Model changing lifetime due to renovation

Use this forum to post Vensim related questions.
LAUJJL
Senior Member
Posts: 1421
Joined: Fri May 23, 2003 10:09 am
Vensim version: DSS

Re: Model changing lifetime due to renovation

Post by LAUJJL »

Hi

After studying Tom's model, it looks better adapted to the problem.

It is based on real age cohorts, which is more natural and calculates for each cohort, how much additional age or capacity is added, that is enough to understand the behaviour and further needs.

It is probably easier to implement too, the real age being easier to follow.

It does not need any special mapping like ordinary ageing chains and cohorts.

The other model with an ageing chain is based on the corrected by retrofitting age or remaining capacity. It delivers some information but one does not have the real age of the cohorts which is embarrassing. To calculate it, it is necessary to add a co flow of the total real age of each cohort and then the average real age. It is too possible to build a co flow of the retrofitted actions.

But all this will not add much compared to Tom's model and will use more complex solutions, mainly mapping.

I then do not find useful to propose anything else, Tom's solution being probably the best to follow and to improve.

But the last decision should be from the end user. Working with different approaches is always useful, but time consuming.

Best regards.

JJ
WayneZhou2009
Senior Member
Posts: 105
Joined: Wed Oct 25, 2017 3:52 pm
Vensim version: PRO

Re: Model changing lifetime due to renovation

Post by WayneZhou2009 »

Dear JJ,

Thank you so much for your comments. I agree with you that Tom's model is elegant and serves the purpose much better.

For the original aging chain model that you corrected, I thought it would still be of value to further explore it by adding a co-flow of efficiency. So I did it. The efficiency is measured in the unit of ton of coal equivalent (tce) per plant, meaning the coal consumption per plant during a year's operation. It is set to be a fixed value, i.e. 300 tce/plant.

In the co-flow, I assume that the retrofits only add the extra lifetime (as already done in your corrected model), and do NOT increase or decrease the efficiency. So, theoretically, whatever the retrofit fraction is (0%, 10%, 30%, etc), the average efficiency of each stock shall remain the same as the original efficiency, i.e. 300 tce/plant, as long as a retrofit does not change the efficiency of a plant.

However, surprisingly, the simulation returns some crazy results for both coal consumption of each stock, as well as the average efficiency of each stock. I thought it might be caused by those very low stock levels (plants) after some years, which are very close to zero. For example, the 3-year-old plants are extremely low after 10 years. So I tried to use IF THEN ELSE to set those low levels to be 0, and then used ZIDZ. However, it did not solve the problem.

I spent many hours today checking the structure of the co-flow but failed to spot the exact reason(s) of getting crazy values.

Could you help take a look at the attached model? Also, I could not get over thinking of adding the retrofit fraction as a co-flow as you previously mentioned. However I am not able to find out how this could be realised. Appreciate if you could enlighten me a bit?

LAUJJL wrote:Hi

After studying Tom's model, it looks better adapted to the problem.

It is based on real age cohorts, which is more natural and calculates for each cohort, how much additional age or capacity is added, that is enough to understand the behaviour and further needs.

It is probably easier to implement too, the real age being easier to follow.

It does not need any special mapping like ordinary ageing chains and cohorts.

The other model with an ageing chain is based on the corrected by retrofitting age or remaining capacity. It delivers some information but one does not have the real age of the cohorts which is embarrassing. To calculate it, it is necessary to add a co flow of the total real age of each cohort and then the average real age. It is too possible to build a co flow of the retrofitted actions.

But all this will not add much compared to Tom's model and will use more complex solutions, mainly mapping.

I then do not find useful to propose anything else, Tom's solution being probably the best to follow and to improve.

But the last decision should be from the end user. Working with different approaches is always useful, but time consuming.

Best regards.

JJ
Attachments
Retrofit reinjection_corrected_with coflow.mdl
(24.67 KiB) Downloaded 192 times
LAUJJL
Senior Member
Posts: 1421
Joined: Fri May 23, 2003 10:09 am
Vensim version: DSS

Re: Model changing lifetime due to renovation

Post by LAUJJL »

Hi

One first question.

What is your job?

Working for an organisation, consultant, student?

What is the purpose of the model? The purpose is to find policies that will be implemented, or is it a simple student work whose purpose is learning?

One must not forget that a model must be understood to be credible and your second model is difficult to understand. The co flow traces the consumption that is the level of a rate, something difficult to trace. It is easier to trace something real like the number of plants, years left than the coal consumption. In the model the levels in the co flow is an amount of coal, that is not a consumption. Than the unit used should not be 'tce' but 'tce/year' .

Another point is what to trace? Is it the tonnage by year or the tonnage by year by plant?

It should be better to change the names of the levels that are misleading.

The plants are not 1 year old but 'as' one year old. It should be better to use 'life time expectancy 5 year' instead of '1 – year-old-plants', same for the consumption levels.

Another point is the ease of understanding. The second model once mapped with subscripts will be totally impossible to understand by any body not used to SD and difficult for SD experts.

What else do you expect that cannot be done with Tom's model.

With it you can easily too trace the efficiency the same way that was traced the capacity and the retrofitted capacity and it will be easier to follow than an ageing chain and a co flow.

If you want absolutely use a co flow, you should take the time to find the right thing for the co flow and preferably trace a state than a rate and derive the efficiency that is a rate from some ratios.

Finally the path followed by a model construction is fundamental like the coordinates of a mathematical problem and choosing the right one is worth the time taken to find it.

I would then definitely work first with Tom's model and see if it can do the job and only then think about finding another way.

Regards.

JJ
WayneZhou2009
Senior Member
Posts: 105
Joined: Wed Oct 25, 2017 3:52 pm
Vensim version: PRO

Re: Model changing lifetime due to renovation

Post by WayneZhou2009 »

Dear JJ,

Thanks so much for your comments. To answer your first question: I am a student. So the model discussed over past few days is for my own interest, only.

By using the co-flow, I want to track how the dynamics of scrappage and retrofit would affect the average efficiency of those plants in use. I followed the generic coflow structure (Sterman's Business Dynamics book). In my case, the efficiency is defined as the annual consumption of coal per plant, hence the correct unit shall be tce per plant per year, as you corrected pointed out. The efficiency can also be termed as "specific coal consumption per plant" to avoid confusion (here, for simplification, I assume that annual operation hours of plants are all the same).

So, the efficiency here is taken as a rate in the model, i.e. the "marginal attribute per unit" in Sterman's generic coflow structure. The purpose of the co-flow is to track the attribute (efficiency). But the real interest is the AVERAGE ATTRIBUTE (average efficiency) of the entire stock, as a function of retrofit, scrappage, and the likely changes of the input efficiency. I set the input efficiency to be a constant, in order to firstly test the model. By doing so, I assume that the average efficiency shall remain unchanged, so long as the input efficiency is a constant, no matter how retrofit/scrappage vary over time. But the results are against this understanding and show some unreasonable numbers.

By way of analogy, suppose I have a stock of cars. The fuel economy of the cars (measured in gas consumption per 100 km) is a fixed level, say, 10 liter per 100 km. So it is an attribute of the car. Over time, there are maintenance and scrappage of my cars, and there are also new purchase of cars that increase my stock. Assuming that (1) maintenance only increases lifetime but does not affect fuel economy; and (2) newly purchased cars have the same fuel economy (10 liter per 100km), then the average fuel economy of my stock shall remain as 10 liter per 100km, no matter how I maintain/scrap my cars. Does this make sense?

LAUJJL wrote:Hi

One first question.

What is your job?

Working for an organisation, consultant, student?

What is the purpose of the model? The purpose is to find policies that will be implemented, or is it a simple student work whose purpose is learning?

One must not forget that a model must be understood to be credible and your second model is difficult to understand. The co flow traces the consumption that is the level of a rate, something difficult to trace. It is easier to trace something real like the number of plants, years left than the coal consumption. In the model the levels in the co flow is an amount of coal, that is not a consumption. Than the unit used should not be 'tce' but 'tce/year' .

Another point is what to trace? Is it the tonnage by year or the tonnage by year by plant?

It should be better to change the names of the levels that are misleading.

The plants are not 1 year old but 'as' one year old. It should be better to use 'life time expectancy 5 year' instead of '1 – year-old-plants', same for the consumption levels.

Another point is the ease of understanding. The second model once mapped with subscripts will be totally impossible to understand by any body not used to SD and difficult for SD experts.

What else do you expect that cannot be done with Tom's model.

With it you can easily too trace the efficiency the same way that was traced the capacity and the retrofitted capacity and it will be easier to follow than an ageing chain and a co flow.

If you want absolutely use a co flow, you should take the time to find the right thing for the co flow and preferably trace a state than a rate and derive the efficiency that is a rate from some ratios.

Finally the path followed by a model construction is fundamental like the coordinates of a mathematical problem and choosing the right one is worth the time taken to find it.

I would then definitely work first with Tom's model and see if it can do the job and only then think about finding another way.

Regards.

JJ
LAUJJL
Senior Member
Posts: 1421
Joined: Fri May 23, 2003 10:09 am
Vensim version: DSS

Re: Model changing lifetime due to renovation

Post by LAUJJL »

Hi

Your calculation of efficiency is biased: your forget the cost of building new plants and the cost of retrofitting them. If you introduce these costs, you will see a difference of 'efficiency' depending on the retrofitting policy even if the coal consumption is the same.

What you want to calculate is the total profit of a set of plants, taken into account the cost of building plants that you can amortize over the expected lifetime, plus the cost of retrofitting that you can amortize over the additional years of the lifetime, plus the current cost of the coal, plus other maintenance costs, and of course the turnover generated by selling the electric power. You can do all that with Tom's model very easily without the need to add any co flow.

Of course if you want to exercise yourself on problems with co flows I would chose a problem where co flows are mandatory.

You can still try to build co flow with that problem, but I cannot help you, because I try to make models simpler and not more difficult. SD is already so complex, that there is no need to increase its complexity, unless you are a consultant and want to look clever and want to justify your fees.

Regards.

JJ
WayneZhou2009
Senior Member
Posts: 105
Joined: Wed Oct 25, 2017 3:52 pm
Vensim version: PRO

Re: Model changing lifetime due to renovation

Post by WayneZhou2009 »

Hi JJ, thank you so much for your further comments. All very much appreciated. As I said before, this is for my own exercise only, nothing to do with consultancy or something else. I do agree with you that models shall be made simpler as far as possible.

Your comments on the efficiency are all correct. I did know that CAPEX and OPEX of the plants shall be duly taken into consideration and that would give a fuller picture and more sensible modelling of the dynamics underlying retrofit activities. But as I clarified previously, I started with the highly simplified scenario where it is assumed that retrofit does not lead to efficiency improvement but only some extra lifetime, just for the purpose of testing my understanding of how the co-flow works.

That said, I totally agree that Tom's model is much better and therefore should be the one to be used and built on further.

Overall, I thank you (and Tom) very much for spending your time offering comments and advices. I found them all very insightful and helpful.
Post Reply