I think they will be using non capacitor systems in the Demo Plant as per the quote below from the same article.
“Non-capacitor systems have the benefit of allowing continuous FJH without a necessary recharging step, offering higher energy delivery despite reduced power output— owing to the long FJH timescales permitted (Fig. 1B). These non-capacitor, continuous systems exist in alternating-current (a.c.) and direct-current (d.c.) varieties
Such systems (AC) are commonly used for pretreatment to reduce the electrical resistivity of a sample before subsequent d.c. flashing
Direct-current systems feature a device that rectifies the a.c. input from the facility to d.c. output to the sample and are the most common type of continuous flash Joule heater”
The article also says (I have shown figure 1h below - a multiple chamber version):
"a US$530 continuous FJH reactor was used to achieve 5 kg h −1 flash graphene synthesis from metallurgical coke at kilogram-scale batch size . Although lab-scale equipment is usually designed for manual operation, mechanical operation was integrated for continuous feedstock loading, FJH and product unloading for reported scale-up prototypes (Fig. 1h)"
I haven’t seen the lab scale figures presented to DARPA, do you have a link?
Interestingly the diagram below shows their lab scale continuous loading system that has 8 chambers
(20min delay)