`
`
`
`
`
`
`Optimized Skid Design for
`Compressor Packages
`
`By
`
`Hongfa Wu, Ph.D., P.Eng.
`Chris Harper, P.Eng.
`Senior Engineer
`Principal Engineer
`hwu@betamachinery.com
`charper@betamachinery.com
`Beta Machinery Analysis
`
`Presented at:
`Gas Machinery Conference 2013
`October 7 - 10, 2013
`Albuquerque, NM
`
`
`Abstract
`
`The majority of compressor packages are now mounted on steel skids or baseplates. Designing a
`skid for a new machinery package can be challenging because of these factors:
` The skids must be designed to avoid resonance and vibration (from dynamic machinery
`forces and couples).
` The industry is looking for lower cost packages. This can drive suppliers to reduce the
`skid cost and associated stiffness, but an inappropriately designed skid will create
`vibration and reliability problems. In some cases, skids are considered too flimsy for the
`required application.
` New designs must consider loading, lifting and transportation issues, as well as weight
`limitations. Pedestal height can also cause problems.
`
`
`This paper will outline the issues and approaches involved in skid design for vibrating loads such
`as reciprocating compressors and pumps.
`
`This paper discusses industry best practices in skid design, including optimized design
`techniques. Two case studies will be used to illustrate different skid designs and the impact on
`cost, performance and reliability. This paper will benefit owners, packagers, and engineering
`companies involved with rotating equipment.
`
`
`
`
`
`www.BetaMachinery.com
`
`
`
`Page 1
`
`Page 1 of 9
`
`

`

`Introduction
`
`Static Loads
`
`1.
`
`When designing a structural steel skid (baseplate) for a compressor or pump package, the design
`must balance stiffness, mass, and cost. High stiffness will help avoid alignment problems due to
`skid deflection during transportation and installation. Heavier skids tend to have lower overall
`vibrations, but can have high deflections when lifted. The challenge when optimizing the design
`is to know where steel can be added or removed to maximize the stiffness and minimize the
`costs.
`
`2. Skid Loads
`
`There is considerable confusion
`about dynamics, quasi-static and
`static analysis. Figure 1 identifies
`the applicable load frequency
`ranges and the design criteria for
`these three analyses.
`
`2.1.
`
`Static skid design focuses on
`evaluating stress and buckling of
`members under constant loads.
`(Constant loads can also be
`described as loads applied at a
`frequency of 0 Hz.) They can also
`focus on deflection of skid
`members, which can affect
`alignment of equipment.
`
`Typically static loads are:
` Dead loads, including weight of permanent equipment.
` Thermal loads which includes forces created by temperature changes and pressure.
` Drive torque of compressors and engines.
` Lifting or dragging loads, when moving the skid with cranes or winches. These loads can
`include a load factor which considers the impact from sudden stops or motion of the
`lifting equipment (e.g., offshore lifts). A load factor of 1.15 to 2.0 is common.
` List angle, which creates horizontal loads when a ship leans to one side.
`
`Vibration
`
`Fatigue
`
`Yielding
`
`Buckling
`
`Deflection
`
`Design Criteria
`
`0
`
`1
`
`10
`
`100
`
`Frequency Spectrum of Loading
`Cycles per Second (Hz)
`
`Static Analysis - Lifting and thermal loads
`
`Quasi-Static Analysis - Loading caused
`by wave, seismic, etc.
`
`Dynamic Analysis - Equipment and other
`forces causing resonance
`Figure 1. Loading type vs. loading frequency and design criteria
`
`
`Guidelines for static skid design include American Institute of Steel Construction (AISC) and
`owner specifications for deflection (e.g., 0.5 inch deflection per 15 feet of skid length when
`lifting).
`
`2.2. Quasi-Static Loads
`
`Quasi-static loads are loads which are periodic, but at a low enough frequency (relative to the
`natural frequencies of the equipment package) so the inertia effects of the structure do not come
`into play. They tend to have a frequency of less than 3 cycle per second or 3 Hertz (Hz).
`
`www.BetaMachinery.com
`
`
`
`Page 2
`
`Page 2 of 9
`
`

`

`
`Typical quasi-static loads are:
` Environmental loads like live loads, wind, current, wave, earthquake, ice, earth
`movement, and hydrostatic pressure. These can occur in any direction.
` Construction loads including loadout, transportation, and installation.
`
`
`Quasi-static skid design focuses on evaluating stress and buckling of members, similar to static
`stress design. However, fatigue analysis may be done on loads like those caused by waves.
`
`Guidelines for quasi-static skid design include API RP 2A-WSD and International Building
`Code (IBC). Owners and equipment manufacturers may also have standards and specifications
`for both static and quasi-static loads and deflections.
`
`
`Lower pressure
`
` Pulsation shaking forces
`Higher pressure
`
` Unbalanced forces and moments
` Crosshead guide forces
` Cylinder horizontal gas forces
`Figure 2. Common Reciprocating Compressor Dynamic Forces
`
`2.3. Dynamic Loads
`
`Dynamic loads can be caused by waves, wind, earthquake or machinery, but it is typically loads
`by the machinery itself that concern the skid designer, as it is most likely to cause resonance.
`Resonance is the condition when the frequency of the dynamic force is within +/-10% of the
`mechanical natural frequency (MNF) of skid, vessels, piping, and structure/foundation. (At the
`design stage, +/-20% is typically used to account for modeling and fabrication uncertainties.)
`
`Figure 2 shows common reciprocating compressor dynamic forces, which include:
` Unbalanced forces created by rotating and reciprocating weights like crankshafts and
`piston assemblies. If the forces are offset, they can create unbalanced moments on the
`equipment. These can be obtained from the compressor or engine/motor manufacturers.
`
`
`
`www.BetaMachinery.com
`
`
`
`Page 3
`
`Page 3 of 9
`
`

`

` Horizontal cylinder gas forces, created by the differential pressure between the head end
`of a cylinder and the crank end.
` Vertical forces on the crosshead guides, created when rotating motion is converted into
`reciprocating motion.
` Pulsation-induced shaking forces, created at elbows and changes in pipe diameter due to
`pressure pulsations.
` Misalignment of the compressor and driver.
` Rolling torque on engines, which can occur at higher orders of engine runspeed (e.g., 4x,
`8x, … or 3x, 6x, …).
` Torsional vibrations, which may cause horizontal vibrations of the compressor frame.
`
`
`These forces are typically harmonic and occur at discrete multiples of equipment runspeed. In
`variable speed machines (e.g., motors with VFDs), the frequency of the force will change with
`the speed of the equipment.
`
`The majority of these forces occur at the first and second order of compressor runspeed, so
`raising the mechanical natural frequencies (MNFs) of major components on the equipment
`package above 2x runspeed is an effective strategy for avoiding resonance. Intertuning (between
`1x and 2x) or detuning (below 1x) may also be possible in selected cases.
`
`In rotating equipment like
`motors, centrifugal
`compressors, and screw
`pumps, the dynamic
`forces are usually just
`unbalance forces and
`flow-induced pulsations,
`which tend to be low.
`Standards like ISO 1940/1
`give recommended
`residual unbalance for
`various rotating
`equipment.
`
`Dynamic skid design
`Figure 3. EFRC Vibration Guidelines Converted to pseudo Peak Overall Velocity
`focuses on evaluating
`vibrations and fatigue. Limiting vibrations of skid members is important to limit the vibration of
`the equipment, vessels and piping that are attached to it. If a skid member has high vibrations
`then the components attached to it will likely have high vibrations also. It is important when
`taking vibration measurements that all components along the load path (described in Section 3.1)
`are measured and compared to guideline.
`
`When discussing guidelines, it is important to distinguish been spectral and overall vibration
`measurements. Overall vibration measurements are the actual deflections (or velocity) of the
`component versus time. Spectral vibration measurements typically require specialized measuring
`
`www.BetaMachinery.com
`
`
`
`Page 4
`
`Page 4 of 9
`
`

`

`equipment, which break the vibration down into peaks at discrete frequencies. It is used for
`troubleshooting to identify and evaluate problem areas.
`
`Vibration guidelines for skid members can vary depending on the situation. Many equipment
`vendors specify a maximum allowable vibration at the equipment mounting locations. A spectral
`guideline of 0.1 in/s peak (25.4 mm/s peak) is common. Figure 3 shows the European Forum
`Reciprocating Compressors (EFRC) overall (OA) screening guideline for different areas of a
`reciprocating compressor. (The guideline is in pseudo Peak because it has been converted from
`an RMS guideline.) Note that the components that are closer to the foundation typically have a
`lower allowable vibration guideline.
`
`3. Best Practices
`
`3.1. Load Path
`
`The structure and foundation underneath a reciprocating compressor are used to absorb the
`energy created by the dynamic forces. When the foundation is heavier, the vibrations are lowered
`due to the formula: acceleration = force / mass. When the support structure is stiffer, the
`vibrations are lowered due to formula: deflection = force / stiffness. Therefore, it is desirable to
`have a massive foundation and stiff structure. (Damping can also lower vibrations, but it is
`usually not practical to add damping to a structure without reducing the stiffness.)
`
`In order to transmit the dynamic forces through the structure to the foundation, the path between
`the dynamic forces and foundation must be as stiff as possible. Since the stiffness of a collection
`of components in series is dominated by the stiffness of the weakest link, care must be taken to
`avoid any excess flexibility in the structure. In other words, the path the dynamic load must take
`to get to a region of high mass or high stiffness must be as direct and stiff as possible.
`
`Steel plate is significantly less stiff in bending than in tension or compression, therefore bending
`of steel plate should be avoided. Full depth gussets should be used in wide flange steel under
`flanges that take dynamic loads.
`
`Table 1. Important Design Considerations Depending on Foundation Type
`
`Type of Foundation
`Concrete foundation or
`block-mounted
`Gravel pad
`
`Pile-mounted
`
`Offshore structure
`
`
`
`Impact on Skid Design
` Grouting ensures stiff connection between skid and foundation.
`Full gravel bed desired.
`
`
`Effective if contact between skid and gravel is ensured through packing and pad design.
`
`Pile locations must be appropriately placed for dynamic loads, in addition to static and quasi-
`static loads.
`Important dynamic load locations include:
`o Under compressor inboard cylinder supports,
`o Under compressor cylinder head-end,
`o Under scrubbers.
` Connection between pile and skid is critical.
` Weight limitations so concrete may not be permitted.
` Connection between skid and structure may not be rigid. Plug weld if required.
`Platform or FPSO deck may not be designed for reciprocating compressors.
`
`
`Important that main full depth deck beam pass perpendicularly under compressor frame.
`Compressors should be located close to vertical columns.
`
`www.BetaMachinery.com
`
`
`
`Page 5
`
`Page 5 of 9
`
`

`

`3.2. Risk
`
` A
`
` finite element analysis (FEA) of the dynamic forces and
`associated vibrations should be done in certain cases including:
` New compressor frame or driver/frame combination.
` Higher horsepower compressor on an existing skid
`design.
` Less massive foundations or less stiff structure, like
`pile-mounted units or offshore units mounted on
`platforms or ship decks.
`
`Loads and welds on beam
`flanges not recommended
`
`Plug welds or T-sections are
`recommended. Loads should
`be in the plane of the beam
`web or gussets should be
`installed
`Figure 4. Recommended design for
`connecting stacked beams
`
`
`Table 1 above outlines the impact the foundation type has on
`the skid design.
`
`3.3. Concrete
`
`Concrete can be used inside the skid itself to add mass and
`stiffness. It is most effective when it is rigidly connected to skid
`beams using rebar or nelson studs, especially the top flange
`where vibration equipment is attached.
`
`Key locations for concrete are in the compressor and
`engine/motor pedestal, in the main skid underneath the
`pedestals and scrubbers.
`
`3.4. Beams
`
`It is usually more advantageous to use taller beams than use
`beams with thicker flanges and/or webs. Taller beams are much
`stiffer in bending than shorter heavier beams.
`
`When stacked beams must be used, the connection between the two beams is important. It is a
`stiffer design to weld a T-section to the top of a wide flange beam (taking care to line up the
`webs) or plug weld beams than to weld two wide flange beams on top of each other at the edge
`of the flange (Figure 4). The loads on beams should be in the plane of the beam web, or gussets
`should be used.
`
`4. Modeling Considerations
`
`When modeling reciprocating compressor skids using finite element analysis (FEA), different
`levels of detail are required for different types of loads. For static loads like lifting loads, one-
`dimensional (1D) beam elements are all that is required. Point masses can be used for equipment.
`For quasi-static analysis, 1D beam elements can also be used, but more detail needs to be added
`to the model because quasi-static loads can be horizontal, and thus act on components like piping
`and vessels. Dynamic analyses often require shell (2D) or solid (3D) elements for certain parts of
`the model to accurately account for local flexibility (e.g., the flexing of a beam flange or the
`impact of gussets). Local flexibility can play a large role in determining mechanical natural
`
`frequencies and vibration amplitudes of equipment packages.
`
`www.BetaMachinery.com
`
`
`
`Page 6
`
`Page 6 of 9
`
`

`

`Figure 5. Existing engine pedestals (left) and new engine rails (right)
`
`5. Case Studies
`
`5.1. Replacement of
`Existing Driver
`Pedestal
`
` A
`
` customer was doing a field
`retrofit to replace an engine
`driving a reciprocating
`compressor. The turnaround
`time was fairly quick, but the
`customer wanted to avoid
`vibration problems. There were
`no skid drawings, so the deck plate had to be removed and the beams field-measured. The goal
`was to remove the four existing engine mounting pedestals and replace them with two new
`engine rails (Figure 5) that could handle the higher horsepower Caterpillar G3608.
`
`
`A finite element model was created, using ANSYS
`software, to try several different designs (Figure 6).
`The design that was eventually installed was a
`W18x106# beam for the two main engine rails. The
`rail was sunk about 3.5” below the top of the main
`skid, and the top flange of the main skid was coped
`so the engine rail could dropped down and be welded
`directly to the skid beam web (Figure 7). Also, this
`allowed the concrete poured in the main skid to
`cover the bottom of the engine pedestal rail, adding
`stiffness and transmitting the engine dynamic forces
`into the concrete.
`
`No cross beams were possible
`in the engine pedestal due to the oil pan on the G3608, so gussets had to be
`installed on the outside of the engine rails. Gussets were added until the
`mechanical natural frequency (MNF) of the engine was above 2.4 times
`maximum engine runspeed. This was done to avoid resonance due to the
`coincidence of the engine MNFs and the engine primary (1x) and
`secondary (2x) forces and moments.
`
`This optimized approach helped focus the design on the areas that required
`stiffness, and avoided excess costs and time in order to meet the tight
`revamp schedule. The engine was installed and has been running without
`issue since 2011.
`
`
`
`
`Engine
`
`Figure 6. Early FEA model (with holes in gussets)
`
`Engine
`rail
`
`Main Skid
`
`
`
`www.BetaMachinery.com
`
`
`
`Page 7
`
`Figure 7. Engine rail design
`
`Page 7 of 9
`
`

`

`5.2. Dynamic Skid Analysis
`& Consequences
`
`
`
`Engine
`
`
`This example illustrates a
`dynamic skid analysis, and the
`consequences of not following
`the design optimization
`recommendation.
`
`The skid design was created
`using ANSYS finite element
`Figure 8. Compressor dynamic skid model (left) and engine mounts (right)
`(FE) software (Figure 8 left).
`The skid dynamics were evaluated and
`both vibration and stress amplitudes were
`compared to guidelines. To reduce
`vibrations to guideline levels, the skid
`design was optimized by adding
`additional anchor bolts in key locations
`and adding gussets near the engine
`mounts (Figure 8 right). The engine
`: 0.31 ips pk
`mounts are circled in Figure 8 and the
`recommended gussets are in blue.
`: 0.12 ips pk
`
`: 0.025 ips pk
`The unit was commissioned and high
`horizontal vibrations were detected.
`: 0.025 ips pk
`Vibration on the non-drive end (NDE) of
`Figure 9. Skid operating deflected shape (ODS) of non-drive end
`engine, at crankshaft height, was 0.47
`of engine
`in/sec peak at 16.7 Hz, which exceeded
`the Caterpillar engine guideline of 0.26 in/sec peak. Since the engine was running at 1000 RPM,
`this vibration occurs at 1x engine runspeed. Horizontal vibration was measured at seven test
`locations on the engine, engine mounts and main skid (Figure 9 and Figure 10). The test
`locations in red have vibration above BETA’s skid vibration
`guideline of 0.1 in/sec peak. It was noted that there was a
`significant increase in vibration between test points  and .
`The mechanical natural frequency (MNF) of the engine was
`also checked, and found to be 18.5 Hz. As described in the
`previous case study, the engine/pedestal assembly MNF should
`be above 2.4x engine runspeed (40 Hz) however leaving this
`mode intertuned may be acceptable if the engine does not run
`below 620 RPM.
`
`There appeared to be high flexibility along the load path from
`the engine to the main skid beam. Upon review, it was
`discovered that the recommended 0.75” thick full depth gussets
`near the engine mounts were not installed during fabrication.
`This flexibility dominated the other well designed and stiff
`components and thus lowered the overall stiffness of the entire
`engine mounting system.
`
` ℄: 0.47 ips pk
`: 0.44 ips pk
`: 0.42 ips pk
`: 0.40 ips pk
`
`
`Engine
`Mounts
`
`
`
`
`
`
`
`Horizontal
`Direction 
`
`
`
`Main
`Skid
`
`Figure 10. Plot of skid ODS
`
`www.BetaMachinery.com
`
`
`
`Page 8
`
`Page 8 of 9
`
`

`

`6. Conclusion
`
`Skid designs must consider static and quasi-static forces, as well as dynamic forces. In most
`cases, a detailed finite element analysis is required to evaluate skid designs.
`
`The main dynamic forces on a reciprocating compressor are the unbalanced forces created by the
`compressor frame, as well as cylinder horizontal gas forces and vertical crosshead guide forces.
`Avoiding having these forces resonant with the main natural frequencies of the equipment,
`vessels, piping, skid, and structure/foundation will lower the chance of resonance and keep skid
`vibrations below the recommended spectral guideline of 0.1 in/s peak.
`
`The dynamic forces can only be controlled if the load path from their sources to the foundation is
`as direct and stiff as possible.
`
`The two examples presented in this paper describe the challenges related to the proper design of
`a skid support structure. They illustrate the need to properly account for the flexibility of the
`different parts of the equipment package skid, in order to achieve an acceptable design.
`
`7. References
`
`
`
`
`
` API Recommended Practice, “Planning, Designing and Constructing Fixed Offshore
`Platforms - Working Stress Design,” API Standard 2A-WSD, 21st Edition, Oct 2007.
`International Building Code (IBC) 2003.
`ISO Balance Standard, “Mechanical vibration -- Balance quality requirements for rotors
`in a constant (rigid) state -- Part 1: Specification and verification of balance tolerances”,
`ISO Standard 1940/1:2003.
` F Newman, T Stephens, and R Harris, “Lateral-Torsional Vibration Coupling in
`Reciprocating Compressors,” presented at Gas Machinery Conference 2012, Oct 2012.
` AJ Smalley, PJ Pantermuehl, “Systems Mounting Guidelines for Separable Reciprocating
`Compressors in Pipeline Service,” SwRI Project No. 18.12083.01.401, Prepared for Gas
`Machinery Research Council, Dec 2006.
`
`www.BetaMachinery.com
`
`
`
`Page 9
`
`Page 9 of 9
`
`