Commercializing Process Technologies

New technologies can provide competitive advantages compared to established processes, but significant effort is required to transition a promising concept into a commercial reality

Many students study chemical engineering with the aspiration of someday inventing an innovative process that could be a true game changer. As we all know, creating a new process is quite complex. Compared to the rapid pace of advancements in other progressive industries, such as consumer electronics, telecommunications, software and automotive, in the world of high-volume, low-margin commodity chemicals, higher-value specialty chemicals and newer sustainable chemicals, process development and commercialization are more methodical and time-consuming.

This article is largely directed at illustrating the development lifecycle for petroleum refining and petrochemical processes, rather than those in the pharmaceutical, biochemical, mining or inorganic chemical industries. The intent is to describe how best to navigate the complex maze of process development, starting from basic research and development (R&D) to technical readiness and commercialization (Figure 1).

FIGURE 1. Demonstrating a promising new technology in a laboratory setting is just one of many complex steps in commercializing a new process technology

Getting started

One of the most difficult challenges facing the chemical process industries (CPI) today is the cost and timeline to commercialize new, innovative technologies. The conventional textbook approach of multiple incremental steps of tenfold scaleup (typically from bench scale to pilot plant to demonstration unit and finally to commercial plant) can take as long as a decade and be prohibitively expensive. Using a staged-gate approach — where process viability and economics are assessed at each step during the development program — is crucial not only for risk mitigation, but also for maintaining management and investor confidence. Because each subsequent step typically requires significantly larger capital and operating cost outlays, the staged-gate approach compels the development team to pause and take an objective view of test results, process design issues and economics.

It is clear that the revamp or building of any chemical plant involves significant investment and market risk. For producers in mature markets, the risk of investing in a large-scale demonstration or full-scale commercial unit for an “unproven” process technology is nearly impossible to justify without a thorough evaluation, including testing at bench and pilot scales. As many readers can attest, few, if any, changes to a chemical plant or process are simple and turn out exactly as calculated at the start of the development program.

Even with justifiable economics and appropriate incentives from the technology developer or owner, most CPI companies are reluctant to be the first, or even the second, adopter of a new process or catalyst system. Most building-block petrochemicals and high-value specialty chemicals produced today are very mature technologies with well-known competitive economics. Being an early adopter can actually have a significant upside because projects can take years to realize. For companies to significantly improve their competitive position, they must move from the standard tried-and-true processes to more forward-looking innovative technologies. Those who choose to “wait and see” run the chance of operating with a sustained disadvantage once the new technology is commercialized. However, on the downside, the risk to the early adopters is failure of the new technology to meet expected performance, thus jeopardizing the entire investment. The chances for success improve with a deeply thought-out and rigorously executed development plan.

Process improvements can take many forms that can save operating cost or capital expenditures, as well as improve product purity and properties. These may include higher catalyst selectivity and conversion, improved reactor productivity, lower catalyst or feedstock costs, reduced environmental impact and more. True stepouts may embody completely new chemistry routes, alternative feedstocks and reactor designs, improved catalyst formulations or enhanced manufacturing techniques, such as improving energy recovery or plant reliability.

The decision to start a technology-development program depends on your perspective. Existing plant owners must protect their market share. They should be constantly evaluating competitive technologies to determine when and if they must purchase or develop new technologies to replace those already in place. Engineering and licensing companies look for advantaged technologies that can be competitive in the marketplace. Their interest is in licensing these technologies and deriving income from license fees, engineering services, proprietary equipment and catalyst sales.

Commercialization steps

Each process-development program requires several project phases, from the initial proof-of-concept through numerous levels of technical readiness, stage gates and eventually to the construction of the commercial plant. These reassessment requirements vary greatly from company to company. As one can imagine, the technical readiness requirements for a major petroleum or chemical company are likely different from those of a technology licensor or engineering company. Even within the same organization, depending on starting point and complexity, each technology-development program has its own path and strategy to get from the initial concept to full-blown commercialization. Additionally, risk mitigation and individual risk tolerances set and drive the scaleup progression and development lifecycle. The textbook rules of process scaleup include steps going from a very small-scale testing apparatus (bench scale) to a more representative pilot plant, then possibly to a demonstration unit and finally to the commercial plant (Figure 2). It should be highlighted that scaleup is not only limited to the key process-unit operations, but also to the catalysts, feedstocks, additives and equipment that form the complete basis for the new process.

FIGURE 2. Bench-scale tests set the foundation for scaling a new technology up to a more representative pilot plant

Defining the required number of steps and their scale greatly influences program cost and timeline. From a capital expenditure standpoint, an automated pilot plant may cost $1–3 million, an integrated demonstration unit in the range of $10–50 million and a full commercial plant could easily be in the hundreds of millions of dollars. These costs must be taken into consideration before undertaking any process-development program.

Deciding on the need for the intermediate demonstration unit step can be one of the toughest decisions. With an outline of the known facts and assumptions, researchers and engineers will have lively debates on the need for this time-consuming and expensive step. Avoiding the demonstration-unit step could reduce the time to commercialization by two or three years, albeit with a potentially higher risk profile. In the end, the final decision may come down to risk tolerance, identifying ways to mitigate risk and adding design safety margins or perhaps even a “gut feel” based on experience. For example, the development team could determine the consequences of lower-than-expected performance. What if the selectivity improvement is only 95% of the expectation? What if the catalyst life is two months shorter or only partially regenerable? Once the most likely performance gaps have been identified, the team should develop a strategy and identify the steps needed to account for or remediate the shortfall.

Simplistically, the number of process-development steps is a function of reaction chemistry complexity, the non-ideality of the components involved and the scalability of the unit operations. While the sequence of process-development steps is outlined here, deciding which steps are mandatory should be determined by an experienced team of researchers and engineers in cooperation with business leaders and plant operations managers.

Catalyst development

For many programs, the earliest project phase involves the development of an alternate chemistry path or an improved catalyst. Nowadays, it is common practice that these improved pathways and catalysts are developed using high-throughput testing methods that study multiple catalyst samples in parallel, significantly shortening the time required to evaluate candidate catalysts or investigate reaction pathways. High-throughput testing is a powerful tool for identifying trends and variables that may lead to fully optimized catalyst formations. During a typical screening project, testing a few dozen or even several hundred materials and conditions is not unique.

As is typical in the early stages of process development, high-throughput testing protocols are performed on a very small scale under isothermal conditions with crushed catalyst. Moderately exothermic reactions or high-activity catalysts typically utilize a catalyst bed diluted with inert material. Not surprisingly, subsequent catalyst testing utilizing a particle size more representative of the commercial catalyst (for example, pellet, cylinder or sphere) operating under simulated commercial-scale conditions may produce considerably different results. While single stage or bed testing is acceptable at this point in the program, managing heat transfer requires careful consideration before moving forward. Options may include multibed reactors, pseudo-isothermal tubular reactors, fluidized-bed reactors, interstage quenching, interstage heat transfer or interstage feed injection.

The most promising results from these early tests are compared to the performance of the existing catalyst and process systems to determine if there is a potential significant advantage. This involves developing a new flowsheet, including the reactor design, and understanding the existing technology. Technical and economic analyses of the new and existing processes result in a decision to proceed to bench-scale testing (Figure 3).

FIGURE 3. Early tests and bench-scale experiments can help to narrow a project’s focus to the most promising catalyst candidates

Bench-scale experiments utilizing the most promising catalyst samples are usually performed on a very small or microscale and could involve glassware, small stirred-tank reactors, microreactors or other systems that use a minimum of amount catalyst and feedstock to obtain proof-of-concept data. The evaluation includes a determination of not only catalyst activity and selectivity, but also gathering preliminary information on catalyst life or regenerability. The experimental results are also compared to those obtained at the earliest phase.

If the results continue to look promising, the next step involves producing the catalyst at a large enough capacity such that it can be considered representative of commercial-scale production. This typically involves working with a catalyst supplier to develop the commercial production techniques required to economically produce a catalyst with performance parameters close to or better than those of the laboratory-scale material. It is common that the catalyst vendor will perform an abbreviated plant trial to produce a sufficient quantity of material. This activity can be the most challenging step because the commercial catalyst must not only have similar activity and selectivity to the sample, but also the physical properties, such as crush strength, required to operate at commercial scale.

Pilot versus demonstration

The terms “pilot plant” and “demonstration unit” are sometimes used interchangeably. For the sake of this article, pilot plants are configured to facilitate fundamental learning, determine rate-controlling steps and collect large amounts of data at a wide range of conditions to ultimately support the design basis for the commercial plant. Within practical limits, the pilot plant should attempt to simulate the envisioned commercial plant and be flexible enough to accommodate projected operating scenarios and numerous experimental runs. Pilot plants are highly instrumented to ensure a complete understanding of the operating conditions and the impact of the equipment on the process being studied. A pilot-plant reactor may have dozens of thermocouples and numerous sample ports to support the understanding of kinetics, reaction mechanisms and catalyst performance and lifetime. A rigorous pilot development program will carefully examine issues well beyond the basic process chemistry and kinetics. Several of these considerations are listed in Table 1.

Conversely, the demonstration unit should be at a scale such that equipment size has little or no impact on the results compared to the envisioned commercial unit. For reactor designs, the configuration should be virtually identical to that expected to be used commercially. Overall, the demonstration unit provides a validation of the commercial process but at a smaller, less risky scale. It should also have some limited built-in flexibility and be designed to operate continuously for long periods of time. Compared to pilot plants, demonstration units typically will have far fewer instruments and connections, which are mainly used for control and safety purposes. Demonstration units should also be a source of valuable information on operational issues (such as startup, shutdown and controllability) and maintenance issues (such as fouling, cleaning intervals and equipment reliability).

For reaction systems, kinetics, fluid dynamics, flow regimes and the reactor type play a central role in determining the appropriate number of scaleup steps. Fixed- and trickle-bed systems, for both liquid or vapor phase, can typically be scaled up from pilot-plant data. Fluidized-bed systems (Figure 4) require larger demonstration-scale testing where reactor size does not constrain the hydrodynamics of the catalyst system. Scaleup of continuous stirred-tank reactors (CSTRs) is done in successively larger vessel sizes because mixing, heat transfer, active reaction zone and gas-fluid contacting are very much dependent on equipment size and geometry.

FIGURE 4. A fluidized-bed reactor requires testing at larger scales than other reactor types, due to the unique hydrodynamics of the system

Notwithstanding, the scalability of supporting unit operations also contributes to the thought process. If vapor-liquid equilibrium (VLE) behavior is well known, piloting of distillation columns is usually not required for determining separation parameters. However, separation of recycle streams and final product are key elements of an integrated pilot system and are critical to the ability to scale up directly to commercial size. The impact of recycle on catalyst performance and life, potential byproduct purge and product quality all factor into the evaluation of ultimate plant performance.

If vapor-pressure and enthalpy data are readily available, heat exchangers will likely not require piloting. Conversely, unit operations, such as extraction, drying and crystallization, typically require more scaleup steps and may involve working directly with equipment suppliers. Processes with significant solids-handling requirements, two-phase flow or non-Newtonian fluids, such as foams or sludge, will also require more steps prior to full-scale commercialization. Similarly, processes that deal with powders or non-free-flowing solids can be particularly challenging and difficult to scale up.

As noted, not all development projects require all four scaleup steps. For something relatively simple, such as an incremental change to a polymer formulation, or if past laboratory results have been validated by previous experiments and practical commercial experience, it may be possible to go directly from bench to commercial scale.

Similarly, for some purely thermal-driven processes, such as ethylene production, where developments are typically aimed at larger capacities, increased yield, improved heat transfer, shorter residence time and longer cycle times, the textbook approach may not apply. In the steam-cracking process, where plant capacities can regularly exceed 1,000,000 metric tons per year, even very small incremental improvements can make a huge impact on a producer’s net margin. Here, commercial implementation of advancements may not require a laboratory or pilot program. Sometimes, technology advancements can be accomplished by using commercially available kinetic-yield model software, along with performing engineering calculations with advanced tools, such as computational fluid dynamic (CFD) simulation.

For the most part, however, because piloting nearly always exposes the unpredictable or unknown, and given the enormous financial risks at the commercial scale, it is almost unthinkable not to pilot a new process. Surprises and revelations can include unexpected catalyst performance, feedstock poisoning, safety issues, equipment corrosion, fouling, foaming and plugging. Most chemical-reaction pathways are complex — their true mechanism may not be fully understood and is likely considerably different from the simplified overall stoichiometric equations. The combination of a properly constructed and instrumented pilot plant, along with a well thought-out experimental program, can efficiently and effectively sort out the myriad issues that need to be addressed for scaleup.

Reaction system design

By and large, the reactor is the core of most chemical plants, and its feeds, products, solvents, catalysts and diluents dictate the downstream processing requirements. Reaction-system fluid dynamics (for example, heterogeneous versus homogeneous or single-phase versus multiphase) and the reactor type play a central role in determining the appropriate number of scaleup steps.

For example, in some cases, the scaleup of a homogeneous CSTR system with simple chemistry can be straightforward, allowing the development to go directly from pilot to commercial scale. Assuming heat transfer is not a controlling issue and mixing characteristics do not significantly affect the kinetic rate, scaleup could be accomplished by using the same operating conditions as the pilot plant, maintaining similar reactor geometry, liquid hold-up volume and residence time. Conversely, a CSTR with multiple phases, complex chemistry, a high heat-transfer load, solids or catalyst addition or solids removal would probably require additional validation beyond the normal pilot scale. Similarly, a bio-based CSTR with live organisms, enzymes or yeast certainly requires multiple steps.

To reduce the overall development timeline, a different tactic may be to incorporate the innovation into an existing operating plant possibly as part of a revamp, debottlenecking or modernization project. Or alternatively, a short, controlled “plant trial” could be attempted. For some researchers, a plant trial with a new catalyst is regarded as a simple “drop in.” Conversely, for plant and operations managers, just the mere thought of this “experiment” can them give nightmares, as any unexpected loss in performance can have an enormous impact on plant economics and profitability. Nevertheless, using existing hardware and infrastructure assets may allow the development to go straight from the pilot-plant stage to commercialization without the expense of a full-blown demonstration unit.

One approach for commercializing new reaction systems is to incorporate an incremental slip-stream reactor in parallel with the existing reactor. While this approach avoids some of the costs associated with a large-scale demonstration, it may present less than ideal conditions for validating the new technology. Examples of limitations could be restrictions on equipment location or plot area, the inability to independently control pressure or flows, difficulties with equipment during startup and shutdown or the inability to maintain steady operations because of moving production-rate targets. Lastly, because the highest priority of plant operating personnel will always be the main production unit, the experimental program and the demonstration unit may not get the attention needed to guarantee success.

Unquestionably, irrespective of the number of development steps, before operating companies fully commit to commercializing improvements, they must be convinced that this new investment provides significantly enhanced economics, the technology has longterm viability and sustainability and the process development was rigorous, covering all aspects of the design and operation.

Scaling up a fluid-bed reactor

One of the more challenging reactor types to scale up is a fluid-bed reactor. Fluid beds are historically used in applications that make beneficial use of high-heat-transfer rates (either highly exothermic or endothermic). Given their inherent high solids-mixing rate, fluid-bed reactors allow very energetic reactions to proceed almost isothermally with minimal temperature gradients either axially or radially. Likewise, fluid-bed reactors are advantageous in catalyst services that age or deactivate rapidly. Because the catalyst can be made to readily flow in any direction (up, down or horizontal), catalyst can be continuously circulated between a reaction vessel where the reaction and deactivation takes place and an external regeneration vessel.

In a fluid-bed reactor, most of the gas is in the form of bubbles rising through a well-mixed bed of suspended solids. The performance of a fluid-bed reactor is directly related to the bubble hydrodynamics, which is a complex function of vapor-phase properties, including density and viscosity, and solid-particle properties (such as particle density and particle-size distribution). Much of this can be difficult to predict from first principles, so empirical correlations are typically used. Bubble hydrodynamics in small fluid beds will be very different from commercial-scale units, as the bubble size is constrained within the walls of the equipment. The bubbles tend to form slugs, and slugging beds perform quite differently than freely bubbling beds. Therefore, scaleup is usually carried out stepwise in increasingly larger equipment, progressing from benchtop through pilot-scale to a demonstration reactor. At a minimum, the last step before full industrial scale should be large enough to be free of the “wall effects” that lead to slugging. A good understanding of the bubble dimensions and other fluid-bed properties, including bed density and solids entrainment rate, is required to ensure the final pre-commercial stage is larger enough to be representative of a full-scale reactor. Additionally, non-reactive “cold flow” models can be a cost-effective supplemental tool to help accumulate this knowledge.

Patent and literature search

After an initial idea is formed, two critical tasks need to be completed: a project justification study and a patent or literature search of previous work in the field prior to embarking on any development program.

Even before proof-of-concept has been fully cemented, the R&D program needs to have economic justification to determine its viability. This usually begins by completing a survey of the current market conditions and evaluating existing commercially available technologies. Pertinent background information and data from trusted sources and dedicated industry reports are helpful in building the business case to move forward. Knowledge on several factors can help support the decision to fund or abandon the program, including the following:

• Market size
• Forecasted growth
• Competition and any emerging technologies
• Estimated production costs
• Worldwide production facilities and announced projects

This analysis can answer key questions: Does the new technology significantly reduce variable operating cost? Is the current market overbuilt with low plant-capacity utilization? What are the price trends and preferred feedstocks? Armed with recent and credible data, this detailed justification study provides the support and confidence to proceed with what will most likely be a multimillion-dollar investment.

Before embarking on any new program, a careful assessment of the patent landscape should be performed to ensure that the new idea is not infringing on the rights of others. If infringement is suspected, the program can be abandoned, or ways to circumvent the claim(s) need to be considered. This initial examination will ultimately serve as the foundation for a comprehensive “Freedom to Operate” (FTO), which is the legal opinion on whether the process, catalyst, equipment or their combination may be infringing on patents held by others. A careful review of past work can also sometimes point the current program to more productive areas of research. If an idea or concept is completely new to the field, it is important to decide early on whether to file for patent protection or keep it as a closely held trade secret. Keep in mind that patenting a novel idea or process that may take 10 years to commercialize is a double-edged sword — while it does grant you the right of unencumbered use, it also alerts competitors to your plans.

Key relationships

Once the initial small-scale catalyst testing and economic analysis have yielded promising results, the leading catalyst formulation is identified and there were no red flags raised by the patent search, it is now time to establish two key relationships — one with an engineering company and one with a catalyst vendor.

It is important to employ an experienced team of researchers and engineers that is capable of executing multiple stages of the development project. It is also essential to select an engineering firm that has pilot facilities of the appropriate scale, the necessary analytical capabilities and experience in process scaleup. A competent engineering contractor should also provide guidance on the scope and requirements of the development program by outlining the matrix of experiments and studies needed to create the design basis for the commercial plant.

The second important relationship to establish is with a catalyst supplier. Process development cannot continue unless the catalyst formulation can be reliably manufactured on a commercial scale. The goal is to select a supplier that has the capabilities to generate catalyst samples at a scale that closely resembles the ultimate commercial version. Typically, this requires engaging the expertise of an established catalyst manufacturer, preferably one with experience that is at least tangentially related to your invention. Catalyst vendors also provide valuable insight into critical matters, such as precious-metal loadings, supports, binder types and target physical properties.

Translating the optimal catalyst formulation, which was tested on a very small scale, directly into a commercial catalyst can often be problematic. In most cases, multiple trials are required before a commercially viable catalyst is produced. Samples from each trial must be evaluated at the pilot scale to determine activity, selectivity, regenerability and catalyst life. Of equal importance is data reproducibility. For the same catalyst batch, experimental results performed under similar conditions should be in close agreement. Eventually, more thorough evaluations are done on the final form of the commercial catalyst to fine-tune the performance envelope and ultimate catalyst life. Overall, developing a commercial version of a laboratory-prepared catalyst is no simple matter, and in some cases can control and significantly extend the entire development effort. This phase of the project is by far the most difficult to predict and can be very frustrating for a development team trying to meet schedule and budget projections.

FIGURE 5. An advanced, automated pilot plant, such as this multibed unit, provides opportunity to efficiently study the process at various operating conditions

Designing an effective pilot plant

Automated pilot plants (Figure 5) truly provide the foundation and are the workhorse of all chemical process development. The benefit of operating a small-scale pilot plant is that it allows numerous operating conditions and variables to be explored efficiently and economically on a relatively short time scale. The primary function of the pilot plant is to demonstrate the viability of the new process and generate the required data to support the plant design.

Determining the appropriate size of a pilot plant is not always a straightforward task. Beyond generating technical data for scaleup, one other function of the pilot plant may be to generate quantities of product samples for analysis and market testing. Another key decision is construction philosophy. Should the unit be skid-mounted, modular-design or stick-built? Should the unit be placed in a dedicated enclosure? Will it eventually be moved to the plant site? Selecting between these options depends on the complexity of the chemistry and process, as well as project schedule, budget, in-house expertise and the projected length of the program.

The design of the pilot plant is based on the conceptual design of the process, which includes a preliminary reactor design and required separations. The key factor in the design is how small the equipment can be to provide accurate and reproducible data for direct scaleup to a commercial plant. Obviously, a smaller unit using smaller flows will require less capital, use less feedstocks and generate less waste. But making the unit too small may introduce scaleup uncertainties for subsequent phases of the program. In small reactors and equipment, phenomena like wall effects, inlet thermal-void reactions, bed bypassing, plugging, heat transfer and heat losses, are not only more difficult to control, but may not be obvious at first and can be difficult to quantify.

The translation of pilot-scale data for direct scaleup is done based on proven methodologies validated through the collection and comparison of data collected on both the pilot and commercial scale. These methodologies have been proven at progressively smaller scales for numerous reaction and separation systems.

Fixed- and trickle-bed vapor- and liquid-phase reactions can typically be done in single-tube reactors as small as¾-inch diameter. However, this is very much dependent on catalyst particle size, reaction energetics and expected commercial thermal profile. Careful sizing and loading of the reactor are required to prevent bridging and uneven flow distribution. Inert materials are sometimes used to dilute the catalyst bed for improved flow distribution and thermal control.

Handling the waste and effluent streams is another factor to consider in determining the size or capacity of the unit. Feedstock and disposal of toxic waste can add considerable cost to a pilot program. Reactor effluents that can be condensed are easier to handle, store and discard. Conversely, effluents containing a high percentage of light gases, such as hydrogen, ethylene, nitrogen and propane, may require onsite oxidation or incineration. Permitting and code issues can place restrictions on the total quantity of hazardous materials that can be stored and used in a facility.

For example, for a pilot unit for propane dehydrogenation where propane conversion is about 40 to 45% and hydrogen and CO₂ are produced as byproducts, designing the system to recycle for propane could be extremely costly to accomplish requiring compression, drying, multiple separations and deep refrigeration. Another extreme case is ethylene epoxidation to ethylene oxide where, even though oxygen is reacted to completion, ethylene per-pass conversion could be 10% or less and the effluent contains significant amounts of _CO2 and methane (ballast). Clearly using feeds on a once-through basis for these two examples would generate large volumes of gases requiring thermal oxidation. While the feedstock costs for these programs would be significant, it would be difficult to justify installing the equipment for recovery and recycling them.

Pilot-plant size and complexity also have a direct bearing on the ability to automate the pilot plant for unattended operation. Staffing a pilot plant 24/7 is usually prohibitively expensive. Most catalyst systems require extensive parametric and life testing, which can require months, if not years, of run time. Automation is essential to bring the cost of operation to a reasonable level, allowing for the most part staffing a day shift only. Experience has shown that the payback for automating a typical pilot plant is less than a year of operation. Automation also improves data quality because measurements are taken at regular, defined intervals so that trends previously difficult to see become evident. In the past, an operator with a clipboard taking hourly readings and then graphing the results was effective, but the tools available through automation are quite valuable, given the close operating margins of today’s markets.

Another potential difference between the pilot and demonstration unit could be the materials of construction. For the pilot plant, it may be expeditious or only a small added cost to use a high-alloy material for the reactor and its associated tubing. In the end, however, researchers need to ensure that metallurgy will not influence the chemistry by catalyzing coking or other undesirable reactions. Materials selection for the demonstration and commercial plant should be based on known reaction and corrosion-rate data available from literature or data generated in the laboratory. Where possible, critical pilot-plant systems are made of the same materials of construction as those projected for commercial scale. For example, the pilot-scale reactor itself acts as the “corrosion coupon” since it was exposed to the fluids, temperatures and pressures of the actual reaction system. Metallographic analysis of the reactor after removal from service provides data crucial to the selection of materials of construction going forward.

Finding equipment capable of efficiently handling flows in the grams-per-hour range is certainly possible, but the number of options can be limited. While off-the-shelf equipment items are certainly available, much of the equipment used at the pilot scale is custom-fabricated and tailored to each unique application.

Frequently, the main emphasis of a pilot program is to validate the chemistry fundamentals and catalyst performance that are usually not available in literature or predictable via calculation. Understanding reaction kinetics and byproduct formation mechanisms help determine which reactor type is best suited for a given chemistry. Establishing the approach to theoretical thermodynamic equilibrium governs the maximum attainable conversion. While heats of reaction fix the energetics of the reaction system, they also greatly influence reactor type and design.

Beyond bench-scale testing, the pilot-plant design should attempt to mimic the commercial plant as much as possible. The key element to this approach is ensuring that the reaction system is designed such that the catalyst bed operates as close as possible to expected commercial conditions. This includes the expected linear and radial temperature profiles, operating pressures, space velocity and number of stages. Linear velocity can be difficult to match, given potential limits on reactor length. Thus, to support scaleup, parametric studies should be done at varying linear velocities to determine whether bulk mass transfer is an issue at the lower velocities of the pilot scale. Additionally, startup and shutdown procedures should be synchronized with those of the future commercial plant and recommended by the catalyst supplier.

Highly exothermic and endothermic reactions, which have large extremes between bed inlet and outlet conditions, can also produce noticeably different results compared to nearly isothermal operation. Per-pass conversion, yield and selectivity can be greatly impacted by all of the above. Consequently, while a prime catalyst candidate can certainly be identified using small-scale, high-throughput testing, alternatives should be available when unforeseen results are encountered.

For example, for the highly endothermic, equilibrium-limited dehydrogenation reaction of ethylbenzene to styrene, when the potassium-promoted iron-oxide catalyst is tested isothermally, an average bed temperature of 580°C is required to obtain 65% conversion. Commercially, the catalyst operates adiabatically and to obtain the same conversion, an average inlet temperature of 620°C is required and results in an outlet temperature of about 560°C. For certain older-generation dehydrogenation catalysts, experience shows that operating below 540°C can result in a significant loss of selectivity due to high CO₂ production. The adiabatic pilot testing allows variables such as steam-to-oil ratio, pressure and space velocity to be adjusted so the reactor outlet temperature is comfortably above the 540°C threshold. Clearly, commercializing a new catalyst of this type based strictly on isothermal data could have resulted in less than expected performance.

The goals and schedule of the pilot plant program should be established by the researchers working collaboratively with the engineering team and catalyst supplier. The team should identify the critical data sets required to support the commercial design and the potential demonstration unit.

In addition to the research program schedule, the team must also develop timelines for pilot-plant construction and commissioning. Pilot-plant construction can take months. The unit must be designed and the process equipment, controls and analytical apparatus procured. During the period leading up to construction, the required permits should be obtained, control logic developed, a hazard and operability (HAZOP) study and layer-of-protection analysis (LOPA) performed and the feedstocks obtained or synthesized.

Feedstock considerations

It is always best to perform pilot testing with industrially produced feedstocks. However, care must be taken to pre-purify these materials before use in the pilot plant, as the very act of drumming or packaging these materials may introduce air, moisture and heavy materials left over from drum manufacture. Scrupulous precleaning of the containers and purging of the sampling lines is critical to minimize feedstock contamination.

Feedstocks should also include representative recycle streams. Generation of recycle streams normally requires that product-separation steps be included in the program. For pilot plants with relatively small flowrates, separations, such as distillation or filtration, can be accomplished in batch mode, but care must be taken to ensure that the level of separation would be feasible at the commercial scale. Better yet would be to perform the separations on a continuous basis according to the projections from process modeling software, so that the recycle streams and product compositions represent commercial operation. These separations can be done in a semi-continuous mode, such that the columns run continuously during the day shift at a high enough capacity to keep up with the 24-hour operation of the catalysis plant. Repeating this over many iterations results in recycle streams close to those that would be seen if the distillation system was directly connected to the reaction system.

If this is not possible or practical, synthetic or reagent-grade feedstocks may be used with certain caveats. Reagent-grade feedstocks with extremely high purities, readily available from many sources, are convenient but may produce overly optimistic results. Typically, these materials will contain an entirely different slate of impurities or poisons that may or may not affect catalyst performance and aging. Chemical suppliers offer a wide range of materials and use the same handling equipment for a variety of substances. As much as they try to clean equipment between uses, trace amounts of different substances may carry over and can contaminate the feedstock. The chemical supplier is only required to meet their certificate of analysis and, for the most part, is not worried about trace impurities. Often, these impurities can be below the detection limits of a supplier’s analytical equipment.

Having access to advanced analytical capacities, such as gas chromatography (GC) methods like thermal conductivity detectors (TCD), flame ionization detectors (FID) or nitrogen-phosphorous detectors (NCD), as well as GC mass spectrometry, are key to validating the catalyst performance data, identifying the effect of trace compounds and ultimately providing the team with confidence in the results (Figure 6). Online GCs and process analyzers can provide realtime data for close monitoring of plant performance. Offline analysis typically provides more detailed and accurate composition data, particularly at extremely low levels. Relationships with local universities and third-party analytical providers can also be helpful when specialized analyses are required, such as nuclear magnetic resonance (NMR). For most moderately sized companies, the use of third parties for specialized analytical work saves both on the initial investment and longterm upkeep needed for highly sophisticated equipment.

FIGURE 6. Access to advanced analytical capabilities is essential in determining feedstock integrity and monitoring plant performance

The importance of having accurate analytical methods cannot be overstressed. New processes typically produce a different mix of byproducts and minor impurities. Developing reproducible analytical methods to identify the minor trace components is crucial for not only determining stoichiometry and kinetics, but also for performing material and carbon balances and maintaining adiabaticity and temperature profiles.

Simulation and modeling

While the researchers are typically focused on reaction kinetics and catalyst life, so that the economics of the integrated plant can be determined and the commercial plant designed in the future, the flowsheet for the entire plant should be defined and a simulation model created. The simulation should include all relevant separations (distillation columns, strippers, dryers, absorbers, decanters, flashes and so on), and heat addition or removal, as well as all recycle, byproduct, vent and purge streams. At the most basic level, model development requires a detailed definition of the components or pseudo-components, including low-level impurities that could affect separations and final product purity. A key undertaking for model development is determining the best methods for estimating thermodynamic properties, equilibrium data and other physical properties. For many chemical compounds, the built-in data banks and estimation methods available in commercially available process-modeling software may be a good starting point. For other lesser-known components, data may have to be generated in the laboratory or determined via literature sources. Likewise, if not available from literature sources, parameters, such as the solubility of water or a solvent in the process streams, may also have to be determined empirically in the laboratory.

Reactor modeling can range from straightforward empirical equations to a full-blown kinetic model with rate constants and equilibrium-based interactions. Constructing a robust kinetic model with frequency factors and activation energies that accounts for temperature, pressure and concentration effects is a large effort and can take months to create. As a result, the process developer needs to decide what level of detail is needed for commercialization. A true kinetic model is the ideal tool for optimization but could be redundant or unnecessary for plant scaleup.

As more pilot data (kinetics and VLE) become available, the model should be used to optimize the flowsheet. However, one of the more important uses of the simulation model is to determine variable operating costs and perform an economic assessment. Raw material and utility consumption figures gleaned from the simulation, along with other variable operating costs, such as catalyst and chemical usage, should provide a representative benchmark when comparing the competitiveness of the new technology versus the incumbent process or other alternatives.

Capital cost estimates

With the information available at the pilot-plant stage, the capital cost of the anticipated commercial unit should be estimated. It goes without saying that the estimate is only as good as the pedigree of the input information. To achieve a robust estimate, one of the first activities should be creating an equipment list encompassing all the equipment on the flowsheet. The equipment list should summarize equipment size and type, dimensions (diameter and length), mechanical design conditions (temperature and pressure) and materials of construction. For pumps and other rotating equipment, volumetric capacity and differential pressure are required to determine driver requirements. Using current equipment pricing or pricing scaled from similar equipment or plants, an order-of-magnitude factored estimate can provide a relatively good idea of inside battery limits (ISBL) investment. Equipment costs can be obtained directly from vendors and past projects, or based on historical cost indices or scaling similar equipment. While having access to accurate equipment pricing is important, knowledge of installation factors and regional location factors, such as construction costs, are also key to determining an accurate investment picture. A more accurate estimate will include a breakdown of direct costs, such as piping, instruments, structural steel, electrical systems, concrete and insulation, as well as construction and field labor.

Armed with the capital estimate, typically±30–40% at this stage, along with the operating cost and calculated return on investment, the economic feasibility of new process can once again be scrutinized. Nowadays commonly referred to as the techno-economic analysis, this evaluation can help business leaders understand the economic upsides, costs, benefits and potential risks. The goal of the analysis should provide the technology developer, business stakeholders and any potential first adopter the confidence that the future plant will be profitable. In parallel, the team should also outline the goals of the demonstration unit, list areas of concern and define future work or pilot studies.

Scaleup and demonstration

Following successful piloting and continued positive economics, the next step toward commercialization could be the construction of a demonstration unit. The time required to design, procure, construct and commission a demonstration unit could be two years or longer. In this interim period, to support the overall program, fine-tune the economics and further reduce risk, it is recommended that the pilot-plant program and other key development activities continue. Even a scaled-back R&D program can reinforce the underlying basic assumptions used in the go-forward decision.

As discussed previously, the need for an intermediate-scale demonstration unit depends on numerous technical factors, as well as commercial requirements. For example, a key step in the commercialization of novel bio-based fuels, polymers and chemicals might be the production of large quantities of materials for market testing and qualification by the end users. Typically, the required volumes for market testing of new products are difficult to produce at the pilot-plant scale.

In many cases, demonstration units can be located near or co-located within an existing operating facility. This should allow the demonstration unit to leverage existing utilities, feedstock handling, waste disposal and permits, as well as operate at higher rates for extended periods. As expected, the capital investment for a demonstration unit can be many times higher than a pilot plant. With sufficient data from the pilot plant, the conceptual design and optimization of the demonstration unit or commercial plant can begin.

Unlike pilot plants, the equipment and flow scheme in a demonstration unit should be nearly identical to the commercial plant. While it is important that equipment scaleup use the pilot data as a guide, direct extrapolation to the demonstration scale is not always possible. Pilot plants typically have higher holdup volumes, different surface-area-to-volume ratios, heat losses and lower fluid linear velocities.

Reactor scaleup

After deciding on the most efficient reactor system, the process and mechanical design of the reactor and ancillary equipment should be completed. For reactors using heterogeneous catalysts, to accurately calculate pressure drop, researchers should be prepared to provide catalyst physical properties, such as bulk density, void faction and a characterization of the pellet’s nominal or wetted diameter. Scaleup of many reactor types can be accomplished using mathematical models and scaling rules.

One of the most important goals of reactor design is to replicate the mass- and heat-transfer characteristics, fluid linear velocities, flow regimes (turbulent versus laminar) or Reynolds numbers used during piloting. CFD modeling of reactor systems is a powerful tool to predict and optimize flow distribution, heat transfer and mixing (Figure 7). For irregular reactor geometries, CFD can provide a reasonable estimate of pressure drop and flow patterns. CFD can highlight problem areas, such as unwanted backmixing, stagnant areas and elements that may be prone to vibration. CFD is particularly helpful in developing reactor details, such as feed injections and mixing elements, which are very difficult to translate directly from the pilot scale.

FIGURE 7. Process simulation, especially CFD, is an important step in estimating process behavior and identifying potential problematic anomalies

In addition to selecting materials of construction, reactor design includes developing details of the reactor internals, including the following:

• Feed or quench spargers
• Bed supports
• Manway access
• Catalyst loading/unloading nozzles
• Internal instrumentation

Less obvious details, such as shipping braces, insulation details, clearances, tolerances and expansion joints, also need to be considered.

Scaling major equipment

In the scaleup of most towers, drums, vessels and heat exchangers, the basic rules of chemical engineering frequently apply. For some unit operations, often the simplest approach to scaleup is to maintain the same or similar geometry and fluid hydrodynamics as the pilot unit. But from a practical standpoint, given the very nature of pilot-plant equipment, direct scalability and geometric similarity is never truly possible. For example, in the pilot plant, the nearly ideal separation of two immiscible liquids could be achieved using a decanter/coalescer based on a nominal holdup time of one hour. Simplistically, using the same basis for the demonstration plant could result in an extremely large horizontal vessel that may not be the most economic or elegant solution. However, knowing the relative liquid densities, average droplet size and the viscosity of the continuous phase, Stokes’ law can be used to estimate the droplet settling rate that may lead to a more cost-effective design and practical solution.

While piloting of heat exchangers is usually not required, there are still numerous important decisions to be made. Considerations include the type of exchanger (for example, shell-and-tube, plate or spiral), fluid allocations (shellside versus tubeside), configuration (horizontal, vertical, thermosyphon, kettle or forced circulation), as well as arrangement details (tube diameter, TEMA head and shell type, number of passes, baffle spacing and so on), most of which can be determined by an experienced engineering contractor.

While many times overlooked, the engineering team will also be responsible for establishing more routine design elements, such as the following:

• Distillation tray efficiencies
• Physical properties
• Fouling factors
• Corrosion allowance
• Control schemes
• Materials of construction
• Overpressure protection
• Effluent or wastewater treatment
• Design codes

Before equipment design can begin in earnest, fluid physical properties must be determined. Properties, such as density, viscosity, thermal conductivity and surface tension, can be found in the literature, estimated by the process simulation software using a validated database such as Design Institute for Physical Properties (DIPPR), calculated using various contribution methods, as outlined in the API Technical Data book or measured in the laboratory.

Procurement and construction

After basic and detailed engineering are complete, the timeline of most large-scale demonstration units follows the norms of a typical chemical plant project. During procurement activities, special attention should be given to equipment items and specialized instrumentation that are critical or unique to the new process. After vendor selection is complete, the engineering team should establish a close relationship with the fabricator. The team should carefully review and comment on vendor prints, ensuring full compliance with the original intent of the design drawings. Then again, many times during equipment fabrication and assembly, vendors have valuable recommendations to improve the design and constructability.

As noted earlier, developing the commercial version of a laboratory catalyst can be challenging, costly and time consuming. During the engineering and construction phases of the demonstration project, the development team should collaborate with the catalyst supplier to ensure large volumes of high-quality, on-specification catalyst can be produced. Consistently producing tons and tons of a new catalyst formulation is quite different and much more challenging than running a short plant trial to produce a drum or two of catalyst for the pilot plant. To avoid unpleasant surprises prior to demonstration unit startup, it is beneficial to benchmark the full production batch of catalyst compared to the catalyst used during piloting.

Demonstration unit startup

Commissioning, startup and operation of the demonstration unit requires a dedicated workforce and laboratory support. After the typical plant-commissioning struggles, once unanticipated issues have been addressed and the unit’s operations have lined out, the gathering, analyzing and reporting of data should become a regular routine. After process performance targets have been met (or exceeded), the larger risks mitigated and the stakeholders convinced that the process is competitive, the new process can be deemed ready for full commercialization. Whether it be for a captive company internal project or broad-based licensing to industry, the next hurdle is convincing customers, management and business partners to invest many millions of dollars for the full-scale commercial unit.

Given that years have probably transpired from that first “ah-ha” moment, with any luck, management will continue to support the project and customers will see the value of the new process. Furthermore, it is important that market demand is still strong, capital-cost estimates have remained reasonable, raw materials are readily accessible, the price of petroleum is stable and environmental permits are obtainable. While it is understandable to be buoyant over the successful demonstration, clearly the challenges and barriers for full process commercialization are far from over.

Each technology has a unique pathway from concept to commercialization. Developing a new chemical process can be one of the most rewarding experiences in one’s chemical engineering career. It demands a committed team effort, hard work, innovative thinking, good timing and a bit of luck. ■

Edited by Mary Page Bailey

Authors

Vincent (Vinny) Welch is the managing director of TechnipFMC’s Boston office (One Financial Center, Boston, MA, 02111; Email: vincent.welch@technipfmc.com). He is responsible for managing engineering, development, licensing and research efforts. With over 35 years of experience, he has spent most of his career licensing and developing process technology for the petrochemical industry. He holds a B.S.Ch.E. degree from Northeastern University.

Joseph Peters is a senior director at TechnipFMC’s Weymouth research facility (56 Woodrock Road, E. Weymouth, MA 02189; Email: joseph.peters@technipfmc.com). He manages the research center and is responsible for the execution of complex development programs aimed at the commercialization of new chemical process technologies, as well as further improvements to TechnipFMC’s existing portfolio of novel process technologies. With over 35 years of experience, he has spent most of his career developing new technologies from concept, through bench, pilot and demonstration scale and finally to commercialization. He holds an M.S.Ch.E. degree from Northeastern University and is a Registered Professional Engineer in Massachusetts.