Most engineers that work with turbomachinery are fairly familiar with centrifugal compressor maps. These maps usually show dimensional head, pressure ratio, or pressure versus flow for various speed lines. Sometimes engineers also use non-dimensional maps that show non-dimensional head coefficient versus flow coefficient, usually on a single line. The dimensional maps also always show a flow-limiting surge line and sometimes a choke restriction line. Some compressor maps may even show efficiency lines that form “islands” around the best efficiency point on a dimensional map. Whether one uses dimensional, non-dimensional, head-flow, pressure ratio-flow, or pressure-flow maps is a matter of convenience since all these maps, if accurately drawn, show identical information and can easily be converted from one to another.

Surprisingly, there is little understanding among industry as to where these maps actually come from and what their underlying physics is. The usual answer I get from engineers is they come from “testing”—while technically correct, it isn’t a very satisfactory response and provides no insight whatsoever. Don’t get us wrong, most compressor maps supplied by compressor manufacturers come from the superimposition of individual impeller stage maps derived from test data and computational fluid dynamics, or they are direct performance measurements from a closed-loop factory or field site test of the actual unit. But more physical understanding is always helpful, so let’s dive a little bit into what makes a centrifugal compressor map.

At its most basic, a centrifugal compressor map can be derived from inlet and discharge flow velocity vector diagrams and associated flow losses for individual stages. Euler’s turbomachinery equation fundamentally relates a compressor’s actual head, i.e., the work input into the gas, to flow components at the compressor impeller inlet and discharge, or axial, radial, and tangential polar velocity vectors in the stationary frame. Each point on the compressor map corresponds to a unique set of flow velocity vectors entering and exiting a compressor’s impeller. These are primarily the axial/radial components in the inlet and the tangential/radial flow components in the discharge. For mixed-flow impellers, the axial component should be considered in the discharge.

An isentropic or polytropic head-flow map is therefore a graphic version of Euler’s turbomachinery equation with some head-loss factor corrections. For example, looking closely at the equation, one can derive that an impeller with purely radial discharge blades should theoretically have horizontal speed lines, and the more back-sweep an impeller has the steeper these lines become. This holds mostly true for flow near the best efficiency point of the compressor but deviates at the map’s extremes. Specifically, for the actual map, we subtract losses from the ideal lines and these losses become more significant at high and low flows where incidence-angle secondary losses and viscous-flow losses start to dominate. Thus, these head-flow lines start to drop at low and high flows based on non-ideal pressure losses. Near surge, the head-flow lines become first horizontal, passing through the peak head, and then start to drop. On the other side of the map, the head-flow lines drop steeply toward zero head after the flow nears the choke region.

For compressors that run at variable speeds, we can construct a family of curves for different speeds from the single basic head-flow curve. Speed lines on the map tend to be similar (nearly parallel) to each other since they are linear extrapolations using the fan law of identically proportioned velocity vector diagrams. Specifically, for a given set of 3D impeller inlet/outlet velocity vectors, there is a near diagonal line on the compressor map of identically proportioned vectors with either larger or smaller total magnitudes but identical flow angles. This is also the reason that, based on the fan laws, the compressor map can be non-dimensionalized (using rotational speed squared and radius squared for head and rotational speed and radius cubed for flow) so that all speed lines collapse onto a single line on a non-dimensional compressor map. This holds mostly true, except for some deviations due to compressibility (Mach Number) and viscous (Reynolds number) effects. Manufacturers usually correct for compressibility and viscosity effects using empirical equations, but there are no industry-wide, universally accepted correction factors or correlations.

The non-dimensional compressor impeller map is often also called the “phi-psi map” based on the Greek letters used for non-dimensional flow and head, respectively. This map shows a single line—or a few nearly perfectly overlapping lines—of efficiency versus non-dimensional flow, which is an indication that efficiency is not drastically affected by the dimensions of the compressor and that a compressor is size-scalable with limited effects on efficiency for moderate changes in size. Specifically, efficiency is primarily affected by the incidence and secondary velocity vectors into and out of the machine while compressibility and boundary-layer friction effects are relatively minor and can be accounted for using Mach and Reynolds number corrections, or by calculating or testing multiple head flow lines at different Mach numbers. This is also the reason that on the dimensional head versus flow or pressure ratio versus flow map the efficiency lines appear as very elongated oval-shaped islands: Lines of constant efficiency are essentially the parabolas that are dictated by the fan law. Based on what is seen on the non-dimensional efficiency map, these lines should appear as elongated parabolas on the dimensional head-flow map that corresponds to the extrapolation of the fan laws. In reality, these lines will close to an oval shape at high head because of the impact of compressibility (Mach number) and viscous flow losses (Reynolds number).

We have discussed compressors at constant or variable speeds but with fixed geometry. Compressors with variable geometry such as variable inlet guide vanes or variable diffuser vanes have different head-flow characteristics for each position of these vanes. An inlet guide vane imparts swirl to the inlet flow, either in the direction of the impeller rotation (pre-swirl) or against it (counter-swirl). Therefore, they impact the work imparted by the impeller according to Euler’s law, thus reducing the head for pre-swirl and increasing the head for counter-swirl. Adjustable vanes in the diffusor move the stability limit of the diffusor, i.e., they allow the compressor to operate at diffusor inlet angles that would lead to diffusor stall for fixed geometry machines. They move head-flow lines almost parallel to lower flows, limited by the onset of impeller rotating stall and clearance losses for the vanes.

Check out part II, a compressor map’s extrema: surge and choke, in the annual 

 2025, which will be released mid-October.

Klaus Brun is the Director of R&D at Elliott Group. He is also the past Chair of the Board of Directors of the ASME International Gas Turbine Institute and the IGTI Oil & Gas applications committee.

Rainer Kurz is a recent retiree from Manager of Gas Compressor Engineering at Solar Turbines Inc. in San Diego, CA. He is an ASME Fellow and has published over 200 articles and papers in the turbomachinery field.

Articles

Part one of this two-part myth on a centrifugal compressor map covers the basic physics of the constant speed head-flow performance lines.

Myth: Part 1: A Centrifugal Compressor Map is Just Test Results

Turbocompressors with side streams offer very complex flow patterns and aerodynamic behaviors. They are used in plants on large, high-power trains for critical services and in larger, more complex arrangements. Historically, compressors with side streams presented difficulties in engineering, simulations, design, testing, commissioning, operation, and monitoring, and often operate at relatively low performance and efficiency.

Turbocompressors with side streams should be operated with tight tolerances on operational parameters compared to conventional compressors. If compressor performance is maintained within certain operational limits, corresponding errors in flow ratio and pressure at the suction and side-stream inlet must be maintained within some bounds. 

As a rule, the recommended variation tolerances for volume flow ratio (side-stream volume to core-flow volume) and inlet pressures (suction inlet and side-stream inlet) are ±4% and ± 2%, respectively. Tighter tolerances are often needed.

Heavy gases—such as propane, propylene, methyl tertiary butyl ether (MTBE), etc.—have very low gas sonic velocities that produce high Mach numbers in aerodynamic flow paths. For these services, each compressor stage features narrow flow maps characterized by limited surge and choke margins.

Compressors with side streams in these applications present great difficulties; the side streams and associated mixing further complicate the operation and performance prediction because the pressure, temperature, flow conditions, and each section’s exit should be within stringent tolerances to optimize the overall efficiency and performance. It’s important to accurately model, simulate, operate, and monitor the performance characteristics of impellers, diffusers, return channels, and all flow-path components.

An important step to operating, evaluating performance, and monitoring a compressor with a side stream is accurate simulation and data verification. Ask the vendor for the compressor curves’ calculation basis. It’s important to know if mixing effects, mixing section losses, and other complex variables are considered. This is different from a conventional compressor, as vendor curves cannot be used for operation, monitoring, and performance evaluations unless the basis and assumptions are fully understood. 

These turbomachines should be operated with tight tolerances of flow, pressure, and temperature at the suction and side-stream inlet. Any deviations in operational conditions at these inlets will result in considerable losses at mixing sections and other key parts. Off-design compressor operation may produce additional effects. The result is a significant drop in overall efficiency and other degradations, abnormalities, and deviations.

A 4.5-MW variable-speed electric-motor driven refrigeration turbocompressor was deployed for a critical refrigeration service with a side stream for plant-optimized operation. 

There was a spread in the performance data, specifically for generated heads and compressor efficiencies. Based on conventional calculations, the measured efficiency of Section 1 (inlet to side-stream) was 2 - 3% lower than the vendor’s proposed efficiency. The efficiency gap was constant over the flow range. For Section 2 (side-stream to discharge), the measured efficiency was much lower than the vendor’s proposed efficiency. There was around a 5% efficiency gap at the minimum flow and more than a 10% efficiency gap at the maximum flow.

The performance data spread showed compressor sensitivity to operating conditions, particularly flow-ratio and pressures.  

Based on the calculation method used, the side-stream losses were attributed to the downstream section (Section 2). This caused lower efficiency and performance compared to Section 1.

The variance reduced as the flow decreased toward surge and increased as the flow rose. Due to this effect, Section 2 efficiency gap (difference of vendor’s efficiency and measured efficiency) was relatively low at the minimum flow. This efficiency gap was constantly increased to more than 10% at the maximum flow.

Fine-tuning the flow ratio (side-stream volume to core-flow volume) and inlet pressures within the proposed limits (±4% and ± 2%, respectively) resulted in better overall efficiency. Section 2’s efficiency after adjustment was within 4% of the vendor’s proposed efficiency for the operating range near the best efficiency point.

A 6.8-MW steam turbine-driven process compressor with a side stream performed a critical process service. The side–stream was a recycle flow that recycled back to the compressor to the downstream reactor. 

Performance data was widely scattered. The first suggestion attributed low efficiencies to poor gas composition control on the compressor’s process flow; however, investigations showed this related to the compressor sensitivity to flow ratio and operational conditions. Section 1’s measured efficiency was 3 - 4% lower than the vendor’s proposed efficiency. The efficiency difference was relatively constant over the flow variation of Section 1. For Section 2, the calculated efficiency was significantly lower than the vendor’s proposed efficiency—around 20% at the minimum flow and more than 25% at the maximum flow.

The large performance data spread showed extreme compressor sensitivity to operating conditions, particularly the flow ratio. For some operating points, the sectional efficiency was calculated as low as 54 - 55%.

Based on the calculation method used, side-stream losses were attributed to the downstream section, causing it to appear very low in performance and efficiency.

The variance increased toward higher flows. However, the efficiency was far below the vendor’s proposed efficiency. For relatively low flows, Section 1’s efficiency gap was scattered around 20% and for high flows, Section 2’s efficiency gap was scattered more than 25%.

These investigations showed the recycle pressure and flow ratio considerably deviated from rated conditions. To solve the problem, recycle pressure and flow ratio were adjusted for the rated conditions. This resulted in operation within 7% of the vendor’s proposed efficiency.

Amin Almasi is a Chartered Professional Engineer in Australia and the U.K. (M.Sc. and B.Sc. in mechanical engineering). He is a senior consultant specializing in rotating equipment, condition monitoring and reliability.

Turbocompressors with side streams are operated with tight tolerances on operational parameters compared to conventional compressors.

Turbocompressors with Side Streams—A Challenging Class of Turbomachinery

Many ammonia plants in the United States and abroad are more than 50 years old. As operators seek to extend the life of these facilities, syngas trains have become a target of modernization projects due to their significant impact on overall plant efficiency, production capacity, and operating expenses. 

In recent years, revamp requests for ammonia syngas trains have focused on upgrades to legacy steam turbine drives. Currently, more than 35 Delaval 103-JAT steam turbines are in operation in Kellogg process-based ammonia plants globally. There are many available upgrades for syngas trains that utilize the 103-JAT frame. Modernization options are diverse and can meet a wide range of plant requirements for uptime, efficiency, costs, and footprint.

Steam path revamps are an effective way to improve a steam turbine’s power, reliability, and efficiency. 

With modern thermodynamic and mechanical design software, in certain cases, it is possible to pass more flow through a turbine and increase its rated output by simply changing the design of the first-stage blades and nozzle plate. Advanced 3D modeling software enables OEMs to implement new designs with greater predictability and fewer challenges, such as vibration. This capability is especially relevant today, with many customers opting for performance guarantees, particularly when it comes to uptime.

In one case study, upgrading multiple steam pathways increased the turbine’s rated power from 13,095 HP to 17,000 HP, in addition to a 6 - 7% efficiency improvement.

New steam paths utilize optimized airfoil designs, improved seals, and a fine-tuned aerodynamic match of stationary to rotating airfoils. Additionally, improvements to the mechanical components such as bearings, seals, and valve operating gear parts can lead to improved reliability.

One revamp to the Delaval 103-JAT is a conversion from a two- to three-wheel configuration. Original two-stage 103-JAT frames can generate around 11,000 HP. Converting to a three-stage design offers several advantages, including an approximate 6% increase in steam rate, higher reliability, and improved rotor dynamics due to thicker shafts and larger journal bearings. Original internals can be kept and used as spares.

Several other minor upgrades can be made during the same scheduled maintenance period. Some examples include but are not limited to:

 Nearly all Delaval turbines use an internal lifting bar to sequentially raise the steam control valves. Older units use flat-bottomed nuts assembled to individual valve stems to set the lifting adjustment and lift the valves. In cases where the valve seat and control valve are misaligned, a bending moment can be introduced into the valve stem during the initial lift—this could eventually lead to valve-stem failure. Installing nuts with a spherical bottom addresses this problem. The nut allows the valve to place itself in the valve seat and eliminates the inherent bending moment. This design has been successfully used for over 20 years.

 On older machines, the main lever arm on the control valve rack is fixed in end support columns by ball or roller bearings. These bearings are in a high-temperature area atop the steam chest and require routine lubrication with high-temperature grease. The location and temperature are a challenge to any form of maintenance and, as a result, the bearings eventually seize or wear prematurely. 

Self-lubricating sleeves can be installed in place of the original bearings to eliminate routine lubrication requirements. These have been proven for years on later-designed turbines. The conversion package consists of sleeves and custom-designed adapter bushings as direct drop-in replacements—no field machining is required. The conversion package includes the parts, plus an updated valve gear assembly drawing showing these parts for future records.

Retrofitting control systems to enable remote monitoring and diagnostic services (RDS).

 Many legacy steam turbines are still operated using mechanical-hydraulic or older electronic controls. Wear and tear on components and an uncertain supply of spare parts for older electronic systems can impair efficient operation. In many cases, RDS can be implemented to enable advanced capabilities, such as early warning of failures, root cause analyses, remote troubleshooting, etc.

One of the most challenging tasks for ammonia plants utilizing steam turbine drives for syngas compressors is to prevent a turbine overspeed triggered by designated turbine monitoring inputs.

Trip and throttle valves (TTVs) are designed to protect personnel and plant assets by rapidly stopping steam flow into the turbine. Many syngas trains today still utilize mechanical latch TTV technology, which was developed in the 1950s. While proven, past experiences overhauling TTVs have shown that many pressure boundary components, like cast valve bodies and yokes, are used well beyond their normal service life due to years of exposure to unknown pressure and temperature transients. Issues such as erosion, corrosion, and stress cracks have been discovered while performing non-destructive testing. These components can usually be repaired or replaced; however, lengthy and costly weld repairs or complete casting replacements can exceed new technology valve replacement costs and extend loss of production.

Siemens Energy’s Gimpel TTVs have been used on a wide range of OEM steam turbine frames since the 1950s. The top inlet Gimpel valves (TITTVs) on the Delaval 103-JAT steam turbine frame have always been challenging to replace. The mechanical latch design is outdated, and standard valve replacement options do not fit the available deck space.

Corrosion and contaminate build-up on TTVs are also problematic, as they can affect trip-time or full close of the valve. In most valve designs, spring compression is used to reset the TTV during partial stroke testing (PST) through direct mechanical movement via handwheel operation. The potential for spurious trips caused by the TTV during PST can lead some operators to avoid and/or delay recommended stroke testing (every 2 - 4 weeks), which presents a significant safety risk. 

Improvements to the Gimpel TTV technology are continually made to address operational pain points and support the increasing demand for plant safety. Currently, Gimpel offers two types of modern valve replacements for TITTVs on Delaval 103-JAT frames, designed with unique steam section geometry to facilitate space constraints on existing deck arrangements.

 Designed for pull-to-close operation, OOTTVs utilize hydraulic oil pressure supplied by the turbine control oil system for valve reset and PST. With hydraulic operation, OOTTVs utilize a stronger closing spring for faster trip time (0.03 second versus 0.05 second for TITTV) and more accurate throttling at start-up. Both oil-operated trip valves (OOTVs) and OOTTVs are available depending on the turbine’s in-operation throttling requirements.

 Designed for pull-to-close or push-to-close operation, EHTTVs have a self-contained hydraulic system and a microprocessor-controlled electro-hydraulic actuator. EHTTVs close in 0.3 seconds or less and provide the most accurate turbine throttling on start-up. No turbine oil is required. The valves can be installed in any orientation based on the available space and are most adaptable for integration into a distributed control system.

Replacement of a mechanical latch TTV with one of these two options can yield a high return on investment. In recent years, Siemens Energy has executed numerous TTV replacements on ammonia syngas trains and received positive feedback from customers. Real-world experience has shown that, in many cases, the costs of installing new technology versus repair costs of a decades-old safety device provide a lower total cost of ownership and improved operational and performance benefits.

In recent years, footprint solutions have emerged as an economical alternative to short-term repairs or partial modernizations. 

A footprint turbine is a replacement machine that can be installed with few—and in some cases, zero—modifications to the existing foundation or piping interfaces. The current oil system, the driven machine, and/or the gear can be reused or adapted according to new operating parameters. The approach is especially appropriate when modifications must be made for new operating data or when the service life of highly stressed components has been exhausted. 

One significant advantage of a footprint turbine is the increase in plant availability resulting from lower production downtime. The components in the new turbine are designed and manufactured according to unique customer requirements. As existing capital and spare parts can still be used, the investment expense is comparatively lower than purchasing a brand-new turbine system.

Compressor revamps are applicable when process conditions have changed sufficiently such that the original compressor aero design is no longer appropriate, i.e., the unit is operating at or near its boundary conditions. While there are cases where the installation of a new compressor package and foundation, piping, ancillary systems, etc., is unavoidable, operators can save on CAPEX and associated downtime by modifying or upgrading the existing compressor. Potential upgrades can include:

Complete re-aero of the compressor internals and conversion of the gearset, including stationary and rotating parts, without changes to the casing, external process connections, and/or existing footprint  

Upgrades to the lube oil or control system to enable remote monitoring  

Material upgrades to O-rings (upgrade to polymer or Teflon) or labyrinth seals (from aluminum to PEEK/PEK)

DETERMINING THE RIGHT MODERNIZATION APPROACH

For any ammonia plant modernization project, a comprehensive analysis of the steam turbine and compressor condition, along with customer objectives, is required to determine the optimal approach. A look at the current and future requirements will quickly reveal which systems, components, and parts can still be used and which have to be adapted and replaced. 

After a preliminary evaluation, a feasibility analysis can be conducted, which allows the operator to evaluate the costs of retrofits or upgrades for the existing turbine(s) and system. Modernization activities can then be scheduled to coincide with planned outages so that customer downtime can be minimized and operators realize a maximum return on investment.

Chris DeHart is the Head of Sales, Footprints modifications and upgrades - Service Americas at Siemens Energy. 

Charlie Danielson is the Critical Service Equipment Development and Technical Sales Professional at Siemens Energy.

Modernization options for ammonia syngas trains can meet a wide range of plant requirements for uptime, efficiency, costs, and footprint. 

Modernization for Ammonia Syngas Trains with Steam Turbine Drives

Monitoring rotating machinery is essential to ensuring operational reliability and performance. However, traditional real-time monitoring methods are often limited in detecting anomalies that can lead to failures because they either rely on predefined thresholds or rules that serve as “alarm” or “trip” values or require human operators to manually review data and identify potential problems. These time-consuming and error-prone methods will eventually lead to unforeseen downtime and costly maintenance interventions.

Vibration analysis is a powerful tool in rotating machinery condition monitoring. Equipment vibrations are captured and processed by techniques, such as Fast Fourier Transform (FFT), to identify how the vibration is distributed over a range of frequencies related to specific faults. This approach allows for early detection of several faults such as imbalance, misalignment, and bearings wear—allowing corrective actions to be implemented before catastrophic failures.

However, the vibrations displayed in the operators’ screens in processing plants are usually amplitude values. The vibration amplitude is useful for monitoring the overall trends or setting alarms and shutdown setpoints; however, it lacks diagnostic depth, as crucial information remains obscured within the frequency domain.

 Predictive maintenance (PdM) strategies have been consequently developed to incorporate periodic vibration routes. PdM technicians periodically collect vibration data in these routes using specialized in-depth analyzers.

Early detection of vibration anomalies significantly aids in avoiding catastrophic failures. Understanding the distinction between vibration amplitude and frequencies enables the detection of abnormal conditions through various methods:

 Specific values configured in the control system that, when reached, trigger alerts for the distributed control system (DCS) operator or immediately trip the machine through the emergency shutdown system.

 Experienced operators, familiar with their equipment, can discern when vibrations are increasing beyond normal levels. Trending the vibration values is useful to monitor the overall condition.

Periodic vibration routes (amplitude- and frequency-related):

 As part of a PdM strategy, vibration technicians periodically collect vibration data using specialized devices for analysis. These routes help monitor equipment health and detect anomalies before they escalate into critical failures.

While these methods complement each other, drawing from both human observations and system interventions, there's always room for improvement. It is essential to thoroughly evaluate these methods to uncover enhancement opportunities. Below are two failure cases that shed light on how conventional anomaly detection methods failed.

A single-stage steam turbine, driving a transfer pump, underwent an overhaul. Following the overhaul, baseline vibration readings on the inboard bearing were at 1.36 mils and 0.9 mils (X & Y probes) at full load, which are considered acceptable. Two months later, however, during warming up at 900 rpm, the turbine tripped on a high-high setpoint of 3.6 mils. Could this have been prevented?

 shows the increased linear vibration trend for this equipment. Considering the three detection methods mentioned earlier: The alarm and shutdown setpoints were not reached until the event, the operator did not observe this linear increase, and the failure happened between the vibration routes (the next vibration route was scheduled a month later). Such cases, albeit rare, tell us that a change in the monitoring strategy is needed—making it less human-dependent and more error-proof.

High-Cycle Impeller Failure of a Centrifugal Compressor

A second case study demonstrates centrifugal compressor failure.

 Two metal pieces were liberated from the impeller trailing edge at the hub side, in addition to consequential damages to diffuser vanes. The root cause was high-cycle fatigue related to running in a choke. What did the vibration tell us?

 shows that the vibration levels (amplitude) were suddenly multiplied by 2, and seven months later they were multiplied by 3. The alarm setpoints were not reached, no operator observed the increase, and the changes were detected during a vibration route eight months after the first increase. Although the detection approach was somewhat proactive, since it was detected during a vibration route, it can be argued that it was late. Eight months of relatively high vibration without any observation or action raises concerns about the condition-monitoring strategy. Would detecting the increase earlier have helped? Perhaps. However, the key point is that failures may develop and go unnoticed, and as a result, they will worsen and cause prolonged downtime.

These case studies demonstrate conventional anomaly detection methods have potential issues that would jeopardize the effectiveness of identifying developing failures:

 By the time a setpoint is reached, it might be too late—permanent damage might have occurred, e.g., in the first case study where the bearing was replaced and the bearing area had minor damages; in the second case study where the compressor impellers were damaged.

 When relying on operators to detect vibration anomalies, experience variation plays a big role. Inexperienced operators might not be familiar with the equipment’s history and trends.

 As the number of equipment increases, collecting vibration data on vibration routes becomes tedious.

USING MACHINE LEARNING TO DETECT DEVELOPING FAILURES

With the widespread adoption of artificial intelligence, maintenance organizations have increasingly turned their attention to applying it in their fields. Numerous models have been proposed to achieve zero-defect operation.

 Machine learning can provide valuable insights into equipment status because of the huge volume of operational data at industrial plants.

 It can address the limitations present in conventional condition-monitoring strategies, rendering them error-proof and more autonomous.

A data-driven approach to PdM is illustrated in 

 The methodology can be broadly divided into four stages: A regression model is used to create a "predicted" signal, which is then compared against real-time data. Alerts are triggered when the deviation between the real and predicted signals exceeds a certain setpoint.

Applying machine-learning algorithms to detect anomalies in Abqaiq Plants showed positive results. So far, real-time models have been developed for 100+ rotating machines. Below are some case studies.

To validate model performance, it was back-tested on a previous trip for a steam turbine on March 17. In 

, the green represents the raw vibration signal as read by the sensor, and the pink is the predicted signal. From left to right, the signals were matching around 0.9 mils until Dec. 5, when the real signal deviated and an alert was generated. Three months later, the trip occurred. This case demonstrates that the developed model is capable of detecting vibration anomalies even though they are far below the alarm setpoint.

Detecting a Developing Anomaly at Early Stages

A developing abnormal condition was detected at an early stage. In 

, the noisy signal is the real-vibration signal, whereas the stable one is the predicted signal. From left to right, the predicted and real signals almost matched at 0.5 mils until preventive maintenance was conducted during a shutdown period. Following that, the real signals started deviating, reaching 1 mil until an alert was generated on July 2. The area’s team was informed, and the issue was fixed during a planned shutdown window on Aug. 28. This shows the benefit of utilizing machine learning to detect slight increases in vibrations that are far below the alarm level where both operational needs and equipment health are met.

An old developing anomaly was discovered. When the model was applied to this equipment—a shipper centrifugal pump— high deviations were found between the real and predicted signal. In 

, the noisy signal is the raw data whereas the stable one is the prediction model. Although the raw data show an increasing average over the year, indicating an actual issue, the discrepancies between the predicted and real signals are worth investigating. Generally, such discrepancies could be caused by selecting model features or physical issues with the asset.

, the two signals matched two years ago until one probe experienced a step-change, followed by a step-change on the other probe. The vibrations continually increased linearly, albeit slowly, and remained below the alarm level. Operations were informed and a suitable shutdown window was selected for inspection. Had this not been detected, a sudden trip would have caused unplanned downtime.

These case studies shed light on the limitations of traditional anomaly detection methods in rotating machinery condition monitoring. These methods may fall short of promptly detecting failures despite the integration of predefined setpoints, operator experience, and periodic vibration routes. This shortcoming could lead to prolonged downtime and operational disruptions.

The introduction of machine learning algorithms offers a promising solution to enhance PdM strategies and mitigate the shortcomings of conventional monitoring approaches. By leveraging real-time operational data and advanced analytics, machine learning models can accurately identify subtle deviations in vibration patterns, even below alarm thresholds, enabling early detection of developing anomalies.

AUTHOR: Abdullah Sofiany is an Associate Engineer of Reliability & Rotating Equipment, at Abqaiq Plants, at Aramco.

[1] Marwala, T. (2012). Condition monitoring using computational intelligence methods: applications in mechanical and electrical systems. Springer Science & Business Media.‏

[2] Moyroud, F., Alas, P., & Libeyre, F. (2018). Impeller High cycle fatigue failure on a Natural Gas Pipeline Compressor Following Choked Flow Operation. Turbomachinery Laboratory, Texas A&M Engineering Experiment Station. Available electronically from 

[3] Wang, K. (2016). Intelligent predictive maintenance (IPdM) system—Industry 4.0 scenario. WIT Transactions on Engineering Sciences, 113, 259-268.

[4] Zhou, D. H., Chen, M. Y., & Xu, Z. G. (2013). Reliability Prediction and Optimal Maintenance Technology. Univ. Sci. Technol. China Press.

[5] Costello, J. J. A., West, G. M., & Mcarthur, S. D. J. (2017). Machine learning model for event-based prognostics in gas circulator condition monitoring. IEEE Transactions on Reliability, 66(4), 1048-1057.

[6] Zhang, W., Yang, D., & Wang, H. (2019). Data-Driven Methods for Predictive Maintenance of Industrial Equipment: A Survey. IEEE Systems Journal, 13(3), 2213-2227. doi:10.1109/JSYST.2019.2905565.

Machine learning techniques offer real-time anomaly detection that proactively identifies potential failures in turbomachinery.

Leveraging Machine Learning to Enhance Anomaly Detection in Rotating Machinery

LNG plays a long-term role in the energy transition. Recent data from Wood Mackenzie’s 

2024 Global Gas in the 2050 Net Zero World

 report shows that gas demand will be resilient for the next 20 years while LNG demand will grow until near 2050. According to Baker Hughes’ first quarter 2024 results, global gas and LNG demand growth will see a compound annual growth rate through 2040 of around 1% for gas and 3.5% for LNG.

Even considering Wood Mackenzie’s report, a case less favorable to fossil fuels, gas is expected to decline far less than oil and coal. In the long run, global LNG demand is expected to be driven by emerging Asian economies, coupled with a limited domestic gas supply which will call for increased LNG imports. In short, LNG is both a transition and a destination fuel.

While in emerging Asian countries gas demand will be propelled by the transition from coal to gas and challenges with scaling up renewables, in developed economies the increasing speed of electrification and the roll-out of intermittent renewable energy sources will require more low-carbon and dispatchable power generation sources, supporting grid stability and flexibility.

The global LNG market is responding with an unprecedented surge of additional export capacity. According to Baker Hughes, over the next three-year period (2024 - 2026) approximately 100 million metric tons per year (MTPA) of liquefaction capacity is expected to reach a final investment decision.

The Institute for Energy Economics and Financial Analysis forecasted in its 

 report that nameplate LNG liquefaction capacity from projects under construction or approved by financially capable backers could add an estimated 193 MTPA of new supply capacity from 2024 to 2028, taking the total capacity to more than 650 MTPA—a 40% increase in five years.

This robust supply growth will facilitate lower prices, increasing competition between export terminals. The most competitive projects will be those able to deliver the most cost-effective LNG while offering flexibility and a lower carbon footprint, which explains the continuous innovation and investment from the LNG supply chain to respond to and overcome these challenges.

Traditionally, the LNG industry has been a capital-intensive market. With significant upfront investments needed and long design-to-construction schedules, building an LNG export terminal has always been a careful and risky venture. In contrast to traditional large-scale liquefaction plants, mid-scale solutions based on multiple smaller modular trains represent an important innovation for the industry.

Modules are plug-and-play assemblies capable of increasing speed to market—the time needed to achieve the first drop of LNG. As such, projects can be developed in phases and gradually add new liquefaction capacity while managing financial exposure from previous phases.

At the same time, more flexibility versus traditional stick-built projects can be obtained by splitting the required total LNG production capacity across multiple independent modules, which optimizes production and minimizes maintenance costs. Plus, pre-constructed turnkey modules help minimize project risk, enhance the level of control over module fabrication, and reduce construction site costs and potential schedule delays.

Significantly fewer construction personnel and less site time enhance overall project safety—a key aspect of any LNG project. Pre-designed modules lower engineering costs, as project design efforts are dedicated to the elements of the LNG plant, which cannot be standardized upfront.

NMBL LNG (pronounced “nimble”) is Baker Hughes’ modular solution for mid-scale gas liquefaction. It consists of modules with the ability to process from 0.8 up to ~2 MTPA each and is based on the single-mixed refrigerant liquefaction process.

Many LNG reserves around the world, both accessible and remote, are too small to be economically viable with the conventional large-scale, stick-built approach—from engineering to construction and start-up. This is especially true in the current market context where LNG supply and demand is growing rapidly and the ability to exploit stranded gas reserves can be a differentiator for monetization.

To seize the opportunity, energy companies need an even greater level of agility and the flexibility to consider onshore and offshore LNG options. Traditional LNG projects often falter providing this flexibility due to long lead times, significant upfront investments, and an inability to right-size production. NMBL is designed for scalability and pre-constructed fit for purpose.

Each NMBL liquefaction train includes the process module, a turbo or electric motor-compressor for the mixed refrigerant, the surge vessel for refrigeration storage, the cold box of brazed aluminum heat exchanger type, and the e-room. Its mid-sized modular design mitigates the challenges faced with traditional approaches to engineering and construction and can reduce risks and increase control during the execution phase. Upon arrival, final installation is significantly faster and requires fewer on-site personnel depending on the project specifics. These factors, along with simplified interconnections and civil works, can dramatically reduce project costs.

NMBL modules are a “design one, build many” concept. Its design can be easily scaled up to serve the range from 0.8 to ~2 MTPA and has been optimized to reduce weight and footprint. With a production capacity of 1 MTPA range of LNG, each module weighs approximately 3,000 tons and has a total plant footprint of approximately 3,000 square meters.

While traditional modular projects have many interfaces among different companies working simultaneously and not always with the same goals, NMBL integrates design, procurement, fabrication, and testing phases to reduce project risk under a single point of responsibility.

Another NMBL feature is the ability to accommodate different mixed refrigerant compressor drivers, either gas turbine or electric motors, without changing module configuration. The liquefaction process produces zero emissions because it is driven by an electric motor compressor.

NMBL also delivers greater project integration with:

a wide range of compact aeroderivative gas turbines, including the LM2500, LM6000, and LM9000 families

a set of electric motors from BRUSH Power Generation, part of the MAX synchronous motors product line, capable of serving a 30 - 95 MW range

mixed-refrigerant centrifugal compressors

With an LM9000 used for the compressor mechanical drive, its high-power engine is coupled with a high level of efficiency (73.5 MW and 44% efficiency, ISO conditions). With the electric motors as prime movers, the eLNG project can instead accommodate a centralized power island working on gas turbines in single- or combined-cycle.

In the latter case, Baker Hughes can supply power blocks with multiple running LM9000s coupled with one SC/type steam turbine. Moving from simple- to combined-cycle alone enables an increase in power generation efficiency by 20% or more, significantly reducing a project’s carbon footprint.

While recommended for both onshore and offshore liquefaction and small- to medium-size operations, NMBL is also suited for remote locations where a reliable power grid may not be available and therefore the power-block concept is a more fitting solution. Because the LM9000 is an aeroderivative gas turbine, its intrinsic ability to accommodate frequent starts and stops is not only compatible with baseload operations but can also compensate for the fluctuations from intermittent sources when the power supply is coming from renewable energy sources.

NMBL modules are also digitally native— that is they can integrate the equipment offering with lifecycle service and software features. The system is fully compatible with commercially available distributed control systems. Modules can be transformed into “smart objects” via the Baker Hughes iCenter digital platform powered by Cordant asset performance management software.

The LNG industry is undergoing a significant transformation driven by the need for cleaner, more flexible, cost-effective solutions. Shaping industry standards and collaborating with partners across the value chain will accelerate the adoption of new, innovative solutions that can drive productivity and deliver emissions reductions at scale.

As the world seeks to balance its growing energy needs with the imperative to decarbonize, modular LNG is poised to play an increasingly important role, and Baker Hughes is well-positioned with NMBL.

Andrea Intieri is the LNG Platform Leader of Gas Technology Equipment, Industrial, and Energy Technology at Baker Hughes.

Baker Hughes’ NMBL LNG delivers flexibility in an increasingly dynamic LNG market. 

Product Spotlight: Baker Hughes' NMBL LNG

As the global energy landscape evolves, recent projections and industry updates highlight the dynamic shifts within the global natural gas sector. In April 2024, the U.S. Energy Information Administration (EIA) anticipated a modest increase in U.S. LNG exports for the year (2%) followed by significant growth in 2025 (an additional 18%). This growth is mirrored by expectations for a rise in U.S. natural gas exports by pipeline (3% in 2024, 4% in 2025) and a slight dip and recovery in pipeline imports in 2024 and 2025, respectively. Concurrently, S&P Global Ratings forecasted stable hydrocarbon prices supported by the Organization of the Petroleum Exporting Countries (OPEC), alongside a year-end rebound in gas prices driven by LNG export expansion and infrastructure developments.

“Low U.S. natural gas prices will likely continue for at least the next 10 years, fueling more LNG developments due to the low base price of natural gas,” said Marybeth McBain Product Line Manager - IGC New Apparatus, Global Sales at Ebara Elliott Energy.

This price point will not only bolster further LNG developments but also the construction of natural gas combined-cycle (NGCC) power plants. “The low and cleaner natural gas fuel will propel more NGCC power plants to be constructed, which provide reliable baseload power to supplement a growing renewables portfolio for the U.S. electric grid—eight new NGCC plants since 2022. This favors all gas-fired turbomachinery supporting the natural gas industry in the United States,” McBain said.

In this context, emerging trends such as modularization and the adoption of e-LNG technologies are reshaping the industry. “LNG projects are requiring modularization to eliminate risk in the field, i.e., stick built, in addition to a decline in available field labor,” said Joel Schubert, Director of Business Development LNG at Siemens Energy. “We also see a trend for future LNG projects going forward with e-LNG (electric motor drives) to reduce emissions and their carbon footprint.”

Moreover, the broader shift toward decarbonization and emissions reduction in LNG projects has started to affect LNG new developments. “Operators are considering greenhouse gas emissions, making changes in design accordingly, and moving toward electrification of some, if not all, of the major compression/pumps in the LNG plant,” McBain said.

Furthermore, Christopher Campos, Area Sales Manager, New Apparatus at Ebara Elliott Energy, said, “The offshore LNG market remains strong despite the current slowdown in the market due to pricing and policy shifts.” He said offshore floating LNG (FLNG) units continue to drive investors to look at capacity space with a limited number of shipyards. “In 2023, we saw Petronas invest in its third FLNG unit while 2024 saw a final investment decision (FID) by Cedar LNG for Canada. Golar is looking at re-purposing Golar Hilli Episeyo for its next area of operation in Argentina. Genting FLNG, an investment by Genting Oil & Gas in Indonesia, contracted out the construction of its first FLNG unit in 2024.”

Two trends are poised to send the global LNG markets into oversupply within two years: lackluster demand growth combined with a massive wave of new export capacity, according to a report from the Institute for Energy Economics and Financial Analysis (IEEFA), 

The global LNG market is slated to experience a shift starting in late 2024, as a wave of new liquefaction projects comes online. IEEFA expects global LNG production capacity (only projects under construction or approved by financially capable backers) to grow by roughly 193 MTPA from 2024 - 2028, rising from ~474 MTPA of nameplate capacity at the beginning of 2024 to 666.5 MTPA by the end of 2028—the fastest capacity growth in the history of the global LNG industry, representing a 40% increase in five years.

During 2024, only 12 MTPA of liquefaction capacity is slated to come online in the Republic of Congo, Mexico, Russia, Mauritania/Senegal, Australia, and the United States. Two of these new projects shipped their first cargoes earlier this year. But late 2024/early 2025 will see a new wave of LNG supply: IEEFA expects ~ 37 MTPA of new LNG facilities to start operations in 2025, with 57 MTPA of new capacity in 2026—the most ever built in a single year. IEEFA sees more LNG capacity additions of (

Note, this excludes any proposed global LNG projects that have not reached FID and the several projects that are moving forward but don’t have a completion date until 2029 or later.

Structural LNG demand growth faces challenges, according to IEEFA: economic, political, financial, and logistical. “The global LNG crisis of the last several years heightened those challenges, spurring some Asian nations to reduce the role of LNG in their development plans and accelerate the development of alternative energy sources.”

Combined imports fell in 2023 in Japan, South Korea, and Europe—a trio of countries that account for more than half of the world’s LNG demand—and IEEFA predicts they will likely continue falling through 2030.

Russia’s invasion of Ukraine spurred a decline in Russian gas supply to Europe that led to a spike in European LNG imports. However, prices have retreated from 2022 levels, largely due to falling demand from developed economies.

The United States, Australia, and Qatar produced three-fifths of the world’s LNG in 2023. Now, for the first time, the United States is no. 1 among global LNG exporters—the first U.S. LNG terminal started up on the Gulf Coast in early 2016. Most of the new LNG capacity slated through 2028 will be concentrated in the United States and Qatar, pushing Australia—the world’s top LNG exporter in 2021 and 2022—to third among global suppliers.

Mark Axford, President of Axford Turbine Consultants, cited in his report from CERAWeek 2024 a disrupting event that completely changed the worldwide energy picture and pushed the United States to the top: “The Nord Stream pipeline blowing up in 2022 was the single-most important occurrence in energy markets in decades. Since then, the United States has gone from third in LNG production to no. 1, with 24% share.”

Today, there are seven operating LNG facilities in the continental United States, with a combined nameplate capacity of 92.3 MTPA of LNG, about one-fifth of the world’s total (

IEEFA said the LNG industry’s capacity to export U.S. gas is set to grow from 94 MTPA to 172 MTPA by 2028. Five U.S. LNG projects, totaling more than 71 MTPA in liquefaction capacity, are currently under construction:

An expansion at Corpus Christi LNG (10 MTPA)

A 1.4-MTPA Mexican project sourced with U.S. gas recently shipped its first cargo and an additional 6 MTPA of U.S.-sourced LNG projects are moving forward in Mexico. IEEFA reports:

The development of the North Field complex will boost Qatar’s liquefaction capacity by 64 MTPA through 2030.

The initial phase of the 20-MTPA Arctic LNG 2 project received its first gas in late 2023.

The country’s first commercial-scale LNG plant is slated to start operations in 2025 or 2026.

Five projects have reached FID or are under construction in Africa.

Campos said that additional LNG carrier fleets have been added to the new-build bookings to support new liquefaction terminals coming online in Qatar, the United States, and Canada. Continued investment in bunkering vessels and LNG-as-a-fuel is still contributing to new-build bookings.

The industry continues to move forward with successful deliveries and progress for liquefaction and pipeline projects.

Siemens Energy successfully delivered, commissioned, and commercially operated its hybrid application for Chart Energy and Chemicals: a nitrogen refrigeration compressor STC-GV (20-3-H) package, powered by a STG-300 2S LE gas turbine. It integrates a synchronous motor/generator into a turbine-compressor configuration and includes online changeover capabilities from motor-to-generator modes and vice versa, supporting a transition to low-emission technology in compliance with local environmental regulations. The integration ensures that surplus power generated by Siemens Energy’s equipment contributes to the local power grid.

Siemens Energy has several recent and upcoming shipments and updates regarding Woodfibre LNG, a 2.1 MTPA export facility with 250,000 cubic meters of floating storage capacity being built near Squamish, British Columbia:

Boil-off gas (BOG) compressors 1 & 2 were successfully delivered on July 16.

Feed gas booster (FGB) compressor will be shipped in mid-August—the company is waiting for date/vessel confirmation from MDR.

BOG/FGB Powerhouses will be delivered at the end of September.

All main refrigeration (MF) and propane (C3) compressors have been successfully tested.

MR LP completed unit test preparation. String test is currently planned for the end of August.

Delivery of the LP MR train is planned for mid-October.

MR HP/C3 Complete Unit Test is planned for the end of November.

Delivery of the MR HP/C3 train is planned for January 2025.

Cheniere said it remains full-speed ahead on its current and future expansion projects. It continues to make progress on its Corpus Christi Liquefaction (CCL) Stage 3 Project and remains optimistic that it will achieve first LNG from the first train by the end of the year.

“We are confident we will secure all the regulatory approvals needed for the Midscale Trains 8 & 9 Project at Corpus Christi and the Sabine Pass Liquefaction (SPL) Expansion Project within our expected timelines,” said Bernardo Fallas, Director of Communications at Cheniere. “We recently received our environmental assessment for Midscale Trains 8 & 9, and we are in pre-filing with FERC for the SPL Expansion Project.”

In terms of pipelines, Fallas said the Sabine Crossing Pipeline Project is proposing a 58-mile pipeline extending east from Texas to the proposed SPL Expansion Project in Cameron, Louisiana, to support the natural gas needs of the project. The proposed pipeline would transport up to a maximum of approximately 2.7 billion cubic feet per day (Bcf/d) of natural gas to SPL from receipt points to be developed.

On July 1, the ADCC Pipeline entered commercial service and can provide approximately 1.7 Bcf/d of natural gas transportation capacity to the Cheniere CCL facility from markets on Whistler Pipeline’s Agua Dulce Header in South Texas.

In late July, Texas LNG Brownsville announced that they converted the two heads of agreement (HOA) with EQT Corp. to a definitive tolling agreement for 2.0 MTPA of LNG offtake. The FID is anticipated by late 2024/early 2025.

Also in late July, Woodside Energy purchased Tellurian Driftwood LNG—it is looking to reach FID in the first quarter of 2025; as part of the FID process, Woodside plans to examine equity sell-down options (up to 50%).

In January 2024, the U.S. Department of Energy (DOE) paused new authorizations for LNG exports to non-free trade agreement (NFTA) nations. The industry pushed back heavily on the pause, calling it “the worst energy decision made so far by the current administration,” according to Mike Sommers, President & CEO of the American Petroleum Institute.

Some projects were affected by the pause: Commonwealth LNG was stalled and is costing the project millions due to inflation while it sits still. In late June, Kimmeridge Texas Gas closed a follow-on control investment in Commonwealth LNG, raising its equity ownership to 90%. After selling his control to Kimmeridge, Founder and Executive Chairman of Commonwealth LNG, Paul Varello, announced his retirement. The pause also impacted Venture Global’s Calcasieu Pass 2 (CP2) LNG in Louisiana; however, on June 27, FERC issued a certificate that allowed CP2 to move forward.

But industry agencies said the pause would not alter the short-term trajectory of the U.S. LNG buildout. In April, IEEFA said: “Five new U.S. liquefaction projects, as well as two Mexican projects sourced with U.S. gas, are under construction and have already been cleared to export to NFTA nations. The projects have a collective capacity of 78 MTPA, meaning—pause or no—the U.S. gas industry is preparing for 85% growth in LNG export capacity by 2028.”

McBain said: “There remains a general acknowledgment in the industry that LNG expansions are continuing to move forward with FERC on a case-by-case basis. Also, the sentiment is that more LNG new licenses will be approved once the election is finalized— regardless of which political party wins.”

Relief came on July 1, when U.S. District Judge James Cain in Lake Charles, LA, ordered the Administration to lift the pause until a lawsuit brought on by 16 Republican-led states—who argued that freezing approvals of LNG exports was “completely without reason or logic”—could be resolved.

“We are encouraged by the recent ruling and expect it will facilitate DOE’s continued processing of LNG export applications,” Fallas said. “We previously stated that we did not believe the DOE pause materially impacted the timelines of our overall projects, and we continue to believe that to be true.”

https://www.eia.gov/todayinenergy/detail.php?id=61863#:~:text=We%20forecast%20that%20U.S.%20LNG,and%20by%204%25%20in%202025

https://www.spglobal.com/ratings/en/research/articles/230627-clear-lng-outlook-could-turn-murky-near-end-of-decade-12770037

The global LNG market has been impacted by new liquefaction capacity, price fluctuations, and political intervention yet it is slated to grow at an unprecedented rate through 2028.

September/October 2024

Myth: Part 1: A Centrifugal Compressor Map is Just Test Results

Turbocompressors with Side Streams—A Challenging Class of Turbomachinery

Modernization for Ammonia Syngas Trains with Steam Turbine Drives

Leveraging Machine Learning to Enhance Anomaly Detection in Rotating Machinery

Product Spotlight: Baker Hughes' NMBL LNG

The Global LNG Market: The Rise of LNG Production

Turbo Expo Showcases Gas Turbines as Part of Decarbonized Future

Turbomachinery International: September/October 2024