The seven stages of chemical technology from concept to industrialization (Issue 27/100) -- Pilot plant construction and operation

Thread Content

This post was last edited by xiouxingzhe on 2026-6-11 at 11:44. The seven stages of chemical technology from concept to industrialization (Issue 27/100) —— Technology development: Pilot plant construction and operation. Dear friends: Hello everyone! In the previous issue, we discussed pilot-scale scale-up design – how to determine the scale and which scaling criteria to use. With the plan in place, the next step is to get to work seriously: build the pilot plant and ensure it operates stably. In this issue, we will discuss several key issues in the construction and operation of pilot plants. I. How to determine the construction standards: An issue prone to errors. The construction standards for pilot-scale plants represent a problem that plagues many project managers. It’s not appropriate to go to either extreme. One extreme is \"It’s just pilot-scale; anything will do\" – pipe welding is completed without inspection, instruments are installed without calibration, and civil work is finished without acceptance. As a result, after the system started operating, there were pipe leaks, instrument drift, and foundation settlement; the data collected showed such large fluctuations that it was impossible to determine whether the problems stemmed from the process itself or from improper equipment installation. Such pilot-scale data packages have little reference value for subsequent design. Money was wasted, time was wasted, and a bunch of unclear \"data\" was left behind. The other extreme is the idea that \"since it’s a pilot plant, nothing should be inferior to that of an industrial production facility\" – almost the same cost as an industrial plant is spent on such pilot plants, stainless steel is used as if it were alloy steel, excessive redundancy is incorporated, and the budget is severely overestimated. The purpose of a pilot plant is to operate for two or three years, collect enough data, and then complete its mission; there is indeed no need to build it with a design life of fifty years in mind. My principle can be summed up in two sentences: anything directly related to data quality must not be compromised ； Those directly related to operational safety cannot be reduced. Save wherever possible, and make use of existing resources whenever feasible. II. What can be saved? Building standards can be reduced. The facility for pilot-scale plants does not need to be designed with a 50-year service life as required for permanent industrial buildings. If renting a factory building in a park, use it directly once the conditions are met ； If it is a new construction, light steel structures or temporary buildings can be chosen, with a design lifespan of five to ten years considered. After being built, the pilot plant fulfills its purpose after two or three years of operation; it may then be demolished or modified. Building it to permanent standards would indeed be a waste. Backup server configuration can save costs. In industrial plants, critical rotating equipment is typically equipped with one in use and one as a backup, or even one in use and two as backups, because the losses resulting from unplanned shutdowns can amount to hundreds of thousands per day. On the pilot plant, most pumps only require a single unit. The pilot production was halted; what was lost was testing time, not profit from the product. Critical safety equipment—such as emergency shut-off valves—still requires redundancy, as the cost of safety incidents is the same whether in industrial plants or pilot plants. Insulation and anti-corrosion standards can save costs. The insulation and corrosion protection of industrial installations are designed for a service life of 15 to 20 years. Pilot-scale units only need to meet the operational requirements for two to three years, so the thickness of the insulation layer can be reduced appropriately—as long as safety measures against heat are met—and an economical version of the anti-corrosion coating can also be used. Utilities can be saved through proper setup. Industrial facilities need to build their own circulating water stations, air compression stations, and boiler rooms. If the pilot plant is built within a industrial park or an existing facility, recycled water, steam, and compressed air can be sourced from the existing pipeline networks as much as possible by installing temporary pipelines, which saves a significant amount of investment compared to building the necessary facilities from scratch. III. What cannot be compromised: Instrument accuracy must not be sacrificed. This is the quality assurance of data, the most critical asset in pilot testing. If a cheap flow meter drifts by a few percentage points, your estimates of yield can be off by several percentage points as well, making it difficult to determine an appropriate safety factor for the design. Critical online analytical instruments such as online gas chromatography, as well as flow meters with high precision requirements such as mass flow meters, cannot be chosen based on cost considerations. Safety interlocks must not be omitted. Reactor over-temperature and over-pressure interlock protection, combustible gas alarm systems, emergency relief systems – these are the infrastructure elements that protect the safety of operators. Although the inventory of hazardous chemicals in a pilot plant is smaller than that in an industrial plant, in the event of an accident, the harm to the personnel on site is not reduced just because it is a pilot plant. The quality of pipeline welding cannot be compromised. Although the pipes in pilot-scale plants have small diameters and may not be under high pressure, the welding quality directly affects the reliability of the plant’s operation as well as the quality of the data generated. If the pipeline leaks, it has to be stopped, and once it’s stopped, the data flow is interrupted. Piping pressure testing and airtightness testing must be carried out in accordance with specifications; it is not acceptable to treat them as mere formality with the attitude of \"it’s just a trial run, so anything will do.\" The material of key equipment cannot be compromised. If the main purpose of pilot testing is to verify the long-term effects of a certain corrosive medium on the material of the equipment, then the material used for this equipment must be the same as that used in future industrial installations. Material data cannot be obtained using glass bottles in pilot tests; it is necessary to use actual materials for long-term testing or to conduct verification under real operating conditions. IV. Utilization of Existing Equipment: In the cost of building pilot-scale facilities, equipment costs usually account for more than half of the total. Making smart use of existing equipment can significantly reduce investment costs. Which devices are suitable for reuse? Standardized general equipment – pumps, heat exchangers, storage tanks, as long as the materials and specifications match ； The instrument has passed calibration ； Steel structural frames, pipes. However, used equipment must undergo strict inspection before use. Pumps and compressors need to be disassembled to check the wear of their key components; heat exchangers must be cleaned and tested under pressure to ensure there are no leaks; instruments require re-calibration; and pressure vessels need to have their historical inspection records checked to confirm they are still within their valid period. Making use of old equipment is not the same as collecting scrap; using a faulty old device comes with data interruptions and security risks, and the costs associated with this are far greater than those of purchasing a new device. V. How to feed materials for the first time: Once the pilot plant is built and single-unit tests as well as integrated tests have been completed, it’s time to feed the materials. For the first pilot-scale feeding, I recommend adopting a \"stepwise\" increase in quantity strategy. Start by operating at 20% of the design load; once stable, increase it to 40%, and after further stabilization, raise it to 60% and 80%, until the pilot plant’s rated load is reached. After each stage has been operating stably for at least two to three hours, proceed to the next stage. Why such caution? Pilot-scale plants are different from lab-scale plants. If there is a problem with the pilot plant setup, you can immediately remove the reaction flask from the heating bath for emergency cooling. On a pilot plant, in a reactor of several hundred liters, even if the heat source is turned off immediately, the heat accumulated inside the reactor and the inertia of the unreacted materials mean that it may take over ten minutes or even longer for the temperature to drop. This means you must identify warning signs earlier than in the laboratory and initiate corrective actions sooner as well. Gradual increases in volume give you time to observe and make judgments. VI. Data recording during operation: The data recorded during the pilot operation phase forms the core of the subsequent data sets. What to record? All key operational parameters—temperature, pressure, flow rate, liquid level, and concentration of key components—must be continuously recorded with timestamps. Record all abnormal conditions at the same time – even minor temperature fluctuations need to have their causes determined. There is a detail that is easily overlooked here: not only should the data from the instruments be recorded, but also the operator’s actions. When which valves were opened, when the set values were adjusted, and when the backup pump was switched in – if these actions are not recorded, it becomes difficult to determine whether data fluctuations are caused by issues with the process itself or by human error when analyzing them later. I usually require pilot plants to be equipped with a set of operation log systems, which can be electronic or paper-based. At the end of each shift, it is necessary to clearly document what happened during that shift, what adjustments were made, and what issues remain unresolved. These operation logs, together with the instrument data, constitute an important part of the pilot plant data package. VII. Depth of problem diagnosis: When it comes to the issues that arise in pilot-scale plants, the worst approach is to merely address the symptoms – for example, if a pipe gets blocked, simply clear it so that operations can continue, without investigating why the blockage occurred ； Seeing that the yield was low, I adjusted the temperature and continued the process without analyzing the reasons for the low yield. Symptom suppression allows you to keep going for the time being, but the purpose of pilot testing is precisely to expose the root causes of these problems. If you overlook the root causes, only fluctuations of \"unknown origin\" remain in the pilot plant data; these fluctuations will reappear with even greater amplitude in subsequent designs. For addressing each issue, it is recommended to follow a complete cycle of \"documenting the phenomenon → analyzing the root cause → formulating a solution → implementing it → verifying it\", with every step being documented in writing. The value of records such as when this problem occurred, what phenomena were observed, who conducted the analysis, how it was handled, and what the results of that handling were, is often greater than that of the normal operation data from pilot tests. Because normal operation data tells you “what performance can be achieved,” while issue logs tell you “why it isn’t possible to achieve that performance sometimes.” VIII. How long does the pilot test take? There is no fixed answer to how long a pilot test should last. It depends on the complexity of the process, the level of scaling risk, and the type of data you need. However, there is a basic standard: the pilot scale operation must proceed until the facility can operate stably and continuously, the key operational parameters no longer exhibit systematic drift over time, and all quality indicators meet the required standards consistently. Stability does not mean \"being stable for those few hours right after the parameters are adjusted.\" Stability means “operating continuously for a considerable period of time, with the parameter curve remaining relatively flat and able to be maintained without any manual intervention”. Only by reaching this level can the collected data be considered representative steady-state data, which can then serve as a basis for preparing the process package. If the process is complex or involves hazardous chemical processes that require industrial-scale testing, the total operating time for pilot testing may amount to several hundred hours or even thousands of hours. But no matter how long it takes, data quality remains the core goal—not \"running for enough time,\" but obtaining usable data. Preview for the next issue: Issue 28 – Sensitivity enhancement experiment: Identifying the controlling physical quantities. The pilot plant is now in operation, and steady-state data is available. But there is another important type of experiment to conduct: the amplification sensitivity experiment. After reactor scaling up, the mixing intensity, heat transfer capacity, and residence time distribution all change. How much impact do these changes have on the reaction results? Which physical quantity is the “controlling” variable for this reaction system? To be continued in the next issue.