Here's Dewar on models:
A model is a simple mechanism that simulates the behavior of a more complex one. A scientific model must simulate the behavior of the universe, or some part of it, while remaining simple enough for us to understand. The test of such a model is purely operational. Does it in fact simulate the behavior of the system being modeled? If not, we have to modify it or replace it with a better one, better in the sense that it simulates the parent system more effectively. There is of course no question of a model being true or false. The same rule applies to scientific theories, which are simply definitions of scientific models. The question, "Is it true?" is meaningless in science. The correct question to ask is "Does it work?"
That there is one of the most accurate encapsulations of the philosophy of chemistry that I have come across. It enunciates all the main features of good chemical models. Chemistry as a science is based much more on models than physics. Crucially, chemical models must be simple enough to be understood by experimentalists while remaining general enough to describe a wide range of situations. This is one reason why reductionism does not work so well in chemistry on a practical basis. Chemists also are more easily reconciled to the value of models as utilitarian constructs rather than representations of some deep and ultimate reality. Physicists worry much more about such things, and the fact that chemists don't does not make them any less scientific or any less enamored of the pursuit of truth. Chemistry much more than physics is about special cases rather than general phenomena. Chemists certainly care about general explanatory frameworks like chemical bonding, steric effects and electron transfer, but even there the thrust is not necessarily to uncover deep reality but to find something that works. The chemist is the perpetual tinkerer, and it is through tinkering that he or she is led to "the truth".