Evaluation of Sequential Testing Strategies for Directional Compressive Strength of 3D-Printed Concrete Using Interlaboratory Data

The compressive strength of 3D-printed concrete (3DPC) varies with the loading direction, and the lowest-strength direction is not fixed across printing groups. Testing in a limited number of directions may miss the governing direction and bias the acceptance decision toward false acceptance. This study used the RILEM TC 304-ADC interlaboratory database comprising 27 laboratories, 34 printing groups, and 907 directional compressive test specimens. Exact sampling-without-replacement enumeration and leave-one-laboratory-out (LOLO) cross-validation were used to evaluate the relationships among directional coverage, false acceptance, and specimen consumption. The observed governing directions of the 34 printing groups were distributed across the U, V, and W directions. At an operational threshold of 40 MPa, the observed false-acceptance rates of the single-direction V3 and two-direction VW6 strategies were 13.7% and 10.6%, respectively, compared with 4.9% for the complete-direction UVW9 baseline. The early-acceptance strategy reduced the expected specimen count from 9.00 to 4.97, a reduction of 44.8%, but the observed false-acceptance rate increased from 4.93% to 9.15%. The risk difference was 4.22 percentage points, exceeding the 2.5-percentage-point screening criterion. Under the restricted-direction strategy, partial-direction results could trigger only early rejection, and final acceptance required complete U, V, and W directional evidence; the strategy therefore produced no acceptance decisions beyond those of the UVW9 baseline. In the present evaluation set, the restricted-direction strategy yielded group-by-group decisions identical to those of the UVW9 baseline, with an expected specimen count of 6.49, a reduction of 27.9% (95% interval 16.8% to 39.9%). Adaptive direction ordering saved 0.47 specimens relative to the original fixed order, and a simple fixed order that tests the most frequently governing direction first achieved the same saving; the overall savings arose primarily from printing groups that did not meet the strength requirement. Overall, the results show that cross-directional historical information can optimize the testing order and identify nonconforming printing groups earlier, whereas final acceptance should continue to require complete directional evidence. In production settings where staged specimen preparation and testing are feasible, this procedure could reduce the cost of destructive testing and bring forward the review and process correction of anomalous printing groups.