Let f be a ring homomorphism from B onto D, such that the kernel is prine. Hence D is also an integral domain. Restrict f to Bu, giving a ring homomorphism from a purely transcendental extension into an integral domain. Call the image Du, and note that this is a finitely generated K algebra. Apply noether normalization, and find indeterminants y1 through yl mapping faithfully into D, while yl+1 through yn map to 0. This is the map from Bt onto Dt, with Bu integral over Bt and mapping to Du integral over Dt, and B integral over Bu mapping to D integral over Du. In other words, D is a transcendental extension of degree l followed by an integral extension, followed by an integral extension. We know that such a ring has dimension l.
Build a chain of l+1 primes in D, starting with 0 and ending with a maximal ideal. This pulls back to a chain of prime ideals in B, starting with the kernel of f and ending with a maximal ideal.
How about the prime ideals inside the kernel? Start with 0 and adjoin yl+1 through yn-1, one at a time. This builds an ascending chain of prime ideals inside the subring Bu. Turn this around so that it is a descending chaing of prime ideals. The largest prime ideal is generated by yl+1 through yn-1, and the last prime ideal is 0. Remember that B is integral over Bu, and is an integral domain. Lift this descending chain to a descending chain in B. All these prime ideals are properly contained in the kernel of f. This because the kernel is a prime ideal lying over yl+1 through yn.
Combine the chain containing the kernel and the chain inside the kernel to build a chain of length n+1 inside B. Since the dimension of B equals its transcendent degree, this is the longest possible chain. It has to run from 0 up to some maximal ideal. Every prime ideal in B participates in a chain of length n+1.
You might be wondering why B has to be an integral domain. The chain of primes inside the kernel was lifted up to B in descending order. Descending chains are harder to lift than ascending chains. The larger ring, in this case B, must be an integral domain, and the smaller ring must be integrally closed. In this case the smaller ring is K adjoin several indeterminants, which is a ufd, which is integrally closed.