Stiffening subfloor with floating noggins

[MortarThePoint]

WARNINGS: NO SUBSTITUTE FOR WHERE FIXED NOGGINS ARE REQUIRED. ALSO, IT'S JUST AN IDEA FOR CONVERSATION AT THE MOMENT.

 

Well I love over analysing things and was wondering about sections of timber fixed under the subfloor at 600mm c/c, but not fixed to the joists. The benefit of this is they don't have to be cut to size and it removes the difficulty of screwing a noggin to the joist. I have modelled sticking 500mm lengths of 2x2 timber under 18mm sheet material and 600mm c/c 47mm perfectly stiff joists. I have used a Young's Modulus (MOE) of 2GPa for the sheet and 8GPa for the timber. OSB3 has a major axis MOE of around 2500MPa and a minor axis of 1250MPa. My tool doesn't allow anisotropic materials (could add ribs) so I plucked for 2000MPa. As for the timber, 8GPa is 'typical' of pine I think.

 

I've added a 5mm thickness of the 2GPa material to model the compliance of the joist fixment. The  bottom surface of that 5mm is rigidly constrained. Zero X-direction displacement constraints on the ends due to symmetry to model and infinite joist length. My model could be a quarter if the size, but...

 

I'm applying a uniform pressure of 0.002 MPa = 2kN/m2 to the top surface of the sheet.

 

Three scenarios:

    No noggin: max displacement 0.52mm midway between joists. Maximum stress 1.2MPa in sheet at joist edge. [modelled by making noggin very soft, MOE 0.001GPa]

    Floating noggin: max displacement of 0.29mm midway between joist and noggins. Max stress of 2.1MPa at edge of noggin bond*.

    Fixed noggin: max displacement of 0.25mm midway between joist and noggins. Max stress of 0.46MPa** at edge of noggin and edge of joist.

 

* not real as there would be some compliance here. I could model a lower MOE material between the two to represent the glue, bit that feels excessive.

** may be underestimated as mesh size large here.

 

I think this could be a good approach as it is pretty quick to fit having precut a load of 500mm 2x2 pieces. Grab adhesive and 4 screws per ' floating noggin'. It's cheap too as uses 1.38m of 2x2 per m2 so, in cheaper times, would cost less than £1/m2.

 

Is this a done thing, what do people think?

Fixed noggin modelled by constraining bottom fact of noggin.

[Big Jimbo]

There are a lot of smart people on here, and i am not One of them. ? Making things stiffer can only be a good thing in my opinion.

[ProDave]

Are you miss understanding the purpose of noggins?

 

You seem to be talking of adding them to stop the bend in the 18mm floor covering, which is inadequate and should be 22mm for 600mm centres.

 

The real purpose of noggins is to stop the joists twisting and so reduce the sag along the length of a joist.

[Big Jimbo]

And we all know that stiff is good, and sagging ain't.

[MortarThePoint]

59 minutes ago, ProDave said:

Are you miss understanding the purpose of noggins?

 

You seem to be talking of adding them to stop the bend in the 18mm floor covering, which is inadequate and should be 22mm for 600mm centres.

 

The real purpose of noggins is to stop the joists twisting and so reduce the sag along the length of a joist.

 

No these are in addition to where real noggins are needed and they must be done properly: "NO SUBSTITUTE FOR WHERE FIXED NOGGINS ARE REQUIRED"

 

I probably should have just called these 'stiffening timbers' rather than 'flying noggins'. As far as I can tell, 18mm OSB3 can be used on 600mm c/c joists:

 

https://www.norbord.co.uk/resources/help-advice/faqs/

https://nhbc-standards.co.uk/6-superstructure-excluding-roofs/6-4-timber-and-concrete-upper-floors/6-4-19-floor-decking/

 

But I wanted to investigate how the sheet material could be stiffened if desired. I might be calling them the wrong thing, but fixed noggins (or strutting) would clearly help stiffen the feeling of the whole floor, but I thought these timbers could help against any fear of bounciness in the sheet material itself. Clearly, strutting out the entire floor with 4x2 would be better, but much more time consuming. Such strutting would help joists act in unison, whereas this approach would only act on the sheet material itself. Normal noggins (>=2/3 of joist height) would still be needed in all the usual places.

[TonyT]

I prefer to use the cut offs of the joists as the noggins not 2x2. 
 

 

[MortarThePoint]

I did model it with a glue layer after all and as suspected the peak stress is lower at 0.85MPa. The orange ball shows where that is and the blue ball shows where the peak displacement is (0.30mm). I've used a coarser mesh away from the area of interest and a finer mesh in the area of interest. I've modelled the glue as 0.1mm thick and having a very low Young's Modulus of 10MPa which is probably a bit low, but even with that low a stress of 1MPa would be a strain of 10%, so 0.010mm.

 

[Marvin]

7 hours ago, TonyT said:

I prefer to use the cut offs of the joists as the noggins not 2x2. 
 

 

I thought they had to be at least 2/3rds of the depth of the joist?

[MortarThePoint]

7 hours ago, Marvin said:

I thought they had to be at least 2/3rds of the depth of the joist?

 

Noggins also need to be attached to the joists.

 

I've confused matters by calling them 'floating noggins'. They are splints really. The concept has nothing to do with noggins. There purpose is to stiffen the sheet membrane not the joists.

Edited May 13, 2021 by MortarThePoint

[Nickfromwales]

Why start the problem in the 1st place by using 18mm chipboard?!? I haven't used that weak weetabix pap for donkey's years. 22mm minimum in every instance for me, and then no silly issues or need for 3D CAD drawings then . If you really are losing sleep over this, go get yourself some pine T&G floorboards and go 'old skool'. 

Ditch the idea of 18mm, and get onto the next problem. And yes, you ARE over-thinking this, but you're focussed on the wrong solution AFAIC.   

[MortarThePoint]

4 minutes ago, Nickfromwales said:

using 18mm chipboard

 

It's 18mm OSB3 which is equivalent to 22mm P5 chipboard

Edited May 13, 2021 by MortarThePoint

[Nickfromwales]

Go and buy one of each, test them, and come back to me  Screws also chew through OSB much worse than with P5, so I don't use it for flooring, only 'racking' on walls. 

[MortarThePoint]

Broadly the idea here is that you can more than double the stiffness of your sheeting for about £1/m2 and <5mins/m2. The improvement on say a 100m2 area for £100 and 4hours work feels worth it.

 

The idea could be used on any sheet material including P5.

Edited May 13, 2021 by MortarThePoint

[MortarThePoint]

3 minutes ago, Nickfromwales said:

Go and buy one of each, test them, and come back to me  Screws also chew through OSB much worse than with P5, so I don't use it for flooring, only 'racking' on walls. 

 

I am effectively doing just that as I'm using 22mm P5 in the garage.

[Nickfromwales]

1 minute ago, MortarThePoint said:

 

I am effectively doing just that as I'm using 22mm P5 in the garage.

I find the screws have much better purchase with P5. I use screws with a part blank shank so they don't 'jack' the boards up off the joists. Vortex screws ( 5.0 x 50mm ) are a favourite. Link. 

[PeterW]

1 hour ago, Nickfromwales said:

Ditch the idea of 18mm, and get onto the next problem. And yes, you ARE over-thinking this, but you're focussed on the wrong solution AFAIC.   

So ... I agree ..!

 

53 minutes ago, MortarThePoint said:

Broadly the idea here is that you can more than double the stiffness of your sheeting for about £1/m2 and <5mins/m2. The improvement on say a 100m2 area for £100 and 4hours work feels worth it.

I don’t think that’s correct as you’re using a single planar deflection. You’re assuming the board can flex but also it will act as a beam and it is a multiple span / support element as a board is fixed across a minimum of 3 joists. 
 

When you look at it like that, you start to realise the complexity of the floor system is actually a modular cassette of beam and floor acting as a single element. Also worth understanding that the floor beam / joist has more of an impact to the structure than the “skin” itself. 
 

This is also a 2 person job as the stress members would need to be added in parallel to the joists so they cannot be added to the boards  ahead of installation so you’re talking about 5min/m x 2 people x £20/hr which is £330 in labour plus £100 materials so £430. 
 

Take into consideration that moving from 600 to 400 centres would add around 8 additional joists, then the £430 would be better spent on that element which has a significant increase on overall stiffness. 
 

There are also additional issues such as the inability to use spreader plates etc so this is possibly a useful remedial activity for a floor that has issues and requires additional stiffening, but as a new build the use of the correct products in the first place would be preferable. 

[MortarThePoint]

39 minutes ago, PeterW said:

I don’t think that’s correct as you’re using a single planar deflection. You’re assuming the board can flex but also it will act as a beam and it is a multiple span / support element as a board is fixed across a minimum of 3 joists.

The constraints of the model should take that in to account, but I can check.

 

39 minutes ago, PeterW said:

This is also a 2 person job as the stress members would need to be added in parallel to the joists so they cannot be added to the boards  ahead of installation

Grab adhesive the timber on to the underside and then screw from above. I have used this approach before for where I would otherwise have needed someone to hold the timber below.

 

39 minutes ago, PeterW said:

Take into consideration that moving from 600 to 400 centres would add around 8 additional joists

I'm not sure where this has come from. The joists remain at 600c/c.

 

39 minutes ago, PeterW said:

There are also additional issues such as the inability to use spreader plates

Yes it would stop you from installing UFH spreader plates, but so would any timbers at right angles to the joists supporting the floor sheeting.

 

39 minutes ago, PeterW said:

but as a new build the use of the correct products in the first place would be preferable

It is using the correct product, perhaps not normally the preferred product. 18mm OSB3 is equivalent to 22mm P5.

 

This approach is a thought for improving the stiffness of any sheet subfloor (P5 or OSB3). I'm interested in OSB3 as that's what I'll be using.

 

Other than the confusion caused by the erroneous use of the word noggin in the name, the issue I currently have with it is that a portion of the bond between the timber splint and the sheet material is in tension. I think the stresses involved are well within the capability of the glue, but would probably pull just a screw into  the sheet material (we're only talking ~0.5mm here). A timber fixed to the joists would only have a compressive loading on the joint with the sheet material so the screws/glue don't play a part then. Ultimately this tension could be the show stopper to the idea.

 

Edited May 13, 2021 by MortarThePoint

[MortarThePoint]

As feared, the tensile strength of OSB3 perpendicular to the board plane is poor. Below is example data for something I could find and it's only 0.3N/mm2 = 0.3MPa. I think the peak stress applied to the OSB3 in this fashion is around 4x that figure. The screw could help, but it would be better of the material properties were compatible with just the glue.

 

It occurs to me that the stress could be reduced by having a 45degree cut at the end of the splint. I don't know if that would be enough though as that out of plane strength is very low.

 

[MortarThePoint]

I've tried the 45degree cut. To aid visualisation, I increased the glue thickness to 1mm, but not changing its MOE. The maximum displacement is now 0.33mm, but the peak tensile stress is much reduced (if it is to be believed) to 10% of the OSB3's out of plane tensile strength. The maximum displacement is 0.33mm.

 

Increasing the MOE of the glue by a factor of 10x to compensate for the increase in thickness (therefore giving the same deformation) puts the deflection back to 0.30mm, but the peak tensile stress to 0.24MPa which is 80% of the OSB3's out of plane tensile strength.

 

The 45 degree cuts help and using a more flexible glue would definitely also help.

 

Glue stiffened to counteract effect if increase thickness used for visualisation:

[MortarThePoint]

I thought I would investigate other timber sizes acting as ribs. I have a limited number of models I can simulate, so am doing this by modifying the Young's Modulus with the cube of thickness and linear with width:

E' = E * (w/47) * (t/47)^3

 

This doesn't very well represent increases in width as the curve of the sheet deflection would be supported further out in the x-direction, so this would give an overestimate of deflection in that case.

 

Using 22 x 100 timber (E' = 1.6GPa) still has a meaningful effect and reduces peak deflection to 0.35mm (0.25mm with a fixed noggin, 0.3mm with 47x47 and 0.52mm with nothing).

 

 

 

Using 38 x 50 timber (E' = 4.5GPa) gives peak deflection of 0.31mm (0.25mm with a fixed noggin, 0.3mm with 47x47 and 0.52mm with nothing).

 

Using a piece of 25 x 38 batten (E' = 0.97GPa) gives peak deflection of 0.37mm (0.25mm with a fixed noggin, 0.3mm with 47x47 and 0.52mm with nothing).

 

So strengthening the long edge joints with even 22x100 timber sections looks to give a significant improvement in stiffness, reducing peak sheet based deflection by 33%. I expect adding another rib mid panel would reduce peak sheet based deflection much further too (below 0.2mm as even in the weakest case above, the deflection above of the 38x25 timber was only 0.2mm), probably reducing peak sheet based deflection by over 60%. If I can be bothered with any of this, I think I would go for 22 x 100 at the joints and perhaps 25 x 38 at the panel midline. I expect that would make the sheet immensely stiff, without putting undue out of plane stress on the OSB3 (No Guarantees though :-).

Edited May 15, 2021 by MortarThePoint

[MortarThePoint]

For reference, the glue tensile stresses in the non-45 degree cut 47 x 47 case at the end of the rib where I expect it to be highest are:

• Not modelling glue layer: 0.51MPa [global max stress was 2.1MPa, not sure where]
• Modelling the glue layer: 0.33MPa [global max stress was 3MPa, not sure where]

Both of these are higher than the OSB3 out of plane tensile strength (0.3 MPa) so would be an issue.

 

Modelling the glue layer and approximating the 22 x 100 gives a glue tensile stress of 0.18MPa. I'm not trusting that value as the solver is doing something weird and violating symmetry (circled). That needs to be halved though due to the way I am modelling the dimension change and so the value becomes 0.09MPa, well under the OSB3 out of plane tensile strength of 0.3MPa. Note, peak deflection in that model is 0.3mm. A hand waving argument would suggest the stiffness per unit width of the rib has gone down to (22/47)^3 = 10% and the glue stress should too for the same displacement. Displacement there will be higher by a factor of (47/22) ~ 2 higher and so the glue stress should be about 20% of the 47 x 47 case. That supports a figure of around 0.06 to 0.1 MPa, under the OSB3 out of plane tensile strength of 0.3 MPa.

 

 

22 x 100: