David H. Sanguinetti Feb. 17, 2016

When developing or applying a new process or energy technology it is generally accepted best practice to build a pilot plant to test the technology and demonstrate that it works before building a full scale plant. It is intuitive that if mistakes are made in the design of a pilot plant, the pilot plant may fail. What is not so obvious is that certain mistakes in the design of the pilot plant can allow the pilot to work well, but cause the full scale plant that is built afterwards to fail. My intent for this article is to outline key questions that need to be considered when designing a pilot plant, in order to ensure that promising technologies are given a fair trial and that the money invested in them is not wasted. Although it is primarily technical in nature there will be value for senior management and investors to read the Reasons for Piloting and Compromise Considerations sections so that they can ask the right questions and protect their investments.

Reasons for Piloting

I firmly believe that you should never do something without understanding the reason, or reasons, why, and in the case of piloting this absolutely holds true. The key thing here is to know all of the reasons why, and not just satisfy yourself that there is at least one good reason and that’s enough. This is because one of the key reasons for piloting is often forgotten until after the pilot is built, and by then it is too late. Before identifying this key consideration let’s look at some of the common reasons given for piloting:

• a) To convince senior management to provide more funds
• b) To convince external parties to invest in the technology and/or the company
• c) To demonstrate the technology to potential customers by proving that it works for their specific circumstances
• d) To learn how the process behaves when operated continuously instead of in batch mode
• e) To demonstrate that the technology is scalable, and works with industrial equipment
• f) To get data for a specific feed stream or set of conditions

All of these reasons can be summarised by the concept of demonstrating: whether to an investor, a potential customer or yourself, you are proving that it works at a given size and in a given set of circumstances. However, in all of these cases there is a critical, unspoken assumption: if everything works you are going to want to build a bigger, often full-scale, plant. In order to build the bigger plant, you are going to have to collect from the pilot plant the data required in order to design that larger plant, and in order to do this you have to have designed the pilot to be able to provide that data. Thus, the reasons for building a pilot plant should be summarised into two requirements:

• 1) To demonstrate the technology at a given capacity and/or with a given feed or range of feeds.
• 2) To obtain the engineering data required in order to design the next, larger plant.

If the second reason is not taken into full consideration (and frequently it isn’t), there is a reasonable likelihood that the pilot will meet the first objective, and the engineers will be tasked with designing a full scale plant without the required design data. At this point in the technology development process there is usually no time or budget to go back, re-design, and re-run the pilot to get the missing data, so the engineers end up designing the full scale plant using assumptions and educated guesses.

Unfortunately, this frequently leads to the full scale plant not delivering on the promise of the pilot, and management and investors abandoning what should have been a successful technology.

With an understanding of these general reasons in mind, the following sections will explore some of the important considerations for designing a pilot plant. The sections are divided by consideration (Demonstration and Engineering Data) although each consideration clearly impacts the other.

Demonstration Considerations

When looking at how to design a pilot for demonstration purposes, the first questions to ask are who is the audience for the demonstration and what are they really looking for. Understanding the answers to these two questions will help you ensure that design criteria are picked intentionally and not randomly.

As an example, I have more than once encountered a situation where at an executive or client meeting someone will say something like “We need to show that this will work continuously, say 100 l/min or so.” but the message that comes out of the meeting to the design team is that 100 l/min is the design criteria. The reality is that the number was pulled out of thin air and the real design criteria is simply to demonstrate continuous flow. If 100 l/min doesn’t cause any problems (e.g. budget, schedule, footprint) and there is no particular benefit to a different flow rate, then you may want to stick with it, but you need to know why it was chosen, so that you understand the implications if you want to change it.

Beyond the whim of a client or senior executive, there are many factors to consider when deciding how big to make your pilot plant before even considering the engineering data requirements. The most obvious of these is flow rate. Is there a standard flow rate that is used for piloting in this industry? Are there other pilot plants that this plant may be installed up or downstream of? Is the product of the plant going to be used for some sort of bulk testing and if so how much is required and how quickly?

A common request for a demonstration plant is to have it mobile or transportable so that it can be tested with various feedstocks without the (sometimes impossible) necessity to transport the feedstocks to the plant. For a fully mobile plant this is typically done by mounting it in a wheeled trailer or in a shipping container, while transportable plants are generally done in multiple shipping containers or skids that can be bolted together at site. The benefits of being able to do this are fairly obvious, but it can significantly increase the cost (in some cases by a factor of 2 or more) so make sure that this is a need to have and not a nice to have: are there multiple destinations confirmed or is it just a concept? I have been involved with more than one mobile pilot plant that was used in a single location, put in storage for a period of time, and then disposed of.

Going back to the question of what is the audience really looking for, it is worth asking whether the entire process needs to be demonstrated, or just one or two unit operations. If this is a case where the vast majority of the proposed process is known technology and there are only a few elements that need to be demonstrated, a lot of money can be saved by piloting only the unit operations in question. Frequently this can be done by placing the pilot unit operations in parallel with an existing process and treating a slip stream. If not, you may need to synthesise the feed to the unit operations, so make sure that this is acceptable to the audience.

Engineering Data Considerations

In my experience, the most likely area for problems in the design of a pilot plant is not ensuring that the plant will produce the complete set of engineering data required to design a full scale plant. Occasionally, the reason for this is flat out not including engineering data in the design considerations. More frequently, the problem is a lack of experience on the part of the team designing the pilot, leading to an incomplete set of data being collected, and leaving out critical parameters that are needed for the design of the full scale plant to be successful.

Since by their nature pilot plants are testing processes that haven’t been run before it would not be possible to cover in a single article all of the parameters that might need to be considered. Rather, I will list below a few of the parameters that should be considered and how to obtain them along with some other design considerations to get you thinking along the right lines. However, once the list of desired design data is complete along with the plan for how to collect it, the best approach is to have at least one engineer experienced in piloting review the entire package. It is important that this experienced engineer is external to the design group. This can be someone from your company who is not involved in the project, or if such an expert isn’t available you can use an outside consultant.

A good place to start in obtaining the engineering data criteria is the proposed equipment list for the full scale plant. Working with the design team, go through the list item by item and figure out how you are going to be designing or specifying each piece of equipment. Consider things such as flow rate, velocity, retention time, all three dimensions (do they increase proportionately or not), thickness of materials of construction (will it affect heat transfer?), etc.

A lot of equipment will be straightforward, for example a pump will be sized based on flow rate which will likely scale up directly from the pilot operation. Similarly, inter-stage tanks are frequently designed based on retention time which can be held constant from the pilot to the full scale. Other types of equipment, however, require more thinking. For example, when scaling up an ion exchange column or other similar packed bed there are significant constraints on bed depth, which means that you can’t maintain all three of the bed volumes per hour, the ratio of height to diameter, and the linear velocity and so need to know which ones will be held constant and which will change. Other examples of issues that need to be considered carefully include:

• Laminar vs. turbulent flow: frequently the dimensions inside pilot equipment are such that flow is in the laminar or transitional region, while it will be fully turbulent in a full scale piece of equipment. This can significantly impact heat and mass transfer and multi-phase flow regimes. You will need to consider whether the equipment that gives the process results you want at the pilot flow rate will also give you the engineering data you need, or whether you will have to be prepared to run separate tests to get the scale up data.
• Wall effects: for many unit operations (e.g. gas-liquid contacting) there is minimum vessel diameter for a given size and style of packing. In order to avoid wall effects, you will likely have to use different packing (or possibly a completely different method of contact) in order to have the process work. With that being the case, make sure you know what data you need in order to size the full scale contactor.
• Agitation: when scaling up agitation you need to know if you are going to maintain power per unit volume, turn-overs per unit time, or something else. In most cases pilot scale agitators simply don’t come with the level of engineering that is used for full scale agitators so if agitation is critical to your process you will have to be careful to make sure you know what data is required and how you will capture it.

Agitation is an excellent example of a case where, if it is critical to your process, it is worth talking to vendors to get their recommendations. I use the plural quite intentionally here, because I have found that different agitator vendors use different methods of scaling up their equipment. It’s always better to make sure you are collecting enough data for more than one vendor’s method of scaling up, otherwise you may accidentally force yourself into sole-sourcing your full scale agitators!

In some cases, you simply won’t be able to scale up the particular piece of equipment directly because what will be used in the full scale isn’t made small enough for a pilot operation. For example, a lot of continuous filtration equipment is not made in a small enough size for a 1 litre per minute pilot flow, and similarly rubber lined slurry pumps aren’t usually available below 50mm x 50 mm which have a min flow > 80 lpm.

In each of the two examples I just gave, the pilot plant will have to be built using a different style of equipment to what will be included in the full scale plant. The challenges presented by them, however, are different: in the case of the filter, you will need to talk to the vendor of the full scale filter to determine what data they will need to size it and what sort of performance it will give. That will likely include taking samples prior to filtration and submitting them to the vendor. You can then match that performance as closely as possible with an off-the-shelf bag or cartridge unit for the purpose of continuous pilot operation. In the case of the pump, the sizing of the full scale unit will be fairly straightforward, but the challenge will be finding a pump that can handle the sort of heavy slurry that requires a slurry pump while pumping at the flow rate that you require. (In the pilots I’ve been involved with that had this problem we’ve found something that works, but not perfectly.)

A similar problem comes when looking at the design of custom pieces of equipment. Frequently such custom pieces of equipment include the core IP of the technology, and when the equipment is being designed for the pilot plant it is the largest that has ever been built for that specific function. As a result, there are frequently challenges to overcome even to make it pilot sized, and overcoming those challenges is the primary focus of the designers. It is imperative, however, that the design team keep in mind that their ultimate goal is to design a full scale plant, so they need to envisage how the full scale piece of equipment will be made and what data they will need in order to do so. More than once I have seen a pilot plant that worked, but that incorporated a proprietary piece of equipment that was as large as could be built without making significant design changes, the design data for which were not possible to collect.

One other thing to consider in the design is pipe and ductwork. As a general rule, the sizing of these is straightforward, but there are a couple of areas where caution is advised. The first is if scaling (i.e. the formation of solids on the walls) might be a concern. If there is even a slight possibility of this, design the piping in such a way that it can be easily disassembled for inspection, particularly in areas of concern like elbows and low-flow spots. The point here is less to watch for pipe blockages in the pilot (although those can happen), and more to see if the full sized plant is going to have to be designed with scale prevention in mind. The second area of concern is if your process involves solids being carried either in a gas or as a liquid slurry. The issue here is the need to determine under what conditions the solids will settle out. There are methods in the literature to calculate this, but nothing beats actual operating data. If it is possible to vary the flow rate while operating in order to determine at what flow rate the actual solids you will be transporting will settle out, this will give you valuable design information.

While discussing the demonstration considerations I pointed out that sometimes it is not necessary to pilot the entire process in order to demonstrate the areas of concern. One thing to watch for if this is being considered is whether or not there are any recycle loops planned for the full scale plant. If there are, it is critical that these be piloted, and for a reasonable length of time, because it is only in this way that you can determine whether there are any species that are accumulating due to the recycle.

Compromise Considerations

So what happens when some other consideration conflicts with one (or a number) of the design criteria you have set based on the demonstration and engineering design considerations above? The cause of this is most often budgetary, although schedule, footprint, and other issues can also be factors. Decisions regarding demonstration conditions are context specific and often made by senior managers or investors, but decisions regarding engineering data collection require the input of the design team.

When faced with the requirement to reduce the amount of data that can be collected the most common response is to say “well, let’s get as much as we can”. However, this is not always the best approach. Rather, I would suggest that the best approach is to ask “what data will be most difficult to approximate or find from other sources?” As an example, although it is better to get filtration data from the product of a continuously operating pilot, it is possible to get an approximation from solids created batch-wise in the lab. However, there may be no other way to get data on a scaled-up proprietary piece of equipment than to build it and run it, and without that data you may be forced into building that equipment for a full-sized plant without any engineering data – a place that you definitely don’t want to be.

Summary

In summary, before designing a pilot plant make sure you understand who you are demonstrating the plant for, what they really want to see, and what engineering data you need to get out of it. Then, have it all reviewed by an outside expert who can challenge you with the difficult questions. Remember: difficult questions at the design stage are nowhere near as difficult as those same questions after the plant is built! Only once you have done all of this can you be confident that you are giving your technology the best chance to succeed at both pilot and full scale.