The seven stages of chemical engineering technology from concept to industrialization (Issue 29/100) -- Pilot-scale calibration and data package

Thread Content

This post was last edited by xiouxingzhe on 2026-6-11 at 11:47. The seven stages of chemical engineering technology from concept to industrialization (Issue 29 out of 100) — Technology development: Pilot-scale testing and data packages. Dear colleagues: Hello everyone! In the previous issue, we discussed the sensitivity amplification experiment and identified the controlling physical quantities. In this issue, we discuss the most important deliverable in the pilot phase: data packets. The pilot plant is now running stably, and a considerable amount of data has been collected under various operating conditions. But these data are scattered, found in the operation logs and gauge records of various positions. You need to systematically organize, calculate, and compile them to create a complete technical document that can be used in the next stage. The quality of this document directly determines the efficiency and reliability of preparing the Phase 4 process package. With well-structured data packets, creating a process package becomes a natural next step. If the data packages are not prepared properly, you will find that many fields in the process package cannot be filled in – the heat load cannot be specified clearly, there are no defined limits for the operating range, and it’s impossible to trace the origin of key parameters. As a result, you have to go back and fill in the missing data, which comes at a high cost. I. When to start calibration? Calibration is not something that is done only at the end of the device’s operation. Many people understand calibration as \"conducting a summary after completion,\" but the risk of this approach is that by the time you realize there isn’t enough data, that a certain operating condition wasn’t recorded, or that a particular parameter wasn’t captured, the device may already have stopped working, been disassembled, or its operating conditions may have changed. Recovering the data later comes at a high cost. The correct approach is to develop the calibration plan before starting the pilot operation. During the operation of the pilot plant, data is collected synchronously in accordance with the requirements of the calibration plan. After the operation is completed, the calibration data are organized, calculated, and compiled. The three actions are performed in one smooth motion, without any breaks. The core of the calibration plan is to answer several questions: which operating conditions need to be calibrated – design conditions, boundary conditions, and part-load conditions ； How long to collect data for each operating condition – it’s not determined arbitrarily, but based on the fluctuation cycle of key indicators. If the fluctuation period of a metric is about half an hour, then the collection duration must cover several complete cycles in order to calculate meaningful average values and standard deviations ； List clearly the parameters to be measured for each operating condition – temperature, pressure, flow rate, composition, and energy consumption ； Who will be responsible for data collection and verification—there must be a person accountable for the data. II. Material balance: The closure rate is a key indicator. Material balance is the most fundamental and essential aspect of pilot-scale calibration. Add up all the feed materials, and also add up all the output materials—products, by-products, waste materials, and unreacted recovered materials. The difference between these two values, divided by the total amount of feed material, gives the unclosed error. What percentage closure rate is considered acceptable? It depends on the device type and complexity. For a system with a single reactor and simple separation, the recovery rate should be over 98%. For complex systems that combine multiple reactors with distillation and recovery, a performance level of over 96% is considered satisfactory, while a level of over 98% is considered excellent. For systems involving solid-phase reactions or emissions, achieving over 94% is already quite good. If the closure rate is below the acceptable standard, the reasons must be investigated. My general order of investigation is as follows: first, I check the accuracy of the analyzed data—retesting the same batch of samples using different personnel or different methods to see if the results are consistent ； Check again whether the sampling is representative – whether the locations of the sampling points and the operating conditions during sampling are the same as those during normal operation ； Then check for any materials that have not been accounted for – micro-leaks from safety valves, sampling discharge, and pump seal flushing can all cause material loss ； Finally, consider whether there are any unrecognized side reactions consuming the raw materials—this could be the most important discovery, as unrecognized side reactions indicate gaps in your understanding of the process. After completing the material balance sheet, every figure must be traceable to its source. The instrument tag number, data acquisition time, and the operating conditions at that time – once this chain of traceability is broken, the credibility of the data is compromised. For example, if the data sheet states “reaction temperature: 180±2°C”, but when you ask whether 180 is a set value or a measured value, which instrument is used for measurement, and when that instrument was calibrated, and if the answers are all “I can’t remember”, then this data is unreliable. III. Energy balance: It’s not just about calculating the overall figure. When it comes to energy balance calculations, many people only focus on the overall figures – how much heat is absorbed, how much heat is released, and what the degree of closure is. This certainly needs to be done, but it’s not enough. For energy balance, it is more important to conduct calculations on a per-unit basis. The reactor is accounted for separately – are the heat released and the heat removed balanced? Separate accounting for the distillation tower – are the heat loads of the reboiler and the condenser matched? Heat exchanger accounted for separately – what is the difference between the measured overall heat transfer coefficient and the design value? Unit-based accounting can reveal many detailed issues. For example, if the measured heat transfer coefficient of a heat exchanger is 30% lower than the design value, this could be an early indication of scaling. The heat load on the reboiler of a certain distillation tower is significantly higher than expected, likely due to improper control of the reflux ratio. The energy balance data for these sub-units serve as the direct basis for equipment data sheets and the optimization of process parameters. IV. Waste data: Waste data are not by-products; they constitute an independent and important part of the pilot-scale data set. Starting from the PFD, identify all the emission points for waste gas, wastewater, and solid waste one by one. For each emission point, it is necessary to determine which device or operational step is the source of the emissions ； What is the emission volume? Is it continuous or intermittent emissions? ； What is its composition, and what are the concentrations of the key pollutants? ； What are the exhaust temperature and pressure? These data directly influence the preparation of subsequent environmental impact assessment reports and the selection of environmental protection facilities. Environmental impact assessments have a significant impact on businesses nowadays; not only must the overall quality be balanced, but it is also necessary to clarify the fate of key elements – which materials contain sulfur, where it ends up, how much goes into wastewater, how much enters exhaust gases, and how much remains in the products. If you fail to collect all this data during the pilot phase and are questioned about it during the environmental impact assessment, you will have to guess and fill in the gaps, which carries a high risk. V. Quality Standards and Delivery List for Data Packs The quality of pilot-scale data packs cannot be judged simply as “good” or “bad”. It needs to be systematically evaluated from several dimensions. The material balance closure rate must meet the required standard; this is the first key indicator. The data traceability chain is complete; every piece of data can be traced back to its source – the instrument, the time, and the operating conditions. This is the second dimension. The operational boundary data is comprehensive; it includes not only the data for the optimal operating points but also data on the boundary conditions – the temperature boundaries cover a range of ±5 to 10 degrees around the optimal temperature, while the load boundaries cover a range of 50% to 110% of the design load. This represents the third dimension. The data on the scaling effect is complete; from pilot scale to pilot plant scale, there is a quantitative comparative analysis of which parameters are \"scaling sensitive\" and which are \"scaling insensitive\" – this constitutes the fourth dimension. A qualified pilot-scale data package should include the following: overall process material balance data, energy balance data for each unit, operational parameters and performance data for all key equipment, data on catalyst performance and lifetime assessment, complete data on waste emissions, a summary of material corrosion issues, scale-up sensitivity analysis, a summary of operational boundaries and flexibility, and a compilation of safety critical parameters. Each data sheet must include the person who prepared it, the person who verified it, and the date, to ensure traceability of responsibilities. When data packets are delivered, they need to undergo internal review. The compiler first checks each item individually, followed by cross-verification by someone from the same field, and finally a comprehensive technical review is conducted by an experienced auditor. Once approved, the data package is officially frozen – subsequent development of the process package will use this frozen data package as the sole source of data. VI. Relationship between pilot-scale data packages and industrial-scale test data packages As mentioned earlier, not all projects require industrial-scale testing. If the process risks of the project are low to moderate and the pilot test data is sufficiently comprehensive, the pilot test data package serves as the direct input for preparing the process package in the fourth phase. However, if the project requires industrial-scale testing – such as for entirely new processes, when the scale of scale-up is very large, or when there are explicit regulatory requirements – then the pilot plant data represents only interim results, and further verification and supplementation are needed through industrial-scale testing. The data packages for industrial-scale tests have a structure similar to those for pilot-scale tests, but the data are closer to actual industrial conditions, providing more robust support for the development of process packages. Regardless of the number of verification steps involved, the data packets delivered in the fourth phase must meet one basic requirement: the data quality must be at a level sufficient to support the preparation of the process package. The data packet does not meet the requirements and cannot proceed further. Preview for the next issue: Issue 30 – The four quality dimensions of data packets. In the previous issue, it was mentioned that the quality of data packets needs to be evaluated from several dimensions, but due to space constraints, this topic was not explored in detail. The next issue will be dedicated to this topic: how to determine the quality of a data packet. What percentage closure rate for material balance is considered acceptable? What are the criteria for a complete data traceability chain? What does it take to consider boundary data handling sufficient? What should the magnification effect data include? It will be detailed in the next issue.