performed. After this analyses and evaluations, the
manufacturing of three compressor rotors was executed. All worksteps was done in accordance to applicable
international codes and standards.

Acknowledgments
The authors gratefully acknowledge Peter Gilardone (Managing Director – TST) and Dr.-Ing. Jürgen
Dünnwald (Technical Director – TST) for their support. The numerical flow simulations presented in this
SPE-183555-MS 13

paper have been carried out with the STAR-CCM+ Software of CD-adapco. Their support is gratefully
acknowledged. The responsibility for the content of this publication lies with the authors.

References
1. N.N. – Age-Hardened Nickel-Based Alloys for Oil and Gas Drilling and Production Equipment –
API Standard 6A718, third Edition – Date N.N.
2. N.N. – Specificatzion for Nickel Base Alloy (UNS NO7718) for Oil and Gas Drilling and

Downloaded from http://onepetro.org/SPEADIP/proceedings-pdf/16ADIP/4-16ADIP/D041S091R005/1423972/spe-183555-ms.pdf/1 by Saudi Aramco user on 04 September 2023

Production Equipment (Amendments/Supplements to API Standard 6A178) – SP-2188, first
Version – Date N.N.
3. N.N. – Specification for Quality Management System Requirements for Product and Service –
SP-1171, third Version – Date 24.06.2012