Kenneth E. Atkins, Engineering Dynamics, Inc., Houston, Texas;

Martin Hinchliff, Dresser-Rand, Painted Post, New York;

and Bruce McCain, Oxy Permian, Sundown, Texas

Reciprocating compressors are usually rated in terms of horsepower, speed and rod load.? While horsepower and speed are easily understood, the term “rod load” is interpreted differently by various users, analysts, OEMs, etc.

“Rod load” is one of the most widely used, but least understood reciprocating compressor descriptors in industry. Typical end users know that rod load is a factor used to “rate” a compressor, but they don’t generally have a good understanding of how this rating is developed and how to utilize it for machinery protection.

The goal here is to discuss the various definitions of rod load, including current API-618 definitions, manufacturer’s ratings, and a typical user’s interpretation.

Basic theory

Consider the typical double-acting compressor cylinder geometry illustrated in Figure 1. The loads (forces) that are generally of concern include the piston rod loads, the connecting rod loads the crosshead pin loads, the crankpin loads, and the frame loads. As the crankshaft undergoes one revolution, all of these loads vary from minimum to maximum values. The loads are generated by both gas and inertia forces, as discussed below.

Gas loads

As the compressor piston moves to compress gas, the differential pressures acting on the piston and stationary components result in gas forces, as illustrated in Figure 2. The pressures acting on the piston faces (head end and crank end) result in forces on the piston rod. The force acting on the piston rod due to the cylinder pressures alone alternates from tension to compression during the course of each crankshaft revolution. It is straightforward to compute the net force on the piston rod due to pressure. The forces due to pressure also act (equal and opposite) on the stationary components.

The maximum compression force due to pressure occurs when the head end is at discharge pressure and the maximum tensile force due to pressure occurs when the crank end is at discharge pressure. Therefore, the equation shown in Figure 2 is often evaluated at the extremes as follows:

(INSERT EQUATION) (1)

(INSERT EQUATION) (2)

Now consider a more realistic pressure versus time diagram, as shown in Figure 3. “Line pressure” refers to the pressure at the line side of the pulsation bottle (suction or discharge). “Flange pressure” refers to the pressure at the cylinder flange. As shown, the in-cylinder discharge pressure exceeds the nominal discharge line pressure, and the in-cylinder suction pressure is less than the nominal suction line pressure due to several effects:

  • Pressure drop due to valve and cylinder passage losses (typically 2-10%)
  • Pressure drop due to pulsation control devices (typically < 1%)
  • Pulsation at cylinder valves (typically <7%)
  • Valve dynamics (inertia, sticktion, flutter, etc.).

API-618 specifies that the internal pressures must be computed, but does not define any procedure for the calculations. There are several methods for accounting for the non-ideal effects. One common method is to model the valve as an orifice and then the pressure drop though the valve (valve loss) is proportional to the square of the piston velocity (flow). Theoretically, it would be more accurate to use the results of the valve dynamics analysis coupled with the digital pulsation simulation to model the instantaneous pressure at the valves. This is not practical to do until all of the piping and valve details are known. In any case, the difference should be small, provided the losses are within the typical values listed above.

Because of these effects, the forces due to differential pressures are higher on both the running gear and the stationary components than those calculated based on nominal line pressures. However, equations 1 and 2 are still applicable as long as the appropriate pressures (discharge pressure higher than nominal discharge pressure, suction pressure lower than nominal line pressure) are used. If the nominal pressures at the suction and discharge cylinder flanges are used for PSuction and PDischarge, then these tension and compression forces represent the term “Flange Loads” as interpreted by some users. Equations 1 and 2 are easy to evaluate and for many years were the basis for rating “rod loads” of reciprocating compressors.

Of course, for the general non-ideal compressor cylinder, the maximum discharge pressure on the head-end will not necessarily occur at the same instant that the minimum suction pressure occurs on the crank-end and vice versa. Therefore, it is common to evaluate the gas forces versus crank angle at discrete steps (e.g. every 5 or 10 degrees). Computing the instantaneous force due to differential gas pressures is easily accomplished with computer-based software. If the actual in-cylinder pressures are used and the extremes are evaluated, these forces are then the “Gas Loads” referred to in the API specifications.

Piston rod loads

The basic slider crank mechanism is illustrated in Figure 4. The exact equation for the position of the crosshead with respect to the x-direction shown is

(INSERT EQUATION) (3)

The piston (crosshead) motion is usually approximated using the first two harmonics of the Taylor series as follows:

(INSERT EQUATION) (4)

The piston rod loads can be evaluated by considering the free body diagram in Figure 5. The forces acting on the piston rod are the gas forces due to differential pressures acting on head end and crank end piston areas plus the inertia forces due to the reciprocating mass. If the reference point is chosen as the crosshead end of the piston rod, then the reciprocating weight will include the piston rod and the piston assembly (piston, rings, rider bands, etc.). The reciprocating inertial force (F=ma) can be computed using the following equation:

(INSERT EQUATION) (5)

where:

mrecip = mass of reciprocating components

r = crank radius

? = angular velocity, (INSERT SYMBOL)

l = connecting rod length.

The combined piston rod load is the sum of the gas force and the inertial force. In accordance with API-618, this value is routinely calculated in the design stage, and used along with the rod area at the minimum cross-section to compute tensile and compressive stresses in the piston rods. The stress in the piston rod is one factor to consider in the design, and in some cases it may be the limiting factor or the “weakest link in the chain.” However, this load is not the rod load to which API-618 refers.

Crosshead pin loads

The free body diagram for the system, including the crosshead pin, is shown in Figure 6. Here, the mass of the crosshead assembly (crosshead, balance weights, crosshead shoes, etc.) must be considered, but the same equations apply. The combination of the gas loads and inertia loads evaluated at the crosshead pin in the direction of piston motion are the “combined rod loads” to which API-618 refers. This load does not consider side forces on the crosshead, or the 1/3 of the connecting rod weight that is usually considered to be reciprocating. Thus, “rod load” by API definition is not really a rod load, but actually a pin load.

Crankpin loads

If the loads and torques throughout the system are evaluated, then the rotating and reciprocating inertias, as well as the side forces, are included. Equations are applied for computing x and y components of crankpin and wrist pin loads, crank throw torques, main bearing loads, etc. All of these loads are typically considered in the design stage. Different OEMs evaluate the loads per their own experience. API guidelines are presented in Table 1.

User perspective

Various OEMs have used other terms such as “maximum allowable frame load,” “maximum allowable gas load,” etc., but the terms Max Allowable Continuous Combined Rod Load (MACCRL) and Max Allowable Continuous Gas Load (MACGL) are defined in API-618, and only since the 4th Edition (1995). It is not clear that all OEMs, users, analysts, operators, analyzer vendors, etc. recognize these terms, and agree that reciprocating machinery should be rated in this manner.

Oxy Permian owns, operates, and maintains in excess of 400,000 horsepower of compression in West Texas and Eastern New Mexico. This includes screw, centrifugal, and reciprocating machines. Most of these machines are reciprocating compressors and range in age from a few months to several decades, with the majority being installed prior to 1986 (i.e., pre-3rd Edition of API-618). The OEM published “rod load” limits range from 18,000 to 225,000 lbf. These machines vary by service, manufacturer, speed, loading, installation, and operating philosophies, and yield an array of equipment configurations. Longevity of service requires many machines to be subjected to a variety of process conditions that result in full utilization of “rod load” capabilities applied to these machines at commissioning. Several factors affect the actual rod load on a machine, including declining field pressures, process changes, improper operation of unloaders, valve degradation, ring failure, process upsets, and machinery modifications.

Oxy Permian performs compressor analyses on the majority of reciprocating compressors at approximately six-week intervals. This snapshot of compressor operation presents the facility with the machinery health at the time of data collection. The rod load is presented, generally in a “percentage of allowable rod load” format, and the facility makes maintenance and operation decisions based on various components of these health reports. The question is: What exactly are we looking at, and how do we compare our measured “rod load” to the OEM recommended maximum?

For practical purposes, a facility is able to measure the gas load on a compressor by using either peak pressures generated inside the cylinder (periodic measurements by experience analyst with portable equipment), or flange pressures (typical pressure gauges, transmitters, etc.), along with the cross sectional area of the piston. If the flange pressures are used, then the resulting loads must be compared to allowable “flange loads.” But that is not an API definition, and all OEMs do not provide such allowable loads. If the appropriate reciprocating weights are known (piston, piston rod, nuts, rings, riders, crosshead, bushings, pins, etc.), then the inertial loads can also be calculated. These inertial loads can be added to the gas load to develop the combined rod load. However, the reciprocating weights are not always known as modifications, and the degree of record-keeping may be in question. A thorough understanding of what the rod load rating implies is generally not apparent to the typical machinery analyst/engineer/operator. The analyzer software will typically display both gas loads and combined rod loads, but only one allowable value is available. Many times the combined rod load is lower than the gas rod load, but this is not always the case as will be shown later. The standard rod load reports do not compare the measured values to both MACCRL and MACGL.

Facilities must continually do more with less, and are generally capital constrained such that end users must obtain every pound of load capability, in addition to all available horsepower, from a given machine. As operations personnel pull liners from cylinders, install new cylinders, and push the machines to the max, they have found that in some cases, they do not actually understand the rod load limits of the compressor. Economic viability of a new project, whether revamping an existing machine, or adding either parallel or series compression, is usually based on three items: power availability, additional throughput, and machinery limitations. It is important that a reputable, experienced OEM be consulted when evaluating design options because the end user is not normally aware of all design limitations on a machine.

For example, the original installation may have been designed with custom distance pieces on one or more throws, and that may limit the load carrying capability of the frame. The following examples serve to illustrate the issues at hand.

Performance study

A group of compressors was being considered for a re-rate project to increase capacity. The economics of the project depended on (among other things) the capital cost to modify the existing compressors if the increased capacity resulted in overload conditions. The compressors in question were from two different OEMs, and were pre-1995 (prior to API-618 4th Edition) vintage. The load ratings for one type of compressor were provided in terms of “rated rod load” and “maximum allowable rod load.” The load rating for the other compressor model were provided in terms of “rated rod load,” “rated flange-to-flange load,” “maximum allowable rod load,” and “maximum allowable flange-to-flange gas load.”

For brand “X,” the rated rod load was 150,000 lb, and the maximum allowable rod load was 180,000 lb. The performance study concluded that overload conditions would occur, based on comparing the calculated pin loads to the 150,000 lb limit. It was later clarified that the MACCRL was 180,000 lb and the MACGL was 180,000 lb for this application.

The brand “Y” machines had a rated rod load of 175,000 lb, a rated flange-to-flange gas load of 187,500 lb, a maximum allowable rod load of 210,000 lb, and a maximum allowable flange-to-flange gas load of 225,000 lb. The performance study also concluded that overload conditions would occur based on comparing the pin loads to the 175,000-lb limit. It was later clarified that the MACCRL was 210,000 lb and the MACGL was 225,000 lb for this application.

The predicted overload situation initially led to proposed modifications to the compressors. This would have added a significant capital cost to the project. After the rod load ratings and operating conditions were reviewed with the OEM (and the current API definitions were applied), it was verified that there was not an overload condition. The project was delayed while the overload issue was resolved and later deferred, due to market conditions.

Combined load

Many users mistakenly assume that the combined rod loads (gas plus inertia) will always be lower than the gas loads. This is not true for low ratio (high volumetric efficiency) applications. As an approximate rule, if the discharge volumetric efficiency (VE) exceeds 50%, the gas load will reach a maximum prior to 90 degrees, while both inertia load and gas load are same sign, and thus are additive. If the discharge VE is less than 50%, then the gas load does not reach a maximum until after 90 degrees. As a result, it is opposite in sign to the inertia load, and the combined rod load will be less than the gas load.

Distorted pressure measurements

Another common mistake is to report distorted rod loads based on distorted pressure measurements. This is most often due to the “channel resonance” effect present in nearly all in-cylinder pressure measurements. The reported rod load is higher when the channel resonance is present.

Conclusions

Compressor rod load ratings are often misunderstood and misapplied. It is important to understand that the API definitions of MACCRL and MACGL are not actually rod loads, but refer to crosshead pin loads and gas loads, respectively. The API definitions help to avoid confusion, but these ratings are not always available for pre-1995 vintage machines.

“Measured” rod loads are actually computed rod loads based on measured pressures. The forces based on the measured pressures are combined with inertia forces based on the weights of reciprocating components that are input into the analysis software. If the pressure measurements are distorted and/or the reciprocating weights are not accurately known, then the combined rod loads reported will be erroneous.

There is some logic in using the simplified gas rod load calculations presented in equations 1 and 2. The trends will be correct (i.e. higher differential pressure results in higher rod load). However, if nominal flange pressures are used to rate a compressor, care must be take to include enough margin to account for the maximum possible in-cylinder pressures due to pressure drop, valve losses, pulsation and valve dynamics. These effects vary for each application.

Acknowledgement

Based on a presentation made at the Gas Machinery Conference, October 3-5, 2005, Covington, Kentucky.