I'll give a two part answer.
1. Regarding how to do something like this in En-ROADS, En-ROADS assumes there is a rate of energy efficiency improvement. That means that, overall, society uses less energy to produce a unit of GDP than it used to produce the same unit of GDP the year before. You could approximate 5G by reducing the Energy Efficiency slider a bit to show that 5G infrastructure would not be as energy efficient. The exact value of the slider would be something you'd have to determine.
2. BUT... This is something to watch for but I feel like I've seen these headlines before. A good example that I remember is the energy use for another tech: Bitcoin. <https://arstechnica.com/tech-policy/2017/12/bitcoins-insane-energy-consumption-explained/>.
It isn't that bitcoin, 5G or another tech won't use a lot of energy or society shouldn't be aware of possible rampant energy consumption. What gives me pause is when today's computing technology is forecast decades out into the future, with no change. We know that simply isn't how tech works. And often the studies have no limits involved, like declining ROI of bitcoin mining, etc.
I don't know much about 5G, but would watch for the towers and router equipment to shift to ARM processors and other lower wattage architectures. Network operators are always on the lookout for ways to reduce their operating costs, including energy use.