The following code will work perfectly.

#include <reg51.h>

#define Xtal 22.1184       // = 22.1184
#define tic12 (12/Xtal)    // = 0.54253...
#define _10msec (unsigned)(65536 - ((1000.0 * 10.0) / tic12))
//=                        65536 - ( 10000.0        / 0.54253...)
//=                        65536 - 18432.0
//=                        47104 = 0xB800

void main (void)
{
float aaa;

aaa = _10msec;

TL1 = _10msec; 
TH1 = _10msec >> 8 ;

while(1);
}

However...

i'm a bit confused, i recently started working with eval version of keil.

The eval version of the Keil tools does not include floating-point support.

Jon