fake data was being written to the binary file. if a link element is nominally closed, even if there's actually flow calculated there, the flow is disregarded upon writing results. so when the flow is read back in, it's zero -- which is different than it would have been if it had stayed in memory. i wonder what that does to mass balance?

this version works, at least with Net3, simplenet, and SampleTown -- but the results are non-canonical, as the current workaround for the issue above writes tiny flow values to the binary file for closed elements. presumably we should check for closed status in the transport() function.

src/hydraul.c

@@ -258,6 +258,9 @@ int  nexthyd(long *tstep)
    else
    {
       Htime++;          /* Force completion of analysis */
+     if (OpenQflag) {
+       Qtime++; // force completion of wq analysis too
+     }
    }
    *tstep = hydstep;
    return(errcode);
@@ -1049,13 +1052,8 @@ void  tanklevels(long tstep)
 
       /* Update the tank's volume & water elevation */
       n = Tank[i].Node;
-
-      // only adjust tank volume if we're in sequential mode.
-      // otherwise, the tankmixing function will do it for us.
-      if (!OpenQflag) {
-        dv = D[n]*tstep;
-        Tank[i].V += dv;
-      }
+      dv = D[n]*tstep;
+      Tank[i].V += dv;
 
       /*** Updated 6/24/02 ***/
       /* Check if tank full/empty within next second */