You are here

Maintenance and Reliability in Plant

As a technician in plant, I learned that Maintenance and Reliability are the two things we should focus on. In most cases, plant personnels are more focused on Corrective actions rather than Preventive and Predictive approach. In this article, I’m going to share with some of the things I’ve learned from my previous job in plant. However, we will focus more on Vibration, Lubrication, Energy and Intercoolers.

Table of contents

1. Types of Maintenance
2. Vibration analysis
3. Oil analysis
4. Monitoring of the coolers
5. Energy consumption

Types of Maintenance

The various analyses are made at frequencies determined by the strategy of maintenance which has been chosen. It is thus appropriate, for that, to define the main types of maintenance met:

Preventive Maintenance

Stops are envisaged systematically, at fixed intervals, of a duration based on the minimal lifespan of the components, determined from experience. We will try to assume the moment at which the failure will occur.

Predictive maintenance

A certain number of parameters are controlled, until the measurement of one of them indicates that it is necessary to intervene. The intervention is carried out according to the evolution of certain parameters announcing a dysfunction.

Corrective maintenance

The intervention is carried out after a dysfunction has occurred.

The fact to opt for one or the other of these methods is determined by a certain number of parameters:

  • The criticality of the machine in the manufacturing process
  • The complexity of the machine
  • The nature of the treated gas
  • The historical review of the machine

Vibration Analysis

Vibratory analysis is one of the most important tools in the maintenance of the revolving machines. With today’s measuring apparatus, it is possible to follow precisely a great number of failures which can affect the machine.

As we seen in the part Vibration, it is advisable to choose the good type of sensor according to their characteristics (pass-band, type of measurement, etc.). Also, to the nature of the phenomenon that is wished to be observed (absolute/relative vibration).

Most of the time, and in a daily context, for practical and financial reasons, vibratory controls are limited to the monitoring of the total threshold of the vibrations. Thanks to sensors which are envisaged and installed by the manufacturer on the machine. The main advantage is to give a general idea of the vibratory levels without for that making advanced investigations.

These measurements can be completed by sharper analysis methods, if necessary, in case of an incident, of doubt or before or after overhaul. The goal of all this is obviously to obtain further information as for the message revealed by the vibratory signal.


In practice, it is impossible to obtain a perfect concentricity of the gravity centres of each element of a rotor. It results in the application of centrifugal forces which warp the rotor, and causes an loss of balance : this is an unbalance.

An unbalance defect is revealed by :
a component of high amplitude at the rotational frequency of the rotor in radial direction.

A phase shift close to 90° between two components corresponding to radial points of measurement on the same bearing of the rotor.

Utility of phase analysis

Many defects are expressed by a component of high amplitude at the rotational frequency, and this is phase analysis which allows to differentiate them. This is the case of the defects coming from a rotary effort (like the unbalance) and of the defects coming from non-revolving directional constraint (effort induced by a too tightened belt, loosening of bearing, eccentricity of a pulley, etc.).

Phase analysis also makes it possible to distinguish a static unbalance (the two bearings will undergo at the same time the centrifugal loads due to imbalance) from a dynamic unbalance (the two bearings will undergo the centrifugal loads in an alternative way). A phase shift close to 180° between the measurements taken on the two bearings is revealing of a dynamic unbalance.


The defect of alignment is one of the main causes of an equipment’s lifespan reduction. It creates great efforts which will involve the fast degradation of the coupling system and of the bearings.

A defect of alignment is revealed by a peak of amplitude prominent at generally 2 times the rotational frequency. It can happen to be 3 or even 4 times the rotational frequency.

A defect of alignment can affect:

  • Two rotors of a machine:
    • The axes of shafts of the two rotors present
    • An angular misalignment
    • A defect of concentricity
  • Two bearings of the same shaft of a machine:
    • The axes of the two bearings a same body of machine are not concentric.

The engagement

The frequency of engagement is made at the engagement rate of teeth according to an engagement frequency fe equal to the rotational frequency multiplied by the number of teeth.

To know if it is the teeth of the LV wheel or of the HV wheel which are damaged, the frequency modulation is to be analysed:

if Δf = 25Hz @LV wheel

if Δf = 115,7Hz @HV wheel

The deterioration of a tooth will produce a periodic shock at the rotational frequency of the wheel considered. On the ray comb obtained, the step corresponds to the rotational frequency of the wheel.

If the two toothed wheels include each one a deteriorated tooth, the spectrum shows not only the two ray combs corresponding to rotational frequencies of each wheel, but also a ray comb of step corresponding to a very low frequency, which reflects the specific shock of the two wheels between them.

When all the teeth is worn or deteriorated, the shocks occur at the crossing of each tooth. The spectrum is made up of a ray comb whose frequency corresponds to the frequency of engagement, but this time with an amplitude usually much higher. Generally, the spectrum also presents rays at the rotational frequency corresponding to imperfect balancing.

Wear of the bearings

The impulse periodic vibrations are called thus in reference to the forces which generate them and to their brutal, short and periodical character. These shocks can be produced by normal events (operation of piston compressor for example) or by abnormal events like the scaling off of the bearings or a defect on gears, an excessive play, …

The spectral analysis of the bearings defects are characterized by:

  • An important number of rays (because of the shocks),
  • At very high frequencies (~30 KHz).

In method SPM (analyze in the noise levels) used to detect the shocks in the bearings, the sensor positioned very close (even in contact) to the bearings, is excited by the shocks and enters in resonance.

It proves to be necessary to define a frequencies filter (envelope) to study the problems of bearings at very high frequencies.

Frame defect : ff = f/2 x (1 – d/D cosΦ)
Ball defect : fb = (D/2d) x f x (1 – (d/D)2 cos2Φ)
External ring defect : fo = N/2 x f x (1 – d/D cosΦ)
Internal ring defect : fi = N/2 x f x (1 + d/D cosΦ)
f = frotation wheel
N = number of balls
d = Diameter of the balls

Lubrication defects

The random vibrations of impulse type can be generated by a defect of lubrication on a bearing, just like the cavitation of a pump or an aeraulic phenomenon.

In a fluid bearing, the shaft is carried by the pressurized oil. Under the effect of rotation, the axis of the shaft reaches an equilibrium position. In comparison to the axis of the bearing, this position is defined by:

  • The distance between the axes of the shaft and the bearing
  • The demeanour angle

This stable position given by the weight of the rotor, the force related to the pressure of oil and the rotation of the shaft, is thus a function of the load, the rotary velocity and the characteristics of the fluid.

A defect of lubrication (swirl, oil lashing) modifies this equilibrium position. Possible defects can be detected by conventional spectrum analysis or by the method of Lissajous

Oil Analysis

The lubricants used allow to improve the tightness, and this in the two ways:

  • On the one hand they prevent external contaminants from polluting the process gas
  • On the other hand they prevent the leaks of the process gas itself.

The physicochemical characteristics of the lubricants used in service evolve in time by degradation and contamination phenomena. These changes finally make them unsuitable for a prolonged use.

Hydrodynamic / unctuous friction
In hydrodynamic mode, a lubricating film with a sufficient thickness is placed between the two surfaces. No metal-metal contact is then possible during the motion. The only forces brought into play to ensure displacement are due to the viscosity of the fluid: the wear is theoretically null.

In an unctuous friction, the lubricant has a particular resistance to expulsion and has a particular aptitude to adhere to surface under severe conditions.

Unctuous friction is typical for the applications of reciprocating compressors, while hydrodynamic friction is more specific for the case of fluid bearings of the dynamic machines. All the other intermediate cases may be possible.

Composition of Lubricants

The additives are integrated to give new properties to mineral or synthetic basic oils, or to improve the already existing characteristics. Other additives also allow to slow down the undesirable or negative changes of the lubricant during its use.

The lubricant additives can be:

  • Chemically active: they have the capacity to be chemically interactive with metals, in order to form a protective film, and with oxidation and degradation products, in order to make them inoffensive.
  • Chemically inert: they improve the physical properties critical for the effective performance of the lubricant.

The main types of most used additives and their function’s are described in the summary table provided in appendix.

Oxidation of the lubricant is a risk common to all our machines. While oxidizing, the oil:

  • forms acids which attack metals (phenomenon moreover promoted by the presence of moisture).
  • forms more or less soluble gums (while precipitating, the gums fill in moving parts which thus lose their mobility or cover them with an insulating varnish which disturbs their cooling).
  • consumes the reacting additives (antioxidant, anti-wear, anti-corrosion).

Lubricant Specifications

The viscosity of a fluid is its resistance to deformability and flowing. Viscosity characterizes the internal friction of an oil.

National and international organizations of standardization:

  • ISO: International Standard Organisation
  • ASTM: American Society for Testing and Materials (USA)
  • API: American Petroleum Institute (USA)
  • DIN: Deutsch Institut für Normung (D)
  • IP: Institute of Petroleum (UK)
  • AFNOR: Association Française de Normalisation (F)

The conformity of the production of lubricants is controlled by physicochemical and performance analyses. All these analyses are also standardised.

Oil Sampling

  • Sampling is a basis of any oil analysis and sampling procedure is of a paramount importance to ensure the representative sample.
  • As a rule, the intake point must be placed in a zone where the flow of lubricant is turbulent.
  • However, there is an exception to this rule applicable to sampling oil from the bottom of a tank. This type of oil analysis can reveal presence of water, precipitated wear and contamination particles.

Sampling is at the base of any oil analysis, that is why the procedure of sampling is of a paramount importance to ensure the representativeness of the sample.

The intake point must imperatively be placed in a zone where the flow of lubricant is turbulent. However, the exception to this rule is the sampling of the bottom of a tank whose occasional analysis can reveal information on the presence of water, or of precipitated wear and contamination particles.

Sampling Procedure

When there is no valve especially envisaged for the sampling of oil, a hand pump is used, and one should ensure, for a given measurement, to always keep the same length of plastic pipe (nine for each take)

Before the operation, the sampling system must be to drained of a quantity of oil greater than twice the dead volume of the circuit.

As the samples often have to be taken from the vats of draining, lines of recycling or large tanks, it is in practice often difficult to carry out this ideal. The insufficiencies of the sampling can nevertheless partially be compensated by regularly taking a sample of the same sampling point, and under the same operating conditions.

The frequency of taking of the samples for a conditional follow-up is determined by:

  • the manufacturer of the machine
  • the use of the equipment
  • the operating conditions (outdoor/indoor, compression ratio)
  • the operating cycle (does the machine work uninterruptedly or not)
  • the role of the lubricant in service
  • the historical review of maintenance

Independently of the equipment and the operating conditions, the frequency of sampling should be monthly for a true conditional follow-up of a machine. A reduction of the frequency of sampling goes hand in hand with a loss of predictive capacity of the follow-up.

Different types of Oil Analysis

Ageing of an oil results in:

  • The increase of the cinematic viscosity
  • The change of the color (it becomes darker)
  • The increase of its acid index
  • The reduction of its antioxidant content

Cinematic viscosity at 40°C

Cinematic viscosity increases if the oil gets oxidized. It must remain within the limits of its ISO viscosity grade, around 15%.

Acidity index

Its increase is the result of an oxidation of oil. Its reduction results from the consumption of the anti-wear/extreme-pressure additives.

Insolubility in n-pentane

The comparison of the rates of insoluble in n-pentane and toluene gives an indication on the quantity of degradation and oxidation products, which precipitate in n-pentane in which they are insoluble.

Infrared Spectrometry

Infra-red spectrography is based on the principle according to which, if a body is crossed by an infra-red radiation, some of its molecular bonds enter in resonance for certain frequencies of the radiation.

The radiation will then undergo an attenuation at these frequencies, characteristic of the bonds of the studied product. The importance of this attenuation is in relation to the concentration of the bond considered, that is to say of the product.

Going back to the definition of an absorptance, one could see that the latter is proportional to the concentration C in absorbing substance, and to the thickness of the container t.
A = ε.t.C

By associating the indices « 1 » with the known initial case, and the indices « 2 » with the case currently measured, one will have :
A1 = ε.t.C1
A2 = ε.t.C2

The concentration of the product could be deduced starting from the absorptance measured, providing that its concentration and absorptance in the initial state are known beforehand.

Oil Contamination Analysis

Water content

The effects of water in the lubrication circuit can have rather important effects on the operation of the machine. This water, in solution or free, can be the consequence of a condensation or a leakage in the cooling circuit. Uncontaminated oils of compressor quickly separate the water, which precipitates in the tank, where it is eliminated by draining. A water content of 0,05% is generally regarded as normal, whereas a water content of 0,1% is fixed alarm level.


Minimal quantities (in the order of a few ppm) of a detergent oil, such as motor oil, are enough to involve the formation of a stable foam in the compressor oil.

Gravimetric pollution

The test is done on a double filtering membrane. The difference in variation of mass between the two membranes after filtration gives the content of solid impurities stated in mg out of 100mL of oil. The stopping power of the double membrane is of 0,8µm.

This type of analysis gives quantitative information on the total quantity of wear or contaminant solid particles, but does not allow to know the size and the shape of these particles, nor their nature.
More precise methods of particles counting, by means of automatic meters make it possible to determine the number of particles present, divided by scales of size.

Emission Spectrometry

The oil sample is beforehand diluted and introduced into a plasma*, in the form of an aerosol mixture with argon. Under the effect of an electromagnetic field generated by the passage of an high frequency alternating current, the oil atoms and the contaminant particles, are excited and emit a light beam. This beam is then diffracted into several monochromatic beams.

Any atom of contaminant element will emit a light of a different color and this color will be always emitted with the same wave length, characteristic of the element. That is thus by measuring the intensity of the light emitted for each wave length that the nature and the quantity of metal present in an oil sample will be determined.

This analysis makes it possible to determine in a fast way the concentrations expressed in ppm (parts per million) in mass, of the various elements present in oils, either in form of additives, or in the form of wear metal particles, or of various solid contaminants, all of size lower than 7-8µm.

Plasma: state of the matter at very high temperature (possibly exceeding one million Celsius degrees) characterized by a partial or complete ionisation of the atoms.


The beginning of an abnormal wear is often associated with the increase of the concentration of the « large » wear particles, of size greater than 10µm. It appears that the data obtained by a spectrometric oil analysis present diagnostic limits.

We indeed saw that spectrometry allowed to detect particles of size going down to 7-8µm. Concerning the ferrography technique, it allows to detect the particles of dimensions going from less than 1µm up to 500µm. It is thus able to bring precise data on the state of the lubricated components by examining the particles contained in the lubricant.

Direct reading ferrography or ferrometry

The metal (or others) particles are fixed in a glass tube under the effect of a magnetic field and positioned according to their magnetic susceptibility and their dimensions.

An optical system allows to measure the optical density of the deposits and to deduce from it the value of the criteria “L” (large particles) and “S” (small particles) specific to the apparatus.
Starting from the results of “L” and “S”, the contamination of oils can be expressed and thus, the severity of wear phenomena.

Analytical ferrography

The deposit of the wear and contamination particles takes place on a plane support (glass or plastic), called “ferrogram”.

The visual identification of wear particles is made thanks to a dichromatic optical microscope.

For information only, the ferrous, non-ferrous particles and the pollutants, that can be determined with this method, are described in appendix.

Analysis of results

For the conditional follow-up of the equipment, it is essential to follow the trends given by the results of analyses. It had been shown that all equipments, even identical, have different wear and tear modes.

This is the reason for which it is necessary to introduce alarm and critical thresholds for each machine.

They are initially fixed from experience, and then by techniques of statistical control. The following values are usually taken:

  • Alarm level: average value + 2 typical deviations
  • Critical level: average value + 3 typical deviations

Monitoring interstage coolers

Our goal is always to approach the isothermal model of compression, which would theoretically give the best efficiencies. This is the reason for which inter-stage coolers are used, in order to evacuate the extra heat produced by compression.

The operating order of inter-stage coolers can be estimated by the measurement of three characteristic parameters:

  • The total exchange coefficient KS, stated in kW/°C, characterizes the heat exchange capacity, which changes with the clogging and varies with the surface of exchange.
  • The difference at cold ends : it will be called approach.
  • The level of pressure drop of the cooler dP on the gas side.

These parameters are not stationary. They strongly change according to the clogging, the load, and are in general only valid for ONE given cooler with its tube bundle.

The KS is a heat exchange coefficient characterizing the efficiency of the operation of the exchanger. In the expression of exchanged heat, the temperature Tm is defined by:

Monitoring of these factors

It passes by the determination of various models, established when the operation of the machine is considered as optimal (at the time of its startup for example)

A model of reference ΔTcold:ref = f(Q) is determined, then the ratio is calculated. This ratio allows to follow the heat exchange capacity of the exchanger.

In the same way, a model of reference ΔPgas:ref is determined, then the ratio from is deducted from it. It gives information about the cogging of the cooler on the gas side (humid air).

Finally a model of reference Ksref. Is determined, then the ratio is established. This last ratio allows to follow the heat exchange capacity of the cooler.

Energy consumption

The purpose of the performance measurement of a machine is to know the energy performance of whole or part of the group at a given moment and to know its evolution in comparison to a frame of reference.

To be able to detect an abnormal drift of a parameter, it is thus appropriate initially to know precisely the law of normal variation of this parameter according to the operating conditions (volume flow rate, etc.): This is what is called the characteristic of the parameter.

This characteristic of reference having to correspond to the normal operation of the machine, it will have to be established preferably when the machine is in good order, when it is new or after it was overhauled.

On the other hand, it is of primary importance that this characteristic accurately represents the real conditions of operation. For that, the curve will be established by regression, starting from operating conditions representative of the operation of the machine (conditions under which is the machine 90% of time, for example).

Energy follow-up

The electric power is the power consumed at the terminals of the electric motor. The thermodynamic power is the effective power of the machine, that is to say the power really transmitted to the gas between the inlet and outlet of the machine. The specific power is defined by:

Qm = mass flow rate

The values of these parameters are given by the purchasing specifications of the machine for at least one given operating point. They normally appear in the handbook of the manufacturer.

The performance curves, preferably stage by stage, allow in theory to know if the machine works correctly under various conditions of load.

Starting from these physical values recorded on site, it is possible to determine experimental points, from which the real characteristic of the machine can be deduced.

Thereafter, an operating point can be calculated, and located compared to recorded characteristics of reference. A possible anomaly can then be detected. This method moreover allows to obtain the performance curves corresponding to each machine established on site, and to very quickly detect an abnormal change in the operation of this one.

Leave a Reply