Feature Report

Commercializing
Process Technologies
New technologies can provide competitive advantages compared to establ
processes, but significant effort is required to transition a promising concep
commercial reality
Vincent Welch

M
any students study chemical *
and Joseph engineering with the aspira­ .
Peters
TechnipFMC
tion of someday inventing
an innovative process that
r
could be a true game changer. As we
all know, creating a new process is quite />
IN BRIEF complex. Compared to the rapid pace
GETTING STARTED of advancements in other progressive
COMMERCIALIZATION
industries, such as consumer electron­
f r £
ics, telecommunications, software and 3
STEPS
automotive, in the world of high-volume,
CATALYST DEVELOPMENTlow-margin commodity chemicals,
1
V- ■ .
vC' iLi
PILOT VERSUS higher-value specialty chemicals and :■
newer sustainable chemicals, process / %
DEMONSTRATION

[i Ifry'^
development and commercialization are

fr^* m
REACTION SYSTEM
more methodical and time-consuming. i
DESIGN
This article is largely directed at il­ %
PATENT AND LITERATURE lustrating the development lifecycle for
SEARCH
KEY RELATIONSHIPS

DESIGNING AN
petroleum refining and petrochemi­
cal processes, rather than those in the
pharmaceutical, biochemical, mining or
inorganic chemical industries. The intent
EFFECTIVE PILOT PLANT is to describe how best to navigate the
FEEDSTOCK complex maze of process development,
ff
J
m
A I i5Jp"\
r\
CONSIDERATIONS starting from basic research and devel­ 3 LAB Fl'Mp
opment (R&D) to technical readiness
SIMULATION AND
MODELING
and commercialization (Figure 1).
*
frn i
CAPITAL COST Getting started FIGURE 1. Demonstrating a promising new technology in a laboratory
ESTIMATES One of the most difficult challenges setting is just one of many complex steps in commercializing a new
process technology
SCALEUP AND facing the chemical process industries
DEMONSTRATION (CPI) today is the cost and timeline to com­ mitigation, but also for maintaining manage­
mercialize new, innovative technologies. The ment and investor confidence. Because each
REACTOR SCALEUP
conventional textbook approach of multiple subsequent step typically requires significantly
SCALING MAJOR incremental steps of tenfold scaleup (typically larger capital and operating cost outlays, the
EQUIPMENT from bench scale to pilot plant to demonstra­staged-gate approach compels the develop­
PROCUREMENT AND tion unit and finally to commercial plant) canment team to pause and take an objective
CONSTRUCTION take as long as a decade and be prohibitively view of test results, process design issues
expensive. Using a staged-gate approach —and economics.
DEMONSTRATION UNIT
where process viability and economics are It is clear that the revamp or building of
STARTUP
assessed at each step during the develop­ any chemical plant involves significant in­
ment program — is crucial not only for risk vestment and market risk. For producers in
40 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
 mature markets, the risk of invest­ development program depends on demonstration unit in the range of
ing in a large-scale demonstration your perspective. Existing plant own­$10-50 million and a full commercial
or full-scale commercial unit for an ers must protect their market share. plant could easily be in the hundreds
“unproven" process technology is They should be constantly evaluat­ of millions of dollars. These costs
nearly impossible to justify without ing competitive technologies to de­ must be taken into consideration
a thorough evaluation, including termine when and if they must pur­ before undertaking any process-
testing at bench and pilot scales. chase or develop new technologies development program.
As many readers can attest, few, if to replace those already in place. Deciding on the need for the in­
any, changes to a chemical plant or Engineering and licensing companies termediate demonstration unit step
process are simple and turn out ex­ look for advantaged technologies can be one of the toughest deci­
actly as calculated at the start of the that can be competitive in the mar­ sions. With an outline of the known
development program. ketplace. Their interest is in licensingfacts and assumptions, researchers
Even with justifiable economics and these technologies and deriving in­ and engineers will have lively debates
appropriate incentives from the tech­come from license fees, engineering on the need for this time-consuming
nology developer or owner, most CPIservices, proprietary equipment and and expensive step. Avoiding the
companies are reluctant to be the catalyst sales. demonstration-unit step could reduce
first, or even the second, adopter of the time to commercialization by two
a new process or catalyst system. Commercialization steps or three years, albeit with a poten­
Most building-block petrochemicals Each process-development programtially higher risk profile. In the end,
and high-value specialty chemicals requires several project phases, fromthe final decision may come down
produced today are very mature tech­ the initial proof-of-concept through to risk tolerance, identifying ways to
nologies with well-known competitivenumerous levels of technical readi­ mitigate risk and adding design safety
economics. Being an early adopter ness, stage gates and eventually to margins or perhaps even a “gut feel”
can actually have a significant upsidethe construction of the commercial based on experience. For example,
because projects can take years to plant. These reassessment require­ the development team could deter­
realize. For companies to significantly ments vary greatly from company to mine the consequences of lower-
improve their competitive position, company. As one can imagine, the than-expected performance. What
they must move from the standard technical readiness requirements forif the selectivity improvement is only
tried-and-true processes to more a major petroleum or chemical com­ 95% of the expectation? What if the
forward-looking innovative technolo­ pany are likely different from those ofcatalyst life is two months shorter or
gies. Those who choose to “wait anda technology licensor or engineering only partially regenerable? Once the
see” run the chance of operating withcompany. Even within the same orga­ most likely performance gaps have
a sustained disadvantage once the nization, depending on starting point been identified, the team should de­
new technology is commercialized. and complexity, each technology-de­velop a strategy and identify the steps
However, on the downside, the risk velopment program has its own pathneeded to account for or remediate
to the early adopters is failure of the and strategy to get from the initial the shortfall.
new technology to meet expected concept to full-blown commercializa­ Simplistically, the number of pro-
performance, thus jeopardizing the tion. Additionally, risk mitigation and cess-development steps is a function
entire investment. The chances individual risk tolerances set and drive of reaction chemistry complexity, the
for success improve with a deeply the scaleup progression and develop­ non-ideality of the components in­
thought-out and rigorously executed ment lifecycle. The textbook rules of volved and the scalability of the unit
development plan. process scaleup include steps goingoperations. While the sequence of
Process improvements can take from a very small-scale testing appa­process-development steps is out­
many forms that can save operat­ ratus (bench scale) to a more repre­ lined here, deciding which steps are
ing cost or capital expenditures, as sentative pilot plant, then possibly to mandatory should be determined by
well as improve product purity and a demonstration unit and finally to the an experienced team of research­
properties. These may include highercommercial plant (Figure 2). It shoulders and engineers in cooperation
catalyst selectivity and conversion, be highlighted that scaleup is not with business leaders and plant
improved reactor productivity, lower only limited to the key process-unit operations managers.
catalyst or feedstock costs, reduced operations, but also to the catalysts,
environmental impact and more. True feedstocks, additives and equipmentCatalyst development
stepouts may embody completely that form the complete basis for the For many programs, the earliest proj­
new chemistry routes, alternative new process. ect phase involves the development
feedstocks and reactor designs, im­ Defining the required number of of an alternate chemistry path or an
proved catalyst formulations or en­ steps and their scale greatly influ­ improved catalyst. Nowadays, it is
hanced manufacturing techniques, ences program cost and timeline. common practice that these improved
such as improving energy recovery or From a capital expenditure stand­ pathways and catalysts are devel­
plant reliability. point, an automated pilot plant may oped using high-throughput testing
The decision to start a technology- cost $1-3 million, an integrated methods that study multiple catalyst
41
CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
 to produce a sufficient quantity of
material. This activity can be the most
challenging step because the com­
i mercial catalyst must not only have
• ■ • •» ft similar activity and selectivity to the
i.Y ■ sample, but also the physical proper­
f/' .1
' ties, such as crush strength, required
% m KM to operate at commercial scale.

r
■»
Pilot versus demonstration
\ i * The terms "pilot plant” and “dem­
onstration unit” are sometimes used

r?5
//
V I
interchangeably. For the sake of this
article, pilot plants are configured to
facilitate fundamental learning, deter­
mine rate-controlling steps and col­
lect large amounts of data at a wide
>1

V range of conditions to ultimately sup­

port the design basis for the commer­
FIGURE 2. Bench-scale tests set the foundation for scaling a new technology up to a more cial
tive pilot plant
plant. Within practical limits, the
representata-
pilot plant should attempt to simulate
samples in parallel, significantly short­ and process systems to determine if the envisioned commercial plant and
ening the time required to evaluate there is a potential significant advan­be flexible enough to accommodate
candidate catalysts or investigate tage. This involves developing a newprojected operating scenarios and
reaction pathways. High-throughput flowsheet, including the reactor de­ numerous experimental runs. Pilot
testing is a powerful tool for identify­ sign, and understanding the existing plants are highly instrumented to en­
ing trends and variables that may lead technology. Technical and economicsure a complete understanding of the
to fully optimized catalyst formations.analyses of the new and existing pro­operating conditions and the impact
During a typical screening project, cesses result in a decision to proceed of the equipment on the process
testing a few dozen or even several to bench-scale testing (Figure 3). being studied. A pilot-plant reactor
hundred materials and conditions is Bench-scale experiments utilizing may have dozens of thermocouples
not unique. the most promising catalyst samplesand numerous sample ports to sup­
As is typical in the early stages of are usually performed on a very small port the understanding of kinetics,
process development, high-through- or microscale and could involve glass­ reaction mechanisms and catalyst
put testing protocols are performed ware, small stirred-tank reactors, mi­performance and lifetime. A rigorous
on a very small scale under isother­ croreactors or other systems that usepilot development program will care­
mal conditions with crushed catalyst.a minimum of amount catalyst and fully examine issues well beyond the
Moderately exothermic reactions or feedstock to obtain proof-of-concept basic process chemistry and kinetics.
high-activity catalysts typically utilize data.
a The evaluation includes a de­ Several of these considerations are
catalyst bed diluted with inert material. termination of not only catalyst activ­listed in Table 1.
Not surprisingly, subsequent catalystity and selectivity, but also gathering Conversely, the demonstration unit
testing utilizing a particle size more preliminary information on catalyst life should be at a scale such that equip­
representative of the commercial or regenerability. The experimental re­ ment size has little or no impact on
catalyst (for example, pellet, cylindersults are also compared to those ob­the results compared to the envi­
or sphere) operating under simulatedtained at the earliest phase. sioned commercial unit. For reactor
commercial-scale conditions may If the results continue to look prom­designs, the configuration should be
produce considerably different results. ising, the next step involves produc­ virtually identical to that expected to
While single stage or bed testing is ac­ ing the catalyst at a large enough ca­be used commercially. Overall, the
ceptable at this point in the program, pacity such that it can be considereddemonstration unit provides a valida­
managing heat transfer requires care­ representative of commercial-scale tion of the commercial process but at
ful consideration before moving for­ production. This typically involves a smaller, less risky scale. It should
ward. Options may include multibed working with a catalyst supplier to also have some limited built-in flex­
reactors, pseudo-isothermal tubular develop the commercial production ibility and be designed to operate
reactors, fluidized-bed reactors, in­ techniques required to economically continuously for long periods of time.
terstage quenching, interstage heat produce a catalyst with performanceCompared to pilot plants, demon­
transfer or interstage feed injection. parameters close to or better than stration units typically will have far
The most promising results from those of the laboratory-scale material. fewer instruments and connections,
these early tests are compared to theIt is common that the catalyst vendorwhich are mainly used for control
performance of the existing catalyst will perform an abbreviated plant trialand safety purposes. Demonstra-
42 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
tion units should also be a source of
valuable information on operational
s I*'*
••_I

issues (such as startup, shutdown "W/i

and controllability) and maintenance
issues (such as fouling, cleaning in­ I
tervals and equipment reliability). h
For reaction systems, kinetics, fluid
N- t: k
dynamics, flow regimes and the re­
actor type play a central role in de­
termining the appropriate number
of scaleup steps. Fixed- and trickle-
bed systems, for both liquid or vapor
phase, can typically be scaled up
m»>
it t
A m
from pilot-plant data. Fluidized-bed
systems (Figure 4) require larger
demonstration-scale testing where
i| } u ^3

reactor size does not constrain the

hydrodynamics of the catalyst sys­
tem. Scaleup of continuous stirred-
tank reactors (CSTRs) is done in suc­ FIGURE 3. Early tests and bench-scale experiments can help to narrow a project's focus to the most prom­
cessively larger vessel sizes because ising catalyst candidates
mixing, heat transfer, active reaction As noted, not all development proj­thinkable not to pilot a new process.
zone and gas-fluid contacting are ects require all four scaleup steps. Surprises and revelations can include
very much dependent on equipment For something relatively simple, suchunexpected catalyst performance,
size and geometry. as an incremental change to a poly­ feedstock poisoning, safety issues,
Notwithstanding, the scalability of mer formulation, or if past labora­ equipment corrosion, fouling, foam­
supporting unit operations also con­ tory results have been validated by ing and plugging. Most chemical-re-
tributes to the thought process. If previous experiments and practical action pathways are complex — their
vapor-liquid equilibrium (VLE) behav­commercial experience, it may be true mechanism may not be fully un­
ior is well known, piloting of distilla­ possible to go directly from bench to derstood and is likely considerably
tion columns is usually not required commercial scale. different from the simplified overall
for determining separation parame­ Similarly, for some purely thermal- stoichiometric equations. The combi­
ters. However, separation of recycle driven processes, such as ethylene nation of a properly constructed and
streams and final product are key el­production, where developments areinstrumented pilot plant, along with
ements of an integrated pilot systemtypically aimed at larger capacities, a well thought-out experimental pro­
and are critical to the ability to scale increased yield, improved heat trans­gram, can efficiently and effectively
up directly to commercial size. The fer, shorter residence time and longersort out the myriad issues that need
impact of recycle on catalyst perfor­ cycle times, the textbook approach to be addressed for scaleup.
mance and life, potential byproduct may not apply. In the steam-cracking
purge and product quality all fac­ process, where plant capacities can Reaction system design
tor into the evaluation of ultimate regularly exceed 1,000,000 metric By and large, the reactor is the core
plant performance. tons per year, even very small in­ of most chemical plants, and its
If vapor-pressure and enthalpy data cremental improvements can make feeds, products, solvents, catalysts
are readily available, heat exchang­ a huge impact on a producer’s net and diluents dictate the downstream
ers will likely not require piloting. margin. Here, commercial implemen­processing requirements. Reaction-
Conversely, unit operations, such astation of advancements may not re­ system fluid dynamics (for example,
extraction, drying and crystallization,quire a laboratory or pilot program. heterogeneous versus homoge­
typically require more scaleup steps Sometimes, technology advance­ neous or single-phase versus multi­
and may involve working directly ments can be accomplished by usingphase) and the reactor type play a
with equipment suppliers. Processescommercially available kinetic-yield central role in determining the appro­
with significant solids-handling re­ model software, along with perform­ priate number of scaleup steps.
quirements, two-phase flow or non- ing engineering calculations with ad­ For example, in some cases, the
Newtonian fluids, such as foams or vanced tools, such as computationalscaleup of a homogeneous CSTR
sludge, will also require more steps fluid dynamic (CFD) simulation. system with simple chemistry can
prior to full-scale commercialization. For the most part, however, be­ be straightforward, allowing the de­
Similarly, processes that deal with cause piloting nearly always exposesvelopment to go directly from pilot
powders or non-free-flowing solids the unpredictable or unknown, and to commercial scale. Assuming heat
can be particularly challenging and given the enormous financial risks attransfer is not a controlling issue and
difficult to scale up. the commercial scale, it is almost un­mixing characteristics do not signifi­

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019 es

 TABLE 1. SELECTED PILOT-PLANT CONSIDERATIONS This usually begins by completing a
Vapor-liquid equilibrium (VIE) survey of the current market condi­
Separations Liquid-liquid equilibrium (LLE) tions and evaluating existing commer­
Vapor-liquid-liquid equilibrium (VLLE)
cially available technologies. Pertinent
Runaways
Process safety
Flammability and toxicity background information and data
Type from trusted sources and dedicated
Corrosion Mechanism industry reports are helpful in build­
Estimated rate ing the business case to move for­
Heat input
Temperature control Heat removal
ward. Knowledge on several factors
Reaction quenching can help support the decision to fund
Replacement or abandon the program, including
Catalyst deactivation Regeneration the following:
Installing pretreatment • Market size
Mixing
Pressure drop • Forecasted growth
Hydrodynamics • Competition and any emerging
Phase separation
Flow distribution technologies
• Estimated production costs
cantly affect the kinetic rate, scaleuptor. While this approach avoids some• Worldwide production facilities and
could be accomplished by using the of the costs associated with a large- announced projects
same operating conditions as the scale demonstration, it may present This analysis can answer key ques­
pilot plant, maintaining similar reac­ less than ideal conditions for validat­tions: Does the new technology sig­
tor geometry, liquid hold-up volume ing the new technology. Examples nificantly reduce variable operating
and residence time. Conversely, a of limitations could be restrictions cost? Is the current market overbuilt
CSTR with multiple phases, com­ on equipment location or plot area, with low plant-capacity utilization?
plex chemistry, a high heat-transfer the inability to independently con­ What are the price trends and pre­
load, solids or catalyst addition or trol pressure or flows, difficulties ferred feedstocks? Armed with re­
solids removal would probably re­ with equipment during startup and cent and credible data, this detailed
quire additional validation beyond shutdown or the inability to maintain justification study provides the sup­
the normal pilot scale. Similarly, a steady operations because of mov­ port and confidence to proceed with
bio-based CSTR with live organisms,ing production-rate targets. Lastly, what will most likely be a multimillion-
enzymes or yeast certainly requires because the highest priority of plant dollar investment.
multiple steps. operating personnel will always be Before embarking on any new pro­
To reduce the overall develop­ the main production unit, the experi­ gram, a careful assessment of the pat­
ment timeline, a different tactic may mental program and the demonstra­ ent landscape should be performed to
be to incorporate the innovation into tion unit may not get the attention ensure that the new idea is not infring­
an existing operating plant possibly needed to guarantee success. ing on the rights of others. If infringe­
as part of a revamp, debottleneck­ Unquestionably, irrespective of ment is suspected, the program can
ing or modernization project. Or al­ the number of development steps, be abandoned, or ways to circumvent
ternatively, a short, controlled “plant before operating companies fully the claim(s) need to be considered.
trial" could be attempted. For some commit to commercializing improve­ This initial examination will ultimately
researchers, a plant trial with a new ments, they must be convinced that serve as the foundation for a compre­
catalyst is regarded as a simple this new investment provides sig­ hensive “Freedom to Operate" (FTO),
“drop in.” Conversely, for plant and nificantly enhanced economics, the which is the legal opinion on whether
operations managers, just the mere technology has longterm viability andthe process, catalyst, equipment or
thought of this “experiment" can sustainability and the process devel­their combination may be infringing on
them give nightmares, as any un­ opment was rigorous, covering all patents held by others. A careful re­
expected loss in performance can aspects of the design and operation.view of past work can also sometimes
have an enormous impact on plant point the current program to more
economics and profitability. Never­ Patent and literature search productive areas of research. If an
theless, using existing hardware andAfter an initial idea is formed, two idea or concept is completely new to
infrastructure assets may allow the critical tasks need to be completed: the field, it is important to decide early
development to go straight from the a project justification study and a on whether to file for patent protec­
pilot-plant stage to commercializationpatent or literature search of previous tion or keep it as a closely held trade
without the expense of a full-blown work in the field prior to embarking secret. Keep in mind that patenting a
demonstration unit. on any development program. novel idea or process that may take
One approach for commercializing Even before proof-of-concept has 10 years to commercialize is a double-
new reaction systems is to incorpo­ been fully cemented, the R&D pro­ edged sword — while it does grant
rate an incremental slip-stream reac­gram needs to have economic jus­ you the right of unencumbered use, it
tor in parallel with the existing reac­ tification to determine its viability. also alerts competitors to your plans.
44 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
Key relationships ment. Eventually, more thorough eval­
Once the initial small-scale catalyst uations are done on the final form of
testing and economic analysis have the commercial catalyst to fine-tune Is
yielded promising results, the leadingthe performance envelope and ulti­ if
catalyst formulation is identified and mate catalyst life. Overall, developing
k
there were no red flags raised by thea commercial version of a laboratory-
patent search, it is now time to estab­ prepared catalyst is no simple matter,
v»
lish two key relationships — one withand in some cases can control and
an engineering company and one significantly extend the entire de­
\ .
with a catalyst vendor. velopment effort. This phase of the
L
It is important to employ an expe­ project is by far the most difficult to
rienced team of researchers and en­predict and can be very frustrating for
gineers that is capable of executing a development team trying to meet
vf
?
multiple stages of the development schedule and budget projections.
project. It is also essential to select an
Designing an effective pilot plant
engineering firm that has pilot facilities
of the appropriate scale, the neces­ Automated pilot plants (Figure 5) truly
sary analytical capabilities and experi­ provide the foundation and are the
ence in process scaleup. A compe­ workhorse of all chemical process de­
tent engineering contractor should velopment. The benefit of operating a
also provide guidance on the scope small-scale pilot plant is that it allows
and requirements of the develop­ numerous operating conditions and FIGURE 4. A fluidized-bed reactor requires testing
at larger scales than other reactor types, due to
ment program by outlining the matrixvariables to be explored efficiently the unique hydrodynamics of the system
of experiments and studies needed and economically on a relatively short
to create the design basis for the time scale. The primary function of ing the unit too small may introduce
commercial plant. the pilot plant is to demonstrate the scaleup uncertainties for subsequent
The second important relationship viability of the new process and gen­phases of the program. In small reac­
to establish is with a catalyst supplier.erate the required data to support thetors and equipment, phenomena like
Process development cannot con­ plant design. wall effects, inlet thermal-void reac­
tinue unless the catalyst formulation Determining the appropriate size oftions, bed bypassing, plugging, heat
can be reliably manufactured on a a pilot plant is not always a straight­ transfer and heat losses, are not only
commercial scale. The goal is to se­ forward task. Beyond generating more difficult to control, but may not
lect a supplier that has the capabili­ technical data for scaleup, one otherbe obvious at first and can be difficult
ties to generate catalyst samples at afunction of the pilot plant may be to to quantify.
scale that closely resembles the ulti­generate quantities of product sam­ The translation of pilot-scale data
mate commercial version. Typically, ples for analysis and market testing. for direct scaleup is done based on
this requires engaging the expertise Another key decision is construction proven methodologies validated
of an established catalyst manufac­ philosophy. Should the unit be skid- through the collection and compari­
turer, preferably one with experiencemounted, modular-design or stick- son of data collected on both the pilot
that is at least tangentially related to built? Should the unit be placed in a and commercial scale. These meth­
your invention. Catalyst vendors alsodedicated enclosure? Will it eventually odologies have been proven at pro­
provide valuable insight into criti­ be moved to the plant site? Selectinggressively smaller scales for numer­
cal matters, such as precious-metal between these options depends on ous reaction and separation systems.
loadings, supports, binder types andthe complexity of the chemistry and Fixed- and trickle-bed vapor- and
target physical properties. process, as well as project schedule,liquid-phase reactions can typically
Translating the optimal catalyst for­budget, in-house expertise and the be done in single-tube reactors as
mulation, which was tested on a veryprojected length of the program. small as %-inch diameter. However,
small scale, directly into a commercial The design of the pilot plant is this is very much dependent on cata­
catalyst can often be problematic. In based on the conceptual design of lyst particle size, reaction energetics
most cases, multiple trials are required the process, which includes a pre­ and expected commercial thermal
before a commercially viable catalystliminary reactor design and required profile. Careful sizing and loading of
is produced. Samples from each trialseparations. The key factor in the the reactor are required to prevent
must be evaluated at the pilot scale design is how small the equipment bridging and uneven flow distribution.
to determine activity, selectivity, re- can be to provide accurate and re­ Inert materials are sometimes used to
generability and catalyst life. Of equal producible data for direct scaleup dilute the catalyst bed for improved
importance is data reproducibility. For to a commercial plant. Obviously, a flow distribution and thermal control.
the same catalyst batch, experimen­smaller unit using smaller flows will re­Handling the waste and effluent
tal results performed under similar quire less capital, use less feedstocks streams is another factor to consider
conditions should be in close agree­ and generate less waste. But mak- in determining the size or capacity of

CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019 45

 \ i* m the tools available through automa­ mimic the commercial plant as much
tion are quite valuable, given the closeas possible. The key element to this
operating margins of today’s approach is ensuring that the reaction
markets.
i*1
n Another potential difference be­ system is designed such that the cat­
, iS\ • 1
*
J tween the pilot and demonstration alyst bed operates as close as pos­
unit could be the materials of con­ sible to expected commercial con­
struction. For the pilot plant, it may ditions. This includes the expected
7ST -f be expeditious or only a small addedlinear and radial temperature profiles,
cost to use a high-alloy material for operating pressures, space velocity
. -j i
the reactor and its associated tub­ and number of stages. Linear veloc­
ing. In the end, however, researchersity can be difficult to match, given
need to ensure that metallurgy will not potential limits on reactor length.
V• influence the chemistry by catalyzingThus, to support scaleup, paramet­
coking or other undesirable reactions. ric studies should be done at varying
I
Materials selection for the linear velocities to determine whether
demonstra­
tion and commercial plant should be bulk mass transfer is an issue at the
t based on known reaction and corro­ lower velocities of the pilot scale.
I
r Additionally, startup and shutdown
sion-rate data available from literature
i or data generated in the laboratory. procedures should be synchronized
m
FIGURE 5. An advanced, automated pilot plant,
Where possible, critical pilot-plant with those of the future commer­
systems are made of the same ma­ cial plant and recommended by the
such as this multibed unit, provides opportunityterials
to of construction as those pro­ catalyst supplier.
efficiently study the process at various operating
conditions
jected for commercial scale. For ex­ Highly exothermic and endothermic
ample, the pilot-scale reactor itself reactions, which have large extremes
the unit. Feedstock and disposal of acts as the “corrosion coupon” since between bed inlet and outlet condi­
toxic waste can add considerable cost it was exposed to the fluids, tempera­tions, can also produce noticeably
to a pilot program. Reactor effluents tures and pressures of the actual re­ different results compared to nearly
that can be condensed are easier to action system. Metallographic analy­isothermal operation. Per-pass con­
handle, store and discard. Conversely, sis of the reactor after removal from version, yield and selectivity can be
effluents containing a high percent­ service provides data crucial to the greatly impacted by all of the above.
age of light gases, such as hydrogen,selection of materials of constructionConsequently, while a prime catalyst
ethylene, nitrogen and propane, maygoing forward. candidate can certainly be identified
require onsite oxidation or incinera­ Finding equipment capable of ef­ using small-scale, high-through­
tion. Permitting and code issues can ficiently handling flows in the grams- put testing, alternatives should be
place restrictions on the total quantityper-hour range is certainly possible, available when unforeseen results
of hazardous materials that can be but the number of options can be are encountered.
stored and used in a facility. limited. While off-the-shelf equipment The goals and schedule of the pilot
Pilot-plant size and complexity alsoitems are certainly available, much ofplant program should be established
have a direct bearing on the ability the equipment used at the pilot scaleby the researchers working collabor-
to automate the pilot plant for unat­ is custom-fabricated and tailored to atively with the engineering team and
tended operation. Staffing a pilot plant each unique application. catalyst supplier. The team should
24/7 is usually prohibitively expensive. Frequently, the main emphasis identify the critical data sets required
Most catalyst systems require exten­of a pilot program is to validate the to support the commercial design
sive parametric and life testing, which chemistry fundamentals and cata­ and the potential demonstration unit.
can require months, if not years, of lyst performance that are usually not In addition to the research program
run time. Automation is essential to available in literature or predictable schedule, the team must also develop
bring the cost of operation to a rea­ via calculation. Understanding reac­ timelines for pilot-plant construction
sonable level, allowing for the most tion kinetics and byproduct formationand commissioning. Pilot-plant con­
part staffing a day shift only. Experi­ mechanisms help determine which struction can take months. The unit
ence has shown that the payback forreactor type is best suited for a givenmust be designed and the process
automating a typical pilot plant is lesschemistry. Establishing the approachequipment, controls and analytical
than a year of operation. Automationto theoretical thermodynamic equilib­apparatus procured. During the pe­
also improves data quality because rium governs the maximum attainable riod leading up to construction, the
measurements are taken at regular, conversion. While heats of reaction fix required permits should be obtained,
defined intervals so that trends previ­the energetics of the reaction system,control logic developed, a hazard and
ously difficult to see become evident.they also greatly influence reactor operability (HAZOP) study and layer-
In the past, an operator with a clip­ type and design. of-protection analysis (LOPA) per­
board taking hourly readings and then Beyond bench-scale testing, the formed and the feedstocks obtained
graphing the results was effective, but pilot-plant design should attempt to or synthesized.
46 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
Feedstock considerations
It is always best to perform pilot test­
ing with industrially produced feed­
stocks. However, care must be taken It y
to pre-purify these materials before
use in the pilot plant, as the very act
of drumming or packaging these ma­V
terials may introduce air, moisture
and heavy materials left over from
drum manufacture. Scrupulous pre­
1T -JE81 i

0CMS#1
r] 7
cleaning of the containers and purg­
ing of the sampling lines is critical to
minimize feedstock contamination.
Feedstocks should also include
representative recycle streams. Gen­ \
eration of recycle streams normally
requires that product-separation
steps be included in the program. For \ J
pilot plants with relatively small flow- FIGURE 6. Access to advanced analytical capabilities is essential in determining feedstock integrity and
monitoring plant performance
rates, separations, such as distillation
or filtration, can be accomplished in from process modeling software, so enough capacity to keep up with the
batch mode, but care must be taken that the recycle streams and product 24-hour operation of the catalysis
to ensure that the level of separationcompositions represent commercial plant. Repeating this over many itera-
would be feasible at the commercial operation. These separations can tions results in recycle streams close
scale. Better yet would be to performbe done in a semi-continuous mode, to those that would be seen if the
the separations on a continuous such that the columns run continu- distillation system was directly con-
basis according to the projections ously during the day shift at a high nected to the reaction system.

For Your Next Clean Air Project

...Go Modular
Get the cost, sourcing and Labor saving benefits r
■Vk
our ScrubPac™ Fully Engineered, "Packaged
u
ft
Emission Control Systems can deliver
JEto
Available ScrubPac” System applications include: •f

• Gas and Fume Scrubbing r

• Particulate Removal ? .ns

V. f
• Aerosol, Mist and N0X Removal Wj

• Odor Control
• Ammonia and Acid Reconcentration
• Chemical By-Product and Waste Heat Recovery
m
With over 40 Years of application expertise...
If it can be scrubbed, we'll scrub it!
a

bBIONOMIC Superior Engineered Systems with Unequalled Performance

Industries Inc.
201-529-1094 • mktg@bionomicind.com • www.bionomicind.com
777 Corporate Drive, Mahwah, NJ 07430

For details visit adlinks.chemengonline.com/73849-03

CHEMICAL ENGINEERING VWWV.CHEMENGONLINE.COM JANUARY 2019 47

 till
flashes and so on), and heat addition
■ 1 or removal, as well as all recycle, by­
product, vent and purge streams. At
the most basic level, model develop­
ment requires a detailed definition of
r the components or pseudo-compo­
I TJJ nents, including low-level impurities
that could affect separations and
M final product purity. A key undertak­
i ■>/ ' ing for model development is de­
l 4
termining the best methods for es­
timating thermodynamic properties,
equilibrium data and other physi­
cal properties. For many chemical
v \ compounds, the built-in data banks
and estimation methods available
in commercially available process­
* modeling software may be a good
*%■?
4^1 a starting point. For other lesser-
known and
FIGURE 7. Process simulation, especially CFD, is an important step in estimating process behavior components, data may have
identifying potential problematic anomalies to be generated in the laboratory or
If this is not possible or practical, plant performance. Offline analysis determined via literature sources.
synthetic or reagent-grade feed­ typically provides more detailed and Likewise, if not available from litera­
stocks may be used with certain accurate composition data, particu­ ture sources, parameters, such as
caveats. Reagent-grade feedstocks larly at extremely low levels. Relation­ the solubility of water or a solvent
with extremely high purities, read­ ships with local universities and third-in the process streams, may also
ily available from many sources, are party analytical providers can also behave to be determined empirically in
convenient but may produce overly helpful when specialized analyses are the laboratory.
optimistic results. Typically, these required, such as nuclear magnetic Reactor modeling can range from
materials will contain an entirely dif­ resonance (NMR). For most mod­ straightforward empirical equa­
ferent slate of impurities or poisons erately sized companies, the use of tions to a full-blown kinetic model
that may or may not affect catalyst third parties for specialized analyticalwith rate constants and equilibrium-
performance and aging. Chemical work saves both on the initial invest­ based interactions. Constructing a
suppliers offer a wide range of materi­ ment and longterm upkeep needed robust kinetic model with frequency
als and use the same handling equip­for highly sophisticated equipment. factors and activation energies that
ment for a variety of substances. As The importance of having accurateaccounts for temperature, pressure
much as they try to clean equipment analytical methods cannot be over­ and concentration effects is a large
between uses, trace amounts of dif­ stressed. New processes typically effort and can take months to create.
ferent substances may carry over and produce a different mix of byprod- As a result, the process developer
can contaminate the feedstock. The ucts and minor impurities. Develop- needs to decide what level of detail
chemical supplier is only required to ing reproducible analytical methods is needed for commercialization. A
meet their certificate of analysis and, to identify the minor trace compo- true kinetic model is the ideal tool for
for the most part, is not worried about nents is crucial for not only determin- optimization but could be redundant
trace impurities. Often, these impuri- ing stoichiometry and kinetics, but or unnecessary for plant scaleup.
ties can be below the detection limits also for performing material and car­ As more pilot data (kinetics and VLE)
ol a supplier's analytical equipment. bon balances and maintaining adia- become available, the model should
Having access to advanced analyt­baticity and temperature profiles. be used to optimize the flowsheet.
ical capacities, such as gas chroma­ However, one of the more important
tography (GC) methods like thermal Simulation and modeling uses of the simulation model is to de­
conductivity detectors (TCD), flame While the researchers are typically fo­ termine variable operating costs and
ionization detectors (FID) or nitrogen-cused on reaction kinetics and cata­ perform an economic assessment.
phosphorous detectors (NCD), as lyst life, so that the economics of theRaw material and utility consumption
well as GC mass spectrometry, are integrated plant can be determined figures gleaned from the simulation,
key to validating the catalyst perfor­ and the commercial plant designed along with other variable operating
mance data, identifying the effect in the future, the flowsheet for the costs, such as catalyst and chemical
of trace compounds and ultimately entire plant should be defined and a usage, should provide a representa­
providing the team with confidence simulation model created. The simu-tive benchmark when comparing the
in the results (Figure 6). Online GCs lation should include all relevant sep-competitiveness of the new technol­
and process analyzers can provide arations (distillation columns, strip­ ogy versus the incumbent process or
realtime data for close monitoring of pers, dryers, absorbers, decanters, other alternatives.
48 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
Capital cost estimates Scaleup and demonstration Reactor scaleup
With the information available at the Following successful piloting and After deciding on the most efficient
pilot-plant stage, the capital cost continued positive economics, the reactor system, the process and
of the anticipated commercial unit next step toward commercialization mechanical design of the reactor
should be estimated. It goes with­ could be the construction of a dem­ and ancillary equipment should be
out saying that the estimate is only onstration unit. The time required completed. For reactors using het­
as good as the pedigree of the input to design, procure, construct and erogeneous catalysts, to accurately
information. To achieve a robust esti­commission a demonstration unit calculate pressure drop, research­
mate, one of the first activities shouldcould be two years or longer. In this ers should be prepared to provide
be creating an equipment list en­ interim period, to support the overall catalyst physical properties, such as
compassing all the equipment on theprogram, fine-tune the economics bulk density, void faction and a char­
flowsheet. The equipment list shouldand further reduce risk, it is recom­ acterization of the pellet’s nominal or
summarize equipment size and type,mended that the pilot-plant program wetted diameter. Scaleup of many
dimensions (diameter and length), and other key development activitiesreactor types can be accomplished
mechanical design conditions (tem­ continue. Even a scaled-back R&D using mathematical models and
perature and pressure) and materi­ program can reinforce the underlyingscaling rules.
als of construction. For pumps and basic assumptions used in the go- One of the most important goals
other rotating equipment, volumetric forward decision. of reactor design is to replicate the
capacity and differential pressure are As discussed previously, the need mass- and heat-transfer character­
required to determine driver require­ for an intermediate-scale demon­ istics, fluid linear velocities, flow re­
ments. Using current equipment stration unit depends on numerous gimes (turbulent versus laminar) or
pricing or pricing scaled from similar technical factors, as well as com­ Reynolds numbers used during pilot­
equipment or plants, an order-of- mercial requirements. For example, aing. CFD modeling of reactor systems
magnitude factored estimate can key step in the commercialization of is a powerful tool to predict and op­
provide a relatively good idea of in­ novel bio-based fuels, polymers and timize flow distribution, heat transfer
side battery limits (ISBL) investment.chemicals might be the production ofand mixing (Figure 7). For irregular re­
Equipment costs can be obtained di­large quantities of materials for mar­ actor geometries, CFD can provide a
rectly from vendors and past projects, ket testing and qualification by the reasonable estimate of pressure drop
or based on historical cost indices orend users. Typically, the required vol­ and flow patterns. CFD can highlight
scaling similar equipment. While hav­umes for market testing of new prod­problem areas, such as unwanted
ing access to accurate equipment ucts are difficult to produce at the backmixing, stagnant areas and ele­
pricing is important, knowledge of in­pilot-plant scale. ments that may be prone to vibration.
stallation factors and regional location In many cases, demonstration CFD is particularly helpful in develop­
factors, such as construction costs, units can be located near or co­ ing reactor details, such as feed injec­
are also key to determining an accu­located within an existing operating tions and mixing elements, which are
rate investment picture. A more ac­ facility. This should allow the dem­ very difficult to translate directly from
curate estimate will include a break­ onstration unit to leverage existing the pilot scale.
down of direct costs, such as piping,utilities, feedstock handling, waste In addition to selecting materials of
disposal and permits, as well as op­ construction, reactor design includes
instruments, structural steel, electrical
systems, concrete and insulation, aserate at higher rates for extended developing details of the reactor inter­
well as construction and field labor. periods. As expected, the capital nals, including the following:
Armed with the capital estimate, investment for a demonstration unit • Feed or quench spargers
typically ±30-40% at this stage, alongcan be many times higher than a • Bed supports
with the operating cost and calcu­ pilot plant. With sufficient data from • Manway access
lated return on investment, the eco­ the pilot plant, the conceptual de­ • Catalyst loading/unloading nozzles
nomic feasibility of new process can sign and optimization of the dem­ • Internal instrumentation
once again be scrutinized. Nowadaysonstration unit or commercial plant Less obvious details, such as ship­
commonly referred to as the techno-can begin. ping braces, insulation details, clear­
economic analysis, this evaluation canUnlike pilot plants, the equipment ances, tolerances and expansion
help business leaders understand the and flow scheme in a demonstra­ joints, also need to be considered.
economic upsides, costs, benefits tion unit should be nearly identical
and potential risks. The goal of the to the commercial plant. While it is Scaling major equipment
analysis should provide the technol­ important that equipment scaleup In the scaleup of most towers, drums,
ogy developer, business stakehold­ use the pilot data as a guide, direct vessels and heat exchangers, the
ers and any potential first adopter theextrapolation to the demonstration basic rules of chemical engineering
confidence that the future plant will bescale is not always possible. Pilot frequently apply. For some unit op­
profitable. In parallel, the team shouldplants typically have higher holdup erations, often the simplest approach
also outline the goals of the demon­ volumes, different surface-area-to- to scaleup is to maintain the same or
stration unit, list areas of concern andvolume ratios, heat losses and lowersimilar geometry and fluid hydrody­
define future work or pilot studies. fluid linear velocities. namics as the pilot unit. But from a
49
CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019
 practical standpoint, given the very as outlined in the API Technical Datathe process is competitive, the new
nature of pilot-plant equipment, di­ book or measured in the laboratory. process can be deemed ready for full
rect scalability and geometric similar­ commercialization. Whether it be for
ity is never truly possible. For exam­ Procurement and construction a captive company internal project or
ple, in the pilot plant, the nearly idealAfter basic and detailed engineering broad-based licensing to industry, the
separation of two immiscible liquids are complete, the timeline of most next hurdle is convincing customers,
could be achieved using a decanter/ large-scale demonstration units fol­ management and business partners
coalescer based on a nominal holduplows the norms of a typical chemical to invest many millions of dollars for
time of one hour. Simplistically, usingplant project. During procurement the full-scale commercial unit.
the same basis for the demonstra­ activities, special attention should be Given that years have probably
tion plant could result in an extremelygiven to equipment items and special­ transpired from that first ’’ah-ha" mo­
large horizontal vessel that may not ized instrumentation that are critical ment, with any luck, management will
be the most economic or elegant so­or unique to the new process. After continue to support the project and
lution. However, knowing the relativevendor selection is complete, the customers will see the value of the
liquid densities, average droplet sizeengineering team should establish a new process. Furthermore, it is impor­
and the viscosity of the continuous close relationship with the fabricator.tant that market demand is still strong,
phase, Stokes’ law can be used to The team should carefully review andcapital-cost estimates have remained
estimate the droplet settling rate thatcomment on vendor prints, ensuring reasonable, raw materials are readily
may lead to a more cost-effective full compliance with the original intentaccessible, the price of petroleum is
design and practical solution. of the design drawings. Then again, stable and environmental permits are
While piloting of heat exchangers many times during equipment fabri­ obtainable. While it is understandable
is usually not required, there are still cation and assembly, vendors have to be buoyant over the successful
numerous important decisions to be valuable recommendations to im­ demonstration, clearly the challenges
made. Considerations include the prove the design and constructability.and barriers for full process commer­
type of exchanger (for example, shell- As noted earlier, developing the cialization are far from over.
and-tube, plate or spiral), fluid allo­ commercial version of a laboratory Each technology has a unique
cations (shellside versus tubeside), catalyst can be challenging, costly pathway from concept to commer­
configuration (horizontal, vertical, and time consuming. During the en­ cialization. Developing a new chemi­
thermosyphon, kettle or forced circu­gineering and construction phases ofcal process can be one of the most
lation), as well as arrangement details the demonstration project, the devel­rewarding experiences in one’s chem­
(tube diameter, TEMA head and shellopment team should collaborate withical engineering career. It demands a
type, number of passes, baffle spac­ the catalyst supplier to ensure large committed team effort, hard work, in­
ing and so on), most of which can volumes of high-quality, on-specifica­novative thinking, good timing and a
be determined by an experienced tion catalyst can be produced. Con­ bit of luck. ■
engineering contractor. sistently producing tons and tons of Edited by Mary Page Bailey
While many times overlooked, a new catalyst formulation is quite
the engineering team will also be different and much more challeng­ Authors

n
responsible for establishing more ing than running a short plant trial to Vincent (Vinny) Welch is the
routine design elements, such as produce a drum or two of catalyst for managing director of Technip-
FMC’s Boston office (One Financial
the following: the pilot plant. To avoid unpleasant Center. Boston, MA. 02111; Email:
• Distillation tray efficiencies surprises prior to demonstration unit \ r vincent.welch@technipfmc.com).
• Physical properties startup, it is beneficial to benchmark He is responsible lor managing
engineering, development, licens­
• Fouling factors
• Corrosion allowance
• Control schemes
the full production batch of cata­
lyst compared to the catalyst used
during piloting.
*M ing and research efforts. With over
35 years of experience, he has
spent most of his career licensing
and developing process technology for the petrochemi­
• Materials of construction cal industry. He holds a B.S.Ch.E. degree from
• Overpressure protection Demonstration unit startup Northeastern University.
« Effluent or wastewater treatment Commissioning, startup and opera­ Joseph Peters is a senior director
af TechnipFMC's Weymouth re­
• Design codes tion of the demonstration unit requires search facility (56 Woodrock Road,
Before equipment design can a dedicated workforce and laboratory E. Weymouth, MA 02189; Email:
begin in earnest, fluid physical prop­ support. After the typical plant-com­ •v > joseph.peters@technipfmc.com).
He manages the research center
erties must be determined. Proper­ missioning struggles, once unantici­ and is responsible for the execu­
ties, such as density, viscosity, ther­ pated issues have been addressed tion of complex development pro­
mal conductivity and surface tension,and the unit’s operations have lined grams aimed at the commercial­
ization of new chemical process
can be found in the literature, es­ out, the gathering, analyzing and technologies, as well as further improvements to Tech­
timated by the process simulation reporting of data should become a nipFMC's existing portfolio of novel process technolo­
software using a validated database regular routine. After process per­ gies. With over 35 years of experience, he has spent
most of his career developing new technologies from
such as Design Institute for Physi­ formance targets have been met (or concept, through bench, pilot and demonstration scale
cal Properties (DIPPR), calculated exceeded), the larger risks mitigatedand finally to commercialization. He holds an M.S.Ch.E.
using various contribution methods, and the stakeholders convinced that degree from Northeastern University and is a Registered
Professional Engineer in Massachusetts.
50 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM JANUARY 2019