February 2, 2019

Decision matrix analysis of layout alternatives

Will the standard gauge O scale layout will the analysis?
With all these layout ideas floating around, Marty suggested, most likely with his tongue firmly in his cheek, to take a relook at the decision matrix. But I was curious, so I did it. The result was somewhat unexpected.

I analyzed 7 options. A, B, C and D are expansions of the Aquia Line. E is an expanded PoLA as I have drawn in the past.

F is a DRGW Soldier Summit based plan that I sketched out in some detail, but never did a final version. It has a double track mainline, CTC, Castle Gate,  a steel mill, the Utah Railway I/C, and several coal mines.

The last, G,  is an O Scale standard gauge plan with a On30 narrow gauge interchange. I just have drafts of this plan, but I do have some rolling stock for it (see image above). The other nice thing about it is that many of the ships, trees, and some structures of my ACW layout could be reused on it. However, the plan is set in the 1920s and trying to find rolling stock for it makes it take a hit for manageability. It is interesting that when I mentioned the Maine idea to my wife, she was excited and said, "ohh good, we can go there to research it." Somehow I think think this means more golf.

The first analysis sums numerical rating of 0 to 3 to rate each G&D. It also assigns a priority to each G&D so I can calculate a weighted sum of ratings by priority. In this analysis, the higher numbers are better. Just looking at the sum of ratings we see that three of the Aquia Line plans are tied. This makes sense as the three designs give up one thing to get another. The winner will be decided by the priority assigned to each G&D. The DRGW plan comes in last. When you take the weighted average by priority, the Fredericksburg Extension is the winner.  The DRGW is still last, but the Aquia double deck is second to last.




For you non-parametric statisticians out there, this next matrix uses ranks instead of priorities and weights. I ranked each layout design in order for each G&D. There were lots of ties in these rankings, that is why you see rankings repeated.  In this analysis, low numbers are better.

The reason E&F get ranked low for reliable engines has to do with DCC programming issues. I really don't cherish the idea of trying to manage a fleet of 20 or so locomotives and their DCC idiosyncrasies.  However, once dialed in, the HO engines with keep alives are good runners. The Fredericksburg Extension ties for the lead with  option B, moving Falmouth to the TV room. All ACW plans win this analysis, even the double deck  Aquia plan. However, now the Maine Central layout finished last. I don't like the rank matrix as much as it doesn't do as good a job at showing how important  a particular parameter is.



Overall, I was surprised because I thought option C would win. But losing the crew lounge was a big hit, as was not having Burnside wharf as part of the plan.  The Fredericksburg Extension is the winner  by a narrow margin over the other Aquia Line expansions. However, even if I zero out the priority for crew lounge, the Fredericksburg extension still wins by a nose.   If I were to drop the Aquia Line,  the PoLA expansion is the way to go.

8 comments:

  1. I’m surprised that C did not compare better with A given that you only had one point difference on the crew lounge.

    Would be interested in seeing the weights used in the first matrix. Are the weights the priority?

    I used to use a Kepner Trevor decision matrix process that was similar and quite useful. The weighted matrix was used to rank wants and desires after each alternative was passed through a must have filter.

    Do you have a list of absolute must haves? These are items that you can not do without. No point in scoring an alternative if it doesn’t pass the must haves.

    As I recall weights for the second step were created before thinking too much about the alternatives in order to reduce bias. I think we used a score of 1 to 10 for each item before factoring the weights. We also would perform a sensitivity analysis to see how much ranking might change if some of the scores changed up or down.

    Finally we would look at adverse or unintended consequences.

    ReplyDelete
    Replies
    1. C compared to A loses points for crew lounge, big bridge and town scene. It gets some points back for longer mainline and better TT&TO opportunities.
      I have not created a list of absolute must haves. I used the priorities for that. I never say never.

      Delete
    2. C also lost points be not being as manageable as A and B, since it requires a lot more work.

      Delete
    3. The weights are listed in the priority column. Multiply priority times rating.

      Delete
  2. Wondering why C crew lounge score is lower than D. Looks the same on the diagrams.

    ReplyDelete
    Replies
    1. D is the double deck layout that has the wide open crew lounge by the stairs. C crew area is in the upper room where Falmouth is now.

      Delete
  3. I'm less concerned about the crew lounge given there are plenty of stools to sit on in the rare instance you're waiting for a train. Also (IMHO) the constant drone of noisy conversation in the crew lounge is a negative, so breaking it up would make for a quieter operation. Just my curmudgeonly observation .

    ReplyDelete
    Replies
    1. I do have several stools in the various rooms for crews to get off their feet. The crew lounge is used more for times when we are not operating. During op sessions everyone is assigned jobs. There are rarely people waiting to go on duty as it other large RRs.

      Delete