This is not an answer. But I will be interested to know wether anyone has one. We searched for it without any success. Yet the number of vascular bundles in an A thaliana petiole is 5 (but certainly more phloem tubes). In a leaf, it might be far more complex, since vascular bundles are branching to become smaller and smaller with a decrementing number phloem tubes.
I don't know the answer to this, but can comment that it is sometimes quicker to cut sections and count for yourself than to delve through the literature. A very detailed anatomical study would take some time, of course, as Robert Degli Agosti has correctly explained "vascular bundles are branching to become smaller and smaller with a decrementing number [of] phloem tubes". Nevertheless, your need appears to be specific enough that a day's work might suffice.
In response to JH Martin Willison. True, in fact there are published electron micrographs of transversal section of Arabidopsis leaves, in which one can see all the cells that form the vascular bundle. The problem is to distinguish the sieve elements that form the phloem tubes from phloem parenchyma, companion cells, etc. Is there a way to tell, just by the morphology/location?
I hadn't looked for images of Arabidopsis leaves/petioles before posting my earlier comment and I have probably underestimated the difficulty of this task. I found the following images after a brief search using Google image and key words 'phloem' and 'arabidopsis':
Based on these, I think you could get a rough estimate using this method, but it will take some practice both with sectioning and cell identification. If you have access to a scanning electron microsope and a helpful technician, you should have less difficulty with identifying cell types, but the cost will be higher (assuming your university recovers costs for work of that sort in the same way that it is done in North America).