3.1.1 Sources of Ideas

In the ideation stage ideas about novel process concepts are generated, ranked and the best are selected to be pursued in the research stage. Novel process ideas may be obtained from open literature, visiting conferences and brainstorm sessions.

However, a major source of ideas for new processes is also by focusing on problems of existing processes. But ways to obtain these problems are in my experience far from obvious. I remember from when I was advising senior technologist in a department of advising technologist at a large chemicals manufacturing site that young process engineers would go to the plant managers and ask them what problems they had, so that he or she could solve them that either they said: “We do not have problems, or plant runs fine”, or they said: “Yes, we have a heat exchanger that rapidly fouls, solve that”. Later the young engineer would then discover that it was nearly impossible to even analyse the nature of the fouling or the cause of the problem and that many advising technologists before him had been given the same problem and were also not able to solve it. The reason for this behaviour of plant personnel is that they know that when the young trainee will start, he/she will ask them enormous amounts of questions and/or he/she will come up with a solution to the problem, that will not work and again they have to spend lots of time to prevent that the bad solution is implemented. May be my observation does not hold for other companies or does not hold anymore anyway.

But still here is a hint for obtaining a list of problems from existing plants. Offer the plant manager that you as process trainee are part of the operating shifts for 3 weeks and be part of all five shift groups. When in the shift groups you explain your purpose and listen to the operating personnel about what they know and think about the process. That traineeship will give you lots of insight in process problems.

Another source of ideas is when a conventional existing process requires additional capacity because of market demands. A de-bottlenecking analysis study should then first be performed to determine the bottleneck and so find the lowest cost solution for the capacity increase. The de-bottlenecking analysis study is another very good reason for visiting the existing plant and again be part of the shift operators for 3 weeks. If a process simulation model is available then the actual process conditions and behaviour can be compared with the simulation model and the model can then first be improved and secondly be used to simulate the conditions and flow rates for the increased capacity. With the list of those conditions the bottlenecks may be identified in discussion with the plant managers and other personnel.

Ranking and selecting the best ideas is not easy as in general little information will be available. In Section 3.2.4 an indication is given for information will be generated in the ideation stage to allow a stage-gate assessment. The budget required to generate this information is widely different between companies. Highly innovative companies such as IBM allow each researcher to spend 10% of their time to research their own ideas, without asking upfront an explanation. Other companies have a game-changer organisation with a budget where researchers explain their idea. The game-changer manager then decides to provide a budget to research the most critical elements of the idea. Step by step the idea can then reach the ideation stage-gate. In small companies the budget to research the idea is in general taken by the company director.

3.1.2 Early Technical Risk Assessment and the Role of Process Concept Design

In this section we concentrate on risk identification and assessment in the ideation stage of innovation. A good way to carry out the risk identification is to make a process concept design of the commercial scale process at this stage. The process concept designer can make a process concept design by extracting knowledge from research chemists involved in the idea and by taking assumptions to the best of his knowledge. This process design can then be analysed on uncertainties and lack of knowledge for each part of the design.

What can help in this analysis are the major risk factors determined by Merrow (1988) from 40 commercial scale solids processing plan start-ups. He found that the major risk factors are the following:

New process steps means here process steps that have not been applied at commercial scale for that particular process. So even if a process step has been used elsewhere for a particular application and is now for the first time used for a different substance or different stream then that process step is new. In my career I have seen several process start-ups with a very long start-up time because the start-up team thought they started up a commercially proven process and therefore did not take precaution measures while in reality they started up a new process because the application of the existing process technology was new.

If the new technology has been applied in a different industry branch and is now for the first time applied in the branch of the intended innovation, the risk can be even larger, because different branches have different ways of operating processes and reliable operation may be obtained in one branch in a totally different way.

Here is an example from my own experience. We started up a bulk chemicals process which included a rotating filter with a filter cloth. The filter technology was obtained from a technology provider, who could show records of reliable operation for years for many applications. After 2 weeks the filter cloth was damaged and crystals leaked through it. The process had to be stopped and the cloth renewed. After another few weeks the cloth was damaged again and the process had to be stopped again. In bulk chemicals processes have to run uninterruptedly for 4–6 years, so these stoppages were really a large problem. The technology provider was consulted about the filter cloth life time. After his/her consultation of the filter users, which were all in the food industry, it appeared that in the food industry applications the filter was cleaned every week and in the cleaning time the filter cloth was also renewed. The real filter cloth life time was unknown, but it was expected to last for a few weeks.

Lack of stream composition knowledge is a major risk factor. It involves every stream in the process – both mass streams and heat streams. Lack of knowledge means furthermore lack on the composition of each stream and lack of the size of each stream. The process concept design can be used to estimate the lack of stream composition knowledge for each stream.

This lack of knowledge causes in general over and under sizing of equipment. It can mean that plant product specification is not met, or design capacity is not met and/or plant energy efficiency design target is not met.

But more importantly this lack of stream composition knowledge can have the consequences that the wrong construction material is chosen, resulting in rapid corrosion, that catalysts fail due to poisoning by trace components, that separations such as distillation or extraction fail; due to foaming or froth formation by components with a preference for interfaces.

What also can happen with grave consequences is that unknown impurities end up in the product and may at first instance not even be noticed by the existing analysis methods. At the client however they can create all kinds of problems. They can for instance cause colourisation of the product. Even parts per billion amounts of impurities can already cause this effect. They can also create fouling or foaming in the process.

Lack of trace component knowledge often happens with waste water treatment steps, where often the waste water feed stream has a composition different from previous feed streams due to a new process implementation with its new waste water effluent (Merrow, 1988).

Lack of knowledge on trace components, that are not noticed in the research stage, but that can build up in recycle streams of the real process. This is an even larger risk. Often this build-up, which can take months, or even years, means in the end that corrosion and or fouling occurs. If the new commercial scale plant is directly build, so with no integrated pilot plant, then this lack of knowledge can threaten the performance of the plant in such a way that the plant continues to operate poorly or needs major re-design (Merrow, 1988).

Solids processing is another important risk factor. It is about crude feeds such as in biomass processing or in metal ore processing is the largest single risk. The solids have the tendency to stick, plug and flow unevenly and go where they should not go. It is also about solids formed in processes, such as in biotechnology processes, where micro-organisms are formed and in polymer processes. The process concept design will reveal all solids processing parts.

The process concept design may be modified to avoid or reduce the sections where solids are present, so that the risk of failure is significantly reduced.

Moving and rotating equipment is also a risk factor. Merrow (1988) also reports most frequent equipment failures summarised in Table 3.1.

What all these equipment types have in common is that they have rotating (moving) parts. Table 3.1 also suggests that processes with a high number of rotating or moving equipment and valves have a higher risk of failure.

The process concept design will show the number of these equipment pieces.

This information is also useful in reducing risks. By making the concept design such that no or few rotating equipment pieces are needed the reliability of the commercial scale process will in the end be higher. This can for instance be achieved by function integration in a single piece of equipment, such as in reactive distillation. This type of design is described in Section 3.2.3.3.

In conclusion risk assessment of commercial scale implementation can be assessed at the early innovation stage using the major risk factor assessment of this section.

3.1.3 Early Stage Knowledge Gap Analysis and Risk Assessment

Each company has its own strengths and weaknesses in process innovations. To assess the risk of a particular innovation it is very important to know the capabilities, strengths and weaknesses of your own company in relation to the envisaged innovation. Determining these is not an easy task because a lot of this knowledge is not written down but is so-called tacit knowledge in the heads of experienced people in your company. It is therefore advised to ask experienced process development people in your company: What do you consider the largest risks of this new process and what are the ways to reduce these risks to acceptable levels. The checklist given in Table 3.2 should also be of help to identify the risks for the specific process innovation.

If the own company scores negative for a certain item and for an identified outside company the score is positive then a bilateral open innovation could be considered. If several items are negative for the own company and positive for outside companies then a multi-partner open innovation could be considered. However, joint developments also have their own risks; see the Section 3.1.4 below. It is therefore recommended to first explore changes to the process concept design in such a way that it falls inside the company’s own expertise and competences. If this is possible without large economic consequences then that is the preferred direction to go. If obtaining the required competences is considered to be too time consuming or costly the project may be terminated.

3.1.4 Co-Operation with Innovation Partners

In answering the question to have an innovation with partners or not it should be considered that joint research and development also contains some risks. The risks are the following:

Therefore the following steps are advised before entering a full joint venture on innovation.

3.1.5 Process Route Selection

Process route selection in the ideation or early research stage is also an important element of efficient innovation. Choosing the best process route early in the research stage means first of all that no effort is wasted on poor process routes, so it is an efficiency measure. However, because it is done at the early research stage little information is available for the selection. The process route selection method presented here is based on the observation that feedstock cost is a major cost element in the overall cost. In chemicals processing they are 70–90% of all cost. This means that product yield on feedstock is a major parameter for selecting process routes.

The second major cost element of processes is capital investment costs. An accurate capital investment cost determination in the early research stage is however very difficult, as a realistic process design cannot be made due to lack of information. However capital cost strongly correlate with the number of reaction steps and with the number of major separations involved. The number of reaction steps involved are in general easily determined and the number of major separations can be derived from the by-product streams involved and whether or not single-pass feedstock conversion is complete or not.

A third element is the research and development cost and time involved to bring the process route to commercial scale implementation. The estimation of these cost and timing correlate with the number of relevant critical aspects involved. The critical aspects can be derived from filling in Table 3.2. An estimate of the cost and timing required for the whole research and development can be done by experienced process research and development employees. Experts outside the company may also be consulted to estimate the research and development time and effort required, but as they, in general, don’t know the company’s innovation competences they will be of limited value.

3.1.6 Ideation Stage-Gate Decision

At the gate between the ideation stage and the research stage a decision has to be taken whether the process idea should be pursued or be terminated. In large companies a decision will be taken on the whole portfolio of new process ideas. The ideas that are sufficiently promising will first be selected. Then a ranking will be carried out and the top options will pass the gate. The total budget available for process research will determine how many process options will be researched.

The selection and ranking will be carried out by a panel of experienced research and development managers and business managers based on the information provided in reports. The reports will in general contain the following (Harmsen and van Eck, 2004):

A process concept design will at least contain a block scheme with input and output flows. Preferably the potential feedstock sources and product destinations are provided (Harmsen et al., 2012). Sometimes the input sources and the output are not completely known but only that the feedstocks and products are available on the market. In cases where the feedstock and or product is not on the market, indications will be given how these feedstock sources and or product destinations will be found. For instance by searching for alliance partners who may be able to provide the feedstock or buy the new product.

The advantages of having a process concept design in this very early ideation stage are that it helps to identify information needed, which helps to plan the research stage. It helps to communicate the idea, and it helps to indicate the technical feasibility and economic advantages over the present conventional process. Figure 3.1 shows roles of process concept design in generating the information flows.

Proof of principle experiments are in general laboratory scale experiments in which the new concept principle is proven in some way. It may show that the reaction of feedstock with the catalyst produces indeed the desired product. It may show that the new separation principle has indeed separation selectivity.

A fit the business strategy has to be shown, if the idea is about a new product. If it is an idea of a new process for a product produced already by the company then it is of value to check, whether the company wants to extend the market. Some caution is however needed here. The business section may say that they have the long-term view that the market will not grow and or also that the market share will not be increased. But this view can often and rapidly change. It is sometimes better to show the large potential of the new idea for product cost reduction and product quality increase to the business and not ask for a long-term business view.

The intellectual property position should also be indicated. A rapid patent search will show whether the idea is novel or not. If it seems to be novel a thorough patent search should be part of the proposed research programme. If patents exist then it should be reported, when the patents were filed and who owns the patents. The research proposal should contain a plan on how to deal with those patents. Should an alliance be made with the owner or should the research be directed to avoid patent infringement?

Also the own intellectual property position and policy should be determined. If the company decides to generate patents to protect their innovations then additional experiments may be required to cover a wider scope for the invention. For technology providers and contract research organisation patent generation is often important to generate new business. For product manufacturers patent protection may be important, but often free to apply their own inventions is even more important. Often if it is decided not to pursue patents the inventions are published to obtain that freedom of action.

A rudimentary economic evaluation in general helps to sell the research plan to obtain budget. Because only a proof of principle experiment is available the economics can only be preliminary and rudimentary. For a novel (catalytic) reaction an indication of a potential feedstock cost reduction compared to the reference case and existing process may be reported. For a novel separation capital cost savings and or energy savings relative to the reference case may be reported.

A preliminary safety, health, environmental (SHE) assessment will be made. A list of components and their SHE properties should be listed and a comparison with the reference case should be made, to show that the new idea has lower SHE risks than the reference case. If the risks are higher then it will be hard to pursue the idea even if on other aspects such as cost the idea is beneficial.

All major critical items for success should be stated in combination with the research items to address these critical items and reduce their risk to an acceptable low level. Identification of these risks for the new idea is not trivial. Experienced process developers should be consulted to obtain the list of critical items. Table 3.2 can be used as an additional checklist to identify critical items.

An estimate of the research effort and timing involved can be made by using the critical items for success as starting points. In general this estimation is too optimistic. It is therefore useful to have regular progress meeting in which the actual progress made is compared with the estimated timing. In review meetings with the management improved estimates of effort and timing will be presented and agreement obtained.

It is also very useful to outline the steps to commercialisation. Options for partnership later in the development can be stated and discussed and preparations can be started.

The report will be send to the panel prior to the meeting. At the meeting the basic idea and a summary of the critical items with the research plan should be presented, followed by a discussion and finally a decision should be made.

3.2.1 Determining Potential Project Showstoppers

Determining potential project showstoppers of a new process is the most important part of the research stage. Because if it appears that these potential showstoppers appear to be real by performing experiments then the process innovation project can be stopped and large amounts of money and research capacity wasted is prevented. Often the process concept is modified in such a way that the showstoppers are removed. If the incentives of the modified concept are still sufficient then the project can be redefined and continue.

Here are some methods to determine potential showstoppers:

In making the process concept design many assumptions will have to be made, because several pieces of information will not be available yet. The assumptions which are key to the success of the process concept are all potential showstoppers.

Experienced process designers have a lot of tacit knowledge that becomes active when showing the process concept design. They will highlight many areas of concern and may also state: This will never work. Their knowledge also concerns the capability of the own company to develop and design the process concept. So this knowledge will also reveal whether the own company can develop this process.

Perhaps a word of warning to invite external academics to judge the feasibility of the concept. Often they will state that the concept is not feasible and then give one or more reasons why the concept is not feasible. The best way of dealing with this advice is to write down all the reasons why it will not work and then test these reasons experimentally at a later stage in the company. What in general is not fruitful is to start a discussion on the reasons why it cannot work.

3.2.2 Testing for Showstoppers

The experimental test of the potential killing element should be done in such a way that it is convincing the experimenter and his/her management that the new idea is based on either a real physical and or a chemical phenomenon or not. This experiment is often called proof of principle experiment. Because in the early research stage little money is available, it will take ingenuity to execute the experiment with little cost. Companies with a prominent innovation strategy often let researchers spend 10% of their time and budget on these types of experiments without writing a proposal upfront. This allows the researcher to feel free to carry out experiments with a low chance of success but potentially with a high commercial impact.

3.2.3 Concept Design

3.2.4 Role of Modelling and Simulation

Predicting the behaviour of processes in relation to design parameters is in general very difficult. This is due to the complex nature of the processes. They are complex because:

The role of process modelling and simulation is now to simulate the process performance in relation to design parameters so that optimum design parameters to obtain the desired performance while minimising on energy and capital cost. In the last 30 years enormous progress has been made in process modelling and simulation to aid the process concept designer. Table 3.4 shows some major types of simulation packages available from vendors, their purpose and their limitations.

Many process flow simulation packages, also called flow sheeters, are now available on the market. They have libraries for unit operations and for predicting physical properties. They have also optimisation routines so that for a defined design target and design constraints they can determine optimum design parameters.

The limitations of these packages are first of all when a novel process step concept involving combinations of functions in a novel way is researched, that then the concept cannot directly be simulated with the package. When knowledge of the mixing, residence time distribution, mass transfer and/or heat transfer is becoming available then by smart combinations of existing library models sometimes still the process simulation package can be used.

Custom made models for these new process sections often can be made by a modelling department inside the company or by an outside modelling contractor company.

Model predictions using flow sheeters are very sensitive to the values of the physical properties such as phase equilibrium. This equilibrium can be predicted with special modules of the simulation packages. However if the mixtures contain polar components and or components that form hydrogen bonds then these model predictions can be wrong. If phase equilibrium for these types of mixtures is predicted by the modules then model validation with dedicated phase equilibrium experiments will be needed.

Also the whole simulation model has to be validated using integrated pilot plant tests with the same process sections and recycle flows, before it can be used to design the commercial process.

Computational fluid dynamics (CFD) packages can now be obtained from vendors to predict complex flow behaviour often in combination with complex geometries. For one-phase flow these packages are in general very suitable.

For multi-phase flow and in particular for gas–liquid and liquid–liquid phase systems however the CFD models are in general very unreliable for predicting bubble and droplet sizes. This is caused by two different phenomena. First of all the real chaotic turbulent behaviour is modelled in the CFD using approximations. For the large scale eddies the model predictions are in general reliable. However for the small scale eddies in particular near the vicinity of a flexible surface the models are inaccurate. Secondly bubble break-up and bubble coalescence are very complex phenomena, where often trace amounts of components present in the two-phase system show a preference for the interface and then either enhance coalescence or prevent coalescence. Because of this the model predictions for bubble and droplet size are often totally wrong.

This field is extensively researched by many academic groups for over 30 years, but the progress is slow. For a comprehensive review of the present status of two-phase flow simulations the readers are referred to review article by van den Akker (2010).

If experiments with the real phase mixture have been carried out then model predictions can be validated. But if the experiments have been carried out for a small scale and the CFD model is used to predict the large commercial scale behaviour, still the model predictions can be wrong. For instance, because in the large scale the time scales for high intense turbulence areas where bubble break-up occurs and the low-intensity turbulence where bubble coalescence occurs are different and the “kinetics” of bubble break-up and coalescence are different from the small scale, where the turbulent field was more homogeneous.

3.2.5 Essential Physical, Chemical and Thermodynamic Property Determination

Physical and chemical properties are needed to perform process concept synthesis and process concept design analysis. Most of these scale-independent properties can be found in all kinds of literature sources.

However for some process steps such as distillation or extraction accurate values of phase equilibrium are needed to determine whether distillation will be feasible for the desired separation performance. If the phase equilibrium properties are inaccurate then it may be concluded that distillation is feasible, while in reality an azeotropic mixture may be formed, so that a one column distillation is not a feasible option. If this is found out at the pilot plant stage then a re-design costly in time and money will be needed.

Another important property is the chemical equilibrium property for reactions. If the physical property prediction wrongly predicts that the reaction equilibrium is far to the product side then a pilot plant reactor shows that the equilibrium is much more to the feedstock side, then the whole process design has to change.

A third important property is the heat of reaction. If the reaction is strongly exothermic or endothermic a special heat exchange reaction system has to be designed. Also the small laboratory scale reactor may already suffer from strong temperature increase or drop effects, due to the high value for the heat of reaction.

A fourth important scale-independent data set is the chemistry of the reaction. This involves first of all the stoichiometry of the reaction. This means simply stated what molecules react to which molecules and how many of molecules are needed to form the desired molecule. Also the stoichiometry of the major by-product reactions is needed, to make a first estimate of the yield of product on feedstock.

Finally the reaction kinetics needs to be determined to some extent. For the reactor-type selection it is needed to know the reaction scheme; if it is a consecutive or a parallel reaction for by-product formation and if it is a parallel reaction what is the relative order of the main reaction to the by-product reaction.

Some indication of the time required to reach the desired conversion is in general sufficient in the ideation stage to determine the reactor size. In the research state accurate kinetic rate expressions are desired to allow reactor design optimisation.

So accurate physical property knowledge is needed in the research stage for certain process steps and for certain physical properties to allow a pilot plant design that has a reasonably high rate of success. If the new process design involves new components then in most cases certain physical properties, critical to success, will have to be determined experimentally. This can be done inside the company or outside by institutes specialised in physical property determination.

3.2.6 Research Stage-Gate Decision

The stage-gate decision between research and development is a very important decision, because the development programme will in general mean spending several millions of dollars. The decision to continue the project with the development stage will in general only be taken if:

The decision to be taken will be based first of all on written documents containing the research results, the commercial scale process design, the evaluation of the commercial scale design, the proposed development programme with its cost and timing provided by the research team. Secondly, it will be based on a challenge session also with some experts outside the research team.

The decision to carry out the development programme will then be taken by management.