Sei sulla pagina 1di 11

Store | Buyers' Guide | Conferences | Webinars | White Papers | The Magazine | Awards | Editorial Submissions

Platinum Edition My Account ex.;Logout


My Che.com www.altairstrickland.com and three other companies are finalists out of 138

PDFs Comments (0)

March 1, 1990

PILOTING PROCESSES INTO COMMERCIAL PLANTS


Suitably designed and well-operated, a pilot plant will ensure the smooth, safe startup of a
commercial plant
Richard P. Palluzi, Exxon Chemicals America

Richard P. Palluzi, Exxon Chemicals America

A pilot plant is a tool for investigating a process, or a process problem, on a manageable scale, and in a
realistic and timely manner. It is normally operated with limited resources. Its equipment is large enough for all
the important factors to be evaluated, yet small and simple enough to be as economical as possible.

In a pilot plant, an engineer studies the effects of variations in operating conditions or on diagnosing problems
and determining corrective actions. A pilot plant and its products are a means to an end — the commercial
facility and its products — rather than an end itself.

A range of sizes

Pilot plants come in bench-top (or micro-unit), integrated and semiwork (also known as prototype) sizes. The
first fits on a bench top (hence the name) or inside a small laboratory hood, and ranges in area from 5 to 10
ft2. The process units are connected by tubing varying in size from 1/16 to 1/4 in.

In the past, such units were operated manually and attended continuously; nowadays, they are frequently
automated to run continuously and mostly unattended.

Integrated pilot plants remain the workhorse units. They vary from occupying several pallets to a small
building, covering areas ranging from 20 to 150 ft2. The equipment is connected by tubing varying from 1/4 to
1 in. dia. and piping from 1/4 to 2 in. dia. Such plants are usually automated for virtually unattended operation.

Semiwork units are the largest (although at the lower end of plant-scale unit size), typically larger than 10,000
ft2, with the equipment connected by commercial-size pipe. They tend to resemble commercial units in
automation and operation.

Simulating or solving

Classifying pilot plants by size is useful for placing their costs and scales in perspective. Size, however, is not
necessarily an indicator of a plant's intrinsic value. A more useful measure of a pilot plant's value is its
purpose.

One purpose is process simulation. In this type of plant, all the major process steps are set up in the actual
order that they are likely to be arranged in the full-size plant. The pilot-plant equipment will be similar, although
smaller.
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 2 of 11

Among the advantages of this type of pilot plant are that any part of a process can be investigated, and the
part does not have to be identified in advance. The plant's investigative scope is encompassing. Because
virtually all of a process can be simulated, there is less chance of a critical step being missed.

Additionally, the data are easier to correlate and safer to scale up than for other types of pilot plants. This pilot
plant can also be readily integrated with the operation of a full-scale plant, which extends its useful life. If
simply a pared-down model of an existing operating plant, a pilot plant can be easily designed, although
complex and costly to build.

A pilot plant for problem solving tends to be narrow in scope because it is limited to those components of a full
-scale plant that are being plagued by problems. However, this limitation tends to make this type of pilot plant
less expensive than others, but data are readily collected because the plant is customized for solving a
particular set of problems.

On the other hand, such a very specialized plant cannot be readily adapted for another purpose, and it would
be unrealistic to construct a pilot plant for each new problem. Of course, this shortcoming can be offset by
adapting the plant to investigating similar problems for several different processes, because the process steps
or equipment are alike. Of course, there would be a calamity if the entire plant were rendered useless because
the problem to be evaluated was incorrectly identified. Despite these disadvantages, this type of pilot plant is
the most common.

A basic research pilot plant is designed for acquiring fundamental knowledge about a process. The character
of the data required and the type of investigation to be pursued determine the design of this type of pilot plant.

The size and complexity of this type of facility qualify it as a pilot plant rather than a laboratory research unit.
Indeed, this is its niche, because accurate data usually cannot be obtained efficiently in a full-size pilot plant or
truly representative data in a laboratory, because operating conditions are normally too complex and too
demanding for laboratory equipment. The primary disadvantage of this pilot plant lies in its narrow focus. The
plant's usefulness is limited to the initial design, because redesigning it is generally not feasible economically.

Design philosophy

Type-casting pilot plants by design philosophy differs from defining them by purpose, in that the former
pertains to the utility of the design, which can be narrow or broad.

A single-purpose plant permits studying only one part of a process. The objective is usually clearly defined,
and the scope of the operating parameters is restricted. A plant can be designed and built cheaply and quickly
because of these qualities. Operating costs also are generally low. Often, such a plant is designed and built so
that it can be easily modified to serve another purpose. If this is not the case, and the initial design basis were
wrong, the plant could be useless.

A multi-purpose plant is designed to accomplish two or more differing objectives. This can make it more cost-
effective than a single-purpose plant. Construction, operating and support costs will be lower than for two
single-utility plants.

On the negative side, this type of plant can be difficult to design, especially if the objectives being sought
require widely differing capabilities. Initial costs are frequently higher than for two single-utility plants because
the design work is more involved, the components are more expensive and the construction work is extensive.
In addition, this pilot plant can be difficult to operate and expensive to maintain because the combined
features can rarely be optimized for more than one objective. Downtimes will be more frequent and support
levels higher because of this plant's complexity.

A multiple-unit pilot plant is a single-purpose plant that is replicated two or more times. The units share
considerable equipment and services (e.g., utilities, instrumentation, supervision), but each can operate as a
separate entity. Thus, the units can operate simultaneously: generating data rapidly by testing similar feeds at
different conditions; verifying data by testing identical feeds at the same conditions; or supporting several
programs by testing a variety of feeds at different conditions.

Large amounts of data can be gathered rapidly in this plant. Operating costs are lower than in other plants
because the same staff can run several units simultaneously. Design and construction costs are also less
because the units are identical. Unless the initial design is well defined, however, errors will be multiplied and
thus be more costly to correct. This problem can be at least partly avoided by starting up each unit before
building the next one.

A general-purpose pilot plant serves a broad purpose. It is designed to operate over a wide range of
feedrates, feedstocks and operating conditions, simulating an entire process. A large variety of studies can be
performed in it at little additional expense. Because its scope is not limited its useful service life can be long. It

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 3 of 11

makes an excellent tool for scaling up equipment and testing construction materials. On the other hand, it is
not very useful for solving specific problems. Because of its size and complexity, building this kind of unit
tends to be capital intensive and operating it labor intensive.

Pilot plant layout

The layout of a pilot plant is critical to optimizing operations. Too often, a pilot plant is simply placed into the
most readily available space without regard for long-term efficiency and costs.

There are four basic layouts:

—Separate building

—Containment (blast) cell

—Open bay

—Laboratory

Pilot plants had long been housed in separate buildings. Initially, they were placed close to the full-scale plant,
in whatever space was available. However, concerns about safety and operational efficiency of the production
units led to pilot plants being located in separate buildings, away from the production units.

Locating pilot plants in different buildings eliminates the possibility that their operations will interfere with each
other's. Still another advantage of separate buildings is that each can be optimized for specific purposes. For
example, the electrical apparatus in a building housing a safe process may qualify for ordinary, rather than
hazardous, classification. In addition, such amenities as air conditioning can be added economically.

Although pilot plants in separate buildings can still be found, this arrangement has fallen out of favor. Isolating
a pilot plant protects it from extraneous hazards but not from its own. Generally, it is safer and more
economical to locate a pilot plant in a containment cell and to equip it well with safety systems. Servicing
separate buildings is rarely economical. The costs of support services (e.g., utilities, sample pickup, mail
delivery) and staffing are greater with numerous buildings than with a single large one.

The idea for isolating pilot plants by means of containment cells came from experimental work with explosives
and rockets. The first barricades were steel walls, which then evolved into reinforced-concrete bunkers.
Typically, the walls or roofs are the blow-out type, designed to vent the force of an explosion to a safe area.
Total-containment cells are much more expensive.

Containment cells are regarded as the safest arrangement if all operations are remotely controlled from
outside the cell and operators only enter the cell when the unit is down. Greater risks can be accepted if only
equipment, not personnel, will be exposed to hazards. If cells are grouped, utilities can be supplied
economically and controls can be unified.

Containment cells are very expensive, with the costs of a 100- to 400-ft2unit ranging from $100,000 to
$1,000,000, depending on the degree of containment and the layout. Installing a pilot plant in a containment
cell is also costly, primarily because of the extensive automation for remote control.

Operating a pilot plant in a cell is also expensive, and operating efficiency is limited by the restricted access.
Total remote control cannot be ensured, except by complex and costly interlock systems. If operators begin
entering the cell routinely, the chief advantage of this arrangement is lost. That a cell's blast rating will not be
exceeded is not easily ensured. Estimating the blast potential of process failures is difficult, and placement in
the cell can affect the rating.

The open bay is becoming the preferred pilot-plant layout because instrumentation, which has been made
more reliable, has reduced the risks of pilot plants, allowing many to be constructed in open and inexpensive
areas. An open bay is a large area having few, or no, vertical partitions. The lack of partitions permits laying
out a pilot plant in any configuration.

The bay arrangement is the least costly to construct. An open warehouse or a prefabricated metal building can
serve as the basic structure. Many utilities can be economically supplied to a bay structure housing several
pilot plants. The openness enhances safety because operators are rarely alone or even out of sight. However,
the addition of another pilot plant next to an existing one should be evaluated carefully, because a spill or fire
in one can easily spread to another.

Laboratory-type pilot plants are a new phenomena brought about by reductions in the sizes of pilot plants.
Many pilot plants can now actually fit inside a standard or a slightly expanded laboratory hood. The chief
advantage of this arrangement is that operations can be conducted in a controlled, well-ventilated
environment.

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 4 of 11

The high rate of ventilation in the hood substantially reduces the hazards of dealing with toxic materials.
However, this feature makes the laboratory pilot plant expensive to construct and even more costly to operate.
Attempts to lessen the operating cost via a heat-recovery system have generally proven to be only marginally
effective. This cost is, of course, even higher if, for safety, the air is not recycled. Relative costs for the
different pilot plant

arrangements are listed in Table 1. These are typical ranges; special circumstances can result in wide cost
variations.

Operating safely

As has been noted, certain layouts enhance safety. However, safety can be achieved in other ways, and a
containment cell or separate building is not a guarantee of safety. Other measures include: intense process
monitoring through high-caliber instrumentation, and limiting inventories of hazardous feedstocks and
products.

One way of estimating the amount of space required for safe pilot plant operation is to divide the unit into
sections and judge the space requirements of each. The space required for storing and handling feedstocks
and products frequently exceeds that for the pilot plant. In allocating this space, ask yourself:

—How much capacity is needed for storage in case of a sudden process upset?

—How should material that must be recycled be stored?

—What will be the lead time for ordering feedstock?

—What will be the holding time for the disposal of product and how much of it will be shipped at a time?

—Will there be enough storage space for such things as cleaning and flushing solvents?

—Will there be sufficient room for storing drums, cans and other containers?

Some of these spaces will have to be suitably prepared. Tanks in which flammable or toxic liquids will be
stored will have to be diked. Some parts of the plant may have to be covered. Rainwater may have to be piped
away. If overtime work is possible, the lighting must be adequate. These represent only a sampling of things
that must be considered.

Tightened regulations has made the control of emissions and the disposal of wastes increasingly more costly.
Emissions of benzene, for example, are severely restricted and subject to strict handling requirements.
Therefore, if benzene is a feedstock, a pilot plant cannot be operated in an open bay or in other layouts
without good ventilation.

All operating and storage equipment must meet the latest code requirements. In the U.S., this includes the
codes of the Occupational Safety and Health Administration (OSHA), Environmental Protection Agency (EPA),
National Fire Protection Assn. (NFPA), American National Standards Institute (ANSI), American Petroleum
Institute (API), as well as provisions of the Resource Conservation and Recovery Act (RCRA), National
Electrical Code (NEC), and state and local fire, safety and health codes.

Estimating construction costs

The cost of designing and building a pilot plant can be estimated via similarity, ratios or detailed data on labor
and materials. Pilot plant cost estimating differs from that for full-size plants because little pilot-plant cost
information has been published; the personnel involved are, in general, inexperienced; and pilot plants are so
diverse in character.

Similarity involves estimating the cost of a pilot plant from the cost to design and construct a similar unit. This
method produces excellent results if the units are almost identical. Differences due to inflation can be
accounted for reasonably through a published cost index, such as the Chemical Engineering Plant Cost Index.
Unfortunately, few pilot plants are similar enough in all aspects to ensure an accurate estimate.

Differences in the type and size of equipment make similarity estimates very inaccurate. Unless the cost base
is large and the estimator experienced, these estimates are rarely more accurate than +/-50%, and frequently
can be less accurate. An estimate of such low accuracy is only useful for a preliminary screening.

In a cost-ratio estimate, the cost of an entire pilot plant (or a part of it) is related to a factor, such as the cost of
a major item of equipment or the number of control loops. Pilot-plant costs are infrequently estimated by
means of such ratios because ratio data for small equipment are not readily available. Therefore, most ratios
are developed by scaling down the known installed cost for the smallest item of equipment for a full-size plant.

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 5 of 11

This frequently results in a highly inaccurate estimate because the cost of equipment below a certain size, and
for its installation, often does not change much. The accuracies of cost estimates based on cost ratios range
from +/-25% to +/-50%, depending on the nature of the ratio and the experience of the estimator.

Estimating from detailed data involves dividing construction into a series of small tasks and estimating the
labor and materials required for each task. The task descriptions may be as encompassing as ``Install
compressor, with piping and wiring'' to as narrow as ``Wire up the compressor alarm.'' The narrower the
scope, the higher the accuracy of the estimate. These accuracies range from +/-10% to +/-25%.

Performing a detailed estimate takes a lot of time. The pilot-plant design and the detailed cost estimate are
sometimes carried out together. Such an estimate generally serves the framework for budgeting and
controlling costs.

Designing a pilot plant

A pilot plant is usually the first engineered version of a laboratory process. All the operating parameters may
not yet have been fully defined before the design of the pilot plant is started. Indeed, one of the primary
purposes of the pilot plant is to define them definitively. However, parameter ranges should be known. If the
operating parameters have not been satisfactorily established before the design is started, one of the following
options may be pursued:

—Return to the laboratory to define the operating parameters. This course is expensive and time consuming

—Construct a modular pilot plant and define the operating ranges for each module, modifying the design as
indicated. This, of course, is only possible if the unit operations are independent. This approach is often
difficult to implement

—Build a small and cheap pilot plant for identifying operating ranges, then proceed to designing and
constructing the proposed pilot plant on the basis of the findings. This route is risky. If the parameters have not
been adequately bracketed, the operation of the larger pilot plant may be delayed for process studies, which
could result in extensive modifications

—Design and build the pilot plant for wide latitudes in operating parameters but equip it with instruments
ranged for the best estimate of the parameter variations. Building the plant this way is costly but allows it to be
modified inexpensively

Pilot plant design typically follows one of two approaches: mimic the traditional method for commercial plants
or tailor the design specifically for the pilot plant. The first can be counted on to result in an operable pilot
plant. The second maximizes the advantages of the pilot plant.

For the first approach, the process steps and the information for designing the equipment are known. Usually,
the process has even been proven beforehand. Unfortunately, this approach is not always feasible. One
problem: The proposed pilot scale may be outside the range of the methods for commercial design, because
of the pilot plant's small size or differing operating conditions or process steps. Another problem: Some full-
scale operations are too expensive to reproduce on the pilot plant scale — e.g., some solids-handling and
complex recycle systems. Still another: The commercial plant may be tied into existing facilities, such as waste
-treating units and emission control equipment, that cannot be duplicated economically in the pilot plant.

For the second approach, conventional design techniques will be used and commercial-scale design
information will be drawn upon, but the equipment will be sized and operations will be conducted so as to
realize the advantages of the pilot plant, by using the smaller scale to bypass problem areas. Foremost among
the advantages are capital and operating savings from small equipment, but also important is flexibility.
However, there are some problems with this approach: Little information has been published on pilot-plant
design. This necessitates a common sense design approach, making the effectiveness of the pilot plant
dependent on the skill level and experience of the design engineer. If problems arise with the final design, the
research program can become side-tracked into a problem-solving project.

Who should design and build?

Except in the large organizations that regularly do a substantial amount of pilot-plant work, skill in designing
pilot plants is scarce. This is mainly due to the small number of pilot plants constructed each year. A company
doing its own design and construction eliminates information transfer out of the company, and reduces startup
time and costs, because the design engineers will be around for the startup of the pilot plant, and even of the
commercial plant.

The company that does not have engineers experienced in the design, construction and operation of pilot
plants must resort to a contractor. Among the drawbacks: Proprietary information would be disclosed to
outsiders, and the contractor may lack pilot plant expertise, despite claims.

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 6 of 11

With planning, the disadvantages can be minimized. First, develop a list of potential contractors, then check
their pilot-plant expertise, the experience of their engineers, and their financial stability. Next, determine the
criteria to be followed in analyzing the bids. This includes addressing such intangible things as working
relationships and professionalism.

Develop a detailed design specification, and supply it to all the contractors asked to bid. Review the design in
detail with operating and staff engineers who have experience in pilot plants and with outside consultants who
are not affiliated with the contractor, to identify possible problems and modify the design as necessary.

Insist on quality equipment, skilled construction mechanics and adequate field supervision. To ensure a quality
plant, set firm but realistic performance specifications. Define the performance requirements in detail. Avoid
vague specifications, such as ``meets all codes.'' The documentation from the contractor should include
copies of piping and electrical diagrams, drawings of equipment (such as vessels, reactors and columns),
equipment and material certifications, installation and operating manuals, and equipment purchase orders.

Designing for safety

Before the designing of the pilot plant has proceeded far, a conceptual safety review is held to identify
potential problems. It serves to ensure that all major safety concerns have been addressed.

Next follows a preappropriation safety review, the purpose of which is to ensure that sufficient funds for all
major safety equipment are included the budget appropriation. The intent here is to discourage the relaxing of
safety standards because an expensive item of safety equipment was overlooked when funding was sought.

A preconstruction safety review, which comes next, is conducted to confirm that all aspects of safety have
been properly addressed in the detailed design. Lastly, members of a pretartup safety review (generally
personnel who have had many years of operating expe rience) examine the finished plant, focusing on the
proposed operating procedures.

Specifying instrumentation

Defining instrumentation for a pilot plant can be difficult because pilot plants generally are not operated long
enough to permit evaluations of instrumentation based on performance comparisons solely in pilot plants
service. Also, the frequent isolation of pilot plants tends to encourage the development of each new one as a
unique entity.

Specification is critical. If too loose, the instrumentation's usefulness may be limited, requiring costly
modifications. If too tight, the instrumentation may be expensive to purchase and maintain. At the least, the
specifications should cover: the number and types of sensors (e.g., orifice, venturi, pitot tube), minimum
accuracy and repeatability, and plant environment (e.g., temperature extremes, exposure to corrosive fluids).

Signals transmitted to indicators and controllers should be as good as the signals put out by the field sensors.
The scanning frequency of a digital system should be fast enough to ensure that all data points can be read
and logged within the required interval. Expansion, operator access, background programs, data reduction
and other operations should not materially slow the scanning frequency. Operator access to information and
controls should be fast and reliable.

Evaluating the reliability of a proposed system's performance in actual operation is difficult because repair
data are rarely available. About the only options available are checking system performance with other
purchasers and by advance testing. Concerning reliability, consider whether the likely frequency of failures
and the cost of downtime can justify a more-expensive system, including one with redundancy.

If trained company personnel will not be available to maintain the instrumentation, service contracts should be
purchased. Although service contracts limit the investment in personnel and equipment, they can be
expensive and unsatisfactory. A contract should specify the response time, which is typically 48 to 72 hours.
Faster response times come at higher costs. An adequate number of service personnel should be local based.
The service company should supply all the replacement parts, which should be available on the same
schedule as the repair workers.

Despite the advantages of contract services, most companies find it worthwhile to have their own maintenance
personnel. Company workers are almost always more readily available. They usually give a more accurate,
unbiased evaluation of a problem. Simple problems are usually remedied more quickly and resolved. Because
most of the work involves replacing a component, little maintenance is done onsite; therefore, a trained
maintenance staff having a reasonably complete selection of spare parts is frequently more cost-effective than
contract service.

Deciding how much is needed

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 7 of 11

Rising labor costs are driving most pilot plants close to unattended operation for extended periods. Although
this reduces operator costs, it boosts instrumentation and maintenance costs. The few operators must be
provided fast access to data by ways of displays and alarms, which should be grouped in a central area. The
instruments must perform many monitoring, control and data-gathering functions previously handled by the
operators. The instrumentation must also be highly flexible and easily modified, so that it can be readily
altered for changing pilot plant requirements.

Most suitable for pilot plants are modular instruments, because these can be easily expanded, modified or
replaced, both completely or partly (i.e., exchanging components for quick maintenance). Low cost is also
important, so that instruments can be liberally installed. Reliability — i.e, a failure rate of less than once per
year — is also desirable. Cost ranges for some pilot-plant instrumentation are listed in Table 2.

Analog instruments have become smaller, less expensive and more flexible. This has made them increasingly
suitable for pilot plants. Recently, microprocessor- or computer-based control systems, hitherto too expensive
for anything but large process plants, have been reduced enough in size and cost to make them suitable for
many pilot plants. Choosing between the two has become difficult.

Analog instruments can be significantly less expensive to install, especially in small installations or where
control rooms and wiring are already present. Maintenance is simple, because such instruments are easily
replaced and sent to a repair shop. Not as complicated as computer-control systems, they can be maintained
by lower-skilled technicians, and are better at continuous monitoring and control, which tend to be important
for rapidly changing processes.

On the debit side, control strategies with analog instruments, being chiefly ``hardwired,'' are less easily
modified than computer systems. Maintenance is more burdensome with large systems. Data display is limited
and largely unorganized.

A computer-control system can be based on a single, large central unit or on a number of small individual
personal computers or microprocessor-based units. For a large installation, it is less expensive than a system
of analog instruments because less hardware and wiring is involved in an installation. It usually can be
expanded less expensively. Because of its flexibility, it is particularly suited to a pilot plant, which usually is
being continually changed. Algorithms can be tailored to handle special objectives, and the display of data can
be organized, relieving the operator of this responsibility.

Among the drawbacks of computer control are its complexity and requirements for specialized programming
and maintenance skills. If the operating area is electrically classified, standard cathode-ray-tube (CRT)
monitors may not be installed, and operators may be left without local instruments. However, many of these
disadvantages can be minimized through proper selection and installation. In general, the advantages of a
computer control system far outweigh the disadvantages.

Types of computer control

There are two types of computer-control systems for pilot plants. The first consists of a centralized computer
that controls all the pilot plants in an area. The second is based on personal computers (PC) or programmable
logic controllers (PLC).

Centralized systems, which are usually based on minicomputers or occasionally on mainframes, represent
true computer installations, with their large storage capacities, substantial system memories and numerous
associated equipment. Although they are becoming less common, centralized systems can be very
economical if they monitor and control numerous units. Oversizing a system initially allows the capability of
each unit to be expanded and applications to be extended to additional pilot plants at a nominal cost.

On the other hand, a failure of the central system shuts down all the pilot plants. Even routine maintenance
becomes difficult to schedule in a large facility consisting of many pilot plants. Installation costs for a
centralized system are usually very high, and environmental constraints are stringent. Annual operating costs
are frequently (although not significantly) higher (and more visible) than for a collection of ``stand alone''
computers.

Gaining acceptance is the ``stand alone'' computer system, based on PCs or PLCs, or a combination.
Expansion is economical and easy through the addition of computers. Individual computers separate the units,
allowing each to be operated and maintained independently. In addition, the initial cost is usually lower.

A ``stand alone'' system should have sufficient capability and memory to fit the largest need envisioned. Room
for expansion should be included. A typical computerized pilot plant grows by 20 to 50% per year during the
first two years, and by 10 to 30% thereafter. When a decision has been made, first select a control and data
acquisition package, and then the computer on which to

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 8 of 11

run it.

A pilot plant can be computer controlled in one of two ways: supervisory or direct digital control. In the first, a
computer supervises another primary control element. In the second, the interfacing element is eliminated and
the controller (e.g., a valve) is regulated directly. New computer-based automatic instrumentation is likely to be
supervisory because it is faster and less expensive.

The primary control element may be a local controller having special abilities for receiving a setpoint from the
computer, or from another computer or programmable logic controller. The supervisory computer typically
retrieves input data from a slave device, converts the data into engineering units (if neces sary), which it then
displays and stores. The supervisory computer passes to the operator all the necessary information and
forwards the operator's inputs (such as setpoint and tuning constants) to the slave element.

Supervisory control reduces the size of computer memory needed because it loads much work onto the slave
controller. It can sometimes be configured to operate while the main computer is down, although usually with
difficulty and reduced capabilities. It usually can be more easily interfaced with existing analog controllers, and
its software requirements are simpler than for direct digital control.

However, the innovative control strategies usually available are limited because the final control algorithm is
located in the slave, and not the host, computer. Also, the algorithm may not be accessible for modification.

Interfacing with other equipment can be complex, especially if the two systems are not compatible. Although
faster overall, the supervisory system may be slower to access and respond to a specific loop because it must
go through the slave controller.

In direct digital control, the computer directly accesses the process input, runs the control algorithm and
produces the output without a slave device. The control function is run as part of the main computer program.

Innovative strategies are possible with direct digital control because the control algorithm is directly
accessible. Not needing an interface, direct control has one fewer element to cause problems. It also allows
access to all phases of the control algorithm and the data-taking procedure.

On the negative side, direct digital control requires more computer memory, which increases the system's
size, complexity and cost. Also, the programming may be more complex, making the system more costly to
purchase, troubleshoot and maintain. In addition, local backup that is not computer based cannot be easily
provided. Although there are units that monitor the process and allow an operator to shift from computer to
local or manual control, these units are expensive for pilot plants and interfacing can become a problem.

Starting up operations

Starting up a pilot plant is different from starting up a commercial process plant in many ways. The smaller
scale alleviates some problems but creates others. A pilot-plant startup typically takes longer because of the
novelty of the process and limited resources.

As with any endeavor, the key to a successful startup is good planning and organization. Startup planning
should begin before the design of the plant has been completed, and the organizing before construction has
been finished.

The design should be checked to determine whether valves permit equipment likely to need a lot of
maintenance to be readily isolated. Likewise, instruments should be easily bypassed for calibration. It should
be possible to vent, purge, flush, pressure-test and leak-test tanks, columns, heat exchangers and pipes.

Begin the startup planning about six weeks before the end of construction by identifying possible problem
areas in time to make changes, and to order flushing solvents, lubricants and spare parts. Develop a detailed
startup sequence for the tasks to be performed. Also estimate the duration of each task and the resources
required for it.

When describing tasks, avoid broad statements, such as: ``Calibrate all sensors.'' Be specific, such as:
``Calibrate flow loop F07 using hexane from a temporary feed system, a pressurized drum. Route the hexane
to an empty drum on a 1,000-lb scale and gauge the actual mass flowrate.'' This will alert the startup team to
order the hexane, scale, drums and request the necessary piping.

The initial startup schedule should be tentative, based on the availability of resources, and it should include
contingency time. Most of the problems that will arise will not be anticipated. The initial schedule need only be
general (Table 3).

Pilot plant size can create operating problems. Such things as samples, purge streams and small leaks can
constitute a significant portion of flowrates, upsetting material balances or controls. Establishing steady state

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 9 of 11

conditions can be difficult because the reactor or other equipment is small. Such ingredients as catalysts,
additives and initiators are added at very low flowrates, making flow control a challenge. Such influences as
wall effects, thermal gradients and transient effects are frequently difficult to eliminate because the equipment
is small. Thermal losses may make adiabatic operation difficult or may create thermal gradients that adversely
affect the process.

Operating a pilot plant can also be difficult for other reasons. The shortness of runs and constantly changing
operating conditions prevent operators and mechanics from developing the high order of skills that comes
from long term experience with a particular plant. Cleaning, flushing or purging between runs can be costly
and time consuming.

Startup resources

Often overlooked are the resources needed for a successful pilot-plant

startup. First, there should be sufficient trained personnel. The people need not be assigned full-time to the
startup but should be available as needed. They should possess skills in instrumentation (metering and
analytical), electricity, piping, computer control, machinery, and process design and operation. Approvals for
the assistance of these people should be obtained.

Select a coordinator and endow the person with sufficient authority to direct the startup. Craft support (piping,
electrical, instrumentation, carpentry, etc.) should be placed under the direct control of the coordinator. A pilot-
plant startup is usually highly stressful. A crash startup can be grueling. This promotes mistakes. The time
allotted to the startup must be realistic.

The raw materials for a dummy run should be on hand. Arrangements for its storage, use and disposal should
be completed in advance. All flushing and cleaning solvents, purge gases and similar materials should be on
hand. Some specialized equipment is likely to be required. If this equipment is not available in the company, it
will have to be rented from an outside firm.

Because data are the chief product of a pilot plant, the instrumentation must be given special attention.
Provisions for instrument calibration must receive high priority. Valves must allow process streams to be
diverted to calibration instruments. Valves and piping should make possible the injection of standard mixtures
into instruments, and sight glasses should permit checking level instruments. Duplicate sensors may be
desirable for verifying critical measurements. All the testing and calibration equipment should be in working
order before the startup is begun.

Turnkey or not?

When a pilot plant is designed and built by an outside contractor, an issue to decide is: who should start it up?
With a turnkey contract, the contractor is responsible for the startup. In effect, the contractor is told to make
the unit work before being paid. This approach ensures that the contractor will devote the best resources to
the design, construction and startup. This seems a safe course for the customer, because unexpected startup
problems and costs due to the design or construction must be corrected, or payment to the contractor can be
delayed.

However, there are pitfalls in a turnkey arrangement. Defining the end of a startup may not be easy. A plant
can be said to ``operate at design conditions'' and yet not operate successfully. A ``specified grade of product''
may be made but at an unaffordable cost. All the mechanical and process systems can be ``proved to be
functioning at design conditions'' yet not do the job for which they were intended. Developing quantitative
measures that define a successful startup can be difficult.

On the other hand, a startup can be an educational experience. If the people who later will run the pilot plant
are not involved in its startup, the cost of their training will be increased and their troubleshooting skills for the
particular plant will not be developed.

No contractor will accept the large cost of a protracted startup. The contractor will seek to put the blame on the
client, dispute the existence of problems, or (in the extreme case) walk away from the project. Any of these
responses can end up with the client incurring substantial costs, even if only in legal fees.

Look after maintenance

Maintenance in a pilot plant often seems too insignificant to warrant organized attention. The seemingly
transitory character of pilot plants appears to suggest that they require little or no maintenance. The reality

is quite different. Few pilot plants operate for so short a time as to require no maintenance.

There are basically four types of maintenance: preventive, scheduled, emergency and corrective. Preventive
maintenance consists of planned repairs to prevent breakdowns. Scheduled maintenance covers general

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555
Chemical Engineering :: PILOTING PROCESSES INTO COMMERCIAL PLANTS Page 10 of 11

repair work covering deficiencies that develop with time, are recognized and planned for the future.
Emergency maintenance, as the name suggests, involves fixing sudden breakdowns. Corrective maintenance
pertains to revamping equipment to eliminate repetitive problems.

A good preventive program is one of the best ways to lower overall maintenance costs. Developing such a
program begins by reviewing all the equipment, reading the related manuals and data sheets, and finally
listing and scheduling the preventive measures for each item of equipment. Preventive maintenance is usually
performed as equipment analyzers (such as vibration monitors) indicate the necessity of it.

Scheduled maintenance involves work not necessarily related to specific equipment, such as flushing, testing
for leaks and replacing gaskets. As with preventive maintenance, scheduled maintenance reduces downtime
and costs because tasks can be grouped, and such activities as flushing, venting and draining are not
duplicated.

Emergency maintenance is, of course, the most costly type of maintenance. It does not allow planning, and
may shut down the plant. The necessary personnel, tools and spare parts must be assembled hurriedly, and
this can be costly. Too often, pilot plant maintenance is the emergency type.

Corrective maintenance is frequently neglected in pilot plants. Its purpose is to correct problems and reduce
long-term maintenance by replacing or modifying equipment that require frequent (scheduled or emergency)
maintenance. Its payback can be larger than the other types of maintenance because it eliminates problems,
not just fixes them.

Good maintenance lengthens the service life of equipment, reducing replacement costs. It lowers overall
maintenance costs, and operating downtime. Typical pilot plant downtime charges range from $200 to $5,000
per day.

Limiting maintenance

Proven techniques for minimizing pilot plant maintenance costs include standardizing components, storing
adequate spare parts, training operators and mechanics, replacing troublesome equipment, staffing the
maintenance organization properly and providing testing and troubleshooting equipment.

Standardizing on proven equipment reduces unexpected, costly breakdowns, training and troubleshooting
costs (operators and mechanics know the equipment), and spare parts costs. Having enough spare parts on
hand lessens the time that maintenance personnel must spend waiting for parts to be delivered, improves
troubleshooting (by making it easy to replace parts in order to pinpoint the source of a problem; this frequently
is the only practical way to troubleshoot microprocessor-based instruments and controllers).

Operator and mechanic training will improve their skills in operating, troubleshooting and repairing equipment.
Equipment needing replacement can be identified by keeping a maintenance log. When an item of equipment
breaks down, note in the log the date, problem, corrective action taken and result. The maintenance staff
should be capable of handling the wide variety of tasks that arise at a typical pilot plant.

An option: contracting

Contract maintenance is a viable alternative for many companies. It allows a small organization to
economically provide a vital service, covering peak workloads effectively and providing access to a broad
base of skills and experience. On the other hand, the bidding process takes time and effort and has to be
repeated at intervals. Moreover, find-

ing a qualified maintenance contractor can be difficult because pilot-plant maintenance is a speciality.
Changing a contractor means losing valuable experience.

The quality of the maintenance may be lower than that from a permanent staff because the contract personnel
will not be as caring. In addition,

the turnover in contract personnel can be high even if the same contractor

is retained; this, of course, lowers productivity.

Cost should not be the sole criterion in selecting a maintenance contractor. Also important is that the
contractor be able to provide skilled people for all the maintenance envisioned. Check a contractor's
references for opinions on past performance. A long-term contract is preferable because it minimizes the work
and expense of going through bidding the process again and increases the gains from experience. Before
signing a contract, have the contractor agree to keeping key personnel involved in the pilot plant for a
stipulated period.

http://www.che.com/articles/1985-1999/Vol97/chevol97_num3_74.html 23/2/2555

Potrebbero piacerti anche