The processes block defines the measurement process itself. You can divide the process by a number of processes and in each process you can set local variables. Each process must have a name that can then be used in calculating the overall uncertainty in the calculation section.
<processes>
<process name="v1">
<variable name="time"> start 10+</variable>
...
<formula>s1 time+</formula>
</process>
<process name="v2">
<variable name="time"> start 20+</variable>
...
<formula>s1 time+ </formula>
</process>
...
</processes>
In the example we define a variable time, which is calculated by the addition of a global variable from the calculation section called start with 10 for process v1 and with 20 for process v2. If the variable time is used in a defintion of a distribution in an instance from the instances section, the value of time of the current process will be used for the evaluation. After all instances returned their results, the formula of the process is evaluated. In the example an instance s1 and the local variable time are added for the result of process v1 and v2 respectively.
It can be that the name of a variable is used in more than one basic model and they overlap and should have different values. In processes you can model that as you can see in the following example:
<instances>
<instance name="v1" model="volume">
<parameters><parameter name="#tmp">temp</parameter></parameters>
</instance>
<instance name="v2" model="volume">
<parameters><parameter name="#tmp">temp</parameter></parameters>
</instance>
</instances>
<processes>
<process name="step1">
<step name="v1">
<variable name="temp">15.2</variable>
</step>
<step name="v2">
<variable name="temp">16.1</variable>
</step>
<formula>v1 v2+</formula>
</process>
</processes>
<calculation>
...
</calculation>
Here both instances of the basic model volume use the variable temp, but in process step1 they should have different values. That can be accomplished using steps inside of process definitions. A step defines variables for a specific instance used in the formula.