The short answer is: it depends.
Some journals have a requirement or a guideline to have experimental results that corroborate the numerical simulations. For example, IEEE Transactions on Microwave Theory and Techniques rarely publishes papers that do not have experimental results in one form or another. However, this is very different for IEEE Transactions on Antennas and Propagation which certainly encourages experimental results, but has many more purely numerical papers. Some topics in this journals overlap, so, sometimes the decision to publish in one vs the other may come from how convenient it is to do/find/use certain experimental results in your paper.
Therefore, one has to look at the common practices both within the field and the particular journal. It is very useful to look at the journal Guidelines for Publications and Editorial Guidelines if those are publicly available.
Certainly, comparing your numerical results to an experiment is a strong (but not necessarily deciding) argument, but not every paper benefits from such a comparison equally. Also, there are things you can do instead of actually doing the physical experiment:
- find if there are prior publications that already have some experimental data. Sometimes, you have to refocus the application of your numerical simulation to the particular case, where the experimental results are available. Reference appropriately. Respect the copyright.
- compare the numerical results with another simulation done with a completely different numerical technique, potentially, with a commercial solver.
- provide evidence that physical laws are not violated (energy balance, various conservation laws, etc)
- do an extra effort on the numerical studies: extra careful convergence study, simulations with different parameters, stability analysis. Sometimes, such techniques as method of manufactured solutions can be helpful in that regard.
- if the numerical technique is applied to a certain level of approximation, convince the reader that the conditions of such approximation are met, and the solution is correct when the more general law is applied.
- making data and the code open-source can also help, and sometimes encouraged by both academics and publishers.
Certainly, it would be hard to publish a paper without experimental data in a journal that explicitly wants them to be present. However, you can do a lot to make your numerical claims stronger while staying purely numerical.