The seven stages of chemical engineering technology from concept to industrialization (Issue 24/100) -- Construction of a pilot-scale continuous plant

Thread Content

This post was last edited by xiouxingzhe on 2026-6-23 at 15:28. The seven stages of chemical technology from concept to industrialization (Issue 24/100) —— Technology development: Setting up pilot-scale continuous systems. Dear friends: Hello everyone! In the previous issue, we discussed the exploration of separation solutions, and both the main solution and the backup solution have been identified. The reaction conditions have been determined, and the separation process has also been established, but all of this is carried out on a small scale in a batch manner: a batch of material is fed in at a time, analyzed after the reaction is complete, and then another batch is fed in. But real industrial production operates continuously 24 hours a day. Materials are in constant flow from entry to exit; if any link in this process stops, it affects both the upstream and downstream areas. Going from intermittent to continuous is not as simple as adding a few pumps and connecting several glass bottles together. In this issue, we’ll discuss how to set up a continuous small-scale testing setup. I. What is the essential difference between intermittent and continuous operation? In intermittent operation, you can carry out inspections, cleanings, and adjustments between batches. The conversion rate for this batch is low; I’ll adjust the temperature slightly for the next batch. The color of this batch of output is a bit dark; reduce the reaction time for the next batch. You have plenty of buffer space. After the process is made continuous, the material flows continuously within the system. When you adjust any upstream parameter, its effects will only become apparent downstream after a certain delay – this delay can range from a few minutes to several hours, depending on the amount of fluid held in the system and its flow rate. This leads to a challenge that is completely absent in batch operations: you can’t finish one batch and then see what happens; you must make judgments and make adjustments in real time. Another difference is that serialization brings to light the connectivity issues between each unit. During batch operation, after the reaction is complete, the material is poured into a storage tank; once the tank is full, it is sent for separation. The various units are \"isolated\" from one another. After serialization, feeding, mixing, reaction, separation, and discharging are all connected in series; any fluctuation in any of the units in between will be transmitted downstream, leading to amplified effects. Therefore, when building a pilot-scale continuous system, the most crucial task is not to ensure that each individual unit can operate, but rather to enable the entire process to function in a coordinated and stable manner. II. How to choose a pump: For pilot-scale continuous systems, I usually prefer metering pumps as the feed pump. The reason is simple: during the pilot stage, the flow rate is low, possibly only a few liters or even a few hundred milliliters per hour, and it is necessary to precisely control the feeding ratio. Centrifugal pumps have very low efficiency at such low flow rates, and their flow stability is also far inferior to that of metering pumps. But there is a problem with metering pumps that is easily overlooked—pulses. Piston pumps or diaphragm pumps do not produce a smooth flow, but rather pulses of flow. The pulse amplitude can be as high as 10% to 20% of the average flow rate. For reactions sensitive to residence time—such as continuous-flow diazotization—such pulses severely affect the reproducibility of the reaction. The solution is to install pulse dampers at the pump outlet, or to use metering pumps with multiple pumping heads connected in parallel, such that the phases of the various heads are offset from each other; this allows the amplitude of the resulting pulses to be reduced to below 5%. If the reaction system contains solid particles—such as suspended catalysts—diaphragm pumps are superior to piston pumps, because the seals in piston pumps wear out rapidly in media containing solids. If the fluid viscosity is high—such as polymer melts or polymer solutions—gear pumps or screw pumps are more suitable. The selection logic for circulation pumps is different. Circulation pumps are typically used in scenarios with low pressure differences and high flow rates—such as the reflux pumps in distillation towers and the circulation pumps in evaporators. In this case, centrifugal pumps are the preferred choice; a high head is not required, but a large flow rate range is necessary. III. How to determine the pipe diameter: When selecting the pipe diameter for pilot-scale continuous systems, there is a mistake that many people make: in order to simplify processing, pipes of the same thickness are used everywhere. The key factor in selecting the pipe diameter is flow velocity. When the flow rate is too low, mixing of the material inside the pipe becomes severe, the residence time distribution widens, and it becomes difficult to interpret the data – it’s hard to determine whether the fluctuations in the exit concentration are due to the reaction itself or to mixing. The flow rate is too high, the pressure drop is excessive, the pump’s head is insufficient, or excessive shear is generated within the microchannels – this is particularly important to consider in reaction systems that are sensitive to shear. For the normal flow rate of liquid transport, I usually use 0.5 to 2 meters per second. After the liquid feed is accurately metered, the flow rate can be appropriately reduced to 0.3 to 1 meter per second—the lower speed helps to minimize dosage fluctuations caused by pulses. The gas flow rate is usually between 5 and 20 meters per second, depending on the pressure and allowable pressure drop. In gas-liquid two-phase flow, special care must be taken to avoid the formation of slug flow; the liquid flow velocity is usually between 0.2 and 0.5 meters per second, while that of the gas is between 1 and 5 meters per second. In microchannels, due to the small scale of the channels, the mixing efficiency is much higher than that in conventional pipelines; therefore, a high flow velocity is not required – 0.01 to 0.5 meters per second is sufficient. The pipe diameter is not chosen independently; it must also be matched to the pump’s flow range. If a metering pump has a low rated flow rate, choosing a larger pipe diameter will result in too low a flow velocity inside the pipe – use a smaller pipe instead. IV. Buffer tank: Providing a “flexible joint” for the system. The buffer tank is an unassuming but extremely important component in pilot-scale continuous systems. What is its function? There is a slight fluctuation in flow rate upstream—such as pulses from the metering pump or minor changes in the feed liquid level—and without a buffer tank, this fluctuation is transmitted directly to the reactor downstream. The residence time, mixing ratio, and temperature distribution in the reactor will all be disrupted. With a buffer tank, these fluctuations are first absorbed by the changes in liquid level within the tank, resulting in a more stable flow rate delivered to the downstream system. But a larger buffer tank is not necessarily better. It is too large, which increases the amount of liquid held in the system; as a result, the time it takes for the entire device to transition from one steady state to another is prolonged, reducing experimental efficiency. It’s too small to provide any cushioning effect. For the effective volume of a buffer tank, a simple empirical formula is: normal upstream flow rate multiplied by the time you wish to buffer. The buffer time is usually set between 3 and 5 minutes. For example, if the upstream flow rate is 2 liters per hour and it is desired that the buffer tank can absorb fluctuations over 5 minutes, then the effective volume is 2 divided by 60 multiplied by 5, which is approximately 0.17 liters. Just choose a 200-milliliter jar. The effective volume is typically taken as 50% to 60% of the total volume of the tank as the operating liquid level range; therefore, the total volume of the tank can be set at 350 to 400 milliliters. V. Temperature control: Much more complex than batch experiments. The temperature control in pilot-scale continuous systems is far more complex than that in batch experiments. In batch experiments, it’s sufficient to place the entire reactor in a constant-temperature bath – magnetic stirring combined with bath temperature control provides sufficient homogeneity in most cases. In continuous systems, the material flows through pipes, and the temperature of each section of pipe needs to be controlled independently. The feed section needs to be preheated to the reaction temperature, the reaction section requires precise temperature control, and the discharge section may need rapid cooling to prevent side reactions – the temperature requirements for each section are completely different. There are three common temperature control schemes. Temperature control via a water bath or oil bath jacket is the most economical option, suitable for applications where the temperature is not very high and moderate precision in temperature control is sufficient. Electric heat tracing temperature controller, suitable for pipeline temperature control; easy to install, but with average temperature control accuracy – proper insulation is necessary. Double-layer jacketed tubes enable circulation-based temperature control, offering the highest precision in temperature regulation; they are suitable for reaction sections that are extremely sensitive to temperature, but they also result in the highest system complexity and cost. There is a detail that is easily overlooked: the location of the temperature measurement points. The thermocouple or thermal resistor must be inserted into the fluid, rather than being attached to the tube wall. There can be a difference of several degrees or even dozens of degrees between the wall temperature and the fluid temperature – the pipe wall may be heated to 200 degrees by the heating medium, but the temperature at the center of the fluid might only be 180 degrees. If only the wall temperature is used to represent the reaction temperature, systematic biases will appear in the data. VI. Data collection: It’s impossible to keep track manually. Once the continuous testing setup is in operation, data is generated continuously. Temperature, pressure, flow rate, liquid level – these parameters change every minute. Relying on manual meter reading is simply not fast enough, and it cannot ensure the continuity and integrity of the data. At the very least, an automated data collection system is required. At least temperature, pressure, flow rate, liquid level, and online analysis signals need to be collected. Regarding the sampling frequency, rapidly changing parameters such as temperature, pressure, and flow rate should be sampled at least once every 10 seconds, while slowly changing parameters such as liquid level can have a sampling interval of 30 seconds to 1 minute. During a 72-hour long operation, the amount of data can reach hundreds of thousands of rows. These data are needed for subsequent material balance calculations, residence time distribution analysis, and operation window validation. If data collection is not done properly, the traceability chain for the pilot-scale data packets is broken. VII. The significance of establishing a complete process flow: With the completion of the pilot continuous setup and the establishment of a full process flow, it means that this process has been tested for the first time in a continuous mode. At this point, you will encounter many problems that do not appear at all in intermittent tests – insufficient flow stability of the feed pump, areas in certain pipelines where temperature control is inadequate, a design of the gas-liquid separator that is too small resulting in liquid being carried downstream, and minor leaks at certain flange connections. These problems do not become apparent in batch experiments, because with batch processing you can inspect, clean, and adjust between batches. But on the other hand, the sooner these problems are exposed, the better. Problems identified on a pilot-scale continuous system can often be resolved in half a day by simply replacing a pipe, changing a pump head, or adding a buffer tank. If the same design flaws are discovered only when moving on to pilot or even industrial-scale facilities, the cost will be far more than just a few hours. Preview for the next issue: Issue 25 – Long-term operation: 72 hours is just the starting point. The pilot continuous system has been set up, and the entire process is now operational. But being able to start running continuously and being able to keep running steadily are two different things. What truly tests the reliability of plant engineering is long-term continuous operation – 72 hours or even longer. The hidden problems that arise in short-term experiments—scaling, clogging, and slight catalyst deactivation—will gradually emerge over prolonged operation. Next time, we’ll discuss what to pay attention to during long-term operation, and how to determine whether to stop or continue. Table of Contents link: Seven-Stage Model of Chemical Technology from Idea to Industrialization (Issue 100) – Table of Contents (updated in real time)