Why did my 28-spoke front wheel lost all tension but rear only some?

The cause is the disc brake braking torque. A wheel having a reduced number of spokes should have larger-flange hubs to withstand torque more efficiently and a stiffer rim to withstand radial loads more efficiently.

This wheel lacks the large-flange hub. This rim also is only very slightly stiffer than my DT Swiss TK 540 replacement rims with 36 spoke holes.

I estimated based on the dimension and cross sectional view of the TK 540 rim that its second moment of area is 3283 mm⁴. I also estimated that the slightly deeper 28-hole rim has a second moment of area of about 4500 mm⁴ (this is only very approximate as I do not have a cross sectional view of the no-brand rim). In comparison, Mavic MA2 second moment of area is according to my calculations 1388 mm⁴.

My bicycle wheel tension simulator shows that the 28-hole wheels have 38% of radial load on one spoke, that the 36-hole TK 540 wheels have 30% of radial load on one spoke and that Mavic MA2 wheels have 39% of radial load on one spoke. Thus the TK 540 wheels better withstand radial load than the others.

We also need to calculate how well a bicycle hub shaft transmits torque. My calculations show about 311 Nm per degree of twist for these hubs. Braking torque for 111 kg rider + 20 kg bicycle at 0.6 g deceleration is 261.02 Nm.

The 28-spoke wheels with their 1-cross spoke pattern twist 0.95426 degrees (left) and 0.29978 degrees (right) to generate 157.94 Nm of torque (left) and 103.08 Nm of torque (right). The torque results in alternating tension change of 626.77 N (left) and 409.07 N (right) for the spoking pattern.

This 626.77 N braking tension change can be combined with the fact that when braking hard on the front brake, all of the 131 kg weight is on the front wheel so there's 1285.1 N of load that reduces the tension of one spoke by 491 N. Thus there's about 1118 N of tension loss on the worst-affected spoke, so clearly there is no safety margin in these 28-spoke wheels if the rider is heavy and brakes hard.

This can be compared with 36-spoke TK540 wheels that twist 0.72795 degrees (left) and 0.25080 degrees (right) to generate 165.43 Nm of torque (left) and 95.584 Nm of torque (right). The torque results in alternating tension change of 465.59 N (left) and 269.01 N (right) for the spoking pattern.

This 466 N tension change is added to the 384 N tension loss in largest load-carrying spoke. Thus the worst-affected spoke has 850 N tension change. This can be combined with typical spoke tension of 1200 N, so with these 36-spoke wheels there actually is some safety margin, albeit small.

The conclusion is that when selecting hubs for disc brake front wheels, best hubs have (1) lots of spokes and (2) large flange. It also helps to have a stiff shaft like these hubs do. Disc brake front rims should be of high quality so that even spoke tension can be achieved, and the spokes need to be evenly tensioned to high tensions.

The rear disc brake in comparison provides minimal forces and is rarely used, so it is not a problem, especially considering that the Cannondale AI frame achieves even spoke tension in the rear.

I would say that your wheels were not initially tensioned properly. The load-unload cycle of riding will work to back off the tension on spokes that aren't tight enough.

That you have disk brakes isn't the root cause, though the forces of disks braking act through the spokes whereas rim brakes are already acting directly on the rims. So braking with disk brakes is stressing the spokes the other way, perhaps doubling the load/unload cycle count for the time you're actively braking.

When the wheel is built, the tension should "preload" each spoke sufficiently that it doesn't stretch as much during riding. If the spoke tension is low but equal, then that will eventually lead to problems, often breaking at the J-bend.