The seven stages of chemical engineering technology from concept to industrialization (Issue 32/100) – The terephthalic diamine continuous flow case study

Thread Content

This post was last edited by xiouxingzhe on 2026-6-11 at 11:50. The seven stages of chemical technology from idea to industrialization (Issue 32/100) —— Technology development: The case of continuous flow production of para-phenylenediamine. Dear friends: Hello everyone! We’ve finished discussing the methodology for the third phase, and also gone over the common pitfalls. This issue concludes with a real-world case study—the entire process of para-phenylenediamine continuous flow production, from the laboratory stage to pilot scale. This is a rather special case: it represents original innovation, rather than an improvement on existing processes, and there are no existing industrial facilities in China that can serve as a reference. Precisely because there was no precedent, many problems that would not arise in mature processes came to light in this project. I think this case will be somewhat helpful for understanding the actual progress of the third phase. I. Project Background: Why Choose Continuous Flow? Para-phenylenediamine is an important organic intermediate, and its traditional production relies on batch processes in reactors. The batch process has a long-standing problem that has plagued the industry: the diazotization reaction releases a large amount of heat and proceeds at a rapid pace, making temperature control difficult under batch conditions and posing certain safety risks. Continuous flow technology—namely microchannel reactors—is well-suited to solving such problems. The heat transfer area-to-volume ratio of microchannels is much larger than that of conventional reactors, resulting in a heat transfer capacity that is an order of magnitude or more higher ； The channel scale is small, resulting in high mixing efficiency, and the residence time can be precisely controlled within seconds. These properties make continuous flow an effective means for handling rapid, exothermic reactions. This direction was first identified during the inspiration-driven phase mentioned in issue 10 – a long-standing pain point in the industry; when it met with cross-disciplinary technical principles, inspiration emerged. Following the dual validation in the innovation incubation phase—scientific validity and commercial viability—after making the Go decision, the project officially enters the technology development stage. II. Pilot stage: Concept validation and condition optimization. The first step in technology development is concept validation – can a continuous diazotization reaction take place? We didn’t buy microchannel reactors right away; instead, we first ran the reactions in the laboratory using simple continuous systems. Called a “simple device,” it actually consists of a few thin tubes, an injection pump, and an ice bath – the cost is very low, but it can answer the most crucial question: Can the diazotization reaction take place under continuous flow conditions? Can the product be detected? The concept validation was successful; the reaction does indeed occur. Next is conditional optimization – temperature range, ratio, and residence time, with single-factor scans performed for each one. Determining the temperature window required considerable effort. The diazotization reaction needs to be carried out at low temperatures, but the actual temperature distribution within the microchannels is not exactly the same as the temperature of the ice bath – as the fluid flows rapidly through the channels, its residence time is short, and the actual temperature is influenced by both the feed temperature and the heat generated by the reaction. We conducted a series of experiments, measuring the temperature of the material at the outlet of the channel at different ice bath temperatures, and ultimately determined the operating window. The main consideration in ratio optimization is to suppress side reactions. An excess of sodium nitrite leads to an increase in by-products, while a deficiency results in insufficient conversion rate. Through a series of comparative experiments, the appropriate ratio range was determined. Residence time is a key parameter in continuous flow processes. It’s too short, resulting in insufficient reaction; it’s too long, and it’s pointless in microchannels—and for a rapid reaction like diazotization, a long residence time isn’t necessary anyway. By varying the feed flow rate and channel length, a set of residence time scans was conducted to determine a suitable range. III. Pilot-scale continuous operation: Identifying engineering issues. After optimizing the conditions, an integrated pilot-scale continuous system was developed—one that covers feeding, mixing, reaction, and discharging, with the entire process connected seamlessly. Then, following the approach of issue 25, a 72-hour continuous operation was carried out. These 72 hours revealed many issues that were completely unforeseen in short-term experiments. One is crystalline blockage. Diazonium salts precipitate from the solution at low temperatures; although we took this into account in advance, the actual rate of precipitation and the location of aggregation did not exactly match our expectations. At one connection point, local low temperatures – due to inadequate insulation – led to the formation of crystal \"aggregation points\"; after operating for over ten hours, this point began to get blocked, causing fluctuations in flow rate. Later, the insulation solution was improved, and additional insulation measures were added at the joints where heat dissipation was easy, which resolved the problem. One is the fluctuation in feed ratio. Although metering pumps have high precision, over long periods of operation, tiny bubbles or impurities in the inlet pipeline can interfere with the pump’s normal suction process, causing the actual flow rate to deviate from the set value. This problem does not exist at all in batch experiments—you just measure it with a measuring cylinder and add it. However, during continuous operation, minor feedstock deviations accumulate, leading to drift in reaction selectivity. Later, a degassing device and a filter were installed at the pump inlet, significantly improving feed stability. Another issue is insufficient cooling capacity. During the 72-hour operation, there was one day with high temperatures; the cooling capacity of the ice bath was insufficient, causing the reaction temperature to rise. This indicates that the cooling system design margin of the pilot plant is insufficient, and problems may arise in summer. This discovery later had a direct impact on the design of the cooling system for the pilot plant, leading to an increase in the cooling capacity at the first stage. IV. Pilot-scale scale-up design: From concept to engineering. Once the pilot tests are successful and sufficient data has been collected, the pilot-scale scale-up design phase begins. The core challenge in pilot-scale testing lies in the selection and scaling of microchannel reactors. The scaling logic for microchannels is completely different from that of traditional reactors – instead of enlarging the channel dimensions in proportion, scale-up is achieved by increasing the number of channels. This means that when selecting a pilot-scale version, it is necessary to take into account factors such as the flux per channel, the total number of channels required, and the even distribution of flow paths. We spent considerable time engaging in technical discussions with several microreactor suppliers, and used pilot test data to have them develop proposals. In the end, the suitable microchannel model along with the accompanying temperature control, feeding, and discharging systems were selected. The plotting of the pilot-scale PID is much more rigorous than that of the lab-scale setup. All control circuits – temperature control, flow ratio control, emergency shutdown – are designed in accordance with industrial standards. Especially the interlock system: in the case of a temperature anomaly during the diazotization reaction, it is necessary to stop the feed and activate emergency cooling within an extremely short time. This interlock logic has been carefully reviewed and verified through simulations. V. Pilot Plant Construction and Operation The construction and operation of the pilot plant followed essentially the approach outlined in Issue 27. In terms of construction standards, instrument accuracy and safety interlocks must not be compromised—this is the minimum requirement for the quality of pilot plant data. Pipeline welding quality is also strictly ensured in accordance with relevant standards – for pipelines exposed to high temperatures or highly corrosive media, failure to meet welding quality requirements poses not only risks of data disruption but also safety hazards. Building standards are appropriately relaxed, and utility systems make use of the existing plant network wherever possible to control construction costs. During operation, several rounds of tests under different conditions were conducted. Data under normal operating conditions, as well as data under boundary conditions—low load, high load, and various combination of boundary conditions—were all collected. Some amplification sensitivity experiments were also conducted – the key amplification factor in microchannels is the uniformity of flow distribution as the number of channels increases. We carried out multiple comparative tests using a pilot-scale device, and the results generally confirmed the effectiveness of the distributor design provided by the supplier. Of course, the entire pilot process was not without challenges either. Once, due to a change in the batch number of the raw material supplied from upstream, the impurity content differed from before, which caused a significant deviation in the reaction selectivity; it took several days to identify the cause. But these problems that are exposed during the pilot-scale phase are precisely what give pilot testing its value — if such problems are only discovered when scaling up to industrial facilities, the consequences would be entirely different. VI. Data Packets and Subsequent Steps: After the pilot-scale calibration is completed, a complete data packet is formed. The material balance closure rate is within the acceptable range for the pilot plant stage; the operating boundary data cover a certain flexibility range around the design load, and the sensitivity analysis data support the selection of reactors for the subsequent industrial installation. Based on this data package, the project entered its fourth phase – technical finalization – and work began on preparing the process package. Every piece of data in the pilot plant dataset directly supports the equipment data sheets, operating parameter settings, and interlock logic design of the process package. Although the project was not able to go through all stages of implementation due to external investment factors, the complete process of preparing the process package and the accumulation of technical documents provided a solid technical foundation for restarting the project in the future. This also confirms a principle I have repeatedly mentioned: the more stable the progress in the early stages, the lower the cost of any subsequent adjustments. VII. Some Insights Looking back at the technical research and development process of continuous-flow para-phenylenediamine production, there are several insights that stand out. First, in the development of originally innovative technologies, the most difficult part is not the core reaction itself, but rather those problems that do not become apparent during the pilot-scale testing phase and only surface during scale-up and long-term operation. Crystal clogging, feed fluctuations, and insufficient cooling — these problems do not occur at all in batch experiments, but they arise one after another during continuous long-term operation. Finding it early is experience ； Finding it late is an accident. Second, the core value of pilot testing is not the production of products, but the generation of data. Data on the amplification effect, on equipment selection, on operational boundaries, and on safety critical conditions – these factors determine how far the distance is from being \"lab feasible\" to being \"industrially feasible\". Third, for original innovation projects, there is no existing reference experience; every parameter and every boundary condition must be explored on one’s own. In this process, data integrity and proactive risk management are two bottom lines that must not be compromised. Any judgment that something “should be more or less the same” can eventually turn into “very different.” Preview for the next issue: Issue 33 – The conclusion of Phase 3: The seeds of the process package hidden within the data package. The technology development phase spans from Issue 19 to Issue 32, progressing from concept validation to pilot-scale data packages. The next issue will cover the conclusion of the third phase, reviewing the core principles of this phase as well as the relationship between data packages and the next phase – technology finalization and the development of process packages. After the end of the third phase, we will enter the fourth phase, which is the most intensive in terms of content within the seven-phase model.