Tolerance-free series production of high-performance concrete modules through transient-interactive coupling of design and production

提出了一种用于超高性能混凝土模组梁式混凝土结构设计和流水生产的交互概念。其目的是通过耦合克服设计和生产的事实和时间上的分离，以智能和交互的方式补偿几何、时间、工艺和材料方面的不精确，并整合端到端的质量保证。这样，就可以在流水作业中实现高效、低浪费的模块化建设。过程中的不精确通过设计中的排列来补偿，反之亦然。模块结构中的“过短”模块可以补偿“过长”模块--其固定规则只是报废超出了容差限制。因此，这两种方法仍然可以使用。通过准实时的交互排列，通过生产过程随机和瞬时地发生相互补偿。与此同时，这种新发现的灵活性被刻意用于提高生产效率。可控交互作用是指在结构、生产过程和时间上整体积累稳健性的交互作用(瞬时交互概念)。在第一个资助期，重点研究了几何形状随机偏差情况下，不同持续时间的热处理和选择性组装对结构尺寸的收缩影响。这是在不同类别的Y-模和相关的特定尺寸的支柱上进行的，这些支柱形成了静态确定的蜂窝状结构。基于一批又一批调整的热处理持续时间和真实尺寸，开发了一个控制系统，限制收缩并产生几何最优结构。在第二个资金期，将考虑模块的多个静态不确定系统。此外，现场组装的模块将被整合，然后实现设计和生产的预测性交互。在外部，模块形成带有四个节点的墙，而内部则根据荷载将它们分解为拓扑最优的支柱。这就产生了一个可变的、可扩展的模块化系统。交互基于现场模块布局和生产规格的耦合启发式。有目的地，同时考虑单个或多个建筑工地。这种相互作用类似于逐个模块的建造过程，并集成了取决于热处理的短期蠕变和由于开裂而产生的软化行为。该研究方法以无批量多面制造为研究对象，采用强化学习进行交互生产和同步过程控制。该性能将在整体、虚拟和真实的真实规模演示器上进行验证。

An interactive concept for design and flow production of beam-like concrete structures made of UHPC modules is developed. The aim is to overcome the factual and temporal separation of design and production through coupling, that intelligently and interactively compensates for imprecisions in geometry, time, process and material and to integrate end-to-end quality assurance. In this way, efficient and low-waste modular construction in flow production is granted. Imprecisions in the process are compensated for by permutations in the design, and vice versa. A “too long” module - with fixed rules just scrap exceeding the tolerance limits - can be compensated for by a “too short” one in the module structure. And thus both can still be used. Mutual compensation happens stochastically and transiently via the production process through interactive permutations in quasi real-time. At the same time, this newfound flexibility is deliberately used to increase the efficiency of production. Controllable interaction is crated that accumulates robustness holistically over structure, production process and time (transient-interactive concept). The efficiency of the construction method is sustainably improved.In the first funding period, the focus was set on the impact of shrinkage on structural dimensions induced by heat treatments of different duration and selective assembly in case of stochastic deviations of geometry. This was investigated on Y-modules of different classes and associated specifically dimensioned struts which form statically determined honeycomb-like structures. Based on from batch to batch adjusted heat treatment durations and true dimensions a control system was developed that limits shrinkage and yields geometrically optimal structures.In the second funding period, multiply statically indeterminate systems of modules will be considered. Additionally, the assembly of modules on-site will be integrated and then grants predictive interaction of design and production. To the outside, the modules form walls with four nodes while internally they are dissolved into topologically optimal struts according to the loading. This yields a variable, scalable modular system. Interaction bases on coupled heuristics of module placement on-site and specifications for production. On purpose, single or multiple construction sites are considered simultaneously. The interaction proceeds in analogy to the module-by-module construction progress and integrates short-term creeping dependent on the heat treatment and softening behavior due to cracking. The research approach aims on lot-free multi-facetted manufacturing employing reinforcement learning for interactive production and simultaneous process control. The performance will be validated on holistic virtual and physically real demonstrators true to scale.