Modeling investments and mandates

[qzaus]

I am interested in modeling some mandates and investment policies. For a policy that mandates investments in electric vehicles, besides increasing the proportion of EV's using UC_COMPRD of imposing lower bounds, will there be any other measures to accommodate such investment or mandates? Should I also modify corresponding demands to reflect these changes? Thank you!

[Antti-L]

For a policy that mandates investments in electric vehicles, besides increasing the proportion of EV's using UC_COMPRD of imposing lower bounds, will there be any other measures to accommodate such investment or mandates?

I have been using market share constraints, for the share of EVs in the total sales of new cars, in addition to market share constraints for EVs in the total passenger car travel demand.  I think at least the former could be a "real-world" policy mandate. But anyway, almost any policy requirement that can be described by convex programming (so basically with linear constraints) should be possible to model with the TIMES UCs. Are you asking for new ideas on possible policies, or help on modelling some known policy requirement?

[qzaus]

For a policy that mandates investments in electric vehicles, besides increasing the proportion of EV's using UC_COMPRD of imposing lower bounds, will there be any other measures to accommodate such investment or mandates?

I have been using market share constraints, for the share of EVs in the total sales of new cars, in addition to market share constraints for EVs in the total passenger car travel demand.  I think at least the former could be a "real-world" policy mandate. But anyway, almost any policy requirement that can be described by convex programming (so basically with linear constraints) should be possible to model with the TIMES UCs. Are you asking for new ideas on possible policies, or help on modelling some known policy requirement?

Thanks Antti. I have specific policies in mind: ZEV mandates (I am thinking to increase market share in future years by parameterizing the RHS values on the UC constraints) and also procurement of EVs for a given sum of investments available (which I am not too sure about how to model).

BTW how to model the sales of EVs in the total sales of new cars? I was only able to model the market share constraints for EVs in the total passenger car (attached my trial UC_scenario). Thank you!

[Antti-L]

> BTW how to model the sales of EVs in the total sales of new cars?

Well, as the car sales are represented by VAR_NCAP, it is quite straightforward. You can refer to VAR_NCAP by UC_NCAP. In the example below (based on the VEDA DEMO) I use one constraint to define the market for new cars, and a second constraint to define the minimum shares for new EV cars in that market.
   
I tested it myself in the DEMO model, and it worked well.

[qzaus]

> BTW how to model the sales of EVs in the total sales of new cars?

Well, as the car sales are represented by VAR_NCAP, it is quite straightforward. You can refer to VAR_NCAP by UC_NCAP. In the example below (based on the VEDA DEMO) I use one constraint to define the market for new cars, and a second constraint to define the minimum shares for new EV cars in that market.

I tested it myself in the DEMO model, and it worked well.

Thanks! This is very helpful!

[AKanudia]

Thanks for this example Antti. I was aware of UC_UCN, but hadn't seen it in action. Makes the constraint a lot more readable.

[Antti-L]

@Amit:  Thanks.
While testing that, at first I had the UC dummies option left activated, and got domain violations for the N constraint dummy commodities. Then I removed the UC dummy option, and it was fine. So, maybe you can check, if there is an oversight in VEDA about that?  Actually I think N constraints would need no dummies ever, would they?

@qzaus: Note that the example generates the constraint for all milestone years (IE=2). You might want to restrict it to some future T≥T1 only, by using IE=5 with UC_RHSRT(T1)=0.

[AKanudia]

you are right Antti, Veda is not supposed to generate dummy variables for N constraints. But perhaps the implementation is not tight enough.

[qzaus]

@Amit:  Thanks.
While testing that, at first I had the UC dummies option left activated, and got domain violations for the N constraint dummy commodities. Then I removed the UC dummy option, and it was fine. So, maybe you can check, if there is an oversight in VEDA about that?  Actually I think N constraints would need no dummies ever, would they?

@qzaus: Note that the example generates the constraint for all milestone years (IE=2). You might want to restrict it to some future T≥T1 only, by using IE=5 with UC_RHSRT(T1)=0.

Thanks Antti for the clarification. If in the years of 2030, 2040 and 2050, I want to limit the market share of all EVs, I should use IE =5, but for the constraint to set up the market for all new cars, should I still leave it as IE =2 since all the milestone years I need to first define the market and then limit the market share as desired?

[qzaus]

> BTW how to model the sales of EVs in the total sales of new cars?

Well, as the car sales are represented by VAR_NCAP, it is quite straightforward. You can refer to VAR_NCAP by UC_NCAP. In the example below (based on the VEDA DEMO) I use one constraint to define the market for new cars, and a second constraint to define the minimum shares for new EV cars in that market.

I tested it myself in the DEMO model, and it worked well.

Hi Antti,

I have tried to set up the constraints as you suggested and tested in the DEMO12, somehow nothing changed as compared to the reference case. Do you mind checking on the constraints I set up? Thank you! I tried the IE option of 5 and 2 and both have no changes.

[Antti-L]

Yes it cannot work like that, you are not following my example. I can only suggest that you take a look at my example more carefully. Moreover, as far as I can see the new cars are mostly named TCAN* and only one is TCAR* in Demo12.  

[qzaus]

Yes it cannot work like that, you are not following my example. I can only suggest that you take a look at my example more carefully. Moreover, as far as I can see the new cars are mostly named TCAN* and only one is TCAR* in Demo12.  

Thanks! Sorry for the typo. I have it revised and did see an increase in VAR_NcapM for EVs, not in VAR_Ncap. There is no VAR_ACT change for EVs. Should I just interpret it as the model endogenously determines these values? I was trying to understand why this is the case since the increase in VAR_NcapM for EVs is drastic.

[Antti-L]

If you don't see Var_Ncap, you still got it wrong.     In my test, I verified that Var_Ncap share for electric cars was exactly 35%, 55% and 75% in 2030, 2040 and 2050, respectively.  That indicates that in my test model the electric cars were still more costly, and the mandate was indeed needed for getting them purchased at the desired proportions.

If you have followed my example, that is what you should be seeing: at least 35%, 55% and 75% share in total new cars in 2030, 2040 and 2050, respectively.  

[Antti-L]

Now I tested my example also with the Demo12 model, simply by putting TCA* as the process filter for all cars and TCANE* for electric cars.  The model results for the Var_Ncap were again fully as expected, as shown in the picture below:
   

[qzaus]

Now I tested my example also with the Demo12 model, simply by putting TCA* as the process filter for all cars and TCANE* for electric cars.  The model results for the Var_Ncap were again fully as expected, as shown in the picture below:

Thank you, Antti! It helps. I have resolved the issue and realized that I did not specify the process name properly earlier. Thanks again for your help and patience