Breaking the simulation barrier

Hydraulic simulation has been around for over 30 years and in some areas is standard practice to ensure performance targets are met. However, in the wider industry it is still rarely used. There appears to be an uncross-able barrier with savings it provides in reduced development time and risk on one side against the time and cost it takes to perform simulations on the other. This article looks in detail at the issues that contribute to this barrier and discusses a new approach for how they can be overcome.

I first saw the power and potential for hydraulic simulation at Bath University in the early 1980s. The Fluid Power Centre were already building mathematical models of hydraulic components that were good enough to predict the exact pressure ripple at any point within a pipe line. I have used a number of different simulation programs since then, usually with impressive results, but when I look around the hydraulics industry today there is still only very limited adoption of computer simulation techniques.

Calculations to design circuits and predict hydraulic performance have always worked and are still a necessity on every new system. Computer simulation is required when these calculations are heavily iterative or dynamic and can no longer be achieved by using MathCad or spreadsheets alone. Prime examples are simple cylinder and load, valve sizing exercises. Several programs exist where you can enter the system parameters as well as the detailed valve dynamic performance and they will accurately predict the system performance in open and closed loop.

These simulations provide vitally important information and because they can be made relatively quickly and accurately, they have become standard practice for servo and proportional valve control system design.

The need for more simulation before build

So why after 30 years of development is it still not possible to check a hydraulic circuit within a reasonable time frame?

If we want to predict the performance or efficiency of a complete hydraulic system with enough accuracy to find issues that a good hydraulics engineer could not predict from experience, then the simplified valve models provided within standard simulation packages are just not detailed enough. It's likely we would need to model a wide range of valves and with enough detail to distinguish the performance of different valve qualities as well as types.

Of course this level of detail is available from tools like the AMESim HCD library. Using this almost every valve can be built from it the individual spool or poppet components however, this level of detailed information is rarely available from the manufacturers and even when it is the final models become too complicated to configure and take too long to run as part of a large simulation.

We need accurate results but without taking incurring on too much extra time and resources. But unlike with simple servo and proportional valve circuits this technology barrier has not been crossed for complete hydraulic circuit design. Modern simulation programs are certainly good enough. They are now very powerful with have all of the flexibility required and are relatively easy to use, however, they struggle to provide better information in less time than a good hydraulics engineer.

It was hoped that valve manufacturers would provide the source data and models from which the simulations could be made. The intention was for users to be able to select the valves they require from a library provided and then simply use them in their models. This does happen to some extent with major OEMs who request simulation models as part of their purchase package. However, there are few signs this will be a route that many will try to follow.

There appears to be an ongoing barrier to using simulation where the amount of work that has to be put in does not justify the results obtained. System models are either too restricted in detail or too time consuming to produce.

Simulating hydraulic circuits is not easy, a lot of data needs to be put in to ensure you get sensible results out. Experienced simulation engineers will tell you that you need a reasonable idea of what you are looking for to stand any hope of finding it. Running general simulations and expecting them results to show you something of which you currently have no idea, is not an effective use of time. It can and does happen but it is not the norm.

For hydraulic simulation to become mainstream we must remove this barrier. We still need the fantastic results simulation can give but this must be possible in the time it takes to build a basic simulation model. After more than 20 years of simulation software development I felt there needed to be a bit of a rethink in the way this software is used.

Working closely with LMS, we had a bit of a rethink about how simulation software could be used. By looking at how we currently design hydraulic circuits we have tried to apply the same strategy to build the simulation models. As with all good solutions we tried to keep the new approach very simple.

Virtually every hydraulic component can be built from a simple combination of an orifice, spring and spool or poppet. We just use these components in different sizes and arrangement to give the functions and performance we require. The functions themselves are relatively easy to and can be defined by 5 different groups e.g. directional, pressure relief/reducing, flow control, pumps and actuators.

So why after 30 years of development is it still not possible to check a hydraulic circuit within hours.

I believed one of the keys was selecting which parameters to use.

When we look at a relief valve for example, the same function can be achieved with a £20 cartridge valve as a £200 industrial valve. Both are designed for purpose but while their components may be similar their performance will be totally different.

So let's look again at how a hydraulics engineer would select the appropriate version for their new machine, using only a product datasheet. Certainly, as we've already mentioned systems containing servo and proportional valve would commonly be simulated and when we look at their datasheets we see all of the information we require. These datasheets contain a dynamic performance curve with all of the frequency response and critical damping information we need. But for more standard products such as relief or directional valves the information is much more limited. In fact, for the lower cost valves they may not wish to mention the less favourable items such as leakage, overshoot or hysteresis.

So if we create our simulation based on models containing only the information from a datasheet, we will find there is little difference in the performance of the £20 and £200 valves and therefore little of value to be gained from the effort of building the simulation.

To move the barrier we need to look at how systems are designed in practice and apply the same techniques to designing our simulation model. An experience hydraulics engineer will for example understand the construction of each valve and have a good feel for its performance. This only comes from testing and experience of using the valves, which is actually the inefficient process we are trying to avoid through using simulation. What we need to do is define and quantify this deeper understanding of the valves performance and build it into our datasheets. As we've said previously this information is unlikely to be available from the valve manufacturer so we need to be able to predict it from the visible construction features or physical lab test results.

Again looking at what works already we can see within AMESim that the 4 way 3 position directional valve works fine for proportional, servo valves and even simple directional valve simulations. The basic model contains all the dynamic elements we require as well as being able to configure each spool land as required. In fact this one model can be configured to suit all 4 way directional valves and only really lacks a little viscosity dependent leakage which can be added externally.

So for directional and proportional valves we need very few model types and we can get most of the information we need from standard datasheets. For the low cost valves where dynamic information is not available then a little bit of experimentation may be required to ensure the simulation model exhibits similar hysteresis and response times etc to the actual valve but this is something than can be achieved reasonably quickly and without expensive hardware.

The essence of this new approach has been to take the above formula and apply it to other valves. Creating a range of simple hydraulic components in which the important parameters can be set. Larger valves or circuit logic can be created by adding a combination of several basic hydraulic valve elements; just as they are in practice e.g. a pilot operated relief valve would have a small flow, fast responding poppet as the pilot and a larger spool or poppet valve for the main stage. Rather than selecting valve codes and expecting someone else to have added it's performance data, the user can build up any combination or piloting arrangement by linking the basic hydraulic components in the correct way.

It's worth pointing out that this approach is still based on complete valve functions rather than individual components. The AMESim HCD library can still be used along side these valves where necessary but is kept to a minimum for the benefit of speed.

A fundamental principal of computer simulation is that you have to know what you are looking for to build a model to find it. So before we decide what parameters each valve model should contain we must first remind ourselves of what we are trying to achieve and the type of problems we see in real hydraulic circuits.

A good comparison is the larger pressure drops and losses permitted for one off industrial system design compared to the higher efficiency required by mobile equipment manufacturers, who generally have longer development programs. In these environmentally conscious days we would hope to achieve far better overall efficiencies in the first off design, greatly reduce the risk when keeping our safety margins to a minimum and always using the most cost effective valve options available.

Experienced hydraulic designers can generally be confident their systems will work but rarely have the opportunity to try other options or the time to optimise the design's performance.

Looking in further details at which parameters should be added to our models we again refer back to the experience gained from actual system development. Poor specification of the basic flow and pressure drop information is rarely the cause of failures, however, the following parameters have a nasty habit of catching designers out.

Customers often don't understand their loads in full and this is probably the biggest source of complications. Modern simulation can dramatically improve our understanding of what the load is doing and new developments with Cosimate allow us to link our existing 3D graphic models directly with our AMESim simulation, thus providing an instant shift in our 'barrier' to quick load model design.

PQ curves, for example, are a fundamental feature of pressure control valves and one area where we need full control. Basic simulation models provide a theoretically perfect performance but this is the key area where lower cost, less perfect valves may or may not be acceptable. We need to be able to enter more representative curves either from datasheet information or from test results, to effectively compare different valve qualities.

Pressure relief valve response times can be critical and hydraulic designers apply various principals when selecting the most appropriate valve. Knowing the correct first and second order differential value to apply to each valve model may take some trial and error but the ability to do this will allow the extremes to be trialled to investigate how important they are in each system.

The effects of hysteresis on system performance can be one of the most difficult parameters to predict. Also as engineers try to design more efficient leak free systems they need to add more seals, which will increase the hysteresis. The value of hysteresis is generally not provided in datasheets but by looking at the valve construction it is possible to predict levels of friction by analysing the number, size, pressure and type of seal.

Many basic valve simulation models do not have enough detail to allow us to identify stability issues within pilot supply and drain lines. However, now that we are able to quickly build complete valves from several small but configurable function models, we can achieve a much better measure of what is happening within this key area of design.

With the ability to build hydraulic control valves from a series of functional elements we can also attach these to variable displacement pumps and motors to create a versatile power source. AMESim already has a powerful library of pipe models to allow comprehensive circuit simulations to be designed.

This new approach to hydraulic simulation could give us enough components to build a complete hydraulic system with sufficient detail to optimise most of the known performance issues.

The time to build a simulation circuit using this technique is greatly reduced and the extra processing power compared to a human brain will provide much more accurate predictions of both performance and efficiency. With the models working we can also trial the affects of extreme environmental or operating conditions and hopefully significantly reduce the overall development time.

One remaining weapon for finally moving removing our barrier to general simulation is the ability to check each systems thermal stability and heat transfer. These facilities are available and should soon be practical within a reasonable development time.

LMS have created an improved range of simulation models that allow large hydraulic circuit simulations to be built and tested in significantly less time. Most hydraulic valve models can be built from a small range of basic valve components. Each component is populated with information that is readily available from most datasheets or can be predicted from a simple analysis of the component design. Model parameters have been heavily based on the factors that experience has shown cause issues as well as the areas on which performance and efficiency are known to depend.

The result is a simulation model that gives a wide range of valuable information in much less time. Removing the barrier to cost effectiveness computer simulation design and making its use to develop and optimise hydraulic systems finally look much more appealing than physically building equipment and sending it out on trials.

Removing the barrier to effective simulation could see much shorter development time, fewer costly mistakes and significantly better operating efficiencies.