J. N. Shirlaw, Land Transport Authority, Singapore The authors have provided a very interesting paper, with a case study, laboratory testing and finite element analysis of the effects of grouting during EPB tunnelling. However, little detail is given on the monitoring and tunnelling in the case study, and further information would help to provide a fuller record of their work. For the monitoring, were settlements measured using an array orthogonal to the direction of advance, to measure the width of the consolidation settlement trough? Based on my experience in Singapore with grout injected after tail void closure, I would expect the settlement troughs during consolidation for arrays B and C, and possibly A, to follow an ‘error function’ shape with a width similar to that of the initial settlement trough. Consolidation settlements due to this type of grouting have the effect of delaying the immediate settlement due to tunnelling, but the final effect is the same as immediate settlement. Using an i value of half the depth to tunnel axis, this would imply that the total volume loss due to tunnelling was 5·7%, 7·5% and 7·9% for cases A, B and C. These are relatively high values, but would be consistent with the closure of a gap of about 60 mm all around the tunnel. It would be useful to know the size of the tail void (the diameter cut by the machine minus the external diameter of the lining), to allow comparison with these volume loss figures. The volume loss figures given above are high for an EPB machine operating in a firm clay. The stability number with no support pressure is 3·49. The total settlements recorded are high for such a stability number, even disregarding the support provided by the EPB machine. It would be useful to know the range of face pressures used during tunnelling, and whether the face was over-pressurised, which could lead to remoulding of the very sensitive clay. The volume loss figures given above can also be expressed as volume losses of 0·405–0·558 m=m of tunnel. Trials B and C involved the injection of 3·8 and 6 m of grout. Comparing these figures, I presume that the conventional tail void grouting was discontinued for several metres of the tunnel, but it is not stated over what length of tunnel the tail void grouting was replaced by the alternative grouting. This is important, because it appears that the settlement over Trials B and C was not just related to the effects of the trials. The surface settlement over the two trials would be due to the cumulative effects of tunnelling from a point about 21 m before the settlement point to about 21 m beyond the settlement point. It would therefore appear that the surface settlements over trials B and C would have included effects both from the trial injection and from the more conventional grouting as used in Trial A. I would note that the ‘immediate’ settlement identified in the paper does not allow for the time necessary for the full development of the immediate settlement trough, owing to this threedimensional effect. The conventional tail void grouting appears to have been more effective than the trial injections at controlling total surface settlement. However, the settlement was still large for this size and type of machine operating in firm clay. The authors do not state whether the tail void grouting was carried out using grout pipes laid along the tail skin, to allow grouting simultaneously with machine advance, or through the lining. The former of these has been found to be much more effective at filling the tail void before it closes (in soft clays) than the latter. I was very interested in the high compressibility of the cement/silicate grout (Type I) used, and the significant reduction in grout compression using the Type II grout. This is clearly an area that warrants further investigation. However, I would appreciate some clarification on some of the items in the paper. The authors report a 120 s gel-hardening time for the Type I grout, but for the identical grout used in the tunnelling trial report a 20 s gel time. In the consolidation testing of the grouts, the authors report that Type I lost 30% of its original volume ‘after hardening’. I am not clear whether the 30% loss was measured on hardened grout, or during the interval between injection and hardening. I also note that with Type I grout the laboratory trials show an initial volume expansion of very close to 100% of the grout volume used in all but one case. For the Type II grout the initial volume expansion is well below 100% of the grout volume used in three out of seven cases, implying some other losses in the system. The authors also do not state the mix used for grouting in trial A: it would be useful to know whether this grout was also prone to significant loss of volume. As a final comment, I question the normalisation of the grout volume by initial soil volume in Figs 10 and 12. Within the laboratory trials the soil volume is defined by the size of the container. However, for wider application, this method of presentation is of limited value. Did the authors consider expressing the data in terms of rþ 3 r=r, where r is the radius of the initial injection cavity, and rþ 3 r is the radius of the expanding bulb (assuming expansion as a sphere)?