Hello 

I have a Exchange 2013 DAG scenario

Production Site:

CAS01

CAS02

MBX01

MBX02

DR Site:

CAS03

MBX03

The MBX Servers are in a 3 member DAG. I am currently testing a complete failure of the Production Site by using the Datacenter Switchover approach. Since it is a 3 member DAG, by loosing MBX01 and MBX02 simultaneously causes the DAG to break. My DAG is in DAC mode.

I have been using this article:

https://technet.microsoft.com/en-us/library/dd351049(v=exchg.150).aspx

Is this the correct approach to use for this scenario???

Correct. With not enough 'votes', then the DAG falls down. This is why we recommend to have a 'symmetrical' datacenter setup.

Don't forget the 'datacenter switchover' PPT. http://blogs.technet.com/b/exchange/archive/2012/10/19/exchange-2010-datacenter-switchover-troubleshooter-now-available.aspx

If I have a symmetrical datacenter, for example 2 MB Servers in Production Site and 2 MB Servers in DR, then I would need to have a third site with a File Share Witness???

If I don't have a third site, where can I put the File Share Witness?

Can I have a File Share Witness in each site? Is that possible?

Also,  

in my first example with 3 MB Servers where 2 is in Production and 1 is in DR, Can I place a File Share Witness in the DR Site to  maintain the DAG if the Production site goes down?????

The idea of having a 3rd site for the FSW is that it is on the 'outside' looking in. 

The FSW is contacted IF a tie needs to be broken. So in your 2 + 2 server setup, the FSW would need to be contacted if 2 of the servers fall down. And if the FSW is in the primary data center and all go down, the 2 servers in DR try and ask the FSW to see if there are enough votes for quorum. Since the tie break of 2 vs. 2 needs to be answered, the FSW needs to answer, but is not available, and thus, the 2 DR servers go down.

(IF the 2 DAG nodes in DR fall down, then the 2 in the primary data center with the FSW stay up and running.)

The 'alternate' FSW is just that, an alternate. It is not a primary/secondary situation like DNS search orders. It is primary and alternate. Use it during a manual data center 'star' over. Which is that PPT document to walk through the steps. 

2010 had the limitation that the FSW was in the same AD site as a DAG node. 2013 lifted that restriction and allows a 3rd AD site, on the outside looking in, to see that the DAG nodes are up and running. 

So yes, we want you to have a FSW in each primary and DR site. But the DR site will not automatically activate. Still a limit of the DAG design. Hence, try and use a 3rd AD site and/or Azure to have seamless fail over without human intervention. 

Makes sense?