Dimensional accuracy in powder bed 3D printing of concrete is strongly influenced by binder distribution, and the resulting geometric deviations can be direction-dependent. This study examines how voxel-wise water dosage influences geometric fidelity and deviation anisotropy. Experiments show that small changes in water content can cause large, systematic deviations, including edge rounding and swelling. We quantify these effects using high-resolution stereophotogrammetry, aligning as-built scans with CAD models. We then compute deviation metrics such as point-wise distance errors and volumetric differences across multiple water-dosage settings, revealing repeatable, directionally biased deformation patterns that intensify with higher water content. Mechanical testing indicates that stiffness and strength change only marginally, with no clear trend in the tested range. This is explained by excess voxel water diffusing into surrounding powder, leaving the effective water-cement ratio largely unchanged. Finally, we demonstrate a design-compensation concept that pre-adjusts digital geometry to counter predictable deviations, improving accuracy without post-processing.

翻译：粉末床混凝土三维打印的尺寸精度受粘结剂分布的显著影响，所产生的几何偏差可能具有方向依赖性。本研究探讨了体素级用水量对几何保真度与偏差各向异性的影响。实验表明，水含量的微小变化会导致显著的系统性偏差，包括边缘圆角化和膨胀。我们采用高分辨率立体摄影测量技术，将实际打印扫描与CAD模型对齐，量化了这些效应。随后计算了多个用水量设置下的点对点距离误差和体积差异等偏差指标，揭示了随水含量升高而增强的、具有重复性和方向偏好的变形模式。力学测试表明，在测试范围内，刚度和强度仅发生微小变化，且无明显趋势。这是因为过量体素水扩散到周围粉末中，导致有效水灰比基本保持不变。最后，我们展示了一种设计补偿概念，通过预调整数字几何来抵消可预测的偏差，从而在不进行后处理的情况下提高精度。