You are on page 1of 161

Optimal Pump


Table of Contents
Principles of P&ID Development....................................................................................................................... 4
The tips provided here will streamline efforts to develop piping & instrumentation diagrams

Facts at Your Fingertips: Positive Displacement Pumps................................................................................. 14

Facts at Your Fingertips: Measurement guide for replacement seals........................................................... 15

Facts at Your Fingertips: Tubing for Peristaltic Dosing Pumps....................................................................... 16

Facts at Your Fingertips: Flow Profile for Reciprocating Pumps.................................................................... 17

Facts at Your Fingertips: Construction- Cost Indices....................................................................................... 18

Facts at Your Fingertips: Vacuum Pumps.......................................................................................................... 19

Optimizing Reciprocating Compressors for CPI Plants.................................................................................. 20

Follow this guidance to improve the design, performance and reliability of these widely used machines

Gear Units in CPI Plants...................................................................................................................................... 25

Follow this guidance to improve the selection and operation of gear units in CPI plants

Fire-Water Pumps for CPI Facilities................................................................................................................... 29

Follow this guidance to improve the selection, design and operation of pumps handling water for firefighting and
related systems

Chemical Process Plants: Plan for Revamps.................................................................................................... 33

Follow this guidance to make the most of engineering upgrades that are designed to improve plant operations or
boost throughput capacity

Variable frequency drives: An Algorithm for Selecting Vfds for Centrifugal Pumps.................................. 39
Using this simple algorithm on a personal computer, engineers can evaluate competing scenarios to identify the
most cost-effective and energy-efficient pump system design

Sizing, Specifying and Selecting Centrifugal Pumps...................................................................................... 45

Follow these tips to determine preliminary pump sizing, to support cost-estimation efforts

Lubricating Rotating Machinery........................................................................................................................ 50

Follow this guidance to improve lubricant selection, process operation and asset reliability

Use Simplified Lifecycle-Cost Computations to Justify Upgrades............................................................... 55

The methodologies presented here can be used to set goals, and will enable performance comparisons among
different plants or industry segments

Remote Thermal Sensing.................................................................................................................................... 59

By making it easy to detect heat anomalies, thermal cameras and infrared thermometers support preventive and
predictive maintenance

Pressurized Piping: Sampling Steam and Water............................................................................................. 64

Without proper systems, analysis of steam and water chemistry can provide erroneous results — with costly

Plot Plan Design: Process Requirements.......................................................................................................... 70

It is important to conceptualize plant layout in terms of both ideal location and optimal geographical positioning of
equipment components
Facts at Your Fingertips: The Impact of Off-BEP Pump Operation................................................................. 76

Lifecycle Costs for Capital Equipment In the CPI............................................................................................ 77

Long-term equipment costs need to be fully considered in capital-cost assessments

A Primer on Reverse Osmosis Technology....................................................................................................... 85

Desalination by reverse osmosis is a key technology for a water-constrained world. Discussed here is its use in
industrial water treatment and drinking-water production

Materials Selection In The CPI............................................................................................................................ 92

An overview of the many factors to be considered when selecting materials of construction

Rotating Machinery: What You Should Know About Operational Problems............................................... 98

Follow this guidance to improve the operation, safety and reliability of rotating machinery in chemical process plants

Condition Monitoring for Rotating Machinery................................................................................................ 103

This valuable insight into the performance of pumps and compressors will help improve operation

Inline Viscosity Measurements.......................................................................................................................... 109

Process viscometers can help keep process control and product quality in check

Piping-System Leak Detection and Monitoring for the CPI........................................................................... 114

Eliminating the potential for leaks is an integral part of the design process that takes place at the very onset of
facility design

Oil-Mist Systems.................................................................................................................................................. 121

This primer discusses the pros and cons of open- versus closed-loop designs for lubricating pump and motor

Making Pump Maintenance Mandatory........................................................................................................... 123

Transfer pumps must be kept in optimum shape to handle harsh chemical processing operations

Improving the Operability of Process Plants.................................................................................................... 129

Turndown and rangeability have a big impact on the flexibility and efficiency of chemical process operations

Facts at Your Fingertips: Variable Frequency Drives........................................................................................ 136

Calculate NPSH with Confidence...................................................................................................................... 138

Determining net positive suction head (NPSH) can be confusing, but with these guidelines, engineers can avoid the
pitfalls of incorrect calculations

The Benefits of Seal-less Pumps for Full Product Containment.................................................................... 144

In cases where full containment of dangerous and hazardous chemicals is necessary, seal-less pumps can provide
many safety and operational benefits

Condition Monitoring Methods for Pumps...................................................................................................... 148

Applying condition monitoring tests to pumps can save costs by optimizing overhaul scheduling

Magnetically Driven Pumps: An overview....................................................................................................... 154

Understanding sealless pump technologies and their potential applications

Mechanical Seals Update: Pharmaceutical and Food Applications.............................................................. 159

For applications that require cleanliness, be sure your centrifugal pumps have the proper seals
Feature Report
Engineering Practice Loop

Piping model

Principles of




P&ID Development


The tips provided here will streamline efforts to Calculations
data sheets

develop piping & instrumentation diagrams




Mohammad Toghraei

Consultant Process
data sheets

he piping and instrumentation instrument lists, cause-and-effect
diagram (P&ID) is often consid- diagrams, control philosophy, de- Figure 1. P&IDs are technically pip-
ered to be the gold standard for scription, alarm-setpoint tables, ing and instrumentation diagrams but
the proper design, operation and line-designation tables (LDT), plot they provide a central repository of es-
sential engineering information that is
maintenance of plants in the chemi- plans, loop diagrams, tie-in lists, relevant to numerous other functions
cal process industries (CPI), includ- and many more (Figure 1). With throughout the planning and operation
ing chemical, oil-and-gas facilities, such universal applicability, P&IDs of most process plants
mining operations, food-processing are often affectionately referred
plants, and water- and wastewater- to as “primary interdisciplinary ment training in academia may re-
treatment plants. The P&ID provides documents.” sult in part from the fact that inher-
important information for manufac- ently, P&ID development involves
turing and installing equipment and Role of the process engineer more art than science. Plus, the
machinery, piping, instrumention, The duties of the process or chemi- content and structure of individual
and safe and appropriate startup cal engineer in a CPI project can be P&IDs tends to vary from company
and correct operation of the plant. broadly split into two categories — to company, and there is a constant
The P&ID is frequently refer- equipment sizing and P&ID devel- stream of new technologies being
enced by various engineering dis- opment. Therefore, most engineers introduced as older ones are retired.
ciplines — during both the design need to have skills in both areas. While volumes could be written on
stages and the operating phase. It The former skill calls for knowl- the development of P&IDs, this ar-
is also referenced in technical meet- edge related to hydraulic calcula- ticle provides a framework of recom-
ings with equipment vendors and tions, pump and compressor siz- mendations for P&ID development.
manufacturers, in hazard and op- ing, vessel and tank sizing, process
erability (HAZOP) studies, in man- safety-valve (PSV) sizing, and heat- P&ID development activities
agement meetings, and during proj- exchanger sizing. Equipment sizing The block flow diagram (BFD) is the
ect scheduling and planning. requires different skill sets, which preliminary document in the devel-
The P&ID is one of the few plant may vary by level of seniority and opment of any CPI project. It out-
documents that is created by multi- by industry segment. lines the most basic, general infor-
ple engineering disciplines working Chemical engineers should have mation related to the project. Then,
in concert. These disciplines include the knowledge that is needed to size it is the job of the process flow dia-
process engineering, instrumenta- specific equipment components re- gram (PFD) to add further details
tion and control (I&C), plot plant lated to their industry segment (for to the design before the final docu-
and piping (PL&P), mechanical, instance, distillation towers for pe- ment — the P&ID — is developed
heat ventilation and air condition- troleum refineries and clarifiers for (Figure 2). In general, the BFD cap-
ing (HVAC), and to a lesser extent water treatment). While equipment- tures the theoretical process steps
civil, structural and architecture sizing skills are routinely taught that are needed to convert a feed
(CSA), and the environmental and during the acquisition of an engi- stream to finished products while
regulatory group. neering degree, the skills needed the PFD goes inside of each of the
Similarly, the information pro- to develop meaningful P&IDs are BFD “blocks” and shows the major
vided by the P&ID allows for the often not formally taught in school, types of equipment that are needed
generation of various other impor- but rather are acquired through “on to meet the goal of each block. The
tant documents, including isomet- the job” training. BFD and PFD only show the main
ric drawings and models for piping, The absence of P&ID-develop- elements of the plant, while the
62 Chemical Engineering April 2014
Block flow diagram Process flow diagram Piping & instrumentation
Reversible system
(BFD) (PFD) diagram (P&ID)

Figure 2. Before a detailed P&ID can be developed, a BFD and PFD must be devel-
oped to identify the major aspects of the process. The BFD identifies primary streams Figure 4. The startup of a reversible
and unit operations. The PFD expands each BFD block, adding tanks, pumps and system often requires a recirculation
some instrumentation. The P&ID pulls it together with fuller details loop; it should be sized appropriately to
minimize costs

A plant with: (to ensure ease of op- All given elements must be de-
• Low capital and eration and flexibility), signed to allow them to be appro-
operating expenses
• Quick construction
while meeting all local priately isolated, drained, vented,
• Proper operation environmental and safety cleaned and flushed (via purging,
Environmental Owner
health and regulatory requirements steaming, or water flushing).
safety (EHS) (Figure 3). 4. Provisions must be made to mini-
codes mize the impact on the rest of
Plant Essential elements plant when an item, equipment or
Ideally, the specific ele- unit is out of operation.
Designer Operator ments captured in any The following points should be con-
P&ID should account for sidered when adding different items
A plant with comercially A plant with: full functionality of the to address any of the above four re-
established design • Ease of operation
procedures • Safe operation
plant in all stages of the quirements:
plant lifecycle, as outlined 1. Make sure that no added element
below: within one stage of the plant’s
Figure 3. CPI facilities require cooperation 1. All given elements — lifecycle will jeopardize another
among three parties. Each has its own responsibili- including equipment item’s function. For example, add-
ties but EHS requirements are common to all and piping items — ing bypass capabilities with a
must operate well and manual block valve for a safety-
P&ID provides more detailed ele- reliably during normal opera- related switching valve (for the
ments, capturing the real plant on tion, within the window of oper- purpose of making the plant op-
paper while ignoring the scale. ating conditions that is expected erational when the switching
Despite the simplified drawing at the plant. A basic process con- valve is out for maintenance, per
shown in Figure 2, P&ID develop- trol system (BPCS) should be Item 4 from the list above) could
ment goes beyond just expanding implemented to bring parameters jeopardize the operation of the
the PFD. There are some small items within normal conditions. The switching valve in an SIS; that is,
that are not shown but that need to five key parameters of chemical the bypass could be left open and
be developed by the designer for the process operations (temperature, therefore create a safety flaw).
P&ID. Still, the development of the pressure, flowrate, level and com- 2. Decide if added items can be
BFD and PFD requires exhaustive position) may need to be “adjusted” “merged” with each other or not.
studies and rigorous calculations continuously by the actions of the This basically involves check-
and simulations. Going through BPCS to ensure that they meet ing if a single shared item can
these “preliminary” efforts and not the requirements at the inlet and address multiple requirements
bypassing BFD and PFD develop- outlet of each component. within the plant lifecycle or not.
ment is essential, because every 2. The element operates well dur- Whenever possible, items should
single decision for main items on ing non-normal conditions, such be “merged” or “shared” to make
the PFD could have a big impact on as under reduced-capacity condi- the most of capital and opera-
the project. tions, and during process upsets, tional costs. In certain cases, this
startup and shutdown. Engineer- can be justified, especially when
The main goal of a facility ing provisions for working reliably an item needs to be added for the
The main goal of a process plant during low-capacity operating purpose of satisfying Item 3 or 4
is to produce desired quantities conditions, the use of safety-in- above. As these specific compo-
of various products while meet- strumented systems (SIS) to shut nents are not in use all the time,
ing stated quality goals. A sound down the system, and safety-relief a good process engineer will at-
plant design will take into consid- valves are examples of the types of tempt to “merge” them with other
eration the owner’s wishes for the items that can address this stage items so they can carry out mul-
plant (for instance, low capital and of plant lifecycle in P&ID develop- tiple functions.
operating expenses, the ability to ment activities. However, this last practice
build it quickly and so forth), the 3. There are enough provisions to cannot be carried out in all situ-
designer’s requirements (that the ensure ease of inspection and ations. From a redundancy point
design procedures can be trustwor- maintenance; these include in- of view, it is not always good to
thy and commercially established) situ inspection, ex-situ inspection, expect one item to carry out mul-
and the operator’s requirements workshop maintenance and more. tiple duties. Technically, one item
Chemical Engineering April 2014 63
Engineering Practice

could be time-shared when it is Table 1. Options for equipment maintenance

meant to carry out different du- In-line Off-line
ties at different times (that is,
In-place By operators doing rounds
with no overlap in duty duration). By the mechanical group
When designing for shared duty, In workshop Not applicable
keep in mind that this setup may
end up creating confusion among
operators, may be more prone to Quick hardware
cross-contamination, and may
Mechanical protection
enable a small failure to lead to
a big shutdown. Meanwhile, de- Major upset
hardware loss
signing components to be dedi- High high
cated (not shared) will drive up Alarm Mild upset
Does not meet
process goal
costs (if items are expensive), High
but they will be easier to trouble-
shoot, should a failure occur. One Interlock
system process
common example is the use of a (SIS)
system Normal
manway pressure-relief valve (specific
(PRV) on one shared nozzle on points) Low
tanks. Alarm Mild upset Does not meet
process goal
Typically, the ability to install
Low low
shared items is most practical Long-term
in batch systems and in systems Major upset hardware loss
with only intermittent operation. Mechanical protection
In such operations, a given item Quick hardware
can be used for different duties loss
during different time spans.
The following discussion explains figure 5. A diagram depicting upset conditions, such as this, can be defined for
the activities that are reqruied for temperature, pressure, level, flowrate or composition of each component
P&ID development for each stage of
a plant lifecycle. following conditions (Each is dis- plant items (such as tanks) have an
1. Normal operation. For normal cussed below): inherently high TDR, while others
operation, each item on the P&ID a. During reduced-capacity opera- (such as equipment with internal
needs to be able to carry out the tion weirs or vessels with internal feed
duty it has been assigned. Since, b. During startup distributors) have a lower TDR. The
in the majority of cases, this is not c. During upset conditions duty of the process engineer is to
achievable through equipment de- Reduced-capacity operation. Oc- provide the required TDR (defined
sign alone, a control system should casionally, actual plant capacity by the client) for each equipment
be implemented on the equipment. should be reduced from the design component and for the entire plant.
The BPCS must ensure that the capacity for a variety of reasons. One method of providing TDR is
design of the equipment will force Such conditions may result from a to install multiple smaller compo-
the equipment to operate within a shortage of raw materials or an ex- nents in parallel, instead of one big
“window” of expected results, typi- cess of production, or from downtime piece of equipment. Another solu-
cally at its best operating point. of a critical equipment component tion is to implement a recirculation
In a broad sense, a control system or unit. The process engineer usu- loop around the equipment to com-
is supposed to bring the five main ally provides some turndown ratio pensate for the low flowrate.
process parameters — flowrate, (TDR) for the plant. TDR is a ratio Startup operations. Startup opera-
pressure, temperature, level and between the normal capacity of the tion could be assumed to be a severe
composition — into the required plant and the minimum running ca- capacity-reduction case. Be cause
range. “Composition” encompasses pacity that is possible without los- process parameters during startup
many relevant parameters, ranging ing the quality of the product. TDR are not necessarily within their
from viscosity, density and conduc- can be defined for the equipment, range, equipment and instrumenta-
tivity to octane number and Brix for a unit, or for the whole plant. tion are not expected to be working
number. All utility distribution and Some owners or operators ex- according to full operating expecta-
collection networks, and heat-con- pect the engineer to provide a TDR tions. However, startup specialists
servation insulation, must also be of around two for their plants (this are often onsite in this stage and
decided at this stage. means they want to be able to oper- can help to compensate for the tem-
2. Non-normal operation. Non- ate the plant at half capacity without porary lack of operability of equip-
normal operations occur under the losing any product quality). Some ment and instruments.
64 Chemical Engineering April 2014
• Each drain can cover a portion of a system; Vents can cover
a bigger portion
• Drain or vent size should be manageable (Minimum size
should be ¾ in.; limit to 2 in., unless inside of dike)
• Multiple drains and vents should be implemented in a cov-
ered area to ensure draining or venting within a reasonable tp = 2 years tm = 1 year tp = 2 years Pi

tm = time between
• When the system is small: drain = vent (usually for pipes <2 each maintenance, inspection or cleaning event
in.), and there is no need for dedicated vents or drains
tp = time between
• The drain or vent should be routed to a safe location (hard each plant overhaul or turnaround
piped, if required)
• For volume < 0.5 m3 (such as pump casings), use ¾-in. tp = 2 years tm = 3 years tp = 2 years
• Vents can be one size smaller than the drain
• Drains for liquids with viscosity higher than 50 cP could be
one size bigger than guidelines stated above Figure 6. When planning for isolation valves, the engineering
team should evaluate data related to the anticipated time for sched-
uled maintenance and anticipated turnaround schedules

Table 2. options for isolating a portion of the process for normal operation for the pur-
from the plant pose of startup recirculation is so
Type Symbol Credibility strong that some process engineers
1 Block valve Not forget to think about the startup op-
(with or without lock) Process acceptable eration during the development of
2 Block valve the P&ID; they simply assume they
(with lock) and blind Process will find a way to accommodate

Safer isolation
3 Double block valve startup somehow without actually
(with lock) and bleed Process planning for it.
Upset conditions. Upset conditions
4 Block valve
can be defined as operation of the
(with lock) and blind Process plant when some of the process pa-
and removable spool rameters are beyond the normal
band. In Figure 5, this situation
is arbitrarily split into two differ-
Table 3. Different methods of removing material ent cases — mild upset and severe
from equipment for inspection or maintenance
upset — for any of the five key pa-
Type of Removal method P&ID rameters (flowrate, pressure, tem-
“dirt” perature, level, and composition). In
1 Solid/ • Manual • Nothing is needed on P&ID both cases, during upset conditions,
semi-solid: •M achine-as- • Do we need “clean-out” doors? the process goals have already been
removal sisted
lost so the immediate goal is to
2 Liquids: • F lushing: By For all the cases , three options are protect the equipment (hardware
Washing water available to show on the P&ID:
conservation) and the health and
•S  teaming out: 1. Only washing valves
By utility steam 2. Washing valves that are hard piped safety of the personnel and neigh-
•C  hemical 3. Hard piped washing system with boring communities.
cleaning: By switching valves for automatic To address point upset conditions,
chemical solu- washing the facility should be equipped with
tion or solvents
an alarm system and a SIS. The
3 Gases: •N eutral gas • If it is by inert gas, the same options alarm setpoints are usually on the
Purging purging for “washing” (above options) are
maximum (or minimum, in some
• Ventilation available here
• For ventilation (by natural draft of cases) value of a parameter, and the
air), imake sure there are at least 2 SIS action will be set to the high-
nozzles are available high (or low-low) level. However,
some additional alarm setpoints
For reversible systems (such as sively large circulation loops, so as or additional SIS setpoints can be
reactors that carry out equilibrium not to waste money for piping that added, too.
reactions), startup operation can be is supposed to be used only during The purpose of this SIS action is
supported by recirculation. If the startup. As much as possible, the to shut down a plant and bring it in
system is not reversible, the startup design should try to use the existing the lowest energy state (in terms of
operations can be more complicated pipe arrangement for the purpose lowest pressure, lowest temperature
and case-specific. Figure 4 shows of startup recirculation, especially and so on) Other than “event-based
the basics of this procedure. when high-bore pipe is needed to SIS” explained above, SIS action(s)
If recirculation is to be used dur- support startup efforts. can also be activated by the opera-
ing the startup procedure, efforts The tendency to use the piping ar- tor. This shutdown is named “opera-
should be made to avoid exces- rangement that was implemented tor activated SIS.”
Chemical Engineering April 2014 65
Table 4. Examples of removable spools (RS)
Item Potential P&ID example
Engineering Practice location of RS
Centrifugal Suction and PG
pump discharge side

The duty of the alarm is to warn

the operator that something has RS RS
gone wrong. If for whatever reason Progressive- Discharge side
the operator fails to respond in a cavity pump
timely manner, the SIS system will RS
initiate the action that the operator
has failed to, or tried without suc- Shell-and- Tube side
cess. This allocation of responsibili-
ties between alarm system and SIS exchanger
is shown in Figure 5.
If, for whatever reason, the SIS Vessels and Lines out of
tanks flanged head RS
cannot mitigate the parameter that or blinded
has deviated from the normal point nozzles
and it has gone beyond high-high
(or low-low) level, then finally a me- Table 5. Options to deal with lost items in a plant
chanical item needs to be triggered
to “tame” the system and regain Option Schematic P&ID example
control. Even though a mechanical The exact
system (as the last line of defense) replica in
can be considered for each of the
five parameters mentioned earlier,
pressure safety valves (PSV) are a A similar item in
popular type of mechanical defense
against a wild parameter. Installing
PSVs, and routing their release to an
Bypassing the absent
appropriate destination, is an essen- item
tial task during P&ID development. Bypass
Winterization is another issue
that is resolved in this stage. Win- Redirecting the in-flow
terization involves implementing to a “reservoir” for later
specific features in a plant design to use
prevent any impact of cold weather Pond
during a plant shutdown. For in-
stance, winterization efforts typi- Upstream tank stores
cally start with provisions to enable this inlet flow and the
downstream tank pro-
“natural internal drainage” of the vides outlet flow for a Furnace
equipment and pipes more tolerant short period of time
items, such as tanks. Other activi-
The inlet flow is sent To flare
ties include heat tracing and insu-
permanently for ulti-
lation of pipes to prevent freezing mate disposal and the
or settling of non-drained (trapped) stream will be wasted Ultimate disposal
fluids, and installing fluid mov-
ers on emergency power sources The absence of an
to provide recirculation to prevent item doesn’t generate
freezing/setting in the case of power any upset in the rest of Steam
plant or whole plant generator
3. Inspection and maintenance. should be shut down
Equipment care can be categorized
into “in-workshop” and “in-place” pair or maintenance. Each requires • Sound: To sense vibration, cavita-
care, and the latter can be catego- different types of provisions for the tion, hammering, PSV release, ex-
rized further into “in-line” or “off- equipment on the P&ID. plosion and more by listening
line” operation. In-place care is Operators making rounds could • Touch: To detect vibration
usually done by operators making be equipped with portable sensors; • Smell: To detect fire, leakage, PSV
rounds, while in-workshop care is if not, then he or she must rely on release to atmosphere, and more
typically carried out in a workshop the use of the senses: To support the work of the operator
by a mechanical group (Table 1). • Sight: To observe, for instance, making rounds, specific items can
Inline care can be considered the leakage, vibration, overflow of be put on the P&ID. These may in-
inspection of operation equipment, tanks, fluid levels, flame color and clude sight glasses to check liquid
while off-line care is equipment re- shape levels, catalyst levels or filtering-
66 Chemical Engineering April 2014
Table 6. An Example of P&ID development for a pump (four phases of operation)
Case P&ID
Normal operation
Putting the pump call-out with the required informa- Call-out should be put on P&ID
tion on top of P&ID sheet
Placing a reducer/expander to match suction and
discharge side of the pump, if needed (may need
a top-flat eccentric reducer at connection)

Adding a permanent strainer to prevent damage to

the pump

Making sure the suction pressure (and tempera- It needs some calculations. The impact on P&ID could be seen
ture) is enough. This reflects the sensitivity of on suitable in upstream container of pump
centrifugal pumps towards NPSH
Showing the pump’s BPCS for capacity control of FT FC

pump FE FV


Adding pump driver control PM




Non-normal condition

Considering a temporary strainer (commissioning) A permanent strainer is already placed

Adding a non-returning valve in the case of reverse PM




Using the minimum flow line on the discharge line FC

with a control valve to protect the pump from flows FV

that are lower than the minimum flow of the pump



Showing the pump SIS and/or alarming system FC

to protect the pump from an abnormal condition FV

(one example is a monitoring system for seal leak- S/S COMMAND



Continues on pg. 69

media levels, or peep holes to check these may include portable pres- may include, for example, a pres-
the color and shape of flames in a sure gages, temperature sensors sure tapping (PT) point, or temper-
furnace or boiler. and so on. The P&ID designer may ature point (TP), to be shown on the
In terms of the use of small, por- decide to provide some “test points” P&ID. An example of PT location
table measuring devices that can be instead of fixed gages, to save some could be the suction side of centrifu-
used by operators making rounds, money in non-critical points. This gal pumps. The decision must be
Chemical Engineering April 2014 67
Engineering Practice

made to either use test points and viding isolation valves is not nec- tions.” This means having provi-
fixed gages that transmit informa- essary for all the equipment in a sions that will allow all five key
tion to the control room, or to imple- plant. Isolation valves are required process parameters to be brought
ment a control loop that depends on to isolate the equipment from the into a safe range:
some parameters based on the criti- rest of the plant if the equipment • Ensuring safe temperatures:
cality of the parameter. is expected to need “off-line care” Options include allowing time
Meanwhile, in-place, off-line care at frequent intervals, in time du- lapses, or options for cooling down
may include chemical or solvent rations that are shorter than the (or warming up, in the case of
cleaning, steaming-out, pigging op- scheduled plant turnaround times). cryogenic services) streams. For
erations and so on. Depending on For instance, if (based on histori- some systems (for instance some
the operation-specific requirements, cal data), the unit expects to need batch operations) that require a
different items should be imple- off-line care every three years but more rapid cooling (or warming)
mented (such as chemical cleaning the entire facility for which you by cooling streams
of valves). are developing the P&ID will need • Making pressure safe: Venting is
For all off-line care a specific ar- planned turnaround work every widely used
rangement must be made to ensure two years, there is no need (at • Ensuring appropriate flowrates:
positive isolation of the system from least theoretically) to put isolation As long as equipment is isolated
the rest of the plant. This arrange- valves upstream and downstream from the rest of the plant, there is
ment typically comprises isolation of the unit. This concept is shown no flow going into it, and it is not
valves, drains, vent valves and so on. in Figure 6. a point of concern
The isolation system is discussed in In some cases, companies don’t • Making levels safe: Drainage op-
greater detail below. provide isolation systems for es- tions are needed for tanks, vessels,
For in-workshop care, the provi- sential equipment, such as heat pump casing and more. Some gen-
sions defined by in the P&ID are exchangers. The logic is that they eral rules for sizing and installing
items that will allow the equipment essentially cannot afford to put the drain and vents are in the Box (p.
to be removed from their founda- heat exchanger out of service, so 65)
tion easily and safely. However, adding isolation valves would be ir- • To ensure safe compositions, the
the characteristics to satisfy this relevant. body of the equipment (external
requirement are not always shown The answer to the second question and/or internal) must be safe in
on P&IDs (mainly to avoid clutter- is that the isolation system should terms of exposure. These provi-
ing of the P&ID). For example, if be added on all downstream or up- sions involve proper cleaning of
equipment needs to be hoisted for stream connecting pipes, as close as the equipment.
removal, this engineering detail is possible to the equipment. However, Table 5 shows options for mak-
often not shown in the P&ID. Items some companies challenge this and ing the composition safe for dif-
that must be shown on the P&ID in- question if there is real needed to ferent types of materials inside
clude the following: put isolation valves on, for example, of the equipment. Washing and
• Isolation valves that allow the a vent pipe to atmosphere or not. purging (through ventilation) are
equipment components to be de- To answer the third question, especially important for walk-in
tached from the rest of the plant it should be stated that there are equipment.
• Drains and vents different type of isolation systems. The last step as mentioned above
• Removable spools (RS) that would Table 3 summarizes these methods. is to provide removable spools (RS).
be used around the equipment to Decision needs to be made about Sometimes required RS are already
allow it to be “untangled” from the the type of isolation method. The present due to previous activities
system by removing the piping sys- isolation method depends on fac- on the P&ID. Table 4 provides some
tem interference; this allows for tors, such as the equipment envi- examples.
easy equipment transfer to ronment (for instance, for confined Allocating a utility station in dif-
the workshop spaces or non-confined spaces), ferent locations of the plant, and de-
When it comes to preparing for the fluid type (aggressive or toxic ciding about the required utilities
off-line care, with regard to de- or not), and the pressure and tem- for each utility station, is another
signing isolation systems, the fol- perature of the system. Usually the activity to address this stage of the
lowing three questions should be first type of isolation (Table 2) does plant lifecycle.
answered: not provide enough “positiveness.” 4. Operability of the plant in the
1. To which equipment should the Possibly the only application of this absence of one item. The designer
isolation elements be added? isolation method is for instruments. needs to decide the impact of equip-
2. Where do they need to be placed In such an application, the isolation ment loss on the rest of plant opera-
“around” the equipment? valve is called a root valve. tions and take engineering steps to
3. Which types of isolation systems The next step for making equip- minimize its impact. The wide range
or elements should be used? ment ready for periodic removal is of answers and decisions should in-
To answer the first question, pro- to bring it to “non-harmful condi- clude the following:
68 Chemical Engineering April 2014
Table 6. An Example of P&ID development for a pump (four phases of operation) (continued)
Case P&ID
Maintenance / Inspection
Adding a pressure gage on discharge and/or
suction side



Adding block valves in the suction and discharge FC

line (such as a gate valve) in order to isolate the FV
pump during maintenance


Consider vent and drain valves in the pump FC

suction and discharge sides and in the pump cas- FV


Consider the use of a piping spool piece to facili- It is already created and exists
tate dismantling
Installing pump insulation for personal protection Service temperature is 40°C and there is no need for personnel
protection insulation
Production interruption
Define the pump sparing philosophy Based on RAM analysis, a second pump with the same arrange-
ment is added (to provide 2 x 100% capacity)

1. A parallel, exactly similar spare to the alternate component in- 5. The storage tanks upstream and
system can take care of flow that stead. Examples include having a downstream of the component
would result from the loss of a given manual throttling valve (such as should have enough residence
component. Examples include spare a globe valve) in the bypass line time to continue operations. This
pumps or spare heat exchangers of a control valve, or placing a by- way, if the component goes out
(in highly fouling services). The pass line for a PSV together with of service, the upstream string
installation of spare equipment a pressure gage (or pressure tap- of equipment can still feed the
is popular for fluid-moving equip- ping) and a globe valve. upstream tank and downstream
ment, since interruption of service 3. The feed to the equipment can be components can still be fed by
in pumps and compressors cannot simply bypassed temporarily with the downstream tank. This ar-
be handled through other below marginal impact on the operation rangement will prevent a surge
options. One important example is of the system. that could impact connected plant
having two fire pumps installed in 4. The feed to the equipment can components.
parallel, with two different types be redirected temporarily to an 6. The feed to the equipment is re-
of drives (for instance one with an “emergency reservoir” (such as a directed temporarily to a waste-
electromotor and the other using a tank or pond), and processed later receiving system or flare.
diesel drive pump). by returning it back to the system. 7. Whole plant or unit should shut
2. A parallel component can be used Usually this option is available for down: This option should be
and the flow can be redirected liquid streams. avoided, if possible. However,
Chemical Engineering April 2014 69
Engineering Practice

Too few tools Too many tools

sometimes this is inevitable when pumps (that may need to
the equipment of interest is a key periodically function as
asset in the facility. spare pumps for all other
Table 5 summarizes these options pumps), the specific pip- Flexibility
for a P&ID. ing arrangements around Badly operating in operation
the pumps will need to be plant
Spare pump options elaborated on the P&ID.
While the discussion below focuses Table 6 shows an exam-
Figure 7. A balance must be sought between
on pumps, the guidelines apply to ple of P&ID development “too much” and “too little” when developing the
any other types of spare equipment for a pump in one case. This engineering details for a P&ID
as well. A spare pump, depend- table only provides the re-
ing on the criticality of the service, quired thought process for
could be “an installed spare” or “a the development of a pump P&ID as CPI facilities should be brought
workshop spare.” A workshop spare an example. closer to the desired operating
pump is not installed but can be point through the use of a control
moved from the workshop and de- Additional important items system. Key parameters that must
ployed within a short period of time In addition to the last four stages be controlled include flowrate and
(say, 24 hours). of the lifecycle of a plant (discussed head for pumps, and heat duty for
Decisions related to any of the above), a few other items must be heat exchangers.
above two options (in terms of in- considered: 2. Check the required temperature
stalled spares or workshop spares) Future plans. If there is any plan and, pressure for each item (inlet
can be based on different param- for expansion, or any prediction for and outlet) and make sure these
eters, including the following: implementing new, under-review are matched with process needs.
• Mean time between failure innovations in future, this needs to If they are not matched, take ac-
(MTBF) of the equipment be addressed in the P&ID to facili- tion to address them.
• Mean time to repair (MTTR) of tate the implementation of the fu- 3. Check the required flowrate for
the equipment ture changes with minimum impact each item. What is the minimum
• Cost of maintenance on the operating plant. flowrate that can be handled
• Value of the “lost production” Insulation to safeguard personnel. without negative impact on the
For installed spares, if the ambient Equipment and pipes with skin process, and what is the mini-
temperature of the space around temperatures that are greater than mum flow that can be accommo-
the a pump is far from the operat- 60–75°C (especially for metallic dated before there is potential
ing temperature of the pump (for items), and those that are located harm to the equipment? If there
instance, differs by 100 to 150°C), in crowded areas within reach of is a chance of harm from low flow,
the pump should be “a hot standby workers, must be insulated. This in- then plan accordingly to protect
pump” (or “a cold standby pump” for sulation is called personnel protec- the equipment from it.
cryogenic service) to make sure it tion (PP) insulation on the P&ID. 4. Check the required composition of
will not experience thermal shock the streams going into the equip-
during the startup. Otherwise the Useful rules-of-thumb ment and note the special care
pump could be installed with no Whether the design engineer (in that should be taken. For exam-
specific “stand-by provisions.” the role of P&ID developer) is ple, a positive-displacement pump
If a spare pump is supposed to capturing general items (such as is prone to plugging if the liquid
“sit” beside more than one operat- containers, fluid movers, heat ex- contains large suspended solids.
ing pump, another feature that changers and so on) or more spe- In this case, a strainer should be
should be decided is whether the cialized items (such as liquid-ex- installed.
spare pump is a common spare traction towers, filter press and so 5. Be sure to account for the required
(to be available for several operat- on), these general rules-of-thumb utilities and their temperatures
ing pumps), or is intended for use can help: and pressures.
with just one dedicated pum; thus 1. Much of the equipment that we 6. What are the weak points of the
a spare should be installed for each buy for any given plant is not item and what requirements
operation pump. “custom built equipment,” so we should be taken when designing a
For spare pumps, the user might cannot expect the components to proper SIS system for the item?
expect that all pumps should be operate exactly according to the 7. Which parameters must be moni-
able to act as both an operating desired operating points. Even for tored by the operator making
pump and a spare pump for any the case of custom made items, we rounds? Think about the five key
of other pumps that may be out usually expect that the equipment parameters: Temperature, pres-
of service (where no specific spare will operate in a pre-determined sure, level, flowrate and composi-
pump has been designated). When “window” of operation. The result tion.
providing common (shared) spare is that almost all equipment in 8. Be sure to acknowledge which as-
70 Chemical Engineering April 2014
pects of each component need in- or will it be captured in other docu- capacity of the system, and ambi-
spection or monitoring. ments?” The P&ID is supposed to ent temperatures and pressures
9. Review any history of item fail- be a common document that can be will most likely differ for each new
ures (in terms of frequency and used by quite a few different disci- project. When developing P&IDs, a
time for maintenance) and act ac- plines. Incompleteness is an inher- previously effective method may be
cordingly to address them. ent feature of it. Furthermore, the entirely ineffective in the current
10. Consider the impact of an item P&ID is supposed to be kept in the project, while a method that has
going out of service. What steps plant, for easy use by operators. If proven useless in the past may work
can be taken to minimize the it is too cluttered, its usefulness is perfectly well this time around. ■
impact of this on the rest of the diminished. Edited by Suzanne Shelley
plant? Is it possible to have a sim- Generally speaking, all process
ilar system as a spare? equipment should be shown in Author
P&IDs. Sometimes, non-process Mohammad Toghraei is
currently an independent
Common challenges related P&IDs (such as gearboxes consultant and is the instruc-
During the development of a and lubrication systems) should tor of several P&ID-related
courses offered through
P&ID, the need to choose between also be shown on the main P&ID or Progress Seminars Inc. (Web-
competing options is a common on auxiliary P&IDs. Meanwhile, if site; Email:
challenge. Here are several common they are not shown on P&IDs, their Toghraei has more than 20
years of experience in pro-
scenarios: details can be found in vendor doc- cess engineering. For the
“Should I show a given detail on the uments. past seven years, he has
held different technical and leadership roles
main body of P&ID by a schematic, All pipes and pipe appurtenances related to oil removal and water treatment for
or can I capture it in the note area?” except bends and elbows are shown steam-assisted gravity drainage (SAGD) proj-
ects. Toghraei holds a B.Sc. in chemical engi-
The P&ID is a pictorial diagram. As on P&IDs. Flanges should be de- neering from Isfahan University of Technology,
much as possible, the P&ID should picted, if there is a specific reason and an M.Sc. in environmental engineering from
the University of Tehran, and is a member of
capture relevant schematic shapes. for them. Specific piping items that APEGA. He is a certified professional engineer
in Alberta, Canada.
“Should I add the item or not?” Items are not shown on the P&ID can be
should be added to give required found on piping models.
flexibly to the operator. A plant with When it comes to instrumenta-

Please visit REMBE®, INC. at PTXi Conference & Exhibition, Rosemont, May 6-8, 2014, booth 1228
insufficient “facility resources” is dif- tion and control system, things
ficult to operate.However, from the become more debatable. The three
other side, this is also the case for a
plant with more than enough pipe
main items of integrated control
and safety system (ICSS) elements
circuits, control valves, alarms and are: Regulatory control system EXPLOSION
SIS actions. For example, a plant
with too many alarms can “overload”
(BPCS), the alarm system and the
SIS. Almost everyone agrees about

Indoor / Outdoor Venting

the operator and result in a loss of basic process control items should With Flame Absorber
urgency from the operator when an be shown on the P&IDs. They are and Dust Retainer
alarm does activate (Figure 7). mainly the elements of the control Q-Box
“Adding more doesn’t hurt.” This is loops. For alarming systems, the
a popular statement when P&ID same clarity exists. The main de-
developers try to “bypass” conduct- bate is usually on SIS systems, in Q-Rohr-3
ing a rigorous evaluation for the terms of the question of “down to
necessity of an item on the system, which level of detail the safety in-
and thus place it with no real ne- terlock loops should be shown on
cessity. However, designers should
remember that in some cases, add-
the P&IDs?”
Different companies follow differ-
ing an item might not necessar- ent directions. CONTROL
ily increase the capital cost of the “Based on my past experience...” The

…for Powder,
project — if the item is small and inherent creativity required in cre- Granulate,
Grain, Flour
relatively inexpensive — but may ating P&IDs may become hindered, and Dust
All rights reserved

still increase the operating cost be- if for every single case one refers
cause of required inspection, main- to “past experience.” As unlikely as
tenance, related utility and chemi- it may seem, the “this is what has
cal usage and more. In addition been done before” mentality is not
to that, any new item added to the the most efficient way of developing ✸✸✸ WE DO IT BETTER ✸✸✸

system provides a new opportunity this document. That being said, the contact us …for Europe …for North America

for mistakes, cross -contamination, technological innovations, availabil-

59929 Brilon/Germany Charlotte, NC 28217
T + 49 (0) 29 61 - 74 05 - 0 T +1 704.716.7022
leaks and other problems. ity of materials, quality of raw ma- F + 49 (0) 29 61 - 5 07 14
F +1 704.716.7025

“Should I add it here on the P&ID terials, required quality of products, Circle 24 on p. 76 or go to

Chemical Engineering April 2014 71

Department Editor: Scott Jenkins Pumps

P umps are essential equipment in the chemical process industries (CPI).

Given the imperatives to reduce production costs and maximize energy
efficiency, pump selection is crucial. Knowledge of pumping basics,
along with the fluid characteristics for a particular application can help
engineers make the best choice.
PD PUMP operating principles
• PD pumps displace the same volume of liquid with each rotation of the
shaft, so flow is proportional to pump speed
• PD pumps are self-priming
• PD-pump mechanics require close-fitting internal parts with some running
While the majority of pumps in use in CPI plants are kinetic energy pumps clearance
(the largest category being centrifugal pumps), positive displacement (PD) • PD pumps require a pressure-relief mechanism (either relief valves or rup-
pumps are an important class of industrial equipment. The following is a ture discs) in case of discharge blockage
collection of information on several types of PD pumps and an outline of the
differences between positive displacement pumps and centrifugal pumps. Types of PD Pumps
Internal gear pumps
Selection Starting Point Internal gear pumps have an outer gear Rotor
called the rotor that is used to drive a
In pump selection, first consider what the expectations of the pump will
smaller inner gear called the idler. The
be. The following parameters must be determined before a pump can be
idler gear rotates on a stationary pin
and operates inside the rotor gear. As
Inlet conditions — To avoid suction problems, the pump should be located the two gears come out of mesh, they
as close as possible to the liquid supply create voids into which the liquid flows.
Flowrate — The flowrate requirements for the pump should be considered When the gears come back into mesh,
Differential pressure — Smaller pipe size and longer pipe runs reduce volumes are reduced and liquid is forced
initial system cost, but the higher pressure differential raises energy consump- out of the discharge port. A “crescent”
tion and reduces pump lifetime is formed between the two gears that
Liquid characteristics — The properties of the fluid to be pumped — includ- functions as a seal between the suction
ing material compatibility, viscosity, sensitivity to shear stress and presence of and discharge by trapping the volume
particulates or solids — are important factors of liquid carried between the teeth of the
rotor and idler.
Internal gear pumps are effective with
PD versus centrifugal Pumps viscous liquids, but do not perform well
for liquids containing solid particles.
PD and centrifugal pumps behave differently. As a means to move liquids,
centrifugal pumps rely on kinetic energy, forcing liquid out of the pump with
External gear pumps
energy imparted to the liquid as it moves toward the outer diameter of a
External gear pumps have a similar
rotating impeller (pressure is created and flow results). PD pumps work by
pumping action to internal gear pumps in
capturing confined amounts of liquid and transferring them from the suction to
that two gears come into and out of mesh
the discharge port (flow is created and pressure results).
to produce flow. The difference is that
The following plots represent examples of the performance behavior differ-
external gear pumps have two identical
ences for centrifugal and PD pump types:
gears rotating against each other. Each
A. Flowrate versus pressure — Centrifugal pumps exhibit variable flow gear is supported by a shaft with bear-
depending on pressure, whereas the flow in PD pumps is largely indepen- ings on both sides of each gear.
dent of pressure External gear pumps work well in high-
pressure applications, such as hydraulics,
B. Efficiency versus viscosity — For centrifugal pumps, efficiency but are not effective in applications
decreases at greater viscosities. Positive displacement pumps are actually requiring critical suction conditions.
more efficient at higher viscosities
C. Flowrate versus viscosity — Centrifugal pumps lose flow as viscosity Lobe pumps
increases, while the flow of a PD pump can actually increase at higher Lobe pumps resemble external gear
viscosities pumps in operation, except that the
pumping elements do not make contact.
D. Efficiency versus pressure — Changes in pressure have minimal effect
Lobe contact is prevented by external
on PD pumps, but have a dramatic effect on centrifugal pumps.
timing gears.
Lobe pumps perform well with liquids
A. Performance B. Viscosity
250 100 that contain solid materials, but do not
perform well with low-viscosity liquids.
200 80
Efficiency, %

Head feet

Vane Pumps Rotor
150 60 Sliding vane pumps have a rotor with
100 40 radial slots, and it is positioned off-center
Centrifugal in a housing bore. Vanes that fit closely
Centrifugal into rotor slots slide in and out as the ro-
50 20
tor turns. Pumping action is caused by the
0 0 expanding and contracting volumes con-
0 50 100 150 0 250 500 750 1,000 tained by the rotor, vanes and housing.
Capacity, gal/min Viscosity, cSt Vane pumps are effective for low-
viscosity liquids, and when dry-priming is
required. They are not ideal for abrasive
C. Flowrate D. Head liquids. Vane Push rod
110 80
100 Positive
Efficiency, %

70 References
Flowrate, %

1. Petersen, J. and Jacoby R., Selecting a Positive Displacement Pump, Chem.
80 Eng., August 2007, pp. 42–46.
70 Centrifugal 2. Viking Pump Inc., When to use a Positive Displacement Pump, The Pump
60 Centrifugal 50 School Website, Viking Pump Inc.,, 2007.
50 3. Soares, C., “Process Engineering Equipment Handbook,” McGraw Hill,
40 40 New York, 2002.
0 100 200 300 400 500 55 80 105 4. “Perry’s Chemical Engineer’s Handbook,” 8th ed. McGraw Hill, New
Viscosity, cSt Feet of head York, 2008.
Measurement guide
for replacement
Department Editor: Scott Jenkins seals

echanical seals play a pivotal role in the chemical Radial space available Shaft Tap depth
process industries (CPI). As the sealing interface around shaft center line or stud
Bolt/stud to top of length
between rotating process equipment and the Bearing bearing
materials contained inside, these devices prevent diameter Sleeve length
product leakage and fugitive emissions, while per- from mounting
mitting process machines, pumps and compressors surface
to operate at ever-increasing pressures, vacuums,
temperatures and speeds. Even the most basic Shaft/sleeve dia.
mechanical seal is a complex assembly. diameter Shaft length Vessel
Correct measurement of the machinery for (if any) from mounting I.D.
replacement seals can be an underappreciated Bearing surface
aspect of maintenance planning. shelf
to shaft Distance to Recess
The following steps outline basic procedures and Bolt/stud bearing from
Bearing center depth
considerations involved in measuring machinery for pattern height line
mounting surface (if any)
replacement seals.

Measuring For shaft seal replacement

Step 1. Shaft diameter MBC MBC
Measure the shaft diameter (ø) where the replacement seal will
be installed. It’s important to actually take a measurement of
the shaft and sleeve, rather than relying on dimensions from the
machinery’s blueprints.
The shaft diameter can vary slightly at different points along
its length, so take the measurement where the seal will be
positioned. This may mean removing the existing seal, or at least
sliding it out of the way.
The shaft may be worn or fretted, so obtain an average
measurement of the shaft. This can be accomplished by taking X
several measurements with a micrometer or caliper, then averag- MBC = X / cos 30 deg
ing them. An alternate way to get an average measurement of Verify MBC = Ø + 2(Z ) Verify MBC = Ø + 2(Z )
the shaft is to use a diameter-reading tape measure. These are
precision tapes with a vernier scale, which allows you to mea-
sure shaft diameters to within 0.001 in. MBC
After taking the most precise measurement possible of the Z3 Z2
shaft and the sleeve, if any, it may be useful to take a photo-
graph of the shaft if it is fretted or worn. Your seal supplier can
be a resource for more information. Y X

Step 2. Radial clearance Z

Using a rule or tape measure, measure the radial clearance around Z4 Z1
the shaft and identify the nearest radial obstruction. If more than
one obstruction exists, measure and mark all on a sketch.
Step 3. Axial clearance
MBC = √X 2 + Y 2 MBC = 2(X)
Measure the axial clearance to the first obstruction. This is usu- Verify MBC: if Z 1 = Z 2 = Z 3 = Z 4 Verify MBC = Ø + 2(Z )
ally the distance from the seal mounting point to the bearing or then MBC = Ø + 2(Z)
drive assembly. There may, however, be other closer obstruc-
tions, such as shaft couplings. A sketch can be helpful in com- MBC
municating your needs.
Step 5. Mounting surface
Step 4. Seal mounting Check the surface on which the seal
Determine how the existing seal is mounted. For example, are will be mounted. Is it flat or perpendic-
there bolts or studs fastening it in place? If so, how many are ular to the shaft? Use a straightedge
there, what size are they, and how are they distributed around and a machinist’s square to verify this
the diameter of the shaft? information, which will be important
Start by measuring the distance between the centers of each when mounting a new seal. Z
pair of adjacent bolts (X). Take this measurement for every pair Note that some equipment has a
of bolts — bolt spacings may not be equal around the circle. machined recess around the shaft
See the accompanying illustrations for how to convert com- cutout, where a boss on the seal is ⎛ ⎛
MBC = 2(X ) ⎜ sin 67.5 deg ⎜
mon bolt patterns to a mounting bolt circle (MBC). You can intended to fit for centering purposes. ⎝ sin 45 deg ⎝
check your work by measuring from the shaft outer diameter to If there is such a recess, measure Verify MBC = Ø + 2(Z )
each bolt center. Twice this distance, plus the shaft diameter, both the diameter and the depth.
should also equal the MBC dimension. Careful attention is
important here because the shaft may not be centered exactly Acknowledgements
within the bolt circle.
If studs are used for fastening, determine their lengths. Measure Material for this month’s Facts at Your Fingertips was provided by Woodex Bear-
lengths and provide details of thread length, as well as whether or ing Co. (Georgetown, Maine; general manager Starkey
not the studs are welded onto or threaded into the machine. Steuernagle, and seal designer Matt King.
Tubing for
Peristaltic Dosing
Department Editor: Scott Jenkins Pumps

eristaltic pumps work by compress- Advantages and disadvantages of peristaltic pumps
ing a tube against a circular pump
housing with rollers on a rotating Advantages Disadvantages
arm. The fluid that is ahead of the roller • Dosing accuracy is high, and is • F lexible tubing tends to de-
gets pushed forward, while new fluid is not affected by line pressure and grade with time and requires
drawn into the tube by the vacuum gener- fluid viscosity periodic replacement
ated as the tube returns to its relaxed • Maintenance can be minimal • T he flow is pulsed, particularly
state. Peristaltic pumps are a type of due to the absence of valves, at low rotational speeds, so
positive displacement pump that can be seals, pipework, strainers and peristaltic pumps are less
used in industrial chemical dosing ap- so on suitable where a smooth, con-
plications and others, including medical • Contamination is virtually elimi- sistent flow is required
applications. The tubing used to convey nated because the only part of
• N
 ot as effective for con-
the material into and out of the pump the pump in contact with the
tinuous process duties, as op-
mechanism is a critical aspect of pump fluid being pumped is the interior
posed to intermittent duties,
performance. The following are consid- of the tube
because hose and coolant
erations for selecting tubing materials for • Handling slurries, highly viscous, replacements are needed
use with a peristaltic pump. shear-sensitive and aggressive
fluids is possible • L argest sizes are limited to
10–15 gal/min
Advantages and disadvantages • Pump design prevents backflow
As dosing pumps, peristaltic-based sys- and syphoning without valves
tems have a number of advantages, along
with some limitations (Table). Reducing the cally for pump tubing rather than for gen- Thicker-walled tubes generate greater
risk of contamination by pump compo- eral use. Tubing that gets an acceptable suction when they return to their original
nents is a distinct advantage of peristaltic rating for general contact with a given shape after being squeezed, so they are
pumps, but the flow is non-uniform, which chemical might not withstand exposure to generally better for pumping more viscous
can present problems in certain applica- the same chemical when subjected to the fluids. For longer tube life, larger-bore tubes
tions requiring continuous flow. physical stresses of peristaltic pumping. at lower pumping speeds should be used.
When using compatibility charts, end-
Tubing materials users should check the compatibility of Pressure capabilities
Peristaltic pump tubing is a key compo- each component of the solution, rather Peristaltic pump applications are typically
nent, and needs to be selected thought- than just the main ingredient. Even trace limited by the pressure capabilities of the
fully. Major considerations for tubing levels of some acids or solvents can be tubing. Typical pump tubing materials
are chemical compatibility, elastomeric enough to destroy pumps with exposure have working pressure ratings from 10
performance and tube life. over longer periods of time. to 40 psi, with softer materials such as
Tubing for peristaltic pumps needs to Chemical resistance decreases as silicone at the low end and firmer materi-
be constructed of an elastomeric material temperature increases. Chemicals that als at the higher end. Recent material
in order to maintain the circular cross- have no effect on the tubing material at advances are expanding the pressure
sectional shape, even after millions of room temperature could attack the tubing ranges for peristaltic pump applications.
squeeze-cycles inside the pump. Because at elevated temperatures. Pressure sources in a fluid-handling
of this requirement, many non-elastomeric system can vary. Backpressure can be
polymer materials that are effective Immersion test generated by the fluid passing through a
at resisting chemical attack must be If information on chemical compatibility filter or by the fluid pushing through the
eliminated from consideration in these cannot be found, or if a plant’s operating flowmeters or the valves. Backpressure
applications. Materials such as PTFE conditions are significantly different from can also come from the fluid pumping
(polytetrafluoroethylene), polyolefins, those used to determine the chemical-re- into a pressurized reaction vessel.
PVDF (polyvinylidene fluoride) and so on sistance ratings, an immersion test can be Peristaltic pumps deliver fixed amounts
should not be considered as material for performed. In an immersion test, a small of fluid with each pass of a roller over
pump tubing unless they are used as a length of tubing is weighed accurately, the tube, so the size of the tube has a
lining of another tubing material. and its diameter and length measured. direct effect on the amount of fluid deliv-
Popular elastomers for pump tubing The tubing is then immersed in a closed ered. Variations in tubing dimensions can
are silicone, PVC (polyvinyl chloride), vessel containing the chemical in question mean compromised consistency and re-
EPDM (ethylene propylene diene for 48 h. The test piece is then rinsed, peatability, so a tighter tubing-dimension
monomer)+polypropylene (as in Santo- dried, weighed and measured again, tolerance is better.
prene), polyurethane and Neoprene. Of and changes are recorded. The tubing
these materials, the EPDM+polypropylene should also be examined for signs of References
(“-prenes”) have the best fatigue resis- softening or embrittlement, which would 1. Hall, J. Process Pump Control. Chem. Eng.,
tance and a wide range of chemical indicate chemical attack on the tubing. November 2010, p. 30–33.
compatibility. Silicone is popular with 2. Ebelhack, A. Peristaltic Pumps: Matching the
water-based fluids, such as in the biop- Tubing to the Fluid. Cole-Parmer Technical Re-
Tube squeezing source Library, article 576. September 2009.
harma industry, but have limited range of The amount of squeeze applied to the tub- Accessed from, March
chemical compatibility in other industries. ing affects pumping performance and the 2012.
To help select tubing materials, many tube life — more squeezing decreases the 3. Cole-Parmer Metering Pump Selection Guide.
tubing suppliers provide chemical com- tubing life dramatically, while less squeez- Cole-Parmer Technical Resource Library, article
patibility charts, but it is important for ing decreases the pumping efficiency, 681. April 2008. Accessed from www.colepar-
engineers to use a chart designed specifi- especially in high-pressure pumping., March 2012.
Flow Profile for
Department Editor: Scott Jenkins Pumps

eciprocating pumps are often used in
the chemical process industries (CPI) Crank 90 deg
because of their ability to generate high
pressures at low velocities. A subcategory of C
positive-displacement pumps, reciprocating Connecting rod
P Piston Cylinder
pumps act through the recipricating motion
of a piston, plunger or diaphragm. Such
pumps work by way of a connecting-rod- 0 deg
and-crank mechanism with a piston.
By nature, reciprocating pumps generate
S 180 deg
pulsing flow, which, when plotted as a func-
tion of time, or of crank angle, produces a A
curve that resembles a sine wave to a first
approximation. For example, manufacturers
of pulsation dampeners and surge suppres- O
sors often use sinusoidal curves for piston
pumps and compressors in their product
literature and sizing formulas. However, a Piston velocity versus crank angle
closer examination of the flow profile for
a piston-and-crank pump or compressor Crank length = 2
Piston velocity, in./deg X 10

reveals the curve to be a significantly dis-
torted sine wave because of the interaction Connecting
0.5000 rod length =
between the crank and the connecting rod. 2.05 (blue line)
0.4000 4 (red line)
Calculating flowrate 10 (green line)
In graphical form, the crank and crankshaft 0.3000
of a reciprocating pump can be visual-
ized by placing the crankshaft center at 0.2000
the 90-deg mark of a 180-deg x-axis, and
placing the crank bearing at the origin (see 0.1000
figure). A connecting rod links the crank to
the piston. 0.0000
Determining the position of the piston at 0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00 180.00 200.00
any crank angle can be accomplished by Crank angle (rotation), deg
measuring on a piston pump, compressor,
or piston engine, or it can be calculated Observations of the plot and pressure drop will be higher, by the
using trigonometric relationships. In an illustrative example, plots of piston square of flowrate
The degree to which the actual flow pro- velocity versus crank angle are shown (see • The higher flowrates and pressure drops
file curve deviates from the sinusoidal curve graph). The ratios of the connecting rod will affect net positive suction head
is determined by the ratio of the connecting length to crank shaft length are 1.05 to (NPSH) and possibly induce vaporization
rod length to the crankshaft length. Smaller 1 (blue line), 2 to 1 (red line) and 5 to 1 • Maximum crank revolutions per minute will
values of the ratio translate into greater (green line). The following observations be lower than what would be allowed by
levels of distortion. As the connecting rod can be made: the pure (non-distorted) sinusoidal curve
becomes very long, the flow profile would 1. At the beginning of the discharge stroke, • Loads experienced by bearings will
approach the sine curve. flowrate approaches zero asympotically, increase somewhat, especially in high-
To calculate the flowrate at a given crank rather than as a sinusoidal curve speed compressors
angle, use the following procedure and 2. Peak flowrates do not occur at the • Stress analysis of the connecting rods will
definitions: 90-deg point, but rather at 95–120 deg, be affected
Crank length = OC depending on the ratio of rod length to • Surge dampeners must handle the sharper
Piston rod length = CP crank length peak of a bell curve, rather than a
For any angle a, Line AC = OC sin (a) 3. Peak flowrates are higher than would be smoother sine curve
Line SA = OC – OC cos (a) predicted with a pure sine curve • Multi-piston pumps and compressors
Line AP = (CP2 – AC2)0.5 4. From 180 to 360 deg (the suction por- would have less “smoothing” effect than
Line SP = AP + SA tion of the pump cycle), the curve mirrors would be predicted because the bell-
the 0-to-180-deg portion shaped curve has a sharper peak
1. Calculate the piston position for two 5. Flowrates during the suction portion
crank angles, perhaps 2 deg apart. of the curve are also higher and occur References
2. The difference in piston positions equals earlier than the 270-deg point 1. McGuire, J.T., “Pumps for Chemical Process-
piston displacement over the time interval ing,” Marcel Dekkar, New York, 1990.
between the two crank angles. The value Effects of distorted sine curve 2. Henshaw, T.E., “Reciprocating Pumps,” Van
is an average over the span of the two Within the areas of fluid flow and mechani- Nostrand Reinhold Co., New York, 1987.
readings, not an instantaneous read- cal pump design, there are a number of
3. Krugler, A., Piston Pumps and Compressors:
ing. As the step size approaches zero, aspects that are affected by the deviation Exploring the Flow Profile, Self-published,
displacement nears the true velocity. of flow profile from a perfect sine curve for 2010.
3. This value can be converted into flow- pumps and compressors. The effects include
rates (gal/min or other units) if the piston the following: Note. Material for this edition of “Facts at Your
Fingertips” was contributed by Arthur Krugler,
diameter and speed (revolutions per • Check valves and passages will have P.E., Krugler Engineering Group Inc., Whittier,
minute, rpm) are known. higher-than-predicted peak flowrates Calif. (
Cost Indices
Department Editor: Scott Jenkins

ost estimation is a critical dimension CEPCI Historical Trends
of project planning in asset-heavy in- 800
dustries such as the chemical process CE INDEX Equipment Buildings
industries (CPI). Construction cost indices 700 Construction Engineering
are useful tools in cost estimation and labor supervision
are used to compare plant construction
costs from one time period to another.
Understanding them can improve the
accuracy of the cost estimates, as well 500
as the effectiveness with which they are
applied. This column discusses the use of 400
cost indices and historical trends.
Indices for the CPI
Cost indices are dimensionless numbers
that compare prices of a class of goods 200
or services to the corresponding prices
at a base period. They are widely used 100
in the construction industry and can be
customized to various industry segments.
Predesign estimates are usually made
1/1/ 75
1/1/ 76
1/1/ 77
1/1/ 78
1/1/ 79
1/1/ 80
1/1/ 81
1/1/ 82
1/1/ 83
1/1/ 84
1/1/ 85
1/1/ 86
1/1/ 87
1/1/ 88
1/1/ 89
1/1/ 90
1/1/ 91
1/1/ 92
1/1/ 93
1/1/ 94
1/1/ 95
1/1/ 96
1/1/ 97
1/1/ 98
1/1/ 99
1/1/ 00
1/1/ 01
1/1/ 02
1/1/ 03
1/1/ 04
1/1/ 05
1/1/ 06
1/1/ 07
1/1/ 08
1/1/ 09
1/1/ 10
1/1/ 11
1/1/ 12
1/1/ 13
1/1/ 14
for equipment and assets that will be built
in the future, but must be assembled from
FIGURE 1. The CEPCI, along with the four sub-indices that contribute to it, has risen unevenly
prices of the past. over the past three decades
The mathematical relationship between
costs and indices is the following: the composite CEPCI index. Annual values sector. Location factors also typically do
for the overall CEPCI and for each of the not consider cost effects that are associated
(Cost at Time 1) ÷ (Cost at Time 2) =
sub-indices are calculated from the arithme- with site-specific conditions, such as local
(Index at Time 2) ÷ (Index at Time 1)
tic mean of the monthly values for that year, climate, earthquake risk and other geologi-
A number of cost indices are relevant to the for each subindex. cal differences. Countries may not have the
CPI, including the Chemical Engineering The overall CEPCI for 1959 was assigned capability to manufacture certain special-
Plant Cost Index (CEPCI) and the Marshall a value of 100 as a benchmark. Figure 1 ized equipment and need to import it, so
& Swift index ( for illustrates the historical escalation of the the degree of importing required needs to
the chemical industry and the Nelson-Farrar CEPCI and its sub-components for the past be part of the calculation.
index (, which is designed 37 years. The CEPCI value, along with those When used, location factors should be
for petroleum refineries. for the sub-indices, has risen unevenly, but reserved for preliminary project evaluations
steadily for the past three decades. — they are not intended to be used when
CEPCI preparing appropriation-quality estimates.
The CEPCI is a composite index, and is Location factors Resources for more information on location
built from four sub-indices: 1) Equipment; Cost indices are traditionally tied to a par- factors include AACE International and
2) Construction Labor; 3) Buildings; and 4) ticular geographic region. As industries and the International Cost Engineering Council
Engineering & Supervision. The Equipment markets globalize, the ability to translate (Sydney, Australia;
subindex is further broken down into seven costs has become increasingly important.
component indices, as follows: The cost engineering professional organiza- References
• Heat exchangers and tanks tion AACE International (formerly Assoc. 1. Ulrich, G.D. and Vasudevan, P.T., Capital
• Process machinery for Advancement of Cost Engineering; Mor- Costs Quickly Calculated, Chem. Eng., April 1,
• Pipe, valves and fittings gantown, W.Va.; defines 2009, pp. 46–52.
• Process instruments a location factor as an estimating factor 2. Humphreys, Kenneth K., and Hamilton, Allen
• Pumps and compressors used to convert the cost of an identical C. Worldwide Sources of Cost and Project In-
• Electrical equipment plant from one location to another. Location formation. Conference Proceedings, Joint Cost
• Structural supports and miscellaneous Management Societies, Morgantown,
factors recognize differences in labor costs,
W. Va., 2000.
The component indices are compiled with engineered equipment and material freight,
3. Vatavuk, William M., Updating the CE Plant Cost
weighting factors and combined to make duty, taxes, engineering, design and project Index, Chem. Eng., January 2002, pp. 62–70.
up the equipment index. The three other administration costs. The cost of land is not
4. Pietlock, Bernard A., Developing Foreign Loca-
sub-indices are compiled independently and included in location factors.
tion Factors, Cost Engineering 34(1), January
also weighted and normalized. Accurately and effectively applying
1992, pp. 7–11.
The CEPCI includes the costs to design, location factors can be challenging and
5. Pietlock, Bernard A., Developing Location Fac-
purchase and install chemical plant equip- doing so requires an understanding of
tors by Factoring, AACE International Recom-
ment, and is weighted as follows: how each factor was derived and exactly
mended Practice No. 28R-03, October 2006.
• 61% equipment, machinery, and supports what it represents. In most cases, location
• 22% construction labor factors are not industry-specific, but rather Editor’s note: The editor would like to acknowledge
• 7% buildings are designed for a broader mix of facility AACE International for allowing access to its virtual
library, and EC Harris LLP (London, U.K.; www.
• 10% engineering and supervision types, so they may not reflect the unique; an Arcadis company), for insight
Combining the four sub-indices gives rise to characteristics of buildings in a particular related to this column.
Department Editor: Kate Torzewski

acuum is any system of reduced Figure 1. Liquid-Ring Pump Figure 2. Rotary-Claw Pump Figure 3.
pressure, relative to local Rotary-
(typically atmospheric) pressure. Lobe
Achieved with a pump, vacuum Pump
systems are commonly used to:
Source: Kurt
• Remove excess air and J. Lesker Co.
its constituents
• Remove excess reactants or Figure 4. Rotary-Screw Pump
unwanted byproducts
• Reduce the boiling point
• Dry solute material
• Create a pressure differential for
initiating transport of material
Liquid-ring and dry pumps offer the
most advantages for the chemical
process industries (CPI). Both of
Source: Medical Gas Info Source: Gardner Denver Source: Kurt J. Lesker Co.
these pump types have bearings
Hanover, Inc.
sealed off from the pumping cham-
ber and do not require any internal toward the discharge side of the void space between the rotors and Rotary Screw. Two long helical
lubrication because the rotors do pump, the pockets decrease in pump housing. On the next rotation, rotors in parallel rotate in opposite
not contact the housing. Both, when size, and the evacuated gas is com- that same trapped sample of gas is directions without touching, syn-
employing a coolant system, prevent pressed, enabling its discharge. compressed and discharged as the chronized by helical timing gears
the coolant from contacting the pro- The ring of liquid not only acts as discharge port opens. (Figure 4). Gas flow moves axially
cess fluid and causing contamina- a seal; it also absorbs the heat of A minimum of three stages in along the screw without any internal
tion, and both use mechanical shaft compression, friction and condensa- series is required to achieve pres- compression from suction to dis-
seals for containment. tion. Popular liquid choices include sures comparable to those of an charge. Pockets of gas are trapped
water, ethylene glycol, mineral oil oil-sealed mechanical pump. Some within the convolutions of the rotors
and organic solvents. dry designs use two technologies in and the casing, and transported to
liquid-ring pumps combination; for example, a rotarly the discharge. Compression occurs
dry pumps lobe as a booster for a claw pump. at the discharge port, where the
In the cylindrical body of the pump, trapped gas must be discharged
a sealant fluid under centrifugal Rotary-claw, rotary-lobe and Rotary Lobe. The rotary-lobe pump
against atmospheric pressure. Each
force forms a ring against the inside rotary-screw pumps dominate as (Figure 3) is typically used as a me-
convolution of the rotor acts similarly
of the casing (Figure 1). dry pumps in the CPI, particularly in chanical booster operating in series
to a stage in series with the one be-
The source of that force is a larger-size pump applications. with an oil-sealed piston or vane
hind it; at least three convoluted gas
multi-bladed impeller whose shaft Rotary Claw. The geometric shape pump to boost pumping capacity at
pockets in the rotor are required to
is mounted so as to be eccentric to of this pump allows for a greater low pressures. achieve acceptable vacuum levels.
the ring of liquid. Because of this compression ratio to be taken This pump consists of two sym-
eccentricity, the pockets bounded across the rotors at higher pressures metrical two-lobe rotors mounted on
by adjacent impeller blades (also (Figure 2). Two claw rotors rotate separate shafts in parallel, which References
called buckets) and the ring increase in opposite directions of rotation rotate in opposite directions to each 1. Vilbert, P., Mechanical Pumps for
Vacuum Processing, Chem. Eng.
in size on the inlet side of the pump, without touching, using timing gears other at high speeds.Timing gears
October 2004, pp. 44–51.
and the resulting suction continually to synchronize the rotation.The gas are used to synchronize the rotation 2. Aliasso, J., Choose the Right
draws gas out of the vessel being enters through an inlet port after it of the lobes to provide constant Vacuum Pump, Chem. Eng. March
evacuated. As the blades rotate has been uncovered and fills the clearance between the two. 1999, pp. 96–100.

Advantages Disadvantages
Simpler design; employs only one rotating assembly •M  ixing of the evacuated gas with
Liquid-Ring Vacuum Pumps

• Can be fabricated from any castable metal the sealing fluid

• Minimal noise and vibration •R  isk of cavitation requires a portion
• Little increase in the temperature of the discharged gas of process load to be noncondens-
• No damage from liquid or small particulates in the process fluid able under operating conditions
• Maintenance and rebuilding are simple •H  igh power requirement to form and
• Slow rotational speed (1,800 rpm or less), maximizing operating life maintain the liquid ring, resulting in
• Can use any type of liquid for the sealant fluid in situations where min- large motors
gling with the process vapor is permissible • Achievable vacuum is limited by the
• No lubricating liquid in the vacuum chamber to be contaminated vapor pressure of sealant fluid at
• Accommodation of both condensable vapors and noncondensables, the operating temperature
while operating as both a vacuum pump and condenser •P  ower consumption
•R  ugged rotor design, constructed of sturdy cast or ductile iron without •C annot handle particulate matter,
any flimsy rotating components nor large slugs of liquid
Dry Vacuum Pumps

• Noncontact design facilitated by timing gears •M ay require a silencer

• High rotational speed reduces the ratio of gas slip to displacement, •M ay discharge gases at high
increases net pumping capacity and reduces ultimate pressure temperatures
• Multiple staging provides inlet pressures below 1-mm Hg absolute while • Most difficult to repair or rebuild
discharging to atmosphere •M ay require a gas purge for cooling,
• No contamination of evacuated gas or to protect the bearings and seals
• Due to lack of condensation, pump can be fabricated of standard, from the process gas
inexpensive cast iron •D ue to high operating temperatures,
some process gases may polymerize
Feature Report
Engineering Practice

Optimizing Reciprocating
Compressors for CPI Plants
Follow this guidance to Connecting rod Piston rod Piston

improve the design, performance Crankshaft

and reliability of these widely Crosshead Distance Piston rod Cylinder
used machines piece

Figure 1. Reciprocating compressors compress air or other

gases using a piston that is driven by a crankshaft. Shown here
are the main parts of a reciprocating compressor
Amin Almasi
WorleyParsons Services Pty. Ltd.

eciprocating compressors — tor of three by adding the proper coat- may consider installing additional
the most commonly used type ing (tungsten carbide is a widely used clearance pocket(s) on the first-stage
of compressor throughout the coating material for piston rods). cylinder(s) and using part-load opera-
chemical process industries Interstage cooling is required when tion via the compressor control logic.
(CPI) — are flexible and efficient, the machine or gas being compressed By selecting the right interstage de-
and they can generate high head has a temperature limit. In this case, sign pressure, users can ensure proper
(from several bar to several thou- as the gas cools, any liquid that may operation in the face of part-load op-
sand bar) independent of gas density. form is separated in interstage fa- eration and variation in suction pres-
Worldwide, the installed recipro- cilities and then the gas is returned sure. In general, the interstage de-
cating compressor horsepower is to next compressor stage for further sign pressures should be around 15%
approximately two times that of cen- compression. Each compressor stage higher than the interstage basic de-
trifugal compressors. may consist one or more cylinders. sign values for applications that are
However, the maintenance costs as- Vendors usually offer a range of in- working with common part-load steps
sociated with reciprocating compres- terstage pressures. The ability to opti- (such as 25%, 50%, 75% and 100%
sors are approximately three times mize interstage pressures can help to capacity) and are expected to experi-
greater than those for centrifugal com- minimize the total cost of ownership ence a variation in suction pressure of
pressors (due to valve, unloader and for the compressor and interstage fa- around +/- 7% during operation.
packing-maintenance requirements). cilities. This optimization can be done In some applications, reciprocating
This article provides practical recom- by evaluating the initial cost and compressors must be designed to op-
mendations for users to consider in an operating costs of compressors and erate reliably in the face of consider-
effort to improve the selection, opera- interstage facilities for various inter- able suction pressure variations while
tion and maintenance of reciprocating stage pressures. still providing full design flow at the
compressors in CPI applications. During operation, interstage pres- desired discharge pressure. These
sures will increase during part-load operating requirements will have a
Compressor design operation (that is, operation at lower direct impact on compressor sizing,
Figure 1 shows the basic design of a flow that results when an unloader especially the frame rating and motor
reciprocating compressor. Close at- device is used; this is discussed below) power required for the unit.
tention to the selection of the piston combined with variation in pressure Figure 2 shows load curves for the
rod packing can improve performance, at the suction inlet. connecting rod of a reciprocating com-
because this is a common source of In a typical reciprocating-compres- pressor in petroleum refining service.
reliability problems associated with sor design, the first stage may contain Variation in suction pressure (in this
reciprocating compressors, and is a one or more cylinders and a clearance case, a roughly 7% reduction in suc-
common path for the leakage of po- pocket. An additional bottle may be tion pressure) results in a higher load
tentially hazardous process gases. added to the cylinder with an actu- on the rod. As a general rule, the com-
Experience shows that packing life ated on/off valve. To avoid unwanted pressor should be designed so that
can be extended by as much as a fac- interstage pressure increases, users the maximum-anticipated rod load
Chemical Engineering December 2010 39
Engineering Practice 0.8

Calculated rod load, kN / allowable rod load, kN

100% Capacity
75% Capacity
does not exceed 80% of the allowable 50% Capacity
rod load. 93% Suction pressure
As shown on the y-axis in Figure 2,
the rod load shall change sign from
negative to positive and then nega-
tive again during one revolution of 0
the crankshaft in order to provide
proper lubrication for the mechanism -0.2
(especially for the cross-head pin). The
duration of rod sign reversal (the pe- -0.4
riod during which load has the oppo-
site sign) should not be less than 15 -0.6
degrees of crank angle. The rod-load
reversal peak (maximum amount of -0.8
0 50 100 150 200 250 300 350
load in the reversed direction) should
not be less than 3% of the actual com- One rotation of the crankshaft angle, deg.
bined load in the opposite direction.
These minimum requirements should FIGURE 2. These rod-load curves are for a reciprocating compressor in refinery
service. This figure shows variation in the ratio of rod load to allowable rod load in
be satisfied under all possible operat- one revolution of the compressor crankshaft (from 0 to 360 deg.). Individual curves
ing conditions (especially in the face of show the rod load for various operating conditions, including 50%, 75% and 100%
suction-pressure variation and part- compressor capacity, as well as full flow with 7% suction reduction (that is, operation
load operation, such as when an un- at 93% of rated suction pressure)
loader device is used to decrease flow
through the compressor). resent compressor torque for normal nance events. For large compressors
In many cases, higher values of full-load capacity and half-load (50%) (that is, those that operate at rela-
rod-reversal duration and peak are capacity, respectively. tively low speeds with high pressure
considered during compressor design A step-less capacity-control system ratios), relatively large-bore ring-
to increase reliability. In Figure 2, uses a hydraulically actuated, finger- type valves (above 100 mm, or 4 in.)
minimum load-reversal duration cor- type unloader. This device unloads the combined with plug-type unloaders
responds to 50% capacity, and the re- suction valve for only a portion of com- should be considered first, to avoid re-
versal duration is more than 70 deg. pression cycle to achieve the desired liability issues associated with finger-
In general, the optimum speed for adjusted capacity. type unloaders. Since ring-type valves
the reliable operation of reciprocat- A finger-type unloader has finger- and plug unloaders are not available
ing compressors is around 350 rpm. shaped parts that act on the cylinder for smaller-sized compressors (those
For compressors operating below 400 valve elements and keep them open that operate at relatively higher
kW, speed on the order of 450 rpm is for a defined duration during the com- speeds), such units typically use plate-
suitable. However, for those operating pression cycle. Users should note that type valves.
below 100 kW, higher speeds (even as these finger-type unloaders have the During operation, the rotating parts
high as 700 rpm) are acceptable. potential to damage the valve-sealing of the compressor, power transmis-
Lubricated cylinders and packing elements and thus may have greater sion and driver will act like springs
may be preferred to extend service maintenance requirements. connected in series. This torsional dy-
life. However, non-lubricated cylin- A step-less capacity control system namic system may create resonance
der compressors should be used when is recommended for larger machines (where one natural frequency of sys-
the possibility of oil contamination (units rated above 2 MW, when large tem coincides with one of excitation
cannot be tolerated in downstream operation variation is expected). In torque). In reciprocating compressor
operations (for instance, when trace these cases, step-less capacity control trains, there is always a risk of tor-
amounts of lubricating oils could (working in the range of 20–100% ca- sional resonance and torsional fatigue
cause catalyst problems in down- pacity) is extensively used due to pro- failure (that is, damage to component
stream reactors). cess requirements. resulting from excessive cyclic loads).
For the optimum operation of re- In general, valves and unloaders are Couplings that connect the driver
ciprocating compressors, sufficient in- responsible for nearly half (roughly to the compressor can be modified to
ertia — provided by a properly sized 45%) of unscheduled reciprocating- tune the system to avoid torsional res-
flywheel — is mandatory to regulate compressor shutdowns, so valve and onance. Several coupling options are
the variable reciprocating torque. unloader selection can have a strong available as follows:
Figure 3 shows brake torque versus impact on reliability. And many con- 1.  Direct, forged-flange rigid connec-
crank angle for one revolution of the sider the automatic cylinder valves tion (no coupling) between driver
crankshaft for a reciprocating com- to be the most critical components of and compressor
pressor used in a petroleum refinery such machines, as they are respon- 2.  High-torsional-stiffness coupling is
setting. The red and blue curves rep- sible for many unscheduled mainte- allowed by torsional analysis. Since
40 Chemical Engineering December 2010
coupling options are limited, it may compressors (around or above 3 manual barring device can be used for
not be possible to find an acceptable MW) relatively small compressors. A pneu-
coupling with the required torsional • High compressor main, and motor matic barring device must be used for
characteristics and service factor, bearing, temperatures (alarm) compressors rated above 750 kW (pro-
especially for large compressors • Valve temperature (monitoring) vided there is no area classification or
(above 3 MW) • Oil temperature, out of compressor power-availability problem).
3.  Flexible coupling (which provides frame (alarm) For larger compressors (2 MW or
more elasticity and damping, but • High jacket-water temperature of larger), these special tools are often
may require greater maintenance each cylinder (alarm) needed to carry out routine mainte-
since elastic elements in such a cou- In addition, proximity probes, typically nance on reciprocating compressors.
pling may need frequent replace- located under the piston rods, provide These tools cannot be easily pur-
ment) alarm capabilities but are not used for chased; they must be specially de-
The most common reasons for prob- shutdown. These are used to measure signed and fabricated based on the
lems caused by torsional vibration the rod position and determine wear actual machine:
are lack of comprehensive torsional- or malfunctions. Such probes can • Bearing extractor
vibration analysis, improper applica- quickly identify problems such as pis- • Piston extractor
tion and maintenance of couplings ton or rider band malfunctions, cracks • Valve extractor
(especially flexible ones) and lack of in the piston rod attachment, a broken • Piston fit-up tool
appropriate monitoring. As a general crosshead shoe or even a liquid carry- • Hydraulic tightening system
rule-of-thumb, the shaft diameter of over to a cylinder. • Crosshead assembling tool
the electric motor should be equal • Special lifting tools
to or greater than the reciprocating Improving maintenance • Partition plate-assembling tools
crankshaft diameter (because the To support regular maintenance, the • Mandrels for wear bands
crankcase is generally forged from a installation of any reciprocating com- During maintenance of compressor
stronger steel grade compared to the pressor must ensure proper access to mechanical components, the following
motor rotor). the entire compressor system, espe- criteria are important:
cially the non-drive end. In particu- • Cylinder clearance for the outboard
Condition monitoring lar, adequate space and work areas end should be around 4–6 mm
Condition monitoring, when done must be provided to enable the com- (0.2–0.3 in.), and for the inboard
properly, can pay for itself by helping plete withdrawal of the piston, re- end, clearance should be around 2–4
operators identify potential systems moval of the cooler bundles or piping mm (0.1–0.2 in.)
malfunctions at an early stage. A rig- spool and laydown area (to carry out • The allowable temperature of the
orous program should include moni- maintenance, dismantling of parts machine bearings, piston rod, con-
toring of these important conditions: and repairs). necting rod bearing and crosshead
Vibration (including continuous Similarly, three crane capacities should be maintained around 85ºC,
vibration monitoring of the com- must be properly identified: The total and for the crosshead pin, it should
pressor and motor casing, provid- capacity of the overhead crane (to lift be maintained around 90ºC
ing both alarm and shutdown ca- components for routine maintenance), • The vibration level of the crank-
pabilities): the maximum maintenance weight (to case should not exceed 100 microns,
• In general, velocity transducers ensure that the heaviest parts, usu- and the expected vibration level of
are preferred over accelerometers ally the motor, can be lifted during the cylinders should be around 150
(because interested frequencies for overhauls), and the maximum instal- microns (these vibration recommen-
monitoring better match with ve- lation weight (maximum skid weight, dations are peak-to-peak vibration
locity-measurement sensors). The usually the compressor skid). readings for an installed, trouble-
optimum configuration for using a For a typical 7-MW API 618 com- free, middle-range machine around
velocity transducer is to install one pressor train for petroleum-refinery 1 MW)
on each end of the crankcase, about service, these crane capacities would • Bearings, piston rings and piston
halfway up from the base plate in be roughly 11 tons, 55 tons and 100 shoes should also be inspected regu-
line with a main bearing, both for tons, respectively, and the required larly.
compressor and motor crane height would be roughly 12 m
• Crosshead accelerometer (alarm) (around 40 ft) Auxiliaries and accessories
Temperature: Any time a given compressor must For auxiliaries and accessories, the
• High gas-discharge temperature for be stopped for an extended time, it optimum configuration is to install a
each cylinder (with both alarm and should be turned a quarter-turn every local panel near the compressor skid
shutdown capabilities) week, using a barring device (this is a (around 250 mm, or one foot away
• Pressure packing piston-rod tem- device that slowly turns the compres- from the compressor skid), and on a
perature (alarm) sor to avoid locking and other prob- standalone skid to minimize the po-
• High crosshead pin temperature lems that often arise during long stop- tential for vibration damage.
(alarm), only for relatively large pages of reciprocating compressors). A The oil system should include two
Chemical Engineering December 2010 41
Engineering Practice Full capacity

1.5 50% Capacity

Ratio of compressor torque, kNm /

oil pumps, both sized for a capacity

electric motor torque, kNm

that is 20% greater than the maximum
oil flow required for the compressor. 1
At a minimum, two pumps should be
used. Either a run-down tank (this is
a stainless-steel tank that allows the
supply oil to safely coast down the
machine in the event that both pumps 0.5
have failed), or a crank-shaft-driven
main oil pump is required. Dual (two)
removable bundle shell-and-tube oil
coolers (TEMA C), double oil filters 0
with a removable element and stain- 0 50 100 150 200 250 300 350
less-steel piping are also necessary. One rotation of the crankshaft angle, degrees
Liquids should not be allowed to ac-
cumulate inside the compressor cyl- FIGURE 3. This plot of reciprocating compressor torque versus crank angle shows
inder. For any application, a suction the variation in compressor torque represented by the ratio of compressor torque/
drum (sized appropriately with regard motor torque, in one revolution of compressor crankshaft (0 to 360 deg.). The
curves show compressor torque in two operation conditions — 50% capacity and
to application-specific retention time, 100% (full) capacity
flow velocity, and, if required, a mist-
collection system to capture contami- • There should be a continuous rise around 85–95% of API 618 (Approach
nants) with a drain provision should from a selected operating point to 3) limits to provide some margin
be provided. It may be part of pulsa- shut off (5–15% of API 618 limit values) to
tion control. To control pulsation, a • There should be proper shut-off mitigate risk during construction and
vertical vessel is sometimes used as pressure compared to a selected op- installation periods, and to cope with
both suction drum and suction pulsa- erating point (preferably 10% but a unanticipated deviations and prob-
tion vessel, but this not recommended minimum of 6%) lems. Similarly, pulsation vessels are
since vertical pulsation vessels cause Note: usually a larger size pump generally fabricated before finaliza-
relatively long piping to the compres- shall be selected to meet these recom- tion of the piping-design-and-pulsa-
sor, which may lead to dynamic prob- mendations. tion study, and enough margin should
lems. Similarly, these are sometimes Any reciprocating compressor sys- be provided to meet potential risks.
conflicting requirements for pulsation tem should be designed with a margin For nearly all applications, horizon-
damping and suction separation, so of excess flow capacity for the cooling tal suction and horizontal discharge
this combined approach is not pre- system, to enable it to respond to situ- vessels are preferred. Long distances
ferred. It may be used only for small ations that deviate from normal oper- between vertical pulsation vessels and
compressors, let’s assume below 250 ation, where the need may arise later compressors increase the likelihood of
kW, with relatively light gases, such for additional cooling flow to remove pulsation problems.
as those lighter than nitrogen. excess generated heat (for example,
Cooling-water systems are gener- unloaded operation when the com- Improving performance
ally used as a heat sink for recipro- pressor is idle, overload conditions, or The maximum predicted discharge
cating compressors, to avoid hot spots future expansion, if applicable). The temperature for any API 618 recip-
and improve machine stability and recommended cooling pump capac- rocating compressor for CPI applica-
reliability. To design a cooling sys- ity margin is 10–25% (that is, pump tions must not exceed 150ºC, and must
tem, first, the generated heat should rated capacity is 10–25% more than not exceed 135ºC for hydrogen-rich
be calculated. Then, the anticipated required normal flow). service. In general, gas discharge tem-
temperature rise should be identified. Users should consider suitably sized peratures below 118ºC tend to lead to
The cooling-water inlet temperature pulsation vessels and correct any po- longer life for the wearing parts.
should be selected between 6°C and tential pulsation resonance in piping When it comes to optimum pressure-
16ºC above the inlet gas temperature. rather than using damping devices, drop values for pulsation dampeners
When selecting the pump to deliver such as orifices, choke tube and so on and suppression devices, the pres-
cooling water to the compressor cylin- to dampen pulsation. Acoustic reviews sure drop maximum is 1% of absolute
der and packings, consider this: should be performed during compres- pressure. For the intercooler, pressure
• With regard to the slope of the op- sor system design to guarantee all an- drop around 0.70 bar or 2% of absolute
erating curve (discussed in greater ticipated combinations of pressures, pressure is recommended.
detail below), the selected oper- speeds and load steps (including the Readers should note that the use
ating point should not be in the use of flow-reduction steps that rely of orifice plates to dampen pulsation,
flat or near-flat part of curve; on unloaders, which can vary the com- especially on high-speed, single-act
rather, enough slope is needed for pressor flow). compressors (that is, those that com-
proper operation Pulsation limits are recommended press gas on only one head of cylin-
42 Chemical Engineering December 2010
der), can contribute to significant tive when suction-pressure variations inders. Assembled cylinders are typi-
pressure drops. are limited). A review of the steepness cally delivered to the site separately
To gain a better understanding of of the proposed load curves can help and installed later. It is common for
reciprocating compressor performance the engineer to quickly identify which the vendor to provide site-supervision
and track ongoing operation, the fol- load curves (and where) are too steep. work for cylinder installation at a ne-
lowing performance curves should In these situations, small changes in gotiated lump-sum price. n
be developed: pressure can have significant changes Edited by Suzanne Shelley
• Suction pressure versus load in load and flow. In general, compres-
• Suction pressure versus flow sors with steep load curves are hard to Author
• Discharge pressure versus load automate and tune. Thus, steep load Amin Almasi is lead rotat-
• Discharge pressure versus flow curves usually indicate improper siz- ing equipment engineer for
WorleyParsons Services Pty
• Suction pressure versus discharge ing of cylinders. Ltd (60 Albert St.; Brisbane
pressure, per load step (that is, for QLD 4000 Australia; Phone:
+61 (0)7 3319 3902; Email:
each flow-reduction step using un- Optimum conditions amin.almasi@worleyparsons.
loaders, typically 50%, 75%, 100% When scoping a reciprocating com- com). He holds a chartered
engineer certificate from IM-
flow; 25% is rarely used because of pressor system, it is absolutely nec- echE (CEng MIMechE) and a
chartered professional engi-
the potential for reliability and load- essary to have a minimum of two neers license from Engineers
reversal issues) technically accepted proposals from Australia (MIEAust CPEng—Mechanical). He
also holds M.Sc. and B.Sc. degrees in mechani-
Such flow curves typically plot the qualified vendors. Small and medium cal engineering. Almasi specializes in rotating
minimum achievable flowrate to the compressors should be delivered fully machines including reciprocating, centrifugal
and screw compressors, gas and steam turbines,
maximum achievable flowrate in spec- fabricated as one skid-mounted pack- process pumps, condition monitoring and reli-
ified increment steps (for instance, in age. Larger compressors are typically ability. He is an active member of IMechE, En-
gineers Australia, SPE, IEEE, ASME, CMVI,
10% steps). (Flow-versus-discharge- delivered as a prefabricated system the Vibration Institute, SMRP and IDGTE, and
has authored several papers and articles dealing
pressure plots of specific suction pres- (including the crankcase, distance with rotating machines, condition monitoring,
sures may be an acceptable alterna- pieces, and so on) with dismantled cyl- offshore operations and reliability.

The Chemical Engineering bookstore offers a

variety of industry topics you will come to rely on.


For a complete list of products, visit the Chemical

Engineering bookstore now.

Chemical Engineering December 2010 43

Feature Report
Engineering Practice

Gear Units David Brown

In CPI Plants FIGURE 1. In this conventional, double-

Follow this guidance to improve the selection helical gear unit (with a welded-steel
casing), double-helical gears are used to

and operation of gear units in CPI plants minimize the axial load (that is, the load
in the direction of the shaft). The complex
casing was fabricated by welding
Amin Almasi and the driven equipment, so the ferent mechanical loads, the shafts
Rotating Equipment Consultant gear unit(s) often receive virtually could displace in the plane perpen-
no attention. However, the client, dicular to the axial direction of each

ear units play a critical role the vendor, and the gear-unit sub- shaft (known as radial shifts). If
in a diverse array of chemical vendor should all pay considerable relatively large radial shifts occur
process industries (CPI) fa- attention to the following aspects of between each equipment shaft
cilities, to match the speed of gear units: and the gear unit shafts, a double-
a selected driver with the speed of • Gear-set design, such as gear jointed coupling is needed, to allow
the driven equipment. Today, they teeth details for some movement between shafts.
are used extensively in pumps, tur- • Gear-unit loading (loads on the This depends on thermal expan-
bines and compressors, in material- gears and shafts) and analysis sions, mechanical displacements
handling units (such as conveyers • Material selection and other design and operation de-
and stackers), and in other equip- • Heat-treatment options cisions.
ment, including mixers, kneaders • Fabrication and assembly Option 2. In the second arrange-
and extruders. • Gear-unit lubrication ment, the driver is attached directly
This article discusses high-power, • Quality and testing to the gear-unit housing. In this sce-
high-speed gear units for large • Commissioning nario, because the driver and the
rotating machines and material- A proper helix angle (the angle be- gear unit both display sufficiently
handling systems. These gear units tween the axis of a helical gear and low axial and radial displacements,
can be used in the power range of an imaginery line tangent to the the coupling (or shaft extension) be-
0.1–100 MW, with gear tip speeds gear tooth) — in the range of 10–30 tween the driver shaft and the gear-
usually between 20–200 m/s. The deg — should be used for the gear unit input shaft could be omitted.
American Gear Manufacturers teeth. This will provide suitable The driven equipment is mounted
Assn.’s (AGMA) “Specification for operation and will considerably re- separately and connected to the
High Speed Helical Gear Units duce the axial forces. gear unit with a coupling. This cou-
(6011)” covers different aspects of Within that range, a relatively pling is necessary because the dif-
high-speed gear units, such as the large helix angle (in the range of, ferent heat-induced expansions of
gear rating, lubrication, vibration, say, 22–30 deg, compared to 10–18 the gear unit and the driven equip-
testing and others. deg), can result in a high contact ment can cause relative shifts in
Large, high-speed gear units for ratio, which could offer better per- the shaft positions.
CPI applications are usually com- formance, smoother operation and Option 3. In the third design, the
prised of case-hardened, double- lower noise and vibration. The con- gear unit is directly connected to
helical gear-teeth in welded-steel tact ratio is the ratio of face advance the driven equipment, in anticipa-
casings (Figure 1). The gear shafts to the circular pitch. This ratio de- tion that no radial or angle shifts
are supported by modern bear- fines how many teeth are in contact can occur. In this scenario, no cou-
ings (typically plain bearings for at all times (on average). pling joint is required between the
low speeds). Ideally, bearing forces gear unit and the driven equip-
should be symmetrical with no tilt- Gear-unit arrangement ment. Figure 2 shows an example of
ing moment, to ensure minimum Several ways of integrating a gear a complex gas turbine coupled to a
axial load (the force in the direction unit into a CPI machinery train are gear unit (for the speed match).
of the shaft). discussed below:
Option 1. In this option, the driver, Planetary gear units
Practical notes on gear units the gear unit and the driven equip- The most commonly used special-
As is often the case, engineers tend ment are all installed separately. purpose gear-unit designs include
to focus their attention on the driver Due to thermal expansions or dif- different models of double-helical
Chemical Engineering December 2013 61

Engineering Practice

gear units. However, planetary gear gear type offers one of the
units can often be employed in high- most compact options.
speed CPI machinery (typically in An important loss in a
the 0.2–30 MW range). These are high-speed gear unit is
popular both for rotating machines windage loss, which rep-
and material-handling units, par- resents the power lost be-
ticularly where large speed ratios cause of the compression
are necessary. of the air-lubricant mix- FIGURE 2. Gas turbines typically have a defined op-
While double-helical gear units ture around teeth roots erating speed range. When used in a machinery train
are simple and popular, planetary during meshing and the (with a generator, compressor or pump train), a well-
gear units offer smaller size and aerodynamic trail of the designed gear unit is used to match the speed of the
turbine shaft to the driven equipment
higher efficiency compared to other teeth in the air-lubricant
types of gear units. Because the mixture. With any planetary-gear planetary gears (for example, in the
transmitted power is divided over type, it is necessary to take into ac- form of flex pins). These will bend
several tooth meshes, planetary count a relatively high windage. under a load, with the result that
gear units are more compact and The best operation and perfor- the planet gears do not skew but are
less expensive than other options. mance occurs when the load is simply minimally displaced parallel
The smaller gear diameters in a distributed as evenly as possible to the pinion or ring-gear axis. This
planetary gear unit also produce across the individual planet gears. solution can ensure an optimum
smaller mass moments of iner- An uneven load balance can create load balance between the meshed
tia, and this substantially reduces damaging effects. teeth and the even load distribution
the acceleration and deceleration The ability to reduce friction can, across the entire width of the tooth
torque during acceleration and theoretically, result in a better-cen- face at both full and partial load.
braking. Generally, high speed ra- tered gear-tooth loading. However,
tios can be achieved (even 80 or in real-life scenarios, some friction Selection criteria
more). The coaxial design permits (and thus some unbalances in the Key factors influencing the choice
a superior arrangement for many loading) should be expected. The of gear units are load capacity, effi-
CPI machineries. However, due to mass inertia of the gear-unit ele- ciency, and successful references (for
operational issues and reliability ments can also offer some dynamic comparable applications). In gen-
risks, planetary gear units will re- loads. It is desirable to reduce the eral, several types of gear types are
quire extra consideration. masses, but there are always practi- available for planetary gear-units,
Figure 3 shows an example of cal limits in the mass-reduction ex- including spur gears, single-helical
a planetary gear unit. They can ercises (some limits to achieve very gears, double-helical gears and oth-
achieve a large speed ratio (even 40 lightweight designs) ers. If single helical gears are used,
to 80) in a very compact unit. The partial-load operations and the two opposing axial forces acting
A planetary gear unit typically particularly the minimum-load on the planetary gear may generate
consists of three coaxially rotating case, can offer some challenges. The a large tilting moment (axial bear-
components: a sun gear, a carrier no-load case occurs when the gear ings could be required if the helix
with several planetary gears, and unit and the associated machinery angle is not selected properly).
a ring gear. Among the variants of train are operating with no load — The double-helical gear set usu-
planetary gear units, the options for example, when a pump unit oper- ally requires free axial adjustabil-
preferred for CPI applications are: ates in the pre-commissioning stage ity to allow users to achieve an even
• The star-gear type, with speed without actually pumping a liquid, load distribution across the two
ratios of around 2–12 and thus, the driven gear unit just tooth halves. However, the two gear
• The planetary-gear type, with rotates with no load transmission. meshes can prevent the necessary
speed ratios of roughly 3–13 Some machinery trains should only movement (see Figure 1). In this
• The star-gear type with the dou- operate a very short time in a no- way, two side-by-side gear meshes
ble planet, with speed ratios of load condition, otherwise damage or do not allow for the tiny move-
around 12–40 reliability issues may arise. ments required for adjustments
• The compound-epicyclic gear type, Usually, most gear units have and smooth operation. In addition,
with speed ratios of about 8–80 several backlash-prone joints and external axial forces may interfere
The star-gear type and the plane- relatively large masses that are un- with the load distribution.
tary-gear type can usually offer a able to center themselves when the Spur gears (or straight-cut gears)
similar ratio range (both can cover load changes, and this may cause a are the simplest type of gear. Their
a speed ratio in the range of 3–12). gear unit to behave unstably. This slightly better efficiency compared
The star-gear type is used when the issue needs considerable attention, to the spur gearing is one advan-
rotating carrier would cause unac- especially for planetary gear units. tage for the helical gearing. For
ceptably high pressure on the bear- For instance, sophisticated elastic planetary gear units, a single-heli-
ing journal. The compound-epicyclic mountings are usually used for the cal gearing with an optimum helix
62 Chemical Engineering December 2013
FIGURE 3. In this planetary gear unit, the transmit-
ted power is divided over several teeth. Planetary
gear units are smaller, more compact and cheaper
than their conventional counterparts

before assembly and allowing it to and manufacturing stage. To achieve

return to the ambient temperature the most-even load distribution pos-
after assembly, employing the phe- sible, adjustments (also called cor-
nomenon of thermal expansion to rections) are usually necessary in
make a joint. the direction of the tooth height and
angle is commonly used (Figure 3). The shrink-fit design for the gear the tooth width. Because the pinion
In some cases, the CPI machin- wheel is usually limited by the in- gets hotter than the gears, the pin-
ery shaft could be connected rigidly fluence of centrifugal forces. For ion’s base circle and base pitch could
(without a coupling) to the plan- instance, the gear-unit codes do expand more. During manufacturing
etary gear-unit shaft. In other de- not accept the shrink-fit design for in a cold condition (say ambient con-
signs, the machinery bearing can be pitch-line velocities of over 140 m/s. ditions), the pinion has to be made
incorporated in the planetary gear- In those cases, the one-piece version with a pitch that has been reduced
unit housing. As a result, the differ- should be applied. In this configura- by a difference (an adjustment) be-
ent heat-induced expansions of the tion, the gear wheel and its shaft are cause of the thermal expansion. The
gear unit and that of the machinery in one piece. In gear units for rating increase in the pitch of the rotating
housing would produce no radial powers of more than 20 MW, the gear as a result of the centrifugal
shift. However, this could cause an one-piece design usually requires force should also be taken into ac-
angle error between the rotor and a large forging with a great weight. count during the manufacturing.
ring-gear axis. This can be absorbed Manufacturing large special-pur- The heat is distributed unevenly
by a single-jointed coupling. pose gear units requires access to across the width of the tooth face,
special forging shops for very large, resulting in uneven expansion of
Gear-unit reliability one-piece components. the gear body. Asymmetrical adjust-
Failures on gear units are usually The gear wheel shaft can be ei- ment techniques should be used for
related to teeth breakage and dam- ther solid or hollow. Because of a better corrections and consequently
age to the tooth flanks (these often better hardenability (during manu- smoother gear unit operation. In
result from impact or fatigue frac- facturing), the hollow-shaft design simple terms, these techniques de-
tures, wear or pitting). Having suf- is often preferred. pend on the accurate prediction of
ficient lubrication and maintaining Material selection and metal- asymmetrical thermal movements
the overall surface condition (par- lurgical issues are very important and the required asymmetrical
ticularly the roughness) of the tooth for any gear unit. Case-hardening- adjustments to achieve a perfect
flanks are important. Sufficient lu- grade gear materials are widely match and smooth operation during
bricant film on the gear flanks can used. Case hardening (or surface actual operating situations.
minimize the impact of pitting and hardening) is the process of hard- Suitable profile adjustments (cor-
micro-pitting. Cracks (particularly ening the surface of the gear while rections) can be employed by the
friction cracks) and erosions can allowing the metal beneath the sur- gear unit vendor to minimize the dy-
also cause gear-unit failures. face to remain soft, thereby forming namic loads and the gear unit noise.
Fractures can occur in areas of a thin layer of harder metal (the These adjustments, if implemented
high stress concentrations, which case) at the surface. properly, can considerably reduce
can result from abrupt changes Two important subjects for gear mesh impacts and shock loads at
in the geometry, localized areas of units deserve special attention for both the beginning and the end of
high strain (particularly bending optimal performance and reliability: meshing (and the transitions from
deformations) or regions experienc- 1. Tooth adjustments and correc- single to double meshing). These
ing thermal extremes. Reliability tions. The prediction of mechanical corrections can also help to achieve
and performance calculations (such and thermal deformations and ad- uniform transmission of the rotary
as the gear-unit rating, gear tooth- justments on the tooth dimensions movement, despite a position-de-
bending strength, gear surface pit- (such as tooth height and width) is pendent tooth deformation.
ting, scuffing and others) should be essential.
carefully reviewed. 2. Heat treatment. Heat treatment Heat-treatment options
The fabrication method also has involves the use of heating or chill- For critical gear units, carbur-
a great influence on reliability and ing, normally to extreme tempera- izing heat-treatment or nitrid-
gear failure modes. Pinions are al- tures, to achieve a desired result ing heat-treatment are commonly
ways made from one-piece fabrica- mostly the hardening of the gear used by vendors. Case hardening
tion, which is generally more reli- material. For gears, carburizing (in the usual form of carburizing
able than other options. heat-treatment or nitriding heat- heat-treatment or nitriding heat-
Shrink fitting is a technique in treatment are commonly employed. treatment) could be considered an
which an interference fit is achieved There is one potential problem economic solution, since it allows
by a relative size change after as- that every engineer working with a good range of adjustments to the
sembly. This is usually achieved by gear units should be aware of. This desired hardness depth of the gear.
heating or cooling one component should be considered in the design These manufacturing and heat-
Chemical Engineering December 2013 63
Engineering Practice

treatment options offer one of the hardening method. With nitriding, It has been proven that the resid-
highest values to ensure long-term the whole heat-treatment process ual stresses can influence the total
resistance to pitting and tooth flex- is carried out below the transfor- strain of the gear sets. In the case
ure. Selection of the most appropri- mation temperature (that is, the of carburizing, residual stresses can
ate option depends on many fac- temperature where a metallurgical be relatively high. An advantage of
tors, such as the application (speed, phase transformation occurs). How- nitriding is that the inner residual
power, operating details and others) ever, only small hardness penetra- stresses are relatively low. n
and many other technical and com- tion depths can be obtained using Edited by Suzanne Shelley
mercial issues. this method. For instance, 0.4–0.7 Author
Carburizing usually improves fa- mm with normal nitriding steels, Amin Almasi is a rotating
tigue resistance. With better fatigue- while this depth could be up to 1.5 equipment consultant in Aus-
tralia (Email: amin.almasi@
resistance and better strength, the mm with special steels. Meanwhile, He previously
carburized gears need relatively nitrided surfaces are usually harder worked at Worley Parsons
Services Pty Ltd. (Brisbane,
lower thicknesses and materials and could show more brittleness in Australia), Technicas Reuni-
and thus gear dimensions can be case of shock strains compared to das (Madrid, Spain) and Fluor
(various offices). He holds
better optimized. As a result, car- carburized ones. Also, the damage chartered professional engi-
neer license from Engineers
burized components can theoreti- curve in the fatigue strength for the Australia (MIEAust CPEng –
cally be designed smaller compared finite life in nitrided gears tends to Mechanical), chartered engineer certificate from
IMechE (CEng MIMechE), RPEQ (Registered
to the nitride-hardened ones. be flat compared to that of carbur- Professional Engineer in Queensland) and he
However, care should be taken ized ones. These factors can influ- also holds M.S. and B.S. in mechanical engineer-
ing. He specializes in rotating machines includ-
when selecting carburizing. Dis- ence the rating of gears for startup, ing centrifugal, screw and reciprocating compres-
tortions of the components during the shock loading the short-circuit sors, gas and steam turbines, pumps, condition
monitoring and reliability. Almasi is an active
quenching can result in some re- torque (for gear units connected an member of Engineers Australia, IMechE, ASME,
Vibration Institute, SPE, IEEE, and IDGTE. He
sidual stresses. By comparison, ni- electric machines) and other tran- has authored more than 60 papers and articles
triding is a relatively low-distortion sient situations. dealing with rotating machines.



• Best Sensitivity Only 3.25" x 6.25" x 2"
• Digital Display • Starter Door • Panel
• Raceway • Wall
• Adjustable Delay Timers • Works on Wide-range of Motors
• Simplifies Installation






No Sensitivity
Power is Linear-Equal Sensitivity For Low Loads
at Both Low and High Loads



Circle 27 on p. 68 or go to
64 Chemical Engineering December 2013

Fire-Water Pumps
for CPI Facilities FIGURE 1 Fire-water pump assem-
blies are typically skid-mounted to ease
installation and operation. Shown here
Follow this guidance to improve the selection, design is an example of a skid-mounted, diesel-
engine-driven, fire-water pump package
and operation of pumps handling water for fire-
ent guidelines have been used to
fighting and related systems estimate the ideal water volume
and flow requirements for CPI fire
Amin Almasi around 20–35% of the insurance- cases, depending on plant specifics,
Consultant deficiency rating points for a CPI applicable codes, regulations and
plant are related to inadequate fire- fire-fighting methods. As a result,

ire-water pumps are criti- water pumping systems. On aver- these specifications will vary based
cal machines that save lives age, 5–10% of all fire-water pumping on whether the plant is using fire
and prevent chemical pro- systems in CPI plants have failed to control, fire suppression, exposure
cess industries (CPI) facili- provide satisfactory services at the cooling and so on.
ties from damage more so than time required (as evidenced during Fire-water demands are usually
any other plant components. Fire- actual fire cases or drill-type exer- calculated based on the maximum
water pumps are nearly always cen- cises). Thus, it is critical to design, rate of water that will be required for
trifugal pumps with capacity from implement, operate and maintain a worst-case scenario — typically a
around 20–3,000 m3/h. Specific re- these critical systems — beyond the potential scenario involving a large,
quirements for fire-water pumps minimum requirements set forth in single-fire incident. The most remote
are briefly noted in fire codes (such published fire codes. unit(s) from the fire-water pumps, or
as NFPA 20 [1]), but this may not the largest unit(s), are typically exam-
be sufficient for specifying high- Fire-water system ined to identify the worse possible fire
performance fire-water pumps in Fire-protection efforts are categorized scenario(s). The potential scenario of
a way that ensures good reliability as passive or active. The primary a vast fire in the largest unit should
and operation as well as optimum passive measures for fire protection be used to define the capacity of fire-
price. This article provides practi- include efforts to ensure sufficient water system. The most-remote fire
cal notes on fire-water pumps to ex- clearances, install protective barri- unit(s) should be used to define maxi-
pand upon the information that can ers, limit and protect fuel sources, mum rated pressure of the fire-water
be found in existing fire codes. and other steps designed to reduce pump.
Various styles and configurations fire risks (such as the use of less CPI fire cases can be very differ-
of fire-water pumps are available hazardous materials, processes and ent, considering the different types
at different prices. In addition to equipment). By comparison, active of materials handled and the types
proper selection, fire-water pumps fire-protection systems are designed of operations carried out at differ-
must be properly integrated into to detect and apply fire-protection ent facilities. Today, computerized
the overall fire-water system, as an measures, which usually rely on some simulations play a critical role in
integral part of the CPI facility. effort to actually extinguish the fire. identifying and modeling potential
Figure 1 shows an example of Commonly used system components fire scenarios, validating fire-fight-
a skid-mounted, diesel-engine- include fire hydrants, monitors, hose ing methodologies, and estimating
driven fire-water pump package. reels, water-spray systems, deluge- the required water capacity.
The performance and reliability of type fire-protection systems or water- When evaluating a potential
a fire-water pumping system is an exposure cooling systems. The fire-wa- fire-fighting scenario, additional
important issue, and details of the ter pump plays an important role in pressure (a safety margin to the
fire-water pumping system are usu- most active fire-protection systems. calculated head) should be added
ally a part of risk studies, HAZOP to maintain the fire-water pres-
and inspection activities. System design and sizing sure in all remote units and critical
Fire-water pumps are important There have been different sets of fire-fighting systems; this is nec-
to different stakeholders, including rules to define the required flow essary to ensure that a fire-water
clients, investors in the CPI plant, and head of fire-water pumps. In stream with adequate pressure can
and insurance providers. Usually other words, a variety of differ- be maintained to support all appli-
Chemical Engineering DECEMBER 2014 53
Environmental Manager

cable fire-fighting systems should • The control of a fire event usu-

there be a fire in a unit. ally requires variable amount
Fire-water and utility-water sys- of water at a relatively constant
tems have sometimes been com- pressure
bined in non-critical plants. In the • Fire-water pumps are typically
event of a fire, the connected util- operated in parallel. A relatively
ity water system would be tripped. flat curve ensures troublefree
However, these combined systems parallel operation
are always risky. Various fire codes Sometimes, a large amount of
recommend that no utility-water water can be required by the fire- FIGURE 2. Shown here are several
connections be made to the fire-wa- water system to battle a vast fire; examples of fire-water pumps; an identi-
ter system. In some special cases, in those cases, the required water cal spare pump is commonly used to in-
the fire-water system may be used could be considerably larger than crease the reliability of fire-water pump-
ing systems
for emergency process-cooling re- the rated flow of the pump. In
quirements, but only as the second- this regard, the fire-water pump ing and power-density exceed a cer-
ary (reserve) supply. overload point (the end operating tain level. As a rough indication,
Fresh (treated) water is always point at the right side of the pump this limit could be 400 kW.
preferred (over seawater, brackish curve) should demonstrate a capac- As noted, the fire-water pumps in-
or untreated water, for instance) for ity of more that 150% of the rated stalled at any given facility should
fire-water systems in all onshore capacity at a head more than 70% be able to operate in parallel. How-
plants. Untreated or brackish water of the rated point. In other words, ever, there are some challenges
can cause many issues such as cor- operation point could move to the and issues in ensuring parallel op-
rosion, which can potentially wreak far right side of the rated point and eration. Even in certain conditions,
havoc on the system components. that point should offer sufficient pumps designed to operate in paral-
In general, engineers should pur- flow and head. lel could be subject to overheating
chase or construct the fire-water A steep pump curve should al- or damage. A well-known danger is
pumping system and the fire-water ways be avoided. As a rough indica- one pump operating at higher flow,
distribution system using proper tion, the average slope of a fire-wa- forcing another pump to operate at
materials (for instance, selecting ter pump curve should preferably lower flow; operation at lower flow
suitable corrosion-resistant mate- be around 10–20% (for instance, an can be damaging to the pump.
rials or proper protective coatings), average slope of 1/10 up to 1/5). When fire-water pumps are oper-
because untreated raw water (such Fire-water pumps can idle ated in parallel, the pump with the
as seawater) could be used as the against closed valves for a short lowest head may work at a reduced
secondary source for extra fire-wa- period of time. In other words, for a flowrate. In this way, the pump
ter capacity, in the case of an unex- short time, the pump should be able could work far from the “best effi-
pected fire event. If this happens, to operate in a closed water system ciency point” with a very low effi-
the fire-water system should be without any fire-water application. ciency, high friction and heat gener-
flushed with treated water after the Check valves should be provided ation, which can result in damage.
incident, to remove residual traces at both the discharge and the suc- Even in identical fire-water pumps,
of untreated source water. tion. The rated pressure of a fire- pumps that have been in use for
water pump could be 4–30 barg. more hours (and thus has probably
Selecting fire-water pumps Single-impeller centrifugal pumps been subjected to more wear), pumps
Centrifugal pumps with a rela- (for applications that require pres- with minor defects, and pumps with
tively flat characteristic perfor- sure below roughly 12 barg), and slightly lower speed could all be
mance curve (a graph of head ver- multi-impeller centrifugal pumps subjected to a reduced flow, which
sus flowrate) are generally selected (for higher-pressure systems) are can create problems during an ac-
for fire-water pumps. Ideally, the also commonly used. tual fire event. Because of this ef-
head should rise continuously from The differential pressure of a fect, operators should rotate pumps
the rated point to the shutoff point, pump is proportional to both the over time, so that each pump works
with only a small increase of head square of the rotating speed and as the main fire-water pump for
(say, a 9–15% rise of the head from the square of the impeller diameter. some period of time; this can help to
rated point to shutoff point). These A discharge pressure of around 10 ensure even wear patterns among
pumps can provide a steady, stable barg can be obtained by a relatively identical pumps in service. Individ-
flow of water at a relatively uniform large, single-impeller pump (with a ual protection against the minimum
pressure over a wide range of re- suitable speed). flow (to ensure a minimum flow for
quired fire-water flowrates. Overhung (OH) pumps have been each pump) is recommended.
A relatively flat performance used for small- and medium-sized Monitoring of the differential
curve is always encouraged for cen- fire-water pumps. Users should temperature of each pump can pro-
trifugal fire pumps for the following consider the between-bearing (BB) vide valuable insight for estimating
reasons: pump design when size, power rat- the parallel operation issue (the re-
54 Chemical Engineering DECEMBER 2014
Environmental Manager

duced-flow problem) and resulting ing details. However, in this au- be considered for pumps above the
inefficient operation (overheat). In thor’s view, NFPA 20 should be con- 350-kW range.
case of reduced flow, the differential sidered as a minimum requirement
temperature (the discharge temper- for a fire-water pump. For critical Fire-water pump arrangement
ature minus the suction tempera- fire pumps, the well-known API- The location for fire-water pumps
ture) would rise and could indicate 610 pump standard [2] is addition- should be selected carefully to mini-
such a malfunction. ally applied. mize various risks and potential
Because of small leakage and hazard situations. Explosions or
small consumption of fire water, the API-610 fire-water pumps high-hazard fires are major con-
pressure in a fire-water network The API-610 pump standard is used cerns, which can disable fire-water
could decrease slightly. Fire-water to ensure the reliable operation of pumps. Ideally, there should be
pump systems are usually designed high-performance pumps, mainly in 40–80 m of clearance between fire-
in a way that spare pumps should the oil-and-gas, petroleum refinery water pumps and a hydrocarbon
be started if fire-water pressure and petrochemical sectors. The API- or chemical process unit or storage
dropped below a certain level. How- 610 is usually considered to be the area. This limit should also be re-
ever, a slight pressure drop should minimum specification for pumps spected for some utility areas, such
not lead always to the startup of a that handle hazardous, flammable, as a power-generation units, gas-
large fire pump, as this could result toxic and explosive liquids since compression units, oxygen-genera-
in many unnecessary on-off oper- any reliability issue associated with tion units, and similar.
ating cycles of the main fire water these pumps could result in a poten- The possibility of an unconfined
pump (Figure 2). tial disaster. API-610 pumps are also vapor-cloud explosion is one of the
On the other hand, small pres- very popular in applications with main concerns, as this could disrupt
sure changes resulting from varia- extreme temperatures, including utilities, damage major support fa-
tions in fire-water consumption pumps for both high-temperature cilities, and damage the fire-water
during a fire incident can result in service (such as boiler-feed-water, or pumping system. Generally, there
an unstable operation of the main BFW, pumps) and low-temperature is a great possibility of the electri-
fire-water pump(s). For instance, applications (for example, pumps cal network or the steam-distribu-
this may lead to unnecessary fast used in liquid petroleum gas (LPG) tion system failing in the event of
changing of the operating point of a liquefied oxygen, and liquefied natu- a major explosion or extensive fire
large pump, which can result in per- ral gas (LNG) service). For critical event. This underscores the critical
formance and reliability issues. (high-risk) CPI units, fire-water role of independent, diesel-engine-
Smaller-capacity pumps (known as pumps are usually specified to com- driven fire-water pumps. Fire-wa-
“jockey” pumps) are usually employed ply with the API-610, to ensure that ter diesel engines should generally
in conjunction with the main pump(s) they are able to achieve a high re- comply with NFPA 37 [3].
to maintain a relatively constant fire- liability level — the same as other Regarding the diesel fuel-tank ca-
water pressure. Jockey pumps usu- pumps in the unit. pacity, typically, a 12-h duration is
ally initiate operation after a rela- Engineers often struggle with specified as the minimum require-
tively small pressure drop (say 0.5–1 whether or not to use API-610- ment. However, some critical CPI
bar) in a fire-water system. compliant fire-water pumps for plants require 24-h-capacity fuel
Main fire-water pumps are typi- a CPI plant. This decision would tanks for each fire-water pump-die-
cally electrically driven and the depend on the application, pump sel engine.
spare (backup or reserve) fire-water head, power rating, capacity, pump Meanwhile, each diesel engine
pumps are typically driven by die- speed and expected reliability. The should be provided with independent
sel engine. A commonly used ar- main variable is CPI service (the auxiliaries and accessories, includ-
rangement for critical CPI facilities CPI plant and expected reliability). ing a dedicated fuel system and fuel
is to install six fire-water pumps, For instance, for critical units han- tank. The startup of the engine is
including two electric-motor-driven dling flammable liquids and gases, commonly managed by a battery sys-
pumps, two diesel-engine-driven API-610-compliant fire-water pumps tem (with two independent barriers).
pumps and two jockey pumps. Fire- are often preferred. The failure of a diesel engine is
water pumps are nearly always pro- For a fire-water pump with dif- usually the result of a problem with
vided on a prefabricated skid. This ferential pressure more than 20 one of the auxiliary systems. Major
packaging concept can help to ease bar, API-610 is usually specified. reasons for such a failure include
the alignment and installation is- The pump power rating is a bit fuel-system issues, a lubrication-
sues and ensure high reliability. tricky, since there are many non- system problem, a starting issue,
As noted earlier, the fire code API fire-water pumps available a wiring problem or component fa-
NFPA 20 is dedicated to fire-water (with successful references) that tigue. Only clean, high-quality die-
pumps. It specifies proper require- are intended for high power ranges sel fuel should be used, and special
ments for pump tests, pump per- in a wide array of industry applica- attention is required for the lubri-
formance curves, pump accessories tions. As a general rule of thumb cation oil selection and supply.
and auxiliaries, and some packag- for many CPI plants, API-610 can Proper overhauls and repairs
Chemical Engineering DECEMBER 2014 55
Environmental Manager

are required, just like for any other dition to the conventional fire-water tightening) is to install dial indica-
properly designed combustion en- pumps — to supply seawater (for tors (or other types of indicators)
gines. Experience has shown that CPI plants located at the coastal re- that monitor movements in critical
the diesel-engine-driven, fire-water gions) or other sources of untreated parts of the machinery train. Usu-
pump is the most reliable option raw water (such as untreated water ally, two dial indicators are used to
currently available for severe loss from a lake or water wells), to observe movement in each machin-
incidents in a CPI plant. The reli- quickly supply additional capacity ery component (such as the driver,
ability and availability of micro- to the plant’s fire-water network. the fire-water pump and the gear
turbines (small gas turbines in the For a seawater-based emergency unit, if any) compared to the base-
50–400 kW range) could be higher fire-water pumping system, the plate or foundation (the main pur-
than diesel engines, but their ef- pumps are usually submerged in pose is to identify improper support
ficiencies are relatively lower (in seawater. For some locations, the of machinery, called “soft-foot”).
terms of lower operating duration seawater level may fluctuate from Two dial indicators can be used
with the same amount of fuel). Cur- –7 m to +16 m. Considering that to monitor critical bearing housing
rently they are not popular for fire- there is often a long distance from movements in the fire-water pump
water pump systems. the sea to the CPI facility, these fire- (usually in x and y directions). Ac-
Fire-water pumps are typically water pumps should be designed ceptable movements should be
arranged for both manual and au- to produce a relatively high head. below 0.04 mm (40 micrometers) to
tomatic startup. Automatic startup These special fire-water pumps are ensure a proper piping-pump con-
is expected to happen rapidly, in a usually multi-impeller pumps. Prop- nection. A similar limit should be
very reliable manner, once a fire erly designed highly reliable pumps applied to movements in all critical
event has been detected. Fire-water are always specified for such com- pump train locations (such as bear-
pumps are usually stopped manu- plex service. Electric-motor-driven ing housing, coupling, machinery
ally at the pump’s local control submersible pumps, or sometimes support and others), and suction and
panel. In other words, operator in- hydraulic-driven pumps are used discharge nozzle flanges (in terms
tervention is usually used to turn for these special applications. These of limiting deformations in all direc-
off the pump, once the situation has are usually down-hole, vertical tur- tions). For special fire-water pumps,
stabilized and the fire is out. bine-type pumps. depending on the machinery design,
A suitable enclosure (or building) In some cases, local regulations speed, power rating and applica-
should be provided for fire-water or plant specifications require three tions, a limit higher or lower than
pumps. Sufficient reinforcement dedicated fire-water pumps (as the the above-mentioned (0.04 mm)
should be considered for the fire- minimum) for special CPI applica- may be specified. n
pump enclosure. This is very im- tions (such as CPI plants that han- Edited by Suzanne Shelley
portant. For example, in the case dle highly explosive or highly flam-
of a major earthquake, fire-water mable materials). In such cases, one References
pumping systems need to be fully of the main concerns is personnel 1 NFPA 20, Standard for the Installation of Sta-
tionary Pumps for Fire Protection, National
operational to respond to fire events safety during a major fire case and Fire Protection Assn. (NFPA), 2013.
resulting from the earthquake. An provisions must be made to ensure 2 API 610, Centrifugal Pumps, American Petro-
open-sided shelter is not recom- a safe personnel evacuation. leum Inst. (API), 2009.
3 NFPA 37, Standard for the Installation and
mended. And, fire-water pumps Use of Stationary Combustion Engines and
should be located at a higher eleva- Installation and commissioning Gas Turbines, National Fire Protection Assn.
(NFPA), 2014.
tion than the majority of the CPI The piping installation and connec-
facility and upwind of it. tion to a fire-water pump can lead
To provide another layer of protec- to relatively high loads on the pump Amin Almasi is a rotating-
tion (in order to avoid common fail- nozzles and to the pump train’s sen- equipment consultant in Aus-
tralia (Email: amin.almasi@
ures), the main fire-water pumps, sitive components, such as bearings, He previously
and any other reserve or support- coupling and rotating assemblies. worked at Worley Parsons Ser-
vices Pty Ltd. (Brisbane, Aus-
ing fire-water pumps should not be The fire-water pumps are often tralia), Technicas Reunidas
located immediately next to each left on standby, therefore any high (Madrid, Spain) and Fluor
Corp. (various offices). He
other. Locating fire-water pumps at nozzle loads or misalignment might holds a chartered professional
two separate locations can improve be left unchecked. This can poten- engineer license from Engi-
neers Australia (MIEAust
both the fire-water system reliabil- tially wreck the pump in the first CPEng – Mechanical), a chartered engineer cer-
tificate from IMechE (CEng MIMechE), RPEQ
ity and overall fire-water network hours of operation in the event of (registered professional engineer in Queensland)
hydraulic behavior. fire. Periodic checks are important. and he also holds M.S. and B.S. degrees in me-
chanical engineering. He specializes in rotating
A well-known method to monitor machines including centrifugal, screw and recip-
Special fire-water pumps the movements and deformations of rocating compressors, gas and steam turbines,
pumps, condition monitoring and reliability. Al-
For critical CPI plants, additional critical machinery components dur- masi is an active member of Engineers Australia,
emergency (or reserve) firewater ing the piping connection (piping IMechE, ASME, Vibration Institute, SPE, IEEE,
and IDGTE. He has authored more than 60 pa-
pumps should be provided — in ad- flange and machinery flange bolt pers and articles dealing with rotating machines.

56 Chemical Engineering DECEMBER 2014

Engineering Practice

Chemical Process Plants: Plan for Revamps

Follow this guidance to make the most of engineering upgrades that are designed to
improve plant operations or boost throughput capacity
Koya Venkata Reddy 600
590 mm (max)
FACT Engineering and Design Organization Efficiency %
570 mm (rated) 70

he chemical process industries (CPI) are func- 530 mm (min)
tioning in an era of globalization, and between the
prevailing economic conditions and upheavals 50
in the energy sector, the number of new invest-

Efficiency, %
Head, m
ments in CPI facilities has fallen in recent years. Many 300 40
industries are seeking cost reductions by revamping
existing plants with minimum investment. The objective 30
is to reduce the cost of production through the use of 200
upgrades and new technologies, to remain competitive 20
in the market. By way of example, if one wants to set 100
up a new complex to produce ammonia and urea, the NPSHR 10
specific capital cost will be on the order of $666/ton of
urea. By comparison, if an existing plant is revamped to 0 0
0 250 500 750 1,000 1,250 1,500 1,750
raise the existing production from 100% to 120% (that Capacity, m3
is, adding 20% additional capacity), this can be done at
an expenditure that is closer to $300/ton to achieve this FIGURE 1. Shown here are typical pump characteristic curves, with three
different impeller sizes, showing capacity versus head, and NPSHR versus
incremental production capacity
This article reviews key concepts, objectives and pro-
cedures that are needed to successfully carry out vari- a turnkey basis. Meticulous planning related to the hook-
ous types of CPI plant revamps. up of tie-in points arising out of expansion schemes can
help to reduce the amount of downtime required to ex-
The need for revamps ecute the revamp schemes and put the plant back on-
Chemical process plant revamps are typically under- line.
taken for the following reasons:
• To change in feedstock composition Targeted revamp capacity, change in process
• To adopt energy-conserving processes in light of in- In general, it is possible to increase the rated capacity
creasing energy costs of a plant by 10%, with very little added expenditure.
• To reduce the fixed-cost components of production, But to increase capacity by 20–50% over the nameplate
by increasing capacity within the existing facility capacity, substantial modifications must be taken into
• To extend the life of a well-maintained process plant consideration that often involve implementing different
Similarly, there are many benefits to conducting an ap- technologies from the ones already applied in the exist-
propriate plant revamp. These include the ability to: ing plant. When seeking such notable increases in pro-
• Increase the reliability of equipment, leading to reduced duction capacity, plant operators and managers must
downtime and maintenance costs not only verify the soundness of the economics, but also
• Reduce energy consumption carefully evaluate the potential drawbacks, if any.
• Extend useful plant life Sometimes the existing process path may have to be
• Reduce the cost of production, thereby improving the changed to enhance the capacity of the plant, since the
overall bottom line for the facility current process may not yield the desired efficiency or
However, experience shows that inefficient implementa- conversion rates. Two cases are discussed below.
tion of proposed revamp options can lead to failure, so Example 1. In the case of units to recover liquefied pe-
care must be taken to avoid this by building the right troleum gas (LPG) from natural gas, such units are de-
team of experts. This team typically includes represen- signed for a certain composition of feed gas. The need
tatives of the process licensor company, engineering for a revamp often arises if the gas composition has
and project-management consultants, and experts from changed and the expected recovery of C3/C4 and higher
the owner company representing diverse fields, such as compounds has become unprofitable. In this case, the
operations, project management and maintenance. If expected recovery of LPG and natural gas liquids (NGLs)
sufficient expertise for the proposed revamp is not avail- can be achieved by compressing the feedstock to higher
able internally, one can hire consultants to carry out the pressures than present levels, or by spiking heavier NGLs
feasibility studies and implementation of the revamp on back to the feed gas stream. Thus, such a revamp re-
48 Chemical Engineering december 2015
quires a study to assess the technical and economic fea- raw materials, utilities and energy per unit of production
sibility of the different process paths being considered. are tabulated. The material-and-energy balance of the
Example 2. A feedstock change from naphtha to natural existing operation, and the required revamp plant load,
gas in ammonia plants, hydrogen plants and methanol are prepared.
plants also necessitates a need for revamp of the re- The existing equipment components are rated for
former section and front end, but in many cases, the the revamp conditions, and then changes and required
existing process path can be retained. In this case, the new equipment are identified. Cost estimates of various
absorbed duty of the reformer — which tends to be the schemes are prepared (after consultation with various
major energy-consuming equipment found in the system vendors). Feasibility studies, followed by detailed project
— and the burner duties required vis-a-vis the required reports (DPR), are also prepared. The potential rates of
reformer absorbed duty are calculated to check their suit- return of various options are studied. The best option
ability. The maximum skin temperature of the reformer available (on the grounds of economic sustainability and
tubes for the feedstock change must be checked. technical feasibility) is then selected, so that the basic
In all cases, the existing process path, along with other engineering design package (BEDP) can be prepared,
options, must be studied in detail to arrive at the most and the revamp project implemented.
economical and technically feasible revamp option. As noted, successful revamps require assembling
the right revamp team. Typically, such a team consists
Lifecycle of the plant of individuals from the process licensor company, con-
The different phases of a plant’s lifecycle must be taken sultants for basic engineering and detailed engineering
into consideration when planning a revamp. Such phases services, contractors for specific electrical-, mechanical-
include the following: and instrumentation-related aspects of the project, and
1. Incubation stage — Initial stabilization period various engineers from the owner’s group (for instance,
2. Growth stage — Optimization and debottlenecking of those who represent specific disciplines and have a con-
operations to improve the efficiency crete understanding of the current operation).
3. Maturity stage — Attainment of stable operation The following planning steps should be undertaken:
4. Declining stage — Realization that plant capacity is 1. Estimate the plant’s inherent capacity from past and re-
not sustainable because of frequent equipment fail- cent data. This can be done by identifying weak areas
ures or excessive maintenance requirements in the plant (for instance, those that are contributing
Revamping the plant during Phases 1, 2 or 3 is relatively to non-realization of rated or required plant capacity),
easy, whereas revamping a plant during Phase 4, when or by conducting an end-to-end survey of the plant.
the facility is already in decline, requires the engineering Once such a study is carried out, efforts should be
team to adapt many of the modern technology options made to predict the potential performance improve-
to an aging infrastructure, and to replace many equip- ments of the plant if the weak areas are rectified.
ment components. 2. Prepare the process scheme and the equipment data
sheets. Carry out feasibility studies of all options (in-
Objectives of a revamp cluding both technical and financial aspects of the pro-
The objectives of a plant revamp should be spelled out posed revamps) and then develop the detailed project
prior to studying the options. Possible objectives could report. Set the target of the revamp in terms of time
be the following: (schedule) and cost.
• Enhance capacity from the present operating level to 3. Implement the approved revamp. Ideally, the revamp
expand capacity to, say, 110%, 120%, 130% of rated activities should be carried out during the annual
capacity scheduled turnaround period for the plant, to minimize
• Reduce production costs unscheduled downtime.
• Reduce pollution
• Reduce the consumption ratios of various raw materi- Estimate plant capacity
als and utilities Many older CPI plants can run at or above the rated ca-
• Reduce maintenance costs and increase the onstream pacity continuously for a week or a month. But due to cer-
factor tain operating limitations, and downtime that may arise
• Upgrade the technology to keep pace with the new from some underperforming equipment, the annual rated
developments, and to increase the plant life capacity is seldom achieved. Analyzing past operating
• Minimize plant shutdown data on a monthly basis (for the past 10 years or so) will
These objectives can be achieved by maximizing efficiency, reveal which equipment components are most often to
yield and conversion of raw materials in various sections. blame for downtime, and are thus affecting overall capac-
Specifically, plant revamps are often implemented to im- ity utilization. Such a study of past data is often called a
prove process optimization, increase energy conserva- weak-area analysis. Similarly, sometimes an end-to-end
tion, improve product quality and expand capacity. survey of the plant (from the plant commissioning to the
present day) is also conducted.
Key revamp procedures Existing equipment poses both opportunities (in the
Every revamp project should start by identifying the goals form of underutilized capabilities) and challenges (in terms
and actual bottlenecks. A material-and-energy balance of limitations). The ability to identify problem areas can
for the base case should be developed to reflect the ac- help the team to prioritize their debottlenecking efforts in
tual operating conditions. The consumption of various order to improve capacity utilization more quickly.

Chemical Engineering december 2015 49

The weak-area analysis TABLE 1. A typical calculation of Cv, before and after a
Understanding current operation is very important for
Unit Before After
the successful revamp of a plant. The plant performance
can be evaluated based on the performance data for the Flowrate m3/h 80 100
past 10 years, if the plants are relatively old. Otherwise Density kg/m3 950 950
the plant performance is studied from the beginning to P kPa 49.03 49.03
the present day (using the end-to-end survey). P kg/cm2 0.5 0.5
Two indices, the plant load factor (PLF), and the on-
N1 unitless 0.0865 0.0865
stream factor (OSF), are important to scientifically evalu-
ate the plant performance. Cv unitless 128.73 160.92
Control valve size in. 4 6
Actual production 100 Pipeline size in. 6 6
(Actual stream days) (Daily rated capacity) (1) • Internal reasons: Recurring. Examples include pro-
cess problems, mechanical breakdown of equipment,
Actual stream days 100 planned shutdowns and more
OSF = • Internal reasons: Non-recurring. Examples include
Annual design on stream days (2) lack of finished product sales, effluent treatment, lack
of byproduct sales and more
PLF OSF • External reasons: Recurring. Examples include utility
Overall capacity utilization = failure, raw-material shortages and more
100 (3) • External reasons: Non-recurring. Examples include
worker strikes, natural calamities and more
Actual annual production 100
Annual design onstream days x daily rated capacity (4) FFS and RLA analysis
In a chemical process plant, critical equipment and pip-
The performance of the plant is studied based on the ing must be evaluated for their fitness for service (FFS),
highest PLF and OSF, on both a yearly and monthly basis. according to API 579 [1], and their potential residual
Data on the highest daily production that is achieved with life analysis (RLA) must also be assessed. The API 579
the present hardware should also be captured. guidelines are designed to ensure that pressurized criti-
In addition to the past production performance of the cal equipment are operated safely. The ability to establish
units, a breakdown of individual equipment must be as- the minimum years of residual life of the critical equip-
sessed to identify the weak areas and arrive at the pre- ment is essential to justify the revamp of the old and well-
dicted performance in the post-revamp implementation maintained plants.
scenario. The best yearly, monthly and daily performance
must be considered in order to find the target capacity of Use of simulation software
the plant and identify the number of stream days that this Simulation software can play an important role during the
target capacity is likely to achieve. evaluation of potential revamp options, so its use is rec-
Analysis of historic downtime factors can also pro- ommended to study the competing process-revamp op-
vide insight. To assess the feasibility of the plant oper- tions. Such modeling can help the team to substantially
ating at higher capacity, the best-achieved PLF (on a reduce the time needed to study the technical feasibility
monthly basis), and the highest load achieved, should of revamp options. However, great care must be taken to
be considered. ensure the use of most appropriate thermodynamic mod-
In any process plant, onstream days are lost due to eling options that are suitable for the plant and its com-
various factors — including process problems, mechani- ponents, fluid properties, process conditions and so on;
cal breakdown of equipment, raw material shortages, otherwise the results can be wrong. Appropriate use of
planned shutdowns, finished product sales, effluent simulation software can reduce the time required to carry
treatment and byproduct sales (if any). Such lost days — out the revamp projects, and help the team to identify an
which contribute to a loss of overall capacity utilization optimized, cost-effective process path, based on an eval-
— should be tabulated, and the associated causative uation of proposed process sequence changes given the
factors noted and tabulated. various constraints.
From the weak-area analysis, one can estimate the The various revamp options are studied from a techni-
inherent capacity potential of the plant and identify in- cal and financial point of view, a suitable process path
dividual equipment components or sections that are is selected and the equipment that create a bottleneck
becoming a bottleneck to maximum capacity utilization. for the desired revamp option are identified. Once the
Sometimes the plant capacity is affected by external additional equipment and piping are identified (per the
circumstances, such as feedstock supply issues (for in- proposed expansion schemes), the required hookup
stance, urea plant capacity is impacted by the capacity points and tie-in connections must be identified. As
of upstream ammonia plants) utility supplies and more. noted, to reduce the impact of these hookups, they should
Dividing these factors into recurring and non-recurring — wherever possible — be undertaken in conjunction
factors will also provide insight into the priorities needed with short shutdowns that are planned for preventive
to address the problem. maintenance.
50 Chemical Engineering december 2015
Environmental and safety impacts TABLE 2. Typical Design Velocities of Fluids in CPI PIPELINEs
Environmental-impact assessment studies should be Type of line Allowable velocity (max), m/s
conducted during the conceptual stage to evaluate the Suction lines for the pump 1
positive and negative impacts of the proposed engineer- Discharge lines for the pump 2–3
ing changes on the environment, and to arrive at the so- Fire water 5
lutions to mitigate the adverse impacts, if any.
Gravity lines 0.6–0.7
Safety is always a paramount consideration. The team
must ensure that the proposed plant revamp, and all re- Low-pressure gas 20
vised process schemes, conform to the latest codes and High-pressure gas 15
safety norms. Hazard operability (Hazop) studies of the Low-pressure steam 20
process schemes during the basic engineering-design High-pressure steam 15
package stage, front-end engineering-design stage, and
the detailed engineering stage should be conducted. Dur-
ing the implementation stage, periodic technical audits the percent flooding velocity with the revamped through-
should be conducted to see that the construction is pro- put. If the flooding velocity is greater than 80%, the pack-
gressing according to design intentions. ings are replaced with ones that offer lower packing factors
Hazardous-area classification drawings of the plant and higher surface area per specified volume. However,
are developed, and existing electrical considerations and adequate wetting of the packing must be ensured, ac-
other instruments are evaluated and changed accord- cording to design guidelines, and circulation rates of liq-
ing to the modified hazardous area classification of the uids must be enhanced accordingly, if needed.
plant. Quantitative risk analysis (QRA) is also conducted Packed towers that contain ceramic packings have a
to submit to the statutory authorities, and any onsite and tendency to flood at lower gas velocities. Hence, in some
offsite emergency plans must be revised, as needed. cases, such packings may be replaced with steel packings
Similarly, a safety integrity level (SIL) analysis should (after conducting the technical suitability check) to help re-
also be conducted according to BS IEC 61511[3] and duce the flooding velocity and increase throughput.
BS IEC 61508 [4]. And, all safety-instrumented functions Pumps. Pumps are very important and often provide a
(SIF) of the instruments are to be SIL 2 (minimum). relatively simple revamp opportunity, to take advantage
of advancements in pump technology. The throughput
Debottlenecking individual equipment systems required at desired plant capacity is determined, and
Different strategies are available to debottleneck different the characteristic head-versus-capacity curves, re-
equipment components and systems. Some examples quired net positive suction head (NPSH), and other key
are discussed below: characteristics should be studied. Normally, pump
Trayed columns. The design data of the distillation manufacturers indicate three impellers (mini-
column should be studied, preferably using process mum, normal, maximum) that are suitable for
simulation software. The column is simulated for both any duty. The possibility of using a larger-sized
the existing operating conditions, and for desired higher impeller diameter should be studied, considering the
throughput or changed feed composition. The liquid and head and capacity requirements (Figure 1).
vapor rates for each tray, along with their physical proper- As the pump capacity increases, required NPSH
ties, are obtained. After obtaining the column profile and (NPSHR) increases. Hence, the available NPSH
liquid-vapor-traffic details in the column, the tray hydrau- (NPSHA)should be checked, to avoid cavitation of the
lics are calculated and suitable recommendations are pump at higher flows. The motor’s suitability should
made, regarding changes made to the weir height, the also be verified. Many successful revamps were car-
number of holes, pitch, the diameter of the holes (con- ried out by changing the impellers to those with larger
sidering the flooding conditions) and more. Tray vendors diameters. The team should also carry out a design
should be contacted when considering revamping the check to ensure that the piping material classification
distillation column trays. The team should ensure that is still suitable for the pump’s discharge piping.
the reboiler and condenser are rated for the maximum Instruments. Instruments such as flowmeters (orifice,
throughput expected. venturi and mass flowmeters), pressure indicators, tem-
Many advanced separation technologies that are avail- perature transmitters, level instruments and so on should
able today allow for higher-capacity trays to be retrofit- be rated and studied in detail for the proposed changed
ted into distillation columns. Similarly, the suitability of condition. Since orifice meters often give rise to higher
advanced structured packings can also be considered pressure drop, they may be replaced with mass flow-
when planning a revamp of distillation columns in petro- meters. Similarly, level instruments based on differential
leum refinery and other critical CPI applications. Many pressure can be replaced with non-contact type, radar-
present-day structured packings can help revamped type level instruments, which tend to be more accurate.
columns to improve capacity by 40–50%, while reducing Normally, the orifice plates in flowmeters are maintained
pressure drop across the column. with  ratios — that is, the ratio of orifice plate bore di-
Packed columns. In the late 1980s, Raschig rings were ameter (d) to pipeline diameter (D) — of 0.3 (minimum) to
popular in chemical process operations. A study of pres- 0.7 (maximum). The orifice meters are rated for the tar-
sure drop of the packed column at the rated capacity get throughput and the pressure drop across the orifice
should be carried out to determine the pressure drop per element is determined. If the pressure drop is too high,
foot of packed column. Such a study should also identify the orifice plates are changed to those of higher  ratios,

Chemical Engineering december 2015 51

to address the pressure drop issue without changing the TABLE 3. Allowable Pressure and Temperature ratings, per [7]
transmitter. To keep the  ratio less than 0.7 for a given Flange rating, Allowable pressure Allowable temperature (max)
pressure drop across primary element, either or both the per ANSI B16.5 (max) kg/cm2
orifice plate and the transmitter is changed. 150 class 18.3 93.3°C /200°F
Control valves. The flow through a control valve depends 300 class# 47.8 93.3°C /200°F
on its capacity, or so-called CV value (Equation 5), which is
defined as the flowrate in m3/h of water at a temperature
of 60°F with a pressure drop across the valve of 1 psi. The with changing the heat exchanger. Also, increasing the
rule-of-thumb rule is that the CV is roughly 10D2 (where number of baffles on the shell side to increase the heat
D is the size of the control valve in inches). For example, transfer coefficient should be considered. In the case of
the CV of a 2-in. control valve is roughly 40. The CV value plate heat exchangers, additional plates can be added
is recalculated according to ISA 75.01.01[2] with the new to increase the heat transfer, in consultation with original
flowrate, inlet pressure and allowable pressure drop. equipment manufacturer.
Normally, the control valves in the original design of the Limitation in line sizes. All of the line sizes are checked
plant are kept one size lower than the pipe line diameter, using the standard velocity criterion. Typical standard ve-
and their rated flow is specified as 1.7 times the normal locity criteria are shown in Table 2.
target flow, or 1.3 times the maximum target flowrate. The lines are checked for pressure drop. In case the line
Since the flowrate is specified as 70% higher normal pressure drop is high, the lines are changed to provide
flowrate, or 30% higher maximum flowrate, the control larger-diameter pipes. Special attention must be given for
valves will be suitable to handle the revamped target gravity-flow lines, as the allowable velocity is in the range
flow, which is 20–30% more than the design flowrate. of 0.6–0.7 m/s and sufficient slope must be ensured.
Hence, for a proposed 20–30% plant load increase, the The piping material thickness (according to ANSI B
existing control valve will normally be sufficient. If the 31.3) and flange ratings (ANSI B16.5) are checked to be
CV of the control valve is not sufficient, the team may sure they comply with higher pressure. In some cases,
consider either changing the trim of the control valve, the flange rating will be sufficient, as there is often a wide
or installing one with a higher CV. Equation 5 is used to margin available, as shown in Table 3.
calculate the CV . Thus, if a line that was designed for 10 kg/cm2 is going
to experience a pressure of 12 kg/cm2 at 90°C, then the
Q / 0 flange rating of 150# need not be changed. However, the
CV = 1
actual pipe thickness should be measured and checked
N1 P (5) for its suitability in the revamped design pressure condi-
tion. Sometimes no piping needs to be changed — for
Where: instance, if the design pressure in the revamped condition
Q = the flowrate through the control valve, m3/h is less than that of the original process. One example is an
N1 = a constant (8.65 x 10-2), from ISA 75.01.01-2007 ammonia synthesis section, where pressures have come
(IEC 60534-2-1 Mod), Table 1 [2] down from 200 kg/cm2 to 140 kg/cm2.
1 = density of the fluid, kg/m3 Pressure safety valve (PSVs). When the plant runs
0 = density of the water at 15°C, kg/m3 at higher revamped capacity, all of the PSVs must be
P = differential pressure, kPa checked according to API 520 [5]. The team must evalu-
Table 1 shows a typical calculation of CV before and ate the nozzle area suitability and the rating of the inlet
after revamp flowrates, and shows how the existing con- and outlet piping, after recalculating the fluid-relieving
trol valve must be changed to the pipeline size for a 20% rates associated with the new throughput. PSVs are
increase in flowrate. changed if they are found to be unsuitable. In the case of
Control valves should also be checked for noise lev- feedstock changeover, PSVs must be also be checked
els. Controllability and rangeabilty are also important for for changes in fluid properties such as molecular weight,
revamping the valve. Revamps involving control valves compressibility factors and so on.
should always involve vendor cooperation. If the revamp Compressors. Various options for revamping the com-
is not able to bring the process into the controllability pressors should be studied initially. Various revamp op-
range, either the valve should be replaced with one of tions include the following:
higher size, or fine feed-control valve can be added par- 1. Installation of a suction booster
allel to the existing control valves. 2. Installation of a parallel compressor
Heat exchangers. The existing heat exchangers should 3. Changing internals in the low-pressure and high-pres-
be checked for any excess available surface area, by rat- sure casing, along with steam turbine upgrading
ing them using standard software modeling packages. In 4. Providing a chiller at the suction inlet and changing the
general, an existing heat exchanger provides enhanced intercoolers. A chiller can be installed to reduce the
heat exchanging capacity if the pressure drop across the gas temperature and increase the volumetric capac-
tube side or shell side is increased. ity of the gas and reduces the power requirement. In
If the heat exchanger is downstream of a pump, the cases where the drive needs to be changed, this can
team should consider increasing pump head, which be applied.
would increase the allowable pressure drop across the 5. Change of compressor type. In older-generation urea
heat exchanger. There may be a tradeoff between the plants, urea reactors operated at 200 kg/cm2, and
operating cost of the pump and fixed cost associated they fed the CO2 to the urea reactor; Historically, CO2
52 Chemical Engineering december 2015
compressors have been reciprocating-type, which sis converter pressures were reduced to 135 kg/cm2
incur high energy costs. As the pressures in present- (from an initial level of 200 kg/cm2), as a result of the
day urea reactors have come down to 135 kg/cm2, introduction of radial basket converters instead of the
centrifugal compressors can be used instead, which older-generation axial converters. By retaining the same
helps to reduce operating costs as well as mainte- high-pressure converter shell, one can change the con-
nance costs). verter baskets to radial ones, which helps to reduce
Effluent treatment plants (ETP). Worldwide, waste- pressure drop.
water-treatment plants are typically designed with high Catalysts play a vital role in enhancing the reaction rate.
safety margins, to cater to shock loading or sudden The use of advanced catalysts should be considered,
peak loading of effluents containing high chemical oxy- where possible. For example, in sulfuric acid plants, va-
gen demand (COD). However, when a plant is stabilized nadium pentoxide (V2O5) is typically used as a catalyst.
and optimized, the generation of wastewater containing If an improved cesium catalyst is added to the reactor,
high COD is drastically reduced. the SO2 to SO3 conversion can be increased, and the
The following methodology should be adopted while emission of SO2 can be reduced, generally to far below
checking the capacity of ETP that are based on an acti- the statutory limits.
vated sludge process during revamp planning: Storage tanks. If the process revamp is based on a
1. Evaluate existing facilities by collecting operating data “more in/more out” concept — that is, more fluids will be
for one month and developing a statistical analysis of flowing into and out of storage tanks — then the team
various parameters. must check the capacity of “breather” valves and emer-
2. Check the design basis and the design volume of the gency vents according to API 2000 [8]. If the breather
aeration basin, thickener and clarifier. valves need to be replaced, the pressure settings may
3. Evaluate the operating case using the above design be adjusted in consultation with vendors, according to
basis. the applicable codes.
4. Calculate the energy requirements of the design and Utilities. During any plant revamp, the capacity of key
operating cases, and quantify the potential for reduc- plant utilities, such as demineralized water, instrument
tion of electrical energy at various loads. air, plant air, steam plants, power, and cooling tower
Flares and knockout drums. Flare systems, including should also be checked to be sure they will support the
knockout drums, must be checked before embarking on proposed revamp. Offsite facilities related to raw-mate-
a plant revamp. Flares are used to ensure plant safety, rials receiving, tank farms, and product-storage capaci-
by flaring hydrocarbons in case of emergency conditions ties must also be studied and related personnel require-
such as power outages, fire or blocked discharge. ments must be ascertained. n
While converting the ammonia plant from a liquid fuel Edited by Suzanne Shelley
(such as naphtha) to natural gas, the properties of the
fluid (such as molecular weight, compressibility factor), References
viscosity and density undergo a drastic change and 1. American Petroleum Inst., API 579: Recommended Practice for Fitness for Service, 2nd
have profound effects on height, flare diameter and flare Ed., July 2007.
tip suitability. Calculations must be performed to verify 2. Instrument Soc. of America, ISA 75.01.01-2007 (IEC 60534-2-1 Mod): Flow Equations
for Sizing Control Valves, 2007.
the new case, according to API 521[6]. The goal is to
see whether the existing flare is suitable to handle the 3. International Electrotechnical Commission (IEC), BS IEC 61511: Functional Safety –
Safety Instrumented Systems for the Process Industry, 2003.
changed load and fluid conditions associated with the
4. International Electrotechnical Commission (IEC), BS IEC 61508: Standard for Functional
proposed revamp. Vendor support should be sought, Safety of Electrical/Electronic/Programmable Electronic Safety-Related Systems, 2010.
if needed, and the flare design can be checked using 5. American Petroleum Inst., API 520: Sizing, Selection and Installation of Pressure Relieving
manual calculations, spreadsheet calculations and flare- Devices, Part 1, 8th Ed., 2008, and Part 2, 5th Ed., 2003.
specific computer software. 6. American Petroleum Inst., API 521: Pressure Relieving and Depressurizing Systems, 5th
Reactors. Reactors are the heart of chemical process Ed., 2007.
operations. Efforts should be made to maximize yield and 7. ASME/ANSI B16.5: Pipe Flanges and Flanged Fittings, April 2013.
conversion rates in the revamp scheme. If, following the 8. American Petroleum Inst., API 2000: Venting Atmospheric and Low Pressure Storage
reaction, raw materials remain unconverted, they must Tanks, 7th Ed., March 2014.
be separated and recycled back to the reactors. This
consumes utilities, thereby increasing energy consump- Author
tion. If conversion rates in the reaction are increased via Koya Venkata Reddy is senior manager, process engineer-
a revamp, the recycle ratios will be drastically reduced. ing, at FACT Engineering & Design Organization (FEDO), a div. of
In one urea plant, a revamp involved the following Fertilizers and Chemicals Travancore Ltd. (FACT; Udyogamandal
683501, Kochi, Kerala, India; Phone: +91-484-2568763; Email:
changes: Introduction of higher-capacity trays in the He has 24 years of experience in chem-
urea reactor in the ammonia plant; changing the con- ical plant operations, including expertise in the fields of process
verter baskets from axial- to radial-type in the ammonia control, process design, process risk analysis, Hazop analysis, pro-
cess simulations, environmental management and plant revamps.
converter in the caprolactam plant; using an enriched- He is a recipient of FACT’s Merit Award. Reddy holds a Bachelor of
oxygen supply to the cyclohexanone reactors with in- Technology degree from Andhra University (Visakhapatnam) and a
troduction of improved safety features. These changes Master of Technology degree in project management from Cochin University of Science
and Technology. He also received an M.B.A. in finance from Indira Gandhi National Open
were able to increase the conversion rate, increase over- University (IGNOU; Delhi). He is a lifetime member of the Indian Inst. of Chemical Engi-
all production and decrease energy consumption. neers (IIChE) and a member of the Institution of Engineers (India).
In the ammonia plant’s synthesis section, the synthe-

Chemical Engineering december 2015 53

Feature Report
Engineering Practice

Variable frequency drives:

An Algorithm for Selecting Vfds for Centrifugal Pumps
Using this simple algorithm on a E
personal computer, engineers can reservoir A

evaluate competing scenarios to B

D H2

identify the most cost-effective and H1

energy-efficient pump system design Static head: H = H2 - H1 P: Pump

ABCDE: Piping schematic V: Throttle valve

D. K. Shukla and D.K. Chaware

Essar Oil Ltd. FIGURE 1. 1.
Figure Shown
Pipinghere is the piping
schematics schematic
(Type I and II) for Type 1
and Type 2 pumping systems. Type 1 systems will have one
operating point (see Figure 2), while Type 2 systems will have
R. B. Swamy two or more than two operating points (see Figure 3) lying on
Litwin PEL LLC the same characteristic curve

his article describes an algo- units achieve greater fuel efficiency, large number of alternative scenarios,
rithm that end users can use to reduced emissions and less wear and which is an important part of the eval-
develop a program on a personal tear compared with continuous opera- uation process but calls for the sys-
computer to assess the technical tion at some pre-determined setpoint. tematic processing of large amounts
and economic feasibility of installing Maintaining a large inventory of re- of data. (This is explained in the case
a variable frequency drive (VFD) on dundant pumps in store or using them study, box, p. 42).
centrifugal pumps. This simple algo- without due consideration for energy Support standardization efforts
rithm serves a variety of useful func- is costly. The use of this algorithm throughout the plant. The use of
tions, helping engineers to carry out can help plant managers to devise a VFDs allow a particular pump or
the following functions: proper system to use surplus or re- suite of pumps to be operated at dif-
Check the technical feasibility and dundant pumps in the most energy- ferent speeds, providing greater op-
evaluate whether or not it is advis- efficient way anywhere in the same erating flexibility for the facility. This
able to use VFD. Such an evaluation plant or elsewhere. algorithm can help engineers in their
is useful for operators considering both Increase the energy efficiency of efforts to evaluate whether it may be
(a) the re-rating of pumps already in centrifugal pumps. In recent years, possible to standardize plant opera-
operation, and (b) the deployment of pressures related to rising fuel costs tions on a few efficient models for the
other, perhaps redundant pumps that and environmental concerns call for entire plant, to help reduce inventory
may be warehoused at the facility. the most energy-efficient operation and maintenance requirements.
During plant expansion and de- of plant equipment. By more closely
bottlenecking activities, operating pa- matching pump operation with actual Methodology
rameters are commonly revised, and usage requirements, the use of a VFD To develop the algorithm, curves
this calls for either the re-rating of can help to improve the pump’s per- showing the available pump hydraulic
existing equipment, the deployment of formance with respect to both energy performance characteristics — spe-
surplus equipment that may be ware- efficiency and emissions. cifically, flow-versus-head, and flow-
housed at the facility, or the purchase Critically evaluate when the use of versus-efficiency curves for a set of
of new equipment. Generally speak- a VFD is appropriate — and when available speed and corresponding di-
ing, plant operators prefer to avoid it is not. While the use of a VFD can ameter for each model — were divided
new equipment purchases, which can provide compelling advantages in into two segments. A simple quadratic
be time consuming and costly, and some applications [1, 2], in other ap- equation is written for each segment.
instead prefer to look for opportuni- plications its use is not warranted This is essential as it is nearly impossi-
ties to re-rate existing and available due to technical or economic reasons, ble to describe the entire performance
equipment at the plant. The addition including the results of a lifecycle curve using a single quadratic equa-
of a VFD provides one option for re- cost (LCC) analysis [3]. The simple tion. The iterative calculation proce-
rating centrifugal pumps to help users computer-based algorithm discussed dure, using different speeds from from
gain better control and to help the below helps the engineer to assess a minimum speed (NMIN) to maximum
38 Chemical Engineering february 2010
m=1 System curve m=2 System curve
As per
H R2 (162)
HR figure 1

Head, m ------>
Head, m ------>

H R1 (70) Operating
H point
H Static head
H Static head
0 0
0 QR 0 Q R1 (31) Capacity, m3/h -----> Q R2 (103)
Capacity, m3/h ----->

FIGURE 2. InFigure
Type 1 pumping systems, the
2. Type I ( Single operating point).
FIGURE 3. This system characteristic curve for Type 2 pumping
pump will be operated only at one point (m = 1), Figure 3.two
system shows Type II ( Two operating
different operatingpoints on same
points system
(m=2), bothcurve)
of which
with required head HR at capacity = QR. Both op- lie on the same system characteristic curve, with required heads =
tions — the use of VFD and the use of FFD — are HR1 and HR2 at respective capacities = QR1 and QR2. In such a sce-
possible to meet these requirements nario, there can be more than two operating points (m>2). However,
all operating points will be lying on same characteristic curves.
Such requirements can be best accommodated by the use of a VFD

speed (NMAX) with predetermined in- stance, the number of units required, related to existing pumps (including
crements (say, 1 rpm), is then applied the pump and motor rating, and power both those that are either already in
to the segments. consumption. The engineer must sys- operation or available as surplus),
First, one of the segments is con- tematically analyze the hydraulic per- data related to new pumps from ap-
sidered, and using Affinity Laws (dis- formance data for each pump model to proved pump vendors can also be
cussed below), another performance evaluate the overall technical benefits stored and used to check the feasibil-
curve is generated at (to start) mini- and feasibility of the solution. ity of new purchases.
mum speed. Specifically, a simple qua- Ideally, end users should check a Using the algorithm, data related to
dratic equation is used to define this large number of pump models for tech- the hydraulic performance of compet-
segment of the performance curve. The nical feasibility (as discussed below in ing pumps that are offered by differ-
quadratic equation for the segment is the case study), which will require the ent approved vendors can quickly be
then used to check if the required duty processing of a large amount of data. analyzed, which may lead to more op-
point lies on the segment or not. In general, VFDs often prove ideal tions, as discussed in the case study.
If the duty point lies on the segment, for systems with zero static head The literature [1, 4] provides flow
then the speed corresponding to this (closed systems) or those with rela- charts and some discussion about
particular curve is established as the tively low static head. However, as the the methodology behind VFD selec-
matching speed. Otherwise the second static head increases, the use of a VFD tion. However, these references do not
segment is checked in a similar way. If may lose its economical advantages provide a precise algorithm with the
that segment does not match with the [2, 4]. A novel method is discussed in necessary equations and formulas to
duty point, then the calculations are Ref. [4] to study the effect of changes write a simple computer program to
repeated for the next higher speeds in static head. Similarly, VFDs may help pump end users evaluate poten-
(NMIN +1, NMIN + 2, and so on) until prove to be less ideal for systems with tial scenarios for the need to develop
the matching speed is determined. operating points at the same head but an algorithm.
If there is no matching speed to at different flows.
achieve the required duty, then the While these aforementioned rules- Pump system operating points
model is not suitable to fit with a VFD of-thumb are handy, they may not be For the purpose of VFD selection, the
and is thus rejected. The process is re- sufficient to justify the selection of a pumping system can be classified into
peated for all the available models to VFD for a given application. By con- three types, whose requirements and
list out those models that can be fitted trast, using the LCC analysis to iden- constraints are described below:
with a VFD (box, page 42). tify the exact cutoff point to confirm Type 1. The pump and VFD system
VFD viability is a more reliable ap- must cater to only one operating point
Take your time in deciding proach, as it takes into consideration (so that m = 1, where m = the number
Because of the potential advantages all the operating conditions. of operating points). The typical piping
that may result from the use of a VFD, Some pump manufacturers and schematics and corresponding system-
many users rush to implement such VFD suppliers may not be equipped characteristics curve are shown in Fig-
a system without conducting a thor- with software to check VFD feasibility. ures 1 and 2. In this case, the required
ough evaluation of its economic and Thus, many will not be able to carry process capacity is QR and the corre-
technical viability. Efforts to carry out out analysis for the redundant or ex- sponding system head is HR. The valve
such a viability check require an eval- isting pumps of some other manufac- settings, piping schematics, liquid lev-
uation of lifecycle costs (LCC) calcula- turer for VFD feasibility. els and pressure acting on the liquid
tions [2, 3]. The algorithm described here can be surface remains unchanged. Because
LCC calculations require techni- used to guide the procurement of new this is a relatively simple pumping sys-
cal-feasibility data related to, for in- pumps, as well. In addition to data tem, the use of fixed frequency drive
Chemical Engineering february 2010 39
System curve
System curve
Engineering Practice Per schematic
ABCDE of figure 4
Per schematic
H R2 ABCFG of figure 4
Discharge H R1
Discharge G reservoir Operating
Suction point
E reservoir

Head, m ------>

B C D H2
H2 H1
H V2
F 0
0 Capacity, m3/h -----> Q R2 Q R1
Static head: H1 = H1 - H
FIGURE 5. The system characteristic curves for a Type 3 pump-
Static head: H2 = H2 - H
ing system shows two different operating points (m=2). Both lie
ABCDE: Piping schematic P: Pump on the different system-characteristic curves with required heads
ABCFG: Piping schematic V1, V2: Throttle valves = HR1 and HR2 at respective capacities = QR1 and QR2. There can
be more than two operating points (m>2); however, all operating
IGURE 4. The piping array for a Type 3 pumping system is
F points will be lying on different characteristic curves. These re-
shown here. In a Type 3 pumping systems, two or more operat- quirements occur most frequently in real life situations and can
ing points all lie Figure
on different system
4. Piping characteristic
schematics (Type III) curves (see be handled by the use of a VFD
Figure 5)

(FFD) — for which a constant-speed • A combination of any of the above the pump manufactuers’s literature.
electric motor driven at 50 or 60 Hz is factors After the details described above
typically used — is generally preferred, The typical piping schematics and are gathered, the required operating
although the use of a VFD is possible. corresponding system-characteristic point(s) (marked as m on the system
Type 2. The pump and VFD system curves are shown in Figures 4 and 5. curves), and, as detailed in Table 1
must cater to two or more operating In this case, the process requirements (values Number 22 through 25), are
points (so that m ≥ 2). The typical change from one operating point gathered into the database.
piping schematics and corresponding (say, QR1) to another (QR2), with cor-
system-characteristic curve for this responding system heads of HR1 and Assumptions for the algorithm
scenario are depicted in Figure 1 and HR2. Such requirements are more The algorithm relies on the following
Figure 3. In this case, the process re- complex. However, they occur fre- assumptions:
quirements change from one operat- quently in real-life situations and can 1. According to the Affinity Laws, flow
ing point to another (say, QR1 to QR2, usually be managed effectively by the (Q) and head (H) are assumed to be
or vice versa). The operating points lie addition of a VFD. proportional to speed (N) and speed
on the same system curve. The valve squared (N2), respectively, for Q-H
setting, piping schematics, liquid lev- Required inputs characteristics from the minimum
els and pressure acting on the liquid To use the algorithm to assist with continuous stable flow (QMCSF) to the
surface remain unchanged. VFD evaluation and compare compet- maximum permissible flow (QMAX).
Such requirements can usually be ing pumping-system scenarios, the It should be noted that the Affinity
best handled by a VFD, as the use of various types of data that must be Laws are most accurately applicable
flow-control-valve throttling to modu- gathered are discussed in this section. for flows close to the pump’s best ef-
late the flow from QR1 to QR2 (or oth- First, data from hydraulic perfor- ficiency point (BEP) flow QBEP, and
erwise, per the process requirement) mance curves are collected and stored can deviate slightly as flow values
is eliminated. in the computer for any pumps for move away from BEP. This slight de-
The case study discussed in the box which the VFD feasibility is to be eval- viation, usually on the order of 3 to 4
(p. 42) provides a typical example of a uated. Table 1 (Numbers 1 through 21) percentage points, is neglected for
Type 2 application. In the absence of shows the database values that need calculation purposes.
a VFD, the required changes in flow to be gathered to use the algorithm. 2. It is known that the pump efficiency
are achieved by throttling, but this These details should be collected from can change with respect to speed.
approach is inherently inefficient and the pump maker. This change can be up to 4 percentage
leads to energy loss. For existing pumps, the details are points. This change is also neglected
Type 3. The pump and VFD system taken from the hydraulic performance for calculation purposes. (It should be
must cater to two or more operating curves, which are based on actual test- noted that there is no universally ac-
points (so that m ≥ 2). The operat- ing that is carried out by the vendor cepted way for the accurate prediction
ing points, however, lie on different before the unit has been dispatched of deviations in value predicted by the
system curves that are generated to the purchaser. For new purchases, Affinity Laws and change in efficiency
due to changes in the following the data are collected from the stan- with respect to speed.)
system attributes: dard hydraulic-performance curves at 3. The pump efficiency read from the
• Valve setting the maximum impeller diameter at a pump’s flow-versus-efficiency curve
• Static head speed corresponding to a 2-, 4- or 6-pole is used to calculate the power input
• Piping schematics, say, from ABCDE motor at 50 or 60 Hz. These standard to the pump (instead of reading the
to ABCFG (see Figure 4) performance curves are available from pump input power directly from the
40 Chemical Engineering february 2010
Table 1. values to be gathered to evaluate competing pumping scenarios
Item Variable (**) Details Item Variable (**) Details
1 n Number of pump models for which 12 H60, m Head, corresponding to Q60
the details below are available 13 BEP60, % Efficiency, corresponding to Q60
2 pump Pump model designation as per the 14 Q110, m3/h 110% of QBEP flow (If QMAX is close to
vendor (such as 80 X 360) Q110 then Q110 is selected as an aver-
3 D2, mm Impeller diameter at which laboratory age of QMAX and QBEP)
test results are available. If laboratory 15 H m Head, corresponding to Q110
test results are not available, then refer
to standard curve from the pump man- 16 BEP 110 % Efficiency at Q110
ufacturer’s book for a particular diam- 17 QMAX, m3/h Maximum permissible flow
eter. If pump manufacturer’s data are 18 H
MAX, m Head, corresponding to QMAX
to be stored for new purchase, then
refer to impeller maximum diameter 19 BEP MAX, % Efficiency at QMAX
20 N rpm * Maximum permissible speed for the
4 N, rpm Speed at which standard laboratory MAX,
tests results are available given pump model
21 N rpm *  Minimum permissible speed for the
5 QBEP, m3/h Flow at best efficiency point (BEP) MIN,
given pump model
6 HBEP, m Head, corresponding to QBEP
22 m Number of operating point(s) at
7 BEP, % Best efficiency, corresponding to QBEP which the pump is expected to run,
8 QMCSF, m3/h Minimum continuous stable flow with the details listed in Items 23, 24
9 HMCSF, m Head, corresponding to QMCSF and 25 for each operating point
23 Q m 3/h Required capacity
10 BEPMCSF, % Efficiency, corresponding to QMCSF R,

11 Q60, m3/h 60% of QBEP flow (If Q60 is close to 24 HR, m  Required head corresponding to
QMCSF then Q60 is selected as an rated capacity
average of QMCSF and QBEP) 25 Sp Gr Specific gravity
* These values are to be collected from the design department of the pump manufacturer, as they are not mentioned in standard performance curves.
** Units of measurement are shown for individual attributes.

pump’s flow-versus-power-input char- known variation makes curve fitting the given operating points (QR, HR
acteristics curve). This is justified, as simple. This is the usual range for and SpGr), the feasibility of using the
the pump efficiency curve nature is pump selection. first model for which data are stored
well-known (that is, it follows a con- This portion of the H-Q curve is in Table 1 is checked, using the it-
tinuous rise from zero flow to flow at further divided into two segments — erative calculations described in the
BEP, and then decreases continuously the first from QMCSF to QBEP, and following steps.
for flow greater than flow at BEP), the second segment from QBEP to Step 2. The iterative calculations for
and the pump efficiency curve is inde- QMAX. The goal is to get better curve selection start with generating the
pendent of the density of liquid being fit and also to take into account the first segment of the Q-H curve (from
pumped. Plus, the industry practice is change in slope of the Q-Efficiency QMCSF to QBEP) at minimum speed,
to refer to the efficiency curve as it is characteristics after QBEP (that is, NMIN, for the first model being con-
easy to read and interpolate. the decrease in efficiency with flow sidered, using data stored in the com-
4. The H-Q characteristics can be ex- after BEP). puter data bank.
pressed using the quadratic Equations The efficiency variation from The three points (QMCSF, Q60 and
(1) and (2) given below: QMCSF to QMAX can be expressed QBEP) required to generate the first
using quadratic Equation (2) for the segment (QMCSF to QBEP) at speed
h = An2 + Bnq + Cq2 (1)
two segments mentioned above. The NMIN are calculated using the Affinity
constants used in the quadratic equa- Laws to points QMCSF, Q60 and QBEP
h = head, m tions for first segment (from QMCSF that are available in the data bank at
n = rotational speed, rpm) to QBEP) are calculated using QMCSF, speed N (see Table 1). The quadratic
q = flow, m3/h Q60 and QBEP and corresponding equation based on these three points
A, B and C = constants for a given head and efficiency values for a given is used to define the first segment at
pump, impeller, and speed speed. The constants used in the qua- speed NMIN. The head for the required
For the typical case of a pump capacity dratic equations for the second seg- flow QR is calculated using the qua-
operating at constant speed, the above ment (from QBEP to QMAX) are calcu- dratic equation and is compared with
equation can be written as standard lated using QBEP, Q110 and QMAX and the required head HR. Then, the second
quadratic equation. corresponding head and efficiency segment is checked in the same way.
values for the speed used for the If the calculated head is within 1% of
h = a + bq + cq2 (2)
first segment. HR (or a different limit chosen by the
The constants a, b and c are for a given user) on any one of the segments, then
speed [5]. The range from minimal The algorithm the speed is taken as VFD speed.
continuous stable flow to maximum The specific steps in the algorithm are Step 3. If matching is not within 1%
permissible flow (QMCSF to QMAX) is shown here: (or a different limit chosen by the
chosen because the nature of the H-Q Step 1. The data mentioned in Table 1 user), then the speed is increased by
curve in this region for most machines for the available pump model(s) being a small increment, say 1 rpm, and the
usually rises continuously (pump evaluated are gathered and stored in next iteration is carried out for speed
head, H, is continuously increasing the computer, per the requirements = (NMIN +1).
with decreases in pump flow). This of the programming language. For Step 4. The process is continued until
Chemical Engineering february 2010 41
Engineering Practice

Case study

he following case study relates to a naphtha splitter application. were selected following a discussion with the VFD manufacturer’s
This study involves the purchase of a new pump, not the rerating technical representative [9]. For the purpose of comparing potential
of an existing or redundant one. Nevertheless, this case study power savings among the competing options (the last column of
demonstrates the efficacy of the algorithm discussed here for identi- Table 2), Option 4 was taken as a benchmark, although any of the
fying the best pump models and determining whether or not the use options could be used as the benchmark for such a comparison.
of a VFD is warranted, after processing a large volume of data to The negative sign shows power saving, while a lack of a nega-
support proper selection. The input data (attributes 1 through 21 of tive sign represents excess power consumption with respect to the
Table 1) for this case are based on standard pump series from two benchmark value.
of the approved vendors. From the options for FFD (Options 1 to 4, Table 2), it is clear that
The pump needed by the facility must accommodate two flows: to achieve the flowrate of 31 m3/h, the single pump will operate
103 m3/h at system head of 162 m, and 31 m3/h at system head at a head considerably higher than 70 m, which represents energy
of 70 m (Figure 3). While the pump must be operated at a flowrate loss during throttling.
of 31 m3/h most of the time, operation at 103 m3/h for smaller, The options for a pump scenarios involving a VFD (Options 5 to
periodic intervals, is also required. 17, Table 2) show that for the same flowrate, the operation at sys-
The exact operating hours for individual duty are required to cal- tem head of 70 m is possible, with substantial savings indicated by
culate LCC. The computer program based on the algorithm checked the negative sign. Thus, it is clear that for flowrate of 31 m3/h the
all the pump models available in the data bank and identified 13 VFD provides increased energy efficiency.
possible options with VFD. These are shown as item Numbers 5 to Looking at impressive savings of 14.95 kW (Option 5 and 14,
17 in Table 2. Table 2), the engineer may be tempted to recommend the same,
Table 2 shows all of the possible equipment scenarios — both with however, an overall review including other flowrates is essential.
and without a VFD — that were identified here. The data marked For instance, for the higher flow requirement of 103 m3/h, we do
with an asterisk refers to duty parameters that can be achieved not see very encouraging results with the addition of a VFD, except
by throttling the flow-control valve, while the data marked with a for Option 11 and 13 ,where savings of 3.89 and 0.4 kW, respec-
double asterisk represent the rated duty parameters required for tively, are observed.
pump selection. As noted in the main text, all options that are potentially viable
In Table 2, Option 4 (using a 37-kW motor) is taken as the from a technical standpoint should also be checked for economical
benchmark for comparison. The pumping liquid’s specific gravity viability as well, using the LCC analysis methodology; while LCC
is 0.63. analysis is not discussed here, it is described in Ref. 3. When car-
The higher flow requirement of 103 m3/h can be achieved in sev- rying out LCC analysis for cases involving operation at the same or
eral ways — by selecting rated flow of 34 m3/h (via three pumps different flows (here 31 m3/h and 103 m3/h), an estimate of the
in parallel), or 51.5 m3/h (via two pumps in parallel), or 103 m3/h number of hours of operation for each flow is critical to carry out
(single pump operation).The flow of 31 m3/h can be achieved with the correct estimation of energy consumption.
a single pump. Due to some specific constraints, engineers at the naphtha split-
The scenarios that required FFD versus VFD to achieve these flows ter described here ended up selecting Option 4 (Table 2), which
are also tabulated in Table 2. The VFD and corresponding motor entailed fixed-frequency operation (that is, no VFD). Nevertheless,

the matching speed or NMAX — which- efficiency value for the required flow operating point(s), the process is re-
ever comes first — is reached. If the is calculated, and the same is used peated for the rest of the models that
matching speed is not available within to calculate the power input to the are under consideration (that is, the
the range of NMIN to NMAX that is ap- pump. pumps whose data are available in the
plicable for the pump model being con- Step 6. If there is more than one oper- data bank).
sidered, then the use of a VFD will not ating point (m ≥ 2, as shown in Figures Step 8. The pump models that can
be suitable for this pump model. 3 and 5), then the feasibility of using best cover the duty parameters are
Step 5. Once the matching speed has the same pump model is checked for tabulated as output.
been established, the flow-versus-effi- all of the operating points, following Finally, once the candidate pump
ciency curve is plotted for the match- Steps 2 through 7. models have been identified using
ing speed. From this curve, using Step 7. After checking the feasibil- this methodology, the compatability of
quadratic equations for segments, the ity of the first pump model for all the using a VFD for each of those pump

D. K. Shukla is the senior vice-president with Deepak K Chaware is the rotating equip- Rajkumar B. Swamy is the senior process en-
Essar Oil Ltd, (Essar Oil Ltd., Refinery Division, ment engineer with Essar Oil Ltd. (Essar Oil gineer with Litwin PEL LLC (Litwin PEL LLC,
Essar House, 1st Floor, 34 KM Jamnagar Okha Ltd., Refinery Division, Essar House, 1st Floor, Emdad Complex, P.O. Box: 44715, Mussafah,
Highway, Vadinar 361 305, Gujrat, India.; Phone: 34 KM Jamnagar Okha Highway, Vadinar 361 Abu Dhabi, U.A.E; Phone: 00971-2-5070121; Fax:
0091-2833 661 444; Mobile: +91 9819730390; 305, Gujrat, India.; Phone: 0091-2833 661 444; 00971-2-5516254; E-mail: Rajkumar.SWAMY@
E-mail: He has Mobile: +91 9909992483 E-mail: Deepak.Cha- He has more than ten years of expe-
more than 30 years of experience in process and He received an M.Tech degree rience in the field of process and system engineer-
systems engineering in the hydrocarbon process- in mechanical engineering from Indian Insti- ing in petroleum refineries, oil-and-gas, and petro-
ing industries, including oil-and-gas production tute of Technology (Chennai). He has more than chemical complexes. He holds a graduate degree in
facilities, and petroleum refining and petrochem- 25 years of experience in the field of rotating chemical engineering from Gulbarga University
ical processes. His previous experience includes equipment design, testing and selection. He has (Karnataka, India). Prior to joining Litwin, his
working with Engineers India Ltd. (New Delhi), worked with several leading pump manufactur- previous experience included working with Foster
M.W. Kellogg (Houston) and BOC Gases (in both ers, including Jyoti Ltd. India, Sulzer Pumps Wheeler Asia Pacific Pte Ltd. (Singapore), Reliance
Murray Hill, N.J., and Guildford, U.K.). India Ltd., and KSB Pumps India Ltd. Engineering Associates Private Ltd. (Jamnagar,
India) and Hindustan Petroleum Corporation Ltd.
Mumbai Refinery (India).

42 Chemical Engineering february 2010

Table 2. A comparison of the FFD and VFD optioNS
Duty parameter Proposed Electric
No. of pumps Pump † Motor VFD Total Power
Pump BkW / motor input /
Option Drive = working Q, H, Speed, model efficiency, efficiency, efficiency, power, saving,
pump rating, pump,
+ standby m3/h m rpm % % % kW kW
kW kW
1 FFD 2 = 1 + 1* 31 164 2,900 80x360 34.00 25.71 37 92 100 27.95 27.95 4.77
    2 = 2 + 0** 51.5 162 2,900   50.00 28.69 37 92 100 31.18 62.37 6.68
2 FFD 2 = 1 + 1* 31 167 2,900 100x360 22.00 40.46 75 91 100 44.46 44.46 21.28
    2 = 1 + 1** 103 162 2,900   58.00 49.47 75 91 100 54.36 54.36 -1.33
3 FFD 3 = 1 + 2* 31 163 2,900 80x360 34.00 25.55 37 92 100 27.78 27.78 4.60
    3 = 3 + 0** 34 162 2,900   36.00 26.31 37 92 100 28.60 85.79 30.10
4 FFD 2 = 1 + 1* 31 168 2,950 80X50E 42.00 21.32 37 92 100 23.18 23.18 0.00
    2 = 2 + 0** 51.5 162 2,950   56.00 25.62 37 92 100 27.84 55.69 0.00
5 VFD 2=1+1 31 70 2,260 50X280 52.00 7.18 45 89 98 8.23 8.23 -14.95
    2=2+0 51.5 162 3,452   54.00 26.57 45 89 98 30.46 60.92 5.23
6 VFD 2=1+1 31 70 1,928 50X360 44.00 8.48 45 89 98 9.72 9.72 -13.46
    2=2+0 51.5 162 2,948   46.50 30.85 45 89 98 35.37 70.74 15.05
7 VFD 2=1+1 31 70 1,493 50X450 43.20 8.64 55 89 98 9.90 9.90 -13.28
    2=2+0 51.5 162 2,297   43.70 32.83 55 89 98 37.64 75.27 19.58
8 VFD 2=1+1 31 70 2,247 80X280 44.50 8.38 55 89 98 9.61 9.61 -13.57
    2=2+0 51.5 162 3,424   48.00 29.89 55 89 98 34.27 68.53 12.84
9 VFD 2=1+1 31 70 1,820 80x360 44.00 8.48 55 89 98 9.72 9.72 -13.46
    2=2+0 51.5 162 2,769   46.00 31.19 55 89 98 35.76 71.51 15.82
10 VFD 2=1+1 31 70 1,528 80X450 40.80 9.15 55 89 98 10.49 10.49 -12.69
    2=2+0 51.5 162 2,331   43.00 33.36 55 89 98 38.25 76.50 20.81
11 VFD 2=1+1 31 70 2,247 80X280 44.50 8.38 90 89 98 9.61 9.61 -13.57
    2=1+1 103 162 3,547   63.50 45.18 90 89 98 51.80 51.80 -3.89
12 VFD 2=1+1 31 70 1,820 80X360 44.00 8.48 75 89 98 9.72 9.72 -13.46
    2=1+1 103 162 2,890   58.20 49.30 75 89 98 56.52 56.52 0.83
13 VFD 2=1+1 31 70 1,528 80X450 40.80 9.15 75 89 98 10.49 10.49 -12.69
    2=1+1 103 162 2,410   59.50 48.22 75 89 98 55.29 55.29 -0.40
14 VFD 3=1+2 31 70 2,260 50X280 52.00 7.18 37 89 98 8.23 8.23 -14.95
    3=3+0 34 162 3,413   44.00 21.52 37 89 98 24.68 74.04 18.35
15 VFD 3=1+2 31 70 1,928 50X360 45.00 8.29 37 89 98 9.51 9.51 -13.67
    3=3+0 34 162 2,897   38.00 24.92 37 89 98 28.58 85.73 30.04
16 VFD 3=1+2 31 70 1,493 50X450 43.20 8.64 37 89 98 9.90 9.90 -13.28
    3=3+0 34 162 2,225   36.00 26.31 37 89 98 30.16 90.49 34.80
17 VFD 3=1+2 31 70 1,528 80X450 40.80 9.15 45 89 98 10.49 10.49 -12.69
    3=3+0 34 162 2,304   31.00 30.55 45 89 98 35.03 105.08 49.39
* Achieved by throttling. ** Rated duty for pump selection. VFD=Variable frequency drive FFD=Fixed-frequency drive. Specific gravity = 0.63
Option 4, which was ultimately selected with 37-kW motor, was taken as the datum for calculating power saving. “-” A negative sign indicates power
saving, while the absence of the sign indicates power loss. † BkW = brake kilowatt (similar to brake horsepower); BkW is used as we are using the meter-
kilogram- second (MPS) system, not the foot-pound-second (FPS) system.

models will need to be verified by the single pump catering to the number operating points with a single pump is
VFD, pump and motor vendors. of operating points, lying on differ- not possible.
From the procedure described above, ent system curves, with maximum ef- The LCC analysis is carried out for
it is clear that selecting VFD for Case ficiency may not be easily available. each of the selected models using the
1 (m = 1, Figure 1) will be relatively Under such circumstances, the user method mentioned in Ref. 3. Selection
simple compared to Case 2 and Case will likely be forced to select the pump is then finalized taking into consider-
3 (m ≥ 2, Figures 3 and 5). Selection that offers optimum efficiency, as ation the advantages, disadvantages,
for Case 3 will be more difficult, as a achieving the highest efficiency for all operational issues, constraints and
problems [1, 6, 7] related to the addi-
References tion of a VFD. It is essential that the
1. “Hydraulic Institute and Europump Variable 6. Nelson, W., Dufour, J., Problem-free pumping
Speed Pumping: A Guide to Successful Ap- system – A guide to holistic design, Chem. feasibility of the selected pump mod-
plications.” ISBN: 1-85617-449-2. Eng., January 1995. els for the new operating conditions be
2. Alfredsson, K., and Bokander, N., Life cycle 7. Case, G., and Marscher, W., Centrifugal checked. The checklist mentioned in
cost calculations applied: Constant or vari- pump mechanical design, analysis and test-
able speed operation, ABS Partners. 1/2002. ing, Proceedings of the 18th International Ref. 8 is a useful tool for doing so. ■
3. Hydraulic Institute and Europump, Pump life Pump Users Symposium. Edited by Suzanne Shelley
cycle costs: A guide to LCC analysis for pump- 8. Shukla, D.K., et al., Centrifugal pump se-
ing systems, 2001. ISBN: 1-880952-58-0. lection and upgrading for revised operating
4. Yongzheng, F., Kegi, W., and Yaqiao, C., In- conditions. Hydrocarbon Processing, May
fluence of back pressure on variable-speed 2005. Acknowledgement
The authors would like to thank the manage-
control, World Pumps, February 2005. 9. Personal discussion with Mr. P. P. Nanavaty ment of Essar Oil Ltd. (Mumbai, India) and
5. Paugh, J., Head vs. capacity characteristics of cen- GM-Marketing, Hi-Rel Electronics Pvt Ltd, Litwin PEL LLC (Abu Dhabi, UAE) for granting
trifugal pumps, Chem. Eng., October 15, 1984. India. permission to publish this article.

Chemical Engineering february 2010 43

Feature Report
Engineering Practice

Sizing, Specifying and Selecting

Centrifugal Pumps
Follow these tips to C-100 C-106

determine preliminary F-102 E-103

pump sizing, to support
cost-estimation efforts 12.5 ft
valve H-105
Filter Heat Furnace
exchangers Distillation
Asif Raza P-101 column
2 ft Pump 3 ft 22 ft
Zeton Inc. Grade

etermining the proper prelimi- Figure 1. Refer to this pump-system sketch for the pump-design calculations
nary size for centrifugal pumps discussed here. Note, a distance of 2 ft between the pump-inlet piping and grade and
3-ft distance between pump-discharge piping and grade is assumed during initial siz-
during the initial stages of any ing. These values should be confirmed once pump selection is done
pump-specification exercise
requires numerous calculations and clude calculation of suction pressure Horsepower and efficiency. In the
assumptions. This article reviews the and rated differential pressure. early stages of the project, you will
steps needed to size centrifugal pumps Net pressure suction head (NPSH) have to provide preliminary horse-
during the early stages of a project, to calculations. During the early stages power to the electrical engineer for
support initial cost-estimation efforts. of the project, the plant layout is not load calculations. From the rated dif-
Such early pump-sizing efforts are im- yet firm. Hence, NPSH available for ferential pressure, calculate the rated
portant steps toward final pump selec- the pump(s) cannot yet be calculated horsepower. Since the pump has not
tion and detailed engineering. with confidence. However, it is okay yet been selected, you can assume a
to carry out preliminary NPSH cal- pump efficiency between 50 and 60%.
Sizing centrifugal pumps culations using information from the Make a judgment call — 50% is typi-
Normal flowrate and rated flow- preliminary layout. Note that NPSH cally a sufficient value for calculating
rate. To define the normal and rated values can be increased by later modi- horsepower during the initial stages
flowrates, refer to the heat and mass fications to the layout. NPSH plays a of a project.
balance of your project. Normal flow- very important role during pump se- Shutoff pressure. Shutoff pressure is
rate is the flow at 100% capacity. lection, and could significantly impact required in early stages of the project
Rated flowrate is the design margin the cost of the pump if a lower-NPSH to determine the flange rating for the
that is added to the normal flowrate (required) pump is specified, since discharge piping of the pump. There
(typically on the order of 10–30%), to pumps with a lower NPSH require- are several ways to calculate shutoff
accommodate potential short-term ex- ment tends to be more expensive. pressure. A more conservative method
cursions in flowrate during operation. The goal is to calculate a preliminary uses the following formula:
Rated flowrate is usually defined by NPSH value and provide it to the
Differential pressure × 1.25 +
the user. If it is not defined, consider pump vendors to get initial feedback.
maximum suction pressure
10% for normal-service pumps, and This allows both parties to determine
20–30% for critical-service pumps, whether a pump with the specified Maximum suction pressure is
such as distillation column reflux NPSH can be achieved or not, can be calculated by:
pumps, reactor feed and furnace-feed achieved with some modification to Pressure safety valve (PSV) set pres-
pumps and other feed pumps that play the plant layout, or can be achieved sure × 10 – 21% accumulation + static
a critical role in the overall process. by selecting a pump with lower NPSH head based on high liquid level
Pressure-drop calculations. The requirements. Based on the vendor’s If a user does not have information
intent of these calculations is to de- feedback, modify your layout to have a such as the PSV set pressure in the
termine suction pressure and rated design margin between the available early stages of the project, a simple
differential pressure. Please refer and required NPSH. A design margin approach for calculating shutoff head
to sample calculation #1 (discussed of 3–4 ft is widely accepted, per indus- is to use this formula:
below) and Figure 1 for performing try standards. See sample calculation
pump-sizing calculations, which in- #2 for calculating NPSH (available). Rated differential head × 1.5

Chemical Engineering FEBRuary 2013 43

TABLE 1. Pump-sizing spreadsheet for calculating suction
pressure and rated differential pressure (Calculation #1)
Engineering Practice Rated flow Normal flow
condition condition
Suction pressure
opecomm Source pressure, psig 30 30
rat en
Head produced by pump

Impeller 1 ing ded Static head = 10.5 ft (ft x specific gravity/2.31), 3.2 3.2
ge* psi
Impeller 2
Suction line loss, psi 0.3 0.3
Impeller 3 Pump suction pressure, psig (30+3.2) (30+3.2)
–0.3 = 32.9 –0.3 = 32.9

Best efficiency
Discharge pressure
point (BEP)** Delivery pressure, psig 100 100
Static head = 19 ft (ft x specific gravity/2.31), psi 5.8 5.8
Flowrate produced
Line loss, psig 33 23
* The recommended operating range is between
70–120% of the BEP flowrate Control-valve pressure drop, psid 38 83.5
* * BEP is a point on the curve, is specific to each
pump, and is provided by the pump manufacturer Filter pressure drop, psid 14.4 10
Heat exchanger 1 pressure drop, psid 14.4 10
Figure 2. This pump-performance
curve shows the best efficiency point Heat exchanger 2 pressure drop, psid 14.4 10
(BEP) and the recommended operating Furnace pressure drop, psid 72 50
range, which are helpful during pump
selection. Note the shaded region rep- Orifice flowmeter pressure drop, psid 2.88 2
resents the operating range for different Contingency, psig 10 10
impeller sizes
Differential pressure, psig
Specifying centrifugal pumps Discharge pressure, psig =100+5.8+3 =100+5.8+2
The following steps must be carried 3+14.4+14. 3+10+10+10
4+14.4+72+ +50+2+83.5
out during the specification of centrif- 2.88+38+10 +10=305
ugal pumps: =305
Gather basic process data. Fill out
Minus suction pressure, psig 32.9 32.9
the pump data sheet with the follow-
ing basic information: Suction pres- Total pump differential pressure, psig 272 272
sure, normal and rated discharge Pump head (psi x 2.31/specific gravity) 900 900
pressures, normal flowrate and rated Hydraulic power = gal/min x pump head (ft) x 382 318
flowrate, fluid properties, such as den- specific gravity/3,960, hP
sity, viscosity and vapor pressure at Efficiency at 3,600 rpm, % 60 60
operating temperature.
Rated power = Hydraulic power/efficiency, hp 637 530
Determine a preliminary value
for NPSH. Please refer to the NPSH
calculations presented below (sample regarding whether to use a sealed or must be based on the end user’s re-
calculation #2). sealless design pump. Consider seal- quirements, process conditions, and
Specify desired materials of con- less pumps for liquids that are flam- the cost of the equipment. In general,
struction. This should come from the mable, toxic and corrosive. For in- the cost increases in this order: Non-
end user or from a metallurgist who is stance, many facilities use traditional ANSI, sub-ANSI, ANSI and API, while
involved in the project and is respon- sealed pumps for pumping water, and non-ANSI pumps have the lowest cost.
sible for specifying the appropriate sealless pumps for pumping acids and A non-ANSI pump usually finds its ap-
metallurgy of the equipment. alkalis and other corrosive liquids. plication in small sizes handling less
Specify sealless versus mechanical Work closely with your vendor and critical service, such as water that is
seals. Magnetic-drive sealless pumps seek guidance based on previous being pumped at relatively low pres-
are desirable for many applications experience in applications with sure and low temperature. An ANSI
since they eliminate the need for me- similar service. If a sealless pump is pump is usually used in applications
chanical seals, and thus eliminate the not available, consider using a double requiring relatively larger sizes (for
inherent risk of leakage and mainte- mechanical seal to minimize the risk instance, more than 10 hP) in chemi-
nance associated with such seals. But of leakage. Rely on the vendor’s expe- cal or hydrocarbon service.
sealless pumps also have drawbacks, rience, as well, in selecting the most However, some ANSI pumps may be
such as an inability to handle larger appropriate mechanical seal for your limited to a maximum casing pressure.
particles in the process fluids. And service. Provide vendors with as much For higher-casing pressures, the user
in some applications, relatively high process data as you have, to ensure may have no choice but to consider a
differential pressure requires high proper seal selection for your service. custom-made pump or an API pump.
torque, which may be beyond the ca- Classify non-ANSI, sub-ANSI and API pumps tend to be considerably
pabilities of the sealless pump. ANSI–API pumps. Ultimately, the more expensive than ANSI pumps, as
Rely on the end-user’s experience selection of specific centrifugal pumps they are typically used in hydrocarbon
44 Chemical Engineering FEBRuary 2013

NPSH– required, ft
Rated flow Efficiency
90 360

80 340 Head
BEP The pump-sizing calculations also
70 320
Head, ft

provide pressure-drop data across the

60 300 control valve, under normal- and rated-
Efficiency, %

50 280 Allowable flow circumstances. Specify control-

operating region valve pressure drop at the rated flow,
40 260
Preferred following the widely accepted rule of
30 240 operating region 150 thumb — that is, pressure drop is 25%

Power, BHP
20 100
of the dynamic head loss at rated flow.
Power In this case, dynamic head loss at the
10 50
rated flow is 151 psig (the sum of pres-
0 0 sure drop across filters, heat exchang-
0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 1,300
ers, furnaces, orifices and line losses.
Flowrate, U.S. gal/m
Hence the differential pressure across
Figure 3. This figure shows a relationship between flowrate and head, power, effi- the control valve in this scenario at
ciency and NPSH. Note the NPSH curve is bowl-shaped. NPSH requirements increase rated flow is 38 psig).
with flow. Make sure you analyze this value carefully and select a NPSH (required)
Note: If the calculated pressure drop
value that corresponds to the rated flowrate
across the control valve is less than 10
service involving high temperatures Sample calculations psig, use a minimum value of 10 psig
and very high pressures. Such pumps Consider the pump system sketch for the control valve at rated flowrate.
are widely used in refinery service. shown in Figure 1. A fluid with a vapor Now adjust the pressure drop across
Specify design pressure and tem- pressure of 45.9 psia at operating tem- the control valve at normal flow and
perature. Specifying design pressure perature of 430°F with a viscosity of try to match the discharge pressure
and temperature based on the design 0.5 cP is pumped at normal flow of until it is equal at normal and rated
conditions of the pump suction vessel 2,000 gal/min. The specific gravity is flow. Pressure drop is directly propor-
may seem to be the easiest approach. 0.7 and the delivery pressure is 100 tional to the square of the flow, hence
But check for any possible upset condi- psig. The operating pressure of col- pressure drop across rated flow is cal-
tions that might warrant an increase umn C-100 is 30 psig. The atmospheric culated using the following formula
in design pressure and temperature, pressure at site is 14.5 psia. Assume (Note that the rated flow is 1.2 times
such as a new process stream entering a rated flowrate of 2,400 gal/min (1.2 the normal flow):
the pump suction in a different pro- times the normal flowrate).
cess operating mode. Note that two sets of calculations Filter pressure drop at normal flow =
Specify the motor requirements. are done for calculating horsepower 10 psig
The required power supply, whether — one for normal flowrate and other Filter pressure drop at rated flow =
it is 460 V at 60 Hz, or 230 V at 50 for rated flowrate. Refer to Figure 1 10 × (1.2 × 1.2) = 14.4 psig
Hz, must be supplied by the end user. for calculating static head. Pressure
Lastly you have to specify the hazard- drop across filters, heat exchangers, In reality, the discharge pressure at
ous area classifications and tempera- orifice meters and furnaces are taken normal flow and rated flow may not be
ture rating of the motor. Check with from actual equipment vendor quotes. the same, but the two values will be
your electrical engineers to identify These quotes may be available from very close. A pump is designed to oper-
the required hazardous area classi- different disciplines, such as mechani- ate at rated flow conditions. However,
fication. Motor temperature rating cal and instrumentation departments. a pump operates at normal flow most
should also come from the end user. If vendor quotes are not available of the time during normal operation.
This is based on the lowest autoigni- during preliminary pump sizing, then
tion temperature of the components assumptions must be made based on Normal operation
involved in the process. Keep in mind, interactions with other disciplines. During normal operation, because the
most of the time, a motor rating up to For instance, it is okay to assume a flow is lower than the rated flow, the
T3A is available at no added cost, but pressure drop of 10 psid across heat pump will try to develop more head.
the cost increases substantially if a exchangers or a pressure drop of 5–10 During this scenario, the control valve
motor with a rating of T4 or higher is psid across a filter. Pressure drop will start closing and will consume
specified. across a vessel filled with catalyst more pressure drop. This will have the
Next, you need to specify whether should be calculated using the Ergun effect of moving the pump back onto
you want a fixed-speed motor or a equation. These pressure-drop values the pump curve.
variable-speed drive motor. If you are are finalized when equipment design Here, you will notice that pressure
controlling the pump flow with a speed is finalized and are used for final pump difference in rated flow is 38 psig
controller, then you must select an sizing, during the detailed engineering (specified by the designer) and is 83.5
inverted-duty, variable-speed motor. phase, to check the rated differential psig in normal flow. The difference be-
However, note that if there is a control pressure and rated brake horsepower. tween these two values is the excess
valve on the pump discharge, then you It is helpful to perform these calcula- dynamic head between normal flow-
must use a fixed-drive motor. tions using an Excel spreadsheet. rate and rated flowrate. The relation-
Chemical Engineering FEBRuary 2013 45
Head capacity curve
Operating points Control valve
pressure drop
Engineering Practice

Head at normal flow

Total head, ft ∆P2
ship between the pump’s head-capac-

Head at rated flow

ity curve and pipe-system relationship
resistance is shown in Figure 4. resistance ∆P2
Pressure drop across the control ∆P1 Dynamic
valve should not be included as a part head loss
of the dynamic head loss. The gap be- Static and discharge
pressure head ∆H piping
tween the head-capacity curve and the differentials
system-resistance curve is available
for throttling (control-valve pressure Capacity (Q), gal/min
drop). Control valve pressure drop at
normal flow is higher than the pres- Figure 4. This figure shows the relationship between a pump's head-capacity
sure drop at rated flow. While pipe curve and pipe system resistance (dynamic head loss). It also shows pressure drop
dynamic head loss increases at higher across a control valve at normal and rated flowrate. Note: Pressure drop across the
flowrates (rated flow), control-valve control valve at rated flow is less than the pressure drop at normal flow
pressure drop decreases. At higher
flowrates, the control valve has to Pump selection sure. Analyze all quotes to see whether
open more and pass larger flow with During this stage of the project, you they meet your specified value of rated
less resistance. should be getting quotes back from flow and differential pressure.
Designers should appreciate the the vendor, based on the pump speci- Material of construction. Does this
importance of specifying the correct fications you have provided. While meet your specified material of con-
pressure drop for the control valve various quotes will vary in their dol- struction?
at different flow conditions, to en- lar value, keep in mind that a more- Analyzing pump curve and effi-
sure a rugged system design. If a sys- expensive pump does not necessarily ciency. Pump efficiency is a very im-
tem is poorly defined, the pump will mean that it is the best pump for the portant value to be considered. Some
never be able to control the flow and job, and the least-expensive pump is vendors may quote a bigger pump
it will never provide proper flow at not worth further consideration. To as- than what is required. In such a case,
the required head. The efficiency will sess competing quotes fairly, develop a the pump efficiency will be reduced.
be low and the pump will consume spreadsheet to gather the following se- Designers should note that a pump
more power. lection criteria. Assign points to each with even 10% higher efficiency will
It is also advisable to install a globe item that meets your specifications. save thousands of dollars in power
valve at the pump discharge, to allow NPSH. Check for NPSH (required) costs over the service life of the pump.
for throttling the flow and adjusting from the pump data sheet provided by It is good practice to examine sev-
the flow and discharge pressure. How- the vendor. How close it is to your esti- eral performance charts at different
ever, please keep in mind that the in- mated value of NPSH (available)? Ask speeds to see if one model satisfies the
stallation of a globe valve will incur a yourself — Can you make this pump requirements more efficiently than an-
constant pressure drop, which must be work by increasing NPSH (available)? other. Whenever possible, the lowest
accounted for during head-loss It cannot be stressed enough that pump speed should be selected, as this
calculations. NPSH is a key parameter during pump will save wear and tear on the rotat-
Ultimately, the calculated control- selection, and insufficient NPSH often ing parts. Efficiency can be found on
valve pressure drop at normal and results in pump cavitation. the pump curve provided by the pump
rated flows will be given to the in- Cavitation occurs when vapor bub- vendor. Refer to Figure 3, which shows
strument engineer who is responsible bles that have formed in areas of low the relationship between efficiency and
for specifying and sizing the control static pressure move along the impel- flowrate. This figure also shows the re-
valves for your project. ler vanes into high-pressure areas, lationship between volumetric flowrate,
Sample calculation #1. Table 1 where they rapidly collapse. The forces head, NPSH and brake horsepower.
shows the results of a pump-sizing ex- produced by these bubbles as they im- Every pump has a best efficiency point
ercise, in which suction pressure and plode erode the impeller vane, result- (BEP), which is the flow/head combina-
rated differential pressure were calcu- ing in progressive pitting to the impel- tion that corresponds to the highest ef-
lated. lers. As a rule of thumb, an acceptable ficiency. The preferred operating region
Sample calculation #2. Static pres- margin between NPSH (available) and is between 70 and 120% of the BEP
sure available at the pump suction NPSH (required) is required to ensure flowrate value [1], although most users
inlet = (Operating pressure of the pump reliability. A minimum margin require the rated flow to fall between
vessel + static head) – suction-piping of 3–4 ft is a widely practiced rule of 80% and 110% of BEP. The allowable
head loss at rated flow. thumb. Since the NPSH requirement operating region varies from pump to
NPSH (available) = Static pressure at increases with increasing flow, it is pump, and is defined as the flow range
the pump suction inlet – vapor pres- important to consider the maximum within which vibrations do not exceed
sure at the operating temperature. expected flow when specifying an ac- the limits established by the American
The results of these sample calcula- ceptable NPSH margin. Petroleum Institute (API) [1].
tions are shown in Table 2. Rated flow and differential pres- Refer to Figure 2, for the recom-
46 Chemical Engineering FEBRuary 2013
Atmospheric pressure, psia 14.5
Specific gravity (SG), dimensionless 0.7
Original pressure (vessel pressure), psig 30 Get Chemical
Static head = 10.5 ft (Note 1) (10.5/2.31) x 0.7 = 3.18 psig
Suction line loss, psig 0.3 Engineering’s
Suction pressure at pump inlet flange, psia =30 + 3.18 – (0.3)
= 32.88 + 14.5 = 47.38 psia
plant cost index
Vapor pressure, psia 45.9 to improve plant
Net pressure suction head (NPSH) = Suc- 47.38 – 45.9 = 1.48 psia
tion pressure – vapor pressure, psia cost estimates…
NPSH available, ft = 1.48 x 2.31/0.7 = 4.89
Note 1: Static head is measured from the low liquid level in the vessel to the center line of the and delivered in
pump-suction flange, or from the vessel bottom nozzle to the center line of the pump-suction
flange. The latter is more conservative approach.
advance of the
mended operating range. The shaded tion-oil coolers, interconnecting piping
region represents the operating range, between the coolers, instruments, such print edition!
that is discussed in the above para- as flow switches on the cooling water
graph. Note in this figure, there are lines and so on. Be sure to compare all For more than 37 years,
three curves for three different impel- the quotes on the same basis. chemical process industries
ler sizes provided by the pump vendor. Price. Before selecting a particular
professionals- engineers,
While selecting an impeller, it is good pump, make sure that you are com-
practice to select a pump with an im- paring “apples to apples,” as different manager and technicians,
peller that can be increased in size, as vendors may have quoted different have used Chemical
this will allow for future increases in pump options in different styles, with Engineering’s Plant Cost
head and capacity. different seal arrangements, using dif- Index to adjust process
Mechanical seal arrangements. ferent assumptions and so on. plant construction costs
When evaluating competing vendor
from one period to another.
quotes, be sure you are comparing Final thoughts
“apples to apples.” For instance, some Close coordination with the pump This database includes all
vendors may have quoted a double me- vendor and developing a solid under- annual archives (1947 to
chanical seal, while your requirement standing of the process requirements
present) and monthly data
was for a single seal. If this happens, are essential steps during pump de-
ask the vendor to revise the quote. sign and selection. By understanding archives (1970 to present).
Motors. Check for the motor sizing, the concepts of rated flow, head, suc- Instead of waiting more
and whether it has been sized for full tion pressure and NPSH, and by un- than two weeks for the
run-out case. Full run-out means that derstanding pump curves, you will be print or online version of
the motor should be sized for the max- on the right track to design and select Chemical Engineering to
imum flowrate the pump can deliver. the most appropriate pump that meets
The stated motor temperature rating all of the process requirements. ■ arrive, subscribers can
and specified electrical area classifica- Edited by Suzanne Shelley access new data as soon as
tion must meet your requirements. it’s calculated.
Physical size of the pump. Check Reference
for the dimension of the pump from 1. American Petroleum Inst. (API), “Centrifugal Resources included with
Pumps for Petroleum, Petrochemicals and
the quotes received. If space is tight, Natural Gas Industries,” 10th Ed., API Stan- Chemical Engineering’s
you may have to consider an inline dards 610 (ISO 13709), Washington, D.C.
Plant Cost Index:
pump, or a high-speed, single-stage
pump over a multistage pump. Author • Electronic notification of
Asif Raza is a lead process monthly updates as soon
Design conditions. Check for design engineer at Zeton Inc. (740
temperature and pressure from the Oval Court, Burlington, ON as they are available
L7L 6A9, Canada; Phone:
vendor quotes and make sure that 905-632-3123; Email: asi- • All annual data archives His
they meet your requirements. work involves the design and (1947 to present)
Design codes. Does the quoted pump manufacture of pilot and
demonstration units for re- • Monthly data archives
meet your specified design codes, such search and development, in
(1970 to present)
as API, ANSI and so on. the areas of petroleum refin-
ing, petrochemicals, gasifica-
Warranty. When evaluating compet- tion and unconventional sources of energy, such • Option to download in
ing pump options, check for the manu- as biomass, gas-to-liquids and coal-to-liquids. Excel format
He has more than 15 years of experience in pro-
facturer’s performance warranties and cess design. His areas of interest include major
equipment sizing and selection, P&ID develop-
for the availability of onsite startup ment, selection of control logics and process
assistance from the vendor, if specified simulation. Before joining Zeton, Raza worked Subscribe today at
as a lead design engineer with engineering, pro-
in the Request for Quotation (RFQ). curement and construction (EPC) companies, in-
Scope of supply for auxiliary cluding Bantrel Inc. and SNC Lavalin Engineers
and Constructors. He holds a B.Tech degree in
equipment. Make sure the vendor’s chemical engineering from Amravati University 17817
quote includes the supply of all the (India). Raza is a registered professional engi-
neer in the province of Ontario, and is a member
accessories required, such as lubrica- of Ontario Society of Professional Engineers.

Chemical Engineering FEBRuary 2013 47

Feature Report
Engineering Practice

Machinery Figure 1. In an oil-injected, twin-screw compressor, the lu-
Follow this guidance to improve bricating oil is injected into the gas stream to absorb the heat
of compression and act as lubricant and sealant. This enables
lubricant selection, process a much higher pressure ratio in a single stage and provides
significant protection against corrosive gases. In multi-stage
machines, inter-cooling is usually not required
operation and asset reliability
Amin Almasi compressors or pumps, the lubricant lent wear protection, excellent lubric-
WorleyParsons Services Pty Ltd. is in contact with moisture (from the ity. They should be non-poisoning to
handled fluid), so the lubricant must catalysts, since some oil will be trans-

ubricants in rotating machines have good demulsibility (that is, resis- ferred downstream by the gas.
reduce friction and wear, dissi- tance to emulsification, or good resis- For reciprocating or screw compres-
pate heat, protect surfaces, keep tance to mixing with water). sors, the lubrication oils that meet all
out foreign contaminants and re- Today, the overwhelming majority of these criteria are mainly synthetic
move wear particles. Commonly used of compressors and pumps are best lubricants formulated with PAG base
liquid lubricants fall into two main served by premium-grade oils with stock. Overall, these provide excellent
categories: mineral (petroleum-based) ISO VG 32 or 46 (sometimes ISO VG oxidative and thermal stability, which
oils and synthetic oils. 68 or 100 grades). However, there are are particularly important for high-
Mineral oils are produced by refining many different types of compressors temperature applications. Relatively
crude petroleum. They usually contain and pumps, and each manufacturer is high viscosity indexes facilitate low-
trace amounts of some unwanted sub- likely to recommend only those lubri- temperature startup and help to main-
stances. By contrast, synthetic oils are cants that have been used successfully tain acceptable viscosity over a wide
engineered, so their properties can be before. Occasionally, compressor lubri- temperature range.
more tightly controlled. Synthetic oil cants have to be formulated for excep- PAG lubricants are highly stable,
lubricants include synthesized hydro- tional severe-service performance. even at sustained high temperatures,
carbons (such as polyalphaolefin, or Reciprocating and screw compres- and thus have very low deposit-form-
PAO), organic esters (such as diesters sors. During operation, lubricants used ing tendencies. And importantly, any
and polyol-esters), polyglycols, polyal- in screw compressors and in the cylin- decomposition products that may form
kylene glycol (PAG), phosphate esters ders of reciprocating compressors are tend to be soluble in the lubricant and
and silicone lubricants. in direct contact with compressed gas. thus do not tend to separate as sludge
Mineral oils are more frequently With conventional mineral oils, such or contribute to the formation of var-
used in chemical process industries gas can become dissolved in the oil. nish or lacquer. Similarly, because of
(CPI) applications, but the impor- Additionally, any oil that becomes dis- their close contact, it is important to
tance and use of synthetic oils has solved in the gas can be carried away, select lubricants that are compatible
been steadily increasing in recent depleting the lubricating film. This can with the elastomer and coatings used
years. In general, they offer superior result in machine component scoring in the compressor’s wetted parts.
performance in terms of higher oxida- and higher wear rates. Specially for- Minimizing lubricant carryover to
tion stability, improved corrosion re- mulated synthetic lubricants generally downstream discharge streams is im-
sistance, and the ability to withstand perform better in these instances. portant for any compressor, particu-
both higher and lower temperatures. When selecting a lubricant for re- larly for screw compressors and recip-
ciprocating or screw compressors, low rocating compressors. In general, gas
Compressors and pumps solubility in compressed gas should be solubility increases roughly linearly
Some compressors operate with gas a key selection criteria. The selected with increasing pressure. As a rough
discharge temperatures exceeding lubricants should also have relatively indication, PAG lubricants typically
160°C. This calls for a lubricating oil high viscosity indexes, minimal vis- have a gas solubility that is less than
that has good oxidation properties and cosity loss, good thermal stability over half of that of mineral oils or PAO syn-
thermal stability. Meanwhile, in some an extended temperature range, excel- thetic oils. High-pressure reciprocat-
44 Chemical Engineering January 2012
Turbine lubrication oil system

FIGURE 2. the same lubricant is often used for

To A typical turbine
ACP: Alternative-current,
bearings as well as internal compres-
motor-driven oil pump
lubrication oil
system TGOP: Turning-gear oil pump
EOP: Emergency oil pump system requires sor lubrication.
JOP: Jacking oil pump
MOP: Main shaft-driven a combined oil Inside screw compressors, the lubri-
oil pump
system that sup- cant is used to lubricate mating metal
Lubrication plies lubrication
oil cooler surfaces and seal compression cham-
oil for bearings
and hydraulic oil
bers (which form as a result of screw
Oil ejector for the turbines mashing) and to cool the compressed
gas. Lubricant films seal and lubricate
all contact lines between male and fe-
ACP TGOP EOP JOP male screws.
Oil viscosity plays an important role
ing compressors (that is, those above cylinders, one of the most important in ensuring machine performance and
300 bar, up to several thousand bar) factors is the rate of oil feed. In general, minimizing power losses. The selected
need special attention in this regard. the risk of over-lubrication is greater lubricants should maintain a depend-
Another important consideration is than that of supplying too little oil. Be- ably strong lubricant film to provide
water solubility. PAG lubricants show cause many problems associated with dependable wear protection, ther-
less than 20% of the water solubility reciprocating compressor operation mal stability, and appropriate viscos-
of typical mineral oils. Generally, min- can be overcome by preventing exces- ity across the temperatures they are
eral oils (or petroleum oils) are complex sive lubrication, proper control of the likely to encounter during operation
mixtures of naturally occurring hydro- supply of oil to the cylinder is the key. (Figure 1). At the same time, the vis-
carbons, but synthetic base fluids have Nearly 45% of reciprocating com- cosity should be appropriate to limit
a controlled molecular structure with pressor shutdowns are due to cylinder power losses.
predictable and advanced properties. valve and unloader problems. About Dynamic compressor and pump lu-
When it comes to reciprocating com- 20% of reciprocating compressor shut- brication. Dynamic compressors and
pressors, the traditional approach has downs result from packing or piston pumps are machines that achieve a
been to ensure that the lube oil that ring problems. Proper selection and pressure rise by adding kinetic energy
is supplied to the crankcase (bearing) use of cylinder and packing lubricant and velocity to a continuous flow of
is the same as the lube oil supplied to is essential to improve the perfor- fluid. These machines deliver relatively
the cylinder and packing. The goal has mance of cylinder valves, unloaders, large volumes of fluid at relatively low
been to keep lubricant types to a small packings, piston rings and bearings. or medium pressure increases. Com-
number, to bring down costs and avoid Operators should examine how mon types of dynamic machines include
lubricant misapplication. However, in closely the applied lubricant’s feedrate centrifugal and axial compressors
a majority of cases, it is more appro- meets the actual cylinder packing’s and pumps.
priate to provide a different lubricant lubricant requirements. This can be Optimal lubricants for dynamic com-
to the cylinder and packing working done by examining internal surfaces, pressors and pumps are usually pre-
space to improve overall performance. such as cylinder walls. When properly mium extreme-pressure (EP), multi-
For instance, lubricants for the lubricated, these surfaces should be purpose oils designed for dependable
crankcase and crankshaft mechanism covered with a thin film of oil. There performance over a wide range of tem-
should be suitable for the particular should be no evidence of oil accumula- peratures and operating conditions
bearing application. In many cases, tion. If the cylinder surfaces are wiped (extreme-pressure lubricants are oils
this will involve using a suitable min- with a piece of paper, oil should stain that can work under shock or sudden
eral oil (or sometimes a synthetic lu- paper evenly, but should not soak in. high pressure rises). The most ap-
bricant that is suitable for bearing ap- If the paper is dry or unevenly spot- propriate lubricants for dynamic ma-
plications) in relatively large volumes ted, the feedrate is too low. If the paper chines tend to be synthetic lubricants
in a closed lubrication loop. is saturated, the feedrate is too high. formulated from PAG base stock, and
On the other hand, cylinder and Lubricant feedrate should be adjusted that have dedicated anti-wear, severe-
packing lubricants must be suitable to provide no more than the minimum service and long-life properties.
for use in contact with gases. In many lubricant requirement. Specifically, when specifying a lu-
cases, separate lubricant-supply sys- There is violent mixing of lubricant bricant for dynamic machinery, users
tems are used for working space (cyl- and gas inside rotary screw compres- should opt for ones that have the fol-
inder packings) and running compo- sors (oil-flooded screw compressors). lowing characteristics:
nents (crankshaft bearings), to enable In general, screw compressors need 1. Superb oxidative and thermal sta-
the most optimal lubrication for each. lubricants with extra oxidation re- bility
Synthetic lubricants with ISO VG sistance to ensure long service life in 2. Relatively high viscosity indexes
100 to 220 ratings are extensively used closed circulatory systems. They also (values depend on the application
for reciprocating compressor cylinder- require lubricants with lower viscos- and can vary considerably)
packing for process services. In the lu- ity compared to cylinder-packing lu- 3. Relatively low pour points for easy
brication of double-acting compressor bricants. In rotary screw compressors, cold temperature startup (again,
Chemical Engineering January 2012 45
Engineering Practice

optimal values depend on applica- tended as a general guideline. If

tion and can vary considerably) the limit is passed, the lubricant
4. Excellent lubricity for enhanced should be replaced, and the prob-
resistance to friction and wear lem root-cause should be studied
5. Extreme pressure lubrication so that required corrective action
6. Good resistance to mechanical may be taken at the first oppor-
breakdown tunity: FIGURE 3. Tilting-pad bearings are com-
monly used for high-speed rotating machines.
7. Good resistance to sludge and var- • Total acid number increase — Each bearing consists of a series of pads.
nish formation The warning limit is 0.3 mg Because oil is basically incompressible, pres-
8. Non-corrosive and stain resistant KOH/g) sure builds within the oil film, which provides a
9. Suitable compatibility with elasto- • Water content — The warning means for the oil film to transfer the load
mers and coatings (particularly seal limit is 0.2%)
system components, gear unit inter- • Cleanliness — It is necessary to find in the case of leakage. Operators
nal paints and so on) the source of particulates (for exam- should adjust or select oil operating
Lubricants with ISO VG 32, 46 and 68 ple, makeup oils, dust or ash entering pressures greater than the cooling-
grades are commonly used in dynamic system, wear and so on), so that steps water’s operating pressure.
compressors and dynamic pumps. can be taken to address the problem 3.  The main contributor to entrained
• The Rotary Bomb Oxidation Test water in the oil system is steam by-
Turbine lubrication (RBOT) warning limit is less than passing the steam seals, and subse-
Turbine lubricants must have excellent half of the test result value of the quently mixing with the lubricant
thermal and oxidation resistance at original oil (RBOT is a method of oil. This is particularly prevalent in
bearing oil temperatures, which can ap- comparing the oxidation life of lu- steam turbines with high back pres-
proach 100°C in typical steam turbines bricants. For more information refer sure or high first-stage pressure, once
or heavy-duty industrial gas turbines, to ASTM 4378 and ASTM D-2272) the seals are worn. It is good practice
and can exceed 200°C in aero-deriva- Steam turbines. A steam-turbine to provide air purge connections on
tive gas turbines. Turbine lubricants oil system is usually required to pro- the bearing seals of steam turbines.
must control the rust and corrosion vide oil for bearings, trip-and-throttle Dry instrument air will provide posi-
that could destroy precision surfaces, valves, governor systems, power cyl- tive pressure in the lube oil area.
and resist foaming and air entrain- inders and similar accessories. Trip- Gas turbine lubrication. In general
ment, which could impair lubrication and-throttle valves have two major there are two classes of gas turbines:
and lead to equipment breakdown. functions — as an emergency shut-off 1.  Heavy-duty gas turbines. Lubricant
Such lubricants must also have suit- valve to trip the steam turbine (that is, selection for these types of gas tur-
ably high viscosity indexes that allow to cut steam flow immediately at the bines is similar to the selection for
more uniform lubricating performance inlet) and to admit and throttle steam steam turbines. Standard compo-
over a wide range of ambient and op- to the steam turbine, particularly dur- nents of these turbines are fairly
erating temperatures. They should ing startup. The use of a combined oil massive and the bearings are typi-
also be easily filterable without addi- unit — one that provides both lubri- cally located at some distance from
tive depletion (additive separation or cation oil for bearings and hydraulic the heat sources. In most cases,
sludge formation). control oil for trip-and-throttle valves, petroleum-based lubricants perform
Turbine lubricants should be ver- governors and similar applications — suitably for these gas turbines.
satile, able to serve as both lubricat- is very common. 2.  Lightweight aero-derivative gas
ing oil and hydraulic fluid for various Steam turbine lubricants must turbines. These turbines tend to be
turbine systems, generators, driven readily shed any water that becomes compact, as they are mainly based
equipment, gear units and other aux- entrained during operation. Water in on aircraft gas turbine engines. As
iliary components. The goal is to sim- the steam-turbine train’s oil reservoir a result, size and weight are ex-
plify lubricant inventories to a rela- typically comes from one of the three tremely important, and the bearings
tively small number of multi-purpose following sources (A water analysis are typically located relatively close
products, thereby minimizing the can usually determine the source): to sources of heat. Aero-derivative
chance of potentially costly lubricant 1. Simple condensation from air within gas turbines tend to require that the
misapplications. Products of interest the reservoir can be minimized by oil not only lubricates under more-
for turbine operators are ISO grades maintaining the manufacturer’s severe thermal and oxidative condi-
32, 46 and 68. specified oil level within the reser- tions, but that the oil also serves as
Steam and gas turbine oils are ex- voir and maintaining good ventila- a heat transfer fluid, to carry heat
pected to provide years of trouble-free tion around the turbine train. away from the bearings and shafts.
operation (Figures 2 and 3). In-service 2.  A leak in the shell-and-tube oil Additionally, aero-derivative gas tur-
monitoring of turbine oils is a valuable cooler(s) may allow cooling water bines are subjected to repeated and
means of assuring optimum oil perfor- into the oil loop. If the oil pressure is rapid starts, as well as hot peaks. The
mance and extended turbine life. The greater than the water pressure, oil extreme operating conditions of aero-
following recommendations are in- will be forced into the cooling water derivative gas turbines generally re-
46 Chemical Engineering January 2012
Case studies
Hot gas fans at a very large refinery complex. At this facility, recommended oil). The bearings of both pumps and motors were
various hot gas fans were experiencing frequent bearing fail- failing because of lubricant starvation.
ure. The fan manufacturer originally recommended mineral oil. In addition to the downtime costs, significant labor and hardware
With the bearing housing typically reaching 130°C, the cooled costs totalling $300,000 were required to restore the unit to nor-
and filtered mineral oil would still overheat to the point of cok- mal operation. After extensive evaluations, a synthetic wax-free
ing. Several bearing failures occurred less than five months lubricant was used in place of the mineral oil. Since making this
from refinery startup. The solution was to switch to a properly change, no cold weather plugging has been experienced.
formulated synthetic oil, which provided superior high-
temperature capabilities. Large gear unit in a power-generating station. The unit was hav-
ing lubrication problems with its originally selected petroleum-
Mid-size gear unit of a special fan unit in a petrochemical plant. based lubricant. Specifically, the following was experienced:
Using mineral oil (ISO VG 160) results in some excessive wear • The lubricant was losing viscosity and had to be changed every
and oil oxidation. Lubricant drain intervals of four months and a four months
gear unit life less than two years were reported. Changeover to an • Air entrainment in the lubricant was causing cavitation in oil
appropriate synthesized hydrocarbon lubricant with appropriate pumps
additives, specifically, an optimized temperature stabilizer, wear • The gear unit was experiencing an unacceptable level of wear
reducer, oxidation inhibitor and anti-corrosive agents, allowed the • On cold mornings the lubricant was so viscous that it had to be
lubricant drain intervals to be extended to more than 20 months heated before the unit could be put in service, and many operators
after lubricant changeover. complained about the impact of significantly varied viscosity
A synthesized hydrocarbon lubricant, with good shear resistance,
Unscheduled shutdowns of centrifugal pumps as a result of cold was selected to address these problems. In addition to solving vis-
weather. Rolling element bearings in 18 centrifugal pumps in a cosity problems, reducing wear and eliminating cavitations, this
large chemical plant were experiencing serious problems that re- newer lubricant helped to reduce power losses by around 8%.
sulted in plant shutdown, due to wax plugging when using ISO With the resulting efficiency gains and reduction in energy cost,
VG-68 grade conventional mineral oil (the pump manufacturer’s the higher cost of the new lubricant was paid off in six months. ❏

quire a high-quality synthetic-base- oils are more expensive compared to Synthesized hydrocarbons (diesters
oil (often one with an ester base). straight mineral oils. Some EP oils and PAOs) are highly recommended
have a relatively short life at operat- for gear units. Other synthetic lubri-
Gear unit lubrication ing temperatures above 75°C. cants, such as polyglycols (high-tem-
Lubricants in gear units are often sub- Compounded oils, that incorporate perature lubricants), phosphate-esters
jected to shock loads and associated several different additives, are also (fire-resistant lubricants) or silicone
overloading. This creates extreme available for gear units. The most com- lubricants (high-temperature and
pressure (EP) requirements for gear- monly available additive is a molybde- heat-resistant lubricants) are not rec-
unit lurication oils. Gears should be num disulfide compound, which has ommended for gear unit applications.
continuously lubricated, and at the been successfully used in some gear This is because of their very high cost
same time, the oil must be kept clean. applications. However, it is difficult (because of high volume of oil required
Viscosity is probably the single most for a gear manufacturer or operator in a typical gear unit), possible reli-
important factor in lubricant selection to recommend these oils since some ability issues and lack of referenced
for a gear unit. A lubrication oil must of these additives have a tendency experience in gear unit applications.
be selected with a viscosity that can to separate from the base-stock. As a In extreme applications (those in-
withstand the anticipated load, speed result, such compounded oils are not volving higher or lower temperatures
and temperature. Other important generally recommended by vendors. or with a need for fire protection), true
factors are: EP additives (relates to Similarly, viscosity improvers in synthetic lubricants (such as polygly-
load and speed), viscosity index (re- gear drives should be used with great cols, phosphate-esters, or similar) may
lates to temperature), and oxidation care. In some cases, these polymer be used for gear units. The user must
stability (relates to temperature and additives can nominally improve the be careful when selecting these lubri-
contamination). viscosity index and extend the operat- cants since some of them remove paint
Lubrication-oil film thickness is ing temperature range of oil. However, and attack rubber seals.
mainly a function of operating speed. what must be remembered is that The more recent synthesized hydro-
Based on experience, high-speed gear polymers are non-Newtonian fluids carbons (again, based on diesters and
units (above 5,000 rpm) often require (so shearing reduces viscosity). A gear PAOs) have many desirable features
heavier oil (for example sometimes, unit is a very high-shear environment, such as compatibility with mineral
heavier than ISO-grade 100). and as a result, the viscosity of the oil oils and excellent high- and low-tem-
Mineral oils. Mineral oils are still will be reduced rapidly if too much perature properties.
widely used for gear units. EP ad- polymer is added.
ditives of the lead-napthenate, sul- Synthesized hydrocarbon lubri- Engine lubrication
phur-phosphate or similar types are cants. Synthesized hydrocarbon lu- Engine manufacturers and lubricant
recommended for gear drives when bricants are gaining more wide-spread manufacturers offer lists that recom-
a lubricant with higher load capacity acceptance in gear unit applications. If mend lubricants that are suitable for
is required. As a general rule, min- properly formulated, synthesized hy- each engine type and model (in general,
eral oils should be used in relatively drocarbon lubricants (typically based the engine makeer’s preferred lubri-
low speed, highly loaded gear drives, on diesters and PAO) can significantly cants must take precedence). If experi-
with a low or medium operating tem- improve gear unit (gears and bear- ence indicates abnormally severe con-
perature (below or around 75°C). EP ings) reliability. ditions, it may be necessary to reduce
Chemical Engineering January 2012 47
Engineering Practice

the oil drain interval or recommend an service engines 1 MW or above. is no need to punch carbon between
oil that provides higher detergency. 2. Premium ash-less lubricant for two major overhauls (for instance, once
Detergent additives are often added cycle engines. during three- to five-year intervals). n
to help keep the engine clean by mini- 3.  Detergent dispersant lubricant Edited by Suzanne Shelley
mizing sludge buildup. For instance, (with around 0.4% ash), recom-
superior engine lubricants are usually mended for most four-cycle engines. Suggested reading
formulated from specially selected, 4. Premium medium-ash lubricant for Bloch, H.P.,”Practical Lubrication for Industrial
Facilities,” 2nd Ed., Fairmont Press, Lilburn,
solvent-extracted naphthenic base lean-burn and cogeneration applica- Ga., 2009.
stocks, which have inherent resistance tions.
to carbon formation in the engine’s For most operators, an engine overhaul
combustion chamber, port and valves. at two- to five-year intervals is com- Author
Generally, engine oils (whether pe- mon (five-year interval is reported for Amin Almasi is a lead
rotating equipment engi-
troleum-based or synthetic, which are low-BMEP gas engines; BMEP means neer at WorleyParsons Ser-
vices Pty Ltd. in Brisbane,
more common) are available in a wide brake mean effective pressure). Pis- Australia (amin.almasi@
range of viscosities and are suitable ton-ring and valve problems are often He pre-
viously worked in Technicas
for both crankcase and cylinder lubri- reported as the main reasons for the Reunidas (Madrid) and Fluor
cation. They should be highly effective unscheduled shutdown of engines, and (various offices). He holds a
chartered professional engi-
to reduce ring-zone suppressing, port this is closely related to proper lubri- neer’s license from Engineers
deposits and the formation of crank- cant selection and use. When using or- Australia (MIEAust CPEng-
Mechanical), and a chartered engineer certificate
case sludge. dinary petroleum-based lubricants, it from IMechE (CEng MIMechE), RPEQ (Regis-
tered Professional Engineer in Queensland). He
The following list indicates desir- is necessary to punch carbon (remove also holds M.S. and B.S. degrees in mechanical
able properties for engine lubricants: carbon deposits) from engines ports engineering. He specializes in rotating machines
including centrifugal, screw and reciprocating
1.  Full synthetic engine oil designed more frequently (let say every 12–18 compressors, gas and steam turbines, pumps,
for superior performance under se- months). When using superior oils condition monitoring and reliability. He has au-
thored more than 45 papers and articles dealing
vere conditions, suitable for critical (mainly synthetic lubricants), there with rotating machines.

rEcEivE full accEss

to all of chemical Engineering’s
facts at Your fingertips art
icles in one convenient location

Each information packEd pdf articlE includes


graphs, charts, tables, equations and columns on the full

chemical engineering processes you deal with on a daily basis.
This is the tool you will come to rely on, referring back to the
information again and again with just the click of a mouse.
Facts at Your Fingertips Topics Include:
 Conservation Economics:  Creating Installed Gain Graphs
Carbon Pricing Impacts  Aboveground and
 Distillation Tray Design Underground Storage Tanks
 Burner Operating  Chemical Resistance of
Characteristics Thermoplastics
 Measurement Guide for  Heat Transfer: System
Replacement Seals Design II
 Steam Tracer Lines and Traps  Adsorption
 Positive Displacement Pumps  Flowmeter Selection
 Low-Pressure Measurement  Specialty Metals
for Control Valves  Plus much, much more… 17872

receive full access today by visiting

48 Chemical Engineering January 2012

Engineering Practice

Use Simplified Lifecycle-Cost

Computations to Justify Upgrades
TABLE 1. Estimated years of
The methodologies presented here can be used to run time before failure of
four principal wear-prone
set goals, and will enable performance comparisons pump components
Pump component Estimated
among different plants or industry segments life, L life for
part, yr
Heinz Bloch Lifecycle cost estimating Mechanical seals, L1 2.5
Consulting Engineer Lifecycle cost estimating is one of the Ball bearings, L2 5
reliability engineer’s most effective im- Couplings, L3 7

irtually every process-plant provement-justification tools. Lifecycle Shafts, L4 15
manager pursues the com- cost estimating takes into account the
mendable goals of safely ex- initial purchase and installation costs (DC) must be added to present acqui-
tending equipment life and of the equipment, auxiliaries and soft- sition and installation costs. Thus, the
maximizing both the availability and ware systems. It assesses the true cost total lifecycle cost (LCC Total) = AC
reliability of plant assets. Achieving of failures, including, of course, the im- + IC + present value of (OC + MC +
these objectives usually requires pact of lost production [1–3]. LP + DC).
upfront effort and money — both of A certain amount of information or A “present worth” value can also be
which can be scarce resources. general data is usually available from calculated. The cumulative present
But even the realistic manager who the plant’s computer-based enterprise- worth factor in Equation (2) can be ob-
knows that reliability comes at an asset management (EAM) or comput- tained from many sources and tables
ufront price may not want to autho- erized maintenance-management sys- as a function of interest rate and time
rize these expenditures on the basis of tem (CMMS). The existence of EAM (yr). It is usually available from the
intuition or guesswork. Instead, he or and CMMS is assumed here because plant’s accounting staff and can also
she may ask for some cost justification modern plants cannot compete with- be obtained as a computer spreadsheet
that is linked to a payback period, a out a CMMS. The plant CMMS is pop- program displaying a present value
cost-to-benefit calculation, a lifecycle- ulated with accurate data related to (PV) function. PV is cost multiplied by
improvement multiplier, or some other work orders, expenditures and failure the cumulative present worth factor:
tangible factor. It is usually at this incidents. All data of interest should
point in the sequence of events that be specific enough to clearly describe
the reliability engineer realizes that the root causes of failures observed.
he or she has no data and the issue The annual cost of parts failure (Cy)
is placed at the bottom of the priority can be assessed using Equation (1): where:
list. Things revert back to status quo i = real annual interest rates, %
Cy = (Cg)(8,760)/(MTBF+MTTR) (1)
and urgent repeat repairs siphon off n = number of years
precious resources. where: Except for data derived from well-de-
Even in the absence of abundant Cy = Annual cost of failures for a com- signed, in-plant EAM-CMMS systems,
data, many methods are available to ponent (or subassembly) system precise failure frequencies and life ex-
allow us to determine, with reasonable Cg = Cost per failure event pectancies are rarely available for pro-
accuracy, the monetary incentives or MTBF* = Mean time between cess machinery and their components.
justification for equipment and com- failure, h There are simply too many variables
ponent upgrading. Such upgrades are MTTR = Mean time to repair or re- that influence these numbers. Nev-
the key to future failure avoidance. place, h ertheless, an experienced reliability
This article describes some options The total lifecycle cost can be ob- professional will not be deterred in his
for determining the value of upgrad- tained by adding the initial acquisition
ing. The narrative and illustrations cost (AC), the initial installation cost *The MTBF of a randomly failing. multiple-com-
ponent, active-redundant system may be evalu-
presented here highlight some meth- (IC), and the recurring yearly costs. ated by the following equation:
odologies that are available to reli- A present value conversion [Equation
ability professionals who are ready (2)] takes into account the time value
to de-emphasize purely intuitive ap- of money. The costs of future operations
proaches, in favor of simple yet effec- (OC), maintenance (MC), lost produc- where the failure rate , and c =
number of parallel
tive numerical pathways. tion (LP) and even decommissioning components

Chemical Engineering January 2013 53


Engineering Practice

or her search for data. Remember, you overall machine MTBF can
want to make the business case for up- be readily visualized from
grading and your managers are only this equation (that is, the
asking for reasonable “ball-park” num- number 1 divided by a large
bers that directionally show which up- number yields a small num-
grades should be pursued. Such num- ber). Even a cursory look at Figure 1. A single-type, heavy-duty mechancial
bers can be found elsewhere [4–6]. Equation (3) — whether or seal for lime slurry service is highlighted in this
For precise estimates, calculated not upgrading is involved pump illustration. Aesseal
life expectancies of various component — will make a significant
categories should ideally be based on difference in the quest to
the experience collected at the reli- improve the life of weak
ability engineer’s facility. But the rel- components.
evant data may never have been col-
lected, or may have been lost when the Centrifugal
source expert left the company. If that pump example
is the case — or whenever realistic Table 1 shows estimated-
life assessments or cost estimates are life values for four different
needed — a reliability engineer may weak, or (relatively) wear-
want to, at least initially, use the data prone, pump components
tables contained in the cited refer- (mechanical seals, ball bear- FIGURE 2. Shown here is a half-section of a modern
ences. As experience is gained, similar ings, couplings and shafts). bearing housing protector seal — with the shaft not
rotating (left) and with the shaft rotating (right)
tables will grow into ever-more-precise We can use these numbers
and locally applicable component-life to calculate MTBF values
databases. Needless to say, once devel- for an entire pump. Of course, our pumps were first marketed in the mid-
oped, these should be preserved and calculation is somewhat general and to late-1980s). But, suppose one later
passed on to others. might pertain only to a particular ap- had the option to convert from grease
plication — say, a given pump size in to liquid oil lubrication. Assume one
Using component life details water service. Calculated MTBF values had also selected a cartridge-style me-
Many modern facilities are finding it refer to the anticipated running time of chanical seal (Figure 1) where compo-
progressively more advantageous to such a pump, if the life expectancies of nent-style seals had been used previ-
collect and classify component data its components are as given in Table ously, and that the user had added an
and to then incorporate these in cal- 1. These pumps had been previously advanced bearing-protector seal (Fig-
culations that predict the probable “upgraded” by converting from sealed ure 2). Suppose these improvements
run length, in terms of MTBF, of an ball bearings to bearings that can be would increase the operational lives
entire machine. Some plants have periodically refilled with fresh grease of mechanical seals and bearings from
successively improved the accuracy of (these are commonly called “regreas- the previous value of 2.5–5 years to
this form of lifecycle cost computation. able” bearings). Using the values for L 3.5–10 years, and would also make the
Basing decisions on improved compu- from Table 1, the estimated MTBF (op- pump more suitable for working in a
tational accuracy has led to greater erating time) of the entire pump was mild lime slurry service. In that case,
visibility and enhanced respect for the calculated with reasonable accuracy the expression in Equation (5) would
diligent contributions of the reliabil- using Equation (3), and the numerical apply, and we would probably have
ity professionals at those plants [5]. result is shown in Equation (4). reason to expect a continuous pump
The monetary value of an improve- operating life of 2.93 yr [6].
ment can be determined from yet
another version of the lifecycle cost
computation. Specifically, improve-
ment value can also be expressed in
even simpler terms, such as a benefit- Then, it is reasonable — and probably
to-cost. Whenever possible, parts that Here, L = estimated life, in years, of quite conservative — to anticipate an
experience wear — the most failure- the component subject to failure [6]. increase in pump MTBF of close to
prone components of a machine — can 40% from these two upgrades. Seeing
be assigned by some experience-based a 40% increase in predicted component
criteria or previously published values life should prompt a more-detailed
of L1, L2 and so on, as shown in Table analysis of the pump’s lifecycle cost.
1. Because of their position in Equation The stipulated 2.11-yr operating life It may be worth paying a certain
(3), low component-life values in Table determined by Equation (4) meets price for upgraded parts if lifecycle
1 will have a real impact on overall ma- the expectations of many reliability costs go down as a result. Calculat-
chine MTBF and the influence or effect engineers in U.S. process plants for ing a simple, straightforward cost-to-
of individual component upgrading on “upgraded” ANSI/ISO pumps (Such benefit ratio would be another way
54 Chemical Engineering January 2013
Bearing failure rate per 1,000 machines at a U.S. chemical plant
Failures per 1,000 pcs rotating equipment 70
60 C UCL C LCL C = Control
50 UCL = Upper control limit
C mean C rate LCL = Lower control limit
30 cally add $500 to the average small or
20 mid-size pump repair cost of $6,700.
UCL = 12.5
10 In sharp contrast, it had been es-
Mean = 6.7
0 timated that removing good pumps
-10 LCL = 1.0 from field locations and taking them
-20 to the shop to implement various up-
grades would cost, on average, $3,470
J FMAM J J A S O N D J FMAM J J A S O N D J FMAM J J A S O N D J FMAM J J A S O N D J FMAM J J A S O N D per pump. That particular option was
2000 2001 2002 2003 2004 obviously far less attractive and was
Calendar time not pursued.
FIGURE 3. Bearing failure rate per 1,000 machines at a U.S. chemical plant, plotted Again, note that the incremental
using the author’s field data cost of $500 per pump pertained only
to pumps that were expected to be
to quantify the value of equipment mental outlay (the cost) of $800 will sent to the shop in the following year.
upgrades. If the incremental cost of have a payback of (800/929)12, or 10.3 An elementary plot, shown in Figure
upgrading is $400 each year and the months. The cost-to-benefit ratio is 4, demonstrates the anticipated reduc-
benefit of an upgrade is avoidance of a 1:1.16 in the first year, and (5 × 929)/ tion in the pump-failure rate; the plot
$4,000 repair each year, then the cost- 800 = 1:5.8 over a 5-year period. That was used to calculate (in the 1990s)
to-benefit ratio is 400/4,000 = 1:10. is a substantial result and is not dif- the cost-to-benefit ratio of bearing
ficult to achieve. protector seals that would replace lip
Cost-to-benefit ratios Meanwhile, a secondary benefit can seals. The calculation was performed
Perhaps the most familiar form of cost be attributed to the systematic exten- by taking total incremental cost per
justification practiced on a wide scale sion of equipment life: Instead of get- year and dividing it by the projected
compares the incremental cost of an ting bogged down in frequent break- value of all avoided pump repairs.
upgrade option with the yearly value down-related maintenance tasks, Attractive and reasonable pro-
of maintenance cost avoidance. With reliability engineers will be able to jections along the lines of what we
that goal in mind, let’s take another devote their attention to other, more- just discussed contributed to wider
look at the pump-upgrade example proactive reliability-improvement use of a variety of different bearing-
discussed above. Recall that for the opportunities, thereby putting their protector seals in the mid-1990s.
proposed upgrades, an older-style me- effort to use to save money for their Then, with time, more-advanced
chanical seal would be replaced with employers over the long run [7]. styles became available. Figure 2
the cartridge seal shown in Figure shows a successful configuration,
1. Also, the bearing protectors with Make use of in-plant data which was first marketed in 2003. If
the lip-seal style in this centrifugal Important reliability-related data are we decided to install it today and used
pump would be discarded and the available [4] and such data can be ef- the same calculation approach, we
bearing-housing-protector seal with fectively used and applied to carry out find its cost-to-benefit ratio surpris-
an advanced design — in this case, a simple MTBF, cost-justification, and ingly attractive.
rotating labyrinth design, as shown in lifecycle-cost studies. However, while
Figure 2 would be used instead [6]. We published data sources are valuable, Upgrading mechanical seals
make the following two assumptions: the use of in-plant data may be even Earlier in this article, we had encour-
• That the two upgraded components more directly applicable and should aged reliability professionals to ex-
would incrementally cost $800 and never be overlooked. tend their horizons by reviewing data
result in shifting the pump MTBF One in-plant data example is dis- published elsewhere. In 1992, a Brit-
from the previous value of 2.11 yr to played in Figure 3. This figure shows ish reliability engineer published the
2.93 yr the reduction in total bearing failures results of failure-reduction programs
• That repairs to a mid-size pump that were actually experienced by a at three petroleum refineries [8]. As
will cost $7,000 by the time materi- U.S. Gulf Coast petrochemical com- shown in Figure 5, Refinery A started
als, labor, overhead, benefits, spare pany over the span of 54 months (4.5 with a pump MTBF of 29 months at
parts procurement, shop supervi- y). Although these improvements were the end of Year 2. The refinery’s pump
sion, planning, vibration monitoring undoubtedly attributable to a combi- MTBF had risen to 71 months at the
and reliability engineering have all nation of procedural, organizational end of Year 7. Accordingly, the run
been factored in and hardware-specific upgrades, the lengths of the pumps there had ex-
Based on these two assumptions, reliability staff made the simplifying perienced an increase of 42 months
our yearly pump-repair cost will assumption that such downturns in in the span of five years. Since these
have dropped from $3,318 (based on the number of bearing replacements increases are attributable to upgrade
$7,000/2.11 yr) to $2,389 (based on related entirely to pumps. It was fur- efforts that went beyond seal improve-
$7,000/2.93 yr). The ensuing cost sav- ther assumed that incorporating im- ments, we will temporarily put them
ings (or benefit) of $929/yr will go on proved bearing-protection components aside and focus instead on Refineries B
for years, while the one-time incre- only during shop repairs would typi- and C, whose reports dealt with
Chemical Engineering January 2013 55
Resonably anticipated pump failure rate reduction
due to upgrades efforts

Engineering Practice 17.7

Failure rate per 1,000 machines

mechanical seals only. configurations and im-
Refinery B documented an increase proved seal materials
in seal-related MTBF, from 57 months would add $1,700 to each 6.7
to 80 months, calculated as (80–57) / pump repair and that
57 = 40%, in four years. Seal-related typical pump repairs,
MTBF values at Refinery C improved using traditional grades
from 33 months to 50 months in the of seals, would cost ap- 0 54
span of two years — an increase in proximately $5,000. Months
MTBF of 51%. To determine this, we Assuming a linear FIGURE 4. A facility starting with 17.7 pump failures per
picked the numbers off of Figure 5 MTBF increase from 1,000 pumps per month might upgrade these 17.7 pumps
and put them into the expression: 28 months presently to and then, over a period of 54 months, reduce its monthly
(50–33)/33 = 0.51. 56 months five years statistics so as to meet a best-of-class number of 6.7 fail-
ures per 1,000 pumps per month
It makes good sense to expect more from now, we could cal-
substantial improvement possibili- culate our yearly repair
ties for the refinery that has the lower cost outlay in the most 80
starting MTBF rate. We note that Re- straightforward man- 70
finery C started with a seal MTBF of ner and list our results B

M.T.B.F. months
33 months, and that “our” refinery (as in tabular format. We A
an arbitrary example) is presently at could pick one of the ap- 50
28 months MTBF. Returning to Re- proaches described ear- 40
finery A and its overall pump MTBF lier in this article and 30
Refinery A
(which had increased from 30 months would review it with 20 Refinery B
at the end of Year 2, to about 71 months one or two competent Refinery C
at the end of Year 7), we would calcu- mechanical seal manu-
late an MTBF increase of (71–30)/30 = facturers — ones that 1 2 3 4 5 6 7 8
36% in 5 years. would agree to a part- Years
If we take into account the obser- nership or alliance that FIGURE 5. Shown here are data demonstrating improve-
vation that refineries starting with rewards them for failure ment in pump MTBF, from experience at three British pe-
MTBF figures of 30 months have ex- reductions instead of troleum refineries
perienced MTBF increases around lowest cost per seal. The
25%/yr it is reasonable to expect that ultimate results will be tangible and tual information to cost-justify equip-
our own plant could go from an MTBF will, after five years, have saved the re- ment improvements. Many reach out
of 28 months to one of 56 months in finery many millions of dollars. for other data sources to augment and
the span of five years. validate in-house data. It has been
Such a reasonable assumption now Different methods shown that data published in the past
allows our refinery operator to embark We have attempted to show how a can form the core material of fairly ac-
on a program to improve mechanical number of straightforward calculation curate savings projections made today.
seal MTBF. As reliability professionals, approaches can be used to determine The methodologies presented in this
we will accede to our management’s re- lifecycle costs, cost-to-benefit ratios, article can be used to set goals, and
quest to develop an appropriately ref- and payback periods for reliability will enable performance comparisons
erenced cost and benefit projection. We improvements in process plants. A re- among different plants or industry
have 1,474 centrifugal pumps at our sourceful reliability professional will, segments. n
plant site. Our seal MTBF was origi- of course, diligently collect and compile Edited by Suzanne Shelley
nally calculated from (1,474 pumps in- failure statistics for equipment and
stalled) × (12 mo/yr)/632 seal failures / components at his or her plant site. At
yr = 28 mo. Furthermore, it is assumed many locations throughout the world, Author
here that upgrading to superior seal competent professionals use this fac- Heinz P. Bloch, P.E., is a
consulting engineer in West-
minster, Colo., (heinzpbloch@ He has held ma-
References gal Pumps, Chem. Eng., November 25, 1985. chinery-oriented staff and line
1. Goble, W.M., and Paul, Brayton, O., Life Cycle 6. Sales and Marketing Literature, AESSEAL, positions with Exxon affiliates
Cost Estimating, Chem. Proc., June 1995. Inc., Rotherham, U.K., and Rockford, Tenn., in the U.S., Italy, Spain, Eng- land, The Netherlands and
2. Paul, Brayton 0., Life Cycle Costing, Chem. Japan, during a career that
Eng., December 1994. 7. Roberts, Woodrow T., The ABC's of Improv-
ing the Reliability of Process Plant Systems, spanned several decades prior
3. Roscoe, Edwin S., “Project Economy,” Richard “Proceedings of 3rd International Confer- to his retirement as Exxon
D. Irwin, Inc., Homewood, Ill., 1960. ence on Improving Reliability in Petroleum Chemical’s regional machin-
Refineries and Chemical Plants,” Houston, ery specialist for the U.S. Bloch is the author of 18
4. Bloch, Heinz P., and Geitner, Fred K., “Ma- comprehensive texts and over 500 publications on
chinery Failure Analysis and Troubleshoot- Tex., 1994
machinery-reliability improvement. He advises
ing,” Butterworth-Heinemann Publishing, 8. David, T.J., A Method of Improving Me- process plants worldwide on strategies and op-
Waltham, Mass., 4th Ed., 2012. chanical Seal Reliability, “Proceedings of the portunities for extending equipment uptime and
5. Bloch, Heinz P., and Johnson, Donald A., Institution of Mechanical Engineers, Fluid reducing maintenance. He is an ASME Life Fel-
Downtime Prompts Upgrading Of Centrifu- Machinery Ownership Costs Seminar,” Man- low and maintains registration as a professional
chester, U.K., September 16, 1992. engineer in Texas and New Jersey.

56 Chemical Engineering January 2013

Feature Report
Engineering Practice

Remote Thermal Sensing

By making it easy to
detect heat anomalies,
thermal cameras and
infrared thermometers
support preventive and
predictive maintenance
Roger Mavrides
General Tools & Instruments

Suzanne Shelley
Figure 1. Rising temperature is often Figure 2. Problems such as over-
Precision Prose, Inc. an indicator of operational problems in heated bearings can be diagnosed with
many types of machinery. The ability to a thermal imaging camera, which pro-

ising temperatures and rapid gather and analyze temperature data in vides an alternative to direct-contact
or excessive heat buildup are realtime using non-contact options can temperature sensors, especially for
help operators to pinpoint issues and components that may be hard to reach,
useful markers for determin- act accordingly inaccessible or potentially hazardous
ing the operational health
of many types of machinery and
components that are used in a wide ance, misalignments, insufficient ous. In this way, these temperature-
array of industrial and manufactur- cleaning or lubrication, friction and monitoring devices help to enable
ing settings. The types of mechani- electrical problems. realtime temperature measure-
cal and electrical systems for which The traditional approach to tem- ment while ensuring worker safety
temperature increases often signal perature monitoring in industrial (Figures 1 and 2).
problems include (but are not lim- settings relies on direct-contact
ited to): rotating machinery, such as temperature sensors, such as ther- Technology options
motors, turbines, compressors, and mocouples and resistance tempera- The two types of remote thermal-
their bearings, couplings and gear- ture detectors (RTDs). While these sensing options — infrared ther-
boxes; other types of process equip- devices are certainly proven and ac- mal-imaging cameras and IRTs
ment, such as pumps, valves, heat curate, they are not appropriate for — are widely used for tempera-
exchangers, steam traps, heaters, use with certain types of equipment ture assessment in industrial and
conveyors belts, rollers, furnaces components or in some types of in- manufacturing facilities. In addi-
and more; steam and electrical dustrial settings. tion, two different types of IRTs are
heat-tracing systems; insulation on By contrast, remote or non-con- available — conventional IRTs, and
pipes and vessels; refractory lining tact temperature-measurement so-called scanning IRTs. Used sepa-
systems for high-temperature sys- devices, such as infrared thermal- rately or together, these devices can
tems and much more. imaging cameras and infrared help users across numerous indus-
Because equipment malfunctions thermometers (IRTs) allow use- try sectors to quickly and easily
and abnormal or fault conditions in ful temperature data to be easily assess the thermal condition of ma-
mechanical and electrical systems gathered from remote locations chinery, process systems, pipelines
are often forewarned by a rise in and thus offer a useful alternative and more.
temperature, the ability to gather to direct-contact temperature sen- Conventional IRTs are best
and analyze temperature data in sors. The ability to safely carry out suited for applications that re-
realtime can often help operators to temperature sensing from a dis- quire accurate spot-temperature
both pinpoint existing performance- tance, using either a thermal cam- readings, while scanning IRTs and
related issues and identify the era or an IRT, is particularly use- thermal cameras are useful for ap-
onset of incipient problems. These ful for machinery components and plications for which knowing the
problems can be caused by an array systems that may be hard to reach, absolute temperature of a surface
of issues, including wear, imbal- inaccessible, or potentially hazard- is less important than knowing the
48 Chemical Engineering January 2014
Figure 4. De-
signed with an
easy-to-use pistol
grip, thermal-im-
aging cameras are
used to diagnose
hot spots in ma-
chinery systems

Figure 3. Thermal-imaging cameras through color variations in the aging cameras, IRTs offer an ideal
measure the surface temperature of the rendered image. way for operators to determine the
objects or areas being analyzed in terms Such visual displays of relative temperature of hot or cold surfaces
of the amount of infrared (IR) energy
that is emitted, transmitted or reflected temperature variations across the remotely, which is especially useful
by the object. The data can then be ren- surface of the objects gives opera- for inaccessible or hard-to-reach ob-
dered as still or video images to help tors and technicians unprecedented jects or areas.
operators interpret the result insight into the health of equip- As noted earlier, two types of
ment and systems, and help to ad- IRTs are available — conventional
temperature of a particular surface dress emerging problems efficiently. “spot” IRTs and so-called scanning
relative to other surfaces around it. Thermograms are especially useful IRTs. Scanning IRTs allow users
The primary advantage of a ther- when thermal imaging is used as to scan an entire area or system
mal camera is its ability to display part of regular inspections, because and quickly identify those sections
the thermal spectrum of an entire they allow engineers to quickly rec- where there is a significant tempera-
area, as seen in Figure 3. ognize changes that may signal an ture differential between the actual
Infrared thermal-imaging cam- emerging problem. Thermal images temperature of that section and a
eras. Portable thermal-imaging that are captured and analyzed over pre-set temperature setpoint value
cameras are easy to use, and typically time for the same component (for that the user has programmed into
come with a pistol-grip design, as instance, a given motor or pump) the device.
seen in Figure 4. They use infrared- can help users to identify locations Using a conventional IRT with
imaging techniques to measure the of incipient malfunction or progres- an appropriate distance-to-spot
surface temperatures of the objects sive wear or deterioration. (D:S) ratio — one that allows the
or areas being analyzed and can ren- Creating such a record of heat measurement “area” to be entirely
der the data in the form of two-di- buildup due to deteriorating con- focused within the object being
mensional images or videos images ditions allows operators or tech- measured — plant personnel can
to illustrate the data. Specifically, nicians to dispatch the most ap- determine the temperature of an
these specialized cameras measure propriate intervention in a timely object at a single spot. A built-in
surface temperature in terms of the manner. These interventions in- laser-beam sighting source helps
amount of infrared (IR) energy that clude detailed inspection, trouble- the user to focus the device on
is emitted, transmitted and reflected shooting and diagnostic efforts, and the target precisely, to ensure
by the object or area being analyzed. strategic maintenance and repair measurement accuracy.
The temperature data are displayed activities. Because thermal imag- While conventional IRTs are use-
as an IR heat spectrum, using a ing is carried out at a distance, it ful for remotely gathering point-
range of colors that are correlated to enables the capture of thermal data source data about the absolute tem-
specific temperature ranges. Today’s from components that are in remote perature of a given spot, scanning
thermal cameras have great sensi- or hazardous areas, thereby ensur- IRTs are useful for applications
tivity, and provide measurement ac- ing worker safety. where it is not necessarily impor-
curacies of up to ±2°F. Infrared thermometers. IRTs are tant to determine the absolute tem-
Thermal cameras produce a portable, non-contact devices — perature of a surface, but it is useful
thermal signature or thermogram, again, typically with a user-friendly, to determine the relative tempera-
which is a two-dimensional vi- pistol-grip design. IRTs use a spe- ture of a surface or area compared
sual display that depicts the rela- cial lens to focus the thermal radia- to its surroundings.
tive temperature variations across tion that is being emitted by the ob- Today’s scanning IRTs are not
the object’s surface. These images ject (in the form of IR energy) onto only very affordable (typically
allow operators to quickly pinpoint an IR sensor. The embedded soft- under a hundred dollars, compared
problem areas, since temperature ware correlates those IR readings to more than a thousand for thermal
excursions, such as areas of heat to the temperature of the object cameras), they are also extremely
loss or heat gain relative to the using information about the mate- easy to use. With a point-and-shoot
surroundings, are easily displayed rial’s emissivity. Like thermal im- design, the user first establishes a
Chemical Engineering January 2014 49
Engineering Practice

baseline temperature that is ap- • Proactive troubleshooting

propriate for the application, after and diagnostic interven-
which an acceptable bandwidth or tion can help to decrease
tolerance range is set — for exam- the likelihood of expensive
ple, ±10 degrees, although tighter or catastrophic equipment
tolerances of ±1 or ±5 degrees failures — especially im-
are possible. portant when considering
When the trigger on a pistol-grip mission-critical assets
scanning IRT is squeezed and the • The ability to plan and ex-
device is moved slowly across the ecute more-strategic re-
target area, the device uses a com- pairs helps to cut material
bination of sound (in the form of and labor costs and extend
Figure 5. When operators are able to monitor
slower versus faster beeping sounds) equipment life, thereby the thermal profile of equipment and systems
and colored lights (for example, red helping to reduce both oper- using a remote monitoring technique, they can
for above range, green for within ating and capital budgets plan for the most appropriate and timely inter-
range, and blue for below range) • Increased uptime allows vention to address a deteriorating condition
to alert the user to any location the facility to maximize
where the temperature falls outside its throughput capacity, product nent’s ideal state under normal
of the user-specified threshold val- yields and profitability. The abil- working conditions. Going forward,
ues that define the setpoint range. ity to carry out strategic main- any departure from normal temper-
The IRT will also provide an abso- tenance activities helps to not atures that appear in the thermal
lute temperature reading for that only save money, but improve images produced for the unit would
particular spot. plant and personnel safety and signify trouble spots that require
While scanning IRTs do not pro- environmental performance closer inspection.
duce a thermal image, they do pro- Proper training and certification
vide a quick, easy, and relatively in- Best practices are extremely beneficial when using
expensive way for facility personnel There are several ways to make IRTs and thermal-imaging cam-
to assess specific mechanical assets best use of thermal-imaging data. eras. While these devices are typi-
and identify those problem areas Trending opportunities can be used cally simple to operate and provide
that may require closer inspection. to the engineer’s advantage. For data and images that are easy to
instance, the thermal signature interpret, both rely on sophisticated
A valuable investment from a given component, such as a technology. As such, to ensure the
While individual thermal cameras particular pump that may be sus- most accurate results, users should
typically cost more than a thousand pected of having a problem, can be gain a good working knowledge of
dollars, they can be a strategic in- compared to the thermal signature the capabilities and limitations of
vestment — and will easily pay for of similar pumps in the facility. This these tools through proper training
themselves over a short amount will help to evaluate its condition and certification. Without proper
of time if their use prevents a relative to other equipment with training, the accuracy of the result-
catastrophic failure. comparable operation. ing thermal data and images could
As noted, the use of remote, IR- In addition, the thermal images be compromised. Following vendor-
based thermal sensing to inspect, generated for a given mechani- recommended operating guidelines
troubleshoot, diagnose and rectify cal asset (say, a particular motor), and proven industry best practices
problems with specific equipment can be strategically captured and is a must for data confidence.
in realtime can help facility opera- cataloged over time in specific in- Today, a variety of third-party
tors to improve the efficiency and tervals. This record can provide groups offer training and certifica-
effectiveness of both component- timely indications of deteriorating tion in the proper use of thermal-
specific and plant-wide operation conditions and help the operator to imaging cameras, including the
and maintenance (Figure 5). Such make reasonable predictions about American Society for Nondestruc-
improvements provide a number of the rate of future deterioration, tive Testing (Columbus, Ohio; www.
opportunities for long-term savings so that the required action — be, the Academy of Infrared
and payback. For instance: it maintenance, repair or replace- Training (Bellingham, Wash.; www.
• The costs associated with un- ment — can be carried out, at the, the Infraspec-
planned downtime in industrial appropriate time, in the most cost- tion Institute (Burlington, N.J.;
facilities, such as manufactur- effective manner., The Snell
ing plants and chemical process When practical, it is a good idea Group (Barre, Vt.; www.thesnell-
facilities can be greatly reduced to carry out baseline thermal-imag- and others.
when thermal imaging is used to ing inspections on new components, Choosing the right thermal imag-
improve preventive and predic- to establish baseline or reference ing camera for the environment has
tive maintenance tasks images that represent the compo- an effect on the equipment’s ability
50 Chemical Engineering January 2014
to best collect accurate data. Ther- differentials resulting from the un- cameras may be available at a
mal imaging cameras are widely wanted ingress or escape of heated lower cost, the desire for cost sav-
used to carry out energy-efficiency or cooled air will be most easily ings alone may “cost” the buyer in
studies and audits in residential, identified by the thermal camera the long run by limiting the num-
commercial and business settings, or scanning IRT assessment. By ber of months over which the cam-
by helping to identify areas through contrast, in southern climates, the era can be reliably utilized to carry
which heated or cooled air is escap- summer months tend to guaran- out energy audits and other types of
ing from a building. Industrial fa- tee the biggest temperature dif- thermal inspections. In general, the
cilities can also use these devices ferential between ambient outdoor higher the resolution of the thermal
to carry out energy audits to iden- temperatures and air-conditioned camera, the more reliably it can
tify further opportunities to reduce settings. depict a thermal difference when
operating costs. This potential seasonal “limita- carrying out a thermal assessment
When this is done, users should tion” can be somewhat overcome — even during those times of the
note that regional climate varia- by selecting a higher-resolution year when the temperature differ-
tions can impact what time of year thermal camera. Today, a variety of ential between indoor and outdoor
the use of a thermal imaging cam- thermal cameras — with a range of temperatures is relatively narrow.
era will be most effective. For in- prices and image-resolution capa- By investing in a higher-resolu-
stance, in northern climates, the bilities — are available. Relatively tion camera, users will be assured
use of thermal imaging to carry out low-end models have sensors with of greater sensitivity and easier
energy audits at a facility tend to a resolution of 60×60 pixels. Mid- thermal assessments no matter
be most accurate when carried out range units have a resolution on the what climate or time of year. This
during winter months (when the order of 160×120 pixels, and high- helps to ensure more accurate re-
temperature differential between end thermal cameras offer a resolu- sults and faster payback for the
indoor and outdoor temperatures tion of 360×280 pixels. camera itself.
is the greatest) and temperature While lower-resolution thermal Another important factor to con-

The Chemical Engineering bookstore offers a

variety of industry topics you will come to rely on.

• Environmental Management: Air-Pollution Control

• Environmental Management: Wastewater and Groundwater Treatment
• Fluid Handling
• Gas-Solid and Liquid-Solid Separation
• Liquid-Liquid and Gas-Liquid Separation
• Managing Bulk Solids

For a complete list of products, visit the Chemical

Engineering bookstore now.

Chemical Engineering January 2014 51

Engineering Practice

sider when using thermal cameras Figure 6. Infrared thermometers let

and IRTs is the relative emissivity users determine the temperature of a
specific point on hot or cold surfaces
values of the materials being sur- remotely, which is especially useful for
veyed. Emissivity is a measure of inaccessible or hard-to-reach objects or
an object’s ability to absorb or re- areas. D:S ratio is an important feature to
flect radiation in the infrared range consider when choosing an IRT
of the electromagnetic spectrum.
As a simplifying assumption, many away from a target will measure the cost-effective repairs and upgrades.
thermal cameras have fixed emis- temperature within a circle that is Such efforts can help to optimize
sivity values that are associated 6 in. in diameter. the productivity and reliability of
with certain commonly encountered Units with a D:S ratio of up to the mechanical assets, maximize
materials programmed into the 100:1 are also available. In general, the uptime of the facility, and mini-
control software. the higher the D:S ratio of the de- mize downtime-related losses and
However, today’s more-sophis- vice, the smaller the zone for tem- expenses and reduce the risk of
ticated thermal-imaging cameras perature capture, because less of catastrophic equipment failures.
allow the user to make adjustments the surrounding area is involved in And, remote IR-based temperature
to the emissivity settings, to more the measurement. IRTs with higher monitoring lets facility personnel
accurately characterize the actual D:S ratios also provide for more ac- carry out such surveillance with-
emissivity values of the materi- curate readings to be gathered from out shutting down the machines,
als being analyzed. Such values greater distances. interrupting the process or put-
can often be found in published ting themselves in harm’s way. This
books and reference articles. This Closing thoughts maintains equipment reliability,
flexibility can help users to im- Remote, non-destructive, ther- the facility’s desired productivity
prove the accuracy of the resulting mal-sensing techniques, using IR levels, as well as personnel safety,
thermal images. thermal cameras and scanning or and thereby helps to protect the
Certain surfaces, such as highly conventional IRTs, provide useful facility’s bottom line. ■
reflective metals, have very low alternatives to direct-contact tem- Edited by Mary Page Bailey
emissivity values and cannot be perature devices based on RTDs and
measured accurately using IR- thermocouples. They provide useful Acknowledgements
based temperature-measurement information on temperature excur- The authors would like to thank
techniques. To overcome this prob- sions that are often the precursor John Javetski, Adrian Gomez,
lem and improve the utility and ac- to operational problems, allowing Kevin Basso and Peter Harper of
curacy of IR-based thermal cameras users to plan the most strategic pre- General Tools & Instruments for
and thermometers, industrial oper- dictive and preventive maintenance their assistance during the develop-
ators often paint the target surfaces activities and to carry out the most ment of this article.
or cover them with electrical tape.
This raises the emissivity values of Authors
the components and improves the Roger Mavrides was for- Suzanne Shelley is the prin-
accuracy of the IR-based thermal merly vice president of en- cipal/owner of Precision Prose,
gineering and product devel- Inc. (65 West 96th St., Suite
imaging techniques. opment for General Tools & 21F, New York, N.Y. 10025;
Instruments (80 White St., Phone: 917-975-2778; Email:
When it comes to evaluating dif- New York, N.Y. 10013; Phone:
ferent IRT models, an important 1-800-697-8665 x222; Email: In that capacity, she provides freelance technical writing,
concept to consider is the distance- He holds a Certificate of Elec- ghostwriting and editing ser-
to-spot (D:S) ratio (Figure 6). This tronic Maintenance (CEM)
from Wentworth Institute of
vices (specializing in science,
engineering, technology and
characteristic of the device provides Technology, a B.S. in electri- business) to magazines and
a measure of the optical resolution cal engineering technology from Northeastern corporate clients. Prior to launching her consultancy
University and an M.B.A. from Anna Maria in 2005, Shelley spent 17 years as a full-time editor
that a particular unit can provide. College. Before joining General Tools, Mavrides at Chemical Engineering magazine, serving as the
was engineering and product manager, test magazine’s managing editor for her last 5 years on
Every IRT model has a stated D:S and measurement, for FLIR Systems (Nashua, staff. As a freelance writer and editor, Shelley serves
ratio, which determines the dis- N.H.), sales and product manager for Nidec/ as a regular freelance contributing editor at Chemi-
Power General (Canton, Mass.), and senior de- cal Engineering and Pharmaceutical Commerce
tance for which the device will pro- sign engineer and project manager for Vishay/ magazines, and as a periodic freelance contribut-
vide the most accurate temperature BLH Electronics (Norwood, Mass.). Mavrides
holds three patents, is a Level 1 Thermographer,
ing editor at Chemical Engineering Progress (CEP;
AIChE). From 2005–2009, she also served as a regu-
reading as well as the diameter and was a team leader during the development lar contributing editor to Turbomachinery Interna-
of the imaging area. For instance, of FLIR’s MeterLink communication protocol. tional magazine. Shelley also provides freelance
He also developed a wireless alternating-cur- writing, ghostwriting and technical editing services
most standard IRTs have a D:S rent circuit identifier that won a Hong Kong to a wide variety of operating and service compa-
Electronic Industries Association (HKEIA) In- nies, consultancies, advertising agencies and trade
ratio of 8:1. This indicates that an novation and Technology Grant Award at the associations throughout the global chemical process
IRT 8 in. away from the object can 2009 HK Electronics Fair, and developed an industries. She holds a B.S. in geology from Colgate
electrically safe video borescope that won the University (Hamilton, N.Y.) and a M.S. in Geology
accurately measure the tempera- Bronze HKEIA Innovation & Technology Grand from the University of South Carolina (Columbia,
ture at a spot that is 1 in. in diam- Award at the 2011 HK Electronics Fair. S.C.), and worked as a deepwater exploration geolo-
gist in the Gulf of Mexico for Amoco Production Co.
eter. Similarly, an IRT that is 48 in. (New Orleans) in the late 1980s.

52 Chemical Engineering January 2014

Feature Report
Engineering Practice

Pressurized Piping:
Sampling Steam and Water
Without proper systems, Isolation Source: Jonas

analysis of steam Flow < 20 ft.

< 200 ft.
and water chemistry Primary

can provide Isokinetic Sampling Expansion

erroneous results — sampling nozzle tubing coil

with costly implications Online

Lee Machemer Analyzer

Sample filter
with bypass Total Pressure Thermal Pressure
Jonas, Inc. flowmeter gauge shut-off reduction
valve valve

pressure Grab Temperature
orrosion and deposition in boil- regulator samples indicator
ers, steam turbines and many
types of process equipment Figure 1. This figure shows an example of a well-designed sampling system for
are among the most expensive extracting and conditioning a representative sample of steam or water. It includes
isokinetic sampling, rapid condensation and cooling, pressure reduction, and
causes of outages in utility and in- process indicators, as well as safety devices to protect online instruments
dustrial steam plants. Deposits and and plant personnel
scale buildup on heat-transfer sur-
faces reduce efficiency, and when industrial steam plants do not or periodically. Proper design of the
allowed to accumulate on steam have properly designed and oper- sampling systems is critical in order
turbines, such buildup can reduce ated sampling systems to monitor to produce samples and analytical
the capacity. Corrosion-related fail- water and steam chemistry. In fact, results that are representative of
ures can result in outages ranging in water chemistry and corrosion the sampled stream [3–9]. Problems
from a few days to several months, control audits, sampling problems with sample withdrawal, transport,
depending on the affected systems, are found in roughly 70% of all collection and handling are often
and can potentially cost tens of mil- plants. As a result, operating deci- major sources of errors that can
lions of dollars. sions are often based on data that lead to incorrect or unnecessary
To reduce the risk of corrosion can have sampling errors as high corrective actions by operators. A
and deposition in water and steam as ± 1,000%. These errors, as well meticulously performed chemical
systems, the standard practice is to as data inconsistencies and concen- analysis is of little value if a bad
monitor cycle chemistry and control tration swings in the analytical re- sample is used. As shown in the box
impurity levels within industry- sults, become commonplace and are on p. 43, there are many potential
and manufacturer-recommended often ignored by plant personnel, causes of sampling errors, some of
limits for the equipment. In steam preventing the timely identification which can cause analytical results
plants, the chemical parameters of of actual chemistry excursions. This to be orders of magnitude higher or
interest include: pH; conductivity; article outlines the principles that lower than the actual concentration
sodium; calcium; magnesium; chlo- must be considered when designing in the process stream.
ride; sulfate; fluoride; phosphate; and operating water- and steam- In high-purity systems, the mea-
acetate; formate; propionate; total sampling systems. sured concentration of impurities
organic carbon (TOC); silica; cop- in many of the process streams is
per; and dissolved and suspended Sampling system design in the low parts-per-billion (ppb)
iron (oxides). Typical target concen- To monitor systems for the ingress range. At such low concentrations,
trations are in the range of <1 part of impurities and for the production the fluid being extracted is very sen-
per billion (ppb) to several parts per and transport of corrosion products, sitive to any deposition or chemical
million (ppm) [1, 2]. several cycle streams are sampled reactions within the sampling sys-
Unfortunately, many utility and and analyzed, either continuously tem. The extraction of non-repre-
42 Chemical Engineering January 2014
Causes of Sampling
System Errors

perators should be mindful of these
common sources of sampling errors
(in order of priority and impact):
• Sample withdrawal — Sample does
not (1) represent the stream (due to
wall effects, stratification, not isoki-
netic, or mixing issues), and (2) repre-
sent all phases (solid, liquid, gas)
• Deposition in the sample line (could
also result in plugging)
FIGURE 2. This weld-in style, single-port isokinetic sampling nozzle meets current
ASTM standards for sampling water and steam. Flanged connections to the process • Release of built-up deposits into the
pipe are also acceptable sample stream (leading to spikes)
• High pressure drop leading to insuf-
sentative samples can lead to large sample line blockage and can cause ficient sample flow (typically due to
sampling errors [3]. Even in lower- unacceptable time lags between long sample lines)
purity systems, sampling errors due sample collection and analysis. A • Changes in sample flowrate (for in-
to improperly designed sampling sample flowing at 2 ft/s through 500 stance, sampling system can take up
systems can be significant. ft of tubing will take over four min- to 6 hours to reach equilibrium from
A well-designed sampling system utes to reach the analyzers. the start of sample flow)
(Figure 1) consists of an isokinetic
• High sample temperature (can lead to
sampling nozzle (discussed below), Why isokinetic sampling? pH and conductivity errors)
isolation valves, sample tubing, a Isokinetic sampling is the extraction
primary cooler (for steam and high- of a representative portion of the • Chemical reactions in sample lines or
temperature liquid samples), a process stream without altering the coolers (reduction of oxygen concen-
tration, change in pH and so on)
secondary sample cooler, pressure- physical and chemical properties of
reduction and total-flow-regulation the sample. In isokinetic sampling, • Corrosion of the sampling system may
valves, a thermal-shutoff valve all phases (solid oxides and precipi- lead to generation of corrosion prod-
(for process temperatures above tates, liquid droplets and vapor) of ucts (as a result of improper materials
100ºF), back-pressure regulator and the sampled fluid enter the sam- of construction)
sample drains. pling nozzle with the same velocity • Filters in the system interfere with de-
Because steam impurities are eas- vector (meaning the same velocity sire to sample suspended solids
ily adsorbed by magnetite (Fe3O4), and direction of flow). The main • Sorption on sample tubing and sus-
the oxide buildup on the inner di- reason isokinetic sampling is nec- pended oxides may remove a por-
ameter of the sampling nozzle and essary is that the sampled stream tion of the chemical species being
tubing should be minimized. For is almost always a two-phase fluid monitored
this reason, all wetted components (gas-liquid, gas-solid, liquid-solid)
of the sampling system should be and the second phase typically has
made from at least Type 316 stain- a very different chemistry composi- is not representative of the condi-
less steel. Carbon and low-alloy tion than the steam or water [2]. In tions in the pipe. Proper sampling
steels should be avoided. addition, the second phase (droplets -nozzle design must consider the ef-
Deposit buildup in the sample or particles) typically has a different fects of flow- and vibration-induced
lines can result in plugging of the density and inertia compared to the forces on the nozzle, as well as the
sample line or seizing of sample primary phase (gas or liquid) and design pressure and temperature.
isolation valves. Even when not therefore would not be proportion- Prior to 2006, the ASTM Standard
directly affecting sample flow, de- ally represented in a sample that D1066 “Standard Practice for Sam-
posits in the sampling system can was not withdrawn isokinetically. pling Steam” [4] included a multi-
affect the sample accuracy. Deposits The benefits of isokinetic sampling port sampling nozzle, which — in
can act as ion-exchange media and have been verified during an Elec- its most basic form — consisted of
adsorb or release impurities during tric Power Research Inst. (EPRI) a piece of pipe with multiple holes
changes in the flow conditions. Even project [3] and through an indepen- in it. The sampling pipe extended
the best sampling-system design is dent analysis [10]. most or all of the way across the
still susceptible to deposition and process pipe and was supposed to
plugging if the cycle chemistry at Sampling nozzle design simultaneously sample from sev-
the plant is not maintained within The design of the isokinetic sam- eral locations across the diameter
industry standards, particularly pling nozzle (Figures 2 and 3) is a of the pipe. However, research has
when high concentrations of corro- critical part of the sampling system, shown that such a multiport design
sion products (such as iron oxide or and should be performed prior to operates non-isokinetically, is prone
copper oxide) are present. Lengthy the selection of the other sampling to plugging, and is susceptible to
sample lines (for instance >100 ft) or system components. As noted, if failure due to vibration [3].
low sample velocities (for instance designed incorrectly, the sampling In many piping applications, the
< 4 ft/s) increase the probability of nozzle could provide a sample that flow in the process pipe is fully tur-
Chemical Engineering January 2014 43
Engineering Practice

bulent. This results in a uniform Source: Jonas VB VB VB

velocity profile across the pipe and
the stream is well mixed, so the
composition is uniform across the
pipe. This makes it unnecessary to
sample at more than one location
along the pipe diameter. In light of
this, a single-port sampling nozzle
was designed in the early 1990s
[3]. The isokinetic sampling nozzles
have a compact design, which has
the advantage of being inserted
only about 12% of the way into the
pipe, compared to multiport nozzles,
which must traverse most (if not
all) of the diameter of the pipe. In
fact, the single-port nozzle design Isokinetic Non-isokinetic
was included in the ASTM standard VB = VN VB < VN VB > VN
in 1996. In 2006, it became the only
recommended sampling nozzle de- FIGURE 3. Shown here is the ideal flow path of particulate matter and droplets into
sign included in the standard. isokinetic and non-isokinetic sampling nozzles. In isokinetic sampling, the extracted
fluid is representative of the composition in the process pipe, including particles and
droplets. When the sampling is non-isokinetic, the concentration of particles andV
Transport of samples droplets can be higher or lower than that found in the process fluid. VB = velocityB of
Almost any fluid will leave or pick process fluid; VN = velocity in the sampling nozzle
up some residue, both while flow-
ing through a tube and while being time required to achieve equilibrium the primary sample cooler should be
stored in a container. As a result, between impurities in the flowing located as close to the sample point
any chemical analysis will become sample and the tubing, the sample as possible (less than 20 ft).
biased due to the loss or gain of con- tubing after the primary cooler/con- The total length of sample tub-
taminants. Several factors contrib- denser should be sized so that the ing should be as short as possible
ute to deposition on the tubing wall, sample flow velocity is maintained to limit both the pressureVN drop and VN
including: crystallization resulting around 5–6 ft/s [3–8, 12]. Several the lag time from when the sam-
from solubility changes, settling studies have shown that both linear ple enters the isokinetic sampling
due to gravity and hydrodynamic velocity and the Reynolds Number nozzle to when it reaches the ana-
forces, and electrostatic attraction (Re; a unitless dimension that de- lyzers. Low sample residence time
of charged particles [6]. scribes the amount of fluid turbu- in the tubing is also preferred to
In any sampling system, there lence) control the net deposition of limit chemical reactions, such as
will be an exchange of contami- particulate matter in sample lines oxygen scavenging and sorption on
nants and particulates between the [13–17]. Therefore, the sample line oxides. The sample velocity should
flowing sample and the sample line should be designed to achieve both be maintained as constant as pos-
surfaces. Eventually, an equilibrium turbulent flow (Re > 4,000) and sible in the interest of maintain-
state will be reached. Whenever the proper velocity (5–6 ft/s). ing equilibrium between deposition
sample is not in equilibrium with In steam-sampling systems, one and re-entrainment of particles,
the surface, the sample composition of the most critical design consider- and chemical equilibrium between
will be changed from its original ations is the size and length of the the sample and deposits. In one
state. In general, the time for new sample line from the sampling noz- research project [16], it was found
sample tubing to reach equilibrium zle to the primary sample cooler/ that it took less than 30 days for a
decreases with smaller tubing (due condenser. Long and oversized sam- newly installed sampling system to
to decreased surface area) and in- ple lines produce a significant pres- reach equilibrium when the sample
creased sample velocity. Even when sure drop and heat loss from the was flowing at 6 ft/s, compared to
a sufficient sample velocity (say, on extracted fluid. In general, both the several years for a sample flowing
the order of 6 ft/s) is maintained, temperature and pressure of the at 1 ft/s.
the equilibration process can take sample should be maintained right In addition to the effects on sam-
up to a month. It is for this reason up to the primary cooler/condenser ple purity, the use of larger-diame-
that sample streams should flow so that desuperheat and condensa- ter tubing can result in an unnec-
continuously rather than be peri- tion occur together. To achieve this, essary waste of sample water (and
odically started and stopped. the sample line should have approx- additional energy required to heat
In order to minimize deposition in imately the same inside diameter as and cool the fluid), require impracti-
the sample lines and to reduce the the isokinetic sampling nozzle, and cal sample-conditioning equipment,
44 Chemical Engineering January 2014
Table 1. Reynolds Number, sampling rate, annual volume, and pressure drop (ΔP) for water
(T = 100°F) flowing through various sizes of tubing at 5 Ft/s
Outer dia. Inner dia. Wall thick- Reynolds Num- Required sampling Annual vol., Estimated ΔP per
(OD), in. (ID), in. ness, in. ber, unitless rate, cm3/min gal/yr) 100 ft of tubing, psi
0.250 0.120 0.065 6.8 × 103 670 93,000 57
0.250 0.152 0.049 8.6 × 103 1,070 148,000 42
0.375 0.245 0.065 1.4 × 104 2,780 386,000 24
0.500 0.370 0.065 2.1 × 104 6,340 879,000 14

and place an extra and expensive vide a minimum change of cross- with allowances for reduced heat
burden on the makeup system. section between the inside diameter transfer due to scale buildup. The
Table 1 compares sampling rate, of the isokinetic sampling nozzle and cooler tubing should be made from
Reynolds number, estimated pres- the orifice of the valve. Large changes Type 316 stainless steel or Inconel.
sure drop, and the annual volume of in cross-section can result in deposi- Sample tubing after the primary
water consumed for several tubing tion within the valve and may even- sample cooler. This tubing should
sizes with a sample flow velocity of tually lead to seizing of the valve, slope downward to allow for com-
5 ft/s. Typically, ¼-in. tubing with which can become a safety issue if plete draining during outages, and
a sampling rate of 1,000 to 1,200 the sample line is damaged during have a minimum number of bends.
cm3/min (condensed) is sufficient to operation. Valves should be made of It should be sized so that the sam-
provide for all online analyzers and Type 316 stainless steel or a higher ple flow velocity is 5 to 6 ft/s.
grab sampling while maintaining alloy. Because valves in steam and Pressure-reduction valve. Such
the required flow velocities. water service are susceptible to de- a valve is used to reduce pressure
position inside the valve (particularly and therefore control the flow of a
Additional considerations for steam service), it is recommended cooled sample in order to protect
When designing a sampling sys- to always have two isolation valves. online instruments. For sample
tem, such as that shown in Figure Sample tubing between the isoki- pressure greater than 500 psig, the
1, follow these recommended prac- netic sampling nozzle and pri- pressure reducer should be a rod-
tices for each of the components dis- mary cooler. Such tubing should in-tube-type orifice or capillary [5].
cussed below: be as short as possible (not longer For sample pressure less than 500
Installation location for the than 20 ft for steam systems) in psig, the pressure reducer should be
sampling nozzle. The preferred order to minimize the pressure drop a needle valve.
location is in long, vertical sections and reduce the possibility of impu- Thermal shut-off valve. This valve
of pipe, away from all flow distur- rity deposition in the sample tubing. protects personnel and downstream
bances (such as bends, valves, and The inside diameter of this sample components by automatically inter-
so on) [4, 6]. Ideally, the sampling tubing should be close to the inside rupting sample flow when the sam-
nozzle should be at least 35 internal diameter size of the isokinetic sam- ple temperature reaches a preset
pipe diameters downstream, and 4 pling nozzle, to minimize changes in limit, in the event of an insufficient
pipe diameters upstream, of any cross-sectional area. This normally amount or loss of cooling water or a
flow disturbances. In many plants requires ¼- or 3/8-in. tubing for liq- fouled sample cooler.
where space is at a premium, this is uid water and medium- to high- Pressure and temperature
not possible, so it is recommended pressure steam systems, and ½-in. gages and flow indicator. Such
that the sampling nozzle be located tube or ½-in. pipe for low-pressure devices provide the operator with
where the ratio of its distance from steam systems. verification that the system is work-
the upstream disturbance to down- The sample line should include a ing properly.
stream disturbance is about 9:1. If series of bends or a coil to allow for Back-pressure regulator. This
a long vertical section is not avail- any movement or expansion of the regulator is used to maintain a
able, the sampling nozzle may be process pipe. Sharp-radius bends slight pressure (~20 psig) in the
installed in a long horizontal sec- should be avoided. The tubing sample tubing before the grab sam-
tion, provided the sampling nozzle should be downward sloping along ple location. This will ensure proper
is installed on the top of the pipe be- the entire length to eliminate any flow to the online, chemical-analysis
tween the “10 o’clock” and “2 o’clock” sections where condensed steam or instruments.
positions to prevent the possibility water can accumulate and result in Inline sample filters. These filters
of water accumulating around the water hammer during startup. should be installed to protect online
sampling nozzle during outages. Primary and secondary sample instruments during commissioning,
Isolation valves. These valves coolers. The coolers should have a or any other time when high concen-
should be rated for the application counterflow design and be sized to trations of corrosion products (iron,
temperature and pressure, and pro- ensure adequate cooling capacity, copper) are present in the sample.
Chemical Engineering January 2014 45
Engineering Practice

They should be installed down- through the system could result in • Check sample flowrates
stream of the grab-sampling line (as insufficient sample flow at the sam- • Check sample temperatures after
shown in Figure 1), or must be by- ple panel, or, the deposition rate in both primary and secondary sam-
passed when obtaining grab samples steam sample lines could be high, ple coolers
for iron and copper analysis. which could result in plugging of • Check sample pressure
Online analyzers. The sample the sample line or a sample that is • Ensure flowrate through online
flowrate, temperature and pres- not representative of the conditions instruments meets manufacturer
sure must all be within the instru- in the pipe. The design must also requirement
ment manufacturers specifications. ensure that the maximum pressure • Check for any vibration at the
A chiller may be required in order recommended by the online instru- sampling nozzle and along the
to cool the sample streams to the ment makers is not exceeded. length of the sample tubing.
proper temperature. ASTM D5127
requires that the sample tempera- Commissioning of the system Operation and maintenance
ture be 25±1°C when measuring After the sampling system is in- Once the sampling system is in-
pH, and ASTM D5391 requires that stalled, the following tasks should stalled, proper operation and main-
the sample temperature be con- be performed to ensure proper op- tenance are required to ensure ac-
trolled to 25±0.2°C when measuring eration of all components: curate sampling, including:
conductivity if specialized tempera- • Check all sampling points to en- Total sampling rate and sam-
ture compensation is not available. sure proper location and sampling pling time. The total sampling
Such strict temperature control nozzle orientation rate should be governed by the rate
may not be practical; therefore, the • Verify that all sample tubing and required for isokinetic sampling,
use of modern pH and conductivity cooling water tubing is properly which is a function of the sampling
analyzers that include temperature sized for the required flowrate nozzle design and the process mass
compensation algorithms may be • Ensure that all valves and flow- flowrate. Even if this sampling rate
an acceptable alternative. meters operate properly exceeds the requirements of online
Booster pumps. These pumps may • Confirm the proper flowrate of analyzers, the total sampling rate
be required for long sample lines cooling water to the primary and should be maintained by routing
(high pressure drop) or low pres- secondary sample coolers excess flow either through the grab-
sure samples (condensate). • Check for leaks along the entire sample location to drain or to the
Once all of the sampling compo- length of sample tubing including condenser hotwell. For high-purity
nents are specified, the estimated the sample panel systems, it can take up to six hours
pressure drop through the system • Perform startup and calibration of isokinetic sample flow to stabilize
should be calculated. The pressure of all online instruments in ac- the sample chemistry. This time can
drop throughout the entire sam- cordance with the original equip- be shorter for lower-purity systems,
pling system (including primary and ment manufacturer’s (OEM) in- but for all sampling systems, con-
secondary coolers, tubing, valves struction manual tinuous flow is preferred.
and elbows) must be low enough to • Verify that online instrument Grab samples. There are many
ensure that there is enough pres- readings agree with readings on opportunities for the grab sample
sure to provide adequate flow veloc- the distributed control system to degrade during collection and
ity (5–6 ft/s) through the tubing to (DCS) or other data-recording sys- storage. This is especially critical
the online instruments and grab- tem, and that alarms are working in samples for pH, conductivity, dis-
“sample tap. A high pressure drop properly solved oxygen and hydrazine anal-

References Palo Alto, Calif., CS-5164, April 1987. Systems for PWR Primary-Coolant Circuits,
8. Jonas, O., and Mancini, J., Sampling savvy, “Proceedings of the Workshop on Corrosion-
1. Jonas, O., Corrosion and water chemistry Product Sampling from Hot-Water Sys-
problems in steam systems — Root causes Power Engineering, May 2005.
tems,” EPRI, Palo Alto, Calif., NP-3402-SR,
and solutions, Materials Performance, 9. Eater, L., Make sure water chemistry sam- March 1984.
December 2001. ples are representative, Power, July 1989.
14. Emory, B., Crud Sample-System Design
2. “Interim Consensus Guidelines on Fossil 10. Binette, V., and others, Impact of Sampling Criteria, “Proceedings of the Workshop on
Plant Chemistry,” EPRI, Palo Alto, Calif., CS- System Design on Superheated Steam Qual- Corrosion-Product Sampling from Hot-Wa-
4629 and other water-chemistry guidelines, ity, “Proceedings of the 5th International ter Systems,” EPRI, Palo Alto, Calif., NP-
June 1986. Conference on Fossil Plant Cycle Chem- 3402-SR, March 1984.
3. “Development of a Steam Sampling System,” istry,” EPRI, Palo Alto, Calif., TR-108459,
Nov. 1997. 15. Sundberg, L., Sampling of Metallic Impuri-
EPRI, Palo Alto, Calif., TR-100196, Dec. ties in BWRs, “Proceedings of the Workshop
1991. 11. Daucik. K., Design of Sampling Devices on Corrosion–Product Sampling from Hot-
4. “Standard Practice for Sampling Steam,”. for Water/Steam Cycle, “Proceedings of Water Systems,” EPRI, Palo Alto, Calif., NP-
ASTM D1066, 2011. the 9th International Conference on Fossil 3402-SR, March 1984.
Plant Cycle Chemistry,” EPRI, TR-1020563,
5. “Standard Practices for Sampling Water Jan. 2010. 16. “Survey of Corrosion-Product Generation,
from Closed Conduits,” ASTM D3370, 2008. Transport, and Deposition in Light-Water
12. McKinney, J., Analyzers and Steam Pan- Nuclear Reactors,” EPRI, Palo Alto, Calif.,
6. “Steam and Water Sampling, Conditioning, els — A Perspective from Both Sides of the
and Analysis in the Power Cycle,” ASME NP-522, March 1979.
Fence, “Proceedings of the 9th International
Performance Test Code (PTC) 19.11, 2008. Conference on Fossil Plant Cycle Chemistry,” 17. Svoboda, R., and others, Trace Analysis of
7. “Guidelines Manual on Instrumentation and EPRI. TR-1020563, Jan. 2010. Corrosion Products by Integrated Sampling
Control for Fossil Plant Chemistry,” EPRI, Techniques, “Water Chemistry 3,” British
13. Bird, L., Requirements for Crud-Sampling Nuclear Energy Systems, London, 1983.
46 Chemical Engineering January 2014
ysis. Special preparation of grab ing all engineering efforts to obtain nozzle, attachment to the process
samples or sample containers may representative samples. pipe, valves and all welds should be
be required, depending upon the Maintaining clean coolers. Pe- periodically inspected for evidence
type of analysis being performed. In riodic cleaning of the cooling water of cracking and other forms of dam-
some cases, chemicals are added to side of the coolers may be required age. For sampling wet steam and
the container before the sample is to maintain proper heat transfer water, the section of process pip-
added to prevent sample degrada- and sample temperature. The fre- ing immediately downstream of the
tion (for instance, samples used for quency of cleaning depends upon sampling nozzle should be periodi-
the analysis of iron or copper). the scaling properties of the water cally inspected for thinning by flow-
Collection methods for samples used for cooling. accelerated corrosion. Installations
to be analyzed for pH, conductiv- Sample tube cleaning. All sam- that sample liquid water should be
ity, dissolved oxygen, ammonia, hy- ple tubing should be periodically checked for cavitation. n
drazine and organics must exclude cleaned by flushing or acid clean- Edited by Suzanne Shelley
contact between the sample and air. ing, or it should be replaced. The
Storage methods and holding times frequency of cleaning depends on Author
of samples from collection to analy- the amount of impurities in the Lee Machemer is president
sis require special consideration to sample streams. One “quick and of Jonas, Inc.(4313 Nebraska
Court, Pomfret, MD 20675,
avoid degradation of samples. dirty” method to test the cleanli- Phone: 301-934-5605; Email:
Calibration and maintenance. ness of the sample line is to shut off and has worked for the com-
These steps should be routinely the sample flow at the sample panel pany for 18 years as a water
chemistry and corrosion con-
performed on all online instruments and then quickly turn on the flow to sultant. Machemer has been
per the manufacturers’ recommen- the maximum sampling rate. If the involved with the design
and development of several
dations. Improperly calibrated and sample is brown or black, there are products used in fossil-fired,
maintained instruments will result deposits in the sampling system. nuclear, and geothermal power-generation facili-
ties. He holds a B.Ch.E. from the University of
in inaccurate measurements, negat- Maintain safety. The sampling Delaware and is a professional engineer.

Get Chemical Engineering’s plant cost index to

improve plant cost estimates… PREMIUM
and delivered in advance of the print edition! EXPLOSION
For more than 37 years, chemical process industries professionals-

Indoor / Outdoor Venting

engineers, manager and technicians, have used Chemical With Flame Absorber
and Dust Retainer
Engineering’s Plant Cost Index to adjust process plant construction
costs from one period to another.
This database includes all annual archives (1947 to present)
and monthly data archives (1970 to present). Instead of waiting Q-Rohr-3
more than two weeks for the print or online version of Chemical
Engineering to arrive, subscribers can access new data as soon
as it’s calculated.
Resources included with Chemical Engineering’s Plant
Cost Index: CONTROL

…for Powder,
• Electronic notification of monthly updates as soon as Granulate,
they are available Grain, Flour
and Dust
All rights reserved

• All annual data archives (1947 to present)

• Monthly data archives (1970 to present)
• Option to download in Excel format
contact us …for Europe …for North America

Subscribe today at REMBE® GMBH SAFETY + CONTROL

59929 Brilon/Germany
T + 49 (0) 29 61 - 74 05 - 0
Charlotte, NC 28217
T +1 704.716.7022
F + 49 (0) 29 61 - 5 07 14 F +1 704.716.7025

Circle 13 on p. 56 or go to

Chemical Engineering January 2014 47

Feature Report
Engineering Practice

Plot Plan Design:

Process Requirements
It is important to conceptualize plant layout in terms of both ideal location and
optimal geographical positioning of equipment components
Mohammad Toghraei a good supply of human resources. ers want to locate plants in loca-
Engrowth Training When plants are established in de- tions with minimum land costs and
sirable locations, they have access minimum applicable taxes.

or any chemical process facility, to a broad pool of qualified, skilled Level 2. During Level 2 efforts,
the cost of improper or sub-ob- people for operation, maintenance stakeholders work to identify the
timal plant layout can be enor- and management of the plant. best method for laying out the dif-
mous. Designers are generally Plant location should also be se- ferent units within the facility. One
well aware of safety constraints, but lected to minimize the potential overarching objective of a plot plan
there is often less recognition about environmental impact of plant op- (but not necessarily the most impor-
other process requirements that erations. For instance, establish- tant one) is to minimize the length
will impact the design. This article ing a plant in an area that is sur- of equipment-connecting elements,
discusses key elements of plant lay- rounded by mountains is often not such as pipes that convey liquids,
out and plot plan design, in two im- a good decision, due to potentially gases and bulk solids. Material
portant levels: poor air-flow characteristics in the conveying systems, such as screw
• Level 1. Plant location. This step area. Similarly, if the treated waste- conveyors, belt conveyors, also fall
involves selecting the best loca- water of a plant needs to be injected within this category.
tion into a disposal well, the plant can- During the development and de-
• Level 2. Plant layout. This step not be very far from an area that sign of a plot plan, requirements
involves the placement of units has a suitable underground geologi- and limitations are identified and
and equipment relative to each cal structure. decisions are made with regard to
other, in an effort to optimize all Economic parameters are also the following considerations:
safety, operations and mainte- important when scouting possible • Process requirements, such as
nance objectives plant locations. Ideally, stakehold- the need to support equipment
Each requires a certain chain of de-
cision making that can impact the Plot plan
success of all efforts (Figure 1). AREA 1: Production tank farm AREA 2: Separation area Heat exchangers AREA 3: Reaction area
O2 Scavenger Acid
Ion exchange beds
An integrated effort
Level 1. During Level 1, engineers Product Product Separator
tank tank
seek to identify the optimal location Separator Caustic

for the plant. Generally speaking, a

chemical process industries (CPI) Finger pipe racks Sand Filters

plant should be in a location that

allows it to easily receive raw ma-
terials and have access to utilities, Future Future Pipe racks

such as water and power. Other key

considerations are related to access
to the infrastructure needed, such
Thickener Thickener
as roads, rail lines and shipping op- Raw water Flush water Feed tank BFW tank Feed tank
tions, to enable both the shipment
of products (and side-products) and
the disposal of waste streams. AREA 4: Tank Farm Off specification water Overflow tank Overflow tank

Other than material resources, Figure 1. Shown here is a typical plot plan
any CPI facility also needs access to
52 Chemical Engineering January 2015
Stacked heat exchangers Figure 2. In many cases, it makes
good engineering sense to incorporate
vertical designs, to take advantage of
gravity-based flow, limited footprint area
• Suitable performance in the facility, and other engineering
• Safety requirements, such as the considerations
need to ensure the proper dis-
tance between the furnace and
oil-storage tanks
• Construction requirements, such
as the need to provide enough ac- Control
cess area to allow construction cabinet
equipment (such as cranes and Centrifuge
lifts) to maneuver around the Mezzanine
equipment level
• Operation, inspection and main-
tenance requirements, to ensure
easy accessibility for operators to
reach each component, system or
monitoring console
• Logistics requirements, to ensure
easy accessibility for service com-
panies and their vehicles, includ-
Truck for
ing chemical-delivery trucks dumped
If the team follows the basic rule sludge
of seeking to minimize pipe length
wherever possible, it will undertake
efforts to minimize the length of
both process piping and utility pip-
ing. To minimize the process piping
length, equipment must be placed
in the order of the process flow dia- Prevailing wind direction at the this typically leads the plot plan de-
gram (PFD) arrangement. However, target location also affects the loca- signer to locate all the air coolers in
the effort to minimize the utility tion of equipment. Thus, a furnace, one location, to economize.
piping length often forces designers for instance, can be placed at the The concept of constructability
to lay down the developed “string” edge of a plant against the direction can affect the location of equipment,
of equipment based on the PFD of the prevailing wind to make sure too. For instance, a piece of equip-
along the perimeter of a circle with any potential flammable leaked ment may be placed relatively far
shared utility units in the center. fuel gases would move away from from the main part of the plant —
Although other requirements (such the plant. This same logic is why in a less-congested area against all
as those related to safety and logis- flares are typically placed far from other requirements — to satisfy the
tics) will usually result in changes human activity. need for ease of access for construc-
to this preliminary arrangement, Another group of safety consid- tion vehicles, such as large cranes.
it will be a useful exercise to group erations could be better classified Operation and inspection require-
equipment components that are as “nuisance issues.” For instance, ments will also impact the location
using common utilities, in a first even though engineering controls of equipment. Such considerations
attempt to minimize the utility are often used to manage noise often require that operator-inten-
pipe lengths. and fugitive emissions, units that sive equipment be grouped and
The designer not only needs to try are inherently noisy or odorous placed in easily accessible areas.
to minimize the pipe length but also should be placed in areas with little There must be specific clearance
to minimize the pipe rack length human presence. around each piece of equipment, to
and the number of finger pipe racks. Construction requirements are provide enough room for operators.
Doing so requires effort to develop equally important during the de- Specific maintenance require-
common routes for different pipes, velopment of a plot plan. Ease-of- ments may require wider clearance
as much as possible. construction rules typically call for around one or all sides. For exam-
Safety requirements call for seg- a plant with a regular shape, pref- ple, a floating bundle shell-and-tube
regating areas within a plot plan in erentially with right angles. heat exchanger needs a clearance of
a way that groups equipment and Meanwhile, specific installation about 1.5 times the tube length in
units with common hazardous char- requirements associated with cer- front of the heat exchanger as an
acteristics. To minimize the risk of tain equipment must also be taken area for bundle removal. Similarly,
releasing flammable, toxic, or lethal into account. For example, air cool- plate-and-frame heat exchangers
liquids from storage tanks, they ers are usually installed in a hori- need enough clearance to allow for
should be grouped based on hazard- zontal direction, at some height plate removal.
ous classification areas and then above the ground. Such an installa- Some equipment — raw material
put within different containments tion will require an additional sup- tanks, silos of additives, chemical-
or diked areas. port structure. The added expense of injection skids that may need fre-
Chemical Engineering January 2015 53
Engineering Practice

quent fillings by trucks — is con- indoors. example, in a small walk-in cabinet,

sidered “logistic-intensive” in the The main group of operator–in- a full firefighting system will not be
sense that it may affect the logistics tensive equipment includes equip- available. Selecting this interim op-
if placed in non-suitable locations. ment used in semi-continuous, tion should be considered very cau-
Placing such units in a poorly con- intermittent or batch operation. Al- tiously, making sure that all safety
sidered spot could also be unsafe though many batch operations are measures are considered carefully.
or might interfere with an opera- designed to be operated fully auto- In very cold areas, such as north-
tor’s daily duties because of the in- matically, frequent operator checks ern Canada, the plot plan of indoor
creased truck traffic that would be are still necessary. equipment sometimes needs to en-
needed to transfer products or raw Consider the case of filtration. It sure that certain sensitive equip-
materials. To address these poten- is popular to place pressure filters ment components are not located
tial issues, such units are usually indoors even though the filtering near doors or windows, to minimize
placed near the edge of the plant. media (such as sand or anthracite) the chance of freezing. In such areas,
Another aspect of plant design are not temperature-sensitive. frequent opening of doors and gates
to consider is where to place ponds Any equipment that, for whatever can increase the chance of freezing
such as storm water ponds that are reason, is not monitored by instru- for the equipment near the door, es-
often used in plants to collect rain ments — and thus requiring more pecially if there is no hot-air curtain
water. Placing such ponds in the operator attention — is usually available.
middle of a plant will decentralize placed indoors. For example, equip- 2. One story versus multi story
the other units and will lengthen ment for which taking samples is plant. The decision must be made
intra-plant travel time for person- important should be placed indoors as to whether or not the plant
nel, so it best to place them at the to ensure a suitable environment should be designed on one level, or
periphery of the overall layout. for the operator to be able to take whether some equipment should be
the samples. However, some compa- placed on a second or third floor. The
Key elements to consider nies may opt to utilize another op- default option is to maintain the en-
The following elements should be tion, which is to use an expensive, tire plant on one floor to minimize
reconciled during the development automatic grab-sampling system the cost, however, that may not
of the plot plan: for some equipment so that they be possible.
1. Indoor versus outdoor instal- can locate it outdoors. Having the plant built on more
lations. If the outside ambient The other consideration during than one story increases the cost
temperature is not extreme (and plot plan development is the need because of additional required
the variation in temperature over for clean rooms, which are used to structures and supports to hold
time is not significant), then the protect processes for which even all equipment systems, especially
least expensive decision is often to trace amounts of impurities could when the equipment is dynamic
place all of the equipment outdoors. impact the quality of the product. rather than static, and reciprocat-
However, this option is not avail- Examples of such processes are ing rather than rotary. For example,
able everywhere. pharmaceutical and microchip man- placing a small tank on a mezza-
The indoor-versus-outdoor deci- ufacturing plants. In general, clean nine level could be less expensive
sion is important because of the rooms need to be located indoors. than placing a centrifugal pump or
added cost of providing buildings It is not the case that only two mixer on that level.
to house the equipment. The first options — indoor or outdoor — However, in some cases having a
choice is always to locate the equip- are available. Having equipment multistory plant is preferable and
ment outdoors unless the equip- partially indoors is one option to justifiable (Figure 2). Consider the
ment is sensitive to very low or high achieve a compromise between cost case for plants involved with sol-
temperatures, or the equipment is and operability. For example, in ids handling, such as those whose
very operator-intensive. In these some cases, the side of the equip- raw materials, products or interim
cases, it makes more sense to locate ment that has all the sampling products are in bulk, granulated
the equipment indoors. points can be located indoors while or powdered form. Solids convey-
Higher costs are associated with the rest of the equipment could ing is typically carried out by
indoor installation not just because be left outdoors. Similarly, when pneumatic conveying systems, belt
of the cost of buildings, but also due equipment has tall towers, instead conveyors, screw conveyors, bucket
to added costs required to meet ju- of raising the roof of the building, conveyors and related systems. One
risdictional building codes (which the tower portion can be left above way to reduce the need for some
tend to be more explicit, and thus (that is, outside) of the building. modes of pneumatic or mechani-
more costly, for indoor facilities). Another interim option is plac- cal conveyance (and hence reduce
Polymeric membrane systems, ing equipment inside of a “shed” or energy requirements) is to ar-
for example, usually cannot tol- walk-in cabinet. The downside to range the equipment vertically and
erate temperatures greater than this approach, however, is that not use gravity.
40–50°C or very low temperatures. all provisions associated with an in- In some cases, vertical arrange-
As a result, they are usually placed door building will be available. For ment can ease cleaning, especially
54 Chemical Engineering January 2015
Content Licensing for
Every Marketing Strategy
Marketing solutions fit for:
• Outdoor

• Direct Mail

• Print Advertising

• Tradeshow/POP Displays

• Social Media

• Radio & Television

Logo Licensing | Reprints | Eprints | Plaques

Leverage branded content from Chemical Engineering to create a more powerful and
sophisticated statement about your product, service, or company in your next marketing
campaign. Contact Wright’s Media to find out more about how we can customize your
acknowledgements and recognitions to enhance your marketing strategies.

For more information, call Wright’s Media at 877.652.5295 or visit our

website at
Engineering Practice

for sticky materials. In this ap- tion acceleration head” term, which severe electrical classification level
proach a simple chute can some- needs to be added to the NPIP by only because of one or a few units
times replace a more complicated, the manufacturer. The magnitude that are handling flammable fluids.
mechanical solid-conveying system. of acceleration head is a function of 5. Inline mixing with or without
Another common example of the speed of reciprocation, the vol- static mixers. When two or more
plants that benefit from vertical ar- ume of cylinder(s) and the length of streams will be mixed in line (pipe)
rangement are those that deal with suction pipe. rather than a mixing tank, require-
liquids in the boiling temperature 4. Chemical compatibility. Dif- ments associated with the static
range (or in the bubble-point range ferent equipment handles different mixer should be defined by the plot
for non-pure liquids). In such cases, chemicals, which are not necessarily plan. The aim of any mixing effort
the designer usually prefers to have compatible. In an ideal world, vari- is to produce a fairly homogenous
a vertical arrangement to prevent ous chemicals are bound by their mixed fluid at the inlet of the down-
unwanted evaporation or flash of respective equipment or containers. stream equipment.
the liquid. It is common to see a However, there is always a chance If the downstream equipment
“stack” of heat exchangers that op- of leakage in an accident or via con- can be placed far from the mixing
erate at temperatures close to the trolled release as the primary step point, a homogenous fluid can often
boiling temperature of liquid, to of maintenance. Plot-plan develop- be attained via regular pipe flow
suppress the flash of liquid caused ment must consider the impact of even without a static mixer. If this
by the weight of the column of liq- the presence (and potential co-min- is not the case, then a static mixer
uid (Figure 2). gling) of all site-specific fluids, in is needed. As a rule of thumb, when
3. Pipe length in the suction side any given location. dealing with relatively watery liq-
of pumps. All pumps, either cen- One specific concern is sumps. uids (viscosity <20 centipoise), a
trifugal or positive-displacement Sumps can receive different liquids pipe length of 100–150 times the
types, are sensitive to low suction and must be designed to store them pipe diameter is usually sufficient
pressure. This sensitivity can be safely for a limited time. The com- and removes the need so that a
quantified using the required net patibility of any liquids that could static mixer can achieve the desired
positive suction head (NPSHR) end up in a given sump should be homogeneity.
term for centrifugal pumps, and the checked to make sure no uncon- 6. Handling time–sensitive flu-
required net positive inlet pressure trolled reactions can occur. If fluids ids. As noted, the relative location
(NPIPR) term for positive-displace- are not compatible, the respective of equipment affects the pipe length
ment pumps. NPSHR is essentially equipment could be moved to an- between units. This could be critical
the required effective liquid column other building, or a dedicated sump for time-sensitive fluids. Although
in the suction side of the pump to could be considered. the travel time in specific sections
guarantee proper operation with Another issue related to poten- of pipe is a matter of minutes, it
minimum cavitation. tual incompatibility of chemicals could still be important for very
This sensitivity can be addressed is electrical area classification. sensitive fluids.
by ensuring an adequate margin on The electrical devices in each area For example, in some flotation
top of NPSHR exists to provide a should be compatible with the na- vessels, the bubbles are generated
suitable available net positive suc- ture of the chemicals in each area. outside of the flotation vessel and
tion head (NPSHA). The goal is to For example, a conventional electri- then bubble-laden liquid is trans-
make sure NPSHA is higher than cal device that may generate sparks ferred to the vessels. If the distance
NPSHR with enough margin. This during its functioning is not suitable between the bubble-generating
can be done by minimizing the suc- to be installed in any area in which mechanism and the flotation vessel
tion pipe length and the number of highly flammable liquids may be is too large, bubbles will coalesce
fittings used. Doing so forces the de- present. This requirement is gen- with each other. And at the vessel
signer to place the source container erally addressed by designating an inlet, instead of having a liquid with
and the pump as close as possible “electrical area class” for each area, small bubbles, there could be liquid
on the plot plan. depending on the type of chemicals with big slugs of gas. This reduces
Although reciprocating pumps in that area. the separation efficiency of the flo-
are typically less sensitive than It should be considered that one tation vessels.
centrifugal pumps toward suction area with a less-stringent electri- 7. Process gravity-flow pipes.
pressure (in other words, they could cal class should not be pushed to a When flow from one equipment
have lower NPIPR than equivalent more-stringent electrical area class component or container to another
NPSHR), their pulsational opera- only because of a few units that are occurs only by gravity, some other
tion, or the fact that pressure in located (unwisely) in the area and issues should be considered. In two
their suction (and discharge side) require more-stringent electrical interconnecting containers, the
can go much below average suction area classification. With proper plot liquid level in the second (down-
pressure makes them vulnerable. plan design, engineers can avoid the stream) container is partly adjusted
The effect of pulsation in their suc- poor (and costly) practice of classify- by the pressure drop in the connect-
tion side is quantified in the “suc- ing a big area or building with more ing pipe. This shows the importance
56 Chemical Engineering January 2015
Engineering Practice

of gravity-pipe pressure drop and operation should be done carefully planned. While the designer of the
its length in the design and opera- and the design should minimize plot plan cannot leave room for the
tion of a unit. The pipe pressure loss attrition. One requirement in such unplanned modifications, he or she
can be minimized by placing these systems is to minimize the transfer should try to reserve enough room
containers as close as possible and line length and use specific fittings, in suitable locations to accommo-
using the minimum number of el- such as long-radius elbows. This re- date those future modifications
bows and other fittings on the con- quires the source and destination that can be planned or anticipated
necting pipe. units to be placed as close to each in some way. n
If the gravity pipe is placed with other as possible, which will affect Edited by Suzanne Shelley
a slope toward the second intercon- the plot plan.
nected container, the importance of 9. Symmetric piping. Providing
minimizing the pipe length becomes a symmetrical piping design to the Author
more obvious. If the pipe is long, the inlet of two similar equipment com- Mohammad Toghraei is an
instructor and consultant
second downstream container may ponents is one passive way to split a with Engrowth Training
need to be placed in a pit to be able stream equally to both of them. The (Web:; Phone:
403-808-8264; Email: moe.
to provide the required slope. This is need for symmetric piping can force, based
not an ideal situation, because deal- the designer to relocate or rotate in Calgary, Alta. He has more
than 20 years of experience in
ing with equipment in a pit provides equipment in the plot plan. As a the process industries, and
has published articles on dif-
some difficulties for operators. rule of thumb, any need for ferent aspects of process op-
8. Hydraulic transferring of symmetrical piping around equip- erations. His main expertise
is in the treatment of produced water and waste-
critical solids. One method of ment should be identified early, water from the oil-and-gas industries. Toghraei
transferring powders, granules as it will impact the plot-plan received a B.Sc. in chemical engineering from
Isfahan University of Technology (Iran), and an
or beads is through the use of hy- design efforts. M.Sc. in environmental engineering from Tehran
draulic transfer systems. Hydrau- 10. Future plans. The basis for University (Iran). He is a member of the Assn. of
Professional Engineers and Geoscientists of Al-
lic transferring can be done by air, future modifications in the plant berta (APEGA), and is a professional engineer,
P.Eng., in the province of Alberta.
water or other fluids. If the integ- could be economic or technical, and
rity of solid beads is important, this they may be either planned or un-

Get Chemical Engineering’s Plant Cost Index to improve plant

cost estimates…and delivered in advance of the print edition!

For more than 40 years, chemical process industries professionals- engineers, manager and
technicians, have used Chemical Engineering’s Plant Cost Index to adjust process plant construction
costs from one period to another.
This database includes all annual archives (1947 to present) and monthly data archives (1970 to present). Instead of waiting
more than two weeks for the print or online version of Chemical Engineering to arrive, subscribers can access new data as
soon as it’s calculated.
Resources included with
Sep ‘06
Aug ‘06
Sep ‘05
Chemical Engineering’s
CE Index 513.1 510.0 467.2
Plant Cost Index:
Equipment 606.5 602.3 541.2
Heat Exchanges and Tanks 565.1 560.9 509.2
• Electronic notification of
Process Machinery 559.6 556.2 521.7 monthly updates as soon
as they are available
Pipe, valves and fittings 734.7 731.7 620.8
Process Instruments 441.4 437.2 379.5
480 • All annual data archives
Pumps and Compressions 788.9 788.3 756.3
(1947 to present)
Electrical equipment 418.9 414.2 374.6
Structural supports 643.7 637.7 579.3
• Monthly data archives
Construction Labor 314.7 312.9 309.1
(1970 to present)
Buildings 476.9 475.2 444.7
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
• Option to download in
Engineering Supervision 350.7 351.9 346.9
Excel format

Subscribe today at


24688 CE PCI House Ad_HalfHorizontal.indd 1 9/4/14 5:34 PM

Chemical Engineering January 2015 57
The Impact of
Off-BEP Pump
Department Editor: Scott Jenkins Operation

Source: ITT Goulds Pumps
est-efficiency point (BEP), the flowrate Venturi
at which a pump operates at its
highest or optimum efficiency for
a specific impeller diameter, is a key
consideration when assessing pump
performance. While most pumps do not
consistently operate at their exact BEP, a
Low velocity/ Increased velocity /
pump that is properly sized will maintain high pressure pressure drop
a flow near peak efficiency (85–105% Bubble Bubble
of BEP). inception Bubble collapse
Operating a pump “off-BEP” means

Static pressure
that the flowrate is either too far above
or below the BEP for a sustained period Vapor pressure
of time, resulting in a number of negative
consequences. This article outlines the
Figure 2. Cavitation can be created for demonstration purposes using a venturi
consequences of operating pumps away
from their BEP for extended periods, as head drop (Figure 1). At that 3% ratio, tion and can lead to catastrophic pump
well as the key questions to ask when a pump is already cavitating. That is failure when portions of the impeller inlet
assessing off-BEP pump operation. why there are common practices in the or discharge vanes fatigue and fail by
industry that require a certain margin to breaking off.
Consequences of off-BEP operation be in place to make sure a pump is not Temperature rise. Temperature rise is one
Vibration and noise. Noise and vibra- running right at the NPSHA, and thus, by of the more severe effects of off-BEP opera-
tion can occur when a pump operates definition, cavitating. tion, because at its most extreme, human
too far to the right of BEP, generating Bearing and seal failure. Bearing and life can be lost. If a pump is allowed to
high-velocity eddy currents that contribute seal failure accounts for more than 80% run at shut-off for an extended period
to the imbalance of pressure and shaft of all premature centrifugal pump fail- of time, enough energy can be applied
deflection. The resulting stress on the ures, and occurs when a pump experi- to the fluid to cause the pressure in the
pump’s internal components can lead to ences increased radial and thrust loads pump to build to a point where it’s greater
poor pump performance, excessive wear during off-BEP operation. Pumps that than the yield strength of the casing, thus
and increased risk of failure. The ideal have a single-volute casing design nor- causing an explosion. Such explosions
noise/vibration point is approximately mally experience a rapid increase in im- have been known to throw motors through
90% of BEP. peller radial and thrust loads as the flow concrete block walls. The chances of this
Cavitation. Cavitation occurs when declines below BEP flow. Dual-volute cas- happening are remote, yet real.
vapor bubbles continuously form and ing designs help to balance radial loads
collapse, creating intense pressure (up and are essential for reliability when a Key questions
to 10,000 psi) and shock waves (Figure pump must operate for a substantial pe- The following application-related ques-
2). This is caused when the net positive riod of time at flows considerably below tions should be considered when assess-
suction head required (NPSHR) increases its BEP flowrate. Modified concentric ing off-BEP pump operation:
beyond the NPSH available (NPSHA), volute-casing designs are an alternative • What type of damage can occur if a
or when the NPSHA drops below the to single volutes — offering reduced pump is run below the BEP?
NPSHR. In determining reliability, if radial loads in off-BEP operation, but • What type of damage can occur if a
the NPSHA in the system drops below giving up a little bit of pump efficiency pump is run above the BEP?
the NPSHR by the pump, the pump will in the process. The impeller loads that • How does off-BEP operation affect a
experience cavitation — eroding the develop during off-BEP operation can pump’s mechanical seal?
impeller, vibrating the bearings and cas- lead to shaft deflection and mechanical • Why does vibration increase in off-
ings, and causing damage that can be seal failures, or overload the bearings BEP operation?
quite severe. The fatigued metal breaks with increased temperatures. • What are some ways pump manufac-
away, creating pitted surfaces, which Discharge and suction recirculation. turers use hydraulic design to mini-
become concentration points for further Discharge and suction recirculation hap- mize loads on bearings and increase
bubble collapse. NPSHR is typically pens when fluid does not flow through bearing life?
based on test standards established by the pump as it was designed, causing • Are proper operational controls in
the Hydraulic Institute (; small flow instabilities called eddies. The place to prevent a catastrophic event?
the definition of it is based on a 3% total damage caused by eddies mirrors cavita-
Source: ITT Goulds Pumps
In pursuit of peak efficiencies and
increased reliability (longer mean time
between failures), pump adjustments are
Total head, ft

often made to align their BEP with the

Constant flow
–3% duty point of the pumping systems. Con-
sider testing a pump if you are uncertain
Of total head about its BEP for a specific application. ■

14 12 10 8 6 4 2 0 –2 –4 –6 –8 –10 Editor’s note: Content for this edition of “Facts at

your Fingertips” was contributed by Rich Nardone,
NPSHA, gage global product manager at ITT Goulds Pumps (Sen-
eca Falls, N.Y.;
Figure 1. Shown here is an example of an NPSH test plot used to determine a pump’s NPSHR
Feature Report

Lifecycle Costs for

Capital Equipment

In the CPI Operation


Longterm equipment costs

need to be fully considered Decommis-

in capital-cost assessments
FIGURE 1. Initial capital costs alone
are too often the primary criteria for se-
Jeff Hoffmann lecting process equipment
Paul O. Abbe

hen considering project pro- ment parts available for both routine Lifecycle cost
posals for new processes in and non-routine maintenance? The purpose of lifecycle-cost (LCC)
the chemical process indus- analysis is to make informed deci-
tries (CPI), capital equip- Minimizing total cost sions based on available alternatives
ment costs often become the primary The emphasis on total operating in order to achieve the most economi-
focus. The purpose of this article is costs over the life of a process does cal process from inception to decom-
to provide a detailed examination of not imply that the initial equipment missioning. LCC takes into account
the total cost of process equipment costs are unimportant. On the con- the design, equipment selection, op-
and the implications that the initial trary, it is precisely the investment eration, maintenance and final dis-
equipment cost has for longterm costs in the correct equipment in the first position costs of a project over its
over the full life of the process. place that is to be examined. The pur- lifespan. LCC is useful for engineers
Aside from equipment costs, other pose of the procurement of process in justifying equipment and process
critical costs to consider include, op- equipment is to perform a particular design based on total costs rather
eration, maintenance and decommis- function within a unit operation. The than the initial purchase price of
sioning (Figure 1). Also, since a pro- goal is not the purchase of a particu- equipment alone.
cess generates revenue only when it lar piece of equipment. If we add the Procurement strategies focused on
is operating, downtime must be added dimension of time, then our defini- lowest initial costs are more likely
to the total costs. When the whole life- tion for process equipment becomes to lead to higher longterm costs. We
time of a process is considered, equip- a piece of equipment that performs a are often directed to reduce costs and
ment costs may account for as little as specific function under various con- work within budgets. In the short run,
5–10% of the total cost (Figure 2). ditions over a prescribed period of this approach can make us and our
There are a number of questions time. Therefore, we should not focus department appear efficient. How-
that should be considered before mov- on equipment with the lowest initial ever, the lower initial capital costs
ing ahead with projects. How should cost, but rather on the realistic long- may come with maintenance or other
you define the product output, quality, term cost of that purchase. problems that eventually will be real-
unit operations, support equipment In the early 1980s, Edward Deming ized by the company shareholders in
and profitability? Who is responsible — the father of quality management the coming years and decades. LCC
for operating and maintaining the — stated that organizations should can help avoid unnecessary downtime
process? Do the demands for process “end the practice of awarding business and help make a process more com-
performance conflict with operating on the basis of price tag alone and, in- petitive and profitable. At the very
and maintenance realities? What is stead, minimize the total cost.” This least, an LCC analysis may prompt
the likelihood that the equipment sentiment is consistent with evaluat- engineers to consider a wider range
will operate trouble-free? Are replace- ing lifecycle cost. of possibilities.
36 Chemical Engineering July 2013
Installation Environmental
5% 3%
Flush water Process
7% requirements

Decommissioning Define unit
all or part of
8% operations

Supplier Define
designs equipment
Capital costs


Operating cost
9% Operation Anticipate
Electric downtime risk
Causes of failure
Maintenance • Equipment design
• Proper operation
Figure 2. Initial capital costs represent a small fraction of the total lifecycle costs Maintenance
• Maintenance
for process equipment. The graph depicts the case of a “worst case” situation with • Parts availability
considerable downtime costs (see Table 1, scenario 1, p. 41)

The remainder of this article pres- • Maintenance costs mission
ents a more-or-less qualitative view • Decommissioning costs
of the LCC analysis process and the These components are further subdi- FIGURE 3. The LCC analysis process
elements that go into LCC. The “Fur- vided (Figure 4). shown here is designed to minimize total
ther reading” list at the end of the cost, even if initial capital costs are higher
article refers readers to several more Steps in LCC analysis
analytic versions of LCC, includ- LCC considers everything in the life tem be modified to accommodate in-
ing Weibull analysis, risk-based cost of a process, starting with a definition creased output of product, changes
analysis, Monte Carlo modeling, and of the process, its unit operations, and in formulation or the addition of a
other “what-if ” analyses. the equipment required to fulfill those step in the process?
The main goals of LCC are: 1) To unit operations, as well as operating • Quantify waste. What percent of
identify risks to process operation costs, maintenance costs and finally waste is acceptable? What is the cost
and efficiency; 2) Quantify these risks decommissioning costs. The following of waste disposal? How can waste be
in terms of downtime; and 3) Deter- are the major steps involved in deter- minimized by a change in the pro-
mine how to avoid these risks and mining LCC. cess? Can off-specification product
subsequent losses early in the design be reprocessed or sold “off-spec” to a
of the system (Figure 3). Assess process requirements. different market?
Tasks to consider when undertaking a
LCC for the CPI new process include the following: Define unit operations
Of all the industries and all the types • Determine present and future ca- This step involves identifying the unit
of manufacturing plants in the world, pacity for the product operations and types of equipment re-
it is safe to say that the process in- • Anticipate the lifetime of the pro- quired by the process.
dustries are some of the most variable cess. Some processes may have a Subcontract. Subcontracting one or
and complex. With more than 70 mil- lifespan of anywhere from a year or more operations in a process is some-
lion identifiable chemicals and a near- two to decades. Anticipating process thing often overlooked, but can in-
infinite number of combinations, and lifetime will either concentrate or crease cost efficiencies and flexibility.
given the number of unit operations extend cost impacts and affect the Few manufacturers of process equip-
possible, there are many opportuni- long-term maintenance and reliabil- ment manufacture everything — mo-
ties to examine process costs. The four ity of the process tors, gear drives and bearings are not
primary components involved in the • Define product quality based on cus- manufactured in-house. Likewise,
LCC are: tomer requirements chemical companies do not manufac-
• Capital equipment costs • Determine process flexibility. How ture all of their raw materials, nor
• Operating costs easily can the equipment and sys- do they necessarily perform all tasks
Chemical Engineering July 2013 37
Feature Report

in-house. Subcontracting is, for most

businesses, a matter of degree rather Equipment
than a yes-or-no decision. Some steps sizing and design features

in a process may not be cost-effective Pipes, valves, Installation

to execute in-house. For example, fitting, electrical rigging
high-pressure reactors, spray-drying Foundations
and structure
or packaging may best be outsourced Warranties
operations, at least until the opera-
tion grows and the investment can be
better justified.
Continuous, batch or a combina- Preventative
tion. The decision for a continuous or supervision

batch process (or a combination of the Raw monitoring Maintenance

material training
two) is sometimes dictated by the pro- supply
cess, and sometimes optional. Within Lifecycle Unexpected
Waste stream Operation Maintenance failure
this decision, a set of factors should treatment and
cost maintenance
be considered: disposal
Parts supply
• Continuous process operations can Utilities: chain
often have much higher output electric, N2, CO2
gas, water inventory
and may require less equipment,
but they may have more vari- Decommis-
ability in quality and reworking sioning
off-spec product in a continuous
process may be difficult
FIGURE 4. Each of Dismantling Demolition
• Batch operations may require more the four main LCC and removal
Sell or scrap
storage and intermediate buffer components can be Cleanup equipment
tanks and larger equipment, but further broken down
into specific cost Waste disposal
they have the advantage of con- (including contaminated
sistency and often have a better pipe, insulation,
and so on)
chance to re-work off-spec product
Storage strategy. In anticipation of
routine or emergency shutdown, a stor-
age strategy should be created. Can the tions. Discuss your requirements with be better cleaned manually. Also,
finished product be stored and, if so, equipment manufacturers and gather it is important to understand what
can the downstream process or packag- information on: performance; design; level of operator exposure to product
ing accommodate a surge in capacity? options; installation; foundation and and cleaning chemicals is acceptable.
Process bottlenecks. Which aspects support requirements; utility require- Other options might include mainte-
of the process have the most variation? ments; mean-time between failures; nance-reducing features, such as ad-
For example, liquid mixing is fairly and recommended spare parts for ditional access hatches, sight glasses
consistent, whereas solids drying can the first few years of operation. This and lights, split seals and bearings
vary considerably with particle size. is also the time to start gathering in- and replaceable wear liners.
Does a dryer need excess capacity? formation on refurbished and used
Evaporator capacity can fall off quickly equipment (discussed later). The steps Equipment installation
due to tube fouling either on the prod- are as follows: Installation costs may equal or exceed
uct or heat-transfer-fluid side. In the 1) Identify suppliers and dealers for equipment costs, depending on the size
example on page 39 (Figure 5), the new versus used versus reconditioned and complexity of the equipment. An
performance of the evaporator falling equipment. Identify alternate designs important consideration during the
below 600 gal/hr can be the result of (for example, shell-and-tube versus layout and installation of equipment
scale build-up or fouling. Investing in plate-and-frame heat exchanger, or is the accessibility to allow preventive
a water demineralization system may fluid-bed versus vacuum dryer). maintenance and future repair. Suffi-
be worthwhile if the bottleneck affects 2) Identify design features that may cient space must be provided for the
productivity and profitability. Like- improve product quality, increase up- extraction of shafts, rotors and motors,
wise, too large an evaporator with low time and reduce maintenance. These as well as to provide access to seals and
velocity may be more prone to fouling. might include automatic lubrication, bearings. Overhead structure should
Bigger is not always better. and monitoring devices for vibration, allow for portable hoisting chains or
over-temperature and low-level pro- permanently installed hoists.
Define required equipment tection. Evaluate whether a clean-in- Although not routine, anticipating
Process equipment has many varia- place (CIP) system would be cost-effec- the removal of large pieces of equip-
tions in basic design and design op- tive, or whether the equipment would ment should not be made impossible
38 Chemical Engineering July 2013
FIGURE 5. • What are the availability of stan-
1,100 Certain aspects dard parts?
of a process
have higher
• What is the availability of special
1,000 Design non-standard parts?
potential to
present process • What is the cost to purchase and
900 bottlenecks than stock the recommended parts?
others • Will your OEM put consigned stock

800 in your facility? What is involved in

administering consigned stock? Are
700 you prepared to safely store and pro-
tect the parts?
600 • What is the cost to stock parts for
catastrophic failures? Some large
500 parts, such as motors, gear drives or
Mixer/ Evaporator Packaging
Metering reactor centrifuge scrolls and bowls, can take
pumps Filtration Dehydrator weeks or months to obtain. The low
probability of failure may be offset by
the very long lead times and may re-
quire investment in costly parts that
by physical constraints. Without clear industry, have to dispose of waste prod- may sit on the shelf for years.
access, preventative maintenance uct or waste streams from washing, off- • Is maintenance staff knowledgeable
may suffer and repair time may be ex- specification product or simply contami- and prepared to identify symptoms
tended. Factors involved in the instal- nated water coming from a wash step. of failure early, and diagnose and
lation cost may include the following: repair issues quickly and correctly
• Machine foundations Maintenance the first time? Check with the OEM
• Accessibility for maintenance and Generally, maintenance can be classi- for guidance and training. Do you
repair fied into two types: preventative and have the installation and operat-
• Support structures and mezzanines repair. Some failures occur randomly ing manuals on file? Have they been
• Piping, valves and fittings and cannot be predicted, but other thoroughly reviewed?
• Instrumentation failures occur as a result of a lack of • What are the anticipated preventa-
• Electrical controls preventative maintenance (PM). tive maintenance intervals?
• Monitoring equipment PM is an area that has evolved into a • What is the expected time between
• Electrical switchgear service that can be subcontracted and failures for components like seals,
may be economical when considering and bearings, and the expected time
Operation the total longterm value provided. PM between belt adjustments and filter
Operation and maintenance are two companies often have superior knowl- replacements?
areas that are critical to avoiding edge of pumps, drives, lubrication and • Should all or some PM be out-
downtime and both are affected by routine maintenance issues, including sourced?
equipment selection, design and oper- good record keeping. The PM record • Does the OEM offer PM services?
ating procedures. keeping can also help support any • What is the repair turnaround time
If the equipment was sized prop- warranty claims and avoid disputes for a specific failure?
erly, there should be no reason to op- with original equipment manufactur-
erate it beyond safe design capacities. ers (OEMs). The cost of subcontracting Decommissioning
Many types of equipment are tested PM must be considered against the The concept of decommissioning is
at, or designed for 150 to 200% of the benefits of avoiding downtime. Parts not something most engineers tend
rated capacity, but operating at these availability is important in avoiding to consider as they are designing a
capacities may risk shortening the downtime both for PM and unexpected plant, but some plants will have finite
life of the equipment. Other aspects failures. Questions to consider in hav- lives of just a few years due to licens-
of operation costs include the train- ing parts available when required are ing agreements, patents, changes in
ing of operations personnel, utilities the following: markets or plans to shift to overseas
(electricity, gas, water, steam and • Do you know the supply chain for production in the future.
cooling tower capacity) and the time the parts you need? Planning for decommissioning a
that the equipment is offline for pre- • Do you know your OEM parts and process plant can vary from simple
ventative maintenance. service contacts? tear-down and selling of equipment
The costs of raw materials, water • Are you considering non-OEM or to preparing for a sophisticated de-
treatment (demineralizing, pH adjust- counterfeit parts? contamination procedure. Chemical
ment), purge gas (N2, CO2) and waste • Do you have a recommend parts list process equipment has special con-
disposal are also key operations costs. for each machine for the first few siderations that can increase the cost
Most CPI processes, even in the food years of operation? of decommissioning. Not only will
Chemical Engineering July 2013 39
Feature Report

discover failure Obtain

and remove Teardown Dagnosis Repair Restart Monitor
from service

FIGURE 6. Downtime represents one of

the most significant and costly issues
for many processes
5 $500 $5,000 $50,000 $500,000 $5,000,000
waste material have to be disposed
of, but piping, insulation and floor-
ing may have to be decontaminated
4 $400 $4,000 $40,000 $400,000 $4,000,000
or treated as hazardous waste.
Other costs of decommissioning Events per year
include dismantling of equipment,
3 $300 $3,000 $30,000 $300,000 $3,000,000
waste disposal of chemicals (unused
chemicals, water-treatment and
cleaning chemicals, as well as those
2 $200 $2,000 $20,000 $200,000 $2,000,000
in above- and below-ground tanks,
evaporation ponds and contaminated
1 $100 $1,000 $10,000 $100,000 $1,000,000
Costs of downtime
Process downtime is one of the most $100 $1,000 $10,000 $100,000 $1,000,000
significant and costly issues for many Cost of event
processes. To properly take into ac-
count the costs of downtime over the
life of the process, engineers must FIGURE 7. A matrix can help quantify the costs of downtime in a process
estimate how much cost is accrued if
the process fails, either in whole or in
part. Further, once it fails, the ques- One factor with a great impact on re- the 10-year life of the process, the sav-
tion becomes how long will it take to ducing downtime is the availability of ing is $890,000.
restore operation? parts. The parts may be common, such Those examples represent the costs
In terms of equipment selection and as O-rings or gaskets, seals or bearings, of just one critical unit operation and
design, which equipment and design or they may be less common, such as one design feature. When consider-
features will be less likely to cause pump housings or drive shafts. ing similar analyses across an entire
downtime? Which will be most eas- If a complete shutdown costs plant, the cost savings can be sub-
ily maintained? How quickly can an $100,000 per day, the expected fre- stantial. The conclusion of the above
expected failure be repaired so the quency of a catastrophic shutdown is that a relatively small upfront in-
equipment can be put back in service? three times per year is a total of vestment may save considerable cost
The risks and costs of process down- $300,000. For a non-critical failure in the long run.
time can be considered in a semi- that reduces productivity, but does
quantitative form by examining the not shut down the process entirely, New, USED, refurbished?
likelihood of an event occurring in a an event cost of $50,000, with a fre- There is a saying that all process
given time period and the cost per unit quency of five times per year would plants run on used equipment, and
time of that failure. total $250,000 (Figure 7). that is true. The LCC analysis is not
The following examples (Figure 8 prejudiced with regard to used or re-
Downtime Cost = f requency of
and Table 1) emphasize maintenance furbished equipment — LCC considers
training and parts availability in the the balance of downtime prevention
x downtime/days
prevention of downtime. Suppose the and investment in equipment and pre-
x $ losses/day
additional cost of training and parts ventative maintenance. If you know
Downtime starts with the failure of is $80,000. With downtime cost at your process requirements and have
the equipment and stops when it is $20,000 per day, the investment of the resources to keep used equipment
put back in service. Better mainte- $80,000 saved $86,000 compared to functioning as reliably as necessary,
nance training can reduce the diagno- without the training and parts after then used or refurbished equipment
sis and repair time significantly. The just one outage event. is the right choice. There are several
basic sequence is the discovery of a In a second example, a $70,000 design LCC issues to consider when deciding
failure, followed by teardown, diagno- feature reduces downtime by making a between new, refurbished and used
sis, obtaining parts, repair, restart and routine and expected part replacement equipment.
monitoriing (Figure 6). faster, from three days to one day. Over Not all buyers are in a position to
40 Chemical Engineering July 2013
Downtime Costs
Dwell time, days
Sequential Scenario Scenario Annual
step Event Preventative action cost
1 2
1 Discover failure 1 0.1 Maintenance training $ 30,000
2 Teardown 1 0.5 Maintenance training
3 Diagnosis 1 0.5 Maintenance training
4 Accquire parts 5 0.1 Parts in stock $ 50,000 $150,000
5 Repair 1 0.5 Maintenance training
6 Re-start 1 0.5 Maintenance training
7 Monitor 1 0.5 Maintenance training
Total days of downtime 11 2.7 $ 80,000
Investment in training $50,000
$0 $80,000
and parts
Downtime cost $20,000/day $220,000 $54,000
Total $220,000 $134,000 $0
Scenario Scenario
1 2
Additional cost of
lack of preparation

FIGURE 8. Different downtime scenarios for availability of parts and other factors can yield variable costs

TABLE 1. The impact of varying downtime costs

Scenario 1 Year
High-pressure 1 2 3 4 5 6 7 8 9 10 Total
reactor mixer
Initial capital costs $380,000 $380,000
Installation and $230,000 $230,000
Utilities - electric $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $600,000
- flush water ($0.04/gal.) $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $300,000
Operating costs $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $400,000
(normal supervision)
Maintenance costs $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $600,000
Downtime costs $144,000 $144,000 $144,000 $144,000 $144,000 $144,000 $144,000 $144,000 $144,000 $144,000 $1,440,000
($48,000/d x 3 d)
Environmental costs $13,000 $13,000 $13,000 $13,000 $13,000 $13,000 $13,000
$13,000 $13,000 $13,000 $130,000
Decommissioning $350,000 $360,000
Total $347,000 $347,000 $347,000 $347,000 $347,000 $347,000 $347,000 $347,000 $347,000 $697,000 $4,430,000

Scenario 2 Year
High-pressure 1 2 3 4 5 6 7 8 9 10 Total
reactor mixer
Initial capital costs $380,000 $450,000
Split seal and bearing $70,000
Installation and $230,000 $230,000
Utilities - electric $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $600,000
- flush water ($0.04/gal) $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $30,000 $300,000
Operating costs $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $40,000 $400,000
(normal supervision)
Maintenance costs $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $60,000 $600,000
Downtime costs $48,000 $48,000 $48,000 $48,000 $148,000 $48,000 $48,000 $48,000 $48,000 $48,000 $480,000
($48,000/d x 1 d)
Environmental costs $13,000 $13,000 $13,000 $13,000 $13,000 $13,000 $13,000
$13,000 $13,000 $13,000 $130,000
Decommissioning $350,000 $350,000
Total $931,000 $251,000 $251,000 $251,000 $251,000 $251,000 $251,000 $251,000 $251,000 $601,000 $3,540,000

Chemical Engineering July 2013 41

Feature Report

purchase new equipment be- Table 2. Comparison of New, Used and REfurbished EquipmenT
cause of cost considerations New Refurbished Used
or time. Used equipment may Application Application definition Limited Limited to none
also be the appropriate alter- assistance and machine design
native when time is a con- Design fea- Unlimited Some variations or None (whatever is
sideration, either in terms of tures modifications possible in stock)
delivery or usage. Used equip- as part of the rebuild
ment is frequently available
Delivery 4–8 months 1–2 months Immediate
for immediate delivery, com-
pared to the relatively long Price 100% 40–50% of new 20–40% of new
lead times that are typical Mechanical 12 months from instal- 90 days to a few None (“as is”)
of new capital equipment. In warranty lation or 18 months
from shipment
these cases, used equipment
Right to return None None 10–30 days
may provide the optimal al-
Parts In-stock or readily The fact that the unit Call OEM and find
ternative (Table 2). availability available is being refurbished out how available
The following scenarios indicates that parts parts are before pur-
favor the purchasing of used are available from the chasing. Parts avail-
or refurbished equipment: OEM. Variable parts ability diminishes
• When price is of prime im- availability with time
portance because of invest- Aftermarket Complete technical Limited None
technical support
ment limitations support
• When the equipment is
needed immediately for an
emerging market tant to be sure you can obtain parts acquire equipment that will accommo-
• When the equipment will be used when needed, especially if the OEM date your process.
for a limited time, such as a feasibil- is located in another country. Trying Mechanical warranty. Mechanical
ity study or short-production run for to get parts for an overseas machine warranties are a certainty with new
a special product or market made 30 years ago, for example, may equipment, but their real purpose
• When the equipment can be eco- be a challenge. Is the company still should not be overestimated. Warran-
nomically modified to fit the pur- in business? Where are their foreign ties are not substitutes for proper op-
pose. This will have a lot to do with offices? Some resourceful companies eration or preventative maintenance
your ability to refurbish and main- have recognized a gap in the supply and should not be construed as pro-
tain the equipment chain and decided to manufacture cess guarantees. Mechanical warran-
• When the process is routine, low parts for older domestic or foreign ties provide benefits especially during
output or low risk. Infrequently equipment. Once you find them, you the initial startup period. If faults
run equipment will have more op- may be in good shape. arise, they will likely occur during the
portunity for PM and will be more Aftermarket technical support. initial warranty period.
“forgiving” With new equipment, the availabil- Avoid surprises and disappointment
ity of good aftersale support is almost by verifying the specifics of the warran-
Aftermarket support assured. But when purchasing used ties before purchasing.
Most companies that manufacture equipment, the OEM may or may not Delivery timing. The delivery time
process equipment would rather sell provide adequate technical support. for used equipment is typically just
new, but most are quite pleased to Find out if drawings, manuals and days, while new equipment will likely
support their older equipment. parts lists are available. They may be months.
Not every company has the same charge $500 to $1,000 for these docu- Design features. Within limits, new
business model. It is important to ments, but it is a good investment to equipment can be outfitted with vir-
know your equipment and the parts ensure you have the right information tually every manner of control, CIP
supply chain. on hand. systems, quick access to internal
The following are some areas of com- Application assistance. There is no parts, and other features to improve
parison that must be considered when doubt that a new equipment manufac- productivity and uptime. Used equip-
deciding between new, refurbished or turer has a vested interest in guiding ment is sold “as-is,” so you will either
used equipment: you toward the correct equipment for need to find a good match or compro-
Aftermarket parts. This is a very- your application. Due to the nature of mise on the features you would like
important consideration for mainte- chemical processing, subtle changes to have. Refurbished equipment may
nance and repair turnaround time. in product characteristics can have present some opportunities for up-
No matter if you are considering new significant effects on the process and grades and modifications.
or used equipment, you should con- the equipment, which is why process Price. New equipment is not expen-
tact the OEM to find out the avail- guarantees are very rare. It is in the sive if you buy into Edward Deming’s
ability of parts. It is especially impor- best interest of the OEM to help you idea that you are purchasing total
42 Chemical Engineering July 2013
value. If you only consider price, then is made and work is undertaken. erating, maintenance and decommis-
new equipment may appear to be more Most used equipment dealers will sioning costs. The other major longterm
costly. LCC is blind to new versus used allow equipment returns within 10 to cost is the cost of downtime compared to
equipment, so let the risk data fall 30 days if it does not work as antici- investments in training, preventative
where they may. pated. All dealers differ, so it is impor- maintenance and spare parts. Lifecycle
Right to return. With new equip- tant to ask specifically before making cost analysis can be done in a rudimen-
ment, once you have placed the the purchase. tary fashion or it can employ complex
order, you are essentially committed what-if algorithms, but in either case,
to the equipment. Backing out after Concluding remarks the benefits of taking a broader view of
the initial deposit has been made Understanding the lifecycle costs of one the factors that may impact the long-
will have some definite costs. If you piece of equipment or an entire process term cost of a process will benefit you
buy refurbished equipment, you requires examining not just the cost of and your company. ■
are also committed once a deposit the capital equipment, but also the op- Edited by Scott Jenkins

Further reading Jeff Hoffmann is a vice
Hydrocarbon Processing Industries, Fourth president at Paul O. Abbe
Abernethy, Robert B. The New Weibull Handbook Co, (735 East Green Street,
(4th ed.). North Palm Beach, Fla., 2002. International Conference on Process Plant
Reliability, Gulf Publishing Company, Bensonville, IL 60106; Phone:
Landers, Richard R. Product Assurance Diction- Houston, Tex., 1995. 630-258-4720; Email: jhoff-
ary, Marlton Publishers, Marlton, N.J., 1996. Hoff-
Goble, William M. Evaluating Control Systems mann has an educational
Bloch, Heinz P. and Fred K. Geitner. Practi- Reliability, Instrument Society of America, background in chemistry and
cal Machinery Management for Process Research Triangle Park, N.C., 1992. a M.S. in industrial and orga-
Plants,Volume 2: Machinery Failure Analysis nizational psychology. During
and Troubleshooting, 2nd Edition, Gulf Pub- Ireson, W. Grant, Clyde F. Coombs Jr., Richard the past 20 years, Hoffmann
lishing Company, Houston, Tex. 1994 Y. Moss. Handbook of Reliability Engineer- has held sales, marketing and
ing and Management, 2nd edition, McGraw- executive positions at several process equipment
Bloch, Heinz P. and Fred K. Geitner. Simplified Hill, New York, 1996. companies. He also holds six U.S. patents for
Life-Cycle Cost Computations Applied in the various process equipment designs.


is more than just a building supplier, we are the Blast Resistant Building authority. We specialize in

assisting you from the CONCEPT to the DESIGN, with your help and your site study. We use that concept and
design to BUILD your solution exactly the way you want. We then place it wherever you desire, when you need it.
To learn more call 877.522.6948 or visit

Circle 2 on p. 56 or go to
Chemical Engineering July 2013 43
Cover Story

A Primer on Reverse
Osmosis Technology
Desalination by reverse osmosis
is a key technology for a
water-constrained world.
Discussed here is
its use in industrial
water treatment and
drinking-water production
Reverse osmosis has
Richard L. Stover become the most efficient and versatile
Desalitech, Inc. technology for desalination and water purification

ater scarcity is one of the Industrial operations consume an from saltwater [4]. Distillation
most serious global chal- estimated 60% of all freshwater plants for desalination still oper-
lenges of our time. Desali- withdrawals in developed countries ate today in some regions of the
nation and water reuse like the U.S. Together, industry world where energy is abundant
are effective and reliable means and agriculture are responsible for and inexpensive. However, the vast
to provide new water resources. about 90% of the freshwater con- majority of desalination and water-
Among the many methods available sumed globally [2]. Efficient and purification plants that have been
to treat water, reverse osmosis (RO) sustainable water treatment meth- constructed recently or are planned
has widely demonstrated superior ods for industrial and agricultural employ RO technology.
reliability and cost-effectiveness at water supply are imperative. Thanks to technological improve-
removing dissolved species, such as Water should be conserved, recy- ments over the last 20 years, the
salts and trace contaminants. This cled and reused to reduce the stress cost to produce freshwater with RO
article reviews the development and on water supplies, but this will not has been reduced by a factor of four
state of the art of RO for industrial meet the increased demand posed or more, and the process has become
water treatment, wastewater treat- by population and economic growth. a reliable component of municipal
ment and drinking water produc- The treatment of salty water re- and industrial infrastructure (Fig-
tion. It also reviews technological sources and wastewater reuse of- ure 1). At present, RO is the most
advances that have improved pro- fers new and reliable sources of energy-efficient and versatile tech-
cess performance, reliability and re- freshwater, without impairing ex- nology for desalination and water
duced cost, and describes currently isting freshwater resources. Today, purification and is the benchmark
available components and methods. an estimated 300 million people in for comparison for any new water-
Finally, it considers the prospects for 150 countries already rely on de- purification technology [5,6].
future advances in the field. salinated water. In 2016, the global RO is most commonly known for
water production by desalination is its use in purifying drinking water
Water scarcity and desalination projected to exceed 10 trillion gal/ from seawater, brackish water or
The world’s freshwater resources yr (38 billion m3/yr) — twice the contaminated water, where RO re-
are under tremendous pressure. rate of global water production by moves salt and other dissolved or
Over one-third of the world’s pop- desalination in 2008 [3]. suspended materials from feedwa-
ulation lives in a water-stressed Early desalination and water pu- ter. Many are surprised to learn
country, and by 2025, this figure rification methods included distil- that its use in other applications is
is predicted to rise to nearly two- lation, which uses large amounts widespread and well-established.
thirds of the global population [1]. of energy to evaporate freshwater For example, RO is used to remove
38 Chemical Engineering July 2014
HP RO systems operate at high recovery
rates, where recovery is defined as
the ratio of permeate flow to source-
Figure 2. In a typical pump water flow. However, high recovery
RO system, a high-pressure increases the risk of membrane con-
(HP) pump forces feedwater through tamination, presents challenges for
membrane elements arrayed in parallel maintaining crossflow, reduces per-
meate quality and can reduce the
minerals from boiler water at power The minimum pressure required flexibility of the process to handle
plants, to clean effluent and brack- to separate pure water from im- feedwater variations.
ish groundwater, and for concen- pure water can be considered as a Many of these challenges have
trating food liquids, such as milk. barrier. The height of this barrier been met with emerging technolo-
depends upon the osmotic pressure gies that enhance the utility of RO.
Reverse osmosis principles or osmotic potential of the water,
Osmosis is the natural movement which in turn depends upon its RO systems
of water from an area of high water salinity and composition. Saltier RO membranes are incorporated
concentration (low salt concentra- water has a higher osmotic poten- into systems that deliver source
tion) through a salt barrier to an tial and requires more pressure water to the membranes, apply
area of low water concentration to desalinate. To drive permeate crossflow that sweeps concentrate
(high salt concentration). Flow is through membranes at reasonable from the membrane surface, and
driven by the difference in osmotic fluxes, the osmotic potential must provide a pathway for conveying
potential of the two solutions, quan- not only be met, but overcome. This water that permeates through the
tified as osmotic pressure. Applying overpressure depends partly upon membranes. The primary com-
an external pressure to reverse the the permeability of the membranes. ponents of an RO system are the
natural flow of water through the In addition, the salt concentration membrane elements, the pumps
barrier is RO. immediately adjacent to the mem- and the devices used to manage
The process of osmosis through a brane surface is elevated, which flow and pressure. Typically, a high-
semipermeable membrane was first increases the osmotic barrier above pressure pump provides both the
observed in 1748 by Jean-Antoine that in the bulk of the saltwater — driving force for the separation and
Nollet. RO was known in the 1950s, a phenomenon known as concentra- the crossflow, but several current
but was not practically demon- tion polarization. Concentration po- process designs use a circulation
strated until the early 1960s, with larization is reduced with crossflow. pump to drive crossflow, so that the
the discovery of asymmetric mem- Other non-idealities in RO sys- role of the high-pressure pump is
branes at the University of Cali- tems that elevate energy require- reduced to pressurizing the system
fornia at Los Angeles (www.ucla. ments include resistance to flow and driving permeate flow.
edu) by Sidney Loeb and Srinivasa through the membranes, mechani-
Sourirajan [7]. These membranes cal and volumetric efficiency losses RO process configurations
were characterized by a thin “skin” in pumps and viscous friction losses A typical system for industrial RO
layer supported atop a highly po- in flowing water. Also, membrane is shown in Figure 2. A high-pres-
rous and much thicker substrate. surfaces can be blocked with con- sure (HP) pump feeds the system.
This basic structure remains the taminants, including organic and Pressure vessels containing mul-
basis of modern RO membranes. inorganic foulants or salts that pre- tiple membrane elements are ar-
When a salt solution is pressur- cipitate out of solution. rayed in parallel. Each pressure
ized against an RO membrane, For RO treatment of brackish, vessel contains six or seven mem-
impurities are retained on the industrial and wastewater (col- brane elements in series, with the
pressurized side of the membrane lectively referred to here as in- concentrate of each element being
as brine and purified water flows dustrial RO) salinity levels and fed to the subsequent element. One
through the membranes as perme- the corresponding feed pressure stage typically achieves 50% re-
ate. RO requires flow across the requirements are typically lower covery, and higher recovery rates
membranes, known as crossflow, to than those for seawater RO. There- require multiple stages. The pro-
keep the membrane surfaces clear fore, energy requirements are usu- ductivity of multiple stages are bal-
of concentrate and allow continu- ally not a major cost consideration. anced by throttling the permeate
ous and almost constant flow of However, the cost and practicality flow from front stages and boosting
permeate. This differs from con- of brine disposal and the need for feed pressure to later stages with a
ventional filtration processes, in the process to handle feed composi- booster pump (BP). The number of
which impurities embed in the fil- tion changes are more significant in pressure vessels arrayed in paral-
ter or build up as a cake that must industrial RO applications. To limit lel in each subsequent stage is re-
be backflushed or removed periodi- brine production and achieve a high duced to maintain sufficient cross-
cally to restore productivity. yield of purified water, industrial flow in later stages. Finally, brine is
Chemical Engineering July 2014 39
Cover Story



Figure 3. Closed-circuit RO processes utilize brine recir- Figure 4. Seawater RO processes require higher pressures that
culation to achieve high recovery without multiple stages those in industrial RO, making energy consumption more important

throttled out of the system through control panel provides substantial (TFC) membranes were developed
a valve or a device that regulates flexibility. The frequent flushing of [12]. TFC membranes exhibit much
system pressure. the system by the brine-feedwater higher intrinsic water permeabili-
At 50% recovery per stage, four exchange also helps suppress foul- ties than cellulose acetate mem-
stages are required to achieve over ing and scaling, such that higher re- branes because of their extremely
90% total recovery. Multiple stages covery rates can be sustained. thin selective layers. Today, nearly
add design and operational com- In seawater RO, higher pressures all RO operations use TFC mem-
plexity and reduce process flexibil- are generally required than those branes [13]. Despite the great im-
ity. Alternately, recovery can be in- used in industrial RO. This makes provements in TFC membrane per-
creased by continuously returning energy consumption more impor- formance and cost, there are still
some of the brine to the membrane tant and it limits recovery rates. shortcomings that hinder their ap-
feed. However, the increase in feed, A typical seawater RO process is plication. These limitations include
brine and permeate salinity caused illustrated in Figure 4. Like in the being prone to fouling and being
by brine recirculation usually un- closed-circuit industrial RO system, susceptible to attack by oxidizing
dermines the benefits it provides. the seawater process uses both a agents, such as chlorine.
Emerging closed-circuit (semi- high pressure pump and a circula-
batch) RO processes, which utilize tion pump to feed the membrane Pumps
brine recirculation in a batch-like array. In addition, an energy recov- The high-pressure pump supplies
operation, provide a new means to ery device (ERD) is used to remove the pressure needed to push water
achieve high recovery without mul- brine and replace it with feedwater through the membranes and, in
tiple stages or reduced permeate while maintaining system pressure. the process illustrated in Figure 1,
quality [8,9]. Such a process is illus- These devices save energy by re- supplies the crossflow that controls
trated in Figure 3. A high-pressure covering hydraulic energy from the concentration polarization. Typi-
pump feeds a closed loop comprised reject brine and returning it to the cal pressures for industrial water
of a single-stage of membrane el- low-pressure feed process, thereby treatment range from 150 to 450
ements and a circulation pump reducing the duty of the HP pump. psi (10 to 31 bars). Seawater RO
(CP). Permeate is produced at a For lower-recovery, higher-pres- pressures range from 750 to 1,200
rate equal to the flowrate of the HP sure applications, such as brine psi (52 to 83 bars). Typically, cen-
pump. Brine is recirculated without concentration and seawater desali- trifugal-type pumps are used, with
depressurization. When a desired nation, an alternative closed-circuit sufficient stages and features to
recovery percentage is reached, process is used to minimize energy meet the pressure requirements.
brine is flushed out of the system, requirements [10,11]. In this pro- In some high-pressure applications,
displaced by feedwater from the cess configuration, brine is displaced positive-displacement pumps are
HP pump. The exchange of brine from the membranes with pressur- preferred because of their generally
and feedwater is executed without ized feedwater from a side cham- higher efficiency.
stopping the HP pump or producing ber. The exchange, emptying and The interstage booster pump
permeate and without depressur- refilling of the side chamber is done shown in Figure 1 adds enough
izing the system. The process then under hydrostatic conditions with pressure to the concentrate stream
returns to closed-circuit operation, almost no loss of pressure energy. to overcome its increased osmotic
during which there is no brine re- pressure in later stages. Like the
ject stream. Membranes HP pump, the booster pump drives
Over 97% recovery has been RO processes are built around both permeate flow and crossflow. It
achieved in closed-circuit RO pro- semipermeable membranes capable is sized and operated to “balance”
cesses. Energy requirements are of filtering out salts. The first mem- flux and flow between the stages so
reduced because the average mem- branes were made of cellulose ac- that all individual membrane ele-
brane feed pressure required to etate and served as the best avail- ments operate within their design
achieve a given recovery rate is lower. able technology until the 1980s, envelopes and are neither over- nor
The ability to change recovery at the when robust thin-film composite under-utilized. Typical boost pres-
40 Chemical Engineering July 2014
Table 1. Approximate Costs of a 367 gal/min (2,000 m3/d) Industrial RO Unit (10-yr life)
Industrial RO 75% Recovery 90% Recovery
Present Cost, Cost per Percent of Present Cost, Percent Cost per
value, 10 yr gal/d 1,000 gal otal Value, 10 yr gal/d of total 1,000 gal
Capex $521,195 $0.99 $0.28 6% $521,195 $0.99 12% $0.28
Energy $480,727 $0.91 $0.26 6% $600,909 $1.14 14% $0.32
O&M $1,148,252 $2.17 $0.62 13% $1,148,252 $2.17 26% $0.62
Brine $6,436,578 $12.18 $3.48 75% $2,145,526 $4.06 49% $1.16
Total $8,586,752 $16.25 $4.64 100% $4,415,882 $8.36 100% $2.39

sures for industrial water treat- (OPEX). CAPEX per volume of reduces brine disposal costs by a
ment range from 40 to 200 psi (3 to water produced depends upon the factor of three, and cuts the overall
14 bars). In some process configu- construction cost and the amortiza- cost of ownership almost in half.
rations, a hydraulic turbocharger tion rate of the plant, the interest
is used as an inter-stage booster rate (the desired yield on the capi- Seawater RO
pump, driven by flow and pressure tal investment) and the plant utili- Seawater RO installations are typi-
of brine from subsequent stages. zation (load factor). OPEX consists cally larger than industrial RO in-
The role of the circulation pump of the fixed costs of staff, insurance stallations, and are tendered for up
shown in Figures 2 and 3 is to pro- and so on, as well as the variable to 25-year expected lifetimes. A typi-
vide crossflow. These pumps only costs of operation and maintenance cal CAPEX requirement for a mid-
supply enough pressure to overcome and repair (including consumables). scale seawater RO plant in 2008 was
friction losses in the flow chan- Energy costs are directly propor- approximately $6 per gallon per day
nels in order to achieve the desired tional to the price of power and in- ($1,585 per m3/d) of permeate out-
crossflow rate. Typical applied pres- clude the pumping energy directly put, installed. Therefore, a 10-mil-
sures range from 10 to 50 psi (0.7 consumed for RO and the energy lion gal/d (38,000 m3/d) plant costs
to 3.4 bars). The circulation pump used for pretreatment, post-treat- approximately $60 million to build.
must be equipped with a shaft seal ment and overall plant operations. OPEX is approximately $0.57 per
and bearings designed to handle the Additional costs can be incurred for 1,000 gal (about $2.8 million/yr for
incoming pressure of the brine. obtaining feedwater and disposing a 10-million gal/day plant [15]. With
of brine concentrate. a typical energy requirement of 3.2
Energy recovery devices kWh/m3, including pre- and post-
Energy recovery devices (ERDs) Industrial RO treatment and a power tariff or en-
are installed in the brine stream of Assuming typical financial and ergy price of $0.10/kWh, the energy
seawater RO processes to recover utilization factors, the CAPEX of cost for a 10 million gal/d plant is
the otherwise wasted hydraulic en- an industrial RO unit depends pri- $4.4 million per year. The resulting
ergy. Initially, these devices were marily upon the composition of the present value costs for 25 years of
reverse-running pumps or turbines water being treated and the cost of operation are presented in Table 2.
that were mechanically coupled to pre- and post-treatment. However, The data in Table 1 indicate that
the HP pump. Net energy-transfer many industrial RO units can be energy comprises more than half
efficiency — the product of turbine installed within existing facilities the total cost of a 25-year seawater
efficiency and pump efficiency — or housed in containers, reduc- RO operation and more than tri-
could be as high as 80% with very ing or eliminating the cost of civil ple the capital cost of building the
large devices. In the early 2000s, work. The approximate costs for a plant. Energy-saving methods and
isobaric ERDs were introduced, al- typical industrial RO unit with a technologies can be easily cost-jus-
lowing direct transfer of pressure permeate production capacity of tified. For example, a 5% reduction
from the reject stream to the mem- 367 gal/min (2,000 m3/d) at a recov- in seawater RO energy consumption
brane feed stream. These devices ery rate of 75%, assuming a 10-year in the 38,000 m3/d (10 million gal/d)
offer net transfer efficiencies of 90% unit life, is summarized in Table 1. plant considered above would save
or more, and reduce the size and of About 70% of the CAPEX indicated over $7 million in present value.
the HP pump [14]. Turbine ERDs is for the equipment, and the bal- This value is equivalent to 12% of
are currently used in some sea- ance is for installation. Operation the cost of ownership of the plant.
water RO applications because of and maintenance (O&M) costs are If a 5% energy savings required a
their relatively low cost. ERDs are estimated at $0.57 per 1,000 gal. 1% increase in the overall cost of
not typically applied in industrial Brine disposal costs are based on the plant, the investment would be
RO installations because the brine an average U.S. commercial sewage paid back in about three years.
stream does not contain enough en- fee of $7.60 per 1,000 gal [17]. For context, it should be noted
ergy to justify their cost. The values in Table 1 clearly indi- that most methods of delivering
cate that brine disposal is the larg- large quantities of water are ex-
Costs est expense item for industrial RO. pensive. Traditional water distribu-
The cost of water generated by an In this example, increasing recov- tion and treatment requires build-
RO plant comprises capital costs ery from 75% to 90% only slightly ing plants and infrastructure. In
(CAPEX) and operating costs increases energy consumption but the developed world, the price to
Chemical Engineering July 2014 41
Table 2. Present Value Costs for a 38,000 m3/d (10 million gal/d) Seawater RO Plant for 25 Years
Seawater RO Typical energy consumption 5% Reduced energy consumption
Present Value, Cost, Cost per Percent Present Value, Cost, Cost per Percent
25 years gal/d 1,000 gal of total 25 years gal/d 1,000 gal of total
Capex $60,004,542 $6.00 $0.66 14% $60,004,542 $6.00 $0.66 15%
Energy $216,953,081 $21.69 $2.38 52% $209,569,720 $20.96 $2.30 51%
O&M $138,571,724 $13.86 $1.52 33% $138,571,724 $13.86 $1.52 34%
Total $415,529,347 $41.55 $4.55 100% $408,145,986 $40.81 $4.47 100%

produce 1,000 gal from traditional lot of attention for their potential to membrane area and therefore save
water supplies ranges from $6 to reduce the pressure needed to drive on capital expense, this would re-
15 ($1.52–3.88 per m3) [17]. A com- permeation, thereby reducing the quire a redesign of membrane ele-
parison of these figures to the costs energy demand of RO [18]. However, ments because concentration polar-
listed above makes it clear that RO high permeability typically reduces ization induced by high water fluxes
can be a cost-competitive means of salt rejection. Assuming salt rejec- already hinders the performance of
water supply. tion can be improved, the amount of current thin-film composite mem-
energy that can be saved by using brane elements. Additionally, mem-
Potential technology advances ultrahigh-permeability membranes brane fouling is exacerbated at
Technology for RO desalination con- is likely to be very small. Current higher water fluxes, further dimin-
tinues to improve. Here are some RO plants already operate near the ishing the value of ultrahigh-perme-
possible future advances for the key thermodynamic limit, with the ap- ability membranes for RO.
components of RO operations, and plied pressure being only 10 to 20% Reducing or eliminating pre-
their potential value. higher than the osmotic pressure of treatment could reduce the cost
Membranes. Ultrahigh-permeabil- the concentrate. Some of this excess of RO, but this would require the
ity membranes, including graphene pressure is needed to drive crossflow development of fouling-resistant
membranes or TFC membranes past whatever membrane material membranes or membrane coatings.
made with carbon nanotubes or is used. Although these membranes Advances in membrane technology
aquaporins, have recently received a might make it possible to use less can also reduce the need for the

The science of inorganic scale

Pure performance
control, the fine art of higher
Introducing Flocon 885, the world’s first
and only phosphorus-free, nitrogen-free,
biodegradable antiscalant for reverse
osmosis. It’s compatible with all major
membranes, including systems using
chlorine and sodium metabisulphite,
and can be used in both fresh and
saltwater. ANSI / NSF Standard 60
certified, with international potable
water approvals, it’s the natural choice

Flocon 885
for efficient membrane operations.

Biodegradable RO Antiscalant Contact us today:

Circle 8 on p. 68 or go to
42 Chemical Engineering July 2014
Cover Story

post-treatment currently required continues to broaden the appli-

for lowering boron and chloride cability of these devices, improve
concentrations in permeate to lev- their performance and reliability,
els suitable for agricultural use. and ultimately reduce the capital
However, membranes with better and operating costs of seawater RO.
selective-ion rejection are generally ERD efficiencies could be increased
less permeable to water, which in- by installing additional ERD units,
creases feed-pressure requirements thereby reducing the flowrate to
and energy consumption. each unit and reducing differen-
Pumps and ERDs. Although there tial pressure losses, by tightening
have been no significant improve- valve seals and by lowering brine
ments in pump or motor efficien- exit velocities [19].
cies specifically for RO applica- Advanced process designs. The
tions, pump and plant designers use of multiple stages can lower
can reduce energy consumption feed pressure requirements to near
with smart process designs. These the theoretical minimum pressure
include the use of larger centrifu- requirements [20] and potentially
gal pumps, with inherently higher allow for more balanced membrane
efficiencies, and variable frequency operation. However, the pressure
drives (VFDs) instead of throttle of the concentrate from the final
valves for flow and pressure control. stage is much higher than the feed DEFINING THE LIMIT
Although VFDs for very large mo-
tors can be cost-prohibitive, the HP
pressure of the first stage, making
it impossible to use isobaric energy- AS STANDARD
pump can be fed with a feed booster recovery devices in multistage
pump that is equipped with a VFD. configurations, undermining some
A properly controlled booster pump of the energy-savings potential of Electric actuators for the oil
can allow both pumps to operate these designs for seawater RO. Also, and gas industry
close to their best efficiency points the additional pumps required add
despite varying process and feed- capital costs. Safe, explosion-proof, tough.
water conditions. In addition, the Closed-circuit RO operations pro- AUMA offer a large portfolio of
use of a feed booster pump allows vide the energy-saving benefits of
the use of a HP pump with a higher multistaging without the need for
actuator and gearbox type ranges.
net positive suction head, which can energy-recovery devices or addi- ■ Automating all types of
increase pump efficiency. tional pumps. They facilitate opera- industrial valves
Despite the apparently advanced tion at high recovery rates, lower-
state of isobaric ERDs, innovation ing brine production, the disposal of ■ High corrosion protection
■ Integration into all commonly
References 10. Efraty, A., U.S. Patent No 7,628,921, 2010a.
used control systems
1. Service, R.F., Science, vol. 313, p. 1,088, 11. Efraty, A., Barak, R. and Gal, Z., Closed
Circuit Desalination — A New Low Energy,
High Recovery Technology without Energy ■ Global certifications and
2. UNESCO, Water for People, Water for Life, Recovery, Desalination and Water Treatment,
United Nations World Water Development vol. 31, pp. 95–101, 2011. approvals
Report, United Nations Educational, Scien-
tific and Cultural Organization, 2003. 12. Larson, R.E., Cadotte, J.E. and Peterson,
R.J, Desalination, vol. 38, p. 473, 1981.
3. Schiermeier, Q., Nature, vol. 452, p. 260
2008. 13. Lee, K.P., Arnot, T.C., Mattia, D., J. Membr.
Sci., vol. 370, p. 1, 2011.
4. National Research Council (U.S.), Committee
on Advancing Desalination Technology, “De- 14. Stover, R.L., Desalination vol. 203, p. 168, 2007.
salination: A National Perspective,” National 15. Sommariva, C., “Desalination and Water
Academies Press, 2008. Treatment Economics and Financing,” Bala-
5. Fritzmann, C., Lowenberg, J., Wintgens, T. ban Desalination Publications, 2010.
and Melin, T., Desalination, vol. 216, p. 1, 16. Water Desalination Report, Media Analytics,
2007. 15 September, [Excerpt], 2008.
6. Semiat, R., Environ. Sci. Technol., vol. 42, p. 17. Global Water Intelligence, vol. 14 (2), Sep-
8,193, 2008. tember 2013.
7. Loeb, S. and Sourirajan, S., “Sea Water De- 18. Elimelech, M. and Phillip, W.A., Science, vol.
mineralization by Means of an Osmotic 333, p. 712, 2011.
Membrane,” Advances in Chemistry Series,
vol. 38, p. 117, 1962. 19. Stover, R.L. and Andrews, W., IDA Journal Discover our
vol. 4 (1), p. 38, 2012.
8. Efraty, A., U.S. Patent No 7,695,614, 2010b. solutions
20. Zhu, A.Z., Christofides, P.D. and Cohen, Y.,
9. Efraty, A., Closed Circuit Desalination Series Ind. Eng. Chem. Res., vol. 48, p. 6,010, 2009. for the oil
No-4: High Recovery, Low Energy Desalina-
tion of Brackish Water by a New Single Stage 21. McGinnis, R.L. and Elimelech, M., Desalina- and gas
Method without any loss of Brine Energy, De- tion, vol. 207, p. 370, 2007. industry
salination and Water Treatment, vol. 42, pp. 22. Cath, T.Y., Hancock, N.T., Lundin, C.D.
262–268, 2012. Hoppe-Jones, C. and Drewes, J.E., J. Membr.
Sci., vol. 362, p. 417, 2010. Circle 6 on p. 68 or go to

Chemical Engineering July 2014 43

anzeige_drittel_Seite_che.indd 1 02.06.2014 09:47
Cover Story

which is the largest cost component system requires two-membrane increases and additional energy
in most industrial RO applications. operation, which adds costs and, savings, particularly with closed-
Forward osmosis. Forward os- for potable water production, the circuit RO process designs.  ■
mosis uses a water-permeable and negative public perception about Edited by Scott Jenkins
salt-rejecting membrane between utilizing wastewater effluent as a
two solutions of different osmotic potable water supply would need to
pressures. These solutions can be be overcome. Author
natural resources, waste streams Richard Stover is the ex-
ecutive vice president at De-
or high-purity solutions. Natural Concluding remarks salitech, Inc. (One Gateway
Center, Suite 250, Newton,
osmosis drives water to perme- Reverse osmosis is a reliable, cost- MA 02458; Phone: 1-510-
ate through the membrane from effective and well-established 333-2767; Email: rstover@ Stover has
the less salty “feed solution” to the means of purifying water for do- over 27 years of professional
more salty “draw solution.” Engi- mestic and industrial use. RO com- experience, specializing in
water technologies. He served
neered draw solutions employ spe- ponent and process designs have as chief technical officer and
cifically selected draw solutes that improved in recent years, lowering vice president for Energy
Recovery, Inc., the leading supplier of energy
are separated from the extracted water treatment costs. For most recovery devices for seawater desalination. He
directed the technical/commercial effort that
water and recycled to facilitate con- industrial RO applications, raising took the company from startup through an IPO.
tinuous forward-osmosis operation. recovery, thereby reducing waste Stover next served as vice president for Oasys
Water, developing forward osmosis for brine
However, draw solution regenera- brine generation, represents the concentration. Stover’s current work with De-
tion is a cost-intensive step [21]. A greatest cost-saving opportunity. salitech increases the efficiency of industrial,
brackish and wastewater treatment with high-
recently proposed design is a hy- For seawater RO, reducing en- recovery RO products. Stover has been granted
brid system that uses a forward-os- ergy consumption has the greatest numerous desalination patents. He was co-recip-
ient of the European Desalination Society’s Sid-
motic contactor to dilute seawater prospect for lowering overall costs. ney Loeb award for outstanding innovation. He
has a Ph.D. from the University of California at
with wastewater effluent prior to Anticipated future technological Berkeley and a B.S. from the University of Texas
RO treatment [22]. However, this advances could provide recovery at Austin, both in chemical engineering.


CHEMICALS September 8-10, 2014
Charleston Marriott
AMERICA 2013 Charleston, South Carolina

Join Us this September in

Historic and Majestic Charleston!
Focused on the Agro & $360 Attendee Rate The South’s
Specialty Chemical Markets: (thru Sept. 1) Includes: Cultural Capital:
• Chemical Intermediates • Access to 120+ Exhibits • #1 U.S. Travel
& Actives • Conference Presentations Destination City
• Custom & Toll (Condé Nast)
• Multiple Networking
Manufacturing Receptions, Lunches, • #4 Food and Wine Travel
• Formulations & Packaging Breakfasts Destination (Trip Advisor)
• Process Equipment • Top Ten Best Shopping
& Services Streets in America
(U.S. News)

*Easily Accessible! More details at

Charleston has non-stop flights
from many major U.S. cities
Circle 26 on p. 68 or go to
44 Chemical Engineering July 2014
Materials Selection
In The CPI
Table 1. Material
An overview of the many factors to be considered Selection Checklist
when selecting materials of construction n Chemical environment
n Impurities
n Microbiologically influenced
corrosion (MIC)

James J. Briem n Corrosion inhibitors
electing a material of construc- n Corrosion accelerators
Briem Engineering tion for process equipment in the n Chemical reaction inhibition
chemical process industries (CPI) n Acidity
is normally not difficult. The material
In Brief selection is often based on past industrial
n Aeration
n Velocity [corrosion erosion and
Resources experience. With few exceptions, past ex- flow-accelerated corrosion (FAC) ]
perience is by far the best way to choose n Temperature
an acceptable material of construction. n Downtime corrosion
However, when past experience cannot n Heat transfer
velocity be used as a guide, material selection can n Internal pressure
be complex and difficult. Unfortunately the n Vacuum loading
complexity of the problem is often not fully n Residual stresses
cleaning/sanitization recognized. The individual making the ma- n Thermal stresses
operations terial choice may fail to consider all of the n Cyclic stresses
factors involved. n Abnormal stresses
Surface finish/ n Safety and health factors
The objective of this article is to provide an
overview of the many factors that should be n Codes and Specifications
downtime corrosion considered when selecting a metallic or non- n Product contamination
n Base material cost
material properties metallic material of construction. A checklist
n Ease of fabrication
is provided in Table 1. Some of the many fac-
safety n Finish appearance
tors are obvious; others are not so obvious.
n Desired life
cost Neglect of the not-so-obvious factors could n Reliability
result in costly equipment failures. n Availability
austenitic stainless This article elaborates briefly on corrosion
steels n Vendor
considerations, stress loading and a number n Ease of repair
concluding of other factors that affect material selection. n Required maintenance
comments Initial and ongoing cost considerations are n Value of lost contents
also reviewed. n Value of downtime

Corrosion resistance is the first and most ob- Corrosive environment
vious factor to consider when selecting a ma- In a corrosive environment, the primary con-
terial of construction for process equipment. stituents and their concentration ranges are
The chemical environment provides the first usually known by the process or project en-
clues to selecting an acceptable material. gineer. Not so obvious, however, is the pres-
As mentioned, past industrial experience is ence of impurities. The classic example of a
the best way to address the corrosion issue. harmful impurity is the chloride ion, which,
Other resources include laboratory corrosion under certain conditions, is known to cause
tests, pilot-plant corrosion tests, published stress-corrosion-cracking (SCC; Figure 1)
corrosion charts, data published by the Na- and pitting of stainless steels. These insidi-
tional Assn. of Corrosion Engineers (NACE; ous forms of corrosion can occur at concen-
Houston;, manufacturers trations as low as a few parts per million.
data, textbooks, articles and materials con- An impurity that is commonly found in
sultants. Most of these sources are available water is bacteria. Bacteria cause a type of
on the Internet. corrosion known as microbiologically influ-
36 Chemical Engineering june 2015
FIGURE 1. This is a polished and etched metallurgical cross-section FIGURE 2. Microbiologically influenced corrosion (MIC) cannot be confirmed from
of a stainless-steel pipe viewed at 200x magnification. The branching, visual inspection. Carbuncles near a weld seam are a strong indication of MIC. Fresh,
transgranular cracks are characteristic of stress corrosion cracking wet carbuncles removed from a newly drained pipe can be cultured to confirm the
presence of microbes and identify species present

enced corrosion, or MIC for short (Figure selection. As a general statement, the
2). This is a form of corrosion in which lower the pH, the more aggressive the
the bacteria can eat through stainless- corrosion. A pH variation of a few points
steel tanks and piping in a few months. could make the difference between using
The bacteria primarily attack the retained, a relatively inexpensive material and a
ferritic, metallic grains in stainless weld- costly material of construction. For ex-
ments, although MIC can occur in the ample, a lower pH environment allows
base metal remote from weldments. The a lower chloride concentration to stress-
bacteria burrow caves (Figure 3). As ob- corrosion-crack stainless steels.
served with a scanning electron micro- The presence or absence of air (ox-
scope, the cave surface has the appear- ygen) may also affect the corrosion
ance of a honeycomb structure (Figure resistance of a material in a specific
4). Unfortunately, most commonly used environment. The presence of O2 or
metals are subject to MIC attack to vary- the occasional exposure to air may be
ing degrees. necessary to maintain the protective
The author’s company is currently in- oxide films on materials, such as stain-
vestigating MIC attack in a new process- less steel, which depend on a chrome-
ing plant at which the hydrostatic test oxide film for corrosion resistance. On
water was left in the stainless-steel piping the other hand, the presence of air may
system for several months after the hydro- destroy the corrosion resistance of a
static test. This stagnant water allowed material that is normally not corroded
bacteria to eat through several million dol- in a reducing, O2-free chemical environ-
lars’ worth of piping in a few months. ment. Boiler feedwater is a good exam-
Some minor constituents of a chemi- ple. A boiler can fail in six months if O2
cal environment can actually inhibit cor- is present in the boiler feedwater. The
rosion. One not too obvious example of same boiler could last six decades with
this is the inhibition of SCC of titanium in deoxygenated feedwater.
methanol. Titanium is subject to SCC in
anhydrous methanol at room tempera- Velocity
ture, but if water is present in concen- Velocity must be evaluated when select-
trations above 1%, the titanium will not ing a material of construction, particularly
stress-corrosion-crack. As another ex- when considering pumps, agitators, and
ample, low concentrations of sulfites pre- other equipment subjected to relatively
vent oxygen pitting in boilers. high fluid velocities. Velocity can mani-
An infrequent, but potentially costly fest itself in a corrosive environment as
mistake is to select a material of con- “corrosion-erosion” or flow-accelerated
struction that, if corroded, could inhibit a corrosion (FAC). Of concern here is the
process chemical reaction. Nickel, cop- fact that most published corrosion data
per and other metallic corrosion products are based on relatively stagnant corro-
have been known to kill a chemical re- sion conditions. Although there are a FIGURE 3. This magnified cross sec-
action that was essential to the chemical few exceptions, corrosion increases as tion of a weld in a stainless-steel pipe
reveals a cave structure that confirms
processing operation. velocity increases. An example is com- MIC. The cave structures typically
The acidity of a chemical environment mercial-grade ambient-temperature sul- occur at the boundary between a weld
may have a significant effect on material furic acid transport in carbon steel pipe. and the base metal

Chemical Engineering june 2015 37

FIGURE 4. The honeycomb struc-
ture observed on the surfaces
inside the cave confirms MIC. This
is a scanning electron microscopic
view at 1,800x magnification

At a velocity below 3 ft/s, the acid can be steam and cleaning agents can corrode
safely handled using carbon-steel pipe. processing equipment even though the
At higher velocities, the protective iron- basic production environment is non-
sulfate layer is stripped away from the corrosive. When designing for a specific
pipe wall and corrosive failure occurs in operating environment, don’t neglect the
a short time. cleaning operation.
One notable exception to the above
is microbiologically influenced corrosion. Surface finish/passivation
Bacteria like stagnant water. Flowing Surface finish is an important consider-
water inhibits MIC. ation particularly when dealing with stain-
less steels used in the pharmaceutical
Temperature and food-processing industries. The sur-
The basic effect of temperature on the face finish must comply with U.S. Food
corrosiveness of an environment is well and Drug Administration (FDA; Washing-
known by most process and project en- ton, D.C.; requirements to
gineers. As a rough estimate, the corro- minimize the possibility of any bacterial
sion rate doubles for every 10°C increase growth on surfaces.
in temperature. This is essentially true up Another important finish consideration
to the boiling point. If the operation is at for stainless steel is the issue of surface
a temperature above the dew point, in a passivation. Process engineers are famil-
gaseous state, the corrosion rate is es- iar with the passivation of stainless steel
sentially nil. with an oxidizing acid, such as nitric or
The above statements explain why citric acid. The passivation treatment pro-
condensing a corrosive gas results in cor- duces a passive protection chrome-oxide
rosive attack that would not occur if the film on stainless steels. This chrome-
operation is kept above the dew-point oxide film is responsible for the overall
temperature. corrosion resistance of the commonly
The effect of temperature must be used austenitic stainless steels, such as
carefully assessed for each environment. 304 and 316. There is some controversy
Many material failures occur because the over the necessity for using a passiva-
temperature evaluated was the mass tion treatment, either initially or at intervals
temperature of the corrosive material in a while the equipment is operating. It is this
vessel, not the higher inside-surface tem- author’s opinion that the main reason for
perature of a steam-jacketed vessel. passivation is not necessarily to form the
protective chrome-oxide film but rather to
Cleaning/sanitizing operations thoroughly clean the surface of any de-
Consideration should be given to the posits on the surface that could initiate
procedure and frequency with which pitting corrosion of the stainless steel.
the process equipment will be cleaned Normally, stainless steel will self-passivate
and sanitized. It is not unusual for the with intermittent exposure to the oxygen
cleaning operation to corrode process- in the atmosphere. If there is no intermit-
ing equipment. Periodic cleaning with tent exposure of the stainless-steel sur-
38 Chemical Engineering june 2015
FIGURE 5. This sprinkler system pipe
sat too long without flushing or oxygen
scavenger maintenance. Premature
failure was caused by grooving corro-
sion of the longitudinal pipe welds

face to oxygen, then intermittent passi- of a vacuum should be considered. Be

vation treatments may be necessary to cautious of potential vacuum loading re-
maintain the passive chrome oxide film sulting from steam collapse in storage
on the stainless-steel surface. If this pas- tanks. Vents frozen shut by ice or cor-
sive film is not maintained, the stainless rosion product have caused many tank
steel becomes active and has a corro- failures when cold weather conditions
sion resistance approximating that of caused steam to condense inside the
carbon steel. vessel, creating a vacuum.
Both metallic and non-metallic equip-
Downtime corrosion ment may contain residual stresses that
Corrosion that occurs when a piece are usually produced during equipment
of equipment is not operational can be fabrication. Forming, welding and ma-
more severe than corrosion that occurs chining metals may result in significant
when the equipment is online. Equip- residual-tensile stresses approaching the
ment that operates in a gaseous environ- yield strength of the metal. These residual
ment at moderate temperatures above stresses are often the primary cause of
the dew point will suffer more corrosion SCC. Residual stresses may contribute
at lower temperatures than when online to corrosion fatigue failures.
at an elevated temperature. An example Thermal stresses are another consid-
of this is a boiler economizer operating in eration that should not be neglected for
fluegas containing SO2. process equipment operating at non-
In a recent project, a boiler firing a bio- ambient temperatures. Thermal stress
fuel gas with high-sulfur content failed in occurs in a monolithic material that op-
less than a year due to frequent shut- erates at different temperatures in differ-
downs. A large fire-suppression sprin- ent locations. Thermal stresses are also
kler system sat stagnant and failed due important when combining different ma-
to grooving corrosion (see Figures 5, 6 terials with different coefficients of ther-
and 7). mal expansion. Consider a plastic-lined
vessel where the thermal expansion of
Material properties the plastic can be an order of magnitude
The next group of factors affecting ma- higher than the substrate steel.
terials selection includes mechanical and Cyclic stress is often overlooked or
physical material considerations. underestimated. Cyclic stresses cause
On occasion, the thermal conductivity fatigue failure. Fatigue is the failure of a
of a material may determine its accept- material under cyclic loading at a stress
ability for process equipment, the most level well below the published tensile
obvious example being heat exchangers. strength of the material. As a very rough
Low thermal conductivity may be desir- estimate, fatigue strength is only about
able for certain equipment. For example, half the tensile strength of the material. In
an un-insulated fiberglass-reinforced a corrosive environment, cyclic stresses
plastic tank may be more economical can result in “corrosion fatigue,” a stress-
than an insulated carbon-steel tank. related material failure that can occur at a
Evaluation of internal pressure is ad- stress below the normal fatigue strength
dressed by standards, such as the ASME of the material.
Boiler & Pressure Vessel Code and the Most published fatigue data are for
ASME Piping Codes, and needs no elab- fatigue strength in an air environment.
oration here. When corrosives are present, the fatigue
The intended or unintended presence strength is even lower and if failure oc-

Chemical Engineering june 2015 39

FIGURE 6. Stagnant water inside the sprinkler pipe caused grooving corrosion FIGURE 7. This metallurgical cross section of a sprinkler system pipe reveals
of the longitudinal weld seam grooving corrosion of the longitudinal weld seam. The grooving corrosion is
caused by the weld seam being anodic (corroding) relative to the rest of the
pipe wall, which is cathodic

curs, the failure mechanism is defined as normal and reasonable first step in evalu-
corrosion fatigue. ating initial cost is an assessment of basic
Abnormal stress loads must be con- material cost. Consider the ease of fab-
sidered. Examples include pump vibra- rication. Are the materials easily formed,
tion loads, loads resulting from agitator cut, bonded, drilled or welded? Is heat
mountings, and, of course, the impact of treatment required?
an errant fork truck. Finish and appearance will impact cost.
Polished stainless steel may be neces-
Safety sary for equipment used in the food and
Today’s increased concern for safety and pharmaceutical industries. Sprayed foam
health has had its impact on materials of insulation is less expensive and less at-
construction. In the area of safety, it’s eas- tractive than lagged, block insulation.
ier to comply with an OSHA (Occupational The life expectancy and desired reliability
Safety and Health Assn.; Washington of the equipment may impact initial cost.
D.C.; guard-rail require- As an example of the result of the trade-
ment on a metal tank than on a plastic offs between initial cost and lifecycle cost,
tank. Asbestos is no longer an acceptable consider the bulk storage of commercial-
material of construction. In the food and grade sulfuric acid. The standard practice
pharmaceutical industries, the material is to construct a new carbon-steel above-
selection is often limited to a material that ground storage tank to API 650 standards
is virtually impervious to chemical attack. using a 1/8 to 1/4 in. corrosion allowance.
Another factor to consider is the degree The steel may be specified as copperized
of hazard to humans and the environment to slow sulfuric acid attack. During the life
that could result from the accidental re- of the tank, periodic thickness surveys are
lease of hazardous materials contained in performed to monitor thinning of the shell.
a vessel or pipe should failure occur. Toward the end of the tank’s life, capacity
The government, industry and individual may be reduced because of thinning of
companies have codes, standards and the lower plate courses. In this case, the
specifications that will often impact or even best solution is a low initial cost with care-
determine the selection of a construction ful monitoring of vessel condition.
material. These codes, standards and Material availability and the choice of a
specifications must be considered when vendor will affect cost. Selecting a relatively
selecting materials of construction. exotic material that is fabricated by only a
few vendors could significantly increase
Cost the cost and extend the delivery time of
Naturally, in selecting a material of con- the equipment. This could impact overall
struction, the objective is to find a mate- project costs if the lack of availability de-
rial that will result in a long safe service lays startup because of extended delivery
life at a reasonable cost. In assessing the times. Consider also the qualifications of a
economics of a material, both initial cost vendor. A sound, basic material poorly fab-
and lifecycle cost should be evaluated. A ricated seldom performs well.
40 Chemical Engineering june 2015
Unfortunately it is not uncommon to more critical than operational stress.
neglect consideration of ongoing cost The corrosive environment primarily re-
factors when selecting a material of con- sponsible for SCC of stainless steels is
struction. Ease of repair, required main- chlorides. The primary variables that ef-
tenance and availability of spare parts fect stress corrosion are stress, chloride
should be considered. concentration, acidity and temperature.
The value and potential danger of the Each of these factors has a limit below
material stored in a tank may impact mate- which SCC will not occur regardless of
rial selection. Some consideration should the other factors. For examples, if the
be given to process downtime should a stress level is low enough then SCC will
material fail. These three factors may af- not occur regardless of the chloride con-
fect the degree of conservatism used in centration, the alloy or the temperature.
selecting a material of construction. Although there are exceptions, SCC rarely
occurs at a temperature below 160°F. If
Austenitic stainless steels one of the listed variables is kept below
Austenitic stainless steels, such as 304 a threshold limit, the equipment will never
and 316 stainless steel, are the most fail from stress corrosion cracking.
commonly used metals in the CPI. As When assessing chloride concentrations
such, austenitic stainless steels deserve it is important to note that what may be
some special attention here, not only stated as being only a few parts per million
because of their extensive use, but also chlorides is not necessarily representative
because austenitic stainless steels are of chloride concentration on the stainless
subject to some unusual forms of cor- surface. Often the processing equipment
rosive attack. is operating in a situation where the chlo-
Austenitic stainless steels are subject rides are concentrated. For example, in
to general corrosion, pitting corrosion, a heat exchanger, the heat may actually
SCC and MIC. Of these, general corro- drive off the water and leave the chlorides
sion, where the surface corrodes rela- such that the chlorides will salt-out to a
tively evenly, is perhaps the most under- concentration of half a percent, which is
stood. With general corrosion, the rate above the threshold limit for SCC.
of corrosion normally slows with time
since the corrosive agent must diffuse Concluding comments
through the rust layer before it can at- The purpose of this article is to make pro-
tack the stainless steel surface. Pitting cess and project engineers more cogni-
corrosion of stainless steel normally oc- zant of the many factors involved in prop-
curs in chloride environments. With pit- erly selecting a material of construction.
ting corrosion, the rate of attack actually If all of the factors reviewed in this article
increases with time because the pit acts are carefully evaluated, the process engi-
to concentrate the corrosive agents and neer is well on the way to selecting the op-
the pit forms a localized galvanic corro- timum material of construction. I mention
sion cell. Both factors increase the rate this not because you need to consider
of pitting corrosive attack. them all for every application but to show
MIC has been discussed above. The you that there is a lot more to selecting
rate of attack can be very aggressive. an acceptable material than looking at a
MIC is capable of causing stainless steels published corrosion chart. n
to fail in a matter of months. Although Edited by Gerald Ondrey
most bacteria are killed by sanitizing the
equipment, at a temperature of about Author
180°F, the caves where the bacteria live James J. Briem, P.E. is the founder of
still remain. Reintroducing the corrodant Briem Engineering (4134 Rider Trail
North, Saint Louis, Missouri 63045;
allows the bacteria to reoccupy homes Phone: 314-298-3773; Email: jimb@
(caves) and the MIC continues. MIC is a, which is a materials
tough problem to deal with. testing and engineering firm established
As noted previously, stainless steels are in 1983. At Briem Engineering, Briem
supervises engineers and technicians
subject to SCC. The two prime factors involved in materials’ investigations and
that cause SCC are stress and, as the tests including: nondestructive testing
name implies, a corrosive environment. and inspection, corrosion evaluation, material selection, fail-
ure analysis, stress testing and mechanical testing. Prior to
The stress that is normally responsible founding Briem Engineering, he was an engineering supervi-
for SCC is residual stress, such as found sor at Mallinckrodt Chemical Inc. and Proctor & Gamble Co.
in the heat-affected zone of a weld or re- He holds a metallurgical engineering degree from Rensselaer
Polytechnic Institute (Troy, N.Y.). He is a member of NACE,
sulting from a forming operation. These ASM, ASTM and ASME. He is a licensed professional engi-
residual stresses are usually significantly neer in Missouri and Ohio.

Chemical Engineering june 2015 41

Engineering Practice

Rotating Machinery: What You Should Know

About Operational Problems
Follow this guidance to improve the operation, safety and reliability of
rotating machinery in chemical process plants

Amin Almasi

roubleshooting problems in
rotating machinery (such as
compressors, pumps, steam
turbines, gas turbines, turbo-
expanders and more) can present
difficult challenges during day-to-day
operations in facilities throughout the
chemical process industries (CPI).
Reliable operation is an important
factor in keeping plants running, par-
ticularly as plants push to operate at
higher capacities. Lost profits due to
persistent machinery problems can
never be recovered.
Rotating machinery often pres-
ents the most difficult operational
problems in a CPI plant. This article FIGURE 1. Shown here is a steam turbine dismantled for repair. This is the driver of the centrifugal com-
provides a practical, in-depth review pressor
of the art and science of machinery
troubleshooting and problem solv- A systematic approach is the key for proach provides greater value, by
ing. Five case studies are presented solving any machinery operational enabling earlier machinery trouble-
to cover a variety of relevant rotat- problem. Comprehensive machinery shooting through improved early
ing-machinery systems, such as assessments should be carried out detection of unusual deviations. This
compressors, pumps, gas turbines using advanced data-collection tools can provide an earlier indication of
and steam turbines. All of the case and methods (Figure 1). For instance, problems that may be developing.
studies presented here were devel- the use of automated machine- In modern operations, expert op-
oped from actual field experience. data-collection software — which erators often opt to manage critical
This article offers a broad working automatically collects a wealth of in- machine components using time-
knowledge of troubleshooting princi- formation specific to the machinery based data trends. Useful trends can
ples and practice, to help operators component — coupled with sophis- be developed when the behavior of
gain insight into different techniques ticated modeling techniques, can individual performance variables,
and various methods for machinery provide great insight to support trou- such as pressure, temperature, flow,
problem solving. bleshooting efforts related to com- lubrication oil characteristics and
plex rotating machinery. Such insight more, are plotted over time on a
A systematic approach can help operators to diagnose the graph. These trends can then be an-
Proper CPI plant operation requires actual root causes of underperform- alyzed in conjunction with additional
sustained attention to the following: ing machinery and premature ma- information, such as that derived
• Identifying opportunities to im- chinery failures. from alarm screens and plant con-
prove machinery reliability and When CPI operators rely too trol-system graphics. By monitoring
safety heavily on alarms and trips, there is trends, operators can anticipate po-
• Developing solutions for chroni- limited time to take remedial action tential problems and propose smart
cally problematic rotating ma- in response to a machinery issue. solutions before any major upset
chines Instead, a proactive monitoring ap- occurs. Trends in relevant data can

74 Chemical Engineering march 2015

support operators and machinery The importance of monitoring
engineers, allowing them to stay In one situation, a CPI company avoided an estimated $10 million in property damage and
aware of situations as they begin business-interruption loss in a year as a result of the early detection of high vibration levels
to arise, diagnose the root cause in its rotating machinery. The high-vibration monitors indicated the deteriorating conditions
of problems and resolve abnormal (a developing problem) within a large, critical compressor train. The developing problem
situations. The ability to gather and resulted in the loss of half of the bolts, a circumferential crack in the coupling spacer (about
80 mm long) and a radial crack emanating from the initial break at the time of shutdown.
analyze trend data can also help op-
If the fault had progressed and all of the bolts had sheared before operator interven-
erators to compare past and present tion, the compressor would have been unrestrained — with potentially disastrous con-
behavior, thereby supporting efforts sequences. This clearly illustrates why, particularly in large CPI plants, it is vital to monitor
to anticipate future changes. critical components on a continuous basis.

Keep an eye on key variables tance thermal detectors (RTDs; also give operators early indication of de-
Vibration monitoring is an important known as resistance thermometers), terioration in the machine’s operating
indicator that can help operators to thermistors, filled thermal systems, condition.
identify machinery problems at an pyrometers, infrared thermography Control valves are vital compo-
early stage. The location of vibration techniques and glass thermometers. nents in every process and in many
sensors is vital, to avoid introducing All are widely used to provide types of equipment packages where
errors during the measurement. In process and machinery temperature fluid is handled and regulated. How-
general, the best location for vibra- measurements. ever, the most common malfunction
tion sensors in rotating machinery The measurement of flowrate pro- with control valves is internal leak-
is at the bearing or bearing housing vides another useful indicator for the age, which cannot be detected eas-
(Figure 2). Vibration measurements health of many processes and ma- ily. Internal leakage can result from
at the machinery casing, in addition chinery systems. Traditionally, tech- one of several factors, including an
to bearing vibration, can provide niques based on differential pres- eroded valve plugs or seats, or insuf-
useful data for identifying and solv- sure, turbine and vortex shedding ficient seat load.
ing machinery problems. have been used to measure fluid
However, monitoring vibration flowrate. However, these methods Five case studies
alone is not sufficient to gauge the might not be suitable if an accurate Case study 1: Insects in an air-
health of rotating machinery espe- measurement is required for gases, compressor inlet filter. In a CPI
cially when large, critical machines since errors can be introduced in the plant in a remote area, insects
are concerned. For instance, in a re- measurement due the compressible caused problems for the inlet-air fil-
cent case, it was observed that one nature of gases. To use these meth- tration unit of a large process-type
electric motor’s temperature and ods with gas flows, it would be nec- air compressor. The accumulation of
current increased when the machine essary to compensate for the gas insects on the filters led to high dif-
was overloaded — well before any density fluctuation in order to reduce ferential pressure in the unit, leading
vibration was detected. This under- the errors. to periodic compressor shutdowns.
scores the importance of monitoring Modern Coriolis mass flowme- Multiple cases of surge and surge-
other parameters (such as tempera- ters are a more appropriate choice related trips occurred because of
ture and current), in addition to vibra- for measuring gas flowrate, which is this issue. Regarding a problem of
tion, to allow for early fault diagnosis usually independent of temperature, persistent insects impacting the air-
of electric machines. pressure, density and composition. compressor, two solutions should be
When vibration analysis is used This meter uses the Coriolis effect to considered:
alone, it may be difficult to estab- measure the amount of mass mov- 1. A multi-sided insect screen. For
lish the real cause of the vibration. ing through the element. In a simple example, three-, four- or five-
In general, it is better to monitor form, the fluid to be measured runs sided insect screens can be con-
both vibration and overall machin- through a U-shaped tube that is structed (for instance, in a cube or
ery performance (such as delivered caused to vibrate in a perpendicu- similar configuration) in the front
pressure and flow in pumps or com- lar direction to the flow. Fluid forces of the air filter, with a surface area
pressors) to obtain a more reliable running through the tube interact roughly three to four times that of
fault diagnosis. And, it is important with the vibration, causing it to twist. the front face of the air filter inlet.
to realize that the vibration produced The greater the angle of the twist, The rationale is that insects will
by the machinery is not a fault on its the higher the flow. These flowme- gather on the larger-surface area
own — but rather a symptom (and ters have no rotating parts. Modern of the screen (far from actual in-
early warning sign) of a developing Coriolis flowmeters are highly accu- take) and thus experience less
failure condition. rate (providing accuracy to around intake air velocity, allowing them
Temperature is another critical ±0.5% of the mass flowrate). to fly away, reducing their oppor-
variable, and the ability to moni- For relatively low-speed rotating tunity to trap and plug the air filter.
tor it accurately is of paramount machines, vibration analysis may not When a screen is built with three
importance in CPI plants. Today, a be suitable to detect degradation times the area of the air intake,
wide range of temperature sensors or operational problems. However, the air velocity across the insect
and measurement systems exists, acoustic-emission sensors, strate- screen would be around 33% of
including thermocouples, resis- gically located on the machine, can the original velocity. When the

Chemical Engineering march 2015 75

map. This test was completed by
hooking proper surge-detection
hardware into the compressor’s vi-
bration-monitoring system. At each
operating speed, the compressor
discharge valve was slowly closed.
Prior to the surge, the surge-de-
tection system (using vibration sen-
sors as part of the compressor vibra-
tion-monitoring system) detected an
increase in low-frequency vibration,
which indicated the onset of surge.
This was used to define the surge
line. The compressor was not actu-
ally surged, just taken to the point
of onset of stall. Once the as-built
FIGURE 2. Tilting pad thrust bearings are widely used in relatively high-speed rotating machinery. These surge line was identified using this
bearings are vulerable and can be damaged easily; shown here is one that was removed for repair technique, it was seen to be different
screen is built at four times the in- fit, by helping to keep rain and dirt from what both the vendor and the
take area, the air velocity across from being driven into the filter inlet sub-vendor indicated in their respec-
the screen would be just 0.25% by wind. tive performance maps.
of the original velocity. Case study 2: Discrepancies in Often, it is common to obtain a
2. Specially designed pulse-clean performance maps. During the site-tested, as-built surge line that
filters. Another possibility is to commissioning stage of a CPI plant, differs from the vendor-calculated
install pulse-clean filters that it was noted that there were two dif- surge line, or even the shop-tested
apply a periodic reverse pulse of ferent sets of performance maps in surge line, because of different com-
air to dislodge the accumulated the manual for a particular, critical pressor piping and arrangement in
dirt and insects from the filter centrifugal compressor — one from the vendor shop test compared to
inlet at a pre-disposed differen- the compressor manufacturer and the final site installation. The site-
tial pressure. This will require a another from the anti-surge-system tested, as-built surge line should be
specialized filter assembly with sub-vendor. The compressor had considered the most reliable data for
a compressed-air supply and been delivered three years earlier any operator.
associated equipment and con- but it was not installed and commis- Case study 3: Re-rating a steam
trols. However, this solution is not sioned upon delivery because of a turbine. The steam turbine driver
usually recommended since this three-year delay in the project and a of a large pump train in a CPI unit
is a complex, expensive and risky stop in the plant construction. Some created a bottleneck that prevented
option. of the engineers who had worked the unit from achieving a desired 5%
In this case study, the first pro- on this machine and associated unit increase in capacity. The operation
posed solution was used and a had left their jobs and there was no team asked to run the steam tur-
four-sided insect screen was built. proper documentation to clarify how bine using 49-barg steam instead
For such an installation, the mesh this discrepancy originated. of 46-barg steam to achieve higher
size should be selected properly to At this facility, there were concerns power generation from the turbine
stop the smallest anticipated insects about surge. During the initial period and higher pump capacity. Simula-
from penetrating, and it should be of operation, the compressor was tions showed that using 49-barg
reasonably rigid to allow for cleaning tripped several times because of steam, it was possible to generate
with a soft brush. If the mesh has a surge or other operational issues re- higher power, and this modeling also
dense or tight weave (say, with holes lated to the existence of two different showed that the pump train could
around 1-mm dia.) to ensure that performance maps and the allow- handle higher power and generate
small insects cannot pass through, able operating range. Any attempt to more flow within the pump curve. In
then it may be necessary to com- identify which performance map and fact, this investigation showed that
pensate with increased surface area surge line were the most appropri- the new pump operating point was a
(so that air intake is not impeded and ate for use failed because of person- bit closer to the best efficiency point
pressure drop is not increased). For nel changes at both the compressor (BEP) than the old operating point. In
this case, a window screen mesh vendor and the anti-surge- system general, significant engineering as-
(with holes roughly 1-3-mm dia.) sub-vendor during the 3-yr period. sessments are required in order to
was selected. Routine inspections In general, turbocompressors re-rate a steam turbine.
by technicians have been required should be surge tested once they The steam turbine casing in this
to clean the screen. The structure of are installed, to develop accurate, case was originally designed and
the insect screen was made rugged as-built surge lines. For this machine, rated to 46 barg. In order to increase
enough to resist wind and ensure it was decided to perform the surge the pressure rating, the design of
longevity of the asset. This insect test at the site to identify the surge the casing was checked by both
screen provides an added bene- line and the correct performance the vendor’s engineering team and

76 Chemical Engineering march 2015

an independent consultant. It was mulated dirt in the air-compressor in order to reduce drag and improve
determined that it could safely and section of a gas turbine can increase the heat rate and power output. Inlet
reliably contain steam at 49 barg. fuel consumption, and lead to more air filters play a significant role in the
The casing was re-rated, after frequent maintenance outages and a degree of fouling that gas turbines
a proper hydro-testing, as per decreased hot-section life. experience. With the original, older-
code requirements. At a given CPI plant, gas turbines style air filters, the operators of this
Using the same rotor, the pressure were used to both generate power plant were faced with a tradeoff: In-
drop per stage was increased. The and drive the gas compressor. The stalling higher-efficiency filters that
existing impellers were checked to original air filters were used in these remove more of the airborne par-
ensure that they were strong enough gas turbine air-inlet systems, which ticles creates a higher initial pressure
and the rotor did not require chang- mandated turbine shutdown and of- drop and typically have a shorter life
ing. Simulations and investigations fline water washing every six months. span compared to lower-efficiency
showed that diametric changes and The dirty air reduced the efficiency filters. The filter pressure drop de-
changes to internal static compo- of the gas turbine fleet between creases the density of the air going
nents within the casing were not cleanings. For instance, the gas tur- to the gas turbine, which reduces
required. The steam turbine was bine trains exhibited efficiency drop the gas turbine’s output power.
re-rated for 49 barg. The steam-tur- of as much as 2.5% on average in Added pressure drop across the air
bine-driven pump train has enjoyed the three-month period before each filters increases the amount of en-
smooth and trouble-free operation offline water washing. The sched- ergy needed to draw air into the gas
since re-rating after the changes uled offline water washing every six turbine, which increases the gas tur-
were made. months on each gas turbine of the bine’s heat rate.
Case study 4: Inlet filter for gas CPI plant (just for cleaning), in addi- In general, there are tradeoffs be-
turbines. In general, gas turbines tion to other outages for inspection tween efficiency, cost, filter life and
“ingest” different materials depend- and maintenance, was wasteful and pressure drop that all need to be
ing on the location and the plant — non-productive. The operation team considered when evaluating which
primarily dirt and dust, but also ice, asked for a solution. air filters may be best for a particu-
rain, snow and salt. The cleanliness Getting maximum performance lar operation. The small particles that
of a gas turbine is the key factor in its and reliability out of a gas turbine cause most of gas turbine fouling
efficiency, reliability and safety. Accu- requires keeping the blades clean are less than 3 micrometers (µm;

2015 ry

He .che

Two-P at Exchange

Written for engineers, by engineers

art Fea

ture Re : m

Advan port

in 3-D ces


Focus g

Housen Industrial
More and more, business in the Chemical Process Industries

• Hea

t exc

(CPI) is not local, it’s global. To keep up with this rapidly



Dimen t Your Finge
sionless rtip

Numb s:
ers evolving marketplace, you need a magazine that covers it all,
Extrem es for
not just one country or region, not just one vertical market,
e Serv

High- amentals o but the entire CPI. With editorial offices around the world,
Shear f Chemical Engineering is well-positioned to keep abreast of all

rsers the latest innovations in the equipment, technology, materials,

122 No

and services used by process plants worldwide. No other

ry 201

publication even comes close.


To subscribe or learn more about membership, please visit

Chemical Engineering march 2015 77

sometimes less than 0.5 µm) in size, they reach their expected lifespan. only be opened to around 78% for
below the standard capabilities of Modern hydrophobic HEPA filters the first normal operating condition
old-fashioned filters. Thus, empha- were installed for all gas turbines case. The investigation showed the
sis should be placed on comparing in this CPI plant. They completely position indicator on the governor
the differences in filter efficiency for eliminated the need to shutdown the was at 90% for this operating case
the removal of particles smaller than gas-turbine trains for offline water (around 12% discrepancy). The initial
3 µm (as well as considering filter washing. The operation has been suggestion was that the governor or
efficiency on particles smaller than satisfactory since the filter change. steam-admission valves might not
0.5 µm) since those particle sizes Case study 5: Problems in tur- be correctly calibrated. The physical
generally contribute more to bine’s steam-admission valves. movement of the admission valves
air-compressor fouling rather than In this case, problems in the steam- has not changed since commission-
larger particles. admission valves used to supply the ing, which indicated that the prob-
Switching to better, higher-effi- steam turbine reduced the required lem was not related to the gradual
ciency air-inlet filters is a good way steam flow into the steam turbine, build-up of deposition or to degrada-
to keep air-compressors cleaner, which impacted the performance of a tion. A highly nonlinear relationship
but they can offer some drawbacks. critical centrifugal compressor driven was reported at low valve-opening
The most appropriate air filter should by the turbine, and this negatively positions (around 0–20% of the rated
be selected with respect to the ef- impacted the plant’s production rate. flow). Below 20% of the governor
ficiency, the filter life and pressure There were four poppet valves in the position, there was very little steam
flow through the steam turbine. This
indicated that there was a problem
There are tradeoffs between efficiency, cost, filter life with the first poppet valve.
There was almost zero increase
and pressure drop that need to be considered when in the steam flow once the governor
evaluating which air filters may be best for a particular position got higher than 91% of the
rated flow. This was evident from
operation. the sharp drop-off at the end of the
stroke-versus-steam-flow curve of
the admission valve. There was also
drop. Today, new filter designs can steam chest of this steam turbine. a problem with the last poppet valve.
keep the gas turbine clean with the These should be staged to open at In other words, the lack of response
same pressure drop and a better fil- different load requirements. during the first 20% of governor
ter life compared to older-style filters. Problems and issues for steam- opening indicated a problem with
Today’s modern filters have a struc- admission valves have been re- the first poppet valve. The lack of re-
ture and layers that allow for the ef- ported extensively for steam turbines sponse to the final 9% of governor
fective capture of even the smallest in different CPI plants. For any steam opening indicated a problem with
particles (below 3 µm and some- turbine governor or admission-valve the last poppet valve. The steam-
times below 0.5 µm) to keep the issues, the following initial investiga- admission valve was overhauled for
gas-turbine intake air clean without tions are recommended: quick replacements of the first and
significant filter plugging. 1. Study the changes in perfor- last valves. The steam turbine opera-
Advanced hydrophobic high-ef- mance of the steam-admission tion has been satisfactory after this
ficiency particulate air (HEPA) filters valve system and governor sys- valve change. n
have the same pressure drop as the tem Edited by Suzanne Shelley
older filters. However, the key to their 2. Evaluate the actual performance
improved filter performance is a novel of steam turbine versus the de- Author
multi-layer media construction that sign specification Amin Almasi is a rotating-equip-
includes a hydrophobic membrane 3. Investigate the possible causes ment consultant in Australia
layer. Usually the first filter layer cap- for discrepancies He previously worked at Worley
tures most of the airborne dirt. The Usually one or more of the follow- Parsons Services Pty Ltd. (Bris-
middle layer — or layers — are most ing issues are reported for steam- bane, Australia), Technicas Reuni-
often made from expanded polytet- admission valves: das (Madrid, Spain) and Fluor
Corp. (various offices). He holds a
rafluoroethylene (PTFE) membranes • Insufficient nozzle-opening size chartered professional engineer
or similar, and they stop both sub- • Blockage inside the admission license from Engineers Australia
micron particles and water that may valve system (MIEAust CPEng – Mechanical), a chartered engineer
certificate from IMechE (CEng MIMechE), RPEQ (regis-
penetrate the first layer. Other layers • Incorrect calibration of the admis- tered professional engineer in Queensland) and he also
are support layers added to provide sion valve system holds M.S. and B.S. degrees in mechanical engineering.
strength and burst pressure resis- • Problem with internal setup of the He specializes in rotating machines, including centrifu-
gal, screw and reciprocating compressors, gas and
tance. These support layers give the steam-admission valve system steam turbines, pumps, condition monitoring and reli-
filter more than twice the required • Gradual buildup of deposition or ability. Almasi is an active member of Engineers Austra-
burst strength of the old-fashioned degradation products lia, IMechE, ASME, Vibration Institute, SPE, IEEE, and
filters. This extra burst strength pre- Based on the steam-turbine man- IDGTE. He has authored more than 60 papers and arti-
cles dealing with rotating machines.
vents filters from breaking before ual for this unit, the governor should

78 Chemical Engineering march 2015

Feature Report
Engineering Practice

Condition Monitoring for

Rotating Machinery
This valuable insight into the
performance of pumps and
compressors will help
improve operation

Amin Almasi
Rotating Machine Consultant

otating machines, such as pumps,
compressors, gas turbines, steam
turbines and electric motors,
play crucial roles in most chemi-
cal process industries (CPI) facilities.
Increasingly, advanced condition mon-
Figure 1.
itoring is being used to improve the Shown here is an
operation and maintenance of these example of a complex
workhorse components. The use of gas turbine in a CPI facil-
well-designed condition-monitoring ity. Modern rotating machines
are very complex and contain many
systems, predictive-maintenance strat- stages and auxiliaries. Stage degradation
egies and failure root-cause analysis and related problems have a cumulative effect.
can improve equipment reliability and A degraded stage creates different exit conditions
bring numerous other technical and and each subsequent stage then operates further
commercial benefits, performance im- away from its design point. Ultimately, degradation
at each stage or component will force all stages to
provements and cost savings. work at off-optimum efficiencies
The goal of any profitable process
plant is to operate rotating machin- own profit by producing machinery tion monitoring and predictive main-
ery with maximum reliability and ef- that meets the user’s project specifica- tenance efforts, to extend the life of
ficiency, and minimum operating and tions and applicable codes at the low- these costly components.
maintenance costs. The first step is est possible cost. The manufacturers’ When carrying out reliability stud-
to ensure proper specification and de- main focus is typically on ensuring ies and condition monitoring, rotat-
sign of rotating machinery. An impor- performance that meets the warranty ing machines should be viewed as a
tant aspect of specification is proper conditions. Thus, machine manufac- complete system, including the driver,
clarification of details up front — prior turers often do not initiate additional transmission system, coupling, all aux-
to the bidding phase and purchase- improvements that could extend reli- iliaries (such as the gear unit, if appli-
order placement. ability and trouble-free operation for cable), lubrication oil system, cooling
Increasingly, once the equipment is a period beyond that of the warranty system (if any) and seal system. Such
in operation, CPI facilities are imple- period. Many manufacturers believe a rotating machine package, regard-
menting condition monitoring and they cannot remain competitive (or less of type is always customized to
predictive-maintenance programs to perhaps even stay in business) if they meet the operator’s site-specific re-
ensure that un-spared (critical) rotat- were to design and produce equipment quirements, which include upstream
ing machines can be operated effec- that could ensure reliable, trouble-free and downstream process facilities,
tively, without an unscheduled shut- operation for a longer period (say 20 site conditions, the machine’s unique
down. At the same time, it should be or 30 years). This makes it even more battery limits and piping arrange-
noted that in general, manufacturers important for operators to exercise ment, foundation details and more.
of rotating machines maximize their great care, through the use of condi- Because every rotating-machinery
Chemical Engineering March 2012 55
Engineering Practice

FIGURE 3. Thermographic monitoring of

mechanical machines is a useful tool during
condition monitoring. Improper lubrication, mis-
alignments, and failing parts are just a few of the
common problems that can be identified using an
infrared scan of mechanical equipment. Properly
scheduled thermal inspections can help to pro-
vide early detection of problems and thus avoid
costly unscheduled shutdowns
FIGURE 2. Disassembly of a gear unit for inspection allows operators to in-
spect gear-contact patterns after installation or during operation. Under ideal
conditions, the contact pattern across each tooth will be uniform around each
gear (indicating full contact). However, machining, assembly errors, distortion
of the gear unit housing and misalignment can create problems. An improper
gear-set contact pattern creates high stresses on the gear teeth

package is essentially a customized terpreting the operating data.

enterprise, each machine produces Revised baseline data
its own unique “operational” signa- should also be collected after
ture. Figure 1 shows an example of a any major maintenance over-
a complex rotating machine in a CPI haul.
setting. Figure 2 shows an example of
a gear unit that has been taken apart What to look for
for inspection. Problematic machines often FIGURE 4. Infrared thermographic inspections of
electrical auxiliary equipment related to plant ma-
generate high vibration (dy- chinery can reveal “hot spots” that are commonly
Condition monitoring namic forces) or hot spots. This found in equipment prior to failure. Loose connec-
Condition monitoring is carried out occurs because machines or tions, faulty components, overloaded parts and high
by analyzing trends in data related to components that are operating friction are some of common problems that can be
key operating parameters. It requires inefficiently or with a problem found during a thermographic scan. All “exceptions”
should be properly documented with thermal and
the strategic use of suitable sensors to often run “hot,” causing compo- digital images that can be used in the condition
track relevant parameters, and the es- nents to wear out or fail pre- monitoring
tablishment of normal (baseline) data maturely. Temperature sensors
so that excursions between measured and infrared thermometers are also dures and represents the most im-
and baseline values can be recognized essential parts of monitoring rotating portant cause of rotating machine
quickly and analyzed to identify con- machines and auxiliary equipment. failure
dition changes. Figures 3 and 4 show examples of • Design, fabrication, assembly, in-
The major components that can ben- thermographic monitoring results. stallation and commissioning is-
efit from condition monitoring include Rotating machines do not fail ran- sues and problems
bearings (including radial and thrust domly. There are root-causes for each • Machine wear-out
bearings), seals and packings, rotors failure mode. To effectively prevent All rotating machines react to operat-
(shaft or crankshaft mechanism) and failure, the reasons leading up to po- ing conditions in the facility. Individ-
auxiliaries. Ideally, baseline informa- tential failure must be known. To ual types of machinery will adjust in
tion should be collected “early in the carry out effective failure analysis, different ways.
life” of the machine components, so that proper measurement parameters, sen- For instance, for positive-displace-
it captures “ideal” operating conditions sors, and setpoints should be defined ment rotating machines (such as
(not conditions that may have already for all key components. Being aware of reciprocating compressors, screw
been subject to some degradation). Too the major reasons for failure and ob- compressors, screw pumps and oth-
often, operators fail to collect data re- serving the conditions that could lead ers), the adjusted flowrate of the ma-
lated to baseline operating conditions to failure enable operators to address chine is not significantly affected by
because the initial weeks or months of the issue and improve reliability. the process system. Thus, flowrate
operation tend to be very busy. A lack Major failure categories include is a very good parameter to monitor
of adequate baseline data will seriously the following: over time, to assess the health of
reduce the value of condition monitor- • Process condition changes, which positive-displacement machines and
ing since there is no reference informa- may include changes in either oper- determine if any problems or wear-out
tion available for comparing and in- ating conditions or operating proce- have arisen.
56 Chemical Engineering March 2012
rapid temperature, pres- Machine troubleshooting
sure and speed changes. Troubleshooting efforts aim to discover
In many cases, the root- and eliminate the root causes of prob-
cause of mechanical dam- lems. Too often, in the rush to define
age to dynamic rotating what the problem is, sufficient time
machines is that the head is not taken to obtain all of the facts,
required by the process and this thwarts any troubleshooting
FIGURE 5. The root cause of rotating machine com- system has exceeded the effort. Any change that has happened
ponent failure is often found in the supporting aux- capability of the machine. recently (or any change compared to
iliary system, particularly the oil or cooling systems. For a given impeller vane design conditions) should be properly
Since change in the coolant supply temperature is slope, the head produced identified for all components in rotat-
a common cause of many machine failures, coolant
temperature sensors (shown here) should be used by a centrifugal pump is ing machine systems. All functions
a function of impeller di- of the components and sub-systems
Dynamic machines (such as cen- ameter and impeller speed. Once the should be clearly identified. It is nec-
trifugal compressors, axial compres- impeller is designed and fabricated, essary to interview all groups, such as
sors, centrifugal pumps and others) it can produce only one value of head operators, maintenance people, manu-
use high-speed rotating parts (such for a given shaft speed and flowrate. facturers, sub-vendors and related
as blades or impellers) to increase the The only circumstance that can cause contractors, during any troubleshoot-
velocity of the fluid, and then use the a lower value of head is if the machine ing exercise.
velocity of the fluid to increase the has experienced mechanical damage Failed components should be care-
fluid pressure. They typically have or has become fouled. Thus pressure fully observed and the mode of fail-
variable flow and pressure character- (particularly differential pressure) is ure should be identified. The vendor’s
istics. As a result, flowrate can vary a very good parameter to monitor in opinion and any consultant advice
with changing operating conditions, assessing dynamic machine health. should be considered. The machine
such as differential pressure or fluid Based on experience, the most com- history should be properly investi-
density, and thus the reliability of mon root cause of failure is changes gated, including both the operating life
dynamic rotating machines (as well to process or operating conditions. before failure and the history of any
as the reliability of their drivers and The second most-common reason re- previous failures.
auxiliaries) is directly affected by the lates to installation and commission- As noted earlier, baseline conditions
process and operating conditions. The ing issues. need to be obtained or established.
rotating machine loading, transmitted Design or manufacturing problems Operator’s logs and reliability data
torques, driver power and auxiliary rank third, but this type of problem should be examined. Special attention
system operation are affected by the usually shows up shortly after startup. is required when monitored param-
process. For example, process require- The main cause of design problems is eters exceed normal values. Trends
ments that call for a higher flowrate that the component has not been de- must be analyzed.
may result in driver overload. signed for the specified operating con- Data related to the failed compo-
The reliability of machine compo- dition. Component wear is most often nent’s supply source (that is, the man-
nents (such as bearings, seals and the result of the problem — not the ufacturer or other supplier), design,
others) is directly related to auxiliary root cause of the problem. Excessive materials, manufacturing details, as-
systems. Changes in the auxiliary sys- wear of the bearing, seal, wear-ring, sembly data and tolerances should
tems supplied fluid temperature (such or similar item usually results from all be considered. Unit piping, piping
as the temperature of lubrication oil process condition changes. Bearings stress analysis, thermal loads, nozzle
or cooling fluid supplied to the ma- often suffer as a result of assembly or loads, temperatures, support condi-
chine) resulting from a change in the installation problems. tions, foundation and generally all
cooling water temperature (for water- In rotating machine applications, surrounding facilities should also be
cooled systems) or in the ambient-air online condition-monitoring systems analyzed during the investigation to
temperature change (for air-cooled allow operators to monitor key oper- ascertain the root cause of a compo-
systems) is a common root cause of ating parameters, such as vibration, nent failure.
component failure. Figure 5 shows the displacement, velocity, air-gap, tem- New modeling methods, advanced
placement of a coolant temperature perature and other parameters in all simulation techniques and numeri-
sensor in a rotating machine. operating conditions (in all operating cal calculations have become impor-
Failure of machines and components cases including normal and special tant tools in modern troubleshoot-
often occurs because the equipment is modes). Figure 6 shows an example of ing and failure root-cause analysis.
subjected to conditions that exceed a shaft-vibration signal for a complex For example, rubbing on the casing
its design values. Most often, machin- vertical-type rotating machine. Figure is a widely reported problem in ro-
ery damage and wear occur during 7 shows an example of gear unit vibra- tating components. Realistic simu-
transient conditions, such as startup tion monitoring. Figure 8 shows an ex- lations of the dynamic and thermal
or shutdown. During these times, ample of the device arrangement used expansion of rotating parts and
the equipment is often subjected to to monitor rotational speed. inner casings are required to carry
Chemical Engineering March 2012 57
Engineering Practice

out a root-cause analysis in such Notes on gear unit inspection

cases. For many reliability studies A broken gear tooth is one of the most
or troubleshooting exercises, the use serious problems that can occur in FIGURE 6. Shown here is an example
of shaft vibration monitoring on a com-
of accurate finite element analysis a gear unit. Because of the relative plex vertical-type rotating machine. The
(FEA) of the machine can also yield weakness of the softer core material radial vibration amplitude of the shaft is
important insight. Figure 9 shows of gears, any gear tooth that has sub- an indicator of the overall mechanical
an example of an FEM model with stantial surface deterioration may be condition. Such a vibration signature al-
fine meshing carried out for a fan at risk of breakage. lows operators to detect many machine
malfunctions, including rotor unbalance,
impeller. The most common surface damage misalignment, bearing wear and rubs.
In some complex rotating machines,
vibrations are not easily detected by
measuring the dynamic motion of the

Unique Products,
shaft relative to that of the bearing. De-
pending on bearing stiffness, vibrations
may be transmitted entirely to the bear-
ing housings. Such a vibration signature

Exceptional Service,
often cannot be detected by conventional
methods for measuring shaft vibration.
In such cases, a piezoelectric acceler-

Technical Support.
ometer or seismic velocity-meter may be
used to measure the absolute severity of
the bearing vibration

mechanism is pitting. Gear pitting

occurs as the result of a combination
of fatigue forces and surface tension.
Any pits on the surface of a gear are
cause for concern because they show
that the tooth loads are far in excess
of the design loads or there is a prob-
lem. Pitting could also indicate serious
misalignment, contact pattern issues
or metallurgical problems.
Poor tooth contact, eccentric load-
ing that causes pitting, and the re-
sultant tooth fractures are the most
frequently experienced gear problems.
Micro-pitting, a less severe form of
surface fatigue damage, sometimes
occurs when there is an inadequate
lubricant film. It shows up where the
high spots of the mating gear surfaces
create pressures sufficient to cause a
Visit us at series of tiny fatigue spalls (resulting
PITTCON in a sandblasted appearance).
Booth #361 The normal progression of damage
in a tooth with excessive loads (or mis-
alignment) is that the micro-pitting
eventually yields to full-scale pitting. If
See it in action! the micro-pitting occurs in bands and
Demos on Tuesday, 3/13 in SR25
is uniform and well distributed, it indi-
at 9:30 AM, 10:30 AM, & 11:30 AM
cates that the gears are heavily loaded
or there were some machining errors.
When micro-pitting is off to one side of
a gear, it indicates there is excessive
misalignment within the unit; this
increases the chance of catastrophic
800-323-4340 failure. Future up-rating or modifica-
tions (particularly speed changes) may
3507 be required for some gear-unit-driven
rotating-machine trains.
Circle 11 on p. 70 or go to
58 Chemical Engineering March 2012
FIGURE 8. Shown here is an example
of a rotational-speed-monitoring system
using gear teeth. Common sensors that
are used to monitor rotating speed and
the phase angle of the rotor are known
as rotational-speed sensors or phase-
angle-reference transducers (most often
FIGURE 7. Continuous (online) vibra-
called key-phasor transducers). This type
tion monitoring should be implemented
of sensor can be used to tie the rotor
for critical gear units. The dynamic forces
vibration data (or other performance or
and vibration that result from gear mesh-
monitoring data) to the shaft rotational
ing often present challenges to unit
motion or phase angle reference. The re-
reliability, causing excessive gear and
sulting information is extremely valuable
bearing wear and adversely affect the
for balancing, vibration condition moni-
driver, driven equipment, steel structures,
toring, torsional monitoring and diagnos-
foundation and even nearby equipment
ing various machine malfunctions

The gear unit could be designed to the early stages to avoid catastrophic be maintained below around 70 mi-
allow for a replacement gear set with damages. While appropriate setpoints crons and 115°C, respectively. For roll-
a different speed ratio, thereby provid- must be selected on a case-by-case ing-element and hydrodynamic thrust
ing the required operating speed for basis, the following rules of thumb bearings, as a rough indication, axial
the alternate operating cases. and practical recommendations gen- displacement should be maintained
erally apply. below around 1.2 mm/s and 0.6 mm,
Practical recommendations Regarding rolling-element bearings, respectively. Lubrication-oil supply
Establishing proper alarm and trip peak housing vibration and tempera- and return temperatures should be
setpoints is very important for proper tures in the bearing housing should maintained below around 55°C and
operation of a CPI rotating machine. be maintained below around 12 mm/s 85°C, respectively.
These setpoints should be properly and 90°C, respectively. For hydrody- Lubrication oil analysis is another
selected to avoid unnecessary alarms namic bearings, peak-to-peak hous- effective tool for evaluating bear-
or shutdowns (trips). Malfunctions ing vibration and temperature in the ing health. It is helpful to measure
and problems must be identified in bearing housing are recommended to the temperature variation (typically

Whether you need to transport, analyze, weigh, batch,

mix, grind, dry, shape or package you’ll find the solution at…

Exhibition & Conference: May 8–10, 2012

Donald E. Stephens Convention Center • Rosemont, IL (adjacent to O’Hare Airport)

Meet Industry Leaders with Thousands of the Latest Processing Solutions:

• Accessories • Feeders • Plant Maintenance/ • Storage
• Conveyors and Elevators • Instrumentation & Controls Safety/Health Products • Thermal Solids Processors
• Dryers • Material Handling & Transportation • Processing/Mixing/ • Weighing Systems & Scales
Blending Equipment
• Dust Collection/Control • Mixers & Blenders …and much more
• Size Reduction
• Energy, Environmental & • Packaging & Bagging
Pollution Control • Particle Enlarger & Formers
• Filtration/Separation

Follow the tag to register or log on to:

Get the free
mobile app at
Produced and managed by: UBM Canon 11444 W. Olympic Blvd. • Los Angeles, CA 90064-1549
21377_CH_PTXi12 Tel: 310/445-4200 • Fax: 310/996-9499

Circle 29 on p. 70 or go to
Chemical Engineering March 2012 59
Engineering Practice

FIGURE 9. Finite element

using a thermographic method), both sure of the air receiver. modeling (FEM) model with
across the gear teeth and around the The machine uses fine meshing for a fan im-
gear sets, to get an idea of the load- sleeve-type bearings peller helps to identify
ing and misalignment in the gear unit. and has suffered from stresses, strains and
displacements in the
The maximum variation should not bearing failure. During an fan impeller
exceed a certain limit (most often in inspection after the failure,
6–10°C range). the bearings were found to be totally
As a rule of thumb for dynamic black and the Babbitt — a special a longer duration of excessive load
machines, if for a given flowrate and material or alloys that is used for the and a corresponding, transient lack
shaft speed the head produced falls bearing surface to provide a thin sur- of lubrication.
below the value predicted (based on face layer in a complex, multi-metal In this case, the compressor ven-
a certified and tested head-flowrate structure and help to resist galling — dor confirmed that this compres-
characteristic curve) by greater than was not found on bearing shells. sor model was originally designed
10–15%, the dynamic machine may be An investigation showed that the for continuous operation. While the
in peril and should be inspected at the plant’s consumption of instrument unit can be operated in long dura-
first opportunity. air had increased four times com- tions of intermittent operation — as
pared to its design value. Since the proposed for the specified condition
Case study flow produced by the compressor during the bidding stage — it cannot
This case study is presented for a remained unchanged, the increased afford four times the usual number
large instrument-air compressor. It flow of instrument air reduced the of starts and stops.
supplies instrument air to an air re- pressure in the air receiver more The proposed solution for this case is
ceiver, which delivers instrument air rapidly, and thus resulted in more- to change the compressor-control phi-
for various continuous and intermit- frequent stops and starts of the ma- losophy by eliminating stop/start op-
tent end uses. The compressor train chine. This put the bearings under eration, and operating the compressor
is started and stopped by the pres- more-transient stresses, resulting in continuously, using a bypass control
valve instead to maintain the desired
air-receiver pressure. In other words,
the compressor was switched to con-
tinuous operation and a new recycle
ADvANCED PROCESS SOLuTIONS loop, from the compressor discharge to
LEADING WORLDWIDE IN MIXING TECHNOLOGIES the compressor suction, was provided
to decrease the discharge pressure.
The discharge pressure needed to be
decreased to relieve stored instrument
air in the air receiver from the dis-
charge to the suction. With this new
mode of operation, there is no need
to stop the compressor. The valve at
the recycle (or bypass) loop is opened,
as needed, to adjust the instrument
air flowrate. n
Having advanced to world market leader, the EKATO GROUP has been providing their
Edited by Suzanne Shelley
customers with the technical excellence and experience for more than 75 years. The
companies within the EKATO GROUP offer a broad spectrum of mixing technologies.
Amin Almasi is a rotat-
From modulary designed industrial agitators to turnkey ing machine consultant in
production plants, the EKATO GROUP provides a range of Brisbane, Australia (amin. He holds
engineering services and custom-made solutions for the a chartered engineer cer-
most challenging customer applications. The synergies tificate from IMechE (CEng
MIMechE), a chartered pro-
within the EKATO GROUP ensure reliable and cost-effective fessional engineer’s license
solutions that meet the highest quality standards for every from Engineers Australia
(MIEAust CPEng – Mechani-
application. This is supported by a global service network. cal) and an EUR ING (Eu-
HALL 5.0, STAND D42 ropean Engineer) certificate
from FEANI. He is a registered professional
engineer in Queensland (RPEO). He also holds
M.Sc. and B.Sc. degrees in mechanical engineer-
Your contact in Europe
Tel.: +49 7622 29-0
GROUP Your contact in the USA
Tel.: +1 201 825 4684
ing. He specializes in rotating machines includ-
ing reciprocating, centrifugal and screw com-
pressors, gas and steam turbines, process pumps,
e-mail: e-mail: condition monitoring and reliability. He has au-
thored more than 50 papers and articles dealing
with rotating machines, condition monitoring
and reliability.
Circle 15 on p. 70 or go to
60 Chemical Engineering March 2012
RZ_AnzChemEng2012.indd 1 09.02.12 15:38
Feature Report

Inline Viscosity
Measurements Figure 1. A Couette-
style viscometer with
defined shear rate
measurements is
typically used when
Process viscometers can help keep measurements that
can be compared to

process control and product quality in check laboratory-type

are needed

Steve Cicchese Plant personnel may have an indica- simulate, in part, what is happening
Brookfield Engineering Laboratories tion of the viscosity or “consistency” of to the fluid during processing. This
a material by looking at it, rubbing it analytical procedure for simulating

nline viscosity measurements can between their fingers, or having it drip the shearing action with an instru-
give continuous, realtime read- off a stick or shovel. This type of prac- ment is the key to predicting flow
ings of a fluid’s viscosity during tical “measurement” of a material’s behavior.
processing and consequently, can characteristic was eventually devel- Rotational viscometers (Figure 1)
provide a means to automate the vis- oped into a somewhat more scientific are widely accepted tools for the mea-
cosity control of process fluids. While approach by using cups with holes in surement of viscosity across most in-
it is difficult to control all factors in the bottom and a stopwatch to measure dustries. The spindle of a rotational
the process that can affect a fluid’s how much time it would take to drain viscometer, when inserted into the
viscosity (such as temperature, air the fluid. The cups (for example efflux liquid, rotates at various fixed speeds,
bubbles, shear history, turbulence cups) are relatively inexpensive and thereby shearing the material contin-
and so on), if these factors are kept easy to use. This type of test uses the uously at defined shear rates. Simul-
relatively constant, then good control force of gravity to drain the fluid out taneously, the viscometer measures
can be achieved. This article presents of the cup. The shearing action on the the amount of torque resistance expe-
the applications for inline viscos- fluid takes place at the orifice on the rienced by the spindle at each speed
ity measurement and the means by bottom of the cup. As the level in the of rotation. This torque measurement
which they are achieved. Let’s first cup goes down, the shear rate at the is quantified as a “shear stress,” which
discuss the subject of viscosity. orifice decreases because the weight of acts across the surface area of the im-
the fluid remaining in the cup is lower. mersed portion of the spindle. These
The basics This type of measurement is referred two key concepts — torque resistance
Viscosity is a property that is often con- to as kinematic viscosity. This method and shearing action — are combined
sidered by process engineers, but sel- was one of the earliest quality control in an equation that defines apparent
dom completely understood. It is gen- (QC) tests that checked viscosity in a or dynamic viscosity as the ratio of
erally not a subject that is covered in quantifiable way. shear stress to shear rate.
much detail in many engineering cur- But the cup method could not al- The unit of measurement used to
ricula. Most engineers know what vis- ways discriminate successfully be- quantify rotational viscosity is the
cosity is, but may have trouble explain- tween materials that proved accept- centipoise (cP) in the western hemi-
ing it or even understanding the full able and those that were marginal or sphere, and the milliPascal second
implications of the measured number. even poor performers because of the (mPa-s) in other countries, although
Scientifically, viscosity is the property varying shear rate. Understanding a there is some degree of overlap in
of a fluid that causes it to resist flow. defined “shear rate” and how it can useage. The good news is that the two
For materials that flow, either while affect the viscosity of the fluid is im- units are interchangeable because
being processed (for pumping, spray- portant. Imagine that the fluid you 1 cP equals 1 mPa-s. There is a way
ing or coating) or in an end-use (like want to test is sandwiched between to correlate viscosity measurements
shampoo, detergent or paint), it is im- two plates separated by a known dis- made with dynamic and kinematic
portant to think about the material’s tance. Keeping the bottom plate sta- methods for materials that are New-
flow characteristics or viscosity. En- tionary and moving the top plate at tonian, using the following equation:
gineers and quality-control personnel a defined velocity, shear rate is the Dynamic viscosity = kinematic vis-
need ways to measure viscosity so that ratio of the moving plate velocity, V, cosity × density (for more on the fun-
they can quantify whether a material to the distance separating the plates, damentals of viscosity, see Viscosity:
will flow the way it needs to for the X. The use of a rotational viscome- The Basics, Chem. Eng., August 2009,
process or for the application. ter running at different speeds can pp. 34–39).
34 Chemical Engineering May 2013
η η

Upper limit

Upper limit

Manual set point Automatic set point

Lower limit Figure 3.
Lower limit
Better product
and process
Reject point control can be
obtained with
t t inline measure-
Time Time ments
Figure 2. With manual control, upper ferent, so it is unlikely that the same sitive to change in the process will
and lower control limits can often be ex-
ceeded, allowing for out-of-specification equipment can be used for both styles allow the engineer to properly control
product of measurement or that the exact the process.
same measured value will be gener-
Process measurements ated. However, if done properly, the Applications
Automatic control of process fluid vis- results of both laboratory and inline Most products are formulated to flow,
cosity ensures consistent product all measurements will follow the same spray or coat in a controlled manner.
the time, reduces product hold times, trend, making inline measurement Monitoring viscosity at critical shear
and can eliminate human errors and useful for ensuring consistent pro- points ensures that the product will
expensive sample testing (Figures duction quality. act the same way every time for the
2 and 3). Also, it provides for a com- Choosing an instrument for in- user. This is the most tangible indi-
plete record of how the process varied line measurement. When evaluat- cator of quality. With the increase
over a span of time, instead of at just ing an instrument for inline viscosity in standardization initiatives, such
one point in time. In a plant environ- control, there are several parameters as ISO 9000 and process analytical
ment, there are many ways that vis- that must be considered to provide technology (PAT), there is an increas-
cosity can be measured, such as by the proper installation. The answers ing use of viscometers to establish
a rotational viscometer, a vibrating to these questions will eliminate some and document the desired properties
element and by a falling object. Un- types of viscometers, and aid in defin- of products. To a much larger extent,
derstanding whether a true, defined ing the specifications of the final in- the use of viscometers for quality con-
shear-rate measurement is needed, strument. These questions include trol, and in particular, the use of in-
or if you are really just looking for the following: line viscometers, wherever possible,
set-point control, is beneficial when • What are the minimum, maxi- to automate the process of controlling
choosing the right type of instrument mum and average pressure and desired fluid properties is on the rise.
for your application. temperature requirements for the Quality, consistency and customer ac-
Process measurements are made application? ceptance require testing and control of
inline or in a flow loop. A bench-top ro- • What is the expected viscosity range, key parameters, of which viscosity is
tational viscometer can be used for off- and control set-points desired? certainly an important one.
line or near-line measurements, where • What are the minimum, maxi- Some typical operations where pro-
a sample of the process fluid is drawn mum and typical flowrates in the cess viscosity control can be important
and tested under controlled conditions process? include the following:
(using the same bath temperature, • What is the area electrical classifi- Determining the endpoint. For
shear history, shear rate and so on). cation (NEMA 4; NEMA 7; ATEX, applications involving chemical re-
Inline viscometers are immersed in for example)? actions, viscosity of a product is con-
the process stream and measure con- • What are the necessary materials of tinuously monitored in-tank and the
tinuously under process conditions. construction, and recommended seal process is either stopped, or the next
Installation can be in a side-stream, and elastomer materials? (This can steps are taken once a specific viscos-
in the main flow stream or in a tank. often be determined based on what ity limit is reached.
It is important to consider how clean- the plant is using for other process In addition to determining the end-
ing and maintenance of these devices equipment in the area, such as point of chemical reactions, this ap-
might occur, if necessary, when decid- pumps.) proach is also used to determine the
ing on the installation. • Where will the instrument be endpoint of blending operations, such
It is also important to make sure mounted? This will determine the as the blending of multiple ingredi-
that a representative sample of the style of instrument to be used. Ex- ents in a batch process. One example is
fluid will be measured. Possible con- amples include a tank/flange mount synthetic-fiber manufacturing. Latex,
cerns about stratification, mixing and (Figure 4); a flow-through housing spandex and other synthetic materials
turbulence should be considered. The (Figure 1); a probe style for barrels are used to manufacture fibers, which
instrument will measure the product (Figure 5); and mounting from the are stretchable, rugged and used in
with which it makes contact, so mak- top of an open tank many applications such as clothing.
ing sure the fluid that the instrument For process control measurements, The manufacturing process is carried
“sees” is the material that you want the critical factors are stability, re- out in a reactor, where both tempera-
it to measure, is a primary consid- peatability and sensitivity to changes ture control and tight viscosity control
eration. The demands of laboratory in viscosity. A stable, repeatable read- are required over the steps and addi-
versus process environments are dif- ing from an instrument that is sen- tions made during the process.
Chemical Engineering May 2013 35
Feature Report

Carrying properties in oil and gas. Figure 4. This

In many oil and natural-gas production vibrating-probe
style viscometer
applications, viscosity is monitored and can be mounted to
controlled to make sure that the fluids a flange on the side
have the proper rheological properties of a tank, for example.
to carry solids. For example, hydraulic Vibrating-probe style
viscometers are typi-
fracturing fluid must have the proper cally used for trending Figure 5.
viscosity under various shear condi- viscosity changes, rather This vibrating-
tions to carry the proppant downhole, than obtaining absolute probe style viscom-
and deposit it at the required location. viscosity values eter can be mounted on
For drilling fluids, the viscosity must the top of a tank, for example
be correct to carry the cuttings away
from the drill bit and out of the hole, as The inline couette viscometer (Fig- cosity of a fluid is controlled
well as to lubricate the bit. ure 1) gives field engineers reliable vis- so that when it is pumped
Field engineers in oil-and-gas drill- cosity measurement, onsite at the well. through a spray nozzle,
ing operations can experience operat- This simplifies complicated test pro- proper atomization of the
ing problems if the viscosity specifica- cedures, minimizes human error and material occurs. Proper
tions of fluids pumped downhole are ensures quality control without delay. atomization through spray
incorrect. This complicates testing The instrument output allows for con- nozzles, which requires con-
procedures, increases the risk of costly stant monitoring and reporting of fluid tinuous and accurate viscosity
errors and wastes time. Consequently, viscosity or for use in ECD (equivalent measurement and control, en-
it is necessary to ensure fast, accurate circulating density) calculations by oil- sures the best combustion efficiency
viscosity measurement, data collec- rig engineers. in oil-delivery systems. To burn fuel
tion and analysis of fluid samples be- Oil-delivery systems, such as for oil at the high-volume flowrates de-
fore they are pumped downhole. burners. In this application, the vis- manded of modern boiler units, the


The Name No other company in the world has more expertise with silicon carbide than
That Makes Saint-Gobain Ceramics. Our Hexoloy sintered alpha silicon carbide is the
material of choice for high performance applications in a variety of chemical
A World processing industries throughout the worldwide market. Hexoloy compo-
Of Difference nents are custom made and offer excellent performance at temperatures
up to 1650°C (3000°F), universal corrosion resistance, excellent wear
resistance, high strength, and high thermal conductivity. Wherever
you’re located in the global market, specify Hexoloy silicon carbide, the
name that delivers performance you can count on.

Saint-Gobain Ceramics Telephone: 716-278-6233

23 Acheson Drive Fax: 716-278-2373
Niagara Falls, New York 14303

Circle 39 on p. 72 or go to
36 Chemical Engineering May 2013
Figure 6. This vibrational viscometer is commonly
used in printing and coating applications

oil must be atomized (dispersed into sign considerations, which may also cations include shampoos, detergents
the furnace as a fine mist). This as- be of interest to other spray-type ap- and yogurts to name a few. In these
sures high-speed vaporization and plications, such as spray-drying oper- cases, too thin a product might ap-
ignition. Most burners atomize oil by ations. Some of these considerations pear to be of poor quality (such as a
shearing the oil into small droplets. are the following: runny yogurt, or a shampoo that pours
Burner manufacturers recommend • Bypass loop for viscometer installa- like water, without any body). Here it
that the oil be supplied to the burn- tion for fail-safe operation is a matter of consumer perception
ers at a specific viscosity to main- • Use of viscosity feedback to control where specifications of final products
tain consistent atomization. Failure the heat rate to the oil-feedline heat are written by companies based upon
to maintain proper atomization re- exchanger consumer test groups, and the product
sults in the following: poor fuel burn- • Output from the viscometer may go must fall within these specifications
ing due to carbon and soot buildup; to a single-loop controller that in- in order to be shipped.
higher fuel consumption and costs; stantly responds to inline viscosity Other quality-related applications
increased stack emissions and possi- changes where viscosity is important are in
ble fines from government agencies. Quality control. To ensure consis- coatings, such as paint applications. A
The inline process viscometer mon- tent quality of many different prod- few more examples follow.
itors and controls viscosity and tem- ucts, it is important that the viscosity Roll-coating thickness control.
perature in pressurized oil-delivery be constantly measured and controlled When dealing in any large-volume
systems. Repeatable viscosity mea- during the production process. In- coating and printing applications
surements are necessary to maximize line measurement ensures consistent where millions of products are being
the efficient atomization and delivery quality control in realtime. It saves on printed per day, the payback on the
of a variety of paraffin-based oils, laboratory testing times, and reduces cost of inline control can be very
asphaltic-based oils, as well as heat- hold-up of product in tanks waiting short when measured against costs of
ing fuels and waste oils. The process for evaluation. Examples of quality- wasted ink, varnishes, or coatings from
CIC-10307 can include
halfp additional
page ad.qxd de-
3/25/07 control
6:19 PMandPage quality-assurance
1 appli- too high a viscosity or from wasted


Collins plastic control valves are
highly responsive control valves
designed for use with corrosive
media and/or corrosive atmos-
Collins valves feature all-plastic
construction with bodies in PVDF,
PP, PVC and Halar in various body
styles from 1/2" - 2" with Globe,
Angle or Corner configurations and
many trim sizes and materials.
Valves may be furnished without
positioner for ON-OFF applications.
Call for more information on our
plastic control valves.

P.O. Box 938 • Angleton, TX 77516

Tel. (979) 849-8266 •
Circle 15 on p. 72 or go to
Chemical Engineering May 2013 37
Feature Report

product if the viscosity (and hence the sistency of a fluid is controlled so that For food batter applications, too thin
color) are too light. when something is dipped in it and a batter will mean improper coating
In printing applications, constant pulled out, it is uniformly and consis- and product quality. Too thick a batter
maintenance of proper ink viscosity tently coated. will mean bad product quality, longer
ensures the quality of the printing, Dipping applications are designed in cook and dry times, and raw mate-
which reduces rejects and waste, while automated systems whereby an item rial waste. You can easily imagine a
also keeping ink costs to a minimum. is brought over a tank or pan, dipped chicken nugget with too much batter.
To assure the uniform application of into the bath, removed and allowed This can be from too viscous a batter
inks on a variety of substrates (boxes, to drip dry before proceeding through during the coating process. ■
newspapers, cartons) it is necessary the process. The main problem with Edited by Dorothy Lozowski
to control viscosity. Continuous moni- the open tank or pan is with the evap- Author
toring and control of the ink reservoir oration of fluids to the environment. Steve Cicchese is the gen-
viscosity using an inline viscometer Viscosity control is used for addition eral manager of process sales
and marketing for Brookfield
(Figure 6) can provide viscosity mea- of water, solvents or other modifiers Engineering Laboratories
surement and control at multiple sta- as needed to control viscosity to a set (11 Commerce Blvd., Mid-
dleboro, MA, 02346; Email:
tions and save money by using less point. In pharmaceutical capsule man- s_cicchese@brookfieldengi
ink. Similar controls are needed in the ufacturing, for example, if the fluid is; Phone: 508-946-
6200) where he has worked
coating of, and the printing on soda too thin, the capsule will break during for the past 12 years. Prior to
that, he spent 15 years with
and other beverage cans. filling, or dissolve too soon when swal- Bird/Baker Process, a manu-
Dip-coating thickness control. The lowed, which would release medicine facturer of liquid-solid separation equipment.
Cicchese holds a B.S.Ch.E. from Northeastern
thickness or consistency of a fluid is in the throat instead of the stomach. If University, an MBA from Babson College and a
controlled to provide a dependable and it is too thick, then there is raw mate- certificate in business administration from Har-
vard University. He is a member of the AIChE
uniformly coated item when dipped, rial waste on millions of capsules that and Society of Petroleum Engineers, and has
then removed, from a coating tank. In will raise product costs, and it may not written numerous articles on process viscosity
measurement in various industries, including
this application, the thickness or con- dissolve properly when swallowed. food, printing, asphalt and oil and gas.

Keep your production

and profitability
Considering a maintenance turnaround or
process improvements? Add Aggreko to
your planning team. We create customized
solutions to minimize downtime and improve
processes with our premium fleet of rental
equipment, dedicated teams of engineers and
industry-specific expertise.


With Aggreko by your side, you’ll have the solution
you need to keep your projects up and running—from
Temporary HVAC
shutdown and turnaround power to seasonal process
Rental Diesel and Natural Gas Generators
and supplemental cooling.
Chillers and Dehumidifiers
Cooling Towers
100% Oil-free Compressed Air 800.348.8370 |

Circle 5 on p. 72 or go to
38 Chemical Engineering May 2013
Cover Story

Piping-System Leak Detection

and Monitoring for the CPI
Eliminating the potential for leaks is an integral part ACRONYMS

of the design process that takes place at the very

AVO = Audio/visual/olfactory
CAA = Clean Air Act
onset of facility design

HAP = Hazardous air pollutants
HON = Hazardous organic NESHAP
W. M. (Bill) Huitt When discussing anything to do LDAR = Leak detection and repair
W.M. Huitt Co. with government regulations, the LUST = Leaking underground
terminology quickly turns into an storage tank

eaks in a chemical process in- “alphabet soup” of acronyms. The NEIC = National Enforcement
dustries (CPI) facility can run box on the right lists, for easy refer- Investigations Center
the gamut from creating a ence, the titles and acronyms that
NESHAP = National Emission
costly waste to prefacing a cat- will be used in this discussion.
Standard for Hazardous Air
astrophic failure.. They can be an Pollutants
annoyance, by creating pools of liq- Leak mechanisms
uid on concrete that can become a Eliminating the potential for leaks NSPS = New Source Performance
possible slipping hazard and house- is an integral part of the design
keeping problem, or a leak that can process that takes place at the very RCRA = Resource Conservation and
emit toxic vapors, causing various onset of facility design. It is woven Recovery Act
degrees of harm to personnel. In into the basic precept of the piping SOCMI = Synthetic organic chemical
some cases a leak may be a simple codes because it is such an elemen- manufacturing industry
housekeeping issue that goes into tal and essential component in the TSDF = Treatment, storage and
the books as a footnote indicating process of designing a safe and de- disposal facilities
that a repair should be made when pendable piping system. UST = Underground storage tank
resources are available. In other Piping systems, as referred to
cases it can become a violation of here, include pipe, valves and other VOC = Volatile organic
regulatory compliance with statu- inline components, as well as the
tory consequences, not to mention equipment needed to hold, move and
a risk to personnel safety and the process chemicals. Why then, if we (CA) is used as an applied factor
possible loss of capital assets. comply with codes and standards, in calculating, among other things,
Understanding the mechanisms and adhere to recommended indus- wall thickness in pipe and pressure
by which leaks can occur and priori- try practices, do we have to concern vessels. The CA value assigned to
tizing piping systems to be checked ourselves with leaks? Quite point- a material is theoretical and predi-
at specific intervals based on a few edly it is because much of what we cated on four essential variables:
simple factors is not only a prag- do in design is theoretical, such as material compatibility with the
matic approach to the preventive material selection for compatibility, fluid, containment pressure, tem-
maintenance of piping systems, but and because in reality, in-process perature of the fluid and velocity
is part of a CPI’s regulatory com- conditions and circumstances do of the fluid. What the determina-
pliance. This includes compliance not always perform as expected. tion of a CA provides, given those
under both the U.S. Environmen- Whether due to human error or variables, is a reasonable guess at
tal Protection Agency (EPA) Clean mechanical deficiencies, leaks are a uniform rate of corrosion. And
Air Act (CAA; 40CFR Parts 50 to a mechanism by which a contained given that, an anticipated loss of
52) and the Resource Conservation fluid finds a point of least resistance material can be assumed over the
and Recovery Act (RCRA; 40CFR and, given time and circumstances, theoretical lifecycle of a pipeline
Parts 260 to 299). We will get into breaches its containment. What we or vessel. It allows a reasonable
more detail with these regulations, look into, somewhat briefly, are two amount of material to be added into
as well as the leak detection and general means by which leaks can the equation, along with mechani-
repair (LDAR) requirement within occur; namely corrosion and me- cal allowances and a mill tolerance
the above mentioned regulations, as chanical joint deficiencies. in performing wall thickness cal-
we move through this discussion. Corrosion. Corrosion allowance culations. The problem is that be-
44 Chemical Engineering May 2014
Table 1. Elements of a Model LDAR Program Pump, compressor and agitator
seals can develop leaks where shaft
Written LDAR compliance First attempt at repair
misalignment plays a part. If the
Training Delay of repair compliance assurance shaft is not installed within recom-
LDAR audits Electronic monitoring and storage of data mended tolerances or if it becomes
misaligned over time there is a
Contractor accountability QA/QC of LDAR data good possibility the seal will begin
Internal leak definitions Calibration/calibration drift assessment to fail.
Less frequent monitoring Records maintenance The LDAR program
Promulgated in 1970 and amended
yond the design, engineering, and areas for abnormal loss of wall in 1977 and 1990, the Clean Air
construction phase of building a thickness, hydrogen stress-corro- Act requires that manufactur-
facility, the in-service reality of cor- sion cracking (HSCC), and others. ers producing or handling VOCs
rosion can be very different. The LDAR program does not develop and maintain an LDAR
Corrosion, in the majority of specify the need to check anything program in accordance with the
cases, does not occur in a uniform other than mechanical joints for po- requirements set forth under the
manner. It will most frequently tential leaks. Monitoring pipe and Clean Air Act. This program moni-
occur in localized areas in the form vessel walls, particularly at welds tors and documents leaks of VOCs
of pits, as erosion at high-impinge- that come in contact with corrosive in accordance with Method 21 —
ment areas, as corrosion under chemicals, is a safety consideration Determination of Volatile Organic
insulation, at heat-affected zones and practical economics. Perform- Compound Leaks.
(HAZ) where welding was improp- ing cursory examinations for such Table 1 provides a listing of key
erly performed, causing a localized points of corrosion where the po- elements that should be contained
change to the mechanical or chemi- tential exists should be made part in an LDAR program. Those ele-
cal properties of the material, and of any quality assurance or quality ments are described as follows:
in many other instances in which control (QA/QC) and preventive Written LDAR compliance. Com-
unforeseen circumstances create maintenance program. pile a written procedure declaring
the potential for corrosion and the Mechanical joints and open- and defining regulatory require-
opportunity for leaks in the pipe ended pipe. Mechanical joints ments that pertain to your specific
itself or in a vessel wall. Because can include such joining methods facility. This should include record-
of that incongruity, corrosion is an as flanges, unions, threaded joints, keeping certifications; monitoring
anomaly that, in reality, cannot valve bonnets, stem seals and clamp and repair procedures; name, title,
wholly be predicted. assemblies. It can also include and work description of each person-
Corrosion-rate values found in pump, compressor and agitator nel assignment on the LDAR team;
various published resources on the seals. Other potential points of tran- required procedures for compiling
topic of material compatibility are sient emissions include open-ended test data; and a listing of all process
based on static testing in which a piping, such as drains, vents, and units subject to federal, state and
material coupon is typically set in the discharge pipe from a pressure- local LDAR regulations.
a vile containing a corrosive chemi- relief device. Any of these joints or Training. Assigned members of
cal. This can be done at varying interfaces can be considered poten- the LDAR team should have some
temperatures and in varying con- tial leak points and require both experience base that includes work
centrations. After a period of time, monitoring and record-keeping doc- performed in or around the types of
the coupon is pulled and the rate umentation in compliance with the piping systems they will be testing
of corrosion is assessed. That is a EPA’s LDAR program. and monitoring under the LDAR
simplification of the process, but Mechanical joints can leak due to program. Their training should in-
you get the point. When a material improper assembly, insufficient or clude familiarization with Method
of construction (MOC) and a po- unequal load on all bolts, improp- 21 and also training as to the cor-
tentially corrosive chemical come erly selected gasket type, sufficient rect procedure for how to examine
together in operational conditions, pressure or temperature swings the various interface connections
the theoretical foundation upon that can cause bolts to exceed their they will be testing. They should
which the material selection was elastic range (diminishing their also receive training on the test
based becomes an ongoing realtime compressive load on the joint), and instrument they will be using and
assessment. This means that due an improperly performed “hot-bolt- how to enter the test data in the
diligence needs to be paid to exam- ing” procedure in which in-service proper manner. All of this needs to
ining areas of particular concern, bolts are replaced while the pipeline be described in the procedure.
depending on operating conditions, remains in service. “Hot bolting” is LDAR audits. An internal audit
such as circumferential pipe welds not a recommended procedure, but team should be established to en-
for cracking, high-impingement is nonetheless done on occasion. sure that the program is being car-
Chemical Engineering May 2014 45
Cover Story

ried out on a routine basis in an ef- ity has consistently demonstrated monitoring equipment should be
ficient and comprehensive manner good performance under monthly made at the end of each monitor-
in accordance with the written pro- testing, then the frequency of test- ing work shift using approximately
cedures. A third-party audit team is ing could be adjusted to a quarterly 500 ppm of calibration gas. If, after
brought in every few years to con- test frequency. the initial calibration, drift assess-
firm that internal audits are being First attempt at repair. Upon de- ment shows a negative drift of more
carried out in the proper manner tection of a leak, most rules will re- than 10% from the previous cali-
and that all equipment that should quire that a first attempt be made bration, all components that were
be included in the monitoring is to repair the leak within five days tested since the last calibration
listed as such. It also ensures that of detection; if unsuccessful, any fol- with a reading greater than 100
the tests are being carried out prop- low-up attempts need to be finalized ppm should be re-tested. Re-test all
erly and that the test results are within 15 days. Should the repair pumps that were tested since the
entered properly. remain unsuccessful within the 15- last calibration having a reading of
Contractor accountability. day time period, the leak must be greater than 500 ppm.
When selecting an outside con- placed on a “delay of repair” list and Records maintenance. Internal
tractor to perform internal LDAR a notation must be made for repair electronic record-keeping and re-
audits for a facility or when bring- or component replacement during porting is an essential component to
ing in an outside contractor to in- the next shutdown of which the a well-implemented LDAR program.
spect the work of the internal audit leaking component is a part. It is an indication to the NEIC that
team, it is recommended that the Delay of repair compliance as- every effort is being made to comply
contract be written in a manner surance. Placing a repair item on with the regulations pertinent to a
that places appropriate responsi- the “delay of repair” list gives assur- facility. It provides ready access to
bility on that contractor. In doing ances that the item justifiably be- the personnel associated with the
so there should be penalties de- longs on the list, that a plan exists program, the test data, leak repair
scribed and assessed as a result to repair the item, and that parts reports and so on.
of insufficient performance or in- are on hand to rectify the problem.
accurate documentation of pre- It is suggested that any item being Testing for leaks
scribed testing and documentation listed in the “delay of repair” list au- Results, when using a leak detec-
procedures. Expectations should tomatically generate a work order tion monitor, are only as accurate
be well defined and any deviation to perform the repair. as its calibration and the manner in
from those prescribed norms by a Electronic monitoring and stor- which it is used. Calibration is dis-
third-party contractor should con- age of data. Entering leak-test cussed in the next section, “Method
stitute a breach of contract. In all data into an electronic database 21.” To use the monitor correctly, the
fairness, both parties must under- system will help in retrieving such auditor will need to place the nozzle
stand exactley what those expecta- data and in utilizing them in ways or end of the probe as close as pos-
tions are. that help provide reports highlight- sible to the flange, threaded joint, or
Internal leak definitions. Inter- ing areas of greater concern to areas seal interface as follows:
nal leak definitions are the maxi- of lesser concern. Such information • In the case of a flange joint test:
mum parts per million, by volume can help direct attention and re- 180 deg around perimeter of the
(ppmv) limits acceptable for valves, sources away from areas of least flange joint at their interface
connectors and seals, as defined by concern, while mobilizing resources • In the case of a threaded joint test:
the CAA regulation governing a fa- to areas of greater concern. This en- 180 deg around perimeter of inter-
cility. For example, a facility may be ables a much more efficient use of face of the male/female fit-up
required to set an internal leak-def- information and resources. • If it is a coupling threaded at both
inition limit of 500 ppm for valves QA/QC of LDAR data. A well ends, check both ends 180 deg
and connectors in light liquid or gas/ written LDAR program will include around the perimeter
vapor fluid service and 2,000 ppm a QA/QC procedure defining the • If it is a threaded union, then
internal leak definition for pumps process by which it is assured that check both ends and the body nut
in light liquid or gas/vapor fluid Method 21 is being adhered to, and 180 deg around the perimeter
service. “Light liquid” is defined that testing is being carried out in • In the case of a valve test:
as a fluid whose vapor pressure is the proper manner and includes the 180 deg around perimeter of
greater than 0.044 psia at 68°F. proper equipment and components. all end connections if anything
Less frequent monitoring. Under This also includes the maintenance other than welded
some regulations it is allowed that of proper documentation. 180 deg around perimeter of
a longer period between testing is Calibration/calibration-drift body flange
acceptable if a facility has consis- assessment. LDAR monitoring 180 deg around perimeter of
tently demonstrated good perfor- equipment should be calibrated in body/bonnet interface
mance (as defined in the applicable accordance with Method 21. Cali- 180 deg around perimeter of
regulation). For example, if a facil- bration-drift assessment of LDAR stem packing at stem
46 Chemical Engineering May 2014
National Cleanup Backlog
(confirmed releases, cleanups completed) 160,000


120,000 by ionizing a sample and then mea-

National backlog

suring the charge (that is, number

of ions) produced.
80,000 Two methods of ionization cur-
rently used are flame ionization
60,000 and photoionization. The flame ion-







40,000 ization detector (FID) theoretically
measures the total carbon content
20,000 of the organic vapor sampled. The
photoionization detector (PID)
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 uses ultraviolet light to ionize the
organic vapors. With both detec-
Fiscal year
tors, the response will vary with
Figure 1. Progress is slowly being made to clean up leaking underground storage the functional group in the organic
tanks under the RCRA program compounds. PIDs have been used to
detect equipment leaks in process
• In the case of a rotating equipment calibration. One is referred to as a units in SOCMI facilities, particu-
shaft seal test: 180 deg around the “zero gas,” defined as air with less larly for compounds such as form-
perimeter of the interface of the than 10 ppmv (parts per million aldehyde, aldehydes and other oxy-
seal and the shaft by volume) VOC. The other cali- genated chemicals that typically do
bration gas, referred to as a “refer- not provide a satisfactory response
Method 21 ence gas,” uses a specified reference on a FID-type unit.
Method 21, under 40 CFR Part 60, compound in an air mixture. The Operation of the non-dispersive
Appendix A, provides rules with concentration of the reference com- infrared (NDIR) detector is based
respect to how VOCs are moni- pound must approximately equal on the principle that light absorp-
tored and measured at potential the leak definition specified in the tion characteristics vary depending
leak points in a facility. Those po- regulation. The leak definition, as on the type of gas. Because of this,
tential leak points include, but are mentioned above, is the threshold NDIR detection can be subject to
not limited to: valves, flanges and standard pertinent to the govern- interference due in large measure
other connections; pumps and com- ing regulation. to such constituents as water vapor
pressors; pressure-relief devices; and CO2, which may absorb light
process drains; open-ended valves;