Study of the simple model Fig. 
reveals several
characteristic properties of an RSFQ biasing line which we believe to
be true for any large (1- or 2- dimensional) periodic
design:
1. If 
long-term slow-decaying negative offsets in bias
current (negative tails in Fig. ) can accumulate over time
and render the entire design inoperational, independently of
its biasing voltage. This effect has a simple physical interpretation:
energy of the induced perturbation is stored mostly in the large
superconductive inductances L of the biasing line and diffusively
(as some power of 
) spreads all over the design.
2. If 
accumulation of negative offsets is finite and given by
the ``worst case'' estimate ().
3. Simultaneously, however, short-term ( 
) positive offsets
become more important (positive exponents in Fig. ) if we
have and () gives the lower limit for l. Choice of
an even bigger value of l (say, 
) further reduces the
mutual influence of the junctions via the power distribution
line. This again is independent of the value of r.
4. Significant (of the order of 10) decrease in power dissipation can
be achieved by reducing the bias voltage with a controllable narrowing
of the bias margin (given by Eq. ()) as a trade-off. At the
same time, it affects the ``worst case'' estimate () which is
very similar to ()).
Effects 4 and 3 can be studied in simulation and when a design is optimized for lower power dissipation inclusion of the full biasing line into simulation and optimization of its parameters is essential. Effects 1 and 2, however, cannot be studied in simulation in the usual way (with PSCAN [10, 11]) since they occur in large designs operating over long periods of time.