Pith. sign in

REVIEW 3 major objections 5 minor 42 references

Modelagem de um Problema de Dimensionamento de Lotes com Demanda Variavel e Deterministica e Efeitos de Learning e Forgetting

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A 0.1% learning discount cut modeled lot-sizing costs by 30%

desk verdict An honest undergraduate monograph whose "learning and forgetting" is modeled as a within-period linear quantity discount, making the headline 30% savings an artifact of the assumed coefficient rather than a demonstrated learning-curve effect. read the letter →

arxiv 1908.10293 v1 pith:BTDLCVYH submitted 2019-08-20 cs.OH

Pedro Cesar Lopes Gerum This is my paper · ORCID
classification cs.OH MSC 90B3090C2090C11
keywords lot-sizinglearningcurveforgettingeffectquadraticprogrammingmixed-integerquadraticallyconstrainedeconomiesofscalevariabledemandproductionplanning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that the learning and forgetting effects, usually ignored in lot-sizing models, materially change both the optimal production plan and the total cost. It extends a standard linear lot-sizing formulation by making the unit production cost fall linearly with the size of the period's lot, then solves the resulting mixed-integer quadratic program for a six-period, single-product case. Across 18 scenarios, even a per-unit discount of one-tenth of a cent (0.01% of unit cost) shifted production quantities and saved about $10,000, while a 0.1% discount produced total-cost reductions of about 30%. The paper's point is that economies of scale from learning deserve to be included, not treated as negligible.

What carries the argument

The load-bearing mechanism is a linear learning/forgetting discount embedded in the unit production cost: $uc_i = 100 - x p_i$, where $x$ is the per-unit discount and $p_i$ is the lot size in period $i$. This replaces the fixed per-unit cost of the classical model with one that rewards larger lots, and the product $uc_i \cdot p_i$ becomes $100 p_i - x p_i^2$, which is why the model is a mixed-integer quadratic program. The forgetting side is represented by resetting the discount each period, so learning does not carry across periods. The machinery's job is to turn the learning curve into a quantity discount that the optimizer can trade against setup and holding costs.

What would settle it

Run the same six-period problem with the discount applied to cumulative production instead of the current period's batch, and solve to proven global optimality; if the 30% saving shrinks or disappears, the linear per-batch discount is doing the work.

Watch

Extended reading notes

Core claim

The central discovery is that adding a per-period linear learning/forgetting discount to the variable production cost changes the lot-sizing solution enough to matter. In the model, unit cost in period $i$ becomes $uc_i = 100 - x p_i$, so the production term in the objective becomes $(100 p_i - x p_i^2)$, making the problem quadratic. With the GAMS solver handling the mixed-integer quadratically constrained program, the author found that the discount coefficient $x = 0.01$ (0.1% of the base unit cost) produced total-cost reductions around 30% compared with the no-discount baseline, and even $x = 0.001$ produced roughly $10,000 in average savings. The paper also reports that variable setup costs helped mainly through strategic timing of production, saving about $1,000, while holding-cost differences could be offset by a slightly larger discount. The author states these solutions are not guaranteed globally optimal because quadratic models may return a local minimum.

Load-bearing premise

The whole result depends on the assumption that learning and forgetting lower the unit cost in a straight line based only on the size of the current batch; real learning curves are usually curved and build on cumulative experience.

Editorial extensions

If this is right

  • Optimal lot sizes shift toward fewer, larger production runs when the discount coefficient is high enough.
  • Total cost falls by amounts that are large relative to the discount: a 0.1% per-unit discount yields about a 30% total-cost reduction in the studied case.
  • Ignoring learning and forgetting overstates cost and can misidentify when to produce.
  • Variable setup costs produce smaller savings, around $1,000, but still improve the plan.
  • The linear-discount formulation requires quadratic programming, so linear lot-sizing solvers are insufficient.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • I read the model's 'forgetting' as a period reset: since unit cost depends only on the current period's $p_i$, the formulation behaves like a quantity discount rather than a cumulative learning curve; switching to cumulative output would likely change which periods produce.
  • The reported 30% figure is tied to this six-period demand pattern; testing on longer horizons, different seasonal peaks, or capacity-constrained settings would show how general the magnitude is.
  • A natural extension is to compare the linear discount against a logarithmic learning curve with explicit forgetting decay; if results are similar, the conclusion is robust, and if not, curvature is decisive.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This manuscript, an undergraduate monograph in Portuguese, proposes an extension of a single-machine, single-product lot-sizing model in which the unit production cost is reduced by a linear term proportional to the lot size in each period (uc_i = 100 - x * p_i). The authors solve the resulting mixed-integer quadratic program with GAMS over a six-period deterministic demand scenario, varying inventory holding costs, setup costs, and the discount coefficient x. They report that including the discount changes the optimal lot sizes and reduces total costs by up to about 30%, and they conclude that learning and forgetting effects are important and should be considered in lot sizing. The paper also notes that the quadratic formulation may yield only local optima and that the assumed linear learning curve is a simplification.

Significance. If the central claim were supported, the paper would provide a simple quantitative illustration that lot-sizing decisions change when production cost declines with cumulative output, and it would complement the existing literature on learning and forgetting in lot sizing. The GAMS model is transparent and easily reproducible, and the parameter sweep across holding costs, setup costs, and discount levels is a reasonable exploratory design. However, the model as specified does not implement learning or forgetting: the unit cost depends only on the current period's production quantity, not on cumulative experience or idle time, so the reported cost savings are a within-period quantity discount. The paper's headline quantitative claim also contains a numerical inconsistency. The core contribution, demonstrating the importance of learning and forgetting, is therefore not established.

major comments (3)
  1. [Section 3.3.2, Eq. (8)] The model's learning/forgetting mechanism is given by uci = 100 - x*pi, which depends only on the production quantity in period i and resets to a base cost of 100 in every period. There is no state variable for cumulative production, no dependence on previous periods' output or idle time, and no decay term that would represent forgetting. Consequently, the model is a within-period quantity-discount model, not a learning-curve model as defined in Section 2.4. The conclusion in Section 5 that the results demonstrate 'que os efeitos de learning e forgetting são sim importantes' is not supported by the equations actually solved.
  2. [Section 5 and Table 3] The statement that 'Um desconto de 0,1% por produto do custo inicial de produção gerou reduções de custo de 30%' is numerically inaccurate: the largest tested value x=0.01 corresponds to 0.01% of the base cost of $100, not 0.1%. Moreover, the 30% cost reduction is computed relative to the x=0.0001 (nearly zero discount) scenario, not relative to a true baseline with x=0, and it is a direct consequence of the assumed linear discount reaching up to 50% of unit cost at the capacity bound of 5000 units. A baseline run without the discount is needed to isolate the effect of the learning/forgetting term; without it, the quantitative importance claim is overstated.
  3. [Section 4 and Appendix A] The paper acknowledges in Section 4 that the quadratic results may represent only local optima, and the GAMS code in Appendix A fixes p('5')=1400 and solves with optcr=0, which does not guarantee global optimality for the nonconvex MIQCP. Since the reported savings are differences between possibly local solutions of different scenarios, the magnitude of the claimed effect is not reliable. A comparison with a known optimal solution for the linear (x=0) problem or a global-optimization certificate would be necessary to support the reported cost reductions.
minor comments (5)
  1. [Abstract and Section 2.4] The abstract and Section 2.4 describe a 'learning curve', but the accepted model is explicitly linear; this discrepancy should be stated more clearly as a deliberate simplification with known limitations, not just in passing.
  2. [List of Symbols and Table 2] The list of symbols defines xi as the discount, but the model text and Table 2 use x without a subscript; the notation should be made consistent.
  3. [Section 3.3.2] The sentence before Eq. (8) says 'A equação número 7 mostra onde o efeito de learning e forgetting foi implementado', but Eq. (7) is the capacity constraint; the new discount equation is Eq. (8). The equation reference should be corrected.
  4. [Section 5] The final paragraph correctly mentions the possibility of a more realistic logarithmic learning curve in future work, but this caveat should also appear where the linear assumption is first introduced, not only in the conclusions.
  5. [References] Several references are incomplete (e.g., missing page ranges or journal identifiers), and one reference to Pinedo (2010) is cited in the text as 'Pinedo (2010)' while the bibliography lists 'Pinedo M. Lot-sizing... 2002'; the citation style should be harmonized.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline 30% cost reduction is built into the assumed linear discount of Eq. (8), so the 'importance of learning and forgetting' largely restates the input assumption.

  1. self definitional [Section 3.3.2, Eq. (8); Section 4, Table 3; Section 5, Considerações Finais]
    "uci = 100 − x ∗ pi i = 1, …, n (8) ... Os resultados obtidos mostraram que os efeitos de learning e forgetting são sim importantes e que devem ser considerados ... Um desconto de 0,1% por produto do custo inicial de produção gerou reduções de custo de 30% no problema estudado."

    Eq. (8) is the entire learning/forgetting mechanism: the unit cost falls linearly with the current period's batch size, with no cumulative-production or idle-time state. The objective therefore contains (100 − x·p_i)·p_i = 100·p_i − x·p_i^2. The paper varies x (0.01, 0.001, 0.0001) and then reports, in Section 5, that the largest x cuts total cost by about 30%. That reduction is a direct algebraic consequence of the assumed quadratic discount; comparing scenarios with different x values is a sensitivity sweep, not a test or prediction of a real learning curve. The conclusion that learning and forgetting are important and should be considered is therefore the input assumption (a batch-size-dependent discount lowers production cost) restated as a result.

full rationale

There are no load-bearing self-citations or imported uniqueness theorems here: the model is solved directly and the references are mostly textbook/background citations. The central problem is structural, not citational. Eq. (8) defines the learning effect as a per-period linear discount, and the paper's quantified claim of a 30% cost reduction is an exact consequence of that definition, not an empirical finding or an estimated prediction. The paper also assumes the linear curve only 'para facilitar a modelagem' and admits the quadratic model may yield only a local optimum, which further weakens the quantitative claim. Nonetheless, the paper is a self-contained modeling exercise, and the setup/inventory tradeoffs are not themselves circular; the circularity is confined to the headline conclusion that learning and forgetting are important because the model assumes their effect as a large cost discount. Score 6 reflects the construction-equivalent nature of the central quantitative result without implying any citation-based circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the linear discount model and the chosen parameters. No new entities are introduced.

free parameters (3)
  • x (learning discount coefficient) = 0.01, 0.001, 0.0001
    Chosen by the author, not fitted. It sets the strength of the learning discount and directly determines the reported cost reductions.
  • Base unit production cost (100) = 100
    Arbitrary constant in the unit cost formula; scaling it changes the absolute savings but not the structure.
  • Per-period capacity limit (5000) = 5000
    Chosen so that all demand could be produced in one period; it constrains the feasible production plans.
assumptions (3)
  • ad hoc to paper Learning and forgetting effects follow a linear pattern (uc = 100 - x * p).
    Section 2.4 explicitly assumes linearity to simplify modeling, despite real learning curves usually being nonlinear.
  • domain assumption Setup time does not influence the problem.
    Stated in Section 2: 'assume-se que o tempo de setup não influencia o problema.' This removes a potentially relevant cost driver.
  • domain assumption Single-machine, single-product, deterministic known demand.
    Scope defined in Sections 1 and 3.1; it simplifies the problem but limits generality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modelagem de um Problema de Dimensionamento de Lotes com Demanda Variavel e Deterministica e Efeitos de Learning e Forgetting." pith.science (2026). https://pith.science/paper/BTDLCVYH

@misc{pith2026190810293,
  author       = {Pith},
  title        = {Pith review of: Modelagem de um Problema de Dimensionamento de Lotes com Demanda Variavel e Deterministica e Efeitos de Learning e Forgetting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BTDLCVYH}},
  note         = {Machine review of arXiv:1908.10293}
}
read the original abstract

The main goal of this paper was to analyze the importance that the effects of learning and forgetting might have in a lot-sizing problem. It assumes that the learning curve and the economies of scale are present in several industries yet are, in most cases, not considered when dealing with a lot-sizing problem. The importance of the effects was demonstrated and quantified, showing that there is still space for developments in this field. However, as the problem becomes quadratic, there is a possibility that the current algorithms are not able to solve the problem to optimality. Thus, future improvements in the algorithms may further improve the results. However, the overall results found with current algorithms show that the contribution of a discount from a learning curve can be very considerable, even if it is a minimal amount.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    Custo de setup: os custos fixos necessários à preparação de uma rodada de fabricação. • Exemplos: mão de obra diretamente aplicada na preparação das máquinas; custos dos materiais e acessórios envolvidos na preparação; outros custos indiretos: administrativos, contábeis, etc

    work page

  2. [2]

    • Exemplos: matérias primas; mão de obra diretamente aplicada na produção, tais como limpeza de máquinas, operadores d e máquinas; custo de tempo de máquinas envolvidos

    Custo variável de fabricação de cada produto : Nesse item são considerados os custos dos insumos básicos diretamente empregados no processo produtivo. • Exemplos: matérias primas; mão de obra diretamente aplicada na produção, tais como limpeza de máquinas, operadores d e máquinas; custo de tempo de máquinas envolvidos

    work page

  3. [3]

    Custo de estoque: a posse do estoque tem um custo que, para a indústria, é bastante significativo e normalmente considerado para cada produto por unidade de tempo de armazenagem. • Exemplos: juros de capit al imobilizado; risco de obsolescência do produto; prêmios de seguro, taxas e impostos; perdas por deterioração; despesas com instalações, aluguéis, il...

    work page 2009

  4. [4]

    A compra de matéria-prima e a produção são instantâneas

    work page

  5. [5]

    Não há restrição de capacidade

    work page

  6. [6]

    Demanda é determinística

    work page

  7. [7]

    Uma ordem de produção gera um custo de setup fixo, sem importar o tamanho do lote

    work page

  8. [8]

    Produtos podem ser analizados individualmente

    work page

Show all 42 references
  1. [9]

    Neste caso, não necessariamente precisa-se saber o custo unitário do produto

    Backorders, ou seja, não-atendimento da demanda, não são permitidos. Neste caso, não necessariamente precisa-se saber o custo unitário do produto. Entretanto, é interessante sabê-lo para poder calcular o custo total de produção O Gráfico 2 mostra, na prática, que o custo de se...

    1998

  2. [10]

    Demanda, que é constante e determinada

  3. [11]

    Custo de estoque de um produto em um período

  4. [12]

    Para se encontrar o número de ordens necessárias para atender a demanda em um período, divide -se a demanda total pelo tamanho do lote

    Custo de setup, ou seja, de fazer o pedido ou começar a rodar a produção; Abaixo, seguindo a demonstração do modelo, Y significa o custo total para se produzir/comprar uma quantidade Q de produtos; D é a demanda constante e determinada que ocorre todo período; c é o custo unit...

  5. [13]

    O custo de setup de cada período

  6. [14]

    O custo unitário de produção/compra

  7. [15]

    A demanda de cada período. 2.1.2.2 Regra da quantidade de pedido fixa Esta regra tem por característica pedir sempre a mesma quantidade, tal como era feito no modelo do EOQ, mas com a diferença que, neste caso, os pontos de refazer o pedido são variados. Esta quantidade pode s...

    2000

  8. [16]

    Isso quer dizer que ou o estoque de i-1 é igual a zero, ou o custo de setup do período i é igual a zero de forma que 𝐴𝑖 ∗ 𝐼𝑖−1 = 0

    Um pedido só pode ser feito quando o nível de estoque atinge zero. Isso quer dizer que ou o estoque de i-1 é igual a zero, ou o custo de setup do período i é igual a zero de forma que 𝐴𝑖 ∗ 𝐼𝑖−1 = 0

  9. [17]

    Isso quer dizer que 𝑝𝑖 ≤ ∑ 𝐷𝑖 𝑗 𝑖=1 O modelo pode ser divido em várias etapas, cada uma delas representando um dos períodos que englobam o total

    Existe um limite superior para o número de períodos para os quais uma encomenda durará. Isso quer dizer que 𝑝𝑖 ≤ ∑ 𝐷𝑖 𝑗 𝑖=1 O modelo pode ser divido em várias etapas, cada uma delas representando um dos períodos que englobam o total. Logo, com 𝑖 = 1, 2, 3, … , 𝑗, precisamos de...

    2011

  10. [18]

    definição do problema

  11. [19]

    construção do modelo

  12. [20]

    um modelo mais preciso, mesmo que resolvido de forma aproximada, pode ser bem mais útil que um modelo menos preciso resolvido de forma exata

    implementação do modelo. 3.1 Definição do problema Morabito e Pureza (2010) ressaltam que é nesta fase que são definidos o escopo e os objetivos do modelo. O modelo deve ser o mais próximo possível da realidade, pois, se o modelo conceitual fugir daquilo que que é real, difici...

    2010

  13. [21]

    e YANASSE H

    ARENALES M., ARMENTANO V., MORABITO R. e YANASSE H. Pesquisa Operacional para cursos de Engenharia, 2007

    2007

  14. [22]

    A state -of-the-art review on lot -sizing, European Journal of Operational Research, vol 188, no

    BISKUP D. A state -of-the-art review on lot -sizing, European Journal of Operational Research, vol 188, no. 2, p. 315–329, 2008

    2008

  15. [23]

    and LIN B.M.T

    CHENG T.C.E., DING Q. and LIN B.M.T. A concise survey of lot-sizing with time- dependent processing times. European Journal of Operational Research , 152:1– 13, 2004

    2004

  16. [24]

    DAGUSPTA S., PAPADIMITRIOU C. H. and VAZIRANI U. V. Algorithms, 2006

    2006

  17. [25]

    FERREIRA P. A. V. Planejamento e análise de sistemas de produção. Notas de aula (2006), 2006. In: TAVARES R. F. N. Proposta de solução de problemas de scheduling considerando possibilid ade de terceirização usando a técnica de otimização por colônia de formigas, 2010

    2006

  18. [26]

    e FERNANDES F

    GODINHO M.F. e FERNANDES F. C. F. Planejamento e controle da produção, Editora Atlas, 2010

    2010

  19. [27]

    C., LUNA H

    GOLDBARG M. C., LUNA H. P. L. Otimização combinatória e programação linear: modelos e algoritmos. Rio de Janeiro : Campus, 2000

    2000

  20. [28]

    GONÇALVES, J. F. Gestão de aprovisionamentos , Editora revista Porto: Publindústria, 2000

    2000

  21. [29]

    GONZALEZ J. L. and GONZÁLEZ D. A senior project submitting in partial fulfillment of the requirements for the degree of Bachelors of Science in Industrial Engineering, California Polytechnic State University San Luis Obispo, 2010 38

    2010

  22. [30]

    HAX A. C. and CANDEA D. Production and Operations Manage ment, Prentice- Hall, Englewood Cliffs, NJ, p. 135, 1984

    1984

  23. [31]

    HILL A. V . The encyclopedia of Operations Management, Pearson Education 2011

    2011

  24. [32]

    and EN -MIN F

    LI-YAN W., JI-BO W., DAN W., NA Y., XUE H. and EN -MIN F. Single-machine lot-sizing with a sum -of-processing-time based learning effect and deteriorating jobs. International Journal of Advanced Manufacturing Technology, 2009

    2009

  25. [33]

    MARTINS R. A. Metodologia de pesquisa em Engenharia de Produção e gestão de operações. In: MIGUEL, P. A. C. (Ed. Elselvier), 2010. cap. Abordagens Quantitativa e Qualitativa, p. 45-62

    2010

  26. [34]

    MAURI G. R. e LORENA L. A. N. Decomposições Lagrangeanas para o problema de programação quadrática binária irrestrita, Pesquisa Operacional, vol.29 no.1. Rio de Janeiro, 2009

    2009

  27. [35]

    e PUREZA V

    MORABITO R. e PUREZA V. A Metodologia de pesquisa em Engenharia de Produção e gestão de operações. In: MIGUEL, P. A. C. (Ed. Elselvier), 2010. cap. Modelagem e Simulação, p. 165-194

    2010

  28. [36]

    Lot-sizing: theory, algorithms, and systems

    PINEDO M. Lot-sizing: theory, algorithms, and systems. Prentice -Hall, Up per Saddle River, 2002

    2002

  29. [37]

    and TAYLOR W

    RUSSELL R. and TAYLOR W. B. Operations Management. Edition 4. Publisher Prentice Hall, 2003. 39

    2003

  30. [38]

    SHARIFUL A. S. N. Historical backgrounf of Operations Research, 2012. Disponível em <http://world24info.blogspot.com.br/2012/02/historical-background- of-operation.html> Acesso em Novembro/2013

    2012

  31. [39]

    R., LEE W

    SHIAU Y. R., LEE W. C., WU C. C. and CHANG C. M. Two machine flow shop lot-sizing to minimize mean flow time under simple linear deterioration. International Journal of Advanced Manufacturing Technology. 34:774–782, 2007

    2007

  32. [40]

    e JOHNSTON R

    SLACK N., CHAMBERS S., HARLAND C., HARRISON A. e JOHNSTON R. Administração da Produção. Editora Atlas, 1996

    1996

  33. [41]

    Disponível em: <http://www.sobrapo.org.br/sitesobrapo.htm> Acesso em: Novembro/2013

    Sociedade Brasileira de Pesquisa Operacional (SOBRAPO). Disponível em: <http://www.sobrapo.org.br/sitesobrapo.htm> Acesso em: Novembro/2013

    2013

  34. [42]

    C., LEE W

    WU C. C., LEE W. C., SHIAU Y. R. Minimizing the total weighted completion time on a single machine under linear deterioration. International Journal of Advanced Manufacturing Technology. 33:1237–1243, 2007. 40 APÊNDICE A - EXEMPLO DE MODELO UTILIZADO NO GAMS sets i period / 1,...

    2007

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.