Article on effectiveness of a DPRK attack on the ROK. Artillery attack Seoul metropolitan area, logistics required, attrition rates (combat and operational attrition) examined. Long but worth a skim for the big points.
Table 1: Summary of Effects
1) No indications there are plans for these events;
2) Assumes most people are at home or in an office i.e. more protection than standing outside in an open field.
Surprise Volley (Primarily counter-force i.e. barracks, military bases)
~2,881 initial volley; mainly soldiers
240 MM MRL170 MM KOKSAN
Surprise Volley (Countervalue and a-strategic i.e. firing directly into population centers)
~29,661 Civilian; likely~790 Foreign nationals~605 Chinese
240 MM MRL170 MM KOKSAN
Counterbattery and counterforcemissions.Very few 240 MM or 170 MM KOKSAN would exist after 1 week.
Expect DPRK to lose these weapons at 1%/hour based on historical rates.
467 ROK aircraft; Possibly 1,200 U.S. aircraft 2,660 Main Battle tanks 1,538 Multiple Rocket Launchers Note: These forces already exist on Korea, or come from Guam and from Carrier Strike Groups in international waters. No need to ask third country permission.
KPA would likely run out of fuel/ammunition within two weeks. NOTE: another study projects KPA can last up to two months. The point is, once started there is very finite amount of fuel and therefore time left.
DPRK needs to drive approximately 2,500 soft-skinned vehicles per day to supply a southward invasion in order to sustain themselves – or spare ROK fuel stores and scavenge from ROK
Conventional Artillery Attack of Seoul
Here is the summarized table of results and what follows below the table is a more detailed description of deriving these numbers.
Table B-1: Conventional Artillery Attack
KPA primarily counter-force
~ 2,811 fatalities initial volley.~ 64,000 first day (majority in first three hours)~ 80,000 one week. Very few KOKSAN and 240 MM MRL last more than one week
2/3 of batteries firing max rate for 5 minutes from likely positions between 5 and 10 km north of DMZ and then sustained rate for ½ of batteries for 24 hours. Batteries destroyed by direct, indirect and counter-battery fire at about 1%/hour. Unrealistic assumption of unlimited ammunition and 100% maintenance rate.
KPA counter-valueLikely indicates KPA desperation
~29,661 fatalities initial volley.Within the range of a previous study by Bennet, Bruce 
2/3 of batteries firing indiscriminately intoSeoulfrom DMZ trace. Most residents at home or office.
Lots of data and looks reasonable, well cited.
And a Time article on evacuation of non-combatants:
There are different accounts of how long the DPRK war machine could operate (fuel, ammunition, attrition absorbance, supplies, etc...) but methinks they could probably operate for a month without resupply from China/Russia. That probably gives the DPRK a few days of advantage (if it generated all of its forces and poured its entire national resources into the attack) before the first waves of the Coalition begin to augment KFOR and begin the push back.
Destroy the ports, airfields and basically all telecommunications as you invade SK and you might be able to stymie a data-dependent force with a high logistical footprint.