need help for my project...

138 views
Skip to first unread message

고재은

unread,
Sep 29, 2019, 10:28:39 PM9/29/19
to crimso...@googlegroups.com

Hi, I’m trying to simulate blood flow in canine portal venous system and wonder is this right approach.

  1. Make my own geometry and mesh with CT DICOM image
  2. Meshing by below values
    1. Mesh size: 0.5 (absolute)
    2. Curvature refinement: 0.2 / 0.3 (minimum factor)
    3. Boundary layer: geometric growth
      1. Fist layer thickness: 0.03
      2. Number of layers: 3
      3. Total thickness: 0.1
  3. Set boundary condition
    1. No slip condition to all wall
    2. Initial pressure
      1. By assuming normal portal venous pressure is 10 mmHg…
      2. 1333.2 g*mm-1*s-1 (approx. 10mmHg)
    3. Inflow: prescribe velocities (analytic)
      1. -14364 (mm3/s)
        1. Inflow face surface area of 95.76mm2
        2. Inflow velocity obtained by Doppler ultrasonography: 150mm/s
    4. Outflow: netlist boundary condition
      1. By assuming that normal portal venous pressure is 10mmHg…
        1. Give value of 1333.2 to R1
        2. Give value of 0 to P1
    5. Solver parameter
      1. Number of time steps: 3000
      2. Time step size: 0.003s
        1. CFL = 1
        2. Time step size = (mesh size) / (velocity) = (0.5)/(150) = 0.003
  4. Visualization

On pressure map and velocity map, there are too much pressures and abnormally high velocities (even higher than initial pressure-10mmHg and inflow boundary condition-150mm/s). I’m not sure which step is the problem.


Arthurs, Christopher

unread,
Sep 30, 2019, 8:36:39 AM9/30/19
to 고재은, crimso...@googlegroups.com
Hello,

Your outflow boundary conditions do not appear to be correct. If I've surmised correctly, you've assigned a single resistor to each of your four outlets. You must determine how much flow, Q, you want to go through each outlet (in mm^3/s) on time-average (say, per cardiac cycle), and then use deltaP = Q R with that Q to determine the resistance. Your deltaP should be 1333.2, as that is your desired mean gradient (as your distal pressure in the Netlist is zero), so your R=1333.2/Q. You need to do this for every outlet.

Regarding the flows being "higher than the inflow boundary condition" - that's perfectly possible, as it's a velocity. You need to look at volumetric flows in order to make meaningful higher/lower comparisons to the inflow (and the outflow will never be greater than the inflow in that case!)

Best wishes,
Chris

--
You received this message because you are subscribed to the Google Groups "CRIMSON Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to crimson-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/crimson-users/SL2P216MB01851E955824EDFE95B721B586820%40SL2P216MB0185.KORP216.PROD.OUTLOOK.COM.


--
Chris Arthurs, D. Phil.
King's Prize Research Fellow
Imaging Sciences and Biomedical Engineering
King's College London
UK
==
CRIMSON Cardiovascular Modelling Environment: http://www.crimson.software/
==
Twitter: @carthu15p

Arthurs, Christopher

unread,
Oct 1, 2019, 6:43:13 AM10/1/19
to 고재은, CRIMSON Users
No, there is not a way to do this. You'll need to either truncate your model at the circles 2 and 3 you have drawn, or you will need to make estimates on what % of flow will go down each of the more distal vessels, based upon e.g. vessel diameter.

Best,
Chris

On Tue, 1 Oct 2019 at 01:29, 고재은 <rhw...@konkuk.ac.kr> wrote:
Unfortunately, it's hard to obtain information about each outflow boundary condition in my lab due to lack of PC-MRI. 

I can obtain flow velocity at the level of 1(inflow), 2(left main branch), and 3(right main branch).

Is there way to apply boundary condition at the middle of geometry? 
보낸 사람: Arthurs, Christopher <christoph...@kcl.ac.uk>
보낸 날짜: 2019년 9월 30일 월요일 오후 9:36
받는 사람: 고재은 <rhw...@konkuk.ac.kr>
참조: crimso...@googlegroups.com <crimso...@googlegroups.com>
제목: Re: need help for my project...
 

Arthurs, Christopher

unread,
Nov 7, 2019, 3:13:13 AM11/7/19
to 고재은, CRIMSON Users
Hello,

Thanks for writing. In general, please send questions to the mailing list, and not to me directly. I'm replying to the list - including your original question - so that there is a record of answers for everyone.

Most likely, your time step size is too large. 10 ms is almost always too large for proper simulations. I'd try 1 or 0.5 ms. It's also possible that your mesh needs to be finer. These are always the first steps to take when a simulation isn't converging.

Otherwise, it's very difficult to work out what is happening here, as this is a very general error. Are you using Netlist boundary conditions?

Best,
Chris

On Thu, 7 Nov 2019, 01:35 고재은, <rhw...@konkuk.ac.kr> wrote:
Thanks to your kind answer.

I made my mesh again with only two main outlets.

I have another questions, about time step and time step size.

I decided to make total 5 seconds of flow, so I write solver with time step size 0.01 / time step 500. But with this option, I got singula matrix error at approx. 200 time steps. Is there any options to fix this error?

보낸 사람: Arthurs, Christopher <christoph...@kcl.ac.uk>
보낸 날짜: 2019년 10월 1일 화요일 오후 7:43
받는 사람: 고재은 <rhw...@konkuk.ac.kr>; CRIMSON Users <crimso...@googlegroups.com>
image.png

고재은

unread,
Dec 2, 2019, 2:39:54 AM12/2/19
to Arthurs, Christopher, CRIMSON Users
again, singular matrix error occured  from below settings

will it be helpful if I attached my project files?
보낸 사람: Arthurs, Christopher <christoph...@kcl.ac.uk>
보낸 날짜: 2019년 11월 7일 목요일 오후 5:13

Arthurs, Christopher

unread,
Dec 2, 2019, 2:47:21 AM12/2/19
to 고재은, CRIMSON Users
Sure, I'll take a quick look.

Chris

Sent from my phone

On Mon, 2 Dec 2019, 07:39 고재은, <rhw...@konkuk.ac.kr> wrote:
again, singular matrix error occured  from below settings

will it be helpful if I attached my project files?
보낸 사람: Arthurs, Christopher <christoph...@kcl.ac.uk>
보낸 날짜: 2019년 11월 7일 목요일 오후 5:13
image.png

고재은

unread,
Feb 10, 2020, 11:54:55 PM2/10/20
to 'Arthurs, Christopher' via CRIMSON Users
Sorry, I missed your reply (I thought my project was neglected...)

I attached my revised case, and brief description about my project and workflow in Crimson.

Thank you for your attention to this matter.
보낸 사람: 'Arthurs, Christopher' via CRIMSON Users <crimso...@googlegroups.com>
보낸 날짜: 2019년 12월 2일 월요일 오후 4:47
받는 사람: 고재은 <rhw...@konkuk.ac.kr>
참조: CRIMSON Users <crimso...@googlegroups.com>

고재은

unread,
Feb 10, 2020, 11:56:14 PM2/10/20
to 'Arthurs, Christopher' via CRIMSON Users
고재은 님이 비즈니스용 OneDrive 파일을 공유했습니다. 해당 파일을 보려면 아래의 링크를 클릭하세요.

보낸 사람: 고재은 <rhw...@konkuk.ac.kr>
보낸 날짜: 2020년 2월 11일 화요일 오후 1:54
받는 사람: 'Arthurs, Christopher' via CRIMSON Users <crimso...@googlegroups.com>

Arthurs, Christopher

unread,
Feb 21, 2020, 9:16:31 PM2/21/20
to 고재은, 'Arthurs, Christopher' via CRIMSON Users
Hello,

I tried looking at the case you sent, but I cannot reproduce the problem. The simulation (Net_list-study) runs fine for 400 time steps without crashing.

One thing I will say is that you have a "no initial pressure" case - you should always have an initial pressure.

There's a lot of different cases in the files you sent - I had to guess which was likely the most complex. Please try to send simple, minimal failing examples where possible.

Best,
Chris

On Mon, Feb 10, 2020 at 11:56 PM 고재은 <rhw...@konkuk.ac.kr> wrote:
고재은 님이 비즈니스용 OneDrive 파일을 공유했습니다. 해당 파일을 보려면 아래의 링크를 클릭하세요.

보낸 사람: 고재은 <rhw...@konkuk.ac.kr>
보낸 날짜: 2020년 2월 11일 화요일 오후 1:54
받는 사람: 'Arthurs, Christopher' via CRIMSON Users <crimso...@googlegroups.com>
Reply all
Reply to author
Forward
0 new messages