IES-VE User Forum

I would like some feedback from my fellow IES users with regards to heat loss calculations.

Basically we have tried several scenarios and profiles to get accurate heat losses and its all a bit hit and miss. I will highlight a few examples below.

1. Steady state heat losses using CIBSE Load.

We have found that adding a time profile under the "heating profile" just complicates matters and opt for an absolute profile in the "simulation heating setpoint"

an example would be to run the heating at 12deg from 00.00 to 07.00 and then rise to 21deg from 07.00 to 17.00 and then back to 12 deg for the rest of the day.

If I tick the box "adjust for intermittent heating using profiles" this should mean my results give a different heat loss for "space conditioning sensible" and "Steady State Heating" however we are finding that the heat losses are the same.

Surely they should be different?

A steady state condition would maybe be used for under floor heating because the heating is always on, however bringing the heating on at a set time would mean a higher heat loss for a short period of time due to the thermal response of the room, ie a room with thick wall would take longer to heat up then a room with light weight cladding.

The steady state heat loss would be fine if IES then told us what factor to add (such as 15%) for certain heat up periods. (Hevacomop does this) Whats the point in modelling a building and adding specific elemental U values only to get such a dumb set of results at the end of it?

Next scenario.

2. Heat loss using Apach Sim.

Now simulations should give us better more accurate results right? Well we are finding that this is not the case.

Take the above example and using the same profiles above and run an Apachi Sim, you will note when viewing the peak day data that the initial heat loss peak is huge (and we should be using the peak condition right?) What I have found is that regardless of what we do to the time profile or temperature profile the software always tries to heat the room up in about 30 minutes. This is unrealistic in the commercial sense, we don't really want all the walls in the room to be covered in radiators do we? Why cant we tell the software that we want a heat up period of say 2 hours? So our emitters could be sized according to this value?

Or even better, wouldn't it be great if the software could calculate the thermal response factor and tell us what the best heat up period would be for a given buildings energy usage? or for a building with a high thermal mass maybe the software could tell us that steady state heating would be better for energy consumption.

When it comes to the crunch, I am finding that using a steady state heat loss is better, then dding my own factor. I find this a tragedy given the time and effort that is spend inputting a building.

Has anyone found a better method or a better set of profiles to us for the heat loss calculations?

I have contacted IES on this matter and I have also undertaken the training. I keep getting told to read the CIBSE documents which I have. Having started out a long time ago producing all calculations by hand I know what needs to be done to accurately calculate heat losses.

Any thoughts would be appreciated.

Phil

Hello Phil

I'll try to answer these in order:

1. CIBSE Steady State

I suspect you are confusing what IES is actually doing with the CIBSE Steady State calculation. It is not running an annual simulation with everything held on a flat line, it is actually carrying out the CIBSE heat loss hand calculation which is by its very nature, based on an instant in time. The idea is that you use it for sizing your boiler and emitters for the peak condition.

The modelling software is not really being activated and all it is doing is working through the same "spreadsheet" calculation that you would have to do if you were carrying out a CIBSE heat loss calc by hand. It takes an extreme outside condition of whatever you choose (-4C for exampe) and then works out the heat losses associated with fabric, infiltration and room side fresh air based on the information in the model (surface areas, u values, air rates etc).

In theory you should be able to work through the entire CIBSE steady state calculation method by hand and reach virtually the same results as the model.

As to the merits of the CIBSE stead state calcs... you'd have to bring that up with CIBSE, but it is expected that safety margins and boost factors are added onto these base figures to account for warm up periods ect.

2. ApacheSim

You can smooth out the heating spikes by setting the warm up period via a timed absolute profile on the heating setpoint. If you have the heating setting back to 12C in the evening and then slowly ramping up to 21C from about 4:00am to 8:30am it should approximate an "optimum start" type system and the heating spikes should be more realistic.
----------
As a side point, the CIBSE heat loss calcs are still the approved method of sizing boilers in buildings under the UK guidance. If you limit this to fabric loss and infiltration, you can then add the air side by hand and add safety margins and boost factors as per official guidance and easily check the component parts to ensure it all makes sense. All of this would be completely defendable in a court of law should you ever find yourself stood infront of a judge.

However, ApacheSim is not official as far as plant sizing is concerned and so you go out on a limb if using it, even though it could be argued as being more accurate. Additionally, ApacheSim is far more complex and prone to errors creeping in which may not be obvious. And finally, ApacheSim includes the offsetting effect of internal gains on the heat loads which can be dangerous when these are often having to be based on assumptions.

In a nutshell, ApacheSim is great for annual energy predictions but it's highly unlikely that anyone will ever know if you are 20% under on that. On the other hand, if your boiler is 20% undersized you can be certain you'll be hearing about it come winter.

Thanks very much for the reply.

So in a nutshell we should be using the CIBSE steady state method which is what we have found to be more accurate anyway.

I know how the calculation is carried out or should I say IES have already told me that the steady state method is in line with the approved CIBSE method which is what we used in the good old days.

I suppose the only point letting the whole process down is that we still need to decide what factor (intermittent operation factor) to add to the room to achieve an acceptable heat up period. In the good old days this was a lengthy set of calculations based on u values, element areas, room volume etc etc (all data available in the model). I don't see why the IES software cant be programmed to give this value. After all this is all part of the approved CIBSE method which IES keeps stating to me.

Thanks again.

That's a good question. In theory, one would presume calculation of boost factors could be done automatically and it might be a nice feature for IES to add. You should send it in as a suggestion and it might make the next build.

In my experience, there is generally a bit of a disconnect between the mechanical engineers and the modellers. Ideally they would be one and the same but in reality, most companies have the guys who do the modelling as a separate entity to the mechanical engineers, who generally have no idea about how the model works and simply have to take the results it spits out and assume the modeller has done a thorough job.

As much as it pains me to say it, a model is only as good as the guy who built it and I've seen a lot of rubbish computer models in my time. And until this situation changes, or the mechanical engineers responsible for the design start to do the modelling themselves, I think keeping sizing of kit as much in the hands of "hand calcs" as possible is probably no bad thing. Fortunately a lot of the mechanical engineers I've spoken to are quite precious about things like plant sizing and boost factors so they normally deal with it themselves.

Thanks again for the reply.

I totally agree. The engineers within our company carry out all thermal modeling, we are very skeptical about the results which is why I started this post. We have had the training but this was clearly geared towards modellers, I could have kept everyone behind for a few hours asking questions but I don't think I would have been popular.

When it comes to sizing plant, I would always do this by hand as there are too many ways to mess up within the model such as using incorrect profiles etc etc. The majority of our projects are design and build contracts, we inherit a lot of models from other companies and the profiles used in some instances are shocking. As you said these have probably been produced by modellers which don't necessarily understand the engineering side of building services.

I will make my suggestion to IES regarding the boost factor, they could perhaps adopt the same options as Hevacomp i.e. certain percentages for certain heat up periods. In reality I would only use this facility for sizing individual room heat emitters or assessing individual room heat losses.

Thanks for your feedback, much appreciated.

I've come across this problem recently on a number of projects.

When I run an ApacheSim with constant heating and a constant set point I'm getting loads in the 20 - 30 W/m2, and when I run the CIBSE calc I get near enough the same values. These are the values I would expect as steady state loads.

But when I run ApacheSim with a temperature profile with night-time and weekend set-back temperatures, and a ramped pre-heat period prior to occupancy, so as to model a dynamic situation with weather compensated optimum start, then the loads treble. This is not what I expect at all.

I've adjusted the pre-heat ramp from 1 to 3 to 7 hours, and although this does lower the loads, they are still some 200% of the steady state, and significantly the peak still always occurs immediately prior to occupation, however long and shallow the pre-heat ramp is.

Am I doing something wrong? Or as suggested in previous posts, is IES simply not good at this? If so that rather begs the questions what is it good at?

Wasted