DigiScale: Digitally driven scale up of chemical processes

化学合成是我们整个社会的基础-无论是在药品、家庭或个人护理用品的生产中，还是在热能和电力的生产中。尽管有几个世纪的经验，但发现新分子并将其投入生产规模的过程仍然是一个挑战。小规模发现的条件可能与将新化学产品推向更广泛社会所需的规模非常不同。缩短这一时间可能意味着更早地将药物推向市场，或为现有产品提供更环保的替代品。数字孪生是在计算框架内的物理过程的表示，允许进行虚拟实验以评估运行过程的最佳条件。这些条件可以选择和加权（例如）最环保的过程，或最便宜的-取决于社会的更广泛的需求。数字孪生也允许动态地解决这个问题。这个项目开发了构建数字孪生的框架。由于所有的化学过程都是不同的，并且在很大程度上取决于特定的合成，我们必须首先了解我们的特定反应。这将在计算机控制下以小规模进行，以缩小该过程的广泛条件范围（例如温度，浓度，用于加速反应的催化剂的选择）。然后，这些信息通知我们的数字孪生模型。在较大规模的情况下，反应堆的条件各不相同-处理数千升的材料与处理发现规模的少量材料非常不同。现在我们有温度梯度，材料混合得更差-这些都可以影响反应。因此，我们利用我们对这些变化的了解，包括运行流动模拟，并将我们的小规模测试的信息映射到上面。我们可以选择我们想要使用的反应器，以及它们的运行条件，并对整个过程做出决定-就像以前一样，通过可持续性，成本，材料纯度等问题来驱动它。这为我们提供了一个框架，通过数字表示跨越材料的物理发展。我们可以通过实验探索比以往任何时候都更广泛的参数空间，并以数字方式评估整个过程。该项目建立在英国用于自优化反应器系统的自动化工具的世界级研究基础上，并拥有一系列制药行业合作伙伴;阿斯利康，辉瑞和UCB Pharma，以及反应器控制软件专家; Perceptive Engineering，确保最大限度地减少行业采用该项目中开发的方法的障碍。

Chemical synthesis underpins our society in general - whether in the production of medicines, household or personal care items or for heat and power. The process of discovering a new molecule and taking it through to production scale remains a challenge, despite centuries of experience. The conditions of small scale discovery can be very different to those at the scales required to bring the new chemical product to wider society. Reducing this time may mean bringing drugs to market earlier, or providing more environmentally friendly alternatives to existing products. A digital twin is a representation of a physical process within a computational framework, allowing virtual experimentation to be carried out to assess the best set of conditions under which to run processes. These conditions can be selected and weighted towards (for example) the most environmentally friendly process, or the cheapest - depending on the wider demands of society. A digital twin also allows this to be addressed dynamically.This project develops the framework for building digital twins. As all chemical processes are different, and depend very much on the particular synthesis, we have to first learn something about our specific reaction. This will be carried out at the small scale under computer control, to narrow down the wide range of conditions (e.g. temperatures, concentrations, choice of catalysts used to speed up the reaction) for the process. This information then informs our digital twin model. At larger scale conditions vary through the reactor - dealing with many thousands of litres of material is very different to dealing with the small quantities at the discovery scale. Now we have gradients of temperature, material is much more poorly mixed - these can all influence the reaction. So we take our knowledge of these variations, which includes running flow simulations, and map onto it the information from our small scale tests. We can select which reactors we want to use, and their conditions that they are run under and make decisions about the overall process - just as before, driving it through questions of sustainability, cost, material purity.This gives us a framework that spans the physical development of material with a digital representation. We can explore a much wider parameter space than ever possible through experimentation, and assess the entire process digitally. This brings unprecedented agility to the manufacturing of products.The project builds on the world-class research that exists in the UK in automated tools used for self-optimising reactor systems and has a range of industrial partners spanning pharmaceuticals; AstraZeneca, Pfizer and UCB Pharma, and reactor control software experts; Perceptive Engineering, ensuring the barriers to adoption by industry of the approach developed in this project are minimised.